title
stringlengths 1
827
⌀ | uuid
stringlengths 36
36
| pmc_id
stringlengths 6
8
| search_term
stringclasses 18
values | text
stringlengths 0
6.94M
|
|---|---|---|---|---|
Health Communication and Adherence to Noninvasive Ventilation in Chronic Hypercapnic Respiratory Failure | 47cd4e2d-bc5d-47ec-9234-4afd1b82ba38 | 11672160 | Health Communication[mh] | Chronic hypercapnic respiratory failure (CHRF) is a complication arising from various pulmonary, cardiovascular, and neuromuscular disorders. CHRF is characterized by the accumulation of bicarbonate (HCO 3 − ) in the venous system. Dyspnea is the hallmark symptom of CHRF, and it substantially impairs sleep quality, vitality, physical function, and psychosocial well-being. Given that the prevalence of CHRF increases with age and CHRF remains a frequent cause of hospital admissions, the associated disease burden is expected to considerably increase in the coming decade due to population aging. Domiciliary noninvasive ventilation (NIV) is a guideline-recommended management strategy for CHRF that delivers positive airway pressure through a facial mask to improve inspiration. Evidence supports its effectiveness in reducing hospital admissions and mortality. Other studies have demonstrated the positive effects of NIV on disease-specific health-related quality of life (HRQoL). , However, NIV nonadherence remains a major problem due to discomfort or injuries from the mask interface, technical difficulties, air leakage, emotional distress, asynchronous breathing, fear of suffocation, treatment dependence, and social unacceptance. , Previous studies have consistently indicated that NIV nonadherence reduces therapeutic benefits related to symptom control, HRQoL, and even mortality. , Maintaining an adherence threshold of at least 4 hours per day has been proven to improve respiratory function, gas exchange, and symptom control. , Research on improving NIV adherence is limited. To our knowledge, only one trial has explored the use of cognitive-behavioral interventions to improve NIV adherence and health outcomes. However, this trial relied on self-reported adherence, which may have been affected by social desirability and recall bias. In addition, the need for frequent sessions with a clinical psychologist limited the intervention’s feasibility, as indicated by the high attrition rate. Another shortcoming was the absence of strategies to promote long-term motivation, which is critical for sustained NIV adherence. The information-motivation-behavioral (IMB) skills model provides a framework for addressing complex factors contributing to poor NIV adherence. In addition to meeting the substantial informational needs of patients with CHRF to support health behaviors, the model can enhance both personal and social motivation. Personal motivation involves cultivating a positive attitude toward NIV use, whereas social motivation focuses on improving patients’ perceived social support and social acceptance in adhering to treatment. Furthermore, the IMB model incorporates a behavioral component that extends beyond treatment-related skill development to increase self-efficacy and perceived control in changing health behaviors. This focus is particularly relevant for addressing maladaptive emotional responses to NIV use. The objective of this randomized clinical trial was to examine the effects of an intervention based on the IMB skills model (ie, the IMB-NIV program) on NIV adherence (primary outcome), serum bicarbonate (HCO 3 − ) levels, patient-reported health outcomes (sleep quality and HRQoL), and health service use (emergency department [ED] and hospital admissions). Study Design This 2-group, multisite, assessor-blinded, randomized clinical trial was conducted in the respiratory clinics of 2 regional hospitals in Hong Kong. The study protocol was approved by the Clinical Research Ethics Committee and adhered to the Declaration of Helsinki. All participants provided written informed consent. The trial followed the Consolidated Standards of Reporting Trials ( CONSORT ) reporting guideline. The trial protocol is available in . Participants A research assistant identified potential participants by screening the electronic health records of all patients who attended the respiratory clinics between January 2022 and March 2023. Patients were eligible if they were diagnosed with CHRF, had a partial pressure of carbon dioxide, arterial (Pa co 2 ), level of 7 kPa (52.5 mm Hg) or greater for at least 4 weeks, had been prescribed domiciliary NIV for 4 or more weeks, and were classified as nonadherent to domiciliary NIV (defined as using NIV for <4 h/night for >70% of days in the last 2 weeks). Eligible patients also needed to provide informed consent to participate. Patients with known psychiatric disorders (except anxiety and depression), diseases with a life expectancy of 1 year or shorter, and active malignant neoplasms were excluded. The effect size was estimated based on our pilot study findings. The pilot study indicated that the IMB-NIV program could improve treatment adherence by nearly 40 percentage points (IMB-NIV vs usual care: 70.6% vs 31.6%) at the 12-month end point. For this trial, we conservatively assumed a between-group difference of 25 percentage points in treatment adherence, with an improvement of 45 percentage points and 20 percentage points in the IMB-NIV and usual care groups, respectively. Power analysis performed using Pass version 14.0 (NCSS) indicated that 62 participants per group would be required to achieve 80% power at a 2-sided 95% CI, accounting for 15% attrition. Participants were randomly assigned to receive either the IMB-NIV program or usual care in a 1:1 ratio through permuted block randomization using block sizes of 4, 6, and 8 with computer-generated sequences. The research assistant used sequentially numbered, opaque, sealed envelopes to ensure allocation concealment. Interventions IMB-NIV program The 6-week IMB-NIV program used a hybrid approach, incorporating home visits, telecare, and clinic visits to improve accessibility for patients with CHRF who are more likely to have decreased activity tolerance. The program was delivered by 3 registered nurses who had strong experience in health counseling. They had received intensive training on using the IMB approach to improve NIV and advanced respiratory care. The training was delivered by advanced practice nurses in respiratory and medical care and nursing academicians who have professional portfolios in using behavioral therapies and motivational interviewing to empower and motivate self-care for chronic disease management. The content covered disease-related knowledge, NIV care and self-care, the IMB model constructs, the role of personal and social motivations in shaping health behaviors, goal-oriented empowerment and motivational strategies, and patient engagement in home visit and telecare. A standardized intervention protocol and toolkit (including information package, goal-setting portfolio, and telephone follow-up documents) were used to guide the practice. For each patient, the same nurse delivered the whole IMB model to ensure continuity of care. They were also supported by a multidisciplinary team, including geriatricians and advanced practice nurses, for program development and implementation. The core of the IMB-NIV program was a person-centered approach aimed at enhancing knowledge and skills for managing CHRF while cultivating a positive attitude and social motivation to promote long-term adherence to domiciliary NIV. For further details, please refer to the study protocol . The program began with a 60-minute home visit during which the nurse assessed the participants’ self-care needs related to CHRF and NIV management. Health education was provided to increase the participants’ awareness of their nonadherent behaviors as well as the underlying misbeliefs and attitudes toward self-care. The potential detrimental health consequences of nonadherence were highlighted to motivate the participants to set self-directed goals and develop an action plan for improving NIV adherence. The goal categories set for the patients are listed in eTable 1 in . A nurse-patient partnership approach was used to address the participants’ concerns, confront negative thoughts, and identify personal and social support resources to facilitate behavioral changes. Skill training focused on proper NIV handling and complication prevention, with an emphasis on optimizing self-care in the home environment. The visit concluded with brainstorming to anticipate barriers to goal attainment and explore potential solutions. If available, family members were involved to provide support, and a telephone consultation hotline, available during office hours, was provided. Two 20-mintue televisits were scheduled in the second and fourth weeks to monitor progress and review goals. Health counseling was provided to address barriers to NIV adherence and any evolving self-care challenges. The program concluded with a 30-minute follow-up clinic visit aimed at reinforcing long-term behavioral changes that support NIV adherence. Health communication during this visit encouraged participants to reflect on improvements in their NIV self-care and the resulting perceived health benefits. Further behavioral skill training, positive reinforcement of the prognostic benefits of optimal NIV adherence, and practical strategies for integrating self-care into daily routines were provided. Usual Care Usual care consisted of regular medical follow-ups at the respiratory clinic where a nursing team provided support for troubleshooting problems related to domiciliary NIV use and delivered relevant health education as needed. The IMB group also received this standard care during their medical follow-ups. Although the use of attention-placebo allows a more stringent evaluation of the active components of the tested intervention, it was challenging to ensure its similar dose as the IMB-NIV program (ie, approximately 2-hour interaction in 4 nurse-patient encounters), with the content perceived as credible by the patients but not related to the self-care and quality-of-life outcomes. Given the education-supportive component of the usual care further limited the content design of the attention placebo, usual care was adopted as the control condition. Outcome Measures All outcomes were assessed at baseline, at IMB-NIV program completion (ie, program exit at the seventh week), and at 3, 6, and 12 months. Data were collected by a nurse blinded to the randomization, either in the clinic or via phone call. For the primary outcome, adherence data, including daily NIV usage over the past 2 weeks, were retrieved from the NIV machine. Adherence was defined as more than 4 hours of nighttime use for more than 70% of days or a mean daily use of more than 5 hours. Secondary outcomes included sleep quality, measured using the Chinese Pittsburgh Sleep Quality Index (CPSQI), and HRQoL, measured using the Chinese Severe Respiratory Insufficiency Questionnaire (CSRI). Both scales include subdomain scores to assess specific aspects of sleep function and the impact of CHRF on physical and psychosocial well-being. Venus HCO 3 − levels were measured at baseline and at 3 and 6 months. Data on respiratory-related ED admissions and hospital admissions were retrieved from electronic health records over the 12-month period. Statistical Analysis A generalized estimating equation (GEE) model was used to examine the effects of the IMB-NIV program on daily NIV usage, CPSQI scores, CSRI scores, and serum HCO 3 − levels following the intention-to-treat principle. The model was adjusted for potential confounders, specifically baseline clinical and demographic characteristics with a 2-sided P < .25. The time × group interaction term in the GEE model was included to compare mean changes in primary and secondary outcomes between the groups from baseline to the various time points across the 12-month period, and effect sizes (Cohen d ) were calculated. The relative ratios of becoming NIV-adherent (ie, >4 hours of nighttime use for >70% of days over the past 4 weeks) were computed for each time point and compared between the groups. Negative binomial regression was used to determine between-group differences in ER visits and hospital admissions because these data were likely over-dispersed in the sample. Cox proportional hazards regression analysis was performed to assess between-group differences in time-to-event outcomes, and corresponding hazard ratios were calculated. Statistical analyses were performed using SPSS software version 29.0 (IBM Corp), with 2-sided α = .05 as the threshold for statistical significance. This 2-group, multisite, assessor-blinded, randomized clinical trial was conducted in the respiratory clinics of 2 regional hospitals in Hong Kong. The study protocol was approved by the Clinical Research Ethics Committee and adhered to the Declaration of Helsinki. All participants provided written informed consent. The trial followed the Consolidated Standards of Reporting Trials ( CONSORT ) reporting guideline. The trial protocol is available in . A research assistant identified potential participants by screening the electronic health records of all patients who attended the respiratory clinics between January 2022 and March 2023. Patients were eligible if they were diagnosed with CHRF, had a partial pressure of carbon dioxide, arterial (Pa co 2 ), level of 7 kPa (52.5 mm Hg) or greater for at least 4 weeks, had been prescribed domiciliary NIV for 4 or more weeks, and were classified as nonadherent to domiciliary NIV (defined as using NIV for <4 h/night for >70% of days in the last 2 weeks). Eligible patients also needed to provide informed consent to participate. Patients with known psychiatric disorders (except anxiety and depression), diseases with a life expectancy of 1 year or shorter, and active malignant neoplasms were excluded. The effect size was estimated based on our pilot study findings. The pilot study indicated that the IMB-NIV program could improve treatment adherence by nearly 40 percentage points (IMB-NIV vs usual care: 70.6% vs 31.6%) at the 12-month end point. For this trial, we conservatively assumed a between-group difference of 25 percentage points in treatment adherence, with an improvement of 45 percentage points and 20 percentage points in the IMB-NIV and usual care groups, respectively. Power analysis performed using Pass version 14.0 (NCSS) indicated that 62 participants per group would be required to achieve 80% power at a 2-sided 95% CI, accounting for 15% attrition. Participants were randomly assigned to receive either the IMB-NIV program or usual care in a 1:1 ratio through permuted block randomization using block sizes of 4, 6, and 8 with computer-generated sequences. The research assistant used sequentially numbered, opaque, sealed envelopes to ensure allocation concealment. IMB-NIV program The 6-week IMB-NIV program used a hybrid approach, incorporating home visits, telecare, and clinic visits to improve accessibility for patients with CHRF who are more likely to have decreased activity tolerance. The program was delivered by 3 registered nurses who had strong experience in health counseling. They had received intensive training on using the IMB approach to improve NIV and advanced respiratory care. The training was delivered by advanced practice nurses in respiratory and medical care and nursing academicians who have professional portfolios in using behavioral therapies and motivational interviewing to empower and motivate self-care for chronic disease management. The content covered disease-related knowledge, NIV care and self-care, the IMB model constructs, the role of personal and social motivations in shaping health behaviors, goal-oriented empowerment and motivational strategies, and patient engagement in home visit and telecare. A standardized intervention protocol and toolkit (including information package, goal-setting portfolio, and telephone follow-up documents) were used to guide the practice. For each patient, the same nurse delivered the whole IMB model to ensure continuity of care. They were also supported by a multidisciplinary team, including geriatricians and advanced practice nurses, for program development and implementation. The core of the IMB-NIV program was a person-centered approach aimed at enhancing knowledge and skills for managing CHRF while cultivating a positive attitude and social motivation to promote long-term adherence to domiciliary NIV. For further details, please refer to the study protocol . The program began with a 60-minute home visit during which the nurse assessed the participants’ self-care needs related to CHRF and NIV management. Health education was provided to increase the participants’ awareness of their nonadherent behaviors as well as the underlying misbeliefs and attitudes toward self-care. The potential detrimental health consequences of nonadherence were highlighted to motivate the participants to set self-directed goals and develop an action plan for improving NIV adherence. The goal categories set for the patients are listed in eTable 1 in . A nurse-patient partnership approach was used to address the participants’ concerns, confront negative thoughts, and identify personal and social support resources to facilitate behavioral changes. Skill training focused on proper NIV handling and complication prevention, with an emphasis on optimizing self-care in the home environment. The visit concluded with brainstorming to anticipate barriers to goal attainment and explore potential solutions. If available, family members were involved to provide support, and a telephone consultation hotline, available during office hours, was provided. Two 20-mintue televisits were scheduled in the second and fourth weeks to monitor progress and review goals. Health counseling was provided to address barriers to NIV adherence and any evolving self-care challenges. The program concluded with a 30-minute follow-up clinic visit aimed at reinforcing long-term behavioral changes that support NIV adherence. Health communication during this visit encouraged participants to reflect on improvements in their NIV self-care and the resulting perceived health benefits. Further behavioral skill training, positive reinforcement of the prognostic benefits of optimal NIV adherence, and practical strategies for integrating self-care into daily routines were provided. Usual Care Usual care consisted of regular medical follow-ups at the respiratory clinic where a nursing team provided support for troubleshooting problems related to domiciliary NIV use and delivered relevant health education as needed. The IMB group also received this standard care during their medical follow-ups. Although the use of attention-placebo allows a more stringent evaluation of the active components of the tested intervention, it was challenging to ensure its similar dose as the IMB-NIV program (ie, approximately 2-hour interaction in 4 nurse-patient encounters), with the content perceived as credible by the patients but not related to the self-care and quality-of-life outcomes. Given the education-supportive component of the usual care further limited the content design of the attention placebo, usual care was adopted as the control condition. The 6-week IMB-NIV program used a hybrid approach, incorporating home visits, telecare, and clinic visits to improve accessibility for patients with CHRF who are more likely to have decreased activity tolerance. The program was delivered by 3 registered nurses who had strong experience in health counseling. They had received intensive training on using the IMB approach to improve NIV and advanced respiratory care. The training was delivered by advanced practice nurses in respiratory and medical care and nursing academicians who have professional portfolios in using behavioral therapies and motivational interviewing to empower and motivate self-care for chronic disease management. The content covered disease-related knowledge, NIV care and self-care, the IMB model constructs, the role of personal and social motivations in shaping health behaviors, goal-oriented empowerment and motivational strategies, and patient engagement in home visit and telecare. A standardized intervention protocol and toolkit (including information package, goal-setting portfolio, and telephone follow-up documents) were used to guide the practice. For each patient, the same nurse delivered the whole IMB model to ensure continuity of care. They were also supported by a multidisciplinary team, including geriatricians and advanced practice nurses, for program development and implementation. The core of the IMB-NIV program was a person-centered approach aimed at enhancing knowledge and skills for managing CHRF while cultivating a positive attitude and social motivation to promote long-term adherence to domiciliary NIV. For further details, please refer to the study protocol . The program began with a 60-minute home visit during which the nurse assessed the participants’ self-care needs related to CHRF and NIV management. Health education was provided to increase the participants’ awareness of their nonadherent behaviors as well as the underlying misbeliefs and attitudes toward self-care. The potential detrimental health consequences of nonadherence were highlighted to motivate the participants to set self-directed goals and develop an action plan for improving NIV adherence. The goal categories set for the patients are listed in eTable 1 in . A nurse-patient partnership approach was used to address the participants’ concerns, confront negative thoughts, and identify personal and social support resources to facilitate behavioral changes. Skill training focused on proper NIV handling and complication prevention, with an emphasis on optimizing self-care in the home environment. The visit concluded with brainstorming to anticipate barriers to goal attainment and explore potential solutions. If available, family members were involved to provide support, and a telephone consultation hotline, available during office hours, was provided. Two 20-mintue televisits were scheduled in the second and fourth weeks to monitor progress and review goals. Health counseling was provided to address barriers to NIV adherence and any evolving self-care challenges. The program concluded with a 30-minute follow-up clinic visit aimed at reinforcing long-term behavioral changes that support NIV adherence. Health communication during this visit encouraged participants to reflect on improvements in their NIV self-care and the resulting perceived health benefits. Further behavioral skill training, positive reinforcement of the prognostic benefits of optimal NIV adherence, and practical strategies for integrating self-care into daily routines were provided. Usual care consisted of regular medical follow-ups at the respiratory clinic where a nursing team provided support for troubleshooting problems related to domiciliary NIV use and delivered relevant health education as needed. The IMB group also received this standard care during their medical follow-ups. Although the use of attention-placebo allows a more stringent evaluation of the active components of the tested intervention, it was challenging to ensure its similar dose as the IMB-NIV program (ie, approximately 2-hour interaction in 4 nurse-patient encounters), with the content perceived as credible by the patients but not related to the self-care and quality-of-life outcomes. Given the education-supportive component of the usual care further limited the content design of the attention placebo, usual care was adopted as the control condition. All outcomes were assessed at baseline, at IMB-NIV program completion (ie, program exit at the seventh week), and at 3, 6, and 12 months. Data were collected by a nurse blinded to the randomization, either in the clinic or via phone call. For the primary outcome, adherence data, including daily NIV usage over the past 2 weeks, were retrieved from the NIV machine. Adherence was defined as more than 4 hours of nighttime use for more than 70% of days or a mean daily use of more than 5 hours. Secondary outcomes included sleep quality, measured using the Chinese Pittsburgh Sleep Quality Index (CPSQI), and HRQoL, measured using the Chinese Severe Respiratory Insufficiency Questionnaire (CSRI). Both scales include subdomain scores to assess specific aspects of sleep function and the impact of CHRF on physical and psychosocial well-being. Venus HCO 3 − levels were measured at baseline and at 3 and 6 months. Data on respiratory-related ED admissions and hospital admissions were retrieved from electronic health records over the 12-month period. A generalized estimating equation (GEE) model was used to examine the effects of the IMB-NIV program on daily NIV usage, CPSQI scores, CSRI scores, and serum HCO 3 − levels following the intention-to-treat principle. The model was adjusted for potential confounders, specifically baseline clinical and demographic characteristics with a 2-sided P < .25. The time × group interaction term in the GEE model was included to compare mean changes in primary and secondary outcomes between the groups from baseline to the various time points across the 12-month period, and effect sizes (Cohen d ) were calculated. The relative ratios of becoming NIV-adherent (ie, >4 hours of nighttime use for >70% of days over the past 4 weeks) were computed for each time point and compared between the groups. Negative binomial regression was used to determine between-group differences in ER visits and hospital admissions because these data were likely over-dispersed in the sample. Cox proportional hazards regression analysis was performed to assess between-group differences in time-to-event outcomes, and corresponding hazard ratios were calculated. Statistical analyses were performed using SPSS software version 29.0 (IBM Corp), with 2-sided α = .05 as the threshold for statistical significance. Study Participants Of 6361 patients who consecutively attended the respiratory clinics at 2 regional hospitals, 124 patients (mean [SD] age, 70.1 [8.0] years; 67 [54.0%] female) were recruited, with 62 randomized to the IMB-NIV group and 62 to the usual care group . Among them, 31 (25.0%) and 64 (51.6%) patients had chronic obstructive airway disease and sleep apnea, respectively. The mean Charlson Comorbidity Index was greater than 4.0 for both groups, indicating a high level of chronic disease burden. Approximately 30% of participants had been hospitalized at least once in the past year. The 2 study groups differed in age, smoking habits, and comorbidity burden; these variables were controlled for in the outcome evaluation. A total of 111 patients completed the 12-month follow-up assessment, with completion rates of 90.3% (56 of 62) and 88.7% (55 of 62) in the IMB-NIV and usual care groups, respectively . No adverse effects related to the IBM-NIV program were reported. Primary Outcome Based on the criterion of nighttime use for more than 4 hours on more than 70% of days, the IMB-NIV group was more likely to adhere to NIV use (32 participants [51.6%] at 7 weeks; 38 [61.3%] at 12 months) compared with the usual care group (10 participants [16.1%] at 7 weeks; 14 [27.4%] at 12 months) at program exit (odds ratio, 3.13; 95% CI, 1.69-5.88). This positive effect of the IMB-NIV program was sustained through the 12-month end point . In addition, the IMB-NIV group showed a significantly greater increase in mean daily NIV use than did the usual care group at program exit (group × time effect: B = 2.09; 95% CI, 1.38-2.81; P < .001). This positive change was maintained through the 12-month evaluation (Cohen d range, 0.724-0.997). Secondary Outcomes The IMB-NIV group demonstrated greater improvements in sleep quality and HRQoL than did the usual care group . In particular, the IMB-NIV group showed significantly greater improvements in the overall score and the 7 domain scores of the CPSQI starting at the 3-month follow-up (eTable 2 in ). These effects were sustained through the 12-month end point (Cohen d range, 0.59-0.92). Similarly, greater improvements in the CSRI summary and domain scores were observed in the IMB-NIV group throughout the evaluation period, from program exit to the 12-month follow-up (Cohen d range, 0.53-0.93). However, no significant differences in serum HCO 3 − levels were noted between the study groups. The IMB-NIV program significantly reduced unplanned ED visits during the 12-month evaluation period. Negative binomial regression indicated that the incidence rate ratio (IRR) for ER admissions in the IMB-NIV group (27 admissions) vs the usual care group (66 admissions) was 0.47 (95% CI, 0.26-0.84; P = .01). However, there was no difference in hospital admissions between the IMB-NIV and usual care groups (44 vs 65; IRR, 0.69; 95% CI, 0.40-1.19; P = .18). and show the survival curves for respiratory-related ED visits and hospital admissions, respectively. Cox proportional hazards regression analysis indicated that the IMB-NIV program significantly delayed the time to respiratory-related ED visits, with a hazard ratio of 0.51 (95% CI, 0.28-0.95) but did not significantly affect time to hospital admissions (hazard ratio, 0.69; 95% CI, 0.38-1.24). Subgroup Analysis As the effects of IMB-NIV program on NIV adherence and hospital services use may vary among patients with different CHRF etiologies (eg, chronic obstructive pulmonary disorder, obstructive sleep apnea alone, obstructive sleep apnea comorbid with other respiratory disorder), subgroup analysis was conducted. The IMB-NIV program significantly improved NIV adherence for all 3 three subgroups (eTables 3-5 in ). As for health service use at 12 months, positive intervention effects were observed in the COPD subgroup, who had greater reduction in ED admissions (IRR, 0.23; 95% CI, 0.09-0.62) and hospital admissions (IRR, 0.40; 95% CI, 0.16-0.99). For the other subgroups, no treatment effect on hospital service use was noted. Of 6361 patients who consecutively attended the respiratory clinics at 2 regional hospitals, 124 patients (mean [SD] age, 70.1 [8.0] years; 67 [54.0%] female) were recruited, with 62 randomized to the IMB-NIV group and 62 to the usual care group . Among them, 31 (25.0%) and 64 (51.6%) patients had chronic obstructive airway disease and sleep apnea, respectively. The mean Charlson Comorbidity Index was greater than 4.0 for both groups, indicating a high level of chronic disease burden. Approximately 30% of participants had been hospitalized at least once in the past year. The 2 study groups differed in age, smoking habits, and comorbidity burden; these variables were controlled for in the outcome evaluation. A total of 111 patients completed the 12-month follow-up assessment, with completion rates of 90.3% (56 of 62) and 88.7% (55 of 62) in the IMB-NIV and usual care groups, respectively . No adverse effects related to the IBM-NIV program were reported. Based on the criterion of nighttime use for more than 4 hours on more than 70% of days, the IMB-NIV group was more likely to adhere to NIV use (32 participants [51.6%] at 7 weeks; 38 [61.3%] at 12 months) compared with the usual care group (10 participants [16.1%] at 7 weeks; 14 [27.4%] at 12 months) at program exit (odds ratio, 3.13; 95% CI, 1.69-5.88). This positive effect of the IMB-NIV program was sustained through the 12-month end point . In addition, the IMB-NIV group showed a significantly greater increase in mean daily NIV use than did the usual care group at program exit (group × time effect: B = 2.09; 95% CI, 1.38-2.81; P < .001). This positive change was maintained through the 12-month evaluation (Cohen d range, 0.724-0.997). The IMB-NIV group demonstrated greater improvements in sleep quality and HRQoL than did the usual care group . In particular, the IMB-NIV group showed significantly greater improvements in the overall score and the 7 domain scores of the CPSQI starting at the 3-month follow-up (eTable 2 in ). These effects were sustained through the 12-month end point (Cohen d range, 0.59-0.92). Similarly, greater improvements in the CSRI summary and domain scores were observed in the IMB-NIV group throughout the evaluation period, from program exit to the 12-month follow-up (Cohen d range, 0.53-0.93). However, no significant differences in serum HCO 3 − levels were noted between the study groups. The IMB-NIV program significantly reduced unplanned ED visits during the 12-month evaluation period. Negative binomial regression indicated that the incidence rate ratio (IRR) for ER admissions in the IMB-NIV group (27 admissions) vs the usual care group (66 admissions) was 0.47 (95% CI, 0.26-0.84; P = .01). However, there was no difference in hospital admissions between the IMB-NIV and usual care groups (44 vs 65; IRR, 0.69; 95% CI, 0.40-1.19; P = .18). and show the survival curves for respiratory-related ED visits and hospital admissions, respectively. Cox proportional hazards regression analysis indicated that the IMB-NIV program significantly delayed the time to respiratory-related ED visits, with a hazard ratio of 0.51 (95% CI, 0.28-0.95) but did not significantly affect time to hospital admissions (hazard ratio, 0.69; 95% CI, 0.38-1.24). As the effects of IMB-NIV program on NIV adherence and hospital services use may vary among patients with different CHRF etiologies (eg, chronic obstructive pulmonary disorder, obstructive sleep apnea alone, obstructive sleep apnea comorbid with other respiratory disorder), subgroup analysis was conducted. The IMB-NIV program significantly improved NIV adherence for all 3 three subgroups (eTables 3-5 in ). As for health service use at 12 months, positive intervention effects were observed in the COPD subgroup, who had greater reduction in ED admissions (IRR, 0.23; 95% CI, 0.09-0.62) and hospital admissions (IRR, 0.40; 95% CI, 0.16-0.99). For the other subgroups, no treatment effect on hospital service use was noted. To our knowledge, this is the first fully powered randomized clinical trial to evaluate the effectiveness of a behavioral strategy aimed at promoting NIV adherence in patients with CHRF. Based on the IMB skills-based model, the IMB-NIV program significantly improved adherence to domiciliary NIV use, and this effect was maintained for up to 12 months. The program also provided extended benefits in sleep quality, symptom control, physical function, and psychosocial adaptation, further reinforcing the therapeutic value of NIV adherence in patients with CHRF. Although no significant improvements were observed in venous HCO 3 − levels or respiratory-related hospital admissions, the reduction in ED visits and the delayed time to these events indicate the program’s additional impact on health service utilization. The hybrid approach, combining in-person visits with telecare, contributed to the program’s full attendance, indicating its feasibility for clinical application. The IMB model suggests the presence of mutual feedback loops between adherence information, motivation, and actual behavioral changes. The 6-week program, which included televisits and clinic visits, offered an effective platform to reinforce these 3 core components continuously. This likely explains the sustained improvement in NIV adherence observed up to the 12-month end point. Conversely, a study that used cognitive-behavioral therapy without a motivational component was unable to maintain the initial increase in NIV adherence. Three major differences in the intervention design compared with our study may explain the variation in the treatment effect. First, our health communication contrasted patients’ self-reported decision-making processes regarding NIV use with corresponding professional perspectives. This interactive, person-centered approach helped increase patients’ self-awareness of any misconceptions or maladaptive behaviors related to disease management. However, the study using CBT adopted a more 1-way communication which may only improve the participants’ cognitive understanding of disease and treatment. Second, our study tried to increase intrinsic motivation to change by providing patients with information on the potential detrimental health consequences associated with their maladaptive behaviors. This arrangement created what is called a fear appeal, which was shown to effectively influence attitudes and intentions toward behavioral changes. In contrast, the CBT study focused on using cognitive restructuring and relaxation to alleviate the anxiety and emotional arousal toward the NIV use. This may enhance extrinsic motivation which is known to be less powerful than intrinsic motivation to secure long-term commitment. Third, even though both studies taught patients behavioral strategies to increase NIV use, our study initiated a patient-directed goal-setting process to tailor behavioral changes, with particular focus placed on combating the common barriers to NIV adherence (including NIV-induced discomfort, emotional response to asynchronized breathing, fear of dependency, and social rejection). Such an experience of goal attainment would increase one’s sense of self-efficacy and self-control, which further motivate positive behavioral changes. , In fact, our findings are consistent with those of a previous study that validated the IMB skills-based model in addressing the complex causes of poor self-care in chronic disease management. Another factor contributing to the long-term improvement in NIV adherence may be enhanced sleep quality, respiratory symptom control, physical functioning, and psychosocial well-being experienced by participants in the IMB-NIV group. These improvements likely reinforced the health communication provided by the nurse regarding the therapeutic benefits of increased NIV adherence. The perceived health gains may have created a positive feedback loop, motivating sustained behavioral changes. Among the various patient-reported outcomes, sleep quality was not significantly improved until the 3-month end point, indicating that these patients may need time to adjust to sleep disturbances associated with increased NIV use. These findings indicate the need for additional support during the initial stages of behavioral changes to address negative thoughts and provide encouragement. The televisits effectively served this role, helping to sustain adherence behaviors. Another notable finding is related to the effects of the IMB-NIV program on health service use. The program significantly delayed the time to respiratory-related ED visits and reduced the number of episodes, but it did not affect hospital admissions. This outcome may be attributable to the less developed system of primary care physicians in Hong Kong, where patients with more debilitating CHRF are more likely to seek ED services during disease exacerbation. The results revealed that the control group was more likely to use ED services even when symptom worsening did not warrant hospital admission. This reduction in unnecessary ED visits in the IMB-NIV group may be associated with improved tidal volume and more favorable breathing patterns resulting from increased NIV adherence. NIV delivers positive airway pressure, which increases lung volume, reduces hyperinflation, and enhances the elastic recoil of the chest wall. These effects, in turn, improve carbon dioxide elimination and reduce perceived breathlessness. Perceived breathlessness was identified as a strong trigger for seeking emergency hospital care. However, because the higher ED attendance in the usual care group did not lead to a significant difference in hospital admissions between the groups, this increased health service use may not have been necessary to address a severe pathophysiological issue. Thus, the IMB-NIV program likely reduced health service use by improving NIV adherence. An alternative explanation for the lack of program effect on hospital admission may be related to the mixed CHRF etiologies in our sample. Indeed, the subgroup analysis indicated that the IMB-NIV program significantly reduced both ED admission and hospital admission of those with COPD. This finding aligns with a systematic review, which reported the positive effects of NIV on hypercapnic patients with COPD. Nevertheless, caution is needed in interpreting the results of the underpowered subgroup analysis for other etiology groups. Future clinical trials on improving NIV adherence need to consider the mixed etiology of the CHRF patients. The IMB-NIV program did not improve serum HCO 3 − levels. This may be because reductions in venous HCO 3 − levels are more likely to be observed when domiciliary NIV is actively in use. , Because blood samples were collected in the clinic, venous HCO 3 − may not have been adequately sensitive to capture the benefits of NIV adherence. Instead, it may be more appropriate to assess ventilatory responses, such as forced expiratory volume in 1 second and total tidal volume, to better reflect these effects. Limitations The study has several limitations. First, the IMB model emphasizes enhancing social motivation by optimizing patients’ social support. However, we did not exclude patients living in singleton or doubleton households, which may have limited the effectiveness of health counseling and threatened the internal validity of our findings. Second, the follow-up period was limited to 12 months, and the small number of patients who died (n = 5) prevented meaningful survival analysis of mortality outcomes. Third, because of the limitations of the NIV machine, we could only retrieve usage data in 2-week intervals. The use of this short reference period necessitates caution when interpreting the study’s findings. Fourth, lack of an attention-placebo as the control may threaten the internal validity of the findings as the effects of the attention from the nurse on the study outcomes cannot be precluded. Fifth, the involvement of a small group of registered nurses to deliver the IMB model also limits the generalizability of the study findings. Additionally, as we did not record the specific time spent with the patients in the intervention arm, its effect cannot be adjusted in the outcome evaluation. The study has several limitations. First, the IMB model emphasizes enhancing social motivation by optimizing patients’ social support. However, we did not exclude patients living in singleton or doubleton households, which may have limited the effectiveness of health counseling and threatened the internal validity of our findings. Second, the follow-up period was limited to 12 months, and the small number of patients who died (n = 5) prevented meaningful survival analysis of mortality outcomes. Third, because of the limitations of the NIV machine, we could only retrieve usage data in 2-week intervals. The use of this short reference period necessitates caution when interpreting the study’s findings. Fourth, lack of an attention-placebo as the control may threaten the internal validity of the findings as the effects of the attention from the nurse on the study outcomes cannot be precluded. Fifth, the involvement of a small group of registered nurses to deliver the IMB model also limits the generalizability of the study findings. Additionally, as we did not record the specific time spent with the patients in the intervention arm, its effect cannot be adjusted in the outcome evaluation. This study demonstrated the positive effect of a behavioral strategy based on the IMB model in improving NIV adherence in patients with CHRF. The extended benefits of the IMB-NIV program for sleep quality, HRQoL, and ED service use, along with the high fidelity of the hybrid approach, suggest its efficacy to enhance CHRF management. Future studies should evaluate its effects on ventilator parameters and validate its overall clinical effectiveness and cost-effectiveness. |
Preoperative estimation of retinal hole location using ultra-wide-field imaging | 7968a6e6-a0e0-4686-a1e9-b015d2903d3b | 10512843 | Ophthalmology[mh] | Rhegmatogenous retinal detachment (RRD) is the most common type of retinal detachment; blindness may occur in the affected eye unless surgery is performed. The primary surgical methods include scleral buckling (SB), pars plana vitrectomy (PPV), and pneumatic retinopexy (PR) . SB is a traditional technique for RRD and can be used as a major surgical method with a high success rate and few complications. SB can also be used as an auxiliary treatment for vitrectomy. In many cases, SB is an effective treatment for RRD, especially for RRD secondary to retinal lattice degeneration with atrophic retinal holes . SB also has a lower incidence of iatrogenic breaks and secondary cataracts than PPV . Despite this, SB causes less disturbance to the vitreous body . However, in recent years, the use of SB in clinical practice has been declining, and it is largely being replaced by PPV . Therefore, some opinions state that vitrectomy should be performed more carefully, and SB should be reassessed . The reasons for the decrease in clinical SB use may be multifaceted; the rapid development of vitrectomy instruments and techniques may be the main reason, but the learning difficulty of SB surgery may also cause it. Accurate localization of retinal holes during surgery determines the success of SB. Currently, most intraoperative localization methods require an indirect ophthalmoscope to observe the hole. The difficulty in mastering this technique is an important reason for the long learning curve of SB. New SB surgery, which is assisted by chandelier endoillumination, makes it easier for surgeons to observe the fundus and locate holes, but it also requires additional scleral incisions and instruments . Although traditional surgery requires surgical experience and skill, it is still necessary for clinicians to master the procedure because its long-term safety profile has been proven. The Optos ultra-wide-field (UWF) imaging system is the latest generation of fundus examination instruments with non-contact and ultra-wide-field characteristics . Surgeons can obtain abundant information about the retina, including data on the number, shape, and approximate position of retinal holes, using preoperative UWF imaging. However, there are few reports on the accurate localization of retinal holes using UWF imaging . The present work aimed to study the feasibility of locating retinal holes using UWF fundus photography before surgery. We expect this to assist surgeons in the localization of retinal holes and shorten the learning curve of SB. This prospective cohort study included patients diagnosed with RRD and treated with SB in the Department of Ophthalmology of the Second Hospital of Hebei Medical University between November 2020 and November 2021. Patients with retinal holes detected by UWF fundus photography were included. The following patients were excluded: those who had undergone or needed cataract surgery and/or vitrectomy; those with PVR C2 or above, giant retinal tears, or dialysis of the ora serrata; and those with RRD where holes could be detected using traditional methods (binocular indirect ophthalmoscope and a three-mirror lens) but not by UWF images. A total of 21 eyes from 21 patients were included. This study adhered to the tenets of the Declaration of Helsinki, and the protocols were approved by the Ethical Committee of The Second Hospital of Hebei Medical University. Written informed consent was obtained from the patients enrolled in the study. Data collection and analysis Preoperative assessment Ultra-wide-angle fundus images were taken using an Optos Ultra-Widefield Fundus Camera (Optos, Dunfermline, UK, 200Tx). These images were taken 1 day preoperatively, 1 week postoperatively, and 1 month postoperatively. Pupils were dilated to the greatest extent before taking images. The images of the central, superior, inferior, nasal, and temporal retina of each eye were taken by the same experienced technician. During the process, patients’ eyelids were lifted by the technician, and they had to try their best to stare in the corresponding directions. Images of the retinal holes were taken in a similar manner. All images were taken when the ‘green in-focus’ signal was obtained on the Optos machine. The images of the inner surface of the eye, which is nearly spherical, cannot be mapped to a flat surface without distortion. The size and shape of some structures can be greatly distorted with further eccentricity in taking images. Therefore, we standardized the images to make them more accurate. We exported images from the Optos and converted them to grayscale in PhotoshopCS3 (Adobe, Inc., San Jose, CA). Images centred on the posterior portion of the eye were inspected for sharpness and lack of systemic distortion across the image. We selected the best image and considered it to be the best image of the periphery. To standardize the resultant images, histogram stretching was performed. Peripheral images were transformed using elastic deformation, and the posterior portions of peripheral images matched the base reference images. Warped images were merged into a montage . The UWF imaging range can reach the ora serrata through eye position guidance. Moreover, pupillary dilation and eyelid lifting can enlarge the imaging range . Therefore, we assumed that the ora serrata corresponded to the blood-free zone at the edge of the obtained image when the eye turned to the limit range. The macular centre was considered the centre, and the horizontal position was determined through the connection between the macular centre and the optic disc centre. Based on the location of the hole, the combined standardized fundus image was divided into four regions: the nasal side (Z1), the superior side (Z2), the temporal side (Z3), and the inferior side (Z4) . We measured the optic disc transverse diameter (D1) using the combined tandardized images. The line between the macular fovea and the centre of the retinal hole was made and extended to the edge of the image. The distance between the centre of the retinal hole and the edge of the extension line was D2 . The standardization of the images and the data measurement were finished independently by two researchers (Li and Shang). Data with large differences were measured again (Gao). The mean was used in the next calculation. The optic disc transverse diameter was measured by an optical coherence tomography scanner (Heidelberg Engineering, Heidelberg, Germany). Automatic software with automated measurement was used to determine the transverse diameter of the optic disc of each eye. We used the transverse diameter values of the optic disc . A Zeiss IOL Master laser interferometer (Optical Biometry, IOL Master; Carl Zeiss Meditec AG, Jena, Germany) was used to measure the axial length (AL). Based on the measured AL, patients were divided into two groups: AL ≥ 26 mm and AL < 26 mm. The optic disc transverse diameter measured using optical coherence tomography was labelled Dd. The preoperative scleral chord length between the limbus and retinal hole was calculated as follows: D = D 2 / D 1 * Dd + D s (where Ds = distance between the ora serrata and limbus). Straatsma et al. measured the distance from the serrated margin to the Schwalbe’s line from the eye of a cadaver as 6.14 ± 0.85 mm superior, 6.20 ± 0.76 mm inferior, 5.73 ± 0.81 mm on the nasal side, and 6.53 ± 0.75 mm on the temporal side . However, the chord length from the limbus to the hole was actually measured from the outside of the eyeball during the operation. The distances between the ora serrata and the limbus used in the calculation were according to personal experience. To simplify the calculation, we added 7 mm to the nasal side, 8 mm to the temporal side, and 7.5 mm to the superior and inferior sides. The preoperative estimation process is shown in a flowchart in . Intraoperative measurement Locating and marking of retinal holes on the sclera were performed by the same experienced surgeon (Duan). The statistics calculated from the images were not shared with the surgeon who performed the surgery. After locating the hole in the surgery, an ophthalmic calliper was used to measure the scleral chord length between the limbus and scleral markers of the retinal holes . Holes were divided into four regions: the superior, inferior, nasal, and temporal. The region and scleral chord length of each hole were recorded. Statistical analyses The preoperative estimated scleral chord length was calculated and compared with the measured scleral chord length during surgery. All analyses were performed using SPSS statistical software (v21; IBM Corp., Armonk, NY, USA). Continuous, normally distributed data were expressed as mean ± standard deviation and were compared using a paired t-test. p values < 0.01 were considered statistically significant. Calculation of preoperative scleral chord length and statistical analysis were completed by another researcher (Zhou) who did not participate in data collection. Preoperative assessment Ultra-wide-angle fundus images were taken using an Optos Ultra-Widefield Fundus Camera (Optos, Dunfermline, UK, 200Tx). These images were taken 1 day preoperatively, 1 week postoperatively, and 1 month postoperatively. Pupils were dilated to the greatest extent before taking images. The images of the central, superior, inferior, nasal, and temporal retina of each eye were taken by the same experienced technician. During the process, patients’ eyelids were lifted by the technician, and they had to try their best to stare in the corresponding directions. Images of the retinal holes were taken in a similar manner. All images were taken when the ‘green in-focus’ signal was obtained on the Optos machine. The images of the inner surface of the eye, which is nearly spherical, cannot be mapped to a flat surface without distortion. The size and shape of some structures can be greatly distorted with further eccentricity in taking images. Therefore, we standardized the images to make them more accurate. We exported images from the Optos and converted them to grayscale in PhotoshopCS3 (Adobe, Inc., San Jose, CA). Images centred on the posterior portion of the eye were inspected for sharpness and lack of systemic distortion across the image. We selected the best image and considered it to be the best image of the periphery. To standardize the resultant images, histogram stretching was performed. Peripheral images were transformed using elastic deformation, and the posterior portions of peripheral images matched the base reference images. Warped images were merged into a montage . The UWF imaging range can reach the ora serrata through eye position guidance. Moreover, pupillary dilation and eyelid lifting can enlarge the imaging range . Therefore, we assumed that the ora serrata corresponded to the blood-free zone at the edge of the obtained image when the eye turned to the limit range. The macular centre was considered the centre, and the horizontal position was determined through the connection between the macular centre and the optic disc centre. Based on the location of the hole, the combined standardized fundus image was divided into four regions: the nasal side (Z1), the superior side (Z2), the temporal side (Z3), and the inferior side (Z4) . We measured the optic disc transverse diameter (D1) using the combined tandardized images. The line between the macular fovea and the centre of the retinal hole was made and extended to the edge of the image. The distance between the centre of the retinal hole and the edge of the extension line was D2 . The standardization of the images and the data measurement were finished independently by two researchers (Li and Shang). Data with large differences were measured again (Gao). The mean was used in the next calculation. The optic disc transverse diameter was measured by an optical coherence tomography scanner (Heidelberg Engineering, Heidelberg, Germany). Automatic software with automated measurement was used to determine the transverse diameter of the optic disc of each eye. We used the transverse diameter values of the optic disc . A Zeiss IOL Master laser interferometer (Optical Biometry, IOL Master; Carl Zeiss Meditec AG, Jena, Germany) was used to measure the axial length (AL). Based on the measured AL, patients were divided into two groups: AL ≥ 26 mm and AL < 26 mm. The optic disc transverse diameter measured using optical coherence tomography was labelled Dd. The preoperative scleral chord length between the limbus and retinal hole was calculated as follows: D = D 2 / D 1 * Dd + D s (where Ds = distance between the ora serrata and limbus). Straatsma et al. measured the distance from the serrated margin to the Schwalbe’s line from the eye of a cadaver as 6.14 ± 0.85 mm superior, 6.20 ± 0.76 mm inferior, 5.73 ± 0.81 mm on the nasal side, and 6.53 ± 0.75 mm on the temporal side . However, the chord length from the limbus to the hole was actually measured from the outside of the eyeball during the operation. The distances between the ora serrata and the limbus used in the calculation were according to personal experience. To simplify the calculation, we added 7 mm to the nasal side, 8 mm to the temporal side, and 7.5 mm to the superior and inferior sides. The preoperative estimation process is shown in a flowchart in . Intraoperative measurement Locating and marking of retinal holes on the sclera were performed by the same experienced surgeon (Duan). The statistics calculated from the images were not shared with the surgeon who performed the surgery. After locating the hole in the surgery, an ophthalmic calliper was used to measure the scleral chord length between the limbus and scleral markers of the retinal holes . Holes were divided into four regions: the superior, inferior, nasal, and temporal. The region and scleral chord length of each hole were recorded. Statistical analyses The preoperative estimated scleral chord length was calculated and compared with the measured scleral chord length during surgery. All analyses were performed using SPSS statistical software (v21; IBM Corp., Armonk, NY, USA). Continuous, normally distributed data were expressed as mean ± standard deviation and were compared using a paired t-test. p values < 0.01 were considered statistically significant. Calculation of preoperative scleral chord length and statistical analysis were completed by another researcher (Zhou) who did not participate in data collection. Ultra-wide-angle fundus images were taken using an Optos Ultra-Widefield Fundus Camera (Optos, Dunfermline, UK, 200Tx). These images were taken 1 day preoperatively, 1 week postoperatively, and 1 month postoperatively. Pupils were dilated to the greatest extent before taking images. The images of the central, superior, inferior, nasal, and temporal retina of each eye were taken by the same experienced technician. During the process, patients’ eyelids were lifted by the technician, and they had to try their best to stare in the corresponding directions. Images of the retinal holes were taken in a similar manner. All images were taken when the ‘green in-focus’ signal was obtained on the Optos machine. The images of the inner surface of the eye, which is nearly spherical, cannot be mapped to a flat surface without distortion. The size and shape of some structures can be greatly distorted with further eccentricity in taking images. Therefore, we standardized the images to make them more accurate. We exported images from the Optos and converted them to grayscale in PhotoshopCS3 (Adobe, Inc., San Jose, CA). Images centred on the posterior portion of the eye were inspected for sharpness and lack of systemic distortion across the image. We selected the best image and considered it to be the best image of the periphery. To standardize the resultant images, histogram stretching was performed. Peripheral images were transformed using elastic deformation, and the posterior portions of peripheral images matched the base reference images. Warped images were merged into a montage . The UWF imaging range can reach the ora serrata through eye position guidance. Moreover, pupillary dilation and eyelid lifting can enlarge the imaging range . Therefore, we assumed that the ora serrata corresponded to the blood-free zone at the edge of the obtained image when the eye turned to the limit range. The macular centre was considered the centre, and the horizontal position was determined through the connection between the macular centre and the optic disc centre. Based on the location of the hole, the combined standardized fundus image was divided into four regions: the nasal side (Z1), the superior side (Z2), the temporal side (Z3), and the inferior side (Z4) . We measured the optic disc transverse diameter (D1) using the combined tandardized images. The line between the macular fovea and the centre of the retinal hole was made and extended to the edge of the image. The distance between the centre of the retinal hole and the edge of the extension line was D2 . The standardization of the images and the data measurement were finished independently by two researchers (Li and Shang). Data with large differences were measured again (Gao). The mean was used in the next calculation. The optic disc transverse diameter was measured by an optical coherence tomography scanner (Heidelberg Engineering, Heidelberg, Germany). Automatic software with automated measurement was used to determine the transverse diameter of the optic disc of each eye. We used the transverse diameter values of the optic disc . A Zeiss IOL Master laser interferometer (Optical Biometry, IOL Master; Carl Zeiss Meditec AG, Jena, Germany) was used to measure the axial length (AL). Based on the measured AL, patients were divided into two groups: AL ≥ 26 mm and AL < 26 mm. The optic disc transverse diameter measured using optical coherence tomography was labelled Dd. The preoperative scleral chord length between the limbus and retinal hole was calculated as follows: D = D 2 / D 1 * Dd + D s (where Ds = distance between the ora serrata and limbus). Straatsma et al. measured the distance from the serrated margin to the Schwalbe’s line from the eye of a cadaver as 6.14 ± 0.85 mm superior, 6.20 ± 0.76 mm inferior, 5.73 ± 0.81 mm on the nasal side, and 6.53 ± 0.75 mm on the temporal side . However, the chord length from the limbus to the hole was actually measured from the outside of the eyeball during the operation. The distances between the ora serrata and the limbus used in the calculation were according to personal experience. To simplify the calculation, we added 7 mm to the nasal side, 8 mm to the temporal side, and 7.5 mm to the superior and inferior sides. The preoperative estimation process is shown in a flowchart in . Locating and marking of retinal holes on the sclera were performed by the same experienced surgeon (Duan). The statistics calculated from the images were not shared with the surgeon who performed the surgery. After locating the hole in the surgery, an ophthalmic calliper was used to measure the scleral chord length between the limbus and scleral markers of the retinal holes . Holes were divided into four regions: the superior, inferior, nasal, and temporal. The region and scleral chord length of each hole were recorded. The preoperative estimated scleral chord length was calculated and compared with the measured scleral chord length during surgery. All analyses were performed using SPSS statistical software (v21; IBM Corp., Armonk, NY, USA). Continuous, normally distributed data were expressed as mean ± standard deviation and were compared using a paired t-test. p values < 0.01 were considered statistically significant. Calculation of preoperative scleral chord length and statistical analysis were completed by another researcher (Zhou) who did not participate in data collection. Patient demographics The data and images of 21 eyes of 21 patients (9 males and 12 females) were collected. The average age of the patients was 27.1 ± 8.6 (range: 16–53) years; 25 retinal holes were observed. The average AL of all eyes was 26.32 ± 1.30 (range: 24.29–29.23) mm. The mean preoperative best-corrected visual acuity (BCVA) was 0.36 ± 0.35 LogMAR, ranging from 1.4 to −0.1 LogMAR. All included eyes were phakic . Determination of retinal hole location based on intra- and postoperative Optos imaging There were 16 holes in the AL ≥ 26 mm group. The average estimated preoperative scleral chord length was 13.99 ± 1.18 mm, and the average actual intraoperative chord length was 13.84 ± 1.19 mm; the difference between them was 0.14 (range: −0.99–1.28) mm and not significant ( p = 0.445 > 0.01). There were 9 holes in the AL < 26 mm group. The average estimated preoperative scleral chord length was 14.01 ± 1.59 mm, and the average actual intraoperative chord length was 13.89 ± 1.76 mm; the difference between them was 0.12 (range: −1.51–1.66) mm and not significant ( p = 0.731 > 0.01). Comparing the partitions, there was no statistically significant difference between the preoperative and intraoperative estimated chord lengths in the Z2 ( p = 0.080 > 0.01), Z3 ( p = 0.751 > 0.01), and Z4 ( p = 0.522 > 0.01) regions. There was only one case that involved the Z1 region, which was Case 9. The estimated preoperative chord length was 16.24 mm, and the actual intraoperative measurement was 15 mm. This data was too small to be analyzed. The data and images of 21 eyes of 21 patients (9 males and 12 females) were collected. The average age of the patients was 27.1 ± 8.6 (range: 16–53) years; 25 retinal holes were observed. The average AL of all eyes was 26.32 ± 1.30 (range: 24.29–29.23) mm. The mean preoperative best-corrected visual acuity (BCVA) was 0.36 ± 0.35 LogMAR, ranging from 1.4 to −0.1 LogMAR. All included eyes were phakic . There were 16 holes in the AL ≥ 26 mm group. The average estimated preoperative scleral chord length was 13.99 ± 1.18 mm, and the average actual intraoperative chord length was 13.84 ± 1.19 mm; the difference between them was 0.14 (range: −0.99–1.28) mm and not significant ( p = 0.445 > 0.01). There were 9 holes in the AL < 26 mm group. The average estimated preoperative scleral chord length was 14.01 ± 1.59 mm, and the average actual intraoperative chord length was 13.89 ± 1.76 mm; the difference between them was 0.12 (range: −1.51–1.66) mm and not significant ( p = 0.731 > 0.01). Comparing the partitions, there was no statistically significant difference between the preoperative and intraoperative estimated chord lengths in the Z2 ( p = 0.080 > 0.01), Z3 ( p = 0.751 > 0.01), and Z4 ( p = 0.522 > 0.01) regions. There was only one case that involved the Z1 region, which was Case 9. The estimated preoperative chord length was 16.24 mm, and the actual intraoperative measurement was 15 mm. This data was too small to be analyzed. RRD is a common ophthalmic disease. SB is a traditional technique for retinal detachment repair, and long-term studies have reported a success rate of 95% . However, recently, owing to the development and popularization of vitrectomy, the use of SB has reduced gradually. A study in the United States showed that in RRD surgery, PPV use increased by 72%. SB without vitrectomy decreased by 69% from 1997 to 2007 . Another study investigated 2000–2014 data and reported that the number of RRD cases in which vitrectomy was performed increased from 13,814 to 19,288, and the number of cases in which SB was performed decreased significantly from 6502 to 1260 . Regarding retinal detachment repair surgery, vitrectomy, SB, and PR were performed in 83%, 5%, and 12% of cases, respectively . However, compared with vitrectomy, SB is suitable for young patients with phakia and especially suitable for RRD secondary to retinal lattice degeneration with atrophic retinal holes . Another advantage of SB is that there are few complications such as cataracts, preretinal proliferation, recurrent traction retinal detachment, and new operation-related holes . The key to successful SB is the closure of the retinal holes, and the key to the closure of the retinal holes is accurate localization. Therefore, accurate location determination of retinal holes during surgery decides the success of SB. Presently, the localization methods of retinal holes are divided into preoperative and intraoperative methods. Preoperatively, surgeons examine the fundus using a binocular indirect ophthalmoscope and a three-mirror lens to detect all possible holes. The location of the holes and the distance between the holes and the limbus are estimated simultaneously. This approach is not quantitative and requires individual experience. Intraoperatively, pressing the sclera and observing by the indirect ophthalmoscope is the most common localization method . However, skill comes from practicing the surgery, as SB has a lengthy learning curve. There have been recent advances in SB to make the learning process easier. SB assisted by 25-G fibre, and chandelier endoillumination has many advantages over traditional methods. For instance, it is easier for surgeons to observe the fundus and even the peripheral retina under a microscope to detect small retinal holes which are difficult to identify with indirect ophthalmoscopy . However, the surgery needs the sclera incision to be extended to the vitreous cavity, which may alter the physiological environment of the vitreous. The risk of postoperative endophthalmitis and other complications, such as retinal holes secondary to vitreous incarceration at the insertion point of the cannula, can be increased . In addition, not all clinics are equipped with chandelier endoillumination. Ultra-wide-angle imaging has greatly enhanced our realization of the peripheral retina preoperatively. A 200° range of the fundus can be observed in one picture. With the patient’s cooperation, the range can be extended to the ora serrata by guiding the eye position (upper, lower, nasal, and temporal directions) . Information regarding retinal holes can be observed preoperatively with UWF imaging, but there is little research analyzing this information quantitatively . We designed this research to locate the retinal hole preoperationally using this technique. In our research, 21 patients were included, and 25 retinal holes were observed. The inner surface of the eye is a three-dimensional sphere, which is difficult to map to a two-dimensional plane without distortion. Deviations measured directly from the wide-angle image and the actual data of the eyeball are uncertain . Therefore, we standardized the images and used the ratio between the transverse diameter of the optic disc and the distance from the measured hole to the edge of the ora serrata to reduce the error caused by direct measurement. Meanwhile, we divided the data from AL and location again to reduce error. From our results, the average difference between the estimated chord length of retinal tears and the actual measured chord length was 0.13 (range: −1.51–1.66) mm. There was no statistically significant difference between the estimated chord length and the actual measured chord length in the AL ≥ 26 mm group ( p = 0.445 > 0.01) and the AL < 26 mm group ( p = 0.731 > 0.01). Regarding the different zones of the holes, statistical analyses based on the location of the retinal holes showed that there was no statistically significant difference in the retinal tears in the Z2, Z3, or Z4 regions. However, only one case involved the Z1 region, and the data was insufficient for analysis. We also observed on the postoperative images that all the retinal holes were on the buckle, suggesting the feasibility of locating retinal holes using UWF photography preoperatively. In clinical practice, we used a relatively simple method of calculation. We used the Optos software to acquire the optic disc transverse diameter (D1) and the distance between the centre of the retinal hole and the ora serrata (D2). The optic disc diameter was estimated to be 1.5 mm. We then used the formula D = D 2 / D 1 * 1.5 + Ds , with Ds being the distance between the ora serrata and limbus. We added 7 mm to the nasal side, 8 mm to the temporal side, and 7.5 mm each to the superior and inferior sides. Supplementary Materials 1 and 2 present the detailed data and our analysis. Interestingly, there were also no statistically significant differences between these simplified preoperative estimate methods and intraoperative estimated chord lengths ( p = 0.119 > 0.01). According to resident training in our study site, our method improves the learning curve of localization of retinal holes skill in SB surgery. There are few studies on calculations of the retinal hole positions using a UWF image. Ishikawa et al. calculated the distance between the posterior edge of the scleral buckle and cornea on a UWF image after SB and compared it to the distance measured during surgery. A regression equation was obtained based on the data. The position of the retinal hole in the preoperative image was brought into the regression equation, and the results fell within the range of the scleral buckle . Their study revealed the feasibility of estimating the intraocular distance using UWF imaging. However, the shape of the eyeball, which could not be estimated accurately, was changed in the area where the buckle was performed. The change in the wide-angle image could not be predicted accurately; this is an unavoidable error when calculating using postoperative images. In our study, preoperative UWF images were used to avoid this problem; this is more consistent with the real therapeutic process. However, our study has some limitations. First, the error caused by the distortion of the wide-angle image could not be completely avoided. Single-shot imaging of the Optos cannot cover the entire retina. We obtained images of the peripheral retina by guiding the direction of the patient’s gaze. Therefore, even a small deviation may significantly impact the final plane image. Second, the chord length from the ora serrata to the limbus that we used in our study (8 mm, 7 mm) was based on estimation and may have individual variations. Third, our data was insufficient for nasal position analysis. Last, the patients in our study might not be representative of all RRD populations; thus, generalizability based on different phakic statuses, large retinal tears, and bullous retinal detachment may be limited, which warrants future investigation. There was no statistically significant difference in the scleral chord length between the retinal holes and the limbus estimated by UWF fundus photography used preoperatively and measured intraoperatively. Preoperative analysis using UWF imaging helps surgeons estimate the scleral chord length between retinal holes and the limbus. Thereby, it may help to improve the learning process for SB surgery. Supplemental Material Click here for additional data file. |
Storylines of family medicine III: core principles—primary care, systems and family | a4f55df7-4ecc-46ac-b646-c95cdbdcdaf6 | 11029207 | Family Medicine[mh] | Family medicine is one of the primary care disciplines of medicine. As such, it is founded on several principles of primary care, each of which is supported by research that demonstrates its importance to a healthcare system. In fact, systems thinking is one of the major factors that distinguish family medicine from other primary care disciplines. Wise family physicians integrate systems into their diagnostic and therapeutic activities. They balance their understanding of patients with recognition of the fact that we all live and work in a complex, interconnected world. In sum, many factors influence the health and well-being of patients, among them patients’ families of origin and chosen families. Tim Joslin and John Saultz Continuity is one of the core values of family medicine. Although continuity has multiple dimensions, the ongoing interpersonal doctor–patient relationship, which continues across a lifespan, is the most defining characteristic of family medicine. Long-term relationships are what allow family physicians entrée into their patients’ lives. Family physicians care for people over the course of years—from cradle to grave, as individuals and as members of multigenerational family groups. Bernice—a single name will suffice. She is a patient in our practice that everyone knows. When I (TAJ) met her halfway through my family medicine residency, she had already experienced a handful of primary care clinicians over the preceding dozen years. Typically, a resident would assume her care at the beginning of training and pass her on to another at graduation. Consistency and continuity lasted for only three years at a time. When I inherited Bernice, the story could have been the same, but something different happened. I stuck around, and so did she. After graduating residency, I stayed on as the chief resident and later as a faculty member. The day I met Bernice, the medical assistant (MA) nudged me just before I entered the examination room. I looked back at her, a puzzled expression on my face. What was that for? I wondered. ’You’ll see,’ the MA said just before I entered the room. As I entered, I saw Bernice in her typical state, conditioned by her persistent mental illness. Her well-worn jacket was no longer waterproof, perforated by cigarette burns. Her sweats were dirty, and she wore sandals over a set of wool socks. Her hair was thin, dark and unkempt, and her fingers were a dark burnt caramel colour—she smoked nearly 100 self-rolled cigarettes per day. I noticed her eyes scanning the room, as if she were trying to physically escape her racing thoughts. Like most second year residents, I tried to wrap my mind around the basics during my first appointment with Bernice. I learnt that she had a history of multiple psychiatric hospital admissions; attempts to get her reconnected with outpatient mental health services had been invariably unsuccessful. Follow-up plans never came to fruition. Her adherence to medication therapy was sporadic, at best. There was no magical connection in our relationship at the beginning, but over weeks, months and years, a connection formed between us. Mutual trust ensued. Through our team’s intense, coordinated and prolonged efforts, Bernice’s life became less chaotic. Her hospital admissions decreased drastically. She often went years between psychiatric hospitalisations. Bernice’s life will never be easy, but the relationship that formed and grew over many years resulted in meaningful improvements in her health and day-to-day existence. The interpersonal domain of continuity, characterised by personal trust, professional intimacy and mutual responsibility, is hard to teach—medical students on their clerkships rarely see people more than once, and family medicine residents commonly spend significant periods away from ambulatory care. Yet, continuous physician–patient relationships are what many family physicians find most satisfying in their careers. Improved continuity of care lowers healthcare costs, reduces risk of hospitalisation, improves patient and clinician satisfaction and improves overall quality of care. Continuity nourishes the seed that is planted when doctors and patients first meet, allowing healing relationships to grow into something wonderful and unexpected . Readings Loxterkamp D. The lost pillar: does continuity of care still matter? Ann Fam Med 2021;19:553–5. doi: 10.1370/afm.2736 Nowak DA, Sheikhan NY, Naidu SC, Kuluski K, Upshur REG. Why does continuity of care with family doctors matter? Review and qualitative synthesis of patient and physician perspectives. Can Fam Physician 2021;67:679–88. doi: 10.46747/cfp.6709679 Saultz JW. Defining and measuring interpersonal continuity of care. Ann Fam Med 2003;1:134–43. doi: 10.1370/afm.23 Loxterkamp D. The lost pillar: does continuity of care still matter? Ann Fam Med 2021;19:553–5. doi: 10.1370/afm.2736 Nowak DA, Sheikhan NY, Naidu SC, Kuluski K, Upshur REG. Why does continuity of care with family doctors matter? Review and qualitative synthesis of patient and physician perspectives. Can Fam Physician 2021;67:679–88. doi: 10.46747/cfp.6709679 Saultz JW. Defining and measuring interpersonal continuity of care. Ann Fam Med 2003;1:134–43. doi: 10.1370/afm.23 Sommer Aldulaimi and Paul Gordon Family physicians are trained to see and manage a variety of different medical illnesses and conditions over the lifetimes of patients in a variety of settings. This comprehensiveness of care is key to improving care, decreasing healthcare costs and working towards health equity. In the same morning, a family physician might see a two year-old for a well-child check; attend to an elderly male for his diabetes, high blood pressure and heart failure; do a Pap smear on a young woman as part of a well-woman examination and discuss birth control options; inject a patient’s knee; treat a urinary tract infection; and deliver a baby. Family physicians are trained to see and manage a variety of different medical illnesses and conditions in a variety of clinical settings; they must manage these illnesses and conditions over the lifetimes of their patients regardless of sex or presenting condition. Such breadth and depth of scope by which family physicians attend to patients’ medical problems is called comprehensiveness . Comprehensiveness is one of the core principles of primary care, though family physicians typically have a scope of practice that is broader than other primary care specialties. Such scope often includes practising in inpatient, outpatient and community settings; attending to maternity, paediatric, adult and elderly patients; and offering services that include procedures, minor surgeries and public health leadership. Why should family physicians and patients care about comprehensive practice? Comprehensiveness is associated with numerous benefits. These include greater efficiency (family physicians can address most patients’ needs without the need for subspecialty consultation) with better health outcomes provided at lower costs. Comprehensiveness results in lower costs for patients and for the healthcare system. National healthcare systems that focus on comprehensive primary care spend much less money and have better outcomes than those systems that are focused on subspecialties. Comprehensive care is also associated with decreased rates of hospitalisations and better self-reported health outcomes by patients themselves. Perhaps even more importantly, in healthcare systems that emphasise comprehensive care, disparities in disease severity decrease while population-based markers of prevention increase. Comprehensiveness and broader scope of practice have been correlated with decreased rates of burnout among family physicians, an important consideration in light of the primary care physician shortage that exists in the USA and in other areas around the world. Comprehensiveness is crucial to improving care and health outcomes, lowering costs of treatment, lessening health disparities and decreasing physician burnout. Family physicians are trained to provide the most comprehensive care of any medical specialty, regardless of geographical location, patient demographic or method of reimbursement for services rendered. Any rational healthcare system should facilitate and foster this type of comprehensiveness in medical care. Readings Cubaka VK, Dyck C, Dawe R, et al . A global picture of family medicine: the view from a WONCA storybooth. BMC Fam Pract 2019;20:129. doi: 10.1186/s12875-019-1017-5 Grumbach K. To be or not to be comprehensive. Ann Fam Med 2015;13:204–5. doi: 10.1370/afm.1788 Weidner AKH, Phillips RL Jr, Fang B, Peterson LE. Burnout and scope of practice in new family physicians. Ann Fam Med 2018;16:200–5. doi:10.1370/afm.2221 Cubaka VK, Dyck C, Dawe R, et al . A global picture of family medicine: the view from a WONCA storybooth. BMC Fam Pract 2019;20:129. doi: 10.1186/s12875-019-1017-5 Grumbach K. To be or not to be comprehensive. Ann Fam Med 2015;13:204–5. doi: 10.1370/afm.1788 Weidner AKH, Phillips RL Jr, Fang B, Peterson LE. Burnout and scope of practice in new family physicians. Ann Fam Med 2018;16:200–5. doi:10.1370/afm.2221 John Lane Prudent generalist practitioners know that coordination of care—addressing issues related to the process of navigating the healthcare system—is a crucial component of their clinical responsibilities. Family physicians sort symptoms and signs into orderly patterns that become diagnoses and treatment plans. Less obvious but equally important is the work of sorting complex interactions with the healthcare system into an orderly and efficient series of interventions and responses. This is called coordination of care . Coordination of care is a core task for all family physicians. Along with other clinical team members, they collect vital signs, review medical records, elicit clinical histories and perform pertinent physical examinations. They order laboratory tests and diagnostic studies. They arrange subspecialty consultations. They gather and synthesise data, develop therapeutic plans, prescribe treatments and recommend appropriate follow-up. In popular imagination, the archetypal family doctor manages these tasks under one roof with minimal complexity. The physician listens to the patient’s chief complaint—let’s say, a sore throat and no cough in an otherwise healthy second grader. On examination, the child has a temperature of 102°F, inflamed tonsils with exudates and anterior cervical adenopathy. The diagnosis is conveyed, questions are answered and the child’s mother receives an antibiotic prescription. She pays the bill at the front desk, and, with her child, heads to the pharmacy. Cure is on the way. Although such to-the-diagnostic-point encounters still exist, most clinical interactions are much more complex. They often involve the following: A chart review that includes many documents from a variety of sources, and often decisions need to be made about information not yet in hand. Decision-making about diagnostic investigations, which is influenced by factors such as availability, cost, insurance and timeliness. As well, orders must be transmitted effectively and efficiently. Patient involvement and needs, including social determinants of health: are there concerns beyond the clinical presentation that may affect clinical attendance, engagement with the therapeutic plan and efficacy of treatments? Referrals and consultations, the arrangements for which often require considerable thought and logistical skills to complete. Consider, then, a more complex case. A family physician sees a patient in follow-up for abdominal pain. The history and physical examination suggest a wide differential diagnosis. Some pertinent imaging results are available from a recent emergency department (ED) visit but not the laboratory results or ED physicians’ impressions. Furthermore, there was a distant, similar medical event in which the patient had an extensive workup and diagnosis, but the findings and impression are presently obscure. The family physician considers a variety of questions. How quickly can previous results be received? At what expense, risk and patient distress can new testing and consultations be obtained? Can the patient attend an important family event before testing and consultations are completed? Can the clinic’s team facilitate the patient’s journey when the above questions have been answered? Successful coordination of care depends on a high-functioning team with sophisticated systems to manage the to-and-from flow of information pertinent to patient care. Such success depends on strategic planning, respectful communication, consistent work habits and people—team members—who understand the patient experience, are forward thinking and practise resiliency daily. This is coordination of care. Done poorly, patients suffer. Done well, it enhances the experience of all participants in the healthcare system. Readings Bodenheimer T. Coordinating care—a perilous journey through the health care system. N Engl J Med 2008;358:1064–71. doi: 10.1056/NEJMhpr0706165 Bodenheimer T, Ghorob A, Willard-Grace R, Grumbach K. The 10 building blocks of high-performing primary care. Ann Fam Med 2014;12:166–71. doi: 10.1370/afm.1616 Phillips C. Care coordination for primary care practice. J Am Board Fam Med 2016;29:649–51. doi: 10.3122/jabfm.2016.06.160312 Bodenheimer T. Coordinating care—a perilous journey through the health care system. N Engl J Med 2008;358:1064–71. doi: 10.1056/NEJMhpr0706165 Bodenheimer T, Ghorob A, Willard-Grace R, Grumbach K. The 10 building blocks of high-performing primary care. Ann Fam Med 2014;12:166–71. doi: 10.1370/afm.1616 Phillips C. Care coordination for primary care practice. J Am Board Fam Med 2016;29:649–51. doi: 10.3122/jabfm.2016.06.160312 Eric Lee and Jake Prunuske Access to healthcare is essential for optimal health. Family physicians improve access to care by offering community-based, continuity-oriented medical services in the context of individuals’ values, family systems and communities. Healthcare access is ‘the possibility to identify healthcare needs, to seek services, to reach resources, to obtain or use healthcare services, and to be offered services appropriate to the needs for care.’ Ideally, the resources of clinicians, health systems and communities meet the needs and align with the abilities of patients. Optimal access requires health literacy, reciprocal trust between clinicians and patients, respect for and integration of personal and cultural values, geographical proximity, a supportive infrastructure and affordable costs. Failure to attend to these factors has led to limited access to healthcare for many people in the USA, where unequal access to healthcare is a major challenge. A strong system of primary care is necessary to ensure that high-quality care is available to every individual and family in every community. Importantly, people must first trust that the doctors who attend to them will understand their needs and perspectives and will work in their best interest. Unfortunately, issues in the USA (such as inequity, racism and biases) have contributed to a culture of distrust. One way to rebuild trust is to improve the diversity of the US healthcare workforce. Relative to other specialties, family medicine has higher racial and ethnic representation and greater presence in ethnically diverse, medically underserved communities. The high cost of healthcare in the USA is another common cause of deferred care and treatment, especially for people who are medically underserved by current healthcare systems. Family doctors’ emphasis on providing a broad scope of care and following principles of continuity, context and community contributes to decreased costs of care and fewer hospitalisations, emergency room visits and surgeries. Much of the cost burden faced by patients could be mitigated or even avoided if more patients regularly saw their family doctor. Family doctors thus play a critical role in addressing multifactorial, intersectional, personal and systemic barriers to access to medical care. They are well equipped to meet the social and cultural needs of populations in their communities. At the heart of family medicine is a drive to provide whole-person care—personalisation, relationship building and continuity are core values of family medicine. The relationships that family doctors develop over years—and in some cases, decades—provide insight into the biological, psychosocial, cultural and political determinants of health that affect their patients. Such breadth of knowledge helps family doctors coordinate care and advocate for patients’ health needs within a multidisciplinary care team. The health of individual patients is, to a certain degree, dependent on access to high-quality healthcare. Along with other community-based generalist clinicians, family physicians help ensure this access. Access to care is vital to improving the overall health of people within communities, and family physicians play an integral part in providing this access ! Readings Levesque JF, Harris MF, Russell G. Patient-centred access to health care: conceptualising access at the interface of health systems and populations. Int J Equity Health 2013;12:18. doi: 10.1186/1475-9276-12-18 National Academies of Sciences, Engineering, and Medicine; Health and Medicine Division; Board on Health Care Services; Committee on Implementing High-Quality Primary Care. Implementing High-Quality Primary Care: Rebuilding the Foundation of Health Care . Consensus Study Report/Highlights. May 2021. Available: https://nap.nationalacademies.org/resource/25983/Highlights_High-Quality/Primary/Care-4.23.21_final.pdf [Accessed 31 Jan 2024]. Nowak DA, Sheikhan NY, Naidu SC, Kuluski K, Upshur REG. Why does continuity of care with family doctors matter? Review and qualitative synthesis of patient and physician perspectives. Can Fam Physician 2021;67:679–88. doi: 10.46747/cfp.6709679 Levesque JF, Harris MF, Russell G. Patient-centred access to health care: conceptualising access at the interface of health systems and populations. Int J Equity Health 2013;12:18. doi: 10.1186/1475-9276-12-18 National Academies of Sciences, Engineering, and Medicine; Health and Medicine Division; Board on Health Care Services; Committee on Implementing High-Quality Primary Care. Implementing High-Quality Primary Care: Rebuilding the Foundation of Health Care . Consensus Study Report/Highlights. May 2021. Available: https://nap.nationalacademies.org/resource/25983/Highlights_High-Quality/Primary/Care-4.23.21_final.pdf [Accessed 31 Jan 2024]. Nowak DA, Sheikhan NY, Naidu SC, Kuluski K, Upshur REG. Why does continuity of care with family doctors matter? Review and qualitative synthesis of patient and physician perspectives. Can Fam Physician 2021;67:679–88. doi: 10.46747/cfp.6709679 Limor Gildenblatt and Mike Friedman Systems theory posits that complicated interaction of unique, inter-related components determines outcomes in a variety of realms. Family medicine has embraced systems theory as a core of the specialty and in so doing has given rise to a new understanding of patient-centred care. General systems theory emerged in the late 1940s and early 1950s to better understand the interactions between individual components of complex problems. Initially targeted to disciplines such as math, science and engineering, systems thinking has come to exert a strong influence on medical education. But that wasn’t always the case. Early medical educators focused narrowly on training students to master the biomedical basis of disease—many still do! This reductionist model derived from the notion that breakdowns in biological processes could fully explain disease and illness; thus, if physicians could reduce problems into discrete, quantifiable units, they could better understand the whole. In time, physicians in all fields saw the limitations of this simple biomedical model. What was missing was context and an appreciation of the complex interactions that exist in our patients’ worlds . Consider the following case: Rosa, a 59-year-old second-generation immigrant from Mexico, was diagnosed with new-onset diabetes and hypertension. Despite appropriate referrals and educational interventions, her problems remained uncontrolled. She worked long hours at an unfulfilling desk job that she didn’t dare leave because she didn’t want to risk her pension or insurance. When asked if her schedule could be modified to accommodate her health needs, Rosa revealed the factors that made such adjustments impossible: her husband didn’t work because of his own health problems; her youngest child, aged 25, couldn’t hold a job; the oldest son, in his 30s, also lived with the family, along with his partner and two young children. Finally, her middle son, separated from his wife, had been driving with their two children when he ran into a highway overpass, instantly killing one of the children. Owing to a greatly elevated blood alcohol level, he was found guilty of involuntary manslaughter and was sentenced to eight years in prison. But his estranged wife and the surviving child moved into the house as well. The biomedical model of healthcare was clearly incapable of capturing her situation. Her diabetes and hypertension had social and cultural influences just as surely as they had a biological component. In the relatively young field of family medicine, where managing individual patients in the context of family and community has always been a core principle, the need to develop an expansive view of systems-based care quickly became evident. Leaders in the discipline embarked on a national strategy to introduce a broader perspective to patient care. They didn’t reject biology but instead incorporated the previously neglected non-biological phenomena into patient care. This approach, which presaged what has come to be known as holistic care, employed the biopsychosocial model of health. It accepted that human pathology arose not simply from failures in biological pathways but was impacted by a complex network of inter-related factors. Applying an understanding of these disparate psychological, social and environmental components not only offered patients a clearer path to health; it also gave physicians the opportunity to better manage their own expectations, work collaboratively with other healthcare professionals and truly provide patient-centred care. Readings Card AJ. The biopsychosociotechnical model: a systems-based framework for human-centered health improvement. Health Syst (Basingstoke) 2022;12:387-407. doi: 10.1080/20476965.2022.2029584 Sturmberg JP. Systems and complexity thinking in general practice: part 1 - clinical application. Aust Fam Physician 2007;36:170–3. Sturmberg JP, Martin CM, Katerndahl DA. Systems and complexity thinking in the general practice literature: an integrative, historical narrative review. Ann Fam Med 2014;12:66–74. doi: 10.1370/afm.1593 Card AJ. The biopsychosociotechnical model: a systems-based framework for human-centered health improvement. Health Syst (Basingstoke) 2022;12:387-407. doi: 10.1080/20476965.2022.2029584 Sturmberg JP. Systems and complexity thinking in general practice: part 1 - clinical application. Aust Fam Physician 2007;36:170–3. Sturmberg JP, Martin CM, Katerndahl DA. Systems and complexity thinking in the general practice literature: an integrative, historical narrative review. Ann Fam Med 2014;12:66–74. doi: 10.1370/afm.1593 Colleen Fogarty and Susan McDaniel Family-oriented practice is the application of general systems theory to families; it focuses on the relationship between significant others—their complementary interactions, shared beliefs and effects on each other’s health and behaviours. What do we mean by family ? What are its characteristics? We define family as the many types of relationships formed by strong social bonds over time—family is ‘any set of intimates with a history and a future’. As to general characteristics, families are rarely uniform in nature. They are commonly affected by cultural expectations about independent living, child rearing and kinship networks. They are often formed and reformed through coupling and uncoupling. They are routinely influenced by previous generations and change over time in response to the realities of human development, from birth through death. Physicians who understand how family matters affect healing learn about their patients’ families. Using a systematic assessment informed by five key questions, family physicians work to identify family beliefs and practices that may explain patients’ symptoms. Identifying these beliefs and practices—aided by family genograms —also helps determine helpful resources for treatment planning. The five questions key to family-oriented practice are as follows : Does anyone in the family have the same problem as you do? Knowing that a condition ‘runs in the family’, by whatever mechanism, helps clinicians gauge patients’ understandings of illness. Hearing, ‘Yeah, my uncle lost his leg from diabetes,’ functions as a learning moment to assess a patient’s concern about avoiding diabetic complications. What do family members think is causing your symptom? Family health beliefs are shared understandings about the causes and treatments of an illness. For example, families may approach treatment of the common cold differently. Some caregivers may make chicken soup as a home remedy for a child with the common cold; others may quickly take a child to the doctor at the first sign of sniffles. Who in the family is most concerned about your problem? If an adult patient appears blasé about a condition, wondering aloud who else might be concerned can reveal a spouse in the waiting room or a worried child across the country. This knowledge can assist in identifying the ‘real’ reason for a visit. Have there been other changes or stressors related to your presenting concerns? Recognising the role of contextual factors—think stress related to a new baby, move, divorce, job loss or death in the family—helps clinicians understand their patients’ perspectives. An example: Mr Morrison has worsening chronic renal failure. During a routine visit, he mentions that his wife and he always argue about who is to do yard work. She hears her husband as critical of her abilities; he, now unable to do basic physical tasks, wants her to know what to do after he dies. Such misunderstandings are common. Who in the family can help you with your health problem? Patients regularly struggle with recommendations for change. Encouraging them to bring significant family members to office visits helps promote adherence to treatment plans. Even when patients come alone to consultations, they bring with them a family system rich with shared understandings about symptoms, the meaning of illness, when to consult with professionals and how to treat problems. Physicians who learn about families can leverage these natural support systems in support of patients’ health and well-being. Readings Felix D, Mauksch L. Improving family interviewing skills using the family centered observation form: an online training module for healthcare providers. 2017. Available: http://www.fcof.us [Accessed 31 Jan 2024 ]. Medalie JH, Cole-Kelly K. The clinical importance of defining family. Am Fam Physician 2002;65:1277–9. Waters I, Watson W, Wetzel W. Genograms. Practical tools for family physicians. Can Fam Physician 1994;40:282–7. Felix D, Mauksch L. Improving family interviewing skills using the family centered observation form: an online training module for healthcare providers. 2017. Available: http://www.fcof.us [Accessed 31 Jan 2024 ]. Medalie JH, Cole-Kelly K. The clinical importance of defining family. Am Fam Physician 2002;65:1277–9. Waters I, Watson W, Wetzel W. Genograms. Practical tools for family physicians. Can Fam Physician 1994;40:282–7. Tessa Rohrberg Providing healing care within the intimacy of the doctor–patient relationship defines the family physician as a family member. As I sit in my office and reflect on what ‘family physician as family member’ means, my eye catches a photo on my bulletin board . It is of Arthur and me. I was new to private practice. Arthur was a World War II vet who routinely drove 30 miles each way to see me for his care. What Arthur wanted was not prescriptions or expensive workups to investigate his chronic cough. Rather, he wanted me to know about him, and he wanted to know about me. I learnt about his passion for farming and shared that my dad owned a tractor implements business. While our visits entailed discussions of changes to his cough, they mostly focused on how his crops were doing. One day, I wheeled next to Arthur and snapped a selfie of us—Arthur in his overalls and baseball cap and I in my newly minted white coat. Of course, we as family physicians rely on our medical knowledge to diagnose and treat patients. We have arguably the broadest of medical knowledge given the variety and complexity of the patients we care for. However, beyond the skill of diagnosing and treating disease, we characterise our specialty by the long-term relationships we have with individual patients. While family physicians often care for generations of families and are trained to appreciate family as a social construct within community, family medicine is not named solely for care of the family unit. Lynn Carmichael, a founder of the Society of Teachers of Family Medicine, maintained that family medicine focused on a close familiarity borne of caring for the whole patient rather than the family as a clinical construct. This familiarity emerged from four components: affinity, continuity, intimacy and reciprocity. As with all human relationships, the doctor–patient relationship is susceptible to ups and downs filled with positive and negative emotions. It is through this reciprocal relationship that people’s humanity is revealed; it is this relationship that brings value to both patients and family physicians. Together, we celebrate successes and grieve losses. We, as family physicians, provide a shoulder for patients to lean on. We work into the late hours reading and learning to aid in diagnosis. We coordinate care. We disagree. We admit mistakes. We listen and forgive. We develop a closeness that can only be described as family—part of ‘any group of intimates with a history and a future’. Eventually, Arthur’s cough led to pneumonia, and he ended up in the hospital. Examinations revealed metastatic cancer. In accordance with his wishes, I signed him up for hospice. Thereafter, it was I who drove the 30 miles to visit Arthur at his home. Beside his bed were pictures of his wife and children, a letter of recognition as a veteran and our photo. I saw it again one more time at his funeral service. Looking at this photo now, I am reminded of my cherished connections with patients. I realise they represent what family physician as family member means. Medicine is my profession, but the enduring relationships built through providing holistic care are what I hold closest to heart in the work I do. Readings McWhinney IR. William Pickles Lecture 1996. The importance of being different. Br J Gen Pract 1996;46:433–6. Carmichael LP. Forty families—A search for the family in family medicine. Fam Syst Med 1983;1:12–6. Carmichael L. Voices from family medicine: Lynn Carmichael. Interview by William B. Ventres and John J. Frey. Fam Med 1992;24:53–7. McWhinney IR. William Pickles Lecture 1996. The importance of being different. Br J Gen Pract 1996;46:433–6. Carmichael LP. Forty families—A search for the family in family medicine. Fam Syst Med 1983;1:12–6. Carmichael L. Voices from family medicine: Lynn Carmichael. Interview by William B. Ventres and John J. Frey. Fam Med 1992;24:53–7. Amy Odom Both patients and physicians are influenced by their family experiences as they interact in the examination room. How often do patients bring their families with them into the examination room? How often do you take your family into the examination room with you? Though it is easy to answer the first question with sometimes, and the latter with never, the answer to both questions is always . Individual patients represent a collective make-up of past experiences and family influences. Similarly, our own values and perspectives have developed through our personal and family experiences. Having a family-oriented approach to patient care means considering how patients’ family contexts influence their reasons for seeking medical care and understandings of medical issues. The importance of this cannot be underestimated. For instance, in spousal relationships, significant others’ worries often prompt appointments; however, the patients themselves may not be similarly alarmed. This can lead to apparent non-adherence. For example, I saw a patient because his wife noted a suspicious lesion on his back. Although the patient was not at all troubled by it, the lesion’s irregularity warranted a biopsy. Then, the patient did not show up for the procedure. At a subsequent visit, the wife (who was also my patient) complained to me in frustration about her husband’s lack of concern. Her father had died from malignant melanoma, and she simply needed the security of the biopsy for reassurance. Knowing this, I was able to have a different conversation with her husband. He proceeded with a biopsy, which thankfully was benign. As family physicians, we are often involved in clinical situations that elicit memories or feelings from past and current family experiences. Our ability to pay attention to these emergent concerns—to be mindfully aware —can enhance our empathy and aid us in setting boundaries. Bill is a 40-year-old patient of mine who is currently unemployed and smokes cigarettes. He is overweight and struggling to get control of his diabetes. The specific details of his presentation seem straightforward. Yet, in listening to his story, I am reminded of my father, who navigated a world of unemployment, obesity and chronic disease. I catch myself wondering whether Bill also uses alcohol to cope with his stressors. It is not wrong that I think of my dad when I see Bill, for the familiarity of context helps me connect with my patient. It is important that I recognise this connection, just as it is important that I separate myself from my dad’s experience so that I may see Bill in a unique light and learn his own very personal story. Family experiences inform how we and our patients approach each other; they frame the lens through which we see the world. To create strong therapeutic relationships with our patients, it is imperative we as family physicians seek to understand the influence of the family on both sides of the stethoscope . Readings Candib LM, Savageau JA, Weinreb L, Reed G. Inquiring into our past: when the doctor is a survivor of abuse. Fam Med 2012;44:416-24. Mengel MB, Mauksch LB. Disarming the family ghost: a family of origin experience. Fam Med 1989;21:45-9. Montgomery L, Loue S, Stange KC. Linking the heart and the head: humanism and professionalism in medical education and practice. Fam Med 2017;49:378–83. Candib LM, Savageau JA, Weinreb L, Reed G. Inquiring into our past: when the doctor is a survivor of abuse. Fam Med 2012;44:416-24. Mengel MB, Mauksch LB. Disarming the family ghost: a family of origin experience. Fam Med 1989;21:45-9. Montgomery L, Loue S, Stange KC. Linking the heart and the head: humanism and professionalism in medical education and practice. Fam Med 2017;49:378–83. |
Changes in intracranial neurophysiology associated with acute COVID-19 infection | a4629937-2fa5-452b-b5dc-103b98057cbf | 9896881 | Physiology[mh] | Conception and design of the study: K.K.S., V.R.R., K.W.S., E.F.C., A.D.K. Acquisition and analysis of data: K.K.S., N.S., D.A.A.M., C.H., A.N.K., J.M.F., V.R.R., K.W.S., E.F.C., A.D.K. Drafting manuscript or figures: K.K.S., V.R.R. KWS receives salary and equity options from Neumora Therapeutics. Other authors have nothing to report. |
Virtual social grooming in macaques and its psychophysiological effects | d446c510-edce-494d-b85b-c519527be2b2 | 11111682 | Physiology[mh] | Macaques spend a considerable amount of their time grooming each other. Besides its hygienic function, grooming is considered instrumental in maintaining social bonds and establishing hierarchical structures within groups , and has been put forward as a social commodity, that can be reciprocated or exchanged for other benefits, such as social support in the face of conflicts . The main hypothesis behind this view is that engaging in allogrooming is relaxing and can alleviate stress. This is supported by physiological evidence in various species of macaque monkeys, such as heart rate reduction , and lower blood cortisol levels in groomed individuals, and lower feacal glucocorticoids concentrations in groomers . Behavioral indicators also point at a reduction of anxiety through allogrooming: reduction in self-directed behaviors and displacement activities have been reported following the giving – , and receiving of grooming . These overall benefits are thought to originate in the activation of C-tactile afferents, triggering the release of key neurotransmitters, such as oxytocin, as well as endorphins , during grooming. However, the question arises as to whether the physiological and social benefits derived from grooming stem directly from body stimulation, or whether other mechanisms come into play. In laboratory settings, monkeys sometimes need to be single-housed and are therefore prevented from performing this core social behavior, despite living in colony rooms and having visual access to congeners. Here, we report a recurrent behavior (hereafter: “glassgrooming”) between two laboratory macaques that seems to mimic grooming through a glass panel separating their cages. We investigated how this behavior manifested itself and was reinforced and whether it induced physiological changes similar to actual allogrooming. Over the course of 55 video recording sessions, spanning a period of 16 months, we identified 10 glassgrooming interactions, lasting on average 473.4 s (min = 37 s, max = 802 s; sd = 292.3). These interactions involved active engagement of a fixed “groomee”, and a “groomer”, with the groomee showing body parts he would like to be groomed, and the groomer performing grooming-like sweeping gestures that flexibly adapt to the other’s postural changes (Fig. and Supplementary Video ). Such interactions were preceded by a behavioral sequence lasting on average 19.7 s (min = 4 s, max = 32 s, sd = 10.47), and triggered by either one of the monkeys, or both simultaneously (Supplementary Video ). This sequence started with affiliative (70%) or agonistic (30%) behaviors, but always ended with mutual affiliative behaviors before the beginning of glassgrooming. Both monkeys showed agitated behavior whether the encounter was affiliative or agonistic, with behaviors such as genital presentations, lipsmacking, jumping and tapping the glass panel. In 2 sessions, external conflict in which the groomer was engaged and the groomee provided support preceded glassgrooming. In the 3 cases in which the groomee started the interaction with aggression, the groomer was resting behind the glass and not looking at or engaging with the other monkey. We did not observe other situations in which the groomee started the interaction with aggression (for instance, the groomee resting behind the glass while the groomer approached). In one session, the groomee began the interaction by approaching the glass while the groomer rested behind it, the groomer responded by showing aggressive behaviors, followed by the usual sequence of mutual affiliative displays and then glassgrooming. Therefore, meetings between the monkeys at the glass interface could have outcomes of opposite valence, which both parties could influence. To assess physiological changes associated with glassgrooming, we recorded heart rate in the groomed animal, in 7 sessions. Our analysis reveals that heart rate during glassgrooming is lower than in control resting periods (Wilcoxon rank-sum test, z = − 3.5, p = 0.0004195) (Fig. ). Heart-rate variability measures showed no significant difference between conditions. Experimenter-groomee glassgrooming We successfully managed to transfer the behavior to the experimenter-groomee interaction, on several occasions, indicating this behavior is generalizable (Supplementary Video ). We collected heart rate data in 4 of these interactions and 4 control resting periods. We show that heart rate is significantly lower during human glassgrooming than in matching resting periods. (Wilcoxon rank-sum test, z = − 4.75, p = 0.000001953). Contrary to glassgrooming between groomer and groomee, experimenter glassgrooming showed significantly higher HF values during human glassgrooming (Wilcoxon rank-sum test, z = − 2.53, p = 0.01135) (Fig. ). Dominance test In order to characterize the dominance relationship of the pair, we performed 4 sessions of a food-grab test. The groomee retrieved the food in 20 out of 23 trials (~ 87%). Interestingly, two out of these four tests induced glassgrooming (Supplementary Video ). We successfully managed to transfer the behavior to the experimenter-groomee interaction, on several occasions, indicating this behavior is generalizable (Supplementary Video ). We collected heart rate data in 4 of these interactions and 4 control resting periods. We show that heart rate is significantly lower during human glassgrooming than in matching resting periods. (Wilcoxon rank-sum test, z = − 4.75, p = 0.000001953). Contrary to glassgrooming between groomer and groomee, experimenter glassgrooming showed significantly higher HF values during human glassgrooming (Wilcoxon rank-sum test, z = − 2.53, p = 0.01135) (Fig. ). In order to characterize the dominance relationship of the pair, we performed 4 sessions of a food-grab test. The groomee retrieved the food in 20 out of 23 trials (~ 87%). Interestingly, two out of these four tests induced glassgrooming (Supplementary Video ). We report the existence of a persistent, until now undescribed form of spontaneous social interaction between two laboratory macaque monkeys. Its triggering in situations of competition and conflict, but also affiliative contexts, indicates that it helps regulate tensions amongst the pair, and could strengthen bonds and foster support in the face of external conflict. Furthermore, we show that during glassgrooming, the groomee’s heart rate is lower than in control conditions. Therefore, it seems that this interaction uses the social codes of allogrooming, adapts them to the restricted social life and environment constrained by laboratory settings, and goes so far as to produce physiological effects found to be present in allogrooming , , hinting at a relaxed state of the groomee. HF measures suggest an accompanying vagal tone in the human-groomee interactions which was not significant for monkey glassgrooming. This may be explained by the ultra-short segment durations used here, which are not the standard for HRV measures , , and the rarity of our data did not allow to investigate this further. Considering the complex social dynamics that seem to underlie this behavior, its high degree of flexibility and generalization to a human groomer with the same associated reduction in heart rate, we argue that it does not qualify as stereotypical behavior. Instead, we explain this behavior’s emergence by the fact that the glass interface is a socially charged region with high incertitude regarding meeting outcomes and thus, glassgrooming arose as a tension-reduction mechanism to overcome the evolving, unstable status of the area. For this to be the case, we argue that this behavior must engage socio-affective and reward brain circuits and neuromodulators (especially oxytocin, serotonin and dopamine) to the same extent as real social touch . Indeed, although the monkeys seemingly derive no immediate benefit from this behavior, the fact that it occurs regularly over months and sometimes lasts more than 10 minutes, suggests that it is reinforced by its positive social and physiological effects, during the interaction, and on a larger timescale. We further suggest that these effects are mediated by high-level multisensory mechanisms previously described in humans. The perception of the body and immediate peripersonal space is indeed highly plastic and can give rise to striking illusions, such as the integration of an artificial limb into the body schema , or the experience of tactile sensations from a visual stimulus as seen in mirror-touch synesthesia , . Integrative circuits in the parietal and frontal cortical regions receiving convergent inputs from visual, somatosensory and other modalities , are generally considered to be the neural substrate of bodily self-awareness . We could therefore hypothesize that glassgrooming induces pleasant sensations in the groomed animal through the activation of these neural ensembles by vision. As to the groomer’s participation in this social interaction, one could suggest that vicarious processes could be at play, and engage similar mechanisms and brain circuits that have been found to support empathic responses to both painful and rewarding stimuli , , and the still poorly understood mechanisms involved in giving grooming . Alternatively, we could argue that what appears to be a calming effect stems from mechanisms that do not involve the feeling of touch. Through the association of tactile sensations and a specific individual, which could in our case have occurred during the pair’s youth as they had experienced shared group-housing at that time, the monkeys may be seeking the interaction for itself. However, it is worth noting that the subjects have not been in direct contact for several years, therefore if such a behavioral association underlies this behavior, it would be surprisingly resilient, highlighting the untiring social nature of these primates. The ability of the groomee to show this behavior and associated effects when interacting with the experimenter hints at the fact that an associated bond might be dispensable, and that the physiological effects might arise on the sole basis of allogrooming being an inherent part of macaque social behavior. Therefore, another possibility is that the visual stimulation previously paired, during past real-life grooming, with C-afferent stimulation, produced, through a process akin to pattern completion, the same pleasant sensations and bodily states. Such effects could be mediated through multimodal limbic structures such as the insula and amygdala, that are involved in the processing of pleasant touch in humans and monkeys – . To sum up, this study points at the possibility of additional mechanisms involved in making social touch beneficial, besides direct body stimulation, and of high-level mechanisms that sustain this adapted form of social behavior. The genesis and maintenance of this singular expression of affiliative behavior sheds new light on the extent to which non-human primates perceive and respond to each other's need for social contact. Subjects All procedures conducted on the animals conformed to European and National Institutes of Health Guidelines for the Care and Use of Laboratory Animals, and complied with ARRIVE guidelines. They were approved by the Regional Ethics Committee (CELYNE). The subjects were two adult male fascicularis macaques, aged 13 at the beginning of the study. They were housed individually, in adjacent cages separated by a 8 mm thick glass panel from which they could see each other and the rest of the colony room. The two monkeys arrived in the lab juvelines from a large breeding colony and were initially housed in a 4-animal, male-only group, until puberty, when permanent conflicts made their separation necessary around 8 years before this study. Since then, groomer and groomee have only been single-housed, but in proximity. Glass interface In order to ensure that vibrations transmitted through the glass couldn’t evoke somatosensory stimulation, we placed a piezoelectric sensor on one side of the glass panel and compared the electrical signal output between touching the sensor directly, and from behind the glass panel as during glassgrooming. When applying equal or greater pressure than the groomer and experimenter during glassgrooming, no vibrations were recorded by the sensor on the other side of the glass panel. Data collection Videos were recorded with 2 GoPro Hero 7 cameras that were placed in front of each of the monkeys’ cages, behind a protective plexiglass panel. External batteries were used to permit long video recordings. The first set of sessions were video-only, and were recorded between February and April 2022. The second set of sessions included telemetric recordings of cardiac data and was collected between the months of January and July 2023. Sessions were recorded in the mornings and afternoons, and lasted up to 4 h. The videos were labelled with BORIS 8.8.1 . We also collected glassgrooming data with the experimenter as a groomer. The glass interface involved in this interaction was located in the front of the groomee’s cage, and was not the same as the one in which the two monkeys perform glassgrooming, despite having the same properties. The monkey would sit on the platform behind the glass interface and voluntarily interact with the experimenter. Telemetry device implantation The groomee was implanted with a PhysioTel® Digital L04 telemetry device (Data Sciences International, DSI, St Paul MN, USA). The animal was pre-anesthetized with ketamine and medetomidine and subsequently maintained on isoflurane (1–2%) throughout the surgical procedure. Vital signs, including heart rate, respiratory rate, and blood oxygen saturation, were continuously monitored. The telemetric device was placed on the lower abdomen through a small incision, and its sensors subcutaneously tunneled to the thoracic region. The subject received postoperative analgesia, antibiotic, anti-inflammatory medication prior to and for a few days after surgery. He was closely monitored during the recovery period to ensure proper and comfortable wound healing. The implant was activated at the beginning of each session by placing a magnet in front of it, which the monkey was trained and rewarded for. Two transceivers were placed above and on the side of the cage of the subject, and collected signal from the implant which was transmitted to a dedicated telemetry computer through the communication link controller. The two animals had previously undergone a separate cranial implant surgical procedure for purposes unrelated to the current study. Data processing and analysis A timestamped event was sent to the DSI software Ponemah (v6.51) at the beginning and end of each session to allow the synchronization of video and physiological data with precision to the second. At the end of each session, heart rate data was processed from raw signal as the reciprocal of the RR intervals and exported with the Ponemah software as their harmonic mean over 1-s periods. Heart rate values that were 3 standard deviations above the mean in each condition were considered outliers, and removed from analysis. Videos were coded in BORIS and timestamps aligned with telemetry data using a custom Matlab R2018b code. Analyses were performed using R and R studio , . To serve as control periods, resting events where the monkey was located at the same position than during glassgrooming were selected. Since the approach and presence of other individuals, and anxiety has been shown to modulate heart rate and especially considering our hypothesis regarding the social importance of the glass interface, we restricted our analyses to resting periods when the groomer was also resting nearby. Since the glassgrooming bouts were preceded by energetic behavioral displays and arousing social situations, heart rate tended to greatly rise, and remain high for the first part of the glassgrooming bout. We did not find matching situations that would allow control periods to integrate this aspect in our analysis, and therefore, using the glassgrooming bouts as a whole for comparison with resting skewed results. Tracing heart rate difference in 10 s bins did show this higher starting point for glassgrooming, and rapid decrease compared to resting periods (see Fig. ). Using the first bin that didn’t show a difference (Wilcoxon rank-sum test) between resting and glassgrooming values as the starting boundary for heart rate comparison, allowed to reduce the effect of the pre-glassgrooming sequence on the overall heart rate during glassgrooming. We bounded the analysis window based on the shortest glassgrooming bout of 120 s to control for the effect of the duration of sitting still on HR, excluding one glassgrooming and one matching control resting bout under 60 s for the results shown in Fig. b (n = 6 glassgrooming, and n = 6 control resting bouts). Note that the co-occurrence of autogrooming by the groomee was repeatedly observed during glassgrooming. Autogrooming tended to generate movements and therefore raise the heart rate, which is consistent with previous reports of heart rate data for macaques’ behavior . For the human-experimenter interactions, we collected cardiac data in 4 human glassgrooming bouts, and 4 matching resting periods. They were processed in the same manner as the groomer-groomee data. We made sure that the behaviors preceding them matched in intensity. We trimmed every bout to the shortest measured one, which was of 92 s, to control for the effect of the duration of sitting still on heart rate. Heart rate variability measures were extracted from the fast-Fourier-transformed raw signal in 10 s windows. The high frequency power band, HF, was used as an indicator of vagal tone. The upper and lower limits of this band were set as 0.15–0.5 Hz based on monkey studies , . Dominance test To study dominance, we performed 4 sessions of a food-grab test . We used a homemade pole to place a piece of fruit or peanut (we confirmed the monkey’s liking of the proposed food and desensitization to the pole prior to the start of the test) above the glass separating the cages of the two monkeys. Each trial began by making the monkeys come to the front of their cages so that they would be at equal distance to the food that was subsequently provided thanks to the pole. Grabbing the reward demanded to climb up to the roof of the cage before the other contestant did, and reaching the food by passing the arm through the cage’s wire. Trials were video recorded with two GoPro Hero 7 cameras, and live-annotated by a second experimenter. All procedures conducted on the animals conformed to European and National Institutes of Health Guidelines for the Care and Use of Laboratory Animals, and complied with ARRIVE guidelines. They were approved by the Regional Ethics Committee (CELYNE). The subjects were two adult male fascicularis macaques, aged 13 at the beginning of the study. They were housed individually, in adjacent cages separated by a 8 mm thick glass panel from which they could see each other and the rest of the colony room. The two monkeys arrived in the lab juvelines from a large breeding colony and were initially housed in a 4-animal, male-only group, until puberty, when permanent conflicts made their separation necessary around 8 years before this study. Since then, groomer and groomee have only been single-housed, but in proximity. In order to ensure that vibrations transmitted through the glass couldn’t evoke somatosensory stimulation, we placed a piezoelectric sensor on one side of the glass panel and compared the electrical signal output between touching the sensor directly, and from behind the glass panel as during glassgrooming. When applying equal or greater pressure than the groomer and experimenter during glassgrooming, no vibrations were recorded by the sensor on the other side of the glass panel. Videos were recorded with 2 GoPro Hero 7 cameras that were placed in front of each of the monkeys’ cages, behind a protective plexiglass panel. External batteries were used to permit long video recordings. The first set of sessions were video-only, and were recorded between February and April 2022. The second set of sessions included telemetric recordings of cardiac data and was collected between the months of January and July 2023. Sessions were recorded in the mornings and afternoons, and lasted up to 4 h. The videos were labelled with BORIS 8.8.1 . We also collected glassgrooming data with the experimenter as a groomer. The glass interface involved in this interaction was located in the front of the groomee’s cage, and was not the same as the one in which the two monkeys perform glassgrooming, despite having the same properties. The monkey would sit on the platform behind the glass interface and voluntarily interact with the experimenter. The groomee was implanted with a PhysioTel® Digital L04 telemetry device (Data Sciences International, DSI, St Paul MN, USA). The animal was pre-anesthetized with ketamine and medetomidine and subsequently maintained on isoflurane (1–2%) throughout the surgical procedure. Vital signs, including heart rate, respiratory rate, and blood oxygen saturation, were continuously monitored. The telemetric device was placed on the lower abdomen through a small incision, and its sensors subcutaneously tunneled to the thoracic region. The subject received postoperative analgesia, antibiotic, anti-inflammatory medication prior to and for a few days after surgery. He was closely monitored during the recovery period to ensure proper and comfortable wound healing. The implant was activated at the beginning of each session by placing a magnet in front of it, which the monkey was trained and rewarded for. Two transceivers were placed above and on the side of the cage of the subject, and collected signal from the implant which was transmitted to a dedicated telemetry computer through the communication link controller. The two animals had previously undergone a separate cranial implant surgical procedure for purposes unrelated to the current study. A timestamped event was sent to the DSI software Ponemah (v6.51) at the beginning and end of each session to allow the synchronization of video and physiological data with precision to the second. At the end of each session, heart rate data was processed from raw signal as the reciprocal of the RR intervals and exported with the Ponemah software as their harmonic mean over 1-s periods. Heart rate values that were 3 standard deviations above the mean in each condition were considered outliers, and removed from analysis. Videos were coded in BORIS and timestamps aligned with telemetry data using a custom Matlab R2018b code. Analyses were performed using R and R studio , . To serve as control periods, resting events where the monkey was located at the same position than during glassgrooming were selected. Since the approach and presence of other individuals, and anxiety has been shown to modulate heart rate and especially considering our hypothesis regarding the social importance of the glass interface, we restricted our analyses to resting periods when the groomer was also resting nearby. Since the glassgrooming bouts were preceded by energetic behavioral displays and arousing social situations, heart rate tended to greatly rise, and remain high for the first part of the glassgrooming bout. We did not find matching situations that would allow control periods to integrate this aspect in our analysis, and therefore, using the glassgrooming bouts as a whole for comparison with resting skewed results. Tracing heart rate difference in 10 s bins did show this higher starting point for glassgrooming, and rapid decrease compared to resting periods (see Fig. ). Using the first bin that didn’t show a difference (Wilcoxon rank-sum test) between resting and glassgrooming values as the starting boundary for heart rate comparison, allowed to reduce the effect of the pre-glassgrooming sequence on the overall heart rate during glassgrooming. We bounded the analysis window based on the shortest glassgrooming bout of 120 s to control for the effect of the duration of sitting still on HR, excluding one glassgrooming and one matching control resting bout under 60 s for the results shown in Fig. b (n = 6 glassgrooming, and n = 6 control resting bouts). Note that the co-occurrence of autogrooming by the groomee was repeatedly observed during glassgrooming. Autogrooming tended to generate movements and therefore raise the heart rate, which is consistent with previous reports of heart rate data for macaques’ behavior . For the human-experimenter interactions, we collected cardiac data in 4 human glassgrooming bouts, and 4 matching resting periods. They were processed in the same manner as the groomer-groomee data. We made sure that the behaviors preceding them matched in intensity. We trimmed every bout to the shortest measured one, which was of 92 s, to control for the effect of the duration of sitting still on heart rate. Heart rate variability measures were extracted from the fast-Fourier-transformed raw signal in 10 s windows. The high frequency power band, HF, was used as an indicator of vagal tone. The upper and lower limits of this band were set as 0.15–0.5 Hz based on monkey studies , . To study dominance, we performed 4 sessions of a food-grab test . We used a homemade pole to place a piece of fruit or peanut (we confirmed the monkey’s liking of the proposed food and desensitization to the pole prior to the start of the test) above the glass separating the cages of the two monkeys. Each trial began by making the monkeys come to the front of their cages so that they would be at equal distance to the food that was subsequently provided thanks to the pole. Grabbing the reward demanded to climb up to the roof of the cage before the other contestant did, and reaching the food by passing the arm through the cage’s wire. Trials were video recorded with two GoPro Hero 7 cameras, and live-annotated by a second experimenter. Supplementary Video 1. Supplementary Video 2. Supplementary Video 3. Supplementary Video 4. |
The Effects of Digital eHealth Versus Onsite 2-Day Group-Based Education in 255 Patients With Irritable Bowel Syndrome: Cohort Study | d89e60c4-804a-40f3-bde7-27b5c66a4024 | 11809937 | Patient Education as Topic[mh] | Irritable bowel syndrome (IBS) is a chronic disorder manifested by recurrent abdominal pain and alterations in stool form or frequency . The condition affects between 4% and 9.2% of the global population and it is highly heterogenous. IBS’ unclear etiology involves multifactorial disturbances of the bidirectional communication between the gut and the brain, including visceral hypersensitivity, low-grade inflammatory responses, intestinal motility disturbances, alterations of central nervous system processing, and alterations in gut microbiota composition . However, no clear biomarker or therapeutic target for IBS has been identified. The condition lacks both a cure and medication that gives sufficient symptom relief, a fact that highlights the necessity of integrating nonpharmacological approaches including patient education in patient care . Patients with IBS require personalized treatment for successful symptom relief. Approaches may include physical therapy, cognitive behavioral therapy ( CBT ), hypnotherapy, mindfulness and exposure therapy, and comprehensive dietary guidance by registered dietitians such as the low FODMAP (fermentable oligosaccharides, disaccharides, monosaccharides, and polyols) diet. However, access to these treatment options is often limited due to lack of trained professionals, travel distances, and cost. Thus, more accessible treatment options are warranted. Web-based interventions, patient education, and self-management has been demonstrated to be effective in patients suffering from chronic diseases, including IBS . A systematic review from 2017 concluded with mixed results regarding the effectiveness of web-based mindfulness-based interventions compared with active control treatment conditions such as CBT. However, the study showed that treatment targeting symptoms of IBS had the largest effect size improvements . A longitudinal qualitative study from 2020 showed that both telephone-based CBT and web-based CBT for IBS were positively received and had lasting positive impacts on participants’ understanding of IBS symptoms, quality of life, and IBS-related behaviors . A recent Japanese randomized controlled trial has shown that a multidisciplinary eHealth self-management program can reduce the severity of IBS-symptoms and improved the quality of life . Indeed, limited health care resources, national priority guidelines, and sparse treatment options may restrict any long-term follow-up that patients with IBS may request from a secondary or tertiary care institution. Since 2012, Haukeland University Hospital has offered patients with IBS an onsite multidisciplinary 2-day group-based education program, a so-called “IBS-school.” The multidisciplinary approach is designed to provide patients with evidence-based information and practical skills that may lead to improved IBS management, reduced IBS symptoms, and enhanced quality of life. Based on our clinical experience with group education, we have developed a novel internet-guided multidisciplinary self-care management program for patients with IBS, from now on referred to as the “eHealth program.” The content in the eHealth program is based on topics covered in the IBS-school, but the patients also have access to digital clinical support during the program. Herein, we hypothesized that the eHealth program would have an equally good effect on symptom severity, quality of life, and patient satisfaction, compared with a cohort of patients attending IBS-school. Data from neither program have been published before. In this study, we aimed to investigate the effects of a digital web-based multidisciplinary eHealth program on the domains of symptom severity (Irritable Bowel Syndrome Symptom Severity Scale [IBS-SSS]), quality of life (irritable bowel syndrome quality of life [IBS-QOL]), anxiety and depression (Hospital Anxiety and Depression Scale [HADS]), and a measure of general client satisfaction (client satisfaction questionnaire [CSQ-8]), compared with an onsite multidisciplinary 2-day group-based education program, IBS-school, in 2 cohorts of 255 patients with IBS. Patient Sample, Randomization, and Treatments In this study , 255 patients between 15 and 70 years were recruited after being accepted for patient education at Haukeland University Hospital in Norway between 2017 and 2019. Random patients on the waiting list to the onsite multidisciplinary 2-day group-based education program, IBS-school, were contacted by phone by a study nurse and offered to attend the novel multidisciplinary digital 5-module eHealth program, from here on referred to as the “eHealth program.” Patients attending the IBS-school were recruited on site when attending the 2-day group-based education program. Patients received oral and written information before giving written consent and sent it to the hospital by post. Furthermore, 123 patients that attended the IBS-school were used for comparison (N=255, 1:1). Here, patients were recruited by a study nurse on site. They received an oral and written information about the study before signing consent . Inclusion criteria included (1) being referred to receiving patient education on IBS by a gastroenterologist or general practitioner after the diagnosis of IBS had been determined ( International Classification of Primary Care, Second Editon [ ICPC-2 ] code D93; International Statistical Classification of Diseases, Tenth Revision [ ICD-10 ] code K58), (2) being between 18 and 70 years, and (3) understanding written and oral Norwegian. There were no specific exclusion criteria. Ethical Considerations Eligible patients gave informed written consent and participants were given the option to withdraw from the study at any time point without a specific reason. The data were deidentified and stored on a secure hospital server, and analysis was performed on anonymous data. None of the participants were compensated. The study was approved by the Regional Ethical Committee of Western Norway (REC-2016/1098). Interventions The Irritable Bowel Syndrome–School Patients attended an onsite multidisciplinary 2-day group-based education program, IBS-school, at Haukeland University Hospital. The number of participants varied between 15 and 40 each month. The program involved lectures and question and answer sessions by 4 health care professionals. Refer to for the intervention outline. On day 1, a gastroenterologist talked about the human body and the gastrointestinal system, what IBS is, what causes it, diagnostics, and treatment options (3 hours). Second, a physiotherapist was giving a lecture on body posture and breathing techniques including demonstrations, introducing pain physiology, and the function of the human nervous system (3 hours). On day 2, a patient representative shared her personal experience with IBS (1 hour), and a clinical dietitian gave a lecture on National Institute for Health and Care Excellence (NICE) guidelines that summarizes the most recent recommendations on IBS in adults in primary care, and the low FODMAP diet (3 hours). Finally, a psychiatrist gave a lecture on coping with life including health worry, social relations, tiring thoughts and feelings, adaptations, and symptoms. A nurse with specialization in gastroenterology hosted the group education program to create a comfortable environment for exchange of personal experiences and participate in group assignments. The eHealth Program Patients who were enrolled in the digital eHealth program had access to a comprehensive, multidisciplinary web-based program that consisted of 5 modules (refer to for the outline). The modules consisted of instructive texts, videos, animations, and images over 150 web-pages on a digital treatment platform by CheckWare AS . In addition, patients carried out “home assignments” based on principles of CBT, a protocol by Ljótsson et al at the Swedish Karolinska Institute, followed by an optional low FODMAP diet intervention guided by a clinical dietitian. The eHealth program allowed a secure login (eg, bank-ID) and digital communication between patient and a clinical dietitian throughout the process, at the patient’s own request and need. The patients were expected to finish the program over a period of 3‐12 weeks, at his or her own pace, all dependent on their motivation and work capacity. Module contents have been described further in this study. Module 1 describes the body and the gastrointestinal system. In this first module, the patient gets an introduction to what IBS is, how it is diagnosed, and what causes it. The patient learns about the function of our digestive system and how it is regulated, and how it can be disturbed in people with IBS. Module 2 describes the posture and breathing techniques. This module focuses on the connection between IBS and musculoskeletal disorders. Many people with IBS may have a “hyperactive” nervous system that can cause pain and physical maladjustments. In this module, a physiotherapist introduces the patient to pain physiology and how the nervous system works. There is a practical section with useful exercises for people with IBS. The module will give the patient an understanding of how long-term pain occurs, why it often persists, and how to influence it. Module 3 consists of diet and lifestyle advices. There is no miracle cure for IBS, but many people experience improvement by following some general advice. In this module, we look at lifestyle advice that has been shown by research to improve symptoms in people with IBS. It is recommended to try this before eliminating other foods or following strict diets. Module 4 focuses on coping with life. In this module, the patient learns techniques from cognitive therapy including “the cognitive diamond,” mindfulness, and practice systematic exposure exercises has previously shown beneficial effects for patients with IBS. The protocol has been described in a study by Ljótsson et al . Module 5 describes the dietary intervention with a low FODMAP diet. This is a dietary treatment that provides symptom relief in approximately 70% of patients with IBS . In this module, the patient will be introduced to the low FODMAP diet in both theory and practice. The patient had access to digital guidance by a clinical dietitian during the entire course of the study. Questionnaires Overview Patients completed questionnaires related to IBS symptom severity (IBS-SSS) , quality of life, (IBS-QOL) , and anxiety and depression (HADS) upon enrollment at baseline and after 3 months. The IBS-SSS is considered the gold standard measure of IBS symptoms and contains 5 questions that measure the frequency of abdominal pain, the severity of abdominal distention, dissatisfaction with bowel habits, and interference with quality of life, scored in the range of 0‐500. A higher score indicating worse condition, scores <175 represent mild IBS symptoms, 175‐300 represents moderate severity, scores >300 represent severe IBS . The IBS-QOL is a condition-specific measure for assessing health-related quality of life in IBS. It consists of 34 items, each with a 5-point response scale in a range of 0‐100, where the higher score indicates a better IBS specific quality of life. There are 8 subscale scores for the IBS-QOL: dysphoria, interference with activity, body image, health worry, food avoidance, social reaction, sexual, and relationships . HADS is a scale of 14 items designed to measure anxiety and depression, 7 items for each subscale (ie, anxiety and depression). The total score is the sum of the 14 items, and for each subscale the score is the sum of the respective 7 items that range from 0 to 21 . CSQ-8 was completed at 3 months. The questionnaire is an 8-item measure with a total score ranging from 8 to 32, with the higher number indicating greater satisfaction with treatment. Primary Outcome Measure and Definition of Intervention Responders In the field of IBS research, a change of 50 points in IBS-SSS score has shown to reliably indicate improvement in IBS symptom severity . Our primary outcome measure was a >50-point reduction in IBS-SSS at 3 months, compared with baseline. Patients reporting a ≥50-point reduction in IBS-SSS at 3 months were categorized as “responders” to the intervention. Patients reporting <50-points on IBS-SSS were categorized as “nonresponders” to the intervention. Statistical Analysis Statistical analyses were performed using SPSS Statistics (version 26, IBM) for Microsoft Windows. Baseline patient characteristics and questionnaires were first presented according to intervention—eHealth group and IBS-school. At 3 months after intervention, patients were categorized as responders or nonresponders. For comparison between groups, unpaired t test were performed for parametric data and Mann-Whitney U for nonparametric data. For comparisons before and after interventions, paired t tests were performed for parametric data and Wilcoxon signed rank test for nonparametric data. A P value <.05 was considered statistically significant. The intervention responder or nonresponder analysis to the eHealth program or IBS-school was carried out performing a paired t test on data from patients who filled out the 3-month questionnaires (n=71 and n=50, respectively). A χ 2 test of independence was used to assess differences between intervention response. In this study , 255 patients between 15 and 70 years were recruited after being accepted for patient education at Haukeland University Hospital in Norway between 2017 and 2019. Random patients on the waiting list to the onsite multidisciplinary 2-day group-based education program, IBS-school, were contacted by phone by a study nurse and offered to attend the novel multidisciplinary digital 5-module eHealth program, from here on referred to as the “eHealth program.” Patients attending the IBS-school were recruited on site when attending the 2-day group-based education program. Patients received oral and written information before giving written consent and sent it to the hospital by post. Furthermore, 123 patients that attended the IBS-school were used for comparison (N=255, 1:1). Here, patients were recruited by a study nurse on site. They received an oral and written information about the study before signing consent . Inclusion criteria included (1) being referred to receiving patient education on IBS by a gastroenterologist or general practitioner after the diagnosis of IBS had been determined ( International Classification of Primary Care, Second Editon [ ICPC-2 ] code D93; International Statistical Classification of Diseases, Tenth Revision [ ICD-10 ] code K58), (2) being between 18 and 70 years, and (3) understanding written and oral Norwegian. There were no specific exclusion criteria. Eligible patients gave informed written consent and participants were given the option to withdraw from the study at any time point without a specific reason. The data were deidentified and stored on a secure hospital server, and analysis was performed on anonymous data. None of the participants were compensated. The study was approved by the Regional Ethical Committee of Western Norway (REC-2016/1098). The Irritable Bowel Syndrome–School Patients attended an onsite multidisciplinary 2-day group-based education program, IBS-school, at Haukeland University Hospital. The number of participants varied between 15 and 40 each month. The program involved lectures and question and answer sessions by 4 health care professionals. Refer to for the intervention outline. On day 1, a gastroenterologist talked about the human body and the gastrointestinal system, what IBS is, what causes it, diagnostics, and treatment options (3 hours). Second, a physiotherapist was giving a lecture on body posture and breathing techniques including demonstrations, introducing pain physiology, and the function of the human nervous system (3 hours). On day 2, a patient representative shared her personal experience with IBS (1 hour), and a clinical dietitian gave a lecture on National Institute for Health and Care Excellence (NICE) guidelines that summarizes the most recent recommendations on IBS in adults in primary care, and the low FODMAP diet (3 hours). Finally, a psychiatrist gave a lecture on coping with life including health worry, social relations, tiring thoughts and feelings, adaptations, and symptoms. A nurse with specialization in gastroenterology hosted the group education program to create a comfortable environment for exchange of personal experiences and participate in group assignments. The eHealth Program Patients who were enrolled in the digital eHealth program had access to a comprehensive, multidisciplinary web-based program that consisted of 5 modules (refer to for the outline). The modules consisted of instructive texts, videos, animations, and images over 150 web-pages on a digital treatment platform by CheckWare AS . In addition, patients carried out “home assignments” based on principles of CBT, a protocol by Ljótsson et al at the Swedish Karolinska Institute, followed by an optional low FODMAP diet intervention guided by a clinical dietitian. The eHealth program allowed a secure login (eg, bank-ID) and digital communication between patient and a clinical dietitian throughout the process, at the patient’s own request and need. The patients were expected to finish the program over a period of 3‐12 weeks, at his or her own pace, all dependent on their motivation and work capacity. Module contents have been described further in this study. Module 1 describes the body and the gastrointestinal system. In this first module, the patient gets an introduction to what IBS is, how it is diagnosed, and what causes it. The patient learns about the function of our digestive system and how it is regulated, and how it can be disturbed in people with IBS. Module 2 describes the posture and breathing techniques. This module focuses on the connection between IBS and musculoskeletal disorders. Many people with IBS may have a “hyperactive” nervous system that can cause pain and physical maladjustments. In this module, a physiotherapist introduces the patient to pain physiology and how the nervous system works. There is a practical section with useful exercises for people with IBS. The module will give the patient an understanding of how long-term pain occurs, why it often persists, and how to influence it. Module 3 consists of diet and lifestyle advices. There is no miracle cure for IBS, but many people experience improvement by following some general advice. In this module, we look at lifestyle advice that has been shown by research to improve symptoms in people with IBS. It is recommended to try this before eliminating other foods or following strict diets. Module 4 focuses on coping with life. In this module, the patient learns techniques from cognitive therapy including “the cognitive diamond,” mindfulness, and practice systematic exposure exercises has previously shown beneficial effects for patients with IBS. The protocol has been described in a study by Ljótsson et al . Module 5 describes the dietary intervention with a low FODMAP diet. This is a dietary treatment that provides symptom relief in approximately 70% of patients with IBS . In this module, the patient will be introduced to the low FODMAP diet in both theory and practice. The patient had access to digital guidance by a clinical dietitian during the entire course of the study. Patients attended an onsite multidisciplinary 2-day group-based education program, IBS-school, at Haukeland University Hospital. The number of participants varied between 15 and 40 each month. The program involved lectures and question and answer sessions by 4 health care professionals. Refer to for the intervention outline. On day 1, a gastroenterologist talked about the human body and the gastrointestinal system, what IBS is, what causes it, diagnostics, and treatment options (3 hours). Second, a physiotherapist was giving a lecture on body posture and breathing techniques including demonstrations, introducing pain physiology, and the function of the human nervous system (3 hours). On day 2, a patient representative shared her personal experience with IBS (1 hour), and a clinical dietitian gave a lecture on National Institute for Health and Care Excellence (NICE) guidelines that summarizes the most recent recommendations on IBS in adults in primary care, and the low FODMAP diet (3 hours). Finally, a psychiatrist gave a lecture on coping with life including health worry, social relations, tiring thoughts and feelings, adaptations, and symptoms. A nurse with specialization in gastroenterology hosted the group education program to create a comfortable environment for exchange of personal experiences and participate in group assignments. Patients who were enrolled in the digital eHealth program had access to a comprehensive, multidisciplinary web-based program that consisted of 5 modules (refer to for the outline). The modules consisted of instructive texts, videos, animations, and images over 150 web-pages on a digital treatment platform by CheckWare AS . In addition, patients carried out “home assignments” based on principles of CBT, a protocol by Ljótsson et al at the Swedish Karolinska Institute, followed by an optional low FODMAP diet intervention guided by a clinical dietitian. The eHealth program allowed a secure login (eg, bank-ID) and digital communication between patient and a clinical dietitian throughout the process, at the patient’s own request and need. The patients were expected to finish the program over a period of 3‐12 weeks, at his or her own pace, all dependent on their motivation and work capacity. Module contents have been described further in this study. Module 1 describes the body and the gastrointestinal system. In this first module, the patient gets an introduction to what IBS is, how it is diagnosed, and what causes it. The patient learns about the function of our digestive system and how it is regulated, and how it can be disturbed in people with IBS. Module 2 describes the posture and breathing techniques. This module focuses on the connection between IBS and musculoskeletal disorders. Many people with IBS may have a “hyperactive” nervous system that can cause pain and physical maladjustments. In this module, a physiotherapist introduces the patient to pain physiology and how the nervous system works. There is a practical section with useful exercises for people with IBS. The module will give the patient an understanding of how long-term pain occurs, why it often persists, and how to influence it. Module 3 consists of diet and lifestyle advices. There is no miracle cure for IBS, but many people experience improvement by following some general advice. In this module, we look at lifestyle advice that has been shown by research to improve symptoms in people with IBS. It is recommended to try this before eliminating other foods or following strict diets. Module 4 focuses on coping with life. In this module, the patient learns techniques from cognitive therapy including “the cognitive diamond,” mindfulness, and practice systematic exposure exercises has previously shown beneficial effects for patients with IBS. The protocol has been described in a study by Ljótsson et al . Module 5 describes the dietary intervention with a low FODMAP diet. This is a dietary treatment that provides symptom relief in approximately 70% of patients with IBS . In this module, the patient will be introduced to the low FODMAP diet in both theory and practice. The patient had access to digital guidance by a clinical dietitian during the entire course of the study. Overview Patients completed questionnaires related to IBS symptom severity (IBS-SSS) , quality of life, (IBS-QOL) , and anxiety and depression (HADS) upon enrollment at baseline and after 3 months. The IBS-SSS is considered the gold standard measure of IBS symptoms and contains 5 questions that measure the frequency of abdominal pain, the severity of abdominal distention, dissatisfaction with bowel habits, and interference with quality of life, scored in the range of 0‐500. A higher score indicating worse condition, scores <175 represent mild IBS symptoms, 175‐300 represents moderate severity, scores >300 represent severe IBS . The IBS-QOL is a condition-specific measure for assessing health-related quality of life in IBS. It consists of 34 items, each with a 5-point response scale in a range of 0‐100, where the higher score indicates a better IBS specific quality of life. There are 8 subscale scores for the IBS-QOL: dysphoria, interference with activity, body image, health worry, food avoidance, social reaction, sexual, and relationships . HADS is a scale of 14 items designed to measure anxiety and depression, 7 items for each subscale (ie, anxiety and depression). The total score is the sum of the 14 items, and for each subscale the score is the sum of the respective 7 items that range from 0 to 21 . CSQ-8 was completed at 3 months. The questionnaire is an 8-item measure with a total score ranging from 8 to 32, with the higher number indicating greater satisfaction with treatment. Primary Outcome Measure and Definition of Intervention Responders In the field of IBS research, a change of 50 points in IBS-SSS score has shown to reliably indicate improvement in IBS symptom severity . Our primary outcome measure was a >50-point reduction in IBS-SSS at 3 months, compared with baseline. Patients reporting a ≥50-point reduction in IBS-SSS at 3 months were categorized as “responders” to the intervention. Patients reporting <50-points on IBS-SSS were categorized as “nonresponders” to the intervention. Patients completed questionnaires related to IBS symptom severity (IBS-SSS) , quality of life, (IBS-QOL) , and anxiety and depression (HADS) upon enrollment at baseline and after 3 months. The IBS-SSS is considered the gold standard measure of IBS symptoms and contains 5 questions that measure the frequency of abdominal pain, the severity of abdominal distention, dissatisfaction with bowel habits, and interference with quality of life, scored in the range of 0‐500. A higher score indicating worse condition, scores <175 represent mild IBS symptoms, 175‐300 represents moderate severity, scores >300 represent severe IBS . The IBS-QOL is a condition-specific measure for assessing health-related quality of life in IBS. It consists of 34 items, each with a 5-point response scale in a range of 0‐100, where the higher score indicates a better IBS specific quality of life. There are 8 subscale scores for the IBS-QOL: dysphoria, interference with activity, body image, health worry, food avoidance, social reaction, sexual, and relationships . HADS is a scale of 14 items designed to measure anxiety and depression, 7 items for each subscale (ie, anxiety and depression). The total score is the sum of the 14 items, and for each subscale the score is the sum of the respective 7 items that range from 0 to 21 . CSQ-8 was completed at 3 months. The questionnaire is an 8-item measure with a total score ranging from 8 to 32, with the higher number indicating greater satisfaction with treatment. In the field of IBS research, a change of 50 points in IBS-SSS score has shown to reliably indicate improvement in IBS symptom severity . Our primary outcome measure was a >50-point reduction in IBS-SSS at 3 months, compared with baseline. Patients reporting a ≥50-point reduction in IBS-SSS at 3 months were categorized as “responders” to the intervention. Patients reporting <50-points on IBS-SSS were categorized as “nonresponders” to the intervention. Statistical analyses were performed using SPSS Statistics (version 26, IBM) for Microsoft Windows. Baseline patient characteristics and questionnaires were first presented according to intervention—eHealth group and IBS-school. At 3 months after intervention, patients were categorized as responders or nonresponders. For comparison between groups, unpaired t test were performed for parametric data and Mann-Whitney U for nonparametric data. For comparisons before and after interventions, paired t tests were performed for parametric data and Wilcoxon signed rank test for nonparametric data. A P value <.05 was considered statistically significant. The intervention responder or nonresponder analysis to the eHealth program or IBS-school was carried out performing a paired t test on data from patients who filled out the 3-month questionnaires (n=71 and n=50, respectively). A χ 2 test of independence was used to assess differences between intervention response. Patients and Characteristics A flowchart of participating patients is illustrated in . Of the 132 eligible patients who gave written consent to participate in the eHealth program group of the study, 7 did not fill out the minimum requirement including the electronic case report form and the IBS-SSS questionnaire. Of the 125 patients who completed the program, 54 patients did not respond to the IBS-SSS questionnaire at 3 months. Hence, the follow-up data from patients attending the eHealth program reduced to n=71 at 3 months. Of the 123 eligible patients who gave written consent to participate in the IBS-school group of the study, 5 did not fill out the minimum requirement including the electronic case report form and the IBS-SSS questionnaire. Thus, of the 118 patients who completed the program, 69 patients did not respond to the IBS-SSS questionnaire at 3 months. Hence, the follow-up data from patients attending the eHealth program reduced to n=49 at 3 months. Participants were predominantly female (190/235, 81%) with a mean age of 38.3 (SD 12.4) years (Characteristics are summarized in ). An unpaired t test showed that there was no difference in age between groups. A Mann-Whitney U test was performed to evaluate whether sex differed between groups, reveling no significant difference between patients attending the eHealth program and patients attending the IBS-school (Table S1 in , U =6407.0, z =−1.394, P =.16). Furthermore, participants at the IBS-school and eHealth program displayed similar baseline characteristics upon enrollment, including moderate to severe IBS symptom severity. There was no significant difference between most relevant features at baseline, except for anxiety and depression, as shown in . Here, patients attending the eHealth program scored an average of 11 (SD 5.2) which corresponds to mild to moderate anxiety, while patients attending the IBS-school scored 9.4 (SD 5.1), corresponding to a mild level of anxiety (2-tailed t 232 = 2.37, P =.02). Thus, both patient groups displayed an anxiety score typical of a clinical case of anxiety (HADS-anxiety ≥8 is considered a clinical case ), and patients attending the eHealth program presented a significantly higher baseline score than patients attending the IBS-school. Furthermore, patients attending the eHealth program reported significantly higher depression scores than patients attending the IBS-school (mean 7, SD 4.1 vs mean 5.9, SD 4, respectively; t 232 =2.01, P =.045). However, both groups displayed an average depression score corresponding to nonclinical severity of depression (HADS-depression ≤8 is considered noncase ) which gives the difference little relevance on group level. Symptom Relief and Enhanced Quality of Life Overview A paired t test revealed that patients attending the eHealth program reported a significant reduction in IBS symptom severity at 3 months, shown in (mean 279.3, SD 80 vs mean 242.7, SD 79.3; t 70 =4.91, P <.001). Comparably, patients attending the IBS-school did not report a significant reduction in symptom severity (mean 248.6, SD 86.6 vs mean 235.4, SD 98.9; t 49 =0.85, P =.40). Refer to Table S2 of for data. However, none of the groups achieved a clinically meaningful improvement of ≥50 points on IBS-SSS at 3 months. Patients in the eHealth program-group reported a 37-point reduction, whereas patients attending IBS-school reported a 13-point reduction in IBS-SSS total score (Table S2 of ). Thus, for further analysis we categorized participants as a “responder” or “non-responder” to the intervention, where the sample response threshold for responders was set as ≥50-point decrease in total IBS-SSS score, a threshold that has been demonstrated to correlate with improvements in clinical symptoms . Responders Of the 71 patients, 29 (41%) attending the eHealth program responded to the intervention with a reduction in IBS-SSS score of ≥50 points, compared with 36% (18/50) who attended the IBS-school after 3 months . The results in this section are summarized in . Patients categorized as responders to the eHealth program reported a significant mean reduction in IBS-SSS score of 103 (SD 72.0) points with a moderate effect size of 0.56 ( t 28 =7.62, Cohen d =1.33, P <.001). In addition, eHealth program–responding patients reported a significant reduction in anxiety, changing from a clinical case to a noncase classification (mean −4.7, SD 6.2; t 28 =6.89, Cohen d =0.85, P <.001). Here, the effect size of 0.39 indicated a moderate effect. On average, these patients also reported an increase in overall quality of life, summarized in (mean 5.80, SD 19.1; t 28 = −2.28, Cohen d =−0.30, P =.03). Here, features such as dysphoria (mean 18, SD 23.3; t 28 = −2.1, Cohen d =−0.85, P <.001) and body image (mean 14.8, SD 18.6; t 28 =−3.9, Cohen d =−0.72, P <.001) increased significantly with a moderate effect size. Features such as food avoidance (mean 6.3, SD 22.5; t 28 =−2.10, Cohen d =−0.13, P= .045), health worry (mean 10.8, SD 23.5; t 28 =−3.44, Cohen d =−0.43, P =.002), interference with activity (mean 13.4, SD 18.8; t 28 =−4.03, Cohen d =−0.63, P <.001), relations (mean 9.5, SD 22.9; t 28 =−2.72, Cohen d =−0.37, P =.01), and social relations (mean 10.7, SD 19.4; t 28 =−3.62, Cohen d =−0.56, P =.001), all improved significantly, compared with baseline. The changes in sexual activity were not significantly different after 3 months (mean 5.4, SD 29.8; t 28 =−1.52, Cohen d =−0.19, P =.14). Patients who responded to the IBS-school intervention reported an average reduction in IBS symptom severity score of 119 (SD 86.2) points (n=18; t 17 =5.54), Cohen d =1.33, P <.001). In addition, patients reported a reduction in depression scores (mean −2.3, SD 4.4; t 15 =2.17, Cohen d , P =.046). However, these patients did not report significant improvements in anxiety or overall quality of life (mean 2.5, SD 6.8; t 15 =1.97, Cohen d =0.43, P =.07 and mean 2.4, SD 27.9; t 7 =−0.57, Cohen d =−0.08, P =.64, respectively). Furthermore, the number of responding patients were too low for further in-depth statistical analysis of quality of life (n=8). Data for these analyses are presented in Table S3 of . Nonresponders Patients who were nonresponders to the eHealth program reported no improvement in symptom severity or quality of life (mean 9.2, SD 70.4; t 41 =−1.73, Cohen d =−0.13, P =.09) and a significant decrease in anxiety scores (mean −2.4, SD 4.8, t 38 =3.65, Cohen d =0.52, P =.001), compared with baseline. However, both changes were of small effect ( r =−0.27 and 0.25, respectively). Results are summarized in . There were no significant changes in depression scores or overall quality of life or the respective sub-categories ( P >.05). Nonresponding patients attending IBS-school reported significantly enhanced symptom scores (mean 46.5, SD 77.8; t 31 =−4.08, Cohen d =−0.57, P <.001). However, it is worth noting that these enhanced symptoms scores were not above the 50-point threshold for clinical relevance. Furthermore, these patients reported a significantly reduced overall quality of life (mean −11.4, SD 23.8; t 23 =3.16, Cohen d =5.52, P =.004) and a significant increase in the domain of relationships (mean 18.7, SD 17; t 23 =−2.11, Cohen d =−0.86, P =.046). Baseline characteristics for these analyses are presented in Table S3 of . Association Between Type of Intervention and Outcome A χ 2 test of independence analysis was used to test whether the type of intervention was independent from the intervention outcome. Results showed that the proportion of patients who responded to the intervention in both groups (29/71 vs 18/50) were the same. Hence, there was no significant association between the type of intervention and response to the intervention (Pearson χ 2 1 =0.2, Φ=0.041, P =.65). Thus, we conclude that there is no difference in treatment response between the eHealth program and IBS-school in affecting symptom severity scores. Data from this analysis is reported in Table S4 of . Patient Satisfaction Patient satisfaction was investigated in both groups 3 months after enrollment (CSQ-8). Patients who attended the IBS-school reported a mean score of 24.2 (SD 3.7), compared with patients attending the eHealth program with a mean score of 23.5 (SD 4.0), which are both equivalent to “good” health care offers . An unpaired t test revealed no significant difference between group scores ( t 115 = −1.032, P =.31; ). A flowchart of participating patients is illustrated in . Of the 132 eligible patients who gave written consent to participate in the eHealth program group of the study, 7 did not fill out the minimum requirement including the electronic case report form and the IBS-SSS questionnaire. Of the 125 patients who completed the program, 54 patients did not respond to the IBS-SSS questionnaire at 3 months. Hence, the follow-up data from patients attending the eHealth program reduced to n=71 at 3 months. Of the 123 eligible patients who gave written consent to participate in the IBS-school group of the study, 5 did not fill out the minimum requirement including the electronic case report form and the IBS-SSS questionnaire. Thus, of the 118 patients who completed the program, 69 patients did not respond to the IBS-SSS questionnaire at 3 months. Hence, the follow-up data from patients attending the eHealth program reduced to n=49 at 3 months. Participants were predominantly female (190/235, 81%) with a mean age of 38.3 (SD 12.4) years (Characteristics are summarized in ). An unpaired t test showed that there was no difference in age between groups. A Mann-Whitney U test was performed to evaluate whether sex differed between groups, reveling no significant difference between patients attending the eHealth program and patients attending the IBS-school (Table S1 in , U =6407.0, z =−1.394, P =.16). Furthermore, participants at the IBS-school and eHealth program displayed similar baseline characteristics upon enrollment, including moderate to severe IBS symptom severity. There was no significant difference between most relevant features at baseline, except for anxiety and depression, as shown in . Here, patients attending the eHealth program scored an average of 11 (SD 5.2) which corresponds to mild to moderate anxiety, while patients attending the IBS-school scored 9.4 (SD 5.1), corresponding to a mild level of anxiety (2-tailed t 232 = 2.37, P =.02). Thus, both patient groups displayed an anxiety score typical of a clinical case of anxiety (HADS-anxiety ≥8 is considered a clinical case ), and patients attending the eHealth program presented a significantly higher baseline score than patients attending the IBS-school. Furthermore, patients attending the eHealth program reported significantly higher depression scores than patients attending the IBS-school (mean 7, SD 4.1 vs mean 5.9, SD 4, respectively; t 232 =2.01, P =.045). However, both groups displayed an average depression score corresponding to nonclinical severity of depression (HADS-depression ≤8 is considered noncase ) which gives the difference little relevance on group level. A paired t test revealed that patients attending the eHealth program reported a significant reduction in IBS symptom severity at 3 months, shown in (mean 279.3, SD 80 vs mean 242.7, SD 79.3; t 70 =4.91, P <.001). Comparably, patients attending the IBS-school did not report a significant reduction in symptom severity (mean 248.6, SD 86.6 vs mean 235.4, SD 98.9; t 49 =0.85, P =.40). Refer to Table S2 of for data. However, none of the groups achieved a clinically meaningful improvement of ≥50 points on IBS-SSS at 3 months. Patients in the eHealth program-group reported a 37-point reduction, whereas patients attending IBS-school reported a 13-point reduction in IBS-SSS total score (Table S2 of ). Thus, for further analysis we categorized participants as a “responder” or “non-responder” to the intervention, where the sample response threshold for responders was set as ≥50-point decrease in total IBS-SSS score, a threshold that has been demonstrated to correlate with improvements in clinical symptoms . Responders Of the 71 patients, 29 (41%) attending the eHealth program responded to the intervention with a reduction in IBS-SSS score of ≥50 points, compared with 36% (18/50) who attended the IBS-school after 3 months . The results in this section are summarized in . Patients categorized as responders to the eHealth program reported a significant mean reduction in IBS-SSS score of 103 (SD 72.0) points with a moderate effect size of 0.56 ( t 28 =7.62, Cohen d =1.33, P <.001). In addition, eHealth program–responding patients reported a significant reduction in anxiety, changing from a clinical case to a noncase classification (mean −4.7, SD 6.2; t 28 =6.89, Cohen d =0.85, P <.001). Here, the effect size of 0.39 indicated a moderate effect. On average, these patients also reported an increase in overall quality of life, summarized in (mean 5.80, SD 19.1; t 28 = −2.28, Cohen d =−0.30, P =.03). Here, features such as dysphoria (mean 18, SD 23.3; t 28 = −2.1, Cohen d =−0.85, P <.001) and body image (mean 14.8, SD 18.6; t 28 =−3.9, Cohen d =−0.72, P <.001) increased significantly with a moderate effect size. Features such as food avoidance (mean 6.3, SD 22.5; t 28 =−2.10, Cohen d =−0.13, P= .045), health worry (mean 10.8, SD 23.5; t 28 =−3.44, Cohen d =−0.43, P =.002), interference with activity (mean 13.4, SD 18.8; t 28 =−4.03, Cohen d =−0.63, P <.001), relations (mean 9.5, SD 22.9; t 28 =−2.72, Cohen d =−0.37, P =.01), and social relations (mean 10.7, SD 19.4; t 28 =−3.62, Cohen d =−0.56, P =.001), all improved significantly, compared with baseline. The changes in sexual activity were not significantly different after 3 months (mean 5.4, SD 29.8; t 28 =−1.52, Cohen d =−0.19, P =.14). Patients who responded to the IBS-school intervention reported an average reduction in IBS symptom severity score of 119 (SD 86.2) points (n=18; t 17 =5.54), Cohen d =1.33, P <.001). In addition, patients reported a reduction in depression scores (mean −2.3, SD 4.4; t 15 =2.17, Cohen d , P =.046). However, these patients did not report significant improvements in anxiety or overall quality of life (mean 2.5, SD 6.8; t 15 =1.97, Cohen d =0.43, P =.07 and mean 2.4, SD 27.9; t 7 =−0.57, Cohen d =−0.08, P =.64, respectively). Furthermore, the number of responding patients were too low for further in-depth statistical analysis of quality of life (n=8). Data for these analyses are presented in Table S3 of . Nonresponders Patients who were nonresponders to the eHealth program reported no improvement in symptom severity or quality of life (mean 9.2, SD 70.4; t 41 =−1.73, Cohen d =−0.13, P =.09) and a significant decrease in anxiety scores (mean −2.4, SD 4.8, t 38 =3.65, Cohen d =0.52, P =.001), compared with baseline. However, both changes were of small effect ( r =−0.27 and 0.25, respectively). Results are summarized in . There were no significant changes in depression scores or overall quality of life or the respective sub-categories ( P >.05). Nonresponding patients attending IBS-school reported significantly enhanced symptom scores (mean 46.5, SD 77.8; t 31 =−4.08, Cohen d =−0.57, P <.001). However, it is worth noting that these enhanced symptoms scores were not above the 50-point threshold for clinical relevance. Furthermore, these patients reported a significantly reduced overall quality of life (mean −11.4, SD 23.8; t 23 =3.16, Cohen d =5.52, P =.004) and a significant increase in the domain of relationships (mean 18.7, SD 17; t 23 =−2.11, Cohen d =−0.86, P =.046). Baseline characteristics for these analyses are presented in Table S3 of . Association Between Type of Intervention and Outcome A χ 2 test of independence analysis was used to test whether the type of intervention was independent from the intervention outcome. Results showed that the proportion of patients who responded to the intervention in both groups (29/71 vs 18/50) were the same. Hence, there was no significant association between the type of intervention and response to the intervention (Pearson χ 2 1 =0.2, Φ=0.041, P =.65). Thus, we conclude that there is no difference in treatment response between the eHealth program and IBS-school in affecting symptom severity scores. Data from this analysis is reported in Table S4 of . Of the 71 patients, 29 (41%) attending the eHealth program responded to the intervention with a reduction in IBS-SSS score of ≥50 points, compared with 36% (18/50) who attended the IBS-school after 3 months . The results in this section are summarized in . Patients categorized as responders to the eHealth program reported a significant mean reduction in IBS-SSS score of 103 (SD 72.0) points with a moderate effect size of 0.56 ( t 28 =7.62, Cohen d =1.33, P <.001). In addition, eHealth program–responding patients reported a significant reduction in anxiety, changing from a clinical case to a noncase classification (mean −4.7, SD 6.2; t 28 =6.89, Cohen d =0.85, P <.001). Here, the effect size of 0.39 indicated a moderate effect. On average, these patients also reported an increase in overall quality of life, summarized in (mean 5.80, SD 19.1; t 28 = −2.28, Cohen d =−0.30, P =.03). Here, features such as dysphoria (mean 18, SD 23.3; t 28 = −2.1, Cohen d =−0.85, P <.001) and body image (mean 14.8, SD 18.6; t 28 =−3.9, Cohen d =−0.72, P <.001) increased significantly with a moderate effect size. Features such as food avoidance (mean 6.3, SD 22.5; t 28 =−2.10, Cohen d =−0.13, P= .045), health worry (mean 10.8, SD 23.5; t 28 =−3.44, Cohen d =−0.43, P =.002), interference with activity (mean 13.4, SD 18.8; t 28 =−4.03, Cohen d =−0.63, P <.001), relations (mean 9.5, SD 22.9; t 28 =−2.72, Cohen d =−0.37, P =.01), and social relations (mean 10.7, SD 19.4; t 28 =−3.62, Cohen d =−0.56, P =.001), all improved significantly, compared with baseline. The changes in sexual activity were not significantly different after 3 months (mean 5.4, SD 29.8; t 28 =−1.52, Cohen d =−0.19, P =.14). Patients who responded to the IBS-school intervention reported an average reduction in IBS symptom severity score of 119 (SD 86.2) points (n=18; t 17 =5.54), Cohen d =1.33, P <.001). In addition, patients reported a reduction in depression scores (mean −2.3, SD 4.4; t 15 =2.17, Cohen d , P =.046). However, these patients did not report significant improvements in anxiety or overall quality of life (mean 2.5, SD 6.8; t 15 =1.97, Cohen d =0.43, P =.07 and mean 2.4, SD 27.9; t 7 =−0.57, Cohen d =−0.08, P =.64, respectively). Furthermore, the number of responding patients were too low for further in-depth statistical analysis of quality of life (n=8). Data for these analyses are presented in Table S3 of . Patients who were nonresponders to the eHealth program reported no improvement in symptom severity or quality of life (mean 9.2, SD 70.4; t 41 =−1.73, Cohen d =−0.13, P =.09) and a significant decrease in anxiety scores (mean −2.4, SD 4.8, t 38 =3.65, Cohen d =0.52, P =.001), compared with baseline. However, both changes were of small effect ( r =−0.27 and 0.25, respectively). Results are summarized in . There were no significant changes in depression scores or overall quality of life or the respective sub-categories ( P >.05). Nonresponding patients attending IBS-school reported significantly enhanced symptom scores (mean 46.5, SD 77.8; t 31 =−4.08, Cohen d =−0.57, P <.001). However, it is worth noting that these enhanced symptoms scores were not above the 50-point threshold for clinical relevance. Furthermore, these patients reported a significantly reduced overall quality of life (mean −11.4, SD 23.8; t 23 =3.16, Cohen d =5.52, P =.004) and a significant increase in the domain of relationships (mean 18.7, SD 17; t 23 =−2.11, Cohen d =−0.86, P =.046). Baseline characteristics for these analyses are presented in Table S3 of . A χ 2 test of independence analysis was used to test whether the type of intervention was independent from the intervention outcome. Results showed that the proportion of patients who responded to the intervention in both groups (29/71 vs 18/50) were the same. Hence, there was no significant association between the type of intervention and response to the intervention (Pearson χ 2 1 =0.2, Φ=0.041, P =.65). Thus, we conclude that there is no difference in treatment response between the eHealth program and IBS-school in affecting symptom severity scores. Data from this analysis is reported in Table S4 of . Patient satisfaction was investigated in both groups 3 months after enrollment (CSQ-8). Patients who attended the IBS-school reported a mean score of 24.2 (SD 3.7), compared with patients attending the eHealth program with a mean score of 23.5 (SD 4.0), which are both equivalent to “good” health care offers . An unpaired t test revealed no significant difference between group scores ( t 115 = −1.032, P =.31; ). Principal Findings In this study, we have shown that the novel digital multidisciplinary eHealth program has a significant reducing effect on IBS symptom severity and is useful as a tool in disease self-management. In total, 41% (29/71) of participants reported significant and clinically relevant symptom relief. Furthermore, we show that the eHealth program is safe, as patients not responding to the intervention reported unchanged symptoms and quality of life at 3 months. In addition, eHealth intervention–responding patients reported significant benefits on multiple domains of IBS-related quality of life such as body image, food avoidance, health worry, interference with activity, relations, and social relations. Levels of anxiety were significantly reduced, and levels of dysphoria were improved. Comparably, 36% (18/50) of participants reported a clinically significant effect to the onsite multidisciplinary 2-day group-based education program, the so-called “IBS-school.” Thus, our results indicate that the digital multidisciplinary eHealth program may be equally effective to the IBS-school, which is often a standard treatment offer to newly diagnosed patients. Furthermore, patients responding to the IBS-school intervention did not report any significant improvements in quality of life or in anxiety, but a small not clinically meaningful decrease in depression scores. The number of IBS-school responders were too low for more in-depth statistical analysis on the domains of quality of life. A χ 2 test of independence showed no difference between intervention outcome in the 2 groups; hence the eHealth program did not have a better intervention response than the IBS-school on the measures of symptom severity. However, the eHealth program had a significant effect on other aspects of IBS symptomatology, including anxiety and quality of life, whereas IBS-school did not. Patients rated both the eHealth program and the IBS-school as good health care offers on measures of patient satisfaction of health care quality, scoring them 23.5 and 24.2 out of a maximum of 32 points, respectively. We believe these results indicate that a digital eHealth approach, designed to provide patients with evidence-based information and practical skills, is preferable to an onsite multidisciplinary 2-day group-based education program covering the same topics. Comparison With Previous Work Initially at baseline in both groups, the greatest impairment in quality of life was observed for the subscale of food avoidance followed by body image, inactivity, and dysphoria. This order of impairment is similar to findings by Drossmann et al in an international study from 2009 (n=1966). However, our findings show lower scores on all subscales except dysphoria and health worries, which are in the same magnitude. In comparison with a newer study on quality of life by Kopczyńska et al (n=87), our baseline results show much lower scores on both overall and all subscales of quality of life. A recent Vietnamese study showed that patient education, lifestyle, and dietary intervention, administered by clinical pharmacists, improved IBS related quality of life compared with standard medical therapy over 8 weeks . These study patients also showed a much higher baseline and postintervention quality of life-scores compared with our study. We speculate that our study participants represent a group with more severe IBS because they are referred to a tertiary health care institution that due to capacity issues has to prioritize those patients who need it the most. However, the low baseline scores on food avoidance in our study indicate that both our interventions with a key focus on diet was the right call. This core problem was targeted because food is a known important trigger of IBS-symptoms . On the domain of food avoidance, we observed an improvement in eHealth program-responders, compared with baseline . No such improvement was observed in patients attending the IBS-school. We speculate that this is due to the differences in comprehensiveness and durability. The eHealth program is designed to engage the patient interactively to learn how to make appropriate changes in diet. It offers a thorough guidance with videos and instructions on how to follow the low FODMAP diet, including the exclusion and reintroduction of foods. In addition, the patient has the option of digital question and answer sessions with a clinical dietitian throughout the program. The IBS-school is also designed for patients to learn how to self-help, but the program is much shorter with its 2 days of physical attendance. In the light of these results, we deduce that the eHealth program is not inferior to the established IBS-school as a health care offer. In fact, we show that an eHealth intervention program can provide the patient with self-help tools that can lead to reduced gastrointestinal symptoms and enhance quality of life for patients with IBS. This aligns with a recent randomized controlled study by Tayama et al (n=40) showing that a multidisciplinary eHealth self-management program leads to an increased intake of FODMAP-groups and subsequently a more extensive diet. Severe food avoidance and dietary restriction is previously reported in 13% (829/955) of IBS-patients and this subgroup of IBS-patients reported more severe IBS-symptoms, reduced quality of life, and reduced intake of nutrients . Thus, providing patients with evidence-based information, practical tools, and support by a clinical dietitian is important in clinical care of patients with IBS. eHealth programs in the form of apps, internet-guided programs, or telehealth has recently accelerated as useful tools in clinical medicine. In an American study on satisfaction during COVID-19, most patients with IBS reported high satisfaction rates and ease of use with telehealth . Here the authors reported on multiple benefits including the patient having to take less time off from work and improved access to the care team. Their most commonly reported challenges with telehealth included feeling impersonal and being unable to address all of their issues or concerns. A majority felt that telehealth was as good as or better than face-to-face visits and would use telehealth for future care. Only approximately 10% (130/1311) of the patients remained dissatisfied. In 2020, a Polish study showed that an educational program combined with elements of behavioral therapy, individualized for patients with IBS, is an important part of therapy . In addition, as a part of a dietetic-led gastroenterology service in primary care, feasibility, acceptability, and cost-efficiency of using webinars to deliver first-line patient education for patients with IBS, has been shown to be successful . A meta-analysis of chronic gastrointestinal illness interventions (19 studies conducted in 8 countries, n=3193) showed that eHealth gastrointestinal interventions improved patients’ quality of life, psychological distress, medication adherence, and illness-related knowledge . The meta-analysis also showed that eHealth gastrointestinal interventions significantly reduced the number of patient visits to the hospital. Taken together with our results, these findings support eHealth interventions holding decent promise in improving outcomes for patients with IBS. Comparably on patient satisfaction, a randomized controlled trial by Lackner et al showed that patients who received 4 gastroenterologist-led patient education sessions over 2 weeks reported 26.6 in patient satisfaction. Here, 43.5% (63/145) of patients reported improvement in addition to a higher satisfaction score than in our study. However, both these and our results reflect that subjective symptom relief may not be required for the patient to experience the treatment as useful. This is also highlighted by our patients attending the IBS-school who did not experience significant enhancement in quality of life nor reduction in symptom severity, but still reported a higher satisfaction than patients who objectively benefited more from the eHealth program. Limitations All 255 recruited patients completed the programs. However, there was a very high percentage that did not submit the 3-month questionnaires. In total, 41% (54/132) percent of patients attending the eHealth program, and 54% (67/123) of the patients who attended the IBS-school did not submit the 3-month follow-up questionnaires on IBS symptoms severity . Unfortunately, this occurred despite multiple efforts and reminders (phone calls, emails, and SMS text messages) from the research team encouraging the participants to respond. First, this high percentage of missing data are significant and may have affected the results of the study even though the rates appear similar across the groups. However, high dropout rates are a common issue in eHealth studies and known as “the law of attrition” . This warrants the need for a larger study where an intention-to-treat analysis can be carried out without diminishing the power of the study. Second, eHealth literacy is defined by Norman and Skinner as “the ability to seek, find, understand, and appraise health information from electronic sources and apply the knowledge gained to addressing or solving a health problem”. The individual patient’s level of health literacy was not mapped during the study and variations in these abilities may have affected our results. Third, the patient population represents all subtypes of IBS and there are no analyses focusing on the differences in response between patients with predominant diarrhea, constipation, or a mix of the 2. Comorbidities or other additional diagnoses are common in IBS but have not been excluded in this study and may have affected the results. Fourth, the use of drugs during the study period have not been reported. Hence, many drugs have side effects such as nausea, vomiting, diarrhea, constipation, flatulence, which are symptoms that the patient may confused with IBS symptoms. Fifth, a high placebo effect, which can be up to 40%, is a known challenge in clinical studies on IBS . Although this study has not been designed with a control group, the placebo effect may have affected our results in either group and have not been adjusted for. However, we may speculate that the phenomenon has affected patients in both groups equally and importantly, the placebo effect may recede after 12 weeks, which was our end point . Sixth, as both programs are broad and cover a variety of information, advice, and treatment including the low FODMAP diet and principles of CBT, another limitation of this study is the unknown specifics patients were responding to in either program. Indeed, there are no measures on cognition verifying the direct effects of the principles of CBT or exposure therapy. This needs to be further investigated in a prospective study, and we acknowledge that an in-depth explanation for our observed benefits after attending the eHealth program remain to be clarified. Thus, these aspects are objectives in our currently ongoing randomized controlled trial . In the light of limited primary and secondary health care resources, it will be useful to develop prediction tools to identify which patients may achieve improvement in both symptom severity and domains of quality of life. For these stratification analyses to be clinically meaningful, the number of participants need to be higher than reported in this study and performed in and randomized controlled trial. Conclusions We conclude that the digital multidisciplinary eHealth program has a significant effect on IBS symptom severity in a portion of patients, and is useful as a tool in disease self-management. In addition, it does not result in worse symptom scores than an onsite multidisciplinary 2-day group-based education program after 3 months. We believe these results indicate that a digital eHealth approach, that include benefits such as 3 months unlimited access to quality assured information and treatment with documented effect, is preferable to an onsite multidisciplinary 2-day group-based education program covering the same topics. In this study, we have shown that the novel digital multidisciplinary eHealth program has a significant reducing effect on IBS symptom severity and is useful as a tool in disease self-management. In total, 41% (29/71) of participants reported significant and clinically relevant symptom relief. Furthermore, we show that the eHealth program is safe, as patients not responding to the intervention reported unchanged symptoms and quality of life at 3 months. In addition, eHealth intervention–responding patients reported significant benefits on multiple domains of IBS-related quality of life such as body image, food avoidance, health worry, interference with activity, relations, and social relations. Levels of anxiety were significantly reduced, and levels of dysphoria were improved. Comparably, 36% (18/50) of participants reported a clinically significant effect to the onsite multidisciplinary 2-day group-based education program, the so-called “IBS-school.” Thus, our results indicate that the digital multidisciplinary eHealth program may be equally effective to the IBS-school, which is often a standard treatment offer to newly diagnosed patients. Furthermore, patients responding to the IBS-school intervention did not report any significant improvements in quality of life or in anxiety, but a small not clinically meaningful decrease in depression scores. The number of IBS-school responders were too low for more in-depth statistical analysis on the domains of quality of life. A χ 2 test of independence showed no difference between intervention outcome in the 2 groups; hence the eHealth program did not have a better intervention response than the IBS-school on the measures of symptom severity. However, the eHealth program had a significant effect on other aspects of IBS symptomatology, including anxiety and quality of life, whereas IBS-school did not. Patients rated both the eHealth program and the IBS-school as good health care offers on measures of patient satisfaction of health care quality, scoring them 23.5 and 24.2 out of a maximum of 32 points, respectively. We believe these results indicate that a digital eHealth approach, designed to provide patients with evidence-based information and practical skills, is preferable to an onsite multidisciplinary 2-day group-based education program covering the same topics. Initially at baseline in both groups, the greatest impairment in quality of life was observed for the subscale of food avoidance followed by body image, inactivity, and dysphoria. This order of impairment is similar to findings by Drossmann et al in an international study from 2009 (n=1966). However, our findings show lower scores on all subscales except dysphoria and health worries, which are in the same magnitude. In comparison with a newer study on quality of life by Kopczyńska et al (n=87), our baseline results show much lower scores on both overall and all subscales of quality of life. A recent Vietnamese study showed that patient education, lifestyle, and dietary intervention, administered by clinical pharmacists, improved IBS related quality of life compared with standard medical therapy over 8 weeks . These study patients also showed a much higher baseline and postintervention quality of life-scores compared with our study. We speculate that our study participants represent a group with more severe IBS because they are referred to a tertiary health care institution that due to capacity issues has to prioritize those patients who need it the most. However, the low baseline scores on food avoidance in our study indicate that both our interventions with a key focus on diet was the right call. This core problem was targeted because food is a known important trigger of IBS-symptoms . On the domain of food avoidance, we observed an improvement in eHealth program-responders, compared with baseline . No such improvement was observed in patients attending the IBS-school. We speculate that this is due to the differences in comprehensiveness and durability. The eHealth program is designed to engage the patient interactively to learn how to make appropriate changes in diet. It offers a thorough guidance with videos and instructions on how to follow the low FODMAP diet, including the exclusion and reintroduction of foods. In addition, the patient has the option of digital question and answer sessions with a clinical dietitian throughout the program. The IBS-school is also designed for patients to learn how to self-help, but the program is much shorter with its 2 days of physical attendance. In the light of these results, we deduce that the eHealth program is not inferior to the established IBS-school as a health care offer. In fact, we show that an eHealth intervention program can provide the patient with self-help tools that can lead to reduced gastrointestinal symptoms and enhance quality of life for patients with IBS. This aligns with a recent randomized controlled study by Tayama et al (n=40) showing that a multidisciplinary eHealth self-management program leads to an increased intake of FODMAP-groups and subsequently a more extensive diet. Severe food avoidance and dietary restriction is previously reported in 13% (829/955) of IBS-patients and this subgroup of IBS-patients reported more severe IBS-symptoms, reduced quality of life, and reduced intake of nutrients . Thus, providing patients with evidence-based information, practical tools, and support by a clinical dietitian is important in clinical care of patients with IBS. eHealth programs in the form of apps, internet-guided programs, or telehealth has recently accelerated as useful tools in clinical medicine. In an American study on satisfaction during COVID-19, most patients with IBS reported high satisfaction rates and ease of use with telehealth . Here the authors reported on multiple benefits including the patient having to take less time off from work and improved access to the care team. Their most commonly reported challenges with telehealth included feeling impersonal and being unable to address all of their issues or concerns. A majority felt that telehealth was as good as or better than face-to-face visits and would use telehealth for future care. Only approximately 10% (130/1311) of the patients remained dissatisfied. In 2020, a Polish study showed that an educational program combined with elements of behavioral therapy, individualized for patients with IBS, is an important part of therapy . In addition, as a part of a dietetic-led gastroenterology service in primary care, feasibility, acceptability, and cost-efficiency of using webinars to deliver first-line patient education for patients with IBS, has been shown to be successful . A meta-analysis of chronic gastrointestinal illness interventions (19 studies conducted in 8 countries, n=3193) showed that eHealth gastrointestinal interventions improved patients’ quality of life, psychological distress, medication adherence, and illness-related knowledge . The meta-analysis also showed that eHealth gastrointestinal interventions significantly reduced the number of patient visits to the hospital. Taken together with our results, these findings support eHealth interventions holding decent promise in improving outcomes for patients with IBS. Comparably on patient satisfaction, a randomized controlled trial by Lackner et al showed that patients who received 4 gastroenterologist-led patient education sessions over 2 weeks reported 26.6 in patient satisfaction. Here, 43.5% (63/145) of patients reported improvement in addition to a higher satisfaction score than in our study. However, both these and our results reflect that subjective symptom relief may not be required for the patient to experience the treatment as useful. This is also highlighted by our patients attending the IBS-school who did not experience significant enhancement in quality of life nor reduction in symptom severity, but still reported a higher satisfaction than patients who objectively benefited more from the eHealth program. All 255 recruited patients completed the programs. However, there was a very high percentage that did not submit the 3-month questionnaires. In total, 41% (54/132) percent of patients attending the eHealth program, and 54% (67/123) of the patients who attended the IBS-school did not submit the 3-month follow-up questionnaires on IBS symptoms severity . Unfortunately, this occurred despite multiple efforts and reminders (phone calls, emails, and SMS text messages) from the research team encouraging the participants to respond. First, this high percentage of missing data are significant and may have affected the results of the study even though the rates appear similar across the groups. However, high dropout rates are a common issue in eHealth studies and known as “the law of attrition” . This warrants the need for a larger study where an intention-to-treat analysis can be carried out without diminishing the power of the study. Second, eHealth literacy is defined by Norman and Skinner as “the ability to seek, find, understand, and appraise health information from electronic sources and apply the knowledge gained to addressing or solving a health problem”. The individual patient’s level of health literacy was not mapped during the study and variations in these abilities may have affected our results. Third, the patient population represents all subtypes of IBS and there are no analyses focusing on the differences in response between patients with predominant diarrhea, constipation, or a mix of the 2. Comorbidities or other additional diagnoses are common in IBS but have not been excluded in this study and may have affected the results. Fourth, the use of drugs during the study period have not been reported. Hence, many drugs have side effects such as nausea, vomiting, diarrhea, constipation, flatulence, which are symptoms that the patient may confused with IBS symptoms. Fifth, a high placebo effect, which can be up to 40%, is a known challenge in clinical studies on IBS . Although this study has not been designed with a control group, the placebo effect may have affected our results in either group and have not been adjusted for. However, we may speculate that the phenomenon has affected patients in both groups equally and importantly, the placebo effect may recede after 12 weeks, which was our end point . Sixth, as both programs are broad and cover a variety of information, advice, and treatment including the low FODMAP diet and principles of CBT, another limitation of this study is the unknown specifics patients were responding to in either program. Indeed, there are no measures on cognition verifying the direct effects of the principles of CBT or exposure therapy. This needs to be further investigated in a prospective study, and we acknowledge that an in-depth explanation for our observed benefits after attending the eHealth program remain to be clarified. Thus, these aspects are objectives in our currently ongoing randomized controlled trial . In the light of limited primary and secondary health care resources, it will be useful to develop prediction tools to identify which patients may achieve improvement in both symptom severity and domains of quality of life. For these stratification analyses to be clinically meaningful, the number of participants need to be higher than reported in this study and performed in and randomized controlled trial. We conclude that the digital multidisciplinary eHealth program has a significant effect on IBS symptom severity in a portion of patients, and is useful as a tool in disease self-management. In addition, it does not result in worse symptom scores than an onsite multidisciplinary 2-day group-based education program after 3 months. We believe these results indicate that a digital eHealth approach, that include benefits such as 3 months unlimited access to quality assured information and treatment with documented effect, is preferable to an onsite multidisciplinary 2-day group-based education program covering the same topics. 10.2196/43618 Multimedia Appendix 1 Supplementary tables. |
Herbert Coddington Major (1850–1921) | f6bc8b2a-a23f-4099-8142-89f728bbcf7d | 10973065 | Pathology[mh] | |
Clinical effects of CYP2D6 phenoconversion in patients with psychosis | 8b7732f6-3ae5-48ec-9600-fd25ea0bc2e2 | 11528948 | Pharmacology[mh] | Antipsychotics, or drugs used in psychosis management, are prescribed for a variety of disorders, most commonly psychotic disorders . Although pharmacological treatment was effective compared to placebo, treatment failure remains common . Discontinuation or switching of medication may occur due to high frequency of or burdensome side effects, or because of a lack of perceived symptom improvement . It is estimated that pharmacological treatment fails in an estimated 30% of patients with schizophrenia starting psychotropic medication, demonstrating a clear need to improve patient care . Heterogeneity at patient, disease and environmental levels is thought to contribute to a lack of response to medication. One source of heterogeneity arises from genes encoding the metabolic enzymes responsible for medication breakdown . Common genetic variants can lead to increased or decreased enzyme activity, resulting in faster or slower metabolism of medication . The hepatic cytochrome P450 (CYP) 2D6 is responsible for the metabolization of 20 drugs for psychosis (Drugbank) ( https://go.drugbank.com/categories/DBCAT002623 ) and approximately 25% of all medications , making it a useful target for personalized medicine. However, CYP2D6 is a complex and highly polymorphic target. To facilitate translation from genotype to phenotype and stimulate standardization, an activity score (AS) system has been proposed . Here, each star-annotated variant encodes for a specific AS, ranging from 0 (no activity) to 1 (normal activity). Summing both allele-predicted AS leads to a predicted phenotype; poor metabolizer (PM) (AS = 0), intermediate metabolizer (IM) (AS = 0.5), normal metabolizer (NM) (AS = 1–2), or ultrarapid metabolizer (UM) (AS > 2, due to copy number variants) . Several groups, including the Dutch Pharmacogenetic Working Group (DPWG) and the Clinical Pharmacogenetics Implementation Consortium, have formulated pharmacogenetic dosing guidelines, including for drugs for psychosis and depression, and (among other pharmacogenes) CYP2D6 . Although pharmacogenetics is suggested as potentially beneficial in clinical practice, its application is still limited. This may be partly explained by the theoretical nature and challenges that arise when working with individual patients with comorbidity and polypharmacy. Polypharmacy can result in phenoconversion; the change in CYP activity due to the use of exogenous substances such as alcohol, caffeine, tobacco or a concomitant medication that can affect CYP enzyme activity, resulting in a different phenotype than would be predicted based on the individual’s DNA. In drug–drug interaction studies, the substrate or ‘victim’ drug is one where the plasma concentrations are altered due to inhibition or induction of the ‘perpetrator’ drug, that is, the drug that affects the involved metabolic pathway . The severity of phenoconversion depends on the strength and timing of the perpetrator’s drug . In psychiatry, simultaneous use of CYP2D6 substrates is not uncommon . The noradrenergic and dopaminergic drug bupropion, and the serotonin reuptake inhibitors (SERTs) fluoxetine and paroxetine are all strong inhibitors, effectively reducing any individual’s metabolic capacity. It is hypothesized that this may result in a situation where patients could be considered IM or PM when multiple medications metabolized through the same CYP are used, even when genetic testing indicates they are NM individuals . As a result, the concentration of the CYP2D6 -metabolized medication exceeds what would be expected based on genotype, increasing the risk of adverse drug effects . Since the prescription of drugs for depression in combination with drugs for psychosis is increasingly common , it would be advantageous to consider phenoconversion in pharmacogenetic studies and it should always be part of clinical application. Not all psychosis treatment drugs would be affected by phenoconversion of CYP2D6, as there are differences in metabolic pathways and involvement of CYP2D6. To be affected by CYP2D6 and possible phenoconversion, it should be a major substrate of the enzyme. This is the case for example, haloperidol, risperidone and aripiprazole, but not for example, quetiapine and olanzapine . The DPWG guidelines only recommend dose alterations due to CYP2D6 for aripiprazole, brexiprazole, haloperidol, pimozide, risperidone and zuclopenthixol . Furthermore, current evidence on the effect of non-NM phenotype in users of drugs for psychosis is mixed . Proper examination of phenoconversion is often lacking, which may contribute to conflicting results as patients can be misclassified into a metabolizer group based on only their genotype. Thus, there is a need for representation of the real-world value of pharmacogenetics including phenoconversion. Here, we retrospectively evaluated the frequency and impact of phenoconversion in a relatively large Dutch cohort of patients with psychosis (GROUP cohort; . As phenoconversion-corrected phenotype (pPT) could be a more accurate representation of the actual metabolic capacity affecting outcome compared to genotype-predicted phenotype (gPT), we hypothesize it may also be more closely associated with symptom severity, side effects and well-being of patients measured using questionnaires. The effect of CYP2D6 is expected only for substrate (‘victim’) drugs, whereas drugs metabolized primarily through other routes should remain largely unaffected. As an exploratory analysis, we also examined phenoconversion in the sample stratified by sex. Sex is often included as a covariate, correcting for any differences that may be related to sex. As a result, information on sex-related differences may be lost. With this, we aim to contribute to a better understanding of differences between men and women in the influence and impact of CYP2D6 phenotype and phenoconversion. Study design and participants A retrospective design was employed using the baseline data from the prospective Genetic Risk And Outcome of Psychosis (GROUP) cohort, the details of which have been described elsewhere . Patients were identified through clinicians active in local psychosis departments or academic centres in the Netherlands and (Dutch speaking part of) Belgium. Both in- and out-patients were eligible. Inclusion criteria were as follows: (1) between 16 and 50 years of age; (2) diagnosed with non-affective psychotic disorder according to the Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition (DSM-IV) , with a maximum period of time since first contact with mental health care of 5 years; (3) good command of the Dutch language; and (4) able and willing to give written informed consent. The study was conducted in accordance with the Declaration of Helsinki and the study protocol was approved by the Ethical Review Board of the University Medical Centre Utrecht centrally (protocol code 04/003, 13 May 2004), and by each local review board per participating institute. Patients who were taking at least one drug for psychosis as a treatment drug at baseline, and for whom gPT was available, were included in the current study. Patients who had an unknown phenotype or who reported chlorpromazine-equivalent dose of their main treatment drug of below 25 mg or above 1000 mg per day were excluded. Measurements Outcome For each outcome, data collected at baseline were used. Abnormal Involuntary Movement Scale (AIMS) Abnormal movements in the facial and oral area, trunk, and extremities are measured by the clinician or research team, to form an assessment of the extrapyramidal symptom of dyskinesia, together with a global assessment . Barnes Akathisia Rating Scale (BARS) Akathisia presents as motor restlessness, part of the extrapyramidal symptom group. The BARS is a commonly used scale to measure clinical features of akathisia based on clinician observation, patient report of restlessness, and patient report of distress due to restlessness, amounting to a total score to indicate an overall assessment of akathisia . The Unified Parkinson’s Disease Rating Scale (UPDRS) Parkinsonian movements can be characterized by slowed movements, rigidity and tremors. The UPDRS is a valid and widely used instrument in Parkinson’s disease research and provides another measure of extrapyramidal symptoms . The Subjective Well-being Under Neuroleptics Scale (SWN20) The SWN20 provides a measure of subjective well-being using five subscales: emotional regulation, mental functioning, physical functioning, social integration and self-control . Positive and Negative Syndrome Scale (PANSS) The PANSS combines the Brief Psychiatric Rating Scale and Psychopathology Rating Schedule to assess positive and negative symptoms of schizophrenia as well as general psychopathology using a 30-item questionnaire. Scale-specific items are summed to calculate a score on the positive and negative symptom subscales, which have been shown to have strong reliability, validity and sensitivity . Covariates Sex, age, ethnicity, number of psychotic episodes, illness duration in years, chlorpromazine-equivalent dose of the main treatment drug, total number of treatment (for psychosis) drugs prescribed, total number of medications prescribed, smoking status and hormonal birth control status were included as covariates. Body mass index (BMI) was not available or computable for this time point and was therefore not included as a covariate. Genotyping From each subject, 20 ml of blood was collected at the participating mental health institutes. The blood sample was sent to the University Medical Centre Utrecht by mail, where DNA extraction was performed from peripheral blood lymphocytes. Quality control procedures were performed using PLINK v1.9 according to standard protocol (Purcell and Chang; ). Genotyping and quality control details have been described elsewhere and in the Supplemental material . For an extensive overview of genotype–phenotype translation of CYP2D6 , we also refer to the supplemental material of . Briefly, genotypes of 570,038 single-nucleotide variants were characterized for 2818 individuals (including patients, siblings, parents, and healthy controls), using a customized Illumina Institute of Psychological Medicine and Clinical Neurology chip array and were subjected to an extensive quality control procedure. Non-genotyped variants were imputed using the 1000 Genome Phase 3 (v5) reference panel using the Michigan Imputation Server . A post-imputation filter of R2 > 0.5 was applied to include only high-quality rare variants. Genotype-predicted UM could not be assessed as structure variants (including copy number variants) were not included in the GWAS panels. CYP2D6 phenotype prediction and phenoconversion calculation Stargazer, a python-based bioinformatics tool, was used to predict metabolizer groups of CYP2D6 . Imputed data for chromosome 22 were used to identify CYP2D6 star alleles (haplotypes), and an AS was generated accordingly. A functional phenotype classification per individual was generated. To improve accuracy, phenotype classifications were cross-checked with the PharmVar genotype–phenotype translation tables and the Pharmacogenomics Knowledgebase ( ; PharmGKB), as explained in more detail by . Concurrent CYP2D6 -inhibitor use was checked per patient. First, all recorded drug information was checked and standardized (i.e. medication names were converted to identical spelling of the generic name, and entries with missing or unclear information on medication type, use, or dose were removed). Known CYP2D6 inhibitors were marked according to strength: weak, moderate and strong, according to . Inducers were not included, since there are no known CYP2D6 inducers (see Supplemental Table S1 ). Following this, a new variable representing pPT was created. All individuals were assigned a pPT. pPT was the same as gPT if the patient was not using any CYP2D6 inhibitors (e.g. the pPT would be NM if gPT was NM and the patient did not use any inhibitors) and was adjusted to a reduced metabolism type according to the methods of , accounting for inhibitor strength. Weak inhibitors do not cause phenoconversion, while strong inhibitors such as paroxetine and fluoxetine have previously been shown to reduce most individuals to PM . Although true conversion to PM can only be confirmed using serum measurements, we used the method of , that is, assigning a pPT based on inhibitor strength, to represent clinical practicality as most clinicians will not have serum measurements available. Thus, individuals taking a strong inhibitor were assigned a poor pPT (pPM) in all cases (i.e. NM and IM individuals were assigned pPM. PM individuals cannot be further reduced, and would remain pPM). If an individual was only taking a weak inhibitor, the pPT remained the same as their gPT. Moderate inhibitors would only affect normal gPT (gNM) individuals, reducing their status to intermediate (pIM), but not reduce intermediate metabolizers (gIM) to pPM. If individuals were taking multiple inhibitors, the highest strength was used to assign pPT. Figure S1 (Supplemental material) provides an overview. Statistical analysis Between-group differences between phenotype groups and between the most commonly reported drug subsamples were examined using Kruskal-Wallis and Dunn tests (for continuous variables) and Fisher’s exact and/or Pearson’s Chi-square tests (for categorical variables). Both differences between gPT groups and pPT groups were examined. Statistical analyses were performed using R version 4.2.3 . To investigate whether pPT is more strongly associated with each outcome compared to gPT, regression models were generated and compared to find the best model, for the most used drugs separately. Linear models were used for continuous outcomes (SWN20, AIMS, UPDRS, PANSS), and a proportional odds cumulative logit model for categorical (BARS), as the BARS has six outcomes (none, minimal, mild, moderate, severe and extreme). Multiple models were explored per outcome, using a backward stepwise variable selection method. This was done twice, once using gPT as a predictor, and once with pPT. The best-fit model per outcome was selected using the Akaike information criterion (AIC) and R 2 . The AIC is an information criteria-based relative fit index developed as an estimation of the accuracy of a model to predict outcomes in samples other than the provided data, and R 2 represents the percentage of total variation in the outcome that is explained by the predictors. According to the AIC rule of thumb, if the difference in AIC between two models is less than two, this suggests the models are indistinguishable. The ‘stats’ package was used for the Mann–Whitney U test, Chi-square test, Kruskal–Wallis rank sum test, Fisher’s exact test for count data, and general linear models . Dunn tests were performed using the dunn_test function from the ‘rstatix’ package . The ‘VGAM: Vector Generalized Linear and Additive Models’ package provided the vglm function for multinomial models. The step4vglm function from this package was used to select the final best multinomial model, using backward direction and specifying the genotype variables to remain included using the scope argument of this function . Similarly, stepAIC from the ‘MASS’ package was used to select the best-fitting GLM models . AIC was automatically compared across models per outcome using the aictab function from the ‘AICcmodavg’ package . A p- value of <0.05 was considered statistically significant. All reported p -values are corrected for multiple testing using the Bonferroni method. GROUP release number 8.0 was used for the current analysis. A retrospective design was employed using the baseline data from the prospective Genetic Risk And Outcome of Psychosis (GROUP) cohort, the details of which have been described elsewhere . Patients were identified through clinicians active in local psychosis departments or academic centres in the Netherlands and (Dutch speaking part of) Belgium. Both in- and out-patients were eligible. Inclusion criteria were as follows: (1) between 16 and 50 years of age; (2) diagnosed with non-affective psychotic disorder according to the Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition (DSM-IV) , with a maximum period of time since first contact with mental health care of 5 years; (3) good command of the Dutch language; and (4) able and willing to give written informed consent. The study was conducted in accordance with the Declaration of Helsinki and the study protocol was approved by the Ethical Review Board of the University Medical Centre Utrecht centrally (protocol code 04/003, 13 May 2004), and by each local review board per participating institute. Patients who were taking at least one drug for psychosis as a treatment drug at baseline, and for whom gPT was available, were included in the current study. Patients who had an unknown phenotype or who reported chlorpromazine-equivalent dose of their main treatment drug of below 25 mg or above 1000 mg per day were excluded. Outcome For each outcome, data collected at baseline were used. Abnormal Involuntary Movement Scale (AIMS) Abnormal movements in the facial and oral area, trunk, and extremities are measured by the clinician or research team, to form an assessment of the extrapyramidal symptom of dyskinesia, together with a global assessment . Barnes Akathisia Rating Scale (BARS) Akathisia presents as motor restlessness, part of the extrapyramidal symptom group. The BARS is a commonly used scale to measure clinical features of akathisia based on clinician observation, patient report of restlessness, and patient report of distress due to restlessness, amounting to a total score to indicate an overall assessment of akathisia . The Unified Parkinson’s Disease Rating Scale (UPDRS) Parkinsonian movements can be characterized by slowed movements, rigidity and tremors. The UPDRS is a valid and widely used instrument in Parkinson’s disease research and provides another measure of extrapyramidal symptoms . The Subjective Well-being Under Neuroleptics Scale (SWN20) The SWN20 provides a measure of subjective well-being using five subscales: emotional regulation, mental functioning, physical functioning, social integration and self-control . Positive and Negative Syndrome Scale (PANSS) The PANSS combines the Brief Psychiatric Rating Scale and Psychopathology Rating Schedule to assess positive and negative symptoms of schizophrenia as well as general psychopathology using a 30-item questionnaire. Scale-specific items are summed to calculate a score on the positive and negative symptom subscales, which have been shown to have strong reliability, validity and sensitivity . Covariates Sex, age, ethnicity, number of psychotic episodes, illness duration in years, chlorpromazine-equivalent dose of the main treatment drug, total number of treatment (for psychosis) drugs prescribed, total number of medications prescribed, smoking status and hormonal birth control status were included as covariates. Body mass index (BMI) was not available or computable for this time point and was therefore not included as a covariate. Genotyping From each subject, 20 ml of blood was collected at the participating mental health institutes. The blood sample was sent to the University Medical Centre Utrecht by mail, where DNA extraction was performed from peripheral blood lymphocytes. Quality control procedures were performed using PLINK v1.9 according to standard protocol (Purcell and Chang; ). Genotyping and quality control details have been described elsewhere and in the Supplemental material . For an extensive overview of genotype–phenotype translation of CYP2D6 , we also refer to the supplemental material of . Briefly, genotypes of 570,038 single-nucleotide variants were characterized for 2818 individuals (including patients, siblings, parents, and healthy controls), using a customized Illumina Institute of Psychological Medicine and Clinical Neurology chip array and were subjected to an extensive quality control procedure. Non-genotyped variants were imputed using the 1000 Genome Phase 3 (v5) reference panel using the Michigan Imputation Server . A post-imputation filter of R2 > 0.5 was applied to include only high-quality rare variants. Genotype-predicted UM could not be assessed as structure variants (including copy number variants) were not included in the GWAS panels. CYP2D6 phenotype prediction and phenoconversion calculation Stargazer, a python-based bioinformatics tool, was used to predict metabolizer groups of CYP2D6 . Imputed data for chromosome 22 were used to identify CYP2D6 star alleles (haplotypes), and an AS was generated accordingly. A functional phenotype classification per individual was generated. To improve accuracy, phenotype classifications were cross-checked with the PharmVar genotype–phenotype translation tables and the Pharmacogenomics Knowledgebase ( ; PharmGKB), as explained in more detail by . Concurrent CYP2D6 -inhibitor use was checked per patient. First, all recorded drug information was checked and standardized (i.e. medication names were converted to identical spelling of the generic name, and entries with missing or unclear information on medication type, use, or dose were removed). Known CYP2D6 inhibitors were marked according to strength: weak, moderate and strong, according to . Inducers were not included, since there are no known CYP2D6 inducers (see Supplemental Table S1 ). Following this, a new variable representing pPT was created. All individuals were assigned a pPT. pPT was the same as gPT if the patient was not using any CYP2D6 inhibitors (e.g. the pPT would be NM if gPT was NM and the patient did not use any inhibitors) and was adjusted to a reduced metabolism type according to the methods of , accounting for inhibitor strength. Weak inhibitors do not cause phenoconversion, while strong inhibitors such as paroxetine and fluoxetine have previously been shown to reduce most individuals to PM . Although true conversion to PM can only be confirmed using serum measurements, we used the method of , that is, assigning a pPT based on inhibitor strength, to represent clinical practicality as most clinicians will not have serum measurements available. Thus, individuals taking a strong inhibitor were assigned a poor pPT (pPM) in all cases (i.e. NM and IM individuals were assigned pPM. PM individuals cannot be further reduced, and would remain pPM). If an individual was only taking a weak inhibitor, the pPT remained the same as their gPT. Moderate inhibitors would only affect normal gPT (gNM) individuals, reducing their status to intermediate (pIM), but not reduce intermediate metabolizers (gIM) to pPM. If individuals were taking multiple inhibitors, the highest strength was used to assign pPT. Figure S1 (Supplemental material) provides an overview. Statistical analysis Between-group differences between phenotype groups and between the most commonly reported drug subsamples were examined using Kruskal-Wallis and Dunn tests (for continuous variables) and Fisher’s exact and/or Pearson’s Chi-square tests (for categorical variables). Both differences between gPT groups and pPT groups were examined. Statistical analyses were performed using R version 4.2.3 . To investigate whether pPT is more strongly associated with each outcome compared to gPT, regression models were generated and compared to find the best model, for the most used drugs separately. Linear models were used for continuous outcomes (SWN20, AIMS, UPDRS, PANSS), and a proportional odds cumulative logit model for categorical (BARS), as the BARS has six outcomes (none, minimal, mild, moderate, severe and extreme). Multiple models were explored per outcome, using a backward stepwise variable selection method. This was done twice, once using gPT as a predictor, and once with pPT. The best-fit model per outcome was selected using the Akaike information criterion (AIC) and R 2 . The AIC is an information criteria-based relative fit index developed as an estimation of the accuracy of a model to predict outcomes in samples other than the provided data, and R 2 represents the percentage of total variation in the outcome that is explained by the predictors. According to the AIC rule of thumb, if the difference in AIC between two models is less than two, this suggests the models are indistinguishable. The ‘stats’ package was used for the Mann–Whitney U test, Chi-square test, Kruskal–Wallis rank sum test, Fisher’s exact test for count data, and general linear models . Dunn tests were performed using the dunn_test function from the ‘rstatix’ package . The ‘VGAM: Vector Generalized Linear and Additive Models’ package provided the vglm function for multinomial models. The step4vglm function from this package was used to select the final best multinomial model, using backward direction and specifying the genotype variables to remain included using the scope argument of this function . Similarly, stepAIC from the ‘MASS’ package was used to select the best-fitting GLM models . AIC was automatically compared across models per outcome using the aictab function from the ‘AICcmodavg’ package . A p- value of <0.05 was considered statistically significant. All reported p -values are corrected for multiple testing using the Bonferroni method. GROUP release number 8.0 was used for the current analysis. For each outcome, data collected at baseline were used. Abnormal Involuntary Movement Scale (AIMS) Abnormal movements in the facial and oral area, trunk, and extremities are measured by the clinician or research team, to form an assessment of the extrapyramidal symptom of dyskinesia, together with a global assessment . Barnes Akathisia Rating Scale (BARS) Akathisia presents as motor restlessness, part of the extrapyramidal symptom group. The BARS is a commonly used scale to measure clinical features of akathisia based on clinician observation, patient report of restlessness, and patient report of distress due to restlessness, amounting to a total score to indicate an overall assessment of akathisia . The Unified Parkinson’s Disease Rating Scale (UPDRS) Parkinsonian movements can be characterized by slowed movements, rigidity and tremors. The UPDRS is a valid and widely used instrument in Parkinson’s disease research and provides another measure of extrapyramidal symptoms . The Subjective Well-being Under Neuroleptics Scale (SWN20) The SWN20 provides a measure of subjective well-being using five subscales: emotional regulation, mental functioning, physical functioning, social integration and self-control . Positive and Negative Syndrome Scale (PANSS) The PANSS combines the Brief Psychiatric Rating Scale and Psychopathology Rating Schedule to assess positive and negative symptoms of schizophrenia as well as general psychopathology using a 30-item questionnaire. Scale-specific items are summed to calculate a score on the positive and negative symptom subscales, which have been shown to have strong reliability, validity and sensitivity . Abnormal movements in the facial and oral area, trunk, and extremities are measured by the clinician or research team, to form an assessment of the extrapyramidal symptom of dyskinesia, together with a global assessment . Akathisia presents as motor restlessness, part of the extrapyramidal symptom group. The BARS is a commonly used scale to measure clinical features of akathisia based on clinician observation, patient report of restlessness, and patient report of distress due to restlessness, amounting to a total score to indicate an overall assessment of akathisia . Parkinsonian movements can be characterized by slowed movements, rigidity and tremors. The UPDRS is a valid and widely used instrument in Parkinson’s disease research and provides another measure of extrapyramidal symptoms . The SWN20 provides a measure of subjective well-being using five subscales: emotional regulation, mental functioning, physical functioning, social integration and self-control . The PANSS combines the Brief Psychiatric Rating Scale and Psychopathology Rating Schedule to assess positive and negative symptoms of schizophrenia as well as general psychopathology using a 30-item questionnaire. Scale-specific items are summed to calculate a score on the positive and negative symptom subscales, which have been shown to have strong reliability, validity and sensitivity . Sex, age, ethnicity, number of psychotic episodes, illness duration in years, chlorpromazine-equivalent dose of the main treatment drug, total number of treatment (for psychosis) drugs prescribed, total number of medications prescribed, smoking status and hormonal birth control status were included as covariates. Body mass index (BMI) was not available or computable for this time point and was therefore not included as a covariate. From each subject, 20 ml of blood was collected at the participating mental health institutes. The blood sample was sent to the University Medical Centre Utrecht by mail, where DNA extraction was performed from peripheral blood lymphocytes. Quality control procedures were performed using PLINK v1.9 according to standard protocol (Purcell and Chang; ). Genotyping and quality control details have been described elsewhere and in the Supplemental material . For an extensive overview of genotype–phenotype translation of CYP2D6 , we also refer to the supplemental material of . Briefly, genotypes of 570,038 single-nucleotide variants were characterized for 2818 individuals (including patients, siblings, parents, and healthy controls), using a customized Illumina Institute of Psychological Medicine and Clinical Neurology chip array and were subjected to an extensive quality control procedure. Non-genotyped variants were imputed using the 1000 Genome Phase 3 (v5) reference panel using the Michigan Imputation Server . A post-imputation filter of R2 > 0.5 was applied to include only high-quality rare variants. Genotype-predicted UM could not be assessed as structure variants (including copy number variants) were not included in the GWAS panels. phenotype prediction and phenoconversion calculation Stargazer, a python-based bioinformatics tool, was used to predict metabolizer groups of CYP2D6 . Imputed data for chromosome 22 were used to identify CYP2D6 star alleles (haplotypes), and an AS was generated accordingly. A functional phenotype classification per individual was generated. To improve accuracy, phenotype classifications were cross-checked with the PharmVar genotype–phenotype translation tables and the Pharmacogenomics Knowledgebase ( ; PharmGKB), as explained in more detail by . Concurrent CYP2D6 -inhibitor use was checked per patient. First, all recorded drug information was checked and standardized (i.e. medication names were converted to identical spelling of the generic name, and entries with missing or unclear information on medication type, use, or dose were removed). Known CYP2D6 inhibitors were marked according to strength: weak, moderate and strong, according to . Inducers were not included, since there are no known CYP2D6 inducers (see Supplemental Table S1 ). Following this, a new variable representing pPT was created. All individuals were assigned a pPT. pPT was the same as gPT if the patient was not using any CYP2D6 inhibitors (e.g. the pPT would be NM if gPT was NM and the patient did not use any inhibitors) and was adjusted to a reduced metabolism type according to the methods of , accounting for inhibitor strength. Weak inhibitors do not cause phenoconversion, while strong inhibitors such as paroxetine and fluoxetine have previously been shown to reduce most individuals to PM . Although true conversion to PM can only be confirmed using serum measurements, we used the method of , that is, assigning a pPT based on inhibitor strength, to represent clinical practicality as most clinicians will not have serum measurements available. Thus, individuals taking a strong inhibitor were assigned a poor pPT (pPM) in all cases (i.e. NM and IM individuals were assigned pPM. PM individuals cannot be further reduced, and would remain pPM). If an individual was only taking a weak inhibitor, the pPT remained the same as their gPT. Moderate inhibitors would only affect normal gPT (gNM) individuals, reducing their status to intermediate (pIM), but not reduce intermediate metabolizers (gIM) to pPM. If individuals were taking multiple inhibitors, the highest strength was used to assign pPT. Figure S1 (Supplemental material) provides an overview. Between-group differences between phenotype groups and between the most commonly reported drug subsamples were examined using Kruskal-Wallis and Dunn tests (for continuous variables) and Fisher’s exact and/or Pearson’s Chi-square tests (for categorical variables). Both differences between gPT groups and pPT groups were examined. Statistical analyses were performed using R version 4.2.3 . To investigate whether pPT is more strongly associated with each outcome compared to gPT, regression models were generated and compared to find the best model, for the most used drugs separately. Linear models were used for continuous outcomes (SWN20, AIMS, UPDRS, PANSS), and a proportional odds cumulative logit model for categorical (BARS), as the BARS has six outcomes (none, minimal, mild, moderate, severe and extreme). Multiple models were explored per outcome, using a backward stepwise variable selection method. This was done twice, once using gPT as a predictor, and once with pPT. The best-fit model per outcome was selected using the Akaike information criterion (AIC) and R 2 . The AIC is an information criteria-based relative fit index developed as an estimation of the accuracy of a model to predict outcomes in samples other than the provided data, and R 2 represents the percentage of total variation in the outcome that is explained by the predictors. According to the AIC rule of thumb, if the difference in AIC between two models is less than two, this suggests the models are indistinguishable. The ‘stats’ package was used for the Mann–Whitney U test, Chi-square test, Kruskal–Wallis rank sum test, Fisher’s exact test for count data, and general linear models . Dunn tests were performed using the dunn_test function from the ‘rstatix’ package . The ‘VGAM: Vector Generalized Linear and Additive Models’ package provided the vglm function for multinomial models. The step4vglm function from this package was used to select the final best multinomial model, using backward direction and specifying the genotype variables to remain included using the scope argument of this function . Similarly, stepAIC from the ‘MASS’ package was used to select the best-fitting GLM models . AIC was automatically compared across models per outcome using the aictab function from the ‘AICcmodavg’ package . A p- value of <0.05 was considered statistically significant. All reported p -values are corrected for multiple testing using the Bonferroni method. GROUP release number 8.0 was used for the current analysis. Patient characteristics In all, 412 patients were included (79.6% male, mean age 27 ± 7.1 years). In total, 228 patients (55.3%) were genotype-predicted NMs (gNM), 167 (40.5%) were IMs (gIM) and 17 (4.1%) were PMs (gPM). Six patients (not included in the described sample) were classified as ‘unknown’ metabolizers and were not included in the analysis. A Chi-square test confirmed this distribution was not different than would be expected in a European population based on existing literature ( p = 0.21). See Table S1 for an overview of patient characteristics of the complete sample. Risperidone ( n = 111), olanzapine ( n = 122), clozapine ( n = 60) and aripiprazole ( n = 49) were the four most reported treatment drugs and were thus selected to investigate the association between genotype prediction and outcome. – provide demographic information on this sample and per drug group. Risperidone and aripiprazole are both metabolized through CYP2D6 to active metabolites 9-hydroxyrisperidone (risperidone) and dehydroaripiprazole (aripiprazole) and CYP2D6 inhibition has been shown to increase relative plasma concentration . By contrast, neither olanzapine nor clozapine is extensively metabolized through CYP2D6 . Due to the differential effects of CYP2D6 inhibitors on each drug, each group was analysed separately. For example, CYP2D6 activity is strongly correlated with risperidone/9-hydroxyrisperidone ratio, but this relationship is less clear for aripiprazole/dehydroaripiprazole and the expected effect on efficacy is uncertain . Frequency of CYP2D6 inhibitors and phenoconversion (complete sample) In all, 102 patients were simultaneously using one or more CYP2D6 inhibitor(s), resulting in 65 instances of phenoconversion (73.8% male, mean age 28.4 ± 7.8). There were 201 (48.8%) phenoconverted NMs (pNM), 129 (31.3%) phenoconverted IMs (pIM) and 82 (19.9%) phenoconverted PMs (pPM). Concomitantly used CYP2D6 inhibitors were paroxetine (60 instances, 47.2%), citalopram (28 instances, 22%), fluoxetine (14 instances, 11%), sertraline (13 instances, 10.2%), escitalopram (7 instances, 5.5%), clomipramine (4 instances, 3.1%) and levomepromazine (one instance, 0.8%). Overall, 24 patients used two or more inhibitors, and the remaining 78 used one. Phenoconversion occurred mainly due to strong inhibitors paroxetine and fluoxetine. In all, 38 gIM and 27 gNM were converted to pPM due to the use of strong CYP2D6 inhibitors. This is a 382% increase in PM ( n = 17 prior to phenoconversion, n = 82 following phenoconversion) and 15.8% of the total sample. When phenoconversion for CYP2D6 was compared between men and women, 6% of women compared to 3.7% of men had gPM. The gIM phenotype was more prevalent in women compared to men (48.8% vs 38.4% ) , but not significant ( p = 0.16). In total, 29.8% of women were concomitantly using a CYP2D6 inhibitor, versus 23.5% of men ( p = 0.36). Although the distribution of inhibitor strength was the same for both sexes (~64% to 69% strong and ~31% to 36% weak), phenoconversion occurred in 20.2% of women compared to 14.6% of men ( p = 0.28). When considering phenoconversion, 26.2% of women was pPM, 36.9% pIM and 36.9% pNM. In comparison, 18.3% of men were pPM, 38.4% pIM and 51.8% pNM. Women were at a 1.4 times higher risk for phenoconversion compared to men. Although the mean total medications was slightly lower in women versus men (3.2 and 3.7, respectively), the most commonly measured number of total medications for women was four, compared to two for men. Women were also at 1.3 times higher risk of taking concomitant medication in general, such as serotonin inhibitors, benzodiazepines and other drugs for psychosis. It should be noted, however, that there were no significant differences between sexes in medication use or phenoconversion rate ( Table S4 ). Subgroups and between-group comparison In the smaller group containing only the four most common drugs for psychosis, 80 patients reported using one or more CYP2D6 inhibitor(s), leading to 51 cases of phenoconversion (78.4% male, mean age 27.8 ± 7.4). The percentage of patients per drug group who were phenoconverted to pPM was not significantly different between groups ( p = 0.277) and ranged from 10.8% (12 individuals; risperidone group) to 22.4% (11 individuals; aripiprazole group). Phenoconversion to pPM occurred at a similar rate in the olanzapine and clozapine groups (14.7%/18 individuals in the olanzapine group; 16.7%/10 individuals in the clozapine group). Age, chlorpromazine-equivalent dose, illness duration, number of treatment drugs for psychosis and medications total, and use of birth control differed between drug groups (see Table S2 ). Users of risperidone were younger and were prescribed lower chlorpromazine-equivalent doses. Similarly, the aripiprazole dose was lower compared to clozapine. Clozapine and aripiprazole users had a longer illness duration, although the number of psychotic episodes did not differ between groups. Risperidone users reported a higher concomitant use of other drugs for psychosis, but not concomitant medication in general. Users of either clozapine or aripiprazole reported taking more medications compared to risperidone and olanzapine users. Last, birth control use differed between aripiprazole and clozapine users. However, there were no birth control users in the clozapine group, and even though the reported use of birth control was proportionally highest in the aripiprazole group (16%), there were still only eight individuals. When both sexes and all drugs were aggregated into one sample of n = 342, the PANSS positive scale was significantly different between pPT groups ( p = 0.0285), where the pIM group had lower PANSS positive scores compared to the pPM group. When stratified by drug, the same result was found in the risperidone group only ( p = 0.0286). Separated by sex, SWN20 scores were significantly lower for male pPM individuals compared to pNM over all drugs ( p = 0.0461), but no drug-specific differences were found. For women, UPDRS scores were slightly higher for NM women using olanzapine in both phenotype groups (p/gNM compared to pPM; p = 0.0476 and compared to gIM; p = 0.0142. It should be noted that there were no gPM women using olanzapine). In the same group, gIM women had higher BARS scores compared to gNM ( p = 0.0307). Association between genotype prediction and outcome Predictors were first checked for multicollinearity, but no variables reached variance inflation factor >2. For each model, only complete cases per drug group were used (see Table S5 ). The model analysis was not repeated segregated by sex due to the small sample size per drug for women. The final models for each outcome in each drug group are presented in and , separated by the role of CYP2D6 in metabolism. AIC and R 2 scores are shown in Table S6 . With the exception of the risperidone group, pPT was included in most outcome models instead of gPT. The AIMS outcome model included pPT in all groups, and the UDPRS included gPT in all groups. It should be noted that the gPT and pPT models for most outcomes demonstrated similar AIC (difference of <2 points). This was not the case in the risperidone group for PANSS positive and UDPRS scores (AIC difference of 4.79 and 5.19, respectively); in the olanzapine group for PANSS positive and BARS scores (difference of 2.24 and 7.48, respectively); in the clozapine group for AIMS and SWN20 scores (difference of 3.36 and 2.87, respectively); and in the aripiprazole group for PANSS negative and UPDRS scores (difference of 2.1 and 2.17, respectively). In the risperidone, olanzapine and clozapine groups, phenotype was significantly associated with outcome. For risperidone, gPM individuals had significantly higher PANSS negative ( p = 0.014) and UPDRS ( p = 0.015) scores, but lower PANSS positive scores compared to gNM individuals ( p = 0.015). pNM clozapine users had lower AIMS scores compared to pIM ( p = 0.03), whereas gNM olanzapine users had better BARS outcomes compared to gIM ( p = 0.003). The selection of predictors in the final model for each outcome differed between drug groups. Higher chlorpromazine-equivalent dose, number of medications, and age, as well as being a smoker and of non-white ethnicity, were consistently associated with worse outcomes (see and ). For illness duration, number of psychotic episodes, use of birth control and sex, and direction of effect differed between drug groups. Longer illness duration in years was associated with worse AIMS outcome in the risperidone group ( p = 0.003), but improved outcome in the aripiprazole group (lower AIMS; p = 0.03, and higher SWN20; p = 0.048). A number of psychotic episodes, however, was associated with increased AIMS (aripiprazole; p = 0.03) and UPDRS scores (clozapine; p = 0.022), but lower PANSS negative scores (olanzapine; p = 0.047). The use of contraceptives was only associated with higher AIMS scores (aripiprazole; p = 0.031) but lower UPDRS scores (olanzapine; p = 0.041). Sex was associated with both PANSS scales yet in different groups (negative, olanzapine; p = 0.015, positive, clozapine; p = 0.005). It was also significant for the final models for UPDRS for risperidone ( p = 0.036), olanzapine ( p = 0.004), and aripiprazole ( p = 0.03), though the model predicted increased UDPRS scores for women in the olanzapine group only, compared to lower scores in the others. In all, 412 patients were included (79.6% male, mean age 27 ± 7.1 years). In total, 228 patients (55.3%) were genotype-predicted NMs (gNM), 167 (40.5%) were IMs (gIM) and 17 (4.1%) were PMs (gPM). Six patients (not included in the described sample) were classified as ‘unknown’ metabolizers and were not included in the analysis. A Chi-square test confirmed this distribution was not different than would be expected in a European population based on existing literature ( p = 0.21). See Table S1 for an overview of patient characteristics of the complete sample. Risperidone ( n = 111), olanzapine ( n = 122), clozapine ( n = 60) and aripiprazole ( n = 49) were the four most reported treatment drugs and were thus selected to investigate the association between genotype prediction and outcome. – provide demographic information on this sample and per drug group. Risperidone and aripiprazole are both metabolized through CYP2D6 to active metabolites 9-hydroxyrisperidone (risperidone) and dehydroaripiprazole (aripiprazole) and CYP2D6 inhibition has been shown to increase relative plasma concentration . By contrast, neither olanzapine nor clozapine is extensively metabolized through CYP2D6 . Due to the differential effects of CYP2D6 inhibitors on each drug, each group was analysed separately. For example, CYP2D6 activity is strongly correlated with risperidone/9-hydroxyrisperidone ratio, but this relationship is less clear for aripiprazole/dehydroaripiprazole and the expected effect on efficacy is uncertain . CYP2D6 inhibitors and phenoconversion (complete sample) In all, 102 patients were simultaneously using one or more CYP2D6 inhibitor(s), resulting in 65 instances of phenoconversion (73.8% male, mean age 28.4 ± 7.8). There were 201 (48.8%) phenoconverted NMs (pNM), 129 (31.3%) phenoconverted IMs (pIM) and 82 (19.9%) phenoconverted PMs (pPM). Concomitantly used CYP2D6 inhibitors were paroxetine (60 instances, 47.2%), citalopram (28 instances, 22%), fluoxetine (14 instances, 11%), sertraline (13 instances, 10.2%), escitalopram (7 instances, 5.5%), clomipramine (4 instances, 3.1%) and levomepromazine (one instance, 0.8%). Overall, 24 patients used two or more inhibitors, and the remaining 78 used one. Phenoconversion occurred mainly due to strong inhibitors paroxetine and fluoxetine. In all, 38 gIM and 27 gNM were converted to pPM due to the use of strong CYP2D6 inhibitors. This is a 382% increase in PM ( n = 17 prior to phenoconversion, n = 82 following phenoconversion) and 15.8% of the total sample. When phenoconversion for CYP2D6 was compared between men and women, 6% of women compared to 3.7% of men had gPM. The gIM phenotype was more prevalent in women compared to men (48.8% vs 38.4% ) , but not significant ( p = 0.16). In total, 29.8% of women were concomitantly using a CYP2D6 inhibitor, versus 23.5% of men ( p = 0.36). Although the distribution of inhibitor strength was the same for both sexes (~64% to 69% strong and ~31% to 36% weak), phenoconversion occurred in 20.2% of women compared to 14.6% of men ( p = 0.28). When considering phenoconversion, 26.2% of women was pPM, 36.9% pIM and 36.9% pNM. In comparison, 18.3% of men were pPM, 38.4% pIM and 51.8% pNM. Women were at a 1.4 times higher risk for phenoconversion compared to men. Although the mean total medications was slightly lower in women versus men (3.2 and 3.7, respectively), the most commonly measured number of total medications for women was four, compared to two for men. Women were also at 1.3 times higher risk of taking concomitant medication in general, such as serotonin inhibitors, benzodiazepines and other drugs for psychosis. It should be noted, however, that there were no significant differences between sexes in medication use or phenoconversion rate ( Table S4 ). In the smaller group containing only the four most common drugs for psychosis, 80 patients reported using one or more CYP2D6 inhibitor(s), leading to 51 cases of phenoconversion (78.4% male, mean age 27.8 ± 7.4). The percentage of patients per drug group who were phenoconverted to pPM was not significantly different between groups ( p = 0.277) and ranged from 10.8% (12 individuals; risperidone group) to 22.4% (11 individuals; aripiprazole group). Phenoconversion to pPM occurred at a similar rate in the olanzapine and clozapine groups (14.7%/18 individuals in the olanzapine group; 16.7%/10 individuals in the clozapine group). Age, chlorpromazine-equivalent dose, illness duration, number of treatment drugs for psychosis and medications total, and use of birth control differed between drug groups (see Table S2 ). Users of risperidone were younger and were prescribed lower chlorpromazine-equivalent doses. Similarly, the aripiprazole dose was lower compared to clozapine. Clozapine and aripiprazole users had a longer illness duration, although the number of psychotic episodes did not differ between groups. Risperidone users reported a higher concomitant use of other drugs for psychosis, but not concomitant medication in general. Users of either clozapine or aripiprazole reported taking more medications compared to risperidone and olanzapine users. Last, birth control use differed between aripiprazole and clozapine users. However, there were no birth control users in the clozapine group, and even though the reported use of birth control was proportionally highest in the aripiprazole group (16%), there were still only eight individuals. When both sexes and all drugs were aggregated into one sample of n = 342, the PANSS positive scale was significantly different between pPT groups ( p = 0.0285), where the pIM group had lower PANSS positive scores compared to the pPM group. When stratified by drug, the same result was found in the risperidone group only ( p = 0.0286). Separated by sex, SWN20 scores were significantly lower for male pPM individuals compared to pNM over all drugs ( p = 0.0461), but no drug-specific differences were found. For women, UPDRS scores were slightly higher for NM women using olanzapine in both phenotype groups (p/gNM compared to pPM; p = 0.0476 and compared to gIM; p = 0.0142. It should be noted that there were no gPM women using olanzapine). In the same group, gIM women had higher BARS scores compared to gNM ( p = 0.0307). Predictors were first checked for multicollinearity, but no variables reached variance inflation factor >2. For each model, only complete cases per drug group were used (see Table S5 ). The model analysis was not repeated segregated by sex due to the small sample size per drug for women. The final models for each outcome in each drug group are presented in and , separated by the role of CYP2D6 in metabolism. AIC and R 2 scores are shown in Table S6 . With the exception of the risperidone group, pPT was included in most outcome models instead of gPT. The AIMS outcome model included pPT in all groups, and the UDPRS included gPT in all groups. It should be noted that the gPT and pPT models for most outcomes demonstrated similar AIC (difference of <2 points). This was not the case in the risperidone group for PANSS positive and UDPRS scores (AIC difference of 4.79 and 5.19, respectively); in the olanzapine group for PANSS positive and BARS scores (difference of 2.24 and 7.48, respectively); in the clozapine group for AIMS and SWN20 scores (difference of 3.36 and 2.87, respectively); and in the aripiprazole group for PANSS negative and UPDRS scores (difference of 2.1 and 2.17, respectively). In the risperidone, olanzapine and clozapine groups, phenotype was significantly associated with outcome. For risperidone, gPM individuals had significantly higher PANSS negative ( p = 0.014) and UPDRS ( p = 0.015) scores, but lower PANSS positive scores compared to gNM individuals ( p = 0.015). pNM clozapine users had lower AIMS scores compared to pIM ( p = 0.03), whereas gNM olanzapine users had better BARS outcomes compared to gIM ( p = 0.003). The selection of predictors in the final model for each outcome differed between drug groups. Higher chlorpromazine-equivalent dose, number of medications, and age, as well as being a smoker and of non-white ethnicity, were consistently associated with worse outcomes (see and ). For illness duration, number of psychotic episodes, use of birth control and sex, and direction of effect differed between drug groups. Longer illness duration in years was associated with worse AIMS outcome in the risperidone group ( p = 0.003), but improved outcome in the aripiprazole group (lower AIMS; p = 0.03, and higher SWN20; p = 0.048). A number of psychotic episodes, however, was associated with increased AIMS (aripiprazole; p = 0.03) and UPDRS scores (clozapine; p = 0.022), but lower PANSS negative scores (olanzapine; p = 0.047). The use of contraceptives was only associated with higher AIMS scores (aripiprazole; p = 0.031) but lower UPDRS scores (olanzapine; p = 0.041). Sex was associated with both PANSS scales yet in different groups (negative, olanzapine; p = 0.015, positive, clozapine; p = 0.005). It was also significant for the final models for UPDRS for risperidone ( p = 0.036), olanzapine ( p = 0.004), and aripiprazole ( p = 0.03), though the model predicted increased UDPRS scores for women in the olanzapine group only, compared to lower scores in the others. The aim of this paper was to investigate the frequency of phenoconversion and examine whether gPT or pPT was more often associated with side effects and PANSS scores. We found no significant differences between gPT groups or pPT groups in any of the outcome measures in the aggregated group, with exception of the pPT and the PANSS positive scale. This was likely driven by the risperidone group, as there were no between-phenotype group differences on any outcome for any other drug group. Across drugs, pPT was included in the final model for most, but not all, measures. Furthermore, only the AIMS and UPDRS were associated with pPT and gPT, respectively, in all subgroups. Phenotype was not significantly associated with the outcome for all drugs for any measure. The similarity in AIC for gPT and pPT models for some outcome measures also indicates that pPT is not superior to gPT to improve model fit in these cases. CYP2D6 substrate (‘victim’) drugs risperidone and aripiprazole were expected to be most affected by CYP2D6 phenotype. Risperidone was indeed the only drug for which phenotype was associated with multiple measures (PANSS negative: gPM estimate = 9.097, p = 0.014; PANSS positive: gNM estimate = 2.49, p = 0.036; UDPRS: gPM estimate = 0.27, p = 0.015), and the only subgroup for which between-phenotype group differences were found (PANSS positive scale, p = 0.0286). However, aripiprazole was unaffected by CYP2D6 genotype and phenoconversion. Although aripiprazole is included in the DPWG guidelines suggesting reduced dose for PMs, as phenotype appears to be correlated to plasma concentration of the sum of aripiprazole and dehydroaripiprazole, there is insufficient or no evidence for an effect on clinical effect of adverse reactions . Similarly, study findings on CYP2D6 phenotype and risperidone adverse reactions are mixed . Contrary to expectations, CYP2D6 phenotype was significantly associated with one model for each olanzapine and clozapine. gNM individuals in the olanzapine group, and pNM individuals in the clozapine group, had better outcomes on the BARS and AIMS (respectively) compared to g/pIM and g/pPMs. The effect coefficients were small, however, and other factors such as sex and smoking status, contributed to a greater extent. This is within the realm of expectations, as they can affect CYP1A2 activity, which plays a main role in the metabolism of both drugs . Our finding may further confirm a limited effect of CYP2D6 in adverse reactions. Alternatively, CYP2D6 metabolizer status affects specific outcomes not appraised in this study. Previously, demonstrated no difference between CYP2D6 phenoconverted phenotype on side effects in general but did report an increase in specific side effects in PM and IM individuals. In both cases, this would also explain the lack of association between phenoconversion and outcome. If CYP2D6 activity has no relationship with treatment outcome of drugs for psychosis in general, it would be expected that a change in activity through phenoconversion has no effect either. Other aspects of the study itself may also explain our findings. Here, aripiprazole presented the smallest subgroup of n = 49, with 3 gPMs and 14 pPMs. Although this was the highest percentage of gPM and pPM in any group (6.2% and 28.6%, respectively), gPT and pPT proportions did not significantly differ between drug groups ( Table S3 ). In general, the relatively small subsample of 51 patients for which phenoconversion occurred may have contributed to the lack of phenoconversion effect. Although our findings highlight the high frequency of phenoconversion and subsequential increase in pPMs, our sample may lack statistical power. The large increase in PM individuals following phenoconversion may have increased within-group variability, potentially distorting the results. With greater variability, statistical power decreases, and the sample size may not have been large enough to compensate for this . Furthermore, there may be patient-related factors associated with the use of concomitant medication that affected the outcome measures on their own. pPM individuals were either gPM or treated with SERTs paroxetine or fluoxetine. Reason for use of this medication, or their associated effect, may have interacted with treatment outcome . Alternatively, patients were taking medication for a longer time and were already prescribed a dose minimizing side effects. The average duration of illness was 4.3 (±3.5) years with 1.7 (±1.03) psychotic episodes, possibly indicating that patients started treatment before the first study measure was collected. Users of clozapine had a significantly longer illness duration of (6.2 ± 2.8) years, which may reflect its status as the first choice for treatment-resistant schizophrenia. In the Netherlands, clozapine is prescribed when treatment with two previous drugs for psychosis fails . Although a subset of patients does not respond well to pharmacotherapy, the majority of patients with a first episode of psychosis respond sufficiently to treatment . In addition, earlier studies found that olanzapine may be associated with later discontinuation and thus longer successful treatment . In this sample, 35.7% of patients were using olanzapine as their main treatment drug, and 17.5% used clozapine. Patient stability may thus have masked between-phenotype group outcomes. Concerning the impact of phenoconversion, we found that approximately 25% of patients in the complete sample reported using a CYP2D6 inhibitor, of which 68% used a strong inhibitor (paroxetine or fluoxetine). In all, 65 instances of phenoconversion from IM or NM to PM occurred, 51 of which were in the analysed sample, resulting in an increase in PM by almost five times (82 vs 17, 20% of the total sample). This high prevalence seems to be in line with previous literature and confirms that the PM phenotype could be more often caused by phenoconversion than by genetic predisposition . In addition, it highlights the potential effects of polypharmacy. On average, participants reported use of about three medications (3.2 ± 2.04). Clozapine users reported the highest number of concomitant medications (4.3 ± 2.2). Although the number of medications was significant only for BARS (clozapine) and PANSS positive (risperidone) scores, when included in a model it was always associated with worse outcomes. Polypharmacy is common in psychiatry and may be associated with worse treatment outcomes, especially increased risk of side effects . However, it can be useful in specific situations, depending on the presentation of symptoms or the severity of side effects. Concomitant prescription of drugs for depression such as SERTs is common to treat negative symptoms, although the reported effect is often small . ‘Antipsychotic polypharmacy’, when two or more drugs for psychosis are prescribed, may even be more effective than monotherapy without increased risk of side effects . In our results, the number of treatment drugs for psychosis was included in some final models, suggesting improved outcomes with antipsychotic polypharmacy. For no model was this significant, however. In our sample, polypharmacy was slightly more prevalent in women. Although no significant differences were found between phenoconversion frequency, number of medications or inhibitor use (see Table S4 ), women appeared to be slightly more at risk of phenoconversion. 18.3% of women and 14.2% of men in the aggregated sample experienced this phenomenon. Women were prescribed more medications on average (3.6 vs 3.2) but were not more likely to report taking more than one medication. It has been suggested that there are sex-related differences in side effect outcomes, and women are at increased risk of overdosing . Our results may suggest that women using olanzapine with lower CYP2D6 activity (pPM/gIM compared to pNM/gNM) experienced fewer parkinsonian side effects but were at increased risk for akathisia. Such a between-phenotype result was not found for men. Sex itself was significant in the final model for UPDRS in the risperidone, olanzapine and aripiprazole groups. Women using aripiprazole or risperidone were predicted to have lower UPDRS scores, and the opposite if they were taking olanzapine. This may be due to the greater number of women non-smokers in this group, as smoking was (non-significantly) associated with reduced UDPRS scores. Differential effect of sex between drugs may also be related to sex-related differences in pharmacokinetics and pharmacodynamics, for example, levels of gastric acidity, body weight and distribution of adipose tissue, blood volume, degree of intestinal motility and renal excretion rates . In addition, although no sex differences have been described for CYP2D6 , evidence exists for increased activity of CYP3A4 in women of reproductive age , and CYP1A2 has been shown to vary with menstrual cycle . The use of contraceptives was associated with lower UPDRS scores in the olanzapine group, which may be related to reduced CYP1A2 fluctuations as olanzapine is primarily metabolized through this enzyme . Evidence on the effect of hormonal contraceptives, especially non-oral, on psychotropic drugs is limited , and it should be noted that the reported use of contraceptives in our sample was low. This is not uncommon in women with schizophrenia . The association found here may suggest a potential interaction with the treatment effect. Our findings may have several implications for clinical practice. The GROUP cohort is a well-characterized real-world sample of patients with a psychotic disorder, unconstrained by the inclusion criteria often specified for randomized clinical trials. This provides a more generalizable sample for clinical practice. It should be noted, however, that these patients do not represent severely ill and uncooperative patients with schizophrenia spectrum disorders. The study also demonstrates another important part of clinical reality that is often ignored in pharmacogenetic studies, the potential impact of polypharmacy and phenoconversion. This information is commonly ignored or included as a confounding variable only. Yet in our sample and practice, a large proportion of psychiatric patients are prescribed multiple medications . We have also demonstrated that polypharmacy leads to phenoconversion in a significant number of patients, which may affect treatment outcomes under certain conditions. The results also indicate, however, a limited role of CYP2D6 activity in treatment outcome. Although CYP2D6 genotyping is not yet standardized in clinical practice, we suggest that consideration of drug–drug interactions should be integral for patient treatment when pharmacogenetic information is available. Genotyping patients for relevant pharmacogenes should be supplemented by an indexation of concomitant medications possibly interfering with enzyme activity. Polypharmacy itself should be carefully considered, especially when new medications are added to the treatment regime. When a prescribing physician considers supplementing the treatment of drugs for psychosis with, for example, drugs for depression, they should review whether the new medication interferes with the treatment medication. Without such consideration, the patient may unnecessarily suffer from changes in the side effects or efficacy of the primary medication. If genotype information is available, phenoconversion should be integrated into standard treatment protocols through the use of regular updating of the predicted phenotype after changes in medication, depending on the inhibition or induction strength of other medications. A practical recommendation for this is the use of the University of Florida’s (UF Health) CYP2D6 Phenoconversion (PROP Pharmacogenetics) Calculator developed by . This calculator is available online and is accessible also without knowledge of pharmacogenetics ( https://precisionmedicine.ufhealth.org/how-to-interpret-results/phenoconversion-calculator/ ). This individual phenotype, rather than the genotype, should then be guiding medication choice and dose to optimize outcome. Our study has some limitations that should be considered. First, the sample size of PMs was small, and it was not possible to analyse UMs due to the constraints of the GWAS panel. Since this metabolizer type is relatively rare, only a small number of UMs are expected to be missed . This should not have affected phenoconversion calculation, as UM individuals are also expected to convert to PM during the use of potent CYP2D6 inhibitors, but does affect genotype-predicted estimations. Second, the number of women in the study was low. This is especially important to take into consideration when interpreting sex-related findings. The overrepresentation of men in schizophrenia and psychosis studies is not uncommon . The low number of women included in the study may be associated with the study design or recruitment methods; patients were recruited through clinicians. Previous research suggests that sex differences exist in the presentation of symptoms, as well as subsequential age of diagnosis and treatment . Possibly, fewer women were considered for potential inclusion due to bias in the recruitment methods. Alternatively, there may be other reasons for lowering the motivation of women to participate. As a result, the findings are not adequately powered to draw sex-specific conclusions about the impact of CYP2D6 and phenoconversion, especially for women. Third, information on lifestyle factors, such as BMI, availability of a social support system, or pre-medication functioning, possibly contributing to treatment outcome, were unavailable at baseline. A lack of social support has also been associated with nonadherence . Nonadherence itself is a known challenge in psychotic disorders and has been associated with variability in treatment outcomes . Fourth, medication use was reported by the patient. Although all medication entries were manually checked and standardized prior to the current study, errors through patient reports or researcher documentation may have occurred, leading to incomplete records, inaccurate reports of dosage or omission of medication. This was considerably higher for documentation of the non-treatment medication. Concomitant inhibitor dose and duration of treatment could not be assessed, though these factors may affect phenoconversion. Although fluoxetine-induced CYP2D6 inhibition has been shown to persist after cessation of fluoxetine administration, the effect is shorter for paroxetine . The strength of inhibition may also increase with dose, although a low dose of 5 mg of paroxetine has been shown to inhibit CYP2D6 activity already . If a patient failed to report using either of these strong inhibitors at a very low dose or was not using the inhibitor at the time despite reporting to do so, they may have been wrongly considered pPM. A related limitation is the lack of therapeutic drug monitoring (TDM) in this sample. TDM is often used in clinical practice to optimize dose in accordance with the therapeutic windows of a drug and can be used to assess medication adherence and CYP2D6 phenotype, as blood plasma concentration of the drug is expected to be higher for PM individuals . However, this is not common practice for psychosis treatment in the Netherlands, where psychosis treatment guidelines specify no need for routine TDM due to a lack of evidence for reliable therapeutic ranges and it is employed only for clozapine and haloperidol when there are signs of treatment resistance or lack of response . In the GROUP study from which the data were derived, medication compliance was assessed by a clinician using a 7-point Likert scale . Thus, there is a possibility that patients were erroneously categorized as compliant, leading to an incorrect classification as pPM (if nonadherence affected the inhibitory medication) or to the incorrect association of the treatment medication with outcome measures (if nonadherence affected treatment). Similarly, phenotype could only be predicted using genotype and was not confirmed using plasma drug concentration through TDM or using probe-specific xenobiotics and urine or blood sampling, as neither method was part of the study protocol nor standardized treatment in The Netherlands . However, genotype–phenotype concordance was previously found to be high for all but UM phenotype, suggesting this problem may have been minimized in the current sample . Regardless, without information on plasma concentration and complete medication regime of treatment and inhibitory drugs, we cannot assure that phenotype prediction, whether based on genotype or phenoconversion, was correct. Last, only CYP2D6 phenotype and inhibitors were considered. Olanzapine and clozapine are mostly metabolized through flavin mono-oxygenase 3 (olanzapine), CYP3A4 (clozapine) and CYP1A2 (both) , which may also be affected by concomitant medication. In addition, CYP2D6 inhibitors are influenced by other enzymes. Paroxetine is a substrate of CYP2D6 and CYP3A4 , and fluoxetine is metabolized by several CYP enzymes including CYP2C9, CYP2C19, CYP2D6 and CYP3A4 . Non-NM status for these enzymes may affect their pharmacokinetic and -dynamic properties . For example, increased metabolization of CYP2D6 inhibitors may affect their strength, and patients are classified erroneously as phenoconverted. These limitations could be overcome in future studies utilizing larger samples, in diverse population. A suggestion to include more women may be to include older participants, as women tend to be diagnosed later than men . Implementing a standardized TDM protocol into the study design would benefit the ascertainment of phenotype, and future studies should ensure thorough documentation of medication use, dose and duration of treatment. Prescription reporting through pharmacies could be used in addition to patient interviews to assess medication regimes. Furthermore, using modern approaches, data analysis allowing for the examination of copy number variants to categorize UM individuals would be of interest. Other CYP enzymes, such as CYP1A2 and CYP3A4 may be included in future analysis and their activity could be used to more accurately estimate the inhibitory activity of concomitant medication. Pharmacogenetic studies should always include polypharmacy and phenoconversion in their analysis and report pPT as well as gPT in their results. We conclude that the prevalence of phenoconversion was high in the GROUP cohort and accounted for a significant increase in PM status due to concomitant use of SERTs. Neither CYP2D6 -predicted nor phenoconversion-corrected phenotype was robustly associated with outcome measures. Risperidone, however, was affected most by CYP2D6 genotype. The results suggest that PM phenotype is more often caused by phenoconversion than genotype, signifying the importance of phenoconversion in this psychiatric sample. sj-docx-1-jop-10.1177_02698811241278844 – Supplemental material for Clinical effects of CYP2D6 phenoconversion in patients with psychosis Supplemental material, sj-docx-1-jop-10.1177_02698811241278844 for Clinical effects of CYP2D6 phenoconversion in patients with psychosis by Emma Y De Brabander, Esmee Breddels, Therese van Amelsvoort and Roos van Westrhenen; GROUP Investigators in Journal of Psychopharmacology |
Assessing Eye Clinic Accessibility: A Study Validating and Applying the SiteWise Survey | ad001daf-422e-4ff8-8e83-7881477dae64 | 11534020 | Ophthalmology[mh] | As the prevalence of visual impairment increases globally, , with an expected surge in the elderly population, the imperative for adaptive healthcare environments is more pronounced than ever. Particularly in the United States, where the number of individuals with significant visual impairment is projected to double by 2050, the challenge extends beyond mere acknowledgment to active accommodation. Accessibility refers to designing environments and services to be usable by all people to the greatest extent possible without the need for adaptation, while universal design is a broader approach that aims to improve human performance, wellness, and social participation by creating inclusive spaces that are accessible to people of all ability levels. This is crucial because older adults with visual impairment face increased risks of functional decline, necessitating comprehensive support that includes enhanced lighting, contrast, and accessibility in healthcare settings to maintain their independence and quality of life. , Despite significant legislative strides, including the Architectural Barriers Act of 1968, the Americans with Disabilities Act of 1990 (ADA), and the United States signing the International Convention on the Rights of Persons with Disabilities in 2009, our healthcare infrastructure often falls short in addressing the nuanced needs of those with low vision. Critical elements such as optimal lighting, clear signage, and accessible information formats remain inadequately addressed, posing daily challenges for this population. – Inaccessible healthcare facilities can result in missed appointments, delayed treatments, and increased anxiety and stress for visually impaired individuals, ultimately compromising their health outcomes. , Recognizing this, the Henry Ford Health system pioneered a comprehensive survey in 2010, probing the efficacy of community facilities in supporting the elderly and visually impaired. , This initiative not only spotlighted frequent key design deficiencies, but also improved the accessibility of their medical center, leading to the birth of the SiteWise low vision accessibility survey—a concerted effort to bridge the gap between legislative intent and practical, inclusive environments. , Unlike the ADA Center's comprehensive but less specific accessibility checklists, the SiteWise Checklist offers a more detailed, vision-focused approach with specific guidelines and scoring systems to achieve higher standards of accessibility for individuals with low vision or blindness. Our study seeks to validate the SiteWise survey through the analysis of inter-surveyor reliability, scrutinize the design features of a leading academic eye center's clinics, and increase awareness regarding the specific needs of patients with visual impairment and the elderly. By demonstrating the effectiveness of the SiteWise survey and providing this vision-focused tool, we advocate for inclusive and responsive designs that elevate the patient experience and improve accessibility standards in healthcare facilities. Development of SiteWise This low vision accessibility tool was developed with feedback from patients attending Henry Ford Health's Center for Vision Rehabilitation, who reported that after training they were more independent performing their activities of daily living at home but often complained of continued difficulty in the community. Difficult tasks included reading in darker restaurants, churches, and libraries, navigating obstacles such as curbs, freestanding signs, and furniture in libraries, community centers, and churches, and writing at the bank. A pilot program to assess low-vision accessibility in the community was developed, and the survey was tested on ten sites in the metro-Detroit area. It was well received by business owners and community leaders, and, with a grant from the Community Foundation for Southeast Michigan, the project was expanded to survey 30 sites in three counties in the metro-Detroit area. Once the survey was complete, each site was provided with recommendations to improve their facilities for seniors and individuals with visual impairments. The original SiteWise accessibility tool was created in 2010 and updated for clarity in 2022. The survey consists of 83 graded items, evaluating eight sites within the outpatient clinical setting: (1) parking lots/sidewalks, (2) entrances/exits, (3) hallways, (4) stairways, (5) waiting areas, (6) customer service areas, (7) restrooms, and (8) examination rooms. Each item can be graded as “Yes,” “No,” or “Not Applicable.” If the question does not apply to the building or room (e.g., no stairs, no elevator, no windows), “Not Applicable” is checked. displays an example of the parking lot/sidewalk assessment (first page) and the scoring checklist with the SiteWise standard threshold for outstanding (gold standard; 90% and higher), adequate (silver standard; 71%–89%), and minimum (bronze standard; 60%–70%) accessibility (last page). The full survey can be found in . Facility Evaluation Methods The Wilmer Eye Institute's seven satellite clinics and two hospital-based clinics were evaluated using the SiteWise checklist. During standard business hours, independent surveyors assessed each facility. Because of the presence of multiple entrances, rooms, and objects at each facility, surveyors predetermined the specific areas to assess before conducting individual assessments. To ensure the selected routes were representative of typical patient experiences, preliminary observations and consultations with facility staff, including ophthalmic technicians and administrative personnel, were conducted to identify the most frequently used paths by patients. These routes included paths from parking areas to main entrances, primary corridors leading to waiting rooms and examination areas, and other high-traffic zones commonly traversed by patients. In conducting these assessments, several key elements were measured. “Contrast” was defined as the differential in color, brightness, or texture between an element (e.g., sign text) and its backdrop, influencing its readability. Graders assessed contrast using examples from the training material . Light intensity in lux was gauged using a digital light meter (Dr.meter LX1330B; Dr.meter, Newark, CA, USA). Additionally, measurements of size and distance were approximated to the nearest half-inch. Surveyor Training and Proficiency Evaluation Two graders without a background in eye care or medicine were trained in the survey methods. This hour-long training ensured both graders grasped the evaluation procedure . To validate their comprehension, the graders independently evaluated a satellite clinic, later comparing findings to establish consistency. This preliminary evaluation was excluded from the study. Throughout the main data collection from nine facilities, graders conducted assessments independently without mutual discussion of results. Statistical Analyses To evaluate inter-grader agreement, Krippendorff's alpha (KA) values, with 95% confidence intervals, were determined across various units: the total sample, clinic type (satellites vs. hospitals), item category (e.g., presence of hospital components; print size, font, boldness; contrast and glare), individual clinics, and specific sites. Scores for each site and clinic were calculated as the proportion of “Yes” responses to the total number of questions answered as “Yes” or “No,” excluding questions that were not applicable. Subsequently, an average score from both judges was computed. Using Generalized Estimated Equations, these proportions were compared between satellites and hospitals, accounting for repeated measures at sites. Differences in mean scores across individual satellite clinics, within the entire sample, and separately within satellites and hospitals were analyzed using analysis of variance (ANOVA). A t -test assessed score variations between individual hospital clinics. All tests were two-tailed with a significance level set at P < 0.05. Analyses were performed in SAS 9.4. This low vision accessibility tool was developed with feedback from patients attending Henry Ford Health's Center for Vision Rehabilitation, who reported that after training they were more independent performing their activities of daily living at home but often complained of continued difficulty in the community. Difficult tasks included reading in darker restaurants, churches, and libraries, navigating obstacles such as curbs, freestanding signs, and furniture in libraries, community centers, and churches, and writing at the bank. A pilot program to assess low-vision accessibility in the community was developed, and the survey was tested on ten sites in the metro-Detroit area. It was well received by business owners and community leaders, and, with a grant from the Community Foundation for Southeast Michigan, the project was expanded to survey 30 sites in three counties in the metro-Detroit area. Once the survey was complete, each site was provided with recommendations to improve their facilities for seniors and individuals with visual impairments. The original SiteWise accessibility tool was created in 2010 and updated for clarity in 2022. The survey consists of 83 graded items, evaluating eight sites within the outpatient clinical setting: (1) parking lots/sidewalks, (2) entrances/exits, (3) hallways, (4) stairways, (5) waiting areas, (6) customer service areas, (7) restrooms, and (8) examination rooms. Each item can be graded as “Yes,” “No,” or “Not Applicable.” If the question does not apply to the building or room (e.g., no stairs, no elevator, no windows), “Not Applicable” is checked. displays an example of the parking lot/sidewalk assessment (first page) and the scoring checklist with the SiteWise standard threshold for outstanding (gold standard; 90% and higher), adequate (silver standard; 71%–89%), and minimum (bronze standard; 60%–70%) accessibility (last page). The full survey can be found in . The Wilmer Eye Institute's seven satellite clinics and two hospital-based clinics were evaluated using the SiteWise checklist. During standard business hours, independent surveyors assessed each facility. Because of the presence of multiple entrances, rooms, and objects at each facility, surveyors predetermined the specific areas to assess before conducting individual assessments. To ensure the selected routes were representative of typical patient experiences, preliminary observations and consultations with facility staff, including ophthalmic technicians and administrative personnel, were conducted to identify the most frequently used paths by patients. These routes included paths from parking areas to main entrances, primary corridors leading to waiting rooms and examination areas, and other high-traffic zones commonly traversed by patients. In conducting these assessments, several key elements were measured. “Contrast” was defined as the differential in color, brightness, or texture between an element (e.g., sign text) and its backdrop, influencing its readability. Graders assessed contrast using examples from the training material . Light intensity in lux was gauged using a digital light meter (Dr.meter LX1330B; Dr.meter, Newark, CA, USA). Additionally, measurements of size and distance were approximated to the nearest half-inch. Two graders without a background in eye care or medicine were trained in the survey methods. This hour-long training ensured both graders grasped the evaluation procedure . To validate their comprehension, the graders independently evaluated a satellite clinic, later comparing findings to establish consistency. This preliminary evaluation was excluded from the study. Throughout the main data collection from nine facilities, graders conducted assessments independently without mutual discussion of results. To evaluate inter-grader agreement, Krippendorff's alpha (KA) values, with 95% confidence intervals, were determined across various units: the total sample, clinic type (satellites vs. hospitals), item category (e.g., presence of hospital components; print size, font, boldness; contrast and glare), individual clinics, and specific sites. Scores for each site and clinic were calculated as the proportion of “Yes” responses to the total number of questions answered as “Yes” or “No,” excluding questions that were not applicable. Subsequently, an average score from both judges was computed. Using Generalized Estimated Equations, these proportions were compared between satellites and hospitals, accounting for repeated measures at sites. Differences in mean scores across individual satellite clinics, within the entire sample, and separately within satellites and hospitals were analyzed using analysis of variance (ANOVA). A t -test assessed score variations between individual hospital clinics. All tests were two-tailed with a significance level set at P < 0.05. Analyses were performed in SAS 9.4. Our study encompassed a diverse array of healthcare facilities, each representing a unique microenvironment within the broader context of visual impairment accessibility challenges. Specifically, the evaluation included seven satellite clinics and two hospital-based clinics within the Wilmer Eye Institute network, each varying in design, patient population, and service offerings. These facilities were carefully selected to provide a comprehensive overview of current accessibility standards and practices, reflecting the real-world experiences of visually impaired patients navigating these spaces. Intergrader agreement was examined across all evaluated facilities, yielding a KA value of 0.99, indicative of almost perfect consensus . This high level of consistency was mirrored in the specific facility evaluations, with agreement values ranging from 0.97 to 1.00. The uniformity in scoring extended to the site and item levels as well, with agreement ranges of 0.96 to 1.00 observed in both categories ( and , respectively). Hospitals recorded an average SiteWise survey score of 78.9%, marginally surpassing the satellite clinics, which averaged at 71.3% . This distinction between hospitals and satellite clinics was statistically significant, with a mean difference of −7.7% ( P = 0.005). Interestingly, the intra-group comparison of survey scores were not statistically significant. Based on the key for SiteWise standards, three of the seven clinics were classified as silver standard, whereas four were rated as bronze standard. Both hospitals were also classified as silver standard. An in-depth analysis of mean site scores across all facilities revealed important distinctions . Within the total sample, the ANOVA highlighted statistically significant differences in mean score across clinic sites ( ; F (7) = 7.5, P < 0.0001). The highest scores, with means of 89.3%, 87.1%, and 79.3%, were observed in “Hallways,” “Waiting Areas,” and “Customer Service Area,” respectively. Conversely, “Parking lot/sidewalks” and “Stairways” registered lower means at 60.9% and 61.1%, respectively. Within satellite clinics, the ANOVA highlighted significant mean site score differences ( ; F (7) = 6.3, P < .0001). Specific sites, like “Hallways” (mean = 86.1%) and “Waiting Areas” (mean = 83.4%), outperformed others, whereas “Parking lot/sidewalks” and “Stairways” scored 56.2% and 61%, respectively. Hospital site scores mirrored these discrepancies . In our assessment, we used the SiteWise tool, a novel instrument specifically developed to evaluate low vision accessibility in outpatient medical clinics. Several previous tools and studies have delved into factors crucial for enhancing access for low vision patients , – ; the SiteWise tool stands out as the first instrument tailored to gauge the accessibility of clinics for visually impaired individuals. Notably, its design emphasizes elements such as lighting and contrast, which could potentially create challenges for this demographic. Intriguingly, these aspects may also be problematic for those without any visual impairment. A standout feature of the SiteWise tool is that it was developed through direct feedback from low-vision patients. Each criterion reflects real-world challenges encountered by patients within a clinical setting, highlighting that trained visual skills and prescribed optical aids are insufficient if the environment remains inaccessible. This approach aligns with the proposed ICF standards by Billiet et al., which stress the significance of considering both environmental and personal factors when aiming to create an inclusive healthcare environment for individuals with vision loss. Moreover, the survey's design facilitates easy administration by medical personnel such as clinic managers or facilities staff—a crucial feature because low vision services may not always be provided by ophthalmologists. Validation of the SiteWise Survey The applicability of this instrument to our sample and the high interobserver reliability demonstrates the instrument's validity. We observed almost perfect agreement (0.96 to 0.99) between two non-expert graders in all facilities, sites, and items ( and ). This high level of agreement between graders may partially reflect the initial training (which included visual examples of contrast) and preliminary evaluation that was completed to validate their comprehension. Additionally, objectively graded items such as low light (measured by lux meter) and font size (measured in inches) may contribute to the level of consensus. Accessibility Evaluation in Ophthalmology Clinics We observed a statistically significant difference in accessibility scores between hospitals and satellite clinics, with hospitals achieving a slightly higher average score of 78.9%, in contrast to satellite clinics which averaged 71.3% . This disparity may be attributed to factors such as the age of the facilities, the design and renovation history, and the scale of services offered, which may affect adherence to contemporary ADA standards. Furthermore, the lower average score for satellite clinics could be influenced by the limited scope of renovations possible outside of and within multipurpose building suites. It is notable that the Johns Hopkins Hospital eye clinic, which includes a clinic specifically tailored for low-vision patients, attained a high score of 82.3%, thereby underscoring the potential impact of specialized design considerations. The observed discrepancy also may be influenced by the small sample size, with only two hospitals versus seven satellite clinics evaluated. Therefore, although all facilities are mandated to be accessible by the ADA, the degree to which they meet or exceed these standards can vary considerably. In addition, our study identified significant variations in accessibility scores across different sites within all evaluated facilities. “Hallways,” “Waiting Areas,” “Customer Service Areas,” and “Restrooms” achieved the highest scores, indicating adequate accessibility (silver standard) by SiteWise Standards . On the other hand, “Parking lot/sidewalks,” “Stairways,” “Entrances/exits,” and “Exam Rooms” scored lower, receiving the lowest accessibility score (bronze standard). These results emphasize that although internal aspects of a facility like hallways and waiting areas may be adequately tailored for low vision accessibility, outdoor and transitional areas require attention the most. These observations also suggest that architectural and design considerations might often be inward-focused, with greater emphasis on internal environments rather than holistic facility design. There was a variation in the item deficits for each low-scoring site. Entrances and exits often lacked sufficient contrast against metal door dividers, complicating the ability of visually impaired individuals to discern different areas, a challenge compounded for the elderly who are at a higher risk for fall-related ocular trauma. Stairways, frequently suffering from inadequate lighting below the recommended 400–600 lux, absence of contrasting railings, and non-contrasted step edges, not only impede visibility for the visually impaired but also significantly increase the risk of falls, which can result in serious ocular injuries. In fact, falls are the leading cause of injury-related emergency department visits and accidental deaths among older adults, with an estimated 3 million emergency department visits, over 950,000 hospitalizations, and approximately 32,000 deaths annually because of fall-related injuries. Parking areas and pathways often missed high-contrast curb edges and clear markings for inclines and declines, which are essential to prevent missteps and falls, whereas unobstructed walk paths are crucial for safe navigation. Examination rooms predominantly displayed undersized room numbers and lacked contrast on chairs, instrument stands, and tables, further contributing to the hazard landscape for visually impaired patients. Overall, these features have critical implications for safety and accessibility, particularly for the visually impaired and elderly. These identified deficits provide actionable insights for facility managers and designers, emphasizing the need for individualized solutions rather than a one-size-fits-all approach. Each area of a facility may require unique modifications, informed by the feedback from those who navigate these spaces regularly—in this instance, the visually impaired patients. For example, the strategic application of contrasting colors and improved lighting could mitigate fall risks, especially during winter months when fall-related injuries peak. Furthermore, our sample revealed that four clinics were classified as bronze according to the SiteWise standard, guiding us to target these clinics for quality improvement. Ultimately, a concerted effort to redesign these spaces with the input of end-users can significantly reduce injury rates and improve overall safety for all patrons. These recommendations for facility-specific improvements, derived from the direct experiences of visually impaired individuals, paves the way for a more expansive conversation on how these principles can be generalized to enhance accessibility across all public and private spaces. Although our research has concentrated on accessibility within eye clinics, the implications likely extend to a variety of public and private spaces frequented by visually impaired individuals. In line with the Inclusive Design framework as detailed by Gomez et al., which prioritizes usability for people regardless of their visual abilities, our findings advocate for design modifications that span beyond healthcare settings. This framework entails creating environments where visual contrasts, tactile cues, auditory signals, and intuitive navigation are integral, thus accommodating the full spectrum of sensory engagement. Such enhancements are not exclusive to those with visual impairments; they benefit a broader population, including the elderly and patients with sensory deficits. Future research should thus examine the degree to which these inclusive design features are present or absent in a range of facilities, comparing these to the deficits we have documented in eye clinics. This would enable a more nuanced understanding of how the built environment can either support or hinder the independence and safety of visually impaired individuals, reinforcing the need for universally accessible design that aligns with the evolving demographics and diverse capabilities of the global population. This study's limitations warrant careful consideration. Primarily, the evaluations were constrained to a single point in time, which may not capture the full spectrum of accessibility challenges. Moreover, our comparative analysis relied solely on non-specialist surveyors; a more comprehensive evaluation would benefit from including both specialists and non-specialists to provide a nuanced understanding of the tool's clinical relevance. Furthermore, the research used a convenience sample drawn from clinics within one healthcare system in a specific geographic area, which may limit the generalizability of our findings to other medical or hospital outpatient facilities, especially in varied locales. These considerations highlight the necessity for expanded research across a broader and more varied range of facilities to thoroughly validate the clinical applicability of our assessment tool. The applicability of this instrument to our sample and the high interobserver reliability demonstrates the instrument's validity. We observed almost perfect agreement (0.96 to 0.99) between two non-expert graders in all facilities, sites, and items ( and ). This high level of agreement between graders may partially reflect the initial training (which included visual examples of contrast) and preliminary evaluation that was completed to validate their comprehension. Additionally, objectively graded items such as low light (measured by lux meter) and font size (measured in inches) may contribute to the level of consensus. We observed a statistically significant difference in accessibility scores between hospitals and satellite clinics, with hospitals achieving a slightly higher average score of 78.9%, in contrast to satellite clinics which averaged 71.3% . This disparity may be attributed to factors such as the age of the facilities, the design and renovation history, and the scale of services offered, which may affect adherence to contemporary ADA standards. Furthermore, the lower average score for satellite clinics could be influenced by the limited scope of renovations possible outside of and within multipurpose building suites. It is notable that the Johns Hopkins Hospital eye clinic, which includes a clinic specifically tailored for low-vision patients, attained a high score of 82.3%, thereby underscoring the potential impact of specialized design considerations. The observed discrepancy also may be influenced by the small sample size, with only two hospitals versus seven satellite clinics evaluated. Therefore, although all facilities are mandated to be accessible by the ADA, the degree to which they meet or exceed these standards can vary considerably. In addition, our study identified significant variations in accessibility scores across different sites within all evaluated facilities. “Hallways,” “Waiting Areas,” “Customer Service Areas,” and “Restrooms” achieved the highest scores, indicating adequate accessibility (silver standard) by SiteWise Standards . On the other hand, “Parking lot/sidewalks,” “Stairways,” “Entrances/exits,” and “Exam Rooms” scored lower, receiving the lowest accessibility score (bronze standard). These results emphasize that although internal aspects of a facility like hallways and waiting areas may be adequately tailored for low vision accessibility, outdoor and transitional areas require attention the most. These observations also suggest that architectural and design considerations might often be inward-focused, with greater emphasis on internal environments rather than holistic facility design. There was a variation in the item deficits for each low-scoring site. Entrances and exits often lacked sufficient contrast against metal door dividers, complicating the ability of visually impaired individuals to discern different areas, a challenge compounded for the elderly who are at a higher risk for fall-related ocular trauma. Stairways, frequently suffering from inadequate lighting below the recommended 400–600 lux, absence of contrasting railings, and non-contrasted step edges, not only impede visibility for the visually impaired but also significantly increase the risk of falls, which can result in serious ocular injuries. In fact, falls are the leading cause of injury-related emergency department visits and accidental deaths among older adults, with an estimated 3 million emergency department visits, over 950,000 hospitalizations, and approximately 32,000 deaths annually because of fall-related injuries. Parking areas and pathways often missed high-contrast curb edges and clear markings for inclines and declines, which are essential to prevent missteps and falls, whereas unobstructed walk paths are crucial for safe navigation. Examination rooms predominantly displayed undersized room numbers and lacked contrast on chairs, instrument stands, and tables, further contributing to the hazard landscape for visually impaired patients. Overall, these features have critical implications for safety and accessibility, particularly for the visually impaired and elderly. These identified deficits provide actionable insights for facility managers and designers, emphasizing the need for individualized solutions rather than a one-size-fits-all approach. Each area of a facility may require unique modifications, informed by the feedback from those who navigate these spaces regularly—in this instance, the visually impaired patients. For example, the strategic application of contrasting colors and improved lighting could mitigate fall risks, especially during winter months when fall-related injuries peak. Furthermore, our sample revealed that four clinics were classified as bronze according to the SiteWise standard, guiding us to target these clinics for quality improvement. Ultimately, a concerted effort to redesign these spaces with the input of end-users can significantly reduce injury rates and improve overall safety for all patrons. These recommendations for facility-specific improvements, derived from the direct experiences of visually impaired individuals, paves the way for a more expansive conversation on how these principles can be generalized to enhance accessibility across all public and private spaces. Although our research has concentrated on accessibility within eye clinics, the implications likely extend to a variety of public and private spaces frequented by visually impaired individuals. In line with the Inclusive Design framework as detailed by Gomez et al., which prioritizes usability for people regardless of their visual abilities, our findings advocate for design modifications that span beyond healthcare settings. This framework entails creating environments where visual contrasts, tactile cues, auditory signals, and intuitive navigation are integral, thus accommodating the full spectrum of sensory engagement. Such enhancements are not exclusive to those with visual impairments; they benefit a broader population, including the elderly and patients with sensory deficits. Future research should thus examine the degree to which these inclusive design features are present or absent in a range of facilities, comparing these to the deficits we have documented in eye clinics. This would enable a more nuanced understanding of how the built environment can either support or hinder the independence and safety of visually impaired individuals, reinforcing the need for universally accessible design that aligns with the evolving demographics and diverse capabilities of the global population. This study's limitations warrant careful consideration. Primarily, the evaluations were constrained to a single point in time, which may not capture the full spectrum of accessibility challenges. Moreover, our comparative analysis relied solely on non-specialist surveyors; a more comprehensive evaluation would benefit from including both specialists and non-specialists to provide a nuanced understanding of the tool's clinical relevance. Furthermore, the research used a convenience sample drawn from clinics within one healthcare system in a specific geographic area, which may limit the generalizability of our findings to other medical or hospital outpatient facilities, especially in varied locales. These considerations highlight the necessity for expanded research across a broader and more varied range of facilities to thoroughly validate the clinical applicability of our assessment tool. The SiteWise tool has revealed key areas for improvement in clinic accessibility for the visually impaired. Although indoor spaces like hallways and waiting rooms are generally accessible, outdoor, and transitional areas in our sample need attention. The disparities in scores between hospitals and satellite clinics further highlight the importance of tailored solutions for different facilities. These findings not only provide actionable insights for eye clinics but also underscore the broader need for inclusive design in various community spaces. Our application of this instrument to our clinics demonstrates its validity and its utility in quality improvement projects. Healthcare clinics that care for the visually impaired must advocate for more accessible design elements, regardless of the services provided. Future studies should broaden their scope, considering diverse facilities and broader participant involvement, to ensure universally accessible environments. Supplement 1 Supplement 2 |
Decade-long impact factors in ophthalmology journals and the effect
of journal characteristics: a longitudinal study | f2bd7b59-3cb0-43f1-b2cf-7fc89f93c73c | 11826576 | Ophthalmology[mh] | |
Clinical utility of tumor-infiltrating lymphocyte evaluation by two different methods in breast cancer patients treated with neoadjuvant chemotherapy | 26d38ccc-7c28-4b7b-ac7e-db3f21c152a9 | 11842476 | Surgical Procedures, Operative[mh] | The tumor immune microenvironment has been shown to have a significant influence on the disease process of breast cancer . Tumor-infiltrating lymphocytes (TILs) are immune cells that migrate from the blood to the tumor microenvironment, where they exert a variety of functions . The immune cells found in the tumor and its stroma include not only lymphocytes, such as T cells and B cells, but also natural killer cells, macrophages, neutrophils, and dendritic cells . Although the composition of immune cells varies depending on tumor character, and the function of these immune cells changes dynamically during the tumor development process, TILs often act in a tumor suppressive manner through the function of tumor suppressor cells, especially CD8 + T cells . Indeed, it has been reported that increased TILs are associated with a better prognosis in breast cancer patients. Moreover, TILs are associated with the efficacy of chemotherapy, and increased TILs are associated with higher pathological complete response (pCR) rates . Furthermore, immune checkpoint inhibitors have recently been introduced in triple-negative breast cancer (TNBC), and the relationship between TILs and the efficacy of immune checkpoint inhibitors has been reported . Thus, in breast cancer, TILs are attracting attention, not only as a prognostic marker, but also as a predictive marker for various therapies. To assess TILs in breast cancer clinically, the International TILs Working Group introduced a standard method of histopathological evaluation of TILs, which has been widely accepted . In this method, the overall trend of TIL distribution is observed by microscopy, and histopathological analysis is performed to assess the average value of TILs in the tumor stromal area. In addition to this standard method, we have previously focused on hot-spot areas of TILs and proposed a simpler method to evaluate TILs using hot-spots, and have reported its clinical usefulness . The advantage of the “hot-spot” method is that it is easy to evaluate differences between cases, even when TILs are very low, as in hormone receptor-positive breast cancer . Based on the above, we believe that the “average” method standardized by the International TILs Working Group is the international standard, and that the hot-spot method can also be used to effectively evaluate TILs for clinical use, depending on the particular case. In recent years, human epidermal growth factor receptor 2 (HER2)-positive breast cancer and TNBC have been increasingly indicated for preoperative chemotherapy with the introduction of the residual disease-guided approach and immune checkpoint inhibitors . Even considering that TILs correlate well with pCR in HER2-positive and TNBCs, the evaluation of TILs utilizing biopsy specimens prior to treatment is expected to be clinically useful. Since a biopsy specimen has a much smaller area than a surgical specimen, this may affect the evaluation of TILs. Although there are many reports evaluating TILs in pre-treatment biopsy specimens and post-treatment surgical specimens, and assessing the clinical utility of each, it has not been adequately verified whether the evaluation of TIL in biopsy specimens is equivalent to that in surgical specimens. The aim of this study was to examine the clinical utility of TILs evaluated by the two different methods of average and hot-spot using biopsy specimens in breast cancer patients who had undergone preoperative chemotherapy. To examine the consistency of TIL assessment in biopsy and surgical specimens, we also studied surgical cases that had not received preoperative treatment. Furthermore, for future development, we aimed to conduct an exploratory study to determine whether it is possible to predict TIL scores using only clinical information, such as blood data and pathological findings obtained in routine clinical practice, without directly measuring TILs. Patient eligibility From August 2016 to October 2023, 907 breast cancer patients underwent surgery at Hyogo Medical University Hospital, and the pathological diagnosis was evaluated for each patient. Of these, 286 patients received preoperative therapies, including neoadjuvant chemotherapy (NAC) in 182 patients and neoadjuvant hormone therapy in 104 patients, and the remaining 621 patients underwent surgery without any preoperative therapies. Out of the 621 surgical cases without preoperative therapy, 367 had formalin-fixed paraffin-embedded tissue samples of both biopsy and surgical specimens in storage and available for examination. From the 182 patients receiving NAC, patients with bilateral breast cancer, recurrent cancer, and Stage IV disease with distant metastasis were excluded, leaving 144 patients for whom clinical information was available included in the study. Similarly, patients with bilateral breast cancer, recurrent cancer, and Stage IV disease with distant metastasis were excluded from the 621 surgical cases without preoperative therapy, and 477 patients for whom clinical information was available were included in the prognostic analysis. The Institutional Review Board of the Hyogo Medical University approved this study (No. 1886), which was planned in accordance with the Declaration of Helsinki. Since this study collected only retrospective clinical data and offered no risk to the patients, they were not asked for their written informed consent. Pathological diagnosis, immunohistochemistry and subtype classification All patients included in the study were histologically diagnosed with breast cancer. Immunohistochemistry for estrogen receptor (ER), HER2 and Ki-67 was performed on formalin-fixed paraffin-embedded tumor tissues obtained from biopsy or surgical specimens. Cases with nuclear staining of 1% or more were classified as ER-positive. Cases were also classified as HER2-negative with either an immunohistochemistry score of 0 to 2+ or with no HER2 amplification, as confirmed by in situ hybridization. Based on immunohistochemistry, we classified ER-positive and HER2-negative breast cancers into luminal A and luminal B according to Ki-67 expression levels (luminal A: Ki-67 < 25%; luminal B: Ki-67 ≥ 25%). Pathological evaluation of TILs Formalin-fixed paraffin-embedded biopsy and surgical specimens were used for preparing hematoxylin and eosin-stained sections, and TIL scores were evaluated under microscopy by the hot-spot method and average method as described previously . Briefly, for the hot-spot method, we microscopically identified lesions containing a relatively high number of invasive cancer cells and lymphocyte infiltration using a low-power field (40×). The hot-spot with the highest lymphocyte infiltration was selected in a medium-power field (100×). Excluding neutrophils, eosinophils, and macrophages, lymphocytes and plasma cells in both the peritumoral and intertumoral stromal regions were evaluated. Finally, the TIL score was calculated as the percentage of area involved in lymphocytes and plasma cells out of the entire tumor and adjacent stroma. For the “average” method, evaluation was performed based on the standardized assessment of TILs in breast cancer by the International Immuno-Oncology Biomarker Working Group ( www.tilsinbreastcancer.org ) . We estimated the proportion of the area infiltrated by lymphocytes to the area of the entire tumor plus adjacent stroma, and classified the TIL scores as low (< 10%), intermediate (10–50%), or high (≥ 50%) in both methods. TIL scores were independently examined by two investigators, and in cases of discrepancies, they were discussed until a consensus was reached. Collection of clinical data for machine learning For the 621 surgical cases without preoperative therapy, data including preoperative blood parameters, height, weight, date of surgery, date of birth, TNM classification, surgical findings, and pathology findings of resection specimens were collected from the medical records of eligible patients. TIL scores (hot-spot and average) were also obtained. Data preprocessing The data was preprocessed as described below to make it ready for use in machine learning. For blood hemoglobin A1c, National Glycohemoglobin Standardization Program values were used, except in patients where only the Japan Diabetes Society standard values were available, in which case the Japan Diabetes Society values plus 0.4 were used. For numerical data, values that were found to be below the detection limit were treated as the detection limit value itself. For the degree of lymphatic and vascular invasion in pathological findings, the pathologist's grading was converted into numerical values. In cases where the degree of lymphatic or vascular invasion was reported to be absent, the item was set to 0. HER2 immunohistochemistry staining was also converted from a grading system to a numerical value. Age at surgery was calculated from the date of surgery and date of birth. T and N from the TNM classification were converted into numerical values according to the conversion described in Supplementary Table 1. Since tissue histological type is a nominal measure, it was converted to a one-hot vector . Finally, we excluded cases in which the histological type was ≤ 1% of all the 621 cases. As a result, 607 cases remained. Learning Python 3.8 and Scikit-learn 0.24.2 were used for machine learning with random forests. Random forests, developed by Tin Kam Ho , Leo Breiman , and others, are a new approach to machine learning and involve an ensemble learning algorithm using decision trees. In the present study, the explanatory variables taken from preoperative blood data were lactate dehydrogenase, carcinoembryonic antigen, carbohydrate antigen 15-3, albumin, low-density lipoprotein cholesterol, triglycerides, hemoglobin A1c, C-reactive protein, white blood cell count, red blood cell count, hemoglobin level, platelet count, ratio of segmented leukocytes, and ratio of lymphocytes. In addition, the following explanatory variables were taken from the pathological findings: degree of lymphatic invasion, vascular invasion, nuclear grade, histological grade, ER status, progesterone receptor status, Ki-67, and HER2-positive status. Other explanatory variables used were height, weight, age at the date of surgery, pT and pN (from the TNM classification) converted to tabulated values, and one-hot vectorized histological type. From the 607 cases, cases with missing explanatory variables were excluded. As a result, 468 cases remained. The 468 cases were divided into study and test cases at a ratio of 7:3. There were 327 training cases and 141 test cases. A random forest regressor was trained using the 327 training cases as supervised data, with TILs (hot spots) or TILs (average) as the objective variable. RandomForestRegressor from Scikit-learn was used as the random forest regressor. The regressors after training were used to predict the objective variable for the 141 test cases, and the obtained predictions were compared with the true values to calculate the root mean squared error. The root mean squared error represents the average absolute value of how far the predicted value deviates from the true value. For example, a root mean squared error of 10 implies that if the true value is 40%, the predicted value is expected to be in the range of 30% to 50%. Statistical analysis A chi-squared test was used to compare data on TILs and clinical findings between the two groups or between multiple groups. Kaplan–Meier plots and log-rank tests of disease-free survival (DFS) or overall survival (OS) were applied for each group. Statistical calculations were performed using JMP Pro 17 (SAS Institute Inc., Cary, NC, USA) and a at p value < 0.05 was considered to indicated a significant difference with a two-tailed test. From August 2016 to October 2023, 907 breast cancer patients underwent surgery at Hyogo Medical University Hospital, and the pathological diagnosis was evaluated for each patient. Of these, 286 patients received preoperative therapies, including neoadjuvant chemotherapy (NAC) in 182 patients and neoadjuvant hormone therapy in 104 patients, and the remaining 621 patients underwent surgery without any preoperative therapies. Out of the 621 surgical cases without preoperative therapy, 367 had formalin-fixed paraffin-embedded tissue samples of both biopsy and surgical specimens in storage and available for examination. From the 182 patients receiving NAC, patients with bilateral breast cancer, recurrent cancer, and Stage IV disease with distant metastasis were excluded, leaving 144 patients for whom clinical information was available included in the study. Similarly, patients with bilateral breast cancer, recurrent cancer, and Stage IV disease with distant metastasis were excluded from the 621 surgical cases without preoperative therapy, and 477 patients for whom clinical information was available were included in the prognostic analysis. The Institutional Review Board of the Hyogo Medical University approved this study (No. 1886), which was planned in accordance with the Declaration of Helsinki. Since this study collected only retrospective clinical data and offered no risk to the patients, they were not asked for their written informed consent. All patients included in the study were histologically diagnosed with breast cancer. Immunohistochemistry for estrogen receptor (ER), HER2 and Ki-67 was performed on formalin-fixed paraffin-embedded tumor tissues obtained from biopsy or surgical specimens. Cases with nuclear staining of 1% or more were classified as ER-positive. Cases were also classified as HER2-negative with either an immunohistochemistry score of 0 to 2+ or with no HER2 amplification, as confirmed by in situ hybridization. Based on immunohistochemistry, we classified ER-positive and HER2-negative breast cancers into luminal A and luminal B according to Ki-67 expression levels (luminal A: Ki-67 < 25%; luminal B: Ki-67 ≥ 25%). Formalin-fixed paraffin-embedded biopsy and surgical specimens were used for preparing hematoxylin and eosin-stained sections, and TIL scores were evaluated under microscopy by the hot-spot method and average method as described previously . Briefly, for the hot-spot method, we microscopically identified lesions containing a relatively high number of invasive cancer cells and lymphocyte infiltration using a low-power field (40×). The hot-spot with the highest lymphocyte infiltration was selected in a medium-power field (100×). Excluding neutrophils, eosinophils, and macrophages, lymphocytes and plasma cells in both the peritumoral and intertumoral stromal regions were evaluated. Finally, the TIL score was calculated as the percentage of area involved in lymphocytes and plasma cells out of the entire tumor and adjacent stroma. For the “average” method, evaluation was performed based on the standardized assessment of TILs in breast cancer by the International Immuno-Oncology Biomarker Working Group ( www.tilsinbreastcancer.org ) . We estimated the proportion of the area infiltrated by lymphocytes to the area of the entire tumor plus adjacent stroma, and classified the TIL scores as low (< 10%), intermediate (10–50%), or high (≥ 50%) in both methods. TIL scores were independently examined by two investigators, and in cases of discrepancies, they were discussed until a consensus was reached. For the 621 surgical cases without preoperative therapy, data including preoperative blood parameters, height, weight, date of surgery, date of birth, TNM classification, surgical findings, and pathology findings of resection specimens were collected from the medical records of eligible patients. TIL scores (hot-spot and average) were also obtained. The data was preprocessed as described below to make it ready for use in machine learning. For blood hemoglobin A1c, National Glycohemoglobin Standardization Program values were used, except in patients where only the Japan Diabetes Society standard values were available, in which case the Japan Diabetes Society values plus 0.4 were used. For numerical data, values that were found to be below the detection limit were treated as the detection limit value itself. For the degree of lymphatic and vascular invasion in pathological findings, the pathologist's grading was converted into numerical values. In cases where the degree of lymphatic or vascular invasion was reported to be absent, the item was set to 0. HER2 immunohistochemistry staining was also converted from a grading system to a numerical value. Age at surgery was calculated from the date of surgery and date of birth. T and N from the TNM classification were converted into numerical values according to the conversion described in Supplementary Table 1. Since tissue histological type is a nominal measure, it was converted to a one-hot vector . Finally, we excluded cases in which the histological type was ≤ 1% of all the 621 cases. As a result, 607 cases remained. Python 3.8 and Scikit-learn 0.24.2 were used for machine learning with random forests. Random forests, developed by Tin Kam Ho , Leo Breiman , and others, are a new approach to machine learning and involve an ensemble learning algorithm using decision trees. In the present study, the explanatory variables taken from preoperative blood data were lactate dehydrogenase, carcinoembryonic antigen, carbohydrate antigen 15-3, albumin, low-density lipoprotein cholesterol, triglycerides, hemoglobin A1c, C-reactive protein, white blood cell count, red blood cell count, hemoglobin level, platelet count, ratio of segmented leukocytes, and ratio of lymphocytes. In addition, the following explanatory variables were taken from the pathological findings: degree of lymphatic invasion, vascular invasion, nuclear grade, histological grade, ER status, progesterone receptor status, Ki-67, and HER2-positive status. Other explanatory variables used were height, weight, age at the date of surgery, pT and pN (from the TNM classification) converted to tabulated values, and one-hot vectorized histological type. From the 607 cases, cases with missing explanatory variables were excluded. As a result, 468 cases remained. The 468 cases were divided into study and test cases at a ratio of 7:3. There were 327 training cases and 141 test cases. A random forest regressor was trained using the 327 training cases as supervised data, with TILs (hot spots) or TILs (average) as the objective variable. RandomForestRegressor from Scikit-learn was used as the random forest regressor. The regressors after training were used to predict the objective variable for the 141 test cases, and the obtained predictions were compared with the true values to calculate the root mean squared error. The root mean squared error represents the average absolute value of how far the predicted value deviates from the true value. For example, a root mean squared error of 10 implies that if the true value is 40%, the predicted value is expected to be in the range of 30% to 50%. A chi-squared test was used to compare data on TILs and clinical findings between the two groups or between multiple groups. Kaplan–Meier plots and log-rank tests of disease-free survival (DFS) or overall survival (OS) were applied for each group. Statistical calculations were performed using JMP Pro 17 (SAS Institute Inc., Cary, NC, USA) and a at p value < 0.05 was considered to indicated a significant difference with a two-tailed test. Consistency of TIL scores between biopsy and surgical specimens using the hot-spot and average methods To determine the consistency of TIL scores evaluated by the hot-spot or average methods between biopsy and surgical specimens in the same patient, we studied specimens from 367 patients undergoing surgery without preoperative therapy in which both biopsy and surgical specimens were available (Fig. a). Using the hot-spot method, surgical specimens showed higher TIL scores than biopsy specimens, resulting in significant differences in the percentages of TIL-score categories ( p < 0.001; Fig. a). On the other hand, biopsy and surgical specimens showed similar percentages of TIL-score categories using the average method ( p = 0.83; Fig. b). The percentage of patients with concordant TIL-score categories for biopsy and surgical specimens (high = high, intermediate = intermediate, or low = low) was 67.0% for the hot-spot method (Fig. c) and 86.1% for the average method (Fig. d). Association between pre-treatment TIL scores and pCR rate in breast cancer patients receiving NAC Next, we evaluated the clinical significance of TIL scores determined by the hot-spot and average methods utilizing biopsy specimens taken for the initial diagnosis in 144 breast cancer patients who underwent NAC. In this patient group, the percentage of patients with high-TILs using the hot-spot method (25.0%) was significantly higher than for the average method (10.4%; p < 0.001; Supplementary Fig. 1). The cancer subtypes for the 144 patients receiving NAC were TNBC in 36 patients (25.0%), HER2-positive in 57 (39.6%), luminal B type in 34 (23.6%), and luminal A type in 17 (11.8%). Looking at the data by the TIL evaluation method and subtype, the percentage of patients with high-TILs tended to be lower in the luminal A subtype compared to other subtypes, although there was no significant difference using either the hot-spot method ( p = 0.44; Fig. a) or the average method ( p = 0.16, Fig. b). A linear relationship existed between the pCR rate and TIL scores determined by the hot-spot method ( p < 0.001, Fig. c), as well as the average method ( p = 0.001, Fig. d). According to subtype, the pCR rate was linearly related to hot-spot TIL scores in HER2-positive subtypes ( p = 0.002) and tended to be related to hot-spot TIL scores in the TNBC and luminal B subtypes ( p = 0.08 and p = 0.06, respectively, Fig. e). Moreover, there was a linear relationship between the pCR rates in TNBC and HER2-positive subtypes and TIL scores determined by the average method ( p = 0.02 and p = 0.04, respectively, Fig. f). In the luminal A subtype, only a few patients showed pCR, and there was no relationship between TILs and pCR using either the hot-spot or average methodologies (Fig. e and f). Association between pre-treatment TIL scores and clinical outcomes in breast cancer patients receiving NAC In this group of 144 breast cancer patients who underwent NAC, we evaluated the association between the TIL score assessed in the pre-NAC biopsy specimen and the patient’s clinical outcome. Kaplan–Meier curves revealed that TILs determined by hot-spot tended to be associated with DFS after surgery for breast cancer patients who underwent NAC ( p = 0.05, Supplementary Fig. 2a), while there was no significant association between TILs determined by the average method and DFS ( p = 0.34, Supplementary Fig. 2b). By subtype, there was no association between TILs determined by hot-spot or average methods and DFS (Supplementary Fig. 2c–j). Moreover, although there was no significant association between TILs determined by hot-spot and OS for these patients (Supplementary Fig. 3a), TILs determined by the average method tended to show an association with OS ( p = 0.08, Supplementary Fig. 3b). Furthermore, by subtype, there was no association between TILs determined by the hot-spot or average methods and OS (Supplementary Fig. 3c–j). Looking carefully at the OS data, all patients with high or intermediate TILs by the average method survived (Supplementary Fig. 3b). Therefore, we re-evaluated the clinical outcomes (DFS and OS) between high/intermediate-TILs and low-TILs for both methods to assess the clinical significance of TILs (Fig. a–d). The high/intermediate-TIL group showed a trend toward better DFS ( p = 0.14, Fig. b) and significantly better OS ( p = 0.03, Fig. d) than the low-TIL group by the average method, although analysis using the hot-spot data did not show any significant difference either in DFS or OS (Fig. a, c). Next, we confirmed the association between pCR and clinical outcomes; breast cancer patients who achieved pCR had significantly better DFS ( p = 0.01; Fig. e) and tended to have better OS ( p = 0.10; Fig. f) than those who did not achieve pCR. We further hypothesized that combining pCR status with TIL scores might better predict clinical outcomes, and we investigated the relationship between the combined index of TILs/pCR and clinical outcomes (DFS and OS) (Fig. ). This additional analysis revealed that, by the average method, patients not achieving pCR with low-TILs had significantly poorer DFS and OS than other patients ( p = 0.03, p = 0.02, respectively; Fig. b, d), although analysis using the hot-spot data did not show any significant difference in either DFS or OS (Fig. a, c). Furthermore, when we analyzed OS by TILs in the patients with recurrence among those receiving NAC, all the patients in the high/intermediate-TILs group were alive, and deaths only occurred in the low-TILs group (Supplementary Fig. 4). In the 477 eligible surgical cases without NAC, we also examined the difference in clinical outcomes (DFS and OS) between the two groups of high/intermediate and low TILs by the hot-spot and average methods (Supplementary Fig. 5). Although TIL scores by both the hot-spot and average methods did not show any association with DFS, high/intermediate TILs showed significantly better OS than low TILs by the hot spot method ( p = 0.04, Supplementary Fig. 5c), and high/intermediate TILs tended to have better OS than low TILs by the average method ( p = 0.11, Supplementary Fig. 5d). Prediction of TIL score by machine learning using clinicopathological data without pathological evaluation of TILs In patients who received NAC, our data confirmed that the TIL score is related to pCR rate and OS. However, to assess the TIL score for each patient in clinical practice, a significant effort would be required for pathologists, which may limit its implementation. Therefore, we conducted an exploratory study to see if it was possible to predict the TIL score using only clinical parameters obtained in routine clinical practice, without the pathological assessment of TILs. To this end, we used 468 surgical cases to predict TIL scores by machine learning using clinicopathological findings; seventy percent of the cases were used as a training set and thirty percent were used as a test set. TIL scores evaluated by the hot-spot method were predicted using a random forest regressor, and the root mean squared error for the training set was 8.143, and that for the test set was 19.962 (Fig. a). Finally, TIL scores evaluated by the average method were predicted using a random forest regressor, and the root mean squared error for the training set was 4.664, and that for the test set was 10.955 (Fig. b). To determine the consistency of TIL scores evaluated by the hot-spot or average methods between biopsy and surgical specimens in the same patient, we studied specimens from 367 patients undergoing surgery without preoperative therapy in which both biopsy and surgical specimens were available (Fig. a). Using the hot-spot method, surgical specimens showed higher TIL scores than biopsy specimens, resulting in significant differences in the percentages of TIL-score categories ( p < 0.001; Fig. a). On the other hand, biopsy and surgical specimens showed similar percentages of TIL-score categories using the average method ( p = 0.83; Fig. b). The percentage of patients with concordant TIL-score categories for biopsy and surgical specimens (high = high, intermediate = intermediate, or low = low) was 67.0% for the hot-spot method (Fig. c) and 86.1% for the average method (Fig. d). Next, we evaluated the clinical significance of TIL scores determined by the hot-spot and average methods utilizing biopsy specimens taken for the initial diagnosis in 144 breast cancer patients who underwent NAC. In this patient group, the percentage of patients with high-TILs using the hot-spot method (25.0%) was significantly higher than for the average method (10.4%; p < 0.001; Supplementary Fig. 1). The cancer subtypes for the 144 patients receiving NAC were TNBC in 36 patients (25.0%), HER2-positive in 57 (39.6%), luminal B type in 34 (23.6%), and luminal A type in 17 (11.8%). Looking at the data by the TIL evaluation method and subtype, the percentage of patients with high-TILs tended to be lower in the luminal A subtype compared to other subtypes, although there was no significant difference using either the hot-spot method ( p = 0.44; Fig. a) or the average method ( p = 0.16, Fig. b). A linear relationship existed between the pCR rate and TIL scores determined by the hot-spot method ( p < 0.001, Fig. c), as well as the average method ( p = 0.001, Fig. d). According to subtype, the pCR rate was linearly related to hot-spot TIL scores in HER2-positive subtypes ( p = 0.002) and tended to be related to hot-spot TIL scores in the TNBC and luminal B subtypes ( p = 0.08 and p = 0.06, respectively, Fig. e). Moreover, there was a linear relationship between the pCR rates in TNBC and HER2-positive subtypes and TIL scores determined by the average method ( p = 0.02 and p = 0.04, respectively, Fig. f). In the luminal A subtype, only a few patients showed pCR, and there was no relationship between TILs and pCR using either the hot-spot or average methodologies (Fig. e and f). In this group of 144 breast cancer patients who underwent NAC, we evaluated the association between the TIL score assessed in the pre-NAC biopsy specimen and the patient’s clinical outcome. Kaplan–Meier curves revealed that TILs determined by hot-spot tended to be associated with DFS after surgery for breast cancer patients who underwent NAC ( p = 0.05, Supplementary Fig. 2a), while there was no significant association between TILs determined by the average method and DFS ( p = 0.34, Supplementary Fig. 2b). By subtype, there was no association between TILs determined by hot-spot or average methods and DFS (Supplementary Fig. 2c–j). Moreover, although there was no significant association between TILs determined by hot-spot and OS for these patients (Supplementary Fig. 3a), TILs determined by the average method tended to show an association with OS ( p = 0.08, Supplementary Fig. 3b). Furthermore, by subtype, there was no association between TILs determined by the hot-spot or average methods and OS (Supplementary Fig. 3c–j). Looking carefully at the OS data, all patients with high or intermediate TILs by the average method survived (Supplementary Fig. 3b). Therefore, we re-evaluated the clinical outcomes (DFS and OS) between high/intermediate-TILs and low-TILs for both methods to assess the clinical significance of TILs (Fig. a–d). The high/intermediate-TIL group showed a trend toward better DFS ( p = 0.14, Fig. b) and significantly better OS ( p = 0.03, Fig. d) than the low-TIL group by the average method, although analysis using the hot-spot data did not show any significant difference either in DFS or OS (Fig. a, c). Next, we confirmed the association between pCR and clinical outcomes; breast cancer patients who achieved pCR had significantly better DFS ( p = 0.01; Fig. e) and tended to have better OS ( p = 0.10; Fig. f) than those who did not achieve pCR. We further hypothesized that combining pCR status with TIL scores might better predict clinical outcomes, and we investigated the relationship between the combined index of TILs/pCR and clinical outcomes (DFS and OS) (Fig. ). This additional analysis revealed that, by the average method, patients not achieving pCR with low-TILs had significantly poorer DFS and OS than other patients ( p = 0.03, p = 0.02, respectively; Fig. b, d), although analysis using the hot-spot data did not show any significant difference in either DFS or OS (Fig. a, c). Furthermore, when we analyzed OS by TILs in the patients with recurrence among those receiving NAC, all the patients in the high/intermediate-TILs group were alive, and deaths only occurred in the low-TILs group (Supplementary Fig. 4). In the 477 eligible surgical cases without NAC, we also examined the difference in clinical outcomes (DFS and OS) between the two groups of high/intermediate and low TILs by the hot-spot and average methods (Supplementary Fig. 5). Although TIL scores by both the hot-spot and average methods did not show any association with DFS, high/intermediate TILs showed significantly better OS than low TILs by the hot spot method ( p = 0.04, Supplementary Fig. 5c), and high/intermediate TILs tended to have better OS than low TILs by the average method ( p = 0.11, Supplementary Fig. 5d). In patients who received NAC, our data confirmed that the TIL score is related to pCR rate and OS. However, to assess the TIL score for each patient in clinical practice, a significant effort would be required for pathologists, which may limit its implementation. Therefore, we conducted an exploratory study to see if it was possible to predict the TIL score using only clinical parameters obtained in routine clinical practice, without the pathological assessment of TILs. To this end, we used 468 surgical cases to predict TIL scores by machine learning using clinicopathological findings; seventy percent of the cases were used as a training set and thirty percent were used as a test set. TIL scores evaluated by the hot-spot method were predicted using a random forest regressor, and the root mean squared error for the training set was 8.143, and that for the test set was 19.962 (Fig. a). Finally, TIL scores evaluated by the average method were predicted using a random forest regressor, and the root mean squared error for the training set was 4.664, and that for the test set was 10.955 (Fig. b). In this study, we demonstrated the usefulness of two different methods of TIL assessment, the hot-spot method and the average method, using biopsy specimens from breast cancer patients treated with NAC. We found that TIL scores obtained by both methods were associated with the pCR rate, suggesting that both methods will be clinically useful. Furthermore, our results suggest that the average method may have an advantage in this setting, as the TIL score by the average method is more consistent between biopsy and surgical specimens, and the TIL score by this method associated better with clinical outcomes compared with the hot-spot method. We also explored the possibility of predicting TIL scores from clinical information in this study and found that machine learning using clinicopathological data may better predict TIL scores assessed by the average method than those assessed by the hot-spot method. TIL scores obtained by both hot-spot and average methods were linearly related to pCR rates in our 144 breast cancer patients treated with NAC. The relationship between increased TIL scores and high pCR rates has been reported many times in previous studies, especially in HER2-positive subtypes and TNBC . Furthermore, in the current study, the combination of pCR and TIL score by the average method related well with clinical outcomes, and patients with low TILs not achieving a pCR were particularly prone to recurrence and death. In breast cancer patients, TILs correlate well, not only with pCR rates, but also with clinical outcomes . In the current study, breast cancer patients who achieved pCR had significantly better DFS and tended to have better OS than those who did not achieve pCR. This suggests that pCR is associated with whether or not the disease recurs. In contrast, TILs are better associated with OS than DFS (Fig. b, d), which may suggest that TILs are also associated with outcome after relapse. Indeed, when we looked at the data only for patients who relapsed after NAC, OS appeared to be better when TILs were high/intermediate (Supplementary Fig. 4), suggesting that the treatment efficacy after relapse may be influenced by TILs. Even in the analysis of the surgical cases without NAC, TILs correlated better with OS than with DFS (Supplementary Fig. 5), which may suggest that post-relapse outcomes differ by TILs. TILs have been reported to be associated with treatment outcome not only in patients receiving NAC, but also in patients with metastatic breast cancer . Taken together, it appears that TILs are not only a factor associated with relapse, but also a factor associated with treatment outcome, which we believe is one reason why TILs are better associated with OS than DFS. Our data suggest that TIL evaluation using the average method may have an advantage for patients treated with NAC, since the TIL score by the average method is more consistent between biopsy and surgical specimens, and the TIL score by this method relates better with clinical outcomes. Due to the nature of the hot-spot method, the larger the area to be evaluated, the greater the number of hot spots and consequently the higher the TIL score. Therefore, it is not surprising that in our results the TIL score by hot-spot was higher in surgical specimens than in biopsy specimens for patients undergoing surgery without preoperative treatment. The current study of patients who received preoperative chemotherapy confirmed the usefulness of the averaging method, given the high proportion of HER2-positive and TNBC patients with relatively high TILs. When studying hormone receptor-positive breast cancer with low overall TILs, the advantages of the hot-spot method should also be kept in mind, as this technique may be able to detect small differences in TILs, and thus assess the patient's condition . TILs can be clinically useful, but it may be practically difficult for pathologists to assess TIL scores in all patients with breast cancer. We explored the possibility of predicting the TIL score from clinical data obtained in routine daily clinical practice, such as blood and biochemical data and pathological findings, and the results suggested that machine learning using clinicopathological data may better predict TIL scores assessed by the average method than those assessed by the hot-spot method. Some researchers are of the opinion that it is difficult to infer TILs and the tumor microenvironment from accessible blood and other systemic data; others indicate it is possible to predict the tumor microenvironment by radiological data . Our results suggest the possibility of indirectly inferring the tumor microenvironment from the analysis of appropriate clinical data from routine pathological diagnosis and laboratory tests, which are widely available. The prediction model we used in this study is lightweight and can be run on ordinary computers, and prediction software based on such a simple model could be developed and widely used in medical facilities. If a more accurate TIL prediction model for clinical use is desired, perhaps the use of imaging data would be ideal, albeit requiring more powerful computers, and further technological advances would be needed to implement such a prediction model widely in society. In the future, it is conceivable that combining these data with imaging and other data may allow TILs to be assessed with a higher degree of accuracy without directly looking at the tumor tissue. This study had limitations; it was a single-center retrospective study, and the number of patients treated with NAC, in particular, was limited. Regarding machine learning, the predictive ability of the random forest method was demonstrated, but it is clear that further improvement in accuracy is needed before it can be used in clinical practice. However, by using a large number of surgical cases without NAC, the present study clarified the differences between the two TIL scoring methods in the evaluation of biopsy and surgical specimens and showed that both scoring methods are clinically beneficial. Furthermore, our results demonstrated the possibility of inferring the TIL score from clinical information which is of value for the future development of the field. In conclusion, TIL scores evaluated by two different methods, the average and hot-spot methods, were associated with pCR, suggesting that both are clinically useful as predictive markers for breast cancer patients treated with NAC. The average method may have an advantage for these patients, since the TIL score using the average method is more consistent between biopsy and surgical specimens, and the TIL score by this method relates better with clinical outcomes. Furthermore, our exploratory study shows that machine learning using clinicopathological data may better predict TIL scores assessed by the average method, suggesting the possibility of indirectly inferring the tumor microenvironment from the analysis of appropriate clinical datasets in the future. Below is the link to the electronic supplementary material. Supplementary file1 (PDF 84 KB) Supplementary file2 Supplementary Fig. 1 The percentages of patients with low (blue), intermediate (green) or high (red) tumor-infiltrating lymphocyte (TIL) scores in biopsy specimens using the hot-spot and average methods. Supplementary Fig. 2 Kaplan-Meier curves showing disease-free survival by pre-treatment tumor-infiltrating lymphocyte (TIL) scores by hot-spot and average methods. Kaplan-Meier curves showing disease-free survival by (a) TIL score (hot-spot) for all 144 patients; (c-f) TIL score (hot-spot) for each subtype; (b) TIL score (average) for all 144 patients; and (g-j) TIL score (average) for each subtype. Int intermediate, TNBC triple-negative breast cancer, HER2 human epidermal growth factor receptor 2. Supplementary Fig. 3 Kaplan-Meier curves showing overall survival by pre-treatment tumor-infiltrating lymphocyte (TIL) scores by hot-spot and average methods. Kaplan-Meier curves showing overall survival by (a) TIL score (hot-spot) for all 144 patients; (c-f) TIL score (hot-spot) for each subtype; (b) TIL score (average) for all 144 patients; and (g-j) TIL score (average) for each subtype. Int intermediate, TNBC triple-negative breast cancer, HER2 human epidermal growth factor receptor 2. Supplementary Fig. 4 Kaplan-Meier curves showing the impact of pre-treatment tumor-infiltrating lymphocytes (TILs) on the prognosis of patients with recurrence among those receiving neoadjuvant chemotherapy. Kaplan-Meier curves showing overall survival (OS) by TIL score (high/intermediate vs. low) using the average method. Supplementary Fig. 5 Kaplan-Meier curves showing the impact of tumor-infiltrating lymphocytes (TILs) on the prognosis of 477 surgical cases without neoadjuvant chemotherapy. Kaplan-Meier curves showing (a) disease-free survival (DFS) by TIL score (high/intermediate vs. low) using the hot-spot method; (b) DFS by TIL score (high/intermediate vs. low) using the average method; (c) overall survival (OS) by TIL score (high/intermediate vs. low) using the hot-spot method; and (d) OS by TIL score (high/intermediate vs. low) using the average method (PDF 117 KB) |
Trends and characteristics of fertility-sparing treatment for atypical endometrial hyperplasia and endometrial cancer in Japan: a survey by the Gynecologic Oncology Committee of the Japan Society of Obstetrics and Gynecology | 493da867-94e7-4fa0-9177-4efd34255356 | 10157339 | Gynaecology[mh] | Due to a tendency for delayed marriage, the age of pregnancy is delayed in Japan . This problem overlaps with the occurrence of gynecological cancer in the reproductive age group. Recently, the number of endometrial cancer (EC) patients younger than 40 years has been increasing . There are approximately 500 patients with EC younger than 40 years per year in Japan. Of them, 77% were stage IA . Using the National Cancer Database, Ruiz et al. reported that the proportion of endometrial cancer patients who were treated with progestin therapy increased from 2.4% in 2004 to 5.9% in 2014. The recommended fertility-sparing (FS) treatments in the National Comprehensive Cancer Network (NCCN) guidelines include hormone therapy (medroxyprogesterone [MPA] and megestrol acetate [MA]) and levonorgestrel-releasing intra-uterine devices (LNG-IUDs) . Japanese treatment guidelines for EC mention that FS treatment is a treatment option for young patients with atypical endometrial hyperplasia (AEH) and EC (endometrioid carcinoma grade 1 [ECG1] and lesion limited to the endometrium) . However, since a variety of FS treatment regimens have been widely adopted, the current trends in FS-treatment are relatively unknown. To elucidate current trends in FS treatment, a questionnaire-style survey regarding FS treatment was performed, in Japan Society of Obstetrics and Gynecology (JSOG) gynecological cancer registered institutions. In addition, this study was performed to identify factors correlated with the clinical response to FS treatment, disease recurrence, pregnancy outcome, and any deviations from the eligibility criteria by analyzing the detailed information of each patient, which were difficult to collect from meta-analysis. In our view, this is the largest-scale evaluation to date in a retrospective nationwide study of FS treatment for AEH and EC patients. 1. Study design and patients This study was conducted by the Committee on Gynecologic Oncology of JSOG in the 2017–2018 fiscal year. A nationwide, retrospective questionnaire style survey—as performed. The survey items included patient demographics (age, body mass index [BMI], complications, family history, desire to have children, etc.), examinations for diagnosis, pathological diagnosis, regimen of FS treatment, adverse events (AEs), presence of myometrial invasion [MI]), maintenance therapy (oral contraceptives/low dose estrogen progestin, estrogen+progestin, or progestin only), outcomes of initial and recurrent FS treatment, and pregnancy outcomes. AEs were assessed using the Common Terminology Criteria for Adverse Events (CTCAE; ver. 4.0; National Institutes of Health, Bethesda, MA, USA). The data of patients with AEH and EC receiving FS treatment between January 2009 and December 2013 were collected from JSOG gynecological cancer registered institutions. These institutions consisted of medical training institutions, cancer specialty hospitals, and local core hospitals. This study was approved by the Institutional Review Board (IRB) of Kurume University and JSOG (IRB registration No. 17310/ UMIN No. 000034254). The present study was conducted after obtaining approval from each IRB. 2. Statistical analysis The endpoint of this study was to examine the current trends in FS treatment for AEH and ECG1 patients in Japan. The secondary objective was to examine the associations of clinical characteristics with the pathological complete remission (CR) rate, recurrence-free survival (RFS), and pregnancy and live birth rates. RFS was measured from the end date of the initial FS treatment to the date recurrence was confirmed. Time to complete remission (TTCR) was measured from the day starting initial treatment to the day achieving CR. TTCR was classified into 2 groups (TTCR <6 and ≥6 months). Survival curves were calculated using the Kaplan-Meier method, and the curves were compared using the log-rank test. A Cox proportional hazards model and logistic regression analysis were used for multivariate analysis. Frequency distributions were compared using the χ 2 test, unless the expected frequency was <5, in which case, Fisher’s exact test was used. All statistical analyses were performed using JMP software (version 14; SAS Institute, Cary, NC, USA). A value of p<0.05 was considered significant. This study was conducted by the Committee on Gynecologic Oncology of JSOG in the 2017–2018 fiscal year. A nationwide, retrospective questionnaire style survey—as performed. The survey items included patient demographics (age, body mass index [BMI], complications, family history, desire to have children, etc.), examinations for diagnosis, pathological diagnosis, regimen of FS treatment, adverse events (AEs), presence of myometrial invasion [MI]), maintenance therapy (oral contraceptives/low dose estrogen progestin, estrogen+progestin, or progestin only), outcomes of initial and recurrent FS treatment, and pregnancy outcomes. AEs were assessed using the Common Terminology Criteria for Adverse Events (CTCAE; ver. 4.0; National Institutes of Health, Bethesda, MA, USA). The data of patients with AEH and EC receiving FS treatment between January 2009 and December 2013 were collected from JSOG gynecological cancer registered institutions. These institutions consisted of medical training institutions, cancer specialty hospitals, and local core hospitals. This study was approved by the Institutional Review Board (IRB) of Kurume University and JSOG (IRB registration No. 17310/ UMIN No. 000034254). The present study was conducted after obtaining approval from each IRB. The endpoint of this study was to examine the current trends in FS treatment for AEH and ECG1 patients in Japan. The secondary objective was to examine the associations of clinical characteristics with the pathological complete remission (CR) rate, recurrence-free survival (RFS), and pregnancy and live birth rates. RFS was measured from the end date of the initial FS treatment to the date recurrence was confirmed. Time to complete remission (TTCR) was measured from the day starting initial treatment to the day achieving CR. TTCR was classified into 2 groups (TTCR <6 and ≥6 months). Survival curves were calculated using the Kaplan-Meier method, and the curves were compared using the log-rank test. A Cox proportional hazards model and logistic regression analysis were used for multivariate analysis. Frequency distributions were compared using the χ 2 test, unless the expected frequency was <5, in which case, Fisher’s exact test was used. All statistical analyses were performed using JMP software (version 14; SAS Institute, Cary, NC, USA). A value of p<0.05 was considered significant. We collected the data of 413 patients from JSOG gynecological cancer registered institutions, consisting of medical training institutions (n=262, 63%), cancer specialty hospitals (n=58, 14%), and local core hospitals (n=93, 22%). A total of 102 institutions had eligible patients. Finally, the clinical information (103 questions/patient) of 413 patients was collected. 1. Patients’ characteristics The median follow-up time was 2,290 days. Patients’ median age and BMI were 35 years and 24.5 kg/m 2 , respectively. Most of the histological types were ECG1 (54.7%) and AEH (41.4%), although there were nine ECG2 patients. Major concomitant conditions were diabetes mellitus (DM) (9%), hypertension (8.7%), and polycystic ovarian syndrome (PCOS) (23.5%). We confirmed family history of cancer (Lynch syndrome suspected clinically) in 30 patients (7.3%). Twenty-six percent of patients had a history of infertility treatment . Thirty-six percent of patients had atypical genital bleeding at the first visit to the hospital. Fifty-two percent of patients had irregular menstrual cycles. 2. Initial treatment The examinations for pre-initial pathological diagnosis included dilatation & curettage (D&C) (80.6%), endometrial biopsy by hysteroscopy (11.9%), blind endometrial biopsy (8.2%), and endometrial cytology (0.2%). MPA was used in 98.8% for initial treatment. In 408 patients treated by MPA, 360 patients (87.2%) used MPA alone, and 48 patients (11.6%) were combined with metformin. The main dosages of MPA were 600 (79.4%) to 400 mg (18.9%). The dosages of metformin were varied from 750 to 2,500 mg . AEs were observed in 4.4% of patients. Grade 3 massive genital bleeding was observed in 2 patients (0.49%). One of the patients underwent hysterectomy to control genital bleeding and discontinued FS treatment. A total of 253 patients (61.3%) took low-dose aspirin during FS treatment to prevent thrombosis. Grade 3 thrombosis was observed in only 1 patient (0.24%), even though she had been taking low-dose aspirin. Body weight gain (20% more) was observed in three patients (0.73%). There were no grade 4 AEs. CR after initial treatment was achieved in 78.2% (323/413) of patients. To accurately determine the response to FS treatment in eligible patients, 360 patients who matched common FS treatment criteria of several guidelines (Pathology: AEH or ECG1 without MI, Treatment: MPA 400 or 600 mg) were selected. Of these 360 selected patients, CR was achieved in 79.1% (285/360). Each treatment response (MPA 400 mg, 600 mg, and MPA + metformin) of initial treatments was higher in AEH patients (78.4%, 83%, and 95.5%) than in ECG1 patients (65.4%, 76.2%, and 83.3%) . We performed univariate and multivariate analysis to examine the relationship between clinicopathological factors (age, BMI, PCOS, family history of cancer, DM, histology, treatment, and treatment period) and the clinical response after initial treatment of AEH and ECG1 patients . On univariate analysis, there were significant differences in BMI (<25 vs ≥25 kg/m 2 : p=0.028), treatment (MPA 400 mg vs. MPA + metformin; p=0.003), and treatment period (<6 vs. ≥6 months; p=0.002). There were no significant differences, but some trends were seen in histology (AEH vs. ECG1; p=0.061) and treatment (MPA 600 mg vs. MPA + metformin; p=0.088 and MPA 400 or 600 mg vs. MPA + metformin; p=0.057, respectively). On multivariate analysis, BMI ≥25 kg/m 2 (hazard ratio [HR]=2.24), ECG1 (HR=2.28) and treatment period <6 months (HR=2.5) were related to a poor response to initial FS treatment. 3. RFS in patients with AEH and ECG1 Univariate and multivariate analyses were performed with clinicopathological factors and RFS of AEH and ECG1 patients without MI treated with MPA 400 or 600 mg . On univariate analysis, there were significant differences in BMI (<25 vs. ≥25 kg/m 2 ; p=0.021), histology (AEH vs. ECG1; p=0.016), TTCR (<6 vs. ≥6 months; p=0.007), maintenance therapy (− vs. +; p<0.001), and pregnancy (− vs. +; p<0.001). On multivariate analysis, ECG1 (HR=1.73), TTCR ≥6 months (HR=1.51), maintenance therapy (−) (HR=2.1), and pregnancy (−) (HR=2.8) were associated with a significantly higher risk of recurrence on multivariate analysis in patients with AEH and ECG1 treated with MPA 400 or 600 mg. shows RFS curves by histology, TTCR, maintenance therapy, and pregnancy. The overall recurrence rate (ORR) was 39.5% in AEH, and 55.0% in ECG1. In terms of TTCR, ORR was 38.5% in TTCR <6 months and 55% in TTCR ≥6 months. In patients on maintenance therapy, ORR after achieving CR to initial treatment was lower (37.4%) than in patients not on maintenance therapy (57.6%). Furthermore, we found lower ORR in patients who could conceive than those who couldn’t. (29.3% vs. 57.3%). To examine the additional therapeutic effect of metformin, we reanalyzed the clinicopathological factors and RFS in patients treated with MPA + metformin or MPA (400 or 600 mg) . On multivariate analysis, there were significant differences in MPA 400 mg versus MPA + metformin (HR=0.33; p=0.026) and MPA 600 mg versus MPA + metformin (HR=0.36; p=0.021), in addition to histology, maintenance therapy, and pregnancy. However, there was no significant difference in TTCR (HR=1.38; p=0.093), which was significantly different in patients treated with MPA 400 or 600 mg. 4. The pathological discrepancies of patients who did not achieve CR after initial treatment Among patients who did not achieve CR after initial treatment, it was suspected that there were some pathological discrepancies between before-and after-initial treatment. Therefore, we examined pathological discrepancies between before-initial treatment (diagnosed by D&C or not D&C) and after initial treatment (diagnosed by hysterectomy). Surprisingly, the pathological discrepancies was 81.3% (13/16) in AEH, while lower in ECG1 (19.6%, 9/46) . The rate of diagnostic discrepancies in AEH (n=16) differed by the method of pathological examination (D&C or not D&C). The rate of diagnostic discrepancies of AEH was 75.0% (9/12; AEH: 3, ECG1: 9) with D&C, and 100.0% (4/4; ECG1: 1, ECG2: 2, ECG3: 1) without D&C (endometrial biopsy). Of note, among patients (AEH + ECG1) who did not achieve CR after initial treatment, 8.1% (5/62) had high-grade carcinoma (endometrioid carcinoma grade3 [ECG3]: 2, clear cell carcinoma: 1, dedifferentiated carcinoma: 2). 5. Cases of deviating from the eligibility criteria (patient with MI and ECG2) In this survey, there were 19 patients (4.6%, 19/413) who were suspected of having MI at pre-initial treatment by pelvic magnetic resonance imaging (MRI). CR rates were 42.1% (8/19) with MI and 74.6% (132/177) without MI. There was a significant difference in the CR rates between the 2 groups (relative risk=2.28; 95% confidence interval=1.4–3.6; p=0.005). The effectiveness of FS treatment in ECG2 was uncertain. In this survey, there were nine patients with ECG2. The CR rate after initial treatment was 88.9% (8/9), and the recurrence rate for those who achieved CR after initial treatment was 55.6% (5/9). 6. Pregnancy outcomes A total of 217 patients desired children after achieving CR following initial treatment. 76% of these patients underwent infertility treatment, whereas 24% did not. In patients with infertility treatment, the pregnancy rate was 58.5%, and the live birth rate was 50.6%. On the other hand, in patients who did not have infertility treatment, the pregnancy and live birth rate were only 11.5%/7.7%. There were significant differences in the pregnancy and live birth rates between infertility and no infertility treatment groups (p<0.010) . 7. Treatments for patients with recurrence Treatments for recurrent AEH and ECG1 patients without MI before initial treatment (n=126) who achieved CR after initial treatment were examined. Most recurrent sites were endometrium (98.4%, 124/126), with only 2 cases outside the uterus (1.6%, 2/126). The treatments for patients with recurrence were repeated FS treatment (42.9%, 54/126), surgery (40.5%, 51/126), repeated FS treatment to surgery (14.3%, 18/126), and unknown (2.4%, 3/126). Furthermore, we examined the efficacy of repeated FS treatment (MPA or MPA + metformin) for recurrent AEH and ECG1. The CR rates in recurrent AEH patients were 91.7% (22/24) with MPA, and 100% (6/6) with MPA + metformin. The CR rates of recurrent ECG1 patients were 90.9% (10/11) with MPA and 100% (1/1) with MPA + metformin. 8. Occurrence of ovarian cancer There were 15 cases of simultaneous ovarian cancer (3.6%) and one case of peritoneal cancer (0.24%). The timing of occurrence of ovarian cancer was before-FS treatment in 7.1%, during FS treatment in 20%, and after-FS treatment in 73.3%. Sixty-seven percent were diagnosed as primary ovarian cancer, whereas 13.3% were diagnosed as metastatic cancer from EMCA. Most pathology was endometrioid adenocarcinoma (85.7%) same as endometrial cancer . 9. Prognosis The prognosis of FS treatment was examined. The rates of patients with no evidence of disease (NED), alive with disease (AWD), and died of disease (DOD) were 95.6% (395/413), 2.7% (11/413), and 1.5% (6/413), respectively. The pathology before initial treatment of patients with DOD (n=6) was AEH in two cases and ECG1 in four cases. All of them were diagnosed by D&C before initial treatment. One patient (17%) was suspected of having MI before initial treatment. Two AEH patients (33.3%) achieved CR after initial treatment, although four ECG1 patients (66.7%) were PD. Five patients (83.3%) underwent surgery; all of their surgical pathology specimens showed high-grade carcinoma (carcinosarcoma: 2, ECG3: 1, clear cell carcinoma: 1, dedifferentiated carcinoma: 1) . The median follow-up time was 2,290 days. Patients’ median age and BMI were 35 years and 24.5 kg/m 2 , respectively. Most of the histological types were ECG1 (54.7%) and AEH (41.4%), although there were nine ECG2 patients. Major concomitant conditions were diabetes mellitus (DM) (9%), hypertension (8.7%), and polycystic ovarian syndrome (PCOS) (23.5%). We confirmed family history of cancer (Lynch syndrome suspected clinically) in 30 patients (7.3%). Twenty-six percent of patients had a history of infertility treatment . Thirty-six percent of patients had atypical genital bleeding at the first visit to the hospital. Fifty-two percent of patients had irregular menstrual cycles. The examinations for pre-initial pathological diagnosis included dilatation & curettage (D&C) (80.6%), endometrial biopsy by hysteroscopy (11.9%), blind endometrial biopsy (8.2%), and endometrial cytology (0.2%). MPA was used in 98.8% for initial treatment. In 408 patients treated by MPA, 360 patients (87.2%) used MPA alone, and 48 patients (11.6%) were combined with metformin. The main dosages of MPA were 600 (79.4%) to 400 mg (18.9%). The dosages of metformin were varied from 750 to 2,500 mg . AEs were observed in 4.4% of patients. Grade 3 massive genital bleeding was observed in 2 patients (0.49%). One of the patients underwent hysterectomy to control genital bleeding and discontinued FS treatment. A total of 253 patients (61.3%) took low-dose aspirin during FS treatment to prevent thrombosis. Grade 3 thrombosis was observed in only 1 patient (0.24%), even though she had been taking low-dose aspirin. Body weight gain (20% more) was observed in three patients (0.73%). There were no grade 4 AEs. CR after initial treatment was achieved in 78.2% (323/413) of patients. To accurately determine the response to FS treatment in eligible patients, 360 patients who matched common FS treatment criteria of several guidelines (Pathology: AEH or ECG1 without MI, Treatment: MPA 400 or 600 mg) were selected. Of these 360 selected patients, CR was achieved in 79.1% (285/360). Each treatment response (MPA 400 mg, 600 mg, and MPA + metformin) of initial treatments was higher in AEH patients (78.4%, 83%, and 95.5%) than in ECG1 patients (65.4%, 76.2%, and 83.3%) . We performed univariate and multivariate analysis to examine the relationship between clinicopathological factors (age, BMI, PCOS, family history of cancer, DM, histology, treatment, and treatment period) and the clinical response after initial treatment of AEH and ECG1 patients . On univariate analysis, there were significant differences in BMI (<25 vs ≥25 kg/m 2 : p=0.028), treatment (MPA 400 mg vs. MPA + metformin; p=0.003), and treatment period (<6 vs. ≥6 months; p=0.002). There were no significant differences, but some trends were seen in histology (AEH vs. ECG1; p=0.061) and treatment (MPA 600 mg vs. MPA + metformin; p=0.088 and MPA 400 or 600 mg vs. MPA + metformin; p=0.057, respectively). On multivariate analysis, BMI ≥25 kg/m 2 (hazard ratio [HR]=2.24), ECG1 (HR=2.28) and treatment period <6 months (HR=2.5) were related to a poor response to initial FS treatment. Univariate and multivariate analyses were performed with clinicopathological factors and RFS of AEH and ECG1 patients without MI treated with MPA 400 or 600 mg . On univariate analysis, there were significant differences in BMI (<25 vs. ≥25 kg/m 2 ; p=0.021), histology (AEH vs. ECG1; p=0.016), TTCR (<6 vs. ≥6 months; p=0.007), maintenance therapy (− vs. +; p<0.001), and pregnancy (− vs. +; p<0.001). On multivariate analysis, ECG1 (HR=1.73), TTCR ≥6 months (HR=1.51), maintenance therapy (−) (HR=2.1), and pregnancy (−) (HR=2.8) were associated with a significantly higher risk of recurrence on multivariate analysis in patients with AEH and ECG1 treated with MPA 400 or 600 mg. shows RFS curves by histology, TTCR, maintenance therapy, and pregnancy. The overall recurrence rate (ORR) was 39.5% in AEH, and 55.0% in ECG1. In terms of TTCR, ORR was 38.5% in TTCR <6 months and 55% in TTCR ≥6 months. In patients on maintenance therapy, ORR after achieving CR to initial treatment was lower (37.4%) than in patients not on maintenance therapy (57.6%). Furthermore, we found lower ORR in patients who could conceive than those who couldn’t. (29.3% vs. 57.3%). To examine the additional therapeutic effect of metformin, we reanalyzed the clinicopathological factors and RFS in patients treated with MPA + metformin or MPA (400 or 600 mg) . On multivariate analysis, there were significant differences in MPA 400 mg versus MPA + metformin (HR=0.33; p=0.026) and MPA 600 mg versus MPA + metformin (HR=0.36; p=0.021), in addition to histology, maintenance therapy, and pregnancy. However, there was no significant difference in TTCR (HR=1.38; p=0.093), which was significantly different in patients treated with MPA 400 or 600 mg. Among patients who did not achieve CR after initial treatment, it was suspected that there were some pathological discrepancies between before-and after-initial treatment. Therefore, we examined pathological discrepancies between before-initial treatment (diagnosed by D&C or not D&C) and after initial treatment (diagnosed by hysterectomy). Surprisingly, the pathological discrepancies was 81.3% (13/16) in AEH, while lower in ECG1 (19.6%, 9/46) . The rate of diagnostic discrepancies in AEH (n=16) differed by the method of pathological examination (D&C or not D&C). The rate of diagnostic discrepancies of AEH was 75.0% (9/12; AEH: 3, ECG1: 9) with D&C, and 100.0% (4/4; ECG1: 1, ECG2: 2, ECG3: 1) without D&C (endometrial biopsy). Of note, among patients (AEH + ECG1) who did not achieve CR after initial treatment, 8.1% (5/62) had high-grade carcinoma (endometrioid carcinoma grade3 [ECG3]: 2, clear cell carcinoma: 1, dedifferentiated carcinoma: 2). In this survey, there were 19 patients (4.6%, 19/413) who were suspected of having MI at pre-initial treatment by pelvic magnetic resonance imaging (MRI). CR rates were 42.1% (8/19) with MI and 74.6% (132/177) without MI. There was a significant difference in the CR rates between the 2 groups (relative risk=2.28; 95% confidence interval=1.4–3.6; p=0.005). The effectiveness of FS treatment in ECG2 was uncertain. In this survey, there were nine patients with ECG2. The CR rate after initial treatment was 88.9% (8/9), and the recurrence rate for those who achieved CR after initial treatment was 55.6% (5/9). A total of 217 patients desired children after achieving CR following initial treatment. 76% of these patients underwent infertility treatment, whereas 24% did not. In patients with infertility treatment, the pregnancy rate was 58.5%, and the live birth rate was 50.6%. On the other hand, in patients who did not have infertility treatment, the pregnancy and live birth rate were only 11.5%/7.7%. There were significant differences in the pregnancy and live birth rates between infertility and no infertility treatment groups (p<0.010) . Treatments for recurrent AEH and ECG1 patients without MI before initial treatment (n=126) who achieved CR after initial treatment were examined. Most recurrent sites were endometrium (98.4%, 124/126), with only 2 cases outside the uterus (1.6%, 2/126). The treatments for patients with recurrence were repeated FS treatment (42.9%, 54/126), surgery (40.5%, 51/126), repeated FS treatment to surgery (14.3%, 18/126), and unknown (2.4%, 3/126). Furthermore, we examined the efficacy of repeated FS treatment (MPA or MPA + metformin) for recurrent AEH and ECG1. The CR rates in recurrent AEH patients were 91.7% (22/24) with MPA, and 100% (6/6) with MPA + metformin. The CR rates of recurrent ECG1 patients were 90.9% (10/11) with MPA and 100% (1/1) with MPA + metformin. There were 15 cases of simultaneous ovarian cancer (3.6%) and one case of peritoneal cancer (0.24%). The timing of occurrence of ovarian cancer was before-FS treatment in 7.1%, during FS treatment in 20%, and after-FS treatment in 73.3%. Sixty-seven percent were diagnosed as primary ovarian cancer, whereas 13.3% were diagnosed as metastatic cancer from EMCA. Most pathology was endometrioid adenocarcinoma (85.7%) same as endometrial cancer . The prognosis of FS treatment was examined. The rates of patients with no evidence of disease (NED), alive with disease (AWD), and died of disease (DOD) were 95.6% (395/413), 2.7% (11/413), and 1.5% (6/413), respectively. The pathology before initial treatment of patients with DOD (n=6) was AEH in two cases and ECG1 in four cases. All of them were diagnosed by D&C before initial treatment. One patient (17%) was suspected of having MI before initial treatment. Two AEH patients (33.3%) achieved CR after initial treatment, although four ECG1 patients (66.7%) were PD. Five patients (83.3%) underwent surgery; all of their surgical pathology specimens showed high-grade carcinoma (carcinosarcoma: 2, ECG3: 1, clear cell carcinoma: 1, dedifferentiated carcinoma: 1) . This study mainly focused on examining the factors that correlated with the response to initial treatment and recurrent risk factors who achieved CR to initial treatment. Furthermore, we specifically examined the pregnancy and live birth rate with and without the introduction of fertility treatment, the concurrent occurrence of ovarian cancer, the histological discrepancy before and after treatment in AEH, and deviations from the eligibility criteria. In terms of the response to initial treatment, we confirmed almost the same remission rate as the previous reports . Li et al. reported that the histology and BMI were significantly associated with a higher likelihood of achieving CR, although age, PCOS, and hormonal agents did not affect CR. In our study, the significant clinicopathological factors correlated to CR were histology (AEH), BMI (<25 kg/m 2 ), and treatment period (≥6 months). There is no consensus on the optimal duration of the treatment period to date . Based on a previous prospective study in Japan in which the treatment period was set at 6 months, it is possible that a certain number of the stable disease and progressive disease cases also completed treatment at six months . Wang et al. examined the treatment period in 68 patients (1 AH and 37 ECG1) to see whether a more extended treatment period affects oncologic results, causing disease progression or recurrence. They found cumulative complete response rates of 59% (≤6 months), 76% (6–9 months), and 95.5% (>9 months), respectively. In the present study, we also confirmed that a more extended treatment period (≥6 months) was correlated with CR. Further investigation of optimal treatment periods is warranted. About recurrent risk factors who achieved CR to initial treatment, we confirmed that ECG1, TTCR ≥6 months, maintenance therapy (−), and pregnancy (−) were associated with a significantly higher recurrent risk on multivariate analysis in the patients. Even in responders, the rate of recurrence was high in EC . The results of four meta-analyses showed 31%–41% of ECG1 patients develop recurrence after the initial response, and the potential risk factors associated with recurrence were BMI ≥25 kg/m 2 , PCOS, and EC . Furthermore, TTCR, family history of cancer, diabetes, pregnancy, progestin type, and maintenance therapy (−) have also been reported in other studies as independent risk factors for recurrence . From this study, we could confirm almost identical recurrent risk factors as previous studies. We confirmed the benefit of maintenance therapy for RFS; 55% of patients without maintenance therapy recurred within 5 years, whereas only 35% of patients on maintenance therapy recurred. About the importance of maintenance therapy to prevent a recurrence, KGOG study also emphasized this point . Maintenance therapy should be recommended for patients who do not express the desire to have children at the time of CR after initial FS treatment. In this study, 48 (11.6%) patients used metformin combined with MPA as initial treatment. Metformin increases insulin sensitivity and activates the AMPK pathway, counteracting the PI3K/ mTOR/Akt/FOXO1 signaling pathway, promoting endometrial proliferation . We found a better response rate to initial treatment and reduced recurrent risk as previous reports . When we analyzed RFS, including with MPA + metformin, TTCR ≥6 months was not a risk factor for RFS, although it was a significant risk factor in patients treated with MPA only. The addition of metformin to MPA might decrease the recurrent risk associated with longer TTCR. Because a small number of patients were treated with metformin, we could not conclude the response of the additional effect of metformin from this study. We would know this exciting answer from the ongoing prospective, randomized study of MPA + metformin versus MPA only (FELICIA trial) . Regarding the pregnancy and live birth rate in the present study, those with infertility treatment (58.5%/50.6%) were better than those for spontaneous pregnancy (11.5%/7.7%). Not only with assisted reproductive technology (ART) but there was also a relatively high pregnancy rate with timing treatment and intrauterine insemination. In previous meta-analysis, the live birth rate of patients who had ART was 39.4%, whereas it was 14.9% in patients who tried to conceive spontaneously . In the present study, the first- and second-year recurrence rates were 25% and 36%, respectively, in ECG1. Therefore, the recommended timing of pregnancy is early after achieving CR to initial treatment. The implementation of in vitro fertilization techniques increases the chance of conception and may also decrease the time to conception . The American Society of Clinical Oncology recommends that physicians “should refer patients who express an interest in fertility preservation to reproductive specialists” . The rate of ovarian cancer during follow-up was reported to be 3.6% in a previous meta-analysis . However, the detailed information of the patients was poorly reported (i.e., primary or metastatic, histology, and response to initial treatment) because the original primary reports did not include these. In the present survey, there were 15 cases of simultaneous ovarian cancer (3.6%). Of these, 67% were diagnosed with primary ovarian cancer, whereas 13% were diagnosed with metastatic cancer from EMCA. The developments of ovarian cancer were mostly detected post-FS treatment (73.3%). In addition, most patients (80%) achieved CR to initial treatment, and the histology of ovarian cancer was endometrioid carcinoma (85.7%), which is the same as endometrial cancer. Two cases of the patients diagnosed with primary ovarian cancer had suspected Lynch syndrome (20%, 2/10). The details of the etiology of ovarian cancer are unknown because genetic testing was not performed in this study. In the future, genetic testing of cases with concurrent ovarian cancer may help to elucidate these factors. Throughout the FS treatment period, gynecologists need to be very careful about the simultaneous occurrence of ovarian cancer. If ovarian tumors are detected during and post FS treatment, patients should be carefully examined by pelvic enhanced MRI and whole-body enhanced CT to rule out or confirm ovarian cancer and metastasis. In this nationwide survey, approximately 10% of patients (patients with ECG2 or MI) did not match the eligibility of standard FS treatment criteria. Through this research study, it was found that these originally off-label cases are being treated. We found initial FS treatment efficacy was equivalent to ECG1 for ECG2 patients but lower for MI cases than those without MI. Although, these are only the results of a few instances, and adhering to the standard indications for FS treatment is still essential. We found that patients with AEH who failed the initial FS treatment had more histologic discrepancies. In particular, the pathological discrepancies between pre- and post-treatment histology were more common in cases where the pre-treatment pathological diagnosis was made without D&C. This histological discrepancy due to differences in testing methods is also reported in the KGOG study . In our study, 8.1% of the patients with initial FS treatment failure had high-grade carcinoma. It is important to perform D&C for histological confirmation pre- and post-treatment. Especially in initial FS treatment failure cases, we need to thoroughly check for hidden high-grade carcinoma. The limitation of the present study is that it was a retrospective, questionnaire-style survey. In addition, this study was performed in a single country. The optimal progestin regimen remains to be elucidated, although MPA and MPA + metformin were mainly analyzed in the present study. The strength of the present study is that it was a nationwide, multi-center survey, and likely, the largest sample size compared with previous reports. Compared with a meta-analysis, although the number of patients was smaller, it was possible to accurately examine potential predictors affecting response to initial therapy, significant risk factors related to recurrence, ovarian cancer patients, and pregnancy data for each assisted reproduction treatment obtained for each patient, which was collected from the same questionnaire data, which covered the patients’ detailed background characteristics. In summary, the present study provides further insight into the current trends of FS treatment in AEH and ECG1 patients who want to have children. The patients who choose this FS treatment should be informed of the relatively low live birth rate, the high chance of recurrence, and the possibility of the occurrence of ovarian cancer, which could be life-threatening. Before FS treatment, detailed pathological examination is needed to rule out high-grade histology by D&C, MI, and concurrent ovarian cancer by MRI. If CR is achieved, gynecologists need to consult with reproductive specialists about infertility treatment, including assisted reproductive treatments, to maximize chances of a live birth. The trends of FS treatment shown in this study are clinically meaningful and may influence the future direction of investigation. |
Phytochemistry and Biological Activities of | c8a8d96e-46e6-4edc-8617-e53466a07e36 | 9786185 | Pharmacology[mh] | The Meliaceae or mahogany family is distributed in tropical and subtropical regions such as Himalaya, South and Central America, Africa, as well as South and Southeast Asia. They consist of over 579 species and 51 genera with the major secondary metabolites being terpenoids and limonoids along with minor compounds such as flavonoids, lignans, chromones, and phenolics . The biological activities of the Meliaceae family include cytotoxic , antiviral , antiplasmodial , antioxidant , antimicrobial , antifeedant , and anti-inflammation . Guarea is one of the largest genera of the American Meliaceae family consisting of over 69 species widely distributed in Mexico and Argentina , while a few species are found in Africa . Initial chemical investigation which commenced in 1962 by Housley et al. isolated a limonoid compound, dihydrogedunin (221), from the ground heartwood of G. thompsonii (Nigerian pearwood). Subsequently, eight classes of secondary metabolites have been identified along with their biological activities, such as cytotoxic, anti-inflammation, antimalarial, antiparasitic, antiprotozoal, antiviral, antimicrobial, insecticidal, antioxidant, and phosphorylation inhibitor. This study was initiated with a literature search related to the Guarea genus, and all the synonym names were confirmed based on a plant database “ www.theplantlist.org (accessed on 28 August 2022)”. Articles related to the biological and phytochemical properties between 1962 and 2022 were collected from the primary literature research through Scifinder ( n = 170), PubMed ( n = 8), Google Scholar ( n = 131), Mendeley ( n = 20), and Scopus ( n = 11) databases and after removing duplicates ( n = 247), 93 records were identified for title and abstract revision . Therefore, at the end of the selection process, 61 articles were screened and 32 articles were included in the systematic review . Guarea belongs to the Meliaceae family which is widely distributed in America and Africa. The diameter of this genus is one meter and its tree is usually 20–45m-tall while the characteristics include leave-pinnate, generative reproduction, and 2–8-valved loculicidal fruit. Its staminal tube is 0.4–1.3 cm in length, and the seeds are often shaped like the segment of an orange, with a fleshy, sometimes vascularized, or mealy sarcotesta, and usually thickened on the adaxial surface . 3.1. Overview of Isolated Compounds Derived from Guarea Genus About 240 compounds have been isolated from the stembark, leaves, fruits, bark, seed, flowering branches, and root of this genus, based on the literature from 1962 to 2022 as shown in . The extract for the isolation process was obtained from various solvents such as n-hexane, chloroform, methanol and n-butanol. The first step of the process is the maceration of the dried sample with solvent, especially methanol or ethanol; after that, MeOH/EtOH extract is diluted with water and partitioned with other solvents for obtaining crude extract. Meanwhile, between the hydrodistillation and isolation process is different. The hydrodistillation process used a fresh sample (part of Guarea ) and submitted to a Clevenger-type apparatus for 4 h for the gained crude oil. The crude extract and crude oil were purified with various techniques such as column chromatography on silica gel or RP-18 silica gel, Sephadex LH-20, preparative TLC, and semipreparative HPLC on RP-18 column for crude extract. The compounds were identified by NMR, mass spectrometry, FTIR, UV, and polarimeter. Moreover, the crude oil was analyzed using a combination of the four techniques of GC, GC/MS, 1 H-, and I3 C-NMR. The compounds identified from the isolation and hydrodistillation processes included 52% sesquiterpenoid, 16% diterpenoid, 15% Triterpenoid, 10% limonoid, as well as 7% non-terpenoid and limonoid. The distribution of the compounds is presented in and the biological activities of the identified compounds are shown in . 3.2. Sesquiterpenoid About 126 sesquiterpenoids have been isolated from the extract and essential oil since 1995 from Guarea guidonia , G. kunthiana , G. thompsonii , G. cedrata , G. macrophylla , G. scabra , G. convergens , and G. sylvatica . They include eudesmane, aromadendrane, guaian, caryophyllene, cadinene derivative, opposite, humulene, germacrene, bicyclogermacrene, cadinene, elemene, bisabolene, longifolene, farnasene, cyclosativene, himachalene, isolongifolane, acorenol, hinesol, cedrane, bourbonene, bergamotene, santalene, drimane, mustakone, and eremophilane as indicated in . Cadinene is a significant sesquiterpenoid from the Guarea genus with twenty-eight compounds. Menut et al. reported that the hydrodistillation of essential oil from G. cedrata bark produced four compounds of cadinene-type, namely γ-muurolene ( 52 ), cadina-1,4-diene ( 58 ), τ-cadinol ( 61 ), and α-muurolene ( 67 ). Moreover, the essential oil of G. macrophylla has been reported as cadinene-type. About twenty-four compounds were also obtained from leaves, fruits, and stem bark essential oil. Lago and Roque discovered two cadinene types, γ-cadinene ( 37 ) and δ-cadinene ( 38 ), isolated from the leaf essential oil of G. macrophylla. In the same year, Lago et al. also obtained cadinene type from the stem bark essential oil of G. macrophylla including ( 38 ), cis -calamenene ( 44 ), cis -cubenol ( 46 ), and trans -cubenol ( 47 ). The other seven compounds isolated from the hydrodistillation of G. macrophylla fruits include cadina-1(6),4-diene ( 54 ), β-cadinene ( 57 ), 1- epi -cubenol ( 60 ), τ-cadinol ( 61 ), τ-muurolol ( 62 ), α-cadinol ( 63 ) with four previous cadinene-type compounds. Furthermore, α-cadinene ( 64 ) and 1-cubenol ( 65 ) were isolated from the leaf essential oil . Ribeiro et al. also discovered γ-amorphene ( 83 ) with four previous cadinene type compounds such as ( 37 ), ( 38 ), ( 52 ), ( 67 ) in 2006. A total of seven other compounds were also obtained from these species such as α-amorphene ( 93 ), trans-muurola-4(14),5-diene ( 94 ), δ-amorphene ( 96 ), α-calacorene ( 97 ), β-calacorene ( 101 ), 1,10-di-epi-cubenol ( 103 ), and cis-cadin-4-en-7-ol ( 106 ) from the leaf essential oil . Núñez and Roque obtained cadinene from stem bark essential oil and other species of G. guidonia . The compounds isolated were trans-4,10(14)-cadinadiene ( 89 ), ( 52 ), and ( 38 ). Six years later, Nunez et al. identified α-muurolol ( 71 ), ( 52 ), ( 37 ), and ( 38 ) from the branch essential oil. One compound from the leaf essential oil of G. scabra was epi-α-cadinol ( 123 ) , and two compounds were isolated from the leaves of G. kunthiana calamenene ( 78 ) and cadalene ( 80 ) . Eudesmane is the second largest sesquiterpenoid from Guarea after the cadinene type with 22 compounds from the hydrodistillation and isolated process. α-eudesmol ( 69 ) was isolated from the bark essential oil of G. cedrata , and the first eudesmane type was reported from this genus . Garcez et al. reported one eudesmane from the wood bark of G. guidonia , namely, voleneol ( 13 ). β-selinene ( 1 ) was also reported in the leaves and essential oil of G. guidonia . Furthermore, several compounds were isolated from the leaves such as eudesm-5,7-dien ( 3 ), eudesm-4,11-diene ( 4 ), 5α,6α-epoxy-eudesm-7-ene ( 5 ), eudesm-6-en-4β-ol ( 6 ), 5α,6α-epoxy-eudesm-7-en-9-ol ( 9 ), 5α,6α,7α,8α-diepoxy-eudesmane ( 10 ), and (2 S *)-eudesm-5,7-dien-2-ol ( 19 ) . About five eudesmane compounds were isolated from the seeds of G. guidonia , including 6α-ethoxyeudesm-4(15)-en-1β-ol ( 21 ), eudesm-4(15)-ene-1β,6α-diol ( 23 ), 5-epi-eudesm-4(15)-ene-1β,6β-diol ( 24 ), eudesm-4(15)-ene-1β,5α-diol ( 25 ), and eudesm-4(15),7-dien-1β-ol ( 26 ) . In addition, 5,6,7,8-diepoxy-eudesmane ( 53 ) and eudesm-5,7-dien-2α-ol ( 8 ) were obtained from leaf essential oil . Ribeiro et al. isolated γ-eudesmol ( 85 ) from the leaf essential oil of G. macrophylla , while Oliveira et al. reported two compounds, namely selina-3,7(11)-diene ( 98 ) and 7-epi-α-eudesmol ( 110 ). Two eudesmane types, α-selinene ( 118 ) and β-eudesmol ( 124 ), were also isolated from branch essential oil of G. convergens and G. silvatica . Furthermore, aromadendrane types such as allo-aromadendrene ( 34 ), viridiflorene ( 42 ), globulol ( 45 ), and epi-globulol ( 59 ) were obtained from the bark essential oil of G. cedrata . Other species, such as G. macrophylla , G. guidonia , G. kunthiana , were found to also contain similar compounds. Spathulenol ( 2 ) and palustrol ( 17 ) were first isolated from the leaves of G. macrophylla while essential oil from the leaves and the stem bark were also reported to contain aromadendrane type. Lago et al. isolated ledol ( 18 ), and α-gurjunene ( 31 ) from the leaves and aromadendrene ( 40 ) from stem bark essential oil . Seven years later, alloaromadendrane-4α,10β-diol ( 88 ) was isolated from the bark . Two aromadendrane types, viridiflorol ( 11 ) and 3-oxo-10-alloaromadendranol ( 12 ), were also obtained from the wood bark of G. guidonia , -4β,10α-aromadendranediol ( 16 ) from the leaves of G. kunthiana , and β-gurjunene ( 115 ) from G. scabra . Furthermore, guai-6-en-10β-ol ( 7 ) was the first guaian type isolated from the leaves of G. macrophylla . Compounds such as cis- β -guaiene ( 55 ), 6,9-guaiadiene ( 91 ), trans- β -guaiene ( 95 ), and guaiol ( 102 ) were isolated from the fruit and leaf essential oil . G. kunthiana also has a guaian type, while alismol ( 14 ) and alismoxide ( 15 ) were identified from the leaves . In addition, α-guaiene ( 51 ) was obtained from the leaf essential oil of G. guidonia . Caryophyllene oxide ( 20 ) and β-caryophyllene ( 33 ) were identified from the bark essential oil of G. cedrata . Núñez and Roque reported isocaryophyllene oxide ( 70 ) from the stem bark essential oil of G. guidonia . Meanwhile, two other species, G. kunthiana and G. macrophylla , were found to contain E-caryophyllene ( 73 ) and 9-epi-β-caryophyllene ( 82 ) . Magalhães et al. also reported two compounds, cis -caryophyllene ( 112 ) and caryophyllene epoxide ( 120 ), from the leaf essential oil of G. scabra and branches of G. humatensis . The derivative compounds from the cadinene type, such as α-cubebene ( 28 ) and β-copaene ( 81 ), were obtained from the leaf and stem bark essential oil of G. macrophylla . Furthermore, α-ylangene ( 29 ) and α-copaene ( 30 ) were first identified from the bark essential oil of G. cedrata , while G. guidonia was found to contain β-cubebene ( 50 ) . The α-humulene ( 32 ) and 6,7-epoxy-2,9-humuladiene ( 72 ) humulene type were identified from the stem bark essential oil of G. guidonia . Furthermore, 1(10)-epoxy-4,7-humuladiene ( 86 ) and 1(10),4-diepoxy-7-humulene were also obtained from the bark ( 87 ) . The latest discovery was performed by Magalhães et al. , where one humulene-type sesquiterpenoid humulene epoxide II ( 122 ) was identified from the branch essential oil of G. silvatica . Nunez and Roque identified germacrene D ( 35 ) from the stem bark essential oil of G. guidonia , while the G. macrophylla species was found to contain germacrene-D-4-ol ( 39 ), germacrene A ( 84 ), and germacrene B ( 100 ) in the leaf essential oil . Moreover, bicyclogermacrene type was also identified from the leaf and stem bark essential oil of G. macrophylla including bicyclogermacrene ( 36 ), cis-bicyclogermacradiene ( 41 ), and trans -bicyclogermacradiene ( 43 ) . The bark essential oil from G. cedrata was reported to contain elemene-type sesquiterpenoid γ-elemene ( 68 ) . β-elemene ( 49 ) was also isolated from the stem bark essential oil of G. guidonia . In 2005, δ-elemene ( 48 ) was reported in the leaf essential oil of this species , while elemol ( 99 ) was identified in the leaf essential oil of G. macrophylla . Eight compounds with bisabolene-type sesquiterpenoids were obtained from four species, namely G. macrophylla , G. kunthiana , G. sylvatica , and G. scabra . β-bisabolene ( 56 ) was obtained from the fruit essential oil of G. macrophylla . Magalhães et al. also identified three compounds, namely (E)-iso-γ-bisabolene ( 119 ) from the branch essential oil of G. silvatica , as well as α- cis -bergamotene ( 113 ) and α- trans -bergamotene ( 126 ) from the leaf essential oil of G. scabra . Eight years later, α-bergamotene ( 74 ), α-curcumene ( 76 ), α-zingiberene ( 77 ), and β-sesquiphellandrene ( 79 ) were isolated from the leaf essential oil of G. kunthiana . Furthermore, minor-type sesquiterpenoids were obtained from this genus, such as two compounds of opposite-type sesquiterpenoid (7 R *)-5-epi-opposit-4(15)-ene-1β,7-diol ( 22 ) and (7 R *)-opposit-4(15)-ene-1β,7-diol ( 27 ) from the seeds of G. guidonia , while longifolene ( 66 ) was isolated from the bark essential oil of G. cedrata . Two compounds of acyclic sesquiterpenoids, β-farnesene ( 75 ) and trans -nerolidol ( 121 ), were identified from the leaf essential oil of G. kunthiana and G. scabra . Moreover, cyclosativene ( 90 ), γ-himachalene ( 92 ), isolongifolan-7-α-ol ( 104 ), α-acorenol ( 105 ), hinesol ( 107 ), cedr-8(15)-en-9α-ol ( 108 ), and valerianol ( 109 ) were isolated from the leaf essential oil of G. macrophylla . Magalhães et al. also reported five other compounds, such as β-bourbonene ( 111 ) from the leaf essential oil of G. scabra ; α-santalene ( 114 ), β-santalene ( 116 ), drima-7,9(11)-diene ( 117 ) from the branches of G. convergens; and mustakone ( 125 ) from G. silvatica . All the sesquiterpenoid structures are shown in . 3.3. Diterpenoid Diterpenoid of 16% was isolated from the Guarea genus with two major types, isopimarane and labdane. One of the diterpenoid types which was first reported by Lago et al. was isopimarane from the leaves of G. macrophylla with three types, namely isopimara-7,15-dien-3-one ( 150 ), isopimara-7,15-dien-3β-ol ( 132 ), and isopimara-7,15-dien-2β-ol ( 151 ). Afterward, five diterpenoids, namely, 7α-hydroperoxy-isopimara-8(14),15-diene-2α,3β-diol ( 148 ), 19-nor-isopimara-7,15,4(18)-trien-3-one ( 149 ), isopimara-7,15-dien-2α-ol ( 152 ), isopimara-7,15-diene ( 158 ), and isopimara-7,15-diene-2α,3β-diol ( 131 ), were isolated and identified from the leaf essential oil of Guarea macrophylla from . Four types of labdane diterpenoids, namely, 3-oxo-labd-8(17),12Z,14-triene ( 133 ), 3α-hydroxylabd-8(17),12Z,14-triene ( 134 ), 3β-hydroxylabd-8(17),12Z,14-triene ( 135 ), and 19-hydroxymanoyloxide ( 135 )—identified from the leaves of G. trichilioides —were reported in 1996 by Furlan et al. . Furthermore, three labdane-type compounds such as manoyl oxide ( 153 ), labda-8,14-dien-13-ol ( 154 ), and labda-8,13-(E)-dien-15-ol ( 159 ), were isolated from the leaves of G. macrophylla , while ent -13-epimanoyloxide ( 147 ) was obtained from the leaves of G. kunthiana . Cneorubin A ( 111 ), B ( 112 ), X ( 113 ), and Y ( 114 ) were isolated from the leaves and the aerial parts of G. guidonia , while three kaurene types of diterpenoid compounds, ent -kaur-16-en-2-one ( 139 ), ent -kaur-16-ene ( 140 ), and ent -3β- and 3α-hydroxykaur-16-ene ( 141 and 142 ), were obtained from the leaves of G. kunthiana . Additionally, Magalhães et al. identified kaurene ( 164 ) from the leaf essential oil of G. sylvatica . Diterpenoids of the sandaracopimaradeine type were identified in the leaves of G. rhophalocarpa . The compounds were ent -8(14),15-sandaracopimaradiene-2α,18-diol ( 156 ), and ent -8(14),15-sandaracopimaradine-2β,18-diol ( 157 ) . Eighteen years later, sandaracopimarinal ( 163 ) was identified from the leaf essential oil of G. macrophylla . Furthermore, two diterpenoids of the clerodane type, -2-oxo-13-hydroxy,3,14-clerodandiene ( 136 ) and 13-hydroxy-3,14-clerodandiene ( 138 ), were obtained from the leaves of G. trichilioides . An investigation to identify three other compounds, including kolavelool ( 143 ), kolavenol ( 144 ), and kolavenal ( 145 ) from the leaves of G. kunthiana , was conducted by Garcez et al. . The acyclic type, phytol ( 155 ), was identified from the leaves of G. macrophylla and G. guidonia . Garcez et al. isolated -nephthenol ( 146 ) from the leaves of G. kunthiana , while one prenylaromadendrane-type boscartol C ( 160 ) was obtained from the aerial parts of G. guidonia . One of the dolabradiene types, 13-epi-dolabradiene ( 145 ), was identified from the leaf essential oil of G. macrophylla , along with phyllocladane ( 146 ) . The diterpenoid structures are presented in detail in . 3.4. Triterpenoid Thirty-five compounds were identified as triterpenoids, such as tirucallane, protolimonoid, lanostane, cycloartane, glabretal, glabretal derivatives, and apotirucallane . Cycloartane was the major triterpenoid type isolated from the Guarea genus. In 1993, seven compounds (cycloart-24-en-3,23-dione ( 173 ), 23-hydroxycycloart-24-en-3-one (epimers) ( 174 and 175 ), 3β-hydroxycycloart-24-en-23-one ( 176 ), 25-hydroxycycloart-23-en-3-one ( 177 ), 3β-21-dihydroxycycloartane ( 178 ), and 3β,21,22,23-tetrahydroxycycloartane-24(31), 25-diene ( 179 )) were identified from the leaves of G. trichilioides . Furthermore, 22,25-dihydroxycycloart-23 E -en-3-one ( 196 ), 24-methylenecycloartane-3β,22-diol ( 197 ), and cycloarta-23,25-dien-3-one ( 192 ) were obtained from the leaves of G. macrophylla , while two cyloartanes, namely (23 S *)-cycloart-24-ene-3β,23-diol ( 193 ) and (23 R *)-cycloart-24-ene-3β,23-diol ( 194 ), were isolated from the leaves of G. guidonia . In the same year, cycloart-23 E -ene-3β,25-diol ( 170 ) was discovered in the leaves of G. macrophylla , while in 2017, Conserva et al. obtained (23 S *,24 S *)-dihydroxycicloart-25-en-3-one ( 171 ). Two lanostane-type compounds, 23-hydroxy-5α-lanosta 7,9(11),24-triene-3-one ( 168 ) and 5α-lanosta-7,9(11),24-triene-3α,23-diol ( 169 ), were obtained from the leaves of Guarea rhophalocarpa , while glabretal ( 172 ) was identified from heartwood of G. glabra . Furthermore, 21,24-epoxy-3α,7α,21,23-tetraacetoxy-25-hydroxy-4α,4β,8β-trimethyl-14,18-cyclo-5α,13α,14α,17α-cholestane ( 181 ), and 21,23-epoxy-3α,7α,21,24,25-pentaacetoxy-4α, 4β,8β-trimethyl-14,18-cyclo-5α,13α,14α,17α-cholestane ( 182 ) as glabretal derivatives were identified from the leaves and twigs of G. jamicensis . The 3,4-secotirucalla-4(28),7,24-trien-3,21-dioic acid ( 165 ) and 3,4-secotirucalla-4(28),7,24-trien-3,21-dioic acid 3-methyl ester ( 166 ) as tirucallane types of triterpenoid were reported by Akinniyi et al. from the bark of G. cedrata . Furthermore, four tirucallane types, guareolide ( 186 ), guareoic acid A ( 187 ) and B ( 188 ), flindissone ( 189 ), as well as picroquassin E ( 190 ), were isolated from the aerial parts of G. guidonia . Jimenez et al. reported that three protolimonoid types, melianone ( 184 ), melianodiol ( 185 ), and 21-α-acetylmelianone ( 191 ), were first isolated from the seeds of G. grandiflora . In 2015, four compounds of this type were also identified, including 3β- O -tigloylmelianol ( 167 ), 3β- O -tigloylmeliantriol ( 198 ), and melianol ( 199 ), from the fruits of G. kunthiana . Moreover, 24-acetoxy-25-hydroxy-3,7-dioxoapotirucalla-14-en-21,23-olide ( 182 ) and 7α,24,25-trihydroxy-3-oxoapotirucalla-14-en-21,23-olide ( 183 ) as apotirucallane types were isolated from the leaves and branches of G. convergens . 3.5. Limonoid Limonoids are classified into many classes based on the type of skeleton , and about eleven classes have been reported from this genus. The first exploration by Housley et al. reported dihydrogedunin ( 221 ) from the heartwood of G. thompsonii . Connollyl et al. also found one andirobine-type limonoid, namely methyl 6-acetoxyangolensate ( 206 ), identified from the bark of G. thompsonii and methyl angolensate ( 214 ) from the fruits of G. kunthiana . Moreover, one of limonoid types which was called with dregeanin ( 207 ) was obtained from the bark of G. thompsonii , and rohituka-type named with 2’-hydroxyrohitukin ( 215 ) was identified from the bark of G. cedrata . The obakunol-type limonoid, 7-acetyldihydronomilin ( 216 ), was isolated from the aerial parts of G. guidonia , and the ecuadorin ( 217 ) which was one of the ecuadorin-types, was found in the aerial parts of G. kunthiana . Prieurianin ( 219 ) and 14,15β-epoxyprieuriani ( 210 ) were found in the root bark of G. guidonia as a prieurianin-type limonoid . Garcez et al. also reported mombasol ( 208 ) from the bark of G. guidonia and the investigation by Lukacova et al. obtained 7-oxo-gedunin ( 218 ) from the root bark, while three gedunin limonoids, 7-deacetoxy-7-oxogedunin ( 200 ), gedunin ( 201 ), and 6α-acetoxygedunin ( 209 ), were isolated from the seeds of G. grandiflora . Zelnik and Rosito discovered one mexicanolide type, called fissinolide ( 220 ), in the seeds of G. trichilioides . Five years later, the seeds were found to also contain angustinolide ( 224 ) . Humilinolide E ( 211 ), methyl 2-hydroxy-3b-tigloyloxy-1-oxomeliac-8(30)-enate ( 212 ), and swietenine acetate ( 213 ) were isolated from the fruits of G. kunthiana . Furthermore, an investigation by Bellone et al. identified 3-(2′-hydroxyisovaleroyl) khasenegasin I ( 205 ) from the stem bark of G. guidonia . The twigs of G. mayombensis produced azadirachtin-type mayombensin ( 222 ) and azadirachtin I ( 223 ) . Meanwhile, three compounds of A2, B, D-seco skeletons such as chisomicine D ( 202 ), chisomicine E ( 203 ), and chisomicine F ( 204 ), were identified from the stem bark of G. guidonia . 3.6. Steroid Ergostane- and pregnane-type steroids were isolated from the Guarea genus, along with general steroid compounds such as β-sitosterol ( 229 ), stigmasterol ( 230 ), and β-sitostenone ( 233 ) . Furthermore, the steroids glycoside stigmasterol glucoside ( 231 ) and β-sitosterol glucoside ( 232 ) were obtained from the twigs of G. mayombensis , while two ergostane-type steroids, ergosta-5,24(24′)-diene-3β,7α,21-triol ( 236 ) and ergosta-5,24(24′)-diene-3β,4β,22S-triol ( 237 ), were identified from the leaves and branches of G. convergens . Garcez et al. also reported two pregnane-type steroids, 2α,3β-dihydroxy-16,17-seco-pregn-17-ene-16-oic acid methyl ester 2β,19-hemiketal ( 234 ) and 2,3:16,17-di-seco-pregn-17-ene-3-oic acid-16-oic acid methyl ester-19-hydroxy-2-carboxylic acid-2,19-lactone ( 235 ), from the trunk bark of G. guidonia . 3.7. Other Compounds Flavonoid, lignan, ceramide, and coumarin were also identified from this plant genus. Quercetin 3- O -β- d -glucopyranoside ( 225 ), quercetin 3- O -β- d -galactopyranoside ( 226 ), and kaempferol 7- O -β- d -glucopyranoside ( 227 ) as glucoside flavonoids were isolated from the flowering branches of G. macrophylla . Furthermore, one neolignane compound, dehydrodiconiferyl alcohol-4-β- d -glucoside ( 228 ), was reported from the same part of this species . Two ceramides, ceramide A ( 238 ) and B ( 239 ), were obtained from the twigs of G. mayombensis , while one coumarin, scopoletin, ( 240 ) was found in the leaves of G. rhopalocarpa . About 240 compounds have been isolated from the stembark, leaves, fruits, bark, seed, flowering branches, and root of this genus, based on the literature from 1962 to 2022 as shown in . The extract for the isolation process was obtained from various solvents such as n-hexane, chloroform, methanol and n-butanol. The first step of the process is the maceration of the dried sample with solvent, especially methanol or ethanol; after that, MeOH/EtOH extract is diluted with water and partitioned with other solvents for obtaining crude extract. Meanwhile, between the hydrodistillation and isolation process is different. The hydrodistillation process used a fresh sample (part of Guarea ) and submitted to a Clevenger-type apparatus for 4 h for the gained crude oil. The crude extract and crude oil were purified with various techniques such as column chromatography on silica gel or RP-18 silica gel, Sephadex LH-20, preparative TLC, and semipreparative HPLC on RP-18 column for crude extract. The compounds were identified by NMR, mass spectrometry, FTIR, UV, and polarimeter. Moreover, the crude oil was analyzed using a combination of the four techniques of GC, GC/MS, 1 H-, and I3 C-NMR. The compounds identified from the isolation and hydrodistillation processes included 52% sesquiterpenoid, 16% diterpenoid, 15% Triterpenoid, 10% limonoid, as well as 7% non-terpenoid and limonoid. The distribution of the compounds is presented in and the biological activities of the identified compounds are shown in . About 126 sesquiterpenoids have been isolated from the extract and essential oil since 1995 from Guarea guidonia , G. kunthiana , G. thompsonii , G. cedrata , G. macrophylla , G. scabra , G. convergens , and G. sylvatica . They include eudesmane, aromadendrane, guaian, caryophyllene, cadinene derivative, opposite, humulene, germacrene, bicyclogermacrene, cadinene, elemene, bisabolene, longifolene, farnasene, cyclosativene, himachalene, isolongifolane, acorenol, hinesol, cedrane, bourbonene, bergamotene, santalene, drimane, mustakone, and eremophilane as indicated in . Cadinene is a significant sesquiterpenoid from the Guarea genus with twenty-eight compounds. Menut et al. reported that the hydrodistillation of essential oil from G. cedrata bark produced four compounds of cadinene-type, namely γ-muurolene ( 52 ), cadina-1,4-diene ( 58 ), τ-cadinol ( 61 ), and α-muurolene ( 67 ). Moreover, the essential oil of G. macrophylla has been reported as cadinene-type. About twenty-four compounds were also obtained from leaves, fruits, and stem bark essential oil. Lago and Roque discovered two cadinene types, γ-cadinene ( 37 ) and δ-cadinene ( 38 ), isolated from the leaf essential oil of G. macrophylla. In the same year, Lago et al. also obtained cadinene type from the stem bark essential oil of G. macrophylla including ( 38 ), cis -calamenene ( 44 ), cis -cubenol ( 46 ), and trans -cubenol ( 47 ). The other seven compounds isolated from the hydrodistillation of G. macrophylla fruits include cadina-1(6),4-diene ( 54 ), β-cadinene ( 57 ), 1- epi -cubenol ( 60 ), τ-cadinol ( 61 ), τ-muurolol ( 62 ), α-cadinol ( 63 ) with four previous cadinene-type compounds. Furthermore, α-cadinene ( 64 ) and 1-cubenol ( 65 ) were isolated from the leaf essential oil . Ribeiro et al. also discovered γ-amorphene ( 83 ) with four previous cadinene type compounds such as ( 37 ), ( 38 ), ( 52 ), ( 67 ) in 2006. A total of seven other compounds were also obtained from these species such as α-amorphene ( 93 ), trans-muurola-4(14),5-diene ( 94 ), δ-amorphene ( 96 ), α-calacorene ( 97 ), β-calacorene ( 101 ), 1,10-di-epi-cubenol ( 103 ), and cis-cadin-4-en-7-ol ( 106 ) from the leaf essential oil . Núñez and Roque obtained cadinene from stem bark essential oil and other species of G. guidonia . The compounds isolated were trans-4,10(14)-cadinadiene ( 89 ), ( 52 ), and ( 38 ). Six years later, Nunez et al. identified α-muurolol ( 71 ), ( 52 ), ( 37 ), and ( 38 ) from the branch essential oil. One compound from the leaf essential oil of G. scabra was epi-α-cadinol ( 123 ) , and two compounds were isolated from the leaves of G. kunthiana calamenene ( 78 ) and cadalene ( 80 ) . Eudesmane is the second largest sesquiterpenoid from Guarea after the cadinene type with 22 compounds from the hydrodistillation and isolated process. α-eudesmol ( 69 ) was isolated from the bark essential oil of G. cedrata , and the first eudesmane type was reported from this genus . Garcez et al. reported one eudesmane from the wood bark of G. guidonia , namely, voleneol ( 13 ). β-selinene ( 1 ) was also reported in the leaves and essential oil of G. guidonia . Furthermore, several compounds were isolated from the leaves such as eudesm-5,7-dien ( 3 ), eudesm-4,11-diene ( 4 ), 5α,6α-epoxy-eudesm-7-ene ( 5 ), eudesm-6-en-4β-ol ( 6 ), 5α,6α-epoxy-eudesm-7-en-9-ol ( 9 ), 5α,6α,7α,8α-diepoxy-eudesmane ( 10 ), and (2 S *)-eudesm-5,7-dien-2-ol ( 19 ) . About five eudesmane compounds were isolated from the seeds of G. guidonia , including 6α-ethoxyeudesm-4(15)-en-1β-ol ( 21 ), eudesm-4(15)-ene-1β,6α-diol ( 23 ), 5-epi-eudesm-4(15)-ene-1β,6β-diol ( 24 ), eudesm-4(15)-ene-1β,5α-diol ( 25 ), and eudesm-4(15),7-dien-1β-ol ( 26 ) . In addition, 5,6,7,8-diepoxy-eudesmane ( 53 ) and eudesm-5,7-dien-2α-ol ( 8 ) were obtained from leaf essential oil . Ribeiro et al. isolated γ-eudesmol ( 85 ) from the leaf essential oil of G. macrophylla , while Oliveira et al. reported two compounds, namely selina-3,7(11)-diene ( 98 ) and 7-epi-α-eudesmol ( 110 ). Two eudesmane types, α-selinene ( 118 ) and β-eudesmol ( 124 ), were also isolated from branch essential oil of G. convergens and G. silvatica . Furthermore, aromadendrane types such as allo-aromadendrene ( 34 ), viridiflorene ( 42 ), globulol ( 45 ), and epi-globulol ( 59 ) were obtained from the bark essential oil of G. cedrata . Other species, such as G. macrophylla , G. guidonia , G. kunthiana , were found to also contain similar compounds. Spathulenol ( 2 ) and palustrol ( 17 ) were first isolated from the leaves of G. macrophylla while essential oil from the leaves and the stem bark were also reported to contain aromadendrane type. Lago et al. isolated ledol ( 18 ), and α-gurjunene ( 31 ) from the leaves and aromadendrene ( 40 ) from stem bark essential oil . Seven years later, alloaromadendrane-4α,10β-diol ( 88 ) was isolated from the bark . Two aromadendrane types, viridiflorol ( 11 ) and 3-oxo-10-alloaromadendranol ( 12 ), were also obtained from the wood bark of G. guidonia , -4β,10α-aromadendranediol ( 16 ) from the leaves of G. kunthiana , and β-gurjunene ( 115 ) from G. scabra . Furthermore, guai-6-en-10β-ol ( 7 ) was the first guaian type isolated from the leaves of G. macrophylla . Compounds such as cis- β -guaiene ( 55 ), 6,9-guaiadiene ( 91 ), trans- β -guaiene ( 95 ), and guaiol ( 102 ) were isolated from the fruit and leaf essential oil . G. kunthiana also has a guaian type, while alismol ( 14 ) and alismoxide ( 15 ) were identified from the leaves . In addition, α-guaiene ( 51 ) was obtained from the leaf essential oil of G. guidonia . Caryophyllene oxide ( 20 ) and β-caryophyllene ( 33 ) were identified from the bark essential oil of G. cedrata . Núñez and Roque reported isocaryophyllene oxide ( 70 ) from the stem bark essential oil of G. guidonia . Meanwhile, two other species, G. kunthiana and G. macrophylla , were found to contain E-caryophyllene ( 73 ) and 9-epi-β-caryophyllene ( 82 ) . Magalhães et al. also reported two compounds, cis -caryophyllene ( 112 ) and caryophyllene epoxide ( 120 ), from the leaf essential oil of G. scabra and branches of G. humatensis . The derivative compounds from the cadinene type, such as α-cubebene ( 28 ) and β-copaene ( 81 ), were obtained from the leaf and stem bark essential oil of G. macrophylla . Furthermore, α-ylangene ( 29 ) and α-copaene ( 30 ) were first identified from the bark essential oil of G. cedrata , while G. guidonia was found to contain β-cubebene ( 50 ) . The α-humulene ( 32 ) and 6,7-epoxy-2,9-humuladiene ( 72 ) humulene type were identified from the stem bark essential oil of G. guidonia . Furthermore, 1(10)-epoxy-4,7-humuladiene ( 86 ) and 1(10),4-diepoxy-7-humulene were also obtained from the bark ( 87 ) . The latest discovery was performed by Magalhães et al. , where one humulene-type sesquiterpenoid humulene epoxide II ( 122 ) was identified from the branch essential oil of G. silvatica . Nunez and Roque identified germacrene D ( 35 ) from the stem bark essential oil of G. guidonia , while the G. macrophylla species was found to contain germacrene-D-4-ol ( 39 ), germacrene A ( 84 ), and germacrene B ( 100 ) in the leaf essential oil . Moreover, bicyclogermacrene type was also identified from the leaf and stem bark essential oil of G. macrophylla including bicyclogermacrene ( 36 ), cis-bicyclogermacradiene ( 41 ), and trans -bicyclogermacradiene ( 43 ) . The bark essential oil from G. cedrata was reported to contain elemene-type sesquiterpenoid γ-elemene ( 68 ) . β-elemene ( 49 ) was also isolated from the stem bark essential oil of G. guidonia . In 2005, δ-elemene ( 48 ) was reported in the leaf essential oil of this species , while elemol ( 99 ) was identified in the leaf essential oil of G. macrophylla . Eight compounds with bisabolene-type sesquiterpenoids were obtained from four species, namely G. macrophylla , G. kunthiana , G. sylvatica , and G. scabra . β-bisabolene ( 56 ) was obtained from the fruit essential oil of G. macrophylla . Magalhães et al. also identified three compounds, namely (E)-iso-γ-bisabolene ( 119 ) from the branch essential oil of G. silvatica , as well as α- cis -bergamotene ( 113 ) and α- trans -bergamotene ( 126 ) from the leaf essential oil of G. scabra . Eight years later, α-bergamotene ( 74 ), α-curcumene ( 76 ), α-zingiberene ( 77 ), and β-sesquiphellandrene ( 79 ) were isolated from the leaf essential oil of G. kunthiana . Furthermore, minor-type sesquiterpenoids were obtained from this genus, such as two compounds of opposite-type sesquiterpenoid (7 R *)-5-epi-opposit-4(15)-ene-1β,7-diol ( 22 ) and (7 R *)-opposit-4(15)-ene-1β,7-diol ( 27 ) from the seeds of G. guidonia , while longifolene ( 66 ) was isolated from the bark essential oil of G. cedrata . Two compounds of acyclic sesquiterpenoids, β-farnesene ( 75 ) and trans -nerolidol ( 121 ), were identified from the leaf essential oil of G. kunthiana and G. scabra . Moreover, cyclosativene ( 90 ), γ-himachalene ( 92 ), isolongifolan-7-α-ol ( 104 ), α-acorenol ( 105 ), hinesol ( 107 ), cedr-8(15)-en-9α-ol ( 108 ), and valerianol ( 109 ) were isolated from the leaf essential oil of G. macrophylla . Magalhães et al. also reported five other compounds, such as β-bourbonene ( 111 ) from the leaf essential oil of G. scabra ; α-santalene ( 114 ), β-santalene ( 116 ), drima-7,9(11)-diene ( 117 ) from the branches of G. convergens; and mustakone ( 125 ) from G. silvatica . All the sesquiterpenoid structures are shown in . Diterpenoid of 16% was isolated from the Guarea genus with two major types, isopimarane and labdane. One of the diterpenoid types which was first reported by Lago et al. was isopimarane from the leaves of G. macrophylla with three types, namely isopimara-7,15-dien-3-one ( 150 ), isopimara-7,15-dien-3β-ol ( 132 ), and isopimara-7,15-dien-2β-ol ( 151 ). Afterward, five diterpenoids, namely, 7α-hydroperoxy-isopimara-8(14),15-diene-2α,3β-diol ( 148 ), 19-nor-isopimara-7,15,4(18)-trien-3-one ( 149 ), isopimara-7,15-dien-2α-ol ( 152 ), isopimara-7,15-diene ( 158 ), and isopimara-7,15-diene-2α,3β-diol ( 131 ), were isolated and identified from the leaf essential oil of Guarea macrophylla from . Four types of labdane diterpenoids, namely, 3-oxo-labd-8(17),12Z,14-triene ( 133 ), 3α-hydroxylabd-8(17),12Z,14-triene ( 134 ), 3β-hydroxylabd-8(17),12Z,14-triene ( 135 ), and 19-hydroxymanoyloxide ( 135 )—identified from the leaves of G. trichilioides —were reported in 1996 by Furlan et al. . Furthermore, three labdane-type compounds such as manoyl oxide ( 153 ), labda-8,14-dien-13-ol ( 154 ), and labda-8,13-(E)-dien-15-ol ( 159 ), were isolated from the leaves of G. macrophylla , while ent -13-epimanoyloxide ( 147 ) was obtained from the leaves of G. kunthiana . Cneorubin A ( 111 ), B ( 112 ), X ( 113 ), and Y ( 114 ) were isolated from the leaves and the aerial parts of G. guidonia , while three kaurene types of diterpenoid compounds, ent -kaur-16-en-2-one ( 139 ), ent -kaur-16-ene ( 140 ), and ent -3β- and 3α-hydroxykaur-16-ene ( 141 and 142 ), were obtained from the leaves of G. kunthiana . Additionally, Magalhães et al. identified kaurene ( 164 ) from the leaf essential oil of G. sylvatica . Diterpenoids of the sandaracopimaradeine type were identified in the leaves of G. rhophalocarpa . The compounds were ent -8(14),15-sandaracopimaradiene-2α,18-diol ( 156 ), and ent -8(14),15-sandaracopimaradine-2β,18-diol ( 157 ) . Eighteen years later, sandaracopimarinal ( 163 ) was identified from the leaf essential oil of G. macrophylla . Furthermore, two diterpenoids of the clerodane type, -2-oxo-13-hydroxy,3,14-clerodandiene ( 136 ) and 13-hydroxy-3,14-clerodandiene ( 138 ), were obtained from the leaves of G. trichilioides . An investigation to identify three other compounds, including kolavelool ( 143 ), kolavenol ( 144 ), and kolavenal ( 145 ) from the leaves of G. kunthiana , was conducted by Garcez et al. . The acyclic type, phytol ( 155 ), was identified from the leaves of G. macrophylla and G. guidonia . Garcez et al. isolated -nephthenol ( 146 ) from the leaves of G. kunthiana , while one prenylaromadendrane-type boscartol C ( 160 ) was obtained from the aerial parts of G. guidonia . One of the dolabradiene types, 13-epi-dolabradiene ( 145 ), was identified from the leaf essential oil of G. macrophylla , along with phyllocladane ( 146 ) . The diterpenoid structures are presented in detail in . Thirty-five compounds were identified as triterpenoids, such as tirucallane, protolimonoid, lanostane, cycloartane, glabretal, glabretal derivatives, and apotirucallane . Cycloartane was the major triterpenoid type isolated from the Guarea genus. In 1993, seven compounds (cycloart-24-en-3,23-dione ( 173 ), 23-hydroxycycloart-24-en-3-one (epimers) ( 174 and 175 ), 3β-hydroxycycloart-24-en-23-one ( 176 ), 25-hydroxycycloart-23-en-3-one ( 177 ), 3β-21-dihydroxycycloartane ( 178 ), and 3β,21,22,23-tetrahydroxycycloartane-24(31), 25-diene ( 179 )) were identified from the leaves of G. trichilioides . Furthermore, 22,25-dihydroxycycloart-23 E -en-3-one ( 196 ), 24-methylenecycloartane-3β,22-diol ( 197 ), and cycloarta-23,25-dien-3-one ( 192 ) were obtained from the leaves of G. macrophylla , while two cyloartanes, namely (23 S *)-cycloart-24-ene-3β,23-diol ( 193 ) and (23 R *)-cycloart-24-ene-3β,23-diol ( 194 ), were isolated from the leaves of G. guidonia . In the same year, cycloart-23 E -ene-3β,25-diol ( 170 ) was discovered in the leaves of G. macrophylla , while in 2017, Conserva et al. obtained (23 S *,24 S *)-dihydroxycicloart-25-en-3-one ( 171 ). Two lanostane-type compounds, 23-hydroxy-5α-lanosta 7,9(11),24-triene-3-one ( 168 ) and 5α-lanosta-7,9(11),24-triene-3α,23-diol ( 169 ), were obtained from the leaves of Guarea rhophalocarpa , while glabretal ( 172 ) was identified from heartwood of G. glabra . Furthermore, 21,24-epoxy-3α,7α,21,23-tetraacetoxy-25-hydroxy-4α,4β,8β-trimethyl-14,18-cyclo-5α,13α,14α,17α-cholestane ( 181 ), and 21,23-epoxy-3α,7α,21,24,25-pentaacetoxy-4α, 4β,8β-trimethyl-14,18-cyclo-5α,13α,14α,17α-cholestane ( 182 ) as glabretal derivatives were identified from the leaves and twigs of G. jamicensis . The 3,4-secotirucalla-4(28),7,24-trien-3,21-dioic acid ( 165 ) and 3,4-secotirucalla-4(28),7,24-trien-3,21-dioic acid 3-methyl ester ( 166 ) as tirucallane types of triterpenoid were reported by Akinniyi et al. from the bark of G. cedrata . Furthermore, four tirucallane types, guareolide ( 186 ), guareoic acid A ( 187 ) and B ( 188 ), flindissone ( 189 ), as well as picroquassin E ( 190 ), were isolated from the aerial parts of G. guidonia . Jimenez et al. reported that three protolimonoid types, melianone ( 184 ), melianodiol ( 185 ), and 21-α-acetylmelianone ( 191 ), were first isolated from the seeds of G. grandiflora . In 2015, four compounds of this type were also identified, including 3β- O -tigloylmelianol ( 167 ), 3β- O -tigloylmeliantriol ( 198 ), and melianol ( 199 ), from the fruits of G. kunthiana . Moreover, 24-acetoxy-25-hydroxy-3,7-dioxoapotirucalla-14-en-21,23-olide ( 182 ) and 7α,24,25-trihydroxy-3-oxoapotirucalla-14-en-21,23-olide ( 183 ) as apotirucallane types were isolated from the leaves and branches of G. convergens . Limonoids are classified into many classes based on the type of skeleton , and about eleven classes have been reported from this genus. The first exploration by Housley et al. reported dihydrogedunin ( 221 ) from the heartwood of G. thompsonii . Connollyl et al. also found one andirobine-type limonoid, namely methyl 6-acetoxyangolensate ( 206 ), identified from the bark of G. thompsonii and methyl angolensate ( 214 ) from the fruits of G. kunthiana . Moreover, one of limonoid types which was called with dregeanin ( 207 ) was obtained from the bark of G. thompsonii , and rohituka-type named with 2’-hydroxyrohitukin ( 215 ) was identified from the bark of G. cedrata . The obakunol-type limonoid, 7-acetyldihydronomilin ( 216 ), was isolated from the aerial parts of G. guidonia , and the ecuadorin ( 217 ) which was one of the ecuadorin-types, was found in the aerial parts of G. kunthiana . Prieurianin ( 219 ) and 14,15β-epoxyprieuriani ( 210 ) were found in the root bark of G. guidonia as a prieurianin-type limonoid . Garcez et al. also reported mombasol ( 208 ) from the bark of G. guidonia and the investigation by Lukacova et al. obtained 7-oxo-gedunin ( 218 ) from the root bark, while three gedunin limonoids, 7-deacetoxy-7-oxogedunin ( 200 ), gedunin ( 201 ), and 6α-acetoxygedunin ( 209 ), were isolated from the seeds of G. grandiflora . Zelnik and Rosito discovered one mexicanolide type, called fissinolide ( 220 ), in the seeds of G. trichilioides . Five years later, the seeds were found to also contain angustinolide ( 224 ) . Humilinolide E ( 211 ), methyl 2-hydroxy-3b-tigloyloxy-1-oxomeliac-8(30)-enate ( 212 ), and swietenine acetate ( 213 ) were isolated from the fruits of G. kunthiana . Furthermore, an investigation by Bellone et al. identified 3-(2′-hydroxyisovaleroyl) khasenegasin I ( 205 ) from the stem bark of G. guidonia . The twigs of G. mayombensis produced azadirachtin-type mayombensin ( 222 ) and azadirachtin I ( 223 ) . Meanwhile, three compounds of A2, B, D-seco skeletons such as chisomicine D ( 202 ), chisomicine E ( 203 ), and chisomicine F ( 204 ), were identified from the stem bark of G. guidonia . Ergostane- and pregnane-type steroids were isolated from the Guarea genus, along with general steroid compounds such as β-sitosterol ( 229 ), stigmasterol ( 230 ), and β-sitostenone ( 233 ) . Furthermore, the steroids glycoside stigmasterol glucoside ( 231 ) and β-sitosterol glucoside ( 232 ) were obtained from the twigs of G. mayombensis , while two ergostane-type steroids, ergosta-5,24(24′)-diene-3β,7α,21-triol ( 236 ) and ergosta-5,24(24′)-diene-3β,4β,22S-triol ( 237 ), were identified from the leaves and branches of G. convergens . Garcez et al. also reported two pregnane-type steroids, 2α,3β-dihydroxy-16,17-seco-pregn-17-ene-16-oic acid methyl ester 2β,19-hemiketal ( 234 ) and 2,3:16,17-di-seco-pregn-17-ene-3-oic acid-16-oic acid methyl ester-19-hydroxy-2-carboxylic acid-2,19-lactone ( 235 ), from the trunk bark of G. guidonia . Flavonoid, lignan, ceramide, and coumarin were also identified from this plant genus. Quercetin 3- O -β- d -glucopyranoside ( 225 ), quercetin 3- O -β- d -galactopyranoside ( 226 ), and kaempferol 7- O -β- d -glucopyranoside ( 227 ) as glucoside flavonoids were isolated from the flowering branches of G. macrophylla . Furthermore, one neolignane compound, dehydrodiconiferyl alcohol-4-β- d -glucoside ( 228 ), was reported from the same part of this species . Two ceramides, ceramide A ( 238 ) and B ( 239 ), were obtained from the twigs of G. mayombensis , while one coumarin, scopoletin, ( 240 ) was found in the leaves of G. rhopalocarpa . Guarea Bioactivity Plants of the genus Guarea have long been used in traditional medicine in several countries for relieving body aches, diarrhea, angina, asthma, and dyspnea. The boiled leaves are used as an emetic . Several biological tests conducted showed that the plant extract has cytotoxic, antimalarial, anti-inflammatory, antimicrobial, insecticidal, antioxidant, antiparasitic, antiprotozoal, antiviral, and phosphorylation inhibitor activities . 4.1. Cytotoxic The cytotoxic activity of the Guarea genus has been studied in many extracts and compounds (diterpenoids, triterpenoids, limonoids, and steroids) using various test methods. The findings could lead to the development of new antitumor and anticancer drugs. The extract and the compounds of four species from the Guarea genus were evaluated in 1962. Lukacova et al. identified three compounds from G. guidonia , including 14,15β-epoxyprieuriani ( 210 ), 7-oxo-gedunin ( 218 ), and prieurianin ( 219 ). The compounds 210 and 219 are active against the leukemia cell line P388 ED 50 0.47–0.74 µg/mL and P388 ED50 4.4–7.8 µg/mL, respectively, while 218 is not active. Furthermore, methylene chloride extract was evaluated against U-937 cell lines; bark and leaf extract of G. polymera each showed a lethal dose (LD 50 ) of 6.1 ± 0.5 µg/mL and 6.1 ± 1.2 µg/mL while the seed of G. guidonia had a LD 50 of 28.8 ± 8.2 µg/mL . The six compounds from G. rhophalacarpa ent -8(14), namely 15-sandaracopimaradiene-2α,18-diol ( 156 ), ent -8(14),15-sandaracopimaradine-2β,18-diol ( 157 ), 23-hydroxy-5α-lanosta 7,9(11),24-triene-3-one ( 168 ), 5α-lanosta-7,9(11),24-triene-3α,23-diol ( 169 ), stigmasterol ( 230 ), and scopoletin ( 240 ), were tested against the KB cell line with an inhibitory concentration (IC 50 ) of 48 µM, 75.8 µM, 30.2 µM, 21.2 µM, > 1272 µM, and 130.2 µM, respectively . Four compounds from G. macrophylla were also tested against the five cancer cell types B16F10-Nex2, A2058, MCF-7, HL-60, and HeLa. Cycloart-23E-ene-3β,25-diol ( 170 ) had the best activity compared to the other three compounds. Meanwhile, the results of the tests against HL-60, HeLa, B16F10-Nex2, A2058, and MCF-7 were 18.3, 52.1, 58.9, 60.7 and 63.5 µM, respectively. Two other compounds, isopimara-7,15-dien-2α,3β-diol ( 131 ) and isopimara-7,15-dien-3β-ol ( 132 ), have activity over 100 µM against five cell lines . Hernandez et al. identified five compounds of which three have an EC 50 under 100 µM. Five compounds were also tested against the Jurkat, HeLa, MCF-7, and PBMC cell lines. Flindissone ( 189 ) showed activity with EC 50 25, 27, 50, and > 100 µM for the Jurkat, HeLa, MCF-7, and PBMC cell lines, while guareoic acid A ( 187 ) had a high EC 50 against the Jurkat cell line with a value of 39 µM. Moreover, picroquassin E ( 190 ), guareolide ( 186 ), and guareoic acid A ( 187 ) showed no activity against PBMC (nontumor human peripheral blood mononuclear cell line). In a recent cytotoxic assay studied by Bellone et al. on four compounds isolated from G. guidonia , chisomicine D ( 202 ) showed inhibitory growth value to U-937 and HeLa cell lines with an IC 50 20 ± 3 µM and > 50 µM, but no activity was found against PBMC. Other compounds (chisomicine E ( 203 ), chisomicine F ( 204 ), and 3-(2′-hydroxyisovaleroyl) khasenegasin I ( 205 )) were also found to be inactive against U-937 and HeLa cell lines. 4.2. Anti-Inflamation Catabolism takes precedence over anabolism in an inflammatory state. It is also a defense mechanism that aids in the elimination of potentially harmful factors and maintains body homeostasis. Because of the increased permeability of capillaries and white blood cells, this causes increased blood flow to the site of inflammation, resulting in symptoms such as redness, swelling, and pain. Oga et al. reported the anti-inflammation activity from ethanol extract of G. guidonia seeds against male Wistar rats. About an 8.0 mL/kg extract dose provided significant inhibition of carrageenin-induced edema, and the effects increased periodically. Similarly, a 5.0 mL/kg extract dose provided effects amounting to 15% on granuloma tissue formation after 2, 4, and 6 days. 4.3. Antimalarial Four extracts from G. multiflora were obtained using petroleum ether, methanol, water, and chloroform. They were collected from leaves, stem bark, and wood, as well as fruits. The extracts showed no significant results as three, namely, petroleum ether from leaves, methanol of stem bark and fruits, as well as chloroform from stem bark, had an IC50 of 50 µg/mL. Meanwhile, other extracts showed an IC 50 of 500 µg/mL and were not active . 4.4. Antiprotozoal Chloroform extract from leaves of G. rhopalocarpa showed high activity against Leishmania donovani with IC 50 45 µg/mL. Moreover, methanol and butanol extracts have IC 50 62.5 µg/mL and 300 µg/mL, while the water extract has the lowest activity. Ent -8(14),15-sandaracopimaradiene-2α,18-diol ( 156 ) was more active than ent -8(14),15-sandaracopimaradine-2β,18-diol ( 157 ) against L. donovani promastigotes with IC 50 of 16,8 and 49.7 µg/mL, respectively. A study on two triterpenoids showed that 23-hydroxy-5α-lanosta 7,9(11),24-triene-3-one ( 168 ) is more active than 5α-lanosta-7,9(11),24-triene-3α,23-diol ( 169 ), tested using L. donovani with an IC 50 of 7.2 µg/mL . Furthermore, Weniger et al. identified methylene chloride extract of bark and leaves of G. polymera which has a selectivity index against Leishmania Viannia panamensis with a lethal dose/effective dose (LD 50 /ED 50 ) of 1.5 µg/mL. The seeds of G. guidonia were also active against Plasmodium falciparum with an LD 50 /IC 50 2.9 µg/mL. Hexane extract obtained from the root of G. kunthiana reportedly had antileishmanial activity on the intracellular parasite, Leishmania donovani . The test was evaluated using the colorimetric method which was an MTT assay and the extract showed an IC 50 of 7.9 ± 1.3 µg/mL . Moreover, the 3β- O -tigloylmelianol ( 167 ) was investigated with larvicide and ecydysis tests against the cattle tick of Rhipicephalus (Boophilus) microplus (Canestrini) (Acari: Ixodidae); the compound showed a significant reduction in the number of oocytes . 4.5. Antiviral Two water extracts from the fruits and leaves of G. guidonia were identified to have antiviral activity against pseudorabies and mouth disease virus in the IB-RS-2 pig cell lines and against bovine herpesvirus 1 (BHV-1) in the GBK bovine cell line. The result of the fruit extract test was more active than the leaves in the IB-RS-2 cell. Meanwhile, the activity of the two extracts increased with an IC 50 of 62.5 and 125 µg/mL in the GBK cell . 4.6. Antimicrobial Several compounds isolated from Guarea have been found to have antimicrobial activity. This activity provides antibiotics against microorganisms that can cause food defects, such as pathogens. A study conducted by Pandini et al. reported the result of antimicrobial activity for essential oil and methanol extracts from G. kunthiana . Methanol extract showed no activity in the MIC or MBC test. Meanwhile, the essential oil evaluated with MIC and MBC against S. infantris , S. tyrphimurium and S. give showed antimicrobial activity amounting to 54.6 µg/mL. The ethyl acetate extract had activity ranging from 100 to 200 µg/mL. 4.7. Insecticidal Activity Four compounds were isolated from G. grandiflora and evaluated against the growth of larva ECB (European corn borer). The results showed that 21-α-acetylmelianone ( 191 ) and melianone ( 184 ) have the activity to inhibit the growth of ECB larvae using the fed control diet. Meanwhile, the pupal weight was not affected by any of the compounds but the percentage of pupation was significantly reduced by melianodiol ( 185 ) . The 10% alcoholic extract from G. kunthiana produced the highest percentage of larval mortality, while the 10% aqueous extract exhibited 14.6%. Moreover, 200 mg/mL of essential oil affected 28.6% of larval mortality . The ethyl acetate extract from G. kunthiana was also evaluated against Aedes aegyptyi with LC 50 and LC 90 values of 105.7 µg/mL and 408, 9 µg/mL, respectively. Melianodiol ( 185 ) exhibited the highest activity with LC 50 14.4 and LC 90 17.54 µg/mL, while meliantriol ( 195 ) showed the activity of over 100 µg/mL . 4.8. Antioxidant and Phosphorylation Inhibitor The antioxidant activity is a defense mechanism that protects our bodies from oxidative stress caused by free radicals and reactive oxygen species (ROS). Oxidative stress can occur as a result of ROS formation and the detoxification of elevated levels of ROS, resulting in impaired cellular function. The compounds which have been isolated from this genus have antioxidant activity . The essential oil, alcoholic, aqueous, and ethyl acetate ex-tracts were evaluated. Based on the results, the alcoholic extract showed an IC 50 of 15.3 µg/mL while ethyl acetate had the lowest activity with an IC 50 176.8 µg/mL. On the other hand, two compounds, 7-deacetoxy-7-oxogedunin ( 200 ) and Gedunin ( 201 ), which were obtained from G. grandiflora, showed 7-deacetoxy-7-oxogedunin up to 350 µM and could inhibit ATP synthase coupled to electron transfer, while the activity of Mg 2+ -ATPase was only slightly inhibited. Meanwhile, the increased concentration of 7-deacetoxy-7-oxogedunin up to 300 µM did not significantly inhibit the ATP hydrolysis process but ATPase activity caused inhibition of 7 and 6% for Mg 2+ and Ca 2+ . Gedunin did not significantly inhibit Ca 2+ - and Mg 2+ -dependent ATPase activities . The cytotoxic activity of the Guarea genus has been studied in many extracts and compounds (diterpenoids, triterpenoids, limonoids, and steroids) using various test methods. The findings could lead to the development of new antitumor and anticancer drugs. The extract and the compounds of four species from the Guarea genus were evaluated in 1962. Lukacova et al. identified three compounds from G. guidonia , including 14,15β-epoxyprieuriani ( 210 ), 7-oxo-gedunin ( 218 ), and prieurianin ( 219 ). The compounds 210 and 219 are active against the leukemia cell line P388 ED 50 0.47–0.74 µg/mL and P388 ED50 4.4–7.8 µg/mL, respectively, while 218 is not active. Furthermore, methylene chloride extract was evaluated against U-937 cell lines; bark and leaf extract of G. polymera each showed a lethal dose (LD 50 ) of 6.1 ± 0.5 µg/mL and 6.1 ± 1.2 µg/mL while the seed of G. guidonia had a LD 50 of 28.8 ± 8.2 µg/mL . The six compounds from G. rhophalacarpa ent -8(14), namely 15-sandaracopimaradiene-2α,18-diol ( 156 ), ent -8(14),15-sandaracopimaradine-2β,18-diol ( 157 ), 23-hydroxy-5α-lanosta 7,9(11),24-triene-3-one ( 168 ), 5α-lanosta-7,9(11),24-triene-3α,23-diol ( 169 ), stigmasterol ( 230 ), and scopoletin ( 240 ), were tested against the KB cell line with an inhibitory concentration (IC 50 ) of 48 µM, 75.8 µM, 30.2 µM, 21.2 µM, > 1272 µM, and 130.2 µM, respectively . Four compounds from G. macrophylla were also tested against the five cancer cell types B16F10-Nex2, A2058, MCF-7, HL-60, and HeLa. Cycloart-23E-ene-3β,25-diol ( 170 ) had the best activity compared to the other three compounds. Meanwhile, the results of the tests against HL-60, HeLa, B16F10-Nex2, A2058, and MCF-7 were 18.3, 52.1, 58.9, 60.7 and 63.5 µM, respectively. Two other compounds, isopimara-7,15-dien-2α,3β-diol ( 131 ) and isopimara-7,15-dien-3β-ol ( 132 ), have activity over 100 µM against five cell lines . Hernandez et al. identified five compounds of which three have an EC 50 under 100 µM. Five compounds were also tested against the Jurkat, HeLa, MCF-7, and PBMC cell lines. Flindissone ( 189 ) showed activity with EC 50 25, 27, 50, and > 100 µM for the Jurkat, HeLa, MCF-7, and PBMC cell lines, while guareoic acid A ( 187 ) had a high EC 50 against the Jurkat cell line with a value of 39 µM. Moreover, picroquassin E ( 190 ), guareolide ( 186 ), and guareoic acid A ( 187 ) showed no activity against PBMC (nontumor human peripheral blood mononuclear cell line). In a recent cytotoxic assay studied by Bellone et al. on four compounds isolated from G. guidonia , chisomicine D ( 202 ) showed inhibitory growth value to U-937 and HeLa cell lines with an IC 50 20 ± 3 µM and > 50 µM, but no activity was found against PBMC. Other compounds (chisomicine E ( 203 ), chisomicine F ( 204 ), and 3-(2′-hydroxyisovaleroyl) khasenegasin I ( 205 )) were also found to be inactive against U-937 and HeLa cell lines. Catabolism takes precedence over anabolism in an inflammatory state. It is also a defense mechanism that aids in the elimination of potentially harmful factors and maintains body homeostasis. Because of the increased permeability of capillaries and white blood cells, this causes increased blood flow to the site of inflammation, resulting in symptoms such as redness, swelling, and pain. Oga et al. reported the anti-inflammation activity from ethanol extract of G. guidonia seeds against male Wistar rats. About an 8.0 mL/kg extract dose provided significant inhibition of carrageenin-induced edema, and the effects increased periodically. Similarly, a 5.0 mL/kg extract dose provided effects amounting to 15% on granuloma tissue formation after 2, 4, and 6 days. Four extracts from G. multiflora were obtained using petroleum ether, methanol, water, and chloroform. They were collected from leaves, stem bark, and wood, as well as fruits. The extracts showed no significant results as three, namely, petroleum ether from leaves, methanol of stem bark and fruits, as well as chloroform from stem bark, had an IC50 of 50 µg/mL. Meanwhile, other extracts showed an IC 50 of 500 µg/mL and were not active . Chloroform extract from leaves of G. rhopalocarpa showed high activity against Leishmania donovani with IC 50 45 µg/mL. Moreover, methanol and butanol extracts have IC 50 62.5 µg/mL and 300 µg/mL, while the water extract has the lowest activity. Ent -8(14),15-sandaracopimaradiene-2α,18-diol ( 156 ) was more active than ent -8(14),15-sandaracopimaradine-2β,18-diol ( 157 ) against L. donovani promastigotes with IC 50 of 16,8 and 49.7 µg/mL, respectively. A study on two triterpenoids showed that 23-hydroxy-5α-lanosta 7,9(11),24-triene-3-one ( 168 ) is more active than 5α-lanosta-7,9(11),24-triene-3α,23-diol ( 169 ), tested using L. donovani with an IC 50 of 7.2 µg/mL . Furthermore, Weniger et al. identified methylene chloride extract of bark and leaves of G. polymera which has a selectivity index against Leishmania Viannia panamensis with a lethal dose/effective dose (LD 50 /ED 50 ) of 1.5 µg/mL. The seeds of G. guidonia were also active against Plasmodium falciparum with an LD 50 /IC 50 2.9 µg/mL. Hexane extract obtained from the root of G. kunthiana reportedly had antileishmanial activity on the intracellular parasite, Leishmania donovani . The test was evaluated using the colorimetric method which was an MTT assay and the extract showed an IC 50 of 7.9 ± 1.3 µg/mL . Moreover, the 3β- O -tigloylmelianol ( 167 ) was investigated with larvicide and ecydysis tests against the cattle tick of Rhipicephalus (Boophilus) microplus (Canestrini) (Acari: Ixodidae); the compound showed a significant reduction in the number of oocytes . Two water extracts from the fruits and leaves of G. guidonia were identified to have antiviral activity against pseudorabies and mouth disease virus in the IB-RS-2 pig cell lines and against bovine herpesvirus 1 (BHV-1) in the GBK bovine cell line. The result of the fruit extract test was more active than the leaves in the IB-RS-2 cell. Meanwhile, the activity of the two extracts increased with an IC 50 of 62.5 and 125 µg/mL in the GBK cell . Several compounds isolated from Guarea have been found to have antimicrobial activity. This activity provides antibiotics against microorganisms that can cause food defects, such as pathogens. A study conducted by Pandini et al. reported the result of antimicrobial activity for essential oil and methanol extracts from G. kunthiana . Methanol extract showed no activity in the MIC or MBC test. Meanwhile, the essential oil evaluated with MIC and MBC against S. infantris , S. tyrphimurium and S. give showed antimicrobial activity amounting to 54.6 µg/mL. The ethyl acetate extract had activity ranging from 100 to 200 µg/mL. Four compounds were isolated from G. grandiflora and evaluated against the growth of larva ECB (European corn borer). The results showed that 21-α-acetylmelianone ( 191 ) and melianone ( 184 ) have the activity to inhibit the growth of ECB larvae using the fed control diet. Meanwhile, the pupal weight was not affected by any of the compounds but the percentage of pupation was significantly reduced by melianodiol ( 185 ) . The 10% alcoholic extract from G. kunthiana produced the highest percentage of larval mortality, while the 10% aqueous extract exhibited 14.6%. Moreover, 200 mg/mL of essential oil affected 28.6% of larval mortality . The ethyl acetate extract from G. kunthiana was also evaluated against Aedes aegyptyi with LC 50 and LC 90 values of 105.7 µg/mL and 408, 9 µg/mL, respectively. Melianodiol ( 185 ) exhibited the highest activity with LC 50 14.4 and LC 90 17.54 µg/mL, while meliantriol ( 195 ) showed the activity of over 100 µg/mL . The antioxidant activity is a defense mechanism that protects our bodies from oxidative stress caused by free radicals and reactive oxygen species (ROS). Oxidative stress can occur as a result of ROS formation and the detoxification of elevated levels of ROS, resulting in impaired cellular function. The compounds which have been isolated from this genus have antioxidant activity . The essential oil, alcoholic, aqueous, and ethyl acetate ex-tracts were evaluated. Based on the results, the alcoholic extract showed an IC 50 of 15.3 µg/mL while ethyl acetate had the lowest activity with an IC 50 176.8 µg/mL. On the other hand, two compounds, 7-deacetoxy-7-oxogedunin ( 200 ) and Gedunin ( 201 ), which were obtained from G. grandiflora, showed 7-deacetoxy-7-oxogedunin up to 350 µM and could inhibit ATP synthase coupled to electron transfer, while the activity of Mg 2+ -ATPase was only slightly inhibited. Meanwhile, the increased concentration of 7-deacetoxy-7-oxogedunin up to 300 µM did not significantly inhibit the ATP hydrolysis process but ATPase activity caused inhibition of 7 and 6% for Mg 2+ and Ca 2+ . Gedunin did not significantly inhibit Ca 2+ - and Mg 2+ -dependent ATPase activities . Guarea is one of the largest genera of the Meliaceae family, and about 240 compounds have been obtained through the hydrodistillation and isolation process with the majority of them being sesquiterpenoids. Furthermore, the bioactivity data show that this plant has a variety of activities, specifically for cytotoxic activity. |
Succession of medico-legal important flesh flies (Diptera: Sarcophagidae) in the temporal gradient of pig decomposition in the Brazilian Cerrado | cdc60ce9-7d28-4a44-84c6-63b0bf6302a7 | 11001974 | Forensic Medicine[mh] | In Brazil, particularly in the Northeast region, the occurrence of criminal incidents involving unresolved deaths, where suspects are either charged or acquitted, is substantial. This situation has sparked significant concern within the population . In the state of Maranhão, which ranks as the second-largest state in terms of territorial extent in the Northeast region (encompassing 329,651.496 square kilometers, spanning 217 municipalities, and hosting an approximate population of 7,153,262 individuals), as well as the eighth largest state among the 26 states and the Federal District , the homicide rate reached 24.1 deaths per 100,000 inhabitants in 2019. This figure surpasses the corresponding rates observed in numerous other states across the country . Remarkably, a considerable portion of these homicides (80 cases) remained categorized as of undetermined cause, leading to a lack of judicial resolution due to insufficient evidence and a scarcity of tools capable of aiding in the resolution of these crimes . Thus, the use of forensic entomology can contribute valuable information about the biology and ecological succession of flies (Diptera), beetles (Coleoptera) and other cadaveric insects, to assist police expertise as evidence in solving crimes involving homicides , . In this context, it is of paramount importance to highlight the significant role of flesh flies, a member of the Sarcophagidae family (Diptera), as tool that can be useful in forensic studies within the medico-legal domain. This contribution is particularly notable in estimating the post-mortem interval in homicide investigations , . This phenomenon stems from the life cycle of flesh flies, which encompasses several developmental stages, commencing with the first instar larva and progressing through the second and third instars, ultimately culminating in the pupal phase. These immature forms primarily develop within carcasses and decomposing cadavers until attaining adult stage – . Multiple investigations have demonstrated that specific species of adult flesh flies appear during distinct phases of the decomposition process, specially in the early, intermediate, and advanced stages , , – . In order to facilitate forensic assessments involving human remains using the aid of flesh flies, preliminary investigations are usually undertaken. These initial inquiries commonly employ carcass models of alternative vertebrate species, such as pigs, rats, rabbits, cats, dogs, among others, to acquire insights into the succession pattern and potential overlap of both adult and immature species across the process of cadaver decomposition. Pig carcasses are the most used in forensic studies, because they have some similarities with humans in size (amount of biomass), integument, proportion and distribution of hair, size of the rib cage, specificities of internal organs and also similarity of the fauna associated with decomposition , , – . These investigations require diverse environmental contexts, encompassing various seasons and regions characterized by distinct land use and coverage, including urban, rural, or natural vegetation areas. This approach allows for data extrapolation in order to establish the chronology of cadaver decomposition , , , , . In particular, when working with areas characterized by native vegetation, distinct biomes, or varying plant formations, there exists a notable diversity in the composition of Sarcophagidae species and the surrounding environmental conditions , , , , – . Consequently, it becomes imperative to investigate these contexts separately in order to enhance the accuracy of extrapolating previous knowledge to cadavers. In the Brazilian context, the limited studies available on flesh fly succession in pig carcasses predominantly focus on the Central-West region, with a particular emphasis on the Southeast region, since there is a greater number of researchers conducting studies for forensic purposes , , , – . There are only a few studies conducted in Cerrado regions , , , , and these have been particularly scarce for the Cerrado of Maranhão, situated in the Northeastern region of the country. Within this specific Cerrado environment in Maranhão, only a study has focused on making an inventory and description of the succession patterns of flesh fly communities across different stages of pig carcass decomposition. Nevertheless, a comprehensive and analytical exploration of this ecological succession process has yet to be conducted . This gap in the literature is especially significant, particularly since this dearth of understanding aligns with one of the most violent regions characterized by a low efficacy in solving homicides . Consequently, it is of utmost importance for further studies involving decomposing carcasses to be conducted in this region, allowing the accumulated data on flesh flies to make a substantial contribution to the field of forensic science. In addition, in the face of a large biodiversity, detecting the pattern of change for a smaller number of species allows for greater agility in investigations and a reduction in the time to obtain answers, which is a great advance. With the objective of elucidating the ecological succession dynamics of flesh flies for forensic applications, the present study was conducted to investigate the temporal progression of these flies throughout different stages of pig carcass decomposition within the Cerrado regions of Northeastern Brazil. Experimental design The research was conducted in Cerrado areas, which resembles a savanna ecosystem, situated in the eastern region of the state of Maranhão, located in Northeastern Brazil. These areas are specifically situated within the municipality of Caxias and fall under the jurisdiction of an established environmental protection zone known as the Municipal Environmental Protection Area of Inhamum ( APA do Inhamum ), previously referred to as the Inhamum Ecological Reserve (REI). Geographically, the area is positioned at coordinates 04°53′54.8′′S and 43°26′32.9′′W. The study site is traversed by the MA-127 highway, connecting Caxias to São João do Soter (Fig. ). Following the Köppen climate classification system for Brazil as outlined by Alvares et al. , the study regions are situated within zones classified as having a Tropical Aw climate—characterized by a dry winter, annual total precipitation ranging between 1300 and 1600 mm, and an average annual temperature surpassing 26 °C. Two distinct experiments were undertaken in this study. The first one was performed during the dry season, during the months of July and August in 2010. This period exhibited minimal precipitation, with a recorded value of 1.7 mm. During this time frame, average temperature and relative humidity stood at 27.5 °C and 71%, respectively. The second experiment was conducted in the rainy season, during the months of March and April in 2011. This season was characterized by a notable precipitation total of 93.8 mm, along with average temperature and relative humidity values of 27 °C and 82%, respectively. Detailed information concerning the study locations has been outlined by Silva et al. . It is noteworthy to mention that APA do Inhamum , according to local news sources, has been sporadically utilized for the disposal of cadavers (Attention, sensitive content: https://youtu.be/ScpUnkv342c ; https://www.youtube.com/watch?v=DBQ778L71N4) – . For experimental purposes, a total of six pigs carcasses ( Sus scrofa Linnaeus; Artiodactyla: Suidae), each weighing 12 kg, were employed as animal models during both the dry and rainy seasons. The carcasses were obtained from a slaughterhouse. Following the ethical guidelines stipulated by the Ethics Commission on Animal Use (CEUA), the pigs were humanely euthanized on-site through a shot to the anterior part of the head, without suffering or pain to the animals, and without the use of chemical products to euthanasia. Subsequently, to prevent disturbance by scavenging vertebrates and to maintain the integrity of the carcasses, the pigs were placed within metal cages (110 × 85 × 85 cm; mesh size 3 × 3 cm). These cages were strategically positioned beneath tree canopies in shaded areas, with an approximate separation of 500 m between them. The present study was conducted with authorization from the Municipal Department of Environment and Preservation of Natural Resources of Caxias, MA (SEMUMA). A suspended trap (measuring 150 × 160 cm), adapted from Rafael and Gorayeb , was placed above each cage. This trap effectively covered over half of the upper section of the cage. The base of the trap was positioned 30 cm above the ground to facilitate the entry and capture of flies attracted to the pig carcasses. The flies captured by the collection cup, equipped with a strip of K-Othrine, a liquid insecticide commonly used for the eradication of predator insects like ants (Hymenoptera: Formicidae), were collected daily between 7 and 10 AM. Flies that were still alive were carefully transferred using an entomological net into a vial containing ethyl acetate for euthanization. The fly collection was authorized by the Biodiversity Authorization and Information System (SISBIO; License No. 12417). The collected flies were preserved in 100 ml vials, meticulously labeled with information on the collection location, date, collector's name, and the specific pig carcass source. These vials contained a solution of 92.8% ethanol and were then deposited within the Zoological Collection of Maranhão (CZMA) at the Caxias Campus of the State University of Maranhão (UEMA) in the municipality of Caxias, located in the state of Maranhão, Northeastern Brazil. The male specimens underwent identification at the Laboratory of Invertebrate Studies (LEI) at UEMA. This process involved employing a stereomicroscope and entomological forceps, utilizing dichotomous keys, revisionary studies, and species descriptions from various sources, such as articles – , book chapters , and theses , . Some specimens were also identified through comparisons with voucher material from CZMA and the Entomological Collection of the Emílio Goeldi Museum (MPEG) in Belém, Pará, Northern Brazil. Identification predominantly involved the examination of male genitalia, facilitated by careful manipulation using entomological forceps and pins within a petri dish containing alcohol. However, female specimens were not identified due to the absence of available literature to support such identifications. After identification, specimens belonging to the same species were meticulously placed in distinct vials, each adequately labeled with sizes proportionate to the quantity of specimens. The voucher material resulting from this research is housed in the collections of both CZMA and MPEG. The comprehensive species list has been previously published by Silva et al. . Data analysis In the scope of the current study, each collection day for each pig carcass in both the dry and rainy seasons was treated as an individual sampling unit. Initially, a Procrustes analysis was conducted to evaluate the correspondence between the collected flesh fly communities in the two distinct seasons, i.e., dry and rainy. Based on the outcome, if the communities gathered during the dry and rainy seasons displayed substantial congruence, implying similarity across the year, the analyses would be pursued in a combined manner. Conversely, if the communities demonstrated incongruity or weak connection, indicating dissimilarity over the period, they would be independently scrutinized, considering the collections from both the dry and rainy seasons. The Procrustes analysis involves a comparison of two data matrices by aligning them using an ordination and rotational adjustment algorithm, generating "M12" values that indicate the extent of concordance between the datasets . A reduced “M12” value indicates a higher level of congruence between the data matrices. To perform the Procrustes analysis, the two matrices containing data of flesh fly species from the dry and rainy seasons underwent a logarithmic transformation using the formula log (x + 1). Subsequently, a Bray–Curtis distance matrix was computed, since the abundance data of the species were used. The logarithmic transformation was implemented to mitigate biases induced by non-normal data within the biological data matrix . The Bray–Curtis distance matrix was employed for community ordination, facilitated by Principal Coordinates Analysis (PCoA) . The PCoA axes were used as input for the Procrustes analysis, utilizing the Monte Carlo permutation test (9999 permutations). To establish congruence between the flesh fly communities across the dry and rainy seasons, two criteria were implemented. Initially, a significance level of 5% was applied; if the obtained result was not statistically significant ( p > 0.05), the matrices would be deemed incongruent. Additionally, even when a significant relationship was present ( p < 0.05), the M12 value needed to be < 0.30 to indicate a substantial level of congruence between the matrices. To investigate alterations in the distribution patterns of visiting flesh flies species and to pinpoint their change points (cp) and distribution trends across the temporal gradient of pig carcass decomposition days, we employed the Threshold Indicator Taxa Analysis (TITAN). This analytical method identifies the change point (nCPA) of species abundance along an environmental gradient (in this case, the time of decomposition in days) and delineates the direction of this shift based on the scores derived from the Individual Indicator Value index ( IndVal ). Should a species exhibit heightened occurrence and increased abundance from the point of change towards the advanced stages of the carcass decomposition gradient, it is categorized as “Z + ”; conversely, if there is a surge in occurrence and abundance from the point of change towards the initial phases of the carcass decomposition gradient, it falls under the “Z− ” category , . The criteria employed to designate a species as Z + or Z− encompass purity and reliability levels of ≥ 90%, along with a p -value ≤ 0.05. These statistical values are ascertained through IndVal scores at each change point via bootstrap and permutation procedures. Less common species were omitted from TITAN analysis due to insufficient occurrences along the gradient to qualify as Z + or Z− indicators. Consequently, solely species exhibiting a minimum of three consecutive days of occurrence, without alternation, and present in at least one carcass, as well as species with more than five collected specimens were considered for inclusion in the TITAN analysis , . The analyses were carried out using the R software . Principal Coordinates Analysis (PCoA) was performed using the vegdist function from the Vegan package , and the cmdscale function from the Stats package . Procrustes analysis was conducted using the protest function from the vegan package . The TITAN analysis was executed using the TITAN function from the TITAN2 package . The research was conducted in Cerrado areas, which resembles a savanna ecosystem, situated in the eastern region of the state of Maranhão, located in Northeastern Brazil. These areas are specifically situated within the municipality of Caxias and fall under the jurisdiction of an established environmental protection zone known as the Municipal Environmental Protection Area of Inhamum ( APA do Inhamum ), previously referred to as the Inhamum Ecological Reserve (REI). Geographically, the area is positioned at coordinates 04°53′54.8′′S and 43°26′32.9′′W. The study site is traversed by the MA-127 highway, connecting Caxias to São João do Soter (Fig. ). Following the Köppen climate classification system for Brazil as outlined by Alvares et al. , the study regions are situated within zones classified as having a Tropical Aw climate—characterized by a dry winter, annual total precipitation ranging between 1300 and 1600 mm, and an average annual temperature surpassing 26 °C. Two distinct experiments were undertaken in this study. The first one was performed during the dry season, during the months of July and August in 2010. This period exhibited minimal precipitation, with a recorded value of 1.7 mm. During this time frame, average temperature and relative humidity stood at 27.5 °C and 71%, respectively. The second experiment was conducted in the rainy season, during the months of March and April in 2011. This season was characterized by a notable precipitation total of 93.8 mm, along with average temperature and relative humidity values of 27 °C and 82%, respectively. Detailed information concerning the study locations has been outlined by Silva et al. . It is noteworthy to mention that APA do Inhamum , according to local news sources, has been sporadically utilized for the disposal of cadavers (Attention, sensitive content: https://youtu.be/ScpUnkv342c ; https://www.youtube.com/watch?v=DBQ778L71N4) – . For experimental purposes, a total of six pigs carcasses ( Sus scrofa Linnaeus; Artiodactyla: Suidae), each weighing 12 kg, were employed as animal models during both the dry and rainy seasons. The carcasses were obtained from a slaughterhouse. Following the ethical guidelines stipulated by the Ethics Commission on Animal Use (CEUA), the pigs were humanely euthanized on-site through a shot to the anterior part of the head, without suffering or pain to the animals, and without the use of chemical products to euthanasia. Subsequently, to prevent disturbance by scavenging vertebrates and to maintain the integrity of the carcasses, the pigs were placed within metal cages (110 × 85 × 85 cm; mesh size 3 × 3 cm). These cages were strategically positioned beneath tree canopies in shaded areas, with an approximate separation of 500 m between them. The present study was conducted with authorization from the Municipal Department of Environment and Preservation of Natural Resources of Caxias, MA (SEMUMA). A suspended trap (measuring 150 × 160 cm), adapted from Rafael and Gorayeb , was placed above each cage. This trap effectively covered over half of the upper section of the cage. The base of the trap was positioned 30 cm above the ground to facilitate the entry and capture of flies attracted to the pig carcasses. The flies captured by the collection cup, equipped with a strip of K-Othrine, a liquid insecticide commonly used for the eradication of predator insects like ants (Hymenoptera: Formicidae), were collected daily between 7 and 10 AM. Flies that were still alive were carefully transferred using an entomological net into a vial containing ethyl acetate for euthanization. The fly collection was authorized by the Biodiversity Authorization and Information System (SISBIO; License No. 12417). The collected flies were preserved in 100 ml vials, meticulously labeled with information on the collection location, date, collector's name, and the specific pig carcass source. These vials contained a solution of 92.8% ethanol and were then deposited within the Zoological Collection of Maranhão (CZMA) at the Caxias Campus of the State University of Maranhão (UEMA) in the municipality of Caxias, located in the state of Maranhão, Northeastern Brazil. The male specimens underwent identification at the Laboratory of Invertebrate Studies (LEI) at UEMA. This process involved employing a stereomicroscope and entomological forceps, utilizing dichotomous keys, revisionary studies, and species descriptions from various sources, such as articles – , book chapters , and theses , . Some specimens were also identified through comparisons with voucher material from CZMA and the Entomological Collection of the Emílio Goeldi Museum (MPEG) in Belém, Pará, Northern Brazil. Identification predominantly involved the examination of male genitalia, facilitated by careful manipulation using entomological forceps and pins within a petri dish containing alcohol. However, female specimens were not identified due to the absence of available literature to support such identifications. After identification, specimens belonging to the same species were meticulously placed in distinct vials, each adequately labeled with sizes proportionate to the quantity of specimens. The voucher material resulting from this research is housed in the collections of both CZMA and MPEG. The comprehensive species list has been previously published by Silva et al. . In the scope of the current study, each collection day for each pig carcass in both the dry and rainy seasons was treated as an individual sampling unit. Initially, a Procrustes analysis was conducted to evaluate the correspondence between the collected flesh fly communities in the two distinct seasons, i.e., dry and rainy. Based on the outcome, if the communities gathered during the dry and rainy seasons displayed substantial congruence, implying similarity across the year, the analyses would be pursued in a combined manner. Conversely, if the communities demonstrated incongruity or weak connection, indicating dissimilarity over the period, they would be independently scrutinized, considering the collections from both the dry and rainy seasons. The Procrustes analysis involves a comparison of two data matrices by aligning them using an ordination and rotational adjustment algorithm, generating "M12" values that indicate the extent of concordance between the datasets . A reduced “M12” value indicates a higher level of congruence between the data matrices. To perform the Procrustes analysis, the two matrices containing data of flesh fly species from the dry and rainy seasons underwent a logarithmic transformation using the formula log (x + 1). Subsequently, a Bray–Curtis distance matrix was computed, since the abundance data of the species were used. The logarithmic transformation was implemented to mitigate biases induced by non-normal data within the biological data matrix . The Bray–Curtis distance matrix was employed for community ordination, facilitated by Principal Coordinates Analysis (PCoA) . The PCoA axes were used as input for the Procrustes analysis, utilizing the Monte Carlo permutation test (9999 permutations). To establish congruence between the flesh fly communities across the dry and rainy seasons, two criteria were implemented. Initially, a significance level of 5% was applied; if the obtained result was not statistically significant ( p > 0.05), the matrices would be deemed incongruent. Additionally, even when a significant relationship was present ( p < 0.05), the M12 value needed to be < 0.30 to indicate a substantial level of congruence between the matrices. To investigate alterations in the distribution patterns of visiting flesh flies species and to pinpoint their change points (cp) and distribution trends across the temporal gradient of pig carcass decomposition days, we employed the Threshold Indicator Taxa Analysis (TITAN). This analytical method identifies the change point (nCPA) of species abundance along an environmental gradient (in this case, the time of decomposition in days) and delineates the direction of this shift based on the scores derived from the Individual Indicator Value index ( IndVal ). Should a species exhibit heightened occurrence and increased abundance from the point of change towards the advanced stages of the carcass decomposition gradient, it is categorized as “Z + ”; conversely, if there is a surge in occurrence and abundance from the point of change towards the initial phases of the carcass decomposition gradient, it falls under the “Z− ” category , . The criteria employed to designate a species as Z + or Z− encompass purity and reliability levels of ≥ 90%, along with a p -value ≤ 0.05. These statistical values are ascertained through IndVal scores at each change point via bootstrap and permutation procedures. Less common species were omitted from TITAN analysis due to insufficient occurrences along the gradient to qualify as Z + or Z− indicators. Consequently, solely species exhibiting a minimum of three consecutive days of occurrence, without alternation, and present in at least one carcass, as well as species with more than five collected specimens were considered for inclusion in the TITAN analysis , . The analyses were carried out using the R software . Principal Coordinates Analysis (PCoA) was performed using the vegdist function from the Vegan package , and the cmdscale function from the Stats package . Procrustes analysis was conducted using the protest function from the vegan package . The TITAN analysis was executed using the TITAN function from the TITAN2 package . Throughout the duration of the carcass decomposition period, which spans approximately 10 days, a total of 10,819 flesh fly specimens were collected. Of these specimens, 40.07% were collected during the dry season, while the remaining 59.93% were gathered during the rainy season. Among the collected specimens, a total of 45 species were identified, with 42 species recorded during the dry season and 37 species during the rainy season. Thirty-four species were common in both seasons (Fig. ; Supplementary Tables S1 and S2 online). The congruence between the flesh fly communities during the dry and rainy seasons was found to be weak (M12 = 0.350, R = 0.807, p = 0.001; Fig. ). Additionally, it was qualitatively noted that among the 45 sampled species, 11 species (24%), were exclusively present during either the dry or rainy season (Fig. ; Supplementary Tables S1 and S2 online). Consequently, in light of this outcome, the analysis of flesh fly communities was performed separately for each respective season throughout the entirety of this study. Nine indicator species were identified using TITAN as representative of carcass decomposition time, comprising three species during the dry season and six species during the rainy season. Specifically, Oxysarcodexia thornax (Walker, 1849) (Change Point: cp = 2.5 days of carcass decomposition), Peckia ( Sarcodexia ) lambens (Wiedemann, 1830) (cp = 3 days), and Ravinia belforti (Prado & Fonseca, 1932) (cp = 2.5 days) were categorized as Z + , signifying an augmentation in occurrence and abundance during the intermediate decomposition phases (Table and Supplementary Table online; Fig. ). No Z− indicator species were found during the dry season. Throughout the rainy season, several species exhibited distinct temporal patterns. Peckia ( Euboettcheria ) collusor (Curran & Walley, 1934) (cp = 2.5 days), R. belforti (cp = 4 days), Tricharaea ( Sarcophagula ) canuta (Wulp, 1896) (cp = 5.5 days), and Tricharaea ( Sarcophagula ) occidua (Fabricius, 1794) (cp = 3 days) were categorized as Z + , showing an increase in occurrence and abundance towards the intermediate stages of decomposition. Conversely, Dexosarcophaga carvalhoi (Lopes, 1980) (cp = 6.5 days) and Peckia ( Sarcodexia ) tridentata (Hall, 1937) (cp = 6.5 days) were classified as Z− , demonstrating increased occurrence and abundance prior to the change point, namely towards the earlier stages of decomposition (Table and Supplementary Table online; Fig. ). This study delved into the succession dynamics of flesh flies on pig carcasses, highlighting distinct sets of indicator species for both the initial and advanced stages of decomposition. Given the marked climatic seasonality within the Maranhão Cerrado region, it became evident that these indicators must be identified separately for the dry and rainy seasons due to inherent variations in flesh fly communities across these periods. Season-specific climatic factors such as temperature, precipitation, and relative humidity play a pivotal role in shaping the cadaver's decay process , thereby influencing the composition of flesh fly communities during each season. Additionally, the limited congruence observed in flesh fly communities across the two seasons underscores the necessity of employing tailored indicator species for accurate assessment in each distinct season. Eight species of flesh flies ( D. carvalhoi , O. thornax , P. ( E .) collusor , P. ( S .) lambens , R. belforti , T . ( S .) canuta , T . ( S .) occidua , and P. ( S .) tridentata ) were identified as indicators of ecological succession along the gradient of carcass decomposition days. These species, except for P . ( S .) tridentata , exhibit extensive distribution not only within various biomes and regions of Brazil, but also across different countries in the Americas , – , , , , – . This widespread distribution underscores their potential applicability across a diverse range of geographical areas. Peckia ( S .) tridentata has been recorded in the two largest biomes of the Neotropical region, namely the Cerrado and the Amazon Rainforest , , , , – . This observation implies a more limited potential for its role as an indicator within these particular biomes. The flesh flies were observed throughout the entire process of carcass decomposition, from the beginning until complete skeletonization . However, only a select few species were identified as being closely associated with specific stages of decomposition. In total, five Z + species were identified, all belonging to the genera Oxysarcodexia Townsend, 1917, Peckia Robineau-Desvoidy, 1830, Ravinia Robineau-Desvoidy, 1863, and Tricharaea Thomson, 1869. These genera encompass numerous species of recognized forensic significance , , , , , , , . Notably, these species exhibited change points and increased abundance values during the initial stages of the carcass decomposition gradient. This occurred within only 2.5 days for the species O. thornax , R. belforti (in the dry season), and P . ( E .) collusor (in the rainy season); within 3 days for P . ( S. ) lambens (in the dry season) and T . ( S .) occidua (in the rainy season); and within 4 days for R. belforti (in the rainy season). The identification of the aforementioned species during the initial stage of the decomposition process, spanning from 2 to 3 days, aligns with the phase characterized by carcass swelling and the noticeable distension and separation of the legs , . By the fourth day, the swelling has subsided, and subtle cracks begin to form on the body. Additionally, this period marks the initial appearance of numerous larvae on the carcass . Typically, on the initial day of decomposition, carbon dioxide, a non-flammable gas, is released due to the activity of aerobic bacteria. Subsequently, between the second and fourth days of decomposition, flammable gases such as hydrogen and hydrocarbons emerge, resulting from the combined metabolic activities of both aerobic and anaerobic bacteria . Consequently, the established change point thresholds for these species substantiate their responsiveness to the morphophysiological alterations inherent to the process of pig carcass decomposition. Ravinia belforti was the only species displaying increased abundance values beyond the change point in both the dry and rainy seasons, making this observation particularly noteworthy. This finding holds substantial implications, as R. belforti could potentially serve as an indicator for estimating the post-mortem interval of cadavers located in Cerrado areas during both climatic periods. However, a distinction existed in the change point days for R. belforti between the two seasons. In the rainy season, the change point occurred on the fourth day, whereas in the dry season, it manifested at 2.5 days. This variation likely stems from differences in temperature, precipitation, and relative humidity experienced during these seasons, thereby influencing the physicochemical transformations inherent to carcass decomposition. Hence, in the dry season, the carcasses provided conditions conducive to the establishment of this species earlier than in the rainy season . Among the Z + species, T . ( S .) canuta exhibited an increase in its abundance values only from 5.5 days onward, a pattern observed during the intermediate phase of carcass decomposition. This specific behavior could potentially be associated to an affinity with the presence of non-flammable gases, such as ammonia and nitrogen, which start to accumulate from the fifth day of decomposition . On this day, noticeable changes were observed in the carcasses, including a visible reduction in body mass compared to preceding days, the appearance of wrinkles in the ventral region, and the emergence of fungal growth. Furthermore, the skeletal components of the body, such as the snout and limbs, began to be exposed. The Z− species identified during the rainy season, categorized under the genera Dexosarcophaga , Townsend, 1917 ( D . carvalhoi ) and Peckia ( P . ( S .) tridentata ), exhibited their highest abundance values when the carcasses had reached an advanced stage of decomposition, with only 1.5 days remaining until they reached the skeletal stage , . Among these species, P. ( S .) tridentata was the sole representative within the three Peckia species displaying the Z− indicator trait. This distinct behavior in comparison to other species within the same genus can be attributed to significant variations in the immature and adult life history of Peckia species. These variations encompass diverse roles, spanning from sarcosaprophages, coprophages, frugivores, omnivores, to parasites of both vertebrates and invertebrates , , , , . Six out of the eight indicator species identified in this study are commonly found and have been observed in forensic studies involving pig carcasses within Cerrado areas of the Central-West and Southeast regions of Brazil ( D. carvalhoi , O. thornax , P . ( E .) collusor , P . ( S .) lambens , R. belforti , and T . ( S .) occidua ) , . Both Barros et al. and Paseto et al. collected specimens of these species from carcasses at comparable stages of decomposition to those observed in the present study, with similar change points and trends in increased abundances. This congruence reinforces our findings and underscores the robust potential of these species as reliable forensic indicators for the Brazilian Cerrado ecosystem. From an ecological perspective, the presence of a species at a specific place and period of the decomposition process adheres to the principles outlined in the Niche Theory, which underscores the association between species distribution and the availability of conditions and resources aligned with their physiological needs . In addition, environmental, spatial and/or temporal heterogeneity enables the coexistence of sarcosaprophagous insect species in local communities, visitors and/or colonizers of carcasses and corpses, being mediated by interactions such as predation, parasitism, aggregation, intra and interspecific competition and dispersal capacity of inferior competitors . Consequently, it is reasonable to anticipate that species to be present and exhibit higher abundance when conditions and resources closely align with their optimal requirements for establishment and development . These conditions and resources tend to transition gradually rather than abruptly across time or space . As a result, analyzing such data in a continuous manner is most suitable, minimizing the loss of information about the ecological succession of fly species in the process of decomposing carcasses and cadavers , . Thus, the adoption of methodologies like TITAN for the study of flesh fly species on pig carcasses is expected to represent the natural fluctuations of these species more accurately across temporal gradients of decomposition compared to analyses based on predefined temporal stages. Hence, this study undertook a comprehensive examination of the ecological succession of flesh fly species (both Z + and Z− ) across the temporal gradient of pig carcass decomposition during both the dry and rainy seasons within the Cerrado biome of Northeastern Brazil. The Z + species demonstrated unique change points, indicating a substantial surge in their prevalence and abundance during the initial phases of decomposition. Conversely, the Z− species exhibited later change points, showcasing heightened occurrences and levels of abundance in the later stages of decomposition. Considering the body proportions of the carcasses analyzed in this study in relation to the cadavers and taking into considering environmental factors, such as climate and phytophysiognomy, which vary greatly between regions and environments, which can alter the ecological succession pattern of the flesh fly species in the temporal gradient of carcass decomposition, we propose the combined utilization of the identified species to enhance the estimation of post-mortem interval in forensic studies. Within this context, noteworthy among them are the Z + species O. thornax , R. belforti, and P . ( S .) lambens , exhibiting potential to estimate the early stages (1–4 days) of post-mortem interval during the dry season. During the rainy season, three Z + species ( P . ( E .) collusor , T . ( S .) occidua , and R . belforti ) could be used for estimating the initial days (1–4 days); whereas two Z + species ( R . belforti and T . ( S .) canuta ), along with two Z− species ( P . ( S .) tridentata and D . carvalhoi ), demonstrate potential to estimate it in intermediate days (4–6 days). Finally, this assemblage of species, excluding R . belforti , holds utility in estimating the advanced stages (from 6 days onwards). In general, the information provided by the present study confirms the existence of an ecological succession of flesh fly species along the gradient of days of decomposition of pig carcasses in Cerrado areas in Northeast Brazil. These data are of great importance for forensic entomology in the medico-legal area, as they can be used as a complement by forensic experts in the elucidation of cases involving the estimation of the post-mortem interval of cadavers. Supplementary Information. |
Surgeons and preventive health: a mixed methods study of current practice, beliefs and attitudes influencing health promotion activities amongst public hospital surgeons | 5ee0d9d6-c5ed-4f9b-ae83-29e75f906f78 | 6555744 | Preventive Medicine[mh] | Chronic non-communicable diseases are the foremost cause of preventable illness, disability and death worldwide . Smoking, diet, and insufficient physical activity are the primary behavioural risk factors behind preventable chronic diseases . The increased prevalence of chronic diseases has influenced demands on the health system , with chronic diseases leading to hospitalisations, long-term disability, and rehabilitation costs . Accordingly, hospitals need to broaden their role from their primary focus on disease treatment towards a position of more integrated health promotion . Hospitals are well situated to play a key role in the delivery of preventive health . As hospital clinicians, surgeons have an important role in advocating for behaviour change for patients with, or at risk of, chronic disease . Due to their extensive medical training and specialisation, surgeons are regarded as reliable sources of medical advice, extending beyond their expertise in surgical care . Surgery is considered a major life event , and individuals are more susceptible to behaviour change in the face of such an event . Surgeons therefore, have potential to be influential in the promotion of lifestyle behaviour change . Surgeons undertake high volumes of non-admitted consultations annually, which provides opportunities to address preventive health directly during routine clinical interactions . In Australian hospitals alone, over 2.2 million elective admissions involving surgery were undertaken in 2015–2016 . In spite of this, there is a scarcity of research investigating preventive health practice in non-admitted surgical practice. Studies examining lifestyle risk management (smoking cessation and/or physical activity promotion) delivered by hospital doctors have consisted largely of cross-sectional studies of self-reported practice . Only two of these studies, which both focused on oncology patients, included hospital surgeons . Findings in all studies demonstrated low rates of preventive health interventions . In addition, hospital doctors report low levels of confidence in their ability to assist patients with health behaviour change and uncertainty over the effectiveness of behaviour change advice . In the studies , no interviews were carried out to probe the survey findings and understand the beliefs and attitudes that might explain the low levels of preventive health interventions undertaken. Given the prevalence of chronic disease and the necessity for hospitals to move to a position of more integrated preventive health practice , it is important to gain insights from hospital surgeons due to the influence they may exert on patient behaviour . Surgeons are clinical leaders with responsibility for clinical performance as well as clinical policy and practice . As such, surgeons maintain autonomy over practice standards , and little is known about the opinions of this professional group concerning preventive health practice. The depth of insight gained from the study of surgeons might offer distinctive perspectives on current preventive health practice and the attitudes and beliefs of these highly professionalised clinicians relating to implementing preventive health into non-admitted surgical practice. Therefore, the aim of this study was to identify which preventive health activities surgeons carry out in non-admitted public hospital clinics and to explore the attitudes of the profession towards preventive health practice. This study used a mixed-methods design to identify which preventive health activities surgeons carry out in non-admitted public hospital clinics and to explore the attitudes of the profession towards preventive health practice. We integrated mixed-methods at the design level using a sequential explanatory design . This two-stage design began with a self-reported clinician survey investigating surgeons’ actual participation in preventive health activities (Fig. ). This was followed by the subsequent collection and analysis of in-depth interviews with surgeons to gain insight into the attitudes of surgeons towards undertaking preventive health activities in non-admitted settings. The protocol for this study has been detailed previously . Ethical approval for the study was gained from the human research and ethics committee of the participating hospital and the associated university. Participants This study targeted all surgeons and their registrars consulting in an elective outpatient clinic of a major tertiary hospital in regional Australia. Participation was offered to all practicing surgeons (general and orthopaedic; n = 20) and registrars ( n = 11) between June 2017 and August 2018. The recruitment strategy has been described elsewhere . In brief, an email containing the link to the clinician survey was sent to all potential participants by their head of department. Participants were informed prior to commencing the survey that informed consent was implied by completing the survey. For the interviews, surgeons were approached individually, in the non-admitted clinic by a project officer to discuss participation. Informed consent was sought from all participants prior to completing the interview. Clinician survey The clinician survey requested detail on surgical practice, including surgical speciality, and number of year of practice. Participants were asked to self-report on proportions of patients who they screened for behavioural risk factors (smoking, diet, physical activity and alcohol), provided verbal and/or written advice and referred to other services for support in changing risk factors (Additional file ). The survey measured surgeons’ knowledge and confidence in screening and managing risk factors, as well as attitudinal measures relating to the delivery of preventive health interventions in surgical care. All survey items were measured on a 5-point Likert scale. Data were analysed using IBM SPSS Statistics for Windows (Version 25; IBM Corp., USA). Semi-structured interviews Following analysis of the clinician survey, face-to-face interviews were conducted with a purposeful sample of surgeons and registrars ( n = 14). Maximum variation sampling was used to ensure that a heterogeneous sample was recruited, to capture the perspectives of orthopaedic and general surgeons and registrars to search for variation in perspectives . In total, 21 surgeons were asked to participate in the interviews, with a participation rate of 71%. We employed a building approach to mixed-methods integration, using the results from our quantitative analyses to inform the data collection of the qualitative component . Quantitative data were used to develop the interview guide (Table ). Interviews were conducted by the first author and covered issues related to preventive health practice in routine practice and the attitudes of surgeons towards preventive health practice. All interviews were audio-taped with participants’ permission and transcribed verbatim by the first author for thematic analysis . Field notes were used to supplement the audio and transcripts to inform the iterative development of interview guides and question-related probes for subsequent interviews. Analyses From the clinician survey, surgeons’ implementation rates in preventive health activities (assessing risk factors, proving information and making referrals) were classified as high, medium or low . High implementation rates defined screening and/or intervention scores in the fourth quartile for responding surgeons. Low implementation rates defined screening and/or intervention scores less than or equal to the first quartile for responding surgeons. Quartile cut-off points were also included for surgeon confidence, knowledge, and attitudinal measures. Spearman’s rank-order correlations were performed to assess the relationships between the dependent variables (preventive health practice) and independent variables (confidence and knowledge in preventive health practice, years of practice, and attitudinal factors). Following this, a Generalized Estimating Equation (GEE) was used to model the associations between independent variables and preventive health practice . The GEE indicates which variables, when added to the model, best predict preventive health practice. Goodness of fit for the GEE model was assessed using the quasi-likelihood under independence model criterion (QIC) . The QIC is a statistic for model selection for GEE models, where lower values indicate better model fit to the data . Data from in-depth interviews were collected and analysed concurrently. Qualitative description was used as the theoretical framework for the qualitative component . Qualitative description provides straightforward, rich descriptions of experiences or events in a language similar to the participant’s own . Transcribed transcripts were analysed and coded line-by-line using the qualitative data analysis software NVivo 10.0 (QSR International, Cambridge, MA, USA). Codes were derived from data rather than being determined beforehand, and a coding scheme was applied to the interview text. Coded text was grouped into more general categories, which were reviewed by the research team and merged into themes to help explain the factors that influence surgeons’ participation in health promotion activities . Two authors (SB 1 and AS) independently coded and analysed the data. To improve reliability and to reach consensus, two additional authors (MK and SB 2 ) reviewed the codebook and samples of transcripts. No new information was found between the twelfth and thirteenth interview, indicating that data saturation was reached by the twelfth interview . To ensure data saturation, one additional participant was interviewed. As this additional interview did not bring forward new information, data saturation was deemed to have occurred , and interviewing was ceased. This study targeted all surgeons and their registrars consulting in an elective outpatient clinic of a major tertiary hospital in regional Australia. Participation was offered to all practicing surgeons (general and orthopaedic; n = 20) and registrars ( n = 11) between June 2017 and August 2018. The recruitment strategy has been described elsewhere . In brief, an email containing the link to the clinician survey was sent to all potential participants by their head of department. Participants were informed prior to commencing the survey that informed consent was implied by completing the survey. For the interviews, surgeons were approached individually, in the non-admitted clinic by a project officer to discuss participation. Informed consent was sought from all participants prior to completing the interview. The clinician survey requested detail on surgical practice, including surgical speciality, and number of year of practice. Participants were asked to self-report on proportions of patients who they screened for behavioural risk factors (smoking, diet, physical activity and alcohol), provided verbal and/or written advice and referred to other services for support in changing risk factors (Additional file ). The survey measured surgeons’ knowledge and confidence in screening and managing risk factors, as well as attitudinal measures relating to the delivery of preventive health interventions in surgical care. All survey items were measured on a 5-point Likert scale. Data were analysed using IBM SPSS Statistics for Windows (Version 25; IBM Corp., USA). Following analysis of the clinician survey, face-to-face interviews were conducted with a purposeful sample of surgeons and registrars ( n = 14). Maximum variation sampling was used to ensure that a heterogeneous sample was recruited, to capture the perspectives of orthopaedic and general surgeons and registrars to search for variation in perspectives . In total, 21 surgeons were asked to participate in the interviews, with a participation rate of 71%. We employed a building approach to mixed-methods integration, using the results from our quantitative analyses to inform the data collection of the qualitative component . Quantitative data were used to develop the interview guide (Table ). Interviews were conducted by the first author and covered issues related to preventive health practice in routine practice and the attitudes of surgeons towards preventive health practice. All interviews were audio-taped with participants’ permission and transcribed verbatim by the first author for thematic analysis . Field notes were used to supplement the audio and transcripts to inform the iterative development of interview guides and question-related probes for subsequent interviews. From the clinician survey, surgeons’ implementation rates in preventive health activities (assessing risk factors, proving information and making referrals) were classified as high, medium or low . High implementation rates defined screening and/or intervention scores in the fourth quartile for responding surgeons. Low implementation rates defined screening and/or intervention scores less than or equal to the first quartile for responding surgeons. Quartile cut-off points were also included for surgeon confidence, knowledge, and attitudinal measures. Spearman’s rank-order correlations were performed to assess the relationships between the dependent variables (preventive health practice) and independent variables (confidence and knowledge in preventive health practice, years of practice, and attitudinal factors). Following this, a Generalized Estimating Equation (GEE) was used to model the associations between independent variables and preventive health practice . The GEE indicates which variables, when added to the model, best predict preventive health practice. Goodness of fit for the GEE model was assessed using the quasi-likelihood under independence model criterion (QIC) . The QIC is a statistic for model selection for GEE models, where lower values indicate better model fit to the data . Data from in-depth interviews were collected and analysed concurrently. Qualitative description was used as the theoretical framework for the qualitative component . Qualitative description provides straightforward, rich descriptions of experiences or events in a language similar to the participant’s own . Transcribed transcripts were analysed and coded line-by-line using the qualitative data analysis software NVivo 10.0 (QSR International, Cambridge, MA, USA). Codes were derived from data rather than being determined beforehand, and a coding scheme was applied to the interview text. Coded text was grouped into more general categories, which were reviewed by the research team and merged into themes to help explain the factors that influence surgeons’ participation in health promotion activities . Two authors (SB 1 and AS) independently coded and analysed the data. To improve reliability and to reach consensus, two additional authors (MK and SB 2 ) reviewed the codebook and samples of transcripts. No new information was found between the twelfth and thirteenth interview, indicating that data saturation was reached by the twelfth interview . To ensure data saturation, one additional participant was interviewed. As this additional interview did not bring forward new information, data saturation was deemed to have occurred , and interviewing was ceased. In total, 16 surgeons completed the survey (response rate of 51%) and interviews were carried out with 14 surgeons (participation rate 71%). The surgeons that participated in the interviews were broadly representative of those completing the survey (Table ). The majority of surgeons worked full-time, and three quarters of surgeons in both the survey and interviews were male. The results of the quantitative and qualitative components are reported in separate sections, using a contiguous narrative approach to integration of mixed-methods data . Clinician survey Table provides preventive health practice rates and attitudes to preventive health amongst the responding surgeons. Overall, all surgeons carried out some preventive health activities, however the majority of surgeons did this at low levels. Asking patients about behavioural risk factors and providing verbal advice were the most undertaken preventive health interventions. Only 2 surgeons reported providing patients with written advice, and 3 surgeons reported having referred patients to other service providers for help with risk factor management. The surgeons self-reported knowledge and confidence in addressing behavioural risk factors was medium to high. In the Spearman’s correlation, significant positive correlations were observed between preventive health practice and clinician confidence (r = 0.635, p = 0.008), knowledge (r = 0.544, p = 0.029), perceived effectiveness of preventive health practice (r = 0.710, p = 0.002), the importance placed on addressing lifestyle changes (r = 0.655, p = 0.006), and the work priority placed on addressing lifestyle changes with patients (r = 0.644, p = 0.007). The GEE model found two factors that together, significantly predicted tendency to undertake preventive health interventions, including number of years of clinical practice (β = 0.26, p = 0.041) and work priority (β = 1.22, p = 0.008) (Table ). The addition of work priority to the model decreased the QIC from 1063 to 736, indicating a more robust fit of data to the model. The lower QIC indicates that the model, with the addition of work priority, contains the best subset of explanatory variables to predict the surgeons undertaking of preventive health interventions. In-depth interviews Four themes were found to influence surgeons’ preventive health practice. The themes, which all centred around the clinical consultation, included: surgeon’s perceptions of their role in preventive health, perceived motivation of patients, the hospital structure, and facilitating factors. The codes, categories and themes are described in Additional file . These themes are expanded upon below using verbatim quotes from participants for illustrative purposes. Additional verbatim quotes for each theme are provided in Additional file . The role of the surgeon in preventive health All surgeons considered preventive health to be important for health. However, the perceived importance did not translate to high rates of preventive health practice. Surgeons who reported undertaking behaviour change discussions with patients reported that their role was to address behaviour change in relation to specific surgical practice, rather than a holistic wellbeing perspective; for example, smoking cessation was advocated to decrease the risk of infection. “I’ll be telling them to either cut back on the smoking or try to aim quitting if it’s possible, at least for the surgery, and then after that if they can continue then great; if they can’t then at least for the time period for the surgery if they can do that that would be great” (Surgeon 3). Some surgeons felt that behaviour change is not part of the role of the surgeon; that surgeons are clinical specialists who have a priority to treat specialist problems, thereby delivering services that no other clinician can, as suggested by the following quote: “So we know what you need to do, but you know, we are trained to do surgery. And other people can’t do that, and that’s what we need to do. If you make us do all of this other stuff, then it’s not a particularly effective use of surgeon time” (Surgeon 9). Viewing themselves as specialist practitioners, surgeons believe their role is best suited to focusing on presenting conditions, rather than taking a holistic approach to the person. “Probably because surgeons don’t feel that it is their job to do that; they are referred a patient for a [specific problem], and they are concentrating on treating that … . they probably zoom in on that pathology rather than looking at the patient as a whole” (Surgeon 13). The motivation of the patients Patient motivation and acceptance of behaviour change interventions by the patient was another important theme. The surgeons described patient acceptance of lifestyle interventions as an important factor that influenced the surgeons’ participation in preventive health practice. Consistent with existing literature , some surgeons reported that patients are not opposed to surgeons raising lifestyle discussions during consultations. This advice however, doesn’t necessarily translate to actual behaviour change, with surgeons reporting that many patients maintain their behaviours despite the provision of advice. “I think a good percentage of patients, you say ‘you need to quit smoking or if you don’t, your risk of infection is higher’ and they are like, ‘ah yeah, whatever’ and you see plenty where you have no response” (Surgeon 1). Surgeons highlighted that the persistence of risky health behaviour by patients despite health advice was a source of frustration, which might decrease the likelihood of surgeons engaging in preventive health activities in the non-admitted setting. “And we say ‘you have to cut down smoking’ and every time they come back to clinic and we go ‘have you cut down on smoking?’ and they are still smoking. I think after a while, you just … you are talking to a brick wall” (Surgeon 3). “I mean smoking we all know in particular when it comes to wound healing and infection, that’s something we all know is not good. However, I don’t always advise them to stop because, I don’t know, I often don’t think they will stop” (Surgeon 6) The surgeons reported that not all patients are appreciative of discussing their lifestyles during surgical consultations. In the surgeons’ experience, many patients attend the consultation seeking specialist advice relating to a particular issue, and are not seeking generalist advice about health behaviours. Previous experience of negative patient reactions may contribute to the surgeons’ narrowing the focus of the consultation to that of the presenting condition only. “ … most people don’t want to talk to a surgeon in an outpatient clinic about their overall wellbeing. They have come here for a problem, so it [consultation] needs to be problem-focused” (Surgeon 12). The hospital structure Surgeons work in busy public hospital clinics that have extensive waiting lists. Surgeons are responsible for clinic throughput, and are accountable to management on such performance indicators. The pressure for volume is a barrier to holistic care. Working under time pressure, curative care is prioritised over preventive health. This issue is compounded by the complexity of the hospital system where there is a disconnect between the absence of preventive health in outpatient clinics and sub-optimal post-operative surgical outcomes. “ … the holistic approach, then probably my extra 5 minutes doing that referral is in the patient’s best interest, I get that. But hospitals don’t always look at the whole picture, they look at the bottom line for them; so the people in clinic here, running the clinic, will be looking at their targets … they don’t really care what they spend up on the surgical ward when the patient gets a wound infection that might be preventable if they weren’t a smoker” (Surgeon 13). One surgeon changed their practice by limiting the number of patients seen in the clinic, allowing increased time with patients. “We have limited it to 18 patients between the 2 of us, and so it’s good, there is enough time if you have enough reviews which are quick, and ‘new’ which are not. Yes, there is enough time [for preventive health]” (Surgeon 7). This surgeon acknowledged that, as a senior hospital clinician, the clinical and institutional influence afforded to him may have permitted such a policy change, but this is ultimately at the discretion of hospital administrators. With respect to generating referrals, surgeons highlighted an absence of specific programs for general behaviour change in the hospital, as well as poor awareness of behaviour change programs in the community. “For people who have been inpatients there [are] options, but there aren’t a lot of things for just young inactive people unfortunately” (Surgeon 12). “I don’t know how you would actually make the referral [to community programs]” (Surgeon 2). Surgeons who refer to internal services, such as exercise physiologists and physiotherapy, use this pathway to address issues that relate to surgical outcomes, increasing muscle strength before surgery for example, rather than increasing physical activity for general health. Surgeons also forego preventive referrals to allied health practitioners due to the demand for rehabilitation services. “ … from a public health system, it’s hard to get them involved in exercise programs. Physiotherapists are often very busy and overworked, and they can’t just be doing exercises with them” (Surgeon 11). Facilitators experienced by surgeons Surgeons were unanimous in their desire for information to give to patients that are specific to their needs (i.e., smoking specific or physical activity specific). The majority of surgeons felt that a referral pathway into specialist behaviour change services, either internally or externally is required to facilitate successful behaviour change. “A flyer would be good … . and I say put that on your fridge, something like that, where you see it every day, and you think ‘oh, the specialist gave it to me’” (Surgeon 8). The surgeons felt that if they had dedicated resources, or referral pathways to offer patients, then they could use their clinical influence to stress the importance of behaviour change, which might increase the likelihood of patients using these services. “I think a clinician handing it [referral] to them, and underlying who they need to see would be much more effective” (Surgeon 13). “We need to be able to say ‘you need to make this change, here is someone who can help’. But it comes from the surgeon as the authorizing environment” (Surgeon 12). Table provides preventive health practice rates and attitudes to preventive health amongst the responding surgeons. Overall, all surgeons carried out some preventive health activities, however the majority of surgeons did this at low levels. Asking patients about behavioural risk factors and providing verbal advice were the most undertaken preventive health interventions. Only 2 surgeons reported providing patients with written advice, and 3 surgeons reported having referred patients to other service providers for help with risk factor management. The surgeons self-reported knowledge and confidence in addressing behavioural risk factors was medium to high. In the Spearman’s correlation, significant positive correlations were observed between preventive health practice and clinician confidence (r = 0.635, p = 0.008), knowledge (r = 0.544, p = 0.029), perceived effectiveness of preventive health practice (r = 0.710, p = 0.002), the importance placed on addressing lifestyle changes (r = 0.655, p = 0.006), and the work priority placed on addressing lifestyle changes with patients (r = 0.644, p = 0.007). The GEE model found two factors that together, significantly predicted tendency to undertake preventive health interventions, including number of years of clinical practice (β = 0.26, p = 0.041) and work priority (β = 1.22, p = 0.008) (Table ). The addition of work priority to the model decreased the QIC from 1063 to 736, indicating a more robust fit of data to the model. The lower QIC indicates that the model, with the addition of work priority, contains the best subset of explanatory variables to predict the surgeons undertaking of preventive health interventions. Four themes were found to influence surgeons’ preventive health practice. The themes, which all centred around the clinical consultation, included: surgeon’s perceptions of their role in preventive health, perceived motivation of patients, the hospital structure, and facilitating factors. The codes, categories and themes are described in Additional file . These themes are expanded upon below using verbatim quotes from participants for illustrative purposes. Additional verbatim quotes for each theme are provided in Additional file . The role of the surgeon in preventive health All surgeons considered preventive health to be important for health. However, the perceived importance did not translate to high rates of preventive health practice. Surgeons who reported undertaking behaviour change discussions with patients reported that their role was to address behaviour change in relation to specific surgical practice, rather than a holistic wellbeing perspective; for example, smoking cessation was advocated to decrease the risk of infection. “I’ll be telling them to either cut back on the smoking or try to aim quitting if it’s possible, at least for the surgery, and then after that if they can continue then great; if they can’t then at least for the time period for the surgery if they can do that that would be great” (Surgeon 3). Some surgeons felt that behaviour change is not part of the role of the surgeon; that surgeons are clinical specialists who have a priority to treat specialist problems, thereby delivering services that no other clinician can, as suggested by the following quote: “So we know what you need to do, but you know, we are trained to do surgery. And other people can’t do that, and that’s what we need to do. If you make us do all of this other stuff, then it’s not a particularly effective use of surgeon time” (Surgeon 9). Viewing themselves as specialist practitioners, surgeons believe their role is best suited to focusing on presenting conditions, rather than taking a holistic approach to the person. “Probably because surgeons don’t feel that it is their job to do that; they are referred a patient for a [specific problem], and they are concentrating on treating that … . they probably zoom in on that pathology rather than looking at the patient as a whole” (Surgeon 13). All surgeons considered preventive health to be important for health. However, the perceived importance did not translate to high rates of preventive health practice. Surgeons who reported undertaking behaviour change discussions with patients reported that their role was to address behaviour change in relation to specific surgical practice, rather than a holistic wellbeing perspective; for example, smoking cessation was advocated to decrease the risk of infection. “I’ll be telling them to either cut back on the smoking or try to aim quitting if it’s possible, at least for the surgery, and then after that if they can continue then great; if they can’t then at least for the time period for the surgery if they can do that that would be great” (Surgeon 3). Some surgeons felt that behaviour change is not part of the role of the surgeon; that surgeons are clinical specialists who have a priority to treat specialist problems, thereby delivering services that no other clinician can, as suggested by the following quote: “So we know what you need to do, but you know, we are trained to do surgery. And other people can’t do that, and that’s what we need to do. If you make us do all of this other stuff, then it’s not a particularly effective use of surgeon time” (Surgeon 9). Viewing themselves as specialist practitioners, surgeons believe their role is best suited to focusing on presenting conditions, rather than taking a holistic approach to the person. “Probably because surgeons don’t feel that it is their job to do that; they are referred a patient for a [specific problem], and they are concentrating on treating that … . they probably zoom in on that pathology rather than looking at the patient as a whole” (Surgeon 13). Patient motivation and acceptance of behaviour change interventions by the patient was another important theme. The surgeons described patient acceptance of lifestyle interventions as an important factor that influenced the surgeons’ participation in preventive health practice. Consistent with existing literature , some surgeons reported that patients are not opposed to surgeons raising lifestyle discussions during consultations. This advice however, doesn’t necessarily translate to actual behaviour change, with surgeons reporting that many patients maintain their behaviours despite the provision of advice. “I think a good percentage of patients, you say ‘you need to quit smoking or if you don’t, your risk of infection is higher’ and they are like, ‘ah yeah, whatever’ and you see plenty where you have no response” (Surgeon 1). Surgeons highlighted that the persistence of risky health behaviour by patients despite health advice was a source of frustration, which might decrease the likelihood of surgeons engaging in preventive health activities in the non-admitted setting. “And we say ‘you have to cut down smoking’ and every time they come back to clinic and we go ‘have you cut down on smoking?’ and they are still smoking. I think after a while, you just … you are talking to a brick wall” (Surgeon 3). “I mean smoking we all know in particular when it comes to wound healing and infection, that’s something we all know is not good. However, I don’t always advise them to stop because, I don’t know, I often don’t think they will stop” (Surgeon 6) The surgeons reported that not all patients are appreciative of discussing their lifestyles during surgical consultations. In the surgeons’ experience, many patients attend the consultation seeking specialist advice relating to a particular issue, and are not seeking generalist advice about health behaviours. Previous experience of negative patient reactions may contribute to the surgeons’ narrowing the focus of the consultation to that of the presenting condition only. “ … most people don’t want to talk to a surgeon in an outpatient clinic about their overall wellbeing. They have come here for a problem, so it [consultation] needs to be problem-focused” (Surgeon 12). Surgeons work in busy public hospital clinics that have extensive waiting lists. Surgeons are responsible for clinic throughput, and are accountable to management on such performance indicators. The pressure for volume is a barrier to holistic care. Working under time pressure, curative care is prioritised over preventive health. This issue is compounded by the complexity of the hospital system where there is a disconnect between the absence of preventive health in outpatient clinics and sub-optimal post-operative surgical outcomes. “ … the holistic approach, then probably my extra 5 minutes doing that referral is in the patient’s best interest, I get that. But hospitals don’t always look at the whole picture, they look at the bottom line for them; so the people in clinic here, running the clinic, will be looking at their targets … they don’t really care what they spend up on the surgical ward when the patient gets a wound infection that might be preventable if they weren’t a smoker” (Surgeon 13). One surgeon changed their practice by limiting the number of patients seen in the clinic, allowing increased time with patients. “We have limited it to 18 patients between the 2 of us, and so it’s good, there is enough time if you have enough reviews which are quick, and ‘new’ which are not. Yes, there is enough time [for preventive health]” (Surgeon 7). This surgeon acknowledged that, as a senior hospital clinician, the clinical and institutional influence afforded to him may have permitted such a policy change, but this is ultimately at the discretion of hospital administrators. With respect to generating referrals, surgeons highlighted an absence of specific programs for general behaviour change in the hospital, as well as poor awareness of behaviour change programs in the community. “For people who have been inpatients there [are] options, but there aren’t a lot of things for just young inactive people unfortunately” (Surgeon 12). “I don’t know how you would actually make the referral [to community programs]” (Surgeon 2). Surgeons who refer to internal services, such as exercise physiologists and physiotherapy, use this pathway to address issues that relate to surgical outcomes, increasing muscle strength before surgery for example, rather than increasing physical activity for general health. Surgeons also forego preventive referrals to allied health practitioners due to the demand for rehabilitation services. “ … from a public health system, it’s hard to get them involved in exercise programs. Physiotherapists are often very busy and overworked, and they can’t just be doing exercises with them” (Surgeon 11). Surgeons were unanimous in their desire for information to give to patients that are specific to their needs (i.e., smoking specific or physical activity specific). The majority of surgeons felt that a referral pathway into specialist behaviour change services, either internally or externally is required to facilitate successful behaviour change. “A flyer would be good … . and I say put that on your fridge, something like that, where you see it every day, and you think ‘oh, the specialist gave it to me’” (Surgeon 8). The surgeons felt that if they had dedicated resources, or referral pathways to offer patients, then they could use their clinical influence to stress the importance of behaviour change, which might increase the likelihood of patients using these services. “I think a clinician handing it [referral] to them, and underlying who they need to see would be much more effective” (Surgeon 13). “We need to be able to say ‘you need to make this change, here is someone who can help’. But it comes from the surgeon as the authorizing environment” (Surgeon 12). This mixed-methods study identified which preventive health activities surgeons carry out in non-admitted public hospital clinics, and explored the attitudes of these professionals towards preventive health practice. The quantitative data suggests that surgeons carried out preventive health interventions at low levels. Face-to-face conversations with patients about behavioural risk factors was the most commonly undertaken intervention. Surgeons were unlikely to provide written advice or refer patients to additional health behaviour change services. Although a number of attitudinal factors individually correlated with rates of preventive health practice undertaken, collectively, years of practice and the work priority placed on addressing lifestyle change were the strongest predictors of preventive health practice identified in the quantitative analyses. The qualitative analysis identified several individual and institutional topics that influenced surgeons undertaking of preventive health practice in non-admitted clinical care, with surgeons preferencing referral pathways into specialist programs to assist patients with behaviour change. In contrast to previous research , lack of knowledge or confidence were not identified as barriers to preventive health practice. Although surgeons’ knowledge and confidence were independently associated with levels of preventive health practice in the quantitative analyses, these variables did not contribute to the model that best predicted preventive health practice rates. The quantitative analysis also highlighted that how important surgeons believe it is to address lifestyle changes with patients was independently associated with preventive health practice in the quantitative analyses, though this did not contribute to the model that best predicted preventive health practice rates either. Quantitative data suggests that although surgeons believe it is important to address lifestyle changes with patients, and are confident and knowledgeable in doing so, these factors do not predict actual rates of preventive health practice. This might reflect the medically oriented work of surgeons, and that surgeons, rather than lacking confidence or knowledge, do not see preventive health as core to their role , which was probed in the subsequent qualitative interviews. The surgeons endorsed this biomedical perspective in the qualitative interviews, preferring to practice under a scope of vision restricted to the presenting issue. Surgeons’ engagement in health discussions predominantly relate to surgical outcomes; they did not consider it part of the surgical role to discuss general wellbeing. In the qualitative interviews, discussions with patients about smoking cessation was the most commonly noted preventive health topic; however, as exemplified by the quotes, smoking cessation was advised due to the operative risk, not for general health. Surgeons’ engagement in risk mitigation through smoking cessation advice is likely to be influenced by the well-publicised literature relating to smoking and post-operative risks . Despite unequivocal evidence that behaviour change interventions are effective in multiple settings , few interventions have targeted hospital surgical patients and very few studies have addressed behaviour change in non-admitted surgical clinics . The lack of published literature on surgeons’ preventive health practice might influence the surgeons’ perception that preventive health does not fit within their role. In the quantitative analyses, surgeons’ perceptions regarding patient acceptance of lifestyle interventions in non-admitted care was not significantly correlated with preventive health practice rates. The acceptance of lifestyle interventions on the part of the patients, and the patients’ motivation to undertake behaviour change was however, repeatedly brought up in the qualitative interviews. Many surgeons reported that although patients are agreeable to receiving lifestyle advice during a non-admitted surgical consult, the provision of advice did not translate to actual behaviour change on the part of the patient. On the other hand, surgeons noted that many patients attend specialist appointments seeking specialist advice, and are not interested in, or motivated by the provision of preventive health in non-admitted surgical care. From the interviews, differences were observed between surgeons as to whether they attempted to use the consultation to illicit patients’ motivation to change. Discussion of risk factors is standard practice for surgeons and in the interviews some surgeons reported using the consultation as an opportunity to link behavioural risk factors to the presenting health issue. Opportunistic health promotion is strongly advocated in chronic disease prevention and management . The quantitative analyses indicated that surgeons who prioritised preventive health were significantly more likely to use the clinical opportunity to undertake preventive health interventions. Not all surgeons however, approached the consultation as a chance to motivate patients, with some surgeons expressing concern about engaging in preventive health discussions with patients in the non-admitted setting. These concerns might reflect a didactic understanding of preventive health, where the passive patient is expected to adhere to the prescriptions of the healthcare expert . Alternative models exist, however, where emphasis is placed on empowering patients over their own health, rather than delivering purely instructive messages . The interviewed surgeons, ambivalent about engaging in behaviour change discussions should be encouraged by research indicating that the majority of patients view hospitals as an appropriate setting for health promotion . Consistent with existing literature, insufficient time was identified as a major barrier to preventive health . Time pressure is institutionally driven, with surgeons under pressure for clinical performance. Surgeons were cognisant of the waiting lists for public services, and the pressure that this places on clinical throughput. Performing under a fixed amount of time, surgeons largely felt that they could not afford to forego time spent in their expert role. In the quantitative analyses, work priority afforded to preventive health was the strongest predictive factor for engaging in preventive health practice, and this finding was subsequently probed in the qualitative interviews. It is important that the surgical profession recognises the role that surgeons can play in preventive health, even in the face of time demands. As little as 3 min of advice can markedly increase a patient’s chance of smoking cessation . The number of years of practice was also a significant predictive factor for engaging in preventive health practice. The qualitative interviews highlighted that one senior surgeon chose to decrease clinical volume, even in the face of service demand. This attests to the aforementioned influence that surgeons might have over institutional practice, and strengthens the argument to engage with surgeons on health promotion policies in the future. The facilitative topics raised by surgeons in the qualitative interviews were unanimous, with surgeons preferencing pathways to refer patients into specific programs tailored for health behaviour change as a means to facilitate preventive health interventions in non-admitted clinical care. The interviewed surgeons believe that information fliers and standardised referral pathways would allow them to engage, in a time-efficient manner, in preventive health with patients, and subsequently offer follow-on services. The provision of dedicated information, as well as referral pathways could offer avenues for surgeons to integrate preventive health into non-admitted care. Further to this, the development of linkages to community-based aftercare resources is likely to improve continuity of patient care , particularly when initiation of behaviour change is driven from the surgical consult . Limitations This study is subject to a number of limitations. First, preventive practices were assessed via self-report. This approach is consistent with numerous previous studies, however, the accuracy in assessing actual behaviour is unclear. Second, although our response rate of 51% was higher than other preventive health research with hospital doctors , the response rate is lower than observed in studies of surgeons’ clinical decision making . While low response rates can increase the possibility of response bias, significant differences were not observed between responding and non-responding doctors in cross-sectional studies . The doctors studied by Kellermen et al., were physicians, and not surgeons, which might limit the generalizability of the findings . Third, the GEE model might be underpowered to show statistical significance in the majority of measured variables . Further research with a larger cohort of participants might result in differing models that best predict preventive health practice by surgeons . Fourth, a degree of selection bias could have resulted from the survey non-response rates and interview non-participation rates, with the participating surgeons potentially more engaged in preventive health than non-responders . However, surgeons were purposefully sampled for the interviews to ensure a variation in the types of specialty areas and experience to providing insights from multiple perspectives. Finally, this study was undertaken in a single hospital. While single-site studies might limit generalizability, the primary aim of this research was to acquire detailed knowledge about context and processes of the studied phenomenon. Steps were taken to maximize rigour and attain theoretical saturation and these should ensure the broad applicability of the findings to other non-admitted, public hospital services. Implications for clinical practice and future research Surgeons undertaking of preventive health activities is influenced by a multitude of factors, with the working structure of the hospital most likely to influence preventive health practice rates . Heavy workload emerged as a core barrier that cannot be ignored . The interviewed surgeons were cognisant of the demands of the clinic, and report practising under a narrow specialist approach, foregoing holistic care. Management support is critical for the availability of time and resources required for surgeons to broaden their practice to increase preventive health practice rates . The interviewed surgeons suggested that in order to increase engagement in preventive health activities in non-admitted care, while managing consultation time, their preference was for the creation of information fliers on behaviour change to give to patients, and for referral pathways that link patients to specialist behaviour change programs available either in-house or in the community. The sheer volume of non-admitted surgical consultations provided annually offers vast potential for opportunistic preventive health in the non-admitted clinical setting . Due to the influence surgeons can exert over patients, it would be valuable to examine how surgeon-initiated referrals to tailored behaviour change programs could be implemented into routine practice, as well as health-related outcomes derived from this pathway. This study is subject to a number of limitations. First, preventive practices were assessed via self-report. This approach is consistent with numerous previous studies, however, the accuracy in assessing actual behaviour is unclear. Second, although our response rate of 51% was higher than other preventive health research with hospital doctors , the response rate is lower than observed in studies of surgeons’ clinical decision making . While low response rates can increase the possibility of response bias, significant differences were not observed between responding and non-responding doctors in cross-sectional studies . The doctors studied by Kellermen et al., were physicians, and not surgeons, which might limit the generalizability of the findings . Third, the GEE model might be underpowered to show statistical significance in the majority of measured variables . Further research with a larger cohort of participants might result in differing models that best predict preventive health practice by surgeons . Fourth, a degree of selection bias could have resulted from the survey non-response rates and interview non-participation rates, with the participating surgeons potentially more engaged in preventive health than non-responders . However, surgeons were purposefully sampled for the interviews to ensure a variation in the types of specialty areas and experience to providing insights from multiple perspectives. Finally, this study was undertaken in a single hospital. While single-site studies might limit generalizability, the primary aim of this research was to acquire detailed knowledge about context and processes of the studied phenomenon. Steps were taken to maximize rigour and attain theoretical saturation and these should ensure the broad applicability of the findings to other non-admitted, public hospital services. Surgeons undertaking of preventive health activities is influenced by a multitude of factors, with the working structure of the hospital most likely to influence preventive health practice rates . Heavy workload emerged as a core barrier that cannot be ignored . The interviewed surgeons were cognisant of the demands of the clinic, and report practising under a narrow specialist approach, foregoing holistic care. Management support is critical for the availability of time and resources required for surgeons to broaden their practice to increase preventive health practice rates . The interviewed surgeons suggested that in order to increase engagement in preventive health activities in non-admitted care, while managing consultation time, their preference was for the creation of information fliers on behaviour change to give to patients, and for referral pathways that link patients to specialist behaviour change programs available either in-house or in the community. The sheer volume of non-admitted surgical consultations provided annually offers vast potential for opportunistic preventive health in the non-admitted clinical setting . Due to the influence surgeons can exert over patients, it would be valuable to examine how surgeon-initiated referrals to tailored behaviour change programs could be implemented into routine practice, as well as health-related outcomes derived from this pathway. This mixed-methods study revealed that the majority of surgeons discuss lifestyle risk factors with their patients at low levels. Surgeons were unlikely to provide written advice or refer patients to ancillary preventive health services. The surgeons largely expressed positive attitudes towards preventive health, and the surgeons who placed the greatest work priority on preventive health were most likely to undertake preventive health practice. To increase preventive health practice, surgeons indicated a preference for pathways to enable referrals into dedicated behaviour change programs that could fit within the scope of non-admitted surgical consultations. Due to the high volume of ambulatory surgical consultations annually, it is important that surgeons remain active participants in preventive health policy. Additional file 1: Surgeons and Preventative Health Survey. (PDF 542 kb) Additional file 2: Codes, categories and themes relating to surgeons’ attitudes and beliefs towards the management of lifestyle risk factors. (DOCX 19 kb) Additional file 3: Themes and corresponding quotes relating to surgeons’ attitudes and beliefs towards the management of lifestyle risk factors. (DOCX 16 kb) |
Conflicts of Interest Among Cardiology Clinical Practice Guideline Authors in Japan | 49e842e6-38f8-4f20-b5ac-a4fc204b5f62 | 11262514 | Internal Medicine[mh] | This analysis of publicly available payment data disclosed by pharmaceutical companies found that 94.4% of Japanese cardiology clinical guideline authors received personal payments from pharmaceutical companies, totaling >US $70.8 million from 2016 to 2020. Leading authors of these cardiology clinical guidelines received larger payments than nonleading authors. More stringent and transparent conflict of interest management strategies are needed for authors of cardiology clinical guidelines in Japan. All data used in this study are available from the Yen For Docs database, managed by the Medical Governance Research Institute ( https://yenfordocs.jp/ ), and from each pharmaceutical company belonging to the Japan Pharmaceutical Manufacturers Association. Due to privacy restrictions on payment recipients, the data sets collected and analyzed during the current study are available from the corresponding author upon reasonable request. As this study was a retrospective analysis of publicly available data and met the definition of nonhuman subjects research, no institutional board review and approval were required. This study followed the Strengthening the Reporting of Observational Studies in Epidemiology guideline. Using a publicly accessible payment database, this study assessed the extent of financial relationships between the pharmaceutical industry and authors of CPGs for cardiovascular diseases in Japan. To comprehensively capture the financial relationships between cardiology CPG authors and pharmaceutical companies, this study included all authors of CPGs developed and published by the JCS from January 2015 to December 2022. This encompassed CPG chairs, development committee authors, review committee authors, systematic review committee authors, and writing supporting authors. We collected data on all authors of CPGs published from 2015 to the latest fiscal year (2022), as the payment database contained data on personal payments starting in 2016. Additionally, several international COI policies recommend that CPG authors should abstain from providing speaking and lecturing services sponsored by pharmaceutical companies for several years following CPG publication. , , , Established in 1935, the JCS represents the preeminent professional organization for cardiologists and cardiovascular researchers in Japan, comprising >32 000 society members. The JCS is responsible for producing CPGs for cardiovascular diseases and board‐certifying cardiologists in Japan. Subsequently, we extracted data on personal payments made by pharmaceutical companies to the CPG authors for the period between 2016 and 2020, using a publicly accessible payment database, consistent with methodologies used in previous studies. , The database included information on speaking fees, consultancy payments, and writing compensation, which pharmaceutical companies provided to individual health care providers from 2016 to 2020, aligning with the data collection methods of prior studies. , , All pharmaceutical firms and their subsidiaries affiliated with the Japan Pharmaceutical Manufacturers Association (JPMA), the foremost trade organization in Japan's pharmaceutical sector, were mandated to disclose payments to individual health care providers with the providers' names. These payments are for activities such as delivering lectures at industry‐sponsored events, offering consultancy services, and creating manuscripts and pamphlets. However, other forms of payments, including royalties, ownership interests, travel and lodging fees, food and beverage fees, and grant and research payments, are not mandated to be disclosed in Japan and have not been disclosed by the companies at individual provider level. This is in contrast to the United States, where the Physician Payments Sunshine Act mandates the disclosure of nearly all nonresearch, research, and ownership payments, making them accessible for review. , , , The disclosed payment data, available on the companies' respective websites, have been voluntarily collated into a searchable database by an independent research organization since 2016. This database, in the most recent iteration, encompasses payment records from 2016 through 2020. We conducted a descriptive analysis of the extracted payment data, including the proportions of CPG authors who received payments and the median and mean of the payments. Furthermore, in line with international COI policies advocating that CPG chairs should be devoid of any financial COIs, , , we separately analyzed the payments made to CPG chairs. We examined the disparity in payment amounts between chair authors and nonchair authors using the Mann–Whitney U test, as the payments per author were not normally distributed. All statistical analyses were conducted using Python 3.9.12 (Python Software Foundation, Beaverton, OR) and Stata version 17.0 (StataCorp, College Station, TX). Given that this study involved a retrospective analysis of publicly available data and designed as nonhuman subjects study, an institutional board review and informed consent were not required in Japan. In our analysis, we identified 929 unique authors from 37 JCS CPGs that were eligible for the study. Among these authors, 275 (29.0%) contributed to the development of 2 or more CPGs. Notably, 877 authors (94.4%) received 1 or more personal payments from pharmaceutical companies between 2016 and 2020 (Table ). The total cumulative payments amounted to US $70 895 253, distributed across 67 618 individual payment transactions. The mean payment per author over this 5‐year period was US $76 314 (SD: US $138663), and the median payment was US $20 792 (interquartile range [IQR]: US $4262–US $76 998). Of the total US $58.3 million in payments, 85.2% (amounting to US $50.7 million) were for lecture compensations, and 10.5% (US $6.2 million) for consulting services. The annual payments varied, ranging from US $14.4 million to US $15.6 million between 2016 and 2019, but showed a decrease to US $11.5 million in 2020. Furthermore, we identified 44 CPG authors who served in the capacities of chairs or vice chairs for CPG development. All 44 of these chairs received personal payments from pharmaceutical companies within the same study period. The median payments to these chairs were significantly higher than those made to nonchair authors, with amounts of US $42 126 versus US $18 978 ( P = 0.002 in the Mann–Whitney U test). The breakdown of payments to authors by specific guidelines is presented in Table . , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , In all 37 eligible JCS CPGs, >80% of authors received personal payments over the 5 years. The proportion of authors receiving personal payments ranged from 84% in the CPG on perioperative cardiovascular assessment and management for noncardiac surgery (published in 2022) to 100% in 13 different CPGs. The CPG on revascularization of stable coronary artery disease recorded the highest median payment amount at US $234 552 (interquartile range: US $63 801–US $309 956). This was followed by the CPGs on nonpharmacotherapy of cardiac arrhythmias (US $144 736), indication and management of pregnancy and delivery in women with heart disease (US $136 398), and management of peripheral arterial disease (US $104 253). This analysis of personal payments to Japanese cardiology CPG authors, as disclosed by pharmaceutical companies, demonstrated that >US $70.8 million were paid to 94% of all Japanese cardiology CPG authors from 2016 to 2020. Notably, all CPG chairs and vice chairs received significantly higher payments than other authors. Moreover, each of the 37 eligible CPGs included in this analysis had >80% of its authors receiving personal payments, offering critical insights into the financial relationships between cardiology CPG authors and the pharmaceutical industry. First, our findings indicate that the total personal payments to all JCS CPG authors exceeded US $70.8 million over 5 years, ranging US $11.5 to US $15.6 million annually. These payments constituted >50% of the personal payments made to all board‐certified cardiologists in Japan. A preliminary study reported that annual personal payments to all 15 048 cardiologists board certified by the JCS ranged from US $27.4 million to US $28.8 million between 2016 and 2019. Additionally, Purkayastha et al assessed nonresearch payments (eg, speaking fees, consulting fees) to authors of CPGs developed by the AHA/ACC from 2014 to 2020, finding that only 29% (169 out of 578 authors) received a total of US $16.8 million over 7 years. Hence, given that our study included only payments for speaking, consulting, and writing services, our analysis indicates that JCS CPG authors received at least 4.2 times more in nonresearch payments than AHA/ACC CPG authors, suggesting stronger financial ties of Japanese cardiology CPG authors to the pharmaceutical industry compared with their US counterparts. Second, this study found that >94% of CPG authors had financial relationships with the pharmaceutical industry, a higher percentage than reported in other countries. For instance, Dudum et al noted that 80% of authors of AHA/ACC CPGs published between 2016 and 2017 received either research or nonresearch payments. Additionally, Purkayastha et al. reported that the majority of AHA/ACC CPG authors published after 2018 did not receive nonresearch payments from health care companies. The fact that the majority of authors received personal payments from the pharmaceutical industry in all JCS CPGs, and that all CPG chairs had substantial financial ties, clearly deviates from current international COI management policies for CPG development. , , The COI policy from the US National Academy of Medicine, which sets widely recognized COI management strategies for CPG development globally, , , , , recommends that health care organizations producing CPGs should predominantly assign experts free from COIs as authors and that CPG chairs should have no financial COIs with the health care industry. Although some CPGs developed in the 2010s did not meet these recommendations, many recent studies have shown improvements in COI management strategies in CPGs developed after the late 2010s. , , , , , , Our study, however, has repeatedly demonstrated that nearly all Japanese CPG authors across specialties received significant payments for activities like delivering lectures and consulting services, leading to direct income. , , , , , , , , , , , , , , , Although the authors acknowledge the importance of collaboration between physicians and the health care industry in improving patient care, it is imperative to develop trustworthy and evidence‐based CPGs without establishing a group where >94% of the experts have substantial financial ties to the pharmaceutical industry. Implications for Future Research and Policy Interventions This study has illuminated the extensive financial relationships between CPG authors and the pharmaceutical industry in Japan. However, the effects of COIs on CPG recommendations, physicians' clinical practices, patient outcomes, and the trust and adherence of physicians and patients to these recommendations remain underexplored. Although individual COIs of CPG authors are documented, less attention has been given to institutional COIs at professional medical societies, universities, departments, and hospitals affiliated with CPG authors. A recent study reported that the JCS received a total of US $10.2 million in sponsorship, donations, and advertising fees from pharmaceutical companies affiliated with the JPMA from 2017 to 2021. Future research should investigate the impact of COIs, both individual and institutional, on clinical practice and patient outcomes. Furthermore, our findings suggest that COI management in Japanese cardiology CPGs does not meet international standards in terms of transparency and rigor. Despite some organizations having stringent COI policies for CPG development, instances of underdeclaration and inaccurate disclosure of COIs by some CPG authors have been reported in previous research. , , , This raises questions about the effectiveness of even the most rigorous and transparent COI policies without a mechanism to verify the accuracy of self‐declared COIs. In the United States, the Physician Payments Sunshine Act mandates that all pharmaceutical and medical device companies report any payments to physicians, both nonresearch and research‐related, on the Open Payments Database. , , , , , This legislation ensures that manufacturers report all research payments to physicians through third parties such as universities and teaching hospitals. , , , Some professional medical societies in the United States, including the American Society of Clinical Oncology and the American Gastroenterological Association, use this database to verify the accuracy of COI declarations by CPG authors. , To enhance COI management globally, regulatory agencies, industry, and professional organizations should consider developing and using similar databases to verify COI declarations outside the United States. A more comprehensive and comparative analysis of COI policies across medical specialties and countries is necessary. Such research could offer detailed insights into effective COI management in CPG development and identify strategies to improve the quality and trustworthiness of CPG recommendations. Limitations This study has several important limitations. First, the publicly accessible payment database developed by the Medical Governance Research Institute includes only nonresearch payments for speaking, consulting, and writing compensations from pharmaceutical companies affiliated with the JPMA. Additionally, due to the absence of a uniform system for payment disclosure and the fact that transparency initiatives are not enforced as rigorously for medical device companies as they are for pharmaceutical companies in Japan, this study was unable to account for payments to CPG authors from medical device companies. Consequently, this study may have underreported the magnitude and extent of financial relationships between CPG authors and the entire health care industry, including payments from companies not affiliated with the JPMA. However, of 104 pharmaceutical companies manufacturing prescription drugs, 73 (70.2%) companies affiliated with the JPMA and disclose their payment data as of 2020. Furthermore, 94% of total prescription drug sales in Japan in 2020 (US $101.0 billion out of US $107.4 billion) were drugs manufactured by the 73 JPMA‐affiliated companies. These figures would support the validity of examination of payments from the JPMA‐affiliated companies to the cardiology CPG authors for our study. Second, the JPMA requires its member companies to disclose payments for speaking, consulting, and writing compensations at the individual level, whereas other types of payments, including those for research, royalties, and ownership interests, were not available for this study. However, because compensation payments are generally paid directly to and can be a direct source of income for individual health care providers, examining the size and fraction of these compensations to the CPG authors is paramount in evaluating the extent of financial relationships between the CPG authors and the pharmaceutical industry for nonresearch purposes. Third, althuogh the JCS is responsible for developing and issuing CPGs for most of cardiovascular diseases in Japan, our study findings based on sampling from a single CPG developing society might not generalize contexts in other disease areas, specialties, and regions. This study has illuminated the extensive financial relationships between CPG authors and the pharmaceutical industry in Japan. However, the effects of COIs on CPG recommendations, physicians' clinical practices, patient outcomes, and the trust and adherence of physicians and patients to these recommendations remain underexplored. Although individual COIs of CPG authors are documented, less attention has been given to institutional COIs at professional medical societies, universities, departments, and hospitals affiliated with CPG authors. A recent study reported that the JCS received a total of US $10.2 million in sponsorship, donations, and advertising fees from pharmaceutical companies affiliated with the JPMA from 2017 to 2021. Future research should investigate the impact of COIs, both individual and institutional, on clinical practice and patient outcomes. Furthermore, our findings suggest that COI management in Japanese cardiology CPGs does not meet international standards in terms of transparency and rigor. Despite some organizations having stringent COI policies for CPG development, instances of underdeclaration and inaccurate disclosure of COIs by some CPG authors have been reported in previous research. , , , This raises questions about the effectiveness of even the most rigorous and transparent COI policies without a mechanism to verify the accuracy of self‐declared COIs. In the United States, the Physician Payments Sunshine Act mandates that all pharmaceutical and medical device companies report any payments to physicians, both nonresearch and research‐related, on the Open Payments Database. , , , , , This legislation ensures that manufacturers report all research payments to physicians through third parties such as universities and teaching hospitals. , , , Some professional medical societies in the United States, including the American Society of Clinical Oncology and the American Gastroenterological Association, use this database to verify the accuracy of COI declarations by CPG authors. , To enhance COI management globally, regulatory agencies, industry, and professional organizations should consider developing and using similar databases to verify COI declarations outside the United States. A more comprehensive and comparative analysis of COI policies across medical specialties and countries is necessary. Such research could offer detailed insights into effective COI management in CPG development and identify strategies to improve the quality and trustworthiness of CPG recommendations. This study has several important limitations. First, the publicly accessible payment database developed by the Medical Governance Research Institute includes only nonresearch payments for speaking, consulting, and writing compensations from pharmaceutical companies affiliated with the JPMA. Additionally, due to the absence of a uniform system for payment disclosure and the fact that transparency initiatives are not enforced as rigorously for medical device companies as they are for pharmaceutical companies in Japan, this study was unable to account for payments to CPG authors from medical device companies. Consequently, this study may have underreported the magnitude and extent of financial relationships between CPG authors and the entire health care industry, including payments from companies not affiliated with the JPMA. However, of 104 pharmaceutical companies manufacturing prescription drugs, 73 (70.2%) companies affiliated with the JPMA and disclose their payment data as of 2020. Furthermore, 94% of total prescription drug sales in Japan in 2020 (US $101.0 billion out of US $107.4 billion) were drugs manufactured by the 73 JPMA‐affiliated companies. These figures would support the validity of examination of payments from the JPMA‐affiliated companies to the cardiology CPG authors for our study. Second, the JPMA requires its member companies to disclose payments for speaking, consulting, and writing compensations at the individual level, whereas other types of payments, including those for research, royalties, and ownership interests, were not available for this study. However, because compensation payments are generally paid directly to and can be a direct source of income for individual health care providers, examining the size and fraction of these compensations to the CPG authors is paramount in evaluating the extent of financial relationships between the CPG authors and the pharmaceutical industry for nonresearch purposes. Third, althuogh the JCS is responsible for developing and issuing CPGs for most of cardiovascular diseases in Japan, our study findings based on sampling from a single CPG developing society might not generalize contexts in other disease areas, specialties, and regions. In conclusion, our findings that at least 94% of cardiology CPG authors in Japan had financial relationships with the pharmaceutical industry for nonresearch purposes, including all chairs of the JCS CPGs published between 2015 and 2022, highlight several deviations from international standards for proper COI management policies. The profound influence of CPG recommendations on physician practice and patient care necessitates the development of trustworthy CPGs that mitigate financial relationships with the pharmaceutical industry. It is crucial for the JCS to implement more transparent and stringent COI management strategies, aligning with the strong recommendations in current international COI policies for CPG development. A more thorough and comparative analysis of COI policies across various medical specialties and countries is warranted. Such research would identify specific strategies to enhance the quality and trustworthiness of CPG recommendations globally, extending beyond the field of cardiology. Future studies should also explore the impact of both individual and institutional COIs on CPG development, as well as their implications for clinical practice and patient outcomes. None. None. |
Enhancing ophthalmology medical record management with multi-modal knowledge graphs | 66cbf5eb-36bc-4412-b31d-8d56eae5167f | 11455959 | Ophthalmology[mh] | The emergence of knowledge graphs has prompted many leading researchers around the world to incorporate them into business systems, thereby improving the accuracy and interpretability of recommendations and bringing more comprehensive data analysis – . A knowledge graph, usually built based on graph databases, provides a powerful method for capturing complex relationships between entities by linking the edges of different entities according to information extracted from various heterogeneous data sources. Multi-modal knowledge graphs contain nodes with diverse data types, such as text, image, and voice information, enabling the joint use of multi-modal data. Knowledge graphs, especially multi-modal knowledge graphs, facilitate richer data information for graph-based neural network-based learning. Ophthalmology, a specialized field dedicated to visual function and eye health, plays a vital role in ensuring the overall well-being of individuals. According to the World Health Organization’s 2019 World Report on Vision, an estimated 2.2 billion people worldwide have vision impairment or blindness, with over 1 billion cases being preventable or treatable conditions that have yet to be addressed . To improve the clinical serviceability of ophthalmology, we have collected a batch of real desensitized ophthalmic medical records and their matched medical images and reports. This dataset has been reviewed by multiple rounds of experts from Beijing Tongren Hospital and will serve as a standard database for the national medical dataset. Previous studies on knowledge graph-based medical record management have predominantly focused on demonstrating enhanced visualization capabilities and indexing speeds , . Additionally, they have highlighted the efficacy of graph-structured Electronic Medical Records (EMRs) in supervised disease diagnosis through graph embedding techniques – . However, these studies have often overlooked the inherent capacity of graph-based data management to facilitate learning potential associations between medical records with little effort. Specifically, the graph structure inherently connects cases through feature nodes, thereby providing implicit feature correlations that AI algorithms can leverage to perform patient feature clustering without the need to construct specific patient cohorts or incorporate external knowledge vectors. This capability is particularly advantageous as it contrasts with the complexities associated with traditional relational database management of EMRs, where significant discrepancies exist in database designs across different countries and even within various hospitals . Information within a single medical record is often dispersed across numerous tables, making unified data structuring challenging. A more coherent data structure schema can be established by adopting a knowledge graph approach aligned with clinical logic, thereby paving the way for more sophisticated AI services. Inspired by this, we propose a schema to manage EMRs using a graph database, which can construct a multi-modal knowledge graph of audited medical records to enhance their management capabilities. Furthermore, we propose a novel contrastive graph attention network-based auxiliary diagnostic model named CGAT-ADM. This method achieves diagnostic clustering based on graph-structured medical data. It provides the ability to make similar diagnostic EMR recommendations, a fundamental aspect of electronic medical record management. The best part is that CGAT-ADM seamlessly integrates with existing data structures, eliminating the need for additional work while providing AI-based medical record recommendations. Our work demonstrates that medical information based on knowledge graphs has better AI service development capabilities. Moreover, in the supplementary documents, we compared the differences between building EMR management systems based on relational and graph databases, including the differences in data visualization capabilities and query time. Our study contributes to the field in the following ways: Taking ophthalmology as an example, we propose an application paradigm and graph schema that can transform structured medical records into a multi-modal knowledge graph that matches the required diagnostic features. We propose an auxiliary diagnostic method based on graph contrastive methods and demonstrate the effectiveness of building AI applications using multi-model knowledge graphs as the data foundation. This study demonstrated the process, structure, and characteristics of establishing a graphical data electronic medical record management system using de-identified real-world medical data. This section reviews relevant work from two perspectives: (i) the development of multi-modal knowledge graphs and their applications in medical information analysis and (ii) auxiliary diagnostic methods using EMRs. Multi-modal knowledge graph in medical research The Multi-modal Knowledge Graph integrates information from multiple modalities, enriches the representation of entities and relationships by labeling images and other forms, and supports multi-modal reasoning. Furthermore, incorporating multi-modal data as additional features can assist in resolving information gaps in some natural language tasks . Significant research has been conducted on integrating EMR information with knowledge graphs, which involves converting patient-related information into a graph structure for analysis. For example, Hasan et al. built a knowledge graph from the registration information of cancer patients for various applications, including visualization analysis and patient queue analysis . Using a graph structure to store patient medical record information has many advantages over relational database storage, including no need for extra code to perform complex searches, faster execution speed, and better connection of databases with different structures. Linfeng et al. present a systematic approach to constructing a medical knowledge graph from EMRs, demonstrating its effectiveness in ranking tasks and graph embedding learning . Aldwairi et al. proposed a new graph-based data management system for effective information storage, retrieval, and processing using healthcare data . However, the application of medical record-driven multi-modal knowledge graphs in clinical systems is relatively limited based on our knowledge. Zheng et al. made an effort to address this gap by constructing a multi-modal knowledge graph that incorporated patients’ lung CT scans, X-rays, and dialogue information between patients and doctors to enhance the diagnostic accuracy of COVID-19 . This study aimed to integrate doctor-patient dialogue and medical imaging into graph structures. However, it is worth noting that the doctor-patient dialogue primarily focused on the patient’s chief complaints, such as fever, dry cough, and muscle pain, without incorporating more extensive patient information. Auxiliary diagnostic methods using EMRs In recent years, research based on actual medical information has become increasingly popular, especially with the release of large-scale medical databases such as the MIMIC-III database , TCGA database, NHANES database, and UK BioBank. Among them, the analysis based on EMRs can be used to build clinical decision support systems to assist physicians in making more accurate clinical decisions, such as patient diagnosis, medication recommendations, surgical recommendations, etc , . Assistive diagnostics using healthcare data is a fundamental topic in analyzing EMRs, which aims to provide a scientifically accurate method for providing diagnostic results or clinical recommendations through AI. For example, Lin et al. constructed a medical knowledge graph from EHRs, ICD-9 ontology, DrugBank, and medical entity descriptions from Wikipedia and developed a patient similarity learning method based on the Siamese CNN model to compute the similarity score between all patient pairs . Pokharel et al. used a structure called “time trees” to represent EHR information and utilized the doc2vec embedding technique to calculate patient similarity scores . Sun et al. embedded a publicly available dataset of patient data into hidden representations and dynamically retrieved patients with the same diagnostic results . Rui Li et al. developed an interpretable and accurate risk prediction model to support EHR data exploration, knowledge graph demonstration, and model interpretation in heart failure risks . The work in constructed a knowledge graph system for EHR and verified that it could improve the utilization efficiency of unused information in clinical routine practice in chronic kidney disease patients. Previous studies have shown that EMR based on graph management performs well in terms of visualization capability and data query speed , . In addition, the potential of using graph-based EMRs for supervised learning was demonstrated – . This article aims to demonstrate another unexplored capability of EMR based on graph management: to achieve AI-based assisted diagnosis through feature associations between audited medical records without the need for additional data annotation or external knowledge feature vectors. This concept is particularly applicable to medical data analysis, as medical data is often limited in quantity and difficult to obtain annotations. By examining the feature associations between reviewed medical records, a rich data graph can be constructed to serve as the basis for training the model. This method helps improve the accuracy of diagnosis and explores potential case associations and disease patterns, providing more support for clinical decision-making. The Multi-modal Knowledge Graph integrates information from multiple modalities, enriches the representation of entities and relationships by labeling images and other forms, and supports multi-modal reasoning. Furthermore, incorporating multi-modal data as additional features can assist in resolving information gaps in some natural language tasks . Significant research has been conducted on integrating EMR information with knowledge graphs, which involves converting patient-related information into a graph structure for analysis. For example, Hasan et al. built a knowledge graph from the registration information of cancer patients for various applications, including visualization analysis and patient queue analysis . Using a graph structure to store patient medical record information has many advantages over relational database storage, including no need for extra code to perform complex searches, faster execution speed, and better connection of databases with different structures. Linfeng et al. present a systematic approach to constructing a medical knowledge graph from EMRs, demonstrating its effectiveness in ranking tasks and graph embedding learning . Aldwairi et al. proposed a new graph-based data management system for effective information storage, retrieval, and processing using healthcare data . However, the application of medical record-driven multi-modal knowledge graphs in clinical systems is relatively limited based on our knowledge. Zheng et al. made an effort to address this gap by constructing a multi-modal knowledge graph that incorporated patients’ lung CT scans, X-rays, and dialogue information between patients and doctors to enhance the diagnostic accuracy of COVID-19 . This study aimed to integrate doctor-patient dialogue and medical imaging into graph structures. However, it is worth noting that the doctor-patient dialogue primarily focused on the patient’s chief complaints, such as fever, dry cough, and muscle pain, without incorporating more extensive patient information. In recent years, research based on actual medical information has become increasingly popular, especially with the release of large-scale medical databases such as the MIMIC-III database , TCGA database, NHANES database, and UK BioBank. Among them, the analysis based on EMRs can be used to build clinical decision support systems to assist physicians in making more accurate clinical decisions, such as patient diagnosis, medication recommendations, surgical recommendations, etc , . Assistive diagnostics using healthcare data is a fundamental topic in analyzing EMRs, which aims to provide a scientifically accurate method for providing diagnostic results or clinical recommendations through AI. For example, Lin et al. constructed a medical knowledge graph from EHRs, ICD-9 ontology, DrugBank, and medical entity descriptions from Wikipedia and developed a patient similarity learning method based on the Siamese CNN model to compute the similarity score between all patient pairs . Pokharel et al. used a structure called “time trees” to represent EHR information and utilized the doc2vec embedding technique to calculate patient similarity scores . Sun et al. embedded a publicly available dataset of patient data into hidden representations and dynamically retrieved patients with the same diagnostic results . Rui Li et al. developed an interpretable and accurate risk prediction model to support EHR data exploration, knowledge graph demonstration, and model interpretation in heart failure risks . The work in constructed a knowledge graph system for EHR and verified that it could improve the utilization efficiency of unused information in clinical routine practice in chronic kidney disease patients. Previous studies have shown that EMR based on graph management performs well in terms of visualization capability and data query speed , . In addition, the potential of using graph-based EMRs for supervised learning was demonstrated – . This article aims to demonstrate another unexplored capability of EMR based on graph management: to achieve AI-based assisted diagnosis through feature associations between audited medical records without the need for additional data annotation or external knowledge feature vectors. This concept is particularly applicable to medical data analysis, as medical data is often limited in quantity and difficult to obtain annotations. By examining the feature associations between reviewed medical records, a rich data graph can be constructed to serve as the basis for training the model. This method helps improve the accuracy of diagnosis and explores potential case associations and disease patterns, providing more support for clinical decision-making. As we introduced in the related work, past research on EMR data modeling primarily employed two forms: sequential or graphical. Moreover, the majority of the work was conducted based on supervised learning , , and classify medical records into two or more categories. In other words, it involves patient queue selection, data processing, and the construction of auxiliary diagnostic or classification models. However, since we can construct knowledge graphs with stronger interpretability based on EMR, we consider using the graph structure of the knowledge graph as the data foundation for providing AI services. Driven by this, we propose a multi-modal knowledge graph-driven ophthalmic medical record management schema, and based on this, we have developed an AI-assisted application that requires no additional data processing. Our study was approved by the Ethics Committee of Beijing Tongren Hospital, Affiliated with Capital Medical University (Approval No. TRECKY2018-056-GZ(2022)-07). All methods were performed in accordance with the relevant guidelines and regulations. We confirm that informed consent was obtained from all subjects participating in the study, or from their legal guardian(s) in the case of minors. In this section, we first introduce the methods of data collection and constructing multi-modal knowledge graphs. On this basis, we introduce a novel auxiliary diagnostic model called CGAT-ADM, which combines metapath2vec , graph contrastive methods , and Siamese GAT model . By inputting clinical information without diagnostic results into the model, medical record recommendations that match the basic factual diagnostic results can be returned from the medical record database. The abbreviations used throughout the text are listed in Table . Construction of multi-model knowledge graph from medical records The information management systems in most hospitals are built on relational databases such as MySQL or SQL Server, and the data source in this study is no exception. A patient’s EMR will be divided and managed across multiple tables in such a structure. For the expert-reviewed and de-identified ophthalmic EMRs, we have developed a set of transformation engines to convert the structured data into a graph structure consisting of heterogeneous nodes and edges. Table summarizes the data statistics of the graph database. The transformation of medical records into a graph structure is based on two fundamental principles: Different medical records are linked through public nodes such as age, gender, diagnosis results, historical diseases, and other common attributes. This design provides a bridge for potential relationships between different medical records. Each medical record includes separate left and right eye nodes, with associated diagnosis results, exam results, ophthalmology history, and medical images connected to their respective eye nodes. Fig. illustrates the basic data structure of a medical record graph. Gender and age nodes function as public nodes that link each patient’s medical record. Systemic diseases, ophthalmic diseases, abnormal ophthalmic examination items, and patient diagnosis results serve as public nodes to connect different patients. Characteristic nodes retain specific textual descriptions of lifestyle habits and disease history. These public nodes provide a bridge for associating medical records of different patients, making data management more efficient and providing a basis for modeling relationships between other patients. All graph data is stored in a Neo4j database. The abnormal examination results from medical imaging are also maintained as feature nodes in the knowledge graph. Our supplementary documents introduce the characteristic information of EMRs directly used for diagnosis under the guidance of clinical experts, as well as the process of data collection. Our publicly available code provides the source code for converting structured EMRs into graphical structured data. Model realization of auxiliary diagnostic search engine We have incorporated a patient’s medical imaging feature information with their background information on a graph. However, we cannot determine the relationships between patients or disease groups. We developed a medical record graph data engine called Contrastive Graph Attention Network-based Auxiliary Diagnostic Model to achieve this. This module is based on metapath2vec , graph contrastive methods , Bert , and GAT . The framework has two layers: Input Layer and Graph Coding Layer. The Input Layer consists of two parts: node-type embedding using the metapath2vec and constructing medical records pairs using contrastive methods. The Graph Coding Layer uses the GAT model as the backbone. The framework of the CGAT-ADM is shown in Fig. . The notations and their descriptions in the figure are shown in Table . Input layer As shown in Fig. , the input layer of the CGAT-ADM model consists of two parts. The first part is the node-type embedding based on metapath2vec, and the second part is the construction of positive and negative samples based on contrastive methods. Node type embedding based on metapath2vec: Metapath2vec is a graph embedding method used for learning graph embeddings to address the problem of entity representation in heterogeneous graph networks. It produces high-quality entity representations using meta-paths as prior knowledge and employing neural networks for learning. Unlike random walks, the path acquisition process of metapath2vec takes into account the type of entity node and forms an explicit entity type path schema. The Skip-gram model is used for path vector modeling during the training process. We first constructed a large-scale sampling based on a multi-model medical graph according to the design of the meta-path. The meta-path starts by traversing all information about a patient from the head node of a medical record. Then, it jumps to the following medical record with the same diagnosis result through the diagnosis result node for further traversal. Each medical record was randomly sampled five times, and 10% of non-critical nodes were removed. We performed 100 rounds of medical record sampling from the entire training dataset. Subsequently, the sampling results were trained using the Skip-gram. The Skip-gram model comprises an input layer, a hidden layer, and an output layer. The input layer receives the unique heat vector of the node, and the output layer gives the conditional probability of each node category under the given central node in the form of the probability distribution. The hidden layer is a fully connected neural network layer responsible for learning the relationship between nodes. After metapath2vec, we obtained 768-dimensional embedding vectors for all node types. These vectors are used to create a node-type vector query table with a size of node-type number \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\times$$\end{document} × 768. Constructing positive and negative samples based on contrastive methods: Contrastive learning was first widely used in Computer Vision – . It involves data augmentation, such as rotating, cropping, and adding noise, to obtain different views of the target sample (image) and take these different views of the image as positive samples. This approach relies on prior knowledge that the augmentation will not affect the image’s label. Graph contrastive learning is a standard data augmentation method for graphs that involves randomly deleting nodes to generate positive and negative samples . Inspired by these studies, we have expanded the existing medical record graphs and generated pairs of medical record samples and their similarity index using contrastive methods. In a patient’s medical record information, the diagnosis result is a core feature. If the diagnosis results of two patients are similar, there is a high potential correlation between them. We utilize the diagnostic results as a monitoring signal to quantify the similarity between the two cases. Considering the potential for partial overlap among cases with multiple diseases, we employ the F1 score as a measure of similarity in negative sample construction and high correlation negative samples. The F1 score is calculated based on the number of true positives (TP), false positives (FP), and false negatives (FN). TP represents the number of consistent items in the diagnostic results of two patients, while FP+FN represents the sum of different diagnostic results of two patients. 1 \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\begin{aligned} \textrm{F1} = \frac{2 \cdot TP}{2 \cdot TP + FP + FN} \end{aligned}$$\end{document} F1 = 2 · T P 2 · T P + F P + F N The process involves applying three algorithms to construct positive, negative, and high-correlation negative samples. We have detailed descriptions of it in the supplementary document and open-source code. Graphic coding layer As illustrated in Fig. , the Graphic Coding Layer retrieves data from the Data Input Layer. It performs encoding to capture the global information of the medical records for achieving multi-modal fusion. We aim to generate a high-dimensional spatial clustering of patients with various diseases at a macro level and obtain spatial coordinates for patients sharing the same disease. Concurrently, these coordinates aim to reflect potential associations between patients, encompassing systemic disease information, specific lesion manifestations, and more. In our proposed CGAT-ADM method, we encode three distinct information types from patient medical records: text mode, numerical mode, and node category mode. In the text mode, we initially establish the principle of text concatenation for each node type and encode the concatenated text information for each node input in the medical record. The text of each node in the graph is fed into MacBERT-base, a Chinese pre-training model based on the BERT-base model . We utilize the 768-dimensional vector of the first token as the text vector representation for the node. MacBERT-base consists of a 12-layer transformer encoder specifically designed for Chinese language processing. Regarding the numerical mode, we utilize a 3-layer fully connected layer with a Sigmoid activation function to convert the numerical information into a 768-dimensional numerical vector representation. We only encode the node types that contain numerical information, such as the patient’s unaided vision, optimal corrected vision, intraocular pressure, and axial length. For each node in the graph, we retrieve the node category from the table obtained through metapath2vec and concatenate it with the corresponding node category vector, which has a size of node count \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\times$$\end{document} × 768. We implemented three matrices of size node count \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\times$$\end{document} × 768. By concatenating these matrices, we obtained a resulting matrix of size 3 \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\times$$\end{document} × node count \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\times$$\end{document} × 768. We applied a fully connected layer and Softmax to simplify the input graph into a node count \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\times$$\end{document} × 768 matrices, representing the node features of the input graph. In Fig. , the graph encoding layer is depicted as a Siamese GAT model with shared parameters. The GAT backbone model consists of a three-layer GAT layer, each with 12 heads of Multi-Head Attention. The node coding obtained from the node encoding layer and the graph’s adjacency matrix are inputs, and the first token is selected as the overall vector representation. We employ the cosine similarity of their respective graph vectors to calculate the similarity between pairs of cases. The established similarity metric is then utilized to guide the Mean Squared Error Loss (MSE). The MSE formula used in our study is as follows: 2 \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\begin{aligned} \textrm{MSE} = \frac{1}{n} \sum _{i=1}^{n} (y_{\text {pred}_i} - y_{\text {true}_i})^2 \end{aligned}$$\end{document} MSE = 1 n ∑ i = 1 n ( y pred i - y true i ) 2 Here, n represents the number of samples, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$y_{\text {pred}_i}$$\end{document} y pred i denotes the predicted value, and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$y_{\text {true}_i}$$\end{document} y true i represents the corresponding actual value. This formula calculates the squared differences between predictions and actual values, averages them over all samples, and yields the MSE value. As previously stated, CGAT-ADM aims to optimize the extraction of medical record information, thereby enhancing the utilization of medical data. The information management systems in most hospitals are built on relational databases such as MySQL or SQL Server, and the data source in this study is no exception. A patient’s EMR will be divided and managed across multiple tables in such a structure. For the expert-reviewed and de-identified ophthalmic EMRs, we have developed a set of transformation engines to convert the structured data into a graph structure consisting of heterogeneous nodes and edges. Table summarizes the data statistics of the graph database. The transformation of medical records into a graph structure is based on two fundamental principles: Different medical records are linked through public nodes such as age, gender, diagnosis results, historical diseases, and other common attributes. This design provides a bridge for potential relationships between different medical records. Each medical record includes separate left and right eye nodes, with associated diagnosis results, exam results, ophthalmology history, and medical images connected to their respective eye nodes. Fig. illustrates the basic data structure of a medical record graph. Gender and age nodes function as public nodes that link each patient’s medical record. Systemic diseases, ophthalmic diseases, abnormal ophthalmic examination items, and patient diagnosis results serve as public nodes to connect different patients. Characteristic nodes retain specific textual descriptions of lifestyle habits and disease history. These public nodes provide a bridge for associating medical records of different patients, making data management more efficient and providing a basis for modeling relationships between other patients. All graph data is stored in a Neo4j database. The abnormal examination results from medical imaging are also maintained as feature nodes in the knowledge graph. Our supplementary documents introduce the characteristic information of EMRs directly used for diagnosis under the guidance of clinical experts, as well as the process of data collection. Our publicly available code provides the source code for converting structured EMRs into graphical structured data. We have incorporated a patient’s medical imaging feature information with their background information on a graph. However, we cannot determine the relationships between patients or disease groups. We developed a medical record graph data engine called Contrastive Graph Attention Network-based Auxiliary Diagnostic Model to achieve this. This module is based on metapath2vec , graph contrastive methods , Bert , and GAT . The framework has two layers: Input Layer and Graph Coding Layer. The Input Layer consists of two parts: node-type embedding using the metapath2vec and constructing medical records pairs using contrastive methods. The Graph Coding Layer uses the GAT model as the backbone. The framework of the CGAT-ADM is shown in Fig. . The notations and their descriptions in the figure are shown in Table . Input layer As shown in Fig. , the input layer of the CGAT-ADM model consists of two parts. The first part is the node-type embedding based on metapath2vec, and the second part is the construction of positive and negative samples based on contrastive methods. Node type embedding based on metapath2vec: Metapath2vec is a graph embedding method used for learning graph embeddings to address the problem of entity representation in heterogeneous graph networks. It produces high-quality entity representations using meta-paths as prior knowledge and employing neural networks for learning. Unlike random walks, the path acquisition process of metapath2vec takes into account the type of entity node and forms an explicit entity type path schema. The Skip-gram model is used for path vector modeling during the training process. We first constructed a large-scale sampling based on a multi-model medical graph according to the design of the meta-path. The meta-path starts by traversing all information about a patient from the head node of a medical record. Then, it jumps to the following medical record with the same diagnosis result through the diagnosis result node for further traversal. Each medical record was randomly sampled five times, and 10% of non-critical nodes were removed. We performed 100 rounds of medical record sampling from the entire training dataset. Subsequently, the sampling results were trained using the Skip-gram. The Skip-gram model comprises an input layer, a hidden layer, and an output layer. The input layer receives the unique heat vector of the node, and the output layer gives the conditional probability of each node category under the given central node in the form of the probability distribution. The hidden layer is a fully connected neural network layer responsible for learning the relationship between nodes. After metapath2vec, we obtained 768-dimensional embedding vectors for all node types. These vectors are used to create a node-type vector query table with a size of node-type number \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\times$$\end{document} × 768. Constructing positive and negative samples based on contrastive methods: Contrastive learning was first widely used in Computer Vision – . It involves data augmentation, such as rotating, cropping, and adding noise, to obtain different views of the target sample (image) and take these different views of the image as positive samples. This approach relies on prior knowledge that the augmentation will not affect the image’s label. Graph contrastive learning is a standard data augmentation method for graphs that involves randomly deleting nodes to generate positive and negative samples . Inspired by these studies, we have expanded the existing medical record graphs and generated pairs of medical record samples and their similarity index using contrastive methods. In a patient’s medical record information, the diagnosis result is a core feature. If the diagnosis results of two patients are similar, there is a high potential correlation between them. We utilize the diagnostic results as a monitoring signal to quantify the similarity between the two cases. Considering the potential for partial overlap among cases with multiple diseases, we employ the F1 score as a measure of similarity in negative sample construction and high correlation negative samples. The F1 score is calculated based on the number of true positives (TP), false positives (FP), and false negatives (FN). TP represents the number of consistent items in the diagnostic results of two patients, while FP+FN represents the sum of different diagnostic results of two patients. 1 \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\begin{aligned} \textrm{F1} = \frac{2 \cdot TP}{2 \cdot TP + FP + FN} \end{aligned}$$\end{document} F1 = 2 · T P 2 · T P + F P + F N The process involves applying three algorithms to construct positive, negative, and high-correlation negative samples. We have detailed descriptions of it in the supplementary document and open-source code. Graphic coding layer As illustrated in Fig. , the Graphic Coding Layer retrieves data from the Data Input Layer. It performs encoding to capture the global information of the medical records for achieving multi-modal fusion. We aim to generate a high-dimensional spatial clustering of patients with various diseases at a macro level and obtain spatial coordinates for patients sharing the same disease. Concurrently, these coordinates aim to reflect potential associations between patients, encompassing systemic disease information, specific lesion manifestations, and more. In our proposed CGAT-ADM method, we encode three distinct information types from patient medical records: text mode, numerical mode, and node category mode. In the text mode, we initially establish the principle of text concatenation for each node type and encode the concatenated text information for each node input in the medical record. The text of each node in the graph is fed into MacBERT-base, a Chinese pre-training model based on the BERT-base model . We utilize the 768-dimensional vector of the first token as the text vector representation for the node. MacBERT-base consists of a 12-layer transformer encoder specifically designed for Chinese language processing. Regarding the numerical mode, we utilize a 3-layer fully connected layer with a Sigmoid activation function to convert the numerical information into a 768-dimensional numerical vector representation. We only encode the node types that contain numerical information, such as the patient’s unaided vision, optimal corrected vision, intraocular pressure, and axial length. For each node in the graph, we retrieve the node category from the table obtained through metapath2vec and concatenate it with the corresponding node category vector, which has a size of node count \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\times$$\end{document} × 768. We implemented three matrices of size node count \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\times$$\end{document} × 768. By concatenating these matrices, we obtained a resulting matrix of size 3 \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\times$$\end{document} × node count \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\times$$\end{document} × 768. We applied a fully connected layer and Softmax to simplify the input graph into a node count \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\times$$\end{document} × 768 matrices, representing the node features of the input graph. In Fig. , the graph encoding layer is depicted as a Siamese GAT model with shared parameters. The GAT backbone model consists of a three-layer GAT layer, each with 12 heads of Multi-Head Attention. The node coding obtained from the node encoding layer and the graph’s adjacency matrix are inputs, and the first token is selected as the overall vector representation. We employ the cosine similarity of their respective graph vectors to calculate the similarity between pairs of cases. The established similarity metric is then utilized to guide the Mean Squared Error Loss (MSE). The MSE formula used in our study is as follows: 2 \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\begin{aligned} \textrm{MSE} = \frac{1}{n} \sum _{i=1}^{n} (y_{\text {pred}_i} - y_{\text {true}_i})^2 \end{aligned}$$\end{document} MSE = 1 n ∑ i = 1 n ( y pred i - y true i ) 2 Here, n represents the number of samples, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$y_{\text {pred}_i}$$\end{document} y pred i denotes the predicted value, and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$y_{\text {true}_i}$$\end{document} y true i represents the corresponding actual value. This formula calculates the squared differences between predictions and actual values, averages them over all samples, and yields the MSE value. As previously stated, CGAT-ADM aims to optimize the extraction of medical record information, thereby enhancing the utilization of medical data. As shown in Fig. , the input layer of the CGAT-ADM model consists of two parts. The first part is the node-type embedding based on metapath2vec, and the second part is the construction of positive and negative samples based on contrastive methods. Node type embedding based on metapath2vec: Metapath2vec is a graph embedding method used for learning graph embeddings to address the problem of entity representation in heterogeneous graph networks. It produces high-quality entity representations using meta-paths as prior knowledge and employing neural networks for learning. Unlike random walks, the path acquisition process of metapath2vec takes into account the type of entity node and forms an explicit entity type path schema. The Skip-gram model is used for path vector modeling during the training process. We first constructed a large-scale sampling based on a multi-model medical graph according to the design of the meta-path. The meta-path starts by traversing all information about a patient from the head node of a medical record. Then, it jumps to the following medical record with the same diagnosis result through the diagnosis result node for further traversal. Each medical record was randomly sampled five times, and 10% of non-critical nodes were removed. We performed 100 rounds of medical record sampling from the entire training dataset. Subsequently, the sampling results were trained using the Skip-gram. The Skip-gram model comprises an input layer, a hidden layer, and an output layer. The input layer receives the unique heat vector of the node, and the output layer gives the conditional probability of each node category under the given central node in the form of the probability distribution. The hidden layer is a fully connected neural network layer responsible for learning the relationship between nodes. After metapath2vec, we obtained 768-dimensional embedding vectors for all node types. These vectors are used to create a node-type vector query table with a size of node-type number \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\times$$\end{document} × 768. Constructing positive and negative samples based on contrastive methods: Contrastive learning was first widely used in Computer Vision – . It involves data augmentation, such as rotating, cropping, and adding noise, to obtain different views of the target sample (image) and take these different views of the image as positive samples. This approach relies on prior knowledge that the augmentation will not affect the image’s label. Graph contrastive learning is a standard data augmentation method for graphs that involves randomly deleting nodes to generate positive and negative samples . Inspired by these studies, we have expanded the existing medical record graphs and generated pairs of medical record samples and their similarity index using contrastive methods. In a patient’s medical record information, the diagnosis result is a core feature. If the diagnosis results of two patients are similar, there is a high potential correlation between them. We utilize the diagnostic results as a monitoring signal to quantify the similarity between the two cases. Considering the potential for partial overlap among cases with multiple diseases, we employ the F1 score as a measure of similarity in negative sample construction and high correlation negative samples. The F1 score is calculated based on the number of true positives (TP), false positives (FP), and false negatives (FN). TP represents the number of consistent items in the diagnostic results of two patients, while FP+FN represents the sum of different diagnostic results of two patients. 1 \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\begin{aligned} \textrm{F1} = \frac{2 \cdot TP}{2 \cdot TP + FP + FN} \end{aligned}$$\end{document} F1 = 2 · T P 2 · T P + F P + F N The process involves applying three algorithms to construct positive, negative, and high-correlation negative samples. We have detailed descriptions of it in the supplementary document and open-source code. As illustrated in Fig. , the Graphic Coding Layer retrieves data from the Data Input Layer. It performs encoding to capture the global information of the medical records for achieving multi-modal fusion. We aim to generate a high-dimensional spatial clustering of patients with various diseases at a macro level and obtain spatial coordinates for patients sharing the same disease. Concurrently, these coordinates aim to reflect potential associations between patients, encompassing systemic disease information, specific lesion manifestations, and more. In our proposed CGAT-ADM method, we encode three distinct information types from patient medical records: text mode, numerical mode, and node category mode. In the text mode, we initially establish the principle of text concatenation for each node type and encode the concatenated text information for each node input in the medical record. The text of each node in the graph is fed into MacBERT-base, a Chinese pre-training model based on the BERT-base model . We utilize the 768-dimensional vector of the first token as the text vector representation for the node. MacBERT-base consists of a 12-layer transformer encoder specifically designed for Chinese language processing. Regarding the numerical mode, we utilize a 3-layer fully connected layer with a Sigmoid activation function to convert the numerical information into a 768-dimensional numerical vector representation. We only encode the node types that contain numerical information, such as the patient’s unaided vision, optimal corrected vision, intraocular pressure, and axial length. For each node in the graph, we retrieve the node category from the table obtained through metapath2vec and concatenate it with the corresponding node category vector, which has a size of node count \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\times$$\end{document} × 768. We implemented three matrices of size node count \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\times$$\end{document} × 768. By concatenating these matrices, we obtained a resulting matrix of size 3 \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\times$$\end{document} × node count \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\times$$\end{document} × 768. We applied a fully connected layer and Softmax to simplify the input graph into a node count \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\times$$\end{document} × 768 matrices, representing the node features of the input graph. In Fig. , the graph encoding layer is depicted as a Siamese GAT model with shared parameters. The GAT backbone model consists of a three-layer GAT layer, each with 12 heads of Multi-Head Attention. The node coding obtained from the node encoding layer and the graph’s adjacency matrix are inputs, and the first token is selected as the overall vector representation. We employ the cosine similarity of their respective graph vectors to calculate the similarity between pairs of cases. The established similarity metric is then utilized to guide the Mean Squared Error Loss (MSE). The MSE formula used in our study is as follows: 2 \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\begin{aligned} \textrm{MSE} = \frac{1}{n} \sum _{i=1}^{n} (y_{\text {pred}_i} - y_{\text {true}_i})^2 \end{aligned}$$\end{document} MSE = 1 n ∑ i = 1 n ( y pred i - y true i ) 2 Here, n represents the number of samples, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$y_{\text {pred}_i}$$\end{document} y pred i denotes the predicted value, and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$y_{\text {true}_i}$$\end{document} y true i represents the corresponding actual value. This formula calculates the squared differences between predictions and actual values, averages them over all samples, and yields the MSE value. As previously stated, CGAT-ADM aims to optimize the extraction of medical record information, thereby enhancing the utilization of medical data. Dataset Our data were collected from Beijing Tongren Hospital from 2022 to 2023. The diagnosis results of patients included seven categories: Choroid Melanoma (CM), Retinal Vein Occlusion (RVO), Age-related Macular Degeneration (AMD), Diabetic Retinopathy (DR), Retinal Detachment (RD), Refractive Error (RE), and Pathological Myopia (PM). Most patients suffer from one disease, while others suffer from multiple diseases simultaneously. In addition to the diagnostic results, the surgical and medication information provided by the doctor is also recorded. Most of the medical records in our dataset contain fundus images centered on the macula and OCT images of the macula area in both eyes of the patient. The fundus image was acquired using the Topcon non-contact fundus camera, while the OCT images were collected using the Heidelberg OCT scanner. To protect patient privacy, sensitive information such as patient names, birthdays, and contact information were removed from the real medical records used in our study. 10,124 patients have completed three rounds of medical expert review, data de-identification, and data structuring. This includes multiple visit records for the same patient. In this study, we only retained the medical records of the first visit for each patient, resulting in a total of 5,367 patients. Among them are 224 patients with the diagnosis label CM, 1,146 patients with RVO, 1,159 patients with AMD, 1,696 patients with DR, 530 patients with RD, 426 patients with RE, and 448 patients with PM. The majority of patients have one type of disease, while some patients are diagnosed with multiple diseases at the same time. There are a total of 2,990 female patients, accounting for 55.72%. The median age of all patients is 59, ranging from 1 to 98. For more information about our workflow, please refer to the supplementary document. To verify the capability of CGAT-ADM, we used a 7:1:2 split method to divide the dataset into a training set, a validation set, and a test set. Fig. illustrates the workflow of data assignment and the distribution of disease labels among patients. Baseline methods In this study, we have implemented a pipeline to provide diagnostic assistance in the form of medical record recommendations using a graph data structure. To evaluate the effectiveness of the proposed CGAT-ADM, we compared it with the following baseline methods. Principal Component Analysis (PCA): An unsupervised algorithm for dimension reduction and feature extraction. We use PCA to map the original high-dimensional data through linear transformation to a low-dimensional space. The number of components is set to 16. Then, we use the cosine distance to measure the similarity between two medical records. Code Sum based Matching (CSM): A method proposed by Choi et al. This method converts ICD diagnostic codes into medical concept vectors. Then, the patient similarity is determined by calculating the cosine distance between patient vectors . In our experiment, the patient node vector obtained through metapath2vec is used as the medical concept vector, and the similarity of patients is determined by calculating the cosine distance between patients. Graph Neural Networks (GNN): GNN is a versatile model for processing graph data, characterized by effective graph modeling, capturing connectivity and dependencies, and leveraging message passing for contextual information exchange among nodes . GNN obtains a comprehensive representation of the entire graph by treating it as a whole, making it particularly suitable for handling non-Euclidean data. GNN’s ability to propagate and aggregate information among nodes enables it to capture the graph’s relational dependencies and global patterns. Graph Convolutional Network (GCN): GCN is a deep learning method that utilizes the underlying graph structure. It primarily employs neighborhood aggregation to perform convolution operations on node features and can dynamically adjust weights through the learning process . GCN extracts and encodes neighborhood information for each node like convolution, allowing for information aggregation from adjacent nodes. Graph Attention Network (GAT): GAT is a graph neural network that leverages the attention mechanism. It incorporates weighted aggregation based on the relationship weight between a node and its neighbors, thereby enhancing the perception of local information . In GAT, each node assigns a weight coefficient to its neighboring nodes, which is learned and reflects the importance of each neighbor to the node. These weights are then used to aggregate the features of the neighboring nodes, resulting in a more refined representation that preserves local information. Evaluation metric For each patient in the test dataset, we calculate the cosine similarity between that patient and all other patients and retrieve the top k most similar patients. The diagnostic similarity between the input patients and their top k most similar patients is evaluated using the precision metric. We denote MP@k as the mean average precision of the top k most similar patients to the query patient of the validation dataset. Since we are building real medical records for ophthalmic clinical practice, which means that a patient can diagnose multiple diseases simultaneously, this is not a multi-classification or multi-label task. We use MP@k to evaluate the quality of search results, which is a commonly used metric used to measure the average accuracy of information retrieval systems in ranking the top k in the returned results. In machine learning, precision refers to the ratio of correctly classified samples by the classifier to the total number of samples classified by the classifier. 3 \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\begin{aligned} \textrm{Precision} = \frac{TP}{TP + FP} \end{aligned}$$\end{document} Precision = TP T P + F P To better evaluate the performance of different models, we set k = 1, 5, 10, and 20, respectively, to retrieve the top 1, 5, 10, and 20 most similar patients in the training database for each medical record graph in the test dataset. Results Table shows the performance comparison results of the models. our CGAT-ADM achieved the highest precision of 0.8940 in MP@1, 0.8742 in MP@5, 0.8686 in MP@10 and 0.8563 in MP@20. Contrastive methods are used to construct medical record pairs, and significant performance improvements have been achieved by incorporating GAT, GCN, and GNN on top of the metapath2vec method. This demonstrates that the graph contrastive strategy can greatly enhance the capability of medical record analysis based on a multi-model knowledge graph. The CGAT-ADM approach significantly improves performance by incorporating more comprehensive information, including medication, medical history, intraocular pressure, and visual acuity. This integration yields a remarkable improvement of over 13.19% in MP@1, 13.25% in MP@5, 13.73% in MP@10, and 12.08% in MP@20 compared to the GAT method. Fig. displays the high-dimensional results of all the EMRs in our dataset obtained through CGAT-ADM, which were then reduced to 2 dimensions using t-SNE and visualized in a scatter plot. In summary, the research findings showcase the potential of CGAT-ADM as a physician-assisted tool in real-world and intricate clinical settings. It also highlights its efficacy in identifying potential disease patterns by fusing medical records with multiple data sources. Ablation study Generally, patient similarity learning based on diagnostic results using EMRs is centered on doing semantic similarity studies with medical text information – or converting patients’ medical data into specific codes or graphs for similarity learning , , . Compared to traditional research, we suggest encoding and integrating different types of information separately to improve the overall performance of medical record similarity analysis. Based on the ablation experiment results in Table , The GAT model based on node types achieved an MP@1 of 0.7898, MP@5 of 0.7719, MP@10 of 0.7637, and MP@20 of 0.7640. Based on node text information, the GAT model achieved an MP@1 of 0.8508, MP@5 of 0.8444, MP@10 of 0.8387, and MP@20 of 0.8318. By fusing text, node types, and numerical information, CGAT-ADM achieved an MP@1 of 0.8940, MP@5 of 0.8742, MP@10 of 0.8686, and MP@20 of 0.8563. The improvement in system performance by adding numerical information is relatively small, mainly due to the limited number of numerical nodes of these medical records. However, we still retain the design of the numerical information channel to enhance the model’s scalability of integrating data for more potential examination indicators. Programming architecture The CGAT-ADM model is trained using PyTorch. The CGAT-ADM model processes all healthcare data to construct their high-dimensional representations stored in the Milvus vector database. The underlying data structure is stored in the Neo4j graph database, while the medical imaging records of patients are saved as URLs in the graph database. This study demonstrates a clustering service that stores multimodal EMR information in a graph structure. We have built a framework for online querying and visualization of record recommendations. The back end of the application is invoked using Node.js. The front end is developed using Vue.js. Doctors or healthcare professionals can input partial or complete patient information through API to retrieve the most similar patients from the database using the CGAT-ADM engine. It is essential to highlight that the data is stored in the Neo4j graph database, while the vector database engine maintains high-dimensional positional vectors for all the records. Our data were collected from Beijing Tongren Hospital from 2022 to 2023. The diagnosis results of patients included seven categories: Choroid Melanoma (CM), Retinal Vein Occlusion (RVO), Age-related Macular Degeneration (AMD), Diabetic Retinopathy (DR), Retinal Detachment (RD), Refractive Error (RE), and Pathological Myopia (PM). Most patients suffer from one disease, while others suffer from multiple diseases simultaneously. In addition to the diagnostic results, the surgical and medication information provided by the doctor is also recorded. Most of the medical records in our dataset contain fundus images centered on the macula and OCT images of the macula area in both eyes of the patient. The fundus image was acquired using the Topcon non-contact fundus camera, while the OCT images were collected using the Heidelberg OCT scanner. To protect patient privacy, sensitive information such as patient names, birthdays, and contact information were removed from the real medical records used in our study. 10,124 patients have completed three rounds of medical expert review, data de-identification, and data structuring. This includes multiple visit records for the same patient. In this study, we only retained the medical records of the first visit for each patient, resulting in a total of 5,367 patients. Among them are 224 patients with the diagnosis label CM, 1,146 patients with RVO, 1,159 patients with AMD, 1,696 patients with DR, 530 patients with RD, 426 patients with RE, and 448 patients with PM. The majority of patients have one type of disease, while some patients are diagnosed with multiple diseases at the same time. There are a total of 2,990 female patients, accounting for 55.72%. The median age of all patients is 59, ranging from 1 to 98. For more information about our workflow, please refer to the supplementary document. To verify the capability of CGAT-ADM, we used a 7:1:2 split method to divide the dataset into a training set, a validation set, and a test set. Fig. illustrates the workflow of data assignment and the distribution of disease labels among patients. In this study, we have implemented a pipeline to provide diagnostic assistance in the form of medical record recommendations using a graph data structure. To evaluate the effectiveness of the proposed CGAT-ADM, we compared it with the following baseline methods. Principal Component Analysis (PCA): An unsupervised algorithm for dimension reduction and feature extraction. We use PCA to map the original high-dimensional data through linear transformation to a low-dimensional space. The number of components is set to 16. Then, we use the cosine distance to measure the similarity between two medical records. Code Sum based Matching (CSM): A method proposed by Choi et al. This method converts ICD diagnostic codes into medical concept vectors. Then, the patient similarity is determined by calculating the cosine distance between patient vectors . In our experiment, the patient node vector obtained through metapath2vec is used as the medical concept vector, and the similarity of patients is determined by calculating the cosine distance between patients. Graph Neural Networks (GNN): GNN is a versatile model for processing graph data, characterized by effective graph modeling, capturing connectivity and dependencies, and leveraging message passing for contextual information exchange among nodes . GNN obtains a comprehensive representation of the entire graph by treating it as a whole, making it particularly suitable for handling non-Euclidean data. GNN’s ability to propagate and aggregate information among nodes enables it to capture the graph’s relational dependencies and global patterns. Graph Convolutional Network (GCN): GCN is a deep learning method that utilizes the underlying graph structure. It primarily employs neighborhood aggregation to perform convolution operations on node features and can dynamically adjust weights through the learning process . GCN extracts and encodes neighborhood information for each node like convolution, allowing for information aggregation from adjacent nodes. Graph Attention Network (GAT): GAT is a graph neural network that leverages the attention mechanism. It incorporates weighted aggregation based on the relationship weight between a node and its neighbors, thereby enhancing the perception of local information . In GAT, each node assigns a weight coefficient to its neighboring nodes, which is learned and reflects the importance of each neighbor to the node. These weights are then used to aggregate the features of the neighboring nodes, resulting in a more refined representation that preserves local information. For each patient in the test dataset, we calculate the cosine similarity between that patient and all other patients and retrieve the top k most similar patients. The diagnostic similarity between the input patients and their top k most similar patients is evaluated using the precision metric. We denote MP@k as the mean average precision of the top k most similar patients to the query patient of the validation dataset. Since we are building real medical records for ophthalmic clinical practice, which means that a patient can diagnose multiple diseases simultaneously, this is not a multi-classification or multi-label task. We use MP@k to evaluate the quality of search results, which is a commonly used metric used to measure the average accuracy of information retrieval systems in ranking the top k in the returned results. In machine learning, precision refers to the ratio of correctly classified samples by the classifier to the total number of samples classified by the classifier. 3 \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\begin{aligned} \textrm{Precision} = \frac{TP}{TP + FP} \end{aligned}$$\end{document} Precision = TP T P + F P To better evaluate the performance of different models, we set k = 1, 5, 10, and 20, respectively, to retrieve the top 1, 5, 10, and 20 most similar patients in the training database for each medical record graph in the test dataset. Table shows the performance comparison results of the models. our CGAT-ADM achieved the highest precision of 0.8940 in MP@1, 0.8742 in MP@5, 0.8686 in MP@10 and 0.8563 in MP@20. Contrastive methods are used to construct medical record pairs, and significant performance improvements have been achieved by incorporating GAT, GCN, and GNN on top of the metapath2vec method. This demonstrates that the graph contrastive strategy can greatly enhance the capability of medical record analysis based on a multi-model knowledge graph. The CGAT-ADM approach significantly improves performance by incorporating more comprehensive information, including medication, medical history, intraocular pressure, and visual acuity. This integration yields a remarkable improvement of over 13.19% in MP@1, 13.25% in MP@5, 13.73% in MP@10, and 12.08% in MP@20 compared to the GAT method. Fig. displays the high-dimensional results of all the EMRs in our dataset obtained through CGAT-ADM, which were then reduced to 2 dimensions using t-SNE and visualized in a scatter plot. In summary, the research findings showcase the potential of CGAT-ADM as a physician-assisted tool in real-world and intricate clinical settings. It also highlights its efficacy in identifying potential disease patterns by fusing medical records with multiple data sources. Generally, patient similarity learning based on diagnostic results using EMRs is centered on doing semantic similarity studies with medical text information – or converting patients’ medical data into specific codes or graphs for similarity learning , , . Compared to traditional research, we suggest encoding and integrating different types of information separately to improve the overall performance of medical record similarity analysis. Based on the ablation experiment results in Table , The GAT model based on node types achieved an MP@1 of 0.7898, MP@5 of 0.7719, MP@10 of 0.7637, and MP@20 of 0.7640. Based on node text information, the GAT model achieved an MP@1 of 0.8508, MP@5 of 0.8444, MP@10 of 0.8387, and MP@20 of 0.8318. By fusing text, node types, and numerical information, CGAT-ADM achieved an MP@1 of 0.8940, MP@5 of 0.8742, MP@10 of 0.8686, and MP@20 of 0.8563. The improvement in system performance by adding numerical information is relatively small, mainly due to the limited number of numerical nodes of these medical records. However, we still retain the design of the numerical information channel to enhance the model’s scalability of integrating data for more potential examination indicators. The CGAT-ADM model is trained using PyTorch. The CGAT-ADM model processes all healthcare data to construct their high-dimensional representations stored in the Milvus vector database. The underlying data structure is stored in the Neo4j graph database, while the medical imaging records of patients are saved as URLs in the graph database. This study demonstrates a clustering service that stores multimodal EMR information in a graph structure. We have built a framework for online querying and visualization of record recommendations. The back end of the application is invoked using Node.js. The front end is developed using Vue.js. Doctors or healthcare professionals can input partial or complete patient information through API to retrieve the most similar patients from the database using the CGAT-ADM engine. It is essential to highlight that the data is stored in the Neo4j graph database, while the vector database engine maintains high-dimensional positional vectors for all the records. This study introduces a schema for converting existing EMRs into graph-structured data. Based on this schema, diagnostic results can be clustered with less further data processing. The CGAT-ADM model we propose is a multi-modal knowledge graph-based medical record-assisted diagnosis and recommendation method. In the field of ophthalmology, doctors primarily rely on patient communication and the observation of medical images to make diagnoses. This study aimed to simulate clinical reasoning by leveraging AI techniques and medical data in a graph structure. Using multi-modal knowledge graphs for medical record management demonstrations improved capabilities for deploying AI applications in this domain. This particularly benefits regions with limited medical resources, scarce doctors, or relatively inexperienced medical professionals. The quality of medical services can be improved by establishing a standardized medical record system that is desensitized and reviewed by experts and utilizing AI methods. In addition, we also presented a visual display of medical records based on graph databases and a comparison of query time between graph data management medical records and relational databases in the supplementary documents. The CGAT-ADM method is defined as an auxiliary diagnostic model because the anchor of the strategy of the graph contrast method is the coarse diagnosis and fine diagnosis results of patients, thus achieving clustering based on diagnostic results. This model can automatically recommend patients similar to their condition based on partial or comprehensive patient information, thereby providing potential diagnostic results and recommendations for similar patients. Through the above work, we demonstrate that managing electronic medical record information in a multi-modal knowledge graph possesses advantages that relational databases do not have, namely the ability to provide AI services with less additional data work. The development of the vast majority of auxiliary diagnostic systems typically involves several necessary processes, including defining classification labels, collecting clinical data, and building classification models. This work is generally supervised learning and requires manual screening of clinical data to construct sequences, graphs, or based on organized medical images. The business logic we demonstrate directly builds auxiliary diagnostics through the underlying data structure of EMR management systems. Previous research has emphasized the advantages of knowledge graph management of medical data, including faster indexing speeds and better visualization capabilities , , , . Conversely, this study showcases the potential of building medical AI services based on the graph structure. It is worth noting that although some studies have proposed using tabular data for classification , , graph structures, due to their unique triple structure, have a distinct advantage when building recommendation systems, which is currently difficult for tabular data to achieve. We are fully aware of the limitations of this study. Firstly, this study integrates over 30 different abnormal results obtained from medical imaging platforms when constructing graph-based EMRs. In the future, we will consider building visual extractors to extract feature values from medical images directly. Moreover, this research is based on ophthalmic medical record data, and different medical disciplines may have different graph data construction schemas. This involves the medical disciplines’ diagnostic processes, clinical challenges, and data modeling methods. Similar to the table design of relational databases, the structure design of graphs also requires in-depth discussions with medical experts and software developers. Our data has undergone expert review, anonymization, and informed consent from patients. However, in real-world applications, especially when using knowledge graphs as the foundation for data management to build artificial intelligence applications, it is necessary to consider more issues, such as patient privacy, especially when using medical data to build artificial intelligence services. Furthermore, this study is based on data that experts have reviewed, and more prospective experiments, including clinical applications, will be beneficial in demonstrating the superiority of our approach. This will be the focus of our next phase of work. We are also considering further exploration of the integration of artificial intelligence and multi-modal knowledge graphs, including integrating external variable datasets such as sequencing information and drug data. This integration will provide new opportunities for advanced clinical analysis, such as determining clinical trial eligibility, evaluating drug treatment efficacy, analyzing patient demographic variables, and identifying risk factors. Supplementary Information. |
Who is the ophthalmologist that developing countries
need? | 5eb5ef75-8ecf-4bd1-b4dd-5d669bf76a12 | 11826786 | Ophthalmology[mh] | |
Navigating Disparities in Dental Health—A Transit‐Based Investigation of Access to Dental Care in Virginia | 9c0aa4c6-4b24-4677-8036-ed3d337bab46 | 11754141 | Dentistry[mh] | Introduction Access to dental care is a national public health crisis in the United States (USA). In 2023, around 21% of Americans are currently living in the dental Health Professional Shortage Areas, measured using the number of dentists relative to the population with consideration of high needs . Based on the 2022 Medical Expenditure Panel Survey data, only 43% of the USA population had a dental visit in the past 12 months . Access to dental care is also a health disparity issue that manifests a significant gap in geographic location and socioeconomic status. In addition to the high cost of dental care, people living in rural areas, having lower income and education, with Medicaid/Children's Health Insurance Plan (CHIP), or being uninsured increases the likelihood of unmet dental needs . The disparities in dental care access also vary across different states in the USA. In 2019, the highest density of dentists was 104 dentists per 100 000 population in the District of Columbia, and the lowest was 40.97 in Alabama . While the density of dentists in Virginia (63.19) was above the national average (61.06), it has a huge disparity across different areas . For example, residents in the Northern and Eastern areas of Virginia are more likely to receive dental care than the residents in the Western and Southern districts . The dentist‐to‐population ratio is generally greater in highly urbanised regions (e.g. Richmond metropolitan area and Northern Virginia) than in rural areas (e.g. Southwest and Southside areas) . Four rural counties including Charles City County, King and Queen County, Surry County, and Sussex County did not have a single licensed dentist . However, limited studies examine geospatial disparities in dental care access, especially in rural areas and small‐sized cities. The existing studies also lack investigation into different accessibility levels in terms of the mode of transit (e.g. driving, public transit, and walking). For example, most studies only apply driving distance and time to measure geospatial accessibility, despite the fact that driving may not be an option for everyone, especially for those who do not have private vehicles or are unable to drive . Thus, other modes of transit, including public transit, should be taken into consideration. Insurance types also play an important role in access to dental care. Only 43% of dentists nationwide participate in Medicaid/CHIP, reflecting burdensome administrative barriers, low reimbursement rates, and missed appointments from patients . This barrier would add a layer of difficulties for patients enrolled in Medicaid to access dental care. In Virginia, only 27% of practicing dentists have treated patients enrolled in Medicaid or Family Access to Medical Insurance Security (FAMIS) in 2022 . Around 50% of Virginians whose household income is below $35 000 per year did not visit a dentist in 2022 compared with one‐third among all Virginians . Given the disparities in accessibility and different healthcare‐seeking behaviours, there has not been a measure of geospatial accessibility among patients using Medicaid insurance considering the use of public transit, given that Medicaid users are more likely to take public transit . To address this knowledge gap, this study aimed to identify vulnerable areas and populations of limited access to dental care in Virginia, through a transit‐based accessibility model, which adheres to public transit schedules and stops. Here, public transit includes fixed‐route transit services, such as commute buses and metros. The aim was measured through two specific objectives: (1) to calculate travel time and accessibility scores to dental care in Virginia using a transit‐based accessibility model for all dental clinics and dental clinics participating in the Medicaid dental program and (2) to estimate factors associated with the accessibility to dental clinics participating in the Medicaid dental program in Virginia. Methods 2.1 Study Region The study comprised nine regions within Virginia, including Greenville, Altavista, Lynchburg, Winchester, Williamsburg, Staunton‐Harrisonburg, Hampton, Richmond, and Northern Virginia (including Alexandria, Arlington, Fairfax County, Loudoun County, and Prince William County). The total population living in our study area is 5 968 587, which is about 69% of the entire population living in Virginia in 2021. Although it would be ideal to include the entire state, many regions in Virginia do not maintain sufficiently detailed public transit data, especially in rural areas. The nine regions were selected as related transit data was available. Each region represented a diverse array of geographical locations and urban–rural mix. Small cities like Greenville and Altavista offered a contrast to the bigger and more urban settings of Richmond, Northern Virginia, and Hampton. Lynchburg, Williamsburg, Winchester, and Staunton‐Harrisonburg are middle‐sized areas. This combination provided us with a comprehensive overview of varying geographic contexts within Virginia. 2.2 Data The study used data from various sources to build a realistic public transit model that plays a key role in accurately assessing transit‐based dental care accessibility. First, in terms of the public transit model, general transit feed specification (GTFS) data were compiled , which is essential for analysing public transit schedules and routes from multiple sources, such as primarily public transit websites and public repositories . The GTFS files contained detailed information about transit routes, schedules, and stops, which are crucial for computing accurate transit‐based travel time between origins and destinations . The GTFS files of nearby transit agencies were aggregated to reflect the real‐world transit service coverage area. To measure the spatial accessibility to dental care services, the study utilised building footprints, a polygon or set of polygons that show the total area of a building, assuming that individuals depart their transit trips from each building footprint to dental care services. Building footprints were used since they are the finest spatial resolution units that can be considered to be occupied by humans. The footprint data were obtained from the Virginia Geographic Information Network (VGIN) . Regarding the destination information, the study used comprehensive dental clinic data, including location and participation in Medicaid in each clinic, from the American Dental Association (ADA) 2022 Office Database . This database includes dentists that participate in pediatric Medicaid or CHIP programs based on a roster of dentists listed in the Insure Kids Now website maintained by the Centers for Medicare and Medicaid Services . 2.3 Statistical Analysis Next, using r5r , an R package designed for transit network analysis , the research team created a high‐resolution schedule‐aware transit network for the study area. For each subregion in the study area, a comprehensive transit network was built, and then a transit‐based origin–destination travel time matrix was computed. The origin included all the geographic coordinates (longitude and latitude) of the centroid of the building footprints in each region, and the destination consisted of the geographic coordinates of the dental care facilities in these regions. Note that the transit‐based origin–destination travel time matrices include detailed trip information, such as in‐vehicle travel time (i.e. ride time) as well as out‐of‐vehicle travel time (e.g. walking to the bus stop or home and waiting time for the bus). Detailed information on travel time estimation was provided in the Data . As a result, accessibility scores were calculated by counting the number of dental clinics that can be reached within 60 min of travel time. The 60‐min threshold was selected considering that it is an acceptable travel time for people using public transit and indicates the point at which accessing healthcare via public transit may impose an undue burden . The accessibility metrics evaluated how easily people could reach dental clinics using public transit . Two accessibility scores were computed and aggregated to the census block group level: (1) access to all dental clinics; (2) access to dental clinics participating in Medicaid. Next, this study examined how residents' primary socio‐economic factors are associated with each accessibility score but focused on dental clinics participating in Medicaid using multiple linear regression models. This group was chosen based on a higher likelihood of Medicaid patients than overall residents taking public transit. Also, only six regions were selected from the included nine regions for this analysis due to the small number of census block groups in Greenville, Altavista, and Lynchburg. Dependent variables were the accessibility scores at the census block group level. Independent variables included primary socio‐economic factors in each region, including population density, poverty percentage and non‐White population that were extracted from the 2021 American Community Survey (ACS) data . R version 4.2.1 was used to conduct rapid transit network analysis and all regression analyses. ArcGIS Pro version 3.2.1 from Esri was used to visualise the accessibility score across the selected cities. The significance level was set at 0.05. Study Region The study comprised nine regions within Virginia, including Greenville, Altavista, Lynchburg, Winchester, Williamsburg, Staunton‐Harrisonburg, Hampton, Richmond, and Northern Virginia (including Alexandria, Arlington, Fairfax County, Loudoun County, and Prince William County). The total population living in our study area is 5 968 587, which is about 69% of the entire population living in Virginia in 2021. Although it would be ideal to include the entire state, many regions in Virginia do not maintain sufficiently detailed public transit data, especially in rural areas. The nine regions were selected as related transit data was available. Each region represented a diverse array of geographical locations and urban–rural mix. Small cities like Greenville and Altavista offered a contrast to the bigger and more urban settings of Richmond, Northern Virginia, and Hampton. Lynchburg, Williamsburg, Winchester, and Staunton‐Harrisonburg are middle‐sized areas. This combination provided us with a comprehensive overview of varying geographic contexts within Virginia. Data The study used data from various sources to build a realistic public transit model that plays a key role in accurately assessing transit‐based dental care accessibility. First, in terms of the public transit model, general transit feed specification (GTFS) data were compiled , which is essential for analysing public transit schedules and routes from multiple sources, such as primarily public transit websites and public repositories . The GTFS files contained detailed information about transit routes, schedules, and stops, which are crucial for computing accurate transit‐based travel time between origins and destinations . The GTFS files of nearby transit agencies were aggregated to reflect the real‐world transit service coverage area. To measure the spatial accessibility to dental care services, the study utilised building footprints, a polygon or set of polygons that show the total area of a building, assuming that individuals depart their transit trips from each building footprint to dental care services. Building footprints were used since they are the finest spatial resolution units that can be considered to be occupied by humans. The footprint data were obtained from the Virginia Geographic Information Network (VGIN) . Regarding the destination information, the study used comprehensive dental clinic data, including location and participation in Medicaid in each clinic, from the American Dental Association (ADA) 2022 Office Database . This database includes dentists that participate in pediatric Medicaid or CHIP programs based on a roster of dentists listed in the Insure Kids Now website maintained by the Centers for Medicare and Medicaid Services . Statistical Analysis Next, using r5r , an R package designed for transit network analysis , the research team created a high‐resolution schedule‐aware transit network for the study area. For each subregion in the study area, a comprehensive transit network was built, and then a transit‐based origin–destination travel time matrix was computed. The origin included all the geographic coordinates (longitude and latitude) of the centroid of the building footprints in each region, and the destination consisted of the geographic coordinates of the dental care facilities in these regions. Note that the transit‐based origin–destination travel time matrices include detailed trip information, such as in‐vehicle travel time (i.e. ride time) as well as out‐of‐vehicle travel time (e.g. walking to the bus stop or home and waiting time for the bus). Detailed information on travel time estimation was provided in the Data . As a result, accessibility scores were calculated by counting the number of dental clinics that can be reached within 60 min of travel time. The 60‐min threshold was selected considering that it is an acceptable travel time for people using public transit and indicates the point at which accessing healthcare via public transit may impose an undue burden . The accessibility metrics evaluated how easily people could reach dental clinics using public transit . Two accessibility scores were computed and aggregated to the census block group level: (1) access to all dental clinics; (2) access to dental clinics participating in Medicaid. Next, this study examined how residents' primary socio‐economic factors are associated with each accessibility score but focused on dental clinics participating in Medicaid using multiple linear regression models. This group was chosen based on a higher likelihood of Medicaid patients than overall residents taking public transit. Also, only six regions were selected from the included nine regions for this analysis due to the small number of census block groups in Greenville, Altavista, and Lynchburg. Dependent variables were the accessibility scores at the census block group level. Independent variables included primary socio‐economic factors in each region, including population density, poverty percentage and non‐White population that were extracted from the 2021 American Community Survey (ACS) data . R version 4.2.1 was used to conduct rapid transit network analysis and all regression analyses. ArcGIS Pro version 3.2.1 from Esri was used to visualise the accessibility score across the selected cities. The significance level was set at 0.05. Results 3.1 Travel Time to the Nearest Dental Care Clinic for All Dental Clinics and Those Participating in Medicaid Table shows the descriptive statistics of one‐way transit‐based travel times to the nearest dental clinic in the study area. The findings demonstrated substantial differences in travel times to dental clinics across different cities. For example, the average transit travel time to the closest dental clinic was 38 min in Greenville (one of the smallest cities in rural Virginia). Medium‐sized cities like Winchester and Staunton‐Harrisonburg reported average travel times of 29 and 33 min, respectively. In contrast, Richmond, a significantly larger city, had an average travel time of 29 min, which is shorter than other small‐ and medium‐sized cities. Across nine cities, residents spent a considerable portion of their trips outside vehicles (i.e. walking and waiting), ranging from 78% to 89%, indicating challenges in accessing public transit. For instance, in Richmond, 80% of transit travel times consisted of out‐of‐vehicle travel. Moreover, the percentage of relatively short trips that can be made only by walking was calculated. This metric indicated another piece of evidence of inadequate accessibility to dental clinics. For instance, larger cities like Hampton Roads and Northern Virginia showed a higher prevalence of walking‐only trips, approximately 53%–60%, in contrast to smaller cities such as Greenville and Altavista with 19% and 13%, respectively. Overall, these findings highlight travel times to dental care substantially vary across different cities, with larger cities benefiting from more densely located dental clinics, well‐developed sidewalks favourable to walking access and well‐covered public transit networks. In terms of Medicaid users, the travel times to dental clinics were consistently more prolonged than overall residents, suggesting more travel burden faced by Medicaid users. 3.2 Transit‐Based Accessibility to All Clinics and Clinics Participating in Medicaid Table and Figure further show the accessibility scores for dental care clinics in the different cities within Virginia. Figure illustrated that people living closer to regional centers (e.g. downtown) had better transit‐based access to dental clinics compared to those living in suburban or peripheral regions because of more densely concentrated clinics, more frequent transit services and better coverage of transit networks, regardless of regions in the study area. Similar to the travel time results described above, accessibility scores varied substantially by city size. Smaller cities like Greenville and Altavista generally had the lowest accessibility to dental clinics, with an average of 1.5 and 0.8 dental clinics reachable within a 60 min travel radius, respectively. For these two cities, only 50.3% (Greenville) and 21.1% (Altavista) of residents had access to at least one clinic within 60 min of transit trips. On the contrary, larger cities generally had better overall accessibility to dental clinics. For instance, Northern Virginia stood out with an average of 122 dental clinics being accessible within 60 min of transit trips, and 83.1% of residents had access to at least one clinic. In comparison, the accessibility to dental clinics participating in Medicaid was notably lower in smaller regions, with an average accessibility score of 0.898 and 0.178 in Greenville and Altavista, respectively. On the other hand, bigger cities such as Richmond, Hampton Roads and Northern Virginia showed relatively higher accessibility to dental clinics participating in Medicaid compared to other smaller cities, with an average accessibility score of 10.108, 6.204, and 51.241, respectively. Regardless of city size, accessibility to dental clinics participating in Medicaid was significantly lower than all dental clinics. 3.3 Socio‐Economic Inequality in Transit‐Based Accessibility to Dental Clinics Participating in Medicaid Table presented the results of multiple linear regression models that examined the association between transit‐based dental care accessibility scores and primary socio‐economic factors for the dental clinics participating in Medicaid in each of the six selected cities and entire regions. The results revealed that population density had a consistently positive relationship with accessibility scores for almost all the regions except for Williamsburg. On the contrary, the poverty variable showed mixed results. For instance, a positive association was found between poverty percentage and accessibility in Northern Virginia [Coefficient: 0.975; standard error (SE): 0.164] and Richmond [Coefficient: 0.262; SE: 0.036], indicating that higher accessibility to dental clinics that participate in Medicaid can be found in the regions with a higher proportion of low‐income populations. However, the association was weaker or did not exist in other areas, such as Winchester (no association). Similarly, the relationship between non‐White population percentage and accessibility scores also varied across different regions. For example, a positive association was found in Winchester, indicating that a higher percentage of non‐White populations had better accessibility to dental clinics than participating in Medicaid [Coefficient: 0.140; SE: 0.041]. In contrast, a negative association was observed in regions like Northern Virginia [Coefficient: −0.552; SE: 0.068] and Richmond [Coefficient: −0.078; SE: 0.017], suggesting that a higher percentage of non‐White populations living in these areas had lower accessibility to dental clinics participating in Medicaid. In addition to the regional differences, the results also underscore the distinction between individual regions and all regions, with the results in each region capturing the unique socio‐demographic and geographic contexts of a specific area, while the results in all selected regions offer a broader but less nuanced view. Travel Time to the Nearest Dental Care Clinic for All Dental Clinics and Those Participating in Medicaid Table shows the descriptive statistics of one‐way transit‐based travel times to the nearest dental clinic in the study area. The findings demonstrated substantial differences in travel times to dental clinics across different cities. For example, the average transit travel time to the closest dental clinic was 38 min in Greenville (one of the smallest cities in rural Virginia). Medium‐sized cities like Winchester and Staunton‐Harrisonburg reported average travel times of 29 and 33 min, respectively. In contrast, Richmond, a significantly larger city, had an average travel time of 29 min, which is shorter than other small‐ and medium‐sized cities. Across nine cities, residents spent a considerable portion of their trips outside vehicles (i.e. walking and waiting), ranging from 78% to 89%, indicating challenges in accessing public transit. For instance, in Richmond, 80% of transit travel times consisted of out‐of‐vehicle travel. Moreover, the percentage of relatively short trips that can be made only by walking was calculated. This metric indicated another piece of evidence of inadequate accessibility to dental clinics. For instance, larger cities like Hampton Roads and Northern Virginia showed a higher prevalence of walking‐only trips, approximately 53%–60%, in contrast to smaller cities such as Greenville and Altavista with 19% and 13%, respectively. Overall, these findings highlight travel times to dental care substantially vary across different cities, with larger cities benefiting from more densely located dental clinics, well‐developed sidewalks favourable to walking access and well‐covered public transit networks. In terms of Medicaid users, the travel times to dental clinics were consistently more prolonged than overall residents, suggesting more travel burden faced by Medicaid users. Transit‐Based Accessibility to All Clinics and Clinics Participating in Medicaid Table and Figure further show the accessibility scores for dental care clinics in the different cities within Virginia. Figure illustrated that people living closer to regional centers (e.g. downtown) had better transit‐based access to dental clinics compared to those living in suburban or peripheral regions because of more densely concentrated clinics, more frequent transit services and better coverage of transit networks, regardless of regions in the study area. Similar to the travel time results described above, accessibility scores varied substantially by city size. Smaller cities like Greenville and Altavista generally had the lowest accessibility to dental clinics, with an average of 1.5 and 0.8 dental clinics reachable within a 60 min travel radius, respectively. For these two cities, only 50.3% (Greenville) and 21.1% (Altavista) of residents had access to at least one clinic within 60 min of transit trips. On the contrary, larger cities generally had better overall accessibility to dental clinics. For instance, Northern Virginia stood out with an average of 122 dental clinics being accessible within 60 min of transit trips, and 83.1% of residents had access to at least one clinic. In comparison, the accessibility to dental clinics participating in Medicaid was notably lower in smaller regions, with an average accessibility score of 0.898 and 0.178 in Greenville and Altavista, respectively. On the other hand, bigger cities such as Richmond, Hampton Roads and Northern Virginia showed relatively higher accessibility to dental clinics participating in Medicaid compared to other smaller cities, with an average accessibility score of 10.108, 6.204, and 51.241, respectively. Regardless of city size, accessibility to dental clinics participating in Medicaid was significantly lower than all dental clinics. Socio‐Economic Inequality in Transit‐Based Accessibility to Dental Clinics Participating in Medicaid Table presented the results of multiple linear regression models that examined the association between transit‐based dental care accessibility scores and primary socio‐economic factors for the dental clinics participating in Medicaid in each of the six selected cities and entire regions. The results revealed that population density had a consistently positive relationship with accessibility scores for almost all the regions except for Williamsburg. On the contrary, the poverty variable showed mixed results. For instance, a positive association was found between poverty percentage and accessibility in Northern Virginia [Coefficient: 0.975; standard error (SE): 0.164] and Richmond [Coefficient: 0.262; SE: 0.036], indicating that higher accessibility to dental clinics that participate in Medicaid can be found in the regions with a higher proportion of low‐income populations. However, the association was weaker or did not exist in other areas, such as Winchester (no association). Similarly, the relationship between non‐White population percentage and accessibility scores also varied across different regions. For example, a positive association was found in Winchester, indicating that a higher percentage of non‐White populations had better accessibility to dental clinics than participating in Medicaid [Coefficient: 0.140; SE: 0.041]. In contrast, a negative association was observed in regions like Northern Virginia [Coefficient: −0.552; SE: 0.068] and Richmond [Coefficient: −0.078; SE: 0.017], suggesting that a higher percentage of non‐White populations living in these areas had lower accessibility to dental clinics participating in Medicaid. In addition to the regional differences, the results also underscore the distinction between individual regions and all regions, with the results in each region capturing the unique socio‐demographic and geographic contexts of a specific area, while the results in all selected regions offer a broader but less nuanced view. Discussion This study analysed transit‐based dental care accessibility, represented by total travel time and its breakdown (e.g. time spent walking and waiting) to the nearest dental clinics and the number of dental clinics that can be reached within 60 min, and examined the associated socio‐economic factors across various regions in Virginia. People who resided in smaller regions with smaller population sizes faced longer travel times and lower accessibility to dental clinics by public transit compared to those residing in larger regions. These findings were consistent with previous research that measured geospatial accessibility to dental care using driving time and straight‐line distance in other states in the USA . However, people choosing to take public transit in Virginia needed to spend a significantly longer time travelling to dental clinics, given that more than three‐fourths of the time was spent waiting for public transit and walking to dental clinics. The extra out‐vehicle travel time might add additional burden to seek dental care and reduce people's willingness to take public transit for dental care. People who used Medicaid would be more likely to rely on public transit for their mobility needs . It is unsurprising that accessing dental care for Medicaid enrolees is more challenging. The study reported notably lower accessibility to dental clinics participating in Medicaid compared with all dental clinics. The long transit travel time to those dental clinics may decrease patients' willingness to seek care or lead to missed appointments. Many states including Virginia have started to increase the investment in Medicaid program to expand the coverage of eligible enrolees and increase reimbursement rates for healthcare providers. However, it has not shown a notable improvement regarding the availability and accessibility of dental care. For example, Virginia expanded Medicaid coverage to include comprehensive adult dental benefits in July 2021 and offered a better reimbursement rate for dental care providers since July 2022 . However, the impact of new policies on improving health outcomes might take years, and a longitudinal analysis might provide additional insights into the effectiveness of those policies. Additionally, though 41% of the dentists enrolled in Medicaid program in Virginia, only 27% of dentists treated Medicaid/FAMIS enrollees . Also, the study only measured accessibility to the nearest dental clinic, which may not be the real choice for everyone as clinics may not accept all types of private insurance and people often have personal preferences in selecting dental care providers. Thus, the actual accessibility may be underestimated compared to those demonstrated in this study. Considering the long transit travel time found in the study, several strategic interventions could be considered. For example, these potential interventions include the development of public transit systems, location choices of new dental clinics, and availability of school‐based dental clinics or co‐location with other medical and social services. Referring to the various relationships between sociodemographic factors and transit‐based accessibility scores across different regions, tailored interventions should be applied, considering sociodemographic and geographic characteristics of each region to improve local dental care access . There are several limitations that future studies can address. First, this study's approach to assessing transit‐based accessibility did not consider the potential mismatch between demand (i.e. the number of dental care users) and supply (i.e. the number of available doctors). In bigger cities, although there are more dental clinics available than in smaller cities, more potential dental care users may compete for the same services. Therefore, the actual accessibility score that considers the supply–demand mismatch, in reality, might be different from the results reported in this study. Also, a recent study introduced the concept of ‘feels‐like’ accessibility, which considers the different impacts of perceived travel time on accessibility regarding different transit trip segments (e.g. in‐vehicle vs. out‐vehicle travel time) . Future studies can address these issues by utilising advanced accessibility models that account for both the demand and supply, and their complex interactions captured by travel time, such as the two‐step floating catchment area (2SFCA) methods, or ‘feels‐like’ accessibility metrics . Second, this study did not consider the temporal variability of transit schedules across time of day and seasons. For instance, smaller cities with high rates of college populations (e.g. Staunton–Harrisonburg) might have transit schedules that are substantially different between semesters and breaks . Additionally, the study calculated transit travel time based on the assumption that transit operated according to its schedule on time, which may not be the case in reality . Fourth, since building footprint data did not include whether buildings are residential or non‐residential which might impact the study results. For instance, neighbourhoods that have adequate levels of transit services in commercial corridors might overestimate transit‐based accessibility despite fewer people living in those neighbourhoods. Future studies can consider obtaining high‐quality land use data to mitigate this challenge. Lastly, this study did not investigate the potential association between dental health accessibility scores and oral health outcomes, which would be explored in a subsequent study. Conclusion Disparities in dental care accessibility exist depending on the size of cities and Medicaid participation in Virginia. People in smaller cities experience longer travel times to dental clinics by public transit compared to those who reside in larger cities. Those who are enrolled in Medicaid also face more serious challenges compared to the general population. The disproportionate amount of time spent waiting for public transit and walking to destinations affects all residents regardless of where they live and what type of insurance they have. To increase accessibility to dental care using public transit, stakeholders should consider the improvement of public transit systems, location choices of new dental clinics, and the availability of school‐based dental clinics or co‐location with other medical and social services, taking into account the sociodemographic and geographic characteristics of each region. The authors declare no conflicts of interest. Data S1. |
Latest Developments in “Adaptive Enrichment” Clinical Trial Designs in Oncology | 0d8a026e-18ff-40e8-a111-162885b7f276 | 11530510 | Internal Medicine[mh] | As cancer has become better understood on the molecular level with the evolution of gene sequencing techniques, considerations for individualized therapy using predictive biomarkers (those associated with a treatment’s effect) have shifted to a new level. Traditional randomized trial designs tend to either oversimplify or overlook differences in patients’ genetic and molecular profiles, either by fully enriching eligibility to a marker subgroup or enrolling all-comers without prospective use of potentially predictive biomarkers. In the former case of marker enrichment, one cannot learn about a marker’s true predictive ability from the trial’s conduct (as marker-negative patients are excluded); in the latter case ignoring the biomarker, the end result may be a “washing out” of the treatment effect when a predictive marker truly does exist within the sampled patient population. In the last decade or so, randomized “adaptive enrichment” clinical trials have become increasingly utilized to strike a balance between enrolling all patients with a given tumor type, versus enrolling only a subpopulation whose tumors are defined by a potential predictive biomarker related to the mechanism of action of the experimental therapy (see for example ). On a high level, adaptive enrichment designs take the form of a clinical trial that begins by randomizing participants to a targeted versus a control therapy regardless of marker value, then adapts through a series of one or more interim analyses to potentially limit subsequent trial recruitment to a marker-defined patient subpopulation that is showing early signals of enhanced treatment benefit. In this review article, we first discuss the “traditional” presentation of both enrichment and adaptive enrichment designs and their decision rules and describe statistical or practical challenges associated with each. Next, we introduce innovative design extensions and adaptations to adaptive enrichment designs proposed during the last few years in the clinical trial methodology literature, both from Bayesian and frequentist perspectives. Finally, we review articles in which different designs within this class are directly compared or features are examined, and we conclude with some comments on future research directions. Enrichment Trial Designs To motivate discussion of adaptive enrichment designs and why they are useful, it is helpful to first understand enrichment trial designs , or designs that focus only on a subset of the patient population from the beginning. Design Details: In the setting of targeted therapies with strong prior evidence or clinical rationale supporting efficacy only within a biomarker-selected subgroup, “marker-enriched” or enrichment trial designs are used to confirm signal or efficacy only in that selected subgroup. In these types of trials, patients are screened and classified into prespecified marker positive and negative subgroups at or prior to enrollment, with only marker positive patients eligible to remain on study and receive protocol-directed targeted therapy. This usually takes the form of a small, single-arm phase II study without a randomized comparator, but in some settings, comparisons against a randomized non-targeted standard of care therapy might be made (see Fig. ). Example: An example of a clinical trial with an enrichment design is the Herceptin Adjuvant (HERA) trial. The HERA trial is a phase III, randomized, three-arm trial that studied the efficacy of 1 year versus 2 years of adjuvant trastuzumab versus control (no additional treatment) in women with human epidermal growth factor receptor 2 (HER2)-positive early breast cancer after completion of locoregional therapy and chemotherapy . HER2 is overexpressed in 15–25% of breast cancer and trastuzumab, a monoclonal antibody, binds the HER2 extracellular receptor . The primary outcome was disease-free survival and using an intention-to-treat analysis, significant treatment benefit was demonstrated for 1 year of trastuzumab compared to the control arm. Limitations: One important limitation of enrichment designs is that a marker’s predictive ability to select patients for treatment is assumed to already be known and cannot be validated from the trial itself. It is theoretically possible that a pre-defined marker-negative subgroup might also benefit from the targeted treatment, but that knowledge won’t be updated with an enrichment design. For example, a pre-clinical study found that trastuzumab can decrease cancer cell proliferation in HER2 negative and HER2 phosphorylation at tyrosine Y877 positive breast cancer cell lines, which is comparable to the drug effect in HER2 positive breast cancer cell lines, showing that the HER2 negative subpopulation may also benefit from trastuzumab . Around the same time, however, the randomized study B-47 conducted by the National Surgical Adjuvant Breast and Bowel Project (NSABP) group showed no effect of trastuzumab in HER2-low patients . Another limitation of enrichment trial designs is the necessity of establishing predefined subgroups during the study planning phase, which becomes complicated when dealing with biomarkers that are measured on a continuous scale, like expression levels or laboratory values. Determining an appropriate threshold to divide patients into “positive” and “negative” groups is not always straightforward, validated, or effective in distinguishing the effect of the targeted treatment. Selecting an incorrect threshold during trial design can result in an ineffective or underpowered study, and revising the decision once the trial has begun accrual is not advisable. Adaptive Enrichment Trial Designs Adaptive enrichment trial designs, on the other hand, are an attractive solution to the inherent weaknesses of a fully enriched trial design. Design Details: An adaptive enrichment trial design initially enrolls patients with any marker value(s) and randomizes them to experimental targeted versus standard (non-targeted) therapy. As the trial progresses, accrual may be subsequently refined or restricted to patients with certain marker values according to those showing initial efficacy on the basis of one or more interim analyses. This design is randomized out of necessity, so that treatment-by-marker interactions may be computed, and adaptations based on differential treatment effects by marker subgroups can be facilitated. At the interim analyses, according to pre-specified decision rules, a trial may stop early for futility or efficacy, either overall, or within a marker-defined subgroup. If the biomarker of interest is not naturally dichotomous, the same interim analyses may also be used to select or revise marker cutpoints (see Fig. ). Example: One real-world example of an adaptive enrichment design is the Morphotek Investigation in Colorectal Cancer: Research of MORAb-004 (MICRO), which is an adaptive, two-stage, phase II study assessing the effect of ontuxizumab versus placebo in patients with advanced metastatic colorectal cancer . Ontuxizumab, a monoclonal antibody treatment targeting endosialin function, was expected to be more effective in patients with endosialin-related biomarkers. Since the biomarkers were continuous in nature and the optimal cutoffs were unknown, the study included an assessment for determining the best cutoffs at an interim analysis, where progression-free survival (PFS) served as the primary endpoint. Initially, the goal was to demonstrate the treatment effect of ontuxizumab either overall or within subgroups defined by biomarkers. However, the interim analysis revealed that none of the biomarkers had a predictive relationship with treatment outcome. Consequently, the design shifted to a non-marker-driven comparison. Additionally, the interim analysis showed early futility for ontuxizumab compared to placebo overall, terminating the trial early due to lack of efficacy. In summary, this adaptive enrichment design concluded both the biomarker assessment and the evaluation of the therapy early, and additional resources and patients were spared. However, it is worth noting that it may have been underpowered to identify modestly-sized interaction effects, had they been present. Limitations: Adaptive enrichment trial designs do have some statistical challenges, including limitations faced in the design of the MICRO trial. These include estimation of subgroup-specific treatment effects, particularly when the marker prevalence is low, as a sufficiently large sample size is required to have enough patient-level information at interim analysis for informative subgroup selection. As a practical consideration, the primary endpoint must be quickly observed relative to the pace of accrual, to allow time for impactful adaptations based on observed outcomes relatively early in the trial. Another challenge is how exactly one should select cutpoints for adaptation of accrual. In the MICRO trial, at the interim analysis, a series of Cox proportional hazards models were fit over a grid of possible cutpoints, and the significance of a marker by treatment interaction term was evaluated. A pre-specified level of statistical significance for the interaction, along with a clinically meaningful effect in the marker “positive” group defined by the interaction, would warrant potential accrual restriction; however, this approach treated truly continuous biomarkers as binary in its implementation, which results (at least theoretically) in a loss of information and potential loss of power. Several groups have attempted to extend or modify the standard adaptive enrichment trial design in various ways to address statistical shortcomings or tailor the strategy to various applications. The remainder of this paper provides an overview of some of these recent developments. While we admit such designations are rather arbitrary, we present this work separately by Bayesian and frequentist approaches, so that structural similarities among them may be readily described and compared. To motivate discussion of adaptive enrichment designs and why they are useful, it is helpful to first understand enrichment trial designs , or designs that focus only on a subset of the patient population from the beginning. Design Details: In the setting of targeted therapies with strong prior evidence or clinical rationale supporting efficacy only within a biomarker-selected subgroup, “marker-enriched” or enrichment trial designs are used to confirm signal or efficacy only in that selected subgroup. In these types of trials, patients are screened and classified into prespecified marker positive and negative subgroups at or prior to enrollment, with only marker positive patients eligible to remain on study and receive protocol-directed targeted therapy. This usually takes the form of a small, single-arm phase II study without a randomized comparator, but in some settings, comparisons against a randomized non-targeted standard of care therapy might be made (see Fig. ). Example: An example of a clinical trial with an enrichment design is the Herceptin Adjuvant (HERA) trial. The HERA trial is a phase III, randomized, three-arm trial that studied the efficacy of 1 year versus 2 years of adjuvant trastuzumab versus control (no additional treatment) in women with human epidermal growth factor receptor 2 (HER2)-positive early breast cancer after completion of locoregional therapy and chemotherapy . HER2 is overexpressed in 15–25% of breast cancer and trastuzumab, a monoclonal antibody, binds the HER2 extracellular receptor . The primary outcome was disease-free survival and using an intention-to-treat analysis, significant treatment benefit was demonstrated for 1 year of trastuzumab compared to the control arm. Limitations: One important limitation of enrichment designs is that a marker’s predictive ability to select patients for treatment is assumed to already be known and cannot be validated from the trial itself. It is theoretically possible that a pre-defined marker-negative subgroup might also benefit from the targeted treatment, but that knowledge won’t be updated with an enrichment design. For example, a pre-clinical study found that trastuzumab can decrease cancer cell proliferation in HER2 negative and HER2 phosphorylation at tyrosine Y877 positive breast cancer cell lines, which is comparable to the drug effect in HER2 positive breast cancer cell lines, showing that the HER2 negative subpopulation may also benefit from trastuzumab . Around the same time, however, the randomized study B-47 conducted by the National Surgical Adjuvant Breast and Bowel Project (NSABP) group showed no effect of trastuzumab in HER2-low patients . Another limitation of enrichment trial designs is the necessity of establishing predefined subgroups during the study planning phase, which becomes complicated when dealing with biomarkers that are measured on a continuous scale, like expression levels or laboratory values. Determining an appropriate threshold to divide patients into “positive” and “negative” groups is not always straightforward, validated, or effective in distinguishing the effect of the targeted treatment. Selecting an incorrect threshold during trial design can result in an ineffective or underpowered study, and revising the decision once the trial has begun accrual is not advisable. Adaptive enrichment trial designs, on the other hand, are an attractive solution to the inherent weaknesses of a fully enriched trial design. Design Details: An adaptive enrichment trial design initially enrolls patients with any marker value(s) and randomizes them to experimental targeted versus standard (non-targeted) therapy. As the trial progresses, accrual may be subsequently refined or restricted to patients with certain marker values according to those showing initial efficacy on the basis of one or more interim analyses. This design is randomized out of necessity, so that treatment-by-marker interactions may be computed, and adaptations based on differential treatment effects by marker subgroups can be facilitated. At the interim analyses, according to pre-specified decision rules, a trial may stop early for futility or efficacy, either overall, or within a marker-defined subgroup. If the biomarker of interest is not naturally dichotomous, the same interim analyses may also be used to select or revise marker cutpoints (see Fig. ). Example: One real-world example of an adaptive enrichment design is the Morphotek Investigation in Colorectal Cancer: Research of MORAb-004 (MICRO), which is an adaptive, two-stage, phase II study assessing the effect of ontuxizumab versus placebo in patients with advanced metastatic colorectal cancer . Ontuxizumab, a monoclonal antibody treatment targeting endosialin function, was expected to be more effective in patients with endosialin-related biomarkers. Since the biomarkers were continuous in nature and the optimal cutoffs were unknown, the study included an assessment for determining the best cutoffs at an interim analysis, where progression-free survival (PFS) served as the primary endpoint. Initially, the goal was to demonstrate the treatment effect of ontuxizumab either overall or within subgroups defined by biomarkers. However, the interim analysis revealed that none of the biomarkers had a predictive relationship with treatment outcome. Consequently, the design shifted to a non-marker-driven comparison. Additionally, the interim analysis showed early futility for ontuxizumab compared to placebo overall, terminating the trial early due to lack of efficacy. In summary, this adaptive enrichment design concluded both the biomarker assessment and the evaluation of the therapy early, and additional resources and patients were spared. However, it is worth noting that it may have been underpowered to identify modestly-sized interaction effects, had they been present. Limitations: Adaptive enrichment trial designs do have some statistical challenges, including limitations faced in the design of the MICRO trial. These include estimation of subgroup-specific treatment effects, particularly when the marker prevalence is low, as a sufficiently large sample size is required to have enough patient-level information at interim analysis for informative subgroup selection. As a practical consideration, the primary endpoint must be quickly observed relative to the pace of accrual, to allow time for impactful adaptations based on observed outcomes relatively early in the trial. Another challenge is how exactly one should select cutpoints for adaptation of accrual. In the MICRO trial, at the interim analysis, a series of Cox proportional hazards models were fit over a grid of possible cutpoints, and the significance of a marker by treatment interaction term was evaluated. A pre-specified level of statistical significance for the interaction, along with a clinically meaningful effect in the marker “positive” group defined by the interaction, would warrant potential accrual restriction; however, this approach treated truly continuous biomarkers as binary in its implementation, which results (at least theoretically) in a loss of information and potential loss of power. Several groups have attempted to extend or modify the standard adaptive enrichment trial design in various ways to address statistical shortcomings or tailor the strategy to various applications. The remainder of this paper provides an overview of some of these recent developments. While we admit such designations are rather arbitrary, we present this work separately by Bayesian and frequentist approaches, so that structural similarities among them may be readily described and compared. Bayesian Approaches Xu et al. proposed an adaptive enrichment randomized two-arm design that combines exploration of treatment benefit subgroups and estimation of subgroup-specific effects in the context of a multilevel target product profile, where both minimal and targeted treatment effect thresholds are investigated . This adaptive subgroup-identification enrichment design (ASIED) opens for all-comers first, and subgroups identified as having enhanced treatment effects are selected at an interim analysis, where pre-set minimum and targeted treatment effects are evaluated against a set of decision criteria for futility or efficacy stopping for all-comers or possible subgroups. A Bayesian random partition (BayRP) model for subgroup-identification is incorporated into ASIED, based on models proposed by Xu et al. and Guo et al. . Due to the flexibility of the BayRP model, biomarkers can be continuous, binary, categorical, or ordinal, and the primary endpoint types can be binary, categorical, or continuous. Per the authors, extensions to count or survival outcomes are also possible. BayRP was implemented due to its robustness, but other Bayesian subgroup identification methods could be used as well, like Bayesian additive regression tree (BART) or random forests for larger sample sizes . A tree-type random partition of biomarkers is used as a prior and an equally spaced k-dimensional grid constructed from k biomarkers is used to represent possible biomarker profiles. The operating characteristics of ASIED as a trial design was evaluated by simulations with 4 continuous biomarkers, a total sample size of 180, an interim analysis after 100 patients were enrolled, a minimum desired treatment effect of 2.37 and target treatment effect of 3.08 on a continuous score scale. ASIED’s recommendations were close to the expected results. However, the number of simulated trials was only 100, which could yield lower precision of the estimated operating characteristics. Another limitation is that the partition of the biomarker profile was limited to at most four biomarker subgroups due to the small sample size in each partition. Another Bayesian randomized group-sequential adaptive enrichment two-arm design incorporating multiple baseline biomarkers was proposed by Park et al. . The design’s primary endpoint is time-to-event, while a binary early response acts as a surrogate endpoint assisting with biomarker pruning and enrichment to a sensitive population at each interim analysis. Initially, the study is open for all-comers and the baseline biomarkers can be binary, continuous, or categorical. The first step at each interim analysis is to jointly select covariates based on both the surrogate and final endpoints by checking each treatment by covariate interaction. The second step is to recalculate the personalized benefit index (PBI), which is a weighted average posterior probability indicating patients with selected biomarkers who benefit more from the experimental treatment. The refitted regression from the variable selection step will redefine the treatment-sensitive patients, and only patients with PBI values larger than some pre-specified cutoff continue to be enrolled to the trial. The third step is to test for futility and efficacy stopping by a Bayesian group sequential test procedure for the previously identified treatment-sensitive subgroups. In simulations, AED was compared with group sequential enriched designs called InterAdapt and GSED, an adaptive enrichment design and all- comers group sequential design . The maximum sample size considered was 400, and patients were accrued by a Poisson process with 100 patients per year. Two interim analyses took place after 200 and 300 patients enrolled, and 10 baseline biomarkers were considered. Across each of the seven scenarios, prevalence of the treatment-sensitive group was set to be 0.65, 0.50, or 0.35. While nearly all the designs controlled the nominal Type I error to 0.05, AED had higher probabilities of identifying the sensitive subgroup and correctly concluding efficacy than other designs. Also, 1000 future patients were simulated and treated by each design’s suggested treatment, and AED had the longest median survival time overall. One stated limitation of this work was its inability to handle high dimensional baseline biomarker covariates, as the authors suggest considering no more than 50 baseline covariates in total. Also, biomarkers in this design are assumed to be independent, though selection adjustment for correlated predictors is mentioned. It is worth noting that early response (as used by this design) has not been validated as a good surrogate for longer-term clinical endpoints. To address the scenario of a single continuous predictive biomarker where the marker-treatment relationship is continuous instead of a step function, Ohwada and Morita proposed a Bayesian adaptive patient enrollment restriction (BAPER) design that can restrict the subsequent enrollment of treatment insensitive biomarker-based subgroups based on interim analyses . The primary endpoint is assumed to be time-to-event, and the relationship between the biomarker and treatment effect is assumed to increase monotonically and is modeled via a four-parameter change-point model within a proportional hazard model. Parameters are assumed to follow non-informative priors, and the posterior distributions are calculated using the partial likelihood of the Cox proportional hazard model. At each interim analysis, decisions can be made for a subgroup or the overall cohort. In addition, treatment-sensitive patients can be selected based on a biomarker cutoff value, which is determined by searching over the range of biomarker values and picking the one with the highest conditional posterior probability of achieving the target treatment effect. Simulations were conducted to compare the proposed method against both a similar method without enrichment and a design using a step-function to model marker-treatment interaction effects without enrichment. The maximum sample size considered was 240 with two interim analyses, and the assumed target hazard ratio was 0.6. The results show that the proposed BAPER method decreases the average number of enrolled patients who will not experience the targeted treatment effect, compared to designs without patient selection. Also, BAPER has a higher probability of correctly identifying the cutoff point that achieves the target hazard ratio. However, BAPER has certain restrictions: the biomarker cannot be prognostic, as the main effect for the biomarker is excluded from the proportional hazard model. Also, the design does not consider the distribution of the biomarker values themselves, so a larger sample size is required when the prevalence of the treatment sensitive (or insensitive) population is small. Focusing on an optimal decision threshold for a binary biomarker which is either potentially predictive or both prognostic and predictive, Krisam and Kieser proposed a new class of interim decision rules for a two-stage, two-arm adaptive enrichment design . This approach is an extension of Jenkins et al.’s design but with a binary endpoint instead of a time-to-event outcome . Initially, their trial randomizes all patients from two distinct subgroups (i.e., a binary biomarker), assuming one subgroup will have greater benefit, and the sample size is fixed per stages by treatment group. At the first interim analysis, the trial might stop early for futility, continue enrolling to only the marker-positive group, or continue enrolling the full population, while using Hochberg multiplicity- corrected p-values for these decisions. When the full population proceeds to the second stage, it remains possible that efficacy testing will be performed both overall and in the treatment-sensitive subgroup if the biomarker is found to be predictive or prognostic, or only within the total population if the biomarker is not predictive. The critical boundaries for subgroup decisions minimize the Bayes risk of a quadratic loss function by setting the roots of partial derivatives as optimal thresholds, assuming the estimated treatment effects follow bivariate normal distributions with design parameters from uniform prior distributions. A relevance threshold for the effect size, which serves as the minimal clinical meaningful effect, also needs to be prespecified. Optimal decision threshold tables are presented for a biomarker that is predictive, both predictive and prognostic, or non-informative, with sample sizes ranging from 20 to 400 and subgroup prevalence values of 0.1, 0.25 and 0.5 considered. In their simulations, the sample size is 200 per group per stage (for a total trial sample size of 800), the treatment effect (response rate) in one of the subgroups is 0.15, and the biomarker is both predictive and prognostic. Optimal decision rules with three different assumptions for the biomarkers (predictive, predictive and prognostic, non-informative) and subgroup prevalence are compared with a rule just based on relevance thresholds. Power is increased under the proposed decision rules when the correct biomarker assumption is made. Since the decision thresholds incorporate sample size and subgroup prevalence information, one major limitation is that knowledge about the biomarkers must be strong enough pre-trial to prespecify the required parameters. Nesting frequentist testing procedures within a Bayesian framework, Simon and Simon proposed a group-sequential randomized adaptive enrichment trial design that uses frequentist hypothesis tests for controlling Type I error but Bayesian modeling to select treatment-sensitive subgroups and estimate effect size . The primary endpoint in their models is binary, and multiple continuous biomarkers are allowed, comprising a vector of covariates for each patient. Patients are sequentially enrolled in a total of K blocks, and enrollment criteria for the next block are refined by a decision function, which is built on the block adaptive enrichment design by Simon and Simon . The final analysis is based on inverse normal combination test statistics using data from the entire trial. A prior for the response rate in each arm needs to be prespecified, which is based on both the biomarker covariates and a utility function. Different utility functions can be applied according to the trial’s goal, and the one adopted here is the expected future patient outcome penalized by accrual time. Using the conditional posterior for the previous block’s information, simulations are conducted to find the optimal enrollment criteria based on the utility function. The expected treatment effect given covariates can be estimated by the posterior predictive distribution for the response rate at the end of trial. In the presented simulation study, there are two continuous biomarkers and 300 patients accrued in two or three enrollment blocks, with three logistic and three cutpoint models for the biomarker-response relationships. An unenriched design and an adaptive enrichment strategy with prespecified fixed cutpoints are compared with the proposed design. The two adaptive enrichment designs have higher power than the unenriched design to detect a treatment sensitive subgroup, and the enrichment designs have higher power when there are three versus two enrollment blocks. Compared with the fixed cutpoint enrichment method, the proposed design generally correctly identifies the treatment-sensitive subgroup while avoiding non-ideal pre-determined cutoff points for the following enrollment criteria. Though the effect size estimation is biased under the proposed design, the bias is more severe under the unenriched design. Graf et al. proposed to optimize design decisions using utility functions from the sponsor and public health points of view in the context of a two-stage adaptive enrichment design with a continuous biomarker . Similar to Simon and Simon’s method, the proposed design’s decisions are based on frequentist hypothesis tests, while the utility functions are evaluated under the Bayesian approach. In this design, patients are classified into marker positive and marker negative groups at enrollment, and decisions can be made with respect to the full population or the marker positive subgroup only. Closed testing procedures along with Hochberg tests are used to control the family wise type I error rate. Parameters called “gain”, which quantify the benefit rendered by the trial to the sponsor and society, need to be pre-specified. The utility function under the sponsor view is the sum of the gain multiplied by the probability of claiming treatment efficacy in the full population or a marker-positive group, respectively. In addition to gain and success probabilities, the public health utility function also considers the true effect sizes in subgroups, and safety risk as a penalization parameter. Prior distributions are used to model treatment effects in each subgroup to account for uncertainty, but the authors assume that only the marker negative group can be ineffective, and only point priors are used, which leads to a single probability that the treatment is effective in just the marker positive subgroup or the full population. This optimized adaptive design is compared with a non-adaptive design when the total sample sizes are the same. The adaptive design provides larger expected utility in both utility functions only when the values are intermediate in gain from treatment efficacy and the prior point probability. One limitation is that those utility functions can only compare designs with the same total sample size and the cost of running a trial is not included. Serving as an extension of Graf et al.’s work by incorporating a term for the trial cost in utility functions, Ondra et al. derived an adaptive two-stage partial enrichment design for a normally distributed outcome with subgroup selection and optimization of the second stage sample size . In a partial enrichment design, the proportion of the marker-positive subjects enrolled does not need to be aligned with the true prevalence. At interim analysis, the trial can be stopped for futility, or continued in only the marker-positive population or the full population. The final analysis is based on the weighted inverse normal function with Bonferroni correction. Utility functions used for optimization are from societal or sponsor perspectives. Expected utility is calculated by numerical integration on the joint sampling distribution of two stage-wise test statistics, with the prior distributions for the treatment effect in each subgroup. The optimal sample size for the second stage maximizes the conditional expected utility given the first stage test statistics and sample size used, and the optimal first stage sample size maximizes the utility using the solved optimal number for the second stage. The optimization function is solved recursively by dynamic programming, and the optimal design in terms of the sample size is obtained. The optimized adaptive enrichment design is compared with an optimized single- stage design for subgroup prevalence ranging from 10 to 90%, with both weak and strong predictive biomarker priors considered. Expected utilities are higher in both sponsor and societal views in the adaptive design. Also, even if the prior distribution for the effect size used in the design differs from the true distribution, the proposed adaptive design is robust in terms of expected utilities when the biomarker’s prevalence is high enough. One limitation is that the endpoint needs to be observed immediately, which might be addressed by a short-term surrogate endpoint—though to date, validated short-term endpoints are rare in oncology. Frequentist Approaches Fisher et al. proposed an adaptive multi-stage enrichment design that allows sub-group selection at an interim analysis with continuous or binary outcomes . Two subpopulations are predefined, and the goal is to claim treatment efficacy in one of the subpopulations or the full population. The cumulative test statistics for the subgroups and the full population are calculated at each interim analysis and compared against efficacy and non-binding futility boundaries. To control the family-wise Type I error rate (FWER), two methods for constructing efficacy boundaries are presented. One is proposed by Rosenblum et al. that spends alpha based on the covariance matrix of test statistics by populations (two subpopulations and the full population) and by interim stages . Another is the alpha reallocation approach . The design parameters, including sample size per stage, futility boundaries, etc., are optimized to minimize the expected number enrolled or expected trial duration using simulated annealing, with constraints on power and Type I error. If the resulting design does not meet the power requirement, the total sample size will be increased until the power requirement is met. The optimized adaptive design is compared with a single-stage design, optimized single-stage design, and a multi-stage group sequential design with O’Brien-Fleming or Pocock boundaries using actual trial data from MISTIE and ADNI . For the MISTIE trail, the proposed designs are optimized by the expected number enrolled, which is lower than for the optimized single-stage design and group-sequential design, but the maximum number enrolled is still lower in the simple single-stage design. In the ADNI trial, when the expected trial duration is optimized, the proposed design has a slightly shorter expected duration but a longer maximum duration than the optimized single-stage design. Similar to the aforementioned Bayesian approaches without predefined sub-populations, Zhang et al. proposed a two-stage adaptive enrichment design that does not require predefined subgroups . The primary outcome is binary, and a collection of baseline covariates, including biomarkers and demographics, is used to define a treatment-sensitive subgroup. The selection criteria are based on a prespecified function modeling the treatment effect and marker by treatment interaction using first stage data. The final treatment effect estimate is a weighted average of estimates in each stage. To minimize the resubstitution bias from using first stage data in subsequent subgroup selection and inference, four methods for estimating the treatment effect and variance for the first stage are discussed: naive approach, cross-validation, nonparametric bootstrap, and parametric bootstrap. To compare those estimation methods, ECHO and THRIVE trial data are used for the simulation with a total sample size of 1000. The first stage has 250, 500 or 750 subjects, and the function used to simulate outcomes is the logistic regression model. The results show that the bootstrap method is more favorable than both the naive estimate (which has a large empirical bias) and the cross-validation method (which is overly conservative). The weight for each stage and first stage sample size need to be selected carefully to reach a small root mean squared error (RMSE) and close-to-nominal one-sided coverage. Though a trial can stop due to inability to recruit to a subset resulting from restricted enrollment, the proposed method does not include an early stopping rule for futility or efficacy. In order to reduce sample size while assessing the treatment effect in the full population, Matsui and Crowley proposed a two-stage subgroup-focused sequential design for time-to-event outcomes, which could extend to multiple stages . In this design, patients are classified into two subgroups by a dichotomized predictive marker, with the assumption that the experimental treatment is more efficacious in the marker-positive subgroup. The trial can proceed to the second stage with one of the subgroups, or the full population, but treatment efficacy is only tested in the marker-positive group or the full population at the final analysis. Choices of testing procedures are fixed-sequence and split-alpha. At the interim analysis, a superiority boundary for the marker-positive subgroup and a futility boundary for the marker-negative subgroup are constructed. The superiority boundary is calculated to control the study-wide alpha level, while the futility boundary is based on a Bayesian posterior probability of efficacy with a non-informative prior. The required sample sizes for each subgroup are calculated separately, and the hazard ratio for the marker-positive subgroup is recommended to be 0.05–0.70 under this application. The proposed design is compared with a traditional all-comers design, an enriched design with only marker-positive subjects, a two-stage enriched design, and a traditional marker-stratified design. Different scenarios are considered including those with no treatment effect, constant treatment effect in both groups with hazard ratio (HR) = 0.75, a nearly qualitative interaction with HRs = 0.65 and 1, and a quantitative interaction with HRs = 0.7 and 0.8. The marker prevalence is set to 0.4, and the accrual rate is 200 patients per year. When using the split-alpha test, the proposed design has greater than 80% power to reject any null hypothesis in the alternative cases, but the traditional marker-stratified design also provides enough power under all cases. The number screened and the number randomized are reduced for the proposed design compared to the traditional marker stratified design, but the reduction is only moderate. To determine whether the full population or only the biomarker-positive subgroup benefit more from the experimental treatment, Uozumi and Hamada proposed a two-stage adaptive population selection design for a time-to-event outcome, an extension of methods from Brannath et al. and Jenkins et al. . The main extension is that the decision-making strategy at the interim analysis incorporates both progression-free survival (PSF) and overall survival (OS) information. Also, OS is decomposed into time-to-progression (TTP) and post-progression survival (PPS) when tumor progression has occurred, to account for the correlation between OS and PFS. The combination test approach is used for the final analysis based on Simes’ procedure . The hypothesis rejection rule for each population is a weighted inverse normal combination function with prespecified weights based on the expected number of OS events in each stage. At the interim analysis, a statistical model from Fleischer et al. under the semi-competing risks framework is applied to account for the correlation between OS and PFS . The interim decision rule uses the predictive power approach in each population, extending Brannath et al.’s method from single endpoint to multiple endpoints with a higher weight on PFS data due to its rapid observation. In the simulation, a dichotomized biomarker is used with a 50% prevalence. Four scenarios are considered, where hazard ratios in the marker-positive subgroup are always 0.5 and are higher in the marker-negative subgroup. For simplicity, the HR is the same for TTP, PPS, and death. FWER is controlled for all cases, but it is a little too conservative when the treatment is effective. The proposed design has a higher probability of identifying the treatment-sensitive population at the interim analysis, particularly when the PPS effect is large, those probabilities are similar between using OS or PFS alone or the combined endpoints when the PFS effect is small. One limitation of this design is that sample size calculations are not considered. Instead of a single primary endpoint, Sinha et al. suggested a two-stage Phase III design with population enrichment for two binary co-primary endpoints, which is an extension of Magnusson and Turnbull’s work with co-primary endpoints . The two binary endpoints are assumed to be independent, and the efficacy goal should be reached in both endpoints. With two distinct predefined subgroups, a set of decision rules stops the non-responsive subgroups using efficient score statistics. The futility and efficacy boundary values, which do not depend on the marker prevalence, are the same for both endpoints due to independence. The lower and upper stopping boundaries are calculated by alpha spending functions, and FWER is strongly controlled. Simulations were conducted assuming biomarker prevalences of 0.25 or 0.75 and weighted subgroup effect sizes of 0, 1, and 2 as the means of efficient score statistics under normal distribution. The results show that the proposed design can reduce false-negative results for heterogeneous treatment effects between subgroups. The authors state the possibility of extending the design to a bivariate continuous outcome, while an extension to bivariate survival would be more challenging. Xu et al. proposed an adaptive enrichment randomized two-arm design that combines exploration of treatment benefit subgroups and estimation of subgroup-specific effects in the context of a multilevel target product profile, where both minimal and targeted treatment effect thresholds are investigated . This adaptive subgroup-identification enrichment design (ASIED) opens for all-comers first, and subgroups identified as having enhanced treatment effects are selected at an interim analysis, where pre-set minimum and targeted treatment effects are evaluated against a set of decision criteria for futility or efficacy stopping for all-comers or possible subgroups. A Bayesian random partition (BayRP) model for subgroup-identification is incorporated into ASIED, based on models proposed by Xu et al. and Guo et al. . Due to the flexibility of the BayRP model, biomarkers can be continuous, binary, categorical, or ordinal, and the primary endpoint types can be binary, categorical, or continuous. Per the authors, extensions to count or survival outcomes are also possible. BayRP was implemented due to its robustness, but other Bayesian subgroup identification methods could be used as well, like Bayesian additive regression tree (BART) or random forests for larger sample sizes . A tree-type random partition of biomarkers is used as a prior and an equally spaced k-dimensional grid constructed from k biomarkers is used to represent possible biomarker profiles. The operating characteristics of ASIED as a trial design was evaluated by simulations with 4 continuous biomarkers, a total sample size of 180, an interim analysis after 100 patients were enrolled, a minimum desired treatment effect of 2.37 and target treatment effect of 3.08 on a continuous score scale. ASIED’s recommendations were close to the expected results. However, the number of simulated trials was only 100, which could yield lower precision of the estimated operating characteristics. Another limitation is that the partition of the biomarker profile was limited to at most four biomarker subgroups due to the small sample size in each partition. Another Bayesian randomized group-sequential adaptive enrichment two-arm design incorporating multiple baseline biomarkers was proposed by Park et al. . The design’s primary endpoint is time-to-event, while a binary early response acts as a surrogate endpoint assisting with biomarker pruning and enrichment to a sensitive population at each interim analysis. Initially, the study is open for all-comers and the baseline biomarkers can be binary, continuous, or categorical. The first step at each interim analysis is to jointly select covariates based on both the surrogate and final endpoints by checking each treatment by covariate interaction. The second step is to recalculate the personalized benefit index (PBI), which is a weighted average posterior probability indicating patients with selected biomarkers who benefit more from the experimental treatment. The refitted regression from the variable selection step will redefine the treatment-sensitive patients, and only patients with PBI values larger than some pre-specified cutoff continue to be enrolled to the trial. The third step is to test for futility and efficacy stopping by a Bayesian group sequential test procedure for the previously identified treatment-sensitive subgroups. In simulations, AED was compared with group sequential enriched designs called InterAdapt and GSED, an adaptive enrichment design and all- comers group sequential design . The maximum sample size considered was 400, and patients were accrued by a Poisson process with 100 patients per year. Two interim analyses took place after 200 and 300 patients enrolled, and 10 baseline biomarkers were considered. Across each of the seven scenarios, prevalence of the treatment-sensitive group was set to be 0.65, 0.50, or 0.35. While nearly all the designs controlled the nominal Type I error to 0.05, AED had higher probabilities of identifying the sensitive subgroup and correctly concluding efficacy than other designs. Also, 1000 future patients were simulated and treated by each design’s suggested treatment, and AED had the longest median survival time overall. One stated limitation of this work was its inability to handle high dimensional baseline biomarker covariates, as the authors suggest considering no more than 50 baseline covariates in total. Also, biomarkers in this design are assumed to be independent, though selection adjustment for correlated predictors is mentioned. It is worth noting that early response (as used by this design) has not been validated as a good surrogate for longer-term clinical endpoints. To address the scenario of a single continuous predictive biomarker where the marker-treatment relationship is continuous instead of a step function, Ohwada and Morita proposed a Bayesian adaptive patient enrollment restriction (BAPER) design that can restrict the subsequent enrollment of treatment insensitive biomarker-based subgroups based on interim analyses . The primary endpoint is assumed to be time-to-event, and the relationship between the biomarker and treatment effect is assumed to increase monotonically and is modeled via a four-parameter change-point model within a proportional hazard model. Parameters are assumed to follow non-informative priors, and the posterior distributions are calculated using the partial likelihood of the Cox proportional hazard model. At each interim analysis, decisions can be made for a subgroup or the overall cohort. In addition, treatment-sensitive patients can be selected based on a biomarker cutoff value, which is determined by searching over the range of biomarker values and picking the one with the highest conditional posterior probability of achieving the target treatment effect. Simulations were conducted to compare the proposed method against both a similar method without enrichment and a design using a step-function to model marker-treatment interaction effects without enrichment. The maximum sample size considered was 240 with two interim analyses, and the assumed target hazard ratio was 0.6. The results show that the proposed BAPER method decreases the average number of enrolled patients who will not experience the targeted treatment effect, compared to designs without patient selection. Also, BAPER has a higher probability of correctly identifying the cutoff point that achieves the target hazard ratio. However, BAPER has certain restrictions: the biomarker cannot be prognostic, as the main effect for the biomarker is excluded from the proportional hazard model. Also, the design does not consider the distribution of the biomarker values themselves, so a larger sample size is required when the prevalence of the treatment sensitive (or insensitive) population is small. Focusing on an optimal decision threshold for a binary biomarker which is either potentially predictive or both prognostic and predictive, Krisam and Kieser proposed a new class of interim decision rules for a two-stage, two-arm adaptive enrichment design . This approach is an extension of Jenkins et al.’s design but with a binary endpoint instead of a time-to-event outcome . Initially, their trial randomizes all patients from two distinct subgroups (i.e., a binary biomarker), assuming one subgroup will have greater benefit, and the sample size is fixed per stages by treatment group. At the first interim analysis, the trial might stop early for futility, continue enrolling to only the marker-positive group, or continue enrolling the full population, while using Hochberg multiplicity- corrected p-values for these decisions. When the full population proceeds to the second stage, it remains possible that efficacy testing will be performed both overall and in the treatment-sensitive subgroup if the biomarker is found to be predictive or prognostic, or only within the total population if the biomarker is not predictive. The critical boundaries for subgroup decisions minimize the Bayes risk of a quadratic loss function by setting the roots of partial derivatives as optimal thresholds, assuming the estimated treatment effects follow bivariate normal distributions with design parameters from uniform prior distributions. A relevance threshold for the effect size, which serves as the minimal clinical meaningful effect, also needs to be prespecified. Optimal decision threshold tables are presented for a biomarker that is predictive, both predictive and prognostic, or non-informative, with sample sizes ranging from 20 to 400 and subgroup prevalence values of 0.1, 0.25 and 0.5 considered. In their simulations, the sample size is 200 per group per stage (for a total trial sample size of 800), the treatment effect (response rate) in one of the subgroups is 0.15, and the biomarker is both predictive and prognostic. Optimal decision rules with three different assumptions for the biomarkers (predictive, predictive and prognostic, non-informative) and subgroup prevalence are compared with a rule just based on relevance thresholds. Power is increased under the proposed decision rules when the correct biomarker assumption is made. Since the decision thresholds incorporate sample size and subgroup prevalence information, one major limitation is that knowledge about the biomarkers must be strong enough pre-trial to prespecify the required parameters. Nesting frequentist testing procedures within a Bayesian framework, Simon and Simon proposed a group-sequential randomized adaptive enrichment trial design that uses frequentist hypothesis tests for controlling Type I error but Bayesian modeling to select treatment-sensitive subgroups and estimate effect size . The primary endpoint in their models is binary, and multiple continuous biomarkers are allowed, comprising a vector of covariates for each patient. Patients are sequentially enrolled in a total of K blocks, and enrollment criteria for the next block are refined by a decision function, which is built on the block adaptive enrichment design by Simon and Simon . The final analysis is based on inverse normal combination test statistics using data from the entire trial. A prior for the response rate in each arm needs to be prespecified, which is based on both the biomarker covariates and a utility function. Different utility functions can be applied according to the trial’s goal, and the one adopted here is the expected future patient outcome penalized by accrual time. Using the conditional posterior for the previous block’s information, simulations are conducted to find the optimal enrollment criteria based on the utility function. The expected treatment effect given covariates can be estimated by the posterior predictive distribution for the response rate at the end of trial. In the presented simulation study, there are two continuous biomarkers and 300 patients accrued in two or three enrollment blocks, with three logistic and three cutpoint models for the biomarker-response relationships. An unenriched design and an adaptive enrichment strategy with prespecified fixed cutpoints are compared with the proposed design. The two adaptive enrichment designs have higher power than the unenriched design to detect a treatment sensitive subgroup, and the enrichment designs have higher power when there are three versus two enrollment blocks. Compared with the fixed cutpoint enrichment method, the proposed design generally correctly identifies the treatment-sensitive subgroup while avoiding non-ideal pre-determined cutoff points for the following enrollment criteria. Though the effect size estimation is biased under the proposed design, the bias is more severe under the unenriched design. Graf et al. proposed to optimize design decisions using utility functions from the sponsor and public health points of view in the context of a two-stage adaptive enrichment design with a continuous biomarker . Similar to Simon and Simon’s method, the proposed design’s decisions are based on frequentist hypothesis tests, while the utility functions are evaluated under the Bayesian approach. In this design, patients are classified into marker positive and marker negative groups at enrollment, and decisions can be made with respect to the full population or the marker positive subgroup only. Closed testing procedures along with Hochberg tests are used to control the family wise type I error rate. Parameters called “gain”, which quantify the benefit rendered by the trial to the sponsor and society, need to be pre-specified. The utility function under the sponsor view is the sum of the gain multiplied by the probability of claiming treatment efficacy in the full population or a marker-positive group, respectively. In addition to gain and success probabilities, the public health utility function also considers the true effect sizes in subgroups, and safety risk as a penalization parameter. Prior distributions are used to model treatment effects in each subgroup to account for uncertainty, but the authors assume that only the marker negative group can be ineffective, and only point priors are used, which leads to a single probability that the treatment is effective in just the marker positive subgroup or the full population. This optimized adaptive design is compared with a non-adaptive design when the total sample sizes are the same. The adaptive design provides larger expected utility in both utility functions only when the values are intermediate in gain from treatment efficacy and the prior point probability. One limitation is that those utility functions can only compare designs with the same total sample size and the cost of running a trial is not included. Serving as an extension of Graf et al.’s work by incorporating a term for the trial cost in utility functions, Ondra et al. derived an adaptive two-stage partial enrichment design for a normally distributed outcome with subgroup selection and optimization of the second stage sample size . In a partial enrichment design, the proportion of the marker-positive subjects enrolled does not need to be aligned with the true prevalence. At interim analysis, the trial can be stopped for futility, or continued in only the marker-positive population or the full population. The final analysis is based on the weighted inverse normal function with Bonferroni correction. Utility functions used for optimization are from societal or sponsor perspectives. Expected utility is calculated by numerical integration on the joint sampling distribution of two stage-wise test statistics, with the prior distributions for the treatment effect in each subgroup. The optimal sample size for the second stage maximizes the conditional expected utility given the first stage test statistics and sample size used, and the optimal first stage sample size maximizes the utility using the solved optimal number for the second stage. The optimization function is solved recursively by dynamic programming, and the optimal design in terms of the sample size is obtained. The optimized adaptive enrichment design is compared with an optimized single- stage design for subgroup prevalence ranging from 10 to 90%, with both weak and strong predictive biomarker priors considered. Expected utilities are higher in both sponsor and societal views in the adaptive design. Also, even if the prior distribution for the effect size used in the design differs from the true distribution, the proposed adaptive design is robust in terms of expected utilities when the biomarker’s prevalence is high enough. One limitation is that the endpoint needs to be observed immediately, which might be addressed by a short-term surrogate endpoint—though to date, validated short-term endpoints are rare in oncology. Fisher et al. proposed an adaptive multi-stage enrichment design that allows sub-group selection at an interim analysis with continuous or binary outcomes . Two subpopulations are predefined, and the goal is to claim treatment efficacy in one of the subpopulations or the full population. The cumulative test statistics for the subgroups and the full population are calculated at each interim analysis and compared against efficacy and non-binding futility boundaries. To control the family-wise Type I error rate (FWER), two methods for constructing efficacy boundaries are presented. One is proposed by Rosenblum et al. that spends alpha based on the covariance matrix of test statistics by populations (two subpopulations and the full population) and by interim stages . Another is the alpha reallocation approach . The design parameters, including sample size per stage, futility boundaries, etc., are optimized to minimize the expected number enrolled or expected trial duration using simulated annealing, with constraints on power and Type I error. If the resulting design does not meet the power requirement, the total sample size will be increased until the power requirement is met. The optimized adaptive design is compared with a single-stage design, optimized single-stage design, and a multi-stage group sequential design with O’Brien-Fleming or Pocock boundaries using actual trial data from MISTIE and ADNI . For the MISTIE trail, the proposed designs are optimized by the expected number enrolled, which is lower than for the optimized single-stage design and group-sequential design, but the maximum number enrolled is still lower in the simple single-stage design. In the ADNI trial, when the expected trial duration is optimized, the proposed design has a slightly shorter expected duration but a longer maximum duration than the optimized single-stage design. Similar to the aforementioned Bayesian approaches without predefined sub-populations, Zhang et al. proposed a two-stage adaptive enrichment design that does not require predefined subgroups . The primary outcome is binary, and a collection of baseline covariates, including biomarkers and demographics, is used to define a treatment-sensitive subgroup. The selection criteria are based on a prespecified function modeling the treatment effect and marker by treatment interaction using first stage data. The final treatment effect estimate is a weighted average of estimates in each stage. To minimize the resubstitution bias from using first stage data in subsequent subgroup selection and inference, four methods for estimating the treatment effect and variance for the first stage are discussed: naive approach, cross-validation, nonparametric bootstrap, and parametric bootstrap. To compare those estimation methods, ECHO and THRIVE trial data are used for the simulation with a total sample size of 1000. The first stage has 250, 500 or 750 subjects, and the function used to simulate outcomes is the logistic regression model. The results show that the bootstrap method is more favorable than both the naive estimate (which has a large empirical bias) and the cross-validation method (which is overly conservative). The weight for each stage and first stage sample size need to be selected carefully to reach a small root mean squared error (RMSE) and close-to-nominal one-sided coverage. Though a trial can stop due to inability to recruit to a subset resulting from restricted enrollment, the proposed method does not include an early stopping rule for futility or efficacy. In order to reduce sample size while assessing the treatment effect in the full population, Matsui and Crowley proposed a two-stage subgroup-focused sequential design for time-to-event outcomes, which could extend to multiple stages . In this design, patients are classified into two subgroups by a dichotomized predictive marker, with the assumption that the experimental treatment is more efficacious in the marker-positive subgroup. The trial can proceed to the second stage with one of the subgroups, or the full population, but treatment efficacy is only tested in the marker-positive group or the full population at the final analysis. Choices of testing procedures are fixed-sequence and split-alpha. At the interim analysis, a superiority boundary for the marker-positive subgroup and a futility boundary for the marker-negative subgroup are constructed. The superiority boundary is calculated to control the study-wide alpha level, while the futility boundary is based on a Bayesian posterior probability of efficacy with a non-informative prior. The required sample sizes for each subgroup are calculated separately, and the hazard ratio for the marker-positive subgroup is recommended to be 0.05–0.70 under this application. The proposed design is compared with a traditional all-comers design, an enriched design with only marker-positive subjects, a two-stage enriched design, and a traditional marker-stratified design. Different scenarios are considered including those with no treatment effect, constant treatment effect in both groups with hazard ratio (HR) = 0.75, a nearly qualitative interaction with HRs = 0.65 and 1, and a quantitative interaction with HRs = 0.7 and 0.8. The marker prevalence is set to 0.4, and the accrual rate is 200 patients per year. When using the split-alpha test, the proposed design has greater than 80% power to reject any null hypothesis in the alternative cases, but the traditional marker-stratified design also provides enough power under all cases. The number screened and the number randomized are reduced for the proposed design compared to the traditional marker stratified design, but the reduction is only moderate. To determine whether the full population or only the biomarker-positive subgroup benefit more from the experimental treatment, Uozumi and Hamada proposed a two-stage adaptive population selection design for a time-to-event outcome, an extension of methods from Brannath et al. and Jenkins et al. . The main extension is that the decision-making strategy at the interim analysis incorporates both progression-free survival (PSF) and overall survival (OS) information. Also, OS is decomposed into time-to-progression (TTP) and post-progression survival (PPS) when tumor progression has occurred, to account for the correlation between OS and PFS. The combination test approach is used for the final analysis based on Simes’ procedure . The hypothesis rejection rule for each population is a weighted inverse normal combination function with prespecified weights based on the expected number of OS events in each stage. At the interim analysis, a statistical model from Fleischer et al. under the semi-competing risks framework is applied to account for the correlation between OS and PFS . The interim decision rule uses the predictive power approach in each population, extending Brannath et al.’s method from single endpoint to multiple endpoints with a higher weight on PFS data due to its rapid observation. In the simulation, a dichotomized biomarker is used with a 50% prevalence. Four scenarios are considered, where hazard ratios in the marker-positive subgroup are always 0.5 and are higher in the marker-negative subgroup. For simplicity, the HR is the same for TTP, PPS, and death. FWER is controlled for all cases, but it is a little too conservative when the treatment is effective. The proposed design has a higher probability of identifying the treatment-sensitive population at the interim analysis, particularly when the PPS effect is large, those probabilities are similar between using OS or PFS alone or the combined endpoints when the PFS effect is small. One limitation of this design is that sample size calculations are not considered. Instead of a single primary endpoint, Sinha et al. suggested a two-stage Phase III design with population enrichment for two binary co-primary endpoints, which is an extension of Magnusson and Turnbull’s work with co-primary endpoints . The two binary endpoints are assumed to be independent, and the efficacy goal should be reached in both endpoints. With two distinct predefined subgroups, a set of decision rules stops the non-responsive subgroups using efficient score statistics. The futility and efficacy boundary values, which do not depend on the marker prevalence, are the same for both endpoints due to independence. The lower and upper stopping boundaries are calculated by alpha spending functions, and FWER is strongly controlled. Simulations were conducted assuming biomarker prevalences of 0.25 or 0.75 and weighted subgroup effect sizes of 0, 1, and 2 as the means of efficient score statistics under normal distribution. The results show that the proposed design can reduce false-negative results for heterogeneous treatment effects between subgroups. The authors state the possibility of extending the design to a bivariate continuous outcome, while an extension to bivariate survival would be more challenging. Kimani, Todd, and Stallard derived a uniformly minimum variance unbiased point estimator (UMVUE) of treatment effect in adaptive two-arm, two-stage enrichment design with a binary biomarker . Based on the Rao-Blackwell theorem, UMVUE for the treatment effect conditional on the selected subgroup is derived with and without prior information on maker prevalence. The proposed estimator is compared with the naive estimator, which is biased but with a lower mean squared error (MSE) when prevalence is known. The estimator is robust, with and without prior information on marker prevalence. Kimani et al. developed estimators for a two-stage adaptive enrichment design with a normally distributed outcome . A predictive continuous biomarker is used to partition the full population into a prespecified number of subgroups, and the cutoff values are determined at the interim analyses based on stage I observations. To estimate the treatment effect after enrichment for the selected subgroup, a naive estimator, uniformly minimum variance conditional unbiased estimator (UMVCUE), unbiased estimator, single- iteration and multiple-iteration biased-adjusted estimators, and two shrinkage estimators are derived and compared. Though no estimator is superior in terms of bias and MSE in all scenarios, UMVUE is recommended by the authors due to its mean unbiasedness. Tang et al. evaluated several proposed adaptive enrichment designs with a binary biomarker against the traditional group sequential design (GSD) for a time-to-event outcome . Type I error is controlled, and the subpopulation is selected by Bayesian predictive power. Adaptive design A selects the subgroup after considering futility and efficacy stopping decision. Design B selects the subgroup when the targeted number of events are observed in full population, which can be earlier than the interim analysis. Design C selects the subgroup only after the full population has reached a futility rule. Design D proceeds with the subgroup or full population by checking the treatment effect in the complementary subgroup, proposed by Wang et al. . When an enhanced treatment effect exists in the subpopulation, all of these adaptive designs could improve study power compared to GSD. Furthermore, Design C generally provides higher power across all scenarios among all the adaptive designs. Benner and Kieser explored how the timing of interim analyses would affect power in adaptive enrichment designs with a fixed total sample size for a continuous outcome and binary marker . Two subgroup selection rules are considered: the estimated treatment effect, or the estimated treatment effect difference between the subgroup and the full population (as opposed to the complement of the subgroup). When using the first selection rule, early timing increases power when the marker prevalence and marker cutoff values are low. However, the interim analysis timing’s impact on power is small when marker prevalence is high. If absolute treatment effect is used instead, earlier timing leads to power loss in general. Power depends more on the marker threshold, prevalence, and treatment effect size when interim timing is later than when half of the total sample size have observed outcomes. Kunzmann et al. investigated the performance of six different estimators besides maximum likelihood estimator (MLE) for a two-stage adaptive enrichment design for a continuous outcome . Those estimators are empirical Bayes estimator (EBE) , parametric bootstrap estimator , conditional moment estimator (CME) , and UMVCUE with MLE and CME as two hybrid estimators . The hybrid UMVCUE and CME estimator could reduce the bias across all considered scenarios, which the authors recommend, though with the cost of larger RMSE. In this review article, we have given an overview of traditional enrichment and adaptive enrichment designs, outlined their limitations, and described recent extensions and modifications to adaptive enrichment design strategies. Both Bayesian and frequentist perspectives in handling statistical issues of these designs were discussed in detail, along with important considerations for design parameters. Although the adaptive enrichment designs we have reviewed contain theoretical benefits such as early subgroup identification and early decision-making resulting in sample size reduction, we caution that selection and implementation of any of these designs requires acceptance of substantial additional trial complexity, and special consideration of the disease setting, endpoints, and markers at hand. For any of these trial designs to possibly have advantages over a simple randomized design followed by retrospective biomarker-focused analyses, the following should be true: the primary endpoint should be quickly observable relative to the pace of accrual; a sufficiently large sample size to detect moderately-sized subgroup effects of clinical interest must be achievable in a reasonable time frame, and the experimental treatment under study must have sufficiently strong preliminary evidence (e.g., from earlier phase studies) of a mechanism of action related to the candidate biomarker(s). If any of these criteria are not met, one runs the serious risk of conducting a study that is far less efficient than a standard design that is not biomarker-driven. In considering use of any design considered here, a trial biostatistician should meet with trial investigators and stakeholders to discuss the assumptions and requirements of different design options. The statistician should also prospectively understand and quantify the impact of any potential deviations from these assumptions while still in the trial planning stage (e.g., by using simulation studies). Each of the designs we discussed also have associated pros and cons, and are more suitable for application in different settings. To guide selection of a particular design for a particular context, we summarize design attributes (e.g., applicable primary endpoint types, number of biomarkers, decision rules, and other structural differences) as well as pros and cons in Table . For example, if there is no predefined biomarker subgroup and predictive biomarker discovery is required, Xu et al. and Zhang et al.’s proposed designs could be considered . Where Bayesian methods for estimation and interim decision-making using utility functions are desired but where final frequentist hypothesis testing is necessary, e.g., for regulatory purposes, the designs by Simon and Simon, Graf et al., or Ondra et al. may be appropriate . Where strong control of Type I error rate is required (e.g., in a later-phase application), designs by Matsui and Crowley, Fisher et al., and Uozumi and Hamada may be referenced . Overall, adaptive enrichment trial designs tend to increase study efficiency while minimizing subsequent study participation among patients showing a low likelihood of benefit based on early trial results . Biomarker-driven designs that reliably identify or validate predictive biomarker relationships and their thresholds with sufficient power to achieve phase II or III objectives continue to be of interest and warrant further development. Designs that make better use of truly continuous (versus dichotomous) marker-efficacy relationships are essential for future research. |
Medicinal Chemistry of Anti-HIV-1 Latency Chemotherapeutics: Biotargets, Binding Modes and Structure-Activity Relationship Investigation | 8e10ee00-5c37-4e92-8d41-39013eb21f91 | 9822059 | Pharmacology[mh] | Acquired immune deficiency syndrome (AIDS) caused by human immunodeficiency virus-1 (HIV-1) infection remains an incurable disease largely due to the existence of a persistent latent reservoir, which has been the last bastion for effective treatment of HIV-1/AIDS. In 1995, Siliciano and co-workers identified for the first time the latent cells in memory resting CD4 + T cells (rCD4s) from HIV-1-infected patients and proposed a term “latent reservoir” to graphically depict an ingenious survival method of the HIV-1 provirus . In addition to rCD4s, the latent cells also exist in monocytes, macrophages, lymphocytes, glial cells, astrocytes, natural killer cells, and multi-functional hematopoietic stem cells. Actually, humans have some immune exemption sites that offer a physiological tissue condition for latency of HIV virus, covering various lymphoid organs and tissues, including but not limited to the spleen, lymph nodes, abdominal and intestinal related lymphoid tissues, central nervous system, testis and other reproductive systems. In this regard, LVR represents a common term for all cells, tissues or any anatomical sites where a virus hides . LVR is a complicated and heterogeneous phenomenon, involving multiple and interlinked factors both at transcriptional and post-transcriptional levels. Accordingly, HIV-1 latency can be divided into two categories: pre-integration latency and post-integration latency. When transcriptional activators are lacking or chromatin structure is concentrated, latency can be maintained at a transcriptional level, while when nuclear RNA transportation or microRNA translation is inhibited, latency is usually maintained at the post-transcriptional level . Currently, the underlying mechanisms of HIV-1 latency are incompletely understood. At least six molecular mechanisms for illustrating the establishment and maintenance of LVRs have been proposed, as displayed in : (i) epigenetic regulation of viral gene expression, e.g., methylation, acetylation, deacetylation, phosphorylation and ubiquitination at histone tails; (ii) accessibility of activation-dependent cellular transcription factors, such as host transcription factors nuclear factor κB (NF-κB), nuclear factor of activated T cells (NFAT), activator protein 1 (AP-1) and positive transcriptional elongation factor b (P-TEFb), etc., which are crucial factors for active HIV transcription; (iii) influence of proviral integration site; (iv) influence of microRNAs (miRNAs) on viral gene transcription; (v) RNA elongation, splicing and transport; and (vi) formation of antisense HIV genomic transcript, which will down-regulate gene expression. In some cases, the aforementioned factors play synergistic and/or antagonistic effects on HIV latency. However, in view of the invisibility and complexity of LVRs, the exact mechanism about why and how HIV virus preferentially establishes latent infections in rCD4s is still not understood, which leaves the efforts to eliminate the latently infected cells unsuccessful to date . Much evidence from both animal and human models have shown that LVRs have been formed within days of HIV-1 infection by directly infecting rCD4s, or initially infecting activated CD4 + T cells, which then convert into a quiescent state. These stable LVRs that harbor an integrated but replication-competent proviruses can remain for a long time in the body, with an estimated half-life as long as 40~44 months. In other words, HIV virus might hole up in host cells for up to 73 years without triggering obvious symptoms, which renders HIV/AIDS an almost incurable disease . Besides, LVRs are extremely difficult to eradicate or reduce, because they are transcriptionally silenced by expressing little or no viral proteins, making them not only readily resistant to combination antiretroviral therapies (cARTs) but difficult to detect and purge for the host immune system. However, cells in LVRs can reactivate at any time and produce more viruses, leading to rapid viral rebound once antiviral treatment is interrupted . Therefore, the exploitation of an effective and safe anti-HIV-1-latency remedy remains a highly prioritized goal. To date, multiple eradication interventions against HIV-1 reservoirs have been put forward, with the expectation of achieving a functional cure for HIV-1/AIDS. Currently, the broadly accepted regimen to combat the LVRs chiefly relies on a ‘shock and kill’ approach, which involves a two-step protocol. First, drugs called latency-reversing agents (LRAs) are utilized to reactivate hiding viruses by stimulating viral protein expression in rCD4s (‘shock’). Second, the reactivated cells, which are now susceptible to cytolytic T lymphocytes or virus-induced cytopathogenicity, can then be readily eliminated by cARTs together with host immune-mediated interventions, or other therapeutic regimens (‘kill’). To ensure the successful implementation of this strategy, the identification of effective and safe LRAs is a prerequisite . Global T cell activators were initially developed to function as LRAs to reactivate proviruses in rCD4s, but severe toxicities (e.g., heart attack and temporary blindness) impelled researchers to seek safer LRAs that merely induce HIV-1 provirus expression without affecting normal immune functions in the body. Benefiting from multiple in vitro and in vivo HIV-1 latently infected models, different sorts of small molecular LRAs with distinct molecular mechanisms have been developed in succession for this purpose, as diagrammed in . Among these identified LRAs, three major types of LRAs with different mechanisms are expanding research hotspots. Molecules of the first type regulate epigenetics and include histone deacetylase inhibitors (HDACIs), DNA methyltransferase inhibitors (DMTIs) and histone methyltransferase inhibitors (HMTIs). Molecules of the second type activate transcriptional factors (e.g., NF-κB and AP1) and include protein kinase C (PKC) activators, CCR5 antagonists, and STAT5 agonists. Molecules of the third type mainly refer to bromodomain and extra-terminal domain inhibitors (BETIs), exerting functions by promoting transcription elongation . So far, albeit with positive progress in developing diverse chemotypes of LRAs with underlying biotargets and mechanisms of action, there has not yet been a significant breakthrough in successfully eliminating the latent proviruses. Most of these LRAs are ineffective in reducing the overall size of LPRs. Thus, novel LRAs with better therapeutic efficacy and lower toxicity are still urgently desirable. In the following subsections, we mainly focus on the “shock” aspect, with the emphasis on the description of potential drug targets, binding patterns as well as SAR perspectives of corresponding LRAs that are relevant to anti-HIV-latency chemotherapies. 2.1. Histone Deacetylase Inhibitors (HDACIs) HDACs are a group of epigenetic enzymes that can significantly affect chromatin topology and the histone deacetylation process by removing functional acetyl groups from the N-terminus of lysine residues and facilitating a high-affinity interaction between histones and DNA backbone, leading to a condensed inactive chromosomal DNA structure and consequent blockage of gene transcription. There are a total of 18 isoforms of mammalian HDACs, which are divided into four classes (class I, II, III and IV), largely based on the sequence homology, cellular location and folding mode of peptide chains. Class I contains three subtypes, IA (HDAC1, HDAC2), IB (HDAC3) and IC (HDAC8). Class II includes two subtypes, class IIa and IIb, in which class IIa consists of HDAC4, HDAC5, HDAC7 and HDAC9, while class IIb includes HDAC6 and HDAC10. Class III HDACs, also known as sirtuins, are nicotinamide adenine nucleotide (NAD) + -dependent enzymes, which contain seven members (sirtuins 1~7). Class IV contains only HDAC11. HDACs generally refer to zinc-dependent class I, class II and class IV isozymes, unless noted otherwise . HDACs contribute to proviral gene silencing of HIV latency by directly deacetylating histones at proviral integration sites (5′-long terminal repeat, 5′-LTR) or by indirectly inducing deacetylation of non-histone proteins (e.g., NF-κB). Thus, inhibiting HDACs can facilitate changes in chromatin architecture and recruitment of host transcription factors to LTR, leading to the acceleration of viral transcription. Using small molecular HDACIs as epigenetic modifiers thereby represents a viable and predominant strategy to eliminate latent reservoirs . Structurally, currently identified HDACIs typically follow a common pharmacophoric feature by mimicking the structure of the natural substrate lysine, which comprises a surface recognition cap moiety that can tolerate structural variability to accommodate the broad hydrophobic region of HDAC (Cap region); a functional zinc-binding group (ZBG) that can orient and coordinate with the catalytic zinc ion; and a linear or cyclic linker with 5~7 atoms (Linker) that traverses the long and narrow tunnel to connect the Cap portion and ZBG . In the past decades, a variety of HDACIs hits or candidates that vary in skeletal structures have been developed by modulating these pharmacophoric fragments and submitting to bioevaluation in various stages, aiming at achieving both elevated potency and isoform selectivity against HDACs-mediated pathological conditions, particularly hematological malignancies and solid malignant tumors. At present, vorinostat (SAHA), which has been approved for the treatment of cutaneous/peripheral T-cell lymphoma, is by far the clinically best-studied HDACI to be fully assessed for latency-related anti-HIV-1 therapies, either as a single regimen or in combination with other types of LRAs. SAHA showed promising in vitro HIV-1 latency-reversing effects in multiple HIV-1 latently infected cell lines (e.g., ACH2, U1 and J-Lat) and a latent provirus that was isolated from resting CD4 + T cells in HAART patients. However, the therapeutic outcome from pan-HDACI SAHA is widely limited by its insufficient selectivity towards specific isoforms, resulting in many unwanted side effects, including dehydration, anorexia, thrombocytopenia, arrhythmia, and also poor pharmacokinetic (PK) profiles. Hence, intensive structural modifications have been carried out to procure more potent HDACIs with improved selectivity and less toxicity . To achieve improved efficacy towards HIV-1 latent reservoirs, Okamoto et al. presented a spectrum of structural mimics of SAHA by employing a structure-based drug design protocol, based on the obtained crystal structure of HDAC in complex with SAHA. Specifically, the hydroxamic acid of SAHA that acts as a ZBG was initially replaced by an acylated thiol group, while the cap moiety was prolonged by inserting a thiazole or phenyl motif. It resulted in HDAC inhibitory potencies of two compounds, NCH-51 ( 1 ) and NCH-51 ( 2 ), increasing by about 2–3 fold over that of prototype SAHA, with selective preferences towards HDAC1/4/6 isoforms. Additionally, compared with congener 1, NCH-51 has more potent HIV latency-reversing effects in latently infected cell lines, OM10.1 (CC 50 = 2.2 µM) and ACH-2 (CC 50 = 2.4 µM), without causing obvious cell death, making it a promising lead. Encouragingly, owing to the introduction of an acylated thiol group instead of hydroxamic acid, which is supposed to have a potential mutagenicity side effect, these two compounds exhibited better PK profiles and lower cytotoxicity than the parent compound SAHA . Considering that the phosphate group can also function as a potential ZBG, Etzkorn et al. hypothesized that the substitution of phosphate analogues (phosphonamidate, phosphonate, phosphinate) for hydroxamic acid might afford equal potency while avoiding the limitations of hydroxamic acid. Keeping this in mind, they retained the cap part of SAHA unchanged and modified the linear linkage with a one-carbon degradation. Unfortunately, the resulting SAHA analogues 3 ~ 5 lost HDAC-binding affinities. One possible reason might be that the negatively charged phosphates may sterically hinder the coordinative contact with the catalytic zinc ion of HDACs, resulting in dramatically impaired efficacies . To discover more SAHA-derived HDACIs, Pflum et al. focused on the bridged linkers to explore chemical diversity and enrich SAR information. To this end, a series of SAHA analogues 6 ~ 9 were synthesized by appending branched substituents with different sizes and dispositions. Surprisingly, except for the benzyl-attached compound 8 that only gave a decent HDAC8 inhibitory affinity, nearly all these compounds furnished significant lowered potencies but exerted preferential selectivity towards HDAC6/8 versus HDAC1/2/3 compared with the archetype SAHA, indicating that structural modifications on bridge portion might be conducive to selective inhibition for HDAC6/8 . As the phosphorus-based ZBGs might largely account for the unfavourable HDAC inhibition, further structural optimizations were then focused on other parts of SAHA. Apicidin, a fungal metabolite bearing a cyclic tetrapeptide unit, is a naturally originated HDACI (IC 50 = 0.7 nM) . Inspired by this, Etzkorn and co-workers postulated that the macrocyclic peptide portion might act as a hydrophobic surface recognition group (Cap), and further modifications were conducted by simplifying the macrocyclic skeleton and constructing a linear chain of five carbon atoms (linker). Considering the strong zinc-binding affinity, the ethyl ketone part of Apicidin was replaced with hydroxamic acid (ZBG). As expected, the yielded hybrid 10 gave a 2-fold improvement in HDAC inhibition and nearly a 5-fold selectivity for HDAC1 over HDAC8. To pursue this lead further, a ring-opening operation was implemented from different locations to expose carboxylic acid and aryl substituent, respectively, aiming to examine the influences of rigid macrocycles and flexibility on activity and selectivity. The resulting compound 11 , however, decreased both activity and selectivity, while another ring-opening product, 12 , gave comparable efficacy only with its precursor, 10, implying that the hydrophobicity and conformational distribution of the cap region may have a significant impact on HDACs inhibition and selectivity, and the restricted macrocycllic ring is more suitable than the flexible side chains for HDAC1-selective inhibition. Likewise, the impaired potency of compound 11 in comparison with 12 might also be attributed to the charged groups . The aforementioned efforts on the structural variations derived from SAHA or Apicidin, as seen in , have provided useful SAR information for further identification of more potent HDACIs, although these newly identified compounds 3 ~ 12 were not submitted to the HIV-1 reactivation evaluation. Actually, similar to SAHA, in the past decades many synthetic and naturally available HDACIs as anticancer agents, including but not limited to the hydroxamic acid-based givinostat (ITF2357), panobinostat (LBH589), nanatinostat (CHR-3996), pracinostat (SB939) and belinostat (PXD101); benzamide-based mocetinostat, entinostat (MS-275), as evinced in , also displayed favorable HIV latency-reversing activities in various latently infected cell lines, and these agents are typically well-tolerated by participants. Among them, thiol-based romidepsin is the most effective LRA to date, which has supported the “proof-of-principle” that latent reservoirs can be safely activated, and perhaps, entirely eliminated . Rasmussen et al. compared the effects on HIV production in latently infected cells (U1 and ACH2) as well as T-cell activation of several hydroxamic acid-based HDACIs that were undergoing clinical development. The results indicated that these HDACIs gave different degrees of HIV-1 reactivation potencies at therapeutic concentrations, with activity order of panobinostat > givinostat ≈ belinostat > SAHA. However, all these HDACIs induced moderate T-cell activation, which hindered their further clinical application . As we know, HDACI treatment in HIV-1-infected individuals generally suffers from abnormal T-cell activation and nonspecific HDAC inhibition, leading to undesired HIV-1 persistence and other side effects by causing clonal expansion of latently infected rCD4s, which has been the major bottleneck for warranting further clinical investigation . Thus, an ideal LRA candidate for achieving the desired ‘shock and kill’ tactics should have the capability of stimulating latent HIV-1 transcription without provoking homeostatic proliferation and/or extensive T-cell activation that is highly correlated with the cytokine release, so as to avoid possible immune hyperactivation and the consequent concomitant cytokine storm, also known as cytokine release syndrome (CRS), as well as acceptable PK properties . Fimepinostat (CUDC-907), a dual inhibitor of class I-selective HDACI and PI3Kα, might be a suitable LRA candidate. It not only displayed comparable latency-reversal activity with romidepsin, the current most effective HDACI tested in anti-HIV-1 trails, at the cellular level, but caused reduced T-cell activation without any negative influence on T cell proliferation . Accumulating evidence indicates that class I-selective HDACIs—especially HDAC-1, -2 and -3, with HDAC3 isoform being the most important, functioning as transcriptional “on switches” of latent viruses and maintaining the deacetylated state of reactivation-related transcription factor nuclear factor κB (NF-κB)—might be more effective than pan-HDACIs in eradicating HIV latency by inducing more latent proviruses . Taking thiol-based orally active pan-HDACI ST7612AA1 ( 13 ), as another example, acting as a prodrug and potent HIV latency activator, ST7612AA1 actually exerts an HIV reactivation effect by transforming into its active form, ST7464AA1 ( 14 ), a class I-selective HDACI, via in vivo hydrolysis . For instance, Lewin SR and co-workers compared the HIV-1 reactivation efficacies of class I-targeted HDACI (entinostat) and three pan-HDACIs, SAHA, panobinostat and oxamflatin (metacept-3, MCT-3); their results also proved that entinostat gave the most potent HIV-1 latency-reversal activity by inducing more viral expression . The fact that HDAC3 highly selective inhibitor BRD3308 not only was active for latent HIV-1 reactivation in 2D10 cell model but could induce viral outgrowth from rCD4s of antiretroviral-treated patients further proved that class I-targeted HDACIs, especially HDAC3 inhibitors, are particularly effective anti-latencyt agents with improved HIV-1 reactivation potencies and fewer effects on other unrelated cellular genes . Similarly, to discover more potent class I (HDAC1/2/3)-selective HDACIs for eliminating a latent reservoir, Yu and co-workers from Merck have develop an array of ethyl ketone-based macrocyclic HDACIs, referring to the skeleton of class I-selective HDACI apicidin via ring expansion while retaining the linker and ZBG (ethyl ketone) unchanged, since the macrocycle structure is believed to have greater binding affinity with HDAC2 subtype, as evinced in . The facts that the two most potent class I-selective macrocyclic HDACIs, 15a and 15b , gave both enhanced class I HDAC inhibition and HIV latency-reversal potencies, supported the strategy for designing macrocyclic class I-selective HDACIs as promising LRAs, to a great extent. However, due to relatively low bioavailability of macrocycles, the PK profiles of these macrocyclic HDACIs still need further improvement . Granted, there are some cases that have proven otherwise. For example, another two effective class I-selective HDACIs, nanatinostat and romidepsin, as shown in , could not induce the generation of viral antigens or particles from rCD4s, partially owing to the lack of effective accumulation of spliced viral transcripts, despite in vitro effectiveness in latency-infected cell lines. Along with this, both of them impaired the function of CD8 + T cells, with romidepsin causing more impairment, which might explain, to a certain degree, the unsatisfactory clinical evaluation of various HDACIs in ARV-suppressed individuals so far . Colletively, as we can see from the above description, although HDACIs, especially class I-HDAC selective inhibitors, are a class of well-acknowledged LRAs with promising in vitro and ex vivo anti-latency effectiveness, which HDAC subtypes actually have a direct impact on HIV-1 latency in vivo is still not yet entirely elucidated. Hence, better investigative tools, whether more potent HDACI-based LRAs or latency screening models, particularly appropriate in vivo evaluation models, that can translate this knowledge into clinical chemotherapies are in urgent demand. 2.2. DNA Methyltransferase Inhibitors (DMTIs) As one of the chief epigenetic modifications, DNA methylation is supposed to play a role in governing HIV latency by epigenetic regulating of 5′-LTR cytosine-phosphate-guanine (CpG) methylation and inhibiting HIV viral transcription initiation. LTR contains two CpG islands and a HIV-1 promoter can be hypermethylated at these two CpG islands, particularly island 2, surrounding HIV transcription initiation sites. Besides, methyl-CpG binding domain protein 2 (MBD2) can specifically bind to methylated DNA and then recruit HDAC1/2, leading to chromatin unwinding and gene silencing and consequently histone deacetylation . The DMTI 5-aza-2′-deoxycytidine (decitabine, 5-aza-CdR, 16 ) which has been approved by FDA for the therapy of myelodysplastic syndrome (MDS), as shown in , has proved to demonstrate weak HIV-1 reactivation activity but displays an intensified promoting effect when utilized in combination with other HIV-activating agents, such as tumor necrosis factor α (TNFα), PKC activator prostratin or HDACIs in most J-Lat cell lines . However, the HIV-1 reactivation ability of decitabine, either used alone or in combination with other types of LRAs, displayed strong cell type dependence, implying a more comprehensive assessment should be carried out when using decitabine as a LRA in HIV reactivation trials. Another DMTI azacitidine (5-azacytidine, 17 ) also could induce latent HIV-1 proviruses . Lint and co-workers have observed that decitabine-induced HIV-1 reactivation was accompanied by a decreased recruitment of ubiquitin-like with an epigenetic integrator, PHD and RING finger domain 1 (UHRF1), to the viral promoter, implying UHRF1 might be a promising pharmacological target for discovering potent LRAs. To provide a demonstration of this proof-of-concept finding, this lab further utilized epigallocatechin-3-gallate (EGCG, 18 ), a polyphenol from green tea, which has been reported to possess certain UHRF1 inhibitory activity , and NSC232003 ( 19 ), a specific UHRF1 inhibitor, as two chemical probes, to ascertain the LRA potential of UHRF1 inhibitors. The results exhibited that EGCG partially reactivates latent proviruses through the inhibition of UHRF1, whereas NSC232003 provoked significant HIV-1 reactivation, highlighting the tight correlation of anti-UHRF1 with latency-reversal potency . Despite some positive latency-reversal outcomes from decitabine and azacitidine, the role of DNA methylation in the formation and maintenance of HIV-1 latency, however, is still in dispute. Blazkova et al. detected very low levels of methylated CpG in HIV-1 infected individuals under antiretroviral treatment, probably because activation is limited due to proviral DNA hypermethylation, highlighting the necessity for a deep understanding of the underlying heterogeneity of DNA methylation on HIV-1 latency as well as a more reasonable assessment of DMTIs as latency activators . All in all, since DNA methylation affects both the viral and host genome, silencing of HIV transcription via inhibition of DNA methylation also causes aberrant methylation of the host genome. This process is irreversible, and vice versa, implying that DNA methyltransferase is likely not the ideal strategy for developing effective HIV-1 reactivating agents. However, the DNA methyltransferase-associated UHRF1 is expected to be an attractive pharmacological target to explore more effective LRAs. 2.3. Histone Methyltransferase Inhibitors (HMTIs) In addition to DNA/CpG methylation, histone methylation—which primarily occurs at the N -terminal arginine or lysine of H3 and H4 histones and is covalently modified by an epigenetic enzyme, histone methyltransferase (HMT)—is also essential for the establishment of the silencing of HIV-1 transcription and maintenance of genome stability. Four important lysine methyl transferases (KMTs), EZH2, SUV39H1, G9a and G9a-like protein (GLP), are the most extensively studied HMTs. EZH2 is located at the promoter of latent HIV-1 provirus in T cells, participates in histone H3 lysine 27 trimethylation (H3K27me3), and plays a major role in chromatin-mediated HIV-1 transcriptional regulation and viral suppression. SUV39H1 primarily participates in H3K9 trimethylation (H3K9me3). G9a, also known as euchromatic histone-lysine N-methyltransferase 2 (EHMT2), is responsible for H3K9 dimethylation (H3K9me2) by catalyzing the addition of a methyl group from S-adenosyl-L-methionine (SAM) to a histone lysine residue. G9a-like protein (GLP), also called EHMT1, is another H3K9 methyltransferase with 80% sequence homology to G9a in the suppressor of variegation 3–9, enhancer of zeste and trithorax (SET) domains . Both G9a and GLP, together with SUV39H1, play pivotal roles in transcriptional silencing in HIV-1 latency. Histone methylation marks H3K9me2, H3K9me3 and H3K27me3 in particular have been validated to prevent lysine from being acetylated and meanwhile keep the chromatin in a dense state in both latency cell lines and primary CD4 + T cell models, and therefore are considered as important histone methylation marks . Quinazoline BIX01294 ( 20 ), a specific G9a inhibitor, was the first identified HMTI with clear in vitro HIV latency-reversing activity from a high-throughput screening protocol . The subsequent resolved X-ray co-crystallographic structure of BIX01294 bound to HMT (PDB code: 3K5K) as well as a SAR investigational outcome of BIX01294 have provided useful clues for further structure-based drug design and chemical optimizations . As indicated in A, HMT contains three active binding grooves: pockets I and II are two functional solvent regions; pocket III, also referred to as a lysine-binding channel, functions as a methyl transfer and transportation channel ( A). The quinazoline core of BIX01294 occupies the central histone peptide binding cavity, while a benzyl-substituted piperidine moiety and a 1,4-diazepane fragment extend into pocket I and II, respectively, and a 7-methoxy group orients towards pocket III, a long and narrow region ( B). Besides, 4-NH, two N atoms of piperidine and a diazepane ring form the key H-bonding force network with three acidic amino acids, Asp1078, Asp1083 and Asp1074, which have significantly contributed to its beneficial HMT inhibitory potency. To further explore the SARs and discover more potent HMTIs, Jin et al. employed several rounds of chemical optimizations on BIX01294. Initially, a 4-amino substituted group (R 1 ) of quinazoline skeleton was investigated. Since the 4-NH forms an important H-bonding force with Asp1083 while the benzyl group has extended beyond the active interfaces without contributing any protein-ligand interactions, the influences of piperidine N and the sizes of substituents on the bioactivity were studied by removing the redundant beznyl portion while retaining the 4-NH unchanged. As a result, compound 21 with 1-methylpiperidine moiety gave an 8-fold improvement in G9a binding affinity (IC 50 = 0.23 μM) compared with the prototype BIX01294 (IC 50 = 1.9 μM), indicating that piperidine N shows great promise to increase the drug-like profiles of quinazoline derivatives by not only providing multiple molecule-protein interactions but also producing ameliorated water solubility and bioavailability. By contrast, introduction of small-sized groups (e.g., cyclopropyl or isopropyl) results in obviously impaired potency. As far as 2-substituent (R 2 ) was concerned, both N-containing 6-membered (e.g., compounds 22 and 23 ) and 7 -membered cycloalkanes were tolerated. When it comes to R 3 group, although it oriented towards the water-exposed tolerant lysine-binding region (pocket III), it did not fit this narrow space properly, which has offered an underexploited but valuable site for further exploitation. Among the investigated hydrophilic side chains, introduction of N , N -dimethylbutane (compound 24 ) gave a marked improvement on G9a affinity by providing an additional water-mediated contact and thus yielding more favorable adaptability to the lysine-binding channel, elevating affinity by about 127 times, in contrast to the prototype BIX01294. The molecular modeling docking result of compound 24 in complex with G9a (PDB: 3K5K), as evinced in , has explained, in a large part, why both the multi-site contacts and N -bearing side chain that inserts into lysine-binding channel greatly account for its high potency. Specifically, in addition to the hydrogen bond formed between 4-NH and ASP1083, the piperazine N atom forms a salt bridge with ASP1074 and an electrostatic interaction with ASP1078. Besides, 1-N of quinazoline ring is supposed to be protonated under physiological pH conditions, forming a salt bridge with ASP1088. More importantly, the protonated N atom of the N , N -dimethylbutane part forms not only a hydrogen bond with Leu1086, but also a cation-π interaction with Tyr1154 (displayed as red circle), which contributed much to the binding affinity . Fuchter et al. examined the influences of a central quinazoline ring and 6,7-dimethoxyl groups of BIX01294 on HMT (G9a and GLP) inhibition. To this end, a scaffold-hopping strategy was conducted by replacing the quinazoline core with various bioisosteric moieties. The result is that two methoxyl groups, especially the 7-methoxyl group, are essential, since the replacement of two methoxyl groups with a dioxalone ring caused markedly impaired potency. Among the varied heterocycles, quinoline-derived derivative 25 gave the most potent affinity, which is largely owing to the basicity of protonated 1-N. Meanwhile, the 3-N atom of the quinazoline skeleton is not a necessity for G9a inhibition . Noticing that there was still room for activity improvement for compound 24 , since N , N -dimethylbutane moiety did not entirely occupy the lysine-binding region, Jin et al. systematically examined how the length and disposition of 7-substituents of quinazoline derivatives affected the potency. As a consequence, compound 26 afforded the most potent G9a inhibition to date, with a 250-fold improved potency over the prototype BIX01294, as evinced in . Albeit with excellent enzymatic activity, compound 26 presented lower cellular activity than BIX01294, which is largely attributable to its poor cell membrane permeability and low lipophicity (log P = 1.9). To overcome this bottleneck, this group further modified 26 by introducing various lipophilic groups, which has led to two promising hits. Compound 27 can effectively reduce the level of dimethylated H3K9 in many cell lines and shows high selectivity for G9a and GLP with low cytotoxicity . Compound 28 not only exhibited high potency and selectivity for G9a and GLP, but has measurable DNMT1 inhibitory activity against DNA methyltransferase, implying that this compound has the potential to become a HMT/DNMT1 dual inhibitor . However, 28 furnished a poor pharmacokinetic profile in an animal assay. The most probable reason might be due to the 2-cyclohexyl group, which is readily oxidized with the action of cytochrome p450 enzyme in vivo, leading to metabolic instability. Given that substituents at the 2-position of the quinazoline ring were well tolerated, this lab further optimized 2-substituents with the expectation of acquiring improved metabolic stability while maintaining high potency and high cellular activities. When 2-cyclohexyl was replaced by a 4,4-difluoropiperidinyl unit, the resulting compound 29 manifested the optimal performance and has been the first chemical probe against G9a and GLP used in animal studies . Sbardella et al. employed a ring expansion strategy by replacing the quinazoline skeleton of 24 with a benzodiazepine framework to afford 1,4-benzodiazepine derivative 30 , which gave about a 35-fold improved DNMT1 potency, comparable G9a inhibition, lowered cytotoxicity and better metabolic stability than the hit 24 , indicating a broad prospect to be developed as a HMT/ DNMT dual inhibitor . Similarly, Oyarzabal et al. utilized compound 28 as a hit to design more potent HMT/DNMT dual inhibitors. The quinazoline scaffold was initially refined, since the 3-N atom actually does not form any interactions with a protein backbone. As expected, both G9a and DNMT1 inhibitions of the obtained compound UNC-0638 ( 31 ) have been greatly improved. To guide further modifications, the binding mode of 31 was investigated by docking it into mouse DNMT1, as shown in . The result evinced that the quinoline core properly occupies the NNMT1 backbone and 1-N forms a H-bonding interaction with Glu1269 (Glu1266 in human DNMT1), 7-O atom 4-NH and contributes to H-bonding contacts with Arg1315 (Arg1312 in human DNMT1) and Ser1233 (Ser1230 in human DNMT1), respectively. As to the 2-cyclohexyl unit, it did not form any interactions with protein, while the 4-terminal isopropyl piperidine group forms a hydrophobic interaction with Met1235 (Met1232 in human DNMT1) and the 6-methoxyl group points towards the 3′-direction of the DNA strand, implying that subsequent structural variations might be conducted at 2- and 4-positions. Guided by this docking result together with previously obtained SAR results, several more potent inhibitors 32 ~ 36 have been identified. However, comparing with their excellent G9a inhibition, there is still room for improving the DNMT1 affinity. Therefore, more detailed and comprehensive SAR information still needs to be replenished. Chaetocin ( 37 ), a fungal mycotoxin from Chaetomium minutum and a specific Suv39H1 inhibitor that works as a competitive inhibitor for SAM, as exhibited in , was capable of inducing 86% latent proviruses from HIV-1 infected HAART-treated patients without affecting T-cell function . By contrast, the control HMTI BIX01294 gave slightly decreased (80%) potency under the same conditions . Furthermore, EZH2 inhibitors, such as GSK343 ( 38 ), 3-deazaneplanocin A (DZNep, 39 ) and tazemetostat (EPZ-6438, 40 ), and EHMT2 inhibitor UNC-0638 were also reported to display strong anti-latency-reversing activities. Intriguingly, two selective EZH2 inhibitors GSK343 and tazemetostat furnished more effective latency-reversing activities than the broad EZH2 inhibitor DZNep, further proving the direct association between specific EZH2 inhibition and HIV-1 reactivation. In addition, the combination of any of these HMTIs, as displayed in , with HDACI (e.g., SAHA, vorinostat) procured a strong synergistic HIV- 1 reactivation effect . Actually, with advancement in various molecular biology techniques, more and more previously unrecognized histone modification-associated factors underlying HIV-1 latency are being disclosed in addition to the widely investigated EZH2, G9a and GLP. For example, since the latency-related histone mark H3K27me3 is catalyzed by polycomb repressive complex 2 (PRC2), while embryonic ectoderm development (EED) is a principle component of PRC2, inhibition of EED is therefore inferred to be largely beneficial to latent provirus reactivation similarly to other HMTIs. James et al. proved that two EED inhibitors (EEDIs), A-395 ( 41 ) and EED226 ( 42 ), not only presented promising HIV-1 reactivation efficacies after treating alone but also gave an additive effect in latently infected 2D10 cells when in combination with EZH2 inhibitors, GSK343 or UNC1999 ( 43 ), further confirming that PRC2-mediated components, such as EZH2, EED and SUZ12, can serve as attractive targets to develop more types of LRAs . Besides, Ott and co-workers disclosed a previously undiscovered lysine methyltransferase, SET and MYND domain-containing protein 2 (SMYD2), which is also a potent target for identifying latency-reversal agents. SMYD2, which works as an epigenetic co-repressor, is found to be tightly linked with HIV-1 latency by inducing monomethylation on histone H4K20me1 at the HIV-1 5′-LTR region, leading to repression of proviral transcription. Accordingly, H4K20me1 can also act as an important histone mark in addition to H3K27me3 and H3K9me2/3. The fact that effective reactivation of latent proviruses with a SMYD2 inhibitor AZ391 ( 44 ) in CD4 + cells further supported the inhibition of the SMYD2’s catalytic activity being directly correlated with HIV-1 reversal activity . Meanwhile, given that H4K20me1 is recognized by a chromatin “reader” protein lethal 3 malignant brain tumor 1 (L3MBTL1), leading to chromatin compaction and consequent transcriptional silencing, L3MBTL1 thereby becomes a possible HMT pathway-related target for exploiting LRAs, although its exact role in modulating HIV-1 latency still needs further verification . Taken together, among the proposed HMT-associated latency factors, EZH2 has stronger associations with HIV-1 reactivating efficacy compared with other HMT-related factors, including but not limited to G9a, SUV39H1, EED, SMYD2 and L3MBTL1. Hence, specific and effective EZH2 inhibitors are strongly encouraged. 2.4. Protein Kinase C Activators The protein kinase C (PKC) family of serine/threonine kinases plays an essential role in reactivating latent HIV-1 through activation of the NF-κB signaling pathway . A host of small molecules derived from natural sources, including but not limited to phorbol esters, ingenol esters, ingols, jatrophanes, and various macrolides, have been proposed to reactivate HIV-1 in latently infected CD4 + T cells, as displayed in . Among them, ingenol derivatives phorbol 12-myristate 13-acetate (PMA, 45 ) and 12-deoxyphorbol-13-acetate (prostratin, 46 ) are two most well-known activators of PKC with nanomolar PKC activation affinities. Although PMA reactivates latent HIV-1 by activating T cells, its clinical utility was limited by tumor-promoting risk and other serious side effects, such as mitotic dysfunction and chromosomal aberrations . Unlike PMA, prostratin, which is isolated from the poisonous New Zealand plant Pimela prostrata, is not a tumor promoter and does not induce cell proliferation by itself. Besides, prostratin can inhibit PMA-induced tumor promotion in a mouse model, illustrating a broad scope in clinical application. Prostratin not only reactivates latent HIV-1 in vitro in a PKC-dependent NF-κB activation manner, but also down-regulates the expressions of HIV-1 receptor CD4 and co-receptor CXCR4, thus avoiding the novo infection of CD4+ cells. However, since prostratin causes overall T cell activation, just as PMA does, further investigation is still needed in order to interpret the suitability of this compound for use in humans . Another two ingenol derivatives, phorbol 13-stearate (P-13S, 47 ) and 12-Deoxyphorbol 13phenylacetate (DPP, 48 ), also presented attractive HIV-1 reactivation potencies. P-13S effectively activates HIV-1 gene expression in the Jurkat-LAT-GFP latency model, with at least a 10-fold increase in potency over that of prostratin. Interestingly, P-13S activates PKC by inducing a translocation of PKC isotypes α and δ to cellular compartments, which is distinctly different from that of prostratin and PMA . As to DPP, a non-tumour-promoting phorbol ester isolated from the West African “candle plant” Euphorbia poissonii and the Moroccan succulent E. resinifera Berg, was reported to induce HIV-1 gene expression in latently infected ACH-2 cells at a 20–40-fold lower concentration than prostratin . The ingenol analogue ingenol-3-angelate (PEP005, 49 ) that was isolated from Euphorbia peplus was validated to reactivate latent HIV-1 through a PKC-dependent NF-κB pathway. Moreover, this compound was able to prevent new rounds of viral infection after HIV-1 reactivation through down-regulating the expression of the HIV-1 receptors CD4 and CXCR4 . The ingenol derivative EK-16A ( 50 ) that was isolated from Euphorbia kansui displayed 200-fold more potent efficacy than prostratin in reactivating latent proviruses in vitro and ex vivo with minimal cytotoxicity on cell viability. Besides, EK-16A could induce synergistic effects with multiple types of LRAs, such as DNMTI 5-Aza, BETIs JQ1 and I-Bet151, as well as HDACIs vorinostat and romidepsin in latently infected cell lines J-Lat 10.6 and 6.3, without any influence on T-cell activation. Mechanistically, EK-16A works as a PKCγ activator to promote HIV-1 transcription HIV-1 reactivation via the activation NF-κB pathway and also to facilitate HIV-1 elongation via the stimulation P-TEFb pathway . Later, three ingenol derivatives, viz. EK-1A ( 51 ), EK-5A ( 52 ) and EK-15A ( 53 ), were isolated from Euphorbia kansui, and not only demonstrated latent reactivation efficacies in vitro and ex vivo at nanomolar concentrations but also could inhibit acute HIV-1 infection via down-regulation of the expression of CCR5 and CXCR4, two cell surface HIV co-receptors. Additionally, these three ingenol derivatives have a synergy with 5-Aza, SAHA, JQ1 or prostatin with little cellular toxicity in T-cells . These data suggested ingenol derivatives derived from Euphorbia species might have great prospects for being developed into successful chemotherapeutic LRAs. Macrocyclic jatrophane diterpenoid compound SJ23B ( 54 ), which is isolated from a Mediterranean plant specimen, E. hyberna, is an activator of PKCα/δ. It could reactivate latent HIV-1 via activation of the NF-κB pathway at a nanomolar level (EC 50 = 50 nM), which is at least 10 times more potent than prostratin. Moreover, SJ23B is not a tumor promoter and displayed strong in vitro anti-HIV-1 activity . The epimeric N , N -dimethylvalinoyl-4α-4-deoxyphorbol derivatives 55 and 56 were isolated for the first time from a medicinal Mexican Croton, and were later found to be potent and isoform-specific activators of PKC with promising HIV-1 reactivating activity . 3,12-Di-O-acetyl-8-O-tigloyl-ingol ( 57 ) that was isolated from Euphorbia lactea, a plant that produces latex with anti-inflammatory activity, was reported to antagonize HIV-1 latency through a PKC-dependent pathway . Likewise, 8-methoxyingol 7,12-diacetate 3-phenylace ( 58 ) which was derived from the latex of Euphorbia officinarum, could also reactivate HIV-1 latency with an EC 50 < 25 µM . Macrolides bryostatins that are isolated from bryozoan are a family of effective PKC activators, among which bryostatin-1 ( 59 ) exhibited the most powerful PKC activating activity at a nanomolar concentration and has been the only PKC agonist that entered clinical evaluation as an effective LRA candidate. Bryostatin-1 not only exhibited significant HIV-1 reactivation efficacy in human astrocytes via a NF-κB and PKC-dependent mechanism, but induced a remarkable decrease in viral production and amyloid beta (Aβ) deposition in myeloid cells. Bryostatin-1 can also activate the mitogen-activated protein kinase (MAPK) pathway and down-regulate the expressions of HIV-1 co-receptors CD4 and CXCR4 without triggering global T cell proliferation, and it has synergy with HDACIs in reactivating latent HIV-1 . All these beneficial properties together with the fact that bryostatin-1 was capable of avoiding de novo infection in HIV-1 in susceptible cells makes it a promising adjunct for the treatment of HIV-1 brain infection . Enlightened by the promising therapeutic potential and SAR outcomes of bryostatin-1, Stone and co-workers prepared two bryostatin-1 analogs, compounds 60 and 61 , which offered comparable to higher PKC-binding affinities compared to the prototypes bryostatin-1 and PMA . In addition, the semi-synthetic ingenol ester 62 was reported to exert HIV-1 reactivation efficacy by activating PKCs and up-regulating the positive transcription elongation factor b (P-TEFb), thus promoting both transcription initiation and elongation of viral genes. Given that ingenol analogues have rich natural resources, compound 62 is therefore a promising LRA candidate to be utilized in clinical practice . Gnidimacrin ( 63 ), a diterpenes PKC activator, can potentially activate latent HIV-1 viruses, with about 10 times more potency than HDACI SAHA at an effective concentration as low as the picomolar level. It is especially noteworthy that 63 can significantly reduce the size of a latent reservoir by decreasing the amount of latently infected cells at a concentration that does not cause overall T cell activation or stimulate production of inflammatory cytokines, indicating its promising clinical application . BL-V8-310 ( 64 ), a benzolactam-related PKC activator, was proved to effectively reactivate latent HIV-1 in latently infected ACH-2 and J-Lat cell lines. Moreover, combining BL-V8-310 with BRD4 inhibitor JQ1 not only showed synergistic latency-reversing activity but also decreased the influence on cytokine secretion from CD4 + T cells induced by BL-V8-310 alone . In brief, PKC activators are still attractive LRAs which are particularly effective in combination with other LRAs with different mechanisms. 2.5. BET Inhibitors (BETIs) Among the four members (BRD2, BRD3, BRD4, and BRDT) of the bromodomain and extra-terminal (BET) protein family, BRD4 is the most concerned and widely studied subtype. The primary structure of BRD4 comprises two highly conserved N-terminal tandem bromodomains (BD1 and BD2) that are responsible for recognizing the acetylation status of lysine residues on histone tails and other proteins, an ET domain, and a C-terminal domain (CTD). Bromodomain protein contains four antiparallel α-helices (αZ, αA, αB, αC) and two hydrophobic loops (ZA loop and BC loop). ZA, BC, and αZ form an essential N -acetylated lysine-binding hydrophobic pocket (Kac site). Besides, a hydrophobic region ‘WPF shelf’ bounded by Trp/Pro/Phe residues (W81, P82, F83) is also important for enhancing BRD4 binding affinity . Acting as an interaction partner with P-TEFb, BRD4 can competitively bind P-TEFb with Tat protein, an exclusive transcriptional transactivator of HIV-1 viruses, thus leading to the silencing of HIV-1 gene transcription. Hence, inhibition over-expression of BET/BRD4 will promote the recruitment of P-TEFb by Tat and subsequent viral transcription elongation, resulting in dissociation of the BET and P-TEFb complex and consequent provirus reactivation . Chemical structures of several representative BET inhibitors (BETIs), as displayed in , are being currently tested in various latently HIV-1-infected evaluation models. Triazolothienodiazepine derivative -JQ1 ( 65 ) is the first BETI discovered by a high-throughput screening protocol, which possesses many good drug-like properties, such as excellent cellular potency, high selectivity for the BRD4 isoform of the BET family, synthetic accessibility, low off-target possibility and a good pharmacokinetic profile. However, JQ1 was less efficient in reactivating latent HIV-1 viruses when used alone and exerted severe cytotoxicity during prolonged treatment . Since the benzodiazepine fragment of JQ1 is validated to be a key pharmacophore that contributes to a good BRD4 affinity, a spectrum of benzodiazepine-derived BETIs was successively identified, including but not limited to OTX015 ( 66 ), CPI-203 ( 67 ), MS417 (GTPL7512, 68 ) and I-BET762 (GSK525762, 69 ). OTX015 ( 66 ) could effectively reactivate latent HIV-1 in various latency models with 1.95~4.34 times improved efficacy over that of JQ1. More interestingly, OTX015 can also increase CDK9 occupancy at HIV-1 promoter, which in turn phosphorylates RNA polymerase II (Pol II) to enable productive elongation and generate full-length HIV transcripts . CPI-203 ( 67 ) is a novel cell-permeable BRD4 inhibitor with good bioavailability when administered orally. Notably, in contrast to JQ1, CPI-203 showed more potent HIV-1reactivation efficacies in various latently infected cell lines and significantly decreased cytotoxicity. CPI-203 also exerted a synergistic effect and alleviated prostratin-induced ‘cytokine storm’, a severe and systemic production of inflammatory cytokines in response to endotoxemic shock, when combined with PKC activators (e.g., prostratin, ingenol-B and bryostatin-1) in the reactivation of latent HIV-1 . MS417 ( 68 ) and I-BET762 ( 69 ) could reactivate HIV-1 from latency in vitro via a P-TEFb-dependent but Tat-independent mechanism, implying BRD2 together with BRD4 cooperatively participate in HIV latency . BETIs PFI-1 ( 70 ) and RVX-208 ( 71 ), whether used alone or in combination with other types of LRAs, exhibited strong HIV-1 reactivation activities through up-regulation of the phosphorylation of CDK9 Thr-186 to increase the expression of P-TEFb in latently infected Jurkat T cells, thus effective activation of HIV-1 transcription. Besides, these two BETIs could also activate HIV-1 transcription in resting CD4 + T cells in cART-treated patients without causing aberrant activation of global immune cells, strengthening the therapeutic values of BETIs in anti-HIV therapy by effective combating HIV-1 latency . Mivebresib ( 72 ) has already been advanced to phase I clinical trials for treating relapsed/refractory acute myeloid leukemia . I-BET151 ( 73 ) could preferentially reactivate HIV-1 gene expression and efficiently increase HIV-1 transcription in monocytic cells, but not in typically HIV-1-infected CD4 + T cells, in cART-treated humanized mice, via a CDK2-dependent mechanism . 8-Methoxy-6-methylquinolin-4-ol (MMQO, 74 ), was identified from a virtual screening that was able to reactivate latent proviruses in vitro and ex vivo by functioning as a BRD4 inhibitor. In addition, MMQO also could potentiate the effects of PKC activators or HDACIs on HIV-1 reactivation. Owing to its minimalistic structure, MMQO might act as an ideal hit for further chemical modification . UMB136 ( 75 ), an imidazo [1,2- a ]pyrazine-derived BETI, exhibited elevated HIV-1 reversal activity compared to JQ1 in multiple latently infected cell models. Besides, UMB136 was able to synergize with PKC activators (e.g., bryostatin-1 and prostratin), and increase viral production and HIV-1 transcription by releasing P-TEFb . The BRD4 selective inhibitor ZL0580 ( 76 ) was capable of reactivating latent HIV-1 and restraining viral replication in CNS reservoirs, indicating the therapeutic potential in treating neuroinflammation or related neurological disorders in HIV-infected individuals . CPI-637 ( 77 ), a BRD4 and TIP60 dual inhibitor, could efficiently reactivate latent proviruses in vitro. Meanwhile, CPI-637 was also able to alleviate overall T cell activation and prevent viral spread to uninfected CD4 + T cells without an obvious toxic effect. Interestingly, CPI-637-mediated TIP60 inhibition, in turn, stimulated BRD4 dissociation from the HIV-1 5′-LTR promoter, causing more effective binding of Tat protein with P-TEFb in comparison with the BRD4 inhibition alone. These data indicated that development of dual-target LRAs, including but not limited to BRD4/TIP60 inhibitors, might be a more promising remedy to effectively eliminate latent reservoirs . According to the X-ray co-crystallographic structures of BETIs-bound BRD4, the vast majority of the aforementioned BETIs share two key pharmacophoric features: a ‘head moiety’ that bears an H-bond donor and/or acceptor to form hydrogen bonds with two key amino residues (Asn140 and Tyr 97), by mimicking the function of the acetyl O atom of ABBV-075, and appropriate hydrophobic substituents to spatially accommodate the ‘WPF shelf’ and ‘ZA channel’, aiming to enhance potency and selectivity towards BRD4 by forming intensive hydrophobic contacts . To identify more potent BETIs, Burns and co-workers believed an appropriate Kac-mimicking pharmacophoric framework, which is capable of providing interactions with the pivotal Kac domain and meanwhile offers strong H-bonds with the key residues in this region, plays a crucial role in achieving desirable BET-binding affinities. To this end, they employed a scaffold-hopping strategy by replacing 6 H -thieno [3,2- f ][1,2,4]triazolo [4,3- a ][1,4]diazepine core of JQ1 with a 1,2,3-triazolobenzodiazepine framework, which was supposed to be capable of forming more favorable binding interactions with target protein. It indicated that, as anticipated, the resulting 1,2,3-triazolobenzodiazepine skeleton ( 78 ) gave a nearly 3-fold increase in BRD4-binding affinity. The introduction of a 2-aminopyridine fragment on the central benzene ring ( 79 ) resulted in further activity improvement. To figure out the possible reason for activity improvements of the 1,2,3-triazolobenzodiazepine nucleus, binding patterns of JQ1 and 79 complexed with BRD4 (PDB entry: 5 UVV) were compared by a molecular modeling method. As evinced in , although both 76 and reference JQ1 adopt similar spatial orientation by docking well into three key binding domains (Kac, ZA, WPF) of BRD4, 79 formed two H-bonds between the 2-aminopyridine portion and two residues (Met398 and Met425), with distances of 2.3 Å and 2.8 Å, respectively. By contrast, JQ1 formed a relatively weak H bond with Tyr390, with a distance of 3.1 Å. Collectively, compound 79 with the new skeleton forms stronger H-bond contacts, which is supposed to largely account for its elevated activity . I-BET151 ( 80 ) is another representative BETI, in which 1 H -imidazo [4,5- c ]21uinoline-2(3 H )-one core is a key factor in affecting HIV-1 reactivation potency by functioning as a mimic of acetylated lysine. The co-crystal structure of I-BET151 complexed with BRD4 (BD1), as evinced in , revealed that the 3,5-dimethylisoxazole part forms a direct H-bond with Asn140 (3.2 Å) and water-bridged hydrogen-bonding interactions with Tyr97 (2.0 Å) in the acetylated lysine binding pocket (Kac site). The 1H-imidazo [4,5- c ]21uinoline-2(3 H )-one nucleus stretches into the hydrophobic ZA channel of protein, and the N atom on the central pyridine ring forms a water-mediated hydrogen bond interaction with Phe83 (2.7 Å). The pyridine ring interacts with the surface of the hydrophobic WPF shelf, which is formed by W81, P82, M149 and I146 residues of BRD4. Based on this binding information, Wang’s lab employed a ‘scaffold-hopping’ strategy to substitute a [6,6,5]tricyclic 5 H -pyrido [4,3- b ]indole (γ-carboline) motif which has a balanced physiochemical profile for the 1 H -imidazo [4,5- c ]21uinoline-2(3 H )-one core of I-BET151, with the expectation of achieving similar BRD4-binding affinity. Among the newly obtained derivatives, compound 81 containing a 3-cyclopropyl-5-methyl-1 H -pyrazole fragment gave the best potency, particular an improved BRD4(2)-binding affinity compared with the reference JQ1, while the replacement of this moiety with a smaller Cl atom ( 82 ) impaired potency . To gain more insight into the activity differences, binding modes were conducted and compared between I-BET151 and compound 81 by using the molecular docking method. As evinced in , although both 81 and I-BET151 adopt very similar spatial orientation and both bind with two key residues (Gln85 and Asn140), 81 offered stronger H-bonding interactions with these two amino acids than I-BET151, which might explain their binding differences to a great extent. In summary, although the exact molecular mechanism of BETIs in reactivating latent proviruses is still not fully understood, the successful development of various BETIs highlights the importance of BET/acetyl-lysine inhibition in regulating HIV-1 transcription and latent reactivation. Nevertheless, so far, none of the BETIs have entered clinical phases as LRAs, primarily attributable to the in vivo inefficiency with no effect on eliminating both latent reservoirs and HIV-1 viremia, although the in vitro latency-reversing activities were satisfactory. This fact leaves the clinical implementations of BETIs as LRAs with a long way to go still. 2.6. P-TEFb Activators P-TEFb, which is composed of cyclin T1 (CycT1) and cyclin-dependent kinase 9 (CDK9), is a CDK9-cyclin T1 heterodimer and is an important part of the super elongation complex (SEC) used by the viral-encoded Tat protein to activate HIV transcription. In this sense, P-TEFb is an indispensable cellular cofactor for Tat protein and an essential elongation factor to realize efficient viral transcriptional elongation and expression. In this regard, P-TEFb activators were able to work as LRAs to reactivate latent viruses . Ye et al. isolated at least five kinds of short-chain fatty acids (SCFAs), namely, butyric acid, isobutyric acid, isovaleric acid, propionic acid and acetic acid, from the saliva and plasma of HIV-infected patients with severe periodontitis undergoing HAART treatment. The subsequent study revealed that, except for acetic acid, the other four SCFAs have the capability to potently reactivate latent HIV-1 in both latent Jurkat cells and resting CD4 + T-cells by activating P-TEFb as well as inducing histone modifications in a dose-dependent manner. Among these five SCFAs, butyric acid presented the strongest activity, while isobutyric acid gave the weakest effect, and the combination of any two SCFAs exhibited an additive effect . PR-957 ( 83 ), also named as ONX-0914, as displayed in , might be a promising LRA candidate with a host of favorable properties. It was found to strongly activate latent viruses in vitro and in vivo by activating P-TEFb without causing overall abnormal immune activation of T cells. It also exhibited relatively low cytotoxicity at a tolerable level and could reduce the expression of HIV receptor CD4 and co-receptors CXCR4 and CCR5. Moreover, it manifested a synergistic effect when combined with known LRA, PKC activator prostratin, and could reduce prostratin-induced T-cell activation. Most notably, it displayed no interference with other anti-viral drugs, thus guaranteeing the efficient implementation of the “shock and kill” strategy and effective elimination of newly produced viruses . Hexamethylene bisacetamide (HMBA, 84 ), which has been dismissed as anticancer candidate in clinical anti-leukemic evaluation, has recently revived interest in anti-HIV therapy due to its remarkable capability in reactivating latent HIV-1 proviruses, whether used alone or in combination with PKC activator prostratin in chronically infected CD4 + T cells . The reactivation occurs at a transcription level and is independent of NF-κB but transiently activates Akt via the inhibition of phosphatidylinositol-3-kinase (PI3K) and relies on Sp1-binding sites in the HIV promoter to recruit P-TEFb to promote the reactivation of viral production from latency . Chalcone analogue Amt-87 ( 85 ) has been proved to significantly reactivate latent HIV-1 provirses and synergistically work with known LRAs, such as PKC activator prostratin and BETI JQ1 via activation of P-TEFb, indicating its development perspective as a potent LRA . 2.7. Polo-like Kinase 1 (PLK1) Inhibitors Acting as a highly conserved Ser/Thr kinase in eukaryotes, PLK1 is proved to be a promising target in identifying potent LRAs. PLK1 inhibitors are expected to have doubled function by not only effectively reactivating latent proviruses but also promoting the death of reservoir cells, based on the fact that PLK1 plays an essential role in HIV-1 Nef-mediated survival of CD4 + T cells . Besides, PLK1 also has a tight correlation with BRD4. On one hand, both BRD4 and PLK1 are able to interact directly with P-TEFb to regulate HIV-1 LTR transcription; on the other hand, BRD4 can also be phosphorylated by PLK1 , which perhaps explains why many PLK1 inhibitors, such as compounds 5-methyl-7,8-dihydropteridin-6(5 H )-one derivatives BI-2536 ( 86 ) and BI-6727 (volasertib, 87 ), as diagrammed in , also function as BRD4 inhibitors. Acting as dual PLK1/BET inhibitors, pteridine-derived compounds 86 and 87 could significantly reactivate silenced HIV-1 proviruses in two latently infected cell lines, ACH2 and U1, at both the mRNA and protein level. Although these two compounds demonstrated more potent PLK inhibitory activities, the latent reactivating effects were directly associated with BET inhibition, further confirming the important role of BETIs in developing promising LRAs. Additionally, BI-2536 could synergistically reactivate HIV-1 proviruses when combined with HDACI SAHA or PKC activator prostratin, in PBMCs from HIV-1-infected patients, verifying the prospective therapeutic value in anti-viral treatment . 2.8. CCR5 Antagonist Maraviroc (MVC, 88 ), as shown in , a selective CCR5 (C-C Motif Chemokine Receptor 5) antagonist with broad-spectrum antiretroviral efficacy that has been used in treatment of HIV-1 infection, was recently validated to be an attractive LRA by inducing NF-κB activation as a result of specific binding of CCR5, whether used alone or in combination with PKC activator bryostatin-1 . However, despite the favorable safety profile in humans as well as positive latent reactivation outcomes in phase II trial, maraviroc could not efficiently reduce the size of latent reservoirs, given the fact that viruses rebounded quickly after antiretroviral therapy was discontinued . All these evidence strongly support that CCR5 antagonists might be successfully developed into effective anti-HIV drugs, which not only function as antiretroviral agents to suppress HIV-1 replication but also act as LRAs to reactivate latent proviruses . 2.9. Inhibitor of Apoptosis Protein (IAP) Antagonist Xevinapant (Debio 1143, AT-406, 89 ), as displayed in , an effective chemo-radio-sensitizer as a first-in-class oral IAP antagonist, currently being assessed in phase II clinical trial for the treatment of squamous cell carcinoma of the head and neck (SCCHN), was recently proved to reverse HIV-1 latency in various latently infected T cell lines via the induction of a non-canonical NF-κB pathway, resulting in the enhancement of HIV-1 transcription . Besides, an obvious synergistic latency-reversal activity was achieved when combined use of IAP antagonist AZD5582 ( 90 ) and BETI I-BET151, rather than largely due to these two LRAs both acting on host cell transcriptome . The favourable latency-reversal efficacies together with the manageable safety profiles may enable IAP antagonists to be developed as promising LRAs. 2.10. Phosphatidylethanolamine-Binding Protein 1 (PEBP1) Inhibitor Zhu et al. found that the PEBP1 gene, also named as RKIP (Raf kinase inhibitor protein) is tightly correlated with the establishment of a latent reservoir as well as the suppression of viral replication by employing a CRISPR-based genetic screen technique. The subsequent depletion of the PEBP1 gene significantly reactivates latent proviruses by regulating the NF-κB pathway, while the up-regulation of PEBP1 expression with epigallocatechin-3-gallate (EGCG, 18 ) inhibits latency reversal by preventing nuclear translocation of NF-κB. These results provided directly proofs that PEBP1 inhibitors might be a type of promising LRAs in effective controlling HIV-1 reservoirs . 2.11. Proteasome Inhibitor (PI) Proteasome inhibitor (PI) thiostrepton (TSR, 91), as demonstrated in , a naturally derived thiazole-containing oligopeptide antibiotic, was recently found to effectively reactivate latent HIV-1 proviruses in vitro and ex vivo with low cytotoxicity. TSR also displayed a significant synergistic effect with prostratin, bryostatin-1 or JQ1 without affecting overall T cell activation or inducing dysfunction of CD8 + T cell, indicating the prospective application as an effective LRA candidate. Mechanistic study revealed TSR activated P-TEFb and NF-κB pathways by up-regulating heat shock proteins (HSPs), resulting in viral reactivation . By using the CRISPR interference-based screening technique, Zhou et al. proved two PIs, bortezomib (PS-341, 92 ) and carfilzomib ( 93 ), could effectively reactivate latent HIV-1 by strongly synergizing with different types of LRAs without inducing CD4 + T cell activation or proliferation. Mechanistic investigation showed that a proteasome inhibitor could promote Tat-transactivation by up-regulating the expressions of ELL2 and ELL2-containing super elongation complexes (ELL2-SECs), leading to reactivation of latent proviruses . Mechanistic results from Liu and co-workers demonstrated that PIs not only could reactivate latent HIV-1 by HSF1-mediated recruitment of complex of HSP90 and P-TEFb, but also may facilitate the human body to the virus by enhancing host innate immune responses . Evidence from Dougherty et al. suggested that PIs, exemplified by bortezomib, clasto-lactacystin β-lactone (CLBL, 94 ) and MG-132 ( 95 ), work as bifunctional antagonists of HIV-1, exerting anti-latency by reactivating latent HIV-1 and anti-replication by inhibiting HIV-1 infectivity . All these data revealed that proteasome can function as a promising and targetable target in developing effective anti-HIV-1 agents . In addition to proteasome, HSF1 and HSP90, which participate in HIV transcription elongation, can also serve as attractive therapeutic targets in developing potent LRAs. Particularly, HSF1 that is tightly associated with HIV-1 5′-LTR might be a targetable target for identifying promising LRAs based on the fact that the majority of LRAs with various mechanism of actions might reactivate latent HIV-1 via an HSF1-involved pathway rather than the previously known pathways . 2.12. Toll-like Receptor (TLR) Agonist Toll-like receptors (TLRs) are transmembrane pathogen-recognition receptors that function as the sentinels of host defense by recognizing a spectrum of pathogen-related molecular components in bacteria, viruses, protozoa and fungi as well as molecules released during cell damage or death. The TLR family contains more than 13 members, of which only 10 (TLR1-TLR10) have been identified in humans . Since TLR members are mainly existing and functional in CD4 + T cells, where the majority of the latent reservoir is harboured, TLR agonists are thereby believed to hold great promise to successfully reactivate latent proviruses by stimulating downstream pathways , such as AP-1, NFAT, IRFs as well as NF-κB, a well-acknowledged regulator of HIV transcription that has a tight correlation with HIV-1 latency . Moreover, TLRs also have high hopes of eliminating activated latent cells by exerting immunologic cytotoxicity, promoting antiviral responses and non-specifically activating T lymphocytes. All these properties render immunostimulatory TLR agonists unique and promising LRAs among the identified LRAs to date, also known as next-generation LRAs . TLR signalling, in particular TL-7/8/9 and TL-1/2 activation, has shown to be notable LRA with multifactorial characteristics. For instance, resiquimod (R848, 96 ), as exhibited in , a imidazoquinoline-based TLR7/8 agonist, showed latency-reversing activity in latently infected cell lines U1 and OM10 by inducing p24 expression . Moreover, resiquimod was capable of diminishing the size of latent reservoirs in HIV-1 infected patients and meanwhile preventing virus replication, suggesting potential clinical value . Vesatolimod (GS-9620, 97 ), a dihydropteridinone-derived TLR7 agonist (EC 50 = 291 nM) could effectively reactivate latent proviruses in PBMCs from HIV-1-infected individuals and also improve immune effector functions at a safe and tolerable clinical dose . Rochat et al. supposed the combination of transcriptional enhancers with a TLR agonist would be more efficient in reversing latency at various levels by generating a Th1 supportive milieu and meanwhile triggering the innate immune system. It was revealed that the combination of thiazoloquinolone-based TLR8 agonist CL075 (3M002, 98 ) with PKC agonist prostratin resulted in significantly augmented latency-reversing activity compared with any single compound, whether in primary cells of HIV-1-infected individuals or on a coculture of J-lat and MDDCs (monocyte-derived dendritic cells). These data implied that the combined use of two LRAs, by which one triggers directly the transcriptional pathway and the other stimulates indirectly the immune system, would be a practicable remedy for efficiently reversing HIV-1 latency . SMU-Z1 ( 99 ), an imidazole-based TLR-1/2 agonist, exhibited excellent latency-reversing potency in HIV-1-infected cells via activation of NF-κB and MAPK pathways in vitro and in PBMCs from HIV-1-infected individuals ex vivo. Besides, SMU-Z1 was also capable of promoting degranulation and gamma interferon (IFN-γ) production in NK cells, as well as increasing TNF-α production in PBMCs without triggering overall T cell activation, indicating the prospect application in eradicating HIV-1 . Considering that some TLR signaling, particularly TLR3, are distributed in microglia cells, which constitute the major HIV-1 reservoir in the brain, Karn and co-workers screened a panel of TLR agonists on different HIV-1 infected cell lines. The results revealed that TLR3 agonists, as anticipated, could efficiently reactivate viral transcription in HIV-1-infected microglia cells, rather than monocytes or T cells, via a previously unreported IRF3, but not the most common NF-κB-mediated mechanism, providing evidence for the therapeutic potential of TLR3 agonists in the treatment of HIV-1 infection in the central nervous system . Despite many beneficial profiles, the precise mechanisms by which the TLR agonists reactivate latent proviruses and modulate immune responses, however, remain incompletely understood. Therefore, more efforts should be made to figure out the detailed mechanisms and identify more robust immunostimulatory TLR agonists against HIV-1 . 2.13. Unclassified LRAs with Undefined Biotargets 2.13.1. AV6 and Analogues Quinoline-derivative antiviral 6 (AV6, 100 ), as displayed in , was identified from a cell-based high-throughput screening protocol, and could reactivate latent HIV-1 in different cell-based latency models without inducing global T-cell activation or proliferation, and could also act synergistically with a HDACI valproic acid (VA) to achieve an additive HIV-1 reactivation . Considering the definite efficacies of HDACI (especially class I HDACI) and AV6, used either alone or in combination, in reactivating latent proviruses, Fang and co-workers expected to develop novel HDACI-derived LRA chemotypes to gain enhanced induction of HIV-1 gene expression. To this end, they investigated various linker lengths and ZBGs, while leaving the N -phenylquinoline skeleton of AV6 unchanged in order to act as a cap region. Among the obtained AV6 derivatives, compounds 101 and 102 exhibited the most potent in vitro latency-reversing activities in J-Lat A2 cells. A preliminary mechanistic study confirmed these two compounds function as dual HIV-1 reactivators by inhibiting HDAC2 isoform and meanwhile facilitating the release of P-TEFb, and 94-dependent HIV-1 gene expression could be obstructed by an immunosuppressant tacrolimus (FK506). However, the anticipated synergetic effect did not occur in combined use of 102 and AV6 . 2.13.2. 2-Acylaminothiazole 2-Acylaminothiazole ( 103 a), as shown in , was able to reactivate HIV-1 gene expression with modest activity (EC 50 = 23 μM) in a high-throughput screening protocol using a dual luciferase reporter cellular assay, which thereby became an ideal hit to develop more chemically diversified LRAs. Among a selection of 2-acylaminothiazole derivatives, 103 b–c, 103 a in particular, gave comparable to increased activity compared to controls, vorinostat and JQ1, in latent CD4 + T cells isolated from cART-treated patients, implying that the introduction of electron-withdrawing groups (e.g., cyano, trifluoromethyl) is favorable to the elevation of HIV-1 reactivation efficacy. Furthermore, synergy could be observed between 103 c and BETI JQ1 in various latent HIV-1 cellular models, indicating the prospects as effective LRAs for treating HIV-positive individuals. However, the exact cellular target(s) of these derivatives were still unclear . 2.13.3. Dihydropyranoindole Derivative Pyranoindole derivative GIBH-LRA002 ( 104 ), as exhibited in , was discovered from a high-throughput screening protocol in the J-Lat cell model, which has a decent HIV-1/SIV reactivation efficacy in resting CD4 + T cells from both chronic SIV-infected rhesus macaques and HIV-1 infected patients. Meanwhile, it possesses relatively low cytotoxicity without causing overall T cell activation, thus avoiding new viral infections, indicating that GIBH-LRA002 might be an attractive candidate for anti-HIV treatment . 2.13.4. Carbazole Derivative Curaxins CBL0137 ( 105 ), as manifested in , has been described as a potent FACT (facilitates chromatin transcription) inhibitor, which exerts anticancer efficacy by suppressing the NF-κB pathway and activating tumor suppressor gene p53 . As previous work has proved the NF-κB pathway plays an important role in disrupting latent HIV-1, while the depletion of FACT also could reactivate HIV-1 proviruses, CBL0137 was thereby submitted to the anti-latency evaluation in post-integrated latency cells, JLAT6.3 and CA5. The result evinced that CBL0137 alone could not only sufficiently potentiate TNF-α stimulated HIV-1 transcription without inducing any cytotoxicity, but also reactivate latent reservoirs in PBMCs from HIV-1-infected patients, suggesting the potential clinical application as an appealing LRA candidate . 2.13.5. Benzazole Derivative Benzazole derivative 106 a (see ) was found to have well-defined HIV-1 reactivation activity and a favorable pharmacokinetic profile in a high-screening protocol in 24STNLSG cells, which thereby was a good hit for further chemical optimization. Among the synthesized benzazole derivatives, 106 b gave the best HIV-1 reactivation effectiveness and the lowest cytotoxicity, which could induce proviral transcription in several latently infected cell types without affecting overall T cell activation. However, the 4-NH2 containing benzazole derivative 106 c provided decreased HIV-1 latency potency in a latently infected ACH-2 cell line. Preliminary SAR outcomes have been summarized in . The preliminary mechanistic study revealed that these benzazole derivatives appear to have a unique mechanism of action. They were not HDACIs and the HIV-1 transcription was driven neither by the NF-κB signaling pathway nor by stimulating of HIV-1 LTR. These data indicated that benzazole might be a promising chemotype to be developed into potent LRAs with appropriate chemical optimizations and further investigation into the mechanism of actions . 2.13.6. Pyridine Derivative Vitamin B3 niacinamide ( 107 ) (see ), a pyridine-derived sirtuin inhibitor, showed notably improved HIV-1 reactivation potency compared to that of the combination of two validated HMTI-based LRAs, chaetocin and BIX01294, in an ex vivo assay for short-term treatment, indicating its clinical potential. Meanwhile, owing to the relatively simple structure, niacinamide might be utilized as a suitable hit to carry out further chemical optimization to find more potent LRAs . 2.13.7. Polyphenols Resveratrol ( 108 ), as revealed in , a natural polyphenol, was capable of reactivating latent HIV-1 without triggering overall T cell activation. Besides, a synergistic HIV reactivation was observed when resveratrol was combined with other conventional LRAs with different mechanisms of action, such as JQ1 (BETI), SAHA (HDACI) and prostratin (PKC activator). A preliminary mechanistic study revealed the latency-reversal activity of resveratrol was due to the activation of HSF1 and increased histone acetylation, but not the activation of silent information regulator 1 (SIRT1), which belongs to a member of the sirtuin family. However, due to relatively low bioavailability, more polyphenol analogues of resveratrol ( 109 ~ 114 ) were successively submitted to the HIV-1 reactivation evaluation. The result was that among these polyphenols, only triacetyl resveratrol ( 109 ) gave comparable latency-reversal potency with the prototype resveratrol, while none of the others could successfully reactivate the latent proviruses in vitro, suggesting more effort is still required to find more potent polyphenol-based LRAs by expanding the chemical diversity of a synthetically or naturally available stilbenoid chemotype. Additionally, the cotreatment with triacetyl resveratrol ( 109 ) with other LRAs also generated a synergistic effect . Q 205 ( 115 ), a synthetic resveratrol analogue, was proved to effectively reactivate latent HIV-1 in vitro without inducing of damaging cytokines. A preliminary mechanistic study revealed that the latency-reversal potency of Q205 was attributable to the activation of P-TEFb and promotion of Tat-mediated HIB-1 transcription and binding of RNAPII to the HIV-1 5′-LTR promoter. The results evinced that identification of promising LRAs from naturally available polyphenols and/or chemically optimized resveratrol derivatives might be a feasible and practicable alternative . HDACs are a group of epigenetic enzymes that can significantly affect chromatin topology and the histone deacetylation process by removing functional acetyl groups from the N-terminus of lysine residues and facilitating a high-affinity interaction between histones and DNA backbone, leading to a condensed inactive chromosomal DNA structure and consequent blockage of gene transcription. There are a total of 18 isoforms of mammalian HDACs, which are divided into four classes (class I, II, III and IV), largely based on the sequence homology, cellular location and folding mode of peptide chains. Class I contains three subtypes, IA (HDAC1, HDAC2), IB (HDAC3) and IC (HDAC8). Class II includes two subtypes, class IIa and IIb, in which class IIa consists of HDAC4, HDAC5, HDAC7 and HDAC9, while class IIb includes HDAC6 and HDAC10. Class III HDACs, also known as sirtuins, are nicotinamide adenine nucleotide (NAD) + -dependent enzymes, which contain seven members (sirtuins 1~7). Class IV contains only HDAC11. HDACs generally refer to zinc-dependent class I, class II and class IV isozymes, unless noted otherwise . HDACs contribute to proviral gene silencing of HIV latency by directly deacetylating histones at proviral integration sites (5′-long terminal repeat, 5′-LTR) or by indirectly inducing deacetylation of non-histone proteins (e.g., NF-κB). Thus, inhibiting HDACs can facilitate changes in chromatin architecture and recruitment of host transcription factors to LTR, leading to the acceleration of viral transcription. Using small molecular HDACIs as epigenetic modifiers thereby represents a viable and predominant strategy to eliminate latent reservoirs . Structurally, currently identified HDACIs typically follow a common pharmacophoric feature by mimicking the structure of the natural substrate lysine, which comprises a surface recognition cap moiety that can tolerate structural variability to accommodate the broad hydrophobic region of HDAC (Cap region); a functional zinc-binding group (ZBG) that can orient and coordinate with the catalytic zinc ion; and a linear or cyclic linker with 5~7 atoms (Linker) that traverses the long and narrow tunnel to connect the Cap portion and ZBG . In the past decades, a variety of HDACIs hits or candidates that vary in skeletal structures have been developed by modulating these pharmacophoric fragments and submitting to bioevaluation in various stages, aiming at achieving both elevated potency and isoform selectivity against HDACs-mediated pathological conditions, particularly hematological malignancies and solid malignant tumors. At present, vorinostat (SAHA), which has been approved for the treatment of cutaneous/peripheral T-cell lymphoma, is by far the clinically best-studied HDACI to be fully assessed for latency-related anti-HIV-1 therapies, either as a single regimen or in combination with other types of LRAs. SAHA showed promising in vitro HIV-1 latency-reversing effects in multiple HIV-1 latently infected cell lines (e.g., ACH2, U1 and J-Lat) and a latent provirus that was isolated from resting CD4 + T cells in HAART patients. However, the therapeutic outcome from pan-HDACI SAHA is widely limited by its insufficient selectivity towards specific isoforms, resulting in many unwanted side effects, including dehydration, anorexia, thrombocytopenia, arrhythmia, and also poor pharmacokinetic (PK) profiles. Hence, intensive structural modifications have been carried out to procure more potent HDACIs with improved selectivity and less toxicity . To achieve improved efficacy towards HIV-1 latent reservoirs, Okamoto et al. presented a spectrum of structural mimics of SAHA by employing a structure-based drug design protocol, based on the obtained crystal structure of HDAC in complex with SAHA. Specifically, the hydroxamic acid of SAHA that acts as a ZBG was initially replaced by an acylated thiol group, while the cap moiety was prolonged by inserting a thiazole or phenyl motif. It resulted in HDAC inhibitory potencies of two compounds, NCH-51 ( 1 ) and NCH-51 ( 2 ), increasing by about 2–3 fold over that of prototype SAHA, with selective preferences towards HDAC1/4/6 isoforms. Additionally, compared with congener 1, NCH-51 has more potent HIV latency-reversing effects in latently infected cell lines, OM10.1 (CC 50 = 2.2 µM) and ACH-2 (CC 50 = 2.4 µM), without causing obvious cell death, making it a promising lead. Encouragingly, owing to the introduction of an acylated thiol group instead of hydroxamic acid, which is supposed to have a potential mutagenicity side effect, these two compounds exhibited better PK profiles and lower cytotoxicity than the parent compound SAHA . Considering that the phosphate group can also function as a potential ZBG, Etzkorn et al. hypothesized that the substitution of phosphate analogues (phosphonamidate, phosphonate, phosphinate) for hydroxamic acid might afford equal potency while avoiding the limitations of hydroxamic acid. Keeping this in mind, they retained the cap part of SAHA unchanged and modified the linear linkage with a one-carbon degradation. Unfortunately, the resulting SAHA analogues 3 ~ 5 lost HDAC-binding affinities. One possible reason might be that the negatively charged phosphates may sterically hinder the coordinative contact with the catalytic zinc ion of HDACs, resulting in dramatically impaired efficacies . To discover more SAHA-derived HDACIs, Pflum et al. focused on the bridged linkers to explore chemical diversity and enrich SAR information. To this end, a series of SAHA analogues 6 ~ 9 were synthesized by appending branched substituents with different sizes and dispositions. Surprisingly, except for the benzyl-attached compound 8 that only gave a decent HDAC8 inhibitory affinity, nearly all these compounds furnished significant lowered potencies but exerted preferential selectivity towards HDAC6/8 versus HDAC1/2/3 compared with the archetype SAHA, indicating that structural modifications on bridge portion might be conducive to selective inhibition for HDAC6/8 . As the phosphorus-based ZBGs might largely account for the unfavourable HDAC inhibition, further structural optimizations were then focused on other parts of SAHA. Apicidin, a fungal metabolite bearing a cyclic tetrapeptide unit, is a naturally originated HDACI (IC 50 = 0.7 nM) . Inspired by this, Etzkorn and co-workers postulated that the macrocyclic peptide portion might act as a hydrophobic surface recognition group (Cap), and further modifications were conducted by simplifying the macrocyclic skeleton and constructing a linear chain of five carbon atoms (linker). Considering the strong zinc-binding affinity, the ethyl ketone part of Apicidin was replaced with hydroxamic acid (ZBG). As expected, the yielded hybrid 10 gave a 2-fold improvement in HDAC inhibition and nearly a 5-fold selectivity for HDAC1 over HDAC8. To pursue this lead further, a ring-opening operation was implemented from different locations to expose carboxylic acid and aryl substituent, respectively, aiming to examine the influences of rigid macrocycles and flexibility on activity and selectivity. The resulting compound 11 , however, decreased both activity and selectivity, while another ring-opening product, 12 , gave comparable efficacy only with its precursor, 10, implying that the hydrophobicity and conformational distribution of the cap region may have a significant impact on HDACs inhibition and selectivity, and the restricted macrocycllic ring is more suitable than the flexible side chains for HDAC1-selective inhibition. Likewise, the impaired potency of compound 11 in comparison with 12 might also be attributed to the charged groups . The aforementioned efforts on the structural variations derived from SAHA or Apicidin, as seen in , have provided useful SAR information for further identification of more potent HDACIs, although these newly identified compounds 3 ~ 12 were not submitted to the HIV-1 reactivation evaluation. Actually, similar to SAHA, in the past decades many synthetic and naturally available HDACIs as anticancer agents, including but not limited to the hydroxamic acid-based givinostat (ITF2357), panobinostat (LBH589), nanatinostat (CHR-3996), pracinostat (SB939) and belinostat (PXD101); benzamide-based mocetinostat, entinostat (MS-275), as evinced in , also displayed favorable HIV latency-reversing activities in various latently infected cell lines, and these agents are typically well-tolerated by participants. Among them, thiol-based romidepsin is the most effective LRA to date, which has supported the “proof-of-principle” that latent reservoirs can be safely activated, and perhaps, entirely eliminated . Rasmussen et al. compared the effects on HIV production in latently infected cells (U1 and ACH2) as well as T-cell activation of several hydroxamic acid-based HDACIs that were undergoing clinical development. The results indicated that these HDACIs gave different degrees of HIV-1 reactivation potencies at therapeutic concentrations, with activity order of panobinostat > givinostat ≈ belinostat > SAHA. However, all these HDACIs induced moderate T-cell activation, which hindered their further clinical application . As we know, HDACI treatment in HIV-1-infected individuals generally suffers from abnormal T-cell activation and nonspecific HDAC inhibition, leading to undesired HIV-1 persistence and other side effects by causing clonal expansion of latently infected rCD4s, which has been the major bottleneck for warranting further clinical investigation . Thus, an ideal LRA candidate for achieving the desired ‘shock and kill’ tactics should have the capability of stimulating latent HIV-1 transcription without provoking homeostatic proliferation and/or extensive T-cell activation that is highly correlated with the cytokine release, so as to avoid possible immune hyperactivation and the consequent concomitant cytokine storm, also known as cytokine release syndrome (CRS), as well as acceptable PK properties . Fimepinostat (CUDC-907), a dual inhibitor of class I-selective HDACI and PI3Kα, might be a suitable LRA candidate. It not only displayed comparable latency-reversal activity with romidepsin, the current most effective HDACI tested in anti-HIV-1 trails, at the cellular level, but caused reduced T-cell activation without any negative influence on T cell proliferation . Accumulating evidence indicates that class I-selective HDACIs—especially HDAC-1, -2 and -3, with HDAC3 isoform being the most important, functioning as transcriptional “on switches” of latent viruses and maintaining the deacetylated state of reactivation-related transcription factor nuclear factor κB (NF-κB)—might be more effective than pan-HDACIs in eradicating HIV latency by inducing more latent proviruses . Taking thiol-based orally active pan-HDACI ST7612AA1 ( 13 ), as another example, acting as a prodrug and potent HIV latency activator, ST7612AA1 actually exerts an HIV reactivation effect by transforming into its active form, ST7464AA1 ( 14 ), a class I-selective HDACI, via in vivo hydrolysis . For instance, Lewin SR and co-workers compared the HIV-1 reactivation efficacies of class I-targeted HDACI (entinostat) and three pan-HDACIs, SAHA, panobinostat and oxamflatin (metacept-3, MCT-3); their results also proved that entinostat gave the most potent HIV-1 latency-reversal activity by inducing more viral expression . The fact that HDAC3 highly selective inhibitor BRD3308 not only was active for latent HIV-1 reactivation in 2D10 cell model but could induce viral outgrowth from rCD4s of antiretroviral-treated patients further proved that class I-targeted HDACIs, especially HDAC3 inhibitors, are particularly effective anti-latencyt agents with improved HIV-1 reactivation potencies and fewer effects on other unrelated cellular genes . Similarly, to discover more potent class I (HDAC1/2/3)-selective HDACIs for eliminating a latent reservoir, Yu and co-workers from Merck have develop an array of ethyl ketone-based macrocyclic HDACIs, referring to the skeleton of class I-selective HDACI apicidin via ring expansion while retaining the linker and ZBG (ethyl ketone) unchanged, since the macrocycle structure is believed to have greater binding affinity with HDAC2 subtype, as evinced in . The facts that the two most potent class I-selective macrocyclic HDACIs, 15a and 15b , gave both enhanced class I HDAC inhibition and HIV latency-reversal potencies, supported the strategy for designing macrocyclic class I-selective HDACIs as promising LRAs, to a great extent. However, due to relatively low bioavailability of macrocycles, the PK profiles of these macrocyclic HDACIs still need further improvement . Granted, there are some cases that have proven otherwise. For example, another two effective class I-selective HDACIs, nanatinostat and romidepsin, as shown in , could not induce the generation of viral antigens or particles from rCD4s, partially owing to the lack of effective accumulation of spliced viral transcripts, despite in vitro effectiveness in latency-infected cell lines. Along with this, both of them impaired the function of CD8 + T cells, with romidepsin causing more impairment, which might explain, to a certain degree, the unsatisfactory clinical evaluation of various HDACIs in ARV-suppressed individuals so far . Colletively, as we can see from the above description, although HDACIs, especially class I-HDAC selective inhibitors, are a class of well-acknowledged LRAs with promising in vitro and ex vivo anti-latency effectiveness, which HDAC subtypes actually have a direct impact on HIV-1 latency in vivo is still not yet entirely elucidated. Hence, better investigative tools, whether more potent HDACI-based LRAs or latency screening models, particularly appropriate in vivo evaluation models, that can translate this knowledge into clinical chemotherapies are in urgent demand. As one of the chief epigenetic modifications, DNA methylation is supposed to play a role in governing HIV latency by epigenetic regulating of 5′-LTR cytosine-phosphate-guanine (CpG) methylation and inhibiting HIV viral transcription initiation. LTR contains two CpG islands and a HIV-1 promoter can be hypermethylated at these two CpG islands, particularly island 2, surrounding HIV transcription initiation sites. Besides, methyl-CpG binding domain protein 2 (MBD2) can specifically bind to methylated DNA and then recruit HDAC1/2, leading to chromatin unwinding and gene silencing and consequently histone deacetylation . The DMTI 5-aza-2′-deoxycytidine (decitabine, 5-aza-CdR, 16 ) which has been approved by FDA for the therapy of myelodysplastic syndrome (MDS), as shown in , has proved to demonstrate weak HIV-1 reactivation activity but displays an intensified promoting effect when utilized in combination with other HIV-activating agents, such as tumor necrosis factor α (TNFα), PKC activator prostratin or HDACIs in most J-Lat cell lines . However, the HIV-1 reactivation ability of decitabine, either used alone or in combination with other types of LRAs, displayed strong cell type dependence, implying a more comprehensive assessment should be carried out when using decitabine as a LRA in HIV reactivation trials. Another DMTI azacitidine (5-azacytidine, 17 ) also could induce latent HIV-1 proviruses . Lint and co-workers have observed that decitabine-induced HIV-1 reactivation was accompanied by a decreased recruitment of ubiquitin-like with an epigenetic integrator, PHD and RING finger domain 1 (UHRF1), to the viral promoter, implying UHRF1 might be a promising pharmacological target for discovering potent LRAs. To provide a demonstration of this proof-of-concept finding, this lab further utilized epigallocatechin-3-gallate (EGCG, 18 ), a polyphenol from green tea, which has been reported to possess certain UHRF1 inhibitory activity , and NSC232003 ( 19 ), a specific UHRF1 inhibitor, as two chemical probes, to ascertain the LRA potential of UHRF1 inhibitors. The results exhibited that EGCG partially reactivates latent proviruses through the inhibition of UHRF1, whereas NSC232003 provoked significant HIV-1 reactivation, highlighting the tight correlation of anti-UHRF1 with latency-reversal potency . Despite some positive latency-reversal outcomes from decitabine and azacitidine, the role of DNA methylation in the formation and maintenance of HIV-1 latency, however, is still in dispute. Blazkova et al. detected very low levels of methylated CpG in HIV-1 infected individuals under antiretroviral treatment, probably because activation is limited due to proviral DNA hypermethylation, highlighting the necessity for a deep understanding of the underlying heterogeneity of DNA methylation on HIV-1 latency as well as a more reasonable assessment of DMTIs as latency activators . All in all, since DNA methylation affects both the viral and host genome, silencing of HIV transcription via inhibition of DNA methylation also causes aberrant methylation of the host genome. This process is irreversible, and vice versa, implying that DNA methyltransferase is likely not the ideal strategy for developing effective HIV-1 reactivating agents. However, the DNA methyltransferase-associated UHRF1 is expected to be an attractive pharmacological target to explore more effective LRAs. In addition to DNA/CpG methylation, histone methylation—which primarily occurs at the N -terminal arginine or lysine of H3 and H4 histones and is covalently modified by an epigenetic enzyme, histone methyltransferase (HMT)—is also essential for the establishment of the silencing of HIV-1 transcription and maintenance of genome stability. Four important lysine methyl transferases (KMTs), EZH2, SUV39H1, G9a and G9a-like protein (GLP), are the most extensively studied HMTs. EZH2 is located at the promoter of latent HIV-1 provirus in T cells, participates in histone H3 lysine 27 trimethylation (H3K27me3), and plays a major role in chromatin-mediated HIV-1 transcriptional regulation and viral suppression. SUV39H1 primarily participates in H3K9 trimethylation (H3K9me3). G9a, also known as euchromatic histone-lysine N-methyltransferase 2 (EHMT2), is responsible for H3K9 dimethylation (H3K9me2) by catalyzing the addition of a methyl group from S-adenosyl-L-methionine (SAM) to a histone lysine residue. G9a-like protein (GLP), also called EHMT1, is another H3K9 methyltransferase with 80% sequence homology to G9a in the suppressor of variegation 3–9, enhancer of zeste and trithorax (SET) domains . Both G9a and GLP, together with SUV39H1, play pivotal roles in transcriptional silencing in HIV-1 latency. Histone methylation marks H3K9me2, H3K9me3 and H3K27me3 in particular have been validated to prevent lysine from being acetylated and meanwhile keep the chromatin in a dense state in both latency cell lines and primary CD4 + T cell models, and therefore are considered as important histone methylation marks . Quinazoline BIX01294 ( 20 ), a specific G9a inhibitor, was the first identified HMTI with clear in vitro HIV latency-reversing activity from a high-throughput screening protocol . The subsequent resolved X-ray co-crystallographic structure of BIX01294 bound to HMT (PDB code: 3K5K) as well as a SAR investigational outcome of BIX01294 have provided useful clues for further structure-based drug design and chemical optimizations . As indicated in A, HMT contains three active binding grooves: pockets I and II are two functional solvent regions; pocket III, also referred to as a lysine-binding channel, functions as a methyl transfer and transportation channel ( A). The quinazoline core of BIX01294 occupies the central histone peptide binding cavity, while a benzyl-substituted piperidine moiety and a 1,4-diazepane fragment extend into pocket I and II, respectively, and a 7-methoxy group orients towards pocket III, a long and narrow region ( B). Besides, 4-NH, two N atoms of piperidine and a diazepane ring form the key H-bonding force network with three acidic amino acids, Asp1078, Asp1083 and Asp1074, which have significantly contributed to its beneficial HMT inhibitory potency. To further explore the SARs and discover more potent HMTIs, Jin et al. employed several rounds of chemical optimizations on BIX01294. Initially, a 4-amino substituted group (R 1 ) of quinazoline skeleton was investigated. Since the 4-NH forms an important H-bonding force with Asp1083 while the benzyl group has extended beyond the active interfaces without contributing any protein-ligand interactions, the influences of piperidine N and the sizes of substituents on the bioactivity were studied by removing the redundant beznyl portion while retaining the 4-NH unchanged. As a result, compound 21 with 1-methylpiperidine moiety gave an 8-fold improvement in G9a binding affinity (IC 50 = 0.23 μM) compared with the prototype BIX01294 (IC 50 = 1.9 μM), indicating that piperidine N shows great promise to increase the drug-like profiles of quinazoline derivatives by not only providing multiple molecule-protein interactions but also producing ameliorated water solubility and bioavailability. By contrast, introduction of small-sized groups (e.g., cyclopropyl or isopropyl) results in obviously impaired potency. As far as 2-substituent (R 2 ) was concerned, both N-containing 6-membered (e.g., compounds 22 and 23 ) and 7 -membered cycloalkanes were tolerated. When it comes to R 3 group, although it oriented towards the water-exposed tolerant lysine-binding region (pocket III), it did not fit this narrow space properly, which has offered an underexploited but valuable site for further exploitation. Among the investigated hydrophilic side chains, introduction of N , N -dimethylbutane (compound 24 ) gave a marked improvement on G9a affinity by providing an additional water-mediated contact and thus yielding more favorable adaptability to the lysine-binding channel, elevating affinity by about 127 times, in contrast to the prototype BIX01294. The molecular modeling docking result of compound 24 in complex with G9a (PDB: 3K5K), as evinced in , has explained, in a large part, why both the multi-site contacts and N -bearing side chain that inserts into lysine-binding channel greatly account for its high potency. Specifically, in addition to the hydrogen bond formed between 4-NH and ASP1083, the piperazine N atom forms a salt bridge with ASP1074 and an electrostatic interaction with ASP1078. Besides, 1-N of quinazoline ring is supposed to be protonated under physiological pH conditions, forming a salt bridge with ASP1088. More importantly, the protonated N atom of the N , N -dimethylbutane part forms not only a hydrogen bond with Leu1086, but also a cation-π interaction with Tyr1154 (displayed as red circle), which contributed much to the binding affinity . Fuchter et al. examined the influences of a central quinazoline ring and 6,7-dimethoxyl groups of BIX01294 on HMT (G9a and GLP) inhibition. To this end, a scaffold-hopping strategy was conducted by replacing the quinazoline core with various bioisosteric moieties. The result is that two methoxyl groups, especially the 7-methoxyl group, are essential, since the replacement of two methoxyl groups with a dioxalone ring caused markedly impaired potency. Among the varied heterocycles, quinoline-derived derivative 25 gave the most potent affinity, which is largely owing to the basicity of protonated 1-N. Meanwhile, the 3-N atom of the quinazoline skeleton is not a necessity for G9a inhibition . Noticing that there was still room for activity improvement for compound 24 , since N , N -dimethylbutane moiety did not entirely occupy the lysine-binding region, Jin et al. systematically examined how the length and disposition of 7-substituents of quinazoline derivatives affected the potency. As a consequence, compound 26 afforded the most potent G9a inhibition to date, with a 250-fold improved potency over the prototype BIX01294, as evinced in . Albeit with excellent enzymatic activity, compound 26 presented lower cellular activity than BIX01294, which is largely attributable to its poor cell membrane permeability and low lipophicity (log P = 1.9). To overcome this bottleneck, this group further modified 26 by introducing various lipophilic groups, which has led to two promising hits. Compound 27 can effectively reduce the level of dimethylated H3K9 in many cell lines and shows high selectivity for G9a and GLP with low cytotoxicity . Compound 28 not only exhibited high potency and selectivity for G9a and GLP, but has measurable DNMT1 inhibitory activity against DNA methyltransferase, implying that this compound has the potential to become a HMT/DNMT1 dual inhibitor . However, 28 furnished a poor pharmacokinetic profile in an animal assay. The most probable reason might be due to the 2-cyclohexyl group, which is readily oxidized with the action of cytochrome p450 enzyme in vivo, leading to metabolic instability. Given that substituents at the 2-position of the quinazoline ring were well tolerated, this lab further optimized 2-substituents with the expectation of acquiring improved metabolic stability while maintaining high potency and high cellular activities. When 2-cyclohexyl was replaced by a 4,4-difluoropiperidinyl unit, the resulting compound 29 manifested the optimal performance and has been the first chemical probe against G9a and GLP used in animal studies . Sbardella et al. employed a ring expansion strategy by replacing the quinazoline skeleton of 24 with a benzodiazepine framework to afford 1,4-benzodiazepine derivative 30 , which gave about a 35-fold improved DNMT1 potency, comparable G9a inhibition, lowered cytotoxicity and better metabolic stability than the hit 24 , indicating a broad prospect to be developed as a HMT/ DNMT dual inhibitor . Similarly, Oyarzabal et al. utilized compound 28 as a hit to design more potent HMT/DNMT dual inhibitors. The quinazoline scaffold was initially refined, since the 3-N atom actually does not form any interactions with a protein backbone. As expected, both G9a and DNMT1 inhibitions of the obtained compound UNC-0638 ( 31 ) have been greatly improved. To guide further modifications, the binding mode of 31 was investigated by docking it into mouse DNMT1, as shown in . The result evinced that the quinoline core properly occupies the NNMT1 backbone and 1-N forms a H-bonding interaction with Glu1269 (Glu1266 in human DNMT1), 7-O atom 4-NH and contributes to H-bonding contacts with Arg1315 (Arg1312 in human DNMT1) and Ser1233 (Ser1230 in human DNMT1), respectively. As to the 2-cyclohexyl unit, it did not form any interactions with protein, while the 4-terminal isopropyl piperidine group forms a hydrophobic interaction with Met1235 (Met1232 in human DNMT1) and the 6-methoxyl group points towards the 3′-direction of the DNA strand, implying that subsequent structural variations might be conducted at 2- and 4-positions. Guided by this docking result together with previously obtained SAR results, several more potent inhibitors 32 ~ 36 have been identified. However, comparing with their excellent G9a inhibition, there is still room for improving the DNMT1 affinity. Therefore, more detailed and comprehensive SAR information still needs to be replenished. Chaetocin ( 37 ), a fungal mycotoxin from Chaetomium minutum and a specific Suv39H1 inhibitor that works as a competitive inhibitor for SAM, as exhibited in , was capable of inducing 86% latent proviruses from HIV-1 infected HAART-treated patients without affecting T-cell function . By contrast, the control HMTI BIX01294 gave slightly decreased (80%) potency under the same conditions . Furthermore, EZH2 inhibitors, such as GSK343 ( 38 ), 3-deazaneplanocin A (DZNep, 39 ) and tazemetostat (EPZ-6438, 40 ), and EHMT2 inhibitor UNC-0638 were also reported to display strong anti-latency-reversing activities. Intriguingly, two selective EZH2 inhibitors GSK343 and tazemetostat furnished more effective latency-reversing activities than the broad EZH2 inhibitor DZNep, further proving the direct association between specific EZH2 inhibition and HIV-1 reactivation. In addition, the combination of any of these HMTIs, as displayed in , with HDACI (e.g., SAHA, vorinostat) procured a strong synergistic HIV- 1 reactivation effect . Actually, with advancement in various molecular biology techniques, more and more previously unrecognized histone modification-associated factors underlying HIV-1 latency are being disclosed in addition to the widely investigated EZH2, G9a and GLP. For example, since the latency-related histone mark H3K27me3 is catalyzed by polycomb repressive complex 2 (PRC2), while embryonic ectoderm development (EED) is a principle component of PRC2, inhibition of EED is therefore inferred to be largely beneficial to latent provirus reactivation similarly to other HMTIs. James et al. proved that two EED inhibitors (EEDIs), A-395 ( 41 ) and EED226 ( 42 ), not only presented promising HIV-1 reactivation efficacies after treating alone but also gave an additive effect in latently infected 2D10 cells when in combination with EZH2 inhibitors, GSK343 or UNC1999 ( 43 ), further confirming that PRC2-mediated components, such as EZH2, EED and SUZ12, can serve as attractive targets to develop more types of LRAs . Besides, Ott and co-workers disclosed a previously undiscovered lysine methyltransferase, SET and MYND domain-containing protein 2 (SMYD2), which is also a potent target for identifying latency-reversal agents. SMYD2, which works as an epigenetic co-repressor, is found to be tightly linked with HIV-1 latency by inducing monomethylation on histone H4K20me1 at the HIV-1 5′-LTR region, leading to repression of proviral transcription. Accordingly, H4K20me1 can also act as an important histone mark in addition to H3K27me3 and H3K9me2/3. The fact that effective reactivation of latent proviruses with a SMYD2 inhibitor AZ391 ( 44 ) in CD4 + cells further supported the inhibition of the SMYD2’s catalytic activity being directly correlated with HIV-1 reversal activity . Meanwhile, given that H4K20me1 is recognized by a chromatin “reader” protein lethal 3 malignant brain tumor 1 (L3MBTL1), leading to chromatin compaction and consequent transcriptional silencing, L3MBTL1 thereby becomes a possible HMT pathway-related target for exploiting LRAs, although its exact role in modulating HIV-1 latency still needs further verification . Taken together, among the proposed HMT-associated latency factors, EZH2 has stronger associations with HIV-1 reactivating efficacy compared with other HMT-related factors, including but not limited to G9a, SUV39H1, EED, SMYD2 and L3MBTL1. Hence, specific and effective EZH2 inhibitors are strongly encouraged. The protein kinase C (PKC) family of serine/threonine kinases plays an essential role in reactivating latent HIV-1 through activation of the NF-κB signaling pathway . A host of small molecules derived from natural sources, including but not limited to phorbol esters, ingenol esters, ingols, jatrophanes, and various macrolides, have been proposed to reactivate HIV-1 in latently infected CD4 + T cells, as displayed in . Among them, ingenol derivatives phorbol 12-myristate 13-acetate (PMA, 45 ) and 12-deoxyphorbol-13-acetate (prostratin, 46 ) are two most well-known activators of PKC with nanomolar PKC activation affinities. Although PMA reactivates latent HIV-1 by activating T cells, its clinical utility was limited by tumor-promoting risk and other serious side effects, such as mitotic dysfunction and chromosomal aberrations . Unlike PMA, prostratin, which is isolated from the poisonous New Zealand plant Pimela prostrata, is not a tumor promoter and does not induce cell proliferation by itself. Besides, prostratin can inhibit PMA-induced tumor promotion in a mouse model, illustrating a broad scope in clinical application. Prostratin not only reactivates latent HIV-1 in vitro in a PKC-dependent NF-κB activation manner, but also down-regulates the expressions of HIV-1 receptor CD4 and co-receptor CXCR4, thus avoiding the novo infection of CD4+ cells. However, since prostratin causes overall T cell activation, just as PMA does, further investigation is still needed in order to interpret the suitability of this compound for use in humans . Another two ingenol derivatives, phorbol 13-stearate (P-13S, 47 ) and 12-Deoxyphorbol 13phenylacetate (DPP, 48 ), also presented attractive HIV-1 reactivation potencies. P-13S effectively activates HIV-1 gene expression in the Jurkat-LAT-GFP latency model, with at least a 10-fold increase in potency over that of prostratin. Interestingly, P-13S activates PKC by inducing a translocation of PKC isotypes α and δ to cellular compartments, which is distinctly different from that of prostratin and PMA . As to DPP, a non-tumour-promoting phorbol ester isolated from the West African “candle plant” Euphorbia poissonii and the Moroccan succulent E. resinifera Berg, was reported to induce HIV-1 gene expression in latently infected ACH-2 cells at a 20–40-fold lower concentration than prostratin . The ingenol analogue ingenol-3-angelate (PEP005, 49 ) that was isolated from Euphorbia peplus was validated to reactivate latent HIV-1 through a PKC-dependent NF-κB pathway. Moreover, this compound was able to prevent new rounds of viral infection after HIV-1 reactivation through down-regulating the expression of the HIV-1 receptors CD4 and CXCR4 . The ingenol derivative EK-16A ( 50 ) that was isolated from Euphorbia kansui displayed 200-fold more potent efficacy than prostratin in reactivating latent proviruses in vitro and ex vivo with minimal cytotoxicity on cell viability. Besides, EK-16A could induce synergistic effects with multiple types of LRAs, such as DNMTI 5-Aza, BETIs JQ1 and I-Bet151, as well as HDACIs vorinostat and romidepsin in latently infected cell lines J-Lat 10.6 and 6.3, without any influence on T-cell activation. Mechanistically, EK-16A works as a PKCγ activator to promote HIV-1 transcription HIV-1 reactivation via the activation NF-κB pathway and also to facilitate HIV-1 elongation via the stimulation P-TEFb pathway . Later, three ingenol derivatives, viz. EK-1A ( 51 ), EK-5A ( 52 ) and EK-15A ( 53 ), were isolated from Euphorbia kansui, and not only demonstrated latent reactivation efficacies in vitro and ex vivo at nanomolar concentrations but also could inhibit acute HIV-1 infection via down-regulation of the expression of CCR5 and CXCR4, two cell surface HIV co-receptors. Additionally, these three ingenol derivatives have a synergy with 5-Aza, SAHA, JQ1 or prostatin with little cellular toxicity in T-cells . These data suggested ingenol derivatives derived from Euphorbia species might have great prospects for being developed into successful chemotherapeutic LRAs. Macrocyclic jatrophane diterpenoid compound SJ23B ( 54 ), which is isolated from a Mediterranean plant specimen, E. hyberna, is an activator of PKCα/δ. It could reactivate latent HIV-1 via activation of the NF-κB pathway at a nanomolar level (EC 50 = 50 nM), which is at least 10 times more potent than prostratin. Moreover, SJ23B is not a tumor promoter and displayed strong in vitro anti-HIV-1 activity . The epimeric N , N -dimethylvalinoyl-4α-4-deoxyphorbol derivatives 55 and 56 were isolated for the first time from a medicinal Mexican Croton, and were later found to be potent and isoform-specific activators of PKC with promising HIV-1 reactivating activity . 3,12-Di-O-acetyl-8-O-tigloyl-ingol ( 57 ) that was isolated from Euphorbia lactea, a plant that produces latex with anti-inflammatory activity, was reported to antagonize HIV-1 latency through a PKC-dependent pathway . Likewise, 8-methoxyingol 7,12-diacetate 3-phenylace ( 58 ) which was derived from the latex of Euphorbia officinarum, could also reactivate HIV-1 latency with an EC 50 < 25 µM . Macrolides bryostatins that are isolated from bryozoan are a family of effective PKC activators, among which bryostatin-1 ( 59 ) exhibited the most powerful PKC activating activity at a nanomolar concentration and has been the only PKC agonist that entered clinical evaluation as an effective LRA candidate. Bryostatin-1 not only exhibited significant HIV-1 reactivation efficacy in human astrocytes via a NF-κB and PKC-dependent mechanism, but induced a remarkable decrease in viral production and amyloid beta (Aβ) deposition in myeloid cells. Bryostatin-1 can also activate the mitogen-activated protein kinase (MAPK) pathway and down-regulate the expressions of HIV-1 co-receptors CD4 and CXCR4 without triggering global T cell proliferation, and it has synergy with HDACIs in reactivating latent HIV-1 . All these beneficial properties together with the fact that bryostatin-1 was capable of avoiding de novo infection in HIV-1 in susceptible cells makes it a promising adjunct for the treatment of HIV-1 brain infection . Enlightened by the promising therapeutic potential and SAR outcomes of bryostatin-1, Stone and co-workers prepared two bryostatin-1 analogs, compounds 60 and 61 , which offered comparable to higher PKC-binding affinities compared to the prototypes bryostatin-1 and PMA . In addition, the semi-synthetic ingenol ester 62 was reported to exert HIV-1 reactivation efficacy by activating PKCs and up-regulating the positive transcription elongation factor b (P-TEFb), thus promoting both transcription initiation and elongation of viral genes. Given that ingenol analogues have rich natural resources, compound 62 is therefore a promising LRA candidate to be utilized in clinical practice . Gnidimacrin ( 63 ), a diterpenes PKC activator, can potentially activate latent HIV-1 viruses, with about 10 times more potency than HDACI SAHA at an effective concentration as low as the picomolar level. It is especially noteworthy that 63 can significantly reduce the size of a latent reservoir by decreasing the amount of latently infected cells at a concentration that does not cause overall T cell activation or stimulate production of inflammatory cytokines, indicating its promising clinical application . BL-V8-310 ( 64 ), a benzolactam-related PKC activator, was proved to effectively reactivate latent HIV-1 in latently infected ACH-2 and J-Lat cell lines. Moreover, combining BL-V8-310 with BRD4 inhibitor JQ1 not only showed synergistic latency-reversing activity but also decreased the influence on cytokine secretion from CD4 + T cells induced by BL-V8-310 alone . In brief, PKC activators are still attractive LRAs which are particularly effective in combination with other LRAs with different mechanisms. Among the four members (BRD2, BRD3, BRD4, and BRDT) of the bromodomain and extra-terminal (BET) protein family, BRD4 is the most concerned and widely studied subtype. The primary structure of BRD4 comprises two highly conserved N-terminal tandem bromodomains (BD1 and BD2) that are responsible for recognizing the acetylation status of lysine residues on histone tails and other proteins, an ET domain, and a C-terminal domain (CTD). Bromodomain protein contains four antiparallel α-helices (αZ, αA, αB, αC) and two hydrophobic loops (ZA loop and BC loop). ZA, BC, and αZ form an essential N -acetylated lysine-binding hydrophobic pocket (Kac site). Besides, a hydrophobic region ‘WPF shelf’ bounded by Trp/Pro/Phe residues (W81, P82, F83) is also important for enhancing BRD4 binding affinity . Acting as an interaction partner with P-TEFb, BRD4 can competitively bind P-TEFb with Tat protein, an exclusive transcriptional transactivator of HIV-1 viruses, thus leading to the silencing of HIV-1 gene transcription. Hence, inhibition over-expression of BET/BRD4 will promote the recruitment of P-TEFb by Tat and subsequent viral transcription elongation, resulting in dissociation of the BET and P-TEFb complex and consequent provirus reactivation . Chemical structures of several representative BET inhibitors (BETIs), as displayed in , are being currently tested in various latently HIV-1-infected evaluation models. Triazolothienodiazepine derivative -JQ1 ( 65 ) is the first BETI discovered by a high-throughput screening protocol, which possesses many good drug-like properties, such as excellent cellular potency, high selectivity for the BRD4 isoform of the BET family, synthetic accessibility, low off-target possibility and a good pharmacokinetic profile. However, JQ1 was less efficient in reactivating latent HIV-1 viruses when used alone and exerted severe cytotoxicity during prolonged treatment . Since the benzodiazepine fragment of JQ1 is validated to be a key pharmacophore that contributes to a good BRD4 affinity, a spectrum of benzodiazepine-derived BETIs was successively identified, including but not limited to OTX015 ( 66 ), CPI-203 ( 67 ), MS417 (GTPL7512, 68 ) and I-BET762 (GSK525762, 69 ). OTX015 ( 66 ) could effectively reactivate latent HIV-1 in various latency models with 1.95~4.34 times improved efficacy over that of JQ1. More interestingly, OTX015 can also increase CDK9 occupancy at HIV-1 promoter, which in turn phosphorylates RNA polymerase II (Pol II) to enable productive elongation and generate full-length HIV transcripts . CPI-203 ( 67 ) is a novel cell-permeable BRD4 inhibitor with good bioavailability when administered orally. Notably, in contrast to JQ1, CPI-203 showed more potent HIV-1reactivation efficacies in various latently infected cell lines and significantly decreased cytotoxicity. CPI-203 also exerted a synergistic effect and alleviated prostratin-induced ‘cytokine storm’, a severe and systemic production of inflammatory cytokines in response to endotoxemic shock, when combined with PKC activators (e.g., prostratin, ingenol-B and bryostatin-1) in the reactivation of latent HIV-1 . MS417 ( 68 ) and I-BET762 ( 69 ) could reactivate HIV-1 from latency in vitro via a P-TEFb-dependent but Tat-independent mechanism, implying BRD2 together with BRD4 cooperatively participate in HIV latency . BETIs PFI-1 ( 70 ) and RVX-208 ( 71 ), whether used alone or in combination with other types of LRAs, exhibited strong HIV-1 reactivation activities through up-regulation of the phosphorylation of CDK9 Thr-186 to increase the expression of P-TEFb in latently infected Jurkat T cells, thus effective activation of HIV-1 transcription. Besides, these two BETIs could also activate HIV-1 transcription in resting CD4 + T cells in cART-treated patients without causing aberrant activation of global immune cells, strengthening the therapeutic values of BETIs in anti-HIV therapy by effective combating HIV-1 latency . Mivebresib ( 72 ) has already been advanced to phase I clinical trials for treating relapsed/refractory acute myeloid leukemia . I-BET151 ( 73 ) could preferentially reactivate HIV-1 gene expression and efficiently increase HIV-1 transcription in monocytic cells, but not in typically HIV-1-infected CD4 + T cells, in cART-treated humanized mice, via a CDK2-dependent mechanism . 8-Methoxy-6-methylquinolin-4-ol (MMQO, 74 ), was identified from a virtual screening that was able to reactivate latent proviruses in vitro and ex vivo by functioning as a BRD4 inhibitor. In addition, MMQO also could potentiate the effects of PKC activators or HDACIs on HIV-1 reactivation. Owing to its minimalistic structure, MMQO might act as an ideal hit for further chemical modification . UMB136 ( 75 ), an imidazo [1,2- a ]pyrazine-derived BETI, exhibited elevated HIV-1 reversal activity compared to JQ1 in multiple latently infected cell models. Besides, UMB136 was able to synergize with PKC activators (e.g., bryostatin-1 and prostratin), and increase viral production and HIV-1 transcription by releasing P-TEFb . The BRD4 selective inhibitor ZL0580 ( 76 ) was capable of reactivating latent HIV-1 and restraining viral replication in CNS reservoirs, indicating the therapeutic potential in treating neuroinflammation or related neurological disorders in HIV-infected individuals . CPI-637 ( 77 ), a BRD4 and TIP60 dual inhibitor, could efficiently reactivate latent proviruses in vitro. Meanwhile, CPI-637 was also able to alleviate overall T cell activation and prevent viral spread to uninfected CD4 + T cells without an obvious toxic effect. Interestingly, CPI-637-mediated TIP60 inhibition, in turn, stimulated BRD4 dissociation from the HIV-1 5′-LTR promoter, causing more effective binding of Tat protein with P-TEFb in comparison with the BRD4 inhibition alone. These data indicated that development of dual-target LRAs, including but not limited to BRD4/TIP60 inhibitors, might be a more promising remedy to effectively eliminate latent reservoirs . According to the X-ray co-crystallographic structures of BETIs-bound BRD4, the vast majority of the aforementioned BETIs share two key pharmacophoric features: a ‘head moiety’ that bears an H-bond donor and/or acceptor to form hydrogen bonds with two key amino residues (Asn140 and Tyr 97), by mimicking the function of the acetyl O atom of ABBV-075, and appropriate hydrophobic substituents to spatially accommodate the ‘WPF shelf’ and ‘ZA channel’, aiming to enhance potency and selectivity towards BRD4 by forming intensive hydrophobic contacts . To identify more potent BETIs, Burns and co-workers believed an appropriate Kac-mimicking pharmacophoric framework, which is capable of providing interactions with the pivotal Kac domain and meanwhile offers strong H-bonds with the key residues in this region, plays a crucial role in achieving desirable BET-binding affinities. To this end, they employed a scaffold-hopping strategy by replacing 6 H -thieno [3,2- f ][1,2,4]triazolo [4,3- a ][1,4]diazepine core of JQ1 with a 1,2,3-triazolobenzodiazepine framework, which was supposed to be capable of forming more favorable binding interactions with target protein. It indicated that, as anticipated, the resulting 1,2,3-triazolobenzodiazepine skeleton ( 78 ) gave a nearly 3-fold increase in BRD4-binding affinity. The introduction of a 2-aminopyridine fragment on the central benzene ring ( 79 ) resulted in further activity improvement. To figure out the possible reason for activity improvements of the 1,2,3-triazolobenzodiazepine nucleus, binding patterns of JQ1 and 79 complexed with BRD4 (PDB entry: 5 UVV) were compared by a molecular modeling method. As evinced in , although both 76 and reference JQ1 adopt similar spatial orientation by docking well into three key binding domains (Kac, ZA, WPF) of BRD4, 79 formed two H-bonds between the 2-aminopyridine portion and two residues (Met398 and Met425), with distances of 2.3 Å and 2.8 Å, respectively. By contrast, JQ1 formed a relatively weak H bond with Tyr390, with a distance of 3.1 Å. Collectively, compound 79 with the new skeleton forms stronger H-bond contacts, which is supposed to largely account for its elevated activity . I-BET151 ( 80 ) is another representative BETI, in which 1 H -imidazo [4,5- c ]21uinoline-2(3 H )-one core is a key factor in affecting HIV-1 reactivation potency by functioning as a mimic of acetylated lysine. The co-crystal structure of I-BET151 complexed with BRD4 (BD1), as evinced in , revealed that the 3,5-dimethylisoxazole part forms a direct H-bond with Asn140 (3.2 Å) and water-bridged hydrogen-bonding interactions with Tyr97 (2.0 Å) in the acetylated lysine binding pocket (Kac site). The 1H-imidazo [4,5- c ]21uinoline-2(3 H )-one nucleus stretches into the hydrophobic ZA channel of protein, and the N atom on the central pyridine ring forms a water-mediated hydrogen bond interaction with Phe83 (2.7 Å). The pyridine ring interacts with the surface of the hydrophobic WPF shelf, which is formed by W81, P82, M149 and I146 residues of BRD4. Based on this binding information, Wang’s lab employed a ‘scaffold-hopping’ strategy to substitute a [6,6,5]tricyclic 5 H -pyrido [4,3- b ]indole (γ-carboline) motif which has a balanced physiochemical profile for the 1 H -imidazo [4,5- c ]21uinoline-2(3 H )-one core of I-BET151, with the expectation of achieving similar BRD4-binding affinity. Among the newly obtained derivatives, compound 81 containing a 3-cyclopropyl-5-methyl-1 H -pyrazole fragment gave the best potency, particular an improved BRD4(2)-binding affinity compared with the reference JQ1, while the replacement of this moiety with a smaller Cl atom ( 82 ) impaired potency . To gain more insight into the activity differences, binding modes were conducted and compared between I-BET151 and compound 81 by using the molecular docking method. As evinced in , although both 81 and I-BET151 adopt very similar spatial orientation and both bind with two key residues (Gln85 and Asn140), 81 offered stronger H-bonding interactions with these two amino acids than I-BET151, which might explain their binding differences to a great extent. In summary, although the exact molecular mechanism of BETIs in reactivating latent proviruses is still not fully understood, the successful development of various BETIs highlights the importance of BET/acetyl-lysine inhibition in regulating HIV-1 transcription and latent reactivation. Nevertheless, so far, none of the BETIs have entered clinical phases as LRAs, primarily attributable to the in vivo inefficiency with no effect on eliminating both latent reservoirs and HIV-1 viremia, although the in vitro latency-reversing activities were satisfactory. This fact leaves the clinical implementations of BETIs as LRAs with a long way to go still. P-TEFb, which is composed of cyclin T1 (CycT1) and cyclin-dependent kinase 9 (CDK9), is a CDK9-cyclin T1 heterodimer and is an important part of the super elongation complex (SEC) used by the viral-encoded Tat protein to activate HIV transcription. In this sense, P-TEFb is an indispensable cellular cofactor for Tat protein and an essential elongation factor to realize efficient viral transcriptional elongation and expression. In this regard, P-TEFb activators were able to work as LRAs to reactivate latent viruses . Ye et al. isolated at least five kinds of short-chain fatty acids (SCFAs), namely, butyric acid, isobutyric acid, isovaleric acid, propionic acid and acetic acid, from the saliva and plasma of HIV-infected patients with severe periodontitis undergoing HAART treatment. The subsequent study revealed that, except for acetic acid, the other four SCFAs have the capability to potently reactivate latent HIV-1 in both latent Jurkat cells and resting CD4 + T-cells by activating P-TEFb as well as inducing histone modifications in a dose-dependent manner. Among these five SCFAs, butyric acid presented the strongest activity, while isobutyric acid gave the weakest effect, and the combination of any two SCFAs exhibited an additive effect . PR-957 ( 83 ), also named as ONX-0914, as displayed in , might be a promising LRA candidate with a host of favorable properties. It was found to strongly activate latent viruses in vitro and in vivo by activating P-TEFb without causing overall abnormal immune activation of T cells. It also exhibited relatively low cytotoxicity at a tolerable level and could reduce the expression of HIV receptor CD4 and co-receptors CXCR4 and CCR5. Moreover, it manifested a synergistic effect when combined with known LRA, PKC activator prostratin, and could reduce prostratin-induced T-cell activation. Most notably, it displayed no interference with other anti-viral drugs, thus guaranteeing the efficient implementation of the “shock and kill” strategy and effective elimination of newly produced viruses . Hexamethylene bisacetamide (HMBA, 84 ), which has been dismissed as anticancer candidate in clinical anti-leukemic evaluation, has recently revived interest in anti-HIV therapy due to its remarkable capability in reactivating latent HIV-1 proviruses, whether used alone or in combination with PKC activator prostratin in chronically infected CD4 + T cells . The reactivation occurs at a transcription level and is independent of NF-κB but transiently activates Akt via the inhibition of phosphatidylinositol-3-kinase (PI3K) and relies on Sp1-binding sites in the HIV promoter to recruit P-TEFb to promote the reactivation of viral production from latency . Chalcone analogue Amt-87 ( 85 ) has been proved to significantly reactivate latent HIV-1 provirses and synergistically work with known LRAs, such as PKC activator prostratin and BETI JQ1 via activation of P-TEFb, indicating its development perspective as a potent LRA . Acting as a highly conserved Ser/Thr kinase in eukaryotes, PLK1 is proved to be a promising target in identifying potent LRAs. PLK1 inhibitors are expected to have doubled function by not only effectively reactivating latent proviruses but also promoting the death of reservoir cells, based on the fact that PLK1 plays an essential role in HIV-1 Nef-mediated survival of CD4 + T cells . Besides, PLK1 also has a tight correlation with BRD4. On one hand, both BRD4 and PLK1 are able to interact directly with P-TEFb to regulate HIV-1 LTR transcription; on the other hand, BRD4 can also be phosphorylated by PLK1 , which perhaps explains why many PLK1 inhibitors, such as compounds 5-methyl-7,8-dihydropteridin-6(5 H )-one derivatives BI-2536 ( 86 ) and BI-6727 (volasertib, 87 ), as diagrammed in , also function as BRD4 inhibitors. Acting as dual PLK1/BET inhibitors, pteridine-derived compounds 86 and 87 could significantly reactivate silenced HIV-1 proviruses in two latently infected cell lines, ACH2 and U1, at both the mRNA and protein level. Although these two compounds demonstrated more potent PLK inhibitory activities, the latent reactivating effects were directly associated with BET inhibition, further confirming the important role of BETIs in developing promising LRAs. Additionally, BI-2536 could synergistically reactivate HIV-1 proviruses when combined with HDACI SAHA or PKC activator prostratin, in PBMCs from HIV-1-infected patients, verifying the prospective therapeutic value in anti-viral treatment . Maraviroc (MVC, 88 ), as shown in , a selective CCR5 (C-C Motif Chemokine Receptor 5) antagonist with broad-spectrum antiretroviral efficacy that has been used in treatment of HIV-1 infection, was recently validated to be an attractive LRA by inducing NF-κB activation as a result of specific binding of CCR5, whether used alone or in combination with PKC activator bryostatin-1 . However, despite the favorable safety profile in humans as well as positive latent reactivation outcomes in phase II trial, maraviroc could not efficiently reduce the size of latent reservoirs, given the fact that viruses rebounded quickly after antiretroviral therapy was discontinued . All these evidence strongly support that CCR5 antagonists might be successfully developed into effective anti-HIV drugs, which not only function as antiretroviral agents to suppress HIV-1 replication but also act as LRAs to reactivate latent proviruses . Xevinapant (Debio 1143, AT-406, 89 ), as displayed in , an effective chemo-radio-sensitizer as a first-in-class oral IAP antagonist, currently being assessed in phase II clinical trial for the treatment of squamous cell carcinoma of the head and neck (SCCHN), was recently proved to reverse HIV-1 latency in various latently infected T cell lines via the induction of a non-canonical NF-κB pathway, resulting in the enhancement of HIV-1 transcription . Besides, an obvious synergistic latency-reversal activity was achieved when combined use of IAP antagonist AZD5582 ( 90 ) and BETI I-BET151, rather than largely due to these two LRAs both acting on host cell transcriptome . The favourable latency-reversal efficacies together with the manageable safety profiles may enable IAP antagonists to be developed as promising LRAs. Zhu et al. found that the PEBP1 gene, also named as RKIP (Raf kinase inhibitor protein) is tightly correlated with the establishment of a latent reservoir as well as the suppression of viral replication by employing a CRISPR-based genetic screen technique. The subsequent depletion of the PEBP1 gene significantly reactivates latent proviruses by regulating the NF-κB pathway, while the up-regulation of PEBP1 expression with epigallocatechin-3-gallate (EGCG, 18 ) inhibits latency reversal by preventing nuclear translocation of NF-κB. These results provided directly proofs that PEBP1 inhibitors might be a type of promising LRAs in effective controlling HIV-1 reservoirs . Proteasome inhibitor (PI) thiostrepton (TSR, 91), as demonstrated in , a naturally derived thiazole-containing oligopeptide antibiotic, was recently found to effectively reactivate latent HIV-1 proviruses in vitro and ex vivo with low cytotoxicity. TSR also displayed a significant synergistic effect with prostratin, bryostatin-1 or JQ1 without affecting overall T cell activation or inducing dysfunction of CD8 + T cell, indicating the prospective application as an effective LRA candidate. Mechanistic study revealed TSR activated P-TEFb and NF-κB pathways by up-regulating heat shock proteins (HSPs), resulting in viral reactivation . By using the CRISPR interference-based screening technique, Zhou et al. proved two PIs, bortezomib (PS-341, 92 ) and carfilzomib ( 93 ), could effectively reactivate latent HIV-1 by strongly synergizing with different types of LRAs without inducing CD4 + T cell activation or proliferation. Mechanistic investigation showed that a proteasome inhibitor could promote Tat-transactivation by up-regulating the expressions of ELL2 and ELL2-containing super elongation complexes (ELL2-SECs), leading to reactivation of latent proviruses . Mechanistic results from Liu and co-workers demonstrated that PIs not only could reactivate latent HIV-1 by HSF1-mediated recruitment of complex of HSP90 and P-TEFb, but also may facilitate the human body to the virus by enhancing host innate immune responses . Evidence from Dougherty et al. suggested that PIs, exemplified by bortezomib, clasto-lactacystin β-lactone (CLBL, 94 ) and MG-132 ( 95 ), work as bifunctional antagonists of HIV-1, exerting anti-latency by reactivating latent HIV-1 and anti-replication by inhibiting HIV-1 infectivity . All these data revealed that proteasome can function as a promising and targetable target in developing effective anti-HIV-1 agents . In addition to proteasome, HSF1 and HSP90, which participate in HIV transcription elongation, can also serve as attractive therapeutic targets in developing potent LRAs. Particularly, HSF1 that is tightly associated with HIV-1 5′-LTR might be a targetable target for identifying promising LRAs based on the fact that the majority of LRAs with various mechanism of actions might reactivate latent HIV-1 via an HSF1-involved pathway rather than the previously known pathways . Toll-like receptors (TLRs) are transmembrane pathogen-recognition receptors that function as the sentinels of host defense by recognizing a spectrum of pathogen-related molecular components in bacteria, viruses, protozoa and fungi as well as molecules released during cell damage or death. The TLR family contains more than 13 members, of which only 10 (TLR1-TLR10) have been identified in humans . Since TLR members are mainly existing and functional in CD4 + T cells, where the majority of the latent reservoir is harboured, TLR agonists are thereby believed to hold great promise to successfully reactivate latent proviruses by stimulating downstream pathways , such as AP-1, NFAT, IRFs as well as NF-κB, a well-acknowledged regulator of HIV transcription that has a tight correlation with HIV-1 latency . Moreover, TLRs also have high hopes of eliminating activated latent cells by exerting immunologic cytotoxicity, promoting antiviral responses and non-specifically activating T lymphocytes. All these properties render immunostimulatory TLR agonists unique and promising LRAs among the identified LRAs to date, also known as next-generation LRAs . TLR signalling, in particular TL-7/8/9 and TL-1/2 activation, has shown to be notable LRA with multifactorial characteristics. For instance, resiquimod (R848, 96 ), as exhibited in , a imidazoquinoline-based TLR7/8 agonist, showed latency-reversing activity in latently infected cell lines U1 and OM10 by inducing p24 expression . Moreover, resiquimod was capable of diminishing the size of latent reservoirs in HIV-1 infected patients and meanwhile preventing virus replication, suggesting potential clinical value . Vesatolimod (GS-9620, 97 ), a dihydropteridinone-derived TLR7 agonist (EC 50 = 291 nM) could effectively reactivate latent proviruses in PBMCs from HIV-1-infected individuals and also improve immune effector functions at a safe and tolerable clinical dose . Rochat et al. supposed the combination of transcriptional enhancers with a TLR agonist would be more efficient in reversing latency at various levels by generating a Th1 supportive milieu and meanwhile triggering the innate immune system. It was revealed that the combination of thiazoloquinolone-based TLR8 agonist CL075 (3M002, 98 ) with PKC agonist prostratin resulted in significantly augmented latency-reversing activity compared with any single compound, whether in primary cells of HIV-1-infected individuals or on a coculture of J-lat and MDDCs (monocyte-derived dendritic cells). These data implied that the combined use of two LRAs, by which one triggers directly the transcriptional pathway and the other stimulates indirectly the immune system, would be a practicable remedy for efficiently reversing HIV-1 latency . SMU-Z1 ( 99 ), an imidazole-based TLR-1/2 agonist, exhibited excellent latency-reversing potency in HIV-1-infected cells via activation of NF-κB and MAPK pathways in vitro and in PBMCs from HIV-1-infected individuals ex vivo. Besides, SMU-Z1 was also capable of promoting degranulation and gamma interferon (IFN-γ) production in NK cells, as well as increasing TNF-α production in PBMCs without triggering overall T cell activation, indicating the prospect application in eradicating HIV-1 . Considering that some TLR signaling, particularly TLR3, are distributed in microglia cells, which constitute the major HIV-1 reservoir in the brain, Karn and co-workers screened a panel of TLR agonists on different HIV-1 infected cell lines. The results revealed that TLR3 agonists, as anticipated, could efficiently reactivate viral transcription in HIV-1-infected microglia cells, rather than monocytes or T cells, via a previously unreported IRF3, but not the most common NF-κB-mediated mechanism, providing evidence for the therapeutic potential of TLR3 agonists in the treatment of HIV-1 infection in the central nervous system . Despite many beneficial profiles, the precise mechanisms by which the TLR agonists reactivate latent proviruses and modulate immune responses, however, remain incompletely understood. Therefore, more efforts should be made to figure out the detailed mechanisms and identify more robust immunostimulatory TLR agonists against HIV-1 . 2.13.1. AV6 and Analogues Quinoline-derivative antiviral 6 (AV6, 100 ), as displayed in , was identified from a cell-based high-throughput screening protocol, and could reactivate latent HIV-1 in different cell-based latency models without inducing global T-cell activation or proliferation, and could also act synergistically with a HDACI valproic acid (VA) to achieve an additive HIV-1 reactivation . Considering the definite efficacies of HDACI (especially class I HDACI) and AV6, used either alone or in combination, in reactivating latent proviruses, Fang and co-workers expected to develop novel HDACI-derived LRA chemotypes to gain enhanced induction of HIV-1 gene expression. To this end, they investigated various linker lengths and ZBGs, while leaving the N -phenylquinoline skeleton of AV6 unchanged in order to act as a cap region. Among the obtained AV6 derivatives, compounds 101 and 102 exhibited the most potent in vitro latency-reversing activities in J-Lat A2 cells. A preliminary mechanistic study confirmed these two compounds function as dual HIV-1 reactivators by inhibiting HDAC2 isoform and meanwhile facilitating the release of P-TEFb, and 94-dependent HIV-1 gene expression could be obstructed by an immunosuppressant tacrolimus (FK506). However, the anticipated synergetic effect did not occur in combined use of 102 and AV6 . 2.13.2. 2-Acylaminothiazole 2-Acylaminothiazole ( 103 a), as shown in , was able to reactivate HIV-1 gene expression with modest activity (EC 50 = 23 μM) in a high-throughput screening protocol using a dual luciferase reporter cellular assay, which thereby became an ideal hit to develop more chemically diversified LRAs. Among a selection of 2-acylaminothiazole derivatives, 103 b–c, 103 a in particular, gave comparable to increased activity compared to controls, vorinostat and JQ1, in latent CD4 + T cells isolated from cART-treated patients, implying that the introduction of electron-withdrawing groups (e.g., cyano, trifluoromethyl) is favorable to the elevation of HIV-1 reactivation efficacy. Furthermore, synergy could be observed between 103 c and BETI JQ1 in various latent HIV-1 cellular models, indicating the prospects as effective LRAs for treating HIV-positive individuals. However, the exact cellular target(s) of these derivatives were still unclear . 2.13.3. Dihydropyranoindole Derivative Pyranoindole derivative GIBH-LRA002 ( 104 ), as exhibited in , was discovered from a high-throughput screening protocol in the J-Lat cell model, which has a decent HIV-1/SIV reactivation efficacy in resting CD4 + T cells from both chronic SIV-infected rhesus macaques and HIV-1 infected patients. Meanwhile, it possesses relatively low cytotoxicity without causing overall T cell activation, thus avoiding new viral infections, indicating that GIBH-LRA002 might be an attractive candidate for anti-HIV treatment . 2.13.4. Carbazole Derivative Curaxins CBL0137 ( 105 ), as manifested in , has been described as a potent FACT (facilitates chromatin transcription) inhibitor, which exerts anticancer efficacy by suppressing the NF-κB pathway and activating tumor suppressor gene p53 . As previous work has proved the NF-κB pathway plays an important role in disrupting latent HIV-1, while the depletion of FACT also could reactivate HIV-1 proviruses, CBL0137 was thereby submitted to the anti-latency evaluation in post-integrated latency cells, JLAT6.3 and CA5. The result evinced that CBL0137 alone could not only sufficiently potentiate TNF-α stimulated HIV-1 transcription without inducing any cytotoxicity, but also reactivate latent reservoirs in PBMCs from HIV-1-infected patients, suggesting the potential clinical application as an appealing LRA candidate . 2.13.5. Benzazole Derivative Benzazole derivative 106 a (see ) was found to have well-defined HIV-1 reactivation activity and a favorable pharmacokinetic profile in a high-screening protocol in 24STNLSG cells, which thereby was a good hit for further chemical optimization. Among the synthesized benzazole derivatives, 106 b gave the best HIV-1 reactivation effectiveness and the lowest cytotoxicity, which could induce proviral transcription in several latently infected cell types without affecting overall T cell activation. However, the 4-NH2 containing benzazole derivative 106 c provided decreased HIV-1 latency potency in a latently infected ACH-2 cell line. Preliminary SAR outcomes have been summarized in . The preliminary mechanistic study revealed that these benzazole derivatives appear to have a unique mechanism of action. They were not HDACIs and the HIV-1 transcription was driven neither by the NF-κB signaling pathway nor by stimulating of HIV-1 LTR. These data indicated that benzazole might be a promising chemotype to be developed into potent LRAs with appropriate chemical optimizations and further investigation into the mechanism of actions . 2.13.6. Pyridine Derivative Vitamin B3 niacinamide ( 107 ) (see ), a pyridine-derived sirtuin inhibitor, showed notably improved HIV-1 reactivation potency compared to that of the combination of two validated HMTI-based LRAs, chaetocin and BIX01294, in an ex vivo assay for short-term treatment, indicating its clinical potential. Meanwhile, owing to the relatively simple structure, niacinamide might be utilized as a suitable hit to carry out further chemical optimization to find more potent LRAs . 2.13.7. Polyphenols Resveratrol ( 108 ), as revealed in , a natural polyphenol, was capable of reactivating latent HIV-1 without triggering overall T cell activation. Besides, a synergistic HIV reactivation was observed when resveratrol was combined with other conventional LRAs with different mechanisms of action, such as JQ1 (BETI), SAHA (HDACI) and prostratin (PKC activator). A preliminary mechanistic study revealed the latency-reversal activity of resveratrol was due to the activation of HSF1 and increased histone acetylation, but not the activation of silent information regulator 1 (SIRT1), which belongs to a member of the sirtuin family. However, due to relatively low bioavailability, more polyphenol analogues of resveratrol ( 109 ~ 114 ) were successively submitted to the HIV-1 reactivation evaluation. The result was that among these polyphenols, only triacetyl resveratrol ( 109 ) gave comparable latency-reversal potency with the prototype resveratrol, while none of the others could successfully reactivate the latent proviruses in vitro, suggesting more effort is still required to find more potent polyphenol-based LRAs by expanding the chemical diversity of a synthetically or naturally available stilbenoid chemotype. Additionally, the cotreatment with triacetyl resveratrol ( 109 ) with other LRAs also generated a synergistic effect . Q 205 ( 115 ), a synthetic resveratrol analogue, was proved to effectively reactivate latent HIV-1 in vitro without inducing of damaging cytokines. A preliminary mechanistic study revealed that the latency-reversal potency of Q205 was attributable to the activation of P-TEFb and promotion of Tat-mediated HIB-1 transcription and binding of RNAPII to the HIV-1 5′-LTR promoter. The results evinced that identification of promising LRAs from naturally available polyphenols and/or chemically optimized resveratrol derivatives might be a feasible and practicable alternative . Quinoline-derivative antiviral 6 (AV6, 100 ), as displayed in , was identified from a cell-based high-throughput screening protocol, and could reactivate latent HIV-1 in different cell-based latency models without inducing global T-cell activation or proliferation, and could also act synergistically with a HDACI valproic acid (VA) to achieve an additive HIV-1 reactivation . Considering the definite efficacies of HDACI (especially class I HDACI) and AV6, used either alone or in combination, in reactivating latent proviruses, Fang and co-workers expected to develop novel HDACI-derived LRA chemotypes to gain enhanced induction of HIV-1 gene expression. To this end, they investigated various linker lengths and ZBGs, while leaving the N -phenylquinoline skeleton of AV6 unchanged in order to act as a cap region. Among the obtained AV6 derivatives, compounds 101 and 102 exhibited the most potent in vitro latency-reversing activities in J-Lat A2 cells. A preliminary mechanistic study confirmed these two compounds function as dual HIV-1 reactivators by inhibiting HDAC2 isoform and meanwhile facilitating the release of P-TEFb, and 94-dependent HIV-1 gene expression could be obstructed by an immunosuppressant tacrolimus (FK506). However, the anticipated synergetic effect did not occur in combined use of 102 and AV6 . 2-Acylaminothiazole ( 103 a), as shown in , was able to reactivate HIV-1 gene expression with modest activity (EC 50 = 23 μM) in a high-throughput screening protocol using a dual luciferase reporter cellular assay, which thereby became an ideal hit to develop more chemically diversified LRAs. Among a selection of 2-acylaminothiazole derivatives, 103 b–c, 103 a in particular, gave comparable to increased activity compared to controls, vorinostat and JQ1, in latent CD4 + T cells isolated from cART-treated patients, implying that the introduction of electron-withdrawing groups (e.g., cyano, trifluoromethyl) is favorable to the elevation of HIV-1 reactivation efficacy. Furthermore, synergy could be observed between 103 c and BETI JQ1 in various latent HIV-1 cellular models, indicating the prospects as effective LRAs for treating HIV-positive individuals. However, the exact cellular target(s) of these derivatives were still unclear . Pyranoindole derivative GIBH-LRA002 ( 104 ), as exhibited in , was discovered from a high-throughput screening protocol in the J-Lat cell model, which has a decent HIV-1/SIV reactivation efficacy in resting CD4 + T cells from both chronic SIV-infected rhesus macaques and HIV-1 infected patients. Meanwhile, it possesses relatively low cytotoxicity without causing overall T cell activation, thus avoiding new viral infections, indicating that GIBH-LRA002 might be an attractive candidate for anti-HIV treatment . Curaxins CBL0137 ( 105 ), as manifested in , has been described as a potent FACT (facilitates chromatin transcription) inhibitor, which exerts anticancer efficacy by suppressing the NF-κB pathway and activating tumor suppressor gene p53 . As previous work has proved the NF-κB pathway plays an important role in disrupting latent HIV-1, while the depletion of FACT also could reactivate HIV-1 proviruses, CBL0137 was thereby submitted to the anti-latency evaluation in post-integrated latency cells, JLAT6.3 and CA5. The result evinced that CBL0137 alone could not only sufficiently potentiate TNF-α stimulated HIV-1 transcription without inducing any cytotoxicity, but also reactivate latent reservoirs in PBMCs from HIV-1-infected patients, suggesting the potential clinical application as an appealing LRA candidate . Benzazole derivative 106 a (see ) was found to have well-defined HIV-1 reactivation activity and a favorable pharmacokinetic profile in a high-screening protocol in 24STNLSG cells, which thereby was a good hit for further chemical optimization. Among the synthesized benzazole derivatives, 106 b gave the best HIV-1 reactivation effectiveness and the lowest cytotoxicity, which could induce proviral transcription in several latently infected cell types without affecting overall T cell activation. However, the 4-NH2 containing benzazole derivative 106 c provided decreased HIV-1 latency potency in a latently infected ACH-2 cell line. Preliminary SAR outcomes have been summarized in . The preliminary mechanistic study revealed that these benzazole derivatives appear to have a unique mechanism of action. They were not HDACIs and the HIV-1 transcription was driven neither by the NF-κB signaling pathway nor by stimulating of HIV-1 LTR. These data indicated that benzazole might be a promising chemotype to be developed into potent LRAs with appropriate chemical optimizations and further investigation into the mechanism of actions . Vitamin B3 niacinamide ( 107 ) (see ), a pyridine-derived sirtuin inhibitor, showed notably improved HIV-1 reactivation potency compared to that of the combination of two validated HMTI-based LRAs, chaetocin and BIX01294, in an ex vivo assay for short-term treatment, indicating its clinical potential. Meanwhile, owing to the relatively simple structure, niacinamide might be utilized as a suitable hit to carry out further chemical optimization to find more potent LRAs . Resveratrol ( 108 ), as revealed in , a natural polyphenol, was capable of reactivating latent HIV-1 without triggering overall T cell activation. Besides, a synergistic HIV reactivation was observed when resveratrol was combined with other conventional LRAs with different mechanisms of action, such as JQ1 (BETI), SAHA (HDACI) and prostratin (PKC activator). A preliminary mechanistic study revealed the latency-reversal activity of resveratrol was due to the activation of HSF1 and increased histone acetylation, but not the activation of silent information regulator 1 (SIRT1), which belongs to a member of the sirtuin family. However, due to relatively low bioavailability, more polyphenol analogues of resveratrol ( 109 ~ 114 ) were successively submitted to the HIV-1 reactivation evaluation. The result was that among these polyphenols, only triacetyl resveratrol ( 109 ) gave comparable latency-reversal potency with the prototype resveratrol, while none of the others could successfully reactivate the latent proviruses in vitro, suggesting more effort is still required to find more potent polyphenol-based LRAs by expanding the chemical diversity of a synthetically or naturally available stilbenoid chemotype. Additionally, the cotreatment with triacetyl resveratrol ( 109 ) with other LRAs also generated a synergistic effect . Q 205 ( 115 ), a synthetic resveratrol analogue, was proved to effectively reactivate latent HIV-1 in vitro without inducing of damaging cytokines. A preliminary mechanistic study revealed that the latency-reversal potency of Q205 was attributable to the activation of P-TEFb and promotion of Tat-mediated HIB-1 transcription and binding of RNAPII to the HIV-1 5′-LTR promoter. The results evinced that identification of promising LRAs from naturally available polyphenols and/or chemically optimized resveratrol derivatives might be a feasible and practicable alternative . Since multiple regulatory pathways have participated in the establishment and maintenance of latent reservoirs, a single LRA is obviously not sufficient to accomplish the global viral reactivation. Thus, the combined use of LRAs with different mechanisms might be a more effective means, by affecting various subsets of signaling regulatory pathways, lowering the dose of each component and reducing the unwanted side effects. Given the fact that the combination of PKC activators and HDACIs has proved to be more effective than any LRAs, the PKC activator/HDACI combinational protocols were by far the most reported. Doria M. and co-workers proved that combination of the PKC activator prostratin with HDACIs would attenuate HDACI toxicity, while the best result was obtained when prostratin was combined with HIDAI romidepsin, by not only stimulating reactivation of latent HIV-1 but also enhancing NKG2D-mediated viral suppression by NK cells . Largazole (SDL 148, 116 ), as evinced in , a macrocyclic class I-selective HDACI, was validated to be a potent LRA with low toxicity by remodelling chromatin at the HIV-1 5′-LTR promoter. Given the fact that the combination of PKC activators and HDACIs generally gave more effective potency than any one LRA component, the effects of the combination of HDACI largazole and two PKC modulators, bryostatin-1 analogues SUW133 ( 117 ) and SUW124 ( 118 ), were determined. As expected, combination of largazole with either of the bryostatin-1 analogues provided remarkable latency-reversal efficacy and induced enhanced levels of proviral expression without triggering overall T cell activation and abnormal cytokine release, indicating that the combination is a potential therapeutic remedy for preclinical advancement . Similarly, the co-administration of the PKC activator gnidimacrin (GM, 63 ) with class I-selective HDACI thiophenyl benzamide (TPB, 119 ), not only gave 2~3-fold HIV-1 reactivation efficacy but decreased the risk of new viral infection, compared to GM alone at non-toxic concentrations . Additionally, the combination of PKC modulator (−)-indolactam V ( 120 ) with a pan-HDACI, hydroxamate-tethered phenylbutyrate AR-42 ( 121 ), resulted in a strong synergistic effect in reversing latent viruses, followed by a significant CD4 down-regulation . In addition to the well-acknowledged PCK activator/HDACI combinations, other combinational LRAs were also reported. For instance, gliotoxin (GTX, 122 ), discovered from a high-throughput screen of fungal metabolites, was validated to be an attractive LRA without clear cytotoxicity by disrupting 7SK small nuclear ribonucleoprotein (7SK snRNP) to release P-TEFb. Moreover, cotreatments with GTX with HDACI SAHA or BRM-associated factors (BAF) inhibitor caffeic acid phenetyl esther ( 123 ) resulted in an unparalleled synergy in reversing HIV-1 latency . Admittedly, combination remedy is not limited to the above-mentioned components, and more and more promising combinations will be identified as the LRA-related research intensifies. Despite promising results of various LRAs with different mechanisms in vitro and ex vivo, in vivo and clinical studies revealed that latent reservoirs cannot be successfully cleared. None of the LRAs discovered to date are capable of reactivating all latent HIV viruses in infected host cells while causing minimal side effects, indicating that HIV-1 reactivation induced by using just LRAs, even with a combination of LRAs with the expectation of targeting different reservoirs, is apparently insufficient to accomplish the “shock” element of the “shock-and-kill” regimen. With further in-depth research on the intrinsic survival and maintenance mechanisms of latent reservoirs, a new “block-and-lock” strategy is proposed alternatively, aiming to use latency-promoting agents (LPAs) to prevent low-level or sporadic transcription of integrated proviruses to realize permanent silencing of viruses and an ultimate functional cure of AIDS. Unlike the “shock-and-kill” strategy that requires reactivation of latent proviruses or elimination of latent infected cells, the “block-and-lock” strategy locks viruses in a latent state permanently; thus, they can never be transcribed even for AIDS patients who have stopped routine antiretroviral treatment. The strategy shows great prospect by exhibiting fewer side effects and less impact on the quality of life of patients, making it possible to accelerate a functional cure of AIDS . Currently, “block and lock” research is mainly focused on two aspects. One is to find new targets for short hairpin RNAs (shRNA) by screening suitable small interfering RNAs (siRNAs) that target the NF-κB motif located in 5′-LTR of virus, then to find candidates with stronger gene silencing abilities . Although siRNA-based drugs are promising, achieving life-long silence to a highly mutated virus that can infect multiple cell types remains a tremendous challenge. Firstly, the effectiveness of various siRNA delivery systems needs to be further confirmed in animal models, and the delivery efficiency also needs to be improved. Secondly, explosive replication of resistant mutant strains should be avoided during application of siRNA. Finally, siRNA-based drugs are expensive, and meanwhile it is infeasible to employ intravenous administration every few days, making it difficult to maintain an effective concentration of siRNA in HIV-1 infected patients for a few years. Therefore, the siRNA-tethered therapy is difficult to popularize extensively, especially in developing and underdeveloped countries. Additional technical breakthroughs, including but not limited to increasing chemical stability, improving pharmacokinetics and delivery efficiency, are still needed before it is widely used in clinical practice. However, for individuals who have been resistant to available antiviral drugs, siRNA-based agents might be optimal alternatives . As described above, there are still numerous difficulties to be dealt with when artificial shRNAs are introduced into HIV-1 infected cells through a suitable delivery system. Therefore, the other attempt to accomplish the “block and lock” strategy is using small molecular LPAs. For example, Zhu et al. found that levosimendan ( 124 , see ), a vasodilator and calcium sensitizer used for heart failure, is a promising LPA by directly screening FDA-approved compound libraries. Levosimendan could inhibit acute viral replication in initial CD4 + T cells and prevent reactivation of latent HIV-1 proviruses . Given that levosimendan has been approved for treating acute heart failure, it is more feasible to be used in clinical promotions as compared to other LPAs. Nonetheless, compared to the well-validated “shock-and-kill” strategy, there is still a long ways to go for the alternative “block and lock” protocol, since there is evidence that the “block and lock” strategy may worsen the immune system in immunodeficient patients and that siRNA-mediated transcriptional silencing can be restrained by Tat protein . The “Shock and kill” strategy will still be the mainstream in the future research on latent reservoirs. However, there are still many difficulties in the development of potent LRAs to postpone the “shock” component. Firstly, there are arguments about the survival and maintenance mechanisms of latent reservoirs, and currently there is still no uniform approach for evaluating the size of latent reservoirs. Hence, deciphering these mechanisms is pivotal in designing more efficient approaches to eliminate latent HIV-1 infection. Secondly, the lack of a suitable in vivo evaluation model is another bottleneck that hinders the quick identification of more effective LRAs for clinical advancement, since most of in vivo latently infected models used to evaluate novel compounds are actually SIV-infected macaque models treated with HAART. Although the SIV virus has a high homology with human HIV-1, and symptoms of SIV-infected macaques are very similar to those of HIV-1 infected humans, there are still many differences in genome sequence between the two viruses. Thirdly, the HIV-1 provirus lurks in different cells which have distinct phenotypes and metabolic characteristics. Affected by a client’s medication history, the latency mechanism may vary from different patients, or vary in different cells from the same patient. Simultaneously, the number and distribution of HIV-1 proviruses in different latently infected cells may change over time; so does the molecular mechanism of latency, enabling latency to be a dynamic process; for that reason, a particular LRA can only be effective for a specific period of time. Finally but not the least, although a variety of novel LRAs have favorable in vitro activities by increasing transcription of viruses in varied latent cell models, there is minimal in vivo HIV-1 reactivation effectiveness in most cases. Therefore, eradicating latent reservoirs in AIDS patients remains a tremendous challenge. To achieve a functional cure of AIDS, future work should consider measures for more specifically modifying viral transcription to identify more effective LRAs by targeting accurately targets, or by using a reasonable LRA combination remedy. |
Molecular Pathology | d8e38f9d-b110-43e3-a656-c3b0249c9478 | 10804512 | Pathology[mh] | |
Joint Statement of the Korean Society for Preventive Medicine and the Korean Society of Epidemiology on the response to the COVID-19 outbreak | e2d85f57-6185-42fa-a367-002d4538c0d5 | 7285423 | Preventive Medicine[mh] | |
Lectin histochemistry in the small intestines of piglets naturally infected with porcine epidemic diarrhea virus | 8956ba9e-fa56-4578-9f03-d8ee2767a4dc | 11450395 | Anatomy[mh] | Porcine epidemic diarrhea virus (PEDV) is a highly pathogenic virus that belongs to the alphacoronavirus genus in the family Coronaviridae. It is the primary cause of severe diarrhea, dehydration, and high death rates in newborn piglets, leading to significant financial losses annually . A mutated variant of PEDV with increased virulence has resulted in severe porcine epidemic diarrhea (PED) outbreaks worldwide since the winter of 2010 . This strain has the potential to cause 100% mortality in piglets that are 7 days old or younger . The villous epithelia of the small intestines are the most common sites of infection for PEDV, which causes blunting of the afflicted villi and disruption of mucosal barrier integrity in vivo . PEDV specifically targets porcine enterocytes, disrupting many cellular activities, such as apoptosis, and intercellular tight junctions . Lectins are classified into numerous families based on their distinct cellular locations and their specificities for various carbohydrate structures, which are determined by the characteristics of their carbohydrate recognition domain modules . Several studies have highlighted the significance of lectins in facilitating interactions between hosts and viruses through the direct recognition of specific oligosaccharides on viral components . Various categories of lectins are capable of either identifying carbohydrates to stimulate the immune system of the host and facilitate the elimination of the virus, or they can utilize those carbohydrates as vulnerabilities to aid in viral entry, replication, or assembly . Glycoconjugates are essential for several functions of viral proteins, such as their folding, stability, and protection. In addition, they are responsible for identifying the precise cellular tropism of the virus . The viral envelope is covered with specific proteins that aid in the virus in binding to host cells, allowing the virus to enter. Prior studies have emphasized that the sugar chains present in glycolipids and glycoproteins in host cells in the intestines have a direct impact on infectious pathogens during the early stages of infection . However, there is a lack of published data that specifically investigate differential changes in glycoconjugates in the intestinal mucosa during PEDV infection. This is surprising given the evidence suggesting significant structural changes in intestinal integrity components as well as disruptions in adherens and tight junctions in the villous epithelium of primary infection sites such as the jejunum and ileum . Several investigations have examined the presence of glycoconjugates and the attachment of bacteria in the small intestines of domestic animals during infections caused by coccidia, including Eimeria tenella , Echinostoma caproni , and Isospora suis . Nevertheless, the influence of viral infections, namely PEDV, on the expression patterns of glycoconjugates in the small intestines has not been examined to date. The objecive of the present study was to assess the alterations in glycoconjugates in the small intestines of piglets naturally infected with PEDV using lectin histochemistry. Animals and tissue preparation Six 7-day-old crossbred Large White × Landrace male piglets were procured from pig farms in Chungcheongnam-do, Republic of Korea. Among them, three piglets that displayed gastroenteritis symptoms were identified as having being infected with PEDV through polymerase chain reaction (PCR) analysis (data not shown). Furthermore, an examination was performed to identify the existence of pathogenic microorganisms in the feces (data not shown). Three healthy piglets from a different herd that had no symptoms of diarrhea served as the control group. These control piglets were confirmed to be PEDV-negative through PCR analysis. After performing a post-mortem examination, parts of the small intestines including the duodenum, jejunum, and ileum were removed and then preserved in 10% buffered formaldehyde solution. The experimental and animal handling methods were performed in compliance with the rules set by the Institutional Animal Care and Use Committee of Chonnam National University (CNU IACUC-YB-2024-70). Histological examination The small intestines including the duodenum, jejunum, and ileum were collected separately. The specimens were immersed in 10% buffered formaldehyde solution and then embedded in paraffin and cut into sections that were 4 μm thick. These steps followed the standard procedures for tissue processing. The deparaffinized sections were then stained with hematoxylin and eosin. Immunohistochemistry The deparaffinized sections were subjected to a reaction with particular antibodies according to the methodology described in a prior work . The sections were subjected to heat at 90°C in citrate buffer (0.01 M, pH 6.0) for 1 h to facilitate antigen retrieval. They were subsequently treated with 0.3% hydrogen peroxide solution in distilled water for a duration of 20 min to block endogenous peroxidase activity. To inhibit nonspecific binding, the sections were exposed to normal horse serum (Vectastain Elite ABC kit; Vector Laboratories, USA) for 1 h, followed by overnight incubation at 4°C with a 1:500 dilution of mouse monoclonal anti-PEDV (Median Diagnostics, Korea) antibody. The slices were then subjected to treatment with a biotinylated horse anti-mouse IgG antibody using the Vectastain Elite ABC kit, followed by the application of the avidin-biotin peroxidase complex using the same kit (Vector Laboratories). The peroxidase reaction was initiated by employing a diaminobenzidine substrate kit (SK-4100; Vector Laboratories). Before being mounted, the sections were stained with hematoxylin as a counterstain. Lectin histochemistry The lectin screening kits, including Biotinylated lectin kits I, II, and III (Cat. No. BK-1000, BK-2000, and BK-3000, respectively), were obtained from Vector Laboratories. provides a comprehensive summary of the lectins used including their acronyms, origins, and specificities. The lectins were classified into different groups based on their specific binding preferences and inhibitory sugars. These groups included N-acetylglucosamine-, mannose-, galactose, N-acetylgalactosamine-, complex type N-glycan-, and fucose-binding lectins. The competitive inhibition experiment utilized sugars obtained from Sigma-Aldrich (USA) and Vector Laboratories, as detailed in . The compounds used in the study included α-methyl mannoside/α-methyl glucoside (Sigma-Aldrich), β-D-galactose (Sigma-Aldrich), lactose (Galβ1, 4Glc; Sigma-Aldrich), melibiose (Galα1, 6Glc; Sigma-Aldrich), N-acetyl-D-galactosamine (α-D-GalNAc; Sigma-Aldrich), N-acetyl-D-glucosamine (β-D-GlcNAc; Sigma-Aldrich), and chitin hydrolysate (Vector Laboratories). Following deparaffinization and rehydration, the sections were subjected to deactivation of endogenous peroxidase activity using 0.3% hydrogen peroxide in methanol. To avoid nonspecific responses, they were then washed with phosphate-buffered saline (PBS) and submerged in 1% bovine serum albumin in PBS. After undergoing an additional round of washing, the sections were incubated at 4°C overnight with biotinylated lectins. They subsequently underwent a 40-min reaction at room temperature using a Vectastain Elite ABC reagent containing an avidin-biotin-peroxidase complex. Prior to mounting, the sections underwent a final counterstaining step using hematoxylin, after which they were exposed to a diaminobenzidine substrate kit (SK-4100; Vector Laboratories). Negative controls for lectin histochemistry were generated by excluding the principal reagent (biotinylated lectins) and pretreating the lectins in 0.2–0.5 M Tris solution with suitable inhibitors for 1 h at room temperature. Lectin staining was assessed using a semiquantitative approach, with the staining intensity categorized as follows: no staining , mild staining , moderate staining , and strong staining . Six 7-day-old crossbred Large White × Landrace male piglets were procured from pig farms in Chungcheongnam-do, Republic of Korea. Among them, three piglets that displayed gastroenteritis symptoms were identified as having being infected with PEDV through polymerase chain reaction (PCR) analysis (data not shown). Furthermore, an examination was performed to identify the existence of pathogenic microorganisms in the feces (data not shown). Three healthy piglets from a different herd that had no symptoms of diarrhea served as the control group. These control piglets were confirmed to be PEDV-negative through PCR analysis. After performing a post-mortem examination, parts of the small intestines including the duodenum, jejunum, and ileum were removed and then preserved in 10% buffered formaldehyde solution. The experimental and animal handling methods were performed in compliance with the rules set by the Institutional Animal Care and Use Committee of Chonnam National University (CNU IACUC-YB-2024-70). The small intestines including the duodenum, jejunum, and ileum were collected separately. The specimens were immersed in 10% buffered formaldehyde solution and then embedded in paraffin and cut into sections that were 4 μm thick. These steps followed the standard procedures for tissue processing. The deparaffinized sections were then stained with hematoxylin and eosin. The deparaffinized sections were subjected to a reaction with particular antibodies according to the methodology described in a prior work . The sections were subjected to heat at 90°C in citrate buffer (0.01 M, pH 6.0) for 1 h to facilitate antigen retrieval. They were subsequently treated with 0.3% hydrogen peroxide solution in distilled water for a duration of 20 min to block endogenous peroxidase activity. To inhibit nonspecific binding, the sections were exposed to normal horse serum (Vectastain Elite ABC kit; Vector Laboratories, USA) for 1 h, followed by overnight incubation at 4°C with a 1:500 dilution of mouse monoclonal anti-PEDV (Median Diagnostics, Korea) antibody. The slices were then subjected to treatment with a biotinylated horse anti-mouse IgG antibody using the Vectastain Elite ABC kit, followed by the application of the avidin-biotin peroxidase complex using the same kit (Vector Laboratories). The peroxidase reaction was initiated by employing a diaminobenzidine substrate kit (SK-4100; Vector Laboratories). Before being mounted, the sections were stained with hematoxylin as a counterstain. The lectin screening kits, including Biotinylated lectin kits I, II, and III (Cat. No. BK-1000, BK-2000, and BK-3000, respectively), were obtained from Vector Laboratories. provides a comprehensive summary of the lectins used including their acronyms, origins, and specificities. The lectins were classified into different groups based on their specific binding preferences and inhibitory sugars. These groups included N-acetylglucosamine-, mannose-, galactose, N-acetylgalactosamine-, complex type N-glycan-, and fucose-binding lectins. The competitive inhibition experiment utilized sugars obtained from Sigma-Aldrich (USA) and Vector Laboratories, as detailed in . The compounds used in the study included α-methyl mannoside/α-methyl glucoside (Sigma-Aldrich), β-D-galactose (Sigma-Aldrich), lactose (Galβ1, 4Glc; Sigma-Aldrich), melibiose (Galα1, 6Glc; Sigma-Aldrich), N-acetyl-D-galactosamine (α-D-GalNAc; Sigma-Aldrich), N-acetyl-D-glucosamine (β-D-GlcNAc; Sigma-Aldrich), and chitin hydrolysate (Vector Laboratories). Following deparaffinization and rehydration, the sections were subjected to deactivation of endogenous peroxidase activity using 0.3% hydrogen peroxide in methanol. To avoid nonspecific responses, they were then washed with phosphate-buffered saline (PBS) and submerged in 1% bovine serum albumin in PBS. After undergoing an additional round of washing, the sections were incubated at 4°C overnight with biotinylated lectins. They subsequently underwent a 40-min reaction at room temperature using a Vectastain Elite ABC reagent containing an avidin-biotin-peroxidase complex. Prior to mounting, the sections underwent a final counterstaining step using hematoxylin, after which they were exposed to a diaminobenzidine substrate kit (SK-4100; Vector Laboratories). Negative controls for lectin histochemistry were generated by excluding the principal reagent (biotinylated lectins) and pretreating the lectins in 0.2–0.5 M Tris solution with suitable inhibitors for 1 h at room temperature. Lectin staining was assessed using a semiquantitative approach, with the staining intensity categorized as follows: no staining , mild staining , moderate staining , and strong staining . Histopathological findings of PEDV infection in the small intestines of piglets The duodenal, jejunal, and ileal sections from the control group had typical histologic characteristics, as depicted in . No discernible disparities were noted in the duodenum between the infected and control groups . Histological examination revealed notable changes in the jejunum and ileum of PEDV-infected piglets. Notably, significant disparities in villous height were seen when comparing the sections from controls and infected animals, and infected animals displayed a reduced villous height. In addition, the jejunum and ileum of the infected piglets exhibited slightly wider villi width than the control piglets. Immunostaining of PEDV in the small intestines of piglets The immunohistochemical examination showed that the small intestines of all infected piglets had the PEDV antigen with different levels of staining intensity . Upon examining various segments of the small intestines, we observed that both non-infected controls and PEDV-infected piglets did not exhibit a positive response in the duodenum. However, in piglets infected with PEDV, the jejunum and ileum displayed a more pronounced and vigorous reaction. In particular, the ends of the villi exhibited the highest concentration of positive reactions, while the crypts did not demonstrate any discernible reactivity. Lectin histochemistry in the duodenum The intensities of 21 different lectins in the duodenum of both the control and PEDV-infected piglets are detailed in . In the non-infected control group, different positive reactions to these lectins were evident in various regions of the duodenum. Compared with the non-infected controls, the PEDV-infected piglets showed significant alterations in three specific groups of lectins, namely N-acetylglucosamine-, galactose-, and fucose-binding lectins, with changes observed in multiple regions of the duodenum. N-acetylglucosamine-binding lectins, such as wheat germ agglutinin (WGA) , Lycopersicon esculentum lectin (LEL) , and Solanum tuberosum lectin (STL) , displayed moderate to strong positive staining at the brush border and intestinal epithelium, except for Bandeiraea simplicifolia lectin (BSL)-II. The levels of succinylated-WGA (s-WGA), LEL , and STL in the brush border or enterocytes tended to decrease, whereas those of WGA and LEL in the goblet cells considerably increased. Most galactose-binding lectins elicited different responses in each duodenal region. The group infected with PEDV had a weaker response than did the non-infected control group when exposed to Ricinus communis agglutinin I (RCA 120 ), Arachis hypogaea (peanut) agglutinin (PNA), BSL-I, and Erythrina cristagalli lectin (ECL) . The fucose-binding lectin Ulex europaeus I was moderately to strongly expressed in the non-infected controls. However, in the PEDV-infected group, its expression was decreased in the brush border, enterocytes, and goblet cells and increased slightly in the lamina propria . Lectin histochemistry in the jejunum summarizes the levels of 21 lectins in the jejunum of both the non-infected controls and PEDV-infected piglets. In the non-infected control group, different positive reactions to these 21 lectins were observed throughout the jejunum. In contrast, the PEDV-infected piglets exhibited significant alterations in the following three specific groups of lectins: N-acetylglucosamine-, galactose-, and N-acetylgalactosamine-binding lectins, with changes observed in multiple regions of the jejunum. The PEDV infection did not affect the expression levels of most N-acetylglucosamine-binding lectins in the brush border. However, the levels of Datura stramonium lectin (DSL), LEL, and STL in the enterocytes decreased, and those of s-WGA and WGA significantly increased in the goblet cells. In the non-infected control group, all galactose-binding lectins, including RCA 120 and PNA , except jacalin, were expressed in the goblet cells. There was a notable increase in their expression in the PEDV-infected group . The N-acetylgalactosamine-binding lectins elicited various responses depending on their location. Glycine maxi (soybean) agglutinin (SBA), in particular, exhibited modest expression in the brush border of the non-infected jejunum . However, in the PEDV-infected group, SBA expression was substantially increased, specifically in the goblet cells . Lectin histochemistry in the ileum presents the intensities of 21 lectins in the ileum of the non-infected controls and PEDV-infected piglets. In the non-infected control group, different positive reactions to these lectins varied across different regions of the ileum. Compared with the controls, the PEDV-infected piglets exhibited significant alterations in three specific groups of lectins, namely N-acetylglucosamine-, galactose-, and complex type N-glycan (complex oligosaccharide)-binding lectins, with changes noted in multiple regions of the ileum. Distinct regional distributions of the labeling patterns of N-acetylglucosamine-binding lectin were observed. The PEDV-infected group exhibited weaker reactivities to LEL and STL in the brush border, enterocytes, and crypts than did the non-infected control group. Galactose-binding lectin histochemistry revealed decreased reactivity of PNA and ECL, with significant changes observed in the brush border and crypts of the PEDV-infected group . The complex type N-glycan-binding lectin PHA-E demonstrated a relative rise in reactivity in the goblet cells, lamina propria, and crypts following PEDV infection. Conversely, PHA-L exhibited a reduction in reactivity, specifically in the brush border. The duodenal, jejunal, and ileal sections from the control group had typical histologic characteristics, as depicted in . No discernible disparities were noted in the duodenum between the infected and control groups . Histological examination revealed notable changes in the jejunum and ileum of PEDV-infected piglets. Notably, significant disparities in villous height were seen when comparing the sections from controls and infected animals, and infected animals displayed a reduced villous height. In addition, the jejunum and ileum of the infected piglets exhibited slightly wider villi width than the control piglets. The immunohistochemical examination showed that the small intestines of all infected piglets had the PEDV antigen with different levels of staining intensity . Upon examining various segments of the small intestines, we observed that both non-infected controls and PEDV-infected piglets did not exhibit a positive response in the duodenum. However, in piglets infected with PEDV, the jejunum and ileum displayed a more pronounced and vigorous reaction. In particular, the ends of the villi exhibited the highest concentration of positive reactions, while the crypts did not demonstrate any discernible reactivity. The intensities of 21 different lectins in the duodenum of both the control and PEDV-infected piglets are detailed in . In the non-infected control group, different positive reactions to these lectins were evident in various regions of the duodenum. Compared with the non-infected controls, the PEDV-infected piglets showed significant alterations in three specific groups of lectins, namely N-acetylglucosamine-, galactose-, and fucose-binding lectins, with changes observed in multiple regions of the duodenum. N-acetylglucosamine-binding lectins, such as wheat germ agglutinin (WGA) , Lycopersicon esculentum lectin (LEL) , and Solanum tuberosum lectin (STL) , displayed moderate to strong positive staining at the brush border and intestinal epithelium, except for Bandeiraea simplicifolia lectin (BSL)-II. The levels of succinylated-WGA (s-WGA), LEL , and STL in the brush border or enterocytes tended to decrease, whereas those of WGA and LEL in the goblet cells considerably increased. Most galactose-binding lectins elicited different responses in each duodenal region. The group infected with PEDV had a weaker response than did the non-infected control group when exposed to Ricinus communis agglutinin I (RCA 120 ), Arachis hypogaea (peanut) agglutinin (PNA), BSL-I, and Erythrina cristagalli lectin (ECL) . The fucose-binding lectin Ulex europaeus I was moderately to strongly expressed in the non-infected controls. However, in the PEDV-infected group, its expression was decreased in the brush border, enterocytes, and goblet cells and increased slightly in the lamina propria . summarizes the levels of 21 lectins in the jejunum of both the non-infected controls and PEDV-infected piglets. In the non-infected control group, different positive reactions to these 21 lectins were observed throughout the jejunum. In contrast, the PEDV-infected piglets exhibited significant alterations in the following three specific groups of lectins: N-acetylglucosamine-, galactose-, and N-acetylgalactosamine-binding lectins, with changes observed in multiple regions of the jejunum. The PEDV infection did not affect the expression levels of most N-acetylglucosamine-binding lectins in the brush border. However, the levels of Datura stramonium lectin (DSL), LEL, and STL in the enterocytes decreased, and those of s-WGA and WGA significantly increased in the goblet cells. In the non-infected control group, all galactose-binding lectins, including RCA 120 and PNA , except jacalin, were expressed in the goblet cells. There was a notable increase in their expression in the PEDV-infected group . The N-acetylgalactosamine-binding lectins elicited various responses depending on their location. Glycine maxi (soybean) agglutinin (SBA), in particular, exhibited modest expression in the brush border of the non-infected jejunum . However, in the PEDV-infected group, SBA expression was substantially increased, specifically in the goblet cells . presents the intensities of 21 lectins in the ileum of the non-infected controls and PEDV-infected piglets. In the non-infected control group, different positive reactions to these lectins varied across different regions of the ileum. Compared with the controls, the PEDV-infected piglets exhibited significant alterations in three specific groups of lectins, namely N-acetylglucosamine-, galactose-, and complex type N-glycan (complex oligosaccharide)-binding lectins, with changes noted in multiple regions of the ileum. Distinct regional distributions of the labeling patterns of N-acetylglucosamine-binding lectin were observed. The PEDV-infected group exhibited weaker reactivities to LEL and STL in the brush border, enterocytes, and crypts than did the non-infected control group. Galactose-binding lectin histochemistry revealed decreased reactivity of PNA and ECL, with significant changes observed in the brush border and crypts of the PEDV-infected group . The complex type N-glycan-binding lectin PHA-E demonstrated a relative rise in reactivity in the goblet cells, lamina propria, and crypts following PEDV infection. Conversely, PHA-L exhibited a reduction in reactivity, specifically in the brush border. In our study, we investigated changes in glycoconjugates in the small intestines of piglets naturally infected with PEDV. Histological analysis revealed villous atrophy, enterocyte destruction, and mucosal erosion, which were particularly severe in the jejunum and ileum. The immunohistochemical analysis results revealed the presence of PEDV-positive enterocytes on the tips and sides of atrophied or fused villi in these regions, consistent with the findings of previous studies . Although the jejunum and ileum differ in physiological roles, tissue structures, and cell compositions, they showed distinct patterns of lectin labeling when infected with PEDV. Notably, changes in lectin labeling were observed in the duodenum even without typical PEDV infection, possibly due to physiological, immunological, and environmental factors. Lectins play a crucial role in pathogen recognition and various physiological processes, and PEDV-induced histological changes in the small intestines affect lectin labeling. However, interpreting the significance of each expression pattern remains challenging. Glycoconjugates are essential components in several biological processes, including viral infections. Several studies have indicated a potential association between glycoconjugates and PEDV infection . PEDV causes PED, a disease that affects pigs of all ages. However, it poses a significant threat to neonatal piglets, resulting in a high mortality rate . Previous studies have shown that the age of the animal has an impact on the lectin-binding profile in the intestines . Some studies have reported that higher levels of specific glycoconjugates are expressed in the small intestines of young pigs . For example, the amount of specific glycoconjugates, such as mucin, in the small intestines of young pigs is larger than that in those of piglets . Furthermore, the distribution pattern of glycoconjugates in the small intestines of mature pigs may vary compared with that in those of young pigs . These variations might impact their resistance or susceptibility to viruses, such as PEDV, and thus contribute significantly to understanding disorders related to intestinal infections . In this study, we examined the intestines of 7-day-old non-infected controls and PEDV-infected piglets to investigate changes in lectin labeling. The results related to lectin expression in the non-infected control group conformed to earlier research findings , but notable differences were observed between the control and infected groups. PEDV infection leads to severe villous atrophy due to infection, necrosis, and detachment of villous epithelial cells . This disrupts the metabolism and protein synthesis of host cells, decreasing the expression of certain glycans recognized by lectins, which is reflected in reduced lectin labeling . The virus targets the brush border and enterocytes, while the presence of immune cells in the infected areas can cause structural changes and inflammation, further decreasing lectin binding . PEDV infection has a substantial impact on the structural and functional integrity of intestinal epithelial cells . Our study results showed that compared with the controls, the PEDV-infected piglets demonstrated significant alterations in N-acetylglucosamine- and galactose-binding lectins (particularly WGA and PNA) in multiple intestinal regions, suggesting that PEDV infection alters glycoconjugate characteristics. Additionally, PEDV infection impacts goblet cells, depleting mucin and compromising the mucin layer, which allows the virus to access the underlying epithelium . A significant reduction in goblet cells was observed in the PEDV-infected piglets . The marked changes in N-acetylglucosamine-, galactose-, and N-acetylgalactosamine-binding lectins (notably WGA, BSL-I, PNA, and SBA) in the goblet cells of the jejunum may reflect a detrimental or compensatory response to mucin depletion. In addition, the potential application of lectins as biomolecules has generated considerable interest because of their capacity to detect carbohydrates on the surface of microorganisms . The ability to identify and avert early adhesion is highly beneficial during the progression of an infection because it is instrumental in preventing the onset of the infection . Consequently, the possibility of using lectins as possible alternatives to antibiotics is being investigated. They have shown potential as antiviral agents owing to their ability to recognize and inactivate high-mannose glycans associated with viruses , making them potent microbicidal agents. For instance, the lectin griffithsin has demonstrated promising antiviral activity against PEDV . Based on our findings, we propose WGA, PNA, and SBA as potential antiviral lectins against PEDV. This highlights the significant potential of lectins in preventing and treating viral infections. However, it is essential to identify critical lectins involved in the immune response to PEDV and utilize glycoconjugate profiling and molecular studies to develop effective prevention and treatment strategies. In conclusion, our research emphasized the modifications observed in glycoconjugates in various parts of the small intestine following PEDV infection. The growing interest in lectins as antiviral agents underscores the need for further investigation regarding anti-PEDV strategies to support our findings. Further research is required to fully understand the role of lectin-binding glycoconjugates in PEDV pathology in different intestinal regions. Advancing our knowledge regarding PEDV infection will ultimately facilitate the development of effective preventive and therapeutic strategies. |
Etablierung und Umsetzung des Nationalen Aktionsplans Gesundheitskompetenz in Deutschland | ae861859-78ae-43ca-b936-8c722fdccf81 | 11868236 | Health Literacy[mh] | Wie die breit gefächerte Literatur und zahlreiche Dokumente der Weltgesundheitsorganisation (WHO) belegen, ist Gesundheitskompetenz – international als „Health Literacy“ bezeichnet – inzwischen zu einem bedeutenden Public-Health-Thema geworden. Definiert wird Health Literacy als „the knowledge, motivation and competences to access, understand, appraise, and apply health information in order to make judgements and take decisions in everyday life concerning healthcare, disease prevention and health promotion“ . Denn zahlreiche Studien zeigen mittlerweile, dass die Gesundheitskompetenz in vielen Ländern – auch in Deutschland – in weiten Teilen der Bevölkerung unzureichend ist . Gleichzeitig stellen Gesundheitskompetenz und ein souveräner Umgang mit Gesundheitsinformationen in modernen komplexen Gesellschaften eine immer wichtiger werdende Voraussetzung für eine selbstbestimmte Lebensgestaltung dar . Die voranschreitende digitale Transformation befördert dies in hohem Tempo. Durch sie sind Gesundheitsinformationen zwar heute einfacher zugänglich als noch vor wenigen Jahren, aber zugleich ist es immer schwerer geworden, relevante Informationen in der unüberschaubaren Informationsfülle zu finden und vor allem, sie einschätzen und nutzen zu können . In Reaktion auf diese Situation wurden in etlichen Ländern politische Strategien, darunter auch „Nationale Aktionspläne“, zur Förderung der Gesundheitskompetenz aufgelegt . Das gilt seit 2018 auch für Deutschland. Über die Entwicklung dieser Strategien und Aktionspläne und auch über die Umsetzung und die ihr zugrundeliegenden Implementationsstrategien und -erfahrungen ist noch immer wenig bekannt. Ziel des vorliegenden Beitrags ist es, die Entstehung und Umsetzung des Nationalen Aktionsplans Gesundheitskompetenz in Deutschland (NAP-GK) darzustellen und zu fragen, welche Erfahrungen damit gesammelt wurden und welche Wirkungen der Plan entfaltet hat. Zuvor wird kurz die Ausgangssituation und die Entwicklung der Gesundheitskompetenzdiskussion in Deutschland betrachtet. Die Idee, einen NAP-GK zu erarbeiten, entstand im Jahr 2015 mit Bekanntwerden der ersten bevölkerungsbezogenen Studienergebnisse für Deutschland . Bis dahin fand das Thema Gesundheitskompetenz in Deutschland kaum Beachtung. Anders war die internationale Situation, insbesondere in den USA. Dort existiert bereits seit den 1990er-/2000er-Jahren eine umfangreiche Forschung zur Gesundheitskompetenz, die auf die Diskussion über die gesundheitlichen Folgen geringer Literalität (unzureichende Lese‑, Schreib- und Rechenfähigkeiten) zurückgeht. Denn bereits in den 1990er-Jahren hatten dort durchgeführte Studien wie der „National Adult Literacy Survey“ (NALS) gezeigt, dass geringe Literalität in den USA ein gravierendes gesellschaftliches Problem darstellt . Das ist es bis heute geblieben: Nach wie vor verfügen ca. 20 % der amerikanischen erwachsenen Bevölkerung nicht über grundlegende Lese- und Schreibfähigkeiten. In Deutschland sind es 12 % der Erwachsenen – das entspricht 6,2 Mio. Menschen . Die Studien weisen zudem darauf hin, dass sich geringe Literalität quer durch alle Gesellschaftsschichten zieht, die unteren sozialen Schichten aber besonders trifft, und ebenso, dass sie gesundheitlich folgenreich ist: Denn sie erschwert den Zugang zum Gesundheitssystem und behindert die Compliance, weil Verschreibungen, Therapiehinweise oder Krankheitsinformationen nicht gelesen und verstanden werden können . In der Regel lag diesen wie auch den anschließend entstandenen Studien ein funktionales, auf Literalität begrenztes Verständnis von Health Literacy zugrunde, das in den folgenden Jahren zahlreiche konzeptionelle Weiterentwicklungen erfahren hat. Zu erwähnen sind insbesondere die Überlegungen zum relationalen Charakter von Gesundheitskompetenz, nach denen Gesundheitskompetenz als Ausdruck sowohl persönlicher Fähigkeiten als auch der gegebenen Umgebungs- und Strukturbedingungen zu verstehen ist . Hervorzuheben sind auch die Bemühungen, zu einem weiter gefassten, Public-Health-orientierten Verständnis von Gesundheitskompetenz zu kommen . Durch diese und andere Impulse hat sich Gesundheitskompetenz/Health Literacy inzwischen zu einem umfassenden, multidimensionalen Public Health-orientierten Konzept entwickelt, das nicht nur auf Compliance, sondern auch auf Krankheitsbewältigung/Versorgung, Prävention und Gesundheitsförderung abstellt und dabei auf die Stärkung selbstbestimmter Entscheidungen von Menschen über ihre Gesundheit zielt . Dieses Verständnis prägt auch den ersten Europäischen Health Literacy Survey (HLS-EU) , mit dem 2012 faktisch der Startschuss für die Forschung zur Gesundheitskompetenz in Deutschland erfolgte. Der HLS-EU ist aber auch deshalb erwähnenswert, weil mit ihm eine neue, die europäische Diskussion bis heute prägende Definition und ein entsprechendes Erhebungsinstrument entwickelt wurden und mit ihm – nicht weniger wichtig – erstmals empirische Befunde für Deutschland vorlagen. Allerdings waren sie auf das Bundesland Nordrhein-Westfalen (NRW) beschränkt. Deutschlandweite Daten fehlten somit, und dies motivierte die Entstehung erster Studien zur Gesundheitskompetenz in Deutschland . Dazu gehört auch die erste Studie zur Gesundheitskompetenz der Bevölkerung in Deutschland (Health Literacy Survey Germany 1 – HLS-GER 1 ), die 2014 nach dem gleichen methodischen Vorgehen wie der HLS-EU durchgeführt wurde. Die Ergebnisse dieser Studie sorgten sowohl in der gesundheitswissenschaftlichen als auch der gesundheitspolitischen Diskussion für eine nachhaltige Veränderung der Aufmerksamkeit für das Thema. Denn sie zeigten, dass mehr als die Hälfte der Bevölkerung in Deutschland – konkret 54,3 % – eine geringe Gesundheitskompetenz aufweist. Sie verdeutlichten zudem, dass Gesundheitskompetenz ungleich verteilt ist und geringe Gesundheitskompetenz u. a. mit niedriger Bildung, niedrigem Sozialstatus und einem höheren Lebensalter assoziiert ist . Anders formuliert: Menschen mit diesen Merkmalen gehören zu den sogenannten „vulnerablen“ Gruppen, die besonders große Schwierigkeiten im Umgang mit Gesundheitsinformationen haben und bei der Förderung von Gesundheitskompetenz spezielle Beachtung erhalten sollten. Das ist umso wichtiger, als geringe Gesundheitskompetenz mit zahlreichen negativen Konsequenzen einhergeht, die von ungesünderen Verhaltensweisen über verminderte Nutzung von Präventionsangeboten, ein höheres Krankheitsrisiko bis hin zu einer intensiveren Nutzung des Gesundheitssystems reichen und erhebliche Kosten verursachen . Ähnliche Tendenzen zeigen sich auch in anderen Ländern (z. B. ) sowie in nachfolgenden Studien zur Gesundheitskompetenz der Bevölkerung und einzelner Bevölkerungsgruppen in Deutschland (z. B. ). Die dargestellten Erkenntnisse zur Gesundheitskompetenz haben international die Entstehung zahlreicher Strategien und Nationaler Aktionspläne zur Verbesserung der Gesundheitskompetenz nach sich gezogen. Auch in Deutschland bildete sich unter dem Eindruck der ersten Studienergebnisse 2015 eine Initiative zur Erarbeitung eines Nationalen Aktionsplans Gesundheitskompetenz. Anders als üblich wurde sie nicht durch eine politische Instanz initiiert, sondern entstand unter der Federführung von Wissenschaftler:innen der Universität Bielefeld als zivilgesellschaftliche Initiative und setzte sich aus einer Gruppe von ausgewiesenen nationalen und internationalen Expert:innen aus unterschiedlichen wissenschaftlichen Disziplinen, Institutionen der Gesundheitsversorgung und der Gesundheitspolitik zusammen. Ende 2015 startete die Erarbeitung des NAP-GK. Sie beruhte auf mehreren Arbeitsschritten und begann mit einer Recherche und Analyse vorliegender Aktionspläne sowie einem Review der vorliegenden Literatur und der existenten Studienbefunde. Es folgten ausgiebige Diskussionen der daraus erwachsenen Konsequenzen für die Kontur und den Inhalt des zu erarbeitenden Aktionsplans für Deutschland – dies im Plenum sowie in unterschiedlichen Arbeitsgruppen der NAP-GK-Expertengruppe. Sie mündeten in der Erstellung eines ersten Entwurfs für den NAP-GK. Daran schloss sich ein umfangreicher Diskussions- und Konsentierungsprozess an: In einem ersten Schritt wurde ein Workshop mit Stakeholdern und Vertreter:innen der Verbände im Gesundheits- und Bildungswesen sowie den Mitgliedern der „Allianz für Gesundheitskompetenz“ (Initiative des Bundesministeriums für Gesundheit, BMG) durchgeführt, um die Empfehlungen des Plans in den unterschiedlichen Handlungsfeldern zu diskutieren und zu ergänzen. Ein zweiter Workshop richtete sich gezielt an Selbsthilfe- und Patientenorganisationen mit dem Ziel, den Entwurf des NAP-GK aus ihrer Perspektive zu kommentieren. Zusätzlich wurden Einzelgespräche mit weiteren relevanten Interessen- und Arbeitsgruppen sowie einzelnen Stakeholdern durchgeführt. Alle Schritte wurden protokolliert, diskutiert und wichtige relevante Ergebnisse anschließend eingearbeitet. Nach mehrfacher Überarbeitung der Rohfassung erfolgte im Spätherbst 2017 die Finalisierung des NAP-GK. Der innerhalb von knapp zwei Jahren erarbeitete NAP-GK enthält 15 aufeinander abgestimmte Empfehlungen in vier Handlungsbereichen, wie die Gesundheitskompetenz in Deutschland gefördert werden kann und welche Maßnahmen dazu angestoßen werden sollten. Die Empfehlungen zielen auf unterschiedliche Bereiche des gesellschaftlichen Lebens und widmen sich zu gleichen Teilen der Stärkung der persönlichen Gesundheitskompetenz wie der Verbesserung der Umgebungs- und Strukturbedingungen. Im Mittelpunkt des ersten Handlungsbereichs stehen die alltäglichen Lebenswelten. Die Empfehlungen konzentrieren sich auf die Stärkung der Gesundheitskompetenz im Erziehungs- und Bildungsbereich, der Wohnumgebung, der Arbeitswelt, der Kommune, in den Medien und im Freizeit- und Konsumbereich. Der zweite Handlungsbereich widmet sich dem Gesundheitssystem. Aufgrund seiner Komplexität, Instanzenvielfalt und Intransparenz stellt es sehr hohe Anforderungen an die Nutzenden. Empfohlen wird, es zu einem gesundheitskompetenten, nutzerfreundlichen und patientenzentrierten System weiterzuentwickeln und dazu die Navigation, Kommunikation, Information und Partizipation zu verbessern. Der dritte Handlungsbereich befasst sich mit dem Leben mit chronischen Erkrankungen, die sich immer weiter ausbreiten. Sie gehen mit zahlreichen Bewältigungsherausforderungen an die Erkrankten und ihre Familien einher. Die Empfehlungen zielen darauf ab, den Betroffenen einen gesundheitskompetenten Umgang mit der Krankheit und ihren zahlreichen Begleiterscheinungen zu ermöglichen, ihre Fähigkeit zum kompetenten Selbstmanagement zu stärken und das Alltagsleben mit chronischer Erkrankung zu erleichtern. Der vierte Handlungsbereich konzentriert sich auf die Verbesserung (und den Ausbau) der Forschung zur Gesundheitskompetenz . Der NAP-GK wurde im Februar 2018 dem damaligen Gesundheitsminister, der zugleich Schirmherr des Vorhabens war, übergeben. Der Plan richtet sich vor allem an die Politik und an relevante Akteur:innen im Gesundheitssystem und in anderen Bereichen der Gesellschaft (Erziehungs- und Bildungssystem, Freizeitbereich, Arbeitswelt etc.). Sein übergeordnetes Ziel besteht darin, gemäß der Prämisse „Health Literacy in all Policies“ ein Kooperationsbündnis aller Bereiche der Gesellschaft zu initiieren und so ein umfassendes Vorgehen bei der Förderung der Gesundheitskompetenz zu ermöglichen. Mit der Veröffentlichung des NAP-GK 2018 trat die Arbeit der Expertengruppe in eine neue Phase ein und begann die Umsetzung des NAP-GK. Mit ihr wurde weitgehend Neuland betreten. Denn zur damaligen Zeit lagen erst wenige Erfahrungen vor, wie die Umsetzung solcher Aktionspläne in die komplexen Strukturen der Politik und der Praxis erfolgen kann. Auch Diskussionen über Implementationsstrategien und -erfahrungen existierten nicht. Sie entstehen erst langsam (z. B. ). Vielfach waren die bis dato existenten Aktionspläne zudem in Ländern mit einem stärker staatlich geprägten Gesundheitssystem entstanden, in denen die Umsetzung von Empfehlungen prinzipiell dadurch möglich ist, dass von übergeordneten Instanzen Richtlinien formuliert werden, deren Umsetzung überprüft und eingefordert werden kann. Mittlerweile wird bezweifelt, dass solche Top-Down-Strategien allein erfolgversprechend sind (, auch ). Analysen der Umsetzung politischer Programme oder gesetzlicher Regelungen zufolge ist anzunehmen, dass sie selbst in zentralistisch gesteuerten Systemen nur sehr bedingt wirkungsvoll sind . Weil Deutschland kein zentral gesteuertes Gesundheitssystem besitzt, sondern als sogenannter „konservativer“ Wohlfahrtsstaat fast alle seine Sicherungs- und Versorgungssysteme nach dem Subsidiaritätsprinzip gestaltet hat, kam für die Umsetzung ohnehin keine Top-Down-Strategie in Frage. Im deutschen Gesundheitssystem kann der Staat nur politische Grundsätze formulieren, deren Konkretisierung und Umsetzung durch eigenständige, selbstorganisierte, relativ autonome Organisationen und Verbände erfolgt. Für die Umsetzung des NAP-GK erwuchs daraus die Notwendigkeit, nicht nur politische Entscheidungsträger, sondern auch die entsprechenden Instanzen, Organisationen, Verbände und Akteur:innen der Selbstverwaltung im Gesundheitssystem und in anderen Infrastrukturbereichen in die Umsetzung und Realisierung der Empfehlungen einzubeziehen. Damit bot sich eine eher Bottom-up ausgerichtete, auf Kooperation setzende Umsetzungsstrategie an, um Institutionen aus möglichst vielen Bereichen zur Umsetzung des Aktionsplans zu motivieren. Das sprach für eine auf Kooperation setzende Implementationsstrategie mit „upstream and downstream actors“ aus unterschiedlichen Bereichen . Unter Rückgriff auf implementationswissenschaftliche Überlegungen wurde die Umsetzung des NAP-GK zudem nicht als ein einmaliger Vorgang, sondern als fortlaufender Prozess konzipiert . Dafür spricht, dass die Umsetzung von Aktionsplänen oder politischen Strategien fast immer auch mit der Einführung von Innovationen verbunden ist, deren Übernahme sich selten ad hoc, sondern meist schrittweise vollzieht. Sie stößt zudem oft zunächst auf Widerstand , dessen Ausräumung Zeit erfordert. Als weiteres wichtiges Merkmal der Umsetzungsstrategie des NAP-GK ist anzuführen, dass sie als Kontinuum von drei unterschiedlichen, aufeinander aufbauenden und sich überlappenden Schritten von Diffusion, Dissemination und Implementation angelegt wurde (; Abb. ). Erster Schritt der Umsetzung: Diffusion Der erste Schritt „Diffusion“ zielte darauf, eine möglichst breite Streuung des NAP-GK und ein an der Umsetzung interessiertes Klima zu schaffen. Dazu wurde der Aktionsplan auf unterschiedlichen Kanälen publik gemacht. Ausgangspunkt war eine Großveranstaltung, an der Akteur:innen aus Politik, Selbstverwaltung, Medien und Wissenschaft teilnahmen. Anschließend wurde der Aktionsplan umfangreich postalisch und digital distribuiert und auf Tagungen vorgestellt. Besonders die Website hat sich als wichtiges Medium zur Distribution des NAP-GK erwiesen. Sie liefert Hintergrundinformationen über den NAP-GK, das Leitungsteam, die Expertengruppe und die Koordinierungsstelle sowie geplante und durchgeführte Veranstaltungen und sonstige Aktivitäten. Auf diese Weise gelang es, den NAP-GK in unterschiedlichen Netzwerken auch in vielen Verbänden und Vereinen bekannt zu machen. Zugleich wurden zahlreiche Presseberichte und Mediennachrichten über den Aktionsplan gestreut. Zweiter Schritt der Umsetzung: Dissemination Bei der parallel eingeleiteten Dissemination ging es um eine gezielte Verbreitung des NAP-GK an ausgewählte, wichtige Zielgruppen und Stakeholder, so u. a. Politiker:innen auf Bundes- und Landesebene, Leitungskräfte von Verbänden und Organisationen des Gesundheitswesens sowie von Wohlfahrtsorganisationen oder Einrichtungen des Erziehungs- und Bildungssektors und Stiftungen. Ziel war es, sie über den NAP-GK und seine Empfehlungen zu informieren, bei ihnen Adoptionsbereitschaft zu wecken und sie zu motivieren, sich in ihrem Handlungsbereich für die Förderung von Gesundheitskompetenz zu engagieren. Dazu erfolgten Publikationen in entsprechenden Fachzeitschriften, ebenso zahlreiche Vorträge über den NAP-GK auf Tagungen, unterschiedlichsten Fachveranstaltungen und Kongressen im Gesundheitssystem und in anderen relevanten Gesellschaftsbereichen. Zudem wurden zahlreiche Beiträge in Fachzeitschriften zum NAP-GK erstellt und veröffentlicht. Dritter Schritt der Umsetzung: Implementation Ein dritter Schritt zielte auf die Implementation des NAP-GK in für die Förderung von Gesundheitskompetenz als wichtig erachtete Handlungsbereiche. Im Mittelpunkt dieses Schritts standen a) kooperative Workshops mit Stakeholdern und Vertreter:innen aus der Politik, Wissenschaft und Praxis und b) eine Zusammenarbeit mit relevanten Netzwerken nach dem Motto „Health Literacy in all Policies“. a) Workshops Die Workshops bilden das Kernelement der Umsetzungsstrategie des NAP-GK. Ziel der Workshops sollte es sein, eine kooperative Weiterbearbeitung der Empfehlungen des NAP-GK in konkrete, direkt umsetzbare Handlungsschritte zu initiieren. Intention war es, auf diese Weise eine Identifikation mit dem NAP-GK und dessen Adoption zu erreichen und die beteiligten Akteur:innen zu motivieren, sich in ihrem Wirkungsfeld für die Umsetzung zu engagieren. Als Ergebnis der Workshops wurden jeweils Strategiepapiere erarbeitet und anschließend distribuiert. Insgesamt wurden neun von der Koordinierungsstelle des NAP-GK organisierte Workshops mit unterschiedlichen Akteur:innen durchgeführt, an denen jeweils etwa 25–30 Teilnehmende mitwirkten. Die Workshops folgten dabei einem festen Ablaufmuster (Abb. ). Von den Workshops wurden ausführliche Protokolle erstellt, die als Datenmaterial für die anschließende kooperative Erarbeitung eines Strategiepapiers dienten. Dazu wurden die wichtigsten Ergebnisse der Workshops anschließend zusammengefasst und erste hypothesenartige Kernaussagen festgelegt. Auf dieser Basis wurde ein von der Leitung der Workshops verantwortetes Strategiepapier entworfen. Es wurde den Workshopteilnehmenden zur Kommentierung zugesendet und anschließend meist mehrfach überarbeitet, bis alle einer Publikation zustimmten. Die Strategiepapiere wurden den Teilnehmenden für ihre Netzwerke zur Verfügung gestellt und parallel über die Website des NAP-GK distribuiert. Teilweise wurden sie ergänzend in Zeitschriften veröffentlicht. Die Strategiepapiere stießen in der Regel auf große Resonanz. In einigen Organisationen führten sie zu intensiven Diskussionen. Sie stimulierten außerdem die Identifikation mit dem NAP-GK und auch die Umsetzungsbereitschaft, was sich u. a. in den Anfragen zu Vorträgen und eigenen Projektplanungen zeigte. Bis heute, nach rund siebenjährigem Bestehen des NAP-GK, ist eine breite Palette an Praxisinitiativen zu finden, die auf den Aktionsplan Bezug nehmen. b) Kooperation mit Netzwerken Ein weiterer Bestandteil der kooperativen Implementationsstrategie bestand in der Zusammenarbeit mit Netzwerken und Mitwirkung in Gremien oder Arbeitsgruppen. Dazu gehörten z. B. die Arbeitsgruppe Gesundheitskompetenz im Nationalen Krebsplan, eine gleichnamige Gruppe in einem Pflegeverband, Beiräte von Institutionen, die entstehenden Netzwerke zur Gesundheitskompetenz wie etwa die AG Gesundheitskompetenz im Deutschen Netzwerk Versorgungsforschung und in der Deutschen Gesellschaft für Sozialmedizin und Prävention oder das Deutsche Netzwerk Gesundheitskompetenz (DNGK). Eine besondere Bedeutung hat die Kooperation mit der „Allianz Gesundheitskompetenz“, die 2017 vom BMG gegründet wurde und das Ziel hat, im Gesundheitswesen Praxisprojekte zur Förderung der Gesundheitskompetenz anzustoßen. Mit dieser Programmatik war und ist die Allianz Gesundheitskompetenz für die Umsetzung des Aktionsplans besonders wichtig. Im Gegenzug konnten die Mitglieder der Allianz dabei unterstützt werden, Ideen für Praxisprojekte zu entwickeln und zu realisieren, denn dazu fehlte es zu Beginn vielfach an Fachexpertise. Der erste Schritt „Diffusion“ zielte darauf, eine möglichst breite Streuung des NAP-GK und ein an der Umsetzung interessiertes Klima zu schaffen. Dazu wurde der Aktionsplan auf unterschiedlichen Kanälen publik gemacht. Ausgangspunkt war eine Großveranstaltung, an der Akteur:innen aus Politik, Selbstverwaltung, Medien und Wissenschaft teilnahmen. Anschließend wurde der Aktionsplan umfangreich postalisch und digital distribuiert und auf Tagungen vorgestellt. Besonders die Website hat sich als wichtiges Medium zur Distribution des NAP-GK erwiesen. Sie liefert Hintergrundinformationen über den NAP-GK, das Leitungsteam, die Expertengruppe und die Koordinierungsstelle sowie geplante und durchgeführte Veranstaltungen und sonstige Aktivitäten. Auf diese Weise gelang es, den NAP-GK in unterschiedlichen Netzwerken auch in vielen Verbänden und Vereinen bekannt zu machen. Zugleich wurden zahlreiche Presseberichte und Mediennachrichten über den Aktionsplan gestreut. Bei der parallel eingeleiteten Dissemination ging es um eine gezielte Verbreitung des NAP-GK an ausgewählte, wichtige Zielgruppen und Stakeholder, so u. a. Politiker:innen auf Bundes- und Landesebene, Leitungskräfte von Verbänden und Organisationen des Gesundheitswesens sowie von Wohlfahrtsorganisationen oder Einrichtungen des Erziehungs- und Bildungssektors und Stiftungen. Ziel war es, sie über den NAP-GK und seine Empfehlungen zu informieren, bei ihnen Adoptionsbereitschaft zu wecken und sie zu motivieren, sich in ihrem Handlungsbereich für die Förderung von Gesundheitskompetenz zu engagieren. Dazu erfolgten Publikationen in entsprechenden Fachzeitschriften, ebenso zahlreiche Vorträge über den NAP-GK auf Tagungen, unterschiedlichsten Fachveranstaltungen und Kongressen im Gesundheitssystem und in anderen relevanten Gesellschaftsbereichen. Zudem wurden zahlreiche Beiträge in Fachzeitschriften zum NAP-GK erstellt und veröffentlicht. Ein dritter Schritt zielte auf die Implementation des NAP-GK in für die Förderung von Gesundheitskompetenz als wichtig erachtete Handlungsbereiche. Im Mittelpunkt dieses Schritts standen a) kooperative Workshops mit Stakeholdern und Vertreter:innen aus der Politik, Wissenschaft und Praxis und b) eine Zusammenarbeit mit relevanten Netzwerken nach dem Motto „Health Literacy in all Policies“. a) Workshops Die Workshops bilden das Kernelement der Umsetzungsstrategie des NAP-GK. Ziel der Workshops sollte es sein, eine kooperative Weiterbearbeitung der Empfehlungen des NAP-GK in konkrete, direkt umsetzbare Handlungsschritte zu initiieren. Intention war es, auf diese Weise eine Identifikation mit dem NAP-GK und dessen Adoption zu erreichen und die beteiligten Akteur:innen zu motivieren, sich in ihrem Wirkungsfeld für die Umsetzung zu engagieren. Als Ergebnis der Workshops wurden jeweils Strategiepapiere erarbeitet und anschließend distribuiert. Insgesamt wurden neun von der Koordinierungsstelle des NAP-GK organisierte Workshops mit unterschiedlichen Akteur:innen durchgeführt, an denen jeweils etwa 25–30 Teilnehmende mitwirkten. Die Workshops folgten dabei einem festen Ablaufmuster (Abb. ). Von den Workshops wurden ausführliche Protokolle erstellt, die als Datenmaterial für die anschließende kooperative Erarbeitung eines Strategiepapiers dienten. Dazu wurden die wichtigsten Ergebnisse der Workshops anschließend zusammengefasst und erste hypothesenartige Kernaussagen festgelegt. Auf dieser Basis wurde ein von der Leitung der Workshops verantwortetes Strategiepapier entworfen. Es wurde den Workshopteilnehmenden zur Kommentierung zugesendet und anschließend meist mehrfach überarbeitet, bis alle einer Publikation zustimmten. Die Strategiepapiere wurden den Teilnehmenden für ihre Netzwerke zur Verfügung gestellt und parallel über die Website des NAP-GK distribuiert. Teilweise wurden sie ergänzend in Zeitschriften veröffentlicht. Die Strategiepapiere stießen in der Regel auf große Resonanz. In einigen Organisationen führten sie zu intensiven Diskussionen. Sie stimulierten außerdem die Identifikation mit dem NAP-GK und auch die Umsetzungsbereitschaft, was sich u. a. in den Anfragen zu Vorträgen und eigenen Projektplanungen zeigte. Bis heute, nach rund siebenjährigem Bestehen des NAP-GK, ist eine breite Palette an Praxisinitiativen zu finden, die auf den Aktionsplan Bezug nehmen. b) Kooperation mit Netzwerken Ein weiterer Bestandteil der kooperativen Implementationsstrategie bestand in der Zusammenarbeit mit Netzwerken und Mitwirkung in Gremien oder Arbeitsgruppen. Dazu gehörten z. B. die Arbeitsgruppe Gesundheitskompetenz im Nationalen Krebsplan, eine gleichnamige Gruppe in einem Pflegeverband, Beiräte von Institutionen, die entstehenden Netzwerke zur Gesundheitskompetenz wie etwa die AG Gesundheitskompetenz im Deutschen Netzwerk Versorgungsforschung und in der Deutschen Gesellschaft für Sozialmedizin und Prävention oder das Deutsche Netzwerk Gesundheitskompetenz (DNGK). Eine besondere Bedeutung hat die Kooperation mit der „Allianz Gesundheitskompetenz“, die 2017 vom BMG gegründet wurde und das Ziel hat, im Gesundheitswesen Praxisprojekte zur Förderung der Gesundheitskompetenz anzustoßen. Mit dieser Programmatik war und ist die Allianz Gesundheitskompetenz für die Umsetzung des Aktionsplans besonders wichtig. Im Gegenzug konnten die Mitglieder der Allianz dabei unterstützt werden, Ideen für Praxisprojekte zu entwickeln und zu realisieren, denn dazu fehlte es zu Beginn vielfach an Fachexpertise. Die Workshops bilden das Kernelement der Umsetzungsstrategie des NAP-GK. Ziel der Workshops sollte es sein, eine kooperative Weiterbearbeitung der Empfehlungen des NAP-GK in konkrete, direkt umsetzbare Handlungsschritte zu initiieren. Intention war es, auf diese Weise eine Identifikation mit dem NAP-GK und dessen Adoption zu erreichen und die beteiligten Akteur:innen zu motivieren, sich in ihrem Wirkungsfeld für die Umsetzung zu engagieren. Als Ergebnis der Workshops wurden jeweils Strategiepapiere erarbeitet und anschließend distribuiert. Insgesamt wurden neun von der Koordinierungsstelle des NAP-GK organisierte Workshops mit unterschiedlichen Akteur:innen durchgeführt, an denen jeweils etwa 25–30 Teilnehmende mitwirkten. Die Workshops folgten dabei einem festen Ablaufmuster (Abb. ). Von den Workshops wurden ausführliche Protokolle erstellt, die als Datenmaterial für die anschließende kooperative Erarbeitung eines Strategiepapiers dienten. Dazu wurden die wichtigsten Ergebnisse der Workshops anschließend zusammengefasst und erste hypothesenartige Kernaussagen festgelegt. Auf dieser Basis wurde ein von der Leitung der Workshops verantwortetes Strategiepapier entworfen. Es wurde den Workshopteilnehmenden zur Kommentierung zugesendet und anschließend meist mehrfach überarbeitet, bis alle einer Publikation zustimmten. Die Strategiepapiere wurden den Teilnehmenden für ihre Netzwerke zur Verfügung gestellt und parallel über die Website des NAP-GK distribuiert. Teilweise wurden sie ergänzend in Zeitschriften veröffentlicht. Die Strategiepapiere stießen in der Regel auf große Resonanz. In einigen Organisationen führten sie zu intensiven Diskussionen. Sie stimulierten außerdem die Identifikation mit dem NAP-GK und auch die Umsetzungsbereitschaft, was sich u. a. in den Anfragen zu Vorträgen und eigenen Projektplanungen zeigte. Bis heute, nach rund siebenjährigem Bestehen des NAP-GK, ist eine breite Palette an Praxisinitiativen zu finden, die auf den Aktionsplan Bezug nehmen. Ein weiterer Bestandteil der kooperativen Implementationsstrategie bestand in der Zusammenarbeit mit Netzwerken und Mitwirkung in Gremien oder Arbeitsgruppen. Dazu gehörten z. B. die Arbeitsgruppe Gesundheitskompetenz im Nationalen Krebsplan, eine gleichnamige Gruppe in einem Pflegeverband, Beiräte von Institutionen, die entstehenden Netzwerke zur Gesundheitskompetenz wie etwa die AG Gesundheitskompetenz im Deutschen Netzwerk Versorgungsforschung und in der Deutschen Gesellschaft für Sozialmedizin und Prävention oder das Deutsche Netzwerk Gesundheitskompetenz (DNGK). Eine besondere Bedeutung hat die Kooperation mit der „Allianz Gesundheitskompetenz“, die 2017 vom BMG gegründet wurde und das Ziel hat, im Gesundheitswesen Praxisprojekte zur Förderung der Gesundheitskompetenz anzustoßen. Mit dieser Programmatik war und ist die Allianz Gesundheitskompetenz für die Umsetzung des Aktionsplans besonders wichtig. Im Gegenzug konnten die Mitglieder der Allianz dabei unterstützt werden, Ideen für Praxisprojekte zu entwickeln und zu realisieren, denn dazu fehlte es zu Beginn vielfach an Fachexpertise. Nach gut drei Jahren war der NAP-GK zu einer Art Referenzwerk für die Förderung von Gesundheitskompetenz in Deutschland geworden und hatte zur Entstehung zahlreicher Initiativen beigetragen – in der Wissenschaft, wie der Praxis oder der Politik. Dann kam die COVID-19-Pandemie, die in fast allen Bereichen gesellschaftlichen Lebens zu einer abrupten Zäsur geführt hat. Auch die Implementationsstrategie des NAP-GK geriet durch sie an Grenzen und musste modifiziert werden, zumal auch die bis dahin erfolgreiche Karriere des Themas Gesundheitskompetenz unerwartet ins Trudeln geriet. Erstaunlich war dies deshalb, weil die Pandemie rasch deutlich zeigte, wie wichtig es in gesundheitlichen Krisen ist, im richtigen Moment die richtigen Informationen zum Umgang mit der unvertrauten Situation zu finden, zu verstehen, einzuordnen und für das eigene Gesundheitsverhalten nutzen zu können. Gerade während der Pandemie wurde die Stärkung der Gesundheitskompetenz daher wichtiger denn je, zumal die Bevölkerung ad hoc mit umfangreichen Informationsherausforderungen konfrontiert war, ohne auf gesichertes Wissen zurückgreifen und ohne sich auf den üblichen Informationspfaden bewegen zu können. Zugleich zeigten die Ergebnisse der zweiten Studie zur Gesundheitskompetenz der Bevölkerung in Deutschland (HLS-GER 2) aus den Jahren 2019/2020, dass sich die Gesundheitskompetenz in Deutschland seit der ersten Studie verschlechtert hat. Auch die ungleiche Verteilung von Gesundheitskompetenz hatte sich verstärkt . Es war also davon auszugehen, dass es um die Gesundheitskompetenz in Deutschland zum Zeitpunkt der Pandemie nicht gut bestellt war. Erste Studien zur coronaspezifischen Gesundheitskompetenz kamen alsbald zu ähnlichen Ergebnissen . Für die NAP-GK-Expertengruppe bedeutete dies, dass die Förderung von Gesundheitskompetenz und die Umsetzung des NAP-GK intensiviert werden mussten. Doch hatten sich die Umsetzungsbedingungen durch die Corona-Pandemie und ihre Begleiterscheinungen wie Kontaktbeschränkungen etc. völlig verändert. Vor allem die Workshopstrategie und die Kooperation mit Gremien und Arbeitsgruppen war kaum noch realisierbar, weil keine persönlichen Zusammenkünfte mehr möglich waren. Daher wurde die Umsetzungsstrategie geändert: Statt Workshops wurde vor allem auf Positionspapiere und Policy Briefs gesetzt, die kooperativ in unterschiedlichen Teams digital erarbeitet wurden. Es entstanden insgesamt fünf solcher Dokumente. Parallel bildete sich in dieser Zeit eine internationale Arbeitsgruppe zu „Health Literacy Policies“ und deren Implementation im „WHO Action Network on Measuring Population and Organizational Health Literacy“ (M-POHL), an der Vertreter:innen des NAP-GK mitwirkten und zur Entstehung eines entsprechenden Policy Guides beitrugen . Zum Ende der Pandemie hin wurde deutlich, dass eine abermalige Modifikation der Umsetzungsstrategie des NAP-GK erforderlich ist. Sie wird aktuell diskutiert und zusammen mit dem Deutschen Netzwerk für Gesundheitskompetenz (DNGK) erstellt. Insgesamt ist es in den sieben Jahren des Bestehens des NAP-GK gelungen, viele der Empfehlungen in die öffentliche Diskussion zu tragen und zahlreiche Impulse zum Agenda-Setting und zur Initiierung von Projekten zur Förderung von Gesundheitskompetenz zu setzen. Die wichtigsten mit dem NAP-GK und seiner Entstehungs- und Umsetzungsstrategie gesammelten Erfahrungen sollen abschließend dargestellt und summierend reflektiert werden. 1. Zivilgesellschaftliche Initiative. Zu den Besonderheiten des deutschen NAP-GK gehört, dass er nicht durch eine von der Regierung eingesetzte Kommission erarbeitet wurde, sondern von einer Gruppe ausgewiesener und von der Wichtigkeit des Themas überzeugter Akteur:innen. Er kann somit als Beispiel dafür gelten, dass zivilgesellschaftliche Initiativen entscheidend zur Entwicklung neuer Politikfelder und zum Kapazitätsaufbau eines neuen gesundheitlich relevanten Themas beitragen können. Zugleich hat diese Konstruktion Vor- und Nachteile. Einerseits war auf diese Weise die Unabhängigkeit und Neutralität gewährleistet, ebenso hohe Motivation. Andererseits birgt diese Konstruktion das Risiko politischer Distanz und schwacher politischer Resonanz und Unterstützung in sich. Sie konnte durch die frühe Übernahme der Schirmherrschaft des Bundesgesundheitsministers über den NAP-GK umgangen werden. Diese verlieh dem NAP-GK besonders in der Anfangszeit eine gewisse politische Legitimität und ein entsprechendes Ansehen. Mit den sich in der Folgezeit vollziehenden Ministerwechseln veränderte sich dies. Doch war der NAP-GK inzwischen zu einer festen Größe geworden, was sich u. a. daran zeigt, dass er in einem kürzlich erschienenen Bericht der Organisation für wirtschaftliche Zusammenarbeit und Entwicklung (OECD), in dem anhand ausgewählter Länderbeispiele unterschiedliche Initiativen zur Verbesserung der Gesundheitskompetenz dargestellt werden, explizit aufgegriffen wurde . 2. Kooperativer Ansatz und Klärung von Missverständnissen. Der kooperative Ansatz bei der Entwicklung und Umsetzung hat wesentlich zum Erfolg des NAP-GK beigetragen: Er ermöglichte die Einbindung vieler wichtiger Akteur:innen, Stakeholder und Interessengruppen. Dadurch gelang es, das Thema in der Fachdiskussion zu platzieren und ihm Akzeptanz zu verschaffen. Zugleich trug dieser Ansatz zur Ausräumung anfänglich bestehender Vorbehalte gegenüber dem Konzept der Gesundheitskompetenz bei. Denn vor allem die Übersetzung von „Health Literacy“ mit „Gesundheitskompetenz“ stieß bei den Beteiligten zunächst auf Missverständnisse, weil sie mit der seit Jahren intensiv geführten Debatte über Gesundheitsförderung und auch über den Kompetenzbegriff und die Kompetenzmessung kollidierte und umfangreiche Klärungsbemühungen erforderte. So wurde beispielsweise der Begriff Gesundheitskompetenz zu Beginn vielfach mit Gesundheitsförderung und der Fähigkeit, sich gesundheitsbewusst zu verhalten, gleichgesetzt. Der für Gesundheitskompetenz zentrale Aspekt – die Fähigkeit zum Umgang mit gesundheitsrelevanten Informationen – wurde vielfach übersehen. Solche Passungsprobleme mit etablierten Konzepten und daraus erwachsener Klärungsbedarf sind keine Seltenheit bei der Einführung von neuen Konzepten und stellen keine nationale Besonderheit dar . Auch international stieß das Konzept zunächst auf Irritation und teilweise auch Aversion . Ähnliches gilt für die den populationsorientierten Studien zugrundeliegende Erhebungsmethodik. Auch sie wurde zunächst (und wird) kritisch betrachtet. Beobachtet werden konnte, dass sich viele solcher Passungsprobleme im Lauf der Zeit abschliffen und auch die Widerstände gegen eine Adoption des Konzepts und der Erhebungsmethodik nachließen. Dennoch bedürfen sie nach wie vor der Beachtung, denn die bestehenden Divergenzen und Nahtstellen zu angrenzenden Konzepten oder die Auseinandersetzung über unterschiedliche Erhebungsmethoden sind bisher keineswegs beendet (z. B. ). 3. Umsetzung frühzeitig mitdenken. Insgesamt zeigen die Erfahrungen mit dem NAP-GK, dass es nicht reicht, Aktionspläne und/oder andere Strategien zur Förderung von Gesundheitskompetenz zu erarbeiten, sondern wichtig ist, den sich anschließenden Schritt – die Umsetzung – frühzeitig mitzudenken und zu planen. Dies geschieht weitaus zu wenig. Egal, ob bei Aktionsplänen, innovativen Politikstrategien, Gesetzen oder Empfehlungen aus Expertenkommissionen – oft wird auf naturwüchsige Umsetzung gesetzt. Doch ist diese Strategie selten erfolgreich und führt zu zahlreichen Umsetzungsdefiziten, wie auch die Literatur aufzeigt (ex. ). Die mit dem NAP-GK gesammelten Erfahrungen bestätigen dies einmal mehr und verdeutlichen, dass vorliegende implementationswissenschaftliche Erkenntnisse offenbar noch größere Beachtung erfahren müssen. Betrachtet man die der NAP-GK Umsetzung zugrundliegenden Schritte, zeigen die gesammelten Erfahrungen, wie essenziell die Schritte Diffusion und Dissemination sind, um neue Themen und Programme bekannt zu machen und um Umsetzungsinteresse und -motivation zu erzeugen. Zugleich wurde retrospektiv deutlich, dass beide Schritte, obschon sehr viel Energie auf sie verwendet wurde, intensivierungsbedürftig sind. Wie Studien mittlerweile zeigen, ist die Bekanntheit des Gesundheitskompetenz-Konzepts, z. B. bei den Gesundheitsprofessionen und damit bei wichtigen Instanzen der Förderung von Gesundheitskompetenz, nach wie vor begrenzt , sodass es weiterer und breitenwirksamerer Aufklärung bedarf. 4. Kooperative Workshops. Der dritte Schritt, die Implementation durch kooperative Workshops zur Operationalisierung der Empfehlungen des NAP-GK mit Politiker:innen und Stakeholdern bildete das Kernstück der Umsetzung. Auch er hat sich als erfolgreich erwiesen, wie allein die Resonanz in den Feedback-Runden und auch die unerwartete Identifikation mit den Strategiepapieren gezeigt haben. Die Durchführung der Workshops und insbesondere die Rekrutierung der Teilnehmenden entpuppten sich zuweilen allerdings als sehr aufwändig und anspruchsvoll – aus terminlichen und organisatorischen Gründen. Ähnlich war es auch während der Corona-Pandemie bei den digitalen Meetings zur Erarbeitung der Policy Briefs bzw. Positionspapiere. Hinzu kam, dass Repräsentanten der Politik nur schwer für eine Teilnahme zu gewinnen waren. Stakeholder machten daher die Mehrheit der Teilnehmenden aus. Zudem zeigte sich bald, dass einmalige Workshops nicht ausreichen und lediglich die Funktion eines „Appetizers“ einnehmen. Sie können die Innovations- und Implementationsbereitschaft zwar anregen, aber keine zeitstabilen Effekte erzeugen. Wichtig ist deshalb – wie die Erfahrungen bestätigten – ein langfristig angelegtes, prozessuales Vorgehen mit Wiederholungsschlaufen und Vertiefungen in unterschiedlichen Formaten, zumal auch die Stimulierung und Umsetzung von Innovationen im Wirkungsfeld der Teilnehmenden selten reibungslos verläuft und Zeit benötigt . Zugleich sollte der kooperative Ansatz bei allen Bemühungen unbedingt beibehalten werden. Er ist zwar aufwändig, hat sich jedoch bewährt und als wirkungsvoll erwiesen. Das gilt auch für die Kooperation mit Gremien, Arbeitsgruppen und Netzwerken. 5. Stärker einzubeziehende Bereiche. Die Dissemination und Implementation des NAP-GK hat sich vornehmlich auf das Gesundheitssystem konzentriert. Das Erziehungs- und Bildungssystem und weitere wichtige Bereiche (Kommune, Ernährung) wurden tendenziell zu wenig einbezogen. Ursache dafür ist, dass dies Ressourcen verlangt, die dem NAP-GK-Projekt nicht zur Verfügung standen. Die Themen „vulnerable Gruppen“ und damit einhergehend „Vermeidung von Ungleichheit“ haben zwar eine bedeutende Rolle bei der Implementation gespielt – allein vier von den neun Workshops widmeten sich einzelnen vulnerablen Gruppen und Ungleichheit wurde wiederum zum wichtigen Querschnittthema erhoben – dennoch sollten auch sie künftig noch intensiver aufgegriffen werden. 6. Monitoring ermöglichen. Wünschenswert wäre zudem eine formative Evaluation zur Ermittlung der Wirkungen des NAP-GK gewesen. Auch sie setzt allerdings mehr Ressourcen und Kapazitäten voraus. Gleichzeitig können vorliegende Bevölkerungsbefragungen zur Evaluation genutzt werden, insbesondere wenn sie als regelmäßiges Monitoring angelegt sind – wie auch von der WHO betont wird . Ein solches Monitoring zu ermöglichen, ist nach wie vor eine der Zukunft vorbehaltene Aufgabe. Denn bislang liegen in Deutschland kaum Wiederholungsbefragungen zur Gesundheitskompetenz vor, die vergleichende Trendanalysen erlauben und zur Evaluation herangezogen werden können. Die wenigen vorliegenden vergleichbaren populationsorientierten Daten deuten jedoch bereits an, dass ein solches Monitoring wichtig sein dürfte, um Veränderungen der Gesundheitskompetenz im Zeitverlauf beobachten und Fördermaßnahmen bewerten zu können. Zu den Besonderheiten des deutschen NAP-GK gehört, dass er nicht durch eine von der Regierung eingesetzte Kommission erarbeitet wurde, sondern von einer Gruppe ausgewiesener und von der Wichtigkeit des Themas überzeugter Akteur:innen. Er kann somit als Beispiel dafür gelten, dass zivilgesellschaftliche Initiativen entscheidend zur Entwicklung neuer Politikfelder und zum Kapazitätsaufbau eines neuen gesundheitlich relevanten Themas beitragen können. Zugleich hat diese Konstruktion Vor- und Nachteile. Einerseits war auf diese Weise die Unabhängigkeit und Neutralität gewährleistet, ebenso hohe Motivation. Andererseits birgt diese Konstruktion das Risiko politischer Distanz und schwacher politischer Resonanz und Unterstützung in sich. Sie konnte durch die frühe Übernahme der Schirmherrschaft des Bundesgesundheitsministers über den NAP-GK umgangen werden. Diese verlieh dem NAP-GK besonders in der Anfangszeit eine gewisse politische Legitimität und ein entsprechendes Ansehen. Mit den sich in der Folgezeit vollziehenden Ministerwechseln veränderte sich dies. Doch war der NAP-GK inzwischen zu einer festen Größe geworden, was sich u. a. daran zeigt, dass er in einem kürzlich erschienenen Bericht der Organisation für wirtschaftliche Zusammenarbeit und Entwicklung (OECD), in dem anhand ausgewählter Länderbeispiele unterschiedliche Initiativen zur Verbesserung der Gesundheitskompetenz dargestellt werden, explizit aufgegriffen wurde . Der kooperative Ansatz bei der Entwicklung und Umsetzung hat wesentlich zum Erfolg des NAP-GK beigetragen: Er ermöglichte die Einbindung vieler wichtiger Akteur:innen, Stakeholder und Interessengruppen. Dadurch gelang es, das Thema in der Fachdiskussion zu platzieren und ihm Akzeptanz zu verschaffen. Zugleich trug dieser Ansatz zur Ausräumung anfänglich bestehender Vorbehalte gegenüber dem Konzept der Gesundheitskompetenz bei. Denn vor allem die Übersetzung von „Health Literacy“ mit „Gesundheitskompetenz“ stieß bei den Beteiligten zunächst auf Missverständnisse, weil sie mit der seit Jahren intensiv geführten Debatte über Gesundheitsförderung und auch über den Kompetenzbegriff und die Kompetenzmessung kollidierte und umfangreiche Klärungsbemühungen erforderte. So wurde beispielsweise der Begriff Gesundheitskompetenz zu Beginn vielfach mit Gesundheitsförderung und der Fähigkeit, sich gesundheitsbewusst zu verhalten, gleichgesetzt. Der für Gesundheitskompetenz zentrale Aspekt – die Fähigkeit zum Umgang mit gesundheitsrelevanten Informationen – wurde vielfach übersehen. Solche Passungsprobleme mit etablierten Konzepten und daraus erwachsener Klärungsbedarf sind keine Seltenheit bei der Einführung von neuen Konzepten und stellen keine nationale Besonderheit dar . Auch international stieß das Konzept zunächst auf Irritation und teilweise auch Aversion . Ähnliches gilt für die den populationsorientierten Studien zugrundeliegende Erhebungsmethodik. Auch sie wurde zunächst (und wird) kritisch betrachtet. Beobachtet werden konnte, dass sich viele solcher Passungsprobleme im Lauf der Zeit abschliffen und auch die Widerstände gegen eine Adoption des Konzepts und der Erhebungsmethodik nachließen. Dennoch bedürfen sie nach wie vor der Beachtung, denn die bestehenden Divergenzen und Nahtstellen zu angrenzenden Konzepten oder die Auseinandersetzung über unterschiedliche Erhebungsmethoden sind bisher keineswegs beendet (z. B. ). Insgesamt zeigen die Erfahrungen mit dem NAP-GK, dass es nicht reicht, Aktionspläne und/oder andere Strategien zur Förderung von Gesundheitskompetenz zu erarbeiten, sondern wichtig ist, den sich anschließenden Schritt – die Umsetzung – frühzeitig mitzudenken und zu planen. Dies geschieht weitaus zu wenig. Egal, ob bei Aktionsplänen, innovativen Politikstrategien, Gesetzen oder Empfehlungen aus Expertenkommissionen – oft wird auf naturwüchsige Umsetzung gesetzt. Doch ist diese Strategie selten erfolgreich und führt zu zahlreichen Umsetzungsdefiziten, wie auch die Literatur aufzeigt (ex. ). Die mit dem NAP-GK gesammelten Erfahrungen bestätigen dies einmal mehr und verdeutlichen, dass vorliegende implementationswissenschaftliche Erkenntnisse offenbar noch größere Beachtung erfahren müssen. Betrachtet man die der NAP-GK Umsetzung zugrundliegenden Schritte, zeigen die gesammelten Erfahrungen, wie essenziell die Schritte Diffusion und Dissemination sind, um neue Themen und Programme bekannt zu machen und um Umsetzungsinteresse und -motivation zu erzeugen. Zugleich wurde retrospektiv deutlich, dass beide Schritte, obschon sehr viel Energie auf sie verwendet wurde, intensivierungsbedürftig sind. Wie Studien mittlerweile zeigen, ist die Bekanntheit des Gesundheitskompetenz-Konzepts, z. B. bei den Gesundheitsprofessionen und damit bei wichtigen Instanzen der Förderung von Gesundheitskompetenz, nach wie vor begrenzt , sodass es weiterer und breitenwirksamerer Aufklärung bedarf. Der dritte Schritt, die Implementation durch kooperative Workshops zur Operationalisierung der Empfehlungen des NAP-GK mit Politiker:innen und Stakeholdern bildete das Kernstück der Umsetzung. Auch er hat sich als erfolgreich erwiesen, wie allein die Resonanz in den Feedback-Runden und auch die unerwartete Identifikation mit den Strategiepapieren gezeigt haben. Die Durchführung der Workshops und insbesondere die Rekrutierung der Teilnehmenden entpuppten sich zuweilen allerdings als sehr aufwändig und anspruchsvoll – aus terminlichen und organisatorischen Gründen. Ähnlich war es auch während der Corona-Pandemie bei den digitalen Meetings zur Erarbeitung der Policy Briefs bzw. Positionspapiere. Hinzu kam, dass Repräsentanten der Politik nur schwer für eine Teilnahme zu gewinnen waren. Stakeholder machten daher die Mehrheit der Teilnehmenden aus. Zudem zeigte sich bald, dass einmalige Workshops nicht ausreichen und lediglich die Funktion eines „Appetizers“ einnehmen. Sie können die Innovations- und Implementationsbereitschaft zwar anregen, aber keine zeitstabilen Effekte erzeugen. Wichtig ist deshalb – wie die Erfahrungen bestätigten – ein langfristig angelegtes, prozessuales Vorgehen mit Wiederholungsschlaufen und Vertiefungen in unterschiedlichen Formaten, zumal auch die Stimulierung und Umsetzung von Innovationen im Wirkungsfeld der Teilnehmenden selten reibungslos verläuft und Zeit benötigt . Zugleich sollte der kooperative Ansatz bei allen Bemühungen unbedingt beibehalten werden. Er ist zwar aufwändig, hat sich jedoch bewährt und als wirkungsvoll erwiesen. Das gilt auch für die Kooperation mit Gremien, Arbeitsgruppen und Netzwerken. Die Dissemination und Implementation des NAP-GK hat sich vornehmlich auf das Gesundheitssystem konzentriert. Das Erziehungs- und Bildungssystem und weitere wichtige Bereiche (Kommune, Ernährung) wurden tendenziell zu wenig einbezogen. Ursache dafür ist, dass dies Ressourcen verlangt, die dem NAP-GK-Projekt nicht zur Verfügung standen. Die Themen „vulnerable Gruppen“ und damit einhergehend „Vermeidung von Ungleichheit“ haben zwar eine bedeutende Rolle bei der Implementation gespielt – allein vier von den neun Workshops widmeten sich einzelnen vulnerablen Gruppen und Ungleichheit wurde wiederum zum wichtigen Querschnittthema erhoben – dennoch sollten auch sie künftig noch intensiver aufgegriffen werden. Wünschenswert wäre zudem eine formative Evaluation zur Ermittlung der Wirkungen des NAP-GK gewesen. Auch sie setzt allerdings mehr Ressourcen und Kapazitäten voraus. Gleichzeitig können vorliegende Bevölkerungsbefragungen zur Evaluation genutzt werden, insbesondere wenn sie als regelmäßiges Monitoring angelegt sind – wie auch von der WHO betont wird . Ein solches Monitoring zu ermöglichen, ist nach wie vor eine der Zukunft vorbehaltene Aufgabe. Denn bislang liegen in Deutschland kaum Wiederholungsbefragungen zur Gesundheitskompetenz vor, die vergleichende Trendanalysen erlauben und zur Evaluation herangezogen werden können. Die wenigen vorliegenden vergleichbaren populationsorientierten Daten deuten jedoch bereits an, dass ein solches Monitoring wichtig sein dürfte, um Veränderungen der Gesundheitskompetenz im Zeitverlauf beobachten und Fördermaßnahmen bewerten zu können. Angesichts anhaltender alter und neuer Krisen mit ihren unterschiedlichen gesellschaftlichen und gesundheitlichen Auswirkungen und Ungewissheiten steigt die Bedeutung von Gesundheitskompetenz weiter. Sah es zunächst so aus, als würde Gesundheitskompetenz rasch zu einem neuen Politikfeld werden, sind inzwischen Zweifel angebracht. Zwar stößt das Thema Gesundheitskompetenz in der Wissenschaft und auch in der Praxis mittlerweile auf bemerkenswerte Aufmerksamkeit. Auf politischer Ebene entspricht die Resonanz jedoch nicht der gesellschaftlichen Bedeutung des Themas, obschon geringe Gesundheitskompetenz in Deutschland nach wie vor ein nicht zu unterschätzendes Public-Health-Problem darstellt, das große Teile der Bevölkerung betrifft. Daher bleibt es eine zentrale Aufgabe, sich weiterhin für ein nachhaltiges Agenda-Setting und eine systematische Interventions- und Forschungsentwicklung zur Gesundheitskompetenz mit ausreichenden finanziellen Ressourcen zu engagieren. Um dies anzustoßen und die weitere Arbeit mit dem Nationalen Aktionsplan auf eine noch breitere Basis zu stellen, ist die zentrale Koordination ab 2025 von der Universität Bielefeld und der Hertie School, Berlin, an das Deutsche Netzwerk für Gesundheitskompetenz übergegangen. |
Adoption of technology enabled care to support the management of children and teenagers in rheumatology services: a protocol for a mixed-methods systematic review | 0afa558a-c7d4-4296-ac41-43c698841fb4 | 10882409 | Internal Medicine[mh] | Context Children and young people (CYP) cared for in paediatric rheumatology services experience inflammatory conditions, with up to two-thirds of CYP continuing with active disease into adulthood. CYP experience inflammatory conditions such as juvenile idiopathic arthritis (JIA), and a range of non-inflammatory chronic musculoskeletal conditions. Consequently, CYP can have painful, stiff, swollen joints and reduced joint function, requiring continuous outpatient assessment and treatment. These chronic conditions cannot be cured and many CYP require long-term management with cytotoxic disease-modifying antirheumatic drugs. CYP with ongoing rheumatological conditions require multi-disciplinary specialist tertiary care, often necessitating long distances to travel, affecting school attendance and parental time off work. Considering these challenges, one potential method of increasing service efficiencies and decreasing the burdensome effect for families within paediatric rheumatology may be to provide more services remotely by optimising the use of healthcare technologies. Gap in knowledge Although some clinical teams were providing remote services (eg, via telephone clinics) prior to COVID-19 alongside normal practice, COVID-19 enforced a steep rise in remote monitoring. During this time National Health Service (NHS) Trusts in the UK strived to reduce face-to-face appointments to 20%. Furthermore, stipulations within NHS England and NHS Improvement (now NHS Impact) required healthcare providers to minimise routine visits, resulting in a rapid move to remote management with limited training for safe implementation. Therefore, there has been considerable variability in technology enabled care (TEC) used to support such rapid changes, with pre-COVID-19 evidence being criticised for not considering patients’ and professionals’ needs, or not being methodologically robust. The adoption (integration and use of a new technology in a workplace ) of caring for patients using TEC is likely to continue. It has been argued that despite having made advances in TEC there is a risk that outpatient services may default back to traditional face-to-face appointments for all outpatient appointments for all patients, rather than capitalising on what COVID-19 taught us in terms of TEC. However, there is a firm commitment in the UK, demonstrated by various governmental publications, to drive technology-based service provision. UK national plans are outlined in the NHS Blueprint and have been built on the NHS Long Term Plan to ensure TEC is provided in a ‘modern way’ to retain what has been argued to be valuable adaptations to NHS services for some patients during COVID-19. Most recently the importance of exploring and implementing technological advances and treatments has been set out as a priority area for the strategic direction of travel in the UK in retaining our NHS workforce and promoting digital literacy for nurses as the largest profession within the workforce. Furthermore, there is a drive to lead, retain and streamline healthcare technology innovations across Europe and to work together globally. There has been some empirical work published describing the changing landscape of remote care. Areas of paediatric focus have been within cardiology, children with complex needs and palliative care. Focusing on paediatric rheumatology, one survey found face-to-face consults more acceptable to parents, with other findings reporting on financial cost-savings for patients but not necessarily for healthcare providers when moving to remote services using TEC. Two international surveys investigating the changes in practice brought by COVID-19 found clinicians reported that patients became more accepting of using smartphones for telemedicine appointments over time and most centres surveyed in the USA (16/18) were using telemedicine for 75%–100% of visits at the height of the pandemic. Objectives Although two systematic reviews have been published on the usability and effectiveness of electronic (e)-health and mobile (m)-health interventions for patients with juvenile idiopathic arthritis, these reviews were limited to studies only reporting on empirical research using quantitative methodologies. To our knowledge, this will be the first mixed-methods systematic review inclusive of all conditions seen within paediatric rheumatology services, all research methodologies, opinion, reviews and grey literature, into the adoption of TEC in the paediatric rheumatology setting. Our review will synthesise all types of papers and reports on the implementation factors. Aim and research questions The overall aim is to provide a comprehensive understanding and evaluation of key factors affecting the adoption of healthcare technologies by children and teenagers in rheumatology services by answering the following research questions: Are healthcare technologies being used to support TEC? If so, how are healthcare technologies being used within TEC? Children and young people (CYP) cared for in paediatric rheumatology services experience inflammatory conditions, with up to two-thirds of CYP continuing with active disease into adulthood. CYP experience inflammatory conditions such as juvenile idiopathic arthritis (JIA), and a range of non-inflammatory chronic musculoskeletal conditions. Consequently, CYP can have painful, stiff, swollen joints and reduced joint function, requiring continuous outpatient assessment and treatment. These chronic conditions cannot be cured and many CYP require long-term management with cytotoxic disease-modifying antirheumatic drugs. CYP with ongoing rheumatological conditions require multi-disciplinary specialist tertiary care, often necessitating long distances to travel, affecting school attendance and parental time off work. Considering these challenges, one potential method of increasing service efficiencies and decreasing the burdensome effect for families within paediatric rheumatology may be to provide more services remotely by optimising the use of healthcare technologies. Although some clinical teams were providing remote services (eg, via telephone clinics) prior to COVID-19 alongside normal practice, COVID-19 enforced a steep rise in remote monitoring. During this time National Health Service (NHS) Trusts in the UK strived to reduce face-to-face appointments to 20%. Furthermore, stipulations within NHS England and NHS Improvement (now NHS Impact) required healthcare providers to minimise routine visits, resulting in a rapid move to remote management with limited training for safe implementation. Therefore, there has been considerable variability in technology enabled care (TEC) used to support such rapid changes, with pre-COVID-19 evidence being criticised for not considering patients’ and professionals’ needs, or not being methodologically robust. The adoption (integration and use of a new technology in a workplace ) of caring for patients using TEC is likely to continue. It has been argued that despite having made advances in TEC there is a risk that outpatient services may default back to traditional face-to-face appointments for all outpatient appointments for all patients, rather than capitalising on what COVID-19 taught us in terms of TEC. However, there is a firm commitment in the UK, demonstrated by various governmental publications, to drive technology-based service provision. UK national plans are outlined in the NHS Blueprint and have been built on the NHS Long Term Plan to ensure TEC is provided in a ‘modern way’ to retain what has been argued to be valuable adaptations to NHS services for some patients during COVID-19. Most recently the importance of exploring and implementing technological advances and treatments has been set out as a priority area for the strategic direction of travel in the UK in retaining our NHS workforce and promoting digital literacy for nurses as the largest profession within the workforce. Furthermore, there is a drive to lead, retain and streamline healthcare technology innovations across Europe and to work together globally. There has been some empirical work published describing the changing landscape of remote care. Areas of paediatric focus have been within cardiology, children with complex needs and palliative care. Focusing on paediatric rheumatology, one survey found face-to-face consults more acceptable to parents, with other findings reporting on financial cost-savings for patients but not necessarily for healthcare providers when moving to remote services using TEC. Two international surveys investigating the changes in practice brought by COVID-19 found clinicians reported that patients became more accepting of using smartphones for telemedicine appointments over time and most centres surveyed in the USA (16/18) were using telemedicine for 75%–100% of visits at the height of the pandemic. Although two systematic reviews have been published on the usability and effectiveness of electronic (e)-health and mobile (m)-health interventions for patients with juvenile idiopathic arthritis, these reviews were limited to studies only reporting on empirical research using quantitative methodologies. To our knowledge, this will be the first mixed-methods systematic review inclusive of all conditions seen within paediatric rheumatology services, all research methodologies, opinion, reviews and grey literature, into the adoption of TEC in the paediatric rheumatology setting. Our review will synthesise all types of papers and reports on the implementation factors. The overall aim is to provide a comprehensive understanding and evaluation of key factors affecting the adoption of healthcare technologies by children and teenagers in rheumatology services by answering the following research questions: Are healthcare technologies being used to support TEC? If so, how are healthcare technologies being used within TEC? Study design A mixed-methods systematic review incorporating quantitative, qualitative, mixed methods and grey literature will be conducted. The protocol has been developed in accordance with the Preferred Reporting Items for Systematic Review and Meta-Analysis Protocols (PRISMA-P) checklist (see ). In using a parallel-results convergent synthesis design, studies reporting on quantitative and qualitative data will be analysed separately and in parallel (ie, the results of one type of methodological approach will not inform the other). Findings of each synthesis will be compared using a narrative approach. See for design process. Significant amendments to the protocol will be updated in PROSPERO and within published review findings. This review will be used to inform future work packages of a doctoral programme of study. The broad definition of ‘Technology Enabled Care’ has been purposefully used at this early stage to shape and focus prospective, related projects. The definition used for the completion of this review, as described by Norwegian colleagues for a systematic review into patient experiences with TEC across healthcare settings at the beginning of the COVID-19 pandemic, will be: 10.1136/bmjopen-2023-082515.supp1 Supplementary data Telecare, telehealth, telemedicine, mobile (m)-, digital- and electronic (e)-health services. (p779) Eligibility criteria Studies meeting all inclusion criteria will be included in the review. Inclusion criteria are presented here. Population To be included, studies must provide data or information related to children and teenagers aged 11–17 years inclusive, who are cared for within paediatric rheumatology services. Intervention All interventions need to be patient facing healthcare technologies that come under the defined TEC umbrella (definition provided by Leonardsen et al in the methodology section). Examples include: Personal digital assistants such as mobile phones or wearable devices. Innovative technologies being brought into the clinical environment (eg, patient-reported outcome instruments used to monitor symptoms). Digital technologies being used or discussed within clinical consultations (eg, digital pedometers). Examples of TEC interventions that would be excluded are those used by healthcare professionals only (eg, digital imaging) and those that can only be used within the physical hospital setting (eg, specialist equipment used for capillaroscopy). Articles will also be excluded if TEC refers only to the method of gathering data (eg, telephone interviews or electronic diaries used as a data capture method and not as part of the overall study aims/objectives). Control All eligible papers will be included regardless of whether they have a control or comparison group. Outcomes For inclusion into the review, study outcomes need to report on at least one of the main outcomes in relation to TEC: Health-related quality of life. Patient satisfaction with paediatric rheumatology services. Adoption of TEC in paediatric rheumatology services. English language articles (inclusive of conference papers, abstracts, posters, theses) or those with English translation, will be included. All countries will be considered. Search strategy Searches were planned to find published and unpublished studies and reports in the following bibliographic databases: Medline (Ovid), Embase (Ovid), CINAHL (EbscoHost), Core Collection (Web of Science), Epistemonikos, PsycInfo (Ovid), HMIC (Ovid). The aim of including grey (unpublished) literature resources is to reduce the risk of publication bias and for inclusion of anticipated small projects or patient involvement work. Although included grey literature is unlikely to be as robust as empirical research papers, the approach taken for the overall project is to be as inclusive as possible and to capture all related projects. For grey literature, the following resources will be searched: E-Theses Online Service (EThOS), Social Care Online, Google Scholar, UK Child Health Technology Conference, NIHR Children and Young People MedTech, International Clinical Trials Registry Platform, in addition to associated paediatric rheumatology charity websites: Lupus UK, Versus Arthritis (VA), National Rheumatoid Arthritis Society, Childrens Chronic Arthritis Association, Juvenile Arthritis Research, Scottish National Arthritis for Children, Olivia’s Vision, Arthur’s Place, Myositis UK. Finally, the British Society of Rheumatology (BSR) and Barbara Ansell National Network for Adolescent Rheumatology will also be searched. The searches will include index terms, synonyms and alternative phrases for the following search concepts: ‘paediatric rheumatology’, ‘aged 11–17 years inclusive’, and ‘technology enabled care interventions’. See for the search strategy. 10.1136/bmjopen-2023-082515.supp2 Supplementary data Reference lists of eligible studies and review articles will be scrutinised and key global researchers in the field will be contacted for any clarifications as required. The search strategy is not restricted by language, year of publication or geographic location. The searching process took place from June 2023 to September 2023 and the screening of titles and abstracts during October 2023. The planned end date for the study is April 2024. Screening and data extraction All studies retrieved from the search will be downloaded into EndNote ( https://endnote.com/ ) to store and organise references. Following duplicate removal, papers will be exported to Covidence online systematic review software ( https://www.covidence.org/ ) for screening of titles and abstracts. HR will independently undertake screening of all papers (titles and abstracts). PL, BD, and AWG will independently screen a selection (approximately a third each) of the papers (titles and abstracts) to ensure that every paper has been independently screened by two authors. In the case of >20% disagreement then review criteria will be discussed between all authors and refined until the consensus threshold is reached. All papers requiring further information to assess criteria will be included. Qualifying papers will be included for full text review. HR and PL will each independently review half of the full text papers. In cases of conflict, HR and PL will meet to discuss and in cases of uncertainty and if in disagreement, discussion will also take place with BD and AWG. Included study authors will be contacted if necessary if further clarification is required. Covidence software will be used to undertake data extraction using a bespoke form within Covidence software. HR and one co-author will together pilot the bespoke data extraction form for the first 10% of studies and agree on data items. Thereon, HR and one co-author who is trained and experienced in extracting and coding data for systematic reviews, will complete the remaining extraction of data (50% each). Authors of individual studies will be contacted via email or ResearchGate ( https://www.researchgate.net/ ) for further information if necessary. In cases of uncertainty around data inclusion, the wider authorship team will be consulted for further direction. The Preferred Reporting Items for Systematic Reviews and Meta-Analyses flow diagram will be used to display the search results. Anticipated data to be extracted where available: Study/paper background details (author, title, journal, year of publication, data collection date, setting details). Population (age, gender, condition, socioeconomic status, ethnicity). Study aim(s) and design. Intervention details (type, aim of use, duration). Factors affecting use/adoptions of intervention. Outcome measures employed. Outcomes and main findings. Quality assessment The Mixed Methods Appraisal Tool (MMAT) will be used to assess risk of bias and quality for individual papers. The MMAT will be effective as guidance for assessing all five types of study approaches (qualitative, quantitative randomised controlled trials, quantitative non-randomised, quantitative descriptive and mixed methods). MMAT is comprised of two parts following two screening questions used to assess whether papers are empirical work. The first part is a checklist and the second, a helpful explanation of all criteria used to inform checklist answers (part 1). Both parts are divided into five sections (ie, for each of the five study methodological approaches) and each subsection allows the reviewer to rate the paper as ‘yes’, ‘no’ or ‘cannot tell’ in regard to each methodological criterion. The checklist criteria will be applied to all included papers independently by two authors. Any disputes will be discussed together and with a member of the PhD supervisory team if there is no consensus. Grey literature (ie, those papers failing the screening section within MMAT) will be assessed using the Authority, Accuracy, Coverage, Objectivity, Date, Significance (ACCODS) checklist as an evaluation and critical appraisal tool for use with grey literature. ACCODS uses five criteria (Authority, Accuracy, Coverage, Objectivity, Date, Significance) for reviewers to assess the quality of grey literature. The same method will be used in terms of reviewer roles as for the MMAT. Evidence synthesis A parallel-results convergent synthesis design is planned, with independent syntheses of quantitative and qualitative data, followed by comparison of the findings of each synthesis using a narrative approach. This narrative interpretation of the relationship between the two sets of evidence will be reported in the discussion section of the final report, with reference to the main research questions. See for an overview of the review design process. Quantitative synthesis For quantitative data, homogenous studies (minimum of two) reporting on comparable aspects of TEC outcomes (eg, quality of life, satisfaction with the service) will be considered for meta-analysis and standardised mean differences using the RevMan systematic review and meta-analysis tool ( https://revman.cochrane.org/info ). If there are insufficient comparable studies, the ‘synthesis without meta-analysis’ method will be used to guide the quantitative synthesis without meta-analysis. Next, the synthesised data from included quantitative papers will be coded into qualitative categories in accordance with normalisation process theory prior to narrative comparison with qualitative data to report the overall review findings. Qualitative synthesis NVivo data analysis software ( https://lumivero.com/products/nvivo/ ) will be used to facilitate organisation and coding of qualitative data using thematic synthesis, for example, data relating to CYP’s attitudes towards TEC, how technologies have changed or affected their care in rheumatology services. Included data will be extracted from the results, discussion and conclusion sections of relevant articles and on which we will develop themes. Although open, inductive coding will be undertaken initially, data relating to normalisation process theory will also be collected to identify implementation issues and enablers for both qualitative data and qualitatively coded quantitative categories. All coding will be undertaken on the extracted data. Subgroup analyses There are many different diagnoses cared for within paediatric rheumatology services, many of which are very rare. Subgroup analyses are planned for different diagnoses seen within the services as diagnosis may affect different experiences of TEC. Examples of specific research questions include: Is the adoption of TEC different according to patient diagnosis? If so, what are these differences? Do demographic factors impact access to TEC (eg, socioeconomic background or ethnicity)? Is TEC accessible to patients who are visually impaired (eg, patients diagnosed with uveitis or cryopyrin-associated periodic syndromes) or those with hearing loss? Subgroup analyses are also planned according to gender and country in which the original paper recruited participants from. An additional sensitivity analysis will be undertaken exclusive of studies at high risk of bias, where available data will allow. Assessing confidence in cumulative evidence The Confidence in the Evidence from Reviews of Qualitative Research (CERQual) tool will be used to assess confidence in qualitative evidence synthesis. Using CERQual will provide a transparent and systematic assessment of qualitative evidence synthesis. HR and one co-author will independently judge on four CERQual components: methodological limitations of included studies, coherence of the review finding, adequacy of data leading to a review finding and relevance of included studies in respect to the study aims and objectives. Patient and public involvement Patient and public involvement (PPI) is defined as research being carried out ‘with’ or ‘by’ members of the public rather than ‘to’, ‘about’ or ‘for’ them. Rather than being about research participants taking part in a study, PPI is about patients with relevant experience of a condition advising or working alongside researchers to influence the design, conduct or dissemination of a project. Well-conducted PPI adds unique insights from those with a lived experience of the condition under investigation, thereby improving the quality and relevance of research projects, resulting in better recruitment and retention rates. PPI has been and will continue to be an important thread within the review; BD is co-author of this paper, and PPI lead for this review and for the wider PhD project. BD is a patient and young person who was diagnosed with JIA and has been cared for under paediatric rheumatology services for several years. As PPI lead for this review, BD’s role has been to actively contribute to designing and reviewing the protocol by meeting with co-authors regularly to discuss and influence important decisions, for example, in defining intervention and inclusion criteria. Other aspects of BD’s role will be to screen papers, meet with co-authors as required regarding any disagreements over article inclusion, and input into dissemination of results as co-author via journal and/or conference presentation. BD’s involvement will ensure that the study will be employed with a patient perspective at every stage and not just the ideas and wishes of the researchers. Additionally, on 4 May 2023, the lead author (HR) consulted with ‘YourRheum’: a national Young Person’s Advisory Group supported by Versus Arthritis ( https://yourrheum.org/ ). Eight YourRheum members aged 12–24 years who have lived experience of having a rheumatological condition attended. HR presented the wider PhD project and more specifically, the present protocol ideas. YourRheum members reported that they felt TEC was an important topic to research. They had strong feelings that the project, ‘should not just be about JIA, as everything is always just about JIA’, which led to the project being designed with all patients seen within paediatric rheumatology services in mind. Another aspect they felt was important to them, was that the project should seek to understand where and how TEC was happening. Hence, the systematic review was designed as a mixed methods protocol. Feedback was received regarding which review search terms members thought should be used and general thoughts about the wider, prospective PhD project. Beyond the present protocol, YourRheum members will be supporting the development of survey questions, topic guides and interview questions for future related projects, to ensure the project’s focus remains aligned with what matters to patients. Protocol validity and registration In accordance with the guidelines, our mixed-methods systematic review protocol is reported according to the PRISMA-P guidelines and registered with the International Prospective Register of Systematic Review (PROSPERO) on 11 July 2023 and was last updated on 17 July 2023 ( https://www.crd.york.ac.uk/prospero/display_record.php?ID=CRD42023443058 ). A mixed-methods systematic review incorporating quantitative, qualitative, mixed methods and grey literature will be conducted. The protocol has been developed in accordance with the Preferred Reporting Items for Systematic Review and Meta-Analysis Protocols (PRISMA-P) checklist (see ). In using a parallel-results convergent synthesis design, studies reporting on quantitative and qualitative data will be analysed separately and in parallel (ie, the results of one type of methodological approach will not inform the other). Findings of each synthesis will be compared using a narrative approach. See for design process. Significant amendments to the protocol will be updated in PROSPERO and within published review findings. This review will be used to inform future work packages of a doctoral programme of study. The broad definition of ‘Technology Enabled Care’ has been purposefully used at this early stage to shape and focus prospective, related projects. The definition used for the completion of this review, as described by Norwegian colleagues for a systematic review into patient experiences with TEC across healthcare settings at the beginning of the COVID-19 pandemic, will be: 10.1136/bmjopen-2023-082515.supp1 Supplementary data Telecare, telehealth, telemedicine, mobile (m)-, digital- and electronic (e)-health services. (p779) Studies meeting all inclusion criteria will be included in the review. Inclusion criteria are presented here. Population To be included, studies must provide data or information related to children and teenagers aged 11–17 years inclusive, who are cared for within paediatric rheumatology services. Intervention All interventions need to be patient facing healthcare technologies that come under the defined TEC umbrella (definition provided by Leonardsen et al in the methodology section). Examples include: Personal digital assistants such as mobile phones or wearable devices. Innovative technologies being brought into the clinical environment (eg, patient-reported outcome instruments used to monitor symptoms). Digital technologies being used or discussed within clinical consultations (eg, digital pedometers). Examples of TEC interventions that would be excluded are those used by healthcare professionals only (eg, digital imaging) and those that can only be used within the physical hospital setting (eg, specialist equipment used for capillaroscopy). Articles will also be excluded if TEC refers only to the method of gathering data (eg, telephone interviews or electronic diaries used as a data capture method and not as part of the overall study aims/objectives). Control All eligible papers will be included regardless of whether they have a control or comparison group. Outcomes For inclusion into the review, study outcomes need to report on at least one of the main outcomes in relation to TEC: Health-related quality of life. Patient satisfaction with paediatric rheumatology services. Adoption of TEC in paediatric rheumatology services. English language articles (inclusive of conference papers, abstracts, posters, theses) or those with English translation, will be included. All countries will be considered. To be included, studies must provide data or information related to children and teenagers aged 11–17 years inclusive, who are cared for within paediatric rheumatology services. All interventions need to be patient facing healthcare technologies that come under the defined TEC umbrella (definition provided by Leonardsen et al in the methodology section). Examples include: Personal digital assistants such as mobile phones or wearable devices. Innovative technologies being brought into the clinical environment (eg, patient-reported outcome instruments used to monitor symptoms). Digital technologies being used or discussed within clinical consultations (eg, digital pedometers). Examples of TEC interventions that would be excluded are those used by healthcare professionals only (eg, digital imaging) and those that can only be used within the physical hospital setting (eg, specialist equipment used for capillaroscopy). Articles will also be excluded if TEC refers only to the method of gathering data (eg, telephone interviews or electronic diaries used as a data capture method and not as part of the overall study aims/objectives). All eligible papers will be included regardless of whether they have a control or comparison group. For inclusion into the review, study outcomes need to report on at least one of the main outcomes in relation to TEC: Health-related quality of life. Patient satisfaction with paediatric rheumatology services. Adoption of TEC in paediatric rheumatology services. English language articles (inclusive of conference papers, abstracts, posters, theses) or those with English translation, will be included. All countries will be considered. Searches were planned to find published and unpublished studies and reports in the following bibliographic databases: Medline (Ovid), Embase (Ovid), CINAHL (EbscoHost), Core Collection (Web of Science), Epistemonikos, PsycInfo (Ovid), HMIC (Ovid). The aim of including grey (unpublished) literature resources is to reduce the risk of publication bias and for inclusion of anticipated small projects or patient involvement work. Although included grey literature is unlikely to be as robust as empirical research papers, the approach taken for the overall project is to be as inclusive as possible and to capture all related projects. For grey literature, the following resources will be searched: E-Theses Online Service (EThOS), Social Care Online, Google Scholar, UK Child Health Technology Conference, NIHR Children and Young People MedTech, International Clinical Trials Registry Platform, in addition to associated paediatric rheumatology charity websites: Lupus UK, Versus Arthritis (VA), National Rheumatoid Arthritis Society, Childrens Chronic Arthritis Association, Juvenile Arthritis Research, Scottish National Arthritis for Children, Olivia’s Vision, Arthur’s Place, Myositis UK. Finally, the British Society of Rheumatology (BSR) and Barbara Ansell National Network for Adolescent Rheumatology will also be searched. The searches will include index terms, synonyms and alternative phrases for the following search concepts: ‘paediatric rheumatology’, ‘aged 11–17 years inclusive’, and ‘technology enabled care interventions’. See for the search strategy. 10.1136/bmjopen-2023-082515.supp2 Supplementary data Reference lists of eligible studies and review articles will be scrutinised and key global researchers in the field will be contacted for any clarifications as required. The search strategy is not restricted by language, year of publication or geographic location. The searching process took place from June 2023 to September 2023 and the screening of titles and abstracts during October 2023. The planned end date for the study is April 2024. All studies retrieved from the search will be downloaded into EndNote ( https://endnote.com/ ) to store and organise references. Following duplicate removal, papers will be exported to Covidence online systematic review software ( https://www.covidence.org/ ) for screening of titles and abstracts. HR will independently undertake screening of all papers (titles and abstracts). PL, BD, and AWG will independently screen a selection (approximately a third each) of the papers (titles and abstracts) to ensure that every paper has been independently screened by two authors. In the case of >20% disagreement then review criteria will be discussed between all authors and refined until the consensus threshold is reached. All papers requiring further information to assess criteria will be included. Qualifying papers will be included for full text review. HR and PL will each independently review half of the full text papers. In cases of conflict, HR and PL will meet to discuss and in cases of uncertainty and if in disagreement, discussion will also take place with BD and AWG. Included study authors will be contacted if necessary if further clarification is required. Covidence software will be used to undertake data extraction using a bespoke form within Covidence software. HR and one co-author will together pilot the bespoke data extraction form for the first 10% of studies and agree on data items. Thereon, HR and one co-author who is trained and experienced in extracting and coding data for systematic reviews, will complete the remaining extraction of data (50% each). Authors of individual studies will be contacted via email or ResearchGate ( https://www.researchgate.net/ ) for further information if necessary. In cases of uncertainty around data inclusion, the wider authorship team will be consulted for further direction. The Preferred Reporting Items for Systematic Reviews and Meta-Analyses flow diagram will be used to display the search results. Anticipated data to be extracted where available: Study/paper background details (author, title, journal, year of publication, data collection date, setting details). Population (age, gender, condition, socioeconomic status, ethnicity). Study aim(s) and design. Intervention details (type, aim of use, duration). Factors affecting use/adoptions of intervention. Outcome measures employed. Outcomes and main findings. The Mixed Methods Appraisal Tool (MMAT) will be used to assess risk of bias and quality for individual papers. The MMAT will be effective as guidance for assessing all five types of study approaches (qualitative, quantitative randomised controlled trials, quantitative non-randomised, quantitative descriptive and mixed methods). MMAT is comprised of two parts following two screening questions used to assess whether papers are empirical work. The first part is a checklist and the second, a helpful explanation of all criteria used to inform checklist answers (part 1). Both parts are divided into five sections (ie, for each of the five study methodological approaches) and each subsection allows the reviewer to rate the paper as ‘yes’, ‘no’ or ‘cannot tell’ in regard to each methodological criterion. The checklist criteria will be applied to all included papers independently by two authors. Any disputes will be discussed together and with a member of the PhD supervisory team if there is no consensus. Grey literature (ie, those papers failing the screening section within MMAT) will be assessed using the Authority, Accuracy, Coverage, Objectivity, Date, Significance (ACCODS) checklist as an evaluation and critical appraisal tool for use with grey literature. ACCODS uses five criteria (Authority, Accuracy, Coverage, Objectivity, Date, Significance) for reviewers to assess the quality of grey literature. The same method will be used in terms of reviewer roles as for the MMAT. A parallel-results convergent synthesis design is planned, with independent syntheses of quantitative and qualitative data, followed by comparison of the findings of each synthesis using a narrative approach. This narrative interpretation of the relationship between the two sets of evidence will be reported in the discussion section of the final report, with reference to the main research questions. See for an overview of the review design process. Quantitative synthesis For quantitative data, homogenous studies (minimum of two) reporting on comparable aspects of TEC outcomes (eg, quality of life, satisfaction with the service) will be considered for meta-analysis and standardised mean differences using the RevMan systematic review and meta-analysis tool ( https://revman.cochrane.org/info ). If there are insufficient comparable studies, the ‘synthesis without meta-analysis’ method will be used to guide the quantitative synthesis without meta-analysis. Next, the synthesised data from included quantitative papers will be coded into qualitative categories in accordance with normalisation process theory prior to narrative comparison with qualitative data to report the overall review findings. Qualitative synthesis NVivo data analysis software ( https://lumivero.com/products/nvivo/ ) will be used to facilitate organisation and coding of qualitative data using thematic synthesis, for example, data relating to CYP’s attitudes towards TEC, how technologies have changed or affected their care in rheumatology services. Included data will be extracted from the results, discussion and conclusion sections of relevant articles and on which we will develop themes. Although open, inductive coding will be undertaken initially, data relating to normalisation process theory will also be collected to identify implementation issues and enablers for both qualitative data and qualitatively coded quantitative categories. All coding will be undertaken on the extracted data. Subgroup analyses There are many different diagnoses cared for within paediatric rheumatology services, many of which are very rare. Subgroup analyses are planned for different diagnoses seen within the services as diagnosis may affect different experiences of TEC. Examples of specific research questions include: Is the adoption of TEC different according to patient diagnosis? If so, what are these differences? Do demographic factors impact access to TEC (eg, socioeconomic background or ethnicity)? Is TEC accessible to patients who are visually impaired (eg, patients diagnosed with uveitis or cryopyrin-associated periodic syndromes) or those with hearing loss? Subgroup analyses are also planned according to gender and country in which the original paper recruited participants from. An additional sensitivity analysis will be undertaken exclusive of studies at high risk of bias, where available data will allow. Assessing confidence in cumulative evidence The Confidence in the Evidence from Reviews of Qualitative Research (CERQual) tool will be used to assess confidence in qualitative evidence synthesis. Using CERQual will provide a transparent and systematic assessment of qualitative evidence synthesis. HR and one co-author will independently judge on four CERQual components: methodological limitations of included studies, coherence of the review finding, adequacy of data leading to a review finding and relevance of included studies in respect to the study aims and objectives. For quantitative data, homogenous studies (minimum of two) reporting on comparable aspects of TEC outcomes (eg, quality of life, satisfaction with the service) will be considered for meta-analysis and standardised mean differences using the RevMan systematic review and meta-analysis tool ( https://revman.cochrane.org/info ). If there are insufficient comparable studies, the ‘synthesis without meta-analysis’ method will be used to guide the quantitative synthesis without meta-analysis. Next, the synthesised data from included quantitative papers will be coded into qualitative categories in accordance with normalisation process theory prior to narrative comparison with qualitative data to report the overall review findings. NVivo data analysis software ( https://lumivero.com/products/nvivo/ ) will be used to facilitate organisation and coding of qualitative data using thematic synthesis, for example, data relating to CYP’s attitudes towards TEC, how technologies have changed or affected their care in rheumatology services. Included data will be extracted from the results, discussion and conclusion sections of relevant articles and on which we will develop themes. Although open, inductive coding will be undertaken initially, data relating to normalisation process theory will also be collected to identify implementation issues and enablers for both qualitative data and qualitatively coded quantitative categories. All coding will be undertaken on the extracted data. There are many different diagnoses cared for within paediatric rheumatology services, many of which are very rare. Subgroup analyses are planned for different diagnoses seen within the services as diagnosis may affect different experiences of TEC. Examples of specific research questions include: Is the adoption of TEC different according to patient diagnosis? If so, what are these differences? Do demographic factors impact access to TEC (eg, socioeconomic background or ethnicity)? Is TEC accessible to patients who are visually impaired (eg, patients diagnosed with uveitis or cryopyrin-associated periodic syndromes) or those with hearing loss? Subgroup analyses are also planned according to gender and country in which the original paper recruited participants from. An additional sensitivity analysis will be undertaken exclusive of studies at high risk of bias, where available data will allow. The Confidence in the Evidence from Reviews of Qualitative Research (CERQual) tool will be used to assess confidence in qualitative evidence synthesis. Using CERQual will provide a transparent and systematic assessment of qualitative evidence synthesis. HR and one co-author will independently judge on four CERQual components: methodological limitations of included studies, coherence of the review finding, adequacy of data leading to a review finding and relevance of included studies in respect to the study aims and objectives. Patient and public involvement (PPI) is defined as research being carried out ‘with’ or ‘by’ members of the public rather than ‘to’, ‘about’ or ‘for’ them. Rather than being about research participants taking part in a study, PPI is about patients with relevant experience of a condition advising or working alongside researchers to influence the design, conduct or dissemination of a project. Well-conducted PPI adds unique insights from those with a lived experience of the condition under investigation, thereby improving the quality and relevance of research projects, resulting in better recruitment and retention rates. PPI has been and will continue to be an important thread within the review; BD is co-author of this paper, and PPI lead for this review and for the wider PhD project. BD is a patient and young person who was diagnosed with JIA and has been cared for under paediatric rheumatology services for several years. As PPI lead for this review, BD’s role has been to actively contribute to designing and reviewing the protocol by meeting with co-authors regularly to discuss and influence important decisions, for example, in defining intervention and inclusion criteria. Other aspects of BD’s role will be to screen papers, meet with co-authors as required regarding any disagreements over article inclusion, and input into dissemination of results as co-author via journal and/or conference presentation. BD’s involvement will ensure that the study will be employed with a patient perspective at every stage and not just the ideas and wishes of the researchers. Additionally, on 4 May 2023, the lead author (HR) consulted with ‘YourRheum’: a national Young Person’s Advisory Group supported by Versus Arthritis ( https://yourrheum.org/ ). Eight YourRheum members aged 12–24 years who have lived experience of having a rheumatological condition attended. HR presented the wider PhD project and more specifically, the present protocol ideas. YourRheum members reported that they felt TEC was an important topic to research. They had strong feelings that the project, ‘should not just be about JIA, as everything is always just about JIA’, which led to the project being designed with all patients seen within paediatric rheumatology services in mind. Another aspect they felt was important to them, was that the project should seek to understand where and how TEC was happening. Hence, the systematic review was designed as a mixed methods protocol. Feedback was received regarding which review search terms members thought should be used and general thoughts about the wider, prospective PhD project. Beyond the present protocol, YourRheum members will be supporting the development of survey questions, topic guides and interview questions for future related projects, to ensure the project’s focus remains aligned with what matters to patients. In accordance with the guidelines, our mixed-methods systematic review protocol is reported according to the PRISMA-P guidelines and registered with the International Prospective Register of Systematic Review (PROSPERO) on 11 July 2023 and was last updated on 17 July 2023 ( https://www.crd.york.ac.uk/prospero/display_record.php?ID=CRD42023443058 ). This mixed-methods systematic review of academic papers and grey literature is exempt from ethical approval because there will be no direct contact with patients or participants. A full report will be produced, and the outcome published in a leading journal in this field (eg, BMJ Open or Health and Technology ). The review report will also be submitted to the British Society for Rheumatology conference and/or Paediatric Rheumatology European Society conference. Members of YourRheum will be asked for their creative ideas regarding dissemination. This protocol paper and subsequent review results are part of a wider programme of work that will combine with expert consensus to create guidelines for implementing future TEC approaches. Reviewer comments Author's manuscript |
The Human Glycome Atlas project for cataloging all glycan-related omics data in human | efd63d9d-2f36-487d-aa8d-84af63c05272 | 11729707 | Biochemistry[mh] | In April 2023, the Human Glycome Atlas (HGA) Project was initiated , marking a significant milestone in the realm of life sciences. Led by three prominent Japanese institutions—the Institute for Glyco-core Research (iGCORE) at Nagoya University and Gifu University, the Exploratory Research Center on Life and Living Systems (ExCELLS) within the National Institutes of Natural Sciences, and the Glycan and Life Systems Integration Center (GaLSIC) at Soka University—this pioneering endeavor is the first large-scale academic Frontiers promotion project supported by the Ministry of Education, Culture, Sports, Science and Technology (MEXT) of Japan in the life sciences field. The primary objective of the project is to establish a comprehensive knowledgebase of human glycans and glycoconjugates as well as glycan-related genes and diseases, setting a standard for the human glycome. Over the course of its initial five years, the project plans to implement a high-throughput pipeline aimed at analyzing 20,000 blood samples. Upon completion of this phase, an access-controlled version of a knowledgebase called TOHSA will be released. By its culmination in the tenth year, TOHSA is envisioned to evolve into a central data resource, seamlessly integrating human glycan data with other omics datasets, thereby facilitating a deeper understanding of disease-related mechanisms involving glycans. Upon completion of this project, an open version of TOHSA, without access restrictions, will also be made available for true FAIRness. While this project has just begun, much progress has been made in the first year and a half, with the establishment of standard operating procedures (SOPs) for glycoproteomics, glycomics, cohort sample processing, etc. These data are being collated into a standardized repository developed in-house to ensure that all metadata is stored appropriately. FAIR principles will be upheld throughout, especially when the access-controlled TOHSA knowledgebase is made available. At that time, it is expected that the SOPs will also be made available such that further collaborations with international institutes can be established. Through this HGA project, we aim to provide the foundation for the field to take the next step in life sciences following the Human Genome and Proteome Projects. By providing information in a FAIR manner, the data in TOHSA will be easily integrated with other omics data, including lipidomics, metabolomics and proteomics. As glycosylation often has been overlooked in medicine, this integration should help propel research in the medical and life sciences forward greatly, especially with advances in artificial intelligence. Ultimately, this project will be able to propose a reference human glycome, combined with other information crucial to understand human diversity and disease. The research plan for HGA consists of four segments: Segment 1: Creation of a detailed glycopeptide list (called a reference map) of all human glycoproteins. This will be produced by preemptively performing deep glycoproteomics coupled with glycopeptide fractionation. In addition to the common MS 2 -based identification , this data acquisition will include the results of a high depth identification using the established IGOT-LC/MS method in combination with MS1-based glycopeptide deductions . This map enables a much larger number of glycopeptides to be identified in a single LC/MS analysis than is possible with MS 2 , and it can fully support the cohort studies in Segment 2. Segment 2: Creation of a human glycan catalog. The human glycoprotein reference map will be utilized to perform glycoproteome characterization of large populations. In addition, through a “Total Glycomics” approach , glycan structures of five categories of N -linked glycans, O -linked glycans, glycosphingolipid (GSL)-glycans, glycosaminoglycans, and free oligosaccharides can be simultaneously and quantitatively obtained for each individual. Furthermore, we have established sialic acid linkage-specific alkylamidation (SALSA) to distinguish sialylated glycan isomers by MS analysis and the SALSA method has been introduced into N -glycans, GSL-glycans, and O-glycan analysis. To obtain these glycan structure data, collaborations will be necessary with many biobanks. These individual glycan structures and proteomic data will then be catalogued together with relevant clinical epidemiological information in a highly secure Cohort sample database developed internally. Since glycans vary between individuals and even within the same individual across tissues and conditions, this information will allow the identification of the characteristics of glycans specific to disease, race, age, gender, etc. Segment 3: Construction of a glycan biosynthesis atlas. Approximately 200 kinds of glycosyltransferases are involved in the biosynthesis of human glycans. Comprehensive information on the gene expression, subcellular distribution, and enzymatic properties (activity and substrate specificity) of these enzymes will be obtained to create a highly accurate glycan biosynthesis simulator. Since most of these enzymes are localized within the Golgi apparatus, the molecular network that defines their location within the Golgi will be highlighted based on bioimaging and omics analyses . These findings will provide the basis for the creation of artificial cells with controlled glycosylation. Segment 4: The data obtained in Segments 1–3 will be stored together with clinical and phenotypic data in the knowledgebase TOHSA, and the open access version will have user and application programming interfaces that can be easily accessed by all researchers and shared worldwide. Initially, around year five of this project, due to the nature of the samples, TOHSA will be only made available in an access-controlled manner, after ethical and confidentiality agreements have been made. At this point, we expect to have analyzed and catalogued approximately 10,000 blood samples from a variety of patient cohorts, including Alzheimer’s, dementia and older aged patients. Near the end of this project, an open-access version of TOHSA will also be made available to the public. During this latter half, this project also aims to analyze other organs and tissues. Segments 1–4 will be executed within Research Goal 1 of the HGA project, “Establishment of Glycan Information Infrastructure”. Moreover, there are two other Research Goals under this project, namely, “Establishment of Equipment and Technology Infrastructure” as Research Goal 2 to support these four segments, and “Establishment of Collaborative Infrastructure” as Research Goal 3. This final research goal is of most importance as it will be the foundation for collaborations at the international level. We hope that the standardized protocols established in this project will be easily distributable and applicable to other cohort samples such that a trustworthy and truly worldwide resource for human glycomics can be established. As this information will also be integrated with other omics information including genomics, metabolomics, proteomics, and others, this will have far-reaching impacts and lasting effects on medical research and our understanding of human biology, which could not have been understood without the information from TOHSA. |
Innovative microscope simulator for cataract patients: Enhancing understanding and comfort | c4844d9d-39e0-4d98-8138-cb3793b30c65 | 11329808 | Ophthalmology[mh] | The authors certify that they have obtained all appropriate patient consent forms. In the form, the patient(s) has/have given his/her/their consent for his/her/their images and other clinical information to be reported in the journal. The patients understand that their names and initials will not be published and due efforts will be made to conceal their identity, but anonymity cannot be guaranteed. Nil. There are no conflicts of interest. |
Development of a real-time PCR assay to detect the single nucleotide polymorphism causing Warmblood Fragile Foal Syndrome | c6533177-a4b0-45dc-94ba-58aa5c1718cb | 8575260 | Pathology[mh] | Warmblood fragile foal syndrome (WFFS) is an autosomal recessive genetic condition affecting the connective tissue of Warmblood foals . The disorder is caused by a mutation in procollagen-lysine, 2-oxoglutarate 5-dioxygenase1 (PLOD1) gene . The PLOD1 gene encodes for the enzyme lysyl hydroxylase, which is responsible for converting lysine to hydroxylysine through hydroxylation. The hydroxyl group of hydroxylysine binds to specific galactose mono- and disaccharides that improve collagen structural strength . Affected foals have a homozygous non-synonymous G>A mutation at nucleotide position 2032 of the PLOD1 gene, resulting in a glycine to arginine substitution at amino acid position 678 . Warmblood Fragile Foal Syndrome is associated with a variety of clinical signs including excessive skin fragility, open skin wounds, flexural deformities, deformities of the spinal canal and perforating lesions of the aorta. Hyper-extensibility of joints, often resulting in dystocia, has frequently been reported. Clinical signs are incompatible with extra uterine life and foals die or are euthanised within hours of birth . As the heterozygous genotype has not been associated with clinical abnormalities, the presence of the mutation in adult horses is often not detected until the birth of an affected foal, which can result in considerable economic losses for the breeder. Genetic testing of breeding mares and stallions for the mutation is vital to prevent the unknown mating of WFFS carriers, thus eliminating the risk of the birth of unviable foals. In previous studies carried out in a number of different locations across the world, prevalence of heterozygous carriers has ranged from less than 5% to 17% . Further prevalence studies are required in order to determine the true prevalence of the mutation in the equine population globally. Although typically associated with Warmblood breeds, the mutation has also been identified in non-Warmblood breeds, including the Thoroughbred , Quarter Horse and American Sport Pony , highlighting the importance of genetic testing for the mutation, in clinically suspicious cases, in non-Warmblood breeds as well as Warmblood breeds. Commercial testing of horses can be costly, particularly the testing of large numbers, as is the case when performing prevalence studies. Affordable and readily accessible genetic testing for the WFFS causative mutation is important for the equine industry. Increased use of genetic testing increases our knowledge of this disorder and allows breeders to make more informed breeding decisions. With this in mind, we have developed a quantitative polymerase chain reaction (qPCR) based assay, as an alternative to the two step PCR and Sanger Sequencing assay currently commercially provided by laboratories working under licence to Cornell University . The protocol described in this peer-reviewed article is published on protocols.io, https://dx.doi.org/10.17504/protocols.io.bw4fpgtn , and is included for printing as with this article. Collection of samples Hair samples were collected with consent from a variety of horse and pony breeds. The hair was manually pulled from the mane to ensure as little discomfort to the horse as possible. The hair was sealed in a biosafety bag and stored at room temperature before testing. The study was conducted under ethical exemption from the University College Dublin Animal Research Ethics Committee: (AREC-E-18-47-Duggan). DNA extraction The Qiagen DNA Blood and Tissue kit was used, and the extraction protocol was adapted from the manufacturer’s instructions. Hair strands (10 per extraction) were cut at a maximum length of 1 cm, including the follicular tag. The hair was placed into a 1.5 ml Rnase, DNase-free micro-centrifuge tube followed by 300 μl of ATL Buffer, 20 μl of Proteinase K and 20ul of a 1 M DTT solution. The sample was incubated at 56°C with occasional vortexing until the hair was completely lysed and could not be seen anymore. After incubation, the sample was vortexed for 15 seconds to ensure complete homogenisation before 300 μl AL Buffer was added and it was again briefly vortexed. This was followed by the addition of 300 μl of ethanol (molecular grade, > 99.5% pure) with brief vortexing again to ensure complete mixing. The sample mixture was pipetted into a DNeasy mini spin column and centrifuged at room temperature at 6000 x g for 1 minute. The flow through was discarded and two wash steps were carried out on the column using the same centrifugation conditions as before–firstly 500 μl of AW1 Buffer was added and the column was centrifuged, followed by 500 μl of buffer AW2 and centrifugation. After each centrifugation, the flow-through was discarded. An extra centrifugation step (dry) was added to remove any excess ethanol before the extracted DNA was eluted into a clean RNase, DNase-free 1.5ml microcentrifuge tube using 100 μl of elution buffer AE. For elution, the buffer was added to the column, allowed to incubate for 1 minute at room temperature and then centrifuged before being re-added to the column, re-incubated for 1 minute and then centrifuged through. This second incubation increased the yield by approximately 15%. Sample concentration and purity was determined using a Nanodrop One instrument before proceeding to the next step. Plasmid controls To provide controls for the PCR and for initial optimisation of the assays, regions of the PLOD1 gene that included the region of interest were synthesized commercially and inserted into a pEX-A128 plasmid vector (Eurofins Genomics). These plasmids were prepared in standard DH5-alpha E-coli using a Qiagen mini-prep kit. This method produced control plasmids that were approximately 60 ng/μl in concentration. Individually the plasmids were used as the homozygous controls, and when mixed in equal concentrations together they acted as the heterozygous control. Real-time PCR design The real-time PCR assay was based on a TaqMan SNP Genotyping assay (Thermo Fisher Scientific). Two primers were designed to amplify the region of equine PLOD1 that contained the SNP of interest. The primers were designed to not amplify human PLOD1, to avoid any potential contamination issues. Two allele specific probes were also designed, that would bind to the exact region where the SNP would be present. One probe sequence matched the wild-type sequence and the other recognised the mutant version (see ). Each probe had a different fluorescent dye (FAM–mutant and VIC–wild-type) bound to it along with a quencher (TAMRA). The principle of the assay is that during the PCR cycle, if the wild-type sequence is present the wild-type probe will bind, DNA polymerase will interact with it during amplification and knock off the fluorescent dye, releasing it from the quencher and allowing a signal to be produced and measured. In contrast, if the mutant sequence is present, the mutant specific probe will bind, and its specific fluorescent signal will be produced. Real-time PCR assay conditions Each PCR was set up in a 96 well plate as outlined in and centrifuged briefly to eliminate air bubbles and to ensure the reaction mixture was in the bottom of the wells. Duplicate plasmid controls for homozygous mutant, homozygous wild-type and heterozygous sequences were included on each run. The plate was sealed with an optical adhesive cover and placed in the Applied Biosystems 7500 Real-Time PCR system which functioned under the reaction conditions detailed in . The sequence of the primers and probes are detailed in ; probe sequences were designed with the help of Thermo Fisher Scientific to create a non-human custom TaqMan SNP Genotyping Assay. A detailed step by step protocol is available in . Validation In Ireland one commercial company performs the genetic test for WFFS and this company was used to validate the newly developed PCR based assay being assessed here. Hair samples were sent to the company for analysis in an anonymous format and results from them were taken as the current ‘gold standard’. Hair samples were collected with consent from a variety of horse and pony breeds. The hair was manually pulled from the mane to ensure as little discomfort to the horse as possible. The hair was sealed in a biosafety bag and stored at room temperature before testing. The study was conducted under ethical exemption from the University College Dublin Animal Research Ethics Committee: (AREC-E-18-47-Duggan). The Qiagen DNA Blood and Tissue kit was used, and the extraction protocol was adapted from the manufacturer’s instructions. Hair strands (10 per extraction) were cut at a maximum length of 1 cm, including the follicular tag. The hair was placed into a 1.5 ml Rnase, DNase-free micro-centrifuge tube followed by 300 μl of ATL Buffer, 20 μl of Proteinase K and 20ul of a 1 M DTT solution. The sample was incubated at 56°C with occasional vortexing until the hair was completely lysed and could not be seen anymore. After incubation, the sample was vortexed for 15 seconds to ensure complete homogenisation before 300 μl AL Buffer was added and it was again briefly vortexed. This was followed by the addition of 300 μl of ethanol (molecular grade, > 99.5% pure) with brief vortexing again to ensure complete mixing. The sample mixture was pipetted into a DNeasy mini spin column and centrifuged at room temperature at 6000 x g for 1 minute. The flow through was discarded and two wash steps were carried out on the column using the same centrifugation conditions as before–firstly 500 μl of AW1 Buffer was added and the column was centrifuged, followed by 500 μl of buffer AW2 and centrifugation. After each centrifugation, the flow-through was discarded. An extra centrifugation step (dry) was added to remove any excess ethanol before the extracted DNA was eluted into a clean RNase, DNase-free 1.5ml microcentrifuge tube using 100 μl of elution buffer AE. For elution, the buffer was added to the column, allowed to incubate for 1 minute at room temperature and then centrifuged before being re-added to the column, re-incubated for 1 minute and then centrifuged through. This second incubation increased the yield by approximately 15%. Sample concentration and purity was determined using a Nanodrop One instrument before proceeding to the next step. To provide controls for the PCR and for initial optimisation of the assays, regions of the PLOD1 gene that included the region of interest were synthesized commercially and inserted into a pEX-A128 plasmid vector (Eurofins Genomics). These plasmids were prepared in standard DH5-alpha E-coli using a Qiagen mini-prep kit. This method produced control plasmids that were approximately 60 ng/μl in concentration. Individually the plasmids were used as the homozygous controls, and when mixed in equal concentrations together they acted as the heterozygous control. The real-time PCR assay was based on a TaqMan SNP Genotyping assay (Thermo Fisher Scientific). Two primers were designed to amplify the region of equine PLOD1 that contained the SNP of interest. The primers were designed to not amplify human PLOD1, to avoid any potential contamination issues. Two allele specific probes were also designed, that would bind to the exact region where the SNP would be present. One probe sequence matched the wild-type sequence and the other recognised the mutant version (see ). Each probe had a different fluorescent dye (FAM–mutant and VIC–wild-type) bound to it along with a quencher (TAMRA). The principle of the assay is that during the PCR cycle, if the wild-type sequence is present the wild-type probe will bind, DNA polymerase will interact with it during amplification and knock off the fluorescent dye, releasing it from the quencher and allowing a signal to be produced and measured. In contrast, if the mutant sequence is present, the mutant specific probe will bind, and its specific fluorescent signal will be produced. Each PCR was set up in a 96 well plate as outlined in and centrifuged briefly to eliminate air bubbles and to ensure the reaction mixture was in the bottom of the wells. Duplicate plasmid controls for homozygous mutant, homozygous wild-type and heterozygous sequences were included on each run. The plate was sealed with an optical adhesive cover and placed in the Applied Biosystems 7500 Real-Time PCR system which functioned under the reaction conditions detailed in . The sequence of the primers and probes are detailed in ; probe sequences were designed with the help of Thermo Fisher Scientific to create a non-human custom TaqMan SNP Genotyping Assay. A detailed step by step protocol is available in . In Ireland one commercial company performs the genetic test for WFFS and this company was used to validate the newly developed PCR based assay being assessed here. Hair samples were sent to the company for analysis in an anonymous format and results from them were taken as the current ‘gold standard’. Optimisation of the PCR PCR conditions were initially optimised using the control plasmids. Varying concentrations of each plasmid were used to individually establish optimal conditions for both the wildtype and mutant sequences. The optimal concentration for each plasmid was 6 ng/μl. Following this, the plasmids were mixed to create a heterozygous situation as would be seen in carriers. Equal concentrations of each plasmid were added to the reaction mix, while maintaining the same volume, and the real–time PCR reaction was shown to effectively distinguish between them, despite only a single nucleotide difference being present . Following optimization using plasmids, DNA was extracted from 20 hair samples, each from a different animal, and the same real-time PCR assay was carried out. Different concentrations of samples were tested and 10 ng/μl was found to be optimal. Samples were run in duplicate each time and repeated at least three times to confirm the robustness of the assay, with a consistent result achieved each time. The assay reliably identified whether a sample was homozygous wild-type or heterozygous. No homozygous mutants from hair samples were identified. Validation of results A total of 18 samples were tested by an external commercial test and compared to the newly developed PCR (example in ). All 18 samples results matched using both assays, 15 samples were homozygous normal on both tests, 3 samples were heterozygous on both tests. PCR conditions were initially optimised using the control plasmids. Varying concentrations of each plasmid were used to individually establish optimal conditions for both the wildtype and mutant sequences. The optimal concentration for each plasmid was 6 ng/μl. Following this, the plasmids were mixed to create a heterozygous situation as would be seen in carriers. Equal concentrations of each plasmid were added to the reaction mix, while maintaining the same volume, and the real–time PCR reaction was shown to effectively distinguish between them, despite only a single nucleotide difference being present . Following optimization using plasmids, DNA was extracted from 20 hair samples, each from a different animal, and the same real-time PCR assay was carried out. Different concentrations of samples were tested and 10 ng/μl was found to be optimal. Samples were run in duplicate each time and repeated at least three times to confirm the robustness of the assay, with a consistent result achieved each time. The assay reliably identified whether a sample was homozygous wild-type or heterozygous. No homozygous mutants from hair samples were identified. A total of 18 samples were tested by an external commercial test and compared to the newly developed PCR (example in ). All 18 samples results matched using both assays, 15 samples were homozygous normal on both tests, 3 samples were heterozygous on both tests. WFFS is a fatal condition for horses that may be economically and emotionally devastating for the owner involved. The mating of heterozygous carriers creates a 1 in 4 chance of creating a homozygous mutant. Independent of the potential positives or negatives, increased understanding of the genotypic status of horses allows informed choices to be made and should be seen as a positive and progressive step towards improving equine health. There is a need to carry out large scale prevalence studies and to encourage owners to screen their animals for this mutation, so inappropriate mating can be avoided. The newly developed WFFS assay reported here is an improvement on what is currently available. It is an efficient, reliable, and rapid test that allows high-throughput analysis of samples thus reducing the associated costs. The new assay is faster than any current test being used: From DNA extraction to PCR result is approximately 6 hours while in our experience, the Sanger sequencing route can take approximately 3 days, while the commercial company we used (only one available in Ireland) took 2 weeks to return results. In terms of cost, both the new qPCR method and a Sanger sequencing route are approximately comparable (excluding start up hardware costs), while the cost of testing a batch of 40 samples commercially was approximately 4 times the cost per sample compared to the qPCR assay described here, again excluding start up hardware costs. There is potential to decrease the price of the real-time PCR based assay further if sample numbers are increased or steps could be automated. In summary, this newly developed assay is an excellent alternative to currently used options and we recommend its use in WFFS associated mutation prevalence studies that are planned in the future. S1 File (PDF) Click here for additional data file. |
Impact of electronic cigarettes (e-cigs) and heat-not-burn/heated tobacco products (HnB/HTP) on asthma and chronic obstructive pulmonary disease: a viewpoint of the Italian Society of Internal Medicine | a2023274-daa8-471a-b5fe-c88e1445aa58 | 11467123 | Internal Medicine[mh] | Asthma and chronic obstructive pulmonary disease (COPD) are the most common chronic respiratory diseases globally, representing a major public health challenge and a leading cause of morbidity and mortality . The prevalence of COPD varies widely due to differences in survey methods, diagnostic criteria, and analytical approaches . According to the Burden of Obstructive Lung Diseases (BOLD) study, the global prevalence of COPD is estimated at about 10.3% and is expected to rise . In Italy, ISTAT (the National Institute of Statistics) reports a COPD prevalence of 5.6%, although this figure might be underestimated due to random diagnoses and hospitalizations for exacerbations . Global prevalence estimates of asthma indicate about 10% in children and adolescents and 6–7% in adults, varying significantly from 2 to 3% in adults in low-income countries to 10% in high-income countries . In Italy, the prevalence of asthma in the population aged over 15 years is about 6.1%, a figure confirmed by the GEIRD study for the age group 20–44 . Recent years show a slight but significant increase in asthma prevalence, especially among children and adolescents . It is important to note that COPD and asthma can overlap in smokers or ex-smokers, making it difficult to differentiate between them in clinical practice due to similarities in symptoms and treatments . Tobacco smoking is a significant environmental risk factor for developing COPD and a common trigger for asthma . The latest survey data from the Istituto Superiore di Sanità-PASSI for 2021–2022 shows that 24.2% of the Italian population are smokers, with the highest prevalence (approximately 29%) among the 18–34 age group . This rate has plateaued after a historical decline (29.8% in 2008, 27.0% in 2014, 24.5% in 2020). Cigarette smoke is a complex aerosol comprising over 7000 chemical components, each possessing toxic and carcinogenic properties in both its gaseous and particulate phases . These components include nicotine, carbon monoxide, carbon dioxide, heavy metals such as nickel, cadmium, chromium, arsenic, formaldehyde, acrolein, acetone, polycyclic aromatic hydrocarbons (like benzo(a)pyrene), ammonia, and tar . Additionally, cigarette smoke contains a significant amount of reactive oxygen species (ROS) derived from oxygen metabolites that compromise the integrity of physical barriers, leading to increased permeability in respiratory epithelial cells and hindering the clearance of mucus by cilia . Table presents a comprehensive list of the most critical chemical components and toxicants found in tobacco smoke. Globally, about 35–45% of COPD patients are current smokers, and around 20% have never smoked . The proportion of current smokers with asthma aligns with the general population , while former smokers range from one quarter to over 40% among asthma patients . Thus, approximately half of the adult asthma population globally are current or former cigarette smokers . Tobacco use, driven by nicotine dependence and behavioral habits, remains a major preventable public health issue. Despite around 70% of smokers wishing to quit, successful long-term cessation often involves multiple attempts: tipically, individuals attempting to quit smoking make about 6 attempts before achieving long-term abstinence. Using a nicotine patch in combination with other nicotine replacement therapy (NRT) products is generally more effective than using a single NRT product alone. As for behavioural support, it can range from brief to intensive and can be effectively delivered in person, or remotely via telephone, text messages, or the internet . Electronic cigarettes have emerged as a potential solution in boosting smoking cessation success rate. However, there is ongoing debate among researchers and public health experts about whether e-cigarettes will significantly enhance smoking cessation efforts, or if they might undermine public health initiatives and potentially increase smoking rates. Another contentious issue is whether e-cigarettes contain fewer and less harmful toxicants compared to conventional combustible cigarettes. Similar claims, regarding their role as an aid in smoking cessation and possessing a less harmful toxicological profile than traditional cigarettes, have also been made for heat-not-burn (HnB) products. A report by Euromonitor International indicated that these products had a retail volume of 233 million units in 2020, accounting for 7% of the overall tobacco market in the country . Additionally, Italy has seen a rise in electronic cigarette use, with an estimated 900,000 users in 2019 . This shift towards alternative smoking methods, coupled with the high prevalence of chronic conditions like COPD and asthma, poses new challenges for the Italian healthcare system. Therefore, a comprehensive understanding of these conditions, their long-term management, and applicable medications is essential. The presence of comorbidities in many of these patients adds complexity to their care . It is crucial to educate patients about self-management, symptom monitoring, adhering to treatments, and lifestyle modifications, including exercise and smoking cessation. Regarding smoking cessation, many patients are curious about the practicality and usefulness of using electronic cigarettes and HnB products as cessation tools. Asthma and COPD, both characterized as chronic inflammatory airway disorders with limited airflow, share similarities yet differ pathologically due to the types of inflammatory cells involved: COPD is predominantly associated with neutrophils and CD8 lymphocytes, while asthma involves eosinophils and CD4 lymphocytes . Reactive oxygen species (ROS) are known to exacerbate inflammation, contributing to COPD progression, and play a role in worsening asthma symptoms. Exposure to cigarette smoke activates damage-associated molecular patterns (DAMPs) and pathogen-associated molecular patterns (PAMPs) within lung epithelial cells and alveolar macrophages. This activation stimulates Toll-like receptors (TLR) and NOD-like receptors (NLR), leading to overproduction of ROS and reactive nitric oxide (RNS). Such processes disrupt the balance between oxidation and antioxidants, heightening oxidative stress that damages key lung components like lipids, proteins, nucleic acids, elastin, and collagen. The resulting damage leads to increased apoptosis, impaired skeletal muscle function, excessive mucus production, and reduced steroid receptor efficacy . Cigarette smoking causes not only irreversible DNA mutations but also potentially reversible changes in the epigenetic landscape, including DNA methylation and chromatin modification, accompanied by chronic lung inflammation . It can also shift the predominant inflammatory mechanisms in asthma to resemble those in COPD . Long-term tobacco use perpetuates oxidative damage, impairs protective responses and DNA repair, disrupts mitochondrial activity, and affects endoplasmic reticulum homeostasis, thereby exacerbating disease progression . Chronic inflammation leads to a continuous influx of inflammatory cells, releasing various mediators including proteases and cytokines, and initiates epithelial–mesenchymal transition (EMT). This constant state induces persistent inflammation and oxidative stress in the lungs, leading to repair and remodelling cycles in both COPD and asthma, promoting mucus hypersecretion, chronic bronchitis, and emphysema . Compared to the complex composition of tobacco smoke, e-liquids have a simpler makeup, primarily consisting of vegetable glycerol (VG), propylene glycol (PG), nicotine, and water. The market offers a wide variety of flavored e-liquids, created from synthetic flavor compounds, natural extracts, or a combination of both . The number of compounds in e-cigarette aerosol varies with the flavorings used, sometimes reaching up to 142 . E-cigarettes produce vapor, not smoke. While the inflammatory response may be less intense than with traditional smoking, inhaling heated aerosol can still trigger airway inflammation, possibly leading to respiratory issues. E-cigarette vapor extracts can exhibit acute cytotoxicity, affect cell proliferation, and alter cell morphology, similar to high-nicotine traditional cigarettes. The toxicity varies with different e-liquids used in the same device . Studies have linked the cytotoxic effects of widely used refill fluids to high concentrations of flavor chemicals . Even flavorless e-cigarettes can reduce cell viability and increase pro-inflammatory cytokines, confirming that e-cigarettes' core components independently exert biological effects . E-cigarette aerosol can stimulate ROS production, causing DNA damage, reducing cell viability in a concentration-dependent manner, and impairing phagocytosis. Analyses reveal activation of apoptosis and programmed necrosis pathways . In another study, e-cigarette users showed distinct changes in proteins associated with membrane and mucus formation, increasing susceptibility to respiratory infections . Comparisons between cigarette smokers and e-cigarette users revealed common downregulated genes related to cilia assembly and movement in both groups . However, different outcomes emerged in studies assessing the impact on epithelial barrier integrity, with cigarette smoke extract compromising this integrity, while e-cigarette aerosol did not, suggesting that only cigarette smoke adversely affects host defence mechanisms . E-cigarette aerosol also triggers the release of IL-8/CXCL8 and MMP-9 and increases neutrophil elastase activity, potentially facilitating neutrophil migration to inflammation sites and exacerbating symptoms in asthma and COPD patients . The variability in e-liquid composition and device types challenges generalization of these effects. It is important to note that, currently, there are no extensive long-term toxicological or safety studies on vaping conducted in humans. HnB products produce both primary and secondary emissions containing harmful chemicals such as nicotine, particulate matter, benzene, acrolein, and tobacco-specific nitrosamines. Though these emissions are lower than those from traditional cigarettes, they still pose potential risks . Extended use of HnB products has been associated with reduced endothelial function, increased oxidative stress, and heightened platelet activation . This exposure can affect mitochondrial function, exacerbate airway inflammation and remodeling, increase oxidative stress, and heighten the risk of respiratory infections due to enhanced microbial adherence . Furthermore, studies indicate that HnB products emit more than electronic cigarettes . Comprehensive studies are necessary to fully understand the health implications of both HnB and e-cigarette products. The authors have undertaken a thorough evaluation and analysis of recently published literature, encompassing various studies and reviews, to comprehensively understand the health impacts of e-cigarettes and HnB devices in relation to asthma and COPD. Through this analysis, the Authors seek to shed light on the complex and nuanced effects these smoking alternatives may have on individuals suffering from these respiratory conditions. Specifically, the aim is to offer a well-informed and reasoned perspective on two critical questions: What is the health impact of e-cigarettes and HnB devices on the outcomes of asthma? What is the health impact of e-cigarettes and HnB devices on the outcomes of COPD? The manuscript is structured as a narrative review, addressing a subject of broad interest, and is grounded in a comprehensive literature review (considering papers published from 2013 to 2023) conducted primarily through PubMed. Relevant keywords including MeSH terms were meticulously defined through a collaborative effort by all co-authors, ensuring the inclusion of pertinent studies. This process entailed multiple iterations to accurately encompass the desired range of studies. These studies were essential for extracting key informative evidence, which was then used to formulate the Discussion and Conclusions sections. Beyond the development of search criteria, a significant aspect of the manuscript's creation was the consensus-building process among the co-authors. This step is critical in scholarly writing, as it guarantees the manuscript's findings, and interpretations accurately represent the collective understanding of the entire research team, thereby reducing the risk of inaccuracies or misinterpretations. This phase entailed an exhaustive review and discussion of the relevant literature until there was unanimous agreement among all co-authors on the interpretation and phrasing of the key findings. References were thoroughly analysed, and the most informative findings/evidence from this analysis were organized and summarized into three distinct tables. The tables report the most relevant literature articles selected by the working group that provide the informative key evidence on the health impact of e-cigarettes/ HnB on general health status (supplementary Table 2), and on asthma (supplementary Table 3) and COPD (supplementary Table 4) clinical outcomes. Effects on health status This section concerns the analysis of the literature regarding general characteristics of e-cig and HnB tobacco products, and potential effects on health status. The evidence referable to this section, listed in supplementary Table 2, reflects data from the literature which may in some cases appear contradictory. These can be summarized in a position which, on the one hand, indicates that alternative products to traditional tobacco contain toxic substances capable to express harmful effects (evidences 1–5 and 7) on the various systems (respiratory and others) and through multiple mechanisms, on the other hand acknowledges that e-cigs and HnB tobacco products do not contain some of the most harmful toxicants produced by smoke from combustion cigarettes (evidences 6 and 8). In the specific field of respiratory diseases, e-cigs in particular can express harmful effects related to nicotine and/or flavoured e-liquids which are expressed through mechanisms of cytotoxicity, oxidative stress, inflammation, airway hyper-reactivity, airway remodelling, mucin production, apoptosis and emphysematous changes (evidences 3 and 4), as well as to increased susceptibility to viral infection through an increase in lung ACE2 expression, a mechanism that we have well known with SARS-CoV-2 infection (evidence 5). For its part, exposure to HnB products has been reported to alter mitochondrial function, which may exaggerate airway inflammation and remodelling (evidence 7). On the other hand, exclusive e-cigarette and HnB products users have lower risk of exposure to tobacco smoke toxicants and carcinogens compared to cigarette smokers (evidences 6 and 8) and consequently undergo less significant pulmonary changes (evidence 9). Finally, we should not forget the potential harm, nicotine-dependent, and demonstrated in murine models, which also affects the use in pregnancy of smoking products alternative to traditional cigarettes (evidence 10). Effects on asthma clinical outcomes In the specific field of asthma, epidemiological studies have documented that among e-cigarettes users there is an increased incidence of the disease compared to non-users (evidence 1, supplementary Table 3). This finding is important considering the huge number of e-cig users among young people who have never smoked. It is known that common flavouring agents are recognized as primary irritants of mucosal tissue of respiratory tract, and the thermal decomposition of propylene glycol and vegetable glycerine (the base constituents of e-liquids) may produce reactive carbonyls, which have known respiratory toxicities. Therefore, it is not surprising that in subjects affected by asthma, the use of e-cigarettes is associated with worse symptomatology and impaired lung function (evidence 2). However, lower odds of negative outcomes have been reported when there is a switch from traditional cigarettes to e-cigs (evidence 3), and this allow to consider that these products can be an option for asthmatic patients who cannot quit smoking by other methods. Heterogeneous findings are available in the literature on the role of e-cigarettes in helping asthmatic subjects to quit smoking (evidence 4). However, as an overall consideration, due to the relatively recent use of e-cigarette (and HnB as well) and the low number of experimental data and consistent evidence, it is hard to draw conclusive indications regarding toxicological and clinical consequences of the use of e-cigarettes (and more generally, of alternative tobacco products) (evidence 5). Effects on COPD clinical outcomes From a qualitative point of view, the considerations expressed for alternative products to traditional tobacco with reference to asthma, are reproducible in the context of the COPD. In particular, also for COPD a higher incidence of the disease is observed in e-cigarettes users compared to non-users (evidence 1, supplementary Table 4) and e-cigarette use leads to an increase of symptoms and impaired lung function in COPD patients (evidence 2). However, e-cigarettes users have been reported to have less negative outcomes if compared to combustible cigarettes or dual smoking (combustible + electronic cigarettes) (evidence 3); further, in the case of HnB products, their use after switch from combustible cigarettes seems associated with reduced exacerbations, improvements in symptomatology and activity level in COPD patients (evidence 4). Inconclusive findings are available on the possible role of e-cigarettes (and HnB products) in helping to reduce or stop smoking in COPD patients (evidence 5). Also in the case of COPD, to express overall considerations on the impact of products alternative to combustible cigarettes is a challenging effort, due to a poor availability of methodologically solid experimental findings, and the relatively recent use of these products combined with the need of decades of chronic smoking for development of COPD (evidence 6). This section concerns the analysis of the literature regarding general characteristics of e-cig and HnB tobacco products, and potential effects on health status. The evidence referable to this section, listed in supplementary Table 2, reflects data from the literature which may in some cases appear contradictory. These can be summarized in a position which, on the one hand, indicates that alternative products to traditional tobacco contain toxic substances capable to express harmful effects (evidences 1–5 and 7) on the various systems (respiratory and others) and through multiple mechanisms, on the other hand acknowledges that e-cigs and HnB tobacco products do not contain some of the most harmful toxicants produced by smoke from combustion cigarettes (evidences 6 and 8). In the specific field of respiratory diseases, e-cigs in particular can express harmful effects related to nicotine and/or flavoured e-liquids which are expressed through mechanisms of cytotoxicity, oxidative stress, inflammation, airway hyper-reactivity, airway remodelling, mucin production, apoptosis and emphysematous changes (evidences 3 and 4), as well as to increased susceptibility to viral infection through an increase in lung ACE2 expression, a mechanism that we have well known with SARS-CoV-2 infection (evidence 5). For its part, exposure to HnB products has been reported to alter mitochondrial function, which may exaggerate airway inflammation and remodelling (evidence 7). On the other hand, exclusive e-cigarette and HnB products users have lower risk of exposure to tobacco smoke toxicants and carcinogens compared to cigarette smokers (evidences 6 and 8) and consequently undergo less significant pulmonary changes (evidence 9). Finally, we should not forget the potential harm, nicotine-dependent, and demonstrated in murine models, which also affects the use in pregnancy of smoking products alternative to traditional cigarettes (evidence 10). In the specific field of asthma, epidemiological studies have documented that among e-cigarettes users there is an increased incidence of the disease compared to non-users (evidence 1, supplementary Table 3). This finding is important considering the huge number of e-cig users among young people who have never smoked. It is known that common flavouring agents are recognized as primary irritants of mucosal tissue of respiratory tract, and the thermal decomposition of propylene glycol and vegetable glycerine (the base constituents of e-liquids) may produce reactive carbonyls, which have known respiratory toxicities. Therefore, it is not surprising that in subjects affected by asthma, the use of e-cigarettes is associated with worse symptomatology and impaired lung function (evidence 2). However, lower odds of negative outcomes have been reported when there is a switch from traditional cigarettes to e-cigs (evidence 3), and this allow to consider that these products can be an option for asthmatic patients who cannot quit smoking by other methods. Heterogeneous findings are available in the literature on the role of e-cigarettes in helping asthmatic subjects to quit smoking (evidence 4). However, as an overall consideration, due to the relatively recent use of e-cigarette (and HnB as well) and the low number of experimental data and consistent evidence, it is hard to draw conclusive indications regarding toxicological and clinical consequences of the use of e-cigarettes (and more generally, of alternative tobacco products) (evidence 5). From a qualitative point of view, the considerations expressed for alternative products to traditional tobacco with reference to asthma, are reproducible in the context of the COPD. In particular, also for COPD a higher incidence of the disease is observed in e-cigarettes users compared to non-users (evidence 1, supplementary Table 4) and e-cigarette use leads to an increase of symptoms and impaired lung function in COPD patients (evidence 2). However, e-cigarettes users have been reported to have less negative outcomes if compared to combustible cigarettes or dual smoking (combustible + electronic cigarettes) (evidence 3); further, in the case of HnB products, their use after switch from combustible cigarettes seems associated with reduced exacerbations, improvements in symptomatology and activity level in COPD patients (evidence 4). Inconclusive findings are available on the possible role of e-cigarettes (and HnB products) in helping to reduce or stop smoking in COPD patients (evidence 5). Also in the case of COPD, to express overall considerations on the impact of products alternative to combustible cigarettes is a challenging effort, due to a poor availability of methodologically solid experimental findings, and the relatively recent use of these products combined with the need of decades of chronic smoking for development of COPD (evidence 6). The well-established connection between cigarette smoking and various severe diseases, including cancer, cardiovascular, and respiratory illnesses, represents a significant public health concern due to its epidemiological, medical, and financial impacts. Internists, who frequently encounter patients with smoking-related diseases, must stay informed about the emerging trends in smoking alternatives such as e-cigarettes and HnB products, along with the ongoing scientific debate surrounding their use. This aligns with the commitment of the Italian Scientific Society of Internal Medicine (SIMI), to analyze the available scientific literature on these products, especially focusing on chronic respiratory diseases like asthma and COPD. While it might seem obvious, a key principle in addressing cigarette smoking is the imperative to prevent the initiation of smoking habits and to encourage cessation. This is particularly crucial for individuals with chronic diseases. For asthma and COPD patients, smoking cessation is known to improve symptoms and slow down the decline in lung function. However, physicians often observe that many of these patients, despite their intentions, do not respond well to conventional smoking cessation interventions, including behavioural support and licensed therapies like nicotine replacement therapy (NRT) and medications, or achieve only limited success. Consequently, many continue to smoke despite experiencing adverse symptoms . In the past decade or so, there has been considerable debate within the scientific community about whether new alternatives to traditional combustion cigarettes, particularly e-cigarettes and HnB products, offer reduced toxicity levels and can be effective tools for quitting or reducing traditional cigarette smoking. Despite the heterogeneity of available data and the challenges in extrapolating results from cellular or animal models to human exposure, the evidence suggests that these alternative products are not risk-free. Their use, compared to not smoking, increases the risk of asthma and/or COPD, exacerbating symptoms and impairing lung function. Although the use of e-cigarettes or HnB products is not without harm and their long-term effects remain uncertain, independent reviews and expert opinions have concluded that these alternatives are likely less harmful than traditional smoking. Studies involving humans have shown that these products, when used by smokers, are associated with significant reductions in blood or urinary biomarkers of tobacco toxicants, particularly when fully switching, and to some extent in case of dual use. While the extent to which these biomarkers represent potential lung toxicity is not entirely clear, several studies have indicated that former smokers who switch to e-cigarettes or HnB products tend to experience lower odds of respiratory outcomes, fewer exacerbations, and improvements in symptoms and physical activity compared to those who continue smoking traditional cigarettes or engage in dual use. Given these findings, the potential of these products as harm reduction options for individuals unable or unwilling to quit smoking merits consideration, especially considering the low long-term cessation rates achieved with pharmacological and behavioural treatments. One hypothesis is that e-cigarettes and HnB products might support smoking reduction or cessation due to better nicotine release, decreasing nicotine concentration overtime, and mimicking the behavioural and sensory aspects of cigarette smoking. Recent publications, including a commentary in Nature and a comprehensive review , have reported higher smoking cessation rates in individuals using nicotine electronic cigarettes compared to those using nicotine replacement therapy. These findings align with what has been previously reported in systematic reviews and randomized controlled and population studies . However, the current data are still insufficient to support unanimous recommendations in this regard . There is a growing public health concern about the increasing use of e-cigarettes and HnB products among never-smokers, particularly adolescents and young people, which may lead to nicotine addiction and potentially increase the likelihood of future conventional smoking . Ironically, while e-cigarettes were introduced as an aid for smoking cessation in adult smokers, epidemiological data indicate that these products often serve as a gateway to tobacco initiation among tobacco-naive adolescents. Factors such as product design, flavours, perceived safety, and targeted marketing strategies on media and social networks have heightened the appeal of e-cigarettes (and HnB products) to younger populations . Recent reports in the United States have shown declines in dual usage rates among youth, offering less concerning trends in this regard , but concerted efforts are still needed to address this public health threat, such as including warnings on product packaging about adverse effects and the risk of nicotine addiction , and promoting awareness campaigns through new media channels. The aforementioned article in Nature asserts that "E-cigarettes are not a panacea for the harms caused by cigarette smoking, but they can contribute to this public health goal. However, the endorsement of e-cigarettes for smoking cessation is likely contingent on continued efforts to restrict access and use by young non-smokers. These two objectives should coexist" . Furthermore, it is crucial for healthcare professionals to receive education, training, and support to improve counselling for patients and the public about the use of tobacco products, their potential health risks, behaviours change, and various cessation options . Smokers with asthma or COPD have compelling reasons to quit or at least reduce cigarette consumption and the associated harms. Clinicians should actively explore all available methods to assist their patients in achieving these goals but should prioritize recommending evidence-based treatments. Given the relatively recent advent of vaping and HnB cigarettes compared to the long duration of chronic smoking required for the development of diseases like COPD, the evolving landscape of these products, and the scarcity of experimental data and consistent evidence, it is not surprising that it is currently difficult to draw definitive conclusions regarding the toxicological, clinical, and public health impact of these alternatives to traditional combustible cigarettes. In this context, high-quality and independent studies with adequate sample sizes and longer follow-up periods are necessary to better document the effects of these products. Tobacco smoking continues to be a major cause of mortality worldwide. Efforts to promote smoking cessation and prevent smoking initiation are crucial. Alternative products to traditional cigarettes tend to produce fewer toxic substances, however they are not risk-free and their use increases the risk of asthma or COPD compared to non-smokers, despite variations in available data and interpretation . Switching from traditional cigarettes to these alternatives can lead to better health outcomes . Although recent data suggest a potential for these combustion-free products to aid in smoking cessation or reduction compared to traditional methods, their potential role in promoting smoking initiation is still up for dispute. High-quality studies with larger sample sizes and longer follow-up periods are essential to thoroughly understand the impact of exposure to alternative tobacco products. Establishing a robust support system for healthcare professionals to provide effective counselling on smoking, including traditional cigarettes, e-cigarettes, and HnB products, is imperative. Below is the link to the electronic supplementary material. Supplementary file1 (DOCX 242 KB) |
Integration of Telemedicine Consultation Into a Tertiary Radiation Oncology Department: Predictors of Use, Treatment Yield, and Effects on Patient Population | 7f099c0d-b2bc-4103-96b4-b69b496c0881 | 11161230 | Internal Medicine[mh] | Telehealth, telemedicine, and related terms are defined by the Centers for Medicare and Medicaid Services as services that a physician or practitioner provides via two-way, interactive technology that substitutes for an in-person (IP) visit. The use of telemedicine rapidly expanded during the COVID-19 pandemic as many health care facilities were forced to halt IP appointments. Telemedicine became integral to ensuring timely, continued care for patients with nonelective medical needs such as cancer. CONTEXT Key Objective What were the characteristics of in-person patients (IPPs) and virtual patients (VPs) who presented to a large cancer center before and during the pandemic and how did telemedicine affect the likelihood of undergoing radiotherapy (yield)? Knowledge Generated In both VPs and IPPs, proximity to clinic was associated with higher yield and higher area deprivation index while VPs received more proton radiation and brachytherapy. Relevance (J.L. Warner) This study describes the impact of telemedicine in the context of radiation oncology, telemedicine allowed for increased reach to rural patients.* *Relevance section written by JCO Clinical Cancer Informatics Editor-in-Chief Jeremy L. Warner, MD, MS, FAMIA, FASCO. Before the pandemic, only 15%-20% of physicians reported experience with telemedicine. As the pandemic accelerated the need for remote medical care, there was a 154% increase in the use of telemedicine in the United States in the first quarter of 2020 when compared with the first quarter of 2019. This was driven by the need for health care during social isolation and supported by policy changes, including the declaration of a public health emergency (PHE) in January 2020. The PHE waived previous telemedicine regulations and allowed for easier implementation and access. Since then, there has been a sustained growth in the use of telemedicine. The impact of telemedicine on access to oncology care has yet to be fully elucidated, especially in the highly specialized, technical, and procedural field of radiation oncology. Hsiao et al found radiation oncology to be the fourth lowest utilizer of telemedicine among 45 specialties between 2020 and 2021. Comparing a population of radiation oncology patients presenting IP before the COVID-19 pandemic with those presenting virtually during the pandemic, one study found telemedicine users to be younger, female, and with higher performance status. Another study evaluating disparities in telemedicine use in a large cancer care network found race, ethnicity, income, and type of insurance most influential to telemedicine utilization. Given the limited understanding of telemedicine in the radiation oncology setting, we aimed to not only further characterize patient utilization of telemedicine in a large radiation oncology center in the United States, but also to understand if the use of telemedicine affected access to radiation oncology care and the likelihood of receiving radiation at our center. Key Objective What were the characteristics of in-person patients (IPPs) and virtual patients (VPs) who presented to a large cancer center before and during the pandemic and how did telemedicine affect the likelihood of undergoing radiotherapy (yield)? Knowledge Generated In both VPs and IPPs, proximity to clinic was associated with higher yield and higher area deprivation index while VPs received more proton radiation and brachytherapy. Relevance (J.L. Warner) This study describes the impact of telemedicine in the context of radiation oncology, telemedicine allowed for increased reach to rural patients.* *Relevance section written by JCO Clinical Cancer Informatics Editor-in-Chief Jeremy L. Warner, MD, MS, FAMIA, FASCO. Selection and Description of Patients We conducted a retrospective, observational study within a large tertiary radiation oncology center in the midwestern United States. All new patient consultations between January 2019 and December 2021 were identified. Before March 2020, all consultations were performed IP. Telemedicine consultations, which started in March 24, 2020, included video conferencing technology or telephone calls. The term telemedicine is used interchangeably with virtual patients (VPs) in this study. A total of 17,915 patients were included. This study was approved by the institutional review board. Demographic, medical, and treatment variables were extracted from the electronic medical record. Medicare Advantage plans were coded as private insurance, while primary Medicare/Medicaid plans were coded as public insurance. If a patient had multiple consultations in the study time range, only their first consultation was included. Miles from clinic, also referred to as distance traveled to clinic or patient distance from clinic, was the distance between a patient's primary address and the radiation oncology center. The clinic and treatment center were at the same location. Those with a primary US address were labeled as US residents. Radiation treatment modalities included photon, proton, brachytherapy, Gamma Knife, and other. The other treatment modalities included intraoperative radiation therapy, orthovoltage, and electron radiotherapy. Treatment yield was defined by whether a patient received radiation treatment at our center by December 31, 2021. The area deprivation index (ADI), as developed by the Health Resources and Services Administration and Kind et al, was used to assess the national health care deprivation percentiles of patients on the basis of zip code. , A higher ADI number indicates higher health care deprivation. To better understand the population of patients served by our clinic, we used a map of county populations by state provided by the 2020 US Census. Statistical Analysis A mean (standard deviation) and number (proportion) are reported for continuous and categorical variables, respectively. Statistical analyses included comparing demographic and treatment variables by type of visit (IP v virtual), yield, and prepandemic versus during the pandemic (starting March 24, 2020). Sex, race, ethnicity, US residency, insurance type (private v public), visit type, consultation year, primary diagnosis, categorized and mean miles from clinic, and treatment modality were assessed for association with yield. A chi square test was used to compare discrete unordered variables, a Wilcoxon rank-sum test (comparison between two groups) or Kruskal-Wallis test (comparison among more than two groups) was used to compare ordered categorical variables, and a T -test assuming equal variances was used for continuous variables (age and mean miles from clinic). Multivariable logistic modeling was used to investigate associations between variables of interest and likelihood for virtual visit and yield. Variables of interest included in the logistic modeling were chosen if they were thought to provide clinical significance in the context of our data. For example, although univariate analysis may have shown race to be significantly different between two groups, our racial distribution was so homogeneous that we did not believe inclusion into our modeling would yield clinically significant results. We also believed many of our variables to be correlated with each other (sex and primary diagnosis, US residency status and miles from clinic, race and ethnicity within our population) and therefore only one of these variables were included. The alpha level was set at P < .05 for statistical significance. SAS version 9.4 was used for analysis. We conducted a retrospective, observational study within a large tertiary radiation oncology center in the midwestern United States. All new patient consultations between January 2019 and December 2021 were identified. Before March 2020, all consultations were performed IP. Telemedicine consultations, which started in March 24, 2020, included video conferencing technology or telephone calls. The term telemedicine is used interchangeably with virtual patients (VPs) in this study. A total of 17,915 patients were included. This study was approved by the institutional review board. Demographic, medical, and treatment variables were extracted from the electronic medical record. Medicare Advantage plans were coded as private insurance, while primary Medicare/Medicaid plans were coded as public insurance. If a patient had multiple consultations in the study time range, only their first consultation was included. Miles from clinic, also referred to as distance traveled to clinic or patient distance from clinic, was the distance between a patient's primary address and the radiation oncology center. The clinic and treatment center were at the same location. Those with a primary US address were labeled as US residents. Radiation treatment modalities included photon, proton, brachytherapy, Gamma Knife, and other. The other treatment modalities included intraoperative radiation therapy, orthovoltage, and electron radiotherapy. Treatment yield was defined by whether a patient received radiation treatment at our center by December 31, 2021. The area deprivation index (ADI), as developed by the Health Resources and Services Administration and Kind et al, was used to assess the national health care deprivation percentiles of patients on the basis of zip code. , A higher ADI number indicates higher health care deprivation. To better understand the population of patients served by our clinic, we used a map of county populations by state provided by the 2020 US Census. A mean (standard deviation) and number (proportion) are reported for continuous and categorical variables, respectively. Statistical analyses included comparing demographic and treatment variables by type of visit (IP v virtual), yield, and prepandemic versus during the pandemic (starting March 24, 2020). Sex, race, ethnicity, US residency, insurance type (private v public), visit type, consultation year, primary diagnosis, categorized and mean miles from clinic, and treatment modality were assessed for association with yield. A chi square test was used to compare discrete unordered variables, a Wilcoxon rank-sum test (comparison between two groups) or Kruskal-Wallis test (comparison among more than two groups) was used to compare ordered categorical variables, and a T -test assuming equal variances was used for continuous variables (age and mean miles from clinic). Multivariable logistic modeling was used to investigate associations between variables of interest and likelihood for virtual visit and yield. Variables of interest included in the logistic modeling were chosen if they were thought to provide clinical significance in the context of our data. For example, although univariate analysis may have shown race to be significantly different between two groups, our racial distribution was so homogeneous that we did not believe inclusion into our modeling would yield clinically significant results. We also believed many of our variables to be correlated with each other (sex and primary diagnosis, US residency status and miles from clinic, race and ethnicity within our population) and therefore only one of these variables were included. The alpha level was set at P < .05 for statistical significance. SAS version 9.4 was used for analysis. Overall Patient Characteristics (January 2019-December 2021) Table displays patient characteristics for the entire population before and during the pandemic. In the study period, 15,977 (88%) patients were seen IP. Treatment yield for all patients was 69%. Most patients were male (56%). Mean age was 63 years (IQR, 55.8-73.1). Most had private insurance (58%). Ninety-three percent were White and 98% were non-Hispanic. Most common malignancies were genitourinary (GU; 25%), breast (14%), and GI (11%). Although 44% of patients lived within 100 miles of clinic, the mean distance traveled to clinic was 260 miles (IQR, 64-291 miles). During the pandemic, VP accounted for 21% of visits, non-US resident patient consultations decreased, the use of private insurance increased, and patients' average miles from clinic decreased. Treatment yield was not affected by the pandemic. Patient Characteristics During the Pandemic (March 2020-December 2021) Table shows differences between IPPs and VPs during the pandemic only. On univariate analysis, VPs were more likely to be male, younger in age, present from further distances, and were 1.5 times less likely to receive treatment at our institution (yield: 48% VPs v 72% IPPs). A significantly higher portion of VPs presented for GU diagnoses (43% VPs v 21% IPPs) and a significantly lower portion presented for palliative diagnosis (2.8% VPs v 10% IPPs). On multivariable logistic modeling, only age and distance from clinic remained significantly different between IPPs and VPs during the pandemic; VPs were more likely to be younger in age and live farther from clinic. Patient Characteristics by Yield During the Pandemic (March 2020-December 2021) IPPs A total of 8,211 patients were seen IP during the pandemic, of whom 5,883 (72%) underwent treatment at our institution (Table ). On univariate analysis, those who received treatment were younger and lived closer to the clinic. We did not appreciate sex or insurance differences on univariate analysis. Patients who self-reported as White or multiracial were the most likely to undergo treatment. By disease category, those with benign, pediatric, and CNS diagnoses were most likely to undergo treatment. Multivariable logistic modeling found male sex and fewer miles from clinic to be associated with higher yield in IPPs, but age was no longer a significant factor. VPs There were 2,143 new VP consultations, of whom 1,026 (48%) underwent treatment at our institution (Table ). In contrast to the IP population, likelihood of treatment was not affected by sex. On univariate analysis, younger patients and those presenting from closer distances were more likely to undergo treatment; however, these distances were still significantly further away than their IP counterparts (205 miles IP v 349 miles VP). Lymphoma, pediatric, and benign diagnoses had the highest yields, albeit significantly lower than their IP counterparts. Upon multivariable logistic modeling, miles from clinic was the only factor contributing to the likelihood of VPs undergoing treatment at our institution, with patients living closer to clinic being more likely to receive treatment at the facility. Compared with IPPs, VPs were 1.9 times more likely to receive proton radiation (23% IP v 44% VP) and had slightly higher use of brachytherapy (4% IP v 6% VP). Proton use for lymphoma and CNS VPs were 2.5 and 2.2 times higher than in IPPs, respectively. GU malignancies accounted for the largest absolute number of VP treatments (188 VPs treated). Census and ADI Appendix Figure A shows a map from the US Census Bureau of counties by persons per square mile, with circles representing incremental 100-mile radii away from the clinic. Apart from one large metropolitan area at the edge of the 100-mile radius, most surrounding counties can be classified as rural, with <500 persons per square mile. As seen in Table , over 60% of all patients presented from within 200 miles of the clinic. Appendix Figure A shows the national ADI for Minnesota and its surrounding states, with red representing the most disadvantaged group. Those presenting IP had significantly higher ADI than those presenting virtually (Table ). In both IPPs and VPs, those who received treatment had significantly higher ADIs (Appendix Tables A and A ). In fact, those who presented IP and underwent treatment had the highest ADI. However, the difference in ADI by yield was most significant in VPs (46 v 43). Table displays patient characteristics for the entire population before and during the pandemic. In the study period, 15,977 (88%) patients were seen IP. Treatment yield for all patients was 69%. Most patients were male (56%). Mean age was 63 years (IQR, 55.8-73.1). Most had private insurance (58%). Ninety-three percent were White and 98% were non-Hispanic. Most common malignancies were genitourinary (GU; 25%), breast (14%), and GI (11%). Although 44% of patients lived within 100 miles of clinic, the mean distance traveled to clinic was 260 miles (IQR, 64-291 miles). During the pandemic, VP accounted for 21% of visits, non-US resident patient consultations decreased, the use of private insurance increased, and patients' average miles from clinic decreased. Treatment yield was not affected by the pandemic. Table shows differences between IPPs and VPs during the pandemic only. On univariate analysis, VPs were more likely to be male, younger in age, present from further distances, and were 1.5 times less likely to receive treatment at our institution (yield: 48% VPs v 72% IPPs). A significantly higher portion of VPs presented for GU diagnoses (43% VPs v 21% IPPs) and a significantly lower portion presented for palliative diagnosis (2.8% VPs v 10% IPPs). On multivariable logistic modeling, only age and distance from clinic remained significantly different between IPPs and VPs during the pandemic; VPs were more likely to be younger in age and live farther from clinic. IPPs A total of 8,211 patients were seen IP during the pandemic, of whom 5,883 (72%) underwent treatment at our institution (Table ). On univariate analysis, those who received treatment were younger and lived closer to the clinic. We did not appreciate sex or insurance differences on univariate analysis. Patients who self-reported as White or multiracial were the most likely to undergo treatment. By disease category, those with benign, pediatric, and CNS diagnoses were most likely to undergo treatment. Multivariable logistic modeling found male sex and fewer miles from clinic to be associated with higher yield in IPPs, but age was no longer a significant factor. VPs There were 2,143 new VP consultations, of whom 1,026 (48%) underwent treatment at our institution (Table ). In contrast to the IP population, likelihood of treatment was not affected by sex. On univariate analysis, younger patients and those presenting from closer distances were more likely to undergo treatment; however, these distances were still significantly further away than their IP counterparts (205 miles IP v 349 miles VP). Lymphoma, pediatric, and benign diagnoses had the highest yields, albeit significantly lower than their IP counterparts. Upon multivariable logistic modeling, miles from clinic was the only factor contributing to the likelihood of VPs undergoing treatment at our institution, with patients living closer to clinic being more likely to receive treatment at the facility. Compared with IPPs, VPs were 1.9 times more likely to receive proton radiation (23% IP v 44% VP) and had slightly higher use of brachytherapy (4% IP v 6% VP). Proton use for lymphoma and CNS VPs were 2.5 and 2.2 times higher than in IPPs, respectively. GU malignancies accounted for the largest absolute number of VP treatments (188 VPs treated). A total of 8,211 patients were seen IP during the pandemic, of whom 5,883 (72%) underwent treatment at our institution (Table ). On univariate analysis, those who received treatment were younger and lived closer to the clinic. We did not appreciate sex or insurance differences on univariate analysis. Patients who self-reported as White or multiracial were the most likely to undergo treatment. By disease category, those with benign, pediatric, and CNS diagnoses were most likely to undergo treatment. Multivariable logistic modeling found male sex and fewer miles from clinic to be associated with higher yield in IPPs, but age was no longer a significant factor. There were 2,143 new VP consultations, of whom 1,026 (48%) underwent treatment at our institution (Table ). In contrast to the IP population, likelihood of treatment was not affected by sex. On univariate analysis, younger patients and those presenting from closer distances were more likely to undergo treatment; however, these distances were still significantly further away than their IP counterparts (205 miles IP v 349 miles VP). Lymphoma, pediatric, and benign diagnoses had the highest yields, albeit significantly lower than their IP counterparts. Upon multivariable logistic modeling, miles from clinic was the only factor contributing to the likelihood of VPs undergoing treatment at our institution, with patients living closer to clinic being more likely to receive treatment at the facility. Compared with IPPs, VPs were 1.9 times more likely to receive proton radiation (23% IP v 44% VP) and had slightly higher use of brachytherapy (4% IP v 6% VP). Proton use for lymphoma and CNS VPs were 2.5 and 2.2 times higher than in IPPs, respectively. GU malignancies accounted for the largest absolute number of VP treatments (188 VPs treated). Appendix Figure A shows a map from the US Census Bureau of counties by persons per square mile, with circles representing incremental 100-mile radii away from the clinic. Apart from one large metropolitan area at the edge of the 100-mile radius, most surrounding counties can be classified as rural, with <500 persons per square mile. As seen in Table , over 60% of all patients presented from within 200 miles of the clinic. Appendix Figure A shows the national ADI for Minnesota and its surrounding states, with red representing the most disadvantaged group. Those presenting IP had significantly higher ADI than those presenting virtually (Table ). In both IPPs and VPs, those who received treatment had significantly higher ADIs (Appendix Tables A and A ). In fact, those who presented IP and underwent treatment had the highest ADI. However, the difference in ADI by yield was most significant in VPs (46 v 43). The COVID-19 pandemic elevated telemedicine to the forefront of medical practice. Despite rapid adoption, the impact of telemedicine on access to cancer care, especially to procedural specialties such as radiation oncology, remains unclear. In our study of 17,915 new patient consultations seen from 2019 to 2021 in a tertiary radiation oncology clinic, we aimed to understand the impact of rapid telemedicine implementation on patient demographics and treatment yield and telemedicine's potential to increase access to radiotherapy. Although VPs were on average younger and lived farther than their IP counterparts, only distance from clinic affected the likelihood of treatment for VPs, with patients who elected to undergo treatment living on average closer than those who did not. Similarly, IPPs were more likely to undergo treatment if they were men and living closer to the clinic. Although treatment yield was higher in the VPs living closer, they remained statistically further from the clinic compared with IPPs. Compared with VPs who did not undergo treatment, VPs undergoing treatment had a higher ADI. These findings suggest that telemedicine was effective in increasing access to specialized radiotherapy for patients living further away, especially those with higher ADI and potentially less access to specialized treatment near their home. At the onset of the pandemic, US ambulatory clinics reported up to 70% reduction in outpatient visits with variable patient volume recovery by 2021. - One study found that almost 41% of patients with cancer experienced decreases in total visits and up to 52% reported decreased IP visits. Another study found that 20% of patients with cancer reported a delay in their cancer care. In comparison, our department experienced just a 5% reduction in new patient volume in 2020, possibly related to swift adoption of telemedicine by the institution and patients. As our institution adapted to the pandemic, new patient consultations in 2021 returned to prepandemic numbers and 22% of visits remained virtual. Our VPs were on average younger and were more likely to be male. In 2021, the CDC found that 37% of adults used telemedicine within the preceding year with higher use reported by females, those older than 65 years, and Non-Hispanic American Indian or Alaskan Natives. By contrast, a report by the Department of Health and Human Services found the highest telemedicine utilizers to be young adults and White patients, with the least utilization by people older than 65 years, and Black, Asian, and Latino populations. Like our study, a medical oncology outpatient cancer with 29% virtual visits in 2020 reported higher utilization among non-Hispanic White patients. The differences in the use of telemedicine in the general population and those presenting for radiotherapy may be at least partially attributable to the older demographic of patients with cancer (who may be less privy to using telemedicine), the increase in men presenting for GU diagnoses, as well as our large catchment of nonmetropolitan patients. Our data showed that VPs presented more frequently for GU diagnoses and less often for palliative consultations. Similarly, a small study reviewing patient satisfaction with telemedicine in radiotherapy reported more use of telemedicine in younger patients and among those presenting for GU and less for palliative diagnoses. Although this may indicate a preference for palliative care closer to home, it may also indicate a need for improved telemedicine infrastructure for patients with palliative needs. Although telemedicine may hold promise for increasing access to care for disadvantaged groups, intentional effort may be needed to best reach these groups. Understanding treatment yield for IPPs and VPs was a primary and novel objective of this study. Our VPs had significantly lower yield than IPPs. VPs with GU diagnoses had the largest reduction in yield compared with IPPs, likely attributed to the decreased acuity and favorable natural history of most prostate cancers that allow for the opportunity to seek multiple medical opinions. VPs who underwent treatment lived closer than those who did not; treated VPs still lived significantly farther away than treated IPPs. This shows telemedicine's ability to reach patients who may have faced challenges accessing care or those seeking a second opinion. With lower ADI in VPs, it could be suggested that intrapandemic telemedicine reached those considered to be less health care–deprived. However, this may be explained by the unexpected increase in demand for telemedicine, which limited direct efforts to reach the medically underprivileged. With increased ADIs among VPs undergoing treatment, we show potential to expand oncology care access to patients who may not have radiotherapy near their home, including those who live in rural and/or health care–deprived areas. Most of our catchment area is considered rural and the majority of those living more than 100 miles away have high ADIs. As telemedicine becomes a more integral part of daily practice, improved identification of underserved populations, those needing telemedicine infrastructure, and those who would benefit from specialty radiotherapy is warranted. A unique aspect of our institution compared with other regional radiation oncology centers is the availability of specialty radiotherapy services such as proton radiotherapy, brachytherapy, and access to a large comprehensive cancer center with clinical trial availability. VPs who underwent treatment used proton radiotherapy 1.9 times more than IPPs. This difference could be explained by significant increases in proton therapy utilization for those with CNS, head/neck/skin, sarcoma, GU, and lymphoma diagnoses, all sites in which proton therapy indications are merited and/or rising. This suggests that the availability of a unique treatment modality may have prompted a greater conversion from consultation to treatment or that consultation was performed for review of indications for proton therapy. This highlights the potential of telemedicine, with improved infrastructure, to improve access to specialty radiation oncology care. This study has several limitations. Our center is a large tertiary cancer center with subspecialized care and treatment modalities, which may not reflect the general radiation oncology practice. Also, our patient population is generally homogenous in race. Therefore, it may not discern telemedicine and care access challenges faced by racial or ethnic minorities. Although our expanded virtual reach leaned in favor of insured, more privileged patients, we were able to treat patients further from clinic with a higher ADI and increase the access to specialized radiotherapy techniques using telemedicine services. Expansion of telemedicine to underprivileged populations is warranted, and a broader study of telemedicine in radiation oncology would benefit from increased racial, gender, and geographic diversity. In determining treatment yield, we only assessed patients who underwent radiotherapy at our center. Some patients may not have been recommended radiotherapy and the proportion may have differed between IP and VPs. Also, patients may have undergone consultation within the study window but initiated radiotherapy outside of the December 2021 study cutoff date. As a result of these factors, the actual treatment yield may have been higher than reported. Our study did not investigate the patient attitudes toward telemedicine, their ability to use telemedicine interfaces, and how attitudes may have changed due to the pandemic. Although studies show general patient satisfaction and equivalent care outcomes with telemedicine, - assessment of patient satisfaction with virtual technologies is merited for continued improvement in quality. Similarly, the effect of telemedicine on cancer care and outcomes is worthy of continued investigation. Previous studies have shown many socioeconomic metrics to be linked to decreased access to care during the COVID-19 pandemic; exploration of other indices would allow for better understanding of patient and technological infrastructure needs. - Development of tools to reach patients with limited access and/or specialty care needs is warranted. Telemedicine in radiation oncology is a developing resource with potential to enhance cancer care. Since the novelty and alarm of the COVID-19 pandemic have subsided, telemedicine has transformed from a necessity to a powerful tool that can enhance access to medical care. Telemedicine can help increase the reach of subspecialty care within radiation oncology, a technical and procedural field, and as a result increase access to other technical medical specialties. Further research is needed to improve telemedicine outreach and infrastructure. |
Evolving Face Mask Guidance During a Pandemic and Potential Harm to Public Perception: Infodemiology Study of Sentiment and Emotion on Twitter | 1714466f-efcc-4254-a29d-6df558a75b55 | 9973548 | Health Communication[mh] | Background Public health recommendations rapidly evolve when contending with a fast-developing pandemic like COVID-19, and optimized communication is critical to positively impact health-related behaviors and outcomes. Effective communication of trustworthy information has proven key to overcoming public health crises in the past, particularly when the coordinated effort of entire populations has been required . During global health crises, public institutions are considered trusted sources of information, but they face challenges in providing evidence-based guidance on real-time preventative measures . The Centers for Disease Control and Prevention (CDC) is one of the leading federal agencies in the United States charged with protecting public health. It provides primary directives for public health measures that are disseminated to the general public via various outlets, including social media platforms . Public Health Communication Messaging strategies are a key tenet of strategic communication. Public health communication in particular is driven by an ecological foundation, recognizing that public health is affected by social, behavioral, political, and environmental factors . As such, it requires multilevel strategies for disseminating information, including “tailored messages at the individual level, targeted messages at the group level, social marketing at the community level, media advocacy at the policy level, and media campaigns at the population level” . In 1993, the director of the CDC established that health communication should be considered an integral component of their prevention programs and created a 10-step messaging framework to promote changes in awareness, attitudes, and beliefs that may ultimately influence health behaviors . This framework has evolved over time, notably with the addition of the crisis and emergency risk communication (CERC) considerations in the aftermath of 9/11 and subsequent anthrax attacks . The CERC strategy generally follows a 5-phase paradigm: (1) the pre-crisis phase, involving potential response preparedness; (2) the initial phase, when the outbreak begins and information is often fluid and possibly confusing; (3) the maintenance phase, involving clarifying information on risk perceptions and correcting misinformation; (4) the resolution phase, when the outbreak is resolved; and (5) the evaluation phase, involving review of lessons learned . Over the past decade, public health organizations have struggled to adequately address public concerns during outbreaks of Ebola, H5N1 avian influenza, and Zika, and these organizations have encountered similar obstacles during the first 3 phases of the COVID-19 pandemic . This is especially apparent when countering misinformation regarding individual-level behaviors . Sentiment and Emotion Growing research demonstrates the association between trust in government and public health organizations and their effectiveness in communicating public health information for optimal individual-level compliance . According to a 2015 poll, only 19% of Americans trusted the US federal government always or most of the time, while 71% of Americans expressed trust in the CDC in 2017 . However, in 2022, trust in the CDC fell to 50% . Considering the stature of the CDC in society, its communications—especially those on social media, where they may get the most amount of attention by the general population—play an essential role in preparedness and response efforts during all phases of disease outbreaks. Health communication generally relies on adapting established theories and models of behavior for each public health campaign. These include the theory of reasoned action, health belief model, social learning/cognitive theory, extended parallel process model, diffusion of innovation, and social marketing . However, these decision-making theories do not effectively consider the influence of attitudes, emotions, and cultural norms on ultimate behaviors, as suggested by an assessment of HIV/AIDS communication campaigns for prevention . Additionally, disseminating evolving and corrective information throughout a communication campaign can also present challenges. As people receive newer information, interpretation of this updated information follows an attitude-consistent manner despite a willingness to accept the information as factual, wherein individuals are more willing to express distrust in the credibility of the information . Analysis of the CDC’s communication campaign during the Zika epidemic suggested that updated or corrective information positively impacted public health perceptions in the initial months of the epidemic and did not affect the credibility of the CDC . However, more recent research from the current COVID-19 pandemic has indicated that the positive effects of updates to information may be short-lived . The Role of Twitter During health crises, Twitter has proven effective at identifying public concerns over the health consequences of emerging disease outbreaks and tracking disease activity based on users’ health behavior . Prior studies have demonstrated that Twitter data can be used to understand public sentiment in real time and tailor individualized public health messages based on user interest and emotion . From February 2020 through May 2021, the CDC changed guidelines on face masks (herein referred to as masks) multiple times, from initially discouraging mask use at the beginning of the pandemic among non–health care workers, to recommending mask use for all individuals, to suggesting masks were optional for individuals who were vaccinated, to again recommending the use of masks for all individuals during another surge in case counts of COVID-19. During a February 2022 press call, CDC Director Rochelle Walensky cautioned that “None of us know what the future holds for us and for this virus.... And we need to be prepared and we need to be ready for whatever comes next. We want to give people a break from things like mask wearing when our levels are low, and then have the ability to reach for them again if things get worse in the future” . Given this evolving messaging toward mask guidelines, it is critical to explore how mask-related decisions thus far during the pandemic have impacted public sentiment and emotion. To evaluate how changes in CDC mask guidelines (ie, recommendation and relaxation) impacted social media discourse, this study applies methods from computational epidemiology and the social sciences to rapidly evaluate vast amounts of publicly accessible social media data . In particular, this study focuses on Twitter, given its role in the proliferation of public health communications . Our research highlights the complex issues of public health communication confronting federal agencies, particularly the CDC, and the public at large. Formally, our objective was to evaluate changes in public perception (as measured through sentiment and emotion expressed on Twitter) surrounding the April 3, 2020, and May 13, 2021, masking guidelines made by the CDC. We investigate the impacts of changing mask guidelines on public sentiment and emotions toward mask use and hypothesize that changing guidelines (1) influence public sentiment toward mask use but (2) do not change perceptions of the CDC’s credibility, specifically trustworthiness. Public health recommendations rapidly evolve when contending with a fast-developing pandemic like COVID-19, and optimized communication is critical to positively impact health-related behaviors and outcomes. Effective communication of trustworthy information has proven key to overcoming public health crises in the past, particularly when the coordinated effort of entire populations has been required . During global health crises, public institutions are considered trusted sources of information, but they face challenges in providing evidence-based guidance on real-time preventative measures . The Centers for Disease Control and Prevention (CDC) is one of the leading federal agencies in the United States charged with protecting public health. It provides primary directives for public health measures that are disseminated to the general public via various outlets, including social media platforms . Messaging strategies are a key tenet of strategic communication. Public health communication in particular is driven by an ecological foundation, recognizing that public health is affected by social, behavioral, political, and environmental factors . As such, it requires multilevel strategies for disseminating information, including “tailored messages at the individual level, targeted messages at the group level, social marketing at the community level, media advocacy at the policy level, and media campaigns at the population level” . In 1993, the director of the CDC established that health communication should be considered an integral component of their prevention programs and created a 10-step messaging framework to promote changes in awareness, attitudes, and beliefs that may ultimately influence health behaviors . This framework has evolved over time, notably with the addition of the crisis and emergency risk communication (CERC) considerations in the aftermath of 9/11 and subsequent anthrax attacks . The CERC strategy generally follows a 5-phase paradigm: (1) the pre-crisis phase, involving potential response preparedness; (2) the initial phase, when the outbreak begins and information is often fluid and possibly confusing; (3) the maintenance phase, involving clarifying information on risk perceptions and correcting misinformation; (4) the resolution phase, when the outbreak is resolved; and (5) the evaluation phase, involving review of lessons learned . Over the past decade, public health organizations have struggled to adequately address public concerns during outbreaks of Ebola, H5N1 avian influenza, and Zika, and these organizations have encountered similar obstacles during the first 3 phases of the COVID-19 pandemic . This is especially apparent when countering misinformation regarding individual-level behaviors . Growing research demonstrates the association between trust in government and public health organizations and their effectiveness in communicating public health information for optimal individual-level compliance . According to a 2015 poll, only 19% of Americans trusted the US federal government always or most of the time, while 71% of Americans expressed trust in the CDC in 2017 . However, in 2022, trust in the CDC fell to 50% . Considering the stature of the CDC in society, its communications—especially those on social media, where they may get the most amount of attention by the general population—play an essential role in preparedness and response efforts during all phases of disease outbreaks. Health communication generally relies on adapting established theories and models of behavior for each public health campaign. These include the theory of reasoned action, health belief model, social learning/cognitive theory, extended parallel process model, diffusion of innovation, and social marketing . However, these decision-making theories do not effectively consider the influence of attitudes, emotions, and cultural norms on ultimate behaviors, as suggested by an assessment of HIV/AIDS communication campaigns for prevention . Additionally, disseminating evolving and corrective information throughout a communication campaign can also present challenges. As people receive newer information, interpretation of this updated information follows an attitude-consistent manner despite a willingness to accept the information as factual, wherein individuals are more willing to express distrust in the credibility of the information . Analysis of the CDC’s communication campaign during the Zika epidemic suggested that updated or corrective information positively impacted public health perceptions in the initial months of the epidemic and did not affect the credibility of the CDC . However, more recent research from the current COVID-19 pandemic has indicated that the positive effects of updates to information may be short-lived . During health crises, Twitter has proven effective at identifying public concerns over the health consequences of emerging disease outbreaks and tracking disease activity based on users’ health behavior . Prior studies have demonstrated that Twitter data can be used to understand public sentiment in real time and tailor individualized public health messages based on user interest and emotion . From February 2020 through May 2021, the CDC changed guidelines on face masks (herein referred to as masks) multiple times, from initially discouraging mask use at the beginning of the pandemic among non–health care workers, to recommending mask use for all individuals, to suggesting masks were optional for individuals who were vaccinated, to again recommending the use of masks for all individuals during another surge in case counts of COVID-19. During a February 2022 press call, CDC Director Rochelle Walensky cautioned that “None of us know what the future holds for us and for this virus.... And we need to be prepared and we need to be ready for whatever comes next. We want to give people a break from things like mask wearing when our levels are low, and then have the ability to reach for them again if things get worse in the future” . Given this evolving messaging toward mask guidelines, it is critical to explore how mask-related decisions thus far during the pandemic have impacted public sentiment and emotion. To evaluate how changes in CDC mask guidelines (ie, recommendation and relaxation) impacted social media discourse, this study applies methods from computational epidemiology and the social sciences to rapidly evaluate vast amounts of publicly accessible social media data . In particular, this study focuses on Twitter, given its role in the proliferation of public health communications . Our research highlights the complex issues of public health communication confronting federal agencies, particularly the CDC, and the public at large. Formally, our objective was to evaluate changes in public perception (as measured through sentiment and emotion expressed on Twitter) surrounding the April 3, 2020, and May 13, 2021, masking guidelines made by the CDC. We investigate the impacts of changing mask guidelines on public sentiment and emotions toward mask use and hypothesize that changing guidelines (1) influence public sentiment toward mask use but (2) do not change perceptions of the CDC’s credibility, specifically trustworthiness. Data Collection Tweets containing at least one COVID-19–related keyword were collected using repeated searches via version 1.1 of the official Twitter application programming interface (API). The API was queried in several steps as part of a separate project conducted by team members at the University of Sydney. Starting on February 10, 2020, the Search Tweets end point was run on an automated schedule every 7 days to collect tweets based on a specific set of COVID-19–related queries . When running, the process would request 100 COVID-19–related tweets from the API, save those tweets to a database, and then request the next 100 tweets until it ran out of tweets to gather. The frequency of requests was 450 times per 15 minutes (due to rate limits imposed by the Twitter API), resulting in 45,000 tweets per 15 minutes. Starting on March 17, 2020, this process was switched to the Twitter Stream API, which had an ongoing open connection with Twitter . In this new process, whenever a tweet matching the keywords of interest was posted by a user, it was sent to the database within seconds. Analysis was restricted to original tweets (ie, retweets were omitted) in English from users based in the United States. The GeoNames geographical database was used to identify user location based on the account location field (ie, the location provided by a Twitter user in their public profile, if any). The data set was then restricted to only tweets that contained mask-related terminology, and these keywords were selected based on the collective expertise of the research team . The comparator data set was generated by first extracting tweets that contained at least one COVID-19 keyword but no mask keywords. A random number of comparator tweets were then selected for each day such that the number of comparator tweets for any given day was equivalent to the number of mask tweets on that day. For example, if there were 500 mask tweets on March 2, 500 random comparator tweets that contained COVID-19 terminology but no mask terminology would be selected for that day. The daily number of tweets used for analysis in 2020 and 2021 are provided in , Table S1, and , Table S2, respectively. Data were evaluated during 2 time periods: March 1, 2020, to June 30, 2020, and April 1, 2021, to June 13, 2021. During the first time period, on April 3, 2020, the CDC set new guidelines that cloth or fabric face coverings (eg, masks) be used as an additional and voluntary preventive measure that could protect others from COVID-19 transmission . This was a reversal of guidelines made during a tweet on February 27, 2020, which stated that the CDC did “not currently recommend the use of masks to help prevent novel coronavirus,” instead encouraging their Twitter followers (4.7 million on the main @CDCGov account as of April 5, 2022) to stay at home when sick and wash hands with soap and water to slow the spread of disease . Amid a shortage of personal protective equipment, CDC officials reasoned that this position might reduce the likelihood of stockpiling by the general public and save hospital-grade masks for health care workers . The second time period of analysis (ie, April 1, 2021, through June 13, 2021) was chosen based on a revision to the guidelines by the CDC, which noted on Twitter, “[i]f you are fully vaccinated against #COVID19, you can resume activities without wearing a mask.” Two months later, this recommendation was revoked amid a surge of the SARS-CoV-2 Delta variant . Sentiment Analysis and Emotion Analysis Links, hashtag symbols, and @ mentions were removed from tweets prior to calculating sentiment scores using the Valence Aware Dictionary and Sentiment Reasoner (VADER). This methodology, which was specifically designed for social media data, incorporates emojis, punctuation, capitalization, and negation when calculating the compound sentiment score (ranging from –1 to 1). Tweets with a score above 0.05 were labeled as “positive” and those below –0.05 were labeled as “negative”; all other tweets were labeled as “neutral” . Each tweet was also mapped to a set of emotions based on the National Research Council of Canada (NRC) Word Emotion Lexicon . The NRC associates each word with at least 1 of 8 emotions—anger, anticipation, disgust, fear, joy, sadness, surprise, and trust—on a scale from 0 to 1. Before calculating emotionality, all HTML escape characters, stop words, punctuation, and numbers were removed, followed by conversion to lower case and tokenization. For a given tweet, the final score corresponding to each emotion was calculated by summing emotion scores across tokens corresponding to that emotion. Statistical Analysis An interrupted time series analysis was used to evaluate the change in sentiment and emotion outcomes around the 2 shifts in guidelines. Each model contained a term for the pre-event trend (ie, recommendation for mask use or relaxation of this recommendation), an instantaneous effect on the day of the event, and a postevent trend. For each year, the outcomes of interest included change in average daily compound sentiment score, percent of tweets with a given sentiment (ie, positive, negative, or neutral, with individual models for each sentiment), and total emotion score (ie, the sum of words tagged with a given emotion of interest, with individual models for each emotion). For all outcomes, models were evaluated individually and relative to the comparator data set. Analysis was conducted in R (version 4.1.2; R Foundation for Statistical Computing) using the RStudio Integrated Development Environment (version 2021.09.0). A P value of less than .05 ( P <.05) was considered statistically significant, and the authors determined there were not enough statistical comparisons to warrant additional hypothesis correction methods. This was due to the exploratory nature of this study and the decision that type II errors (eg, failing to identify a true association) were more deleterious than type I errors (eg, identifying a spurious association) . Tweets containing at least one COVID-19–related keyword were collected using repeated searches via version 1.1 of the official Twitter application programming interface (API). The API was queried in several steps as part of a separate project conducted by team members at the University of Sydney. Starting on February 10, 2020, the Search Tweets end point was run on an automated schedule every 7 days to collect tweets based on a specific set of COVID-19–related queries . When running, the process would request 100 COVID-19–related tweets from the API, save those tweets to a database, and then request the next 100 tweets until it ran out of tweets to gather. The frequency of requests was 450 times per 15 minutes (due to rate limits imposed by the Twitter API), resulting in 45,000 tweets per 15 minutes. Starting on March 17, 2020, this process was switched to the Twitter Stream API, which had an ongoing open connection with Twitter . In this new process, whenever a tweet matching the keywords of interest was posted by a user, it was sent to the database within seconds. Analysis was restricted to original tweets (ie, retweets were omitted) in English from users based in the United States. The GeoNames geographical database was used to identify user location based on the account location field (ie, the location provided by a Twitter user in their public profile, if any). The data set was then restricted to only tweets that contained mask-related terminology, and these keywords were selected based on the collective expertise of the research team . The comparator data set was generated by first extracting tweets that contained at least one COVID-19 keyword but no mask keywords. A random number of comparator tweets were then selected for each day such that the number of comparator tweets for any given day was equivalent to the number of mask tweets on that day. For example, if there were 500 mask tweets on March 2, 500 random comparator tweets that contained COVID-19 terminology but no mask terminology would be selected for that day. The daily number of tweets used for analysis in 2020 and 2021 are provided in , Table S1, and , Table S2, respectively. Data were evaluated during 2 time periods: March 1, 2020, to June 30, 2020, and April 1, 2021, to June 13, 2021. During the first time period, on April 3, 2020, the CDC set new guidelines that cloth or fabric face coverings (eg, masks) be used as an additional and voluntary preventive measure that could protect others from COVID-19 transmission . This was a reversal of guidelines made during a tweet on February 27, 2020, which stated that the CDC did “not currently recommend the use of masks to help prevent novel coronavirus,” instead encouraging their Twitter followers (4.7 million on the main @CDCGov account as of April 5, 2022) to stay at home when sick and wash hands with soap and water to slow the spread of disease . Amid a shortage of personal protective equipment, CDC officials reasoned that this position might reduce the likelihood of stockpiling by the general public and save hospital-grade masks for health care workers . The second time period of analysis (ie, April 1, 2021, through June 13, 2021) was chosen based on a revision to the guidelines by the CDC, which noted on Twitter, “[i]f you are fully vaccinated against #COVID19, you can resume activities without wearing a mask.” Two months later, this recommendation was revoked amid a surge of the SARS-CoV-2 Delta variant . Links, hashtag symbols, and @ mentions were removed from tweets prior to calculating sentiment scores using the Valence Aware Dictionary and Sentiment Reasoner (VADER). This methodology, which was specifically designed for social media data, incorporates emojis, punctuation, capitalization, and negation when calculating the compound sentiment score (ranging from –1 to 1). Tweets with a score above 0.05 were labeled as “positive” and those below –0.05 were labeled as “negative”; all other tweets were labeled as “neutral” . Each tweet was also mapped to a set of emotions based on the National Research Council of Canada (NRC) Word Emotion Lexicon . The NRC associates each word with at least 1 of 8 emotions—anger, anticipation, disgust, fear, joy, sadness, surprise, and trust—on a scale from 0 to 1. Before calculating emotionality, all HTML escape characters, stop words, punctuation, and numbers were removed, followed by conversion to lower case and tokenization. For a given tweet, the final score corresponding to each emotion was calculated by summing emotion scores across tokens corresponding to that emotion. An interrupted time series analysis was used to evaluate the change in sentiment and emotion outcomes around the 2 shifts in guidelines. Each model contained a term for the pre-event trend (ie, recommendation for mask use or relaxation of this recommendation), an instantaneous effect on the day of the event, and a postevent trend. For each year, the outcomes of interest included change in average daily compound sentiment score, percent of tweets with a given sentiment (ie, positive, negative, or neutral, with individual models for each sentiment), and total emotion score (ie, the sum of words tagged with a given emotion of interest, with individual models for each emotion). For all outcomes, models were evaluated individually and relative to the comparator data set. Analysis was conducted in R (version 4.1.2; R Foundation for Statistical Computing) using the RStudio Integrated Development Environment (version 2021.09.0). A P value of less than .05 ( P <.05) was considered statistically significant, and the authors determined there were not enough statistical comparisons to warrant additional hypothesis correction methods. This was due to the exploratory nature of this study and the decision that type II errors (eg, failing to identify a true association) were more deleterious than type I errors (eg, identifying a spurious association) . April 3, 2020, CDC Mask Recommendation Guideline There were 1,106,756 mask-related tweets during the 4-month period surrounding the first guideline (ie, the CDC mask recommendation) with an equivalent quantity collected for the comparator. Between February 29, 2020, and June 30, 2020, mask-related tweets were more positive than comparator COVID-19 tweets (β=.06, 95% CI .05-.07; P <.001; ). In particular, the percent of positive tweets on any given day was 4.43 percentage points higher than concurrently observed in the comparator (95% CI 3.82-5.03; P <.001), while the percent of neutral tweets was lower (β=–3.94, 95% CI –4.68 to –3.21; P <.001). After the mask recommendation on April 3, 2020, the proportion of negative tweets within the mask-related data set increased (β=.51, 95% CI .43-.59; P <.001). However, the average number of negative tweets on any given day was not substantially different from the comparator (β=–.49, 95% CI –1.31 to .33; P =.24; ). In terms of emotion, mask-related tweets expressed an increasing level of trust (β=.004, 95% CI .003-.004; P <.001) but decreasing levels of both sadness (β=–.003, 95% CI –.004 to –.002 P <.001) and surprise (β=–.001, 95% CI –.001 to 0; P =.005) during the period preceding the April 3, 2020, CDC recommendation. However, the levels of sadness (β=.004, 95% CI .003-.005; P <.001) and surprise (β=.001, 95% CI 0-.001; P =.003) expressed in mask-related tweets increased following the CDC recommendation, while trust decreased (β=–.004, 95% CI –.004 to –.003; P <.001). The levels of anger, anticipation, disgust, or joy expressed on any given day did not substantially differ between the mask-related data set and the comparator. However, mask-related tweets expressed a higher level of trust (β=.131, 95% CI .122-.140; P <.001), but less sadness (β=–.042, 95% CI –.053 to –.031; P <.001) and surprise (β=–.026, 95% CI –.03 to –.021; P <.001) relative to the comparator data set. May 13, 2021, CDC Mask Relaxation Guideline There were 321,119 mask-related tweets during the 10-week period surrounding the second guideline shift (ie, the CDC mask relaxation), with an equivalent amount in the comparator. On any given day between April 1, 2021, and June 13, 2021, sentiment expressed in mask-related tweets was more negative than the comparator (β=–.06, 95% CI –.05 to –.06; P <.001; ). In particular, the proportion of negative tweets within the mask-related data set was 9.50 percentage points higher on average than in the comparator (95% CI 8.74-10.3; P <.001). During the same time period, the proportion of neutral tweets was 8.74 percentage points lower (95% CI –9.31 to –8.17; P <.001), and the proportion of positive tweets was 0.76 percentage points lower (95% CI –1.37 to –0.15; P =.02). Immediately after the mask relaxation on May 13, the proportion of negative tweets increased (β=3.43, 95% CI 1.61-5.26; P <.001), whereas the percent of neutral tweets decreased (β=–4.46, 95% CI –7.07 to –1.84; P =.001) On any given day, and in all categories except the emotion of surprise (β=–.004, 95% CI –.009 to .001; P =.09), mask-related tweets expressed higher levels of emotion than tweets in the comparator. Before the mask recommendation was revoked, the levels of anger (β=.001, 95% CI 0-.001; P =.007), fear (β=.001, 95% CI .001-.002; P <.001), sadness (β=.001, 95% CI 0-.002; P =.001), and trust (β=.001, 95% CI 0-.001; P <.001) expressed in mask-related tweets increased daily. Following the mask recommendation relaxation, the level of anger continued to increase (β=.001, 95% CI 0-.002; P =.02), whereas trust decreased (β=–.001, 95% CI –.002 to 0; P =.008). There were 1,106,756 mask-related tweets during the 4-month period surrounding the first guideline (ie, the CDC mask recommendation) with an equivalent quantity collected for the comparator. Between February 29, 2020, and June 30, 2020, mask-related tweets were more positive than comparator COVID-19 tweets (β=.06, 95% CI .05-.07; P <.001; ). In particular, the percent of positive tweets on any given day was 4.43 percentage points higher than concurrently observed in the comparator (95% CI 3.82-5.03; P <.001), while the percent of neutral tweets was lower (β=–3.94, 95% CI –4.68 to –3.21; P <.001). After the mask recommendation on April 3, 2020, the proportion of negative tweets within the mask-related data set increased (β=.51, 95% CI .43-.59; P <.001). However, the average number of negative tweets on any given day was not substantially different from the comparator (β=–.49, 95% CI –1.31 to .33; P =.24; ). In terms of emotion, mask-related tweets expressed an increasing level of trust (β=.004, 95% CI .003-.004; P <.001) but decreasing levels of both sadness (β=–.003, 95% CI –.004 to –.002 P <.001) and surprise (β=–.001, 95% CI –.001 to 0; P =.005) during the period preceding the April 3, 2020, CDC recommendation. However, the levels of sadness (β=.004, 95% CI .003-.005; P <.001) and surprise (β=.001, 95% CI 0-.001; P =.003) expressed in mask-related tweets increased following the CDC recommendation, while trust decreased (β=–.004, 95% CI –.004 to –.003; P <.001). The levels of anger, anticipation, disgust, or joy expressed on any given day did not substantially differ between the mask-related data set and the comparator. However, mask-related tweets expressed a higher level of trust (β=.131, 95% CI .122-.140; P <.001), but less sadness (β=–.042, 95% CI –.053 to –.031; P <.001) and surprise (β=–.026, 95% CI –.03 to –.021; P <.001) relative to the comparator data set. There were 321,119 mask-related tweets during the 10-week period surrounding the second guideline shift (ie, the CDC mask relaxation), with an equivalent amount in the comparator. On any given day between April 1, 2021, and June 13, 2021, sentiment expressed in mask-related tweets was more negative than the comparator (β=–.06, 95% CI –.05 to –.06; P <.001; ). In particular, the proportion of negative tweets within the mask-related data set was 9.50 percentage points higher on average than in the comparator (95% CI 8.74-10.3; P <.001). During the same time period, the proportion of neutral tweets was 8.74 percentage points lower (95% CI –9.31 to –8.17; P <.001), and the proportion of positive tweets was 0.76 percentage points lower (95% CI –1.37 to –0.15; P =.02). Immediately after the mask relaxation on May 13, the proportion of negative tweets increased (β=3.43, 95% CI 1.61-5.26; P <.001), whereas the percent of neutral tweets decreased (β=–4.46, 95% CI –7.07 to –1.84; P =.001) On any given day, and in all categories except the emotion of surprise (β=–.004, 95% CI –.009 to .001; P =.09), mask-related tweets expressed higher levels of emotion than tweets in the comparator. Before the mask recommendation was revoked, the levels of anger (β=.001, 95% CI 0-.001; P =.007), fear (β=.001, 95% CI .001-.002; P <.001), sadness (β=.001, 95% CI 0-.002; P =.001), and trust (β=.001, 95% CI 0-.001; P <.001) expressed in mask-related tweets increased daily. Following the mask recommendation relaxation, the level of anger continued to increase (β=.001, 95% CI 0-.002; P =.02), whereas trust decreased (β=–.001, 95% CI –.002 to 0; P =.008). Principal Findings This study is among the first to characterize the evolution of mask-related content on Twitter surrounding the recommendation and relaxation of mask guidelines by the CDC during the COVID-19 pandemic. In summary, our study found that after both the 2020 mask recommendation and the 2021 mask relaxation a pronounced decrease in neutral tweets occurred. Following the 2020 mask recommendation, sentiment expressed in mask-related tweets was substantially more positive than in other COVID-19 tweets. In contrast, sentiment expressed in mask-related tweets following the 2021 mask relaxation was more negative. Furthermore, both mask-related data sets suggested higher levels of emotions than other COVID-19 tweets. In particular, both time periods were marked by a higher proportion of tweets expressing disgust before the change in guidelines and lower proportion of tweets expressing trust following the change. Our main findings suggest that shifts in guidelines emanating from the CDC may have a tangible, negative impact on the perception of mask use among United States–based Twitter users, with implications for the design of mask-wearing policies and other similar preventative health measures in the future. Masks are a crucial public health tool to fight the spread of infections such as SARS-CoV-2. High adherence to mask-wearing policies may help reduce transmission during severe disease outbreaks, including pandemics . However, mask use in the United States has become increasingly politicized and polarizing. Recent work evaluating the state of mask-related discourse on Twitter found that corresponding tweets expressed increasingly negative sentiment between March and July 2020, although that research did not focus on CDC announcements as interventions or include an extended time period after the relaxation . Other research suggests that anti-mask rhetoric accounted for 10% of mask-related content between January and October 2020, with varying volume around key US guideline shifts . These results corroborate our findings, namely that the mask-related discourse on Twitter was increasingly more polarized after the CDC announced the mask recommendation on April 3, 2020. As online information-seeking behaviors increase, so do access and exposure to conflicting information and political infighting . False information quickly and easily spreads via online social networks and, in tandem with fluctuating and confusing messaging during the initial phase of a public health emergency, promotes negative public sentiments and difficulties in preserving public trust . Recent research indicates that efforts to disseminate corrective information during the maintenance phase of a public health crisis are ineffective at both countering misconceptions and gaining support for the adoption of preventive health-related behaviors . This finding suggests that, despite the quickly changing atmosphere, concise and consistent messaging is critical in the precrisis and initial phases of a public health emergency for highest individual-level adherence to preparedness and prevention measures. While the CDC attempted to provide clear messaging regarding mask use, its response was perceived as slow relative to the speed at which clinical findings were released. Furthermore, this perceived slow response, coupled with positions that conflicted with other global health organizations, such as the World Health Organization, may have inadvertently contributed to feelings of confusion and mistrust among the general public . This effect may have been captured within our data set as the decreased levels of trust-related terminology expressed within tweets following each shift in guidelines. Furthermore, the fact that mask tweets within our data were substantially more negative than the comparator in 2021 may suggest a high degree of preexisting mask fatigue, and the subsequent additional increase in negative tweets following the relaxation recommendation on May 13 may indicate discontent at the lack of transparency from the CDC. Health Communications Recommendations Although Twitter and other social media platforms can be leveraged to rapidly inform the public of important recommendations, this study suggests that there may be negative consequences for public support when such messages are not communicated effectively. In our study, this is illustrated by the decrease in levels of trust expressed by United States–based Twitter users following both guideline shifts in 2020 and 2021 . Based on these findings, we believe that there are several communication strategies that should be considered during future health emergencies to ensure that the general public maintains trust in government agencies. First, it is imperative that a consistent message is embraced by diverse, respected professionals in the field. Along with trusted government agencies like the CDC, this may also include public health and medical experts, research scientists, politicians, science communications specialists, and even popular influencers and celebrities in order to reach multiple demographics . This message should be authentic and transparent about the fact that information will likely evolve, especially during ongoing crises. Second, it is important for government agencies to monitor social media engagement and promote dialogue to understand perceptions and motives for health practice. Each social media platform reaches a different target audience, so multiple accounts across platforms may be warranted to ensure that as many individual opinions are considered as possible. While social media is not generalizable to the entire population, it can help supplement traditional epidemiologic measures of data collection, such as representative surveys, that may be more reliable but are more costly to coordinate. Third, it may be salient for government agencies to develop educational materials that directly address and correct incorrect perceptions, attitudes, and behaviors. These materials must be “living” documents that are continuously updated as new misperceptions emerge. They should also be made widely accessible and promoted through multiple media outlets, including social media. Taken together, the increased transparency and access afforded by consistent messaging, increased social media engagement, and easily understood education materials could help ensure that the general public continues to look to government agencies for guidance during future health emergencies, especially those that are tumultuous. Limitations and Future Directions Our study is the first to evaluate the sentiment and emotion of mask-related tweets in the United States surrounding 2 key guideline shifts made by the CDC relative to a matched comparator data set of other COVID-19 tweets during the same period. However, there are several limitations to note. First, the reliance on keywords to collect relevant tweets may introduce some selection bias. Specifically, filtering tweets with keywords may exclude tweets that discuss the topic of interest but contain a misspelling. Additionally, some tweets, such as automated advertisements, may contain the appropriate keywords but are not relevant to public opinion. Given the persuasive nature of advertising, it is likely that their inadvertent inclusion might have biased our results and skewed the estimation of positive sentiment to be higher than that which was present in the general public. Future work could use the –is:nullcast filter, which was not available in the version of the Twitter API that was used to collect the data for this study (version 1.1), to ensure that these tweets were removed. Second, tweets were restricted to those posted by users located within the United States based on the geotag in the user profile. However, users reporting location information in their profile may be different from those without such content. Future work should attempt to identify and leverage other methods to assess where Twitter users are located. Third, sociodemographic data were not available, which may impact generalizability. While social media studies can provide rapid insights during health emergencies, they are not necessarily representative of the overall US population; specifically, Twitter users tend to be younger, more educated, and have a higher average income than the general US population . Fourth, findings are based on aggregate analysis at the national level, and future work could characterize patterns at a state level. Lastly, future work could employ alternative natural language processing and sentiment analysis methods, such as emoji analysis or word embeddings, to understand how results may change. Conclusions Our study supports findings from prior research on the importance of formulating clear public health communications and disseminating accurate public health guidance on social media. Specifically, we found that tweets surrounding the 2020 mask recommendation and 2021 mask relaxation were more polarizing and contained less trust-related terminology than those before the guidelines were announced. Furthermore, while mask-related tweets posted in 2020 were more positive than other COVID-19 tweets, mask-related tweets in 2021 were more negative. The change in sentiment observed in 2021 may signal frustration among Twitter users about public health discourse centered around masks and recognition that the initial mask relaxation change may have been premature. Gaining insight into how the general public engages on social media platforms, perceives preventative public health measures imposed during the COVID-19 pandemic, and reacts to shifts in guidelines declared by the US government is of utmost importance for policy makers, health workers, and interested stakeholders. Official communications that include concise information backed by systematic data are critical to ensure widespread adoption and sustained adherence to public health interventions. However, the rapid spread of COVID-19 and the evolving evidence around its mitigation led to confusion from the public surrounding the fluctuating mask guidelines. When messaging remains unclear and lacks direction, public sentiment and trust in authoritative entities erode. This is especially true for masks, where policy recommendations pertaining to mask use constantly shifted throughout 2020 and 2021, sometimes without clear evidence presented to the public . Given that health officials have noted that mask guidelines may serve as a recurring tool to mitigate contagion spread during peak infection (both in the current pandemic and in response to future pandemic threats or emerging biothreats) it is imperative that institutions such as the CDC use consistent, clear communication strategies that align with other major health organizations and the broader scientific community. This will ensure that the potential for polarization is minimized while trust in the government and adherence to preventive measures is maximized. This study is among the first to characterize the evolution of mask-related content on Twitter surrounding the recommendation and relaxation of mask guidelines by the CDC during the COVID-19 pandemic. In summary, our study found that after both the 2020 mask recommendation and the 2021 mask relaxation a pronounced decrease in neutral tweets occurred. Following the 2020 mask recommendation, sentiment expressed in mask-related tweets was substantially more positive than in other COVID-19 tweets. In contrast, sentiment expressed in mask-related tweets following the 2021 mask relaxation was more negative. Furthermore, both mask-related data sets suggested higher levels of emotions than other COVID-19 tweets. In particular, both time periods were marked by a higher proportion of tweets expressing disgust before the change in guidelines and lower proportion of tweets expressing trust following the change. Our main findings suggest that shifts in guidelines emanating from the CDC may have a tangible, negative impact on the perception of mask use among United States–based Twitter users, with implications for the design of mask-wearing policies and other similar preventative health measures in the future. Masks are a crucial public health tool to fight the spread of infections such as SARS-CoV-2. High adherence to mask-wearing policies may help reduce transmission during severe disease outbreaks, including pandemics . However, mask use in the United States has become increasingly politicized and polarizing. Recent work evaluating the state of mask-related discourse on Twitter found that corresponding tweets expressed increasingly negative sentiment between March and July 2020, although that research did not focus on CDC announcements as interventions or include an extended time period after the relaxation . Other research suggests that anti-mask rhetoric accounted for 10% of mask-related content between January and October 2020, with varying volume around key US guideline shifts . These results corroborate our findings, namely that the mask-related discourse on Twitter was increasingly more polarized after the CDC announced the mask recommendation on April 3, 2020. As online information-seeking behaviors increase, so do access and exposure to conflicting information and political infighting . False information quickly and easily spreads via online social networks and, in tandem with fluctuating and confusing messaging during the initial phase of a public health emergency, promotes negative public sentiments and difficulties in preserving public trust . Recent research indicates that efforts to disseminate corrective information during the maintenance phase of a public health crisis are ineffective at both countering misconceptions and gaining support for the adoption of preventive health-related behaviors . This finding suggests that, despite the quickly changing atmosphere, concise and consistent messaging is critical in the precrisis and initial phases of a public health emergency for highest individual-level adherence to preparedness and prevention measures. While the CDC attempted to provide clear messaging regarding mask use, its response was perceived as slow relative to the speed at which clinical findings were released. Furthermore, this perceived slow response, coupled with positions that conflicted with other global health organizations, such as the World Health Organization, may have inadvertently contributed to feelings of confusion and mistrust among the general public . This effect may have been captured within our data set as the decreased levels of trust-related terminology expressed within tweets following each shift in guidelines. Furthermore, the fact that mask tweets within our data were substantially more negative than the comparator in 2021 may suggest a high degree of preexisting mask fatigue, and the subsequent additional increase in negative tweets following the relaxation recommendation on May 13 may indicate discontent at the lack of transparency from the CDC. Although Twitter and other social media platforms can be leveraged to rapidly inform the public of important recommendations, this study suggests that there may be negative consequences for public support when such messages are not communicated effectively. In our study, this is illustrated by the decrease in levels of trust expressed by United States–based Twitter users following both guideline shifts in 2020 and 2021 . Based on these findings, we believe that there are several communication strategies that should be considered during future health emergencies to ensure that the general public maintains trust in government agencies. First, it is imperative that a consistent message is embraced by diverse, respected professionals in the field. Along with trusted government agencies like the CDC, this may also include public health and medical experts, research scientists, politicians, science communications specialists, and even popular influencers and celebrities in order to reach multiple demographics . This message should be authentic and transparent about the fact that information will likely evolve, especially during ongoing crises. Second, it is important for government agencies to monitor social media engagement and promote dialogue to understand perceptions and motives for health practice. Each social media platform reaches a different target audience, so multiple accounts across platforms may be warranted to ensure that as many individual opinions are considered as possible. While social media is not generalizable to the entire population, it can help supplement traditional epidemiologic measures of data collection, such as representative surveys, that may be more reliable but are more costly to coordinate. Third, it may be salient for government agencies to develop educational materials that directly address and correct incorrect perceptions, attitudes, and behaviors. These materials must be “living” documents that are continuously updated as new misperceptions emerge. They should also be made widely accessible and promoted through multiple media outlets, including social media. Taken together, the increased transparency and access afforded by consistent messaging, increased social media engagement, and easily understood education materials could help ensure that the general public continues to look to government agencies for guidance during future health emergencies, especially those that are tumultuous. Our study is the first to evaluate the sentiment and emotion of mask-related tweets in the United States surrounding 2 key guideline shifts made by the CDC relative to a matched comparator data set of other COVID-19 tweets during the same period. However, there are several limitations to note. First, the reliance on keywords to collect relevant tweets may introduce some selection bias. Specifically, filtering tweets with keywords may exclude tweets that discuss the topic of interest but contain a misspelling. Additionally, some tweets, such as automated advertisements, may contain the appropriate keywords but are not relevant to public opinion. Given the persuasive nature of advertising, it is likely that their inadvertent inclusion might have biased our results and skewed the estimation of positive sentiment to be higher than that which was present in the general public. Future work could use the –is:nullcast filter, which was not available in the version of the Twitter API that was used to collect the data for this study (version 1.1), to ensure that these tweets were removed. Second, tweets were restricted to those posted by users located within the United States based on the geotag in the user profile. However, users reporting location information in their profile may be different from those without such content. Future work should attempt to identify and leverage other methods to assess where Twitter users are located. Third, sociodemographic data were not available, which may impact generalizability. While social media studies can provide rapid insights during health emergencies, they are not necessarily representative of the overall US population; specifically, Twitter users tend to be younger, more educated, and have a higher average income than the general US population . Fourth, findings are based on aggregate analysis at the national level, and future work could characterize patterns at a state level. Lastly, future work could employ alternative natural language processing and sentiment analysis methods, such as emoji analysis or word embeddings, to understand how results may change. Our study supports findings from prior research on the importance of formulating clear public health communications and disseminating accurate public health guidance on social media. Specifically, we found that tweets surrounding the 2020 mask recommendation and 2021 mask relaxation were more polarizing and contained less trust-related terminology than those before the guidelines were announced. Furthermore, while mask-related tweets posted in 2020 were more positive than other COVID-19 tweets, mask-related tweets in 2021 were more negative. The change in sentiment observed in 2021 may signal frustration among Twitter users about public health discourse centered around masks and recognition that the initial mask relaxation change may have been premature. Gaining insight into how the general public engages on social media platforms, perceives preventative public health measures imposed during the COVID-19 pandemic, and reacts to shifts in guidelines declared by the US government is of utmost importance for policy makers, health workers, and interested stakeholders. Official communications that include concise information backed by systematic data are critical to ensure widespread adoption and sustained adherence to public health interventions. However, the rapid spread of COVID-19 and the evolving evidence around its mitigation led to confusion from the public surrounding the fluctuating mask guidelines. When messaging remains unclear and lacks direction, public sentiment and trust in authoritative entities erode. This is especially true for masks, where policy recommendations pertaining to mask use constantly shifted throughout 2020 and 2021, sometimes without clear evidence presented to the public . Given that health officials have noted that mask guidelines may serve as a recurring tool to mitigate contagion spread during peak infection (both in the current pandemic and in response to future pandemic threats or emerging biothreats) it is imperative that institutions such as the CDC use consistent, clear communication strategies that align with other major health organizations and the broader scientific community. This will ensure that the potential for polarization is minimized while trust in the government and adherence to preventive measures is maximized. |
Noise-induced hearing loss in the pre-industrial era: early contributions in | 3b11412d-b503-4ce3-906c-fef5e546aca5 | 10772019 | Preventive Medicine[mh] | It is well established that continual exposure to loud sounds at the workplace can cause hearing disorders. Noise-induced hearing loss is the second most common cause of sensorineural hearing loss after age-related hearing loss and affects approximately 5 per cent of the population. This disorder is the most prevalent occupational disease worldwide. An estimated 16 per cent of adult hearing loss cases are associated with exposure to noise in the workplace, and more than 10 per cent of workers in developed countries may suffer from noise-induced hearing loss. , Otolaryngologists can diagnose these disorders and work with occupational health physicians to define and facilitate workplace accommodations. Noise-induced hearing loss is historically associated with the introduction of manufacturing machines powered from steam engines, starting with the Industrial Revolution. However, noise-induced hearing loss has been described in pre-industrial literature, particularly in the work of the Italian physician Bernardino Ramazzini, who is universally credited as the founder of occupational medicine. – Ramazzini was born in Carpi, Italy, in 1633. After his medical graduation in 1659 in Parma, he worked as a physician in Rome and in the Duchy of Castro, a vassal state to the Papal States. In 1682, he was appointed Professor of Theoretical Medicine at the University of Modena by the Duke Francesco II d'Este (1660–1694). While in this role, he wrote several medical works, with a particular focus on population studies. At the end of the century, Ramazzini started to investigate the influence of occupations on workers’ health and taught a class on this subject at the University of Modena. He published the treatise De Morbis Artificum in 1700, which is the first publication on work-related health problems in different occupations. Each chapter of the treatise contains a description of the disease associated with a particular work activity followed by a literature analysis, workplace description, questions for the workers, disease description, remedies and advice. He analysed the effects of chemicals, dusts, repetitive motions and awkward postures on the health of different classes of workers and modernly sustained that rulers should protect the health of workers to preserve the workforce and the productivity of their states. In 1700, Ramazzini's renown enabled him to move to the University of Padua, where he was appointed as Professor of Practical Medicine. Towards the end of his life, he continued to teach students and published a new edition of De Morbis Artificum (1713), which included new occupations. He died in Padua in 1714, but his work had an immediate and lasting impact, inspiring the studies on occupational health in the following centuries. Noise-induced hearing loss in coppersmiths and corn millers In Chapter XXII (‘Diseases of bakers and millers’) of De Morbis Artificum , Ramazzini reported that corn millers often suffered from hearing loss. In particular, he wrote, ‘nearly all of them are half-deaf [ surdastri ] because they spend all night and day surrounded by the repetitive noise of wheels and millstones and the roar of water falling from a height; the tympanum, subjected to more powerful noise than it can tolerate and unremitting blows loses its tone’. Another accurate description of the effects of exposure to noise in the workplace can be found in the second edition of De Morbis Artificum , dated back to 1713. Regarding the diseases of coppersmiths (Chapter V), Ramazzini stated: one can observe these workers […] bent over all day hammering, first with wood, then with iron, working the newly mined copper until it is as ductile as required. It is obvious that this constant din hurts both their ears and heads. In fact, these workers become half deaf [ surdastri ] and, if they continue this profession, stone-deaf [ et, ubi in hoc opera consenuerint, omnino surdi ]. Because of that constant percussion, the tympanum of the ear loses its natural tension and the repercussion of the internal air [ aeris interni ] towards the outside weakens and subverts the entire hearing apparatus [ auditus organa ]. These texts demonstrated that Ramazzini was aware of the effects of noise on hearing, distinguishing hearing loss ( surdastri ) from deafness ( surdi ). He also recognised that this disorder is irreversible and progressive while the exposure to noise continues (‘if they continue this profession’). Ramazzini believed that hearing loss occurred when the tympanum was damaged by constant loud sound, reducing its natural tension and rendering it ‘weak’. The Italian physician was apparently ignorant of studies on the ossicles of the middle ear by Renaissance anatomists (Vesalius, Falloppius, Eustachi and Ingrassia) and the works of his contemporaries, such as Antonio Maria Valsalva (1666–1723), who published his masterpiece De aure humana tractatus in 1704. , This apparent lack of knowledge is not surprising; Ramazzini was an adherent of the neo-Hippocratism doctrine and followed the teachings of Hippocrates and other ancient physicians. This approach was inspired by the English physician Thomas Sydenham (1624–1689) and the French scholar Guillaume de Baillou (1538–1616). Ramazzini based his thinking on the doctrines of Hippocrates, who first described the tympanic membrane in his treatise De carnibus , and on those of Aristotle, who argued that audition took place within a completely closed ear cavity filled with air ( aer innatus ). The Aristotelian concept of aer innatus seems to be referenced by Ramazzini's use of the words aer internus . In his accounts of the health conditions of coppersmiths, Ramazzini seems to describe tinnitus, a symptom that is strongly associated with hearing loss. As he stated, ‘the workers frequently complain of a ringing in their ears [ aurium sonitus ], which they believe to be a bad omen, but only because according to Hippocrates, such noises are a dangerous symptom. It is not at all surprising that the hearing of coppersmiths has been impaired and […] such noises are accentuated’. Finally, Ramazzini proved to be pioneering in his suggestion of preventive measures and treatment for hearing disorders in coppersmiths. He suggests that ‘their ear could be plugged with cotton to protect the internal parts and, when they are bruised and battered by the constant noise, they could be rubbed with sweet almond oil’. In Chapter XXII (‘Diseases of bakers and millers’) of De Morbis Artificum , Ramazzini reported that corn millers often suffered from hearing loss. In particular, he wrote, ‘nearly all of them are half-deaf [ surdastri ] because they spend all night and day surrounded by the repetitive noise of wheels and millstones and the roar of water falling from a height; the tympanum, subjected to more powerful noise than it can tolerate and unremitting blows loses its tone’. Another accurate description of the effects of exposure to noise in the workplace can be found in the second edition of De Morbis Artificum , dated back to 1713. Regarding the diseases of coppersmiths (Chapter V), Ramazzini stated: one can observe these workers […] bent over all day hammering, first with wood, then with iron, working the newly mined copper until it is as ductile as required. It is obvious that this constant din hurts both their ears and heads. In fact, these workers become half deaf [ surdastri ] and, if they continue this profession, stone-deaf [ et, ubi in hoc opera consenuerint, omnino surdi ]. Because of that constant percussion, the tympanum of the ear loses its natural tension and the repercussion of the internal air [ aeris interni ] towards the outside weakens and subverts the entire hearing apparatus [ auditus organa ]. These texts demonstrated that Ramazzini was aware of the effects of noise on hearing, distinguishing hearing loss ( surdastri ) from deafness ( surdi ). He also recognised that this disorder is irreversible and progressive while the exposure to noise continues (‘if they continue this profession’). Ramazzini believed that hearing loss occurred when the tympanum was damaged by constant loud sound, reducing its natural tension and rendering it ‘weak’. The Italian physician was apparently ignorant of studies on the ossicles of the middle ear by Renaissance anatomists (Vesalius, Falloppius, Eustachi and Ingrassia) and the works of his contemporaries, such as Antonio Maria Valsalva (1666–1723), who published his masterpiece De aure humana tractatus in 1704. , This apparent lack of knowledge is not surprising; Ramazzini was an adherent of the neo-Hippocratism doctrine and followed the teachings of Hippocrates and other ancient physicians. This approach was inspired by the English physician Thomas Sydenham (1624–1689) and the French scholar Guillaume de Baillou (1538–1616). Ramazzini based his thinking on the doctrines of Hippocrates, who first described the tympanic membrane in his treatise De carnibus , and on those of Aristotle, who argued that audition took place within a completely closed ear cavity filled with air ( aer innatus ). The Aristotelian concept of aer innatus seems to be referenced by Ramazzini's use of the words aer internus . In his accounts of the health conditions of coppersmiths, Ramazzini seems to describe tinnitus, a symptom that is strongly associated with hearing loss. As he stated, ‘the workers frequently complain of a ringing in their ears [ aurium sonitus ], which they believe to be a bad omen, but only because according to Hippocrates, such noises are a dangerous symptom. It is not at all surprising that the hearing of coppersmiths has been impaired and […] such noises are accentuated’. Finally, Ramazzini proved to be pioneering in his suggestion of preventive measures and treatment for hearing disorders in coppersmiths. He suggests that ‘their ear could be plugged with cotton to protect the internal parts and, when they are bruised and battered by the constant noise, they could be rubbed with sweet almond oil’. Bernardino Ramazzini can be credited as the first author to accurately describe noise-induced hearing loss among specific classes of workers in the pre-industrial era. Although his knowledge remained anchored in ancient medicine and was not current with the discoveries in physiology and anatomy that occurred during the Renaissance, Ramazzini demonstrated unprecedented attention to workers’ suffering and suggested remedies and preventive measures such as archaic earplugs. Recently, Thurston criticised Ramazzini for not discussing the risk of blacksmiths becoming deaf ‘even though they were actively producing loud impact sounds with their hammer blows, and were surrounded by other noises common to smithies’. This criticism may be correct if one considers that Ramazzini was always very careful and detailed in his description of the diseases that affected workers. As noted in the present paper, Ramazzini added a chapter on the diseases of coppersmiths in the supplement of the second edition of De Morbis Artificum , published in Padua in 1713. Perhaps the Italian physician, aware of the serious oversight of hearing disorders in blacksmiths, added this chapter, dedicated to a numerically very limited number of workers, implying that the disorders of coppersmiths could also extend to other craftsmen working with metal, including blacksmiths. Indeed, Ramazzini is keen to note in that chapter he considers ‘those who work with copper in their city workshops, not copper miners who I have already mentioned in the first chapter on metal miners’. Finally, it should be noted that in his masterpiece, Ramazzini describes another occupational disease relevant to otolaryngologists. Voice disorders among occupational voice users, that is, workers whose voice is essential to their job and who rely on their voice as a means of employability (e.g. teachers, singers, actors), is an emerging problem, even though insurance companies recognise voice disorders less than occupational hearing disorders. In the chapter XXXVIII of De Morbis Artificum , the Italian physician describes ‘diseases that generally affect music teachers, singers and the like’ . Although including this category of workers was innovative and pioneering for his time, Ramazzini did not recognise that these problems were related to laryngeal alteration. Instead, he stated that ‘head colds and hoarseness are common ills in singers and actors, once again owing to the muscle tension that causes more lymph than necessary to be secreted by the salivary glands’. However, a modern reader should remember that, at that time, the role of the vocal folds and the larynx in general in phonation was not yet known, although contemporaneous contributions of the French naturalist Denis Dodart (1634–1707) were beginning to provide insight. Despite these limitations, the words of Bernardino Ramazzini appear pioneering for his time and represent an important milestone in the history of otology and laryngology. |
null | e9b78df3-45b5-4915-bee6-fbdd3bec14e9 | 10180103 | Pharmacology[mh] | Fruits, vegetables, and herbs provide the body with many valuable specialized metabolites, often with pro-health properties. These include polyphenols, alkaloids, terpenoids, and essential oils, all of which display specific biological activities. Plants have been utilized in folk medicine in many countries for centuries because of their long-known therapeutic benefits. Not only have the beneficial medicinal properties of plants have been known for thousands of years, consumer interest in phytotherapy/herbal medicines and natural food supplements continues to grow, and in many countries, traditional medicine is the only mode of treatment for many diseases. Although herbal materials have many ethnomedicinal benefits, their toxicity or potential side effects remain relatively unexplored, and their medical potential frequently lacks a scientific basis. By determining the chemical composition of plant extracts, it is possible to estimate their safety and biological activities and hence their potential as natural drugs. The use of medicinal plants as natural sources of compounds with inter alia antioxidant, anti-inflammatory, and anti-diabetic properties has drawn the attention of many researchers. Oxidative stress and the inflammatory response are associated with many neurological diseases (Alzheimer’s disease, Parkinson’s disease, amyotrophic lateral sclerosis), cardiovascular diseases (atherosclerosis), hepatic conditions, gastrointestinal diseases, and cancers . Recent years have seen a greater interest in the genera Dipsacus L. (teasel in English) and Scabiosa L. (pincushions in English) , with the latter being poorly understood. The first mention of phytochemical studies on Dipsacus spp. dates to the 1920s . In 2011, Zhao and Shi reviewed the chemical composition and biological properties of the specialized metabolites of some Dipsacus species. Ten years later, Tao et al. published a review of the literature regarding one Dipsacus species, D. asper Wall. ex C.B.Clarke, its chemical constituents, selected pharmacological activities, and pharmacokinetics. In 2018, Pinto et al. provided a review of flavonoids and terpenoid derivatives identified in some species from Scabiosa and their biological effects. In recent years, a significant number of reports have been published on the chemical composition and biological properties of the two genera. Therefore, the present review aims to provide an overview of the current literature (until December 2022) regarding the phytochemistry, biological activities, and toxicology of selected species of Dipsacus and Scabiosa . Several online databases, including PubMed, Google Scholar, Scopus, and ScienceDirect, were searched in the current review. These two genera include many synonymous Latin species names, meaning the same species, which can lead to confusion. In the present review, the species names are cited according to the authors of publications. The natural occurrence of Dipsacus and Scabiosa species as well as synonymous species names are described based on the data from The World Flora Online, Plants of the World Online, The Global Biodiversity Information Facility, and Flora of China . The genera Dipsacus and Scabiosa currently belong to the Caprifoliaceae Juss. family (honeysuckle family) . They were previously classified taxonomically in the Dipsaceae . According to Plants of the World Online database , the Caprifoliaceae includes 33 accepted genera. This family has been divided into six subfamilies and one genus. Both genera Dipsacus and Scabiosa are classified into the subfamily Dipsacoideae and order Dipsacales . The genus Dipsacus is widely distributed in Europe, North Africa, and Asia in North Myanmar and comprises 21 accepted species . The native range of the genus Scabiosa includes Eurasia, Macaronesia to North Africa, Eritrea, and South Africa, with 66 accepted species . In particular, numerous representatives of European Scabiosa species appear in the Mediterranean region . presents accepted species of the genera Dipsacus and Scabiosa . Dipsacus includes various ornamental plants used in floristry for their decorative dried inflorescences. The dried inflorescences were previously used in the textile industry to clean and lift the nap on woolen fabrics . The name dipsacus itself is believed to be derived from the Greek word for dipsa or thirst . The genus name scabiosa is derived from the Latin word scabies or itch . According to Akar , Scabiosa plants were traditionally used to treat scabies, skin sores, and other skin infections. Many Dipsacus and Scabiosa species are known in various traditional medicines, including traditional Chinese medicine. So far, among over 20 species of the genus Dipsacus and 60 species of Scabiosa , the phytochemical profiles and biological properties of only a few species are known. As such, there is a need to better understand the chemical composition of these genera and their potential medicinal value. 2.1. Dipsacus spp. Numerous, recent reports have examined the potential therapeutic effects of Dipsacus genera. However, most phytochemical and biological studies have focused on the biological activities and health-promoting factors of D. asperoides C.Y.Cheng et T.M.Ai (= D. asper Wall. ex C.B.Clarke) . The oldest report on Dipsacus genus can be found in Shen Nong’s Herbal Classic ( Shen Nong Ben Cao Jing in Chinese) . D. asper is widespread in the southern and northern regions of China such as the Hunan, Yunnan, Gansu, and Shanxi Provinces . The growing demand for this species and its mass harvesting has significantly weakened its population in a natural state . D. asperoides is cultivated on a large scale, mainly in Hefeng City, Hubei Province. This species is also cultivated in other Chinese cities, including Xichang, Sichuan, Xifeng, Guizhou, Jianchuan, and Yunnan Provinces as well as Jiangxi and Guangxi Provinces in China . The roots of D. asperoides in China are commonly known as Xu Duan or Himalayan Teasel Roots . Traditionally, in China and Korea, Dipsaci radix is known as the raw material used in treating joint disease (rheumatic arthritis) and bone diseases (osteoporosis, bone fractures), lumbar and knee pain, arthralgia, traumatic hematoma, uterine bleeding, and gynecological diseases; it is also used to strengthen muscles and improve liver and kidney functions . The raw material is collected in autumn. Dipsaci radix is 5–15 cm in length and 0.5–2 cm in diameter, with slightly twisted or twisted longitudinal wrinkles and furrows. It is greyish-brown or yellowish-brown in color . Dipsaci radix has a spicy, bitter, slightly sweet, then astringent taste . Dipsaci radix can be subjected to diaphoretic-, salt-, and wine-processing methods . Some researchers suggest that the procedure of herb processing may result in differences in specialized metabolite content and biological effects. Some studies indicate that wine processing yields higher levels of key compounds (e.g., asperosaponin V and VI, dipsacoside A and B, dipsacussaponin B, loganic acid, and chlorogenic acid) than crude Dipsaci radix . Materials processed by rice wine may promote anti-osteoporosis, anti-inflammatory, anti-coagulant, and analgesic activities or have a beneficial effect on blood circulation . Dipsaci radix is commonly available in the Chinese herbal medicine market. This herb is mainly produced in China, in the Provinces of Sichuan, Yunnan, Hubei, Hunan, Xizang, Jiangxi, Guizhou, and Guangxi . Dipsaci radix has a pharmacopeial monograph in the Korean Herbal Pharmacopoeia (Korea Food and Drug Administration 2007) and the 10th Chinese Pharmacopoeia, 2015 edition. The raw material is standardized to contain akebia saponin D whose level should be above 2% . Chun et al. indicate that the traditional use of D. asperoides is associated with its analgesic and anti-inflammatory properties for treating inter alia rheumatoid arthritis and bone fractures. The Chinese Pharmacopoeia, 2015 edition, recommends 9–15 g as the daily dose of D. asper for humans . In the Korean herbal medicine markets, Phlomidis radix from Phlomis umbrosa Turczaninow is often sold instead of Dipsaci radix or both raw materials are mixed . This mistake results from the morphological similarity of these two dried raw materials and their names . Both species have been used in the Korean medicine Sok-dan for the treatment of bone- and arthritis-related diseases and are listed in the Korean Herbal Pharmacopoeia . HPLC/UV profiles of four samples of Phlomidis radix and 17 samples of Dipsaci radix demonstrated that loganin, sweroside, dipsanoside A, 3- O -[ β - d -glu-(1→4)][α- l -rha-(1→3)]- β - d -glu(1→3)-α- l -rha-(1→2)-α- l -ara-hed 28- O - β - d -glu-(1→6)-β- d -glu ester, and akebia saponin D were not detected in Phlomidis radix . The levels of these compounds in Dipsaci radix varied depending on the origin of the material and the extraction methods . Akebia saponin D (a quality indicator of Dipsaci radix ) predominated, with a content range of 0.73–10.96% ( w / w ) . Both herbs showed anti-osteoarthritis ability in a Sprague-Dawley rat model induced by monosodium iodoacetate intra-articular knee injections . However, Phlomidis radix was found to be more effective, suggesting that it may be used as an alternative to Dipsaci radix . It was found that the 70% ethanol extract (200 mg/kg/day) administered as oral gavage for 21 days restored the weight-bearing ability of the hind paw, suppressed histopathological changes of the osteoarthritic knee, inhibited the serum levels of inflammation mediators tumor necrosis factor α (TNF-α) and interleukin-1β (IL-1β), and inhibited the over-expression of the gene encoding matrix-degrading metalloproteinases MMP-9 and MMP-13 in the knee joint tissue . In addition, the ethanol extract of Dipsaci radix displayed a protective effect against the destruction of articular cartilage, elevated myeloperoxidase (MPO) and down-regulated dystonin gene expression, modulated WNT/β-catenin signaling pathway, suppressed gene expression of Adamts4, and increased the expression of cartilage collagen genes (e.g., Col2A1, Col9A1 , and Col11A1) and SOX5, SOX9, and Frzb genes . Another species of Dipsacus is D. fullonum L., commonly known as teasel or wild teasel . Traditionally, D. fullonum has been used to treat Lyme disease and eye infections in cattle . D. fullonum is distributed naturally in Europe to the Caucasus and North-Western Africa . A rich source of valuable ingredients used in the traditional medicine of the Kashmir Himalayas is D. inermis Wall., also known as Wopal haakh/Wopal Hak in the Kashmiri language. It is used in treating cold, fever, cough, sore throat, general fatigue, and body pain and has demonstrated stomachic and carminative properties . The roots of D. japonicus Miq., commonly known as Tuc doan in Vietnam, show tonic, anodyne, and demulcent activities. The recommended daily dose is 10–20 g as a decoction, alcoholic maceration, powder, or pill . Decoction of D. japonicus roots has been used in the traditional medicine of China and Vietnam for rheumatism, sprains, trauma, fractures, relieving joint pain and ostealgia, and hepatic and renal hypofunction . D. japonicus is widespread in central and northern China, Korea, Japan, and Vietnam . The leaves of D. sativus (L.) Honck. are used as an infusion for the treatment of cardiovascular diseases. This species was originally cultured in Europe and was introduced from Japan to China in 1929 . 2.2. Scabiosa spp. With respect to Scabiosa genera, many studies indicate that species of this genus show antioxidant , anti-inflammatory , anti-diabetic , anti-hepatic fibrosis , anti-cancer , and antibacterial properties. Many species of Scabiosa grow naturally in the Mediterranean region . In Tunisia, species from Scabiosa were commonly applied for skin treatment . In the Iberian Peninsula, an infusion of S. atropurpurea L. inflorescence is used externally on the skin as an anti-acne treatment and orally for measles, rubeola, and scarlet fever . In northern Peru, the aerial parts have been used for menstrual regulation and in Iberia, as a veterinary diuretic . S. atropurpurea is distributed throughout the Mediterranean, Europe, Asia, and southern Africa . Genetic investigations show that S. atropurpurea has 97% similarity with S. tschiliensis and thus is also called Japanese scabiosa . This species is known as Mor uyuzotu or Şeytanotu in Turkey, Ambarina in Northern Peru, and Escabiosa in the Iberian Peninsula . The leaves and flowers of S. stellata L. were used in the traditional medicine of Morocco to treat heel cracks . S. stellata is an endemic plant in North Africa and is commonly known as starflower pincushions . Another endemic species in North Africa (Algeria, Egypt, Libya, Morocco, and Tunisia) is S. arenaria Forssk. . In traditional Mongolian medicine, the inflorescences of S. comosa Fisch. ex Roem. et Schult. and S. tschilliensis Grunning (known as Lanpenhua in Chinese) are used for liver diseases . S. tschiliensis is widespread in China (in Hebei Province) and the Inner Mongolia autonomy district and is locally called Meng Gu Shan Luo Bo . Qingganjiuwei powder (composed of nine herbal materials including S. comosa ) is commonly used as an anti-fibrosis drug in patients with chronic hepatic disease in Inner Mongolia . This drug is accepted by the Inner Mongolia Region Drug Administration . The flowers of S. comosa ( Scabiosae flos ) are also an ingredient of Gurigumu-7 used in traditional Mongolian and Tibetan medicine to treat liver diseases. This preparation is in the form of a bitter and astringent powder . Numerous, recent reports have examined the potential therapeutic effects of Dipsacus genera. However, most phytochemical and biological studies have focused on the biological activities and health-promoting factors of D. asperoides C.Y.Cheng et T.M.Ai (= D. asper Wall. ex C.B.Clarke) . The oldest report on Dipsacus genus can be found in Shen Nong’s Herbal Classic ( Shen Nong Ben Cao Jing in Chinese) . D. asper is widespread in the southern and northern regions of China such as the Hunan, Yunnan, Gansu, and Shanxi Provinces . The growing demand for this species and its mass harvesting has significantly weakened its population in a natural state . D. asperoides is cultivated on a large scale, mainly in Hefeng City, Hubei Province. This species is also cultivated in other Chinese cities, including Xichang, Sichuan, Xifeng, Guizhou, Jianchuan, and Yunnan Provinces as well as Jiangxi and Guangxi Provinces in China . The roots of D. asperoides in China are commonly known as Xu Duan or Himalayan Teasel Roots . Traditionally, in China and Korea, Dipsaci radix is known as the raw material used in treating joint disease (rheumatic arthritis) and bone diseases (osteoporosis, bone fractures), lumbar and knee pain, arthralgia, traumatic hematoma, uterine bleeding, and gynecological diseases; it is also used to strengthen muscles and improve liver and kidney functions . The raw material is collected in autumn. Dipsaci radix is 5–15 cm in length and 0.5–2 cm in diameter, with slightly twisted or twisted longitudinal wrinkles and furrows. It is greyish-brown or yellowish-brown in color . Dipsaci radix has a spicy, bitter, slightly sweet, then astringent taste . Dipsaci radix can be subjected to diaphoretic-, salt-, and wine-processing methods . Some researchers suggest that the procedure of herb processing may result in differences in specialized metabolite content and biological effects. Some studies indicate that wine processing yields higher levels of key compounds (e.g., asperosaponin V and VI, dipsacoside A and B, dipsacussaponin B, loganic acid, and chlorogenic acid) than crude Dipsaci radix . Materials processed by rice wine may promote anti-osteoporosis, anti-inflammatory, anti-coagulant, and analgesic activities or have a beneficial effect on blood circulation . Dipsaci radix is commonly available in the Chinese herbal medicine market. This herb is mainly produced in China, in the Provinces of Sichuan, Yunnan, Hubei, Hunan, Xizang, Jiangxi, Guizhou, and Guangxi . Dipsaci radix has a pharmacopeial monograph in the Korean Herbal Pharmacopoeia (Korea Food and Drug Administration 2007) and the 10th Chinese Pharmacopoeia, 2015 edition. The raw material is standardized to contain akebia saponin D whose level should be above 2% . Chun et al. indicate that the traditional use of D. asperoides is associated with its analgesic and anti-inflammatory properties for treating inter alia rheumatoid arthritis and bone fractures. The Chinese Pharmacopoeia, 2015 edition, recommends 9–15 g as the daily dose of D. asper for humans . In the Korean herbal medicine markets, Phlomidis radix from Phlomis umbrosa Turczaninow is often sold instead of Dipsaci radix or both raw materials are mixed . This mistake results from the morphological similarity of these two dried raw materials and their names . Both species have been used in the Korean medicine Sok-dan for the treatment of bone- and arthritis-related diseases and are listed in the Korean Herbal Pharmacopoeia . HPLC/UV profiles of four samples of Phlomidis radix and 17 samples of Dipsaci radix demonstrated that loganin, sweroside, dipsanoside A, 3- O -[ β - d -glu-(1→4)][α- l -rha-(1→3)]- β - d -glu(1→3)-α- l -rha-(1→2)-α- l -ara-hed 28- O - β - d -glu-(1→6)-β- d -glu ester, and akebia saponin D were not detected in Phlomidis radix . The levels of these compounds in Dipsaci radix varied depending on the origin of the material and the extraction methods . Akebia saponin D (a quality indicator of Dipsaci radix ) predominated, with a content range of 0.73–10.96% ( w / w ) . Both herbs showed anti-osteoarthritis ability in a Sprague-Dawley rat model induced by monosodium iodoacetate intra-articular knee injections . However, Phlomidis radix was found to be more effective, suggesting that it may be used as an alternative to Dipsaci radix . It was found that the 70% ethanol extract (200 mg/kg/day) administered as oral gavage for 21 days restored the weight-bearing ability of the hind paw, suppressed histopathological changes of the osteoarthritic knee, inhibited the serum levels of inflammation mediators tumor necrosis factor α (TNF-α) and interleukin-1β (IL-1β), and inhibited the over-expression of the gene encoding matrix-degrading metalloproteinases MMP-9 and MMP-13 in the knee joint tissue . In addition, the ethanol extract of Dipsaci radix displayed a protective effect against the destruction of articular cartilage, elevated myeloperoxidase (MPO) and down-regulated dystonin gene expression, modulated WNT/β-catenin signaling pathway, suppressed gene expression of Adamts4, and increased the expression of cartilage collagen genes (e.g., Col2A1, Col9A1 , and Col11A1) and SOX5, SOX9, and Frzb genes . Another species of Dipsacus is D. fullonum L., commonly known as teasel or wild teasel . Traditionally, D. fullonum has been used to treat Lyme disease and eye infections in cattle . D. fullonum is distributed naturally in Europe to the Caucasus and North-Western Africa . A rich source of valuable ingredients used in the traditional medicine of the Kashmir Himalayas is D. inermis Wall., also known as Wopal haakh/Wopal Hak in the Kashmiri language. It is used in treating cold, fever, cough, sore throat, general fatigue, and body pain and has demonstrated stomachic and carminative properties . The roots of D. japonicus Miq., commonly known as Tuc doan in Vietnam, show tonic, anodyne, and demulcent activities. The recommended daily dose is 10–20 g as a decoction, alcoholic maceration, powder, or pill . Decoction of D. japonicus roots has been used in the traditional medicine of China and Vietnam for rheumatism, sprains, trauma, fractures, relieving joint pain and ostealgia, and hepatic and renal hypofunction . D. japonicus is widespread in central and northern China, Korea, Japan, and Vietnam . The leaves of D. sativus (L.) Honck. are used as an infusion for the treatment of cardiovascular diseases. This species was originally cultured in Europe and was introduced from Japan to China in 1929 . With respect to Scabiosa genera, many studies indicate that species of this genus show antioxidant , anti-inflammatory , anti-diabetic , anti-hepatic fibrosis , anti-cancer , and antibacterial properties. Many species of Scabiosa grow naturally in the Mediterranean region . In Tunisia, species from Scabiosa were commonly applied for skin treatment . In the Iberian Peninsula, an infusion of S. atropurpurea L. inflorescence is used externally on the skin as an anti-acne treatment and orally for measles, rubeola, and scarlet fever . In northern Peru, the aerial parts have been used for menstrual regulation and in Iberia, as a veterinary diuretic . S. atropurpurea is distributed throughout the Mediterranean, Europe, Asia, and southern Africa . Genetic investigations show that S. atropurpurea has 97% similarity with S. tschiliensis and thus is also called Japanese scabiosa . This species is known as Mor uyuzotu or Şeytanotu in Turkey, Ambarina in Northern Peru, and Escabiosa in the Iberian Peninsula . The leaves and flowers of S. stellata L. were used in the traditional medicine of Morocco to treat heel cracks . S. stellata is an endemic plant in North Africa and is commonly known as starflower pincushions . Another endemic species in North Africa (Algeria, Egypt, Libya, Morocco, and Tunisia) is S. arenaria Forssk. . In traditional Mongolian medicine, the inflorescences of S. comosa Fisch. ex Roem. et Schult. and S. tschilliensis Grunning (known as Lanpenhua in Chinese) are used for liver diseases . S. tschiliensis is widespread in China (in Hebei Province) and the Inner Mongolia autonomy district and is locally called Meng Gu Shan Luo Bo . Qingganjiuwei powder (composed of nine herbal materials including S. comosa ) is commonly used as an anti-fibrosis drug in patients with chronic hepatic disease in Inner Mongolia . This drug is accepted by the Inner Mongolia Region Drug Administration . The flowers of S. comosa ( Scabiosae flos ) are also an ingredient of Gurigumu-7 used in traditional Mongolian and Tibetan medicine to treat liver diseases. This preparation is in the form of a bitter and astringent powder . Dipsacus and Scabiosa Species Dipsacus and Scabiosa have closely related phytochemical profiles and include over 200 specialized metabolites. Some representatives of Dipsacus and Scabiosa biosynthesize chemical compounds of various classes, mainly triterpenoid derivatives , which possess a variety of bioactivities. In addition, iridoids , phenolic acids , and flavonoids have also been reported. Further, few alkaloids have been found in D. asper such as cantleyine, venoterpine, gentianine, dipsaperine ((3 S ,5 S )-5-carboxystrictosidic acid 22-loganin ester), 3 β ,5α-tetrahydrodesoxycordifoline lactam, and (3 R ,5 S )-5-carboxyvincosidic acid 22-loganin ester , and lignans have been detected . Scabiosa species are also rich sources of flavonoids, mainly in the aerial parts and flowers . Essential oils were also isolated from S. arenaria , S. atropurpurea , D. fullonum , and D. japonicus . Fatty acids have also been detected in Scabiosa spp. and D. asper . The specialized metabolites identified in Dipsacus and Scabiosa are listed in , , , , , and . The structures of selected compounds that have displayed some biological activities in a number of studies are shown in . 3.1. Terpenoid Derivatives The most diverse group of specialized metabolites in Dipsacus and Scabiosa are the triterpenoid derivatives, which can be divided into oleanane-type, hederagenin-type, or ursane-type . Some compounds are derived from pomoic acid (scabiosaponin H-I) . In Dipsacus , the main group of triterpenoids identified is hederagenin and its related saponins, while the rarest is ursane-type terpenoids . The oleanane-type triterpenoids were also common in Scabiosa genus (for example, oleanolic acid, scabiosaponins A-G, scabiostellatosides A-F, and hookeroside A and B) . Some of these specialized metabolites such as oleanolic acid and ursolic acid were detected in the species of both genera . Some of them were genus specific, such as hookeroside A and B and scabiosaponin A-K identified in S. tschiliensis whole plants and palustroside III and scabiostellatosides A-H detected in S. stellata whole plants . The new, tentatively detected triterpenoid derivatives in S. atropurpurea subsp. maritima leaves were oleanolic acid-pentosyl-rhamnosyl-pentosyl-glucosyl-di-glucoside, oleanolic acid-pentosyl-rhamnosyl-glucosyl-glucosyl-di-glucoside, and maslinic acid-pentosyl-rhamnosyl-glucosyl-glucosyl . Yu et al. reported some new arborinane-type triterpenoid (25-acetoxy-28-dehydroxyrubiarbonone E), ursane-type triterpenoids (2 α ,3 β ,24-trihydroxy-23-norurs-12-en-28-oic acid and 2 α ,3 β -dihydroxy-23-norurs-4(24),11,13(18)-trien-28-oic acid), and oleanane-type triterpenoids (2′,3′- O -diacetyl-3- O - α - l -arabinopyranosyl-23-hydroxyolea-12-en-28-oic acid, 2′,4′- O -diacetyl-3- O - α - l -arabinopyranosyl-23-hydroxyolea-12-en-28-oic acid, and 3- O - β - d -glucopyranosyl-(1→3)-α- l -rhamnopyranosyl-(1→2)-α- l -arabinopyranosyl-23-hydroxyolean-18-en-28-oic acid 28- O - β - d -glucopyranosyl-(1→6)-β- d -glucopyranosyl ester) in D. asper roots. A new compound elucidated in D. japonicus roots was saponin XII . In addition, this species has presented the only example to date of japondipsaponin E1 . The indicator of the quality of Dipsaci radix is akebia saponin D (asperosaponin VI) . It is known that environmental conditions, geographic location, the growth stage of the plants, or the year of harvest affect the content of the specialized metabolites in plants. Jin et al. demonstrated that the content of akebia saponin D varied in D. asperoides roots collected from different geographical regions of China (i.e., Guizhou, Hubei, Sichuan, and Yunnan Provinces). The highest akebia saponin D content (about 6% of dry weight) was noted in roots collected from Hubei. A similar observation was noted by Du et al. . The level of akebia saponin D ranged from 1.61% to 15.19% in samples of different origin. Iridoids Iridoids were found in both genera . The first iridoids isolated from D. asper were loganin and sweroside . The presence of iridoid glycosides including loganin and loganic acid was widespread in the roots or leaves of Dipsacus spp. and flowers of S. atropurpurea subsp. maritima . Dipsanosides A-N, dipsasperoside A and B, lisianthioside, triplostoside A, and cocculoside were characteristic only for roots of D. asper . Sweroside and sylvestrosides III-IV were found in two Dipsacus species: D. asper roots and D. fullonum leaves and roots . Sweroside and sylvestrosides I and II were also detected in some Scabiosa species, such as S. stellata whole plants, S. tschilliensis , and S. atropurpurea subsp. maritima flowers . 7- O -caffeoyl-sylvestroside I and 7- O -( p -coumaroyl)-sylvestroside I were isolated as new compounds from the whole plants of S. stellata . Interestingly, three iridoid-like compounds, viz., eustomoside, eustomoruside, and septemfidoside, were identified only in S. stellata . It should be also noted that these iridoids have not been previously described in Caprifoliaceae . Atropurpurin A-B and secologanin-methyl-hemiacetal were reported for the first time in Scabiosa spp. in S. atropurpurea subsp. maritima . Lehbili et al. suggested that Dipsacus and Scabiosa are closely related to one another; septemfidoside, sylvestroside I, and its derivatives (e.g., 7- O -caffeoyl-sylvestroside I and 7- O -[ p -coumaroyl]-sylvestroside I) isolated from S. stellata whole plants are closely related to the bis -iridoids, cantleyoside identified in S. atropurpurea subsp. maritima , Dipsaci radix , and D. fullonum as well as dipsanosides C-G detected in D. asper . The bis -iridoids identified in Dipsacus spp. possess a secoiridoid/iridoid subtype skeleton consisting of secologanic acid condensed to the 7-OH of loganin or loganin-like iridoids . The levels of the main iridoids in Dipsaci radix , e.g., loganic acid, loganin, sweroside, and dipsanosides A and B, varied according to location and collection time. The highest level of loganic acid was 31.223 mg/g; loganin, 4.411 mg/g; sweroside, 8.364 mg/g; dipsanoside A, 4.513 mg/g; and dipsanoside B, 7.426 mg/g . Du et al. also demonstrated that the content of loganic acid (0.71–1.10%) and loganin (0.29–0.61%) varied between samples from different origins in China. In addition, the level of some iridoids in the leaves and roots including loganic acid, loganin, sweroside, sylvestroside III, and cantleyoside detected in D. fullonum varied according to the time of year of harvesting . The leaves of the plants collected in the first year of vegetation contained a higher total iridoid content than the roots and leaves harvested in the second year. The most abundant iridoids were sylvestroside III in the leaves and cantleyoside in the roots. Moreover, sweroside (0.67 mg/g dry weight) and sylvestroside III (34.8 mg/g d.w.) were present in higher amounts in leaves, while the levels of loganic acid (5.27 mg/g d.w.), loganin (3.02 mg/g d.w.), and cantleyoside (21.41 mg/g d.w.) were higher in the roots . 3.2. Phenolic Acids Many phenolic acids have been noted in Dipsacus and Scabiosa species . Mainly derivatives of hydroxycinnamic acid were identified, with chlorogenic acid as the most abundant component. This phenolic acid was detected in Dipsaci radix , D. fullonum leaves and roots , S. comosa and S. tschiliensis inflorescences , S. atropurpurea stems , and S. atropurpurea subsp. maritima leaves . In addition, other mono-caffeoylquinic acid derivatives (1- O -caffeoylquinic acid, 3- O -caffeoylquinic acid methyl ester, and 4- O -caffeoylquinic acid) and dicaffeoylquinic acid derivatives (3,4-di- O -caffeoylquinic acid, 3,5-di- O -caffeoylquinic acid, 4,5-di- O -caffeoylquinic acid, and their methyl derivative) have been identified . The hydroxycinnamic acid derivatives included caffeic acid, caffeic acid methyl ester, p -coumaric acid, p -coumaric acid 3-glucoside, 2,6-dihydroxycinnamic acid, 5- O -ferruloylqunic acid, p -hydroxycinnamic acid, and sinapic acid . The benzoic acid derivatives were 3,4-dihydroxybenzoic acid, protocatechuic acid, protocatechuic acid 3-glucoside, and vanillic acid . The phenolic acid content in Dipsaci radix varied according to the geographical region of China . The total level of phenolic acids ranged from 0.98 mg/g to 49.55 mg/g, with the highest level observed in plant material collected from Guizhou Province . The predominant phenolic acid was chlorogenic acid (0.186–19.174 mg/g; 2.02–8.28%) , which was also an abundant phenolic in D. fullonum . However, qualitative differences were reported between the leaves and the roots, with the highest level being found in the leaves obtained from plants in the second year of vegetation (28.44 mg/g d.w.). Wang et al. also found differences in the content of chlorogenic acid in plants of S. tschiliensis between the pre-flowering, flowering, and fruiting stages. The level was also dependent on the extraction solvent; the richest source of chlorogenic acid, 45.35 mg/g d.w., was found in the ethyl acetate fraction from plants at the pre-flowering stage. 3.3. Flavonoids The Scabiosa genus is a rich source of flavonoids, which are identified mainly in the flowers but also in the leaves, stems, and roots . Many studies note that apigenin, luteolin, and their derivatives are particularly common . The new flavonoids identified for the first time in Scabiosa spp. were diosmetin-7- O -glucoside, luteolin-7,3′-diglucoside, luteolin 3′-glucoside, and quercetin 3,4′-diglucoside in S. atropurpurea subsp. maritima leaves ; quercimeritrin in S. atropurpurea stems; quercitrin, rutin, kaempferol-3- O - β - d -6- O -( p -hydroxycinnamoyl)-glucopyranoside, and kaempferol-3- O - β - d -(3,6-di- p -(hydroxycinnamoyl)-glucopyranoside in S. tschilliensis flowers ; and tamarixetin derivative 3- β - l -rhamnosyl-(1→2)[ β - l -rhamnosyl-(1→6)] β - d -glucoside] and tiliroside in S. stellata whole plants . Isoorientin, isovitexin, orientin, saponarin, and saponaretin were detected for the first time in Dipsacus spp., in D. fullonum leaves . 3.4. Lignans Lignans were identified only in Dipsaci radix and include prinsepiol, fraxiresinol, and their derivatives such as dipsalignans A-D . Dipsaci radix also demonstrated derivatives of pinoresinol and syringaresinol . Syringaresinol hexoside was detected as a new compound in S. atropurpurea stems . 3.5. Polysaccharides Dipsacus spp. also include polysaccharides . The water-soluble polysaccharide from D. asperoides roots (ADAPW) has a molecular weight of 16 kDa and contains glucose, rhamnose, arabinose, and mannose in a molar ratio of 8.54:1.83:1.04:0.42 . The polysaccharide WDRAP-1, with a molecular weight of 61 kDa, was composed of glucose, mannose, galactose, arabinose, and rhamnose in a molar ratio of 3.1:0.9:5.2:1.1:0.3. The predominant monosaccharides were glucose (29.2%) and galactose (49.1%) . Xu et al. found two polysaccharides in D. asperoides roots, DAI-1 and DAI-2, which consisted only of glucose and had respective molecular masses of 17 and 4 kDa. Sun et al. isolated from D. asper roots a homogenous polysaccharide (DAP) with a molecular weight of 26.1 kDa that was composed of galactose and mannose in a molar ratio of 1:1. 3.6. Essential Oils Essential oils are present in some species of Dipsacus and Scabiosa . The essential oils isolated by hydrodistillation from dried and fresh roots and leaves of D. fullonum were rich in many components; however, quantitative and qualitative differences were noted. The dominant compound in the essential oils from leaves, regardless of fresh or dried materials, was phytol (branched-chain unsaturated diterpene alcohol; precursor for vitamins E and K1), whose content ranged from 61.08% to 72.31%. It was not detected in root essential oil. In addition, the main components in fresh leaf essential oil, i.e., with a level above 5%, were 9,12,15-octadecatrienoic acid methyl ester and cyclohexane, cyclopropylidene. On the other hand, in the essential oils from the dry and fresh roots, the main component was n -hexadecanoic acid, especially in the dried material (16.00%), which was also rich in 11,14,17-eicosatrienoic acid, methyl ester (15.86%). n -hexadecanoic acid was also identified in the essential oils from dried and fresh leaves but at much lower levels (2.01–2.34%) . In the essential oil from the flowering aerial parts of D. japonicus , linalool (11.78%), trans -geraniol (8.58%), and 1,8-cineole (7.91%) predominated. The other main ingredients present at above 5% were β -caryophyllene (5.58%), α -terpineol (5.32%), β -selinene (5.15%), and spathulenol (5.04%) . Qualitative and quantitative differences were also observed in the essential oils isolated from different parts of S. arenaria . The main compounds detected in the oil from the aerial parts and flower oils were chrysanthenone (23.43–38.52%), camphor (11.75–12.98%), and α-thujone (9.5–10.7%), while α-thujone (34.39%), camphor (17.48%), and β-thujone (15.29%) predominated in the fruit oil. In addition, longifelone (2.41–3.96%) and filifolone (1.99–3.72%) were only identified in oils from the vegetative parts of the plants and the flowers. In S. atropurpurea stems, in volatile fractions VF1 (extracted by hexane) and VF2 (extracted by chloroform), the most abundant ingredients were 1,8 cineole (8.1–33.8%), tetradecene (5.7–24.1%), and (E)-β-ionone (5.9–20.7%). It is worth mentioning that dihydroactinidiolide, which was present in significant amounts (26.1%), was identified only in the chloroform fraction . 3.7. Fatty Acids The presence of fatty acids has only been investigated in S. stellata aerial parts and D. asper roots . Thirteen fatty acids including saturated and unsaturated acids have been identified. The fatty acids present in the hexane extract of S. stellata aerial parts accounted for 87% of the total content with linolenic acid, palmitic acids, and linoleic acids predominating . The most diverse group of specialized metabolites in Dipsacus and Scabiosa are the triterpenoid derivatives, which can be divided into oleanane-type, hederagenin-type, or ursane-type . Some compounds are derived from pomoic acid (scabiosaponin H-I) . In Dipsacus , the main group of triterpenoids identified is hederagenin and its related saponins, while the rarest is ursane-type terpenoids . The oleanane-type triterpenoids were also common in Scabiosa genus (for example, oleanolic acid, scabiosaponins A-G, scabiostellatosides A-F, and hookeroside A and B) . Some of these specialized metabolites such as oleanolic acid and ursolic acid were detected in the species of both genera . Some of them were genus specific, such as hookeroside A and B and scabiosaponin A-K identified in S. tschiliensis whole plants and palustroside III and scabiostellatosides A-H detected in S. stellata whole plants . The new, tentatively detected triterpenoid derivatives in S. atropurpurea subsp. maritima leaves were oleanolic acid-pentosyl-rhamnosyl-pentosyl-glucosyl-di-glucoside, oleanolic acid-pentosyl-rhamnosyl-glucosyl-glucosyl-di-glucoside, and maslinic acid-pentosyl-rhamnosyl-glucosyl-glucosyl . Yu et al. reported some new arborinane-type triterpenoid (25-acetoxy-28-dehydroxyrubiarbonone E), ursane-type triterpenoids (2 α ,3 β ,24-trihydroxy-23-norurs-12-en-28-oic acid and 2 α ,3 β -dihydroxy-23-norurs-4(24),11,13(18)-trien-28-oic acid), and oleanane-type triterpenoids (2′,3′- O -diacetyl-3- O - α - l -arabinopyranosyl-23-hydroxyolea-12-en-28-oic acid, 2′,4′- O -diacetyl-3- O - α - l -arabinopyranosyl-23-hydroxyolea-12-en-28-oic acid, and 3- O - β - d -glucopyranosyl-(1→3)-α- l -rhamnopyranosyl-(1→2)-α- l -arabinopyranosyl-23-hydroxyolean-18-en-28-oic acid 28- O - β - d -glucopyranosyl-(1→6)-β- d -glucopyranosyl ester) in D. asper roots. A new compound elucidated in D. japonicus roots was saponin XII . In addition, this species has presented the only example to date of japondipsaponin E1 . The indicator of the quality of Dipsaci radix is akebia saponin D (asperosaponin VI) . It is known that environmental conditions, geographic location, the growth stage of the plants, or the year of harvest affect the content of the specialized metabolites in plants. Jin et al. demonstrated that the content of akebia saponin D varied in D. asperoides roots collected from different geographical regions of China (i.e., Guizhou, Hubei, Sichuan, and Yunnan Provinces). The highest akebia saponin D content (about 6% of dry weight) was noted in roots collected from Hubei. A similar observation was noted by Du et al. . The level of akebia saponin D ranged from 1.61% to 15.19% in samples of different origin. Iridoids Iridoids were found in both genera . The first iridoids isolated from D. asper were loganin and sweroside . The presence of iridoid glycosides including loganin and loganic acid was widespread in the roots or leaves of Dipsacus spp. and flowers of S. atropurpurea subsp. maritima . Dipsanosides A-N, dipsasperoside A and B, lisianthioside, triplostoside A, and cocculoside were characteristic only for roots of D. asper . Sweroside and sylvestrosides III-IV were found in two Dipsacus species: D. asper roots and D. fullonum leaves and roots . Sweroside and sylvestrosides I and II were also detected in some Scabiosa species, such as S. stellata whole plants, S. tschilliensis , and S. atropurpurea subsp. maritima flowers . 7- O -caffeoyl-sylvestroside I and 7- O -( p -coumaroyl)-sylvestroside I were isolated as new compounds from the whole plants of S. stellata . Interestingly, three iridoid-like compounds, viz., eustomoside, eustomoruside, and septemfidoside, were identified only in S. stellata . It should be also noted that these iridoids have not been previously described in Caprifoliaceae . Atropurpurin A-B and secologanin-methyl-hemiacetal were reported for the first time in Scabiosa spp. in S. atropurpurea subsp. maritima . Lehbili et al. suggested that Dipsacus and Scabiosa are closely related to one another; septemfidoside, sylvestroside I, and its derivatives (e.g., 7- O -caffeoyl-sylvestroside I and 7- O -[ p -coumaroyl]-sylvestroside I) isolated from S. stellata whole plants are closely related to the bis -iridoids, cantleyoside identified in S. atropurpurea subsp. maritima , Dipsaci radix , and D. fullonum as well as dipsanosides C-G detected in D. asper . The bis -iridoids identified in Dipsacus spp. possess a secoiridoid/iridoid subtype skeleton consisting of secologanic acid condensed to the 7-OH of loganin or loganin-like iridoids . The levels of the main iridoids in Dipsaci radix , e.g., loganic acid, loganin, sweroside, and dipsanosides A and B, varied according to location and collection time. The highest level of loganic acid was 31.223 mg/g; loganin, 4.411 mg/g; sweroside, 8.364 mg/g; dipsanoside A, 4.513 mg/g; and dipsanoside B, 7.426 mg/g . Du et al. also demonstrated that the content of loganic acid (0.71–1.10%) and loganin (0.29–0.61%) varied between samples from different origins in China. In addition, the level of some iridoids in the leaves and roots including loganic acid, loganin, sweroside, sylvestroside III, and cantleyoside detected in D. fullonum varied according to the time of year of harvesting . The leaves of the plants collected in the first year of vegetation contained a higher total iridoid content than the roots and leaves harvested in the second year. The most abundant iridoids were sylvestroside III in the leaves and cantleyoside in the roots. Moreover, sweroside (0.67 mg/g dry weight) and sylvestroside III (34.8 mg/g d.w.) were present in higher amounts in leaves, while the levels of loganic acid (5.27 mg/g d.w.), loganin (3.02 mg/g d.w.), and cantleyoside (21.41 mg/g d.w.) were higher in the roots . Iridoids were found in both genera . The first iridoids isolated from D. asper were loganin and sweroside . The presence of iridoid glycosides including loganin and loganic acid was widespread in the roots or leaves of Dipsacus spp. and flowers of S. atropurpurea subsp. maritima . Dipsanosides A-N, dipsasperoside A and B, lisianthioside, triplostoside A, and cocculoside were characteristic only for roots of D. asper . Sweroside and sylvestrosides III-IV were found in two Dipsacus species: D. asper roots and D. fullonum leaves and roots . Sweroside and sylvestrosides I and II were also detected in some Scabiosa species, such as S. stellata whole plants, S. tschilliensis , and S. atropurpurea subsp. maritima flowers . 7- O -caffeoyl-sylvestroside I and 7- O -( p -coumaroyl)-sylvestroside I were isolated as new compounds from the whole plants of S. stellata . Interestingly, three iridoid-like compounds, viz., eustomoside, eustomoruside, and septemfidoside, were identified only in S. stellata . It should be also noted that these iridoids have not been previously described in Caprifoliaceae . Atropurpurin A-B and secologanin-methyl-hemiacetal were reported for the first time in Scabiosa spp. in S. atropurpurea subsp. maritima . Lehbili et al. suggested that Dipsacus and Scabiosa are closely related to one another; septemfidoside, sylvestroside I, and its derivatives (e.g., 7- O -caffeoyl-sylvestroside I and 7- O -[ p -coumaroyl]-sylvestroside I) isolated from S. stellata whole plants are closely related to the bis -iridoids, cantleyoside identified in S. atropurpurea subsp. maritima , Dipsaci radix , and D. fullonum as well as dipsanosides C-G detected in D. asper . The bis -iridoids identified in Dipsacus spp. possess a secoiridoid/iridoid subtype skeleton consisting of secologanic acid condensed to the 7-OH of loganin or loganin-like iridoids . The levels of the main iridoids in Dipsaci radix , e.g., loganic acid, loganin, sweroside, and dipsanosides A and B, varied according to location and collection time. The highest level of loganic acid was 31.223 mg/g; loganin, 4.411 mg/g; sweroside, 8.364 mg/g; dipsanoside A, 4.513 mg/g; and dipsanoside B, 7.426 mg/g . Du et al. also demonstrated that the content of loganic acid (0.71–1.10%) and loganin (0.29–0.61%) varied between samples from different origins in China. In addition, the level of some iridoids in the leaves and roots including loganic acid, loganin, sweroside, sylvestroside III, and cantleyoside detected in D. fullonum varied according to the time of year of harvesting . The leaves of the plants collected in the first year of vegetation contained a higher total iridoid content than the roots and leaves harvested in the second year. The most abundant iridoids were sylvestroside III in the leaves and cantleyoside in the roots. Moreover, sweroside (0.67 mg/g dry weight) and sylvestroside III (34.8 mg/g d.w.) were present in higher amounts in leaves, while the levels of loganic acid (5.27 mg/g d.w.), loganin (3.02 mg/g d.w.), and cantleyoside (21.41 mg/g d.w.) were higher in the roots . Many phenolic acids have been noted in Dipsacus and Scabiosa species . Mainly derivatives of hydroxycinnamic acid were identified, with chlorogenic acid as the most abundant component. This phenolic acid was detected in Dipsaci radix , D. fullonum leaves and roots , S. comosa and S. tschiliensis inflorescences , S. atropurpurea stems , and S. atropurpurea subsp. maritima leaves . In addition, other mono-caffeoylquinic acid derivatives (1- O -caffeoylquinic acid, 3- O -caffeoylquinic acid methyl ester, and 4- O -caffeoylquinic acid) and dicaffeoylquinic acid derivatives (3,4-di- O -caffeoylquinic acid, 3,5-di- O -caffeoylquinic acid, 4,5-di- O -caffeoylquinic acid, and their methyl derivative) have been identified . The hydroxycinnamic acid derivatives included caffeic acid, caffeic acid methyl ester, p -coumaric acid, p -coumaric acid 3-glucoside, 2,6-dihydroxycinnamic acid, 5- O -ferruloylqunic acid, p -hydroxycinnamic acid, and sinapic acid . The benzoic acid derivatives were 3,4-dihydroxybenzoic acid, protocatechuic acid, protocatechuic acid 3-glucoside, and vanillic acid . The phenolic acid content in Dipsaci radix varied according to the geographical region of China . The total level of phenolic acids ranged from 0.98 mg/g to 49.55 mg/g, with the highest level observed in plant material collected from Guizhou Province . The predominant phenolic acid was chlorogenic acid (0.186–19.174 mg/g; 2.02–8.28%) , which was also an abundant phenolic in D. fullonum . However, qualitative differences were reported between the leaves and the roots, with the highest level being found in the leaves obtained from plants in the second year of vegetation (28.44 mg/g d.w.). Wang et al. also found differences in the content of chlorogenic acid in plants of S. tschiliensis between the pre-flowering, flowering, and fruiting stages. The level was also dependent on the extraction solvent; the richest source of chlorogenic acid, 45.35 mg/g d.w., was found in the ethyl acetate fraction from plants at the pre-flowering stage. The Scabiosa genus is a rich source of flavonoids, which are identified mainly in the flowers but also in the leaves, stems, and roots . Many studies note that apigenin, luteolin, and their derivatives are particularly common . The new flavonoids identified for the first time in Scabiosa spp. were diosmetin-7- O -glucoside, luteolin-7,3′-diglucoside, luteolin 3′-glucoside, and quercetin 3,4′-diglucoside in S. atropurpurea subsp. maritima leaves ; quercimeritrin in S. atropurpurea stems; quercitrin, rutin, kaempferol-3- O - β - d -6- O -( p -hydroxycinnamoyl)-glucopyranoside, and kaempferol-3- O - β - d -(3,6-di- p -(hydroxycinnamoyl)-glucopyranoside in S. tschilliensis flowers ; and tamarixetin derivative 3- β - l -rhamnosyl-(1→2)[ β - l -rhamnosyl-(1→6)] β - d -glucoside] and tiliroside in S. stellata whole plants . Isoorientin, isovitexin, orientin, saponarin, and saponaretin were detected for the first time in Dipsacus spp., in D. fullonum leaves . Lignans were identified only in Dipsaci radix and include prinsepiol, fraxiresinol, and their derivatives such as dipsalignans A-D . Dipsaci radix also demonstrated derivatives of pinoresinol and syringaresinol . Syringaresinol hexoside was detected as a new compound in S. atropurpurea stems . Dipsacus spp. also include polysaccharides . The water-soluble polysaccharide from D. asperoides roots (ADAPW) has a molecular weight of 16 kDa and contains glucose, rhamnose, arabinose, and mannose in a molar ratio of 8.54:1.83:1.04:0.42 . The polysaccharide WDRAP-1, with a molecular weight of 61 kDa, was composed of glucose, mannose, galactose, arabinose, and rhamnose in a molar ratio of 3.1:0.9:5.2:1.1:0.3. The predominant monosaccharides were glucose (29.2%) and galactose (49.1%) . Xu et al. found two polysaccharides in D. asperoides roots, DAI-1 and DAI-2, which consisted only of glucose and had respective molecular masses of 17 and 4 kDa. Sun et al. isolated from D. asper roots a homogenous polysaccharide (DAP) with a molecular weight of 26.1 kDa that was composed of galactose and mannose in a molar ratio of 1:1. Essential oils are present in some species of Dipsacus and Scabiosa . The essential oils isolated by hydrodistillation from dried and fresh roots and leaves of D. fullonum were rich in many components; however, quantitative and qualitative differences were noted. The dominant compound in the essential oils from leaves, regardless of fresh or dried materials, was phytol (branched-chain unsaturated diterpene alcohol; precursor for vitamins E and K1), whose content ranged from 61.08% to 72.31%. It was not detected in root essential oil. In addition, the main components in fresh leaf essential oil, i.e., with a level above 5%, were 9,12,15-octadecatrienoic acid methyl ester and cyclohexane, cyclopropylidene. On the other hand, in the essential oils from the dry and fresh roots, the main component was n -hexadecanoic acid, especially in the dried material (16.00%), which was also rich in 11,14,17-eicosatrienoic acid, methyl ester (15.86%). n -hexadecanoic acid was also identified in the essential oils from dried and fresh leaves but at much lower levels (2.01–2.34%) . In the essential oil from the flowering aerial parts of D. japonicus , linalool (11.78%), trans -geraniol (8.58%), and 1,8-cineole (7.91%) predominated. The other main ingredients present at above 5% were β -caryophyllene (5.58%), α -terpineol (5.32%), β -selinene (5.15%), and spathulenol (5.04%) . Qualitative and quantitative differences were also observed in the essential oils isolated from different parts of S. arenaria . The main compounds detected in the oil from the aerial parts and flower oils were chrysanthenone (23.43–38.52%), camphor (11.75–12.98%), and α-thujone (9.5–10.7%), while α-thujone (34.39%), camphor (17.48%), and β-thujone (15.29%) predominated in the fruit oil. In addition, longifelone (2.41–3.96%) and filifolone (1.99–3.72%) were only identified in oils from the vegetative parts of the plants and the flowers. In S. atropurpurea stems, in volatile fractions VF1 (extracted by hexane) and VF2 (extracted by chloroform), the most abundant ingredients were 1,8 cineole (8.1–33.8%), tetradecene (5.7–24.1%), and (E)-β-ionone (5.9–20.7%). It is worth mentioning that dihydroactinidiolide, which was present in significant amounts (26.1%), was identified only in the chloroform fraction . The presence of fatty acids has only been investigated in S. stellata aerial parts and D. asper roots . Thirteen fatty acids including saturated and unsaturated acids have been identified. The fatty acids present in the hexane extract of S. stellata aerial parts accounted for 87% of the total content with linolenic acid, palmitic acids, and linoleic acids predominating . The use of traditional medicines, herbs, and supplements of plant origin is becoming more popular. However, it is necessary to determine the side effects that they may cause or their toxicity to use safely. An in vitro study showed that the extract of D. asperoides was not toxic for normal cells at a concentration up to 500 µg/mL . The viability of RAW 264.7 macrophages and periodontal ligament stem cells after treatment with water or ethanol extracts for 24 h was about 90–100% . However, the periodontal ligament stem cells treated for 7–21 days with 500 µg/mL of D. asper ethanol extract showed morphology changes . The cell viability of J774A.1 murine macrophage was also not affected by the methanol extract of D. inermis at a concentration up to 100 µg/mL . The concentration above 300 µg/mL decreased the cell viability below 80%. In addition, Dipsaci radix stimulated the proliferation of MC3T3-E1 and primary osteoblastic cells in the concentration range of 3–300 µg/mL after 24 h and 48 h . It should be also noted that akebia saponin D, a quality indicator of Dipsaci radix , at a concentration of 25–200 µM has no cytotoxic effect in mouse primary chondrocytes after 24 h . In addition, saponin favored the proliferation of rat bone marrow stomal cells on days 4 and 7 in a dose-dependent manner (0.01–10 µM) and enhanced the proliferation of human mesenchymal stem cells at a concentration up to 1 mg/L after 3 and 5 days . The toxicity of Dipsacus or Scabiosa plants has been tested not only in vitro but also in vivo. According to Zhou et al. , the clinical safety of D. asper has been evaluated by Zhan et al. . The total saponin extract (0.28 g/tablet) was administered to volunteers for six months, resulting in only mild side effects such as abdominal discomfort, constipation, swollen gums, raised level of blood alanine aminotransferase (ALT), and dysphoria. A study using F344 rats which were orally treated with the aqueous extract of Dipsaci radix at doses of 0.125, 0.25, 0.5, 1, or 2 g/kg body weight (b.w.)/day for 13 weeks resulted in no deaths or pathophysiological changes . On the other hand, Xiao et al. found that an extract of D. asper roots may have an adverse effect at a concentration of 2–32 g/kg/day. Dipsaci radix enhanced fetal malformation in a dose-dependent manner in pregnant ICR mice. In addition, the extract at a dosage of 32 g/kg/day, i.e., 17-times higher than recommended for an adult human, was toxic to the fetus; it led to abnormalities of fetal skeletal development including malformed limbs (polydactylia) and sternum (hypoplasia and split) as well as inhibited mineralization cartilaginous tissue and osteogenesis. It is worth emphasizing that the extract also inhibited the mouse embryogenic stem cells and 3T3 cells growth in a dose-dependent manner (0.1–125 mg/mL) with IC 50 values of 6.83 mg/mL and 5.13 mg/mL, respectively . The ethyl acetate extract of S. stellata whole plants orally administered for 14 days at a concentration of 0.5–2 g/kg in a single dose was not toxic for albino Wistar rats and did not induce animal mortality . Moreover, no changes in respiration and urination or in hematological and serum biochemical parameters such as ALT, aspartate aminotransferase (AST), total bilirubin, urea, creatinine, cholesterol, triglycerides, and glucose levels were observed in comparison to those in the control animals . Mouffouk et al. estimated the cytotoxicity of the hydroethanolic extract, petroleum ether, ethyl acetate, and n -butanolic fractions of S. stellata whole plants in larvae of brine shrimp lethality method. A dose-dependent pattern (at a concentration of 10–100 µg/mL) in the mortality rate of the brine shrimp nauplii was noted. Only the n -butanolic extract caused mortality above 50% at a concentration of above 80 µg/mL. The 70% ethanol extract of aerial parts of S. atropurpurea possessed an LD 50 value for adult male albino rats of 5 g/kg b.w. . Dipsaci radix The synergistic effect of numerous, various specialized metabolites of plants is responsible for the medicinal properties of plants. Therefore, knowing the chemical composition of herbal materials is the basis for understanding the mechanisms of their action . It is important to know the pharmacokinetics of drugs to understand the toxicity of their preparations or their therapeutic potential. The therapeutic effectiveness of the preparations/compounds is related to their bioavailability. The bioavailability of compounds depends on many different parameters, such as digestion, absorption, or metabolism . Akebia saponin D, the main ingredient of Dipsaci radix , shows low absorption. Therefore, its therapeutic effect is limited . Wang et al. suggested that the microcrystalline form of akebia saponin D obtained by antisolvent precipitation may enhance the bioavailability of akebia saponin D. A few specialized metabolites identified in the roots of D. asper , e.g., 4- O -caffeoylquinic acid, 3- O -caffeoylquinic acid, 3,5-di- O -caffeoylquinic acid, loganic acid, loganin, sweroside, dipsacoside B, and asperosaponin VI, showed rapid absorption after intragastric administration to Sprague-Dawley rats at a concentration of 75.6 g/kg; these ingredients reached the maximum plasma concentration in an hour . In addition, it was shown that sauteing with rice wine of D. asper roots enhanced the bioavailability of these specialized metabolites, indicated by a significant increase in maximum plasma concentration and area under the curve for the plasma concentration from zero to the last quantifiable time-point as well as an increase in the level of bioactive compounds in rat liver and kidney tissues compared with those in the crude material (aqueous extract) . Dipsacus and Scabiosa Species 6.1. Strengthening the Bone Tissue and Antiarthritic Activity The roots of D. asperoides , known in Chinese as Xu Duan, means “connects broken bones” . Until now, numerous in vitro and in vivo studies have demonstrated that Dipsaci radix or the pure compounds isolated from this plant material can be potential agents for promoting osteoblast formation and may have an anabolic systemic skeletal effect. Dipsaci radix can improve bone density and affect bone histomorphology. Dipsaci radix has also shown osteoprotective properties in ovariectomized animals . In healthy BALB/c mice, an increase in the bone trabeculae density, bone volume/tissue volume ratio, bone surface/tissue volume, and trabecular number and the depletion in the trabecular separation on the proximal tibia after drinking water extract of Dipsaci radix were observed . Osteoporosis is characterized by an increase in bone fragility. Osteoporosis can usually occur with aging and after menopause due to estrogen deficiency, which contributes to the reduction of bone density and bone mass and degradation of the microstructure . The hormone replacement therapy that is used in osteoporosis treatment increases bone density. However, the use of hormone replacement therapy for a long time should be limited due to the serious side effects of its use . Modern therapies enhance bone metabolism by promoting osteoblast activity and by inhibiting the effects of osteoclasts . The ethanol extract of D. asperoides roots showed concentration-dependent progestogenic activity (40–100 μg/mL) in the T47D progesterone receptor-positive human mammary adenocarcinoma cell line; 100 μg/mL of extract demonstrated the equivalent of 31.45 ng/mL of progesterone treatment. These results indicate that D. asperoides roots can be used as an option for progestins . Moreover, asperosaponin D may promote the osteogenic differentiation of human mesenchymal stem cells through the estrogen signaling pathway . Liu et al. reported that Dipsaci radix decoction has anti-osteoporosis properties in ovariectomized Wistar or Sprague-Dawley rats (oral treatment at a dose of 100–500 mg/kg b.w./day) by increasing trabecular bone formation and bone mineral density and preventing bone mass loss and trabecular structure changes; it also decreases the serum alkaline phosphatase (ALP) level and the level of bone turnover markers, e.g., serum osteocalcin and urinary deoxypyridinoline/creatinine ratio, and receptor activator for nuclear factor κB ligand (RANKL) in osteoblasts and bone marrow stromal cells of the tibia. A recent study showed that Dipsaci radix may be able to control osteoblast differentiation, osteoclast proliferation, and mineralization via regulating mitogen-activated protein kinases (MAPK), nuclear-kappa B factor (NF-κb), TNF-α, and Toll-like receptor (TLR4) signaling pathways . Intragastric treatment of bilaterally ovariectomized Wistar rats with wine processed Dipsaci radix (with a dose of 75.6 g/kg/day) resulted in protection from an increase in urine Ca/creatinine and P/creatinine levels and serum ALP and osteocalcin concentrations and increased the femur bone mineral density. These effects were comparable to that observed in rats treated with 17 β -estradiol . In traditional Chinese medicine, Xian-Ling-Gu-Bao capsules have been used to prevent and treat osteoporosis, osteoarthritis, aseptic bone necrosis, or climacteric syndrome. Xian-Ling-Gu-Bao was officially approved in 2002 by the China Food and Drug Administration as an over-the-counter drug for the treatment of osteoporosis . This product is composed of the raw material of six plant species: Epimedii herba (70%), Dipsaci asperoidis radix (10%), Anemarrhenae rhizoma (5%), Psoraleae fructus (5%), Rehmanniae radix (5%), and Salviae miltiorrhizae radix (5%) . Xian-Ling-Gu-Bao administered to ovariectomized C57/BL6 mice for six weeks displayed anti-osteoporosis effects by enhancing bone mineral density and bone strength and by decreasing the serum level of the bone formation marker procollagen type I N-terminal propeptide (PINP) and the bone resorption marker C-terminal telopeptide of type I collagen (CTX) . Wu et al. also reported that Xian-Ling-Gu-Bao showed the ability to prevent osteoporosis in two osteoporosis models, prednisolone-treated zebrafish ( Danio rerio ) and ovariectomized Sprague-Dawley rats. Xian-Ling-Gu-Bao altered the protein levels of osteoprotegerin and RANKL. An increase in the OPG/RANKL ratio may suppress bone loss. It is worth mentioning that a dose of 1800 mg/kg (a concentration of six-times the recommended daily dose) did not cause toxicity or adverse effects in the heart, kidney, liver, stomach, or small intestine . Another formulation containing Dipsaci radix used in traditional Chinese medicine for treating kidney disease, osteoporosis, strengthening bones, and bone fractures is Du-Zhong-Wan. This preparation is composed of Eucommiae cortex (salted bark of Eucommia ulmoides Oliv.) and Dipsaci radix in an equal weight ratio (1:1) . It was shown that Du-Zhong-Wan displayed anti-osteoporotic activity in the Sprague-Dawley rat osteopenia model. After treatment with Du-Zhong-Wan at a dose of 2–6 g/kg/day for 12 weeks, an increase in the bone mineral density of the femur was observed, together with an improvement of trabecular bone mass and microarchitecture, reduction of the bone resorption and tartrate-resistant acid phosphatase 5b (TRACP-5b) level, and higher serum level of osteocalein, and serum and endometrium level of estrogen . Tian et al. reported that Du-Zhong-Wan favored fractured callus formation and improved osteoblastogenesis and angiogenesis by increasing the level of the H-type vessel endothelium markers (CD31 and endomucin) and proangiogenic factor SLIT3 in C57BL/6 mice after ovariectomy with the open transverse femoral fracture. Niu et al. showed that the 60% ethanolic extract and total saponins of Dipsaci radix at a concentration of 50–500 mg/kg/day may have anti-osteoporotic properties in Sprague-Dawley female rats after bilateral ovariectomy or hindlimb unloading Sprague-Dawley rat model by preventing a decrease in bone mass and by improving bone mineral density, biomechanical strength, and trabecular bone architecture. In addition, Dipsaci radix suppressed osteoclastogenesis through a reduction of the serum level of bone turnover marker (osteocalcin) and urine concentration of phosphorus, calcium, and deoxypyridinoline/creatinine ratio. In vitro studies demonstrated that Dipsaci radix stimulated osteoblastic proliferation, maturation, and differentiation via bone morphogenetic protein-2 (BMP-2)/MAPK/Smad1/5/8-dependent Runt-related transcription factor 2 (Runx2) signaling pathway. Dipsaci radix inhibited osteoclastogenesis by an increase in OPG/RANKL ratio in MC3T3-E1 murine preosteoblasts and primary osteoblastic cells . The ethanol extract of D. asper enhanced osteogenic differentiation of periodontal ligament stem cells through activation of the vascular endothelial growth factor (VEGF)/PI3K/Akt pathway. In addition, periodontal ligament stem cells displayed greater mineralization and mRNA expression of osteogenesis-related genes such as Col-1, ALP, Runx2, and osteocalcin . Treatment with the triterpenoid akebia saponin D also inhibited the osteoclastic gene RANKL in bone marrow mesenchymal stem cells, suppressing osteoclastogenesis and promoting osteogenesis. In addition, the combination of akebia saponin D and BMP-2 immobilized in 2-N, 6-O-sulfated chitosan alleviated osteoclastic formation, enhanced osteogenesis, and promoted angiogenesis by stimulating SMADs, TGF-β1, VEGFA, and OPG/RANKL signaling pathways . Moreover, loganic acid enhanced osteoblastic differentiation in preosteoblast MC3T3-E1 cells and suppressed osteoclast differentiation of primary-cultured monocytes derived from mouse bone marrow . Akebia saponin D was also found to have a potential application in osteoarthritis therapy. Gu et al. demonstrated that it showed anti-inflammatory activity in mouse primary chondrocytes and alleviated osteoarthritis following the surgical destabilization of a medial meniscus model of osteoarthritis in C57BL/6 male wild-type mice. Akebia saponin D was able to inhibit the production of cyclooxygenase-2 (COX-2), inducible nitric oxide synthase (iNOS), nitric oxide (NO), prostaglandin E2 (PGE2), IL-6, TNF-α, and NF-κB; it also activated the nuclear factor erythroid-2-related factor 2 (Nrf2)/Heme oxygenase 1 (HO-1) pathway and suppressed the expression of matrix synthesis degradation-related proteins such as disintegrin, metalloproteinase with thrombospondin motifs 5 (ADAMTS-5), as well as MMP13 in IL-1β treated chondrocytes. It was also found to enhance the expression of Aggrecan and Collagen II . A recent study found it to also enhance the proliferation and differentiation of human mesenchymal stem cells into nucleus pulposus-like cells through p-ERK1/2 and p-Smad2/3 activation, which may prevent intervertebral disc degeneration . Akebia saponin D also favored the proliferation and osteogenic differentiation of rat bone marrow stromal cells through the phosphatidylinositol-3 kinase/AKT signaling pathway, thus inhibiting osteoporosis. This active compound elevated osteogenic differentiation markers such as ALP activity and calcium deposit formation and mRNA level of osteogenic-related genes (ALP, osteocalcin, type 1 collagen (COL 1), and RUNX2) . Another terpenoid compound, hederagenin, also alleviated the progression of osteoarthritis and inhibited inflammation and cartilage degradation. In an in vitro study, hederagenin exerted chondroprotective and anti-inflammatory effects by suppressing the JAK2/STAT3/MAPK pathway, inhibited extracellular matrix degradation, elevated Aggrecan and Collagen II levels, and reduced levels of MMPs and ADAMTS5. In addition, this pentacyclic triterpenoid saponin inhibited cartilage destruction in rats induced by monosodium iodoacetate . Hederagenin 3- O -(2- O -acetyl)-α- l -arabinopyranoside and the dichloromethane fraction of Dipsaci radix also improved osteoblastic differentiation of human alveolar bone marrow-derived mesenchymal stem cells. This specialized metabolite caused the formation of calcified nodules and enhanced the level of bone differentiation protein expression such as sialoprotein and osteocalcin similar to dexamethasone . Moreover, the main iridoid glycoside isolated from Dipsaci radix , sweroside (at a concentration of 1 µM), enhanced rat osteoblast-like UMR 106 cell proliferation, while loganic acid, loganin, and sweroside favored mineralization. Loganin and sweroside suppressed the formation of adipocytes in 3T3-L1 cells . The anti-osteoporotic property of sweroside was also confirmed by Wu et al. . This compound enhanced mineralization of MC3T3-E1 cells by increasing the protein expression of the membrane estrogen receptor-α and G protein-coupled receptor 30 (GPR30), which activate the p38 signaling pathway. MC3T3-E1 mouse embryonic osteoblast proliferation and differentiation were also promoted by various concentrations (25, 50, or 100 μg/mL) of a 17 kDa polysaccharide (DAI-1) isolated from D. asperoides in high glucose concentrations; this appeared to act via the stimulation of the bone morphogenetic protein 2 (BMP-2)/Smad/runt-related transcription factor 2 (Runx2)/Osterix signaling pathway. The polysaccharide also enhanced osteocalcin level and the mRNA and protein levels of BMP-2 and Runx2 . A homogenous polysaccharide with a molecular weight of 26.1 kDa isolated from the roots of D. asper (at a dose of 50 or 200 mg/kg b.w.) was also able to enhance mRNA and protein levels of VEGF and osteoprotegin, suppress mRNA and protein expression levels of RANKL, and activate the PI3K/Akt/eNOS signaling pathway in ovariectomized rats . The water-soluble polysaccharide (ADAPW), with an average molecular weight of 16 kDa, inhibited the viability of the human osteosarcoma HOS cells by induction of apoptosis cells and inhibition of the PI3K/Akt signaling pathway . An increase in bone resorption may also lead to osteo- or rheumatoid arthritis. Rheumatoid arthritis is a chronic autoimmune and inflammatory disease of the connective tissue . Dipsaci radix has been also applied for many years in traditional Chinese medicine to treat other bone diseases such as rheumatic arthritis . The aqueous extract of D. asperoides roots at a concentration of 50 mg/kg and 100 mg/kg, administered orally once a day for 21 days, displayed antiarthritic effects in collagen-induced rheumatoid arthritis in male DBA/1 mice by enhancement of the ankle joint architecture and suppression of arthritis score (synovitis, pannus, and bone erosion scores) and serum levels of anti-CII IgG2a antibody and the inflammatory mediators (TNF-α, IL-1β, and IL-6). These effects were comparable or stronger to those after treatment with 1 mg/kg of the anti-rheumatoid drug indomethacin . Akebia saponin D at a concentration of 10 or 20 mg/kg/day also showed anti-osteoclastogenic activity in the arthritic joints of male BALB/c and DBA/1 with collagen-induced arthritis. The saponin reduced mRNA expression of osteoclastogenesis markers such as TRAP, CtsK, MMP-9, and β3-integrin. In addition, akebia saponin D also inhibited phosphorylation of Akt, p38, and JNK and mRNA and protein levels of osteoclastogenesis markers in RANKL-induced osteoclastogenesis bone marrow-derived monocytes . Dipsaus saponins also inhibited chondrocyte apoptosis in a rat model of osteoarthritis in a dose-dependent pattern by decreasing expression of Bax, caspase-3, and caspase-9 and by increasing expression of Bcl-2 . Cantleyoside, an iridoid identified in D. asper roots, inhibited proliferation of human rheumatoid arthritis fibroblast synovial cells (HFLS-RA) and induced cell apoptosis through AMPK/Sirt 1/NF-κB pathway activation . Moreover, the protective effect of sweroside was observed in IL-1β-induced inflammation in rat articular chondrocytes. The anti-inflammatory effect of this iridoid was mediated by the inhibition of NF-κB and mTORC1 signaling pathways . These above findings suggest that Dipsaci radix may have a beneficial therapeutic effect in the treatment of postmenopausal osteoporosis and may protect against arthritis. 6.2. Anti-Neurodegenerative Activity Alzheimer’s disease is a neurodegenerative disease that destroys memory and deteriorates cognitive function . Five-month administration of 4 g/kg ethanol extract of D. asper roots improved neurocognitive dysfunction in the passive avoidance task and diminished expression of hippocampal β-amyloid protein (Aβ) positive cells in aluminum chloride-treated male Sprague-Dawley rats. This effect increased with the time of treatment (1–5 months). It may be important in the treatment of Alzheimer’s disease and memory system dysfunction . Akebia saponin D, at a dose of 30–270 mg/kg administered for four weeks, also showed a preventive effect against memory cognitive impairment in ibotenic acid-exposed male Sprague-Dawley rats . Moreover, it was shown that the saponin can protect against learning and memory dysfunction in rats induced by bilateral intracerebroventricular injections of Aβ1–42 in the Y-maze and Morris water-maze tests. Akebia saponin D attenuated the activation of microglia and astroglia and protein expression level of IL-1, COX-2, TNF-α, NF-κB, and Akt phosphorylation in the rat brain . The total saponin of D. asperoides also possessed neuroprotective properties in rat cortical and hippocampal neurons against damage induced with β-amyloid protein, ameliorated cell viability, and decreased lactate dehydrogenase (LDH) release and lipid peroxidation at a concentration of 150–300 mg/L . Other triterpene compounds, such as oleanolic acid, ursolic acid, and hederagenin, also showed neuroprotective activity in multiple brain disorders . For example, hederagenin diminished Aβ deposition in the head area of Caenorhabditis elegans , ameliorated cognitive impairment or pathological changes in APP/PS1 mice, and induced PPARα/TFEB-dependent autophagy of BV2 cells . Moreover, treatment with sweroside, a secoiridoid glycoside, alleviated memory deficits in scopolamine-induced Zebrafish ( Danio rerio ) in behavioral tests such as the tank diving test, the Y-maze, and the novel object recognition test . The flavonoid apigenin also suppressed neurotoxicity and cognitive function in LPS-induced mice. The compound protected against neuronal degenerative changes in mice hippocampi . It was also found that 50% methanol extracts from leaves and roots of D. fullonum demonstrated anti-acetylcholinesterase activity . The ethyl acetate and n -butanolic fractions of 70% methanol extract from S. stellata whole plants, at a concentration of 200 μg/mL, exhibited moderate (30.8%) and low (10.9%) acetylcholinesterase (AChE) inhibitory activity, respectively . On the other hand, the ethyl acetate and n -butanolic fractions of stems and leaves from S. arenaria (at a dose of 1 mg/mL) were able to inhibit AChE in 97.61% and 90.47%, respectively. The value of IC 50 ranged from 0.016 mg/mL to 0.029 mg/mL. In comparison to eserine (positive control), the IC 50 value was 0.0029 µg/mL . The root extract also displayed potent activity. The strongest effect was demonstrated by the n -butanolic fraction (87.61% AChE inhibition and IC 50 value = 0.02 mg/mL) . Yu et al. showed the effectiveness of some compounds isolated from the roots of D. asper ; for example, oleanane triterpenoid saponin (3- O - β - d -glucopyranosyl-(1→3)-α- l -rhamnopyranosyl-(1→2)-α- l -arabinopyranosyl-23-hydroxyolean-18-en-28-oic acid 28- O - β - d -glucopyranosyl-(1→6)- β - d -glucopyranosyl ester) was found to inhibit acetylcholinesterase with an IC 50 value of 15.8 μM. Other terpenoid compounds isolated from Dipsaci radix , such as dipsacus saponin IV, dipsacus saponin XI, and dipsacus saponin X, displayed strong AChE inhibitory activity, while cauloside A, dipsacus saponin C, and dipsacus saponin XI were more effective against butyrylcholinesterase. These activities were higher than that of the positive control, berberine. Moreover, the saponins were found to be more effective than the iridoids, loganic acid and sweroside. The terpenoids also inhibited β-site amyloid precursor protein cleaving enzyme 1 (BACE1) and advanced glycation end-product (AGE) formation . 6.3. Hepatoprotective Activity Liver fibrosis is a chronic liver disease caused by many agents such as hepatitis B virus (HBV), hepatitis C virus (HCV), non-alcoholic steatohepatitis, or alcoholic fatty liver disease . The flavonoids and phenolic acids from Scabiosa spp. have demonstrated anti-hepatic fibrosis potential in male Wistar or male Sprague-Dawley rats treated intraperitoneally with carbon tetrachloride (CCl 4 ) (a selective hepatotoxic drug); similar effects were also noted for drugs used in Mongolian medicine, such as Qingganjiuwei and Gurigumu-7 (composed of various herbs including S. comosa flowers) . Qingganjiuwei, a drug commonly used in Inner Mongolia in patients with chronic hepatic disease, improved liver morphology and structure in CCl 4 -treated SD rats when administered at 1.575–4.725 g/kg/day for eight weeks. The drug reduced hepatocyte necrosis, lymphocytic infiltration, and pseudolobuli and lowered COL1, tissue inhibitor of metalloproteinase1 (TIMP1), and α-smooth muscle actin (α-SMA) expression. The drug also activated the MAPK pathway in the liver through the suppression of extracellular signal-regulated kinase (ERK), C-Jun amino-terminal kinases (JNKs), and stress-activated protein kinase-2 (p38 proteins) . Qingganjiuwei also increased mRNA and protein expressions of MMP2 and MMP9 and inhibited the levels of the serum aminotransferases (ALT and AST) . The methanol-eluted fraction of Gurigumu-7 extract (0.264 g/kg) displayed a more potent hepatoprotective effect in mice with CCl 4 -induced liver damage compared to crude Gurigumu-7 extract, even when applied at a four-times higher concentration (1.084 mg/kg). The methanol fraction alleviated histopathological changes in the liver and serum ALT, ASP, and liver malonyldialdehyde (MDA) levels and enhanced the liver superoxide dismutase (SOD) level in a dose-dependent manner (66, 132, and 264 mg/kg) . Anti-hepatic fibrosis was also reported for S. comosa and S. tschilliensis . Some specialized metabolites inhibited the viability of hepatic stellate LX-2 cells at concentrations of 12.5–200 μM; this included those belonging to flavonoids, which accounted for about 60% of the total identified compounds in the inflorescences of Scabiosa plants. The flavonoids enhanced the expression of Stat1, Pparg, Hsp90aa1 genes, signal transduction and transcriptional activator 1 (STAT1), and peroxisome proliferator-activated receptor G (PPARG) proteins, which play key roles in the pathogenesis of liver fibrosis . Apigenin exhibited the strongest ability to inhibit cell proliferation . The flavonoid-rich extract of S. comosa inflorescences at concentrations of 100 and 200 mg/kg also suppressed hepatic fibrosis in Wistar rats pre-treated with CCl 4 ; the extract inhibited the level of biochemical parameters in blood serum (ALT, AST, ALP, and hyaluronic acid), the markers of liver fibrosis (laminin, amino-terminal propeptide of type III procollagen (PIIINP), collagen IV, collagen deposition in the liver tissues), and expression of α-SMA, collagen I, and fibronectin . Moreover, the extract attenuated phosphorylation of Smad3 in liver tissue and TGF-β1-pre-treatment primary mouse hepatic stellate cells. In the latter, it was observed that the expression of the fibrotic genes (α-SMA, collagen I, and fibronectin) was suppressed in a dose-dependent manner . Ethanol extract of aerial parts of S. atropurpurea and hexane, ethyl acetate, n -butanolic, and chloroform fractions were able to decrease the levels of serum ALT, AST, and ALP in CCl 4 -induced liver damage in albino rats after treatment with a dose of 100 mg/kg . Sweroside demonstrated a protective effect against liver fibrosis in mouse models treated with CCl 4 and methionine-choline-deficient diet-induced non-alcoholic steatohepatitis . Sweroside treatment yielded an anti-fibrotic effect through the FXR-miR29a pathway . Sweroside also improved NASH symptoms by inhibiting the activation of the hepatic NLRP3 inflammasome . Apigenin was found to inhibit palmitic acid-induced pyroptosis by regulating the pyrin domain containing 3 (NLRP3) inflammasome in HepG2 cells and primary mouse hepatic cells . Other polyphenols, derivatives of caffeoylquinic acid such as 1,5-di- O -caffeoylquinic acid, isochlorogenic acid C, isochlorogenic acid B, chlorogenic acid, isochlorogenic acid A, neochlorogenic acid, and caffeic acid also suppressed LX-2 hepatic stellate cell growth . The two predominant caffeoylquinic acid derivatives in inflorescences of S. comosa and S. tschilliensis , viz., 3,5-di- O -caffeoylquinic acid and chlorogenic acid, demonstrated anti-hepatitis C virus (HCV) activity in the Huh-7.5 cell line infected with HCV . In addition, some triterpenoid derivatives isolated from the whole plants of S. tschiliensis , such as scabiosaponins E, F, G, I, and J; hookerosides A and B; and prosapogenin 1b, showed strong inhibition of pancreatic lipase. Moreover, 0.12 mg/mL prosapogenin 1b exerted similar inhibitory properties as the lipase inhibitor orlistat at a concentration of 0.005 mg/mL . Akebia saponin D reduced lipid droplet accumulation in BRL cells, attenuated hepatic steatosis, and elevated the expression of Bcl-2/adenovirus E1B 19-kDa interacting protein 3 (BNip3) and phospho-AMPP; it also improved mitochondrial function and autophagy modulation, inhibited rotenone-induced BRL cell apoptosis, elevated Bcl-2/Bax ratio, and suppressed the level of intracellular reactive oxygen species (ROS) and mitochondrial membrane potential loss in rotenone-treated BRL cells and rat liver mitochondria . 6.4. Cardioprotective Activity Cardiovascular diseases are one of the leading causes of death worldwide . The anti-atherosclerotic effect of akebia saponin D was studied in vitro in H 2 O 2 -treated human umbilical vein endothelial cells (HUVECs) and in vivo in ApoE −/− mice . It was shown that this saponin protected against H 2 O 2 -induced cytotoxicity in HUVEC, inhibited ROS level, the mitochondrial membrane potential disruption, and apoptosis in oxidative stress-induced endothelial cells in a dose-dependent manner (50–200 µM); the mechanism was believed to involve increasing Bcl-2 family protein levels and decreasing caspase-3 and Bax activation. Moreover, doses of 150 mg/kg/day and 450 mg/kg/day reduced aortic plaque formation in mice as well as aortic and liver apoptosis, serum triglyceride (TG), total cholesterol (TC), low-density lipoprotein cholesterol (LDL-C), and lipid deposition in the liver, as well as atherosclerotic lesion size; it also enhanced the expression of antioxidant enzymes (SOD, catalase (CAT), and glutathione (GSH)) in vascular tissue and liver . Li et al. demonstrated that sweroside has a protective effect on ischemia-reperfusion-induced myocardial injury by inhibiting oxidative stress and pyroptosis partially via modulation of the Kelch-like ECH-associated protein 1 (Keap1)/Nrf2 pathway. The iridoid also suppressed aconitine-induced cardiac toxicity in the H9c2 cardiomyoblast cell line . Long-term gavage (for six weeks) of akebia saponin D protected against fibrosis myocardial ischemia injury, inhibited cardiac dysfunction, and reduced infarct size in a Sprague-Dawley rat model with chronic myocardial infarction induced by permanent ligation of the left coronary artery. Treatment with this saponin decreased hydroxyproline level and changed the activity of the oxidative stress enzymes by elevating SOD and DSH-peroxidase (GSH-Px) levels and by reducing MDA content. Moreover, akebia saponin D regulated inflammatory mediators by diminishing the levels of TNF-a and IL-6 and by elevating the level of IL-10 . Several in vitro and in vivo studies found flavonoids detected in Scabiosa spp., such as apigenin and luteolin, to have cardioprotective properties . However, Song et al. demonstrated that D. asper roots, drunk widely as a tea for beneficial health effects, or dipsacus saponin D may have an undesirable effect on platelets and may increase the risk of thrombosis. D. asper roots were found to be an herb with procoagulant activities on platelets and prothrombotic properties. Dipsacus saponin C elevated procoagulant activity in a dose- and time-dependent manner, elevated intracellular calcium level, and decreased ATP. In addition, it caused translocation of Bax and Bak, cytochrome c release, caspase-3 activation, and the disruption of mitochondrial membrane potential. The oral administration of 10 mg/kg and 25 mg/kg dipsacus saponin C also resulted in an increase in thrombus formation in a rat venous thrombosis model . 6.5. Renal and Gastritis Protection A 61 kDa polysaccharide (WDRAP-1) isolated from D. asperoides roots showed protective activity against oxidative stress generated in renal ischemia-reperfusion injury in male Wistar rats and displayed strong superoxide and hydroxyl radical scavenging activities in vitro. It was shown that oral pre-treatment of rats with the polysaccharide at a concentration of 50–200 mg/kg b.w. for 14 days before ischemia-reperfusion may improve renal injury (especially at the highest dose); treatment suppressed the level of renal injury indicators including creatinine, blood urea nitrogen, lactate dehydrogenase (LDH), and serum MDA and enhanced serum SOD and some renal tissue antioxidant enzyme activities (SOD, GSH-Px, and CAT) . Hederagenin was also found to protect against renal fibrosis. This terpenoid compound attenuated the proliferation and fibrosis of TGF-β-treated NRK-49 F cells by targeting the muscarinic acetylcholine receptor . Dipsacus saponin C, a saponin isolated from the roots of D. asper , was found to have a protective effect against HCl·ethanol-induced gastritis and indomethacin-induced gastric ulcers in male Sprague-Dawley rats. It was shown that treatment with dipsacus saponin C caused a decrease in gastric secretion volume and gastric acid production in pylorus-ligated rats. It was found that this compound had a moderate effect on colonization and growth inhibition of Helicobacter pylori at a concentration of 50–100 µM. In addition, dipsacus saponin C also showed a cytotoxic effect for SNU638 and AGS human gastric cancer cells with an IC 50 at 54.6 mM and 37.3 mM, respectively . Moreover, akebia saponin D may exert a therapeutic role by regulating the intestinal microbiome and protecting intestinal epithelial cells from external damage. It was found to achieve this by inhibiting oxidative damage to the intestinal barrier by downregulating PPAR-γ/FABP4 in the human intestinal cell line FHs74 Int . Luteolin was also found to have therapeutic effects against interstitial fibrosis-induced renal anemia in vitro and in vivo. This activity was mediated via the SIRT1/forkhead box O3 (FOXO3) pathway . 6.6. Anti-Asthmatic Effect Dipsaci radix alleviated the asthmatic response in BALB/c mice with allergic asthma induced by an ovalbumin. Dipsaci radix treatment at a dose of 20 m/kg or 40 mg/kg resulted in attenuation of the methacholine response, inflammatory cell infiltration, and mucus secretion in the bronchial airway and a decrease in the levels of pro-inflammatory cytokines (IL-5 and IL-13) in bronchoalveolar lavage fluid, eotaxin, serum total IgE, expression of iNOS, and NF-κB phosphorylation in lungs . Similarly, apigenin, one of the flavonoids identified in Scabiosa spp., was found to inhibit inflammatory mediators and eosinophilia in lung and airway tissues in in vivo acute lung injury and asthma models . 6.7. Anti-Diabetic Activity A polysaccharide (DAP) isolated from D. asper roots demonstrated beneficial effects on renal function and renal pathological changes as well as antihyperglycemic, hypolipidemic, and antioxidant activities in streptozotocin-induced diabetic Wistar rats . Four weeks of intragastric administration (100 mg/kg and 300 mg/kg per day) in type 2 diabetic rat model resulted in a reduction of glycosylated, fasting blood glucose, serum creatinine, blood urea nitrogen, urine protein, and urinary albumin excretion. Moreover, oral administration of the polysaccharide (300 mg/kg) suppressed the serum level of TC, TG, LDL, and renal AGE-RAGE formation (advanced glycation end products-receptor for advanced glycation end products) and enhanced SOD, CAT, and GSH activities in the kidney of rats with diabetic nephropathy . Similar antilipemic effects were also observed for hederagenin. This pentacyclic triterpene exerts its potential through the p38MAPK pathway in oleic acid-induced HepG2 cells and in hyperlipidemic Sprague-Dawley rats . The ethanol extract and the hexane, ethyl acetate, n -butanolic, and chloroform fractions of the aerial parts of S. atropurpurea demonstrated anti-hyperglycemic activity by decreasing the blood glucose level in albino rats with alloxan-induced hyperglycemia . The methanol extract of S. atropurpurea subsp. maritima whole plants demonstrated α-glucosidase inhibitory activity with an IC 50 value = 100 μg/mL. This effect was higher than that found for the positive control, the anti-diabetic drug acarbose (IC 50 = 196 μg/mL) . Methanolic extracts from the fresh leaves and roots of D. fullonum also inhibited porcine pancreatic α-amylase activity . These extracts demonstrated low effectiveness in this study, with the strongest activity being found to be IC 50 = 86.01 µg/mL for the dried leaf extract. This activity was more than 100-times lower compared to that of acarbose (IC 50 = 0.69 μg/mL). 6.8. Anti-Inflammatory Activity Several plant species of Dipsacus and Scabiosa or some specialized metabolites isolated from them can be valuable, new anti-inflammatory agents . The water extract of D. asper roots demonstrated anti-inflammatory properties in the lipopolysaccharide (LPS)-activated murine macrophage cell line RAW 264.7 by suppressing NO production with an IC 50 = 45.1 µg/mL . In a later study , an aqueous extract of D. asperoides roots at a dose of 50–500 µg/mL showed inhibitory potential on inflammation and oxidative stress in RAW 264.7 macrophages exposed to LPS; it was found to act by lowering NF-κB and ERK1/2 phosphorylation, nuclear translocation of NF-κB, and activation of Nrf2/HO-1. The extract reduced the levels of inflammatory mediators (iNOS, COX-2, and cytokines IL-6 and IL-1β) as well as ROS levels . The methanol extract of D. inermis leaves also showed the ability to inhibit the production of NO, COX-2, PGE2, pro-inflammatory mediators (IL-1β and IL-6, and TNF-α), intracellular ROS level, and phosphorylation of NF-κBp65 and IκBα in a dose-dependent manner (25–100 µg/mL) in the LPS-induced murine macrophage cell line J774A.1 . The anti-inflammatory potential of D. inermis leaf extract was also confirmed in vivo in Wistar albino rats; it protected against vascular permeability (caused by acetic acid) and paw oedema (induced by carrageenan) in a concentration- and time-dependent manner. In addition, the serum levels of TNF-α, IL-1β, and IL-6 were significantly reduced while IL-10 level was enhanced after oral administration of the extract at a concentration of 50–100 mg/kg b.w. . Anti-inflammatory properties in a Wistar rat model of carrageenan-induced paw oedema also demonstrated the ethyl acetate extract of S. stellata whole plants. The highest activity was observed in the first hour after treatment with the extract at a concentration of 50 mg/kg (72.73% of inhibition). This effect was stronger than that of diclofenac (about 45% of inhibition). In addition, the anti-inflammatory effect of the plant extract lasted up to 24 h . Some compounds belonging to iridoids, saponins, or phenolic acids (for example, dipsasperoside A, dipsanoside A and B, dipsacus saponin A, akebia saponin D, or caffeic acid) isolated from the roots of D. asper also were able to reduce the production of NO in RAW 264.7 cells. The potent activity was demonstrated by akebia saponin D and dipasperoside A, with IC 50 values of 12.7 µM and 15.2 µM, respectively; these values were higher than those for the positive control, a nonselective NOS inhibitor, N G -monomethyl-L-arginine (IC 50 = 22.6 µM) . Reduced NO levels and iNOS expression have also been observed in LPS-induced RAW 264.7 cells after treatment with akebia saponin D (at a concentration of 25–100 µM) . Akebia saponin D also suppressed the expression of DNA methyltransferase (DNMT) 3b, the levels of PGE2 and p-STAT3, as well as the protein and mRNA levels of IL-6 and TNF-α . The levels of the inflammatory indicators, prostaglandin E2, i-NOS, COX-2, TNF-α, IL-1β, and IL-6, were also decreased after treatment with 40 and 80 µM sweroside in LPS-induced RAW264.7 cells. In addition, sweroside suppressed inflammation through the sirtulin 1 (SIRT1)/NF-κB and SIRT1/Forkhead transcription factor O1 signaling pathways . Akebia saponin D also showed anti-inflammatory activity in vivo by reducing paw oedema in carrageenan-induced Sprague Dawley rats and by inhibiting xylene-induced ear swelling in mice. It also lowered the level of NO in rat plasma in a carrageenan-induced rat paw oedema model . Anti-inflammatory potential was also demonstrated by apigenin, which was believed to act by the modulation of the p38/MAPK, PI3K/Akt and NF-κB pathways . 6.9. Antioxidant Activity The most frequently used methods to determine the antioxidant properties of Dipsacus and Scabiosa plant extracts were the DPPH (2,2-diphenyl-1-picrylhydrazyl radical scavenging assay), ABTS (2,2′-azinobis(3-ethylbenzothiazoline-6-sulfonic acid) radical cation assay), ORAC (oxygen radical absorbance capacity), FRAP (ferric ion reducing antioxidant assay), and CUPRAC assays (cupric-reducing antioxidant capacity). The antioxidant activity varied depending on the plant material (whole plants, leaves, and roots), the solvent, extraction time, and the extraction method . The antioxidant activity of the methanolic extract from D. asper roots was confirmed in DPPH and Cu 2+ -mediated LDL oxidation with IC 50 values of 90.2 and 134.4 µg/mL, respectively. This activity may be attributed to caffeoylquinic acid derivatives identified in root extract such as 3,4-di- O -caffeoylquinic acid, 4,5-di- O -caffeoylquinic acid, 3,5-di- O -caffeoylquinic acid, and their methyl derivatives. These specialized metabolites showed potent antioxidant activity against DPPH radical formation and Cu 2+ -mediated LDL oxidation with IC 50 values of 10.4–18.2 µM and 1.8–2.3 µM, respectively . The content of the total polyphenols in the acetone/water extract (7:3) of D. fullonum whole plant was 19.52 mg GAE (gallic acid equivalents)/g d.w. of plant material; it displayed an antioxidant capacity lower than 5 mmol TEAC (Trolox equivalent antioxidant capacity)/100 g d.w. plant material in the DPPH and ABTS assays . The antioxidant properties were also demonstrated in 50% methanolic extract of leaves and roots (ultrasound assisted extraction) in the ORAC assay with values of 14.78 mmol TEAC/100 g d.w. and 10.87 mmol TEAC/100 g d.w. for the leaf and root extracts, respectively . A similar observation was found by Saar-Reismaa et al. . The crude 70% ethanol extract of D. fullonum leaves showed antioxidant activity in the ORAC assay (10.8 mmol TEAC/100 mL). The fraction NP7 of crude leaf extract that was rich in two chlorogenic acid derivatives, saponarin and isoorientin, also displayed antioxidant activity (12.5 mmol TEAC/100 mL), while the fraction NP2 containing bis -iridoids was ineffective (0.78 mmol TEAC/100 mL) . On the other hand, the aqueous extract of D. fullonum leaves obtained by ultrasound assisted extraction (extraction time, four hours) showed high radical scavenging activity (RSA) (73.81%) in DPPH . The procedure extraction with the aqueous solution of amino acid ionic liquids, viz., triethanolammonium salts of two amino acids methionine ([TEAH] + [Met] − ) and threonine ([TEAH] + [Thr] − ), resulted in a beneficial effect on the antioxidant activity of leaf extracts in FRAP and CUPRAC assays compared to the extract after extraction with the pure water. In addition, extraction with the aqueous solution of [TEAH] + [Thr] − (extraction time, two hours) increased the total polyphenol content to 8.16 mg GAE/g raw material. The use of [TEAH] + [Met] − in aqueous solution and reducing the extraction time to one hour resulted in a similar level of polyphenol (7.38 mg GAE per g of raw material) . Antioxidant effect was also observed for the extract of D. sativus leaf. The extract at a dose of 300 mg/kg/day in ICR mice pre-treated with D-galactose to induce oxidative stress resulted in an increase in the level of SOD and a decrease in the level of MDA in the peripheral blood plasma . S. stellata whole plants are a rich source of polyphenolic compounds. The ethyl acetate and n -butanolic fractions have moderate antioxidant activity in simple, chemical, antioxidant tests such as DPPH, ABTS assays, and FRAP test. The n -butanolic fraction displayed stronger potential in the DPPH assay, i.e., FRS 50 (free radical scavenge) = 64.46 µg/mL, compared to ascorbic acid FRS 50 = 8.21 µg/mL . The dichloromethane fraction did not possess significant activity in antioxidant assays, which may be related to the low content (below 1 mg of gallic acid/g dry extract) of phenolic compounds . A 70% ethanol extract of S. stellata whole plants in DPPH also displayed antioxidant activity, with an IC 50 value of 86 μg/mL . The petroleum ether, ethyl acetate, and n -butanolic fractions from S. stellata whole plants showed various antioxidant potential in different chemical models such as DPPH and ABTS, FRAP, CUPRAC, β-carotene, phosphomolybdate assay, ferrous ions, and metal chelating assays . The strongest effect was demonstrated by the n -butanolic fraction in DPPH (IC 50 = 21.22 μg/mL) and chelation in ferrous iron assay (EC 50 = 1.65 mg/mL), while the ethyl acetate fraction was most active in ABTS (IC 50 at 14 μg/mL), CUPRAC (A 0 . 50 = 28.5 µg/mL), and β-carotene assays (IC 50 = 10.34 μg/mL). In addition, the n -butanolic fraction in DPPH had a higher reducing power than butylated hydroxyl toluene (BHT) (22.32 µg/mL) but lower than α-tocopherol (13.02 µg/mL), butylated hydroxy anisole (BHA) (6.82 µg/mL), and ascorbic acid (3.1 µg/mL) . The highest protein denaturation inhibition was found for the ethyl acetate extract, which showed 78.86% inhibition at the maximal tested concentration (1 mg/mL). Ibuprofen (standard drug) at the same concentration caused 100% inhibition . Antioxidant activity was also reported for 70% ethanol extracts of S. comosa and S. tschilliensis inflorescences. S. tschilliensis had a stronger antioxidant activity than S. comosa in DPPH, ABTS, and FRAP assays. For example, the IC 50 values in DPPH were 272.8 µg/mL and 331.1 µg/mL, respectively . Wang et al. reported that the 90% ethanol extract of S. tschilliensis at a concentration of 26.5 µg/mL scavenged 50% of DPPH free radicals (IC 50 value for ascorbic acid was 5.41 µg/mL). The crude extract (95% ethanol) and four solvent partitioned fractions (water, n -butanol, ethyl acetate, and petroleum ether) from S. tschiliensis whole plants at various growing stages (pre-flowering, flowering, and fruiting stage) showed different antioxidant activities in DPPH, ABTS, inhibition of lipid peroxidation, or OH scavenging activity . The IC 50 values for the crude extract were in the range of 25.65–86.79 µg/mL while the ethyl acetate fraction from the pre-flowering stage of plants had the highest antioxidant capacity (IC 50 8.47 µg/mL) in DPPH. This value was comparable to that of vitamin C (7.6 µg/mL). The ethyl acetate fraction from the pre-flowering stage of plants also possessed the highest ABTS (58.76 µg/mL), hydroxyl radical scavenging ability (67.64 µg/mL), and lipid-peroxidation-inhibition activity . The n -butanolic and ethyl acetate fractions of S. arenaria roots also displayed strong antioxidant activity in four assays: DPPH, ABTS, reducing power, and β-carotene bleaching inhibition activity. The n -butanolic fraction demonstrated excellent ability, mainly in the β-carotene bleaching inhibition assay (IC 50 = 0.018 mg/mL). This effect was stronger than that obtained for BHT (IC 50 = 0.04 mg/mL). In the DPPH and ABTS assays, the IC 50 values for both fractions were comparable to BHT . The ethyl acetate fractions of the roots, flowers, fruits, and aerial parts (stems and leaves) of S. arenaria showed the beneficial antioxidant ability in DPPH, with IC 50 values of 0.017–0.019 mg/mL; the best properties were observed for the flowers . Other species of Scabiosa , S. artropurpurea and S. atropurpurea subsp. maritima , also demonstrated antioxidant properties. In comparison to ascorbic acid (IC 50 = 0.084 mg/mL), among four tested extracts, the ethanol extract of S. artropurpurea stems exhibited the best antioxidant capacity, with IC 50 = 0.1383 mg/mL in DPPH assay. In addition, the hexanoic volatile fraction (VF1) and ethyl acetate extract displayed similar effects with IC 50 values of 0.4798 mg/mL and 0.4806 mg/mL, respectively . The antioxidant activity of 70% ethanol extract of aerial parts of S. atropurpurea and hexane, ethyl acetate, n -butanol, and chloroform fractions were also demonstrated by increasing blood glutathione in diabetic albino rats . Silver nanoparticles with S. atropurpurea subsp. maritima water fruit extract also was found as a promising antioxidant agent in DPPH and FRAP assays. The IC 50 values were 0.112 mg/mL and 0.036 mg ascorbic acid equivalent antioxidant capacity/g d.w., respectively, and were comparable to that for ascorbic acid . The pure compounds isolated from D. asper roots, i.e., six dicaffeoylquinic acid derivatives (3,4-di- O -caffeoylquinic acid, methyl 3,4-di- O -caffeoyl quinate, 3,5-di- O -caffeoylquinic acid, methyl 3,5-di- O -caffeoyl quinate, 4,5-di- O -caffeoylquinic acid, and methyl 4,5-di- O -caffeoyl quinate) exhibited strong antioxidant capacity in the DPPH assay (10.4–18.2 µM) and displayed inhibitory activity against Cu 2+ -mediated LDL oxidation (1.8–2.3 µM), stronger than those obtained for the positive controls, BHT and caffeic acid . In another study, 3,5-di- O -caffeoylquinic acid also demonstrated significant antioxidant properties in the DPPH assay, with an IC 50 value of 3.63 µg/mL . Two secoiridoid glucosides, eustomoruside and eustomoside, and one flavonoid, isoorientin, isolated from the whole plant S. stellata Cav. also displayed strong radical scavenging activities in DPPH with an IC 50 value of 7.1–8.5 μg/mL compared to that for ascorbic acid (IC 50 = 6.3 μg/mL) . The polysaccharide fraction from the roots of D. asperoides also demonstrated antioxidant effects in DPPH and ABTS assays but with little potency; the EC 50 value was 0.355 mg/mL in DPPH free radical scavenging activity and 5.867 mg/mL in ABT . 6.10. Anticancer Activity The synthetic drugs used in chemotherapy not only have strong toxic effects on cancer cells, but they also have strong adverse side effects in chemotherapy. Many plant specialized metabolites are used in cancer therapies and new substances, and new plant species with potential anti-cancer activity are being sought. Considerable attention has been noted regarding the cytotoxicity of Dipsacus and Scabiosa against various cancer cell lines including lung carcinoma A549, hepatoma Bel7402 and Hep3B, gastric carcinoma BGC-823, AGS, KATO III, MKN-45, and SNU-638, liver H157 and HepG2, colon cancer HCT-8, ovary cancer A2780, breast MCF-7, human breast cancer MCF-7 and MDB-MB-231, acute myeloid leukemia OCI-AML3, osteosarcoma HOS, or fibrosarcoma HT1080 cell lines. Some pure specialized metabolites of different classes identified in the plant extracts have also demonstrated potent or promising cytotoxic effects in vitro . An aqueous extract of D. asperoides roots inhibited the viability of human mammary carcinoma-derived triple negative MDA-MB-231 cells (with IC 50 = 15 μg/mL) and arrested the cell cycle in the G 2 /M phase; it also induced apoptosis by increasing pro-apoptotic caspase 3/7 activity and by suppressing the expression of BRAF, p-ERK, MEK, pPI3K, pAKT, and cyclin-dependent kinase 4/6 in a dose-dependent manner . The cytotoxic activity of bis -iridoid glycosides fraction of D. fullonum leaf methanol extract (with sylvestroside III and IV as the main compounds) was evaluated against human breast cancer cell lines MCF7 and MDB-MB-231 and human cervical cancer HeLa cell line . The two breast cancer cell lines were most sensitive to the fraction, resulting in a viability of 64.0% for MCF7 cells and 69.5% for MDB-MD-231 cells . In addition, the ethanolic extracts of the aerial parts and flowers of D. fullonum have antiproliferative activity on the human hepatocellular carcinoma Hep3B cell line with an IC 50 value above 100 µg/mL . The methanol extract of S. atropurpurea subsp. maritima leaves at a concentration of IC 10 , IC 20 , or IC 30 enhanced the toxicity of doxorubicin in human epithelial colorectal adenocarcinoma Caco-2 cells with IC 50 = 1.04 µg/mL (vs. 2.41 µg/mL when the cells were treated only with doxorubicin) . In addition, the combination of doxorubicin with S. atropurpurea extracts at a concentration of IC 50 and IC 10 , respectively, increased the percentage of apoptotic cells, the percentage of caspase-activated cells, mRNA levels of the apoptosis related-genes (Bax, caspase-3, p21), and decreased the expression level of anti-apoptotic genes (Bcl-2). It was a stronger effect than that obtained for doxorubicin or S. atropurpurea extract alone. The plant methanol extract also reversed P-glycoprotein or multidrug resistance-associated protein in Caco-2 cells . The use of silver nanoparticles with S. atropurpurea subsp. maritima water fruit extract was found to be promising anticancer agents with cytotoxic activity against the human multiple myeloma U266 cell line and the human breast cancer cell line MDA-MB-231. The silver nanoparticles inhibited the growth of cells in a concentration-dependent manner with IC 50 values of 10 and 12 µg/mL, respectively . Some specialized metabolites such as phenolic acids, triterpenoid derivatives, or iridoids isolated from Dipsacus or Scabiosa revealed cytotoxic effects in various cancer cell lines . Phenolic acids, such as caffeic acid, 2,6-dihydroxycinnamic acid, vanillic acid, 2′- O -caffeoyl- d -glucopyranoside ester, and caffeoylquinic acid, demonstrated cytotoxic activity against five cancer cell lines (A549, Bel7402, BGC-823, HCT-8, and A2780) with IC 50 values ranging from 3.883 µg/mL to 7.395 µg/mL. The positive control, fluorouracil (a known cytostatic compound), had an IC 50 value of 0.177–0.695 µg/mL . Akebia saponin PA from D. asperoides caused the death of various human gastric cancer cell lines (AGS, MKN-45, SNU-638, and KATO III) via both apoptosis and autophagy. The IC 50 values were 24.1 µM (MKN-45 cells), 27.6 µM (SNU-638), 30.3 µM (AGS), and 36.5 µM (KATO III). In addition, akebia saponin PA increased the AGS cell number in the sub-G 1 phase and activated caspase-3, cleavage of PARP-1, MAPK, and p38/c-Jun N-terminal kinase. Autophagy was induced through the PI3K/AKT/mTOR and AMPK/mTOR pathways . Another saponin, akebia saponin D, was also found to induce cytotoxicity of the human monocyte-like histiocytic U937 cells in a concentration-dependent manner (0.1–1000 µM); it also enhanced the percentage of sub-G 1 cells and increased Bax and p53 gene expression . Saponin XII isolated from the roots of D. japonicus (1–2 µg/mL) suppressed the growth of acute myeloid leukemia OCI-AML3 cells; it stimulated apoptosis, increased the number of cells in the G 0 /G 1 phase of the cell cycle, decreased the number of cells in the S and G 2 /M phases, and activated caspase-3 . Some triterpenoid saponins and iridoids isolated from S. stellata whole plants were found to have cytotoxic effects against the fibrosarcoma HT1080 cell line . Scabiostellatoside F, at a concentration of 12.0 mM, was able to inhibit HT1080 cell growth by 50% . Other triterpenoid saponins, scabiostellatoside B, D, E, and H, were found to have an IC 50 of 38–49 µM. In addition, scabiostellatoside A, C, and G were not cytotoxic at a concentration of 50 mM . Yu et al. found that some compounds isolated from D. asper roots such as ursane and oleanane type triterpenoids with a feruloyloxy group or an arabinosyl moiety at C-3 were more cytotoxic than arboinane-type triterpenoids against four tumor cell lines: lung A549, liver H157 and HepG2, and breast MCF-7. Moreover, the highest activity was shown by an ursane-type triterpenoid (3 β - O -trans-feruloyl-2 α -hydroxy-urs-12-en-28-oic acid) with IC 50 values of 5.66 μM (H157), 9.36 μM (MCF-7), 9.5 μM (HepG2), and 12.8 μM (A549). The oleanane-type triterpenoid arabinoglycosides with a diacetylated sugar unit displayed cytotoxicity against A549 and H157 cell lines with IC 50 values below 10 μM. The compounds with a free or monoacetylated sugar moiety demonstrated cytotoxic activity with IC 50 values above 20 μM . Another oleanane-type triterpenoid saponin isolated from D. asper roots (3- O -[ β - d -xylopyranosyl-(1→4)- β - d -glucopyranosyl-(1→4)][α- l -rhamnopyranosyl-(1→3)]- β - d -glucopyranosyl-(1→3)-α- l -rhamnopyranosyl-(1→2)-α- l -arabinopyranosylhederagenin) displayed cytotoxicity against two lung cancer cells lines, A549 and H157, with IC 50 values of 6.94 and 9.06 μM, respectively . The 16 kDa water-soluble polysaccharide (ADAPW) isolated from D. asperoides roots had the ability to inhibit the growth of human osteosarcoma cell line HOS and induce apoptosis in a concentration-dependent manner (100, 200, and 400 µg/mL) after 24 h. It was also found to down-regulate PI3K and pAkt protein levels, reduce mitochondrial membrane potential, and increase intracellular ROS level . However, a number of iridoid glycosides (dipsanosides C-G, 3′- O - β - d -glucopyranosyl sweroside, loganin, cantleyoside, triplostoside A, lisianthioside, and 6′- O - β - d -apiofuranosyl sweroside) had no cytotoxic effect on a set of tested cell lines, including lung carcinoma A549, hepatoma Bel7402, gastric carcinoma BGC-823, colon cancer HCT-8, and ovary cancer A2780 . Similarly, 7- O -( E - p -coumaroyl)-sylvestroside I isolated from the whole plants of S. stellata also was not cytotoxic (IC 50 > 100 μg/mL) to fibrosarcoma HT1080 cells. However, 7- O -( E -caffeoyl)-sylvestroside I showed moderate activity, with an IC 50 value of 35.9 μg/mL . Taken together, these above results indicated that some Dipsacus and Scabiosa plants or some specialized metabolites may display anticancer activity and may be useful as chemopreventive agents. 6.11. Antimicrobial and Anti-Insecticidal Activity An increase in bacterial resistance to antibiotics has caused researchers to look for alternative solutions, which may be natural antibiotics . Recent studies confirmed that extracts or essential oils from Dipsacus or Scabiosa spp. such as D. asper , D. fullonum , D. japonicus , S. stellata , S. arenaria , or S. atropurpurea subsp. maritima have antimicrobial activity . Traditionally, D. fullonum is known as the remedy for Lyme disease caused by Borrelia burgdorferi whose vectors are ticks. The anti- Borrelia activity of D. fullonum / D. sylvestris extracts were evaluated in only a few studies in the recent ten years . A 70% ethanol extract of D. fullonum leaves and its fractions showed significant anti- Borrelia activity against the stationary phase of B. burgdorferi strain B31 . The strongest growth inhibition was found for a crude ethanol extract, which suppressed the cell viability by about 80% at a concentration of 305.5 mg/L. The NP5 fraction, containing loganic acid, and NP7, rich in saponarin, isoorientin, and two chlorogenic acid derivatives, were also effective, with a residual viability of 23.4–29.8% at a concentration of 332.8 mg/L and 340.2 mg/L, respectively; these values were comparable to that of the positive control, the triple antibiotic combination (doxycylin, cefoperazone, and daptomycin at a dose of 22.2 mg/L, 33.4 mg/L, and 80.1 mg/L, respectively) . In contrast, Feng et al. found that the 45% ethanolic extract of D. fullonum (accidentally mixed with a sample of D. asper ) at a concentration of 0.25–1% was not active against either the non-growing stationary phase or growing B. burgdorferi , with residual viability of 84–90% and MIC > 2%. Among three tested extracts (70% ethanolic, ethyl acetate, and dichloromethane extracts) from D. sylvestris roots, only the ethanol extract was inactive against B. burgdorferi while the ethyl acetate extract showed the strongest ability . A 50% methanolic extracts of D. fullonum leaves and roots were also tested against other microorganisms, including bacteria ( Bacillus subtilis B5, Escherichia coli ATCC 10536, Pseudomonas aeruginosa DSM 939 , P. fluorescens W1, and Staphylococcus aureus DSM 799) and yeasts ( Candida famata AII4b, C. tropicalis ATCC 60557, C. sphaerica FII7A, Saccharomyces cerevisiae SV30, and Yarrowia lipolytica PII6a). It was found that the cell growth inhibitory activity differed among plant materials and bacteria strains. The greatest effect of growth inhibition zones was observed for the root extract against E. coli ATCC 10536 and S. aureus DSM 799 . The antibacterial potential was also found for S. arenaria . Various degree of antibacterial activity was related to the type of plant material (stems and leaves, roots, flowers, and fruits) and solvent used (crude extract and its fractions such as ethyl acetate, n -butanol, and aqueous). It was found that the highest antibacterial effect was noted for the n -butanolic fraction of fruits. In this case, MIC values for two Escherichia coli strains and two Pseudomonas aeruginosa strains were 0.019 mg/mL and 0.156 mg/mL, respectively. The butanolic fractions of the aerial parts and flowers were active against only E. coli strains with MIC values of 0.078 mg/mL and 0.156 mg/mL, respectively. In addition, Staphylococcus aureus ATCC 25923 and S. saprophyticus were sensitive to the butanolic fraction of fruits (MIC = 0.625 mg/mL). Among four tested strains of Candida spp. ( C. albicans ATCC 90028, C. glabrata ATCC 90030, C. parapsilosis ATCC 22019, and C. krusei ATCC 6258), the most sensitive was C. albicans ATCC 90028 with MIC = 0.0195 mg/mL. E. coli ATCC 25922 and C. albicans ATCC 90028 were also sensitive to eleven subfractions from the butanolic fraction of the aerial part (MIC = 0.0195 mg/mL) . A 70% ethanol extract of the whole plant S. stellata showed the highest antibacterial activity against Streptococcus pyogenes with MIC = 1.2 mg/mL (in comparison to gentamicin MIC = 2 µg/mL). For other strains of Gram-positive bacteria ( Bacillus subtilis , Enterococcus faecalis ATCC 1034, Staphylococcus aureus 8325-4, S. aureus CIP 53.154, S. epidermidis , Micrococcus luteus , and Listeria innocua ), Gram-negative bacteria ( Escherichia coli CIP 54.127, Enterobacter cloacae , Salmonella enterica , Serratia marcescens , Proteus vulgaris , Klebsiella pneumoniae , Providencia stuartii , Pseudomonas aeruginosa ATCC 9027, and Shigella sonnei ) and five yeasts ( Candida albicans , C. glabrata , C. tropicalis , C. kefyr , and Cryptococcus neoformans ), MIC ranged from 2.5 mg/mL to above 10 mg/mL . The highest antimicrobial activity was found for fractions B and C obtained after eluting from a Diaion HP-20 column with 25% and 50% methanol. Staphylococcus spp., Candida spp. ( C. albicans , C. tropicalis , and C. kefyr ), and Cryptococcus neoformans were the most sensitive microorganisms to both fractions with MIC values of 0.6–1.5 mg/mL, while E. faecalis ATCC 1034 , M. luteus , and S. pyogenes were also sensitive to fraction B . The ethyl acetate, n -butanol, and the petroleum ether extracts from S. stellata whole plants were also tested for antibacterial activity in the agar disk diffusion assay against ten bacterial strains including four Gram-positive ( Staphylococcus aureus ATCC 25923, S. albus , Enterococcus spp., and Streptococcus D) and six Gram-negative bacteria ( Escherichia coli ATCC 35218 , Pseudomonas aeruginosa ATCC 15442 , Acinetobacter baumannii, Proteus mirabilis, Salmonella typhimurium, and Enterobacter sakazaki ) . Three bacterial strains, S. albus , P. aregionosa ATCC 15442, and S. typhimurium , were the most resistant strains to all extracts. The highest activity was exhibited by the ethyl acetate extract against the clinical strain of P. mirabilis (16–20 mm of the growth inhibition zones at a concentration of 0.0625–1 mg/mL). In addition, this extract also was active against five other bacterial strains, including S. aureus ATCC 25923, A. baumannii , E. coli ATCC 35218, Enterococcus sp., and Streptococcus D. The petroleum ether extract showed inhibitory activity against S. aureus (ATCC 25923) and E. coli (ATCC 35218) while the n -buthanol extract against A. baumannii and E. sakazaki . In another study, the antibacterial and antifungal activities of the silver nanoparticles with S. atropurpurea subsp. maritima water extract from fruit against bacteria ( Escherichia coli , Micrococcus luteus , Staphylococcus aureus , and Klebsiella pneumoniae ) and fungal pathogens including Candida clinical strains ( C. albicans , C. tropicalis , and C. glabrata ), Microsporum canis , Trichophytom rubrum, and Trichophytom interdigitale were also reported. The silver nanoparticles inhibited the cell growth of bacteria and Candida sp., as evidenced by the zone inhibition (19.3–28 mm) and the MIC value (3.9–15.62 µg/mL). The lowest MIC value was found for two dermatophyte species, T. rubrum and T. interdigitale . In addition, the antifungal potential of the silver nanoparticles was associated with the disruption of membrane integrity and attenuation of the biofilm and hyphae formation . D. asper crude extract from the roots also displayed antifungal activity in vivo in a whole-plant assay. This property was evaluated against seven plant pathogenic fungi such as Magnaporthe grisea causing rice blast, Rhizoctonia solani causing rice sheath blight, Botrytis cinerea causing tomato gray mold, Phytophthora infestans causing tomato late blight, Puccinia recondita causing wheat leaf rust, Blumeria graminis f. sp. hordei causing barley powdery mildew, and Colletotrichum coccodes causing red pepper anthracnose. It was shown that the activity was dependent on the fungal pathogens and the solvent used ( n -hexane, ethyl acetate, acetone, methylene chloride, and methanol) for extraction. The fungi causing the tomato late blight and the tomato gray mold were the most sensitive to Dipsacus root extract. The greatest anti-fungal effect was demonstrated by the ethyl acetate and acetone extracts at a concentration of 1–2 mg/mL that inhibited tomato diseases by 90% . Antifungal activity was also demonstrated by the pure compounds isolated from the roots of D. asper such as cauloside A (the main compound of the extract). Cauloside A was most effective against fungal pathogens causing the tomato late blight, the rice blast, and the tomato gray mold at a dose of 0.5 mg/mL. Colchiside inhibited the growth of Phytophthora infestans while three sterols (campesterol, β-sitosterol, and stigmasterol) displayed the weakest antifungal activity . Among twelve specialized metabolites isolated from the whole plant S. stellata , two iridoids, viz., 7- O -caffeoyl-sylvestroside I and 7 -O -( p -coumaroyl)-sylvestroside I, showed the highest antimicrobial activity with an MIC value of 31.2 µg/mL against Enterococcus faecalis ATCC 1054 and Staphyllococcus epidermis ; sylvestroside I was also able to inhibit the growth of E. coli CIP 54.127 (MIC = 62.5 µg/mL). These iridoids also inhibited the growth of S. aureus CIP 53.154 with an MIC value of 62.5 µg/mL . 2′,4′- O -diacetyl-3- O - α - l -arabinopyranosyl-23-hydroxyolea-12-en-28-oic acid and hederagonic acid, isolated from D. asper roots, inhibited the growth of S. aureus ATCC 25923 with IC 50 values of 12.3 and 10.3 µM, respectively. Furthermore, 2 α ,3 β ,24-trihydroxy-23-norurs-12-en-28-oic acid and 2 α ,3 β -dihydroxy-23-norurs-4(24),11,13(18)-trien-28-oic acid also exhibited antimicrobial activity but the IC 50 value was three-times higher . The other triterpenoid derivative, oleanolic acid, found in some Dipsacus and Scabiosa species, showed weaker antibacterial properties against E. coli ATCC 25922, Pseudomonas aeruginosa ATCC 27853, and Candida albicans ATCC 90028 with an IC 50 value ranging from 170 µM to 680 µM . It is well known that essential oils and their ingredients have potent antimicrobial properties . The essential oil isolated from flowers of S. arenaria showed a strong ability (stronger than the positive control, thymol; MIC = 0.2 mg/mL) to inhibit the growth of cells of two Staphylococcus aureus strains with an MIC = 0.1562 mg/mL . Notably, the essential oil isolated from fruits was found to be an anticandidal agent against Candida albicans ATCC 90028, C. parapsilosis ATCC 27853, C. kreusei ATCC 6258, and C. glabrata ATCC 90030 (MIC = 0.625 mg/mL) . The essential oil isolated from D. japonicus flowering aerial parts can be used as a promising, natural insecticidal agent against stored-product insects such as adult red flour beetles ( Tribolium castaneum ) and maize weevils ( Sitophilus zeamais ), and they displayed contact toxicity with LD 50 values of 13.45 μg/adult and 18.32 μg/adult, respectively. This essential oil also had fumigant activity against adult insects with LC 50 5.26 mg/l air for T. castaneum and 10.11 mg/l air for S. zeamais . The strongest fumigant toxicity was possessed by one of the abundant ingredients in D. japonicus essential oil, i.e., 1,8-cineole . Regarding antiviral activity, only one study showed that dipsalignan A, -1-hydroxy-2,6-bis- epi -pinoresinol, and dipsanosides M-N displayed inhibitory activities against human immunodeficiency virus-1 (HIV-1) integrase. The IC 50 values were 53.26 μM, 61.74 μM, 84.03 μM, and 92.67 μM, respectively. The positive control, baicalein, had a value of 1.37 μM . 6.12. Others Akebia saponin D was also found to be a potential antidepressant agent. Intraperitoneal injection (40 mg/kg/d) alleviated LPS-induced microglia-mediated neuroinflammatory response in mice by inhibiting the TLR4/NF-κB signaling pathway in the hippocampus and prefrontal cortex . It also ameliorated chronic mild stress-induced depressive-like behaviors in C57BL/6 mice by inducing a neuroprotective microglial phenotype in the hippocampus through the PPAR-γ pathway . A similar antidepressant effect was also found for apigenin in a mouse model of chronic mild stress . Gong et al. found akebia saponin D to be effective against pain. It displayed an anti-nociceptive effect in SPF KM mice by shortening the licking time in the formalin test, increasing the reaction time to heat stimuli, and inhibiting acetic acid-induced writhing in mice. Akebia saponin D activated the expression of the progesterone receptor in primary decidual cells and the Notch signaling pathway. Gao et al. proposed that Dipsaci radix and its main ingredient, akebia saponin D, may promote decidualization in pregnant women. Bushen Antai, a Chinese herbal medicine preparation containing Dipsaci radix , was found to reduce the pregnancy loss caused by mifepristone administration . This preparation may stimulate estrogen and progesterone receptors through Akt and Erk1/2 signaling pathways in the maternal–fetal interface of pregnant rats. The roots of D. asperoides , known in Chinese as Xu Duan, means “connects broken bones” . Until now, numerous in vitro and in vivo studies have demonstrated that Dipsaci radix or the pure compounds isolated from this plant material can be potential agents for promoting osteoblast formation and may have an anabolic systemic skeletal effect. Dipsaci radix can improve bone density and affect bone histomorphology. Dipsaci radix has also shown osteoprotective properties in ovariectomized animals . In healthy BALB/c mice, an increase in the bone trabeculae density, bone volume/tissue volume ratio, bone surface/tissue volume, and trabecular number and the depletion in the trabecular separation on the proximal tibia after drinking water extract of Dipsaci radix were observed . Osteoporosis is characterized by an increase in bone fragility. Osteoporosis can usually occur with aging and after menopause due to estrogen deficiency, which contributes to the reduction of bone density and bone mass and degradation of the microstructure . The hormone replacement therapy that is used in osteoporosis treatment increases bone density. However, the use of hormone replacement therapy for a long time should be limited due to the serious side effects of its use . Modern therapies enhance bone metabolism by promoting osteoblast activity and by inhibiting the effects of osteoclasts . The ethanol extract of D. asperoides roots showed concentration-dependent progestogenic activity (40–100 μg/mL) in the T47D progesterone receptor-positive human mammary adenocarcinoma cell line; 100 μg/mL of extract demonstrated the equivalent of 31.45 ng/mL of progesterone treatment. These results indicate that D. asperoides roots can be used as an option for progestins . Moreover, asperosaponin D may promote the osteogenic differentiation of human mesenchymal stem cells through the estrogen signaling pathway . Liu et al. reported that Dipsaci radix decoction has anti-osteoporosis properties in ovariectomized Wistar or Sprague-Dawley rats (oral treatment at a dose of 100–500 mg/kg b.w./day) by increasing trabecular bone formation and bone mineral density and preventing bone mass loss and trabecular structure changes; it also decreases the serum alkaline phosphatase (ALP) level and the level of bone turnover markers, e.g., serum osteocalcin and urinary deoxypyridinoline/creatinine ratio, and receptor activator for nuclear factor κB ligand (RANKL) in osteoblasts and bone marrow stromal cells of the tibia. A recent study showed that Dipsaci radix may be able to control osteoblast differentiation, osteoclast proliferation, and mineralization via regulating mitogen-activated protein kinases (MAPK), nuclear-kappa B factor (NF-κb), TNF-α, and Toll-like receptor (TLR4) signaling pathways . Intragastric treatment of bilaterally ovariectomized Wistar rats with wine processed Dipsaci radix (with a dose of 75.6 g/kg/day) resulted in protection from an increase in urine Ca/creatinine and P/creatinine levels and serum ALP and osteocalcin concentrations and increased the femur bone mineral density. These effects were comparable to that observed in rats treated with 17 β -estradiol . In traditional Chinese medicine, Xian-Ling-Gu-Bao capsules have been used to prevent and treat osteoporosis, osteoarthritis, aseptic bone necrosis, or climacteric syndrome. Xian-Ling-Gu-Bao was officially approved in 2002 by the China Food and Drug Administration as an over-the-counter drug for the treatment of osteoporosis . This product is composed of the raw material of six plant species: Epimedii herba (70%), Dipsaci asperoidis radix (10%), Anemarrhenae rhizoma (5%), Psoraleae fructus (5%), Rehmanniae radix (5%), and Salviae miltiorrhizae radix (5%) . Xian-Ling-Gu-Bao administered to ovariectomized C57/BL6 mice for six weeks displayed anti-osteoporosis effects by enhancing bone mineral density and bone strength and by decreasing the serum level of the bone formation marker procollagen type I N-terminal propeptide (PINP) and the bone resorption marker C-terminal telopeptide of type I collagen (CTX) . Wu et al. also reported that Xian-Ling-Gu-Bao showed the ability to prevent osteoporosis in two osteoporosis models, prednisolone-treated zebrafish ( Danio rerio ) and ovariectomized Sprague-Dawley rats. Xian-Ling-Gu-Bao altered the protein levels of osteoprotegerin and RANKL. An increase in the OPG/RANKL ratio may suppress bone loss. It is worth mentioning that a dose of 1800 mg/kg (a concentration of six-times the recommended daily dose) did not cause toxicity or adverse effects in the heart, kidney, liver, stomach, or small intestine . Another formulation containing Dipsaci radix used in traditional Chinese medicine for treating kidney disease, osteoporosis, strengthening bones, and bone fractures is Du-Zhong-Wan. This preparation is composed of Eucommiae cortex (salted bark of Eucommia ulmoides Oliv.) and Dipsaci radix in an equal weight ratio (1:1) . It was shown that Du-Zhong-Wan displayed anti-osteoporotic activity in the Sprague-Dawley rat osteopenia model. After treatment with Du-Zhong-Wan at a dose of 2–6 g/kg/day for 12 weeks, an increase in the bone mineral density of the femur was observed, together with an improvement of trabecular bone mass and microarchitecture, reduction of the bone resorption and tartrate-resistant acid phosphatase 5b (TRACP-5b) level, and higher serum level of osteocalein, and serum and endometrium level of estrogen . Tian et al. reported that Du-Zhong-Wan favored fractured callus formation and improved osteoblastogenesis and angiogenesis by increasing the level of the H-type vessel endothelium markers (CD31 and endomucin) and proangiogenic factor SLIT3 in C57BL/6 mice after ovariectomy with the open transverse femoral fracture. Niu et al. showed that the 60% ethanolic extract and total saponins of Dipsaci radix at a concentration of 50–500 mg/kg/day may have anti-osteoporotic properties in Sprague-Dawley female rats after bilateral ovariectomy or hindlimb unloading Sprague-Dawley rat model by preventing a decrease in bone mass and by improving bone mineral density, biomechanical strength, and trabecular bone architecture. In addition, Dipsaci radix suppressed osteoclastogenesis through a reduction of the serum level of bone turnover marker (osteocalcin) and urine concentration of phosphorus, calcium, and deoxypyridinoline/creatinine ratio. In vitro studies demonstrated that Dipsaci radix stimulated osteoblastic proliferation, maturation, and differentiation via bone morphogenetic protein-2 (BMP-2)/MAPK/Smad1/5/8-dependent Runt-related transcription factor 2 (Runx2) signaling pathway. Dipsaci radix inhibited osteoclastogenesis by an increase in OPG/RANKL ratio in MC3T3-E1 murine preosteoblasts and primary osteoblastic cells . The ethanol extract of D. asper enhanced osteogenic differentiation of periodontal ligament stem cells through activation of the vascular endothelial growth factor (VEGF)/PI3K/Akt pathway. In addition, periodontal ligament stem cells displayed greater mineralization and mRNA expression of osteogenesis-related genes such as Col-1, ALP, Runx2, and osteocalcin . Treatment with the triterpenoid akebia saponin D also inhibited the osteoclastic gene RANKL in bone marrow mesenchymal stem cells, suppressing osteoclastogenesis and promoting osteogenesis. In addition, the combination of akebia saponin D and BMP-2 immobilized in 2-N, 6-O-sulfated chitosan alleviated osteoclastic formation, enhanced osteogenesis, and promoted angiogenesis by stimulating SMADs, TGF-β1, VEGFA, and OPG/RANKL signaling pathways . Moreover, loganic acid enhanced osteoblastic differentiation in preosteoblast MC3T3-E1 cells and suppressed osteoclast differentiation of primary-cultured monocytes derived from mouse bone marrow . Akebia saponin D was also found to have a potential application in osteoarthritis therapy. Gu et al. demonstrated that it showed anti-inflammatory activity in mouse primary chondrocytes and alleviated osteoarthritis following the surgical destabilization of a medial meniscus model of osteoarthritis in C57BL/6 male wild-type mice. Akebia saponin D was able to inhibit the production of cyclooxygenase-2 (COX-2), inducible nitric oxide synthase (iNOS), nitric oxide (NO), prostaglandin E2 (PGE2), IL-6, TNF-α, and NF-κB; it also activated the nuclear factor erythroid-2-related factor 2 (Nrf2)/Heme oxygenase 1 (HO-1) pathway and suppressed the expression of matrix synthesis degradation-related proteins such as disintegrin, metalloproteinase with thrombospondin motifs 5 (ADAMTS-5), as well as MMP13 in IL-1β treated chondrocytes. It was also found to enhance the expression of Aggrecan and Collagen II . A recent study found it to also enhance the proliferation and differentiation of human mesenchymal stem cells into nucleus pulposus-like cells through p-ERK1/2 and p-Smad2/3 activation, which may prevent intervertebral disc degeneration . Akebia saponin D also favored the proliferation and osteogenic differentiation of rat bone marrow stromal cells through the phosphatidylinositol-3 kinase/AKT signaling pathway, thus inhibiting osteoporosis. This active compound elevated osteogenic differentiation markers such as ALP activity and calcium deposit formation and mRNA level of osteogenic-related genes (ALP, osteocalcin, type 1 collagen (COL 1), and RUNX2) . Another terpenoid compound, hederagenin, also alleviated the progression of osteoarthritis and inhibited inflammation and cartilage degradation. In an in vitro study, hederagenin exerted chondroprotective and anti-inflammatory effects by suppressing the JAK2/STAT3/MAPK pathway, inhibited extracellular matrix degradation, elevated Aggrecan and Collagen II levels, and reduced levels of MMPs and ADAMTS5. In addition, this pentacyclic triterpenoid saponin inhibited cartilage destruction in rats induced by monosodium iodoacetate . Hederagenin 3- O -(2- O -acetyl)-α- l -arabinopyranoside and the dichloromethane fraction of Dipsaci radix also improved osteoblastic differentiation of human alveolar bone marrow-derived mesenchymal stem cells. This specialized metabolite caused the formation of calcified nodules and enhanced the level of bone differentiation protein expression such as sialoprotein and osteocalcin similar to dexamethasone . Moreover, the main iridoid glycoside isolated from Dipsaci radix , sweroside (at a concentration of 1 µM), enhanced rat osteoblast-like UMR 106 cell proliferation, while loganic acid, loganin, and sweroside favored mineralization. Loganin and sweroside suppressed the formation of adipocytes in 3T3-L1 cells . The anti-osteoporotic property of sweroside was also confirmed by Wu et al. . This compound enhanced mineralization of MC3T3-E1 cells by increasing the protein expression of the membrane estrogen receptor-α and G protein-coupled receptor 30 (GPR30), which activate the p38 signaling pathway. MC3T3-E1 mouse embryonic osteoblast proliferation and differentiation were also promoted by various concentrations (25, 50, or 100 μg/mL) of a 17 kDa polysaccharide (DAI-1) isolated from D. asperoides in high glucose concentrations; this appeared to act via the stimulation of the bone morphogenetic protein 2 (BMP-2)/Smad/runt-related transcription factor 2 (Runx2)/Osterix signaling pathway. The polysaccharide also enhanced osteocalcin level and the mRNA and protein levels of BMP-2 and Runx2 . A homogenous polysaccharide with a molecular weight of 26.1 kDa isolated from the roots of D. asper (at a dose of 50 or 200 mg/kg b.w.) was also able to enhance mRNA and protein levels of VEGF and osteoprotegin, suppress mRNA and protein expression levels of RANKL, and activate the PI3K/Akt/eNOS signaling pathway in ovariectomized rats . The water-soluble polysaccharide (ADAPW), with an average molecular weight of 16 kDa, inhibited the viability of the human osteosarcoma HOS cells by induction of apoptosis cells and inhibition of the PI3K/Akt signaling pathway . An increase in bone resorption may also lead to osteo- or rheumatoid arthritis. Rheumatoid arthritis is a chronic autoimmune and inflammatory disease of the connective tissue . Dipsaci radix has been also applied for many years in traditional Chinese medicine to treat other bone diseases such as rheumatic arthritis . The aqueous extract of D. asperoides roots at a concentration of 50 mg/kg and 100 mg/kg, administered orally once a day for 21 days, displayed antiarthritic effects in collagen-induced rheumatoid arthritis in male DBA/1 mice by enhancement of the ankle joint architecture and suppression of arthritis score (synovitis, pannus, and bone erosion scores) and serum levels of anti-CII IgG2a antibody and the inflammatory mediators (TNF-α, IL-1β, and IL-6). These effects were comparable or stronger to those after treatment with 1 mg/kg of the anti-rheumatoid drug indomethacin . Akebia saponin D at a concentration of 10 or 20 mg/kg/day also showed anti-osteoclastogenic activity in the arthritic joints of male BALB/c and DBA/1 with collagen-induced arthritis. The saponin reduced mRNA expression of osteoclastogenesis markers such as TRAP, CtsK, MMP-9, and β3-integrin. In addition, akebia saponin D also inhibited phosphorylation of Akt, p38, and JNK and mRNA and protein levels of osteoclastogenesis markers in RANKL-induced osteoclastogenesis bone marrow-derived monocytes . Dipsaus saponins also inhibited chondrocyte apoptosis in a rat model of osteoarthritis in a dose-dependent pattern by decreasing expression of Bax, caspase-3, and caspase-9 and by increasing expression of Bcl-2 . Cantleyoside, an iridoid identified in D. asper roots, inhibited proliferation of human rheumatoid arthritis fibroblast synovial cells (HFLS-RA) and induced cell apoptosis through AMPK/Sirt 1/NF-κB pathway activation . Moreover, the protective effect of sweroside was observed in IL-1β-induced inflammation in rat articular chondrocytes. The anti-inflammatory effect of this iridoid was mediated by the inhibition of NF-κB and mTORC1 signaling pathways . These above findings suggest that Dipsaci radix may have a beneficial therapeutic effect in the treatment of postmenopausal osteoporosis and may protect against arthritis. Alzheimer’s disease is a neurodegenerative disease that destroys memory and deteriorates cognitive function . Five-month administration of 4 g/kg ethanol extract of D. asper roots improved neurocognitive dysfunction in the passive avoidance task and diminished expression of hippocampal β-amyloid protein (Aβ) positive cells in aluminum chloride-treated male Sprague-Dawley rats. This effect increased with the time of treatment (1–5 months). It may be important in the treatment of Alzheimer’s disease and memory system dysfunction . Akebia saponin D, at a dose of 30–270 mg/kg administered for four weeks, also showed a preventive effect against memory cognitive impairment in ibotenic acid-exposed male Sprague-Dawley rats . Moreover, it was shown that the saponin can protect against learning and memory dysfunction in rats induced by bilateral intracerebroventricular injections of Aβ1–42 in the Y-maze and Morris water-maze tests. Akebia saponin D attenuated the activation of microglia and astroglia and protein expression level of IL-1, COX-2, TNF-α, NF-κB, and Akt phosphorylation in the rat brain . The total saponin of D. asperoides also possessed neuroprotective properties in rat cortical and hippocampal neurons against damage induced with β-amyloid protein, ameliorated cell viability, and decreased lactate dehydrogenase (LDH) release and lipid peroxidation at a concentration of 150–300 mg/L . Other triterpene compounds, such as oleanolic acid, ursolic acid, and hederagenin, also showed neuroprotective activity in multiple brain disorders . For example, hederagenin diminished Aβ deposition in the head area of Caenorhabditis elegans , ameliorated cognitive impairment or pathological changes in APP/PS1 mice, and induced PPARα/TFEB-dependent autophagy of BV2 cells . Moreover, treatment with sweroside, a secoiridoid glycoside, alleviated memory deficits in scopolamine-induced Zebrafish ( Danio rerio ) in behavioral tests such as the tank diving test, the Y-maze, and the novel object recognition test . The flavonoid apigenin also suppressed neurotoxicity and cognitive function in LPS-induced mice. The compound protected against neuronal degenerative changes in mice hippocampi . It was also found that 50% methanol extracts from leaves and roots of D. fullonum demonstrated anti-acetylcholinesterase activity . The ethyl acetate and n -butanolic fractions of 70% methanol extract from S. stellata whole plants, at a concentration of 200 μg/mL, exhibited moderate (30.8%) and low (10.9%) acetylcholinesterase (AChE) inhibitory activity, respectively . On the other hand, the ethyl acetate and n -butanolic fractions of stems and leaves from S. arenaria (at a dose of 1 mg/mL) were able to inhibit AChE in 97.61% and 90.47%, respectively. The value of IC 50 ranged from 0.016 mg/mL to 0.029 mg/mL. In comparison to eserine (positive control), the IC 50 value was 0.0029 µg/mL . The root extract also displayed potent activity. The strongest effect was demonstrated by the n -butanolic fraction (87.61% AChE inhibition and IC 50 value = 0.02 mg/mL) . Yu et al. showed the effectiveness of some compounds isolated from the roots of D. asper ; for example, oleanane triterpenoid saponin (3- O - β - d -glucopyranosyl-(1→3)-α- l -rhamnopyranosyl-(1→2)-α- l -arabinopyranosyl-23-hydroxyolean-18-en-28-oic acid 28- O - β - d -glucopyranosyl-(1→6)- β - d -glucopyranosyl ester) was found to inhibit acetylcholinesterase with an IC 50 value of 15.8 μM. Other terpenoid compounds isolated from Dipsaci radix , such as dipsacus saponin IV, dipsacus saponin XI, and dipsacus saponin X, displayed strong AChE inhibitory activity, while cauloside A, dipsacus saponin C, and dipsacus saponin XI were more effective against butyrylcholinesterase. These activities were higher than that of the positive control, berberine. Moreover, the saponins were found to be more effective than the iridoids, loganic acid and sweroside. The terpenoids also inhibited β-site amyloid precursor protein cleaving enzyme 1 (BACE1) and advanced glycation end-product (AGE) formation . Liver fibrosis is a chronic liver disease caused by many agents such as hepatitis B virus (HBV), hepatitis C virus (HCV), non-alcoholic steatohepatitis, or alcoholic fatty liver disease . The flavonoids and phenolic acids from Scabiosa spp. have demonstrated anti-hepatic fibrosis potential in male Wistar or male Sprague-Dawley rats treated intraperitoneally with carbon tetrachloride (CCl 4 ) (a selective hepatotoxic drug); similar effects were also noted for drugs used in Mongolian medicine, such as Qingganjiuwei and Gurigumu-7 (composed of various herbs including S. comosa flowers) . Qingganjiuwei, a drug commonly used in Inner Mongolia in patients with chronic hepatic disease, improved liver morphology and structure in CCl 4 -treated SD rats when administered at 1.575–4.725 g/kg/day for eight weeks. The drug reduced hepatocyte necrosis, lymphocytic infiltration, and pseudolobuli and lowered COL1, tissue inhibitor of metalloproteinase1 (TIMP1), and α-smooth muscle actin (α-SMA) expression. The drug also activated the MAPK pathway in the liver through the suppression of extracellular signal-regulated kinase (ERK), C-Jun amino-terminal kinases (JNKs), and stress-activated protein kinase-2 (p38 proteins) . Qingganjiuwei also increased mRNA and protein expressions of MMP2 and MMP9 and inhibited the levels of the serum aminotransferases (ALT and AST) . The methanol-eluted fraction of Gurigumu-7 extract (0.264 g/kg) displayed a more potent hepatoprotective effect in mice with CCl 4 -induced liver damage compared to crude Gurigumu-7 extract, even when applied at a four-times higher concentration (1.084 mg/kg). The methanol fraction alleviated histopathological changes in the liver and serum ALT, ASP, and liver malonyldialdehyde (MDA) levels and enhanced the liver superoxide dismutase (SOD) level in a dose-dependent manner (66, 132, and 264 mg/kg) . Anti-hepatic fibrosis was also reported for S. comosa and S. tschilliensis . Some specialized metabolites inhibited the viability of hepatic stellate LX-2 cells at concentrations of 12.5–200 μM; this included those belonging to flavonoids, which accounted for about 60% of the total identified compounds in the inflorescences of Scabiosa plants. The flavonoids enhanced the expression of Stat1, Pparg, Hsp90aa1 genes, signal transduction and transcriptional activator 1 (STAT1), and peroxisome proliferator-activated receptor G (PPARG) proteins, which play key roles in the pathogenesis of liver fibrosis . Apigenin exhibited the strongest ability to inhibit cell proliferation . The flavonoid-rich extract of S. comosa inflorescences at concentrations of 100 and 200 mg/kg also suppressed hepatic fibrosis in Wistar rats pre-treated with CCl 4 ; the extract inhibited the level of biochemical parameters in blood serum (ALT, AST, ALP, and hyaluronic acid), the markers of liver fibrosis (laminin, amino-terminal propeptide of type III procollagen (PIIINP), collagen IV, collagen deposition in the liver tissues), and expression of α-SMA, collagen I, and fibronectin . Moreover, the extract attenuated phosphorylation of Smad3 in liver tissue and TGF-β1-pre-treatment primary mouse hepatic stellate cells. In the latter, it was observed that the expression of the fibrotic genes (α-SMA, collagen I, and fibronectin) was suppressed in a dose-dependent manner . Ethanol extract of aerial parts of S. atropurpurea and hexane, ethyl acetate, n -butanolic, and chloroform fractions were able to decrease the levels of serum ALT, AST, and ALP in CCl 4 -induced liver damage in albino rats after treatment with a dose of 100 mg/kg . Sweroside demonstrated a protective effect against liver fibrosis in mouse models treated with CCl 4 and methionine-choline-deficient diet-induced non-alcoholic steatohepatitis . Sweroside treatment yielded an anti-fibrotic effect through the FXR-miR29a pathway . Sweroside also improved NASH symptoms by inhibiting the activation of the hepatic NLRP3 inflammasome . Apigenin was found to inhibit palmitic acid-induced pyroptosis by regulating the pyrin domain containing 3 (NLRP3) inflammasome in HepG2 cells and primary mouse hepatic cells . Other polyphenols, derivatives of caffeoylquinic acid such as 1,5-di- O -caffeoylquinic acid, isochlorogenic acid C, isochlorogenic acid B, chlorogenic acid, isochlorogenic acid A, neochlorogenic acid, and caffeic acid also suppressed LX-2 hepatic stellate cell growth . The two predominant caffeoylquinic acid derivatives in inflorescences of S. comosa and S. tschilliensis , viz., 3,5-di- O -caffeoylquinic acid and chlorogenic acid, demonstrated anti-hepatitis C virus (HCV) activity in the Huh-7.5 cell line infected with HCV . In addition, some triterpenoid derivatives isolated from the whole plants of S. tschiliensis , such as scabiosaponins E, F, G, I, and J; hookerosides A and B; and prosapogenin 1b, showed strong inhibition of pancreatic lipase. Moreover, 0.12 mg/mL prosapogenin 1b exerted similar inhibitory properties as the lipase inhibitor orlistat at a concentration of 0.005 mg/mL . Akebia saponin D reduced lipid droplet accumulation in BRL cells, attenuated hepatic steatosis, and elevated the expression of Bcl-2/adenovirus E1B 19-kDa interacting protein 3 (BNip3) and phospho-AMPP; it also improved mitochondrial function and autophagy modulation, inhibited rotenone-induced BRL cell apoptosis, elevated Bcl-2/Bax ratio, and suppressed the level of intracellular reactive oxygen species (ROS) and mitochondrial membrane potential loss in rotenone-treated BRL cells and rat liver mitochondria . Cardiovascular diseases are one of the leading causes of death worldwide . The anti-atherosclerotic effect of akebia saponin D was studied in vitro in H 2 O 2 -treated human umbilical vein endothelial cells (HUVECs) and in vivo in ApoE −/− mice . It was shown that this saponin protected against H 2 O 2 -induced cytotoxicity in HUVEC, inhibited ROS level, the mitochondrial membrane potential disruption, and apoptosis in oxidative stress-induced endothelial cells in a dose-dependent manner (50–200 µM); the mechanism was believed to involve increasing Bcl-2 family protein levels and decreasing caspase-3 and Bax activation. Moreover, doses of 150 mg/kg/day and 450 mg/kg/day reduced aortic plaque formation in mice as well as aortic and liver apoptosis, serum triglyceride (TG), total cholesterol (TC), low-density lipoprotein cholesterol (LDL-C), and lipid deposition in the liver, as well as atherosclerotic lesion size; it also enhanced the expression of antioxidant enzymes (SOD, catalase (CAT), and glutathione (GSH)) in vascular tissue and liver . Li et al. demonstrated that sweroside has a protective effect on ischemia-reperfusion-induced myocardial injury by inhibiting oxidative stress and pyroptosis partially via modulation of the Kelch-like ECH-associated protein 1 (Keap1)/Nrf2 pathway. The iridoid also suppressed aconitine-induced cardiac toxicity in the H9c2 cardiomyoblast cell line . Long-term gavage (for six weeks) of akebia saponin D protected against fibrosis myocardial ischemia injury, inhibited cardiac dysfunction, and reduced infarct size in a Sprague-Dawley rat model with chronic myocardial infarction induced by permanent ligation of the left coronary artery. Treatment with this saponin decreased hydroxyproline level and changed the activity of the oxidative stress enzymes by elevating SOD and DSH-peroxidase (GSH-Px) levels and by reducing MDA content. Moreover, akebia saponin D regulated inflammatory mediators by diminishing the levels of TNF-a and IL-6 and by elevating the level of IL-10 . Several in vitro and in vivo studies found flavonoids detected in Scabiosa spp., such as apigenin and luteolin, to have cardioprotective properties . However, Song et al. demonstrated that D. asper roots, drunk widely as a tea for beneficial health effects, or dipsacus saponin D may have an undesirable effect on platelets and may increase the risk of thrombosis. D. asper roots were found to be an herb with procoagulant activities on platelets and prothrombotic properties. Dipsacus saponin C elevated procoagulant activity in a dose- and time-dependent manner, elevated intracellular calcium level, and decreased ATP. In addition, it caused translocation of Bax and Bak, cytochrome c release, caspase-3 activation, and the disruption of mitochondrial membrane potential. The oral administration of 10 mg/kg and 25 mg/kg dipsacus saponin C also resulted in an increase in thrombus formation in a rat venous thrombosis model . A 61 kDa polysaccharide (WDRAP-1) isolated from D. asperoides roots showed protective activity against oxidative stress generated in renal ischemia-reperfusion injury in male Wistar rats and displayed strong superoxide and hydroxyl radical scavenging activities in vitro. It was shown that oral pre-treatment of rats with the polysaccharide at a concentration of 50–200 mg/kg b.w. for 14 days before ischemia-reperfusion may improve renal injury (especially at the highest dose); treatment suppressed the level of renal injury indicators including creatinine, blood urea nitrogen, lactate dehydrogenase (LDH), and serum MDA and enhanced serum SOD and some renal tissue antioxidant enzyme activities (SOD, GSH-Px, and CAT) . Hederagenin was also found to protect against renal fibrosis. This terpenoid compound attenuated the proliferation and fibrosis of TGF-β-treated NRK-49 F cells by targeting the muscarinic acetylcholine receptor . Dipsacus saponin C, a saponin isolated from the roots of D. asper , was found to have a protective effect against HCl·ethanol-induced gastritis and indomethacin-induced gastric ulcers in male Sprague-Dawley rats. It was shown that treatment with dipsacus saponin C caused a decrease in gastric secretion volume and gastric acid production in pylorus-ligated rats. It was found that this compound had a moderate effect on colonization and growth inhibition of Helicobacter pylori at a concentration of 50–100 µM. In addition, dipsacus saponin C also showed a cytotoxic effect for SNU638 and AGS human gastric cancer cells with an IC 50 at 54.6 mM and 37.3 mM, respectively . Moreover, akebia saponin D may exert a therapeutic role by regulating the intestinal microbiome and protecting intestinal epithelial cells from external damage. It was found to achieve this by inhibiting oxidative damage to the intestinal barrier by downregulating PPAR-γ/FABP4 in the human intestinal cell line FHs74 Int . Luteolin was also found to have therapeutic effects against interstitial fibrosis-induced renal anemia in vitro and in vivo. This activity was mediated via the SIRT1/forkhead box O3 (FOXO3) pathway . Dipsaci radix alleviated the asthmatic response in BALB/c mice with allergic asthma induced by an ovalbumin. Dipsaci radix treatment at a dose of 20 m/kg or 40 mg/kg resulted in attenuation of the methacholine response, inflammatory cell infiltration, and mucus secretion in the bronchial airway and a decrease in the levels of pro-inflammatory cytokines (IL-5 and IL-13) in bronchoalveolar lavage fluid, eotaxin, serum total IgE, expression of iNOS, and NF-κB phosphorylation in lungs . Similarly, apigenin, one of the flavonoids identified in Scabiosa spp., was found to inhibit inflammatory mediators and eosinophilia in lung and airway tissues in in vivo acute lung injury and asthma models . A polysaccharide (DAP) isolated from D. asper roots demonstrated beneficial effects on renal function and renal pathological changes as well as antihyperglycemic, hypolipidemic, and antioxidant activities in streptozotocin-induced diabetic Wistar rats . Four weeks of intragastric administration (100 mg/kg and 300 mg/kg per day) in type 2 diabetic rat model resulted in a reduction of glycosylated, fasting blood glucose, serum creatinine, blood urea nitrogen, urine protein, and urinary albumin excretion. Moreover, oral administration of the polysaccharide (300 mg/kg) suppressed the serum level of TC, TG, LDL, and renal AGE-RAGE formation (advanced glycation end products-receptor for advanced glycation end products) and enhanced SOD, CAT, and GSH activities in the kidney of rats with diabetic nephropathy . Similar antilipemic effects were also observed for hederagenin. This pentacyclic triterpene exerts its potential through the p38MAPK pathway in oleic acid-induced HepG2 cells and in hyperlipidemic Sprague-Dawley rats . The ethanol extract and the hexane, ethyl acetate, n -butanolic, and chloroform fractions of the aerial parts of S. atropurpurea demonstrated anti-hyperglycemic activity by decreasing the blood glucose level in albino rats with alloxan-induced hyperglycemia . The methanol extract of S. atropurpurea subsp. maritima whole plants demonstrated α-glucosidase inhibitory activity with an IC 50 value = 100 μg/mL. This effect was higher than that found for the positive control, the anti-diabetic drug acarbose (IC 50 = 196 μg/mL) . Methanolic extracts from the fresh leaves and roots of D. fullonum also inhibited porcine pancreatic α-amylase activity . These extracts demonstrated low effectiveness in this study, with the strongest activity being found to be IC 50 = 86.01 µg/mL for the dried leaf extract. This activity was more than 100-times lower compared to that of acarbose (IC 50 = 0.69 μg/mL). Several plant species of Dipsacus and Scabiosa or some specialized metabolites isolated from them can be valuable, new anti-inflammatory agents . The water extract of D. asper roots demonstrated anti-inflammatory properties in the lipopolysaccharide (LPS)-activated murine macrophage cell line RAW 264.7 by suppressing NO production with an IC 50 = 45.1 µg/mL . In a later study , an aqueous extract of D. asperoides roots at a dose of 50–500 µg/mL showed inhibitory potential on inflammation and oxidative stress in RAW 264.7 macrophages exposed to LPS; it was found to act by lowering NF-κB and ERK1/2 phosphorylation, nuclear translocation of NF-κB, and activation of Nrf2/HO-1. The extract reduced the levels of inflammatory mediators (iNOS, COX-2, and cytokines IL-6 and IL-1β) as well as ROS levels . The methanol extract of D. inermis leaves also showed the ability to inhibit the production of NO, COX-2, PGE2, pro-inflammatory mediators (IL-1β and IL-6, and TNF-α), intracellular ROS level, and phosphorylation of NF-κBp65 and IκBα in a dose-dependent manner (25–100 µg/mL) in the LPS-induced murine macrophage cell line J774A.1 . The anti-inflammatory potential of D. inermis leaf extract was also confirmed in vivo in Wistar albino rats; it protected against vascular permeability (caused by acetic acid) and paw oedema (induced by carrageenan) in a concentration- and time-dependent manner. In addition, the serum levels of TNF-α, IL-1β, and IL-6 were significantly reduced while IL-10 level was enhanced after oral administration of the extract at a concentration of 50–100 mg/kg b.w. . Anti-inflammatory properties in a Wistar rat model of carrageenan-induced paw oedema also demonstrated the ethyl acetate extract of S. stellata whole plants. The highest activity was observed in the first hour after treatment with the extract at a concentration of 50 mg/kg (72.73% of inhibition). This effect was stronger than that of diclofenac (about 45% of inhibition). In addition, the anti-inflammatory effect of the plant extract lasted up to 24 h . Some compounds belonging to iridoids, saponins, or phenolic acids (for example, dipsasperoside A, dipsanoside A and B, dipsacus saponin A, akebia saponin D, or caffeic acid) isolated from the roots of D. asper also were able to reduce the production of NO in RAW 264.7 cells. The potent activity was demonstrated by akebia saponin D and dipasperoside A, with IC 50 values of 12.7 µM and 15.2 µM, respectively; these values were higher than those for the positive control, a nonselective NOS inhibitor, N G -monomethyl-L-arginine (IC 50 = 22.6 µM) . Reduced NO levels and iNOS expression have also been observed in LPS-induced RAW 264.7 cells after treatment with akebia saponin D (at a concentration of 25–100 µM) . Akebia saponin D also suppressed the expression of DNA methyltransferase (DNMT) 3b, the levels of PGE2 and p-STAT3, as well as the protein and mRNA levels of IL-6 and TNF-α . The levels of the inflammatory indicators, prostaglandin E2, i-NOS, COX-2, TNF-α, IL-1β, and IL-6, were also decreased after treatment with 40 and 80 µM sweroside in LPS-induced RAW264.7 cells. In addition, sweroside suppressed inflammation through the sirtulin 1 (SIRT1)/NF-κB and SIRT1/Forkhead transcription factor O1 signaling pathways . Akebia saponin D also showed anti-inflammatory activity in vivo by reducing paw oedema in carrageenan-induced Sprague Dawley rats and by inhibiting xylene-induced ear swelling in mice. It also lowered the level of NO in rat plasma in a carrageenan-induced rat paw oedema model . Anti-inflammatory potential was also demonstrated by apigenin, which was believed to act by the modulation of the p38/MAPK, PI3K/Akt and NF-κB pathways . The most frequently used methods to determine the antioxidant properties of Dipsacus and Scabiosa plant extracts were the DPPH (2,2-diphenyl-1-picrylhydrazyl radical scavenging assay), ABTS (2,2′-azinobis(3-ethylbenzothiazoline-6-sulfonic acid) radical cation assay), ORAC (oxygen radical absorbance capacity), FRAP (ferric ion reducing antioxidant assay), and CUPRAC assays (cupric-reducing antioxidant capacity). The antioxidant activity varied depending on the plant material (whole plants, leaves, and roots), the solvent, extraction time, and the extraction method . The antioxidant activity of the methanolic extract from D. asper roots was confirmed in DPPH and Cu 2+ -mediated LDL oxidation with IC 50 values of 90.2 and 134.4 µg/mL, respectively. This activity may be attributed to caffeoylquinic acid derivatives identified in root extract such as 3,4-di- O -caffeoylquinic acid, 4,5-di- O -caffeoylquinic acid, 3,5-di- O -caffeoylquinic acid, and their methyl derivatives. These specialized metabolites showed potent antioxidant activity against DPPH radical formation and Cu 2+ -mediated LDL oxidation with IC 50 values of 10.4–18.2 µM and 1.8–2.3 µM, respectively . The content of the total polyphenols in the acetone/water extract (7:3) of D. fullonum whole plant was 19.52 mg GAE (gallic acid equivalents)/g d.w. of plant material; it displayed an antioxidant capacity lower than 5 mmol TEAC (Trolox equivalent antioxidant capacity)/100 g d.w. plant material in the DPPH and ABTS assays . The antioxidant properties were also demonstrated in 50% methanolic extract of leaves and roots (ultrasound assisted extraction) in the ORAC assay with values of 14.78 mmol TEAC/100 g d.w. and 10.87 mmol TEAC/100 g d.w. for the leaf and root extracts, respectively . A similar observation was found by Saar-Reismaa et al. . The crude 70% ethanol extract of D. fullonum leaves showed antioxidant activity in the ORAC assay (10.8 mmol TEAC/100 mL). The fraction NP7 of crude leaf extract that was rich in two chlorogenic acid derivatives, saponarin and isoorientin, also displayed antioxidant activity (12.5 mmol TEAC/100 mL), while the fraction NP2 containing bis -iridoids was ineffective (0.78 mmol TEAC/100 mL) . On the other hand, the aqueous extract of D. fullonum leaves obtained by ultrasound assisted extraction (extraction time, four hours) showed high radical scavenging activity (RSA) (73.81%) in DPPH . The procedure extraction with the aqueous solution of amino acid ionic liquids, viz., triethanolammonium salts of two amino acids methionine ([TEAH] + [Met] − ) and threonine ([TEAH] + [Thr] − ), resulted in a beneficial effect on the antioxidant activity of leaf extracts in FRAP and CUPRAC assays compared to the extract after extraction with the pure water. In addition, extraction with the aqueous solution of [TEAH] + [Thr] − (extraction time, two hours) increased the total polyphenol content to 8.16 mg GAE/g raw material. The use of [TEAH] + [Met] − in aqueous solution and reducing the extraction time to one hour resulted in a similar level of polyphenol (7.38 mg GAE per g of raw material) . Antioxidant effect was also observed for the extract of D. sativus leaf. The extract at a dose of 300 mg/kg/day in ICR mice pre-treated with D-galactose to induce oxidative stress resulted in an increase in the level of SOD and a decrease in the level of MDA in the peripheral blood plasma . S. stellata whole plants are a rich source of polyphenolic compounds. The ethyl acetate and n -butanolic fractions have moderate antioxidant activity in simple, chemical, antioxidant tests such as DPPH, ABTS assays, and FRAP test. The n -butanolic fraction displayed stronger potential in the DPPH assay, i.e., FRS 50 (free radical scavenge) = 64.46 µg/mL, compared to ascorbic acid FRS 50 = 8.21 µg/mL . The dichloromethane fraction did not possess significant activity in antioxidant assays, which may be related to the low content (below 1 mg of gallic acid/g dry extract) of phenolic compounds . A 70% ethanol extract of S. stellata whole plants in DPPH also displayed antioxidant activity, with an IC 50 value of 86 μg/mL . The petroleum ether, ethyl acetate, and n -butanolic fractions from S. stellata whole plants showed various antioxidant potential in different chemical models such as DPPH and ABTS, FRAP, CUPRAC, β-carotene, phosphomolybdate assay, ferrous ions, and metal chelating assays . The strongest effect was demonstrated by the n -butanolic fraction in DPPH (IC 50 = 21.22 μg/mL) and chelation in ferrous iron assay (EC 50 = 1.65 mg/mL), while the ethyl acetate fraction was most active in ABTS (IC 50 at 14 μg/mL), CUPRAC (A 0 . 50 = 28.5 µg/mL), and β-carotene assays (IC 50 = 10.34 μg/mL). In addition, the n -butanolic fraction in DPPH had a higher reducing power than butylated hydroxyl toluene (BHT) (22.32 µg/mL) but lower than α-tocopherol (13.02 µg/mL), butylated hydroxy anisole (BHA) (6.82 µg/mL), and ascorbic acid (3.1 µg/mL) . The highest protein denaturation inhibition was found for the ethyl acetate extract, which showed 78.86% inhibition at the maximal tested concentration (1 mg/mL). Ibuprofen (standard drug) at the same concentration caused 100% inhibition . Antioxidant activity was also reported for 70% ethanol extracts of S. comosa and S. tschilliensis inflorescences. S. tschilliensis had a stronger antioxidant activity than S. comosa in DPPH, ABTS, and FRAP assays. For example, the IC 50 values in DPPH were 272.8 µg/mL and 331.1 µg/mL, respectively . Wang et al. reported that the 90% ethanol extract of S. tschilliensis at a concentration of 26.5 µg/mL scavenged 50% of DPPH free radicals (IC 50 value for ascorbic acid was 5.41 µg/mL). The crude extract (95% ethanol) and four solvent partitioned fractions (water, n -butanol, ethyl acetate, and petroleum ether) from S. tschiliensis whole plants at various growing stages (pre-flowering, flowering, and fruiting stage) showed different antioxidant activities in DPPH, ABTS, inhibition of lipid peroxidation, or OH scavenging activity . The IC 50 values for the crude extract were in the range of 25.65–86.79 µg/mL while the ethyl acetate fraction from the pre-flowering stage of plants had the highest antioxidant capacity (IC 50 8.47 µg/mL) in DPPH. This value was comparable to that of vitamin C (7.6 µg/mL). The ethyl acetate fraction from the pre-flowering stage of plants also possessed the highest ABTS (58.76 µg/mL), hydroxyl radical scavenging ability (67.64 µg/mL), and lipid-peroxidation-inhibition activity . The n -butanolic and ethyl acetate fractions of S. arenaria roots also displayed strong antioxidant activity in four assays: DPPH, ABTS, reducing power, and β-carotene bleaching inhibition activity. The n -butanolic fraction demonstrated excellent ability, mainly in the β-carotene bleaching inhibition assay (IC 50 = 0.018 mg/mL). This effect was stronger than that obtained for BHT (IC 50 = 0.04 mg/mL). In the DPPH and ABTS assays, the IC 50 values for both fractions were comparable to BHT . The ethyl acetate fractions of the roots, flowers, fruits, and aerial parts (stems and leaves) of S. arenaria showed the beneficial antioxidant ability in DPPH, with IC 50 values of 0.017–0.019 mg/mL; the best properties were observed for the flowers . Other species of Scabiosa , S. artropurpurea and S. atropurpurea subsp. maritima , also demonstrated antioxidant properties. In comparison to ascorbic acid (IC 50 = 0.084 mg/mL), among four tested extracts, the ethanol extract of S. artropurpurea stems exhibited the best antioxidant capacity, with IC 50 = 0.1383 mg/mL in DPPH assay. In addition, the hexanoic volatile fraction (VF1) and ethyl acetate extract displayed similar effects with IC 50 values of 0.4798 mg/mL and 0.4806 mg/mL, respectively . The antioxidant activity of 70% ethanol extract of aerial parts of S. atropurpurea and hexane, ethyl acetate, n -butanol, and chloroform fractions were also demonstrated by increasing blood glutathione in diabetic albino rats . Silver nanoparticles with S. atropurpurea subsp. maritima water fruit extract also was found as a promising antioxidant agent in DPPH and FRAP assays. The IC 50 values were 0.112 mg/mL and 0.036 mg ascorbic acid equivalent antioxidant capacity/g d.w., respectively, and were comparable to that for ascorbic acid . The pure compounds isolated from D. asper roots, i.e., six dicaffeoylquinic acid derivatives (3,4-di- O -caffeoylquinic acid, methyl 3,4-di- O -caffeoyl quinate, 3,5-di- O -caffeoylquinic acid, methyl 3,5-di- O -caffeoyl quinate, 4,5-di- O -caffeoylquinic acid, and methyl 4,5-di- O -caffeoyl quinate) exhibited strong antioxidant capacity in the DPPH assay (10.4–18.2 µM) and displayed inhibitory activity against Cu 2+ -mediated LDL oxidation (1.8–2.3 µM), stronger than those obtained for the positive controls, BHT and caffeic acid . In another study, 3,5-di- O -caffeoylquinic acid also demonstrated significant antioxidant properties in the DPPH assay, with an IC 50 value of 3.63 µg/mL . Two secoiridoid glucosides, eustomoruside and eustomoside, and one flavonoid, isoorientin, isolated from the whole plant S. stellata Cav. also displayed strong radical scavenging activities in DPPH with an IC 50 value of 7.1–8.5 μg/mL compared to that for ascorbic acid (IC 50 = 6.3 μg/mL) . The polysaccharide fraction from the roots of D. asperoides also demonstrated antioxidant effects in DPPH and ABTS assays but with little potency; the EC 50 value was 0.355 mg/mL in DPPH free radical scavenging activity and 5.867 mg/mL in ABT . The synthetic drugs used in chemotherapy not only have strong toxic effects on cancer cells, but they also have strong adverse side effects in chemotherapy. Many plant specialized metabolites are used in cancer therapies and new substances, and new plant species with potential anti-cancer activity are being sought. Considerable attention has been noted regarding the cytotoxicity of Dipsacus and Scabiosa against various cancer cell lines including lung carcinoma A549, hepatoma Bel7402 and Hep3B, gastric carcinoma BGC-823, AGS, KATO III, MKN-45, and SNU-638, liver H157 and HepG2, colon cancer HCT-8, ovary cancer A2780, breast MCF-7, human breast cancer MCF-7 and MDB-MB-231, acute myeloid leukemia OCI-AML3, osteosarcoma HOS, or fibrosarcoma HT1080 cell lines. Some pure specialized metabolites of different classes identified in the plant extracts have also demonstrated potent or promising cytotoxic effects in vitro . An aqueous extract of D. asperoides roots inhibited the viability of human mammary carcinoma-derived triple negative MDA-MB-231 cells (with IC 50 = 15 μg/mL) and arrested the cell cycle in the G 2 /M phase; it also induced apoptosis by increasing pro-apoptotic caspase 3/7 activity and by suppressing the expression of BRAF, p-ERK, MEK, pPI3K, pAKT, and cyclin-dependent kinase 4/6 in a dose-dependent manner . The cytotoxic activity of bis -iridoid glycosides fraction of D. fullonum leaf methanol extract (with sylvestroside III and IV as the main compounds) was evaluated against human breast cancer cell lines MCF7 and MDB-MB-231 and human cervical cancer HeLa cell line . The two breast cancer cell lines were most sensitive to the fraction, resulting in a viability of 64.0% for MCF7 cells and 69.5% for MDB-MD-231 cells . In addition, the ethanolic extracts of the aerial parts and flowers of D. fullonum have antiproliferative activity on the human hepatocellular carcinoma Hep3B cell line with an IC 50 value above 100 µg/mL . The methanol extract of S. atropurpurea subsp. maritima leaves at a concentration of IC 10 , IC 20 , or IC 30 enhanced the toxicity of doxorubicin in human epithelial colorectal adenocarcinoma Caco-2 cells with IC 50 = 1.04 µg/mL (vs. 2.41 µg/mL when the cells were treated only with doxorubicin) . In addition, the combination of doxorubicin with S. atropurpurea extracts at a concentration of IC 50 and IC 10 , respectively, increased the percentage of apoptotic cells, the percentage of caspase-activated cells, mRNA levels of the apoptosis related-genes (Bax, caspase-3, p21), and decreased the expression level of anti-apoptotic genes (Bcl-2). It was a stronger effect than that obtained for doxorubicin or S. atropurpurea extract alone. The plant methanol extract also reversed P-glycoprotein or multidrug resistance-associated protein in Caco-2 cells . The use of silver nanoparticles with S. atropurpurea subsp. maritima water fruit extract was found to be promising anticancer agents with cytotoxic activity against the human multiple myeloma U266 cell line and the human breast cancer cell line MDA-MB-231. The silver nanoparticles inhibited the growth of cells in a concentration-dependent manner with IC 50 values of 10 and 12 µg/mL, respectively . Some specialized metabolites such as phenolic acids, triterpenoid derivatives, or iridoids isolated from Dipsacus or Scabiosa revealed cytotoxic effects in various cancer cell lines . Phenolic acids, such as caffeic acid, 2,6-dihydroxycinnamic acid, vanillic acid, 2′- O -caffeoyl- d -glucopyranoside ester, and caffeoylquinic acid, demonstrated cytotoxic activity against five cancer cell lines (A549, Bel7402, BGC-823, HCT-8, and A2780) with IC 50 values ranging from 3.883 µg/mL to 7.395 µg/mL. The positive control, fluorouracil (a known cytostatic compound), had an IC 50 value of 0.177–0.695 µg/mL . Akebia saponin PA from D. asperoides caused the death of various human gastric cancer cell lines (AGS, MKN-45, SNU-638, and KATO III) via both apoptosis and autophagy. The IC 50 values were 24.1 µM (MKN-45 cells), 27.6 µM (SNU-638), 30.3 µM (AGS), and 36.5 µM (KATO III). In addition, akebia saponin PA increased the AGS cell number in the sub-G 1 phase and activated caspase-3, cleavage of PARP-1, MAPK, and p38/c-Jun N-terminal kinase. Autophagy was induced through the PI3K/AKT/mTOR and AMPK/mTOR pathways . Another saponin, akebia saponin D, was also found to induce cytotoxicity of the human monocyte-like histiocytic U937 cells in a concentration-dependent manner (0.1–1000 µM); it also enhanced the percentage of sub-G 1 cells and increased Bax and p53 gene expression . Saponin XII isolated from the roots of D. japonicus (1–2 µg/mL) suppressed the growth of acute myeloid leukemia OCI-AML3 cells; it stimulated apoptosis, increased the number of cells in the G 0 /G 1 phase of the cell cycle, decreased the number of cells in the S and G 2 /M phases, and activated caspase-3 . Some triterpenoid saponins and iridoids isolated from S. stellata whole plants were found to have cytotoxic effects against the fibrosarcoma HT1080 cell line . Scabiostellatoside F, at a concentration of 12.0 mM, was able to inhibit HT1080 cell growth by 50% . Other triterpenoid saponins, scabiostellatoside B, D, E, and H, were found to have an IC 50 of 38–49 µM. In addition, scabiostellatoside A, C, and G were not cytotoxic at a concentration of 50 mM . Yu et al. found that some compounds isolated from D. asper roots such as ursane and oleanane type triterpenoids with a feruloyloxy group or an arabinosyl moiety at C-3 were more cytotoxic than arboinane-type triterpenoids against four tumor cell lines: lung A549, liver H157 and HepG2, and breast MCF-7. Moreover, the highest activity was shown by an ursane-type triterpenoid (3 β - O -trans-feruloyl-2 α -hydroxy-urs-12-en-28-oic acid) with IC 50 values of 5.66 μM (H157), 9.36 μM (MCF-7), 9.5 μM (HepG2), and 12.8 μM (A549). The oleanane-type triterpenoid arabinoglycosides with a diacetylated sugar unit displayed cytotoxicity against A549 and H157 cell lines with IC 50 values below 10 μM. The compounds with a free or monoacetylated sugar moiety demonstrated cytotoxic activity with IC 50 values above 20 μM . Another oleanane-type triterpenoid saponin isolated from D. asper roots (3- O -[ β - d -xylopyranosyl-(1→4)- β - d -glucopyranosyl-(1→4)][α- l -rhamnopyranosyl-(1→3)]- β - d -glucopyranosyl-(1→3)-α- l -rhamnopyranosyl-(1→2)-α- l -arabinopyranosylhederagenin) displayed cytotoxicity against two lung cancer cells lines, A549 and H157, with IC 50 values of 6.94 and 9.06 μM, respectively . The 16 kDa water-soluble polysaccharide (ADAPW) isolated from D. asperoides roots had the ability to inhibit the growth of human osteosarcoma cell line HOS and induce apoptosis in a concentration-dependent manner (100, 200, and 400 µg/mL) after 24 h. It was also found to down-regulate PI3K and pAkt protein levels, reduce mitochondrial membrane potential, and increase intracellular ROS level . However, a number of iridoid glycosides (dipsanosides C-G, 3′- O - β - d -glucopyranosyl sweroside, loganin, cantleyoside, triplostoside A, lisianthioside, and 6′- O - β - d -apiofuranosyl sweroside) had no cytotoxic effect on a set of tested cell lines, including lung carcinoma A549, hepatoma Bel7402, gastric carcinoma BGC-823, colon cancer HCT-8, and ovary cancer A2780 . Similarly, 7- O -( E - p -coumaroyl)-sylvestroside I isolated from the whole plants of S. stellata also was not cytotoxic (IC 50 > 100 μg/mL) to fibrosarcoma HT1080 cells. However, 7- O -( E -caffeoyl)-sylvestroside I showed moderate activity, with an IC 50 value of 35.9 μg/mL . Taken together, these above results indicated that some Dipsacus and Scabiosa plants or some specialized metabolites may display anticancer activity and may be useful as chemopreventive agents. An increase in bacterial resistance to antibiotics has caused researchers to look for alternative solutions, which may be natural antibiotics . Recent studies confirmed that extracts or essential oils from Dipsacus or Scabiosa spp. such as D. asper , D. fullonum , D. japonicus , S. stellata , S. arenaria , or S. atropurpurea subsp. maritima have antimicrobial activity . Traditionally, D. fullonum is known as the remedy for Lyme disease caused by Borrelia burgdorferi whose vectors are ticks. The anti- Borrelia activity of D. fullonum / D. sylvestris extracts were evaluated in only a few studies in the recent ten years . A 70% ethanol extract of D. fullonum leaves and its fractions showed significant anti- Borrelia activity against the stationary phase of B. burgdorferi strain B31 . The strongest growth inhibition was found for a crude ethanol extract, which suppressed the cell viability by about 80% at a concentration of 305.5 mg/L. The NP5 fraction, containing loganic acid, and NP7, rich in saponarin, isoorientin, and two chlorogenic acid derivatives, were also effective, with a residual viability of 23.4–29.8% at a concentration of 332.8 mg/L and 340.2 mg/L, respectively; these values were comparable to that of the positive control, the triple antibiotic combination (doxycylin, cefoperazone, and daptomycin at a dose of 22.2 mg/L, 33.4 mg/L, and 80.1 mg/L, respectively) . In contrast, Feng et al. found that the 45% ethanolic extract of D. fullonum (accidentally mixed with a sample of D. asper ) at a concentration of 0.25–1% was not active against either the non-growing stationary phase or growing B. burgdorferi , with residual viability of 84–90% and MIC > 2%. Among three tested extracts (70% ethanolic, ethyl acetate, and dichloromethane extracts) from D. sylvestris roots, only the ethanol extract was inactive against B. burgdorferi while the ethyl acetate extract showed the strongest ability . A 50% methanolic extracts of D. fullonum leaves and roots were also tested against other microorganisms, including bacteria ( Bacillus subtilis B5, Escherichia coli ATCC 10536, Pseudomonas aeruginosa DSM 939 , P. fluorescens W1, and Staphylococcus aureus DSM 799) and yeasts ( Candida famata AII4b, C. tropicalis ATCC 60557, C. sphaerica FII7A, Saccharomyces cerevisiae SV30, and Yarrowia lipolytica PII6a). It was found that the cell growth inhibitory activity differed among plant materials and bacteria strains. The greatest effect of growth inhibition zones was observed for the root extract against E. coli ATCC 10536 and S. aureus DSM 799 . The antibacterial potential was also found for S. arenaria . Various degree of antibacterial activity was related to the type of plant material (stems and leaves, roots, flowers, and fruits) and solvent used (crude extract and its fractions such as ethyl acetate, n -butanol, and aqueous). It was found that the highest antibacterial effect was noted for the n -butanolic fraction of fruits. In this case, MIC values for two Escherichia coli strains and two Pseudomonas aeruginosa strains were 0.019 mg/mL and 0.156 mg/mL, respectively. The butanolic fractions of the aerial parts and flowers were active against only E. coli strains with MIC values of 0.078 mg/mL and 0.156 mg/mL, respectively. In addition, Staphylococcus aureus ATCC 25923 and S. saprophyticus were sensitive to the butanolic fraction of fruits (MIC = 0.625 mg/mL). Among four tested strains of Candida spp. ( C. albicans ATCC 90028, C. glabrata ATCC 90030, C. parapsilosis ATCC 22019, and C. krusei ATCC 6258), the most sensitive was C. albicans ATCC 90028 with MIC = 0.0195 mg/mL. E. coli ATCC 25922 and C. albicans ATCC 90028 were also sensitive to eleven subfractions from the butanolic fraction of the aerial part (MIC = 0.0195 mg/mL) . A 70% ethanol extract of the whole plant S. stellata showed the highest antibacterial activity against Streptococcus pyogenes with MIC = 1.2 mg/mL (in comparison to gentamicin MIC = 2 µg/mL). For other strains of Gram-positive bacteria ( Bacillus subtilis , Enterococcus faecalis ATCC 1034, Staphylococcus aureus 8325-4, S. aureus CIP 53.154, S. epidermidis , Micrococcus luteus , and Listeria innocua ), Gram-negative bacteria ( Escherichia coli CIP 54.127, Enterobacter cloacae , Salmonella enterica , Serratia marcescens , Proteus vulgaris , Klebsiella pneumoniae , Providencia stuartii , Pseudomonas aeruginosa ATCC 9027, and Shigella sonnei ) and five yeasts ( Candida albicans , C. glabrata , C. tropicalis , C. kefyr , and Cryptococcus neoformans ), MIC ranged from 2.5 mg/mL to above 10 mg/mL . The highest antimicrobial activity was found for fractions B and C obtained after eluting from a Diaion HP-20 column with 25% and 50% methanol. Staphylococcus spp., Candida spp. ( C. albicans , C. tropicalis , and C. kefyr ), and Cryptococcus neoformans were the most sensitive microorganisms to both fractions with MIC values of 0.6–1.5 mg/mL, while E. faecalis ATCC 1034 , M. luteus , and S. pyogenes were also sensitive to fraction B . The ethyl acetate, n -butanol, and the petroleum ether extracts from S. stellata whole plants were also tested for antibacterial activity in the agar disk diffusion assay against ten bacterial strains including four Gram-positive ( Staphylococcus aureus ATCC 25923, S. albus , Enterococcus spp., and Streptococcus D) and six Gram-negative bacteria ( Escherichia coli ATCC 35218 , Pseudomonas aeruginosa ATCC 15442 , Acinetobacter baumannii, Proteus mirabilis, Salmonella typhimurium, and Enterobacter sakazaki ) . Three bacterial strains, S. albus , P. aregionosa ATCC 15442, and S. typhimurium , were the most resistant strains to all extracts. The highest activity was exhibited by the ethyl acetate extract against the clinical strain of P. mirabilis (16–20 mm of the growth inhibition zones at a concentration of 0.0625–1 mg/mL). In addition, this extract also was active against five other bacterial strains, including S. aureus ATCC 25923, A. baumannii , E. coli ATCC 35218, Enterococcus sp., and Streptococcus D. The petroleum ether extract showed inhibitory activity against S. aureus (ATCC 25923) and E. coli (ATCC 35218) while the n -buthanol extract against A. baumannii and E. sakazaki . In another study, the antibacterial and antifungal activities of the silver nanoparticles with S. atropurpurea subsp. maritima water extract from fruit against bacteria ( Escherichia coli , Micrococcus luteus , Staphylococcus aureus , and Klebsiella pneumoniae ) and fungal pathogens including Candida clinical strains ( C. albicans , C. tropicalis , and C. glabrata ), Microsporum canis , Trichophytom rubrum, and Trichophytom interdigitale were also reported. The silver nanoparticles inhibited the cell growth of bacteria and Candida sp., as evidenced by the zone inhibition (19.3–28 mm) and the MIC value (3.9–15.62 µg/mL). The lowest MIC value was found for two dermatophyte species, T. rubrum and T. interdigitale . In addition, the antifungal potential of the silver nanoparticles was associated with the disruption of membrane integrity and attenuation of the biofilm and hyphae formation . D. asper crude extract from the roots also displayed antifungal activity in vivo in a whole-plant assay. This property was evaluated against seven plant pathogenic fungi such as Magnaporthe grisea causing rice blast, Rhizoctonia solani causing rice sheath blight, Botrytis cinerea causing tomato gray mold, Phytophthora infestans causing tomato late blight, Puccinia recondita causing wheat leaf rust, Blumeria graminis f. sp. hordei causing barley powdery mildew, and Colletotrichum coccodes causing red pepper anthracnose. It was shown that the activity was dependent on the fungal pathogens and the solvent used ( n -hexane, ethyl acetate, acetone, methylene chloride, and methanol) for extraction. The fungi causing the tomato late blight and the tomato gray mold were the most sensitive to Dipsacus root extract. The greatest anti-fungal effect was demonstrated by the ethyl acetate and acetone extracts at a concentration of 1–2 mg/mL that inhibited tomato diseases by 90% . Antifungal activity was also demonstrated by the pure compounds isolated from the roots of D. asper such as cauloside A (the main compound of the extract). Cauloside A was most effective against fungal pathogens causing the tomato late blight, the rice blast, and the tomato gray mold at a dose of 0.5 mg/mL. Colchiside inhibited the growth of Phytophthora infestans while three sterols (campesterol, β-sitosterol, and stigmasterol) displayed the weakest antifungal activity . Among twelve specialized metabolites isolated from the whole plant S. stellata , two iridoids, viz., 7- O -caffeoyl-sylvestroside I and 7 -O -( p -coumaroyl)-sylvestroside I, showed the highest antimicrobial activity with an MIC value of 31.2 µg/mL against Enterococcus faecalis ATCC 1054 and Staphyllococcus epidermis ; sylvestroside I was also able to inhibit the growth of E. coli CIP 54.127 (MIC = 62.5 µg/mL). These iridoids also inhibited the growth of S. aureus CIP 53.154 with an MIC value of 62.5 µg/mL . 2′,4′- O -diacetyl-3- O - α - l -arabinopyranosyl-23-hydroxyolea-12-en-28-oic acid and hederagonic acid, isolated from D. asper roots, inhibited the growth of S. aureus ATCC 25923 with IC 50 values of 12.3 and 10.3 µM, respectively. Furthermore, 2 α ,3 β ,24-trihydroxy-23-norurs-12-en-28-oic acid and 2 α ,3 β -dihydroxy-23-norurs-4(24),11,13(18)-trien-28-oic acid also exhibited antimicrobial activity but the IC 50 value was three-times higher . The other triterpenoid derivative, oleanolic acid, found in some Dipsacus and Scabiosa species, showed weaker antibacterial properties against E. coli ATCC 25922, Pseudomonas aeruginosa ATCC 27853, and Candida albicans ATCC 90028 with an IC 50 value ranging from 170 µM to 680 µM . It is well known that essential oils and their ingredients have potent antimicrobial properties . The essential oil isolated from flowers of S. arenaria showed a strong ability (stronger than the positive control, thymol; MIC = 0.2 mg/mL) to inhibit the growth of cells of two Staphylococcus aureus strains with an MIC = 0.1562 mg/mL . Notably, the essential oil isolated from fruits was found to be an anticandidal agent against Candida albicans ATCC 90028, C. parapsilosis ATCC 27853, C. kreusei ATCC 6258, and C. glabrata ATCC 90030 (MIC = 0.625 mg/mL) . The essential oil isolated from D. japonicus flowering aerial parts can be used as a promising, natural insecticidal agent against stored-product insects such as adult red flour beetles ( Tribolium castaneum ) and maize weevils ( Sitophilus zeamais ), and they displayed contact toxicity with LD 50 values of 13.45 μg/adult and 18.32 μg/adult, respectively. This essential oil also had fumigant activity against adult insects with LC 50 5.26 mg/l air for T. castaneum and 10.11 mg/l air for S. zeamais . The strongest fumigant toxicity was possessed by one of the abundant ingredients in D. japonicus essential oil, i.e., 1,8-cineole . Regarding antiviral activity, only one study showed that dipsalignan A, -1-hydroxy-2,6-bis- epi -pinoresinol, and dipsanosides M-N displayed inhibitory activities against human immunodeficiency virus-1 (HIV-1) integrase. The IC 50 values were 53.26 μM, 61.74 μM, 84.03 μM, and 92.67 μM, respectively. The positive control, baicalein, had a value of 1.37 μM . Akebia saponin D was also found to be a potential antidepressant agent. Intraperitoneal injection (40 mg/kg/d) alleviated LPS-induced microglia-mediated neuroinflammatory response in mice by inhibiting the TLR4/NF-κB signaling pathway in the hippocampus and prefrontal cortex . It also ameliorated chronic mild stress-induced depressive-like behaviors in C57BL/6 mice by inducing a neuroprotective microglial phenotype in the hippocampus through the PPAR-γ pathway . A similar antidepressant effect was also found for apigenin in a mouse model of chronic mild stress . Gong et al. found akebia saponin D to be effective against pain. It displayed an anti-nociceptive effect in SPF KM mice by shortening the licking time in the formalin test, increasing the reaction time to heat stimuli, and inhibiting acetic acid-induced writhing in mice. Akebia saponin D activated the expression of the progesterone receptor in primary decidual cells and the Notch signaling pathway. Gao et al. proposed that Dipsaci radix and its main ingredient, akebia saponin D, may promote decidualization in pregnant women. Bushen Antai, a Chinese herbal medicine preparation containing Dipsaci radix , was found to reduce the pregnancy loss caused by mifepristone administration . This preparation may stimulate estrogen and progesterone receptors through Akt and Erk1/2 signaling pathways in the maternal–fetal interface of pregnant rats. The present review broadens the knowledge of the phytochemistry of some species of Dipsacus and Scabiosa genera as well as their biological properties. The phytochemical analyses showed qualitative similarities in some specialized metabolites, especially iridoids, between species of both genera. Some species of Dipsacus and Scabiosa contain above 200 different compounds belonging to iridoids, triterpenoids derivatives, flavonoids, or phenolic acids with caffeoylquinic acid derivatives. Dipsacus spp. were predominated by terpenoid saponins while Scabiosa spp. were rich sources of iridoids and flavonoids. Apigenin, luteolin, and their derivatives were particularly common. Hederagenin and its related saponins are the main group of triterpenoids identified in Dipsacus . The oleanane-type triterpenoids were also common in Scabiosa genus. The wine-processing method has a beneficial effect on the biological activity of Dipsaci radix and the level of some specialized metabolites with its quality indicator, akebia saponin D. Dipsacus, and Scabiosa species and their constituent compounds possess beneficial biological activities. Many in vitro and in vivo studies confirmed their traditional medicinal uses. Dipsaci radix and akebia saponin D demonstrated anti-osteoporosis and antiarthritic properties. Scabiosa spp. showed anti-hepatic fibrosis potential. In addition, akebia saponin D displayed cardioprotective activity. Antioxidant, antimicrobial, and anti-inflammatory activities of both genera were also confirmed. Some newly identified specialized metabolites such as polysaccharides displayed promising biological properties. Thus, in the future, it is worth paying more attention to their pharmacological activities. The varied biological activities of extracts from Dipsacus and Scabiosa as well as the pure compounds isolated from them indicate their potential use in the future as effective, natural herbal drugs in the treatment of various diseases in official medical applications. |
Variability of functional and biodiversity responses to perturbations is predictable and informative | ec5c1d63-92fd-42bd-a5c3-f3a2bd46f48c | 11604961 | Microbiology[mh] | Describing aggregate properties of ecosystems and predicting their behaviour in the face of perturbations is a major goal of contemporary ecology. If consistent patterns emerged when considering aggregate-level responses, ecologists could aim for data-based predictions and provide clear, practical recommendations , . However, there are many relevant aggregate properties to consider, from diversity metrics to ecosystem functions, that may all respond in different ways to perturbations , . As there is no obvious way to organize this variability, the hopes for general predictions of community-level responses to perturbations can seem slim. The importance and origin of species diversity was a central theme of late 20th century ecology – , which led to a proliferation of metrics to define and measure diversity based on the richness, evenness and rarity of species – . Since then, understanding how species collectively perform a function has become a prominent area of research – , with clear implications for our understanding of concrete issues regarding productivity, carbon sequestration, pollination, or nutrient cycling of natural or engineered ecosystems. In light of rapid anthropogenic global change, there is currently increased focus on understanding how aggregate ecological properties will respond to perturbations such as land-use change, invasive species, climate change and pollution , – . Ecologists are very aware that different aggregate properties, such as diversity metrics or ecosystem functions, describe very different aspects of communities and may thus respond in completely different ways to a given environmental perturbation , , . For instance, the many different diversity metrics employed by ecologists describe different facets of community structure , . If a perturbation caused the extinction of rare species while making the overall abundance distribution of the community more even, species richness would decrease, but a measure of evenness (e.g. Simpson’s index) would increase. Similarly, ecosystem functioning takes many forms, and can be measured in a myriad of ways. Some functions, such as biomass production or respiration, are broad functions: they are performed by most or all species in a community. Other functions, such as the breakdown of specific chemicals or the production of specific enzymes, are narrow in the sense that they require the presence of particular species, or combinations of species, to be performed , . The great variety of ecosystem functions—in what they do, how broad or narrow they are, how species contribute to them, and how they respond to perturbations—has motivated the rapid development of multifunctional ecology where multiple functions are considered at once to more accurately characterize the state of an ecosystem – . In the face of this inherent ecological complexity, what can be learned from the variability of functional and biodiversity responses to perturbations? Here we claim that this variability can be used to explore hidden features of ecosystems and of perturbations. To make this point we analyse data from global change experiments conducted in microbial soil systems (Box and Fig. ). Focusing on three diversity metrics, two broad ecosystem functions, and eight narrow ecosystem functions, we explore patterns of mismatches between functional and diversity responses to global change factors (such as pollution, environmental events or land-use change, all seen here as perturbations). Concretely, we look at the proportion of cases where one aggregate property responds negatively to a perturbation while the other responds positively to it. As expected, we find a great degree of variability in responses to perturbations. This variability, however, is not random, but instead shows a recognizable degree of structure. Aggregate properties that are thought to describe ecosystems in similar ways (e.g. production of beta-xylosidase and production of cellobiohydrolase, enzymes that contribute to carbon cycling) have a lower proportion of mismatches than would be expected by chance (modules of blue squares, Fig. A). On the other hand, diversity metrics and ecosystem functions tend to systematically differ in how they respond to perturbations (dominance of red squares between diversity and ecosystem functions, Fig. A). Our intuitions about how mechanistically similar aggregate properties are (i.e. how we ordered the observations Fig. A) thus provide a useful starting point for understanding ecosystem’s response variability, and we also find that there exist generic diversity-function response patterns. Motivated by the findings of this empirical synthesis, we propose a framework that helps us glean useful, hidden information from the variability of functional and diversity responses to perturbations. To do so, we convert the ecological problem into a simpler geometrical one by representing perturbations as displacement vectors and community aggregate properties as directions in community state-space (the high-dimensional space whose axis reports the biomass of all constituent species). The central ingredient of our framework is a geometrical definition of collinearity between two aggregate properties which quantifies their similarity and predicts whether they will respond to a perturbation in the same way (Fig. ). This prediction assumes a high response diversity at the species level, and depends on how species’ responses to perturbations scale with their biomass. Here, coarse-grained assumptions about population-level responses are used to better understand ecosystem functions. Conversely, we show that with some knowledge of the aggregate properties used to observe the ecological impacts of perturbations, the variability of these observations can be leveraged to gain information about species response diversity and how species’ responses scale with their biomass. Armed with our geometrical framework we then reanalyse the empirical data from microbial soil systems to gain new insights on soil microbial ecosystem functions and how they are being impacted by anthropogenic global change. As well as proposing novel methods for validating and applying our framework to ecological data (outlined in an online tutorial at https://jamesaorr.github.io/community-properties-tutorial ), we more broadly aim to inspire new approaches to studying complex ecological systems that embrace the variability of community-level responses to perturbations, using perturbations as probes to reveal hidden features of ecosystem dynamics and functioning. Box 1 Initial analysis of empirical data To quantify the variability of functional and biodiversity responses to perturbations we analysed a dataset of global change experiments conducted in microbial soil systems . This dataset contained 1235 perturbations from 341 publications. Perturbations included warming, elevated carbon dioxide levels, altered precipitation, nutrient enrichment, land-use change, or combinations of these factors. The effect of each perturbation in a given experiment was quantified using the natural logarithm-transformed response ratio: Box 1 Eq 1 \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${{{\rm{RR}}}}=\ln \left(\frac{{X}_{{{\rm{t}}}}}{{X}_{{{\rm{c}}}}}\right)$$\end{document} RR = ln X t X c where X t and X c are the means of the treatment and control groups for a given aggregate property. The variances of these effect sizes are also available in the dataset, but we do not require them for this initial analysis as we do not exclude points based on some statistical cutoff. Indeed, following our geometric approach, there is no reason to expect that the proportion of mismatches between two aggregate properties would be different for data points with or without statistically significant results. Each individual perturbation was quantified using multiple aggregate properties covering a wide range of ecosystem functions and measures of diversity. We focused on aggregate properties where all pairs had at least ten observations in the dataset so that the proportion of mismatches between them could be estimated with some robustness. This arbitrary number of observations was chosen to strike a balance between having enough observations to estimate proportions of mismatches reliably and having enough pairs of aggregate properties to see general patterns across broad functions, narrow functions, and diversity metrics. Choosing other cut-offs does not qualitatively change the results (demonstrated in the R markdown at 10.5281/zenodo.13985015). This filtering of the data returned 1015 perturbations that were measured with at least two of thirteen aggregate properties including three measures of diversity (richness, Shannon index, and Chao index), two broad ecosystem functions (biomass and respiration), and eight narrow ecosystem functions subdivided into P-cycling enzymes (phosphatase), N-cycling enzymes ( N -acetyl-beta-glucosaminidase), hydrolytic C-cycling enzymes (beta-xylosidase, cellobiodydrolase, beta-glucosidase, and alpha-glucosidase), and oxidative C-cycling enzymes (peroxidase, phenol oxidase). Details of how the functions were measured (e.g. whether respiration was calculated in the laboratory or the field, or whether enzymes were measured using colourimetric or microplate assays) can be found in Zhou et al. (2020). This list of aggregate properties was sorted a priori based on intuitions about their underlying mechanisms (grouped by diversity metrics, broad functions and narrow functions based on Zhou et al. (2020)) and a heatmap was made to visualize the proportion of mismatches between each pair (Fig. A). If the variability between aggregate properties was just random (i.e. if the heatmap was all white or just showed random distributions of red and blue) there might not be much more to say, but if the heatmap showed some structure there could be useful information to gain from the variability. Indeed, the modularity of the heatmap shows that aggregate properties that are thought to be similar tend to respond to perturbations similarly (e.g. relatively low proportion of mismatches—ranging from 0.16 to 0.28—between measures of diversity). Conversely, groups of aggregate properties that describe different aspects of a community can systematically differ in their responses to perturbations (e.g. an abundance of red between diversity metrics and ecosystem functions, with the proportion of mismatches going as high as 0.73). We will return to these empirical results after we have outlined our geometrical approach for quantifying the notion of similarity between aggregate properties. In fact, we can use our framework to reinterpret these empirical data to gain useful insights into how the perturbations in these experiments impacted these communities and also into how the species in these communities contribute to the different ecosystem functions. Our geometrical arguments—outlined in the Methods and in Fig. —are well supported by our simulation results and can be used to refine the analysis of the empirical data. By simulating perturbation experiments on species-rich communities, we show how mismatches in the observations of two functions can be used to quantify the similarity of those functions and can be used to estimate a notion of response diversity (Fig. ). We also found that mismatches in the observations of a function and a diversity metric can be used to quantify the scaling of perturbations by species biomass (Fig. ). Returning to the empirical data, we applied a validation test (formally described in Supplementary Note ) to show that the data, when grouped by biome, meet the assumptions of our geometrical arguments. We could then quantify (i) the similarity and broadness of empirically measured functions, and (ii) the response diversity and biomass scaling of key global change factors (Fig. ). Mismatches between functions In theory, the proportion of qualitative response mismatches between two linear ecosystem functions directly depends on their collinearity (Eq. ; that is, the angle between their respective directions in phase-space (the high-dimensional space whose axis reports the biomass of all constituent species). This is confirmed by our simulations, whose outcomes are represented in Fig. E. This basic result, however, hinges upon the assumption that perturbations are unbiased at the population level; meaning that approximately half the species show positive responses and half the species show negative responses to any given perturbation. If population-level responses are biased towards positive or negative, the geometrical prediction overestimates the proportion of mismatches (Fig. ). This effect occurs because when perturbation effects on species are mostly negative (or mostly positive), they tend to fall in the areas of phase space where functions will necessarily observe the same responses (top right and bottom left quadrants in Fig. D). Because this systematic overestimation indicates that a key assumption is violated, it informs us about population-level effects of perturbations. We can therefore deduce a link between mismatches in observations at the community level and information on population-level response diversity (Fig. B). Deviations from our predictions reveal a degree of population-level response diversity to the perturbation considered. Mismatches between functions and diversity metrics The probability of mismatches between ecosystem functions and diversity metrics can be predicted by considering the angle between the function and the gradient of the diversity metric (Fig. A). Again, consistency of responses at the population level causes the prediction to overestimate the actual proportion of mismatches. We note, however, that the angle between the direction associated with a positive function and a diversity metric can exceed 90° leading to a systematic bias towards qualitative response mismatches. This intriguing result is connected to a second piece of population-level information: the scaling of perturbations by species biomass (Box Eq ). When the effect of perturbations is larger for more abundant species, function and diversity show qualitatively different responses (only the larger points are above the red line in Fig. A). If a perturbation causes the biomass of abundant species to decrease, total biomass will decrease but a diversity measure related to evenness will increase. If on the other hand, a perturbation causes the biomass of abundant species to increase, total biomass will increase but evenness will decrease. This means that the degree of scaling of species responses to perturbations by their biomass can be predicted based on the observed proportion of mismatches between total biomass and diversity measures (Fig. B). Empirical results The validation test of the geometrical framework (outlined in Supplementary Note ) with the entire Zhou et al. datasets (1235 perturbations tested across a huge diversity of biomes including agricultural systems, tundra, desert, and wetlands) was negative. We found no correlation between actual mismatches between two given functions and predicted mismatches based on the mismatches with other functions, if the latter could be seen as vectors in a given phase space . However, validation tests with data from either grassland systems ( n = 367) or forest systems ( n = 435) provided very conclusive support (strong correlation between predicted and realized mismatches) for the use of our geometrical framework (Fig. A). The fact that the test was inconclusive when pooling all data together should not be surprising, since the notion of unique phase space to position the different systems does not make sense. Only when grouping by biome can this fundamental assumption stand a chance of being a useful approximation (but it could very well have failed as the systems remain very different: unlike simulation experiments, the data does not represent repeated perturbations of the same system). A network depicting the similarity of functions using the grasslands dataset further reinforced that mismatch data coincides with our mechanistic understanding of these ecosystem functions (Fig. B). The only two broad functions in the network, biomass and respiration, are beside each other in the network and the seven narrow functions (production of different enzymes) grouped together as would have been expected a priori . Of the ecosystem functions with enough observations to make estimates of their broadness, respiration was the broadest, followed by net nitrogen mineralization rate, and then by three specific enzymes related either to carbon cycling (phenol oxidase and beta-1,4-glucosidase) or to phosphorus cycling (phosphatase). These quantitative estimates support our basic biological intuitions about these systems: a few species contribute to the production of a specific enzyme, more species are involved in the mineralization of nitrogen, and more species still contribute to whole ecosystem respiration. The estimates of broadness for beta-1,4-glucosidase and phosphatase were almost identical for forests and grasslands but respiration was estimated to be more broad in grasslands than in forests (Fig. C). In Supplementary Note we further show that the estimated broadness of functions based on their mismatches with total biomass can even be used to predict their actual proportion of mismatches. The perturbations in the dataset that had enough observations for us to examine their response diversity and/or biomass scaling were warming, carbon dioxide enrichment, phosphorous addition, nitrogen addition, phosphorus and nitrogen addition combined, nitrogen addition and increased precipitation combined, and four types of land-use change: conversion from native ecosystems to agriculture, to pasture, to plantation or to secondary ecosystems. For grasslands, conversion to secondary ecosystems or the addition of nitrogen or carbon dioxide had relatively low response diversity while warming had relatively high response diversity. For forests, land-use change (particularly conversion to pastures) had relatively low response diversity while warming and the addition of nitrogen had relatively high response diversity (Fig. D). For grasslands, land-use change typically showed strong biomass scaling while nutrient enrichment and warming showed relatively weak biomass scaling. For forests, conversion to pasture had relatively low biomass scaling while conversion to secondary ecosystems had relatively high biomass scaling with the other perturbation types falling in between the two (Fig. E). In general, land-use change perturbations had low response diversity and high biomass scaling indicating that species respond in the same way (presumably negatively) and have absolute changes relative to their biomass (e.g. consistent with a perturbation decreasing 50% of all species). Perturbations like warming and nutrient enrichment, on the other hand, typically had high response diversity and low biomass scaling indicating that some species responded negatively while some responded positively to these perturbations and that absolute responses were not completely proportional to initial biomass. In theory, the proportion of qualitative response mismatches between two linear ecosystem functions directly depends on their collinearity (Eq. ; that is, the angle between their respective directions in phase-space (the high-dimensional space whose axis reports the biomass of all constituent species). This is confirmed by our simulations, whose outcomes are represented in Fig. E. This basic result, however, hinges upon the assumption that perturbations are unbiased at the population level; meaning that approximately half the species show positive responses and half the species show negative responses to any given perturbation. If population-level responses are biased towards positive or negative, the geometrical prediction overestimates the proportion of mismatches (Fig. ). This effect occurs because when perturbation effects on species are mostly negative (or mostly positive), they tend to fall in the areas of phase space where functions will necessarily observe the same responses (top right and bottom left quadrants in Fig. D). Because this systematic overestimation indicates that a key assumption is violated, it informs us about population-level effects of perturbations. We can therefore deduce a link between mismatches in observations at the community level and information on population-level response diversity (Fig. B). Deviations from our predictions reveal a degree of population-level response diversity to the perturbation considered. The probability of mismatches between ecosystem functions and diversity metrics can be predicted by considering the angle between the function and the gradient of the diversity metric (Fig. A). Again, consistency of responses at the population level causes the prediction to overestimate the actual proportion of mismatches. We note, however, that the angle between the direction associated with a positive function and a diversity metric can exceed 90° leading to a systematic bias towards qualitative response mismatches. This intriguing result is connected to a second piece of population-level information: the scaling of perturbations by species biomass (Box Eq ). When the effect of perturbations is larger for more abundant species, function and diversity show qualitatively different responses (only the larger points are above the red line in Fig. A). If a perturbation causes the biomass of abundant species to decrease, total biomass will decrease but a diversity measure related to evenness will increase. If on the other hand, a perturbation causes the biomass of abundant species to increase, total biomass will increase but evenness will decrease. This means that the degree of scaling of species responses to perturbations by their biomass can be predicted based on the observed proportion of mismatches between total biomass and diversity measures (Fig. B). The validation test of the geometrical framework (outlined in Supplementary Note ) with the entire Zhou et al. datasets (1235 perturbations tested across a huge diversity of biomes including agricultural systems, tundra, desert, and wetlands) was negative. We found no correlation between actual mismatches between two given functions and predicted mismatches based on the mismatches with other functions, if the latter could be seen as vectors in a given phase space . However, validation tests with data from either grassland systems ( n = 367) or forest systems ( n = 435) provided very conclusive support (strong correlation between predicted and realized mismatches) for the use of our geometrical framework (Fig. A). The fact that the test was inconclusive when pooling all data together should not be surprising, since the notion of unique phase space to position the different systems does not make sense. Only when grouping by biome can this fundamental assumption stand a chance of being a useful approximation (but it could very well have failed as the systems remain very different: unlike simulation experiments, the data does not represent repeated perturbations of the same system). A network depicting the similarity of functions using the grasslands dataset further reinforced that mismatch data coincides with our mechanistic understanding of these ecosystem functions (Fig. B). The only two broad functions in the network, biomass and respiration, are beside each other in the network and the seven narrow functions (production of different enzymes) grouped together as would have been expected a priori . Of the ecosystem functions with enough observations to make estimates of their broadness, respiration was the broadest, followed by net nitrogen mineralization rate, and then by three specific enzymes related either to carbon cycling (phenol oxidase and beta-1,4-glucosidase) or to phosphorus cycling (phosphatase). These quantitative estimates support our basic biological intuitions about these systems: a few species contribute to the production of a specific enzyme, more species are involved in the mineralization of nitrogen, and more species still contribute to whole ecosystem respiration. The estimates of broadness for beta-1,4-glucosidase and phosphatase were almost identical for forests and grasslands but respiration was estimated to be more broad in grasslands than in forests (Fig. C). In Supplementary Note we further show that the estimated broadness of functions based on their mismatches with total biomass can even be used to predict their actual proportion of mismatches. The perturbations in the dataset that had enough observations for us to examine their response diversity and/or biomass scaling were warming, carbon dioxide enrichment, phosphorous addition, nitrogen addition, phosphorus and nitrogen addition combined, nitrogen addition and increased precipitation combined, and four types of land-use change: conversion from native ecosystems to agriculture, to pasture, to plantation or to secondary ecosystems. For grasslands, conversion to secondary ecosystems or the addition of nitrogen or carbon dioxide had relatively low response diversity while warming had relatively high response diversity. For forests, land-use change (particularly conversion to pastures) had relatively low response diversity while warming and the addition of nitrogen had relatively high response diversity (Fig. D). For grasslands, land-use change typically showed strong biomass scaling while nutrient enrichment and warming showed relatively weak biomass scaling. For forests, conversion to pasture had relatively low biomass scaling while conversion to secondary ecosystems had relatively high biomass scaling with the other perturbation types falling in between the two (Fig. E). In general, land-use change perturbations had low response diversity and high biomass scaling indicating that species respond in the same way (presumably negatively) and have absolute changes relative to their biomass (e.g. consistent with a perturbation decreasing 50% of all species). Perturbations like warming and nutrient enrichment, on the other hand, typically had high response diversity and low biomass scaling indicating that some species responded negatively while some responded positively to these perturbations and that absolute responses were not completely proportional to initial biomass. Variability of results, or “context-dependency”, is pervasive in ecology . While this is partly what makes ecosystems so fascinating to study—indeed there is great interest in the mechanistic underpinning of contrasting responses of diversity and function to perturbations —it could also be viewed as an obstacle to the synthesis of previous results and to the prediction of future impacts. Our research has focused on some of this variability (the variability between the responses of community aggregate properties to a given perturbation) and found that it is predictable and also a rich source of information. Mismatches between the responses of different aggregate properties to a class of perturbations (e.g. land-use change) can give us previously hidden information about the aggregate properties themselves (i.e. similarity and broadness of ecosystem functions) and about how such perturbations impact the species that constitute the community (i.e. response diversity and biomass scaling). Ecological research is typically reductionist, using information about individuals and populations to understand communities and ecosystems . Our work demonstrates the reverse approach by using information about communities to understand population-level responses. In this paper we have reported two analyses of the microbial soil system dataset: (i) an initial, naive synthesis that we used to motivate our work (Fig. ), and (ii) a more detailed analysis informed by our geometrical framework (Fig. ). Our geometrical approach helped to explain some of the interesting patterns in the initial analysis—such as the relatively high levels of mismatches between broad and narrow functions and between functions and diversity—but more importantly, it allowed us to take our biological interpretations further and to extract new information from the data using a novel type of analysis. For instance, we found that perturbations associated with global change vary greatly in their response diversity (Fig. D). Land-use change typically had relatively low response diversity (i.e. most species responded in the same direction), while warming showed relatively high response diversity (i.e. some species increased in abundance while others decreased in abundance). Furthermore, we found that biomass scaling is a prominent feature of anthropogenic perturbations of these ecosystems. The proportion of mismatches between total biomass and Shannon diversity—positively correlated with the biomass scaling exponent (Fig. B)—ranged from ∼0.3 for warming in grassland systems all the way up to ∼0.8 for some land-use change perturbations. In other words, species that initially represent a large proportion of the overall biomass in these microbial systems also represent a large proportion of the variation in biomass caused by global change factors. In the next two sections, we will first outline in more detail the general empirical applications of our proposed framework and we will then discuss the future research directions that our geometrical perspective of aggregate properties could lead to. Empirical applications Although we have shown that variability of community-level responses to perturbations can be predicted, our geometrical framework does not attempt to predict how specific aggregate properties will respond to specific perturbations. Instead, based on the assumption that functions can be seen as directions (which amounts to assuming that per-capita contributions of species to functions are fixed), it can be used to generate null expectations for when aggregate properties should and shouldn’t respond in the same way to a perturbation. From a practical perspective, our framework therefore offers a novel set of methods (demonstrated in the tutorial available at: https://jamesaorr.github.io/community-properties-tutorial/ ) that ecologists can use to study species’ contributions to ecosystem functions and the population-level effects of perturbations. As the central ingredient of our framework is the proportion of mismatches in the observations of different aggregate properties, increasing the volume of data will lead to more robust estimates. Indeed, the size of the points in Fig. C–E indicates the confidence of those estimates. However, there is a trade-off between the volume of data used to quantify proportions of mismatches and the consistency of the underlying systems; the validation test was inconclusive when we pooled data from all biomes but gave very convincing results when we focused on either the grasslands or forests systems. Given the requirements for moderate to high volumes of data, our framework is probably best suited for use in research synthesis, where it can be used to complement traditional tools like meta-analyses. The geometrical view of aggregate properties allows us to use perturbations as probes to better understand how species influence the functioning of ecosystems. We found that the proportion of mismatches between functions can be used to quantify their similarity in terms of which species contribute to them. This was demonstrated by the modularity of the heat map in Fig. A and of the network in Fig. B. This observation is certainly reassuring, as it confirms that mechanistic understandings at the chemical level of microbial functions are consistent with ecosystem-level observations. Furthermore, given that total biomass is by definition the broadest function, we can now use mismatches between a function of interest and total biomass to quantify the broadness of that function (demonstrated by numerical evidence in Fig. E inset and empirical evidence in Fig. C). If an ecologist was interested in a new ecosystem function they could quickly compare it to other functions based on how it responds to perturbations to estimate how broad it was and to identify which species were contributing to it (based on it’s similarity to functions with more information about their species’ contributions). Understanding the links between community composition and functioning has far-reaching implications for many sectors including ecosystem management, agriculture, forestry and medicine – and our approach contributes to recent efforts to study ecosystem functions in their natural context, in contrast to the traditional reductionist approach of using controlled experiments where populations or even organisms are studied in isolation , . Our framework can also be used to study population-level responses to perturbations from the top down by comparing the observations of different functions. Response diversity—the variation between species responses to a perturbation—can be measured in different ways and is a key mechanism underlying ecological stability and the biological insurance hypothesis – . Although the information we can gain using our geometrical approach (i.e. the proportion of species responding positively or negatively—see Fig. ) is a coarse measure of response diversity, it can be accessed by just comparing the observations of different functions (e.g. total biomass and respiration) rather than actually measuring each species’ response. The easiest approach is to take two functions and compare their proportion of mismatches over different perturbations (or different systems or different contexts) to gain a relative measure of response diversity (as we did in Fig. D). However if the collinearity between two functions is known (for well-studied functions, or by using our approximations based on the estimated broadness of the functions), then we can use the deviations from our null expectation to quantify the population-level response diversity (Fig. B). Another useful piece of information that can be gained with our top-down approach is the biomass scaling of a perturbation (i.e. whether the direct effect of a perturbation is proportional to the biomass of each species). This feature of perturbations controls the relative importance of rare or common species in determining the community’s temporal variability (“environmental perturbations” sensu Arnoldi et al., ). Using mismatches between any function and any diversity metric can be used to rank perturbations based on their biomass scaling (Fig. E). Furthermore, the proportion of mismatches between diversity and total biomass is actually a very good proxy for the biomass scaling exponent itself Fig. B). If the responses of each species to the perturbations is available then biomass scaling (and response diversity) can be extracted from the data directly. However, it is very common for measures of diversity to be estimated from data without measuring species-level responses (there are 221 observations in the Zhou et al., dataset where OTU richness is the only measure of diversity). In these cases, biomass scaling cannot be measured directly from the data, but it can be estimated using our framework. Comparing multiple community-level observations—measuring responses of more functions allows for more pairwise comparisons and therefore more detailed insights—allows us to describe these features of perturbations without ever having to collect information directly at the population level, which could therefore be an efficient and cost-effective tool for research synthesis or the analysis of biomonitoring data. Future directions Our work has so far overlooked the temporal dynamics of responses to perturbations. As we only needed to consider the initial and perturbed states of ecosystems for our geometrical approach (perturbations as displacement vectors in Fig. A), we haven’t made the distinction between press and pulse perturbations and we also haven’t considered non-linear responses. However, to consider a community’s trajectory during and after a perturbation, our framework could be applied in future studies to test if response diversity and biomass scaling of perturbations change over time. Practically this would involve comparing the responses of two (or more) aggregate properties to a perturbation over time and checking if there was a change in the proportion of mismatches (e.g. over-replicates in an experimental treatment). For example, if the proportion of mismatches between a set of ecosystem functions was initially very low following a perturbation but then increased over time, this would be consistent with a scenario where most species initially responded negatively to that perturbation but then some species increased in abundance (e.g. due to competitive release). Changes in the proportion of mismatches between diversity and function over time would likewise imply changes in the biomass scaling of a perturbation. It seems likely that this new geometrical perspective could be combined with tools in the ecological stability literature to study dynamic ecological responses to perturbations. In our work, we did not explicitly consider biotic interactions, yet they nonetheless play a role. The state that an ecosystem reaches after a perturbation undoubtedly depends on species interactions, especially if the time scale considered is long enough to allow community dynamics to play out. The classic example is the trophic cascade . If a perturbation directly impacts the top of a food chain (e.g. species invasion), it will in time also affect its base, following the alternating sign pattern characteristic of a cascade. Put in the context of our work, biotic interactions play a role in what we call “features of perturbations” like biomass scaling and response diversity. An exciting future direction would therefore be to seek for recognizable signatures of species interactions in the variations through time of those perturbation features. For instance, we can hypothesise that strong mutualistic interactions would generate increasingly coherent responses as time grows (corresponding to a reduction of response diversity). Furthermore, our framework makes the simplifying assumption that species per capita contributions to functions are fixed, but in reality how a species contributes to a function may be dependent on its interactions with other species (although the fact that our validation test was conclusive implies that this assumption is not a bad approximation). We propose to see our work as a first step of a more general program: using perturbations as “probes”, where ecosystem functions are macroscopic “observables”, to better understand the dynamics of natural ecosystems. Given the generality of our framework, our work touches many areas of contemporary ecology. For multifunctional ecologists, it helps to explain how different functions can respond in different ways to global change . For ecologists interested in multiple perturbations, our work can be used to understand variability in how community-level properties observe the interactions (antagonistic or synergistic) between perturbations . For biodiversity-ecosystem functioning research, the opposing responses of diversity and function to perturbations (which we explained) should be considered when understanding how perturbations influence biodiversity-ecosystem functioning relationships . Our work can be used in disturbance ecology to link studies across disparate systems and may even help to interpret trade-offs between biodiversity and crop yield under different farming practices . When studying complex systems such as ecosystems, it is important to have baseline expectations for their behaviour. We have found that the variability between community-level responses to perturbations does not just limit synthesis and prediction in ecology. Instead, this variability is predictable and can be leveraged to gain useful information about species’ responses to perturbations and species’ contributions to ecosystem functioning. Our work provides a solid platform from which the complexity of community-level responses to anthropogenic global change can be better understood. Although we have shown that variability of community-level responses to perturbations can be predicted, our geometrical framework does not attempt to predict how specific aggregate properties will respond to specific perturbations. Instead, based on the assumption that functions can be seen as directions (which amounts to assuming that per-capita contributions of species to functions are fixed), it can be used to generate null expectations for when aggregate properties should and shouldn’t respond in the same way to a perturbation. From a practical perspective, our framework therefore offers a novel set of methods (demonstrated in the tutorial available at: https://jamesaorr.github.io/community-properties-tutorial/ ) that ecologists can use to study species’ contributions to ecosystem functions and the population-level effects of perturbations. As the central ingredient of our framework is the proportion of mismatches in the observations of different aggregate properties, increasing the volume of data will lead to more robust estimates. Indeed, the size of the points in Fig. C–E indicates the confidence of those estimates. However, there is a trade-off between the volume of data used to quantify proportions of mismatches and the consistency of the underlying systems; the validation test was inconclusive when we pooled data from all biomes but gave very convincing results when we focused on either the grasslands or forests systems. Given the requirements for moderate to high volumes of data, our framework is probably best suited for use in research synthesis, where it can be used to complement traditional tools like meta-analyses. The geometrical view of aggregate properties allows us to use perturbations as probes to better understand how species influence the functioning of ecosystems. We found that the proportion of mismatches between functions can be used to quantify their similarity in terms of which species contribute to them. This was demonstrated by the modularity of the heat map in Fig. A and of the network in Fig. B. This observation is certainly reassuring, as it confirms that mechanistic understandings at the chemical level of microbial functions are consistent with ecosystem-level observations. Furthermore, given that total biomass is by definition the broadest function, we can now use mismatches between a function of interest and total biomass to quantify the broadness of that function (demonstrated by numerical evidence in Fig. E inset and empirical evidence in Fig. C). If an ecologist was interested in a new ecosystem function they could quickly compare it to other functions based on how it responds to perturbations to estimate how broad it was and to identify which species were contributing to it (based on it’s similarity to functions with more information about their species’ contributions). Understanding the links between community composition and functioning has far-reaching implications for many sectors including ecosystem management, agriculture, forestry and medicine – and our approach contributes to recent efforts to study ecosystem functions in their natural context, in contrast to the traditional reductionist approach of using controlled experiments where populations or even organisms are studied in isolation , . Our framework can also be used to study population-level responses to perturbations from the top down by comparing the observations of different functions. Response diversity—the variation between species responses to a perturbation—can be measured in different ways and is a key mechanism underlying ecological stability and the biological insurance hypothesis – . Although the information we can gain using our geometrical approach (i.e. the proportion of species responding positively or negatively—see Fig. ) is a coarse measure of response diversity, it can be accessed by just comparing the observations of different functions (e.g. total biomass and respiration) rather than actually measuring each species’ response. The easiest approach is to take two functions and compare their proportion of mismatches over different perturbations (or different systems or different contexts) to gain a relative measure of response diversity (as we did in Fig. D). However if the collinearity between two functions is known (for well-studied functions, or by using our approximations based on the estimated broadness of the functions), then we can use the deviations from our null expectation to quantify the population-level response diversity (Fig. B). Another useful piece of information that can be gained with our top-down approach is the biomass scaling of a perturbation (i.e. whether the direct effect of a perturbation is proportional to the biomass of each species). This feature of perturbations controls the relative importance of rare or common species in determining the community’s temporal variability (“environmental perturbations” sensu Arnoldi et al., ). Using mismatches between any function and any diversity metric can be used to rank perturbations based on their biomass scaling (Fig. E). Furthermore, the proportion of mismatches between diversity and total biomass is actually a very good proxy for the biomass scaling exponent itself Fig. B). If the responses of each species to the perturbations is available then biomass scaling (and response diversity) can be extracted from the data directly. However, it is very common for measures of diversity to be estimated from data without measuring species-level responses (there are 221 observations in the Zhou et al., dataset where OTU richness is the only measure of diversity). In these cases, biomass scaling cannot be measured directly from the data, but it can be estimated using our framework. Comparing multiple community-level observations—measuring responses of more functions allows for more pairwise comparisons and therefore more detailed insights—allows us to describe these features of perturbations without ever having to collect information directly at the population level, which could therefore be an efficient and cost-effective tool for research synthesis or the analysis of biomonitoring data. Our work has so far overlooked the temporal dynamics of responses to perturbations. As we only needed to consider the initial and perturbed states of ecosystems for our geometrical approach (perturbations as displacement vectors in Fig. A), we haven’t made the distinction between press and pulse perturbations and we also haven’t considered non-linear responses. However, to consider a community’s trajectory during and after a perturbation, our framework could be applied in future studies to test if response diversity and biomass scaling of perturbations change over time. Practically this would involve comparing the responses of two (or more) aggregate properties to a perturbation over time and checking if there was a change in the proportion of mismatches (e.g. over-replicates in an experimental treatment). For example, if the proportion of mismatches between a set of ecosystem functions was initially very low following a perturbation but then increased over time, this would be consistent with a scenario where most species initially responded negatively to that perturbation but then some species increased in abundance (e.g. due to competitive release). Changes in the proportion of mismatches between diversity and function over time would likewise imply changes in the biomass scaling of a perturbation. It seems likely that this new geometrical perspective could be combined with tools in the ecological stability literature to study dynamic ecological responses to perturbations. In our work, we did not explicitly consider biotic interactions, yet they nonetheless play a role. The state that an ecosystem reaches after a perturbation undoubtedly depends on species interactions, especially if the time scale considered is long enough to allow community dynamics to play out. The classic example is the trophic cascade . If a perturbation directly impacts the top of a food chain (e.g. species invasion), it will in time also affect its base, following the alternating sign pattern characteristic of a cascade. Put in the context of our work, biotic interactions play a role in what we call “features of perturbations” like biomass scaling and response diversity. An exciting future direction would therefore be to seek for recognizable signatures of species interactions in the variations through time of those perturbation features. For instance, we can hypothesise that strong mutualistic interactions would generate increasingly coherent responses as time grows (corresponding to a reduction of response diversity). Furthermore, our framework makes the simplifying assumption that species per capita contributions to functions are fixed, but in reality how a species contributes to a function may be dependent on its interactions with other species (although the fact that our validation test was conclusive implies that this assumption is not a bad approximation). We propose to see our work as a first step of a more general program: using perturbations as “probes”, where ecosystem functions are macroscopic “observables”, to better understand the dynamics of natural ecosystems. Given the generality of our framework, our work touches many areas of contemporary ecology. For multifunctional ecologists, it helps to explain how different functions can respond in different ways to global change . For ecologists interested in multiple perturbations, our work can be used to understand variability in how community-level properties observe the interactions (antagonistic or synergistic) between perturbations . For biodiversity-ecosystem functioning research, the opposing responses of diversity and function to perturbations (which we explained) should be considered when understanding how perturbations influence biodiversity-ecosystem functioning relationships . Our work can be used in disturbance ecology to link studies across disparate systems and may even help to interpret trade-offs between biodiversity and crop yield under different farming practices . When studying complex systems such as ecosystems, it is important to have baseline expectations for their behaviour. We have found that the variability between community-level responses to perturbations does not just limit synthesis and prediction in ecology. Instead, this variability is predictable and can be leveraged to gain useful information about species’ responses to perturbations and species’ contributions to ecosystem functioning. Our work provides a solid platform from which the complexity of community-level responses to anthropogenic global change can be better understood. Geometrical approach To understand what can be learned from the variability of aggregate properties’ responses to perturbations, we transpose the ecological problem to a more abstract, but simpler, geometrical setting (described more formally in Box ). First, we consider the effects of perturbations on populations as displacement vectors in the ecosystem’s state-space, where axes report the biomass of all constituent species (Fig. A). This vector is the difference between initial and perturbed states. It encodes the response to the perturbation at the population level at a given time and can be applied to both press perturbations (where the community may be expected to stay at the perturbed state for some time) and pulse perturbations (where the community may be expected to return to the initial state from the perturbed state). We then see ecosystem functions as positive directions in this same state space (Fig. B). Total biomass for example is the sum of all the species’ biomass and its direction lies exactly between all the axes, giving equal weight to all species. Other functions may not be influenced by the biomass of all species equally. In the hypothetical example shown in Fig. B, general decomposition is slightly more sensitive to the biomass of fungi than to the biomass of bacteria, plastic decomposition is primarily carried out by bacteria, and chemical production is primarily carried out by fungi. In general, a positive direction is spanned by a vector of positive values representing the per-capita contribution of each species to the function of interest. Our approach therefore aligns with Grime’s “biomass-ratio hypothesis” where species contributions to ecosystem functions increase with increasing biomass . The “broadest” function, total biomass, is made up entirely of ones. The “narrowest” functions, are made up entirely of zeroes, except on the entry associated with the only contributing species . Next, we combine these two levels of abstraction to model how functions “observe” perturbations. We recenter the state space so that the axes now represent the response of each species, with the origin consequently being the initial state of the community (Fig. C). Projecting the displacement vector (multi-dimensional vector describing species responses to a perturbation) onto the direction of an ecosystem function (one dimensional vector made up of species contributions to the function) gives the “observation” of that function (see blue and red lines coming from perturbed states A and B in Fig. C). For each function, drawing a line through the origin and perpendicular to the direction of the function delineates two zones. One where the projection is negative, and thus the function observes a negative response and the other where the projection is positive and thus the function observes a positive response. If the two directions associated to the two functions are not perfectly collinear, there will be zones of state-space where responses to perturbations will be qualitatively different when observed by one function or the other. These zones are the two symmetrical cones centred on the origin, formed by the delineation lines of the functions, perpendicular to their respective directions (red zones in Fig. D). The larger the angle between two functions, the larger the zones of mismatches. Consequently, if species’ responses were random and unbiased, the probability of finding a qualitative mismatch between two functions is: 1 \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\mathbb{P}}(\,{\mbox{Mismatch}})=\frac{\theta }{\pi }$$\end{document} P ( Mismatch ) = θ π where θ is the angle between the two functions measured in radians. This collinearity of functions allows us to quantify their similarity. The similarity between functions, defined in this way, is related to their respective broadness, which quantifies the evenness of species per-capita functional contributions . Indeed, in a community of S species and functions f and g : 2 \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\cos \theta \approx \cos {\theta }_{{div}}=\sqrt{{\scriptstyle{2}\atop}\!D(f)/S\,\times {\scriptstyle{2}\atop}\!D(g)/S}$$\end{document} cos θ ≈ cos θ d i v = 2 D ( f ) / S × 2 D ( g ) / S where 1/ S ≤ 2 D ( f )/ S ≤ 1 is the broadness of the function f (same for function g ), defined here as the Gini–Simpson diversity index of the vector of species contributions to the function, and normalized by species richness S . Expression quantifies the intuitive expectation that two broad functions ought to be highly collinear, whereas two narrow functions can be independent (i.e. orthogonal to one another) if they are not performed by the same set of species. There is a straightforward, yet very useful application to this reasoning that we will use in our data analysis: because total biomass is the broadest function by definition (corresponding to a value of 1), we can use the proportion of mismatches P f ,bio between total biomass and a given function f to estimate the latter’s broadness. Indeed, if perturbations are random, we have, for any positive function: 3 \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\cos {({P}_{{f},{{bio}}}\times \pi )}^{2}{=}^{2}D(f)/S$$\end{document} cos ( P f , b i o × π ) 2 = 2 D ( f ) / S We illustrate this relationship between broadness and mismatches with total biomass in Fig. E (inset). Our final level of abstraction is the realization that measures of diversity, which are highly non-linear functions of species biomass (in the mathematical sense of a function of variables, not in the sense of ecological functioning), can still be placed into this geometrical setting by considering their (state-dependent) gradients (outlined in more detail in Box ). The gradient of a diversity metric is a state-dependent vector encoding how small variations in each species’ biomass change that diversity metric. The collinearity between diversity metrics and ecosystem functions can therefore be quantified by measuring the angle between the gradient of a diversity metric and the direction of an ecosystem function. Importantly, gradients of diversity metrics span non-positive directions in state space because increasing the biomass of some species (the more abundant ones) decreases diversity. This allows for the angle between diversity metrics and ecosystem functions to exceed 90°. Box 2 Formalizing the variability of observed responses to a perturbation We formalize the process of observing the ecosystem-level impact of a given perturbation, based on aggregate features of functioning or diversity. Our goal is to explain what controls the probability that two scalar observations of the same perturbed ecosystem give opposite results. Here bold symbols denote S —dimensional vectors, where S is the species richness of the community. Let N c be the initial (or control) state of a community: the vector of species biomass prior to the perturbation. Let N p be the perturbed community state. The observed response, quantified via an ecosystem function f ( N ), is Box 2 Eq 1 \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\Delta f=f({{\boldsymbol{N}}}^{p})-f({{\boldsymbol{N}}}^{c}).$$\end{document} Δ f = f ( N p ) − f ( N c ) . For a linear function, there exists a constant f 0 (because we will consider changes in functioning, and not absolute levels of functioning, this constant will play no role in what follows) and a vector φ —the gradient—such that Box 2 Eq 2 \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$f({{\boldsymbol{N}}})={f}_{0}+\left\langle {{{\boldsymbol{\varphi }}}},{{\boldsymbol{N}}}\right\rangle$$\end{document} f ( N ) = f 0 + φ , N with \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\left\langle \cdot,\cdot \right\rangle$$\end{document} ⋅ , ⋅ the scalar product of vectors. The elements of the gradient vector φ encode the per capita contribution of species to the function. For us it will not matter what those exact contributions are. Only relative species contributions, which determine the direction spanned by the vector φ , are required for our framework. A positive function is such that the elements of the gradient are positive. If we rewrite the response of the function to the perturbation, we get that Box 2 Eq 3 \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\Delta f=\left\langle {{{\boldsymbol{\varphi }}}},{{{\mathbf{\Delta }}}}{{\boldsymbol{N}}}\right\rangle$$\end{document} Δ f = φ , Δ N where Δ N = N p − N c is the vector of population-level responses. For non-linear aggregate properties, such as diversity metrics, the (state dependent) gradient vector can be computed as \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\varphi }_{i}({{{{\boldsymbol{N}}}}}^{c})=\frac{\partial f}{\partial {N}_{i}}{| }_{{{{{\boldsymbol{N}}}}}^{c}}$$\end{document} φ i ( N c ) = ∂ f ∂ N i ∣ N c . In this case, expression (Box 2 Eq 3) will be an approximation, accurate for weak perturbations for which the state-dependent gradient vector is still relevant. Now, for two functions, f , g associated with two directions spanned by the two gradient vectors φ and ϕ , we define their collinearity as the angle 0 ≤ θ < 2 π whose cosine is Box 2 Eq 4 \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\cos \theta=\frac{\left\langle {{{\boldsymbol{\varphi }}}},{{{\boldsymbol{\phi }}}}\right\rangle }{\left\Vert {{{\boldsymbol{\varphi }}}}\right\Vert \left\Vert {{{\boldsymbol{\phi }}}}\right\Vert }$$\end{document} cos θ = φ , ϕ φ ϕ where \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\left\Vert \cdot \right\Vert$$\end{document} ⋅ denotes the Euclidian norm of vectors. A graphical argument (Fig. D) tells us that the fraction of perturbation vectors Δ N that will lead to a mismatch between the observations of f and g is Box 2 Eq 5 \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\mathbb{P}}(\,{{\mbox{sign}}}(\Delta f) \, \ne \ {{\mbox{sign}}}\,(\Delta g))=\frac{\theta }{\pi }$$\end{document} P ( sign ( Δ f ) ≠ sign ( Δ g ) ) = θ π In such cases, one of the functions will observe a positive response, while the other function will observe a negative response. Generically, we can evaluate the cosine of the angle based on a notion of functional broadness. Indeed, given a random choice of positive functions Box 2 Eq 6 \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\frac{\left\langle {{{{\boldsymbol{\varphi}}}}},{{{{\boldsymbol{\phi}}}}}\right\rangle}{\| {{{{\boldsymbol{\varphi}}}}}\| \| {{{{\boldsymbol{\phi}}}}}\| }\approx \frac{1}{S}\frac{{\sum}\varphi_{i}{\sum }\phi_{i}}{\sqrt{{\sum}\varphi_{i}^{2}{\sum }\phi_{i}^{2}}}=\frac{1}{S}\sqrt{\frac{1}{\sum{\left(\frac{\varphi_{i}}{{\sum}\varphi_{i}}\right)}^{2}}\frac{1}{\sum{\left(\frac{\phi_{i}}{{\sum}\phi_{i}}\right)^{2}}}}=\sqrt{\frac{{2\atop}D_{f}}{S}\frac{{2\atop}D_{g}}{S}}$$\end{document} φ , ϕ ∥ φ ∥ ∥ ϕ ∥ ≈ 1 S ∑ φ i ∑ ϕ i ∑ φ i 2 ∑ ϕ i 2 = 1 S 1 ∑ φ i ∑ φ i 2 1 ∑ ϕ i ∑ ϕ i 2 = 2 D f S 2 D g S where q D denotes Hill’s diversity index. We will call the fraction \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\frac{{2\atop}D_{f}}{S}$$\end{document} 2 D f S the broadness of the function f , which is maximal (and equal to one) if all species contribute equally to the function (i.e. total biomass). We can modify the above theory to account for an additional piece of population-level information in the form of a biomass scaling of population-level responses. It is indeed reasonable to expect that more abundant species will, in absolute terms, show a larger response to some types of perturbations (e.g. habitat loss of 50% may decrease biomass of all species by 50%, so the most abundant species will experience the greatest absolute losses). For some scaling exponent α ≥ 0, if we denote Λ the diagonal matrix whose elements are the species biomass prior to the perturbation, we may assume that the perturbation displacement vector takes the form Δ N = Λ α Δ . We then have that Box 2 Eq 7 \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\Delta f=\left\langle {\Lambda }^{\alpha }{{{\boldsymbol{\varphi }}}},{{{\mathbf{\Delta }}}}\right\rangle$$\end{document} Δ f = Λ α φ , Δ the relevant angle to consider then becomes Box 2 Eq 8 \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\cos {\theta }_{\alpha }=\frac{\left\langle {{{\boldsymbol{\varphi }}}},{\Lambda }^{2\alpha }{{{\boldsymbol{\phi }}}}\right\rangle }{\left\Vert {\Lambda }^{\alpha }{{{\boldsymbol{\varphi }}}}\right\Vert \left\Vert {\Lambda }^{\alpha }{{{\boldsymbol{\phi }}}}\right\Vert }$$\end{document} cos θ α = φ , Λ 2 α ϕ Λ α φ Λ α ϕ giving the fraction of rescaled vectors Δ that would lead to a qualitative mismatch. Simulation model for perturbation experiments To test, explore and illustrate the geometrical ideas outlined above, we conducted numerical experiments where ecological communities were perturbed and their responses were observed using different aggregate properties. We did not ask our simulations to have complex, realistic underpinnings. We simply defined a protocol to generate a wide range of initial and perturbed states, and a wide range of aggregate properties (representing ecosystem functions or diversity measures) that we then used to quantify the ecosystem-level impacts of the perturbations. Initial states were vectors N of length S (chosen uniformly between S = 20 and S = 100) whose elements N i are the initial species abundance or biomass. Those were drawn from log-normal distributions with zero mean and standard deviation (uniformly chosen between 1/2 and 2), thus generating a wide range of communities while also mimicking empirical abundance distribution patterns. For each initial state, 500 perturbations were generated as vectors Δ N of length S (perturbed states are N + Δ N ) whose elements Δ N i were generated in the following way. First, for each species, we drew a value x i from a normal distribution with unit standard deviation and mean μ . For a given initial state, μ is a fixed value uniformly chosen between −0.3 and 0.3. It determines the qualitative consistency of population-level responses (more on this below). We then normalized the set of values x i by \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${(\frac{1}{S}\sum {x}_{j}^{2})}^{1/2}$$\end{document} ( 1 S ∑ x j 2 ) 1 / 2 , which gave us a set of values y i that we used to define the actual response of species as 4 \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\Delta {N}_{i}={{{\rm{intensity}}}}\times {y}_{i}\times {N}_{i}^{\alpha }$$\end{document} Δ N i = intensity × y i × N i α For a given perturbation, its intensity was drawn uniformly between 0 and 0.1. We also allowed the impacts of perturbations to scale with the initial abundance (or biomass, in this toy model there is no difference) of species. For each perturbation, the biomass scaling exponent ( α ) was uniformly chosen between 0 and 1. When α = 1, the population response to the perturbation is, on average over the community, proportional to the species initial biomass. The other basic population-level feature that we considered is a notion of response diversity (i.e. whether the perturbation impacted most species positively or negatively). As mentioned above, this feature is set by the parameter μ . Indeed, if we define the population-level response consistency as 5 \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${{{\rm{bias}}}}=| \frac{1}{2}-\frac{\#\{i,\Delta {N}_{i} < 0\}}{S}|$$\end{document} bias = ∣ 1 2 − # { i , Δ N i < 0 } S ∣ ( # denotes the number of elements in a discrete set, here the set of species whose abundances are reduced by the perturbation); then, the expected fraction of negative population responses in the above expression is Φ(− μ ) where Φ( x ) is the cumulative function of a standard normal distribution. Ecosystem functions, which we used to “observe” the ecosystem-level response to perturbations, were represented by positive directions in an S -dimensional space, spanned by vectors φ whose elements φ i represent species’ per-capita functional contributions. For a given state N , its level of functioning is then f ( N ) = ∑ φ i N i (see Box ). The per-capita contributions φ i were drawn from a log-normal distribution with a standard deviation uniformly chosen between 0 and 1.3. When the standard deviation was small, the functions were broad as the per-capita contributions of each species were similar. When the standard deviation was large, however, the functions were more narrow, with a large variation in the per-capita contributions of each species to the function. Diversity metrics were taken from the family of Hill diversity that define the effective number of species as: 6 \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${}^{q}D({{{\boldsymbol{N}}}})={\left({\sum}_{i=1}^{S}{p}_{i}^{q}\right)}^{1/(1-q)}$$\end{document} D q ( N ) = ∑ i = 1 S p i q 1 / ( 1 − q ) where S is richness, p i is the relative abundance (or biomass) of species i and q is the hill number that determines the sensitivity of the diversity index to rare or to abundant species. This general equation encompasses species richness ( q = 0), the Shannon index ( q = 1) and the Gini–Simpson index ( q = 2) , , , . To apply our geometrical framework to diversity observations we considered the directions spanned by their gradients (the vector of partial derivatives \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\frac{{\partial }^{q}D}{\partial {N}_{i}}$$\end{document} ∂ q D ∂ N i ), evaluated at the initial state, which take the form q φ = ( q φ i ) with 7 \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${}^{q}{\varphi }_{i}=\frac{q}{1-q}\left({p}_{i}^{q-1}-{\sum}_{j=1}^{S}{p}_{j}^{q}\right)$$\end{document} φ i q = q 1 − q p i q − 1 − ∑ j = 1 S p j q For each perturbation experiment and each pair of aggregate properties f , g —either two positive linear functions, or a diversity metric and a function (for two diversity metrics see Supplementary Note )—we checked the consistency of their responses. That is, we looked at the sign of f ( N + Δ N ) − f ( N ) and compared it to the sign of g ( N + Δ N ) − g ( N ). If they do not coincide, there is a qualitative mismatch between the two ways of observing the ecosystem’s response to the perturbation. For the simulations, 1000 communities (i.e. initial states) were generated and each one experienced 500 different perturbations. For Figs. E and , two ecosystem functions of varying broadness were generated for each community and used to observe the community-level responses to the perturbations. The angle between the directions defined by the functions was calculated, divided by π (Eq. , and plotted against the realised proportion of mismatches over the 500 perturbations, while recording the relative deviation from the prediction. For Fig. E all perturbations were unbiased at the population level, but for Fig. perturbations could vary in their population-level consistency. The angle between each pair of functions was also estimated using only the knowledge of their broadness based on their mismatches with biomass. For Fig. A, total biomass (positive direction whose elements are all 1) and Hill-Simpson ( 2 D ) were used to observe the ecosystem-level responses to the perturbations. The effective angle between total biomass and the state-dependent gradient of the diversity index, based on (Box Eq 8), was calculated, divided by π , and plotted against the actual proportion of mismatches. Detailed analysis of empirical data Equipped with our geometrical framework for understanding the variability of functional and biodiversity responses to perturbations we can return to the empirical data from Box to uncover novel insights. However, before we can use our framework to learn more about species contributions to ecosystem functions and about the structure of perturbations, we can first confirm that viewing functions as directions and equating their mismatches to their collinearity is a valid approach for a given dataset. To do this we can perform a validation test, formally described in Supplementary Note , where we try to predict the mismatches between two functions (i.e. their collinearity) based on the mismatches between all other pairs of functions. Indeed, if we know the respective angles that two chosen directions make with the remaining set of directions, we should be able to estimate, in a specific way , the angle between the chosen pair. This test involves matrix operations that can introduce artefacts into the results, meaning that an inconclusive test does not necessarily invalidate the application of our framework to a given dataset. However, a conclusive test—mismatches between two functions being well predicted by mismatches between all other pairs of functions—is very strong support for the view of aggregate properties as directions in state space and gives a green light for further exploration of the data using our geometrical arguments. To better understand species’ contributions to ecosystem functions we can use the mismatch data (i.e. matrix in Fig. ) to examine both the similarity of functions and their relative broadness. Firstly, the matrix of mismatches can be used as an adjacency matrix for a network that groups functions based on their similarity. A force-directed layout algorithm, such as the Kamada–Kawai path length cost-function , will generate networks where distance corresponds to the similarity of functions. Secondly, we can use total biomass (the broadest ecosystem function by definition) as a baseline to quantify the broadness of other functions. The angle between total biomass and other broad functions will be small so, over many perturbations with unbiased population-level effects, the proportion of mismatches will therefore be low. Narrower functions will have larger angles with biomass, which will result in higher proportions of mismatches (Fig. ). Working in reverse, we can use the proportion of mismatches between some function and total biomass (directly available from the data) to predict the broadness of that function. For a fair estimate of broadness, the proportion of mismatches between the function and biomass should be quantified over a large pool of perturbations that collectively have random effects. Here, we therefore do not consider perturbations of nutrients for ecosystem functions related to that nutrient—these perturbations have systematic effects rather than random effects—and we only consider cases where the proportion of mismatches between a function and biomass is based on at least twenty perturbations including at least five types of global change factors. So far we have used perturbations to gain insights into species contributions to ecosystem functions. However, we can also use the mismatches between functions to gain useful information about the population-level effects of the perturbations themselves. We can compare the proportion of mismatches between two ecosystem functions (e.g. total biomass and respiration) across different perturbations to quantify the relative response diversity of those perturbations. If perturbations have low response diversity (i.e. most species respond in the same direction), then perturbations will be biased in their directions in state-space towards the fully negative or fully positive areas of state-space (bottom left quadrant or top right quadrant of Fig. D, respectively), and would avoid the cones of mismatches for functions with positive directions. We can therefore use the proportion of mismatches for a given pair of functions to rank perturbations based on their response diversity. We can also use mismatch data to ask if a perturbation’s population-level effects are independent of biomass or if more abundant species have larger absolute changes in biomass (i.e. biomass scaling of a perturbation). If a perturbation causes the biomass of abundant species to decrease, total biomass will decrease but diversity will increase. If on the other hand, a perturbation causes the biomass of abundant species to increase, total biomass will increase but diversity will decrease. As such, when perturbations are scaled by biomass, there will be a higher proportion of mismatches between functions and diversity. As a result, we can use the proportion of mismatches between a function and a diversity metric to rank perturbations based on their biomass scaling. Here, we only made estimates for the response diversity or biomass scaling of perturbations if there were at least five shared observations of those perturbations for the relevant pair of aggregate properties in the dataset. A detailed tutorial, aimed at empirical ecologists interested in applying this geometrical framework to their data, is available at https://jamesaorr.github.io/community-properties-tutorial/ . The tutorial contains useful snippets of code and detailed descriptions of all stages of the analysis from (i) data preparation, (ii) validation test, (iii) exploring species contributions to functions, and (iv) exploring the population-level effects of perturbations. Reporting summary Further information on research design is available in the linked to this article. To understand what can be learned from the variability of aggregate properties’ responses to perturbations, we transpose the ecological problem to a more abstract, but simpler, geometrical setting (described more formally in Box ). First, we consider the effects of perturbations on populations as displacement vectors in the ecosystem’s state-space, where axes report the biomass of all constituent species (Fig. A). This vector is the difference between initial and perturbed states. It encodes the response to the perturbation at the population level at a given time and can be applied to both press perturbations (where the community may be expected to stay at the perturbed state for some time) and pulse perturbations (where the community may be expected to return to the initial state from the perturbed state). We then see ecosystem functions as positive directions in this same state space (Fig. B). Total biomass for example is the sum of all the species’ biomass and its direction lies exactly between all the axes, giving equal weight to all species. Other functions may not be influenced by the biomass of all species equally. In the hypothetical example shown in Fig. B, general decomposition is slightly more sensitive to the biomass of fungi than to the biomass of bacteria, plastic decomposition is primarily carried out by bacteria, and chemical production is primarily carried out by fungi. In general, a positive direction is spanned by a vector of positive values representing the per-capita contribution of each species to the function of interest. Our approach therefore aligns with Grime’s “biomass-ratio hypothesis” where species contributions to ecosystem functions increase with increasing biomass . The “broadest” function, total biomass, is made up entirely of ones. The “narrowest” functions, are made up entirely of zeroes, except on the entry associated with the only contributing species . Next, we combine these two levels of abstraction to model how functions “observe” perturbations. We recenter the state space so that the axes now represent the response of each species, with the origin consequently being the initial state of the community (Fig. C). Projecting the displacement vector (multi-dimensional vector describing species responses to a perturbation) onto the direction of an ecosystem function (one dimensional vector made up of species contributions to the function) gives the “observation” of that function (see blue and red lines coming from perturbed states A and B in Fig. C). For each function, drawing a line through the origin and perpendicular to the direction of the function delineates two zones. One where the projection is negative, and thus the function observes a negative response and the other where the projection is positive and thus the function observes a positive response. If the two directions associated to the two functions are not perfectly collinear, there will be zones of state-space where responses to perturbations will be qualitatively different when observed by one function or the other. These zones are the two symmetrical cones centred on the origin, formed by the delineation lines of the functions, perpendicular to their respective directions (red zones in Fig. D). The larger the angle between two functions, the larger the zones of mismatches. Consequently, if species’ responses were random and unbiased, the probability of finding a qualitative mismatch between two functions is: 1 \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\mathbb{P}}(\,{\mbox{Mismatch}})=\frac{\theta }{\pi }$$\end{document} P ( Mismatch ) = θ π where θ is the angle between the two functions measured in radians. This collinearity of functions allows us to quantify their similarity. The similarity between functions, defined in this way, is related to their respective broadness, which quantifies the evenness of species per-capita functional contributions . Indeed, in a community of S species and functions f and g : 2 \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\cos \theta \approx \cos {\theta }_{{div}}=\sqrt{{\scriptstyle{2}\atop}\!D(f)/S\,\times {\scriptstyle{2}\atop}\!D(g)/S}$$\end{document} cos θ ≈ cos θ d i v = 2 D ( f ) / S × 2 D ( g ) / S where 1/ S ≤ 2 D ( f )/ S ≤ 1 is the broadness of the function f (same for function g ), defined here as the Gini–Simpson diversity index of the vector of species contributions to the function, and normalized by species richness S . Expression quantifies the intuitive expectation that two broad functions ought to be highly collinear, whereas two narrow functions can be independent (i.e. orthogonal to one another) if they are not performed by the same set of species. There is a straightforward, yet very useful application to this reasoning that we will use in our data analysis: because total biomass is the broadest function by definition (corresponding to a value of 1), we can use the proportion of mismatches P f ,bio between total biomass and a given function f to estimate the latter’s broadness. Indeed, if perturbations are random, we have, for any positive function: 3 \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\cos {({P}_{{f},{{bio}}}\times \pi )}^{2}{=}^{2}D(f)/S$$\end{document} cos ( P f , b i o × π ) 2 = 2 D ( f ) / S We illustrate this relationship between broadness and mismatches with total biomass in Fig. E (inset). Our final level of abstraction is the realization that measures of diversity, which are highly non-linear functions of species biomass (in the mathematical sense of a function of variables, not in the sense of ecological functioning), can still be placed into this geometrical setting by considering their (state-dependent) gradients (outlined in more detail in Box ). The gradient of a diversity metric is a state-dependent vector encoding how small variations in each species’ biomass change that diversity metric. The collinearity between diversity metrics and ecosystem functions can therefore be quantified by measuring the angle between the gradient of a diversity metric and the direction of an ecosystem function. Importantly, gradients of diversity metrics span non-positive directions in state space because increasing the biomass of some species (the more abundant ones) decreases diversity. This allows for the angle between diversity metrics and ecosystem functions to exceed 90°. To test, explore and illustrate the geometrical ideas outlined above, we conducted numerical experiments where ecological communities were perturbed and their responses were observed using different aggregate properties. We did not ask our simulations to have complex, realistic underpinnings. We simply defined a protocol to generate a wide range of initial and perturbed states, and a wide range of aggregate properties (representing ecosystem functions or diversity measures) that we then used to quantify the ecosystem-level impacts of the perturbations. Initial states were vectors N of length S (chosen uniformly between S = 20 and S = 100) whose elements N i are the initial species abundance or biomass. Those were drawn from log-normal distributions with zero mean and standard deviation (uniformly chosen between 1/2 and 2), thus generating a wide range of communities while also mimicking empirical abundance distribution patterns. For each initial state, 500 perturbations were generated as vectors Δ N of length S (perturbed states are N + Δ N ) whose elements Δ N i were generated in the following way. First, for each species, we drew a value x i from a normal distribution with unit standard deviation and mean μ . For a given initial state, μ is a fixed value uniformly chosen between −0.3 and 0.3. It determines the qualitative consistency of population-level responses (more on this below). We then normalized the set of values x i by \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${(\frac{1}{S}\sum {x}_{j}^{2})}^{1/2}$$\end{document} ( 1 S ∑ x j 2 ) 1 / 2 , which gave us a set of values y i that we used to define the actual response of species as 4 \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\Delta {N}_{i}={{{\rm{intensity}}}}\times {y}_{i}\times {N}_{i}^{\alpha }$$\end{document} Δ N i = intensity × y i × N i α For a given perturbation, its intensity was drawn uniformly between 0 and 0.1. We also allowed the impacts of perturbations to scale with the initial abundance (or biomass, in this toy model there is no difference) of species. For each perturbation, the biomass scaling exponent ( α ) was uniformly chosen between 0 and 1. When α = 1, the population response to the perturbation is, on average over the community, proportional to the species initial biomass. The other basic population-level feature that we considered is a notion of response diversity (i.e. whether the perturbation impacted most species positively or negatively). As mentioned above, this feature is set by the parameter μ . Indeed, if we define the population-level response consistency as 5 \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${{{\rm{bias}}}}=| \frac{1}{2}-\frac{\#\{i,\Delta {N}_{i} < 0\}}{S}|$$\end{document} bias = ∣ 1 2 − # { i , Δ N i < 0 } S ∣ ( # denotes the number of elements in a discrete set, here the set of species whose abundances are reduced by the perturbation); then, the expected fraction of negative population responses in the above expression is Φ(− μ ) where Φ( x ) is the cumulative function of a standard normal distribution. Ecosystem functions, which we used to “observe” the ecosystem-level response to perturbations, were represented by positive directions in an S -dimensional space, spanned by vectors φ whose elements φ i represent species’ per-capita functional contributions. For a given state N , its level of functioning is then f ( N ) = ∑ φ i N i (see Box ). The per-capita contributions φ i were drawn from a log-normal distribution with a standard deviation uniformly chosen between 0 and 1.3. When the standard deviation was small, the functions were broad as the per-capita contributions of each species were similar. When the standard deviation was large, however, the functions were more narrow, with a large variation in the per-capita contributions of each species to the function. Diversity metrics were taken from the family of Hill diversity that define the effective number of species as: 6 \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${}^{q}D({{{\boldsymbol{N}}}})={\left({\sum}_{i=1}^{S}{p}_{i}^{q}\right)}^{1/(1-q)}$$\end{document} D q ( N ) = ∑ i = 1 S p i q 1 / ( 1 − q ) where S is richness, p i is the relative abundance (or biomass) of species i and q is the hill number that determines the sensitivity of the diversity index to rare or to abundant species. This general equation encompasses species richness ( q = 0), the Shannon index ( q = 1) and the Gini–Simpson index ( q = 2) , , , . To apply our geometrical framework to diversity observations we considered the directions spanned by their gradients (the vector of partial derivatives \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\frac{{\partial }^{q}D}{\partial {N}_{i}}$$\end{document} ∂ q D ∂ N i ), evaluated at the initial state, which take the form q φ = ( q φ i ) with 7 \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${}^{q}{\varphi }_{i}=\frac{q}{1-q}\left({p}_{i}^{q-1}-{\sum}_{j=1}^{S}{p}_{j}^{q}\right)$$\end{document} φ i q = q 1 − q p i q − 1 − ∑ j = 1 S p j q For each perturbation experiment and each pair of aggregate properties f , g —either two positive linear functions, or a diversity metric and a function (for two diversity metrics see Supplementary Note )—we checked the consistency of their responses. That is, we looked at the sign of f ( N + Δ N ) − f ( N ) and compared it to the sign of g ( N + Δ N ) − g ( N ). If they do not coincide, there is a qualitative mismatch between the two ways of observing the ecosystem’s response to the perturbation. For the simulations, 1000 communities (i.e. initial states) were generated and each one experienced 500 different perturbations. For Figs. E and , two ecosystem functions of varying broadness were generated for each community and used to observe the community-level responses to the perturbations. The angle between the directions defined by the functions was calculated, divided by π (Eq. , and plotted against the realised proportion of mismatches over the 500 perturbations, while recording the relative deviation from the prediction. For Fig. E all perturbations were unbiased at the population level, but for Fig. perturbations could vary in their population-level consistency. The angle between each pair of functions was also estimated using only the knowledge of their broadness based on their mismatches with biomass. For Fig. A, total biomass (positive direction whose elements are all 1) and Hill-Simpson ( 2 D ) were used to observe the ecosystem-level responses to the perturbations. The effective angle between total biomass and the state-dependent gradient of the diversity index, based on (Box Eq 8), was calculated, divided by π , and plotted against the actual proportion of mismatches. Equipped with our geometrical framework for understanding the variability of functional and biodiversity responses to perturbations we can return to the empirical data from Box to uncover novel insights. However, before we can use our framework to learn more about species contributions to ecosystem functions and about the structure of perturbations, we can first confirm that viewing functions as directions and equating their mismatches to their collinearity is a valid approach for a given dataset. To do this we can perform a validation test, formally described in Supplementary Note , where we try to predict the mismatches between two functions (i.e. their collinearity) based on the mismatches between all other pairs of functions. Indeed, if we know the respective angles that two chosen directions make with the remaining set of directions, we should be able to estimate, in a specific way , the angle between the chosen pair. This test involves matrix operations that can introduce artefacts into the results, meaning that an inconclusive test does not necessarily invalidate the application of our framework to a given dataset. However, a conclusive test—mismatches between two functions being well predicted by mismatches between all other pairs of functions—is very strong support for the view of aggregate properties as directions in state space and gives a green light for further exploration of the data using our geometrical arguments. To better understand species’ contributions to ecosystem functions we can use the mismatch data (i.e. matrix in Fig. ) to examine both the similarity of functions and their relative broadness. Firstly, the matrix of mismatches can be used as an adjacency matrix for a network that groups functions based on their similarity. A force-directed layout algorithm, such as the Kamada–Kawai path length cost-function , will generate networks where distance corresponds to the similarity of functions. Secondly, we can use total biomass (the broadest ecosystem function by definition) as a baseline to quantify the broadness of other functions. The angle between total biomass and other broad functions will be small so, over many perturbations with unbiased population-level effects, the proportion of mismatches will therefore be low. Narrower functions will have larger angles with biomass, which will result in higher proportions of mismatches (Fig. ). Working in reverse, we can use the proportion of mismatches between some function and total biomass (directly available from the data) to predict the broadness of that function. For a fair estimate of broadness, the proportion of mismatches between the function and biomass should be quantified over a large pool of perturbations that collectively have random effects. Here, we therefore do not consider perturbations of nutrients for ecosystem functions related to that nutrient—these perturbations have systematic effects rather than random effects—and we only consider cases where the proportion of mismatches between a function and biomass is based on at least twenty perturbations including at least five types of global change factors. So far we have used perturbations to gain insights into species contributions to ecosystem functions. However, we can also use the mismatches between functions to gain useful information about the population-level effects of the perturbations themselves. We can compare the proportion of mismatches between two ecosystem functions (e.g. total biomass and respiration) across different perturbations to quantify the relative response diversity of those perturbations. If perturbations have low response diversity (i.e. most species respond in the same direction), then perturbations will be biased in their directions in state-space towards the fully negative or fully positive areas of state-space (bottom left quadrant or top right quadrant of Fig. D, respectively), and would avoid the cones of mismatches for functions with positive directions. We can therefore use the proportion of mismatches for a given pair of functions to rank perturbations based on their response diversity. We can also use mismatch data to ask if a perturbation’s population-level effects are independent of biomass or if more abundant species have larger absolute changes in biomass (i.e. biomass scaling of a perturbation). If a perturbation causes the biomass of abundant species to decrease, total biomass will decrease but diversity will increase. If on the other hand, a perturbation causes the biomass of abundant species to increase, total biomass will increase but diversity will decrease. As such, when perturbations are scaled by biomass, there will be a higher proportion of mismatches between functions and diversity. As a result, we can use the proportion of mismatches between a function and a diversity metric to rank perturbations based on their biomass scaling. Here, we only made estimates for the response diversity or biomass scaling of perturbations if there were at least five shared observations of those perturbations for the relevant pair of aggregate properties in the dataset. A detailed tutorial, aimed at empirical ecologists interested in applying this geometrical framework to their data, is available at https://jamesaorr.github.io/community-properties-tutorial/ . The tutorial contains useful snippets of code and detailed descriptions of all stages of the analysis from (i) data preparation, (ii) validation test, (iii) exploring species contributions to functions, and (iv) exploring the population-level effects of perturbations. Further information on research design is available in the linked to this article. Supplementary Information Reporting Summary Transparent Peer Review file |
Crowdsourced Feedback to Improve Resident Physician Error Disclosure Skills | f647ff66-9d4d-46ce-9a18-b446fe6c912b | 11307134 | Internal Medicine[mh] | Following harmful medical errors, physicians often feel ill-equipped to communicate with patients and families. , , , Incomplete or poor physician communication magnifies the pain and uncertainty experienced by patients and impairs efforts to improve patient safety. , To better prepare physicians, the Accreditation Council for Graduate Medical Education requires that all residents receive training and practice in adverse event disclosure to patients. However, 23% of US residencies provided no such training in 2021. Most other programs provided only informal training or lectures, approaches that are necessary but likely insufficient. Lectures do not ensure communication skill acquisition, and informal training falls short because real-life disclosure is unpredictable and often concludes without formative feedback from supervisors or harmed patients. , , , To supplement lectures and bedside learning, educators need practical tools for residents to practice simulated medical error disclosure and receive reliable, patient-centered formative feedback. The video-based communication assessment (VCA) is software for this purpose, but limited evidence exists regarding its effectiveness. The VCA provides physicians with practice and feedback on their communication skills. It presents videos of vignettes and prompts users to audio-record what they would say to the patient. Recorded responses are rated by web-based panels of laypeople responding as if they were the patient in the scenario. The laypeople are recruited via Amazon Mechanical Turk (MTurk), a crowdsourcing website with a large and diverse participant population. , Physicians receive feedback reports with summary ratings of their performance, average peer scores, learning points derived from raters’ comments, and audio of highly rated peer responses. VCA feedback reports are designed to support self-directed communication skill learning through multiple aspects of deliberate practice. , , First, learning points reinforce desired behaviors and help learners to reconstruct task knowledge around the approach desired by patients. Second, listening to exemplars aids the conceptualization of ideal performance on specific communication subtasks. For example, cases are organized around challenging questions raised by patients that physicians may struggle to address without training and practice (eg, “Why did this happen?” or “Who is going to pay for this care?”). Third, personal ratings help learners to gauge relative performance and determine areas for further practice. In prior studies, , the VCA proved highly acceptable and feasible for preparing learners for common communication scenarios, and raters generated high-quality, actionable feedback. For VCA cases presenting harmful medical errors, panels of crowdsourced laypeople provided ratings that were consistent with those of patients with personal experience with harmful error. In a single-site pre-post pilot study involving paid resident volunteers from 3 specialties, standalone VCA practice without a didactic curriculum was associated with an increase in ratings of residents’ error disclosure skills. Because the effectiveness of the VCA has not been assessed, we sought to test the effect of formative feedback delivered by VCA with a large multisite cohort as part of an error disclosure curriculum. This article describes a randomized clinical trial to test the hypothesis that residents’ error disclosure skills, as assessed by laypeople, would improve after reviewing reports with personal performance feedback and recommendations for effective error disclosure. From July 2022 through May 2023, we conducted a single-blinded, multicenter, randomized clinical trial of the effect of crowdsourced ratings and feedback on postgraduate year 2 (PGY2) internal medicine (IM) and family medicine (FM) resident physicians’ medical error communication skills (see the trial protocol in ). The University of Washington institutional review board ruled this study exempt from review. Participants were not compensated. No VCA results were shared with residency faculty. Risks and benefits were explained verbally; participation was considered to indicate consent. Residents could participate in the training and opt out of research. This report follows the Consolidated Standards of Reporting Trials ( CONSORT ) reporting guideline for randomized studies. Setting and Participants Participants attended IM and FM residencies at 7 US academic medical centers: University of Washington, Seattle (IM and FM); University of Washington, Boise (IM); Washington State University, Everett (IM); Beaumont University (IM at Dearborn and Royal Oak, FM at Wayne and Troy); Dartmouth-Hitchcock Medical Center (IM); University of Massachusetts, Worcester (IM); and Washington University, St. Louis (IM). Each residency participated during a 4- to 8-week window chosen by program leaders to optimize PGY2 residents’ availability. Before the study, none of the residencies provided programwide required error disclosure training. We chose IM and FM residencies because of their large size and shared familiarity with medical cases involving adults. We enrolled only PGY2 residents to control for years of training and simplify scheduling. Residents were eligible for the study if they were on any clinical or nonclinical rotation that provided protected time to attend the teaching conference chosen by their program for VCA practice. Residents were not eligible if they were on leave at the time of the study. Error Disclosure Training Programs assigned all eligible PGY2 residents to attend a 75-minute teaching session at time 1, consisting of 50 minutes of lecture about communication with patients after medical errors, 20 minutes of VCA practice with 2 cases (containing 4 and 3 sequenced vignettes, respectively), and 5 minutes of debrief. At time 2, residents attended a session consisting of 25 minutes of lecture about institutional programs to support clinicians with error disclosure and 20 minutes of VCA practice with 2 additional cases (3 sequenced vignettes each). The recommended duration between time 1 and time 2 was 4 weeks, although the conference schedule at 2 residencies required an interval of 5 to 8 weeks for some residents. The training took place during regularly scheduled conferences for PGY2 residents. The lectures were delivered over video conference by investigators experienced with communication skills training (A.A.W. and T.H.G.). The lecture was adapted from published curricula and modified to highlight site-specific event review policies and clinician support systems. , Residents were encouraged to complete the VCA during the allocated conference time, but could complete it within 5 days if necessary. The study ended when all teaching conferences organized by programs had concluded. Intervention Residents who completed the VCA at time 1 were randomized in 1:1 fashion to either receive feedback before time 2 (intervention) or after time 2 (control) . Block randomization was performed centrally in variable block sizes, before time 1 responses were scored, by a coinvestigator (A.E.D.) with access to lists of the nonidentifying coded usernames of residents who completed time 1. Investigators and raters were blinded to assignments. Residents were unblinded after feedback was released. Intervention residents received automated emails when their feedback was available, instructing them to review it in the application (app) before the next teaching session and VCA practice. Feedback was typically provided 2 weeks after VCA use to allow for completion of rating and data quality checks. Reports presented an interactive feedback display within the VCA app for each vignette . We asked residents receiving the intervention not to discuss feedback with colleagues to avoid contamination. VCA App and Cases The VCA app used in this study has been described previously. , Users entered the app with a personal login and password to access vignettes or review feedback. This study used 4 cases, including 2 previously described cases (a delayed diagnosis of breast cancer and an anticoagulant overdose). , We created 2 new cases depicting a delayed diagnosis of sepsis and the development of a pressure sore (eTable 1 in ). The cases were tested and refined with feedback from 6 faculty members in IM or FM to improve relevance, clarity, and believability. We designed all cases to reflect serious safety events of equivalent preventability and harm severity. Professional actors portrayed each patient or family member. Audio Collection and Rating Residents provided audio responses to each vignette through the VCA software. Audio responses were bundled into rating tasks on MTurk for raters who were US residents aged 18 years or older and able to speak and read English. Raters answered demographic questions, read a vignette description in lay language, viewed the patient video, and listened to resident responses. They rated each response on 6 items covering domains related to accountability, honesty, apology, empathy, caring, and overall response, using a previously described instrument. Items used a 5-point scale anchored with the labels poor, fair, good, very good, and excellent. After rating a set of responses, the rater responded in free text to the question, “What would you want the provider to say if you were the patient in this situation?” A power analysis based on previous research with a moderate η p 2 of 0.09, determined that a sample of 96 PGY2 residents was needed to achieve a power of 0.85 at α = .05 for the analysis of covariance (ANCOVA) to effectively test the study hypotheses. We sought at least 6 raters per response after removing raters with indications of low contributions to reliability. To eliminate inattentive raters from quantitative analysis, open-ended responses were analyzed for quality. One analyst reviewed all responses and flagged responses that bypassed the question (eg, none, good, or NA [not applicable]), were generic, repetitive for multiple vignettes, or were copied and pasted from the ratings task questions (eg, “the provider understood how I was feeling”). A second analyst reviewed and confirmed all exclusions. Resident Surveys Residents completed questionnaires in the VCA application before proceeding to cases. The survey at time 1 asked about age, gender, race, the number of times the resident had personally participated in disclosure of a harmful error to a patient or family, and their highest level of involvement during disclosure of a harmful medical error. Data on race were included in this study because this information would be valuable for future analyses to address racial concordance between users and raters. Before time 2, residents who had received feedback were asked, “Approximately how many minutes did you spend reviewing your feedback?” (response options in 5-minute ranges), “How many of your own responses did you replay?”, and “How many of the exemplar (highly rated peer) responses did you play?” (response options of 0, 1-2, 3-4, and ≥5). Residents responded to 4 additional items about the usefulness of each feedback component (scores, personal recordings, exemplar recordings, and learning points) using a 5-point scale with labels from not at all to extremely. Statistical Analysis Data analysis was performed from July to December 2023. We averaged ratings across items and raters to create an overall rating of each response. We then averaged response ratings across all 7 vignettes at time 1 to create an overall time 1 score, and across all 6 vignettes at time 2 to create a time 2 score. We created a dichotomous disclosure exposure variable by combining disclosure involvement level and the number of times participated in disclosure. To address our primary study question about the effect of the intervention (ie, access to VCA feedback), we conducted a factorial ANCOVA examining the impact that the intervention and prior disclosure exposure had on the primary outcome, time 2 scores, while adjusting for time 1 scores. We conducted a modified intention-to-treat analysis, including all residents with both time 1 and time 2 data. However, those who did not complete time 2 were necessarily excluded from analysis because they did not provide data for the main outcome. Post hoc tests examining the difference between the intervention and control group for each level of prior disclosure exposure were conducted using the Bonferroni correction. We used a Wilcoxon rank sum test to compare performance across specialties on overall scores. We used logistic regression to investigate whether time 1 scores were associated with the likelihood that participants returned for time 2. All statistical analysis was performed in R statistical software version 4.1.2 (R Project for Statistical Computing), with a 2-sided P < .05, except with ANCOVA, which is inherently 1-sided. Participants attended IM and FM residencies at 7 US academic medical centers: University of Washington, Seattle (IM and FM); University of Washington, Boise (IM); Washington State University, Everett (IM); Beaumont University (IM at Dearborn and Royal Oak, FM at Wayne and Troy); Dartmouth-Hitchcock Medical Center (IM); University of Massachusetts, Worcester (IM); and Washington University, St. Louis (IM). Each residency participated during a 4- to 8-week window chosen by program leaders to optimize PGY2 residents’ availability. Before the study, none of the residencies provided programwide required error disclosure training. We chose IM and FM residencies because of their large size and shared familiarity with medical cases involving adults. We enrolled only PGY2 residents to control for years of training and simplify scheduling. Residents were eligible for the study if they were on any clinical or nonclinical rotation that provided protected time to attend the teaching conference chosen by their program for VCA practice. Residents were not eligible if they were on leave at the time of the study. Programs assigned all eligible PGY2 residents to attend a 75-minute teaching session at time 1, consisting of 50 minutes of lecture about communication with patients after medical errors, 20 minutes of VCA practice with 2 cases (containing 4 and 3 sequenced vignettes, respectively), and 5 minutes of debrief. At time 2, residents attended a session consisting of 25 minutes of lecture about institutional programs to support clinicians with error disclosure and 20 minutes of VCA practice with 2 additional cases (3 sequenced vignettes each). The recommended duration between time 1 and time 2 was 4 weeks, although the conference schedule at 2 residencies required an interval of 5 to 8 weeks for some residents. The training took place during regularly scheduled conferences for PGY2 residents. The lectures were delivered over video conference by investigators experienced with communication skills training (A.A.W. and T.H.G.). The lecture was adapted from published curricula and modified to highlight site-specific event review policies and clinician support systems. , Residents were encouraged to complete the VCA during the allocated conference time, but could complete it within 5 days if necessary. The study ended when all teaching conferences organized by programs had concluded. Residents who completed the VCA at time 1 were randomized in 1:1 fashion to either receive feedback before time 2 (intervention) or after time 2 (control) . Block randomization was performed centrally in variable block sizes, before time 1 responses were scored, by a coinvestigator (A.E.D.) with access to lists of the nonidentifying coded usernames of residents who completed time 1. Investigators and raters were blinded to assignments. Residents were unblinded after feedback was released. Intervention residents received automated emails when their feedback was available, instructing them to review it in the application (app) before the next teaching session and VCA practice. Feedback was typically provided 2 weeks after VCA use to allow for completion of rating and data quality checks. Reports presented an interactive feedback display within the VCA app for each vignette . We asked residents receiving the intervention not to discuss feedback with colleagues to avoid contamination. The VCA app used in this study has been described previously. , Users entered the app with a personal login and password to access vignettes or review feedback. This study used 4 cases, including 2 previously described cases (a delayed diagnosis of breast cancer and an anticoagulant overdose). , We created 2 new cases depicting a delayed diagnosis of sepsis and the development of a pressure sore (eTable 1 in ). The cases were tested and refined with feedback from 6 faculty members in IM or FM to improve relevance, clarity, and believability. We designed all cases to reflect serious safety events of equivalent preventability and harm severity. Professional actors portrayed each patient or family member. Residents provided audio responses to each vignette through the VCA software. Audio responses were bundled into rating tasks on MTurk for raters who were US residents aged 18 years or older and able to speak and read English. Raters answered demographic questions, read a vignette description in lay language, viewed the patient video, and listened to resident responses. They rated each response on 6 items covering domains related to accountability, honesty, apology, empathy, caring, and overall response, using a previously described instrument. Items used a 5-point scale anchored with the labels poor, fair, good, very good, and excellent. After rating a set of responses, the rater responded in free text to the question, “What would you want the provider to say if you were the patient in this situation?” A power analysis based on previous research with a moderate η p 2 of 0.09, determined that a sample of 96 PGY2 residents was needed to achieve a power of 0.85 at α = .05 for the analysis of covariance (ANCOVA) to effectively test the study hypotheses. We sought at least 6 raters per response after removing raters with indications of low contributions to reliability. To eliminate inattentive raters from quantitative analysis, open-ended responses were analyzed for quality. One analyst reviewed all responses and flagged responses that bypassed the question (eg, none, good, or NA [not applicable]), were generic, repetitive for multiple vignettes, or were copied and pasted from the ratings task questions (eg, “the provider understood how I was feeling”). A second analyst reviewed and confirmed all exclusions. Residents completed questionnaires in the VCA application before proceeding to cases. The survey at time 1 asked about age, gender, race, the number of times the resident had personally participated in disclosure of a harmful error to a patient or family, and their highest level of involvement during disclosure of a harmful medical error. Data on race were included in this study because this information would be valuable for future analyses to address racial concordance between users and raters. Before time 2, residents who had received feedback were asked, “Approximately how many minutes did you spend reviewing your feedback?” (response options in 5-minute ranges), “How many of your own responses did you replay?”, and “How many of the exemplar (highly rated peer) responses did you play?” (response options of 0, 1-2, 3-4, and ≥5). Residents responded to 4 additional items about the usefulness of each feedback component (scores, personal recordings, exemplar recordings, and learning points) using a 5-point scale with labels from not at all to extremely. Data analysis was performed from July to December 2023. We averaged ratings across items and raters to create an overall rating of each response. We then averaged response ratings across all 7 vignettes at time 1 to create an overall time 1 score, and across all 6 vignettes at time 2 to create a time 2 score. We created a dichotomous disclosure exposure variable by combining disclosure involvement level and the number of times participated in disclosure. To address our primary study question about the effect of the intervention (ie, access to VCA feedback), we conducted a factorial ANCOVA examining the impact that the intervention and prior disclosure exposure had on the primary outcome, time 2 scores, while adjusting for time 1 scores. We conducted a modified intention-to-treat analysis, including all residents with both time 1 and time 2 data. However, those who did not complete time 2 were necessarily excluded from analysis because they did not provide data for the main outcome. Post hoc tests examining the difference between the intervention and control group for each level of prior disclosure exposure were conducted using the Bonferroni correction. We used a Wilcoxon rank sum test to compare performance across specialties on overall scores. We used logistic regression to investigate whether time 1 scores were associated with the likelihood that participants returned for time 2. All statistical analysis was performed in R statistical software version 4.1.2 (R Project for Statistical Computing), with a 2-sided P < .05, except with ANCOVA, which is inherently 1-sided. Participant Characteristics Programs identified 181 PGY2 residents available to attend educational conferences protected for VCA use (25 FM and 156 IM). Of these, 146 completed the VCA at time 1 before randomization (87 [60.0%] aged 25-29 years; 60 female [41.0%]; 77 male [53.0%]; 2 nonbinary [1.0%]) . Of the 146 residents randomized, 103 (70.5%) completed the VCA at time 2 (53 randomized to intervention, and 50 randomized to control). All responses of these 103 residents were rated by at least 6 raters. Of the 43 who only completed time 1, we omitted 10 whose responses were rated by 5 or fewer raters to ensure adequate reliability of scores. shows participants’ demographic characteristics. We recruited 592 raters via MTurk. Of these, 187 (32.0%) were removed for providing poor-quality data consistent with inattentiveness, resulting in a final rater sample of 405 (eTable 2 in ). After removing inattentive raters, each response was rated by 6 to 18 laypeople (mean [SD], 9.50 [1.60] individuals). The 53 participants in the intervention group completed surveys about interacting with the VCA feedback available before time 2. Two surveys lacked data because of electronic storage errors. Of the 51 residents with survey data, 28 (54.9%) reported that they had reviewed their feedback before the survey, reporting variable total periods of time in review; 7 (13.7%) spent less than 5 minutes, 12 (23.5%) spent 6 to 10 minutes, 5 (9.8%) spent 11 to 16 minutes, 3 (5.9%) spent 16 to 20 minutes, and 1 (2.0%) spent 21 to 25 minutes in review. Residents reported listening to variable numbers of their own or exemplar responses, but reported listening to more exemplar responses . Residents rated the usefulness of the 4 feedback components similarly . Communication Rating Outcomes displays the distribution of crowdsourced ratings by intervention assignment (eTable 3 in presents time 1 ratings). High performers were rated 2 points higher than low performers on a 5-point scale. The ANCOVA model, which included time 1 scores as a covariate, showed a significant main effect of the intervention; the mean (SD) time 2 overall scores were 3.26 (0.45) for the intervention group and 3.14 (0.39) for the control group (difference, 0.12; 95% CI, 0.08-0.48; η p 2 = 0.04; P = .01). We also detected a significant interaction between the intervention (ie, feedback availability) and prior exposure to disclosure conversation (η p 2 = 0.05; P = .03) after adjusting for time 1 scores (eTable 4 in ). Post hoc comparisons using Bonferroni correction revealed that when residents had no prior disclosure exposure, those in the feedback intervention group scored significantly higher than those in the control group (mean [SD] score, 3.33 [0.43] vs 3.09 [0.44]; difference, 0.24; 95% CI, 0.01-0.48; P = .007) at time 2. We did not observe a significant difference in communication skill performance between IM and FM residents (mean [SD] score, 3.24 [0.44] vs 3.26 [0.27]). Logistic regression found a significant association between time 1 scores and the likelihood that a participant returned for time 2, such that a 1-unit increase in time 1 scores corresponded to a 2.89-fold increase in the odds of participants completing time 2 (odds ratio, 2.89; 95% CI, 1.06-7.84; P = .04). Programs identified 181 PGY2 residents available to attend educational conferences protected for VCA use (25 FM and 156 IM). Of these, 146 completed the VCA at time 1 before randomization (87 [60.0%] aged 25-29 years; 60 female [41.0%]; 77 male [53.0%]; 2 nonbinary [1.0%]) . Of the 146 residents randomized, 103 (70.5%) completed the VCA at time 2 (53 randomized to intervention, and 50 randomized to control). All responses of these 103 residents were rated by at least 6 raters. Of the 43 who only completed time 1, we omitted 10 whose responses were rated by 5 or fewer raters to ensure adequate reliability of scores. shows participants’ demographic characteristics. We recruited 592 raters via MTurk. Of these, 187 (32.0%) were removed for providing poor-quality data consistent with inattentiveness, resulting in a final rater sample of 405 (eTable 2 in ). After removing inattentive raters, each response was rated by 6 to 18 laypeople (mean [SD], 9.50 [1.60] individuals). The 53 participants in the intervention group completed surveys about interacting with the VCA feedback available before time 2. Two surveys lacked data because of electronic storage errors. Of the 51 residents with survey data, 28 (54.9%) reported that they had reviewed their feedback before the survey, reporting variable total periods of time in review; 7 (13.7%) spent less than 5 minutes, 12 (23.5%) spent 6 to 10 minutes, 5 (9.8%) spent 11 to 16 minutes, 3 (5.9%) spent 16 to 20 minutes, and 1 (2.0%) spent 21 to 25 minutes in review. Residents reported listening to variable numbers of their own or exemplar responses, but reported listening to more exemplar responses . Residents rated the usefulness of the 4 feedback components similarly . displays the distribution of crowdsourced ratings by intervention assignment (eTable 3 in presents time 1 ratings). High performers were rated 2 points higher than low performers on a 5-point scale. The ANCOVA model, which included time 1 scores as a covariate, showed a significant main effect of the intervention; the mean (SD) time 2 overall scores were 3.26 (0.45) for the intervention group and 3.14 (0.39) for the control group (difference, 0.12; 95% CI, 0.08-0.48; η p 2 = 0.04; P = .01). We also detected a significant interaction between the intervention (ie, feedback availability) and prior exposure to disclosure conversation (η p 2 = 0.05; P = .03) after adjusting for time 1 scores (eTable 4 in ). Post hoc comparisons using Bonferroni correction revealed that when residents had no prior disclosure exposure, those in the feedback intervention group scored significantly higher than those in the control group (mean [SD] score, 3.33 [0.43] vs 3.09 [0.44]; difference, 0.24; 95% CI, 0.01-0.48; P = .007) at time 2. We did not observe a significant difference in communication skill performance between IM and FM residents (mean [SD] score, 3.24 [0.44] vs 3.26 [0.27]). Logistic regression found a significant association between time 1 scores and the likelihood that a participant returned for time 2, such that a 1-unit increase in time 1 scores corresponded to a 2.89-fold increase in the odds of participants completing time 2 (odds ratio, 2.89; 95% CI, 1.06-7.84; P = .04). This multisite, randomized clinical trial found that using VCA to provide crowdsourced feedback to PGY2 IM and FM residents about error disclosure skills was associated with an improvement in these skills. Feedback was most impactful among residents who reported they had not been exposed to error disclosure in clinical care, suggesting this intervention could be particularly beneficial at an earlier phase of training. Our findings highlight the potential for the VCA as a scalable practice tool for training that would be logistically challenging to replicate with standardized patients. Despite these encouraging findings, surveys revealed that many residents either did not review or spent minimal time reviewing their feedback, which likely blunted the intervention’s effect. To optimize the VCA’s efficacy, future research should investigate and resolve barriers to residents’ use of crowdsourced feedback. Possible barriers in this trial included the delay between practice and feedback, the lack of protected time to review feedback, a need for adjunctive coaching, unidentified shortcomings of the feedback content and presentation, or the need for more practice repetitions. If confirmed, some of these potential barriers can be addressed with technical or curricular changes, such as providing dedicated time for feedback review or a paired faculty coach. However, using crowdsourcing to incorporate the layperson’s voice in statistically reliable feedback currently requires at least 2 to 3 days, making it difficult to provide instantaneous results. To our knowledge, this study represents the largest assessment of medical error communication skills among IM and FM residents using a standardized instrument. Although all participants received a lecture on practical error disclosure skills, we observed significant variation in their performance, with high performers rated 2 points higher than low performers on a 5-point scale. Self-reported disclosure exposure did not explain this variation. These findings suggest that common teaching approaches leave at least a subset of residents unprepared for effective error disclosure and affirms the Accreditation Council for Graduate Medical Education’s requirement that residents practice these skills. The reliability of VCA scores for comparative assessment may be useful for residency directors evaluating milestone progress within their programs. Yet, the VCA was intended for formative use by individuals and should first be optimized for uses that residents find engaging and psychologically safe. Of particular concern, we found that worse performance at time 1 was associated with not completing time 2. One possible explanation is that participants who found the exercise difficult left discouraged. The departure of individuals with the most room for improvement highlights a difficulty for educators and health system leaders tasked with preparing all physicians for effective error disclosure. Future work should determine approaches that better engage low performers in deliberate practice, including repetition and coaching. Limitations Our work has limitations. First, statistical power was reduced by both nonparticipation with the intervention and dropout before the second VCA use. Second, survey results may be affected by social desirability and recall bias. Third, there is no established score benchmark for competence or mastery, limiting contextualization of the observed effect size. Fourth, we relied on self-report of feedback review, rather than direct measurement; the software does not currently track time spent in feedback activities. Because of the timing of survey administration, residents who reported not reviewing feedback could have theoretically chosen to delay taking the VCA to review feedback instead. However, if this had occurred, it would have diminished, not increased, the effect size. Fifth, reviewing layperson responses to remove those with low contribution to reliability requires effort that may not scale to very widespread use. Sixth, the crowdsourced laypeople were predominantly White and non-Hispanic; lack of racial diversity in the rating pool may introduce unmeasured bias in the results. Seventh, unmeasured confounders missing from analysis may have affected the results. The study has important strengths, including a large geographically diverse cohort with robust participation, suggesting the findings may generalize to other IM and FM residencies. Our work has limitations. First, statistical power was reduced by both nonparticipation with the intervention and dropout before the second VCA use. Second, survey results may be affected by social desirability and recall bias. Third, there is no established score benchmark for competence or mastery, limiting contextualization of the observed effect size. Fourth, we relied on self-report of feedback review, rather than direct measurement; the software does not currently track time spent in feedback activities. Because of the timing of survey administration, residents who reported not reviewing feedback could have theoretically chosen to delay taking the VCA to review feedback instead. However, if this had occurred, it would have diminished, not increased, the effect size. Fifth, reviewing layperson responses to remove those with low contribution to reliability requires effort that may not scale to very widespread use. Sixth, the crowdsourced laypeople were predominantly White and non-Hispanic; lack of racial diversity in the rating pool may introduce unmeasured bias in the results. Seventh, unmeasured confounders missing from analysis may have affected the results. The study has important strengths, including a large geographically diverse cohort with robust participation, suggesting the findings may generalize to other IM and FM residencies. In summary, this study found that self-directed review of crowdsourced feedback was associated with error disclosure skill improvement in IM and FM residents who had already received a lecture on the topic. The VCA has the potential to solve a widely unmet need for graduate medical education patient safety educators. Future work should determine the viewpoints of residency leaders and residents about how the tool can be improved for curricular adoption, and eventually to evaluate its effect on patient-reported communication outcomes. |
Assessing pharmacogenomic literacy in China through validation of the Chinese version of the Minnesota Assessment of Pharmacogenomic Literacy | e68d44df-28d7-4ad1-be0c-ff7ca3f84fb4 | 10651651 | Pharmacology[mh] | Pharmacogenomics (PGx) is a rapidly advancing field that investigates how an individual's genetics affect medication response and tolerability. By tailoring treatment regimens based on this information, PGx may help clinicians optimize clinical outcomes and improve patient care. The growing application of PGx in clinical care is supported by more than 30 years of translational research. This progress is evident in several pioneering healthcare institutions from North America and Europe where PGx has been successfully integrated into routine clinical practice. , These initiatives are strongly backed by the accumulating evidence‐based PGx clinical guidelines and supporting resources developed by professional organizations and consortia, such as the Clinical Pharmacogenetics Implementation Consortium in the United States, the Dutch Pharmacogenetics Working Group, the Canadian Pharmacogenomics Network for Drug Safety, and the French Network of Pharmacogenetics. In the past decade, there has been a notable surge in PGx research specifically targeting non‐European populations. This emphasis has significantly expanded the clinical application of PGx in countries with predominantly non‐European populations, aiming to better serve diverse populations. Converging findings highlight significant differences in the frequency and distribution of genetic variants affecting drug response across different ancestral groups. One striking example is the Han Chinese population, which represents the world's largest ancestral group native to China. Studies have revealed that this population exhibits significantly higher frequencies of actionable genetic variants and altered phenotypes in certain clinically important pharmacogenes compared to non‐Asian populations. , , Prominent examples include CYP2C19 , HLA‐B , and VKORC1 . Overlooking these differences can lead to suboptimal treatment outcomes and elevated risks of adverse reactions. Some implicated medications are commonly prescribed among the Chinese, such as clopidogrel, warfarin, carbamazepine, and allopurinol. , , , , The clinical significance of actionable drug‐gene pairs has recently emerged as a catalyst for the implementation of PGx testing into healthcare systems in China. , , This trend is further compounded by the increasing availability of commercial PGx testing products offered by genetic testing companies. A nationwide survey conducted in 2019 revealed that ~10% of hospitals across the country provide PGx testing services. The majority of these hospitals conduct tests for specific medications, with warfarin and clopidogrel being the most commonly tested. , The extent of clinical implementation, however, varies significantly based on the types of institution and geographic locations, with tertiary hospitals in municipalities and provincial capitals exhibiting higher programmatic utilization of PGx. , , Healthcare professionals in China have increasingly demonstrated a positive attitude toward the clinical application of PGx. , Unfortunately, despite the enthusiasm and potential surrounding PGx, several barriers to its widespread adoption in China still require attention. These obstacles include cost, infrastructural support, awareness, and education, all of which have previously hindered the broader implementation of PGx in other countries and regions, regardless of their current stages of adoption. , Addressing these barriers is crucial for advancing the clinical uptake of PGx in China. To promote the effective adoption of PGx, it is important to raise awareness and provide education to various stakeholders. Research assessing knowledge of PGx is emerging worldwide, including in countries like China, using diverse methodologies ranging from self‐report assessments to customized instruments. , , , , However, these efforts have predominantly focused on evaluating the readiness of health professionals to offer PGx services, rather than assessing the understanding of PGx among patients or test recipients. Shared decision making and counseling about medical tests and treatments are essential components of clinical care and personalizing treatment. In the context of PGx testing, effective patient counseling necessitates that patients have sufficient understanding to actively participate in the decision‐making process of whether to obtain a test and to appreciate potential benefits and limitations. , Recognizing the need for a PGx‐specific instrument to assess patients' knowledge of key PGx concepts, our group recently developed the Minnesota Assessment of Pharmacogenomic Literacy (MAPL). This validated instrument is the first to quantitatively measure the knowledge that test recipients have about PGx in clinical and research settings. It is available for clinical, academic, and research use. Based on findings from a scoping review to identify areas important for patient education followed by a confirmatory focus group assessment, the foundation of the MAPL consists of 15 items designed to assess four primary knowledge domains: underlying concepts, limitations, benefits, and privacy, which are crucial for test recipients to comprehend their results (Table ). Initially developed in English, the instrument was validated among 646 participants in the United States, with subsequent efforts underway to confirm the generalizability of the MAPL and expanding its accessibility across different populations. In this study, we translated and adapted the MAPL into Chinese, considering that Chinese is one of the most widely spoken languages globally. With over 1.3 billion native speakers, representing over 16% of the global population, Chinese holds significant linguistic importance. The translated version of the MAPL TM , named MAPL‐C TM , was validated using a study sample of native Mandarin Chinese speakers. By validating the MAPL‐C, our aim was to bridge the knowledge gap and assess the overall level of PGx literacy among the Chinese public. Additionally, we investigated the potential influences of sociodemographic and clinical characteristics on PGx knowledge. Survey development The approaches that the original MAPL applied to assess patient literacy have been used in assessing medical genetics literacy. The conversion and cultural adaption of all original 15 MAPL items from English to Chinese followed established cross‐cultural translation guidelines. Fifteen items were integrated into a survey for subsequent validation in Chinese. Additional revisions were made based on feedback from a pilot survey prior to the full‐scale administration. These detailed procedures are available in the . The validation survey for MAPL‐C was conducted online using the Research Electronic Data Capture (REDCap) platform. The survey included a total of 42 questions including the 15 prevalidated MAPL‐C items with response options of “Yes,” “No,” and “I don't know.” Twenty‐seven other questions were included to gather participants' sociodemographic and healthcare‐related characteristics, as well as their health literacy assessed with the All Aspects of Health Literacy Scale (see for details). Survey administration Eligibility and recruitment The full‐scale survey administration took place from January 20 to February 24, 2023, utilizing a combination of simple random sampling and snowball sampling techniques through major Chinese social media platforms. The target population consisted of adults who self‐identified as being of Chinese descent and possessed native proficiency in Chinese. To participate in the survey, individuals self‐identified by responding to posters and social media advertisements. A questionnaire to verify eligibility was administered prior to obtaining consent. An introductory screen was presented on their personal devices through a common uniform resource locator, providing an overview of the study, including the inclusion/exclusion criteria and relevant consent information. Exclusion criteria for participation included being less than 18 years of age, not of Chinese descent, lacking native proficiency in Chinese, and being unable or unwilling to provide informed consent in Chinese. After providing consent, participants were instructed to proceed with the survey. Data collection To ensure comprehensive data collection, all questions in the survey were configured as mandatory, requiring participants to answer all fields. All participant responses were collected and securely stored in the REDCap database. To ensure data accuracy and quality, a research staff member verified the collected data. This research project was reviewed and approved by the University of Minnesota Institutional Review Board #00017889 on December 21, 2022. Data analyses Data cleaning Completed survey responses were exported from REDCap and reviewed by research team members. Responses that raised doubts or uncertainties were flagged for further discussion. The final decision regarding inclusion was made based on agreement between the first and senior authors. Responses that exhibited straight‐line patterns (consistently selecting the same response option, such as all “True,” all “False,” or all “I don't know”) across MAPL‐C items were excluded from analyses. Additionally, duplicate survey entries were identified, confirmed, and excluded from the analyses. Please refer to Figure for more information about the data cleaning process. Statistical analyses Participants' characteristics were summarized using descriptive statistics. The proportion of correct responses was calculated to assess the difficulty level and performance of the MAPL‐C items. Each item was scored as 1 for a correct response and 0 for an incorrect or a “do not know” response. Similar to the approach employed with the original MAPL, items with a correct or incorrect rate exceeding 95% were re‐evaluated for inclusion in the final version of the MAPL‐C. This was because they were deemed either “too easy” or “too hard” and lacked the discriminatory capacity to effectively assess a respondent's real knowledge level. A MAPL‐C total score was calculated by summing the scores across individual items, resulting in a composite measure of PGx literacy. The relationships between participants' characteristics and MAPL‐C total scores were examined using paired t ‐tests, analysis of variance, Spearman correlation, and linear regression analyses, depending on the nature of the variables under investigation. Psychometric analysis Tetrachoric correlation was used to assess the correlation across the correctness of individual MAPL‐C items. To explore the latent structure of participants' performance on the MAPL‐C, we initially conducted exploratory factor analysis (EFA) with varimax rotation. This allowed us to identify the underlying relationships between individual MAPL‐C items and empirically determine the number of factors that explain the correlation pattern within the item set. The resulting latent factor structure was subsequently validated using confirmatory factor analysis (CFA) with weighted least square mean and variance adjusted estimation. The fit of the CFA model was evaluated using the root mean square error of approximation (RMSEA; good fit if <0.05, acceptable fit if <0.08), comparative fit index (CFI; good fit if >0.9 and adequate fit if >0.8), and Tucker‐Lewis index (TLI; good fit if >0.9 and mediocre fit if >0.8). The determination of dimensionality (i.e., the optimal number of factors) was obtained through EFA and CFA, which further guided interpretations of internal consistency. The alpha values ranged from 0 to 1 with the value between 0.7 and 0.9 indicating good internal consistency of the instrument. The CFA was performed in Mplus 8 and other analyses were performed in R version 4.2.1 using the following packages (“tidyverse,” “psych,” and “stats”). The approaches that the original MAPL applied to assess patient literacy have been used in assessing medical genetics literacy. The conversion and cultural adaption of all original 15 MAPL items from English to Chinese followed established cross‐cultural translation guidelines. Fifteen items were integrated into a survey for subsequent validation in Chinese. Additional revisions were made based on feedback from a pilot survey prior to the full‐scale administration. These detailed procedures are available in the . The validation survey for MAPL‐C was conducted online using the Research Electronic Data Capture (REDCap) platform. The survey included a total of 42 questions including the 15 prevalidated MAPL‐C items with response options of “Yes,” “No,” and “I don't know.” Twenty‐seven other questions were included to gather participants' sociodemographic and healthcare‐related characteristics, as well as their health literacy assessed with the All Aspects of Health Literacy Scale (see for details). Eligibility and recruitment The full‐scale survey administration took place from January 20 to February 24, 2023, utilizing a combination of simple random sampling and snowball sampling techniques through major Chinese social media platforms. The target population consisted of adults who self‐identified as being of Chinese descent and possessed native proficiency in Chinese. To participate in the survey, individuals self‐identified by responding to posters and social media advertisements. A questionnaire to verify eligibility was administered prior to obtaining consent. An introductory screen was presented on their personal devices through a common uniform resource locator, providing an overview of the study, including the inclusion/exclusion criteria and relevant consent information. Exclusion criteria for participation included being less than 18 years of age, not of Chinese descent, lacking native proficiency in Chinese, and being unable or unwilling to provide informed consent in Chinese. After providing consent, participants were instructed to proceed with the survey. Data collection To ensure comprehensive data collection, all questions in the survey were configured as mandatory, requiring participants to answer all fields. All participant responses were collected and securely stored in the REDCap database. To ensure data accuracy and quality, a research staff member verified the collected data. This research project was reviewed and approved by the University of Minnesota Institutional Review Board #00017889 on December 21, 2022. The full‐scale survey administration took place from January 20 to February 24, 2023, utilizing a combination of simple random sampling and snowball sampling techniques through major Chinese social media platforms. The target population consisted of adults who self‐identified as being of Chinese descent and possessed native proficiency in Chinese. To participate in the survey, individuals self‐identified by responding to posters and social media advertisements. A questionnaire to verify eligibility was administered prior to obtaining consent. An introductory screen was presented on their personal devices through a common uniform resource locator, providing an overview of the study, including the inclusion/exclusion criteria and relevant consent information. Exclusion criteria for participation included being less than 18 years of age, not of Chinese descent, lacking native proficiency in Chinese, and being unable or unwilling to provide informed consent in Chinese. After providing consent, participants were instructed to proceed with the survey. To ensure comprehensive data collection, all questions in the survey were configured as mandatory, requiring participants to answer all fields. All participant responses were collected and securely stored in the REDCap database. To ensure data accuracy and quality, a research staff member verified the collected data. This research project was reviewed and approved by the University of Minnesota Institutional Review Board #00017889 on December 21, 2022. Data cleaning Completed survey responses were exported from REDCap and reviewed by research team members. Responses that raised doubts or uncertainties were flagged for further discussion. The final decision regarding inclusion was made based on agreement between the first and senior authors. Responses that exhibited straight‐line patterns (consistently selecting the same response option, such as all “True,” all “False,” or all “I don't know”) across MAPL‐C items were excluded from analyses. Additionally, duplicate survey entries were identified, confirmed, and excluded from the analyses. Please refer to Figure for more information about the data cleaning process. Statistical analyses Participants' characteristics were summarized using descriptive statistics. The proportion of correct responses was calculated to assess the difficulty level and performance of the MAPL‐C items. Each item was scored as 1 for a correct response and 0 for an incorrect or a “do not know” response. Similar to the approach employed with the original MAPL, items with a correct or incorrect rate exceeding 95% were re‐evaluated for inclusion in the final version of the MAPL‐C. This was because they were deemed either “too easy” or “too hard” and lacked the discriminatory capacity to effectively assess a respondent's real knowledge level. A MAPL‐C total score was calculated by summing the scores across individual items, resulting in a composite measure of PGx literacy. The relationships between participants' characteristics and MAPL‐C total scores were examined using paired t ‐tests, analysis of variance, Spearman correlation, and linear regression analyses, depending on the nature of the variables under investigation. Psychometric analysis Tetrachoric correlation was used to assess the correlation across the correctness of individual MAPL‐C items. To explore the latent structure of participants' performance on the MAPL‐C, we initially conducted exploratory factor analysis (EFA) with varimax rotation. This allowed us to identify the underlying relationships between individual MAPL‐C items and empirically determine the number of factors that explain the correlation pattern within the item set. The resulting latent factor structure was subsequently validated using confirmatory factor analysis (CFA) with weighted least square mean and variance adjusted estimation. The fit of the CFA model was evaluated using the root mean square error of approximation (RMSEA; good fit if <0.05, acceptable fit if <0.08), comparative fit index (CFI; good fit if >0.9 and adequate fit if >0.8), and Tucker‐Lewis index (TLI; good fit if >0.9 and mediocre fit if >0.8). The determination of dimensionality (i.e., the optimal number of factors) was obtained through EFA and CFA, which further guided interpretations of internal consistency. The alpha values ranged from 0 to 1 with the value between 0.7 and 0.9 indicating good internal consistency of the instrument. The CFA was performed in Mplus 8 and other analyses were performed in R version 4.2.1 using the following packages (“tidyverse,” “psych,” and “stats”). Completed survey responses were exported from REDCap and reviewed by research team members. Responses that raised doubts or uncertainties were flagged for further discussion. The final decision regarding inclusion was made based on agreement between the first and senior authors. Responses that exhibited straight‐line patterns (consistently selecting the same response option, such as all “True,” all “False,” or all “I don't know”) across MAPL‐C items were excluded from analyses. Additionally, duplicate survey entries were identified, confirmed, and excluded from the analyses. Please refer to Figure for more information about the data cleaning process. Participants' characteristics were summarized using descriptive statistics. The proportion of correct responses was calculated to assess the difficulty level and performance of the MAPL‐C items. Each item was scored as 1 for a correct response and 0 for an incorrect or a “do not know” response. Similar to the approach employed with the original MAPL, items with a correct or incorrect rate exceeding 95% were re‐evaluated for inclusion in the final version of the MAPL‐C. This was because they were deemed either “too easy” or “too hard” and lacked the discriminatory capacity to effectively assess a respondent's real knowledge level. A MAPL‐C total score was calculated by summing the scores across individual items, resulting in a composite measure of PGx literacy. The relationships between participants' characteristics and MAPL‐C total scores were examined using paired t ‐tests, analysis of variance, Spearman correlation, and linear regression analyses, depending on the nature of the variables under investigation. Tetrachoric correlation was used to assess the correlation across the correctness of individual MAPL‐C items. To explore the latent structure of participants' performance on the MAPL‐C, we initially conducted exploratory factor analysis (EFA) with varimax rotation. This allowed us to identify the underlying relationships between individual MAPL‐C items and empirically determine the number of factors that explain the correlation pattern within the item set. The resulting latent factor structure was subsequently validated using confirmatory factor analysis (CFA) with weighted least square mean and variance adjusted estimation. The fit of the CFA model was evaluated using the root mean square error of approximation (RMSEA; good fit if <0.05, acceptable fit if <0.08), comparative fit index (CFI; good fit if >0.9 and adequate fit if >0.8), and Tucker‐Lewis index (TLI; good fit if >0.9 and mediocre fit if >0.8). The determination of dimensionality (i.e., the optimal number of factors) was obtained through EFA and CFA, which further guided interpretations of internal consistency. The alpha values ranged from 0 to 1 with the value between 0.7 and 0.9 indicating good internal consistency of the instrument. The CFA was performed in Mplus 8 and other analyses were performed in R version 4.2.1 using the following packages (“tidyverse,” “psych,” and “stats”). Characteristics of survey respondents Out of the 1043 completed surveys, high‐quality responses from 959 adult respondents were included in the analyses (Figure ). All respondents were born in China with 96.3% self‐identifying as Han Chinese (Table ). The proportion of men (54.5%) slightly exceeded that of women. The median age was 25 years (interquartile range = 10) and ~71% of participants were less than 30 years of age. Over 97% of participants resided in China, representing all seven geographic regions. Ninety‐five percent of respondents had attained at least some college education, and 22% had received or were pursuing graduate degrees. Approximately 43% identified as students or trainees. Over 30% were involved in healthcare‐related professions. A significant majority (94.6%) reported having health insurance. Excluding medications taken for short‐term use, such as those for treating symptoms of coronavirus disease 2019 (COVID‐19) infection, 45.7% of participants reported taking at least one prescription medication, and 57.8% reported taking at least one non‐prescription medication in the past 30 days (refer to Table for more details). Regarding prior experience with genetic testing, nearly 91% of participants had heard about it, and ~17% had undergone previous genetic testing. Comparisons of MAPL‐C with the original MAPL Fifteen MAPL‐C items are summarized in Table , along with their corresponding English MAPL items. Both manual and computerized‐based backward translations showed a strong similarity to the original MAPL, indicating that the meaning and intent of each item were well‐preserved in the MAPL‐C (Table ). Figure presents the distribution of participants' responses across 15 MAPL‐C items. The difficulty level of each item was determined by the correct answer rate, with a lower rate indicating higher difficulty. Chinese participants' responses to 13 out of 15 MAPL‐C items closely matched the difficulty ranking of the corresponding MAPL items among US participants (i.e., within ±1 ranking difference) across the entire assessment (Figure ). Consistent with the MAPL validation, responses to Items 1 and 6 on the MAPL‐C received correct responses from over 95% of respondents. Consequently, they were determined to be uninformative and excluded from the assessment. This resulted in a 13‐item version of the MAPL‐C in direct alignment with the original MAPL. Similar to previous MAPL validations, the distribution of MAPL‐C total scores exhibited a unimodal symmetric pattern (Figure ). The MAPL‐C total scores shared the same median score of 7 with the MAPL total scores but exhibited higher kurtosis (MAPL‐C: 4.176 vs. MAPL: 2.593). There was no statistically significant difference ( t (1160) = −1.334, p = 0.183) in the average performance between Chinese respondents on the MAPL‐C (total score mean = 6.864, SD = 1.961) and US respondents on the MAPL (total score mean = 7.020, SD = 2.492; Figure ). The MAPL‐C total score exhibited significant correlations with all individual items except for Item 12—“Pharmacogenomic testing will help determine your diagnosis” (Table ). This may be attributed to its low correct response rate (5.03%), which is over five‐fold lower than that of the original MAPL (26.2%). Factor structure of Chinese participants' performance on the MAPL‐C Figure a presents a correlation matrix quantifying pairwise relationships across individual MAPL‐C item scores. Based on the similarities of statistical correlations and domain knowledge, three distinct groups were identified, which was further confirmed by a three‐factor EFA (see Table for factor loadings). All items, except for Item 2 and Item 3, exhibited loadings above 0.45 on their respective factors, signifying at least moderate correlations between individual items and factors. They were retained in the first factor considering their semantic meaning, intent, and prespecified knowledge domains, along with five other items (4, 7, 8, 9, and 15) to represent performance on the concepts and function of PGx. The second factor comprised four items (5, 10, 11, and 12), whereas the third factor included two items (13 and 14), representing participants' performance on limitations and privacy of PGx testing, respectively. To validate the three‐factor solution, CFA was conducted, demonstrating good model performance (RMSEA = 0.044, CFI = 0.888, and TLI = 0.859). Figure illustrates the structural model, with all items showing positive and statistically significant loadings on their respective factors ( p < 0.05). Participants' knowledge of privacy in PGx testing was significantly correlated with their knowledge of concepts and function of PGx (effect size = 2.994, p = 0.003) as well as limitations of PGx testing (effect size = 2.701, p = 0.003). There were no significant correlations between understanding the limitations of PGx testing and knowledge of concepts and function of PGx among Chinese participants. Due to the three‐factor solution and the small number of items within each factor, assessing internal consistency using Cronbach's alpha was not applicable in this study. Assessment of PGx literacy in the Chinese There were no significant differences in PGx literacy based on gender, ethnic groups, health insurance status, or number of medications used in the past month among the participants (Figures ). Age, education attainment, current student or trainee status, involvement in a healthcare‐related profession, and previous experience with genetic testing were each found to be associated with PGx literacy. In a multivariable regression analysis considering these factors together, younger age (under 45 years), pursuing or holding a bachelor's degree or higher, and prior knowledge of genetic testing were all significantly associated with higher levels of PGx literacy among the Chinese individuals assessed here. These associations remained independent of the positive association between overall health literacy and PGx literacy (Table ). Out of the 1043 completed surveys, high‐quality responses from 959 adult respondents were included in the analyses (Figure ). All respondents were born in China with 96.3% self‐identifying as Han Chinese (Table ). The proportion of men (54.5%) slightly exceeded that of women. The median age was 25 years (interquartile range = 10) and ~71% of participants were less than 30 years of age. Over 97% of participants resided in China, representing all seven geographic regions. Ninety‐five percent of respondents had attained at least some college education, and 22% had received or were pursuing graduate degrees. Approximately 43% identified as students or trainees. Over 30% were involved in healthcare‐related professions. A significant majority (94.6%) reported having health insurance. Excluding medications taken for short‐term use, such as those for treating symptoms of coronavirus disease 2019 (COVID‐19) infection, 45.7% of participants reported taking at least one prescription medication, and 57.8% reported taking at least one non‐prescription medication in the past 30 days (refer to Table for more details). Regarding prior experience with genetic testing, nearly 91% of participants had heard about it, and ~17% had undergone previous genetic testing. MAPL‐C with the original MAPL Fifteen MAPL‐C items are summarized in Table , along with their corresponding English MAPL items. Both manual and computerized‐based backward translations showed a strong similarity to the original MAPL, indicating that the meaning and intent of each item were well‐preserved in the MAPL‐C (Table ). Figure presents the distribution of participants' responses across 15 MAPL‐C items. The difficulty level of each item was determined by the correct answer rate, with a lower rate indicating higher difficulty. Chinese participants' responses to 13 out of 15 MAPL‐C items closely matched the difficulty ranking of the corresponding MAPL items among US participants (i.e., within ±1 ranking difference) across the entire assessment (Figure ). Consistent with the MAPL validation, responses to Items 1 and 6 on the MAPL‐C received correct responses from over 95% of respondents. Consequently, they were determined to be uninformative and excluded from the assessment. This resulted in a 13‐item version of the MAPL‐C in direct alignment with the original MAPL. Similar to previous MAPL validations, the distribution of MAPL‐C total scores exhibited a unimodal symmetric pattern (Figure ). The MAPL‐C total scores shared the same median score of 7 with the MAPL total scores but exhibited higher kurtosis (MAPL‐C: 4.176 vs. MAPL: 2.593). There was no statistically significant difference ( t (1160) = −1.334, p = 0.183) in the average performance between Chinese respondents on the MAPL‐C (total score mean = 6.864, SD = 1.961) and US respondents on the MAPL (total score mean = 7.020, SD = 2.492; Figure ). The MAPL‐C total score exhibited significant correlations with all individual items except for Item 12—“Pharmacogenomic testing will help determine your diagnosis” (Table ). This may be attributed to its low correct response rate (5.03%), which is over five‐fold lower than that of the original MAPL (26.2%). MAPL‐C Figure a presents a correlation matrix quantifying pairwise relationships across individual MAPL‐C item scores. Based on the similarities of statistical correlations and domain knowledge, three distinct groups were identified, which was further confirmed by a three‐factor EFA (see Table for factor loadings). All items, except for Item 2 and Item 3, exhibited loadings above 0.45 on their respective factors, signifying at least moderate correlations between individual items and factors. They were retained in the first factor considering their semantic meaning, intent, and prespecified knowledge domains, along with five other items (4, 7, 8, 9, and 15) to represent performance on the concepts and function of PGx. The second factor comprised four items (5, 10, 11, and 12), whereas the third factor included two items (13 and 14), representing participants' performance on limitations and privacy of PGx testing, respectively. To validate the three‐factor solution, CFA was conducted, demonstrating good model performance (RMSEA = 0.044, CFI = 0.888, and TLI = 0.859). Figure illustrates the structural model, with all items showing positive and statistically significant loadings on their respective factors ( p < 0.05). Participants' knowledge of privacy in PGx testing was significantly correlated with their knowledge of concepts and function of PGx (effect size = 2.994, p = 0.003) as well as limitations of PGx testing (effect size = 2.701, p = 0.003). There were no significant correlations between understanding the limitations of PGx testing and knowledge of concepts and function of PGx among Chinese participants. Due to the three‐factor solution and the small number of items within each factor, assessing internal consistency using Cronbach's alpha was not applicable in this study. PGx literacy in the Chinese There were no significant differences in PGx literacy based on gender, ethnic groups, health insurance status, or number of medications used in the past month among the participants (Figures ). Age, education attainment, current student or trainee status, involvement in a healthcare‐related profession, and previous experience with genetic testing were each found to be associated with PGx literacy. In a multivariable regression analysis considering these factors together, younger age (under 45 years), pursuing or holding a bachelor's degree or higher, and prior knowledge of genetic testing were all significantly associated with higher levels of PGx literacy among the Chinese individuals assessed here. These associations remained independent of the positive association between overall health literacy and PGx literacy (Table ). In this study, we successfully adapted and validated the MAPL‐C, derived from the original English MAPL, by using rigorous linguistic translation, cultural adaptation, and psychometric validation processes that are widely recognized and utilized in this type of assessment tools to measure human genetic knowledge and health literacy. , Through this process we confirmed the reliability and cultural appropriateness for assessing knowledge of PGx in native Chinese speakers. Performance on the MAPL‐C was represented by a three‐factor structure, indicating varying levels of understanding of the concepts and function of PGx, limitations of PGx testing, and privacy of PGx testing among the Chinese. Furthermore, we identified several sociodemographic factors, such as age, education, healthcare‐related occupation, and prior experience with genetic testing, that were associated with PGx literacy. The MAPL‐C, therefore, has potential significance in facilitating the development of targeted interventions and educational programs to address the diversity of PGx knowledge gaps and enhance PGx literacy among the Chinese population. Multidimensional PGx literacy in the Chinese population: Variations, knowledge gaps, and cultural considerations The present findings suggest that PGx literacy in the Chinese participants assessed in this study is multidimensional, showing notable variations in knowledge levels across various aspects of PGx. This stands in contrast to the unidimensional structure observed in the performance pattern of US participants on the MAPL. For example, participants demonstrated relatively low performance on average when answering questions related to the limitations of PGx testing, compared to their understanding of underlying concepts and function of PGx, as well as privacy considerations. The correct answer rates for the four items concerning limitations (i.e., 5, 10, 11, and 12) ranged from 5.0% to 14.8%, indicating notable knowledge gaps in this domain. These gaps were also observed among US participants using the English MAPL, but they appear to be more pronounced among the Chinese participants surveyed herein. These findings underscore how PGx knowledge and response patterns may differ across populations. By understanding a patient's misunderstanding and potential confusion, healthcare providers can tailor their education to facilitate informed decision‐making regarding testing and its potential benefits and limitations. This information enables healthcare providers to proactively address any misconceptions through the provision of educational materials and discussions. Low medication literacy among the public may contribute to confusion and misunderstandings regarding certain aspects of PGx literacy among the Chinese population. Recent studies highlight a concerning pattern of insufficient medication literacy among patients in China who receive pharmacotherapy for chronic medical conditions. , , These studies across different disease states consistently reveal that only one‐third of patients possess adequate knowledge about their medications. This percentage is likely to be even lower among the general population, particularly those with less experience with medications. Understanding the complexities of PGx testing and its limitations requires familiarity with medication‐related concepts and terminology. Individuals who have limited knowledge about medications, including the mechanisms of action and the biological factors that can impact effectiveness and tolerability, may encounter difficulties in accurately comprehending the scope and limitations of PGx testing. Furthermore, traditional Chinese medicine (TCM) holds deep cultural significance in China and is practiced alongside with contemporary Western medicine. TCM emphasizes a holistic approach to understanding body function and disease processes, which conceptualizes disease diagnoses and treatment distinctly compared to Western medicine. Pharmacokinetics (PK), the branch of pharmacology that explores how biological factors impact the absorption, distribution, metabolism, and excretion of drugs in the body, is not traditionally emphasized within the framework of TCM. This cultural context may create inherent barriers for some Chinese individuals to recognize and appreciate the role of genetic factors in influencing these PK processes, where PGx has substantial clinical relevance. Enhancing overall medication literacy can be accomplished through direct patient education and medication counseling. In the United States, medication therapy management (MTM) has emerged as a crucial healthcare service where pharmacists provide education and counseling to patients who are on multiple medications to optimize therapeutic outcomes. The increased availability and incorporation of PGx information in MTM enables clinicians to further refine patients' treatment regimens by considering drug‐gene interactions. In China, both MTM and PGx implementation are still in their early stages but experiencing rapid growth. , The exploration of an integrated model that combines these approaches in a culturally sensitive fashion holds great potential to expedite their development and generate synergistic effects, ultimately leading to more optimized and personalized treatment outcomes for patients. Sociodemographic and clinical relationships with PGx literacy When examining the potential influences of sociodemographic and clinical characteristics, we observed that younger age and higher education level exhibited the strongest associations with higher PGx literacy among our Chinese study cohort. These results are consistent with the findings from the US cohort as well as other studies that highlight the relationships between these factors and health literacy across different population groups in various regions of China. , , , In the present study, a statistically significant positive correlation was observed between overall health literacy and PGx literacy among the participants. However, the small effect size (Spearman rho = 0.140, p < 0.001) suggests that this correlation may have limited clinical significance. Although individuals with higher health literacy may have a better grasp of general health information, this does not necessarily translate into adequate knowledge about PGx testing. PGx involves specific knowledge concerning genetics and its impact on medication outcomes, which extends beyond the concepts of general health. On the other hand, a higher level of PGx literacy was observed among participants who have prior knowledge or experience with genetic testing, suggesting that familiarity with genetic testing concepts can contribute to a better understanding of PGx. Tailoring patient educational to PGx literacy In clinical practice, it is important to recognize that having higher health literacy alone does not automatically guarantee sufficient knowledge about all components of PGx testing. Clinicians should address knowledge gaps and misunderstandings surrounding specific PGx knowledge domains through tailored education and counseling. When using the MAPL‐C to assess PGx knowledge in practice, it is conceivable that an individual may lack understanding in certain domains while having good comprehension in others. Therefore, a case‐specific educational approach is necessary. Our population‐level findings reveal that Chinese individuals, in particular, may have a higher tendency to misunderstand the limitations of PGx testing compared to other aspects. Therefore, emphasizing what is within and outside the scope of PGx testing during the counseling section becomes particularly crucial. This approach ensures test recipients acquire necessary knowledge to make informed decisions, accurately interpret the test results, and actively participate in their personalized healthcare plans. Variations of PGx literacy among healthcare professionals in China In addition to age, education, and prior experience with genetic testing, our study also suggests that healthcare‐related occupation is independently associated with higher performance on the MAPL‐C (Table ). However, the difference in overall PGx literacy between healthcare professionals and non‐healthcare individuals was relatively small (mean difference = 0.69, same median score of 7) and unlikely to be clinically relevant (Figure ). It is also important to note that there was a wide range of MAPL‐C scores among the health professionals included in this study. The slightly better PGx literacy observed among healthcare professionals compared to other participants in our study may be attributed to their higher overall health literacy acquired through healthcare education and professional experience (Table ). However, this does not necessarily guarantee they all possess comprehensive PGx knowledge. Recent studies have consistently highlighted that healthcare workers across various disciplines in China perceive and demonstrate their knowledge of PGx to be inadequate. , These findings are in line with our study results, confirming the consistency of self‐assessed and actual knowledge gaps of PGx testing among Chinese health professionals. Therefore, significant efforts should be made to enhance PGx knowledge and competency among Chinese health professionals through ongoing education, training, and professional development programs. This will ensure that health professionals are equipped with the necessary skills and understanding to effectively integrate PGx into routine clinical practice. Study design considerations, limitations, and implications for future endeavors Study design considerations and limitations are important for interpreting and generalizing the findings. Participant enrollment and survey delivery through social media platforms might introduce selection bias, as those who chose to engage with this type of survey delivery may have different characteristics or motivations compared to those who did not participate. The study sample showed an over‐representation of relatively younger individuals and those with higher levels of education, which may limit the generalizability of the findings to the older population. Additionally, the use and engagement with social media can vary across different geographic locations and between urban and rural populations in China, further influencing the representativeness of the study sample. Another limitation is the lack of stratified sampling, which could result in certain geographic regions being over‐represented or under‐represented in the study sample. Although all seven geographic regions in China were included, the sample may not equally represent the actual population distribution. Last, the study primarily relied on cross‐sectional data, making it difficult to establish causality or determine changes in PGx literacy over time. Despite these limitations, this study represents a significant first step in understanding PGx literacy in the Chinese population. Precision medicine has been an integral part of the national agenda, specifically the “Healthy China 2030” initiative, which seeks to address major healthcare challenges in China through the application of precision medicine. This study is in line with that component of the national blueprint, offering valuable insights and serving as a notable example for future research in assessing PGx literacy and implementing community educational programs to promote precision medicine. Moving forward, it will be important for future studies to explore opportunities to engage older populations and expand outreach efforts to more rural geographic locations. By doing so, a more comprehensive understanding of PGx literacy can be achieved, leading to improved healthcare outcomes and better implementation of PGx in clinical practice. In conclusion, we successfully validated the MAPL‐C, a tool to assess knowledge in various aspects of PGx testing among Chinese speakers. The MAPL‐C fills a notable gap in evaluating PGx knowledge within this population, demonstrating effectiveness in determining the PGx literacy of this Chinese cohort, while also identifying response patterns and knowledge gaps. The MAPL‐C holds great promise as a tool for future clinical practice and research, enabling the assessment of PGx literacy and guiding tailored educational interventions to enhance understanding and utilization of PGx testing among Chinese individuals. PGx literacy in the Chinese population: Variations, knowledge gaps, and cultural considerations The present findings suggest that PGx literacy in the Chinese participants assessed in this study is multidimensional, showing notable variations in knowledge levels across various aspects of PGx. This stands in contrast to the unidimensional structure observed in the performance pattern of US participants on the MAPL. For example, participants demonstrated relatively low performance on average when answering questions related to the limitations of PGx testing, compared to their understanding of underlying concepts and function of PGx, as well as privacy considerations. The correct answer rates for the four items concerning limitations (i.e., 5, 10, 11, and 12) ranged from 5.0% to 14.8%, indicating notable knowledge gaps in this domain. These gaps were also observed among US participants using the English MAPL, but they appear to be more pronounced among the Chinese participants surveyed herein. These findings underscore how PGx knowledge and response patterns may differ across populations. By understanding a patient's misunderstanding and potential confusion, healthcare providers can tailor their education to facilitate informed decision‐making regarding testing and its potential benefits and limitations. This information enables healthcare providers to proactively address any misconceptions through the provision of educational materials and discussions. Low medication literacy among the public may contribute to confusion and misunderstandings regarding certain aspects of PGx literacy among the Chinese population. Recent studies highlight a concerning pattern of insufficient medication literacy among patients in China who receive pharmacotherapy for chronic medical conditions. , , These studies across different disease states consistently reveal that only one‐third of patients possess adequate knowledge about their medications. This percentage is likely to be even lower among the general population, particularly those with less experience with medications. Understanding the complexities of PGx testing and its limitations requires familiarity with medication‐related concepts and terminology. Individuals who have limited knowledge about medications, including the mechanisms of action and the biological factors that can impact effectiveness and tolerability, may encounter difficulties in accurately comprehending the scope and limitations of PGx testing. Furthermore, traditional Chinese medicine (TCM) holds deep cultural significance in China and is practiced alongside with contemporary Western medicine. TCM emphasizes a holistic approach to understanding body function and disease processes, which conceptualizes disease diagnoses and treatment distinctly compared to Western medicine. Pharmacokinetics (PK), the branch of pharmacology that explores how biological factors impact the absorption, distribution, metabolism, and excretion of drugs in the body, is not traditionally emphasized within the framework of TCM. This cultural context may create inherent barriers for some Chinese individuals to recognize and appreciate the role of genetic factors in influencing these PK processes, where PGx has substantial clinical relevance. Enhancing overall medication literacy can be accomplished through direct patient education and medication counseling. In the United States, medication therapy management (MTM) has emerged as a crucial healthcare service where pharmacists provide education and counseling to patients who are on multiple medications to optimize therapeutic outcomes. The increased availability and incorporation of PGx information in MTM enables clinicians to further refine patients' treatment regimens by considering drug‐gene interactions. In China, both MTM and PGx implementation are still in their early stages but experiencing rapid growth. , The exploration of an integrated model that combines these approaches in a culturally sensitive fashion holds great potential to expedite their development and generate synergistic effects, ultimately leading to more optimized and personalized treatment outcomes for patients. PGx literacy When examining the potential influences of sociodemographic and clinical characteristics, we observed that younger age and higher education level exhibited the strongest associations with higher PGx literacy among our Chinese study cohort. These results are consistent with the findings from the US cohort as well as other studies that highlight the relationships between these factors and health literacy across different population groups in various regions of China. , , , In the present study, a statistically significant positive correlation was observed between overall health literacy and PGx literacy among the participants. However, the small effect size (Spearman rho = 0.140, p < 0.001) suggests that this correlation may have limited clinical significance. Although individuals with higher health literacy may have a better grasp of general health information, this does not necessarily translate into adequate knowledge about PGx testing. PGx involves specific knowledge concerning genetics and its impact on medication outcomes, which extends beyond the concepts of general health. On the other hand, a higher level of PGx literacy was observed among participants who have prior knowledge or experience with genetic testing, suggesting that familiarity with genetic testing concepts can contribute to a better understanding of PGx. PGx literacy In clinical practice, it is important to recognize that having higher health literacy alone does not automatically guarantee sufficient knowledge about all components of PGx testing. Clinicians should address knowledge gaps and misunderstandings surrounding specific PGx knowledge domains through tailored education and counseling. When using the MAPL‐C to assess PGx knowledge in practice, it is conceivable that an individual may lack understanding in certain domains while having good comprehension in others. Therefore, a case‐specific educational approach is necessary. Our population‐level findings reveal that Chinese individuals, in particular, may have a higher tendency to misunderstand the limitations of PGx testing compared to other aspects. Therefore, emphasizing what is within and outside the scope of PGx testing during the counseling section becomes particularly crucial. This approach ensures test recipients acquire necessary knowledge to make informed decisions, accurately interpret the test results, and actively participate in their personalized healthcare plans. PGx literacy among healthcare professionals in China In addition to age, education, and prior experience with genetic testing, our study also suggests that healthcare‐related occupation is independently associated with higher performance on the MAPL‐C (Table ). However, the difference in overall PGx literacy between healthcare professionals and non‐healthcare individuals was relatively small (mean difference = 0.69, same median score of 7) and unlikely to be clinically relevant (Figure ). It is also important to note that there was a wide range of MAPL‐C scores among the health professionals included in this study. The slightly better PGx literacy observed among healthcare professionals compared to other participants in our study may be attributed to their higher overall health literacy acquired through healthcare education and professional experience (Table ). However, this does not necessarily guarantee they all possess comprehensive PGx knowledge. Recent studies have consistently highlighted that healthcare workers across various disciplines in China perceive and demonstrate their knowledge of PGx to be inadequate. , These findings are in line with our study results, confirming the consistency of self‐assessed and actual knowledge gaps of PGx testing among Chinese health professionals. Therefore, significant efforts should be made to enhance PGx knowledge and competency among Chinese health professionals through ongoing education, training, and professional development programs. This will ensure that health professionals are equipped with the necessary skills and understanding to effectively integrate PGx into routine clinical practice. Study design considerations and limitations are important for interpreting and generalizing the findings. Participant enrollment and survey delivery through social media platforms might introduce selection bias, as those who chose to engage with this type of survey delivery may have different characteristics or motivations compared to those who did not participate. The study sample showed an over‐representation of relatively younger individuals and those with higher levels of education, which may limit the generalizability of the findings to the older population. Additionally, the use and engagement with social media can vary across different geographic locations and between urban and rural populations in China, further influencing the representativeness of the study sample. Another limitation is the lack of stratified sampling, which could result in certain geographic regions being over‐represented or under‐represented in the study sample. Although all seven geographic regions in China were included, the sample may not equally represent the actual population distribution. Last, the study primarily relied on cross‐sectional data, making it difficult to establish causality or determine changes in PGx literacy over time. Despite these limitations, this study represents a significant first step in understanding PGx literacy in the Chinese population. Precision medicine has been an integral part of the national agenda, specifically the “Healthy China 2030” initiative, which seeks to address major healthcare challenges in China through the application of precision medicine. This study is in line with that component of the national blueprint, offering valuable insights and serving as a notable example for future research in assessing PGx literacy and implementing community educational programs to promote precision medicine. Moving forward, it will be important for future studies to explore opportunities to engage older populations and expand outreach efforts to more rural geographic locations. By doing so, a more comprehensive understanding of PGx literacy can be achieved, leading to improved healthcare outcomes and better implementation of PGx in clinical practice. In conclusion, we successfully validated the MAPL‐C, a tool to assess knowledge in various aspects of PGx testing among Chinese speakers. The MAPL‐C fills a notable gap in evaluating PGx knowledge within this population, demonstrating effectiveness in determining the PGx literacy of this Chinese cohort, while also identifying response patterns and knowledge gaps. The MAPL‐C holds great promise as a tool for future clinical practice and research, enabling the assessment of PGx literacy and guiding tailored educational interventions to enhance understanding and utilization of PGx testing among Chinese individuals. All authors wrote the manuscript. L.Z. and J.R.B. designed the research. L.Z., J.R.B., S.Z., and F.W. performed the research. L.Z. analyzed the data. J.A., J.R.B., and A.L.P. developed the original MAPL questions. Information on licensing and terms of use of the MAPL and MAPL‐C instruments may be accessed at https://license.umn.edu/product/minnesota‐assessment‐of‐pharmacogenomic‐literacy‐mapl . This work was supported by the University of Minnesota's College of Pharmacy Hadsall‐Uden Award for Pharmacy Advancement (L.Z.). J.D.A. has served as a consultant to Tempus Labs, Inc., which offers pharmacogenomics testing. J.R.B. has served as a consultant to OptumRx for drug information activities unrelated to pharmacogenomics. All other authors declared no competing interests for this work. Data S1 Click here for additional data file. |
Dimethyl Sulfoxide Dentin Pretreatments Do Not Improve Bonding of a Universal Adhesive in Etch-and-Rinse or Self-etch Modes | 3cf7df0c-25d4-4887-94ec-86e71765e91c | 11734310 | Dentistry[mh] | Specification of Materials, Producing DMSO Solutions shows the materials used in the present study, their composition, and methods of use. DMSO was solubilized in distilled water or ethanol by diluting it in these solutions to obtain a final concentration of 50% (v/v). The rationale for using 50% (v/v) DMSO was based on three studies by Stape et al – which reported significant improvements in resin-dentin interactions when using conventional adhesives. The molecular weight was considered for calculating the appropriate concentration. DMSO and distilled water or ethanol were added to an Eppendorf tube with a pipette, vortexed for 1 min, and used promptly. The pH of the solutions was measured in triplicate with a microelectrode (Model 2A14, Analyser Instrumentação Analítica; São Paulo, SP, Brazil) and a pH meter (Model MPA 210, MS Tecnopon Instrumentação; Piracicaba, SP, Brazil), obtaining pH values of 10.52 for DMSO, 6.64 for DMSO solubilized in water, and 8.74 for DMSO solubilized in ethanol. The pH of the universal adhesive was measured to be pH 2.92. Selection of Teeth, Dentin Blocks, Dentin Pretreatments After approval of the study by the local ethics committee (CAAE 12345519.2.0000.5374/29386619.9.0000.5374), 80 sound, recently extracted human third molars without cracks or any changes in enamel and/or dentin were selected and frozen immediately afterwards. The specimens were obtained by cleaning the teeth externally with periodontal curettes, sectioning the teeth perpendicular to their long axis, and removing the enamel from the occlusal surface to obtain a flat surface of superficial dentin. Next, the surfaces were polished with 600-grit silicon carbide paper disks in a rotating electric polisher (Aropol 2V, Arotec; Cotia, SP, Brazil) to obtain a standardized surface with smear layer formation. The specimens were sectioned again 3 mm below the cementoenamel junction, exposing the pulp chamber. The pulp was cleaned with dentin excavators, and an adhesive (Scotchbond Universal, 3M Oral Care; St. Paul, MN, USA) was applied in the pulp chamber, light polymerizsed with an LED photocuring device (Valo Cordless, Ultradent; South Jordan, UT, USA) and filled with resin composite (Filtek Z350 XT, 3M Oral Care). The dentin blocks were randomly distributed into eight groups to apply the treatments (n = 10). The etch-and-rinse adhesive strategy was used in four of these groups by applying phosphoric acid and the universal adhesive according to the manufacturer’s instructions . In the other four groups, the self-etch adhesive strategy was used. The groups received the following treatments for each adhesive strategy: Control (CON): the universal adhesive was applied according to the manufacturer’s instructions . Photoactivation was performed with an LED photocuring device (Valo Cordless, Ultradent), in standard mode with 1000 mW/cm 2 . Pretreatment with DMSO (DMSO): DMSO was applied with a disposable applicator (Micro Brush, KG Sorensen; Cotia, SP, Brazil) for 60 s, prior to the application of the universal adhesive. – The excess was removed with absorbent paper, and the adhesive was applied. Pretreatment with DMSO solubilized in water (DMSO/water): DMSO was applied with a disposable applicator (Micro Brush) for 60 s, prior to the application of the universal adhesive. – The excess was removed with absorbent paper, and the adhesive was applied. Pretreatment with DMSO solubilized in ethanol (DMSO/ethanol): DMSO was applied with a disposable applicator (Micro Brush) for 60 s, prior to the application of the universal adhesive. – The excess was removed with absorbent paper, and the adhesive was applied. Next, the resin composite was inserted using the incremental technique with increments of up to 2 mm, then light polymerized with the LED curing device (Valo Cordless), in standard mode with 1000 mW/cm 2 . Blocks of 5 mm x 5 mm were obtained using a nanohybrid composite (Filtek Z350 XT, shade A2B, 3M Oral Care). Microtensile Bond Strength Tests The teeth were stored under humid conditions in an incubator at 37°C for 24 h, and then sectioned perpendicular to the dentin/resin interface with a double-sided diamond disk in a cutting machine (Isomet 1000, Buehler; Lake Bluff, IL, USA) to obtain stick-shaped specimns. A slice of each tooth was selected for micromorphological evaluation of the dentin/resin interface. The resulting sticks were stored immersed in simulated body fluid (SBF) solution containing 50 mmol/l HEPES, 5 mmol/l CaCl 2 2H 2 O, 0.001 mmol/l ZnCl 2 , 150 mmol/l NaCl, and 3 mmol/l sodium azide at pH 7.4, in an incubator at 37°C, which was changed weekly. The bonding area of the sticks in mm 2 was measured using a digital caliper (Mitutoyo Sul Americana; Suzano, SP, Brazil). Half of the sticks obtained from each tooth were evaluated after 24 h, while the other half were evaluated after 6 months of storage. The sticks were fixed in a metallic device with cyanoacrylate glue (Super Bonder; Itapevi, São Paulo, Brazil), and subjected to microtensile bond strength testing in a universal testing machine (EZ Test, Shimadzu; Kyoto, Japan) with a 50-N load cell at a speed of 0.5 mm/min until fracture. The microtensile bond strength by group was obtained by averaging the values measured on different sticks obtained from the same tooth, according to the interface area, using the following formula: microtensile bond strength (in MPa) = load (in N)/area (in mm 2 ). Fracture Mode Analysis The surfaces fractured after the microtensile bond strength tests were evaluated using a stereoscopic magnifying glass (EK3ST, Eikonal Equip Óticos e Analíticos; São Paulo, SP, Brazil) at 40X magnification to determine the fracture mode. The fractures were classified as adhesive, cohesive in dentin, cohesive in resin, or mixed. Micromorphological Analysis of the Dentin/Resin Interface The slices of each tooth obtained after cutting the resin/dentin interface were prepared for SEM micromorphological analysis. The surfaces were polished with sandpaper of decreasing abrasive granulations (600- and 1200-grit) (Imperial Wetordry, 3M Oral Care), then felt and polishing pastes of four different granulations. The surfaces were copiously rinsed, then demineralized for 30 s with 6N hydrochloric acid (HCl), and washed again. Then they were deproteinized with 3% sodium hypochlorite solution for 10 min, followed by washing with distilled water for 15 s, drying with absorbent paper, and dehydration in an ascending series of ethanol concentrations (25%, 50%, 75% and 100%). The slices were sputter-coated with gold for 60 s and examined in a scanning electron microscope (Leo 440i, LEO Electron Microscopy; Oxford, Cambridge, UK), at a voltage of 10 Kv and 2000X magnification. Differences in hybrid layer formation were evaluated descriptively according to the groups. Statistical Analysis The data were analyzed using the R program. Initially, descriptive and exploratory analyses were performed. Since the microtensile bond strength data did not meet the assumptions for parametric analysis, generalized linear models were used, considering the design of subdivided plots. Each tooth was considered an experimental unit, and the average values of the sticks per tooth were considered in the analysis. The prematurely failed sticks were assigned a bond strength of zero and included in this analysis. The associations of treatment with premature failure and fracture mode were analyzed using Fisher’s exact test. The significance level for all analyses was set at 5%. shows the materials used in the present study, their composition, and methods of use. DMSO was solubilized in distilled water or ethanol by diluting it in these solutions to obtain a final concentration of 50% (v/v). The rationale for using 50% (v/v) DMSO was based on three studies by Stape et al – which reported significant improvements in resin-dentin interactions when using conventional adhesives. The molecular weight was considered for calculating the appropriate concentration. DMSO and distilled water or ethanol were added to an Eppendorf tube with a pipette, vortexed for 1 min, and used promptly. The pH of the solutions was measured in triplicate with a microelectrode (Model 2A14, Analyser Instrumentação Analítica; São Paulo, SP, Brazil) and a pH meter (Model MPA 210, MS Tecnopon Instrumentação; Piracicaba, SP, Brazil), obtaining pH values of 10.52 for DMSO, 6.64 for DMSO solubilized in water, and 8.74 for DMSO solubilized in ethanol. The pH of the universal adhesive was measured to be pH 2.92. After approval of the study by the local ethics committee (CAAE 12345519.2.0000.5374/29386619.9.0000.5374), 80 sound, recently extracted human third molars without cracks or any changes in enamel and/or dentin were selected and frozen immediately afterwards. The specimens were obtained by cleaning the teeth externally with periodontal curettes, sectioning the teeth perpendicular to their long axis, and removing the enamel from the occlusal surface to obtain a flat surface of superficial dentin. Next, the surfaces were polished with 600-grit silicon carbide paper disks in a rotating electric polisher (Aropol 2V, Arotec; Cotia, SP, Brazil) to obtain a standardized surface with smear layer formation. The specimens were sectioned again 3 mm below the cementoenamel junction, exposing the pulp chamber. The pulp was cleaned with dentin excavators, and an adhesive (Scotchbond Universal, 3M Oral Care; St. Paul, MN, USA) was applied in the pulp chamber, light polymerizsed with an LED photocuring device (Valo Cordless, Ultradent; South Jordan, UT, USA) and filled with resin composite (Filtek Z350 XT, 3M Oral Care). The dentin blocks were randomly distributed into eight groups to apply the treatments (n = 10). The etch-and-rinse adhesive strategy was used in four of these groups by applying phosphoric acid and the universal adhesive according to the manufacturer’s instructions . In the other four groups, the self-etch adhesive strategy was used. The groups received the following treatments for each adhesive strategy: Control (CON): the universal adhesive was applied according to the manufacturer’s instructions . Photoactivation was performed with an LED photocuring device (Valo Cordless, Ultradent), in standard mode with 1000 mW/cm 2 . Pretreatment with DMSO (DMSO): DMSO was applied with a disposable applicator (Micro Brush, KG Sorensen; Cotia, SP, Brazil) for 60 s, prior to the application of the universal adhesive. – The excess was removed with absorbent paper, and the adhesive was applied. Pretreatment with DMSO solubilized in water (DMSO/water): DMSO was applied with a disposable applicator (Micro Brush) for 60 s, prior to the application of the universal adhesive. – The excess was removed with absorbent paper, and the adhesive was applied. Pretreatment with DMSO solubilized in ethanol (DMSO/ethanol): DMSO was applied with a disposable applicator (Micro Brush) for 60 s, prior to the application of the universal adhesive. – The excess was removed with absorbent paper, and the adhesive was applied. Next, the resin composite was inserted using the incremental technique with increments of up to 2 mm, then light polymerized with the LED curing device (Valo Cordless), in standard mode with 1000 mW/cm 2 . Blocks of 5 mm x 5 mm were obtained using a nanohybrid composite (Filtek Z350 XT, shade A2B, 3M Oral Care). The teeth were stored under humid conditions in an incubator at 37°C for 24 h, and then sectioned perpendicular to the dentin/resin interface with a double-sided diamond disk in a cutting machine (Isomet 1000, Buehler; Lake Bluff, IL, USA) to obtain stick-shaped specimns. A slice of each tooth was selected for micromorphological evaluation of the dentin/resin interface. The resulting sticks were stored immersed in simulated body fluid (SBF) solution containing 50 mmol/l HEPES, 5 mmol/l CaCl 2 2H 2 O, 0.001 mmol/l ZnCl 2 , 150 mmol/l NaCl, and 3 mmol/l sodium azide at pH 7.4, in an incubator at 37°C, which was changed weekly. The bonding area of the sticks in mm 2 was measured using a digital caliper (Mitutoyo Sul Americana; Suzano, SP, Brazil). Half of the sticks obtained from each tooth were evaluated after 24 h, while the other half were evaluated after 6 months of storage. The sticks were fixed in a metallic device with cyanoacrylate glue (Super Bonder; Itapevi, São Paulo, Brazil), and subjected to microtensile bond strength testing in a universal testing machine (EZ Test, Shimadzu; Kyoto, Japan) with a 50-N load cell at a speed of 0.5 mm/min until fracture. The microtensile bond strength by group was obtained by averaging the values measured on different sticks obtained from the same tooth, according to the interface area, using the following formula: microtensile bond strength (in MPa) = load (in N)/area (in mm 2 ). The surfaces fractured after the microtensile bond strength tests were evaluated using a stereoscopic magnifying glass (EK3ST, Eikonal Equip Óticos e Analíticos; São Paulo, SP, Brazil) at 40X magnification to determine the fracture mode. The fractures were classified as adhesive, cohesive in dentin, cohesive in resin, or mixed. The slices of each tooth obtained after cutting the resin/dentin interface were prepared for SEM micromorphological analysis. The surfaces were polished with sandpaper of decreasing abrasive granulations (600- and 1200-grit) (Imperial Wetordry, 3M Oral Care), then felt and polishing pastes of four different granulations. The surfaces were copiously rinsed, then demineralized for 30 s with 6N hydrochloric acid (HCl), and washed again. Then they were deproteinized with 3% sodium hypochlorite solution for 10 min, followed by washing with distilled water for 15 s, drying with absorbent paper, and dehydration in an ascending series of ethanol concentrations (25%, 50%, 75% and 100%). The slices were sputter-coated with gold for 60 s and examined in a scanning electron microscope (Leo 440i, LEO Electron Microscopy; Oxford, Cambridge, UK), at a voltage of 10 Kv and 2000X magnification. Differences in hybrid layer formation were evaluated descriptively according to the groups. The data were analyzed using the R program. Initially, descriptive and exploratory analyses were performed. Since the microtensile bond strength data did not meet the assumptions for parametric analysis, generalized linear models were used, considering the design of subdivided plots. Each tooth was considered an experimental unit, and the average values of the sticks per tooth were considered in the analysis. The prematurely failed sticks were assigned a bond strength of zero and included in this analysis. The associations of treatment with premature failure and fracture mode were analyzed using Fisher’s exact test. The significance level for all analyses was set at 5%. There was no significant association between premature failure and pretreatment using either etch-and-rinse or self-etch strategies at either time point (p > 0.05) . There was a higher percentage of premature failure after 6 months of storage. No significant difference was observed among the dentin pretreatments regarding microtensile bond strength for the same bonding strategy (p > 0.05) . However, the etch-and-rinse strategy yielded higher bond strengths than did the self-etch mode (p < 0.05) at both evaluation time points. Microtensile bond strengths were lower at 6 months than at 24 h for the etch-and-rinse strategy and all dentin pretreatments (p < 0.05). No significant association was found between the fracture mode and the dentin pretreatment (p = 0.8172) in the etch-and-rinse strategy at 24 h; the majority of sticks presented adhesive failure . In contrast, the association was significant (p = 0.0222) at 6 months. Although most of the failures were adhesive for all experimental pretreatments (DMSO, DMSO/water or DMSO/ethanol), cohesive resin failures were predominant in the CON group. As for the self-etch strategy, an association existed between the fracture mode and the pretreatment only at 24 h (p < 0.0001) . The majority of failures for the CON group were adhesive (61.9%), whereas pretreatments with DMSO and DMSO/water had a majority of cohesive in resin failures, and pretreatment DMSO/ethanol exhibited mainly mixed failures. At 6 months, there was no significant association between failure type and dentin pretreatment (p = 0.8439). The resin/dentin interface images show that the hybrid layer had longer and more numerous resin tags when the etch-and-rinse strategy was used, and fewer, shorter tags when the self-etch mode was used . Micromorphological differences existed between the dentin pretreatments in each bonding strategy. Simplified protocols for universal adhesives and bonding agents have been developed to achieve greater stability of the hybrid layer, by promoting more effective infiltration of the adhesive resin monomers between the collagen fibrils demineralized by the adhesive. , , Solutions containing DMSO have been found to be effective in enhancing the penetration ability of adhesive monomer into the dentin matrix of conventional adhesives, , , and increasing bond strength to dentin. , , However, the present study showed that DMSO (whether or not solubilized in water or ethanol) did not promote higher bond strength to dentin by using a universal adhesive with etch-and-rinse and self-etch strategies. Thus, the results failed to reject the first null hypothesis of the study. The solvents (ethanol and water) in universal adhesives both provide and promote penetration of the adhesive’s resin monomers by eliminating dentin moisture trapped among the collagen fibers after the volatilization process. In this respect, the presence of water in the dentin layer is required to form a stable hybrid layer. On the other hand, water is also responsible for the long-term hydrolysis and deterioration of the bonded interface. , Recognizing that the use of the etch-and-rinse strategy leaves more residual water in the hybrid layer than does the self-etch strategy, the use of pretreatments with DMSO, whether solubilized in water or ethanol or not solubilized, could lead to the replacement or displacement of residual water more effectively, in a relatively short application period. , , DMSO competes with water molecules in collagen interpeptide hydrogen bonding, and increases collagen interfibrillar spacing. However, the dentin pretreatment with DMSO (solubilized in water or ethanol or not solubilized) in the present study may made the evaporation of these solvents more difficult, since DMSO is a polar aprotic compound that can also absorb water, because it is characteristically attracted to hydrogen molecules. Since the amount of water and ethanol in Scotchbond Universal adhesive is about 30%, the use of a dentin pretreatment containing yet another a solvent in addition to DMSO, water and ethanol, may have hindered evaporation of the solvents in both strategies in the universal adhesive, since the hydrophobic and hydrophilic content were applied in the same procedure (unlike a conventional 3-step adhesive). Whether solubilized in water or ethanol, pretreatment with DMSO may have contributed to forming residual moisture, which not only impairs volatilization of the DMSO solvent, but fails to improve the bond strength of the respective groups using either strategy. The same outcome could be expected even if pretreatment were applied in a manner to allow volatilization of the solvents, eg, by performing circular scrubbing movements for 60 s. – Higher microtensile bond strength was achieved at both time periods when the etch-and-rinse strategy was used, as corroborated in the studies by Wagner et al, Luque-Martinez et al, and Dačić et al. The micromorphology of the interface shows the presence of longer tags and a thicker hybrid layer, since the acid conditioning increases the surface energy and removes the smear layer. On the other hand, the universal adhesive used is classified as “ultra-mild,” with a pH of 2.92. It promotes the demineralization of superficial dentin, which allows maintaining the part of the hydroxyapatite that is bound to collagen fibers, and also chemical bonding to the functional monomer. , Although resin tags are known not to contribute significantly to resin-dentin bonding strength, a higher frequency of premature failures can be expected when the self-etch strategy vs the etch-and-rinse strategy is used, because of the thinner hybrid layer, especially at the 24-h time point. After 6 months of storage, a higher frequency of premature failures was observed in both strategies, owing to the hydrolytic and enzymatic degradation promoted by the storage medium. This can be expected especially because of the hydrophilic character of the monomeric units of the polymers in the adhesive, including hydroxyethyl methacrylate (HEMA), which favors water sorption. After 6 months of storage, a greater number of cohesive failures in resin were also observed, as a result of hydrolytic degradation caused by water sorption. After 6 months of storage, the resin composite absorbed water and released unpolymerized monomers that did not react after photocuring. This may have compromised not only the physical and mechanical properties of the resin composite, owing mostly to the breaking of the hydrolytic bond between the silane and the inorganic particles, but also the cohesive strength of the resin. After 6 months of storage, the hydrolytic and proteolytic degradation caused by the storage medium when using the etch-and-rinse strategy was significant, resulting in lower bond strengths than those observed at 24 h. DMSO is known to decrease collagen degradation in the hybrid layer by inhibiting the activity of MMP-2 and MMP-9. , Nevertheless, the storage medium used (SBF), containing calcium and zinc, may have promoted higher MMP activity, and consequent loss of stiffness and solubilization of the dentinal matrix collagen. This is especially relevant to the etch-and-rinse strategy, which denudes the collagen fibrils, and may provide greater enzymatic activity. When performing acid etching, smear layer removal exposes the dentin collagen, and leads to increased dentin permeability. This makes the bonding interface vulnerable to hydrolytic and enzymatic degradation. The residual water and non-evaporated solvent in the adhesive could remain trapped after polymerization, thus compromising the bond strength and the mechanical properties of the hybrid layer. However, pretreatments with DMSO did not benefit bond strength longevity (it remained the same as the control group). These results failed to reject the second null hypothesis of the present study. Regarding the self-etch strategy, the fact that bond strengths were maintained over time can be attributed to the 10-MDP functional group interacting ionically with the calcium in the dentin, and forming a relatively long 10-MDP-calcium carbonyl chain hydrophobically on the dentin surface. , This chain plays an important role in increasing the durability of the universal adhesive dentin bond in the self-etch strategy. , Stape et al also performed pretreatment with 50% DMSO incorporated with ethanol or water in a universal adhesive using a self-etch strategy, with no significant influence on the microtensile bond strength. This can be explained by the reduced availability of crosslinked dimethacrylate monomers in universal adhesives, which penetrate better when used with DMSO, but which have limited bond strength to the same extent as they do with a three-step etch-and-rinse adhesive. However, treatment with DMSO, whether solubilized in water or ethanol or not solubilized, influenced the failure mode when the self-etch strategy was used. This was reflected in the higher prevalence of cohesive resin failures in the self-etch group than in the control group after 24 h, thereby leading to the rejection of the third null hypothesis of the study. The DMSO groups (whether or not solubilized in water or ethanol) had a higher prevalence of cohesive resin failures, which may suggest that pretreatments with DMSO led to an improvement in hybrid layer integrity, as reported by Stape et al. Although pretreatment with DMSO, whether solubilized in water or ethanol or not solubilized, has been found to improve the encapsulation of collagen fibers as well as the quality of the collagen-resin interface, , , , , – , our study suggests that no benefit is derived from using DMSO as a pretreatment for a universal adhesive with either etch-and-rinse or self-etch strategies, since our study showed that there was no increase in bond strength to dentin, not even after 6 months. The use of dentin pretreatments containing DMSO did not improve the bonding or the micromorphology of a universal adhesive system in etch-and-rinse or self-etch modes. |
Efficacy of human umbilical cord mesenchymal stem cell in the treatment of neuromyelitis optica spectrum disorders: an animal study | 6f29535e-b7a8-4296-8950-3a068be8fa27 | 11806600 | Surgical Procedures, Operative[mh] | Neuromyelitis optica spectrum disorder (NMOSD) is a chronic, severe, autoimmune demyelinating disease of the central nervous system (CNS) that primarily affects the spinal cord and optic nerves . It is characterized by recurrent optic neuritis and longitudinally extensive transverse myelitis . NMOSD is a rare disease with a global incidence of approximately 0.039–0.73/100,000 and is more prevalent in young and middle-aged females, with a male-to-female incidence ratio of about 1:9 . Although the incidence of NMOSD is low, it is associated with high relapse rates and disability. Disease relapse leads to aggravated neurological dysfunction and decreased quality of life, which imposes a heavy burden on both the individual and society . In most cases, NMOSD is caused by autoantibodies targeting the aquaporin 4 water channel (AQP4), which is mainly expressed on astrocytes in the CNS . Due to impaired immune tolerance, AQP4 is recognized as a foreign antigen, and naïve B cells are abnormally activated, transforming into plasma cells and secreting AQP4 IgG . AQP4 IgG enters the CNS and binds to the AQP4 antigen on the surface of astrocytes, leading to antibody- and complement-dependent cytotoxicity . Simultaneously, the inflammatory process also damages surrounding oligodendrocytes and neurons, leading to demyelination and axonal damage . NMOSD therapy includes acute treatment and prevention of long-term relapse. Long-term relapse prevention is important in NMOSD treatment to decrease the relapse rate and inhibit aggravation of neurological dysfunction . Most drugs, such as azathioprine and mycophenolate work by suppressing the immune response. There are also an increasing number of monoclonal drugs targeting B cells, complement activation pathway, IL-6 pathway, et al. . Despite the wide variety of drugs used to prevent recurrence, some patients experience relapse, leading to severe functional disabilities. Moreover, most current drugs are immunosuppressive, which may cause side effects, such as infection . Therefore, it is important to explore novel drugs for the treatment of NMOSD patients. Mesenchymal stem cells (MSCs) are stem cells derived from the mesoderm with self-renewal capacity and multipotency. MSCs are widely distributed and can be found in the umbilical cord, bone marrow, and adipose tissue, et al. . In recent years, MSCs have been increasingly explored for the treatment of autoimmune diseases because of their low immunogenicity, anti-inflammatory and anti-apoptotic features, and their safety and efficacy have been explored in basic research and clinical studies of autoimmune diseases, such as systemic lupus erythematosus, type I diabetes, and multiple sclerosis . Human umbilical cord mesenchymal stem cells (hUC-MSCs) are promising for clinical applications owing to the following advantages: (1) wide range of sources and easy access, (2) weaker immunogenicity; (3) stronger immunomodulatory effects and (4) stronger proliferation ability . Two pilot, observational clinical studies have explored the safety and efficacy of administering bone marrow MSCs or hUC-MSCs in NMOSD patients and showed potential therapeutic effects . However, evidence regarding the efficacy and mechanism of hUC-MSC therapy in NMOSD animal model remains insufficient. In this study, we aimed to investigate the therapeutic effects of hUC-MSCs in a systemic NMOSD mouse model. We successfully established an NMOSD mouse model and first applied hUC-MSCs therapy in this model. Our results demonstrated that hUC-MSCs significantly alleviated motor disorders and pathological manifestations as well as decreasing spinal cord inflammatory cell infiltration and blood–brain barrier (BBB) disruption. Additionally, our study revealed that hUC-MSCs inhibited AQP4 IgG- and complement-induced astrocyte apoptosis for the first time. Collectively, our study highlights that hUC-MSC treatment may be an effective therapy for NMOSD. Animals In this study, 8–10 weeks-old female C57BL/6 mice were purchased from Charles River and housed in the animal facilities of Shanghai Model Organisms. All mice were housed in a temperature- and humidity-controlled room and were kept in regular cages under a 12/12 h light/dark cycle. All mice were fed with standard food and water. No sample size calculation was performed; sample sizes were chosen based on previous in vivo experiments done in the laboratory to allow for statistical power. Mice were allocated to groups randomly using a random number generator. The order of treatments and measurements for each mouse were randomly assigned. A total of 65 C57BL/6 mice were used. 12 C57BL/6 mice were used for the purpose of establishing the NMOSD mice model (EAE: n = 6; NMOSD: n = 6) and 53 C57BL/6 mice were used for the purpose of assessing the therapeutic effects of hUC-MSCs (CFA + PTX: n = 12, NMOSD + PBS: n = 18; NMOSD + MSC: n = 23). All animals were used in the analysis. During the outcome evaluation and data analysis stage, the group allocation was made known. All procedures have been approved by the ethics committee of Shanghai Model Organisms (IACUC 2024-0021-1). The work has been reported in line with the ARRIVE guidelines 2.0. Extraction of human IgG In this study, the plasma exchange fluid of an NMOSD patient and the plasma of a healthy donor were collected. The patient was in the acute phase of an attack and received double-filtration plasmapheresis. A total of 200 ml plasma exchange fluid was collected. The informed consents were obtained. The plasma exchange fluid and the plasma were diluted 1:10 and 1:6 respectively with 1 × Melon Gel Purification Buffer (Thermo Fisher, USA). IgG was extracted using IgG purification kit (Thermo Fisher, USA) according to instructions. The concentration of IgG from plasma exchange fluid of NMOSD patients or plasma of healthy donors was quantified using BCA Protein Assay Kits (Thermo Fisher, USA) and the concentration is 9–10 mg/ml. The tilter of AQP4 IgG is 1:1000 using cell-based indirect immunofluorescence assays in Simplegen institution. Then purified IgG was inactivated at 56 °C. Finally, IgG was filtered with 0.22 μm sterile filters and stored at −80 °C. The extracted antibody was used within 1 year after extraction. hUC-MSCs acquisition and characterization hUC-MSCs were purchased from Guangzhou SALIAI Stem cell Science and Technology Co.Ltd. The processes including isolation, culture, identification, quality control, and storage of hUC-MSCs were strictly conformed to standard operating procedures. P6 hUC-MSCs were used for in vivo and in vitro studies. All hUC-MSCs used in this study were from one healthy donor. The characterization of hUC-MSCs was assessed by flow cytometer. Briefly, P6 hUC-MSCs were collected and labeled with the following antibodies: FITC-anti-HLA-DR, -anti-CD45, -anti-CD73, -anti-CD34, APC-anti-CD90, -anti-CD19, -anti-CD105, -anti-CD11b (Biolegend, USA, 1:400) at 4 °C for 30 min. After two washing steps, cells were acquired with a flow cytometer (BD FACS CELESTA, USA) and analyzed with Flowjo 10 software. Construction of EAE/NMOSD model and MSC treatment NMOSD mice model was constructed by continuous intraperitoneal injection of AQP4 IgG into EAE mice. For EAE model, 0.3 mg/ml MOG 35–55 (GL Biochem, China) was emulsified in Complete Freund’s Adjuvant (CFA) containing 8 mg/ml H37Ra Mycobacterium tuberculosis (BD DIFCO, USA). Mice were anesthetized by intraperitoneal injection of 2.5% Avertin (Meilunbio, China). Then 200 μl emulsion was injected subcutaneously at two sites in the right and left flank at once. 200 ng pertussis toxin (PTX) (Sigma-Aldrich, USA) in 100ul PBS was injected intraperitoneally at day 0 and day 2. For NMOSD model, EAE mice were intraperitoneally injected with 400 μl AQP IgG from day 7 to day 18. On day 10 and day 14, 10 6 hUC-MSCs prepared in 200 μl PBS were injected into NMOSD mice intravenously. Control mice received 200 μl PBS. Animals’ clinical scores were assigned as follows: 0 = normal; 0.5 = partial tail paralysis; 1 = entire tail paralysis; 1.5 = entire tail paralysis with unsteady gait; 2 = partial paralysis in hind limbs with the body below the abdomen powerless to land; 2.5 = entire paralysis in one hind limb; 3 = entire paralysis in one hind limb with partial paralysis in another hind limb; 3.5 = entire paralysis in two hind limbs; 4 = entire paralysis in two hind limbs with one forelimb paralysis; 4.5 = entire paralysis in four limbs; 5 = death. On day 19, mice were anesthetized with Avertin and euthanized by cardiac perfusion with PBS. Pathological tissues were obtained for HE staining, LFB staining, immunofluorescence staining and TUNEL staining after PBS and 4% paraformaldehyde perfusion. Spinal cord for protein and RNA extraction was obtained after PBS perfusion. HE staining and LFB staining Paraffin sections (2 μm) of lumbar spinal cord was used for HE and LFB staining according to instructions respectively. Briefly, after being dewaxed and rehydrated, the sections were sequentially placed in eosin solution and hematoxylin solution for HE staining, and were incubated in LFB solution at 56 °C overnight and then hydrolyzed for LFB staining. The complete spinal cord was scanned using high throughput digital slice scanning system (Hamamatsu, Japan) and images were obtained under 5× magnification using NDP.view2 software. For HE staining, the level of inflammatory infiltration was assessed as follows: 0: no inflammatory infiltration; 1: only a slight infiltration of inflammatory cells around the blood vessels or spinal cord capsule; 2: small amount of inflammatory cell infiltration in the spinal cord parenchyma; 3: moderate infiltration of inflammatory cells in the spinal cord parenchyma; 4: large infiltration of inflammatory cells in the spinal cord parenchyma. For LFB staining, the area of demyelinating area and white matter were measured using Image J software. Immunofluorescence staining After being dewaxed and rehydrated, the sections of lumbar spinal cord were incubated with Tris-EDTA antigen retrieval solution (Solarbio, China) for 30 min at 100 °C and naturally cooled to room temperature. Then, sections were blocked with QuickBlock™ Blocking Buffer (Beyotime, China), incubated with primary antibody overnight at 4 °C and secondary antibody for 2 h at room temperature. Cell nuclei were stained with DAPI (Beyotime, China). The following primary antibodies were used: rabbit anti-AQP4 (1:800, Sigma-Aldrich, USA, MABN2527); mouse anti-glial fibrillary acidic protein (GFAP) (1:400, Cell Signaling Technology, USA, 3670 T); rabbit anti-myelin basic protein (MBP) (1:200, Abcam, USA, ab7349); rabbit anti-NeuN (1:400, Abcam, USA, ab177487). The following fluorescent secondary antibodies were used: goat anti-mouse 488 (1:1000, Beyotime, China); goat anti-rabbit 488 (1:1000, Beyotime, China); goat anti-rabbit 555 (1:1000, Thermo Fisher, USA). Sections were observed by fluorescence microscope (Olympus, USA). The loss area of AQP4, GFAP, MBP and the number of NeuN+ cells were measured using Image J software. TUNEL staining After being dewaxed and rehydrated, the sections of lumbar spinal cord were incubated with 20 μg/ml proteinase K (Beyotime, China). TUNEL working solution was prepared according to instructions (Beyotime, China). Then, sections were incubated with working solution for 1 h at 37 °C. Cell nuclei were stained with DAPI (Beyotime, China). Sections were observed by fluorescence microscope (Olympus, USA). The number of TUNEL+ cells were counted using Image J software. Magnetic Resonance Imaging (MRI) scanning MRI scanning was conducted on day 19 after modeling. Mice were anesthetized using isoflurane (RWD, China) during MRI. MRI scanning of spinal cord and optic nerves were performed with a 7-T small-animal MRI instrument (BioSpec70/20USR, Bruker, Biospin, Ettlingen, Germany). T2-weighted images were acquired. Quantitative real time-PCR Total RNA was extracted with Trizol (Abconal, China) according to instructions from spinal cord and spleen cells. Reverse transcription was conducted using Hiscript III RT Supermix for PCR (Vazyme, China) and real-time PCR was performed using qPCR SYBR Master Mix (Vazyme, China) according to instructions. GAPDH served as an internal reference. Relative expression of genes was calculated as 2 −ΔΔCt . Primer sequences are listed in Table . Western blotting The total protein content of spinal cord tissues was isolated using radioimmunoprecipitation assay lysis buffer and then qualified using a BCA kit (Thermo Fisher, USA). Equal amounts of protein were loaded onto SDS–polyacrylamide gel and then transferred to polyvinylidene difluoride membranes membranes. Membranes were blocked with blocking solution (Epizyme Biotech, China) and incubated with primary antibody (Occludin, 1:2500, abmart, China, T55997; GAPDH, 1:10,000, abcam, USA, ab8245) overnight at 4 °C, followed by a 1 h incubation of secondary antibody. An enhanced chemiluminescence kit (Epizyme Biotech, China) was used to detect the protein signals. Protein was quantified using Image J software. Astrocytes isolation and culture Astrocytes were generated from the cerebral cortex of 1-day-old mice. A total of 20–30 mice were used for the experiment. Briefly, cerebral hemispheres were isolated, cut into pieces and incubated in 0.125% trypsin–EDTA at 37 °C for 15 min. Mixed cortical glial cells were passed through a 70-um cell strainer, and then centrifuged and resuspended in Dulbecco's Modified Eagle Medium containing 4.5 g/L D-Glucose, 4 mM L-Glutamine,10% FBS and 1% penicillin/streptomycin. Cells grown at 37 °C in a 5% CO 2 incubator. After confluence, cells were purified by shaking in a rotator at 260 rpm for 5 h. P1 astrocytes were used for the study. Cell purity was identified by GFAP immunofluorescence staining and GFAP + cells > 90% (Additional file : Figure S1). Complement-dependent cytotoxicity and hUC-MSC coculture Astrocytes were exposed to 10% healthy control or AQP4 IgG for 1 h at 4 °C . Then, 2% human complement (Innovative Research, USA) was added and cells were incubated at 37 °C for 18 h for the CCK8 test. Astrocytes were seeded to the lower chamber of the transwell on the day before coculture. When astrocytes were exposed to AQP4 IgG and complement, hUC-MSCs were cocultured with astrocytes for 48 h. CCK8 The culture medium was replaced with fresh Dulbecco's Modified Eagle Medium containing 10% CCK8 (Dojindo, Japan) and cells were incubated at 37 °C for 2–4 h. Absorbance was read at 450 nm with a Microplate Reader. Cell viability was calculated as follows: test group—blank group/control group—blank group. Annexin V/PI staining Cells were harvested with accutase, collected into the centrifugation tube and washed with PBS. Cells were stained with the annexin V/PI detection kit (Multi Sciences, China) according to instructions. Data were acquired with flow cytometry (BD FACS CELESTA, USA) and were analyzed using Flowjo 10 software. Statistical analysis Statistical analysis was performed with GraphPad Prism 9 software. The results are expressed as mean ± SEM. For two-group comparison, t-test was used for normally distributed data and nonparametric test was used for non-normally distributed data. For multiple comparisons, ANOVA (one-way) was performed. p < 0.05 was considered statistically significant. In this study, 8–10 weeks-old female C57BL/6 mice were purchased from Charles River and housed in the animal facilities of Shanghai Model Organisms. All mice were housed in a temperature- and humidity-controlled room and were kept in regular cages under a 12/12 h light/dark cycle. All mice were fed with standard food and water. No sample size calculation was performed; sample sizes were chosen based on previous in vivo experiments done in the laboratory to allow for statistical power. Mice were allocated to groups randomly using a random number generator. The order of treatments and measurements for each mouse were randomly assigned. A total of 65 C57BL/6 mice were used. 12 C57BL/6 mice were used for the purpose of establishing the NMOSD mice model (EAE: n = 6; NMOSD: n = 6) and 53 C57BL/6 mice were used for the purpose of assessing the therapeutic effects of hUC-MSCs (CFA + PTX: n = 12, NMOSD + PBS: n = 18; NMOSD + MSC: n = 23). All animals were used in the analysis. During the outcome evaluation and data analysis stage, the group allocation was made known. All procedures have been approved by the ethics committee of Shanghai Model Organisms (IACUC 2024-0021-1). The work has been reported in line with the ARRIVE guidelines 2.0. In this study, the plasma exchange fluid of an NMOSD patient and the plasma of a healthy donor were collected. The patient was in the acute phase of an attack and received double-filtration plasmapheresis. A total of 200 ml plasma exchange fluid was collected. The informed consents were obtained. The plasma exchange fluid and the plasma were diluted 1:10 and 1:6 respectively with 1 × Melon Gel Purification Buffer (Thermo Fisher, USA). IgG was extracted using IgG purification kit (Thermo Fisher, USA) according to instructions. The concentration of IgG from plasma exchange fluid of NMOSD patients or plasma of healthy donors was quantified using BCA Protein Assay Kits (Thermo Fisher, USA) and the concentration is 9–10 mg/ml. The tilter of AQP4 IgG is 1:1000 using cell-based indirect immunofluorescence assays in Simplegen institution. Then purified IgG was inactivated at 56 °C. Finally, IgG was filtered with 0.22 μm sterile filters and stored at −80 °C. The extracted antibody was used within 1 year after extraction. hUC-MSCs were purchased from Guangzhou SALIAI Stem cell Science and Technology Co.Ltd. The processes including isolation, culture, identification, quality control, and storage of hUC-MSCs were strictly conformed to standard operating procedures. P6 hUC-MSCs were used for in vivo and in vitro studies. All hUC-MSCs used in this study were from one healthy donor. The characterization of hUC-MSCs was assessed by flow cytometer. Briefly, P6 hUC-MSCs were collected and labeled with the following antibodies: FITC-anti-HLA-DR, -anti-CD45, -anti-CD73, -anti-CD34, APC-anti-CD90, -anti-CD19, -anti-CD105, -anti-CD11b (Biolegend, USA, 1:400) at 4 °C for 30 min. After two washing steps, cells were acquired with a flow cytometer (BD FACS CELESTA, USA) and analyzed with Flowjo 10 software. NMOSD mice model was constructed by continuous intraperitoneal injection of AQP4 IgG into EAE mice. For EAE model, 0.3 mg/ml MOG 35–55 (GL Biochem, China) was emulsified in Complete Freund’s Adjuvant (CFA) containing 8 mg/ml H37Ra Mycobacterium tuberculosis (BD DIFCO, USA). Mice were anesthetized by intraperitoneal injection of 2.5% Avertin (Meilunbio, China). Then 200 μl emulsion was injected subcutaneously at two sites in the right and left flank at once. 200 ng pertussis toxin (PTX) (Sigma-Aldrich, USA) in 100ul PBS was injected intraperitoneally at day 0 and day 2. For NMOSD model, EAE mice were intraperitoneally injected with 400 μl AQP IgG from day 7 to day 18. On day 10 and day 14, 10 6 hUC-MSCs prepared in 200 μl PBS were injected into NMOSD mice intravenously. Control mice received 200 μl PBS. Animals’ clinical scores were assigned as follows: 0 = normal; 0.5 = partial tail paralysis; 1 = entire tail paralysis; 1.5 = entire tail paralysis with unsteady gait; 2 = partial paralysis in hind limbs with the body below the abdomen powerless to land; 2.5 = entire paralysis in one hind limb; 3 = entire paralysis in one hind limb with partial paralysis in another hind limb; 3.5 = entire paralysis in two hind limbs; 4 = entire paralysis in two hind limbs with one forelimb paralysis; 4.5 = entire paralysis in four limbs; 5 = death. On day 19, mice were anesthetized with Avertin and euthanized by cardiac perfusion with PBS. Pathological tissues were obtained for HE staining, LFB staining, immunofluorescence staining and TUNEL staining after PBS and 4% paraformaldehyde perfusion. Spinal cord for protein and RNA extraction was obtained after PBS perfusion. Paraffin sections (2 μm) of lumbar spinal cord was used for HE and LFB staining according to instructions respectively. Briefly, after being dewaxed and rehydrated, the sections were sequentially placed in eosin solution and hematoxylin solution for HE staining, and were incubated in LFB solution at 56 °C overnight and then hydrolyzed for LFB staining. The complete spinal cord was scanned using high throughput digital slice scanning system (Hamamatsu, Japan) and images were obtained under 5× magnification using NDP.view2 software. For HE staining, the level of inflammatory infiltration was assessed as follows: 0: no inflammatory infiltration; 1: only a slight infiltration of inflammatory cells around the blood vessels or spinal cord capsule; 2: small amount of inflammatory cell infiltration in the spinal cord parenchyma; 3: moderate infiltration of inflammatory cells in the spinal cord parenchyma; 4: large infiltration of inflammatory cells in the spinal cord parenchyma. For LFB staining, the area of demyelinating area and white matter were measured using Image J software. After being dewaxed and rehydrated, the sections of lumbar spinal cord were incubated with Tris-EDTA antigen retrieval solution (Solarbio, China) for 30 min at 100 °C and naturally cooled to room temperature. Then, sections were blocked with QuickBlock™ Blocking Buffer (Beyotime, China), incubated with primary antibody overnight at 4 °C and secondary antibody for 2 h at room temperature. Cell nuclei were stained with DAPI (Beyotime, China). The following primary antibodies were used: rabbit anti-AQP4 (1:800, Sigma-Aldrich, USA, MABN2527); mouse anti-glial fibrillary acidic protein (GFAP) (1:400, Cell Signaling Technology, USA, 3670 T); rabbit anti-myelin basic protein (MBP) (1:200, Abcam, USA, ab7349); rabbit anti-NeuN (1:400, Abcam, USA, ab177487). The following fluorescent secondary antibodies were used: goat anti-mouse 488 (1:1000, Beyotime, China); goat anti-rabbit 488 (1:1000, Beyotime, China); goat anti-rabbit 555 (1:1000, Thermo Fisher, USA). Sections were observed by fluorescence microscope (Olympus, USA). The loss area of AQP4, GFAP, MBP and the number of NeuN+ cells were measured using Image J software. After being dewaxed and rehydrated, the sections of lumbar spinal cord were incubated with 20 μg/ml proteinase K (Beyotime, China). TUNEL working solution was prepared according to instructions (Beyotime, China). Then, sections were incubated with working solution for 1 h at 37 °C. Cell nuclei were stained with DAPI (Beyotime, China). Sections were observed by fluorescence microscope (Olympus, USA). The number of TUNEL+ cells were counted using Image J software. MRI scanning was conducted on day 19 after modeling. Mice were anesthetized using isoflurane (RWD, China) during MRI. MRI scanning of spinal cord and optic nerves were performed with a 7-T small-animal MRI instrument (BioSpec70/20USR, Bruker, Biospin, Ettlingen, Germany). T2-weighted images were acquired. Total RNA was extracted with Trizol (Abconal, China) according to instructions from spinal cord and spleen cells. Reverse transcription was conducted using Hiscript III RT Supermix for PCR (Vazyme, China) and real-time PCR was performed using qPCR SYBR Master Mix (Vazyme, China) according to instructions. GAPDH served as an internal reference. Relative expression of genes was calculated as 2 −ΔΔCt . Primer sequences are listed in Table . The total protein content of spinal cord tissues was isolated using radioimmunoprecipitation assay lysis buffer and then qualified using a BCA kit (Thermo Fisher, USA). Equal amounts of protein were loaded onto SDS–polyacrylamide gel and then transferred to polyvinylidene difluoride membranes membranes. Membranes were blocked with blocking solution (Epizyme Biotech, China) and incubated with primary antibody (Occludin, 1:2500, abmart, China, T55997; GAPDH, 1:10,000, abcam, USA, ab8245) overnight at 4 °C, followed by a 1 h incubation of secondary antibody. An enhanced chemiluminescence kit (Epizyme Biotech, China) was used to detect the protein signals. Protein was quantified using Image J software. Astrocytes were generated from the cerebral cortex of 1-day-old mice. A total of 20–30 mice were used for the experiment. Briefly, cerebral hemispheres were isolated, cut into pieces and incubated in 0.125% trypsin–EDTA at 37 °C for 15 min. Mixed cortical glial cells were passed through a 70-um cell strainer, and then centrifuged and resuspended in Dulbecco's Modified Eagle Medium containing 4.5 g/L D-Glucose, 4 mM L-Glutamine,10% FBS and 1% penicillin/streptomycin. Cells grown at 37 °C in a 5% CO 2 incubator. After confluence, cells were purified by shaking in a rotator at 260 rpm for 5 h. P1 astrocytes were used for the study. Cell purity was identified by GFAP immunofluorescence staining and GFAP + cells > 90% (Additional file : Figure S1). Astrocytes were exposed to 10% healthy control or AQP4 IgG for 1 h at 4 °C . Then, 2% human complement (Innovative Research, USA) was added and cells were incubated at 37 °C for 18 h for the CCK8 test. Astrocytes were seeded to the lower chamber of the transwell on the day before coculture. When astrocytes were exposed to AQP4 IgG and complement, hUC-MSCs were cocultured with astrocytes for 48 h. The culture medium was replaced with fresh Dulbecco's Modified Eagle Medium containing 10% CCK8 (Dojindo, Japan) and cells were incubated at 37 °C for 2–4 h. Absorbance was read at 450 nm with a Microplate Reader. Cell viability was calculated as follows: test group—blank group/control group—blank group. Cells were harvested with accutase, collected into the centrifugation tube and washed with PBS. Cells were stained with the annexin V/PI detection kit (Multi Sciences, China) according to instructions. Data were acquired with flow cytometry (BD FACS CELESTA, USA) and were analyzed using Flowjo 10 software. Statistical analysis was performed with GraphPad Prism 9 software. The results are expressed as mean ± SEM. For two-group comparison, t-test was used for normally distributed data and nonparametric test was used for non-normally distributed data. For multiple comparisons, ANOVA (one-way) was performed. p < 0.05 was considered statistically significant. Establishment of NMOSD mouse model Plasma exchange is an effective treatment for the acute phase of NMOSD, and high titers of AQP4 antibodies can be found in plasma exchange fluids. AQP4 IgG was extracted from plasma exchange fluids of an NMOSD patient. Then, we established the NMOSD model by continuous injection of AQP4 IgG intraperitoneally into EAE mice (Fig. A). Both EAE and NMOSD mice showed typical motor dysfunction, and no significant differences in clinical scores were observed between the two groups (Fig. B). The mice were sacrificed on day 19. As indicated by the white arrow, the immunofluorescence result showed a prominent loss of AQP4 and GFAP expression in NMOSD mice compared to that in EAE mice, which is a characteristic pathological feature of NMOSD (Fig. C). Therefore, the NMOSD mice model was successfully established. hUC-MSC treatment ameliorates the disease progression in NMOSD mice The characterization of hUC-MSCs was assessed by flow cytometer (Additional file : Figure S2). 10 6 P6 hUC-MSCs were prepared in PBS and injected into the tail vein on day 10 and 14 after NMOSD induction. Clinical signs of the disease were monitored daily. We found that the incidence was 100% in PBS-treated mice but only 50% in MSC-treated mice. MSC-treated NMOSD mice showed a significant improvement in clinical scores (Fig. A). To investigate the effects of hUC-MSCs on demyelination and inflammation in the spinal cord of NMOSD mice, the mice were sacrificed on day 19. HE staining showed that MSC-treated mice had fewer infiltrating inflammatory cells in the lumbar spinal cord and the inflammation score was lower than that of PBS-treated mice (Fig. B, E). In addition, LFB and MBP immunofluorescence were conducted to examine changes in demyelination. As shown in Fig. C, D, both LFB and immunofluorescence of MBP showed improved myelin loss in MSC-treated mice. Furthermore, the area of myelin loss in the spinal cord of MSC-treated mice was significantly reduced (Fig. F, G). On day 19, a 7.0-T small-animal MRI was performed to examine optic neuritis and myelitis. MRI revealed longitudinally high signals on T2WI in the spinal cord of NMOSD mice, whereas the lesion was unclear in MSC-treated mice (Fig. A). Moreover, high signals were observed in the optic nerves of NMOSD mice and the signal intensity of the lesions was attenuated in MSC-treated mice (Fig. B). Collectively, these findings showed that hUC-MSC treatment ameliorated disease progression in NMOSD mice. hUC-MSC treatment attenuates astrocyte and neuron injury in NMOSD mice Astrocyte impairment is the core pathological feature of NMOSD. Compared to healthy mice, immunofluorescence revealed a marked decrease in AQP4 and GFAP levels in lumbar spinal cord of NMOSD mice as indicated by the white arrow. However, a significant reduction of AQP4 and GFAP loss was observed in MSC-treated NMOSD mice (Fig. A, B, p = 0.03). We performed immunofluorescence staining for NeuN to assess the level of neuron loss in NMOSD mice. We found that the number of NeuN+ cells was significantly lower in NMOSD mice than that in healthy mice, whereas the number of NeuN+ cells was elevated in MSC-treated NMOSD mice (Fig. C, D). These results showed that hUC-MSCs attenuated astrocyte and neuron injury in NMOSD mice. hUC-MSC treatment inhibits central inflammatory infiltration and protects the blood–brain barrier The immunomodulatory property is a function feature of MSCs. Thus, we examined the expression levels of inflammatory cytokines in the spleen cells using qRT-PCR. The expression levels of Il-6 , Ifn-γ , Il-17a , and Il-10 were significantly increased in the spleen cells of NMOSD mice compared with healthy mice. However, the levels of these inflammatory cytokines in MSC-treated NMOSD mice were not different from those in PBS-treated NMOSD mice, suggesting that hUC-MSC treatment did not attenuate the inflammatory process in the spleen (Fig. A–D). We extracted RNA from mouse spinal cord tissues and measured the expression levels of inflammatory cytokines using qRT-PCR. Results showed that the expression levels of Il-6 , Ifn-γ , Il-17a , and Il-10 in the spinal cord of NMOSD mice were increased compared with healthy mice, whereas the expression levels of Il-6 , Ifn-γ , and Il-17a in MSC-treated NMOSD mice were significantly decreased compared with PBS-treated NMOSD mice (Fig. E–H). These results indicated that hUC-MSCs may inhibit the infiltration of peripheral inflammatory cells into the spinal cord, thereby reducing central inflammation and exerting therapeutic effects. The BBB plays a crucial role in inhibiting central inflammatory infiltration. We further examined the expression levels of the BBB permeability indicator: occludin using western blotting. The result showed that the expression level of occludin in the spinal cords of MSC-treated NMOSD mice was significantly higher than that in PBS-treated NMOSD mice, indicating that hUC-MSCs play a role in protecting the BBB (Fig. I, J). Overall, these data showed that hUC-MSCs did not inhibit the peripheral inflammatory process, but reduced CNS inflammation and attenuated the blood–brain barrier disruption. hUC-MSC treatment inhibits AQP4 IgG-induced astrocyte apoptosis In-vivo study showed that hUC-MSC treatment mitigated astrocyte damage in NMOSD mice. We further investigated whether hUC-MSCs protected against AQP4 IgG-induced astrocyte injury in vitro. Consistent with previous findings , we found that AQP4 IgG significantly reduced astrocyte viability only in the presence of complement (Fig. A). Primary astrocytes were co-cultured with hUC-MSCs for 48 h in a transwell system. The CCK8 assay revealed that hUC-MSC co-culture significantly enhanced the viability of astrocytes (Fig. B). In addition, the level of cell death and apoptosis were investigated by annexin V/PI staining. Compared to healthy control IgG, AQP4 IgG induced a large amount of cell apoptosis after 48 h of incubation, whereas no significant difference was observed in the proportion of cell death (Fig. C–E). Co-culture with hUC-MSCs significantly reduced the apoptosis level of astrocytes (Fig. C, E). We further performed TUNEL staining on mouse spinal cord slices to detect the level of apoptosis in vivo. The results showed that TUNEL + apoptotic cells in the spinal cord were significantly reduced in MSC-treated NMOSD mice compared with PBS-treated NMOSD mice, which is consistent with the results of the in vitro studies (Fig. F, G). Overall, these results indicate that hUC-MSCs suppressed AQP4 IgG and complement-induced apoptosis in vitro and inhibited apoptosis in the spinal cord of NMOSD mice in vivo. Plasma exchange is an effective treatment for the acute phase of NMOSD, and high titers of AQP4 antibodies can be found in plasma exchange fluids. AQP4 IgG was extracted from plasma exchange fluids of an NMOSD patient. Then, we established the NMOSD model by continuous injection of AQP4 IgG intraperitoneally into EAE mice (Fig. A). Both EAE and NMOSD mice showed typical motor dysfunction, and no significant differences in clinical scores were observed between the two groups (Fig. B). The mice were sacrificed on day 19. As indicated by the white arrow, the immunofluorescence result showed a prominent loss of AQP4 and GFAP expression in NMOSD mice compared to that in EAE mice, which is a characteristic pathological feature of NMOSD (Fig. C). Therefore, the NMOSD mice model was successfully established. The characterization of hUC-MSCs was assessed by flow cytometer (Additional file : Figure S2). 10 6 P6 hUC-MSCs were prepared in PBS and injected into the tail vein on day 10 and 14 after NMOSD induction. Clinical signs of the disease were monitored daily. We found that the incidence was 100% in PBS-treated mice but only 50% in MSC-treated mice. MSC-treated NMOSD mice showed a significant improvement in clinical scores (Fig. A). To investigate the effects of hUC-MSCs on demyelination and inflammation in the spinal cord of NMOSD mice, the mice were sacrificed on day 19. HE staining showed that MSC-treated mice had fewer infiltrating inflammatory cells in the lumbar spinal cord and the inflammation score was lower than that of PBS-treated mice (Fig. B, E). In addition, LFB and MBP immunofluorescence were conducted to examine changes in demyelination. As shown in Fig. C, D, both LFB and immunofluorescence of MBP showed improved myelin loss in MSC-treated mice. Furthermore, the area of myelin loss in the spinal cord of MSC-treated mice was significantly reduced (Fig. F, G). On day 19, a 7.0-T small-animal MRI was performed to examine optic neuritis and myelitis. MRI revealed longitudinally high signals on T2WI in the spinal cord of NMOSD mice, whereas the lesion was unclear in MSC-treated mice (Fig. A). Moreover, high signals were observed in the optic nerves of NMOSD mice and the signal intensity of the lesions was attenuated in MSC-treated mice (Fig. B). Collectively, these findings showed that hUC-MSC treatment ameliorated disease progression in NMOSD mice. Astrocyte impairment is the core pathological feature of NMOSD. Compared to healthy mice, immunofluorescence revealed a marked decrease in AQP4 and GFAP levels in lumbar spinal cord of NMOSD mice as indicated by the white arrow. However, a significant reduction of AQP4 and GFAP loss was observed in MSC-treated NMOSD mice (Fig. A, B, p = 0.03). We performed immunofluorescence staining for NeuN to assess the level of neuron loss in NMOSD mice. We found that the number of NeuN+ cells was significantly lower in NMOSD mice than that in healthy mice, whereas the number of NeuN+ cells was elevated in MSC-treated NMOSD mice (Fig. C, D). These results showed that hUC-MSCs attenuated astrocyte and neuron injury in NMOSD mice. The immunomodulatory property is a function feature of MSCs. Thus, we examined the expression levels of inflammatory cytokines in the spleen cells using qRT-PCR. The expression levels of Il-6 , Ifn-γ , Il-17a , and Il-10 were significantly increased in the spleen cells of NMOSD mice compared with healthy mice. However, the levels of these inflammatory cytokines in MSC-treated NMOSD mice were not different from those in PBS-treated NMOSD mice, suggesting that hUC-MSC treatment did not attenuate the inflammatory process in the spleen (Fig. A–D). We extracted RNA from mouse spinal cord tissues and measured the expression levels of inflammatory cytokines using qRT-PCR. Results showed that the expression levels of Il-6 , Ifn-γ , Il-17a , and Il-10 in the spinal cord of NMOSD mice were increased compared with healthy mice, whereas the expression levels of Il-6 , Ifn-γ , and Il-17a in MSC-treated NMOSD mice were significantly decreased compared with PBS-treated NMOSD mice (Fig. E–H). These results indicated that hUC-MSCs may inhibit the infiltration of peripheral inflammatory cells into the spinal cord, thereby reducing central inflammation and exerting therapeutic effects. The BBB plays a crucial role in inhibiting central inflammatory infiltration. We further examined the expression levels of the BBB permeability indicator: occludin using western blotting. The result showed that the expression level of occludin in the spinal cords of MSC-treated NMOSD mice was significantly higher than that in PBS-treated NMOSD mice, indicating that hUC-MSCs play a role in protecting the BBB (Fig. I, J). Overall, these data showed that hUC-MSCs did not inhibit the peripheral inflammatory process, but reduced CNS inflammation and attenuated the blood–brain barrier disruption. In-vivo study showed that hUC-MSC treatment mitigated astrocyte damage in NMOSD mice. We further investigated whether hUC-MSCs protected against AQP4 IgG-induced astrocyte injury in vitro. Consistent with previous findings , we found that AQP4 IgG significantly reduced astrocyte viability only in the presence of complement (Fig. A). Primary astrocytes were co-cultured with hUC-MSCs for 48 h in a transwell system. The CCK8 assay revealed that hUC-MSC co-culture significantly enhanced the viability of astrocytes (Fig. B). In addition, the level of cell death and apoptosis were investigated by annexin V/PI staining. Compared to healthy control IgG, AQP4 IgG induced a large amount of cell apoptosis after 48 h of incubation, whereas no significant difference was observed in the proportion of cell death (Fig. C–E). Co-culture with hUC-MSCs significantly reduced the apoptosis level of astrocytes (Fig. C, E). We further performed TUNEL staining on mouse spinal cord slices to detect the level of apoptosis in vivo. The results showed that TUNEL + apoptotic cells in the spinal cord were significantly reduced in MSC-treated NMOSD mice compared with PBS-treated NMOSD mice, which is consistent with the results of the in vitro studies (Fig. F, G). Overall, these results indicate that hUC-MSCs suppressed AQP4 IgG and complement-induced apoptosis in vitro and inhibited apoptosis in the spinal cord of NMOSD mice in vivo. NMOSD is characterized by a high relapse rate and prevalence of disability. Generic oral immunosuppressants and monoclonal antibodies that target B cells, the IL-6 pathway, or the complement activation pathway are the major long-term relapse prevention treatments. However, some patients still can’t benefit from these treatments and experience relapse, resulting in severe disability. Several studies have shown that MSCs exhibit therapeutic potential in various autoimmune diseases, such as multiple sclerosis, systemic lupus erythematosus, and rheumatoid arthritis . Besides, several clinical studies applied bone marrow MSCs or hUC-MSCs in the treatment of NMOSD patients and showed potential therapeutic effects . Previous studies reported the therapeutic effects of hUC-MSC transplantation in animal models of multiple sclerosis: EAE . However, there is relatively little research on the role of hUC-MSCs in the disease course in animal models of NMOSD. It is reported that hUC-MSCs ameliorated motor dysfunction in an NMOSD animal model via passive transfer of AQP4 IgG to mice with BBB damage induced by CFA and PTX . In our study, we constructed an NMOSD mouse model by continuous intraperitoneal injection of AQP4 IgG based on EAE. Although it cannot completely mimic the production of AQP4 IgG, it can partially simulate AQP4 IgG-mediated astrocyte injury, and mice have typical motor dysfunction, which makes it easy to observe the therapeutic effects of hUC-MSCs. Similar to previous findings, hUC-MSCs significantly reduced the incidence and improved motor dysfunction in NMOSD mice . In addition, 7.0 T MRI was performed to obtain a three-dimensional view of the overall lesion. We observed that NMOSD mice exhibited long-segment myelitis and optic neuritis, whereas hUC-MSC-treated NMOSD mice showed a significant reduction in both myelitis and optic neuritis, further providing strong evidence for the therapeutic effect of hUC-MSCs on NMOSD. Immunomodulatory function is a property of MSCs. In an in vivo study of experimental autoimmune cholangitis treated with hUC-MSCs, significant down-regulation of serum levels of IFN-γ and IL-17A was observed . In vitro, hUC-MSCs reduced the expression of pro-inflammatory cytokines, such as IFN-γ and IL-6 in activated T-lymphocytes . These findings demonstrated the immunomodulatory potential of hUC-MSCs. However, we found that hUC-MSC treatment did not affect the inflammatory response in the spleen by examining the levels of cytokine expression, which is unexpected. Previous studies have shown that the immunoregulatory function of MSCs is affected by their inflammatory environment , . For example, IFN-γ enhanced the immunoregulatory function of MSCs by promoting the secretion of indoleamine-2,3-dioxygenase , whereas transforming growth factor-β could inhibit the secretion of inducible nitric oxide synthase and weaken the immunoregulatory potential of MSCs . Therefore, we hypothesized that the immunoregulatory capacity of hUC-MSCs is diminished in the inflammatory environment of the NMOSD mouse model and thus failed to modulate peripheral immune responses. However, the factors that influence the immunomodulatory capacity of hUC-MSCs remain unclear and require further investigation. Although hUC-MSCs did not affect peripheral inflammatory responses in NMOSD mice, we observed reduced infiltrated inflammatory cells and decreased inflammatory factor expression in the spinal cord of MSC-treated mice. These results indicate that hUC-MSCs inhibit inflammatory infiltration in the spinal cord of NMOSD mice. The BBB plays a key role in this process and we found that the level of BBB disruption was significantly reduced in the spinal cord of MSC-treated NMOSD mice, indicating the protective role of hUC-MSCs in the BBB. This result is consistent with previous studies showing that hUC-MSCs protect the BBB in other disease models, such as spinal cord injury and EAE . hUC-MSCs have been reported to increase the expression of tight junction proteins and decrease the expression of matrix metalloproteinases, which have a destructive effect on the BBB . Collectively, the protective effect of hUC-MSCs on the BBB effectively inhibits the infiltration of inflammatory factors from the periphery into the spinal cord, thus reducing the level of inflammation and damage to CNS-resident cells. Another property of MSCs is their capacity for tissue repair. AQP4 IgG-induced astrocyte injury is a core pathological feature of NMOSD. Therefore, we investigated the role of hUC-MSCs in this process. We found that AQP4 IgG induced astrocyte cytotoxicity in a complement-dependent manner, which is consistent with previous studies . Furthermore, in vitro hUC-MSC co-culture significantly promoted astrocyte survival and reduced apoptosis. Treatment with hUC-MSCs also suppressed apoptosis in the spinal cords of NMOSD mice in vivo. These results indicated that hUC-MSCs played a protective role in the process of AQP4 IgG- and complement-induced astrocyte injury and showed anti-apoptotic effects. Previous studies have mostly focused on the protective role of MSCs in promoting oligodendrocyte and neuron differentiation as well as suppressing oligodendrocyte and neuron apoptosis , while their effects on astrocyte apoptosis are less explored. BM-MSCs have been reported to inhibit astrocyte apoptosis in a glyoxylate stripping model in vitro , which is consistent with our results. Collectively, hUC-MSCs showed anti-apoptotic effects and great potential for the treatment of NMOSD. Collectively, our study first applied hUC-MSCs therapy in this NMOSD animal model and showed therapeutic effects of hUC-MSCs on NMOSD. Besides, our study indicated that hUC-MSCs suppressed AQP4 IgG- and complement- induced astrocytes apoptosis for the first time. However, this study has some limitations. The mechanisms underlying the anti-apoptotic effects of hUC-MSCs on astrocytes need further exploration. The NMOSD model that we used is a passive immune model. The model also cannot mimic the production of anti-AQP4 antibodies by plasma cells in patients with NMOSD. Whether hUC-MSCs inhibit plasma cell differentiation and antibody production in NMOSD should be further explored using an active immune model. In summary, we found that hUC-MSCs significantly attenuated motor dysfunction and pathological and imaging manifestations in an animal model of NMOSD. Meanwhile, we revealed that hUC-MSCs did not attenuate peripheral inflammation, but significantly attenuated inflammatory infiltration and BBB disruption in the spinal cord. Furthermore, hUC-MSCs were found to have a protective effect against AQP4 IgG- and complement-mediated astrocyte injury and inhibited apoptosis. Collectively, our study provides further evidence for the therapeutic application of hUC-MSCs in NMOSD, and reveals, for the first time, their protective effects against AQP4 IgG- and complement-induced astrocyte apoptosis. Additional file 1. Additional file 2. Additional file 3. |
Regional antimicrobial resistance gene flow among the One Health sectors in China | b4247a09-3fb4-403c-a739-f94b9e3c86a0 | 11705761 | Microbiology[mh] | The increase in antimicrobial resistance (AMR) poses a significant challenge to healthcare systems, raising concerns regarding global public health, and food safety . The dissemination and spread of AMR is complex, involving antimicrobial resistance gene (ARG) flow across all three One Health sectors (animals, humans, and environments). Understanding the emergence and transmission of ARGs and/or antimicrobial-resistant bacteria (ARB) among these sectors as well as their interfaces such as food and water, is a prerequisite for better control of AMR. Through culture-dependent and -independent approaches, specific ARG dissemination pathways especially those from animals to humans have been proposed. For example, plasmid-mediated colistin resistance was first identified mainly in isolates from animals and then was found widely spread across animals, humans, and environments . The dissemination of AMR among the One Health sectors can be largely attributed to the mobility of ARGs carried by mobile genetic elements (MGEs), which can then be vertically propagated through clonal spread. This process has occurred for many clinically important ARGs, including the carbapenemase gene bla NDM-1 , the colistin resistance gene mcr-1 , and the tigecycline resistance genes tet (X3) and tet (X4) . Diverse MGEs, such as plasmid and foodborne pathogens (especially members of the Enterobacteriaceae ), are frequently evidenced to be involved in the dissemination of these ARGs . With the development of high-throughput sequencing, genomics, and metagenomics have been increasingly incorporated into One Health studies, enabling investigation of the abundance, transmission, and distribution of ARGs across different sources . Through whole-genome sequencing analyses of Escherichia coli isolates, the Australian silver gull was found to host different carbapenemase genes that may be anthropogenic sources , while the KPC-2-producing Serratia marcescens clone found in farm animals could potentially contribute to the clinical isolates present in nosocomial settings (i.e., healthcare environments where infections are acquired during hospital stays) . Recently, the global distribution of ARGs was profiled by collecting thousands of metagenomic sequencing datasets from six types of habitats, and nearly 24% of the detected ARGs were concluded to pose a health risk to humans . Although these findings enhance our understanding of AMR dynamics, most of the studies were based on large areas and/or time scales, which may not reveal the ecology of AMR at high resolution. Additionally, relatively fewer AMR studies have focused on the dissemination and spread of AMR at One Health interfaces in geographically proximate ecosystems. Assessment of the risk of AMR acquisition in humans, as a result of human behaviors such as dietary habits and occupational exposure, is continually needed. In this study, we investigated regional AMR transmission in a Chinese city by analyzing the metagenomic sequencing data and the genomes of carbapenem-resistant isolates collected from humans, food, and the environment. We identified key MGEs and bacteria in mediating ARG transmission and proposed potential new transmission pathways for ARG transmission from food and the environment to humans. Habitat-specific profiles of the antibiotic resistome We collected 592 samples from nine different human subgroups, three food subgroups, and six environmental subgroups in Dengfeng, Henan Province, China for antibiotic resistome and ARG flow analyses (Fig. a and Additional file : Table S1). We identified 40 ARG types and 743 ARG subtypes. The most abundant ARG types identified were multidrug resistance genes (151 ARG subtypes; accounting for 27.5% of the total ARG abundance), followed by the macrolide-lincosamide-streptogramin (MLS; 90 subtypes; 24.6%), tetracycline (52 subtypes; 14.2%), aminoglycoside (78 subtypes; 7.7%), and beta-lactam resistance genes (138 subtypes; 6.3%) (Fig. b and Additional file : Table S2). Samples of different origins (human, food, and the environment) displayed different resistome profiles (Additional file : Fig. S1a and Additional file : Table S3). The most obvious difference was that compared with human samples, food (pork, chicken, and vegetable/fruit), soil, surface water, and fly microbes showed a very high load of multidrug resistance genes (from 1.460 to 5.236 ARG copies per 16S rRNA gene copy). Interestingly, microbes from food and the environment contained a significantly higher load of ARGs than those from human feces (Kruskal–Wallis test, P = 2.9 × 10 −14 ; Fig. c), especially for multidrug ( P < 2.2 × 10 −16 ), aminoglycoside ( P = 7.0 × 10 −16 ), unclassified ( P < 2.2 × 10 −16 ), and bacitracin ( P < 2.2 × 10 −16 ) resistance genes (Additional file : Fig. S1b and Additional file : Table S4). In humans, dietary habits and occupational exposure were found to affect ARG abundance, especially for the top 20 most abundant ARG types (Kruskal–Wallis tests, P < 0.05; 90.0%, 18/20; Additional file : Fig. S1c and Additional file : Table S5). For example, samples from boarding students showed a higher load of beta-lactam resistance genes; samples from pork abstainers exhibited a lower load of glycopeptide resistance genes; samples from livestock farmers carried a higher load of phenicol resistance genes. Specific aminoglycoside and tetracycline resistance genes were detected at higher levels in the human fecal samples, including tetM , tetW , tet(W/NW) , tetO , tet40 , AAC(6’)-Ie-APH(2’’)-Ia , BANAP , and APH(2’’)-If (Fig. d). Diversity analyses indicated that ARG alpha diversity, as measured by the Shannon index, was lower in human fecal samples than in samples from other groups (Kruskal–Wallis test, P < 2.2 × 10 −16 ; Fig. e), while at the subgroup level, samples from poultry feces, flies, and wastewater had higher Shannon indices (Kruskal–Wallis test, P < 2.2 × 10 −16 ; Additional file : Fig. S2a); among the human samples, compared with omnivores, vegan communes exhibited higher ARG alpha diversity ( P = 1.9 × 10 −4 ). Beta diversity analysis revealed the clustering of samples from human, food, and environmental sources into three distinct groups (adonis, P < 0.001, R 2 = 36.4%; Fig. f and Additional file : Table S6), highlighting the influence of habitat on the ARG composition. Principal coordinate analysis (PCoA) revealed significant differences in ARG subtypes according to the Axis1 values between the human and animal fecal samples and the other samples (Kruskal–Wallis test, P < 2.2 × 10 −16 ; Additional file : Fig. S2b), with the exception of wastewater samples. We then compared the Bray–Curtis distances between the resistome profiles of the human fecal samples and other samples. Our PCoA analyses were supported by the comparison of Bray–Curtis distances, which indicated human fecal samples were more similar to animal feces and wastewater samples compared to other samples (Additional file : Fig. S2c). A further analysis of the abundance of the Bacteroides bacteriophages B40-8 and crAssphage, indicators of fecal contamination , were highly represented in the fecal and wastewater samples (Kruskal–Wallis test, P < 2.2 × 10 −16 ; Fig. g and Additional file : Table S7). Pseudomonadota members are the primary drivers that shape the resistome We next analyzed the microbiome profiles from the metagenomic sequencing data to establish the connections between microbial and ARG compositions. At the phylum level, a total of 31 phyla with varied abundance among samples from different sources was identified (Fig. a). Bacillota (with a relative abundance of 53.0%) and Bacteroidota (34.7%) were the dominant phyla in human fecal samples, whereas the Pseudomonadota occupied a higher proportion in food (43.0%) and environmental samples (31.3%). By binning the metagenomic contigs, we generated 6067 strain-level metagenome-assembled genomes (MAGs, 99.0% average nucleotide identity [ANI]) and 1302 species-level MAGs (95.0% ANI) from a total of 14,787 MAGs. Among these species, 230 species were identified as putative novel species (Additional file : Fig. S3a and Additional file : Table S8). Diversity analyses revealed significantly lower Shannon indices for food samples than for human and environmental samples (Kruskal–Wallis test, P < 2.2 × 10 −16 ; Fig. b), and the microbial compositions differed significantly among the three groups of samples (PERMANOVA, P = 0.001, R 2 = 10.9%; Fig. c and Additional file : Table S9). We then performed Procrustes analysis to assess the extent to which the microbial composition influenced the resistome profiles and found a significant correlation between the resistome profiles and the microbial composition ( P = 0.001, R 2 = 74.0%; Fig. d). Further, we applied a random forest model to identify the main contributors (at different taxonomic levels) influencing the resistome profiles. The results showed that E. coli , a typical member of Pseudomonadota at the species level, was determined to be the main taxon distinguishing the resistome profiles (Fig. e, f and Additional file : Tables S10–S11). Differences in abundance between the various groups of samples revealed the following: the food samples were higher in the abundance of Gammaproteobacteria ( P < 2.2 × 10 −16 ), Enterobacterales ( P = 2.5 × 10 −12 ), and Enterobacteriaceae ( P = 1.2 × 10 −6 ) than the other two groups of samples, while the environmental samples exhibited higher abundances of Pseudomonadota ( P < 2.2 × 10 −16 ) and Gammaproteobacteria ( P < 2.2 × 10 −16 ) than the human fecal samples, and E. coli ( P = 1.5 × 10 −7 ) and Escherichia ( P = 6.3 × 10 −7 ) were more abundant at the species and genus levels in the human fecal samples (Fig. f and Additional file : Table S11). Spearman’s correlation analyses further verified the close associations between the abundance of these taxa and the abundance and diversity of the ARGs ( P < 0.05; Fig. g and Additional file : Table S12). We compared the relative contributions of vertical and horizontal transfer of ARGs, and found that in food samples, the resistome was more affected by MGEs, whereas in environmental samples, it was more associated with the microbiome (Additional file : Fig. S3b). To reveal the putative bacterial host of ARGs, we assembled the sequence reads into contigs and used these contigs for taxonomic assignment. After sequence assembly and ARG annotation, we identified a total of 162,001 ARG-carrying contigs (ACCs) containing 183,720 ARGs from all 592 samples. Taxonomic assignment based on sequence information classified 20.4% of the ACCs at the genus or species level; however, one-third of the ACCs could not be classified, even at the phylum level (Additional file : Fig. S3c). A total of 54,831 ACCs, accounting for 33.8% of the total ACCs, could be binned into MAGs, of which, 14,367 had accurate taxonomic assignments according to the taxonomic information of the associated MAGs (Additional file : Fig. S3d). Overall, we finally assigned ~ 25.0% of the ACCs to the species or genus level (Additional file : Fig. S3e and Additional file : Table S13). We then computed the ARG load index at different taxonomic levels, i.e., the proportion of ARGs assigned to each taxon divided by the average relative abundance of the corresponding taxon. At the phylum level, Pseudomonadota was found to be the main ARG carrier (load index was 1.5), followed by Bacillota (0.8), Actinomycetota (0.3), and Bacteroidota (0.2) (Additional file : Table S14). At the class and order levels, Gammaproteobacteria (1.6) and Enterobacterales (1.1), respectively, both belonging to Pseudomonadota , contributed to the highest load of ARGs. The enrichment of ARGs in Enterobacteriaceae plasmids and prophages/phages MGEs such as plasmids and prophages/phages are known as a reservoir for ARGs. A total of 13.0% (21,134/162,001) of our assembled ACCs were predicted to be plasmid sequences, and 2.0% (3234/162,001) of ACCs contained prophage/phage sequences (Fig. a, b). These sequences together carried 12.3% (22,571/183,720) of the total ARGs. The proportions of plasmid sequences differed among human, food, and environmental samples (Kruskal–Wallis test, P < 2.2 × 10 −16 ), with food samples exhibiting the highest ratio of plasmid sequences compared with human fecal and environmental samples (Fig. c). Seen from the ARG composition, the plasmid sequences in food, fly, and surface water samples were found to carry more multidrug resistance genes (Additional file : Fig. S4a). Replicon-based typing of the plasmid sequences identified 113 plasmid types that were associated with 102 ARG subtypes (Additional file : Fig. S4b), with IncQ1 ( n = 20), and rep_cluster_1232 ( n = 10) were the top frequently encountered replicons. To investigate the potential of these plasmid types in disseminating ARGs, we searched the PLSDB database for the replicons we identified in our samples and annotated the ARGs carried by the known plasmid sequences (Additional file : Fig. S4c). In total, 291 ARG subtypes were found to be carried by the 113 plasmid types, and plasmids with replicon of IncHI2A/rep_cluster_1088 carried the highest number of ARG subtypes ( n = 106), followed by IncFIB ( n = 101) and IncR ( n = 77). Among these 291 subtypes in the database, 24.4% ( n = 71) were found carried by the plasmids found in our samples. Interestingly, we found eight ARG subtypes carried by our plasmids that have not been reported previously (Fig. d and Additional file : Table S15). For example, we showed that tetL gene encoding tetracycline efflux protein can be carried by rep_cluster_1018/rep_cluster_1118, rep_clsuter_1118, and rep_cluster_2013 type plasmids. The prophage/phage fragments found in our samples were predominantly from the family Peduoviridae (11.3% in human fecal samples, 33.8% in food samples, and 21.5% in environmental samples; Additional file : Fig. S5a–c). Glycopeptide and peptide resistance genes were found to be more abundant in prophages/phages from both human and fecal samples, while the unclassified DNA binding protein H-NS was highly represented in phages from food samples (Additional file : Fig. S5d and Additional file : Table S16). The glycopeptide resistance gene vanU and the peptide resistance gene ugd were the two common phage/prophage ARGs found in all three groups of samples. Taking human fecal samples as an example, we found that the frequency of occurrence of each ARG among our samples was highly consistent with that in the Gut Phage Database (GPD) of human gut phages (Spearman’s correlation, R 2 = 0.852, P = 5.8 × 10 −12 ; Fig. e). We then predicted the hosts of the prophages/phages and found that Bacillota and Pseudomonadota were found to be the main hosts of the phages at the phylum level (Additional file : Fig. S5e), and Enterobacteriaceae was the main host at the family level (Fig. f), especially for prophages/phages in samples from chickens, flies, and surface water (Additional file : Fig. S5f). A total of 24 ARG subtypes were found carried by the Enterobacteriaceae prophages/phages, including genes encoding multidrug and tetracycline resistance. Regional AMR dissemination mediated by horizontal gene transfer (HGT) and strain transmission We then investigated ARG transmission among different habitats by clustering the ARG nucleotide acid sequences from various sources at 100.0% identity and constructed ARG-sharing networks (Additional file : Fig. S6a and Additional file : Tables S17–S19). According to the ARG risk ranks classified previously , we categorized the 328 ARG subtypes into six groups (from high- to low-risk ranks): Q1 ( n = 88), Q2 ( n = 45), Q3 ( n = 35), Q4 ( n = 42), RI = 0 ( n = 79), and Unknown ( n = 41) (Additional file : Table S20). We found that ARG subtypes in Q1 had higher sharing frequencies than ARG subtypes in other categories (Kruskal–Wallis test, P < 2.8 × 10 −13 ; Additional file : Fig. S6b and Additional file : Table S21). Additionally, there was a higher ratio of ARGs shared between human fecal samples and food, animal fecal, wastewater, and fly samples (Fig. a), especially those in the Q1 category (Additional file : Fig. S6c). This was also the case when different human subgroups were separately compared with the food and environmental subgroups (Fig. b). Typically, compared with other human subgroups, the human fecal samples from the vegetarian subgroup exhibited the highest ratio of ARGs shared with vegetables or fruits (41.9%), while the lowest ratio with pork (4.2%). A total of 78 ARGs (Q1) were found to be shared in at least one comparison pair between fly, pork, vegetables/fruits, and different human subgroups (Fig. c and Additional file : Table S22), and interestingly, flies shared a relatively high ratio of ARGs with food processing workers. Taken together, these results highlight the important role of the high-risk ARGs in AMR spread. We then used StrainPhlAn to create maps of strain-sharing events between samples to reveal regional strain transmission. We showed that 257 species-level genome bins (SGBs) were present in at least 20 samples (Additional file : Table S23). The most prevalent SGB was Agathobacter rectalis ( n = 328) from the family Lachnospiraceae , followed by Blautia wexlerae ( n = 311), Faecalibacterium prausnitzii ( n = 299), Prevotella copri ( n = 298), and Blautia_A faecis ( n = 298). Among six species of Enterobacteriaceae , E. coli (SGB10068) was the most prevalent species in our samples ( n = 249), consisting of three different strain lineages (I to III). Lineage I was the largest group present in 235 samples from multiple sources, across human, fly, food (pork, chicken, vegetables, and fruits), and wastewater. For other species in Enterobacteriaceae , the same strain lineage of Klebsiella pneumoniae , K. michiganensis , K. ornithinolytica , Clostridium freundii , and Enterobacter hormaechei was also found shared among samples of different origins, mainly human, food (chicken, pork, vegetables, and fruits) and fly. Notably, flies may play an important role in the transmission of Enterobacteriaceae strains, as samples from flies appeared in almost all of the main strain lineages in the phylogenetic trees (Fig. d). Apart from the Enterobacteriaceae , strain transmission events between human and other samples were also detected for species from other families, including C. perfringens , Enterococcus faecium , Weissella confusa , and Streptococcus pasteurianus (Additional file : Fig. S7). For example, one C. perfringens strain was found in 51 samples across human, vegetable/fruit, fly, pork, and chicken, representing a typical regional strain transmission event of opportunistic pathogen. The regional dissemination of carbapenemase genes To further reveal regional ARG transmission, we selected carbapenemase genes as representative ARGs. We identified a total of 111 subtypes of carbapenemase genes in the ACCs from all of our samples (Additional file : Table S24). OXA-347 ( n = 308) was found to be the most prevalent gene carried by the ACCs (Additional file : Fig. S8a), being predominantly present in human and animal fecal samples, and wastewater (Additional file : Fig. S8b, c). To uncover the potential transmission of carbapenemase genes, we analyzed the sharing events of these genes across various sources. A total of 200 unique sequence types were identified from the top 20 frequently encountered carbapenemase genes (1 to 30 distinct sequence types per gene), clustering with a cut-off of 100.0% nucleotide identity (Additional file : Fig. S9). Among these 20 genes, 11 genes were found to be shared among samples of different sources, suggesting complex dissemination pathways of these genes. For example, OXA-347 , OXA-1 , OXA-85 , and CfiA2 showed a high sharing ratio between feces and wastewater. Interestingly, ACT-12 and ORN-1 showed a high sharing ratio between food and fly samples, suggesting that flies may serve as mediators for ARG cross-transmission. Of note, a sequence type of OXA-50 was detected in poultry feces, livestock farmers, and food processing workers. There were rare instances of sharing events between chicken or pork and poultry or pork abstainers. Interestingly, vegans who live together were more likely to carry specific carbapenemase genes, such as SHV-27 , ORN-1 , and ACT-28 , possibly due to person-to-person transmission. We next isolated carbapenem-resistant Enterobacteriaceae strains from our 592 samples. In total, 45 resistant isolates affiliated with nine bacterial species were identified from 42 samples (Fig. a and Additional file : Table S25). The annotation results indicated that 17 carbapenemase gene subtypes were carried by these isolates, among which six subtypes, namely, IMP-4 , NDM-3 , NDM-5 , OXA-805 , SFO-1 , and SHV-61 , were only found in the pure culture isolates, not in the metagenomic data. The carbapenemase genes recovered from these isolates accounted for only a small fraction of the subtypes found in the metagenomic data ( n = 111), indicating the presence of other carriers besides Enterobacteriaceae members. We then clustered each species by including the corresponding genomes of the isolates and MAGs that were assembled from the metagenomic sequencing data. We observed potential transmission of the same strain (ANI ≥ 99.0%) harboring different carbapenemase genes across different sample sources (Fig. b and Additional file : Table S26). For example, the E. coli strain cluster 1 carrying different carbapenemase genes (such as NDM-1 , NDM-5 , OXA-1 , OXA-10 , and CTX-M-27 ) was widely distributed among 141 samples covering human fecal samples and another seven subgroups (except for samples from surface water and wastewater). Similar situations were also found for E. coli strain clusters 2 and 3 containing NDM-1 or NDM-5 , and the K. pneumoniae strain cluster carrying SHV-61 or SHV-33 . A strain of K. ornithinolytica harboring three carbapenemase genes ( SFO-1 , ORN-1 , and NDM-1 ) was found to only be present in fly and human samples, again suggesting the significant role of fly-mediated transmission of ARGs. Our metagenomic data showed that carbapenemase gene OXA-347 was prevalent in human and animal fecal samples, however, we did not recover isolates carrying carbapenemase gene OXA-347 , probably because the SuperCARBA medium used mainly targets Enterobacteriaceae bacteria. To determine the host range and potential dissemination of OXA-347 , we searched for this gene among the complete genomes in the database (Additional file : Fig. S10). The OXA-347 gene was present in 12 bacterial species associated with two known types of plasmid replicons (rep_cluster_663 and rep_cluster_1097) in three Bacteroides species ( B. fragilis , B. thetaiotaomicron , and B. xylanisolvens ) (Fig. c and Additional file : Table S27) and two unknown types of plasmids in Myroides albus and Sphingobacterium faecium . Additionally, the chromosome-located OXA-347 gene was found to be frequently flanked by insertion sequences such as IS1380, IS91, and IS1595. These results suggest a high spread risk of OXA-347 via horizontal gene transfer. Out of the 12 bacterial species possessing OXA-347 , four were found to be common inhabitants in both human feces and wastewater samples (Fig. d). Prediction of carbapenem-resistant strains using a machine learning model We finally constructed random forest models to evaluate the potential of utilizing metagenomic sequencing data for predicting carbapenem-resistant strains. The results showed that all prediction models could achieve an area under the curve (AUC) exceeding 0.90 when constructed based on microbiome profiles, resistome profiles, or a combination of both (Fig. a–c). Notably, microbiome profiles proved to be robust indicators with an AUC of 0.939 and a specificity of 0.595. Among the identified predictors, Aeromonas hydrophila , Bacilli bacterium , Alistipes putredinis , and Blautia wexlerae were considered strong contributors (Fig. d). We collected 592 samples from nine different human subgroups, three food subgroups, and six environmental subgroups in Dengfeng, Henan Province, China for antibiotic resistome and ARG flow analyses (Fig. a and Additional file : Table S1). We identified 40 ARG types and 743 ARG subtypes. The most abundant ARG types identified were multidrug resistance genes (151 ARG subtypes; accounting for 27.5% of the total ARG abundance), followed by the macrolide-lincosamide-streptogramin (MLS; 90 subtypes; 24.6%), tetracycline (52 subtypes; 14.2%), aminoglycoside (78 subtypes; 7.7%), and beta-lactam resistance genes (138 subtypes; 6.3%) (Fig. b and Additional file : Table S2). Samples of different origins (human, food, and the environment) displayed different resistome profiles (Additional file : Fig. S1a and Additional file : Table S3). The most obvious difference was that compared with human samples, food (pork, chicken, and vegetable/fruit), soil, surface water, and fly microbes showed a very high load of multidrug resistance genes (from 1.460 to 5.236 ARG copies per 16S rRNA gene copy). Interestingly, microbes from food and the environment contained a significantly higher load of ARGs than those from human feces (Kruskal–Wallis test, P = 2.9 × 10 −14 ; Fig. c), especially for multidrug ( P < 2.2 × 10 −16 ), aminoglycoside ( P = 7.0 × 10 −16 ), unclassified ( P < 2.2 × 10 −16 ), and bacitracin ( P < 2.2 × 10 −16 ) resistance genes (Additional file : Fig. S1b and Additional file : Table S4). In humans, dietary habits and occupational exposure were found to affect ARG abundance, especially for the top 20 most abundant ARG types (Kruskal–Wallis tests, P < 0.05; 90.0%, 18/20; Additional file : Fig. S1c and Additional file : Table S5). For example, samples from boarding students showed a higher load of beta-lactam resistance genes; samples from pork abstainers exhibited a lower load of glycopeptide resistance genes; samples from livestock farmers carried a higher load of phenicol resistance genes. Specific aminoglycoside and tetracycline resistance genes were detected at higher levels in the human fecal samples, including tetM , tetW , tet(W/NW) , tetO , tet40 , AAC(6’)-Ie-APH(2’’)-Ia , BANAP , and APH(2’’)-If (Fig. d). Diversity analyses indicated that ARG alpha diversity, as measured by the Shannon index, was lower in human fecal samples than in samples from other groups (Kruskal–Wallis test, P < 2.2 × 10 −16 ; Fig. e), while at the subgroup level, samples from poultry feces, flies, and wastewater had higher Shannon indices (Kruskal–Wallis test, P < 2.2 × 10 −16 ; Additional file : Fig. S2a); among the human samples, compared with omnivores, vegan communes exhibited higher ARG alpha diversity ( P = 1.9 × 10 −4 ). Beta diversity analysis revealed the clustering of samples from human, food, and environmental sources into three distinct groups (adonis, P < 0.001, R 2 = 36.4%; Fig. f and Additional file : Table S6), highlighting the influence of habitat on the ARG composition. Principal coordinate analysis (PCoA) revealed significant differences in ARG subtypes according to the Axis1 values between the human and animal fecal samples and the other samples (Kruskal–Wallis test, P < 2.2 × 10 −16 ; Additional file : Fig. S2b), with the exception of wastewater samples. We then compared the Bray–Curtis distances between the resistome profiles of the human fecal samples and other samples. Our PCoA analyses were supported by the comparison of Bray–Curtis distances, which indicated human fecal samples were more similar to animal feces and wastewater samples compared to other samples (Additional file : Fig. S2c). A further analysis of the abundance of the Bacteroides bacteriophages B40-8 and crAssphage, indicators of fecal contamination , were highly represented in the fecal and wastewater samples (Kruskal–Wallis test, P < 2.2 × 10 −16 ; Fig. g and Additional file : Table S7). We next analyzed the microbiome profiles from the metagenomic sequencing data to establish the connections between microbial and ARG compositions. At the phylum level, a total of 31 phyla with varied abundance among samples from different sources was identified (Fig. a). Bacillota (with a relative abundance of 53.0%) and Bacteroidota (34.7%) were the dominant phyla in human fecal samples, whereas the Pseudomonadota occupied a higher proportion in food (43.0%) and environmental samples (31.3%). By binning the metagenomic contigs, we generated 6067 strain-level metagenome-assembled genomes (MAGs, 99.0% average nucleotide identity [ANI]) and 1302 species-level MAGs (95.0% ANI) from a total of 14,787 MAGs. Among these species, 230 species were identified as putative novel species (Additional file : Fig. S3a and Additional file : Table S8). Diversity analyses revealed significantly lower Shannon indices for food samples than for human and environmental samples (Kruskal–Wallis test, P < 2.2 × 10 −16 ; Fig. b), and the microbial compositions differed significantly among the three groups of samples (PERMANOVA, P = 0.001, R 2 = 10.9%; Fig. c and Additional file : Table S9). We then performed Procrustes analysis to assess the extent to which the microbial composition influenced the resistome profiles and found a significant correlation between the resistome profiles and the microbial composition ( P = 0.001, R 2 = 74.0%; Fig. d). Further, we applied a random forest model to identify the main contributors (at different taxonomic levels) influencing the resistome profiles. The results showed that E. coli , a typical member of Pseudomonadota at the species level, was determined to be the main taxon distinguishing the resistome profiles (Fig. e, f and Additional file : Tables S10–S11). Differences in abundance between the various groups of samples revealed the following: the food samples were higher in the abundance of Gammaproteobacteria ( P < 2.2 × 10 −16 ), Enterobacterales ( P = 2.5 × 10 −12 ), and Enterobacteriaceae ( P = 1.2 × 10 −6 ) than the other two groups of samples, while the environmental samples exhibited higher abundances of Pseudomonadota ( P < 2.2 × 10 −16 ) and Gammaproteobacteria ( P < 2.2 × 10 −16 ) than the human fecal samples, and E. coli ( P = 1.5 × 10 −7 ) and Escherichia ( P = 6.3 × 10 −7 ) were more abundant at the species and genus levels in the human fecal samples (Fig. f and Additional file : Table S11). Spearman’s correlation analyses further verified the close associations between the abundance of these taxa and the abundance and diversity of the ARGs ( P < 0.05; Fig. g and Additional file : Table S12). We compared the relative contributions of vertical and horizontal transfer of ARGs, and found that in food samples, the resistome was more affected by MGEs, whereas in environmental samples, it was more associated with the microbiome (Additional file : Fig. S3b). To reveal the putative bacterial host of ARGs, we assembled the sequence reads into contigs and used these contigs for taxonomic assignment. After sequence assembly and ARG annotation, we identified a total of 162,001 ARG-carrying contigs (ACCs) containing 183,720 ARGs from all 592 samples. Taxonomic assignment based on sequence information classified 20.4% of the ACCs at the genus or species level; however, one-third of the ACCs could not be classified, even at the phylum level (Additional file : Fig. S3c). A total of 54,831 ACCs, accounting for 33.8% of the total ACCs, could be binned into MAGs, of which, 14,367 had accurate taxonomic assignments according to the taxonomic information of the associated MAGs (Additional file : Fig. S3d). Overall, we finally assigned ~ 25.0% of the ACCs to the species or genus level (Additional file : Fig. S3e and Additional file : Table S13). We then computed the ARG load index at different taxonomic levels, i.e., the proportion of ARGs assigned to each taxon divided by the average relative abundance of the corresponding taxon. At the phylum level, Pseudomonadota was found to be the main ARG carrier (load index was 1.5), followed by Bacillota (0.8), Actinomycetota (0.3), and Bacteroidota (0.2) (Additional file : Table S14). At the class and order levels, Gammaproteobacteria (1.6) and Enterobacterales (1.1), respectively, both belonging to Pseudomonadota , contributed to the highest load of ARGs. Enterobacteriaceae plasmids and prophages/phages MGEs such as plasmids and prophages/phages are known as a reservoir for ARGs. A total of 13.0% (21,134/162,001) of our assembled ACCs were predicted to be plasmid sequences, and 2.0% (3234/162,001) of ACCs contained prophage/phage sequences (Fig. a, b). These sequences together carried 12.3% (22,571/183,720) of the total ARGs. The proportions of plasmid sequences differed among human, food, and environmental samples (Kruskal–Wallis test, P < 2.2 × 10 −16 ), with food samples exhibiting the highest ratio of plasmid sequences compared with human fecal and environmental samples (Fig. c). Seen from the ARG composition, the plasmid sequences in food, fly, and surface water samples were found to carry more multidrug resistance genes (Additional file : Fig. S4a). Replicon-based typing of the plasmid sequences identified 113 plasmid types that were associated with 102 ARG subtypes (Additional file : Fig. S4b), with IncQ1 ( n = 20), and rep_cluster_1232 ( n = 10) were the top frequently encountered replicons. To investigate the potential of these plasmid types in disseminating ARGs, we searched the PLSDB database for the replicons we identified in our samples and annotated the ARGs carried by the known plasmid sequences (Additional file : Fig. S4c). In total, 291 ARG subtypes were found to be carried by the 113 plasmid types, and plasmids with replicon of IncHI2A/rep_cluster_1088 carried the highest number of ARG subtypes ( n = 106), followed by IncFIB ( n = 101) and IncR ( n = 77). Among these 291 subtypes in the database, 24.4% ( n = 71) were found carried by the plasmids found in our samples. Interestingly, we found eight ARG subtypes carried by our plasmids that have not been reported previously (Fig. d and Additional file : Table S15). For example, we showed that tetL gene encoding tetracycline efflux protein can be carried by rep_cluster_1018/rep_cluster_1118, rep_clsuter_1118, and rep_cluster_2013 type plasmids. The prophage/phage fragments found in our samples were predominantly from the family Peduoviridae (11.3% in human fecal samples, 33.8% in food samples, and 21.5% in environmental samples; Additional file : Fig. S5a–c). Glycopeptide and peptide resistance genes were found to be more abundant in prophages/phages from both human and fecal samples, while the unclassified DNA binding protein H-NS was highly represented in phages from food samples (Additional file : Fig. S5d and Additional file : Table S16). The glycopeptide resistance gene vanU and the peptide resistance gene ugd were the two common phage/prophage ARGs found in all three groups of samples. Taking human fecal samples as an example, we found that the frequency of occurrence of each ARG among our samples was highly consistent with that in the Gut Phage Database (GPD) of human gut phages (Spearman’s correlation, R 2 = 0.852, P = 5.8 × 10 −12 ; Fig. e). We then predicted the hosts of the prophages/phages and found that Bacillota and Pseudomonadota were found to be the main hosts of the phages at the phylum level (Additional file : Fig. S5e), and Enterobacteriaceae was the main host at the family level (Fig. f), especially for prophages/phages in samples from chickens, flies, and surface water (Additional file : Fig. S5f). A total of 24 ARG subtypes were found carried by the Enterobacteriaceae prophages/phages, including genes encoding multidrug and tetracycline resistance. We then investigated ARG transmission among different habitats by clustering the ARG nucleotide acid sequences from various sources at 100.0% identity and constructed ARG-sharing networks (Additional file : Fig. S6a and Additional file : Tables S17–S19). According to the ARG risk ranks classified previously , we categorized the 328 ARG subtypes into six groups (from high- to low-risk ranks): Q1 ( n = 88), Q2 ( n = 45), Q3 ( n = 35), Q4 ( n = 42), RI = 0 ( n = 79), and Unknown ( n = 41) (Additional file : Table S20). We found that ARG subtypes in Q1 had higher sharing frequencies than ARG subtypes in other categories (Kruskal–Wallis test, P < 2.8 × 10 −13 ; Additional file : Fig. S6b and Additional file : Table S21). Additionally, there was a higher ratio of ARGs shared between human fecal samples and food, animal fecal, wastewater, and fly samples (Fig. a), especially those in the Q1 category (Additional file : Fig. S6c). This was also the case when different human subgroups were separately compared with the food and environmental subgroups (Fig. b). Typically, compared with other human subgroups, the human fecal samples from the vegetarian subgroup exhibited the highest ratio of ARGs shared with vegetables or fruits (41.9%), while the lowest ratio with pork (4.2%). A total of 78 ARGs (Q1) were found to be shared in at least one comparison pair between fly, pork, vegetables/fruits, and different human subgroups (Fig. c and Additional file : Table S22), and interestingly, flies shared a relatively high ratio of ARGs with food processing workers. Taken together, these results highlight the important role of the high-risk ARGs in AMR spread. We then used StrainPhlAn to create maps of strain-sharing events between samples to reveal regional strain transmission. We showed that 257 species-level genome bins (SGBs) were present in at least 20 samples (Additional file : Table S23). The most prevalent SGB was Agathobacter rectalis ( n = 328) from the family Lachnospiraceae , followed by Blautia wexlerae ( n = 311), Faecalibacterium prausnitzii ( n = 299), Prevotella copri ( n = 298), and Blautia_A faecis ( n = 298). Among six species of Enterobacteriaceae , E. coli (SGB10068) was the most prevalent species in our samples ( n = 249), consisting of three different strain lineages (I to III). Lineage I was the largest group present in 235 samples from multiple sources, across human, fly, food (pork, chicken, vegetables, and fruits), and wastewater. For other species in Enterobacteriaceae , the same strain lineage of Klebsiella pneumoniae , K. michiganensis , K. ornithinolytica , Clostridium freundii , and Enterobacter hormaechei was also found shared among samples of different origins, mainly human, food (chicken, pork, vegetables, and fruits) and fly. Notably, flies may play an important role in the transmission of Enterobacteriaceae strains, as samples from flies appeared in almost all of the main strain lineages in the phylogenetic trees (Fig. d). Apart from the Enterobacteriaceae , strain transmission events between human and other samples were also detected for species from other families, including C. perfringens , Enterococcus faecium , Weissella confusa , and Streptococcus pasteurianus (Additional file : Fig. S7). For example, one C. perfringens strain was found in 51 samples across human, vegetable/fruit, fly, pork, and chicken, representing a typical regional strain transmission event of opportunistic pathogen. To further reveal regional ARG transmission, we selected carbapenemase genes as representative ARGs. We identified a total of 111 subtypes of carbapenemase genes in the ACCs from all of our samples (Additional file : Table S24). OXA-347 ( n = 308) was found to be the most prevalent gene carried by the ACCs (Additional file : Fig. S8a), being predominantly present in human and animal fecal samples, and wastewater (Additional file : Fig. S8b, c). To uncover the potential transmission of carbapenemase genes, we analyzed the sharing events of these genes across various sources. A total of 200 unique sequence types were identified from the top 20 frequently encountered carbapenemase genes (1 to 30 distinct sequence types per gene), clustering with a cut-off of 100.0% nucleotide identity (Additional file : Fig. S9). Among these 20 genes, 11 genes were found to be shared among samples of different sources, suggesting complex dissemination pathways of these genes. For example, OXA-347 , OXA-1 , OXA-85 , and CfiA2 showed a high sharing ratio between feces and wastewater. Interestingly, ACT-12 and ORN-1 showed a high sharing ratio between food and fly samples, suggesting that flies may serve as mediators for ARG cross-transmission. Of note, a sequence type of OXA-50 was detected in poultry feces, livestock farmers, and food processing workers. There were rare instances of sharing events between chicken or pork and poultry or pork abstainers. Interestingly, vegans who live together were more likely to carry specific carbapenemase genes, such as SHV-27 , ORN-1 , and ACT-28 , possibly due to person-to-person transmission. We next isolated carbapenem-resistant Enterobacteriaceae strains from our 592 samples. In total, 45 resistant isolates affiliated with nine bacterial species were identified from 42 samples (Fig. a and Additional file : Table S25). The annotation results indicated that 17 carbapenemase gene subtypes were carried by these isolates, among which six subtypes, namely, IMP-4 , NDM-3 , NDM-5 , OXA-805 , SFO-1 , and SHV-61 , were only found in the pure culture isolates, not in the metagenomic data. The carbapenemase genes recovered from these isolates accounted for only a small fraction of the subtypes found in the metagenomic data ( n = 111), indicating the presence of other carriers besides Enterobacteriaceae members. We then clustered each species by including the corresponding genomes of the isolates and MAGs that were assembled from the metagenomic sequencing data. We observed potential transmission of the same strain (ANI ≥ 99.0%) harboring different carbapenemase genes across different sample sources (Fig. b and Additional file : Table S26). For example, the E. coli strain cluster 1 carrying different carbapenemase genes (such as NDM-1 , NDM-5 , OXA-1 , OXA-10 , and CTX-M-27 ) was widely distributed among 141 samples covering human fecal samples and another seven subgroups (except for samples from surface water and wastewater). Similar situations were also found for E. coli strain clusters 2 and 3 containing NDM-1 or NDM-5 , and the K. pneumoniae strain cluster carrying SHV-61 or SHV-33 . A strain of K. ornithinolytica harboring three carbapenemase genes ( SFO-1 , ORN-1 , and NDM-1 ) was found to only be present in fly and human samples, again suggesting the significant role of fly-mediated transmission of ARGs. Our metagenomic data showed that carbapenemase gene OXA-347 was prevalent in human and animal fecal samples, however, we did not recover isolates carrying carbapenemase gene OXA-347 , probably because the SuperCARBA medium used mainly targets Enterobacteriaceae bacteria. To determine the host range and potential dissemination of OXA-347 , we searched for this gene among the complete genomes in the database (Additional file : Fig. S10). The OXA-347 gene was present in 12 bacterial species associated with two known types of plasmid replicons (rep_cluster_663 and rep_cluster_1097) in three Bacteroides species ( B. fragilis , B. thetaiotaomicron , and B. xylanisolvens ) (Fig. c and Additional file : Table S27) and two unknown types of plasmids in Myroides albus and Sphingobacterium faecium . Additionally, the chromosome-located OXA-347 gene was found to be frequently flanked by insertion sequences such as IS1380, IS91, and IS1595. These results suggest a high spread risk of OXA-347 via horizontal gene transfer. Out of the 12 bacterial species possessing OXA-347 , four were found to be common inhabitants in both human feces and wastewater samples (Fig. d). We finally constructed random forest models to evaluate the potential of utilizing metagenomic sequencing data for predicting carbapenem-resistant strains. The results showed that all prediction models could achieve an area under the curve (AUC) exceeding 0.90 when constructed based on microbiome profiles, resistome profiles, or a combination of both (Fig. a–c). Notably, microbiome profiles proved to be robust indicators with an AUC of 0.939 and a specificity of 0.595. Among the identified predictors, Aeromonas hydrophila , Bacilli bacterium , Alistipes putredinis , and Blautia wexlerae were considered strong contributors (Fig. d). In this study, we revealed the regional ARG flow among humans, food, and the environment in a county-level city in China by applying a One Health sampling approach. The key findings from this study were (1) antibiotic resistomes are habitat-specific, and human or animal fecal contamination is an important factor that influences the wastewater ARG composition; (2) Enterobacteriaceae bacteria together with their plasmids and bacteriophages are the main ARG carriers; (3) flies and food may be important mediators for the regional spread of ARGs; (4) HGT and strain transmission independently or jointly contribute to regional AMR dissemination; (5) the carbapenemase gene OXA-347 is widely present among human and animal gut microbiomes; and (6) a microbiome profile-based machine learning model can predict the presence of carbapenem-resistant strains in metagenomic samples. We found that Pseudomonadota bacteria, especially those from the Enterobacteriaceae , are the primary drivers that shape the resistome, which is consistent with a previous study . The Enterobacteriaceae family encompasses numerous pathogens commonly identified in clinical cultures and poses a major risk to human health as a result of their role as primary reservoirs for ARGs . Our findings indicate that members of the Enterobacteriaceae are also the major ARG carriers in non-clinical settings, supporting the use of Enterobacteriaceae as indicators for AMR surveillance worldwide . Additionally, we showed that Enterobacteriaceae carrying carbapenemase genes ( SHV-61 , NDM-1 , OXA-1 ) were present in surface water, pork, and fly samples, highlighting the potential of regional Enterobacteriaceae -mediated dissemination of clinically important ARGs. HGT mediated by MGEs is recognized as the major reason for AMR dissemination. ARGs found in MGEs are even more “global” than microbes, as illustrated by the ability of mobile ARGs to cross habitat boundaries . The higher ratio of plasmids sequences in food samples we presented here suggests the higher risk of ARG transmission from food to humans (Fig. c). Besides plasmids, we showed that Enterobacteriaceae prophages/phages were also important ARG carriers in both our samples and the GPD database (Fig. e–f), supporting the significant role of phages in the transfer of ARGs . Although HGT greatly contributes to ARG dissemination among microbial communities, strain transmission or clone spread can be regarded as a major route for regional epidemics of both AMR and infections. This was strongly supported by our finding that a single E. coli strain (SGB10068 Lineage I) was present in 235 samples from multiple sources, and an E. coli strain (cluster 1) carrying carbapenemase genes was widely distributed among 141 samples. Notably, different carbapenemase genes were found to be carried by the same strain [i.e., E. coli strain (cluster 1)], suggesting HGT events (Fig. d). These findings imply that regional AMR dissemination involves strain transmission accompanied by HGT and ARG exchange on MGEs such as plasmids and prophages/phages. Additionally, the underestimation of carbapenemase genes by metagenomics may reflect the fact that many are harbored in microbes, especially ARB that are difficult to cultivate. In this study, we observed a high prevalence of the carbapenemase gene OXA-347 in human and animal feces, as well as in wastewater. OXA-347 is associated with phenotypic resistance to penicillin, cephalosporins, and imipenem . This gene has been found in species from the genera Bacteroides , Myroides , and Capnocytophaga from various sources . In wastewater samples from the swine feedlots, OXA-347 was found the most abundant beta-lactam resistance gene , and compared with swine fecal samples, the abundance of OXA-347 was significantly higher in the human gut microbiome . Notably, consistent with our results, recent genomic analyses suggest that OXA-347 is likely located on MGEs, indicating its potential ability to move between microbes. Although OXA-347 is currently not considered a major clinical concern, given the fact that the administration of antibiotics led to an increase in the abundance of OXA-347 , possibly via HGT under selection pressure in the human microbiome , future research to evaluate its potential risk to human health should be a priority. The dissemination of AMR among humans, animals, and environments includes both direct and indirect pathways involving various One Health interfaces . Studies have suggested that food may act as a reservoir of ARB , indicating the critical role of diet in the transmission of ARGs from food to humans. This was evidenced by our finding that vegetarians shared more ARGs from vegetables or fruits than from pork. We also showed that vegetable/fruit microbes carried a relatively higher abundance of ARGs, probably because vegetables are accessible to microbial contamination via different routes such as manure fertilization and wastewater irrigation . In addition to diet, we identified flies as a key environmental factor that facilitates ARG dissemination. Fly microbes were found to carry the highest relative abundance of ARGs among all the subgroup samples, and frequently shared Enterobacteriaceae strains as well as carbapenemase genes with other samples (Fig. d). As a result of their omnivorous diet and breeding habits , flies may play a significant but neglected role in the spread and transmission of pathogens and ARGs among the One Health sectors. Our findings highlight the importance of the vegetable- and fly-mediated ARG transmission routes and strongly suggest that these may be new points of intervention for the control of ARG spread. Additionally, we found potential person-to-person transmission of carbapenemase genes among vegan communes and showed the flow of the carbapenemase gene along the food supply chain (Additional file : Fig. S9), providing specific examples of the impact of lifestyle, dietary habits, and occupational exposure on the transmission of ARG to humans. Taken together, these results highlight the potential routes of the AMR spread, especially through the food-chain dissemination pathway and occupational exposure. In summary, we give a landscape of regional ARG flow among humans, food, and the environment in a city in China. We highlight that the resistome profiles in samples of different origins showed habitat specificity, and Pseudomonadota members are the major contributors shaping the resistome through strain transmission and/or MGEs (plasmids and prophages or phages)-mediated HGT. We suggest that lifestyle, dietary habits, and occupational exposure are all risk factors that contribute to the spread of ARGs to humans. We also have evidence that the integration of metagenomic sequencing data and machine learning techniques could serve as a valuable approach for the surveillance of carbapenem-resistant strains. Collectively, the regional antimicrobial resistance gene flow we presented here provides real evidence of the AMR dynamics among the One Health sectors, and our findings highlight new points of focus for AMR surveillance and control in the future. We should stress that efforts are still needed to probe seasonal influences on the regional antimicrobial resistance gene flow, which was not taken into account in this study. Sample collection From October 2018 to April 2019, we collected a total of 592 samples from Dengfeng, Henan Province, China (Additional file ), deriving from humans, food, and the environment. In detail, we collected fecal samples from individuals with various dietary patterns, including omnivores ( n = 49), vegetarians ( n = 49), pork abstainers ( n = 50), chicken abstainers ( n = 46), and aquatic product abstainers ( n = 44). We also collected fecal samples from humans with different lifestyles or occupational exposures, including vegan communes ( n = 68), boarding students ( n = 48), food processing workers ( n = 50), and livestock farmers ( n = 50). Additionally, 51 samples from various food sources were collected, including pork ( n = 13), chicken ( n = 11), and vegetables and fruits ( n = 27); and 87 samples from different environments were collected, including soil ( n = 10), surface water ( n = 9), wastewater ( n = 4), flies ( n = 20), poultry feces ( n = 24), and swine feces ( n = 20). All samples were used for culturing carbapenem-resistant microbes and metagenomic sequencing. DNA extraction and metagenomic sequencing For metagenomic sequencing, DNA extraction was performed using the QIAamp Power Fecal DNA Kit (Qiagen, Hilden, Germany), following the manufacturer’s instructions. The DNA concentration was measured using a Qubit dsDNA Assay Kit and Qubit 2.0 fluorometer (Life Technologies, CA, USA), and the integrity was assessed using 0.8% agarose gel electrophoresis. Libraries were constructed using the MGIEasy FS DNA Library Prep Set (MGI Tech, Shenzhen, China). A paired-end library with an insert size of ~ 350 bp was constructed for each sample and sequenced on the MGISEQ-2000RS platform (MGI Tech). To avoid potential contamination indicated in a previous study , we also incorporated negative controls during the processes of library construction and metagenomic sequencing. The metagenomic sequencing of the 592 samples generated a total of 3492.7 Gb (6.0 Gb per sample after quality control, with a standard deviation of 1.9 Gb). Microbiome profiling and resistome profiling For the metagenomic sequencing data, we excluded low-quality bases and residual adapter contamination using Trim Galore ( https://github.com/FelixKrueger/TrimGalore ). To remove human DNA contamination, the sequencing data were mapped to the human genome (hg38) using Bowtie2 v2.3.5.1 . Taxonomic profiling of the metagenomic sequencing samples was performed using MetaPhlAn v4.0.2 with default parameters. The construction of the sample-specific strains of all species was performed on the samples using StrainPhlAn v4.0.2 . We considered SGBs that met the following criterion: present in at least three samples in one subgroup of both human and food or environmental origin. To detect strain-sharing events, we first calculated the SGB-specific normalized phylogenetic distance (nGD) using pyphlan ( https://github.com/SegataLab/pyphlan ). Strain boundaries were set to below the threshold of 0.1 of the nGD. Resistome profiling of the samples was performed using deepARG v2 with the pipeline short-read pipeline to predict ARGs directly from short reads. The diversity of the microbiome and the resistome was calculated using the R package vegan v2.6–4. The presence of the crAssphage genome was regarded as an indicator of human or animal fecal contamination . We used Bowtie v2.3.5.1 to align the clean metagenomic sequencing reads to the reference genomes of Carjivirus communis (JQ9955537.1) and Bacteroides phage B40-8 (NC_011222.1), as recommended previously . The average coverage of each phage genome was calculated using SAMtools v1.9 . Subsequently, the phage coverage was normalized by the data size of each sample (copies/Gb) to compare the profiles among the samples from various sources. ExtrARG , a machine-learning approach using the extremely randomized tree algorithm, was utilized to identify discriminatory ARGs. Values with an importance greater than 0.004 were considered biomarkers from various sources. To identify the taxa that were associated with the resistome profiles, we applied feature selection by sorting the mean decrease in Gini values generated by the R package randomForest v4.7–1.1. Metagenomic assembly and genome binning The clean reads from each set of metagenomic sequencing data were independently assembled using MEGAHIT v1.1.3 . For metagenomic binning, three methods, namely MetaBAT2 v2.12.1 , Maxbin2 v2.2.6 , and Concoct v1.0.0 , were used. A superior bin set from multiple original binning predictions was produced using the bin_refinement module of metaWRAP v1.3.2 . This module combines the three original binning predictions. The completeness and contamination of each bin from the superior bin set were evaluated using CheckM v1.0.12 . Then, bins with a completeness of ≥ 70.0% and a contamination rate of ≤ 10.0% were retained. All MAGs were dereplicated at 99.0% ANI and 95.0% ANI using dRep v2.6.2 . The bin annotation pipeline of CAT v5.2.3 was used to assign taxonomy to MAGs. A genome was classified as a novel species if the ANI output was less than 95.0%. We also compared the MAGs with the isolate genomes by clustering at 99.0% ANI. The phylogenetic trees were inferred using PhyloPhlAn v3.0.60 . Correlations between the MGEs and the microbiome To determine the correlations between the resistome and both the microbiome (at the phylum level) and the MGEs in Additional file : Fig. S3b, the abundance matrices were analyzed through Procrustes analysis and a “protest” test. Here, the abundance of MGEs and the 16S rRNA sequence for each sample was calculated by mapping to the MobileGeneticElement and SILVA databases, respectively. The MGE profiles were normalized by the copies of MGEs per 16S rRNA gene. ARG annotation and taxonomic assignment of ACCs The gene contents in the assembled contigs were predicted using GeneMark-HM v2.07 , and ARGs were further predicted using the “DeepARG-LS” mode of deepARG. To maintain consistency with the taxonomic assignment of the MAGs, we also utilized the contig annotation tool in CAT to predict the taxonomic assignment of the ACCs. To reveal the risk ranks of the discovered ARGs, the ARGs identified by deepARG were re-annotated against CARD to ensure consistency with the risk rank assignments made in a previous study . The ARG load index at different taxonomic levels was calculated using the following formula: the proportion of ARGs assigned to each taxon divided by the average relative abundance of the corresponding taxon. Analyses of plasmid and prophage/phage sequences We used geNomad v1.5.2 , a classification and annotation framework, to find plasmid and prophage/phage sequences from the ACCs. Taxonomic assignment of the phage fragments generated from ACCs was performed using PhaGCN2 , followed by host prediction using HostG . Because some segments of the ACCs contained both prophage fragments and microbial genomes, we extracted the ARGs located within the prophage fragments using local scripts. The predicted plasmid sequences were typed using the mob_typer method implemented in MOB-suite v3.1.0 based on the replicon, relaxase, and cluster assignment . This method provides in silico predictions of the replicon family, relaxase type, mate-pair formation type, and predicted transferability of the plasmid. The ARG annotation of the predicted prophage/phage-related and plasmid-related ACCs followed the aforementioned guidelines. To reveal the prophage/phage-associated ARGs on a global scale, we downloaded the GPD , a collection of 142,809 non-redundant viral genomes obtained by mining a dataset of 28,060 globally distributed human gut microbiomes. The ARG annotation followed the same pipeline as mentioned above. We established the potential connections between the plasmid replicon types and the ARGs through further analysis of the PLSDB database , a plasmid database obtained from the NCBI nucleotide database, which comprises 34,513 sequences. Determination of ARG-sharing networks Over a shorter time span, the substitution rate of bacteria typically falls within the range of ~ 1 single nucleotide polymorphism per genome per year . Thus, as suggested by a previous study , the recently transmitted ARGs should have identical sequences (100.0% identity) between two sources, considering the length of the ARGs. To identify potential ARG transmission among various sources, we conducted the following steps: (1) clustering of the DNA sequences of the ARGs discovered in the metagenomic assemblies from each source, with 100.0% identity and 85.0% coverage, using the scripts provided by checkV v1.0.1 to generate the non-redundant set of ARGs from various sources; and (2) comparing the non-redundant sets of ARGs from various sources and constructing the ARG-sharing network between human and other samples. Isolation and antimicrobial susceptibility testing of carbapenem-resistant strains SuperCARBA (CHROMagar, Paris, France) was used to isolate carbapenem-resistant bacteria, mainly Enterobacteriaceae . Colonies with various morphologies were selected from each Petri dish. A matrix-assisted laser desorption/ionization time-of-flight mass spectrometry system (Zybio Inc., Chongqing, China) was used for the early identification of isolates. Antimicrobial susceptibility testing of 28 antimicrobial agents (amikacin, amoxicillin-clavulanate, ampicillin-sulbactam, aztreonam, cefazolin, cefepime, cefoperazone-sulbactam, cefoxitin, ceftazidime, ceftriaxone, cefuroxime, chloramphenicol, ciprofloxacin, colistin, ertapenem, gentamicin, imipenem, levofloxacin, meropenem, minocycline, moxifloxacin, nitrofurantoin, norfloxacin, piperacillin-tazobactam, tetracycline, tigecycline, tobramycin, and trimethoprim-sulfamethoxazole) was performed using the agar dilution method following the standards and guidelines of the Clinical and Laboratory Standards Institute M100-2022 (V32) . E. coli ATCC 25922 and Enterococcus faecalis ATCC 29212 were used as quality control strains in each run. To obtain a more comprehensive view of carbapenem resistance gene carriage in bacteria, we analyzed the carbapenemase genes in complete genomes (41,185 genomes) downloaded from the National Center for Biotechnology Information (NCBI, July 29th, 2023) by CARD. We then used ISEScan v1.7.2.3 to identify the insertion sequence elements surrounding the carbapenemase gene OXA-347 . To determine the association between insertion sequences and ARGs on plasmids, the adjacent sequences, i.e. 5 kb upstream and downstream of ARGs, were extracted as previously suggested . Genome sequencing and bioinformatic analyses of the isolates For whole-genome sequencing, the genomic DNA of the 49 carbapenem-resistant isolates was extracted using a Wizard Genomic DNA Extraction Kit (Promega, Madison, WI, USA). The DNA concentration was measured using a Qubit dsDNA Assay Kit and a Qubit 2.0 fluorometer (Life Technologies), and the integrity was assessed using 1.0% agarose gel electrophoresis. Libraries were constructed using the MGIEasy FS DNA Library Prep Set (MGI Tech, Shenzhen, China). A paired-end library with an insert size of 350 bp was constructed for each sample and sequenced on the MGISEQ-200RS platform (MGI Tech, Shenzhen, China). Before assembling the microbial genomes, we performed quality control using Trim Galore ( https://github.com/FelixKrueger/TrimGalore ). The sequence reads were assembled using Unicycler v0.4.8 with default parameters, followed by annotation using Prokka v1.13.3 . To remove duplicate genomes from the same sample, we estimated genome similarity by calculating the pairwise Mash distance using Mash v2.3 and removing the duplicated genomes with MASH distances less than 0.001. Phylogenetic trees of all isolates were constructed using PhyloPhlAn with the default parameters. We estimated maximum likelihood trees for microbial genomes of the same species from isolates and MAGs using IQ-TREE 2 v2.1.4-beta . The best-fitting substitution model was automatically selected using the ModelFinder program implemented in IQ-TREE 2 and performed with 1000 bootstrap replicates. The ARGs from the microbial genomes were annotated against CARD. All draft genomes of the isolates and MAGs belonging to the same strain were clustered at 99.0% ANI using dRep. Predictive random forest models Random forest models were utilized to forecast the presence of carbapenem-resistant isolates based on microbiome profiles, resistome profiles, or a combination of both, across 592 samples. To assess the predictive performance of these random forest classification models, we used three indicators (AUC, sensitivity, and specificity), derived from a tenfold cross-validation in this study. In this process, the dataset was randomly divided into 10 equal subsets. During each iteration, nine subsets were used for model training, while the remaining subset was used for prediction. The values of each iteration were recorded. These analyses were performed using the R packages randomForest v4.7–1.1 and caret v6.0–94. Statistical analysis and visualization Differential analyses were performed by Wilcoxon rank sum tests or Kruskal–Wallis tests. When analyzing more than two groups, multiple comparisons were conducted using the R package agricolae v1.3–7. For beta diversity statistical analyses, the “adonis” function from the R package vegan v2.6–4 and the “pairwise.adonis” function from the R package pairwiseAdonis v0.4.1 were utilized. We employed Procrustes analysis to establish correlations among various profiles using the “procrustes” function of the R package vegan v2.6–4. The correlations were determined by the “protest” function of the R package vegan v2.6–4. The correlations between two variables in this study were calculated using the R function “cor.test”. The intersecting sets were analyzed using the R package UpSetR v1.4.0 . The co-occurrence networks were visualized using Cytoscape v3.10.1 . The heatmaps were visualized using the R package ComplexHeatmap v2.16.0 . The phylogenetic trees were visualized using the R package ggtree v3.8.2 or iTOL v5 . Figures were visualized primarily using the R packages ggplot2 v3.4.4 and patchwork v1.1.3. From October 2018 to April 2019, we collected a total of 592 samples from Dengfeng, Henan Province, China (Additional file ), deriving from humans, food, and the environment. In detail, we collected fecal samples from individuals with various dietary patterns, including omnivores ( n = 49), vegetarians ( n = 49), pork abstainers ( n = 50), chicken abstainers ( n = 46), and aquatic product abstainers ( n = 44). We also collected fecal samples from humans with different lifestyles or occupational exposures, including vegan communes ( n = 68), boarding students ( n = 48), food processing workers ( n = 50), and livestock farmers ( n = 50). Additionally, 51 samples from various food sources were collected, including pork ( n = 13), chicken ( n = 11), and vegetables and fruits ( n = 27); and 87 samples from different environments were collected, including soil ( n = 10), surface water ( n = 9), wastewater ( n = 4), flies ( n = 20), poultry feces ( n = 24), and swine feces ( n = 20). All samples were used for culturing carbapenem-resistant microbes and metagenomic sequencing. For metagenomic sequencing, DNA extraction was performed using the QIAamp Power Fecal DNA Kit (Qiagen, Hilden, Germany), following the manufacturer’s instructions. The DNA concentration was measured using a Qubit dsDNA Assay Kit and Qubit 2.0 fluorometer (Life Technologies, CA, USA), and the integrity was assessed using 0.8% agarose gel electrophoresis. Libraries were constructed using the MGIEasy FS DNA Library Prep Set (MGI Tech, Shenzhen, China). A paired-end library with an insert size of ~ 350 bp was constructed for each sample and sequenced on the MGISEQ-2000RS platform (MGI Tech). To avoid potential contamination indicated in a previous study , we also incorporated negative controls during the processes of library construction and metagenomic sequencing. The metagenomic sequencing of the 592 samples generated a total of 3492.7 Gb (6.0 Gb per sample after quality control, with a standard deviation of 1.9 Gb). For the metagenomic sequencing data, we excluded low-quality bases and residual adapter contamination using Trim Galore ( https://github.com/FelixKrueger/TrimGalore ). To remove human DNA contamination, the sequencing data were mapped to the human genome (hg38) using Bowtie2 v2.3.5.1 . Taxonomic profiling of the metagenomic sequencing samples was performed using MetaPhlAn v4.0.2 with default parameters. The construction of the sample-specific strains of all species was performed on the samples using StrainPhlAn v4.0.2 . We considered SGBs that met the following criterion: present in at least three samples in one subgroup of both human and food or environmental origin. To detect strain-sharing events, we first calculated the SGB-specific normalized phylogenetic distance (nGD) using pyphlan ( https://github.com/SegataLab/pyphlan ). Strain boundaries were set to below the threshold of 0.1 of the nGD. Resistome profiling of the samples was performed using deepARG v2 with the pipeline short-read pipeline to predict ARGs directly from short reads. The diversity of the microbiome and the resistome was calculated using the R package vegan v2.6–4. The presence of the crAssphage genome was regarded as an indicator of human or animal fecal contamination . We used Bowtie v2.3.5.1 to align the clean metagenomic sequencing reads to the reference genomes of Carjivirus communis (JQ9955537.1) and Bacteroides phage B40-8 (NC_011222.1), as recommended previously . The average coverage of each phage genome was calculated using SAMtools v1.9 . Subsequently, the phage coverage was normalized by the data size of each sample (copies/Gb) to compare the profiles among the samples from various sources. ExtrARG , a machine-learning approach using the extremely randomized tree algorithm, was utilized to identify discriminatory ARGs. Values with an importance greater than 0.004 were considered biomarkers from various sources. To identify the taxa that were associated with the resistome profiles, we applied feature selection by sorting the mean decrease in Gini values generated by the R package randomForest v4.7–1.1. The clean reads from each set of metagenomic sequencing data were independently assembled using MEGAHIT v1.1.3 . For metagenomic binning, three methods, namely MetaBAT2 v2.12.1 , Maxbin2 v2.2.6 , and Concoct v1.0.0 , were used. A superior bin set from multiple original binning predictions was produced using the bin_refinement module of metaWRAP v1.3.2 . This module combines the three original binning predictions. The completeness and contamination of each bin from the superior bin set were evaluated using CheckM v1.0.12 . Then, bins with a completeness of ≥ 70.0% and a contamination rate of ≤ 10.0% were retained. All MAGs were dereplicated at 99.0% ANI and 95.0% ANI using dRep v2.6.2 . The bin annotation pipeline of CAT v5.2.3 was used to assign taxonomy to MAGs. A genome was classified as a novel species if the ANI output was less than 95.0%. We also compared the MAGs with the isolate genomes by clustering at 99.0% ANI. The phylogenetic trees were inferred using PhyloPhlAn v3.0.60 . To determine the correlations between the resistome and both the microbiome (at the phylum level) and the MGEs in Additional file : Fig. S3b, the abundance matrices were analyzed through Procrustes analysis and a “protest” test. Here, the abundance of MGEs and the 16S rRNA sequence for each sample was calculated by mapping to the MobileGeneticElement and SILVA databases, respectively. The MGE profiles were normalized by the copies of MGEs per 16S rRNA gene. The gene contents in the assembled contigs were predicted using GeneMark-HM v2.07 , and ARGs were further predicted using the “DeepARG-LS” mode of deepARG. To maintain consistency with the taxonomic assignment of the MAGs, we also utilized the contig annotation tool in CAT to predict the taxonomic assignment of the ACCs. To reveal the risk ranks of the discovered ARGs, the ARGs identified by deepARG were re-annotated against CARD to ensure consistency with the risk rank assignments made in a previous study . The ARG load index at different taxonomic levels was calculated using the following formula: the proportion of ARGs assigned to each taxon divided by the average relative abundance of the corresponding taxon. We used geNomad v1.5.2 , a classification and annotation framework, to find plasmid and prophage/phage sequences from the ACCs. Taxonomic assignment of the phage fragments generated from ACCs was performed using PhaGCN2 , followed by host prediction using HostG . Because some segments of the ACCs contained both prophage fragments and microbial genomes, we extracted the ARGs located within the prophage fragments using local scripts. The predicted plasmid sequences were typed using the mob_typer method implemented in MOB-suite v3.1.0 based on the replicon, relaxase, and cluster assignment . This method provides in silico predictions of the replicon family, relaxase type, mate-pair formation type, and predicted transferability of the plasmid. The ARG annotation of the predicted prophage/phage-related and plasmid-related ACCs followed the aforementioned guidelines. To reveal the prophage/phage-associated ARGs on a global scale, we downloaded the GPD , a collection of 142,809 non-redundant viral genomes obtained by mining a dataset of 28,060 globally distributed human gut microbiomes. The ARG annotation followed the same pipeline as mentioned above. We established the potential connections between the plasmid replicon types and the ARGs through further analysis of the PLSDB database , a plasmid database obtained from the NCBI nucleotide database, which comprises 34,513 sequences. Over a shorter time span, the substitution rate of bacteria typically falls within the range of ~ 1 single nucleotide polymorphism per genome per year . Thus, as suggested by a previous study , the recently transmitted ARGs should have identical sequences (100.0% identity) between two sources, considering the length of the ARGs. To identify potential ARG transmission among various sources, we conducted the following steps: (1) clustering of the DNA sequences of the ARGs discovered in the metagenomic assemblies from each source, with 100.0% identity and 85.0% coverage, using the scripts provided by checkV v1.0.1 to generate the non-redundant set of ARGs from various sources; and (2) comparing the non-redundant sets of ARGs from various sources and constructing the ARG-sharing network between human and other samples. SuperCARBA (CHROMagar, Paris, France) was used to isolate carbapenem-resistant bacteria, mainly Enterobacteriaceae . Colonies with various morphologies were selected from each Petri dish. A matrix-assisted laser desorption/ionization time-of-flight mass spectrometry system (Zybio Inc., Chongqing, China) was used for the early identification of isolates. Antimicrobial susceptibility testing of 28 antimicrobial agents (amikacin, amoxicillin-clavulanate, ampicillin-sulbactam, aztreonam, cefazolin, cefepime, cefoperazone-sulbactam, cefoxitin, ceftazidime, ceftriaxone, cefuroxime, chloramphenicol, ciprofloxacin, colistin, ertapenem, gentamicin, imipenem, levofloxacin, meropenem, minocycline, moxifloxacin, nitrofurantoin, norfloxacin, piperacillin-tazobactam, tetracycline, tigecycline, tobramycin, and trimethoprim-sulfamethoxazole) was performed using the agar dilution method following the standards and guidelines of the Clinical and Laboratory Standards Institute M100-2022 (V32) . E. coli ATCC 25922 and Enterococcus faecalis ATCC 29212 were used as quality control strains in each run. To obtain a more comprehensive view of carbapenem resistance gene carriage in bacteria, we analyzed the carbapenemase genes in complete genomes (41,185 genomes) downloaded from the National Center for Biotechnology Information (NCBI, July 29th, 2023) by CARD. We then used ISEScan v1.7.2.3 to identify the insertion sequence elements surrounding the carbapenemase gene OXA-347 . To determine the association between insertion sequences and ARGs on plasmids, the adjacent sequences, i.e. 5 kb upstream and downstream of ARGs, were extracted as previously suggested . For whole-genome sequencing, the genomic DNA of the 49 carbapenem-resistant isolates was extracted using a Wizard Genomic DNA Extraction Kit (Promega, Madison, WI, USA). The DNA concentration was measured using a Qubit dsDNA Assay Kit and a Qubit 2.0 fluorometer (Life Technologies), and the integrity was assessed using 1.0% agarose gel electrophoresis. Libraries were constructed using the MGIEasy FS DNA Library Prep Set (MGI Tech, Shenzhen, China). A paired-end library with an insert size of 350 bp was constructed for each sample and sequenced on the MGISEQ-200RS platform (MGI Tech, Shenzhen, China). Before assembling the microbial genomes, we performed quality control using Trim Galore ( https://github.com/FelixKrueger/TrimGalore ). The sequence reads were assembled using Unicycler v0.4.8 with default parameters, followed by annotation using Prokka v1.13.3 . To remove duplicate genomes from the same sample, we estimated genome similarity by calculating the pairwise Mash distance using Mash v2.3 and removing the duplicated genomes with MASH distances less than 0.001. Phylogenetic trees of all isolates were constructed using PhyloPhlAn with the default parameters. We estimated maximum likelihood trees for microbial genomes of the same species from isolates and MAGs using IQ-TREE 2 v2.1.4-beta . The best-fitting substitution model was automatically selected using the ModelFinder program implemented in IQ-TREE 2 and performed with 1000 bootstrap replicates. The ARGs from the microbial genomes were annotated against CARD. All draft genomes of the isolates and MAGs belonging to the same strain were clustered at 99.0% ANI using dRep. Random forest models were utilized to forecast the presence of carbapenem-resistant isolates based on microbiome profiles, resistome profiles, or a combination of both, across 592 samples. To assess the predictive performance of these random forest classification models, we used three indicators (AUC, sensitivity, and specificity), derived from a tenfold cross-validation in this study. In this process, the dataset was randomly divided into 10 equal subsets. During each iteration, nine subsets were used for model training, while the remaining subset was used for prediction. The values of each iteration were recorded. These analyses were performed using the R packages randomForest v4.7–1.1 and caret v6.0–94. Differential analyses were performed by Wilcoxon rank sum tests or Kruskal–Wallis tests. When analyzing more than two groups, multiple comparisons were conducted using the R package agricolae v1.3–7. For beta diversity statistical analyses, the “adonis” function from the R package vegan v2.6–4 and the “pairwise.adonis” function from the R package pairwiseAdonis v0.4.1 were utilized. We employed Procrustes analysis to establish correlations among various profiles using the “procrustes” function of the R package vegan v2.6–4. The correlations were determined by the “protest” function of the R package vegan v2.6–4. The correlations between two variables in this study were calculated using the R function “cor.test”. The intersecting sets were analyzed using the R package UpSetR v1.4.0 . The co-occurrence networks were visualized using Cytoscape v3.10.1 . The heatmaps were visualized using the R package ComplexHeatmap v2.16.0 . The phylogenetic trees were visualized using the R package ggtree v3.8.2 or iTOL v5 . Figures were visualized primarily using the R packages ggplot2 v3.4.4 and patchwork v1.1.3. Additional file 1: Fig. S1: The abundance of the ARG types. (a) Average abundance of the main ARG types in various subgroups. (b) The abundance of the top 10 ARG types in the three groups. (c) The abundance of the 40 ARG types in all subgroups. Left panels: the ARG types (n = 20) with relative higher abundance; right panels: the ARG types (n = 20) with relative lower abundance. Fig. S2: Diversity analysis of the resistome profiles. (a) The Shannon indices of the resistome profiles. (b) Axis 1 values of PCoA of the resistome profiles from various subgroups. (c) Bray–Curtis distances between the resistome profiles of various human fecal samples and those from food and environmental samples. Figure S3: Metagenome-assembled genomes and ACC taxonomic assignment. (a) Phylogenetic tree of the 1,302 MAGs at the species level. A total of 14,787 MAGs (completeness ≥ 70.0% and contamination ≤ 10.0%) were constructed in this study. The novel species are labeled with light blue bars in the outer ring. The color of the inner ring represents the class level of the MAG. (b) Procrustes analysis of the correlations between the resistome and both MGEs and the microbiome in the three groups. The correlations were calculated by the function “protest” in R package vegan. (c) Proportion of taxonomic assignment of the ACCs by the software CAT. CAT exploits homology searches of individual open reading frames to classify ACCs directly. (d) The taxonomic assignment performance of the two methods based on the 54,831 ACCs, which were binned into MAGs out of 162,001 ACCs. (e) Proportion of the taxonomic assignment of the ACCs with the integration of taxonomic information of MAGs. Figure S4: Plasmid-associated ARGs. (a) The proportion of ARG types carried by plasmids in the different subgroups. (b) Co-occurrence network between plasmid replicon types and the ARGs revealed by the ACCs. The brown rounded rectangles represent the plasmid replicons, the colored circle nodes represent the ARG types, and the gray lines represent the associations between the ARGs and the plasmid replicons. (c) An expanded association between the ARGs identified in (b) and the plasmid types identified by integrating the results from the PLSDB database. The brown squares represent the plasmid replicons; the colored circles represent the ARG types; and the gray lines represent the associations between the ARGs and the plasmid replicons. Figure S5: Analysis of the prophages/phages identified from ACCs. The ratio of different taxa of prophage/phage identified in samples from human (a), food (b), and environmental (c) sources. (d) Proportion of ARGs carried by phage/prophage fragments. Host prediction of prophage fragments at the phylum (e) and family (f) levels. Figure S6: Analysis of the ARG-sharing events. (a) ARG-sharing network. A cluster of shapes represents an ARG subtype shared among certain subgroups. The color of each node indicates the subgroup, while the different shapes represent human, food, and environmental samples. The size of the circle corresponds to the frequency of shared events. (b) Frequency of ARG-sharing events in the different risk ranks. Unknown represents the ARGs without a classification. (c) The ratio of ARGs found in food and environmental subgroups shared with human fecal groups. Figure S7: Strain-level phylogenies of species present in all 592 samples. Each node represents the main strain of a sample. Different colors represent various subgroups. Light blue rectangles indicate phylogenetic trees of the same strain (nGD < 0.1). The annotations are the number of samples harboring the same strain. Figure S8: Distribution of carbapenemase genes found in the metagenomic sequencing data. (a) The prevalence rates of the top 20 carbapenemase genes in all 592 samples. The ratio of positive samples for carbapenemase genes in the three groups (b) and the 17 subgroups (c). The numbers labeled on cells represent the ratio of positive samples for each carbapenemase gene. Figure S9: Phylogenetic trees of the top 20 carbapenemase genes. In the phylogenetic tree, each branch represents a sequence type of the carbapenemase gene. Orange circle nodes in the phylogenetic trees represent the reference carbapenemase gene sequences from CARD. Heatmap illustrates the prevalence of each sequence type in the different subgroups, and the color of the cells reflects the number of positive samples. The red border of each cell represents the sharing events of the carbapenemase gene across the different subgroups. Figure S10: Co-occurrence network of carbapenemase genes and their associated hosts identified in complete genomes. The carbapenemase genes includes those from both metagenomic sequencing data and the strain genomic data in our study. The blue nodes represent microbes; the orange nodes represent plasmid replicons, the green nodes represent the ARGs, and the gray lines represent the locations of the ARGs. Additional file 2: Table S1: Sample information and sequencing statistics, related to Fig. a. Table S2: Statistics of the top 10 ARG types, related to Fig. b. Table S3: Average abundance of the Top 10 ARG types in different subgroups (copies per copy of 16S rRNA gene), related to Fig. S1a. Table S4: Statistical tests for the abundance of ARG types among different groups, related to Fig. S1b. Table S5: Statistical tests for the abundance of the 40 ARG subtypes (Kruskal–Wallis test), related to Fig. S1c. Table S6: Pairwise permutational multivariate ANOVA (PERMANOVA) for resistome profiles in the three groups, related to Fig. f. Table S7: Abundance of bacteriophages B40-8 and crAssphage, related to Fig. g. Table S8: Taxonomic assignment of the MAGs, related to Fig. a. Table S9: Pairwise permutational multivariate ANOVA (PERMANOVA) for the microbial compositions, related to Fig. c. Table S10: Variable importance, related to Fig. e. Table S10 Variable importance, related to Fig. e. Table S11: Statistical tests for the abundance of the six taxa, related to Fig. f. Table S12: Correlations between the abundance of the six taxa and the abundance/alpha-diversity of the resistome profiles, related to Fig. g. Table S13: Raw and improved taxonomic assignment of ACCs, related to Fig. e. Table S14: ARG load index of the ACCs at different taxonomic levels. Table S15: Plasmid replicons and their associated ARGs shared between human and food/environment, related to Fig. d. Table S16: Main ARGs carried by the phages/prophages, related to Fig. S5d. Table S17: Statistics of ARG sequence types in different subgroups. Table S18: Sharing events in the ARG-sharing network, related to Fig. S6a. Table S19: Statistics of the ARGs in the ARG-sharing network, related to Fig. S6a. Table 20: Number of ARGs from different risk ranks. Table S21: Statistics of ARGs shared between human and both food and environmental samples. Table S22: Number of sharing events between human and three potential sources, related to Fig. c. Table S23: Prevalence of SGBs detected in the present study, related to Fig. d and Fig. S7. Table S24: Number of carbapenemase genes found in the ACCs, related to Fig. S8a. Table S25: Antimicrobial susceptibility testing of carbapenem-resistant strains. Table S26: Clusters of the isolate genomes and the MAGs, related to Fig. b. Table S27: Host and genetic environment of OXA-347 revealed in the 14,875 complete genomes, related to Fig. c Additional file 3: Introduction to Dengfeng |
Analysis of perioperative autonomic nervous system activity to visualize stress in pediatric patients undergoing alveolar bone graft surgery | 8c8ea92e-10b7-4f77-8f33-0b5d8c44272b | 11821712 | Dentistry[mh] | Surgery under general anesthesia is stressful for pediatric patients . In addition to the stress caused by surgical invasion and environmental changes resulting from hospitalization, patients may develop complications such as nausea, vomiting, and pain . For example, alveolar bone grafting, which is commonly performed in pediatric patients with a history of cleft lip and palate, often involves grafting of the cancellous bone from the anterior iliac crest into the maxillary alveolar cleft under general anesthesia ; it may result in postoperative pain and swelling of the maxilla and the iliac donor site. Moreover, gait disturbance has been reported for an average of 6.6 days postoperatively . In surgeries that have perioperative stress as a concern, the stress of the patient throughout the perioperative period should be visualized and understood. In general, interviews, questionnaires and circulatory dynamics are used to assess perioperative stress, but pediatric patients often find it more difficult to express their condition than adults . Additionally, self-report tools, such as face scale (FS) scores, have limitations related to children’s comprehension and communication skills . Behavioral observation methods can introduce variability between assessors. In addition, blood sampling is considered invasive and stressful for children, and is not a suitable method for stress assessment in pediatric patients. Thus, it is often difficult to visualize and understand the extent of perioperative stress in pediatric patients; therefore, a non-invasive and objective method to monitor children’s stress should be established. Stress is associated with autonomic nervous system (ANS) activity . There are several methods for analyzing ANS activity, and heart rate variability (HRV) analysis using electrocardiogram (ECG) data enables monitoring ANS activity without the requirement for invasive procedures in patients . Although changes in ANS activity due to general anesthesia have been discussed in previous reports , no reports have evaluated ANS activity during the perioperative period in pediatric patients who underwent uniform surgical procedures, and there are many unclear points about perioperative ANS activity in pediatric patients undergoing surgery under general anesthesia. In this study, in addition to circulatory dynamics and psychological status, which have been used as indicators for stress assessment, we used HRV analysis to observe changes in ANS activity in pediatric patients undergoing alveolar bone graft surgery under general anesthesia. We hypothesized that this observation would enable us to visualize the perioperative stress and analyzed changes in ANS activity, circulatory dynamics, and psychological status during the perioperative period. Study design/sample This prospective observational study was approved by the Kagoshima University Hospital Ethics Review Committee (approval no. 190086) and adhered to the tenets of the Declaration of Helsinki. The study was explained to the parents, and age-appropriate explanations were provided to the participants. Written informed consent for study participation was obtained from the parents. This study included patients aged 8–12 years with a diagnosis of cleft lip and palate who visited the Department of Oral and Maxillofacial Surgery at the Kagoshima University Hospital and were scheduled to undergo alveolar bone grafting under general anesthesia with sevoflurane at the same hospital between October 2019 and February 2022. All patients were admitted the day before surgery, and data acquisition was performed during hospitalization. The following patients were excluded: patients with conditions affecting the ANS, those receiving regular medications, those with a history of drug allergy, those with psychosis or psychiatric symptoms that would make study participation difficult, and those who were otherwise deemed ineligible by the dentist. Patients were permitted to eat up to 8.5 h before surgery, drink clear liquids up to 2 h before surgery, and, then discontinue all intake 2 h before surgery. Data collection environment and protocol In this study, ANS and circulatory dynamics were evaluated in the supine position, either on the operating bed or on the hospital bed at rest. ANS and circulatory dynamics were evaluated the day before surgery, during general anesthesia, 2 h postoperatively, 24 h postoperatively, and before discharge (7 days postoperatively) (Fig. ). ANS activity was recorded the day before surgery, 2 h postoperatively, 24 h postoperatively, and the day before discharge (postoperative day 7) in the hospital room for 5 min and from the beginning to the end of anesthesia in the operating room. The measurement duration in the hospital room adhered to HRV guidelines . To evaluate circulatory dynamics, heart rate (HR) and systolic blood pressure (SBP) were recorded. When measuring postoperative ANS activity, we also evaluated pain, postoperative nausea and vomiting (PONV), and fever. Face scale (FS) scores were recorded the day before surgery, before anesthesia, and at the time of postoperative ANS activity measurement. HRV ANS activity was assessed via HRV analysis using three-point trigger ECG data. ECG data were acquired using the MWM01 ECG monitor (GMS, Tokyo, Japan) and analyzed using Memcalc-Mackin2 (GMS). For this study, we employed the frequency-domain analysis method, which is preferred for short-term recordings of ANS activity . The high frequency (HF) (0.15–0.4 Hz; mediated by the parasympathetic nervous system) and low frequency (LF) (0.04–0.15 Hz; mediated by the parasympathetic and sympathetic nervous system) components were analyzed to quantify the activities of the sympathetic and parasympathetic nervous systems from the HRV parameters obtained from the ECG data. The evaluation items were LF/HF (the parameter of sympathetic nervous system activity) and HF (the parameter of parasympathetic nervous system activity) . FS score evaluation Since most patients were aged around 10 years and could find it difficult to express pain and their psychological state to the interviewer, the FS was used with HRV measurements (Fig. ). As an evaluation method using the FS, we asked the patients to select the picture that most closely represented their feelings from six pictures presenting very happy to very unhappy facial expressions. The happiest facial expression was evaluated as 1 point, and the saddest facial expression was evaluated as 6 points. Anesthesia Anesthesia was administered via slow induction with 5% sevoflurane or rapid induction with propofol 1–2 mg/kg. Acetated Ringer’s solution with 1% glucose fluid replacement and remifentanil 0.3 µg/kg/min were administered after intravenous catheterization. Rocuronium 0.8 mg/kg was administered, and oral intubation was performed using a Macintosh laryngoscope. Anesthesia was maintained with 1.5–2% sevoflurane and continuous intraoperative remifentanil at a dose of 0.1–0.5 µg/kg/min. As a local anesthetic, 0.5% lidocaine containing 1:200,000 epinephrine was administered by the surgeon in the alveolar cleft region and iliac donor site before the start of surgery. Postoperative analgesia involved intravenous administration of 15 mg/kg acetaminophen over 15 min when the surgeon began suturing the iliac region, 1 mg/kg flurbiprofen axetil over 5 min after all sutures were completed, and 3 mg/kg (upper limit 75 mg) of a local anesthetic (ropivacaine hydrochloride hydrate) at the iliac donor site at the end of suturing. Operative procedure Cancellous bone was grafted from the anterior iliac crest into the maxillary alveolar cleft in all cases. The surgery was performed by the same team, including an oral surgeon, using the procedure reported by Boyne and Sands . The bone harvesting technique was based on the report by Robertson and Barron . Postoperative management From the time the patient returned to the ward until 24 h postoperatively, 15 mg/kg acetaminophen was administered intravenously every 6 h. Ibuprofen was prescribed if the patients experienced pain. Patients undergoing alveolar bone graft surgery at our hospital were instructed to rest in the bed on the first postoperative day. They were allowed to move using a wheelchair from 2 days postoperatively and walk with a walker from 4 days postoperatively. Unassisted walking was allowed 6 days postoperatively, and patients were discharged 8 days postoperatively. Oral splints were provided for wound healing postoperatively. Statistical analysis The required sample size was calculated using a power analysis (α = 0.05; β = 0.2) and set at 40 to account for dropouts. All statistical analyses were performed using GraphPad Prism version 6 (San Diego, CA, USA). Comparisons of changes over time in preoperative values for LF/HF, HF, HR, SBP, and FS scores were performed using the Friedman test and Steel–Dwass test as a post hoc test. The level of statistical significance was set at P values < 0.05. This prospective observational study was approved by the Kagoshima University Hospital Ethics Review Committee (approval no. 190086) and adhered to the tenets of the Declaration of Helsinki. The study was explained to the parents, and age-appropriate explanations were provided to the participants. Written informed consent for study participation was obtained from the parents. This study included patients aged 8–12 years with a diagnosis of cleft lip and palate who visited the Department of Oral and Maxillofacial Surgery at the Kagoshima University Hospital and were scheduled to undergo alveolar bone grafting under general anesthesia with sevoflurane at the same hospital between October 2019 and February 2022. All patients were admitted the day before surgery, and data acquisition was performed during hospitalization. The following patients were excluded: patients with conditions affecting the ANS, those receiving regular medications, those with a history of drug allergy, those with psychosis or psychiatric symptoms that would make study participation difficult, and those who were otherwise deemed ineligible by the dentist. Patients were permitted to eat up to 8.5 h before surgery, drink clear liquids up to 2 h before surgery, and, then discontinue all intake 2 h before surgery. In this study, ANS and circulatory dynamics were evaluated in the supine position, either on the operating bed or on the hospital bed at rest. ANS and circulatory dynamics were evaluated the day before surgery, during general anesthesia, 2 h postoperatively, 24 h postoperatively, and before discharge (7 days postoperatively) (Fig. ). ANS activity was recorded the day before surgery, 2 h postoperatively, 24 h postoperatively, and the day before discharge (postoperative day 7) in the hospital room for 5 min and from the beginning to the end of anesthesia in the operating room. The measurement duration in the hospital room adhered to HRV guidelines . To evaluate circulatory dynamics, heart rate (HR) and systolic blood pressure (SBP) were recorded. When measuring postoperative ANS activity, we also evaluated pain, postoperative nausea and vomiting (PONV), and fever. Face scale (FS) scores were recorded the day before surgery, before anesthesia, and at the time of postoperative ANS activity measurement. ANS activity was assessed via HRV analysis using three-point trigger ECG data. ECG data were acquired using the MWM01 ECG monitor (GMS, Tokyo, Japan) and analyzed using Memcalc-Mackin2 (GMS). For this study, we employed the frequency-domain analysis method, which is preferred for short-term recordings of ANS activity . The high frequency (HF) (0.15–0.4 Hz; mediated by the parasympathetic nervous system) and low frequency (LF) (0.04–0.15 Hz; mediated by the parasympathetic and sympathetic nervous system) components were analyzed to quantify the activities of the sympathetic and parasympathetic nervous systems from the HRV parameters obtained from the ECG data. The evaluation items were LF/HF (the parameter of sympathetic nervous system activity) and HF (the parameter of parasympathetic nervous system activity) . Since most patients were aged around 10 years and could find it difficult to express pain and their psychological state to the interviewer, the FS was used with HRV measurements (Fig. ). As an evaluation method using the FS, we asked the patients to select the picture that most closely represented their feelings from six pictures presenting very happy to very unhappy facial expressions. The happiest facial expression was evaluated as 1 point, and the saddest facial expression was evaluated as 6 points. Anesthesia was administered via slow induction with 5% sevoflurane or rapid induction with propofol 1–2 mg/kg. Acetated Ringer’s solution with 1% glucose fluid replacement and remifentanil 0.3 µg/kg/min were administered after intravenous catheterization. Rocuronium 0.8 mg/kg was administered, and oral intubation was performed using a Macintosh laryngoscope. Anesthesia was maintained with 1.5–2% sevoflurane and continuous intraoperative remifentanil at a dose of 0.1–0.5 µg/kg/min. As a local anesthetic, 0.5% lidocaine containing 1:200,000 epinephrine was administered by the surgeon in the alveolar cleft region and iliac donor site before the start of surgery. Postoperative analgesia involved intravenous administration of 15 mg/kg acetaminophen over 15 min when the surgeon began suturing the iliac region, 1 mg/kg flurbiprofen axetil over 5 min after all sutures were completed, and 3 mg/kg (upper limit 75 mg) of a local anesthetic (ropivacaine hydrochloride hydrate) at the iliac donor site at the end of suturing. Cancellous bone was grafted from the anterior iliac crest into the maxillary alveolar cleft in all cases. The surgery was performed by the same team, including an oral surgeon, using the procedure reported by Boyne and Sands . The bone harvesting technique was based on the report by Robertson and Barron . From the time the patient returned to the ward until 24 h postoperatively, 15 mg/kg acetaminophen was administered intravenously every 6 h. Ibuprofen was prescribed if the patients experienced pain. Patients undergoing alveolar bone graft surgery at our hospital were instructed to rest in the bed on the first postoperative day. They were allowed to move using a wheelchair from 2 days postoperatively and walk with a walker from 4 days postoperatively. Unassisted walking was allowed 6 days postoperatively, and patients were discharged 8 days postoperatively. Oral splints were provided for wound healing postoperatively. The required sample size was calculated using a power analysis (α = 0.05; β = 0.2) and set at 40 to account for dropouts. All statistical analyses were performed using GraphPad Prism version 6 (San Diego, CA, USA). Comparisons of changes over time in preoperative values for LF/HF, HF, HR, SBP, and FS scores were performed using the Friedman test and Steel–Dwass test as a post hoc test. The level of statistical significance was set at P values < 0.05. Among the 40 eligible patients, nine were excluded because their parents refused to consent to data collection; thus, we could not obtain ECG recordings, interviews, or other data from their hospital rooms. Thus, the data of 31 patients were analyzed. The mean age, weight, and height of the analyzed patients were 9.87 ± 1.07 years, 30.29 ± 5.98 kg, and 134.62 ± 8.17 cm, respectively (Table ). LF/HF, HF, HR, SBP, and FS scores were presented as relative ratios to preoperative values (Fig. ). LF/HF during general anesthesia (3.89 ± 0.85) was significantly higher than the preoperative value ( P < 0.01). Thereafter, LF/HF remained significantly higher at 2 h postoperatively (2.25 ± 0.35, P < 0.01), 24 h postoperatively (3.61 ± 0.92, P < 0.01), and the day before discharge (2.23 ± 0.26, P < 0.01) than the preoperative values (Fig. a). HF significantly decreased from the day before surgery to during general anesthesia (0.10 ± 0.02, P < 0.01). The values 2 h postoperatively (0.68 ± 0.12), 24 h postoperatively (0.68 ± 0.16), and before discharge (0.63 ± 0.10) were significantly higher than the value during general anesthesia ( P < 0.01), but remained significantly lower than the preoperative value (2 h postoperatively: P < 0.05, 24 h postoperatively: P < 0.01, before discharge: P < 0.01) (Fig. b). HR during general anesthesia (1.03 ± 0.02) was not significantly different from the preoperative value. However, it significantly increased 2 h postoperatively (1.13 ± 0.03, P < 0.01) and 24 h postoperatively (1.15 ± 0.03, P < 0.01) compared with the preoperative value. HR on the day before discharge (1.06 ± 0.0.03) was not significantly different from the preoperative value (Fig. c). SBP increased significantly during general anesthesia (1.13 ± 0.03, P < 0.01) and 2 h postoperatively (1.09 ± 0.02, P < 0.01) compared with the preoperative value; however, SBP values 24 h postoperatively (1.04 ± 0.03) and on the day before discharge (1.00 ± 0.02) were not significantly different from the preoperative value (Fig. d). FS scores before anesthesia (2.19 ± 0.2), 2 h postoperatively (2.22 ± 0.24), and 24 h postoperatively (2.07 ± 0.23) were significantly higher than the preoperative value ( P < 0.01). However, the value on the day before discharge (1.16 ± 0.11) was not significantly different from the preoperative value (Fig. e). Table presents the results of pain, PONV, and fever among the patients at different time points. Patients were most likely to complain of pain and fever at 24 h postoperatively, with only a few patients having persistent pain, PONV, and fever until the day before discharge. In this study, ANS activity was observed with uniform surgical procedures and perioperative management in order to visualize perioperative stress in pediatric patients. In addition to vital signs and interviews, which are commonly used to assess stress, ANS activity assessment was also used to evaluate perioperative stress from multiple perspectives. As a result, after alveolar bone graft surgery under general anesthesia, HR, SBP, and FS score were significantly increased compared to preoperative values. Regarding ANS activity, LF/HF significantly increased and HF significantly decreased compared to preoperative values. Significant increases in FS scores and either circulatory dynamics (SBP or HR) were observed compared to preoperative values during general anesthesia, 2 h postoperatively, and 24 h postoperatively. These results suggest the existence of perioperative stress after surgery under general anesthesia. Furthermore, at that time, LF/HF increased and HF decreased, and the ANS activity also supported the existence of perioperative stress. On the other hand, before discharge, LF/HF increased and HF decreased, but other indicators such as SBP, HR, and FS score had recovered to preoperative values. This suggests that ANS activity indicators may be useful for monitoring potential stress, which is difficult to assess using vital signs or interviews. This study revealed that fluctuations in the ANS activity after alveolar bone grafting under general anesthesia persisted until the day before discharge, which was 7 days postoperatively. The ANS activity after general anesthesia has been studied in several reports . However, there are no reports evaluating long-term ANS activity in pediatric patients undergoing the same surgical procedure. Therefore, we evaluated ANS activity over time based on previous studies and limited the target population to pediatric patients undergoing alveolar bone grafting. Previous studies reported that general anesthesia decreases HRV parameters , and surgical invasion or stress increases sympathetic nervous system activity (LF/HF) and decreases parasympathetic nervous system activity (HF), resulting in an imbalance in ANS activity . In this study, the sympathetic nervous system activity increased, whereas the parasympathetic nervous system activity decreased from the preoperative period during anesthesia. This may be because of the suppression of ANS activity by anesthetics and the increase in sympathetic nervous system activity due to various stimuli, such as tracheal intubation and surgical invasion . Previous studies examining changes in HRV indices following surgery under general anesthesia showed that ANS activity gradually returned to baseline over a period of 2 to 8 h postoperatively in adults and within 24 h postoperatively in children younger than 8 years. In our study, wherein patients aged 8–12 years underwent alveolar bone grafting, the ANS activity (LF/HF and HF) on the day before discharge were both significantly different from the preoperative values, indicating that fluctuations in ANS activity due to surgery under general anesthesia persisted until the day before discharge. On the other hand, SBP returned to preoperative values 24 h postoperatively, and HR and FS score returned to preoperative values on the day before discharge. Few patients reported pain before discharge, and none of the patients reported PONV. In a previous study on stress assessment, stress evaluation of on-duty anesthesiologists was conducted using HRV analysis and psychological tests. The results showed that the ANS activity was suppressed under stress; however, the results of the questionnaire assessment remained unchanged. Therefore, it has been reported that HRV analysis may be useful for potential stress monitoring . Similarly, in this study, only fluctuations in the ANS activity (LF/HF and HF) persisted until before discharge. This suggests that analysis of ANS activity may be useful in visualizing perioperative stress in pediatric patients, which is difficult to assess via an interview or vital signs. It has been reported that fluctuations in the ANS activity are associated with stress . Inpatients are exposed to a variety of stresses, including unfamiliar environments, loss of independence, threats from their disease, and complications such as pain and PONV . Furthermore, at our hospital, all patients who undergo alveolar bone grafting are instructed bed rest for 2 days postoperatively and to wear an oral splint for wound healing. Moreover, they are not allowed to walk independently for at least 6 days. According to previous studies on alveolar bone grafting at other institutions, gait disturbance persists for an average of 6.6 days . Patients were able to resume school from an average of 12.6 days postoperatively, and sports activities were resumed from 1 month postoperatively . Thus, alveolar bone grafting under general anesthesia may impose long-term physical limitations and emotional distress on patients, which may cause ANS fluctuations. Previous reports have shown that music listening and video games can effectively reduce stress in children undergoing general anesthesia and surgery . We propose that these methods may be effective in improving perioperative stress in pediatric patients. Additionally, regular ANS monitoring throughout the perioperative period might allow early identification of stress and interventions, resulting in improved patient satisfaction during hospitalization. The strength of this study is that we were able to observe perioperative ANS activity in pediatric patients over 1 week, with uniform surgical procedures and perioperative management. Another advantage was that we were able to assess perioperative stress from multiple perspectives by evaluating ANS activity as well as circulatory dynamics and psychological status. However, one limitation of this study was that ECG recordings were unavailable after the day of discharge, making it impossible to assess whether the ANS activity subsequently recovered to the preoperative levels. Furthermore, while HRV is used to assess ANS activity, it is influenced by factors, such as age, measurement time, environment, pain, and anxiety . The purpose of this study was to observe the changes in ANS activity resulting from these various factors, therefore the direct association between individual factors and HRV was not examined. In conclusion, the present study showed that fluctuations in vital signs, psychological status, and ANS activity were observed after alveolar bone graft surgery under general anesthesia in pediatric patients, and the results of ANS activity analysis also suggested the presence of perioperative stress. However, on the day before discharge, although vital signs and psychological state had recovered to preoperative states, fluctuations in ANS activity persisted. Only ANS activity did not recover by the day of discharge, suggesting that ANS activity can be used to assess stress not captured by other measures and may lead to improved quality of perioperative management. If a specific method for monitoring ANS activity is established, analysis of ANS activity may be useful in visualizing perioperative stress in pediatric patients, which is difficult to assess via an interview or vital signs. The relationship between ANS activity and perioperative stress requires further investigation. |
How Should we Teach Medicinal Chemistry in Higher Education to Prepare Students for a Future Career as Medicinal Chemists and Drug Designers? – A Teacher's Perspective | ad96c27f-0d3b-4030-ae30-cb4ad1163aab | 11733470 | Pharmacology[mh] | In 1998 medicinal chemistry has been defined by Medicinal Chemistry Section (now part of Division VII: Chemistry and Human Health) of the International Union of Pure and Applied Chemistry (IUPAC) as “Medicinal chemistry is a chemistry‐based discipline, also involving aspects of biological, medical and pharmaceutical sciences. It is concerned with the invention, discovery, design, identification and preparation of biologically active compounds, the study of their metabolism, the interpretation of their mode of action at the molecular level and the construction of structure‐activity relationships.” Although the need for new and in particular more specific definitions has been formulated, this early general definition holds true and underlines two important but somewhat contradictory aspects of medicinal chemistry: the clear allocation as a historically chemistry‐based discipline and the highly interdisciplinary nature of medicinal chemistry comprising multiple methods from a broad variety of different associated natural science disciplines. This makes teaching medicinal chemistry challenging as teachers coming from the organic chemistry field must transmit a broad range of knowledge from outside their field. In the last decades, the interdisciplinary nature of medicinal chemistry has expanded significantly comprising a growing influence of structural biology, genetic modulation techniques, virtual screening, and modeling techniques, computer‐aided drug design, and data sciences including artificial intelligence and machine learning tools, bioinformatic and chemoinformatic analysis. Furthermore, technological advances in all disciplines involved in the drug discovery and development process have boosted new developments in the field of medicinal chemistry and initiated several paradigm shifts currently changing the field drastically. While still part of the methodology portfolio in drug discovery, pure pharmacology‐guided phenotypic approaches significantly lost ground in favor for more rational “Target‐First, Pharmacology‐Second” drug design approaches enabled by a growing knowledge feed from genome‐wide association studies (GWAS) and deep sequencing of DNA. As a consequence, the target space in drug discovery has significantly expanded and beyond classical drug targets such as enzymes and receptors, now regulatory or structural proteins are on the drug discovery menu list often targeted by modulation or inhibition of their protein‐protein interactions. New hit‐finding technologies (e. g. new High‐Throughput Screening (HTS) technologies, virtual screening (VS), focused screens, fragment‐based drug discovery, DNA‐encoded library screens, MS‐based affinity selection) have emerged boosting the success of medicinal chemistry projects. Furthermore, the steadily evolving discipline of chemical biology representing the counter side to medicinal chemistry in the scientific continuum of drug discovery has contributed a broad variety of tool compounds and techniques such as affinity or activity‐based protein profiling (ABPP) to identify and understand drug targets significantly complementing the process of drug discovery. These drastic changes and the broadening of the scope of drug discovery campaigns need to be reflected in the teaching of medicinal chemistry in higher education, which raises challenges to select content for teaching without overwhelming students and provoke them to avoid too difficult courses. Furthermore, two major design rules in medicinal chemistry have been questioned in the last decade leading to paradigm changes following a dramatic expansion of the scope of medicinal chemistry. Historically and based on hepatotoxic properties of covalent binding metabolites of drugs such as acetaminophen (Paracetamol), investigated in the 1970s, drugs binding covalently to their targets were associated with concerns about potential for off‐target activity and for years controversially discussed for their role in the pathogenesis of idiosyncratic drug‐related toxicity. Despite multiple successful examples (e. g. esomeprazol /Nexium; AstraZeneca, clopridogrel /Plavix; Sanofi‐Aventis/Bristol‐Myers Squibb) of selective and safe drugs beyond acetylsalicylic acid, antimicrobial compounds such as β‐lactam antibiotics or conventional chemotherapeutics, covalent binding had been considered a risk factor and, as a consequence, covalent mode‐of‐actions were avoided in drug discovery programs. In the 2000s, the distinct strength of the combination of covalent and non‐covalent modes of actions was recognized and the concept of targeted covalent inhibitors (TCIs) evolved rapidly with rational drug design approaches with nowadays multiple examples of successfully market drugs, especially in the area of kinase inhibitors. The second paradigm change was an even more pronounced drumbeat as the long standing Lipinsky's Rule of Five (Ro5) was overthrown, which had influenced the drug designers for over 25 years. In 1997, Lipinsky and coworkers analyzed 2245 drug candidates, which had progressed beyond early clinical trials and therefore were assumed to have achieved sufficient levels of systemic exposure in early clinical trials. They assessed four selected physicochemical properties of these compounds, defined cut‐offs so that 90 % of the compounds fall inside them and introduced the Rule of Five (Ro5) predicting that poor gastrointestinal absorption and membrane permeation for small molecules (SMOLs) is more likely with more than 5 hydrogen bond donors (expressed as the sum of OHs and NHs), more than 10 hydrogen bond acceptors (expressed as the sum of Ns and Os), a molecular weight over 500 g/mol and a LogP over 5. Although, Lipinsky and coworkers cautioned that the Ro5 should merely be used as an alert tool for newly synthesized SMOLs rather than a hard rule and already made exclusion for certain compound classes as well as noticed that natural product like drug classes did not comply with the Ro5, it quickly gained attention in the industrial and also academic medicinal chemistry community and became an overinterpreted rule shaping the thinking of a generation of medicinal chemists. In some pharma companies the Ro5 became dogma introducing an era of property‐based drug design. Early concerns on the general validity of the Ro5 led to expansions and variants of Ro5 such as the Ghose's or Egan's filter for the prediction of “ druglikeness ”, Veber's rule, particularly questioning the hard molecular weight cut‐off and accounting for the influence of polar surface area (PSA), Muegge's drug like criteria and the truncation to the Rule of 3 (Ro3) for fragment‐based drug discovery or fragment‐based design by Jhoti and coworkers. However, all these approaches were not considering metabolism of drugs, compensating effects between different drug properties or the impact of high potency and the corresponding rules were applied too restrictively. Even in recent years, medicinal chemists desperately try to find standardized criteria for “druglikeness” but all these rigorous models narrow the available chemical space and should only be used in Lipinsky's original intention as an alert tool for drug design. The physicochemical understanding and acceptance of drug space “ beyond rule of 5 ” (bRo5) in the medicinal chemistry community in recent years have culminated in the development of protein‐targeting chimeras (PROTACs). These bifunctional SMOLs occupy bRo5 drug space and consist of a high affinity binder for a protein target and E3‐ligase active war head, which induce proximity between the protein of interest (POI), the PROTAC and an E3‐ligase by formation of a ternary complex inducing ubiquitination. The ubiquitylated POI then gets degraded intracellularly by the proteasome. The PROTAC concept has been demonstrated to be particularly successful in addressing difficult targets and although no PROTACs have been approved yet, several clinical studies are ongoing. Carefully designed PROTACs can be orally active and even pass the blood‐brain barrier convincing experts in the field that PROTACs can be a game changers for several diseases. In general targets seem to benefit from bRO5 drugs if they have complex hot spot structures within their binding site leading to either a positive correlation between binding affinity and molecular weight (e. g. protein‐protein interactions (PPIs)) or the need for increased selectivity within a target family (e. g. protein kinases). Considering that not too long ago it seemed that SMOL drugs were on the decline as advances in biotechnology enabled pharma companies to cost‐effectively generate a range biologics such as large peptides, recombinant proteins, monoclonal antibodies, fusion proteins and vaccines, now SMOL drug discovery is having its moment and the rise of an excitingly innovative time is tangible. Although Ro5 compliant SMOLs remain a corner stone of modern drug discovery, new chemical modalities (such as PROTACs, other bRo5 SMOLs and peptides (e. g. cyclopeptides and macrocycles ), oligonucleotide therapeutics (including small interfering RNA (siRNAs), antisense oligonucleotides (ASOs), microRNAs and aptamers), biologics (e. g. antibodies), CRISPR‐cas9‐based therapeutics, SMOL‐radionuclide conjugates and mixed molecular conjugates for extracellular‐targeted drug delivery (e. g., antibody‐drug conjugates, antibody‐degrader conjugates and peptide‐drug conjugates)) have emerged with successful Proof‐of‐Principles (PoPs), initiated clinical studies and approved drugs now significantly expanding the scope of medicinal chemistry. A new era of drug design has been heralded and it seems that terms like “undruggable” or “non‐ligandable” are terms of the past. Our excitement about these new opportunities is something we should transport when teaching medicinal chemistry in higher education to facilitate intrinsic learning motivation and foster creative thinking in students. This makes it also fundamental to create an inclusive, safe and autonomy‐supportive learning atmosphere allowing students to be creative, to think out of the box and expand the boundaries of the state‐of‐the‐art. However, a deep understanding of physicochemical guidelines and their limitations seems to be key to support the students in considering pharmacodynamic and pharmacokinetic safety aspects of drugs. Similar to the Rule of Five (Ro5), another pivotal guideline in drug discovery, the pan‐assay interference compounds (PAINS) rule, has recently faced scrutiny regarding the rigidity of its application. The identification of hit compounds through high‐throughput screenings (HTS) is a critical and powerful technique for expediting the drug discovery process by evaluating large compound libraries, providing initial leads for drug development efforts. However, the emergence of false‐positive hit compounds, especially frequent hitters (FHs), which are repeatedly identified as hits across multiple unrelated assays, presents significant challenges. The unproductive optimization of these non‐progressible compounds can consume substantial research time and resources. In 2010, Baell and colleagues published a study identifying specific structural subclasses of compounds with diverse chemical reactivity and apparent biological activity as potential sources of false‐positive results in biochemical assays. They proposed the PAINS rule as an electronic filter to alert medicinal chemists to compounds likely to be non‐optimizable and non‐progressible. Incorporating PAINS substructures as electronic filters into HTS enabled the rapid identification of PAINS compounds in large screening libraries, earning widespread recognition in the drug discovery community for seemingly simplifying the critical process of compound triage. As a result, the PAINS rule has become a key component in biological screening campaigns. In fact, nine editors‐in‐chief of prominent journals highlighted the importance of PAINS alerts in ensuring the quality of reported chemotype‐biology associations, and the Journal of Medicinal Chemistry recommended that authors include PAINS alerts for active compounds in submitted screening results. However, it has become increasingly evident in recent years that this seemingly straightforward solution to the complex issue of false‐positive hits caused by frequent hitters is far from flawless. Even more concerning is that the overly strict or indiscriminate use of these filters can lead to the inappropriate exclusion of valuable compounds, while also promoting unproductive dark chemical matter compounds as candidates for development, thus undermining the intended simplification of compound triage in drug discovery efforts. Indeed, 87 FDA‐approved small‐molecule drugs contain PAINS motifs. Physicochemical properties such as hydrophilicity, which may mitigate assay interference, are not accounted for by simple substructural filters. Furthermore, factors like the influence of contaminants or decomposition products‐elements much harder to predict‐are often overlooked. It has become clear that more refined filters are required, utilizing larger datasets and incorporating machine learning approaches. Nonetheless, the role of these filters should be limited to providing alerts, while individual analysis of compound behavior, combined with purity checks and orthogonal resynthesis, should be integrated into a thorough hit validation process. This process should be rooted in an efficient and comprehensive screening strategy, including orthogonal experiments. In this context, it has been recommended to focus on assay limitations (GAINs=give attention to limitations in assays) rather than restricting drug discovery by excluding hits with complex biological mechanisms, potentially overlooking compounds that could serve as innovative treatments for diseases. The discussed changes in the drug discovery landscape are accompanied by an increasing synthetic complexity of lead compounds with a growing share of 3‐dimensionality and chirality, which has also triggered the implementation of new synthetic methods such as direct C−H‐functionalization of advanced lead compounds, photo‐redox‐catalysis, electro chemical transformations, flow chemistry and high‐throughput experimentation for reaction condition screening and optimization setting the stage for machine learning approaches for synthetic route predictions and late‐stage functionalization approaches. Furthermore, there is a growing importance of sustainable medicinal chemistry with greener active pharmaceutical ingredients (APIs). Teaching in medicinal chemistry always needs to be anchored in organic synthesis aimed to provide methodology to synthesize the targeted compounds and best allow the students to develop retrosynthetic routes on their own. Finally, in terms of automation and digitalization, the paradigm shift in pharmaceutical industry seems to be on‐going as digital tools such as machine learning (ML) supported prediction of physicochemical compounds properties, in vitro and in vivo absorption‐distribution‐metabolism‐excretion‐toxicity (ADMET) properties and artificial intelligence (AI)‐based retrosynthesis and late‐stage functionalization tools find first application in medicinal chemistry projects and being further developed, while tools like AlphaFold, clearly start to make differences in the accuracy of protein folding predictions. Medicinal chemistry teaching in higher education needs to reflect these changes in automation and digitalization by training students on automated devices and in the use of AI based tools without neglecting the teaching of the underlaying manual analog principles and concepts. The described changes in medicinal chemistry have led to an evolving skill set of distinct soft and hard skills for scientists in medicinal chemistry, which need to be considered when teaching medicinal chemistry to the next generation. However, this task goes far beyond the continuous adaptation of medicinal chemistry education to changes, and the obvious challenge to select the content for a time limited lecture from this broadened variety of topics. As current academic medicinal chemistry education seems not to meet the needs of pharmaceutical industry, one of the most important stakeholders, it is time to rethink the way we teach medicinal chemistry! To provide a comprehensive picture and analyze which skill sets and learning outcomes need to be aimed at in medicinal chemistry teaching, it is important to identify the different stakeholder groups in medicinal chemistry teaching in higher education. The results of this analysis are summarized in Figure . The following analysis is structured by the different stakeholder groups identified in literature or by my practice as teacher as well as academic and industrial researcher in the field. Skills derived from the needs of the different stakeholders are highlighted in bold. The term stakeholders refers to groups with relevant interests which need to be considered in teaching of medicinal chemistry in higher education. It is however important to note that this analysis is not meant to be complete. Its primary aim is to be thought‐provoking, stimulating further inquiry and discussion in the scientific community and among medicinal chemistry teachers. Students Pursuing Medicinal Chemistry Education Within a constructivist student‐centered learning approach for medicinal chemistry teaching, the primary stakeholder group to mention should be the students pursuing a medicinal chemistry education. In general, this group is highly diverse in terms of educational background coming from first, second and third cycle chemistry, medicinal chemistry, pharmacy, and chemical biology study programs. Although the expectations of medicinal chemistry students can vary based on their academic experience, career goal, and depth of knowledge they seek, all might have a direct interest in the quality of education, the curriculum content, and the relevance of the teaching material to their future careers. First cycle students seek to be introduced to fundamental concepts of medicinal chemistry and the drug discovery process and aim to learn basic laboratory skills including commonly used techniques to lay the groundwork for further second cycle studies. Second cycle students might already have a more distinct perception of their areas of interest looking for a broader variety of advanced topics such as specific computational methods or more specialized therapeutic areas. Second cycle students often aim to gain hands‐on experience in medicinal chemistry research to prepare for internships in both academic and industrial medicinal chemistry. Furthermore, expectations of this stakeholder group might include opportunities to network with medicinal chemistry professionals from industry. This wish for network opportunities seems to increase among third cycle students. Many of them are becoming experts in specific areas of medicinal chemistry on their own. However, their expectations might be related to future career preparation, including the expansion of knowledge in areas beyond their thesis topic and transferable skills for both academic and industry postdoctoral or industrial entry‐level positions. Although the overall expectations on “How medicinal chemistry should be taught?” from the perspective of the stakeholder groups of students might be less defined, it has been demonstrated that students in higher education clearly prefer student‐centered interactive teaching activities over classical information‐transmitting teacher‐centered approaches. Teachers of Medicinal Chemistry in Higher Education and Academic Medchem Community The faculty and instructors responsible for teaching medicinal chemistry are crucial stakeholders as they are directly involved in designing the curriculum and associated teaching activities in medicinal chemistry in higher education. Teachers in medicinal chemistry education do not only have to address the pedagogical challenges arising with the multidisciplinary field but also hold the important role to mediate and address the needs of all stakeholder groups within the curriculum of medicinal chemistry courses. They should have an interest in providing relevant topics, practical lab course experiences and teaching relevant skills to their students, preparing them for future tasks as medicinal chemists and drug designers. Therefore, they need access to state‐of‐the‐art laboratories and teaching facilities, cutting‐edge technologies, and proper equipment to enable effective teaching. In addition, this group should have an emphasis on engaging effective teaching methods to capture students’ interest, facilitate understanding of complex medicinal chemistry concepts and create an inclusive and supportive learning environment for the subject. One challenge for the teachers in medicinal chemistry in higher education might also be to balance the interest of some of the other stakeholder groups by alignment of their individual teaching goals with the evolving needs of pharmaceutical industry, broader society‐derived needs, and goals of the educational institution they are representing to ensure a cohesive and effective educational experience for the students. As they often conduct medicinal chemistry projects in their own academic research groups, this stakeholder group of academic teachers of medicinal chemistry has a huge overlap with the academic medicinal chemistry community. Both share a strong interest in the proper education of future PhDs and coworkers with creative and critical thinking driving future innovation in their labs. In their role within the higher education system, medicinal chemistry teachers strongly depend on their educational institutions. Educational Institutions Therefore, educational institutions such as universities and technical colleges offering medicinal chemistry programs are stakeholders closely connected to the stakeholder group of medicinal chemistry teachers. They have an interest in maintaining the quality and reputation of their educational programs, attracting students, and meeting accreditation standards. Within this role they are responsible for providing the required facilities, equipment, and proper framework for teaching. Future Employers and the Pharmaceutical Industry A further very important stakeholder group are the future employers of graduates of medicinal chemistry study programs, a group being strongly dominated by the pharmaceutical industry. The pharma discovery ecosystem has faced major changes from mainly big pharma companies in the 1990s towards a pharma landscape today comprised of big and medium sized pharma companies, contract research organizations (CROs) providing top‐notch lab infrastructure and scientific expertise, and a broad range of biotech companies and start‐ups. Therefore, this stakeholder group might be even more diverse than expected on the first glance and expectations of this stakeholder group towards medicinal chemistry education might differ based on actual work tasks and structural frameworks within these organizations. While the actual responsibility of specific medicinal chemist within bigger pharma companies with more employees might be more specialized or can be centered around pure synthetic chemistry, in mid‐size pharma and biotech companies who have overall smaller medicinal chemistry departments or less coworkers, the expected skill set for a medicinal chemist might be even broader, simply as it cannot be afforded to exclusively dedicate internal chemist to synthetic chemistry task only. The basic skill sets for medicinal chemists in industry have been controversy discussed multiple times, especially in the context of the reorganization of European study programs within the Bologna process. The participants of the European Medicinal Chemistry Leaders in Industry (EMCL) meeting, despite all the differences, represent a substantial part of this stakeholder group and identified as key competences for medicinal chemists a solid education in organic synthesis with the ability to efficiently access complex molecules and a clear understanding how drug properties are linked to a molecular structure. In addition, an understanding of the fundamental biology and new treatments concepts is of growing importance to be able to navigate securely through the broadened scope of medicinal chemistry and make valuable decisions within drug discovery's Design‐Make‐Test‐Analyze (DMTA) Cycles. As pointed out by the European Federation of Medicinal Chemistry and Chemical Biology (EFMC) the competence in and understanding of basic chemical biology techniques seems to be key for modern drug discovery to enable medicinal chemists to design and develop bioactive molecular tools to modulate and study biological processes, optimize chemical probes and imaging tools (e. g. fluorescent molecules) to explore biological systems and identify molecular targets or utilize omics approaches to identify a compound's effects on native proteins expression and function or the biomolecular target of a small molecule. Furthermore, digital competence in the use of tools for prediction of binding interaction or physicochemical chemical properties or even hands‐on digital capabilities such as coding skills will gain more importance in the future. Beyond that, a distinct set of soft skills is of growing importance for industrial medicinal chemists working in highly diverse and interdisciplinary teams. These include good communication skills, being able to reach out to diverse audiences, proper conflict handling, the ability to work in an intercultural, diversity‐oriented, collaboration‐driven environment, and management skills are seen as important as the scientific know‐how listed above. Teachers in medicinal chemistry need to review if and how such skills are reflected in the curricula and should aim for appropriate teaching activities fostering students’ development towards these highly ambitious learning outcomes. The Society The stakeholder group of the future employers, particularly the pharmaceutical industry, is of special importance and a medicinal chemistry education ignoring the needs and interest of these stakeholders would probably fail to educate the next generation of drug designers. However, it also needs to be considered that the needs of society and the aims of economically driven companies do not necessarily always match. One specific example underlying this mismatch situation is the current antibiotic crisis. After the “ golden era of antibiotic drug discovery ” from the 1930s to mid‐1960 s in which most of the antibiotics used today have been discovered and developed to enter the drug market, a huge innovation gap of over 40 years followed in which no new antibiotics have been developed. At the same time, the occurrence and spread of resistances against marketed antibiotics has risen tremendously leading to a situation today in which we see ourselves confronted with nosocomial infections with multi‐drug‐resistant (MDR) bacterial pathogens where there are no effective treatments available anymore. This situation has been amplified by the fact that most big pharma companies have stopped their antimicrobial research programs and stepped out of the non‐lucrative business of the development of antibiotics for economic reasons. From a purely economic standpoint of view, an understandable decision. While the development cost and time of an antibiotic is comparable to those of a cardiovascular or anticancer drug, the economic risks for the company are significantly higher. Not only that in pronounced contrast to other medical indications, the efficacy of antibacterial drugs deteriorates over time as bacterial resistance is an intrinsic factor of bacterial evolution in consequence to the antibiotics impact on the bacterial resistome. Furthermore, additional limitations such as reduced prescription for drugs classified as antibiotics of last resort or comparably short treatment times of infections contribute to a difficult economic exploitation of antibiotics. Despite all understanding of the problematic situation, despite the clear need for a political solution addressing the requirements of functional business models for the pharmaceutical industry, and despite recent single positive examples of antibiotic drugs entering the marked, the social responsibility of the pharmaceutical industry as key player in the development of new antibiotics to counteract the antibiotic crisis is, from ethical stand point of view, currently not fulfilled. Similarly, the need for drug discovery to counteract other global health issues such as neglected and tropical diseases that pose a bigger problem in low‐ and middle‐income countries (LMICs) such as malaria, tuberculosis, dengue fever, leishmaniasis or AIDS or the global problem of multi‐drug resistant fungal infections is currently not met. In this context, alternative funding and collaboration models for drug discovery through public‐private partnerships (PPPs, such as the Innovative Medicines Initiative's (IMI): New Drugs for Bad Bugs (ND4BB) ), publicly‐supported non‐profit organizations (such as Drugs for Neglected Diseases initiative (DNDi) or The Global Antibiotic Research & Development Partnership (GARDP) ), multi‐national, multi‐disciplinary research consortia (such as The Tuberculosis Drug Accelerator (TBDA) or The Malaria Drug Accelerator (MalDA) ) and dedicated funding bodies like CARB‐X (Combating Antibiotic‐Resistant Bacteria) have become important in catalyzing breaking areas of science within high‐risk or global public health strategies that may otherwise have not progressed. In this context, the society needs to be considered as a stakeholder with a strong interest in medicinal chemistry to maintain public health through development of effective, safe, and more sustainable drugs. In medicinal chemistry teaching in higher education, we need to identify potential conflicts of interest and balance them out by incorporating the needs of society into an industry‐oriented teaching of medicinal chemistry. Finally, society has a strong interest in the education of creative and critical thinkers, challenging the status quo and being able to provide innovative solutions for current and future problems of humanity. Miscellaneous Stakeholders Furthermore, partially overlapping additional stakeholder groups are governments and policymakers providing support for educational programs with an interest in workforce development, regulatory authorities providing framework for drug development, or non‐governmental organizations such as patience associations or professional organizations related to medicinal chemistry setting standards and advocating for their fields. For example, The European Federation for Medicinal Chemistry and Chemic al Biology (EFMC) has established a work group focused on Best Practice in Medicinal Chemistry ( https://www.efmc.info/best‐practices ). Each of these stakeholders plays a role in shaping and influencing the landscape of medicinal chemistry education and defining the educational mandate with a distinct skill set of scientific hard skills and required soft skills to succeed as future drug designers. Within a constructivist student‐centered learning approach for medicinal chemistry teaching, the primary stakeholder group to mention should be the students pursuing a medicinal chemistry education. In general, this group is highly diverse in terms of educational background coming from first, second and third cycle chemistry, medicinal chemistry, pharmacy, and chemical biology study programs. Although the expectations of medicinal chemistry students can vary based on their academic experience, career goal, and depth of knowledge they seek, all might have a direct interest in the quality of education, the curriculum content, and the relevance of the teaching material to their future careers. First cycle students seek to be introduced to fundamental concepts of medicinal chemistry and the drug discovery process and aim to learn basic laboratory skills including commonly used techniques to lay the groundwork for further second cycle studies. Second cycle students might already have a more distinct perception of their areas of interest looking for a broader variety of advanced topics such as specific computational methods or more specialized therapeutic areas. Second cycle students often aim to gain hands‐on experience in medicinal chemistry research to prepare for internships in both academic and industrial medicinal chemistry. Furthermore, expectations of this stakeholder group might include opportunities to network with medicinal chemistry professionals from industry. This wish for network opportunities seems to increase among third cycle students. Many of them are becoming experts in specific areas of medicinal chemistry on their own. However, their expectations might be related to future career preparation, including the expansion of knowledge in areas beyond their thesis topic and transferable skills for both academic and industry postdoctoral or industrial entry‐level positions. Although the overall expectations on “How medicinal chemistry should be taught?” from the perspective of the stakeholder groups of students might be less defined, it has been demonstrated that students in higher education clearly prefer student‐centered interactive teaching activities over classical information‐transmitting teacher‐centered approaches. The faculty and instructors responsible for teaching medicinal chemistry are crucial stakeholders as they are directly involved in designing the curriculum and associated teaching activities in medicinal chemistry in higher education. Teachers in medicinal chemistry education do not only have to address the pedagogical challenges arising with the multidisciplinary field but also hold the important role to mediate and address the needs of all stakeholder groups within the curriculum of medicinal chemistry courses. They should have an interest in providing relevant topics, practical lab course experiences and teaching relevant skills to their students, preparing them for future tasks as medicinal chemists and drug designers. Therefore, they need access to state‐of‐the‐art laboratories and teaching facilities, cutting‐edge technologies, and proper equipment to enable effective teaching. In addition, this group should have an emphasis on engaging effective teaching methods to capture students’ interest, facilitate understanding of complex medicinal chemistry concepts and create an inclusive and supportive learning environment for the subject. One challenge for the teachers in medicinal chemistry in higher education might also be to balance the interest of some of the other stakeholder groups by alignment of their individual teaching goals with the evolving needs of pharmaceutical industry, broader society‐derived needs, and goals of the educational institution they are representing to ensure a cohesive and effective educational experience for the students. As they often conduct medicinal chemistry projects in their own academic research groups, this stakeholder group of academic teachers of medicinal chemistry has a huge overlap with the academic medicinal chemistry community. Both share a strong interest in the proper education of future PhDs and coworkers with creative and critical thinking driving future innovation in their labs. In their role within the higher education system, medicinal chemistry teachers strongly depend on their educational institutions. Therefore, educational institutions such as universities and technical colleges offering medicinal chemistry programs are stakeholders closely connected to the stakeholder group of medicinal chemistry teachers. They have an interest in maintaining the quality and reputation of their educational programs, attracting students, and meeting accreditation standards. Within this role they are responsible for providing the required facilities, equipment, and proper framework for teaching. A further very important stakeholder group are the future employers of graduates of medicinal chemistry study programs, a group being strongly dominated by the pharmaceutical industry. The pharma discovery ecosystem has faced major changes from mainly big pharma companies in the 1990s towards a pharma landscape today comprised of big and medium sized pharma companies, contract research organizations (CROs) providing top‐notch lab infrastructure and scientific expertise, and a broad range of biotech companies and start‐ups. Therefore, this stakeholder group might be even more diverse than expected on the first glance and expectations of this stakeholder group towards medicinal chemistry education might differ based on actual work tasks and structural frameworks within these organizations. While the actual responsibility of specific medicinal chemist within bigger pharma companies with more employees might be more specialized or can be centered around pure synthetic chemistry, in mid‐size pharma and biotech companies who have overall smaller medicinal chemistry departments or less coworkers, the expected skill set for a medicinal chemist might be even broader, simply as it cannot be afforded to exclusively dedicate internal chemist to synthetic chemistry task only. The basic skill sets for medicinal chemists in industry have been controversy discussed multiple times, especially in the context of the reorganization of European study programs within the Bologna process. The participants of the European Medicinal Chemistry Leaders in Industry (EMCL) meeting, despite all the differences, represent a substantial part of this stakeholder group and identified as key competences for medicinal chemists a solid education in organic synthesis with the ability to efficiently access complex molecules and a clear understanding how drug properties are linked to a molecular structure. In addition, an understanding of the fundamental biology and new treatments concepts is of growing importance to be able to navigate securely through the broadened scope of medicinal chemistry and make valuable decisions within drug discovery's Design‐Make‐Test‐Analyze (DMTA) Cycles. As pointed out by the European Federation of Medicinal Chemistry and Chemical Biology (EFMC) the competence in and understanding of basic chemical biology techniques seems to be key for modern drug discovery to enable medicinal chemists to design and develop bioactive molecular tools to modulate and study biological processes, optimize chemical probes and imaging tools (e. g. fluorescent molecules) to explore biological systems and identify molecular targets or utilize omics approaches to identify a compound's effects on native proteins expression and function or the biomolecular target of a small molecule. Furthermore, digital competence in the use of tools for prediction of binding interaction or physicochemical chemical properties or even hands‐on digital capabilities such as coding skills will gain more importance in the future. Beyond that, a distinct set of soft skills is of growing importance for industrial medicinal chemists working in highly diverse and interdisciplinary teams. These include good communication skills, being able to reach out to diverse audiences, proper conflict handling, the ability to work in an intercultural, diversity‐oriented, collaboration‐driven environment, and management skills are seen as important as the scientific know‐how listed above. Teachers in medicinal chemistry need to review if and how such skills are reflected in the curricula and should aim for appropriate teaching activities fostering students’ development towards these highly ambitious learning outcomes. The stakeholder group of the future employers, particularly the pharmaceutical industry, is of special importance and a medicinal chemistry education ignoring the needs and interest of these stakeholders would probably fail to educate the next generation of drug designers. However, it also needs to be considered that the needs of society and the aims of economically driven companies do not necessarily always match. One specific example underlying this mismatch situation is the current antibiotic crisis. After the “ golden era of antibiotic drug discovery ” from the 1930s to mid‐1960 s in which most of the antibiotics used today have been discovered and developed to enter the drug market, a huge innovation gap of over 40 years followed in which no new antibiotics have been developed. At the same time, the occurrence and spread of resistances against marketed antibiotics has risen tremendously leading to a situation today in which we see ourselves confronted with nosocomial infections with multi‐drug‐resistant (MDR) bacterial pathogens where there are no effective treatments available anymore. This situation has been amplified by the fact that most big pharma companies have stopped their antimicrobial research programs and stepped out of the non‐lucrative business of the development of antibiotics for economic reasons. From a purely economic standpoint of view, an understandable decision. While the development cost and time of an antibiotic is comparable to those of a cardiovascular or anticancer drug, the economic risks for the company are significantly higher. Not only that in pronounced contrast to other medical indications, the efficacy of antibacterial drugs deteriorates over time as bacterial resistance is an intrinsic factor of bacterial evolution in consequence to the antibiotics impact on the bacterial resistome. Furthermore, additional limitations such as reduced prescription for drugs classified as antibiotics of last resort or comparably short treatment times of infections contribute to a difficult economic exploitation of antibiotics. Despite all understanding of the problematic situation, despite the clear need for a political solution addressing the requirements of functional business models for the pharmaceutical industry, and despite recent single positive examples of antibiotic drugs entering the marked, the social responsibility of the pharmaceutical industry as key player in the development of new antibiotics to counteract the antibiotic crisis is, from ethical stand point of view, currently not fulfilled. Similarly, the need for drug discovery to counteract other global health issues such as neglected and tropical diseases that pose a bigger problem in low‐ and middle‐income countries (LMICs) such as malaria, tuberculosis, dengue fever, leishmaniasis or AIDS or the global problem of multi‐drug resistant fungal infections is currently not met. In this context, alternative funding and collaboration models for drug discovery through public‐private partnerships (PPPs, such as the Innovative Medicines Initiative's (IMI): New Drugs for Bad Bugs (ND4BB) ), publicly‐supported non‐profit organizations (such as Drugs for Neglected Diseases initiative (DNDi) or The Global Antibiotic Research & Development Partnership (GARDP) ), multi‐national, multi‐disciplinary research consortia (such as The Tuberculosis Drug Accelerator (TBDA) or The Malaria Drug Accelerator (MalDA) ) and dedicated funding bodies like CARB‐X (Combating Antibiotic‐Resistant Bacteria) have become important in catalyzing breaking areas of science within high‐risk or global public health strategies that may otherwise have not progressed. In this context, the society needs to be considered as a stakeholder with a strong interest in medicinal chemistry to maintain public health through development of effective, safe, and more sustainable drugs. In medicinal chemistry teaching in higher education, we need to identify potential conflicts of interest and balance them out by incorporating the needs of society into an industry‐oriented teaching of medicinal chemistry. Finally, society has a strong interest in the education of creative and critical thinkers, challenging the status quo and being able to provide innovative solutions for current and future problems of humanity. Furthermore, partially overlapping additional stakeholder groups are governments and policymakers providing support for educational programs with an interest in workforce development, regulatory authorities providing framework for drug development, or non‐governmental organizations such as patience associations or professional organizations related to medicinal chemistry setting standards and advocating for their fields. For example, The European Federation for Medicinal Chemistry and Chemic al Biology (EFMC) has established a work group focused on Best Practice in Medicinal Chemistry ( https://www.efmc.info/best‐practices ). Each of these stakeholders plays a role in shaping and influencing the landscape of medicinal chemistry education and defining the educational mandate with a distinct skill set of scientific hard skills and required soft skills to succeed as future drug designers. Unfortunately, traditional lecture formats are still a very common approach in teaching natural sciences. Although this teacher‐centered format can work reasonably well with certain charismatic teachers and highly self‐motivated students, who are both attentive in class and additionally invest the required hours after lectured to revise and learn the material, most classroom realities look different. Even if we assume all teachers were charismatic and all students were highly self‐motivated, it is highly questionable whether it is an efficient use of classroom time when we “read out” information to students considering that large amounts of information are easily available. Beyond that, it is unquestionable that this approach neither improve student's scientific communication and collaboration abilities nor help them to develop scientific and critical thinking and problem‐solving skills. Looking at the skills sets for medicinal chemists identified in Section 2 and summarized in Figure , this seems to be a clearly mismatching approach to teach medicinal chemistry and science in general. Thus, we need to adjust the way we are teaching medicinal chemistry to meet the learning outcomes required for the next generation of medicinal chemists to tackle the problems of our future. A Constructivist Active‐Learning Approach for Teaching Medicinal Chemistry Will Foster Critical Thinking and Understanding The learning theory has been long established, the teaching tools and their functional evidence exists. Constructivism learning theory tells us that learning is depending on understanding and that learners actively need to construct their own meaning to understand taught subjects. Learning is an active process and more than passive listening to fact‐based lectures. Thus, a learning environment must provide opportunities for active learning. Learners need to challenge and adjust their current understanding depending on what they encounter in the new learning situation. If the encounter is inconsistent with their current understanding, their understanding can change to accommodate new experiences. Learners remain active throughout this process: they apply current understandings, note relevant elements in new learning experiences, judge the consistency of prior and emerging knowledge, and based on that judgment, they can modify knowledge. Constructivist active learning activities such as flipped classroom settings, problem‐based learning (PBL) approaches or interrupted case‐studies put the focus of learning on critical thinking and understanding rather than on basic memorization, which allow the students to create organizing principles that they transfer more easily to other learning settings and subjects. Problem solving is furthermore fostering creative thinking. Learning in Medicinal Chemistry Needs to be Collaborative and Cooperative We have known for decades that social interactions (between students and teachers as well as among students) play a key role in the learning process. Learning situations therefore should be social, cooperative and collaborative. The importance on the emphasis of team environment within student‐centered teaching approaches has recently be highlighted by the implementation of “student centered team based learning” (SCTBL) in medicinal chemistry teaching. Creating a class room environment that emphasizes collaboration and exchange of ideas promotes social and communication skills as students learn how to articulate ideas clearly, negotiate with peers and evaluate their contributions in a socially acceptable manner to collaborate effectively on a task. In most medicinal chemistry classes, we can take advantage of the usually highly diverse student body to mimic the diverse and intercultural teams in medicinal chemistry departments of the pharmaceutical industry, if we actively involve the students into the teaching and learning process by fostering group discussions and peer‐to‐peer teaching activities. Looking at the teaching body for highly interdisciplinary medicinal chemistry courses, which is usually less diverse, it might help to incorporate multiple teachers into the course to break stereotyped role models, make teaching more inclusive and take advantage of specific expertise by incorporating colleagues from theoretical chemistry or biology. Furthermore, embedding of course parts led by researchers from industry is a valuable opportunity to include very specialized medicinal chemistry expertise, promote exchange between industry representatives and students, and align educational aims with industry needs. Finally, authentic learning tasks are crucial for meaningful learning as they stimulate students natural curiosity to the matter and engage students into critical thinking. Teachers will support the understanding of, for example, physicochemical values of drug candidates way better if students can experience their effects in real‐world examples of interrupted case‐studies on e. g. Hit‐to‐Lead optimizations taken from recent industry or academic medicinal research examples. The functional, scientific evidence of active student‐centered learning is massive and, as phrased perfectly by Clarissa Dirks, former co‐chair of the US National Academies Scientific Teaching Alliance: “At this point it would be unethical to teach any other way.” As implication for teaching medicinal chemistry courses, it should be our priority aim to design a comprehensive education in a constructivist active way that students can learn and apply the mandatory scientific hard skills and train the required soft skills in collaborative student‐centered active learning settings. A common counter argument against the implementation of collaborative active learning activities is that they consume more classroom time compared to classical lecture formats and would limit the content which could be covered in class. However, for example flipped classroom events can actually help to provide both time to cover course content and getting sufficient time to work through more complex problems in class by refreshing existing knowledge prior to class and reduce the in‐class cognitive load for the students. A More Holistic Approach: Teaching of Concepts and Train Dmta‐Cycle Competence The course content of medicinal chemistry lectures needs to be reduced, we need to leave such boring pharmacy lectures behind us which consist of getting lost in detail and reading through hundreds of examples of marked drug classes. On the contrary, we should concentrate on overarching principles and concepts to create room for topics like new modalities, covalent drugs, or compound classes bRo5. Therefore, a solid introduction into the general concepts of pharmacodynamics, pharmacokinetics and general design strategies is important at the beginning of the course. This might even be partially achieved using lecture style formats, but with implementation of flipped classroom events and significant interruption by active exercises enabling the students to apply and thereby understand the content. During such an introduction, the marketed drug classes still find their place but as perfect examples for the taught, general concepts. Course accompanying tutorials might give an additional focus on the synthesis of specific marketed drug classes. The major part of the courses should have a clear focus on design and synthesis of current drugs utilizing multiple active learning events to cover different topics. Thus, interrupted case studies based on recent industry and academic medicinal chemistry campaigns including multiparameter optimizations hold potential to enable both the coverage of different drug classes (e. g. SMOLs, PROTACs, cyclopeptide drugs, natural product derived antibiotics, etc.) as well as aspects of pharmacokinetic and drug design problems at different stages of the drug discovery process (e. g. hit‐finding, hit‐to‐lead optimization, lead optimization, etc.). All case studies should, first, challenge the students to retrosynthetically dissect the discussed molecules and suggest synthesis routes to facilitate the understanding on how drugs can be made and, further, embed the medicinal chemistry course as essential within organic chemistry. A smart selection of examples can ensure that multiple synthetic aspects and strategies (e. g. photo‐redox catalysis, electro chemical late‐stage derivatization, CH‐activation, combinatorial chemistry, solid‐phase supported synthesis or flow chemistry) are covered sufficiently. Such a more holistic approach would allow the students to explore effects of physicochemical compound properties and structural modifications and gain a clear understanding of DMTA cycles by evaluating their own decisions and strategies while discussing with their peers. Short introductions to the general field, the targeted diseases or bio pathological mechanism and market drugs in the indication could provide context for the case studies and might also give opportunities to visualize drug‐target interactions based on available co‐crystal structures using 3D‐illustration software. Recently, Lewis D. Pennigton published an article in which he outlined the analogues conceptual frameworks of total synthesis and medicinal chemistry stating that total synthesis and the retrosynthetic disconnection of natural products might be the perfect training for medicinal chemistry. Considering this case study examples derived from active natural products might be smoothly integrated into the teaching approach described above. Additionally, a flipped classroom type case study in which groups of students dissect a given publication on a drug discovery campaign guided by questions into a digestible core and present the result to their peers could not only help to manifest understanding and prepare the students for examinations but further help the teacher to identify aspects which need more clarification. Real‐Problem Laboratories with Hands‐On DMTA Cycle Decisions Laboratory parts should be mandatory for medicinal chemistry classes to connect theory with practice. However, although lab courses are by definition active, often they are conducted as classical cook‐book laboratories in which students follow a given procedure to conduct an experiment, collect data and partially draw conclusions from it. Certainly, students might learn how to conduct a specific experiment in such settings, but cook‐book labs leave out important parts of the scientific research process such as the definition of the research question, the background research, the formulation of a hypothesis, and the design of the experiment to prove the hypothesis. If we aim to teach students how to design and synthesize drugs and justify expensive lab courses, we need to provide a more comprehensive lab experience allowing students to experience at least a glance of what drug discovery means and how DMTA cycles in industry work in a more holistic scenario. A good set up for such a lab course could be problem‐based learning (PBL) or guided inquiry‐based lab (IBL) approaches, in which groups of students are confronted with structures and biological data of compounds in fictional drug discovery campaign as a real‐life and contextual problem to solve motivating them to fill the knowledge gap through collaborative knowledge building. As such approaches emphasize both process and content as the learning objectives, they are often perceived by students as more useful, and substantial evidence exists that such lab courses increase students’ motivation and outcomes with respect to data analysis and experimental design as well as their ability to ask questions. Furthermore, there are multiple successful implementations of such lab courses across different areas of chemistry underlining their efficacy. Thus, using PBL or guided IBL lab course design, a clear focus on DMTA cycles and the connection of structure and physicochemical properties can be set to connect to the medicinal chemistry content of the theoretical part of the medicinal chemistry course. The synthesis of potential drug candidates should be a central element in such lab courses, but embedded in a hit‐finding, hit‐to‐lead optimization, or lead‐optimization context, in which students must collaboratively work together in groups to: derive a hypothesis/identifying a research question from on given structure of potential drug candidates and their biological data identify proper assays to access derivative's performance design modified compounds based on structural information of a biological target synthesize/make modified compounds through retrosynthetic analysis, literature research, planning and conducting of a synthesis route test the compounds in simple biochemical, biological or spectroscopic assays, which might also include the generation of a tool compound to visualize the activity of the synthesized compounds by chemical biology means and critically analyze their results to make recommendations/decisions thus, going through a complete DMTA cycle, giving students to a realistic view of collaborative work in drug discovery projects and allowing for intentional development of team‐working skills. By this means, it is less important if the individual student groups have designed substantially improved compounds, as positive as well as negative DMTA cycle decision will support understanding of the overall drug discovery process once properly discussed and reflected upon. A good level of creative freedom in this process might even be an incentive to facilitate the students’ engagement in the task. This should further be supported engaging students in the use of state‐of‐the‐art virtual screening techniques and molecular docking utilizing available co‐crystal target structures to help the students to understand the structure and functionality of the biological target and promote their understanding of drug‐target interactions. Students in a medicinal chemistry lab course should learn how they can plan and conduct the synthesis of new derivatives. A clear understanding of retrosynthesis aspects from earlier advanced organic chemistry courses should be the basis for own retrosynthetic analysis of envisaged compounds. Thereby the use of chemical reaction research tools and databases like Scifinder or Reaxys, but also AI‐supported retrosynthesis software like (Synthia or others) should encourage students to touch base with these tools and enable critical reflection of opportunities such as evaluation of the environmental sustainability of selected synthesis routes. Due to time restrictions or cost limitations in medicinal chemistry lab courses, it might not always be possible to conduct complex biological assays within lab courses. An alternative could be simple colorimetric or fluorometric assays (e. g. GOD‐PAP or PUB assays) conducted with plate readers in well plates giving the students the opportunity to access the properties of the compounds they aim to change first‐hand. Additionally, data of more complex assays could be provided by the teacher to enable the simulation of complex, multi‐parametric DMTA cycles in drug discovery. Medicinal chemistry relevance of the lab course will be key to fostering the student's engagement and deepening the taught knowledge. Easy to implement and simple experiments such as NMR‐based LogP determination experiments in combination with property prediction by virtual property prediction tools could help students to understand the link between structural moieties in the drug candidate and observed ADME properties. Furthermore, the introduction of combined activity, physicochemical and pharmacokinetic property profiles might assist students to orientate within a multiparametric optimization study. Beyond that, splitting the course lab into two group of students with the first group generating derivatives of an enzyme inhibitor, while the second group deals with the design of a chemical probe enable the read‐out and visualization of the inhibitor's activity could also reflect the importance of chemical biology in modern drug discovery processes within the curriculum. During the lab course, groups of students should make decisions collaboratively based on scientific discussions guided by the lab assistant. Thereby, we can take advantage of the high diversity of current student bodies to train communication skills in intercultural and diverse teams. Finally, results of the lab courses might be documented by the students in publication style or even as scientific posters with incorporation of peer‐review cycles to improve scientific writing and proper communication. State‐of‐the‐Art Lab Equipment is Key for Good Medicinal Chemistry Education For such lab courses it will be of great importance to provide access to state‐of‐the‐art educational laboratories equipped with appropriate research devices (e. g. microwave or photo reactors, peptide synthesizers and automated chromatography devices for synthesis, or pipettes, plate readers, gel electrophoresis devices for the conduction of biochemical and biological assays) and student licenses for modelling, prediction, and retrosynthesis software. Beyond the associated cost, a common challenge in this regard might be that often educational laboratories are used for different lab courses and therefore need to be flexible in their equipment and the arrangement of lab and hood space. A simple win‐win solution to reduce equipment associated cost and retain flexibility might be to involve suppliers of synthesis and analytical devices providing the equipment during the lab course chargeless for demonstration. For the supplier, the chance to get future decision makers trained on company devices seems to be reward enough to engage in such deals. A recent example at the University of Gothenburg showed that this can lead to high‐end, automated equipment for course labs, very much appreciated by students and finally enabling time and space for additional biological experiments in the course. The learning theory has been long established, the teaching tools and their functional evidence exists. Constructivism learning theory tells us that learning is depending on understanding and that learners actively need to construct their own meaning to understand taught subjects. Learning is an active process and more than passive listening to fact‐based lectures. Thus, a learning environment must provide opportunities for active learning. Learners need to challenge and adjust their current understanding depending on what they encounter in the new learning situation. If the encounter is inconsistent with their current understanding, their understanding can change to accommodate new experiences. Learners remain active throughout this process: they apply current understandings, note relevant elements in new learning experiences, judge the consistency of prior and emerging knowledge, and based on that judgment, they can modify knowledge. Constructivist active learning activities such as flipped classroom settings, problem‐based learning (PBL) approaches or interrupted case‐studies put the focus of learning on critical thinking and understanding rather than on basic memorization, which allow the students to create organizing principles that they transfer more easily to other learning settings and subjects. Problem solving is furthermore fostering creative thinking. We have known for decades that social interactions (between students and teachers as well as among students) play a key role in the learning process. Learning situations therefore should be social, cooperative and collaborative. The importance on the emphasis of team environment within student‐centered teaching approaches has recently be highlighted by the implementation of “student centered team based learning” (SCTBL) in medicinal chemistry teaching. Creating a class room environment that emphasizes collaboration and exchange of ideas promotes social and communication skills as students learn how to articulate ideas clearly, negotiate with peers and evaluate their contributions in a socially acceptable manner to collaborate effectively on a task. In most medicinal chemistry classes, we can take advantage of the usually highly diverse student body to mimic the diverse and intercultural teams in medicinal chemistry departments of the pharmaceutical industry, if we actively involve the students into the teaching and learning process by fostering group discussions and peer‐to‐peer teaching activities. Looking at the teaching body for highly interdisciplinary medicinal chemistry courses, which is usually less diverse, it might help to incorporate multiple teachers into the course to break stereotyped role models, make teaching more inclusive and take advantage of specific expertise by incorporating colleagues from theoretical chemistry or biology. Furthermore, embedding of course parts led by researchers from industry is a valuable opportunity to include very specialized medicinal chemistry expertise, promote exchange between industry representatives and students, and align educational aims with industry needs. Finally, authentic learning tasks are crucial for meaningful learning as they stimulate students natural curiosity to the matter and engage students into critical thinking. Teachers will support the understanding of, for example, physicochemical values of drug candidates way better if students can experience their effects in real‐world examples of interrupted case‐studies on e. g. Hit‐to‐Lead optimizations taken from recent industry or academic medicinal research examples. The functional, scientific evidence of active student‐centered learning is massive and, as phrased perfectly by Clarissa Dirks, former co‐chair of the US National Academies Scientific Teaching Alliance: “At this point it would be unethical to teach any other way.” As implication for teaching medicinal chemistry courses, it should be our priority aim to design a comprehensive education in a constructivist active way that students can learn and apply the mandatory scientific hard skills and train the required soft skills in collaborative student‐centered active learning settings. A common counter argument against the implementation of collaborative active learning activities is that they consume more classroom time compared to classical lecture formats and would limit the content which could be covered in class. However, for example flipped classroom events can actually help to provide both time to cover course content and getting sufficient time to work through more complex problems in class by refreshing existing knowledge prior to class and reduce the in‐class cognitive load for the students. The course content of medicinal chemistry lectures needs to be reduced, we need to leave such boring pharmacy lectures behind us which consist of getting lost in detail and reading through hundreds of examples of marked drug classes. On the contrary, we should concentrate on overarching principles and concepts to create room for topics like new modalities, covalent drugs, or compound classes bRo5. Therefore, a solid introduction into the general concepts of pharmacodynamics, pharmacokinetics and general design strategies is important at the beginning of the course. This might even be partially achieved using lecture style formats, but with implementation of flipped classroom events and significant interruption by active exercises enabling the students to apply and thereby understand the content. During such an introduction, the marketed drug classes still find their place but as perfect examples for the taught, general concepts. Course accompanying tutorials might give an additional focus on the synthesis of specific marketed drug classes. The major part of the courses should have a clear focus on design and synthesis of current drugs utilizing multiple active learning events to cover different topics. Thus, interrupted case studies based on recent industry and academic medicinal chemistry campaigns including multiparameter optimizations hold potential to enable both the coverage of different drug classes (e. g. SMOLs, PROTACs, cyclopeptide drugs, natural product derived antibiotics, etc.) as well as aspects of pharmacokinetic and drug design problems at different stages of the drug discovery process (e. g. hit‐finding, hit‐to‐lead optimization, lead optimization, etc.). All case studies should, first, challenge the students to retrosynthetically dissect the discussed molecules and suggest synthesis routes to facilitate the understanding on how drugs can be made and, further, embed the medicinal chemistry course as essential within organic chemistry. A smart selection of examples can ensure that multiple synthetic aspects and strategies (e. g. photo‐redox catalysis, electro chemical late‐stage derivatization, CH‐activation, combinatorial chemistry, solid‐phase supported synthesis or flow chemistry) are covered sufficiently. Such a more holistic approach would allow the students to explore effects of physicochemical compound properties and structural modifications and gain a clear understanding of DMTA cycles by evaluating their own decisions and strategies while discussing with their peers. Short introductions to the general field, the targeted diseases or bio pathological mechanism and market drugs in the indication could provide context for the case studies and might also give opportunities to visualize drug‐target interactions based on available co‐crystal structures using 3D‐illustration software. Recently, Lewis D. Pennigton published an article in which he outlined the analogues conceptual frameworks of total synthesis and medicinal chemistry stating that total synthesis and the retrosynthetic disconnection of natural products might be the perfect training for medicinal chemistry. Considering this case study examples derived from active natural products might be smoothly integrated into the teaching approach described above. Additionally, a flipped classroom type case study in which groups of students dissect a given publication on a drug discovery campaign guided by questions into a digestible core and present the result to their peers could not only help to manifest understanding and prepare the students for examinations but further help the teacher to identify aspects which need more clarification. Laboratory parts should be mandatory for medicinal chemistry classes to connect theory with practice. However, although lab courses are by definition active, often they are conducted as classical cook‐book laboratories in which students follow a given procedure to conduct an experiment, collect data and partially draw conclusions from it. Certainly, students might learn how to conduct a specific experiment in such settings, but cook‐book labs leave out important parts of the scientific research process such as the definition of the research question, the background research, the formulation of a hypothesis, and the design of the experiment to prove the hypothesis. If we aim to teach students how to design and synthesize drugs and justify expensive lab courses, we need to provide a more comprehensive lab experience allowing students to experience at least a glance of what drug discovery means and how DMTA cycles in industry work in a more holistic scenario. A good set up for such a lab course could be problem‐based learning (PBL) or guided inquiry‐based lab (IBL) approaches, in which groups of students are confronted with structures and biological data of compounds in fictional drug discovery campaign as a real‐life and contextual problem to solve motivating them to fill the knowledge gap through collaborative knowledge building. As such approaches emphasize both process and content as the learning objectives, they are often perceived by students as more useful, and substantial evidence exists that such lab courses increase students’ motivation and outcomes with respect to data analysis and experimental design as well as their ability to ask questions. Furthermore, there are multiple successful implementations of such lab courses across different areas of chemistry underlining their efficacy. Thus, using PBL or guided IBL lab course design, a clear focus on DMTA cycles and the connection of structure and physicochemical properties can be set to connect to the medicinal chemistry content of the theoretical part of the medicinal chemistry course. The synthesis of potential drug candidates should be a central element in such lab courses, but embedded in a hit‐finding, hit‐to‐lead optimization, or lead‐optimization context, in which students must collaboratively work together in groups to: derive a hypothesis/identifying a research question from on given structure of potential drug candidates and their biological data identify proper assays to access derivative's performance design modified compounds based on structural information of a biological target synthesize/make modified compounds through retrosynthetic analysis, literature research, planning and conducting of a synthesis route test the compounds in simple biochemical, biological or spectroscopic assays, which might also include the generation of a tool compound to visualize the activity of the synthesized compounds by chemical biology means and critically analyze their results to make recommendations/decisions thus, going through a complete DMTA cycle, giving students to a realistic view of collaborative work in drug discovery projects and allowing for intentional development of team‐working skills. By this means, it is less important if the individual student groups have designed substantially improved compounds, as positive as well as negative DMTA cycle decision will support understanding of the overall drug discovery process once properly discussed and reflected upon. A good level of creative freedom in this process might even be an incentive to facilitate the students’ engagement in the task. This should further be supported engaging students in the use of state‐of‐the‐art virtual screening techniques and molecular docking utilizing available co‐crystal target structures to help the students to understand the structure and functionality of the biological target and promote their understanding of drug‐target interactions. Students in a medicinal chemistry lab course should learn how they can plan and conduct the synthesis of new derivatives. A clear understanding of retrosynthesis aspects from earlier advanced organic chemistry courses should be the basis for own retrosynthetic analysis of envisaged compounds. Thereby the use of chemical reaction research tools and databases like Scifinder or Reaxys, but also AI‐supported retrosynthesis software like (Synthia or others) should encourage students to touch base with these tools and enable critical reflection of opportunities such as evaluation of the environmental sustainability of selected synthesis routes. Due to time restrictions or cost limitations in medicinal chemistry lab courses, it might not always be possible to conduct complex biological assays within lab courses. An alternative could be simple colorimetric or fluorometric assays (e. g. GOD‐PAP or PUB assays) conducted with plate readers in well plates giving the students the opportunity to access the properties of the compounds they aim to change first‐hand. Additionally, data of more complex assays could be provided by the teacher to enable the simulation of complex, multi‐parametric DMTA cycles in drug discovery. Medicinal chemistry relevance of the lab course will be key to fostering the student's engagement and deepening the taught knowledge. Easy to implement and simple experiments such as NMR‐based LogP determination experiments in combination with property prediction by virtual property prediction tools could help students to understand the link between structural moieties in the drug candidate and observed ADME properties. Furthermore, the introduction of combined activity, physicochemical and pharmacokinetic property profiles might assist students to orientate within a multiparametric optimization study. Beyond that, splitting the course lab into two group of students with the first group generating derivatives of an enzyme inhibitor, while the second group deals with the design of a chemical probe enable the read‐out and visualization of the inhibitor's activity could also reflect the importance of chemical biology in modern drug discovery processes within the curriculum. During the lab course, groups of students should make decisions collaboratively based on scientific discussions guided by the lab assistant. Thereby, we can take advantage of the high diversity of current student bodies to train communication skills in intercultural and diverse teams. Finally, results of the lab courses might be documented by the students in publication style or even as scientific posters with incorporation of peer‐review cycles to improve scientific writing and proper communication. For such lab courses it will be of great importance to provide access to state‐of‐the‐art educational laboratories equipped with appropriate research devices (e. g. microwave or photo reactors, peptide synthesizers and automated chromatography devices for synthesis, or pipettes, plate readers, gel electrophoresis devices for the conduction of biochemical and biological assays) and student licenses for modelling, prediction, and retrosynthesis software. Beyond the associated cost, a common challenge in this regard might be that often educational laboratories are used for different lab courses and therefore need to be flexible in their equipment and the arrangement of lab and hood space. A simple win‐win solution to reduce equipment associated cost and retain flexibility might be to involve suppliers of synthesis and analytical devices providing the equipment during the lab course chargeless for demonstration. For the supplier, the chance to get future decision makers trained on company devices seems to be reward enough to engage in such deals. A recent example at the University of Gothenburg showed that this can lead to high‐end, automated equipment for course labs, very much appreciated by students and finally enabling time and space for additional biological experiments in the course. Following the recognition of the discussed paradigm changes in medicinal chemistry, it is time to not only adjust the curricula of medicinal chemistry courses regarding their content but furthermore revise courses towards constructivist teaching approaches fostering active and sustainable learning techniques justifying in class time and lab courses by additional value beyond pure knowledge resources. Medicinal chemistry teaching should be embedded in organic chemistry and chemical biology to address the core requirements of industry as the main stakeholder. Student‐centered, collaborative active learning techniques can make the crucial difference in medicinal chemistry teaching enabling the learning of mandatory scientific hard skills while applying and training the industry‐required soft skills. The authors declare no conflict of interest. |
Cardiotocography in practice: a qualitative study to explore obstetrical care professionals’ experiences with using cardiotocography information in Dutch practice | 2f7b4671-c915-4311-abef-6592a186ecea | 10277076 | Gynaecology[mh] | Monitoring fetal well-being by using a cardiotocograph (CTG) has become a standard practice of intrapartum care. CTG records the fetal heart rate (FHR) and uterine contractions to identify signs of intrapartum fetal hypoxia so that a timely intervention, for example, operative delivery, can take place and prenatal asphyxia may be prevented. This tool is heavily dependent on care professionals’ interpretation of its traces. CTG traces can have ambiguous clinical meanings and considerable interobserver and intraobserver variability has often been demonstrated, even among groups of experts. Still, 50 years after its introduction, interpretation of the FHR by CTG remains a challenge. In practice, working with information provided by CTG registrations is complex. CTG monitoring of the fetal and maternal condition not only requires frequent classification of the CTG tracing. Interpretation of this information leading to decision-making involves taking personal needs and circumstances of high-risk women in labour into account by obstetric healthcare providers working in interprofessional teams in dynamic care setting. Use of (international) interpretation guidelines and training programmes are often recommended to standardise interpretation and improve skills needed for classification of CTG tracings. The body of literature on classification skills, interventions to improve CTG use and related educational programmes tends to pay particular attention to the individual, without including the social and material context in which CTG is being used. To fully understand how obstetric caregivers, working in the context of dynamic teams and settings, use CTG information in their practice, more emphasis should be placed on the interaction between individuals and their practice context. Therefore, this study aimed to explore obstetric care professionals’ experiences with using CTG information and how they employed this tool in their practice. We have focused specifically on interpretation and interprofessional interactions, as insight in that aspect is deemed necessary to integrate educational interventions or other improvement efforts related to the use of CTG. Study design This qualitative study was based on an interpretivist research paradigm, focusing on understanding perspectives from various care professionals. This study combined semi-structured interviews with focus groups sessions. The Consolidated criteria for Reporting Qualitative research criteria were used for reporting qualitative research. Setting This study was conducted in the Netherlands. The Dutch Society of Obstetrics and Gynaecology (NVOG) recommends continuous intrapartum monitoring for women whose babiesare at risk for perinatal asphyxia and to use the modified classification of the International Federation of Obstetrics and Gynaecology (FIGO). This modified FIGO classification involves four categories of CTG classification: normal, suboptimal, abnormal, preterminal. Furthermore, the NVOG guideline states that interpretation and classification of the CTG, in the context of medical history and information on current pregnancy, should occur every hour during the first stage of labour and every 15 min during the second stage. If the CTG is classified as suboptimal or abnormal, interpretation has to be more frequent. There is no policy requirement for another clinician to independently review CTGs on a regular basis. However, in practice, both residents, clinical midwives and obstetricians jointly monitor active CTG registrations of women in labour. Clinical information from (ante)natal notes is part of this assessment of the CTG. In the Netherlands, CTG is used in hospital settings where obstetrical teams take care of women with increased risk of adverse fetal or maternal outcomes. More specialised care for complex and very preterm cases is provided in academic hospitals. Women with lower risks of pathology are cared for by independent primary-care midwives at home and in midwifery practices in the locality. The current study was performed at the Amsterdam University Medical Center (UMC), an academic hospital with two locations in an urban setting. With a total of approximately 2600 childbirths annually, Amsterdam UMC is the largest academic hospital in the Netherlands. CTG monitoring, according to the Dutch guideline, is applied to all women in labour under the care of an obstetrician, since it is a high-risk population. Central fetal monitoring is in use with fetal blood sampling as an additional technique to gain more insight in the fetal condition. Education on FHR monitoring is incorporated at daily handovers and regular meetings for staff. Participants, sampling and data collection Participants involved obstetricians, residents in obstetrics and gynaecology, junior physicians working in obstetrics and gynaecology (who are not in a postgraduate training programme), clinical midwives (also called hospital-based midwives) and nurses. For the semi-structured interviews, 30 care professionals were invited (15 care professionals from each location). We expected that we would need less than 30 interviews to reach data sufficiency, which we defined as no substantial additions to the themes that resulted from the interviews. Initial analysis started during data collection. After 20 interviews no new themes were found in the data. At that point, we decided to finish data collection after all 30 interviews were completed. All semi-structured interviews were conducted by the first author (AR). The interviews focused on care professionals’ experiences with using CTG information in practice as well as the learning process, involving six main topics: attitude towards the use of CTG, interpretation of CTG, decision-making process, consulting with colleagues, learning process and improvement efforts. The interview topics were derived from literature as well as authors (PB and PWT) with experience in using CTG in daily practice. The interview protocol was pilot tested. Since the interview topic guide barely changed after the pilot, the pilot interview was included in the current research sample. The interview topic guide is given as . Each interview lasted approximately 60 min. The interviews were conducted between May and October 2018. 10.1136/bmjopen-2022-068162.supp1 Supplementary data Care professionals with different professional backgrounds were approached for two focus group sessions in order to discuss and reflect on findings of the semi-structured interviews. Participation in an interview was no exclusion criterion for participation in the focus group. We included heterogeneous groups of all care professions to explore topics using different perspectives. The focus group sessions were facilitated by an independent experienced moderator with the first author (AR) present as an observer. Based on the main results of the semi-structured interviews, participants discussed the interview findings and used that discussion to jointly formulate opportunities for improvement of practical use of CTG. Each focus groups session lasted approximately 90 min. The focus group sessions were conducted in April 2019. All participants in this study were invited by email and gave informed consent for tape-recording the semi-structured interviews and focus group discussions. Within 4 weeks the participants received a summary for verification purposes. None of the participants requested any changes. Analysis All recordings of the semi-structured interviews were transcribed verbatim. These qualitative data were analysed using a conventional content analysis approach. All interviews were repeatedly read to capture the essence of the data. Relevant text fragments were highlighted and coded. No codes or categories were identified before analysis. An initial coding scheme was created by AR through line-by-line open coding of five interviews (one of each profession). Codes were sorted into categories based on how different codes were related and linked, discussed by all authors. For example, care professionals’ experiences on basic information of CTG components (code) as well as guidelines (code) were linked to knowledge (subcategory) as part of the main category ‘Individuals’. Further analysis led to modifications of this coding scheme. After 20 interviews no new codes were generated by the researchers and the final coding scheme was used to analyse the remaining interviews. Data from focus group sessions were not part of the initial content analysis, but were used to further refine the categories and understand how they were related to each other. Analysis was done using Atlas.ti V.8.0 software. Patient and public involvement None. This qualitative study was based on an interpretivist research paradigm, focusing on understanding perspectives from various care professionals. This study combined semi-structured interviews with focus groups sessions. The Consolidated criteria for Reporting Qualitative research criteria were used for reporting qualitative research. This study was conducted in the Netherlands. The Dutch Society of Obstetrics and Gynaecology (NVOG) recommends continuous intrapartum monitoring for women whose babiesare at risk for perinatal asphyxia and to use the modified classification of the International Federation of Obstetrics and Gynaecology (FIGO). This modified FIGO classification involves four categories of CTG classification: normal, suboptimal, abnormal, preterminal. Furthermore, the NVOG guideline states that interpretation and classification of the CTG, in the context of medical history and information on current pregnancy, should occur every hour during the first stage of labour and every 15 min during the second stage. If the CTG is classified as suboptimal or abnormal, interpretation has to be more frequent. There is no policy requirement for another clinician to independently review CTGs on a regular basis. However, in practice, both residents, clinical midwives and obstetricians jointly monitor active CTG registrations of women in labour. Clinical information from (ante)natal notes is part of this assessment of the CTG. In the Netherlands, CTG is used in hospital settings where obstetrical teams take care of women with increased risk of adverse fetal or maternal outcomes. More specialised care for complex and very preterm cases is provided in academic hospitals. Women with lower risks of pathology are cared for by independent primary-care midwives at home and in midwifery practices in the locality. The current study was performed at the Amsterdam University Medical Center (UMC), an academic hospital with two locations in an urban setting. With a total of approximately 2600 childbirths annually, Amsterdam UMC is the largest academic hospital in the Netherlands. CTG monitoring, according to the Dutch guideline, is applied to all women in labour under the care of an obstetrician, since it is a high-risk population. Central fetal monitoring is in use with fetal blood sampling as an additional technique to gain more insight in the fetal condition. Education on FHR monitoring is incorporated at daily handovers and regular meetings for staff. Participants involved obstetricians, residents in obstetrics and gynaecology, junior physicians working in obstetrics and gynaecology (who are not in a postgraduate training programme), clinical midwives (also called hospital-based midwives) and nurses. For the semi-structured interviews, 30 care professionals were invited (15 care professionals from each location). We expected that we would need less than 30 interviews to reach data sufficiency, which we defined as no substantial additions to the themes that resulted from the interviews. Initial analysis started during data collection. After 20 interviews no new themes were found in the data. At that point, we decided to finish data collection after all 30 interviews were completed. All semi-structured interviews were conducted by the first author (AR). The interviews focused on care professionals’ experiences with using CTG information in practice as well as the learning process, involving six main topics: attitude towards the use of CTG, interpretation of CTG, decision-making process, consulting with colleagues, learning process and improvement efforts. The interview topics were derived from literature as well as authors (PB and PWT) with experience in using CTG in daily practice. The interview protocol was pilot tested. Since the interview topic guide barely changed after the pilot, the pilot interview was included in the current research sample. The interview topic guide is given as . Each interview lasted approximately 60 min. The interviews were conducted between May and October 2018. 10.1136/bmjopen-2022-068162.supp1 Supplementary data Care professionals with different professional backgrounds were approached for two focus group sessions in order to discuss and reflect on findings of the semi-structured interviews. Participation in an interview was no exclusion criterion for participation in the focus group. We included heterogeneous groups of all care professions to explore topics using different perspectives. The focus group sessions were facilitated by an independent experienced moderator with the first author (AR) present as an observer. Based on the main results of the semi-structured interviews, participants discussed the interview findings and used that discussion to jointly formulate opportunities for improvement of practical use of CTG. Each focus groups session lasted approximately 90 min. The focus group sessions were conducted in April 2019. All participants in this study were invited by email and gave informed consent for tape-recording the semi-structured interviews and focus group discussions. Within 4 weeks the participants received a summary for verification purposes. None of the participants requested any changes. All recordings of the semi-structured interviews were transcribed verbatim. These qualitative data were analysed using a conventional content analysis approach. All interviews were repeatedly read to capture the essence of the data. Relevant text fragments were highlighted and coded. No codes or categories were identified before analysis. An initial coding scheme was created by AR through line-by-line open coding of five interviews (one of each profession). Codes were sorted into categories based on how different codes were related and linked, discussed by all authors. For example, care professionals’ experiences on basic information of CTG components (code) as well as guidelines (code) were linked to knowledge (subcategory) as part of the main category ‘Individuals’. Further analysis led to modifications of this coding scheme. After 20 interviews no new codes were generated by the researchers and the final coding scheme was used to analyse the remaining interviews. Data from focus group sessions were not part of the initial content analysis, but were used to further refine the categories and understand how they were related to each other. Analysis was done using Atlas.ti V.8.0 software. None. In this qualitative study, 30 care professionals participated in semi-structured interviews. In addition, 15 care professionals participated in two focus group sessions, of whom two participants were also involved in the interviews (from location A). In total, 43 care professionals were included in this study. shows the respondent characteristics. The majority of the respondents were female, representative for the team composition at the hospital. Based on our analysis, we have organised obstetrical care professionals’ experiences with using CTG information and how they employ this tool in their practice in three main categories: (1) individual characteristics, (2) teams, (3) work environment. shows the main categories and the related subcategories. Individual characteristics This category refers to individual characteristics of care professionals. Three characteristics were distinguished: knowledge, experience and personal beliefs . First, using CTG information in daily practice required knowledge on technical aspects and basic CTG features (baseline, variability, accelerations and decelerations) for classification purposes. I’ve learned to work with CTG by doing, seeing and being taught. Where to look at? What do you see and why? Basically, knowledge transfer (P2) Moreover, participants stated that knowledge on physiological mechanisms and clinical context information is necessary for a comprehensive interpretation and adequate management. Participants acknowledged the importance of consensus on the definition of CTG characteristics defined in guidelines, yet struggled to be consistent. Participants recognised that definitions were not always used appropriately which could lead to confusion about a patient’s situation. A common example given by participants was the definition of decelerations, by not specifying the type of deceleration or simple referring to ‘a slight dip’ in the fetal heart rate. Second, all participants emphasised that the ability to use CTG in clinical context required experience . CTG is a matter of doing. In theory, there are ways to describe a CTG, good and bad scenario’s. However, in practice there are many variations. (…) And you need exposure to learn about these variations, the outcomes and management. (P1) Being involved in the interpretation and decision-making process and by questioning or discussing CTG patterns with team members participants developed practical know-how. Participants explained that there was no substitute for practical experience; it helped build confidence in one’s skills and develop intuition by incorporating clinical context information. Moreover, participants mentioned that increasing experience made them more alert for (sudden) pattern changes, enabled a better understanding of appropriate management and a broader range of possible interventions. On the contrary, less experienced participants mentioned that they relied heavily on the assessment of their team members, expressed feelings of insecurity about their own skills and worried about possible diagnostic errors, making them more prone to choose obstetrical interventions (according to them). All participants referred to experience as a dynamic frame of reference, based on pattern recognition. It was often mentioned that experience helped to recognise recurrent scenarios, related to the outcome of childbirth and clinical context. Mentioned drawbacks of this pattern recognition were false reassurance as well as a shift of frame of reference after (several) negative experiences. Third, participants expressed different personal beliefs about the effectiveness of CTG in adequately monitoring fetal well-being. Some indicated to have faith in the technology, others were reluctant to trust CTG interpretation alone and appreciated additional tests, and few were sceptical of the value of CTG and questioned whether CTG should be used at all. It was suggested by participants that personal beliefs influenced interpretation and decision-making in practice. And it also has to do with, whether you have a defensive attitude and you want to avoid making a wrong assessment at all costs and, therefore, act on all deviations. Or are you willing to accept variations in nature and to see how something develops without interference? And I believe more in the latter. (P12) Teams A much-discussed topic in the interviews and focus groups was the importance of teamwork when using CTG in practice, involving collaboration between different obstetrical team members in and between shifts. Regardless of the level of experience, participants described some degree of uncertainty when interpreting a CTG pattern or determining appropriate management. Therefore, easily approachable team members were highly valued by all care professions to help with and discuss interpretation of CTG patterns. If I have any doubts about something, I will discuss it. I think you look at it [CTG] with several people, and sometimes I notice something which I will point out to someone and another time someone else does that for me. You have to do it together. (P8) Participants mentioned various functions of collaboration, a continuum ranging from making an assessment together for learning purposes, looking for confirmation, a second opinion or requesting consultation for an (expected) deteriorating case. In a request for help with (expected) severe and acute cases participants stated that they would involve a more experienced team member and/or supervisor. However, in non-urgent situations most participants indicated that profession was irrelevant and they simply wanted to share their thinking with a team member. Many respondents emphasised that collaboration, in any form, created a sense of shared responsibility. Sometimes you just need someone to spar with or consult, and I actually think that you often find a solution and that that consultation is sometimes initiated to feel supported or strengthened in a decision. (P19) According to participants, team composition could have an influence on collaboration which they often illustrated by the difference between day and night shifts . In general, participants referred to day shifts as being more dynamic and crowded as more team members were available for a shared assessment of CTG. Though, some participants stated that more input, distractions and stress were perceived as limitations for practical use of CTG during day shifts. On the contrary, night shifts were relatively more quiet and participants could focus better on CTG traces in a smaller team. However, in this setting, a supervisor was not always physically present and phoning for help with CTG interpretation was perceived as a barrier. In addition, many participants recognised that especially during night shifts there was a risk of ‘tunnel vision’ regarding CTG interpretation. Care professionals could feel more insecure about CTG interpretation at night and, affected by fatigue and team composition, even postpone a decision for non-urgent interventions until the day shift arrived. Moreover, participants identified shift handovers as important moments for CTG interpretation and decision-making. CTG interpretation by team members of a new shift was described as offering a fresh perspective. Many respondents experienced that this fresh perspective could lead to a different interpretation or policy. And yet, I don’t know whether its objective or subjective, but still you start a new shift with a more fresh perspective. You are just … well if it is quicker I don’t know (…) but your helicopter view is just wider than if you have been on it for eight hours (P18) Work environment This category refers to environmental factors which influenced the use of CTG in practice. First, a prerequisite for working with CTG was functional equipment . Participants found it essential that CTG traces were visible at multiple screens in the labour ward. Functional equipment is important of course. Well-functioning, complete and visible on screen (P15) Second, participants referred often to norms and values of the maternity unit. Participants who mentioned their previous workplace were most explicit about difference in cultural aspects as they compared uses of CTG in practice. Examples of differences in perceived cultur e were; time to respond to suspicious or abnormal CTG traces and related interventions, frequency of interdisciplinary case reviews based on CTG tracings (eg, during shift handovers) and request for help and speaking up. Participants trusted that their colleagues would speak up on concerns about CTG traces and that team members would accept unsolicited input. Although this was widely supported, many participants were able to recall situations where they struggled to or did not speak up to a supervisor or a more experienced team member. And that [speak-up] is easy if you have a young colleague, for example a junior physician, but for me it’s something else when it comes to a more experienced colleague (…) a midwife or obstetrician who has been doing this work for much longer than I do, and can cause friction, at least … I find it quite uncomfortable to provide unsolicited input. Because it also gives, a kind of, sign of lack of trust … (P9) Finally, participants stated that working with CTG implied a continues learning process requiring education. In many interviews and the focus groups sessions it was argued that a basic course should be compulsory, for all who start working with CTG regardless the profession. Subsequently, participants stressed the importance of recurrent training and frequent CTG discussion and reviews with colleagues, preferable in a multidisciplinary setting. This could mean different things, ranging from shared assessments of a CTG trace, explicit attention for interpretation and decision-making based on CTG information during handover as well as specifically planned meetings to discuss case scenarios. You should organise regular meetings and give colleagues the opportunity, like i said, for example to share cases of their own shifts to learn and learn from each other. (P14). Discussion This qualitative study explored obstetric care professionals’ experiences with using CTG information and how they employed this tool in their practice. Based on semi-structured interviews and focus group sessions, three categories were identified; individual characteristics, teams and work environment. Smith et al performed a systematic review and thematic analysis on professionals’ views of electronic fetal monitoring. Four themes were identified, reassurance, technology, communication/education and midwife by proxy. The current study used different methodology, yet some similar results were found. Individual factors were related to views as well as experiences with using CTG information, for example the personal beliefs regarding the technology. Moreover, in both studies communication and education were considered as relevant categories. Smith et al reported concerns that the technology might hinder effective communication with women in labour. Our study focused on communication between team members and highlighted the importance of collaboration in and between shifts. Participants emphasised the importance of collaboration considering the need for frequent (interprofessional and intraprofessional) consultations, reassurance as well as establishing shared responsibility for maternal and fetal outcomes. These results relate to the concept of situational awareness and the need to include perceptions of team members in decision-making processes. Therefore, comparable to previous findings it is recommended to embed these team factors in educational efforts to improve CTG interpretation and management. Our study highlights that use of CTG involves a continuous learning process. Development of individual skills is a prerequisite (regardless of what guideline is in use), however, use of CTG in practice is perceived as a team-based activity which requires regular multidisciplinary meetings to learn from colleagues’ perspectives. Finally, this study provided insight into vulnerabilities related to use of CTG in practice. For example, care professionals with little CTG experience can feel insecure about their own skills and sometimes perceive a barrier to phone for help. Young et al observed that junior medical staff is a group most frequently involved in suboptimal care. Moreover, Ugwumadu et al pointed out in a commentary that ‘often critical decisions are left to trainees or junior staff members, while those who have developed expertise over the years are not even aware that there is a problem’. Our results suggested that supervisors and more experienced team members can (and should) work to create a culture that enables a low threshold for collaboration and a safe learning environment. Another finding was the positive impact of ‘a fresh perspective’ at shift handovers. Our participants reported that this is influenced by fatigue, team composition and the availability of resources. This was most apparent comparing day to night shifts, according to many participants. Organisations need to be aware of these issues and further research is necessary on the impact of ‘a fresh perspective’ on patient safety in obstetrics. Limitations This study was conducted as a single centre study in one hospital institution, with two locations, in the Netherlands. Therefore, transferability of findings may be limited to settings where composition of the team or workplace environment is structured similarly. The study aimed to be explorative, yielding insight in various factors and their interactions. Findings, however, cannot be used to know what effect changes in one category might have on the other categories. Conclusions In conclusion, in this study the use of CTG in practice was perceived as a team-effort rather than an individual task. In practice, individual skills interact with environmental factors and collaboration within multi-professional teams, resulting in CTG interpretation and subsequent decision-making. There is a particular need to create shared responsibility among team members, which should be addressed in educational programmes and regular multidisciplinary meetings to learn from colleagues’ perspectives. Despite its limitations, CTG is still the most commonly used tool worldwide to monitor the condition of the fetus during labour. The findings of this study can contribute to improve how CTG is employed in practice. This category refers to individual characteristics of care professionals. Three characteristics were distinguished: knowledge, experience and personal beliefs . First, using CTG information in daily practice required knowledge on technical aspects and basic CTG features (baseline, variability, accelerations and decelerations) for classification purposes. I’ve learned to work with CTG by doing, seeing and being taught. Where to look at? What do you see and why? Basically, knowledge transfer (P2) Moreover, participants stated that knowledge on physiological mechanisms and clinical context information is necessary for a comprehensive interpretation and adequate management. Participants acknowledged the importance of consensus on the definition of CTG characteristics defined in guidelines, yet struggled to be consistent. Participants recognised that definitions were not always used appropriately which could lead to confusion about a patient’s situation. A common example given by participants was the definition of decelerations, by not specifying the type of deceleration or simple referring to ‘a slight dip’ in the fetal heart rate. Second, all participants emphasised that the ability to use CTG in clinical context required experience . CTG is a matter of doing. In theory, there are ways to describe a CTG, good and bad scenario’s. However, in practice there are many variations. (…) And you need exposure to learn about these variations, the outcomes and management. (P1) Being involved in the interpretation and decision-making process and by questioning or discussing CTG patterns with team members participants developed practical know-how. Participants explained that there was no substitute for practical experience; it helped build confidence in one’s skills and develop intuition by incorporating clinical context information. Moreover, participants mentioned that increasing experience made them more alert for (sudden) pattern changes, enabled a better understanding of appropriate management and a broader range of possible interventions. On the contrary, less experienced participants mentioned that they relied heavily on the assessment of their team members, expressed feelings of insecurity about their own skills and worried about possible diagnostic errors, making them more prone to choose obstetrical interventions (according to them). All participants referred to experience as a dynamic frame of reference, based on pattern recognition. It was often mentioned that experience helped to recognise recurrent scenarios, related to the outcome of childbirth and clinical context. Mentioned drawbacks of this pattern recognition were false reassurance as well as a shift of frame of reference after (several) negative experiences. Third, participants expressed different personal beliefs about the effectiveness of CTG in adequately monitoring fetal well-being. Some indicated to have faith in the technology, others were reluctant to trust CTG interpretation alone and appreciated additional tests, and few were sceptical of the value of CTG and questioned whether CTG should be used at all. It was suggested by participants that personal beliefs influenced interpretation and decision-making in practice. And it also has to do with, whether you have a defensive attitude and you want to avoid making a wrong assessment at all costs and, therefore, act on all deviations. Or are you willing to accept variations in nature and to see how something develops without interference? And I believe more in the latter. (P12) A much-discussed topic in the interviews and focus groups was the importance of teamwork when using CTG in practice, involving collaboration between different obstetrical team members in and between shifts. Regardless of the level of experience, participants described some degree of uncertainty when interpreting a CTG pattern or determining appropriate management. Therefore, easily approachable team members were highly valued by all care professions to help with and discuss interpretation of CTG patterns. If I have any doubts about something, I will discuss it. I think you look at it [CTG] with several people, and sometimes I notice something which I will point out to someone and another time someone else does that for me. You have to do it together. (P8) Participants mentioned various functions of collaboration, a continuum ranging from making an assessment together for learning purposes, looking for confirmation, a second opinion or requesting consultation for an (expected) deteriorating case. In a request for help with (expected) severe and acute cases participants stated that they would involve a more experienced team member and/or supervisor. However, in non-urgent situations most participants indicated that profession was irrelevant and they simply wanted to share their thinking with a team member. Many respondents emphasised that collaboration, in any form, created a sense of shared responsibility. Sometimes you just need someone to spar with or consult, and I actually think that you often find a solution and that that consultation is sometimes initiated to feel supported or strengthened in a decision. (P19) According to participants, team composition could have an influence on collaboration which they often illustrated by the difference between day and night shifts . In general, participants referred to day shifts as being more dynamic and crowded as more team members were available for a shared assessment of CTG. Though, some participants stated that more input, distractions and stress were perceived as limitations for practical use of CTG during day shifts. On the contrary, night shifts were relatively more quiet and participants could focus better on CTG traces in a smaller team. However, in this setting, a supervisor was not always physically present and phoning for help with CTG interpretation was perceived as a barrier. In addition, many participants recognised that especially during night shifts there was a risk of ‘tunnel vision’ regarding CTG interpretation. Care professionals could feel more insecure about CTG interpretation at night and, affected by fatigue and team composition, even postpone a decision for non-urgent interventions until the day shift arrived. Moreover, participants identified shift handovers as important moments for CTG interpretation and decision-making. CTG interpretation by team members of a new shift was described as offering a fresh perspective. Many respondents experienced that this fresh perspective could lead to a different interpretation or policy. And yet, I don’t know whether its objective or subjective, but still you start a new shift with a more fresh perspective. You are just … well if it is quicker I don’t know (…) but your helicopter view is just wider than if you have been on it for eight hours (P18) This category refers to environmental factors which influenced the use of CTG in practice. First, a prerequisite for working with CTG was functional equipment . Participants found it essential that CTG traces were visible at multiple screens in the labour ward. Functional equipment is important of course. Well-functioning, complete and visible on screen (P15) Second, participants referred often to norms and values of the maternity unit. Participants who mentioned their previous workplace were most explicit about difference in cultural aspects as they compared uses of CTG in practice. Examples of differences in perceived cultur e were; time to respond to suspicious or abnormal CTG traces and related interventions, frequency of interdisciplinary case reviews based on CTG tracings (eg, during shift handovers) and request for help and speaking up. Participants trusted that their colleagues would speak up on concerns about CTG traces and that team members would accept unsolicited input. Although this was widely supported, many participants were able to recall situations where they struggled to or did not speak up to a supervisor or a more experienced team member. And that [speak-up] is easy if you have a young colleague, for example a junior physician, but for me it’s something else when it comes to a more experienced colleague (…) a midwife or obstetrician who has been doing this work for much longer than I do, and can cause friction, at least … I find it quite uncomfortable to provide unsolicited input. Because it also gives, a kind of, sign of lack of trust … (P9) Finally, participants stated that working with CTG implied a continues learning process requiring education. In many interviews and the focus groups sessions it was argued that a basic course should be compulsory, for all who start working with CTG regardless the profession. Subsequently, participants stressed the importance of recurrent training and frequent CTG discussion and reviews with colleagues, preferable in a multidisciplinary setting. This could mean different things, ranging from shared assessments of a CTG trace, explicit attention for interpretation and decision-making based on CTG information during handover as well as specifically planned meetings to discuss case scenarios. You should organise regular meetings and give colleagues the opportunity, like i said, for example to share cases of their own shifts to learn and learn from each other. (P14). This qualitative study explored obstetric care professionals’ experiences with using CTG information and how they employed this tool in their practice. Based on semi-structured interviews and focus group sessions, three categories were identified; individual characteristics, teams and work environment. Smith et al performed a systematic review and thematic analysis on professionals’ views of electronic fetal monitoring. Four themes were identified, reassurance, technology, communication/education and midwife by proxy. The current study used different methodology, yet some similar results were found. Individual factors were related to views as well as experiences with using CTG information, for example the personal beliefs regarding the technology. Moreover, in both studies communication and education were considered as relevant categories. Smith et al reported concerns that the technology might hinder effective communication with women in labour. Our study focused on communication between team members and highlighted the importance of collaboration in and between shifts. Participants emphasised the importance of collaboration considering the need for frequent (interprofessional and intraprofessional) consultations, reassurance as well as establishing shared responsibility for maternal and fetal outcomes. These results relate to the concept of situational awareness and the need to include perceptions of team members in decision-making processes. Therefore, comparable to previous findings it is recommended to embed these team factors in educational efforts to improve CTG interpretation and management. Our study highlights that use of CTG involves a continuous learning process. Development of individual skills is a prerequisite (regardless of what guideline is in use), however, use of CTG in practice is perceived as a team-based activity which requires regular multidisciplinary meetings to learn from colleagues’ perspectives. Finally, this study provided insight into vulnerabilities related to use of CTG in practice. For example, care professionals with little CTG experience can feel insecure about their own skills and sometimes perceive a barrier to phone for help. Young et al observed that junior medical staff is a group most frequently involved in suboptimal care. Moreover, Ugwumadu et al pointed out in a commentary that ‘often critical decisions are left to trainees or junior staff members, while those who have developed expertise over the years are not even aware that there is a problem’. Our results suggested that supervisors and more experienced team members can (and should) work to create a culture that enables a low threshold for collaboration and a safe learning environment. Another finding was the positive impact of ‘a fresh perspective’ at shift handovers. Our participants reported that this is influenced by fatigue, team composition and the availability of resources. This was most apparent comparing day to night shifts, according to many participants. Organisations need to be aware of these issues and further research is necessary on the impact of ‘a fresh perspective’ on patient safety in obstetrics. This study was conducted as a single centre study in one hospital institution, with two locations, in the Netherlands. Therefore, transferability of findings may be limited to settings where composition of the team or workplace environment is structured similarly. The study aimed to be explorative, yielding insight in various factors and their interactions. Findings, however, cannot be used to know what effect changes in one category might have on the other categories. In conclusion, in this study the use of CTG in practice was perceived as a team-effort rather than an individual task. In practice, individual skills interact with environmental factors and collaboration within multi-professional teams, resulting in CTG interpretation and subsequent decision-making. There is a particular need to create shared responsibility among team members, which should be addressed in educational programmes and regular multidisciplinary meetings to learn from colleagues’ perspectives. Despite its limitations, CTG is still the most commonly used tool worldwide to monitor the condition of the fetus during labour. The findings of this study can contribute to improve how CTG is employed in practice. Reviewer comments Author's manuscript |
Recommendations to improve use and dissemination of patient versions of oncological clinical practice guidelines in Germany: results of a multi-stakeholder workshop | 7c6e41c6-6028-48fe-90df-76097ae6daef | 11373279 | Internal Medicine[mh] | Oncological patients have high information needs that are often unmet . Unfulfilled information needs might be related to quality of life, level of depression and anxiety as well as physical symptoms . Patients’ needs range from the basic need for medical information and documentation, to the need for additional information and explanation to complement that provided by health professionals, to the need for support, assistance and advice depending on the difficulties encountered, to the need for listening and psychological support . Patient versions of clinical practice guidelines (PVGs) as a special form of evidence-based practice information might help to address the basic need for medical information and documentation as well as need for additional information and explanation and they often provide contact addresses for additional support. PVGs translate clinical practice guidelines (CPG) into common speech . CPG provide evidence-based recommendations with regard to medical conditions and mainly aim to help health care providers in the decision-making process regarding appropriate care . Patients can also use them as a source of information . The transformation into a PVGs helps to increase the understandability for laypersons, since the concept of CPGs is sometimes difficult for them to understand . However, PVGs do not only include a translation of the CPG, they often also contain further and explanatory information. Examples are the introduction to the grading of recommendations, but also background information on the disease comprehensible to patients or further addresses, for example to self-help groups. In Germany, the German Guideline Program in Oncology (GGPO) develops oncological PVGs for various diseases. Currently, 30 oncological PVG have been published by the GGPO and more are being developed. These PVGs are available in PDF format and as a printed brochure. The printed brochures can be ordered at the website of the German Cancer Aid . The PDF versions can also be downloaded from this website as well as from the website of the GGPO . Both are available free of charge. The development process follows a strict methodology . To our knowledge, there is only scarce information on the use and applicability of oncological PVGs in Germany. To obtain information on this as well as on possible ways for improvements, a large multi-phase study was carried out. The study included a review to assess international methods and approaches of PVGs , qualitative interviews on experiences of international guideline producers as well as qualitative interviews and focus groups to analyse the national perspective on the implementation and dissemination of PVGs. Further information on the study can be found in the protocol . The last stage of the study was the development of recommendations based on the previous study results and on the knowledge of several stakeholders within a workshop. The aim was to formulate recommendations that can improve the use and dissemination of PVGs based on the results of the main project and the consultation of several stakeholders. This will help to transfer the results into practice. Design To formulate recommendations for improvement, a one-day workshop was held consisting of a World Café and subsequent voting on the recommendations. The World Café is a method for engaging people in discussions on diverse topics . Unfortunately, we are not aware of any reporting guidelines for publications following a workshop. Accordingly, we did not use any reporting guideline for the preparation of the manuscript. Recruitment Invitations were sent by e-mail to 50 organisations from German-speaking countries. Users of PVGs (patients, medical staff, and multipliers) were included as well as creators and initiators/funding organisations of PVGs and organisations with methodological expertise in the development of clinical practice guidelines or in patient health information. The organisations were selected in cooperation with the project partners: the GGPO, the Association of the Scientific Medical Societies in Germany - Institute for Medical Knowledge Management (AWMF-IMWi), the German Agency for Quality in Medicine (ÄZQ), and two German self-help groups focusing on prostate cancer (Bundesverband Prostatakrebs Selbsthilfe [BPS]) and cancer in women (Frauenselbsthilfe Krebs–Bundesverband [FSH]). A save the date was sent in June 2022, followed by the initial invitation in October 2022, and a reminder in November 2022, if no response was received. If unsuccessful, personal contacts were used where possible. The organisations themselves decided whom to register for participation. However, only one person per organisation could participate. Preparation of the workshop The researchers involved drafted recommendations based on the project results. To achieve this, they summarised the project results. In a brainstorming session, they discussed which recommendations for action can be derived from the results. Some recommendations for action were based on clear results, but in other cases the results contradicted each other. An assessment was made as to which recommendations were based on clear results within the project and whose implementation was considered practicable (feasible) and which were not (to be discussed). The recommendations were then made available to the project partners together with an accompanying explanatory text. The project partners were asked to take part in an online survey to vote on the extent to which the classification in the “feasible” category was appropriate. The recommendations were then re-categorised in line with the voting results, if necessary, and the comments were added anonymously as additional background information. The document, consisting of recommendations, accompanying explanatory text and possible comments from the project partners, was then made available to all workshop participants in advance. Workshop The workshop was conducted in April 2023. Information on the entire process related to the workshop can be found in Fig. . The Workshop was facilitated by JB (Introduction/Presentation of recommendations and rationale) and MN (neutral moderation/voting). It began with a short introduction of the project and its results to ensure a common knowledge of the topic. After this, the process of the World Café was introduced. Four tables on the topics (1) dissemination, (2) dissemination and use of the PVG in collaboration between healthcare providers and patients, (3) format/design and (digital) links, and (4) digitalisation/up-to-dateness were prepared in advance. Recommendations voted as “to be discussed” in advance were printed as a basis for discussion. Each table was hosted by a member of the project team (JH, NK, SW, JB) to ensure a focused discussion, because of the high number of recommendations assigned to each table. Furthermore, the host wrote down the discussion points on a flip chart and ensured that all group member were involved in the discussion. Participants were assigned to the groups in advance to ensure a heterogeneous composition. After 25 min, the groups switched to the next table. The host summarized the discussion points of the previous groups at the beginning. After the first two tables, there was a lunch break to promote personal exchange. Following the lunch break the discussion was continued until each group had discussed on every topic. After the hosts presented the results of the tables to the whole group, the voting was conducted. Participants voted in blocks on all recommendations from each topic that were a priori assigned to the “feasible” category. Recommendations assigned to the “to be discussed” category were voted individually. Participants could agree, disagree, or abstain from voting. Voting was open using coloured cards. Participants who abstained from voting were excluded from the total population for the calculation of the approval rate. Recommendations with agreement of > 75% were approved, in case of ≤ 75% agreement, recommendations were rejected. If organisations were unable to send a representative or its representative had to leave before voting was conducted for all recommendations, the right to vote could be transferred to the representative of another organisation. Data synthesis Following the workshop, the project team added the discussion points and the voting results to the recommendations developed in advance. The updated recommendations were uploaded to the project website and sent to all workshop participants as well as people who had expressed an interest in the project results. To formulate recommendations for improvement, a one-day workshop was held consisting of a World Café and subsequent voting on the recommendations. The World Café is a method for engaging people in discussions on diverse topics . Unfortunately, we are not aware of any reporting guidelines for publications following a workshop. Accordingly, we did not use any reporting guideline for the preparation of the manuscript. Invitations were sent by e-mail to 50 organisations from German-speaking countries. Users of PVGs (patients, medical staff, and multipliers) were included as well as creators and initiators/funding organisations of PVGs and organisations with methodological expertise in the development of clinical practice guidelines or in patient health information. The organisations were selected in cooperation with the project partners: the GGPO, the Association of the Scientific Medical Societies in Germany - Institute for Medical Knowledge Management (AWMF-IMWi), the German Agency for Quality in Medicine (ÄZQ), and two German self-help groups focusing on prostate cancer (Bundesverband Prostatakrebs Selbsthilfe [BPS]) and cancer in women (Frauenselbsthilfe Krebs–Bundesverband [FSH]). A save the date was sent in June 2022, followed by the initial invitation in October 2022, and a reminder in November 2022, if no response was received. If unsuccessful, personal contacts were used where possible. The organisations themselves decided whom to register for participation. However, only one person per organisation could participate. The researchers involved drafted recommendations based on the project results. To achieve this, they summarised the project results. In a brainstorming session, they discussed which recommendations for action can be derived from the results. Some recommendations for action were based on clear results, but in other cases the results contradicted each other. An assessment was made as to which recommendations were based on clear results within the project and whose implementation was considered practicable (feasible) and which were not (to be discussed). The recommendations were then made available to the project partners together with an accompanying explanatory text. The project partners were asked to take part in an online survey to vote on the extent to which the classification in the “feasible” category was appropriate. The recommendations were then re-categorised in line with the voting results, if necessary, and the comments were added anonymously as additional background information. The document, consisting of recommendations, accompanying explanatory text and possible comments from the project partners, was then made available to all workshop participants in advance. The workshop was conducted in April 2023. Information on the entire process related to the workshop can be found in Fig. . The Workshop was facilitated by JB (Introduction/Presentation of recommendations and rationale) and MN (neutral moderation/voting). It began with a short introduction of the project and its results to ensure a common knowledge of the topic. After this, the process of the World Café was introduced. Four tables on the topics (1) dissemination, (2) dissemination and use of the PVG in collaboration between healthcare providers and patients, (3) format/design and (digital) links, and (4) digitalisation/up-to-dateness were prepared in advance. Recommendations voted as “to be discussed” in advance were printed as a basis for discussion. Each table was hosted by a member of the project team (JH, NK, SW, JB) to ensure a focused discussion, because of the high number of recommendations assigned to each table. Furthermore, the host wrote down the discussion points on a flip chart and ensured that all group member were involved in the discussion. Participants were assigned to the groups in advance to ensure a heterogeneous composition. After 25 min, the groups switched to the next table. The host summarized the discussion points of the previous groups at the beginning. After the first two tables, there was a lunch break to promote personal exchange. Following the lunch break the discussion was continued until each group had discussed on every topic. After the hosts presented the results of the tables to the whole group, the voting was conducted. Participants voted in blocks on all recommendations from each topic that were a priori assigned to the “feasible” category. Recommendations assigned to the “to be discussed” category were voted individually. Participants could agree, disagree, or abstain from voting. Voting was open using coloured cards. Participants who abstained from voting were excluded from the total population for the calculation of the approval rate. Recommendations with agreement of > 75% were approved, in case of ≤ 75% agreement, recommendations were rejected. If organisations were unable to send a representative or its representative had to leave before voting was conducted for all recommendations, the right to vote could be transferred to the representative of another organisation. Following the workshop, the project team added the discussion points and the voting results to the recommendations developed in advance. The updated recommendations were uploaded to the project website and sent to all workshop participants as well as people who had expressed an interest in the project results. Participants The five-hour workshop took place on 24th April 2023 in Cologne, Germany. 23 people from 24 organisations participated in the discussion. Four representatives from four organisations had to cancel their participation at short notice for various reasons. Users of PVGs (patients ( n = 6), medical staff ( n = 4), and multipliers ( n = 2)) were included as well as creators ( n = 4) and initiators/funding organisations ( n = 3) of PVGs and organisations with methodological expertise in the development of clinical practice guidelines ( n = 2) or in patient health information ( n = 3). In addition, one initiating organisation was not able to send a representative and therefore transferred its voting right for the whole workshop. Recommendations on dissemination In the topic area of dissemination, 13 recommendations were available for voting (three of which were in the “feasible” category). Two out of the 13 recommendations were rejected. The recommendations are shown in Table . The use of already existing structures for the dissemination of the PVGs was evaluated very positively. For example, they could be integrated into existing modules of training and continuing education curricula (on communication and evidence-based medicine for service providers (1.4). In addition, to use PVGs could be explicitly mentioned in the requirement catalogue of the oncological centres certified by the German Cancer Society; the current version of the catalogue only refers to patient information in general (1.5). The participants emphasized that this should not displace other high quality information materials. According to the participants, indexing the PVGs for search engine optimisation is very time-consuming because it is a complex technical process (1.6). The use of intuitive terminology on the cover page could already improve the search if necessary (2.4). The participants had a controversial discussion about the use of multilingual information materials such as flyers (1.7). In particular, the use of artificial intelligence was considered beneficial for translation into plain language. When providing information on the PVG in relevant scientific journals, it was assessed important to use free announcements and articles and no advertisement that has to be paid (1.8). Pointing out that congresses aimed at healthcare professionals are already adequately covered using fair stands for information, participants were in favour of presenting PVG at congresses for patients and patient representatives (1.9). The dissemination of the PVG via social media was rejected mainly for the perceived lack of resources (1.11). First, the establishment of structures for the collection of target group-specific media strategies was deemed necessary. The unsolicited sending of flyers and/or printed version of the PVG to relevant healthcare facilities was rejected with the argument that it would be a waste of resources (funds and material; 1.12). The future significance of digital health applications was discussed and partly doubted. Nevertheless, the reference to the PVG in existing digital health applications was evaluated positively (1.13). Recommendations on design and format In the topic area of design and format, seven recommendations were available for voting (2 of which were in the “feasible” category). Three out of the seven recommendations were rejected. The recommendations are shown in Table . Participants pointed out that product neutrality is sometimes difficult to ensure, especially in the case of photos as distinct from images (2.3). Because the term PVG is not intuitively understandable, intuitive terminology is to be added to the term PVG on the cover page (2.4). Participants controversially discussed proposals for intuitive terminology and pointed out that it should be assessed in advance which terms are understandable for patients. Three recommendations were rejected in view of the high effort that would be involved (2.5–2.7). Recommendations on (digital) links In the topic area of (digital) links, five recommendations were available for voting (2 of which were in the “feasible” category). None of the recommendations was rejected. The recommendations are shown in Table . Participants emphasised that recurring cross-references to the explanations of the grading of recommendations are feasible in PDF brochures but not in printed brochures (3.2). They discussed various ways to optimise existing links to target websites (3.4) such as verification on update, annual verification or, in perspective, automated verification. It was assumed that the amount of resources required to create a digital one-pager that lists, among other things, updates to PVG content would be high (3.5). In addition, it was noted that an acceleration of editorial processes is needed to include new content in the one-pager in a timely manner. Recommendations on digitalisation In the topic area of digitalisation, four recommendations were available for voting. One of the recommendations was rejected. The recommendations are shown in Table . The participants appreciated the transfer of the content to an app (4.1). There are already plans for implementation. The transfer of the PVG into a digital health application was rejected due to the efforts associated with the benefit assessment in the development of a new digital health application (4.4). In Germany, digital health applications can be prescribed if proven beneficial. With regard to the integration into already existing digital health applications (1.13), the recommendation was rejected as redundant. Linking the PVG to the electronic patient record that is accessible for patients and medical personnel was appreciated (4.3). When linking the PVG to the electronic patient record, some participants emphasised the right of ignorance, so that consent to display the PVG should first be given first. Participants very much welcomed a voice output to make the PVG available to disadvantaged groups of people with physical disabilities (4.2). In case of foreign languages or plain language, a voice output in different languages is needed. According to the participants, this is already feasible for English, but the technology still needs further development for other languages. Recommendations on up-to-dateness In the topic area of up-to-dateness, three recommendations were available for voting. One of the recommendations was rejected. The recommendations are shown in Table . If the underlying CPG is a living CPG, there should be a transition of the PVG to a living PVG (5.1). This requires a simplification of the editorial structures in the development of PVGs. A corresponding simplification is also necessary if there is no living CPG, but the PVG updating process is still to be optimised (5.1.1). Overall, participants again advocated for a general acceleration/optimization of editorial structures in order to integrate new/relevant content into the PVGs more quickly. The display of notifications on the phone must be set individually by the user. Since the organisations that produce PVGs have no influence on this, the recommendation on push notifications (5.2) was rejected. Recommendations on use of the PVG in collaboration between healthcare providers and patients In the topic area of use of the PVG in collaboration between healthcare providers and patients, three recommendations were available for voting (one of which was in the “feasible” category). All recommendations were approved. The recommendations are shown in Table . The distribution of the PVG to patients was considered useful. First, PVG should be offered by physicians (6.2), and the multidisciplinary team should offer it in the subsequent healthcare process (6.3). On the one hand, participants emphasised the improvement of the physician-patient relationship through the implementation of active and/or passive handovers by the physicians as required (6.2); on the other hand, it was stressed that knowledge of the existence of the PVG is a prerequisite for this. Active handover refers to the delivery of the printed PVG, which is briefly introduced in a conversation. Passive handing over refers to the simple handing over of the PVG without any further reference. The latter can be used particularly when physicians feel that an active handover would be overwhelming at this point. However, a passive handover requires an active handover at a later stage. This is separate from the continuous reminder by other health professionals recommended in Recommendation 6.3. The five-hour workshop took place on 24th April 2023 in Cologne, Germany. 23 people from 24 organisations participated in the discussion. Four representatives from four organisations had to cancel their participation at short notice for various reasons. Users of PVGs (patients ( n = 6), medical staff ( n = 4), and multipliers ( n = 2)) were included as well as creators ( n = 4) and initiators/funding organisations ( n = 3) of PVGs and organisations with methodological expertise in the development of clinical practice guidelines ( n = 2) or in patient health information ( n = 3). In addition, one initiating organisation was not able to send a representative and therefore transferred its voting right for the whole workshop. In the topic area of dissemination, 13 recommendations were available for voting (three of which were in the “feasible” category). Two out of the 13 recommendations were rejected. The recommendations are shown in Table . The use of already existing structures for the dissemination of the PVGs was evaluated very positively. For example, they could be integrated into existing modules of training and continuing education curricula (on communication and evidence-based medicine for service providers (1.4). In addition, to use PVGs could be explicitly mentioned in the requirement catalogue of the oncological centres certified by the German Cancer Society; the current version of the catalogue only refers to patient information in general (1.5). The participants emphasized that this should not displace other high quality information materials. According to the participants, indexing the PVGs for search engine optimisation is very time-consuming because it is a complex technical process (1.6). The use of intuitive terminology on the cover page could already improve the search if necessary (2.4). The participants had a controversial discussion about the use of multilingual information materials such as flyers (1.7). In particular, the use of artificial intelligence was considered beneficial for translation into plain language. When providing information on the PVG in relevant scientific journals, it was assessed important to use free announcements and articles and no advertisement that has to be paid (1.8). Pointing out that congresses aimed at healthcare professionals are already adequately covered using fair stands for information, participants were in favour of presenting PVG at congresses for patients and patient representatives (1.9). The dissemination of the PVG via social media was rejected mainly for the perceived lack of resources (1.11). First, the establishment of structures for the collection of target group-specific media strategies was deemed necessary. The unsolicited sending of flyers and/or printed version of the PVG to relevant healthcare facilities was rejected with the argument that it would be a waste of resources (funds and material; 1.12). The future significance of digital health applications was discussed and partly doubted. Nevertheless, the reference to the PVG in existing digital health applications was evaluated positively (1.13). In the topic area of design and format, seven recommendations were available for voting (2 of which were in the “feasible” category). Three out of the seven recommendations were rejected. The recommendations are shown in Table . Participants pointed out that product neutrality is sometimes difficult to ensure, especially in the case of photos as distinct from images (2.3). Because the term PVG is not intuitively understandable, intuitive terminology is to be added to the term PVG on the cover page (2.4). Participants controversially discussed proposals for intuitive terminology and pointed out that it should be assessed in advance which terms are understandable for patients. Three recommendations were rejected in view of the high effort that would be involved (2.5–2.7). In the topic area of (digital) links, five recommendations were available for voting (2 of which were in the “feasible” category). None of the recommendations was rejected. The recommendations are shown in Table . Participants emphasised that recurring cross-references to the explanations of the grading of recommendations are feasible in PDF brochures but not in printed brochures (3.2). They discussed various ways to optimise existing links to target websites (3.4) such as verification on update, annual verification or, in perspective, automated verification. It was assumed that the amount of resources required to create a digital one-pager that lists, among other things, updates to PVG content would be high (3.5). In addition, it was noted that an acceleration of editorial processes is needed to include new content in the one-pager in a timely manner. In the topic area of digitalisation, four recommendations were available for voting. One of the recommendations was rejected. The recommendations are shown in Table . The participants appreciated the transfer of the content to an app (4.1). There are already plans for implementation. The transfer of the PVG into a digital health application was rejected due to the efforts associated with the benefit assessment in the development of a new digital health application (4.4). In Germany, digital health applications can be prescribed if proven beneficial. With regard to the integration into already existing digital health applications (1.13), the recommendation was rejected as redundant. Linking the PVG to the electronic patient record that is accessible for patients and medical personnel was appreciated (4.3). When linking the PVG to the electronic patient record, some participants emphasised the right of ignorance, so that consent to display the PVG should first be given first. Participants very much welcomed a voice output to make the PVG available to disadvantaged groups of people with physical disabilities (4.2). In case of foreign languages or plain language, a voice output in different languages is needed. According to the participants, this is already feasible for English, but the technology still needs further development for other languages. In the topic area of up-to-dateness, three recommendations were available for voting. One of the recommendations was rejected. The recommendations are shown in Table . If the underlying CPG is a living CPG, there should be a transition of the PVG to a living PVG (5.1). This requires a simplification of the editorial structures in the development of PVGs. A corresponding simplification is also necessary if there is no living CPG, but the PVG updating process is still to be optimised (5.1.1). Overall, participants again advocated for a general acceleration/optimization of editorial structures in order to integrate new/relevant content into the PVGs more quickly. The display of notifications on the phone must be set individually by the user. Since the organisations that produce PVGs have no influence on this, the recommendation on push notifications (5.2) was rejected. In the topic area of use of the PVG in collaboration between healthcare providers and patients, three recommendations were available for voting (one of which was in the “feasible” category). All recommendations were approved. The recommendations are shown in Table . The distribution of the PVG to patients was considered useful. First, PVG should be offered by physicians (6.2), and the multidisciplinary team should offer it in the subsequent healthcare process (6.3). On the one hand, participants emphasised the improvement of the physician-patient relationship through the implementation of active and/or passive handovers by the physicians as required (6.2); on the other hand, it was stressed that knowledge of the existence of the PVG is a prerequisite for this. Active handover refers to the delivery of the printed PVG, which is briefly introduced in a conversation. Passive handing over refers to the simple handing over of the PVG without any further reference. The latter can be used particularly when physicians feel that an active handover would be overwhelming at this point. However, a passive handover requires an active handover at a later stage. This is separate from the continuous reminder by other health professionals recommended in Recommendation 6.3. All in all, 35 recommendations were part of the voting procedure. Of these, 28 recommendations were approved. The recommendations referred to the topics dissemination, design and format, (digital) links, digitalisation, up-to-dateness, and use of the PVG in collaboration between healthcare providers and patients. The recommendations address different stakeholders such as PVG creators, but also healthcare professionals. Many recommendations refer to the dissemination of PVGs. This is particularly relevant in view of the insufficient awareness observed during the project. A number of participants (patients and healthcare providers) in the qualitative part of study indicated that they appreciated the concept of PVGs but had no awareness about it beforehand . Considering that many participants perceived the PVG as a helpful tool for informed decision-making (data not yet published), the aim should be to increase awareness about the PVGs and their use. To this regard, we provide several recommendations with different approaches addressing patients or healthcare professionals. Different contexts such as training, certification and information policies addressing healthcare professionals are included. Furthermore, providing patient information in healthcare facilities, self-help groups and measures to increase the visibility of PVG in the context of the internet are addressed. A review found that many patients search the Internet for health information and that they most often use a search engine as a starting point . Accordingly, it is important that PVGs can be found well when searching the internet via a search engine. However, many recommendations refer to the integration of PVG into already existing structures. Even though the recommendations provide a practical basis due to the involvement of divergent stakeholders, their implementation in practice is highly important. In this context, further research is needed. For example, a training session to teach physicians on how to integrate PVGs into the doctor-patient conversation could be developed. As a positive side effect, this would also increase doctors’ awareness of the PVGs. Possible obstacles to dealing with the PVGs during the doctor-patient-conversation in detail could be time restrictions experienced by the doctors or the patient-physician relationship. Furthermore, decreased cognitive capacities because of anxiety or stress can also play a role in the ability to perceive information . The timing of the handover might play an important role in this context. According to a qualitative study oncological patients require relevant health information from a very early start . The fact that this time is associated with a high level of emotion, particularly in the case of oncological diseases, can be a challenge in terms of handover. Appropriate training for service providers regarding the PVG could also provide assistance in this regard. After its implementation, such a training session could be integrated in the certification system of the German Cancer Society. This would lead to a higher awareness and use of the PVG on the physicians’ as well as the patients’ side. The implementation of some recommendations would also enable people to use PVG who were previously unable or only partially able to do so due to various circumstances. In our recommendations, we refer to non-native speakers as well as people with impaired vision. There are further groups of people such as patients with intellectual disabilities who may not be addressed by the PVGs. Because our project did not provide any results on this, they are not mentioned in the recommendations. Nevertheless, we want to emphasize the importance of addressing all target groups. This is certainly very challenging due to highly divergent needs. Information needs differ in terms of what information is needed, in what form and in what level of detail. A survey found that the Internet is the most frequently sought source of health information by both men and women . However, the frequency of searching on the Internet also depends on underlying sociodemographic factors such as the socioeconomic status. In the course of a systematic review, it was found that information needs (of patients, relatives and the general population) vary in type and scope . Beyond topics such as treatment, diagnosis, prevention and health promotion, aetiology, and prognosis, where information needs are high, information on topics such as rehabilitation and impact on social life was in demand less frequently. In this context, it would be helpful to individualise PVGs to a greater extent. The user test of a PVG also showed that some needs are so heterogeneous that individualisation, if possible, should be attempted . If all potential information needs were addressed in a PVG, the scope would be far too large for many patients. This trade-off is a challenge that might be addressed by (digital) links or the use of different formats. The integration of PGVs in different formats such as apps or the electronic patient record could enable a staggered integration of the content. At least at present, the implementation of a part of the recommendations is only possible for some of the formats offered. One example is the recurrent explanation of the grading of recommendations. This is easy to implement for PVGs in PDF format, but difficult in case of a printed PVG. However, printed PVGs continue to be very popular even though there might be differences between patient populations (e.g. age ). Especially in the context of digitalisation, there are likely to be some opportunities for further development of PVGs in the future. Some of them are directly taken into account in our recommendations; others have been discussed in the context of the implementation of individual recommendations. This was the case, for example, with the recommendation to optimize existing links to target internet sites. Participants mentioned that this could be done automatically in the future. Strength and limitations Some limitations have to be considered when interpreting our results. Even though the list of participants is not exhaustive, the big players in the German field of PVGs and patient information took part. Due to time constraints, the recommendations from the “feasible” category could not be discussed in detail during the World Café. In the course of the voting procedure, there was restricted additional time for discussion, if necessary. Nevertheless, it became apparent that there was no need for discussion for most of the recommendations from the feasible category. However, for some of the other recommendations more discussion time would have been helpful. On the other hand, we assume that the time restriction could increase the participation rate by allowing the participants to arrive and depart on the day itself. The World Café was chosen to enable all stakeholder to participate in the discussion and to share their point of view. On the other hand, this left less time for discussion in the plenary session. Additionally, some participants of the workshop gave the feedback, that they would have preferred an anonymous online voting procedure. Since there were also some delays in counting, especially when people changed their minds or needed longer time to think about it, the project team would use online voting procedure in the future. In the course of the discussion, some participants expressed that they did not want to prescribe to specific addressees (e.g. German Cancer Aid) what they must do. The moderator then clarified that these were only recommendations and not obligations. Nevertheless, it cannot be ruled out that this misinterpretation may have influenced the voting behaviour of some participants. Our project referred to oncological PVG only, therefore the majority of recommendations can only be applied to oncological PVG. This becomes clear, for example, in the recommendation 1.5 (integration of the PVG into the certification system of the German Cancer Society) or 1.10 (Prominent positioning of the PVG on the German Cancer Aid website) since they are specifically targeted at relevant stakeholders in the field of oncology. However, a number of the recommendations without named addressees also clearly belong to the field of oncological PVG. One example is recommendation 2.1 including the clearer presentation of the medical recommendation by using bold front. For oncological PVG, unlike other PVG, an italicized font has been used to date. However, this was described by some participants as not striking enough. Other recommendations may also apply to non-oncological PVG. Since the concept of PVGs was not well known, it can be assumed that this is a fundamental circumstance and not exclusively related to the field of oncology. Accordingly, it should be investigated to what extent measures to disseminate PVG, for example, can be implemented beyond oncology. Considering all this, the project achieved practical recommendations under consideration of various perspectives. This can help to improve use and dissemination of (oncological) PVGs in Germany. Some limitations have to be considered when interpreting our results. Even though the list of participants is not exhaustive, the big players in the German field of PVGs and patient information took part. Due to time constraints, the recommendations from the “feasible” category could not be discussed in detail during the World Café. In the course of the voting procedure, there was restricted additional time for discussion, if necessary. Nevertheless, it became apparent that there was no need for discussion for most of the recommendations from the feasible category. However, for some of the other recommendations more discussion time would have been helpful. On the other hand, we assume that the time restriction could increase the participation rate by allowing the participants to arrive and depart on the day itself. The World Café was chosen to enable all stakeholder to participate in the discussion and to share their point of view. On the other hand, this left less time for discussion in the plenary session. Additionally, some participants of the workshop gave the feedback, that they would have preferred an anonymous online voting procedure. Since there were also some delays in counting, especially when people changed their minds or needed longer time to think about it, the project team would use online voting procedure in the future. In the course of the discussion, some participants expressed that they did not want to prescribe to specific addressees (e.g. German Cancer Aid) what they must do. The moderator then clarified that these were only recommendations and not obligations. Nevertheless, it cannot be ruled out that this misinterpretation may have influenced the voting behaviour of some participants. Our project referred to oncological PVG only, therefore the majority of recommendations can only be applied to oncological PVG. This becomes clear, for example, in the recommendation 1.5 (integration of the PVG into the certification system of the German Cancer Society) or 1.10 (Prominent positioning of the PVG on the German Cancer Aid website) since they are specifically targeted at relevant stakeholders in the field of oncology. However, a number of the recommendations without named addressees also clearly belong to the field of oncological PVG. One example is recommendation 2.1 including the clearer presentation of the medical recommendation by using bold front. For oncological PVG, unlike other PVG, an italicized font has been used to date. However, this was described by some participants as not striking enough. Other recommendations may also apply to non-oncological PVG. Since the concept of PVGs was not well known, it can be assumed that this is a fundamental circumstance and not exclusively related to the field of oncology. Accordingly, it should be investigated to what extent measures to disseminate PVG, for example, can be implemented beyond oncology. Considering all this, the project achieved practical recommendations under consideration of various perspectives. This can help to improve use and dissemination of (oncological) PVGs in Germany. Overall, 35 recommendations were part of the voting procedure. Of these, 28 recommendations were approved. The recommendations referred to the topics dissemination, design and format, (digital) links, digitalisation, up-to-dateness, and use of the PVG in collaboration between healthcare providers and patients. The practical recommendations consider various perspectives and can help to improve use and dissemination of (oncological) PVGs in Germany. |
Immunohistochemical study of histone protein 3 modification in pediatric osteosarcoma identifies reduced H3K27me3 as a marker of poor treatment response | 5e51d514-79b8-4c7f-b22d-14234ac94d99 | 11581320 | Anatomy[mh] | Osteosarcoma (OS) is the most common primary malignant bone tumor in the pediatric age group, and has a bimodal distribution with a second peak in older individuals . Specifically within the pediatric and young adult population, OS peaks in the teenage years . In pediatric and young adults <24 years of age, the incidence ranges from 2–5 per million per year in the USA . Clinical presentation related to primary tumor sites may include pain, swelling or mass, constitutional symptoms and/or pathologic fracture. Historically, OS has been treated after an initial diagnostic biopsy with a standard chemotherapy protocol (MAP; methotrexate, doxorubicin, and cisplatin), and the 5-year survival rate has remained suboptimal compared to other pediatric cancers, with survival being 60–70% for those with localized disease . Unfortunately, up to 20% of patients may have metastatic disease the time of diagnosis, for which 5-year survival outcomes remains dismal at <30% . Classification of OS is typically made according to the World Health Organization (WHO), which identifies different types of OS based on location, radiology, and pathologic findings . Most OS are considered conventional high grade tumors that are intra-osseus and often break through the cortex, alter the periosteum, and invade into the adjacent soft tissues. Conventional OS can demonstrate different morphologic appearances and matrix components, and while many pathologists provide a description of the morphologic appearances within the diagnosis of conventional OS, the significance of this morphologic assessment is unclear. A recent study suggested that the telangiectatic and unspecified types of conventional OS were favorable, while the chondroblastic subtype was less favorable . The classic description for the underlying chromosomal instability seen in OS is chromothripsis, which is seen in 77% of OS, with 61% of these showing complex and haphazard copy number variations across at least 5 chromosomes . Two genes are frequently mutated in OS, and both have predisposition syndromes leading to OS, namely Li-Fraumeni ( TP53 mutation; 3–5% of OS have a germline mutation in TP53 , while somatic mutations are seen in up to 95% of OS ) and hereditary retinoblastoma ( RB1 mutation; <1% of OS have a germline mutation in RB1 , while somatic mutations are seen in up to 64% of OS ). Interestingly, a recent study found that in fact up to one-quarter of 1,244 unselected OS patients demonstrated an underlying germline cancer predisposition variant, including genes not previously linked to OS . A recent review looking at the molecular mechanisms in OS highlights a number of genes and cellular mechanisms that may be critical and oncogenesis and potentially therapeutically targetable, but limited work has been done in the epigenetics of OS . Epigenetics involves complex processes of regulating gene expression and activity without directly altering the DNA sequence and may occur in a different pattern from cell to cell and tissue to tissue. This regulation may include post-translational histone modifications (PTHM) which allows for a particular gene to be expressed or silenced without an underlying change to the DNA sequence. Histones are proteins that are bundled in DNA to form nucleosomes, a core structure of chromatin, resulting in a ‘beads on a string’ structure. The N-terminal tails that protrude from these proteins can be remodeled with covalently attached residues such as methyl or phosphoryl groups . Between these residues, molecular interactions configure euchromatin, where the DNA is readable and in the ‘beads on a string’ form, and heterochromatin, where DNA is unreadable and in the ‘30-nm fiber’ form. These chromatin remodeling mechanisms give modulatory control of gene expression within a cell, by modifying access for RNA polymerases to transcribe DNA. Despite the classical description in oncogenesis as a group of diseases born of genetic aberration, epigenetic perturbations, such as those due to chromatin modifiers, DNA methylation, and histone modifications, have been found to underlie several malignancies . The value of histone research translating to diagnostic pathology has been recently demonstrated, notably in cases of malignant peripheral nerve sheath tumors (MPNST). Cleven et al., showed that the diagnostic accuracy for MPNSTs arising from a neurofibroma could be improved by an immunohistochemistry (IHC) stain for the tri-methylation of the 27th residue (lysine) on the histone protein 3 (H3K27me3) . Other studies have found that other methylation patterns in MPNST, such as dimethyl loss at H3K27, may be more specific when using immunohistochemistry . Since its role in MPNST, H3K27me3 has been implicated in a number of different tumors for diagnostic and prognostic purpose, often associated with a poor prognosis or aggressive behavior, including rhabdomyosarcoma , chordoma , central nervous system (CNS) tumors , amongst others. There are other markers of post-translational modification of H3, which are generally associated with DNA replication in the S-phase of the cell cycle . A select number of these residue alterations have been reported in the literature . The function of these specific PTHMs are usually context-specific but have been studied in depth for their abilities to form areas of closed or open chromatin at varying times in development, throughout the cell cycle, or by modulating the actions of other epigenetic factors. Therefore, these dynamic controlling mechanisms make PTHMs particularly interesting as developmental or cell cycle dysregulation are both hallmarks of malignancies. In this current study, our primary aim was to survey selected epigenetic markers in a series of diagnostic OS biopsies to determine if any particular patterns of open or closed chromatin may predict neoadjuvant necrosis response, overall survival and other important clinical outcomes such as post-therapy response and event free survival. Our secondary aim was to determine if there were any specific changes seen when the diagnostic biopsy was compared to a matched metastatic or recurrent biopsy. 2.1 Sample Identification This project was approved by the British Columbia (BC) Children’s and Women’s Research Ethics Board with a waiver of consent. Data was de-identified and no personal health information or identifying data was collected. The pathology laboratory information system (LIS) was queried using the term “osteosarcoma” and “osteogenic sarcoma” between January 1, 2002 and December 31, 2021, which identified primary diagnostic biopsies and metastatic and recurrent specimens. Clinical information was collected from the LIS and electronic clinical records from January 1, 2002 and December 31, 2021, which included demographical data, disease characteristics and investigations, therapy, and disease events, last known medical follow up, and survival status is known. Patient data was accessed from July 28, 2022 to July 7, 2023. Cases were excluded from analysis if the formalin-fixed, paraffin-embedded (FFPE) block(s) were not available. 2.2 Tissue microarray and immunohistochemistry Tissue microarrays (TMAs) were built using the Manual Tissue Arrayer (Beecher Instruments Incorporated, Sun Prairie, WI). Hematoxylin and eosin (H&E) slides from the available FFPE blocks were reviewed and areas selected for TMA inclusion were selected that were most representative of the tumor and with no necrosis. In samples that showed a variety of morphologies, areas of morphologic heterogeneity were intentionally selected. TMAs were constructed with duplicate 1.0 mm cores for each case in each TMA, and two TMAs were constructed for the primary and event tumors each (four total TMAs). Non-malignant tissues were interspersed in the TMA, including normal cortical bone, bone marrow, osteochondroma, chondroblastoma, and osteoblastoma. After reviewing the TMA H&E slides, the TMAs were stained using the following IHC antibodies: H3K4me3 (Abcam, ab8580; 1:1000), H3K9me3 (Abcam, ab176916; 1:500), H3K27me2 (Abcam, ab24684; 1:200) H3K27me3 (Cell Signaling Technology, C36B11 9733; 1:200), H3S28phos (Abcam, ab32388; 1:10,000), and H3S10T11phos (Abcam, ab32107; 1:100). Antibodies were optimized using the known positive and negative controls listed by the manufacturer. Scoring and interpretation of the TMAs was done using QuPath under the guidance of the study pathologist (JWB). Relative positivity values used in the comparison of primary tumors and their paired first events (metastasis or recurrence) were calculated using the default positive cell detection function. We used the highest interpretable cores to analyze our primary as well as their paired first event tumors. This follows the default scoring system for many studies involving TMAs . Staining in known normal tissues were excluded from analysis. Each antibody was tested against two cores from each tumor block to address tumor heterogeneity. The scoring system used a dichotomous scale of low (<25%) and high (≥25%) tumor nuclei expression . Dropped cores from the TMA or cores that did not include tumor were excluded from analysis. 2.3 Statistical analysis Outcomes including post-chemotherapeutic response at the time of definitive resection, represented by tumor necrosis percentage, were reformatted to binary values, being either a good (≥90% necrosis) or poor (<90% necrosis) response to treatment. TMA scoring categories were also formatted to binary values of low expression (<25%) and high expression (≥25%) . Pairwise deletion was used in these cases of missing data, whether it be unavailable due to missing information in the clinical charts, or randomly lost data (i.e. dropped TMA cores). Data were summarized through descriptive statistics including frequencies and percentages for categorical variables. χ 2 contingency analyses were performed and p-values of <0.05 were deemed to be statistically significant for this study. Fisher’s exact test was utilized in cases where the expected counts violated the assumptions of the analysis. The relationship between survival and histone modification was summarized Kaplan-Meier survival curves . A paired t-test was used for analysis between the primary and the recurrent or metastatic samples . Statistical analysis was performed using R statistical software (R Core team 2023, Vienna, Austria). This project was approved by the British Columbia (BC) Children’s and Women’s Research Ethics Board with a waiver of consent. Data was de-identified and no personal health information or identifying data was collected. The pathology laboratory information system (LIS) was queried using the term “osteosarcoma” and “osteogenic sarcoma” between January 1, 2002 and December 31, 2021, which identified primary diagnostic biopsies and metastatic and recurrent specimens. Clinical information was collected from the LIS and electronic clinical records from January 1, 2002 and December 31, 2021, which included demographical data, disease characteristics and investigations, therapy, and disease events, last known medical follow up, and survival status is known. Patient data was accessed from July 28, 2022 to July 7, 2023. Cases were excluded from analysis if the formalin-fixed, paraffin-embedded (FFPE) block(s) were not available. Tissue microarrays (TMAs) were built using the Manual Tissue Arrayer (Beecher Instruments Incorporated, Sun Prairie, WI). Hematoxylin and eosin (H&E) slides from the available FFPE blocks were reviewed and areas selected for TMA inclusion were selected that were most representative of the tumor and with no necrosis. In samples that showed a variety of morphologies, areas of morphologic heterogeneity were intentionally selected. TMAs were constructed with duplicate 1.0 mm cores for each case in each TMA, and two TMAs were constructed for the primary and event tumors each (four total TMAs). Non-malignant tissues were interspersed in the TMA, including normal cortical bone, bone marrow, osteochondroma, chondroblastoma, and osteoblastoma. After reviewing the TMA H&E slides, the TMAs were stained using the following IHC antibodies: H3K4me3 (Abcam, ab8580; 1:1000), H3K9me3 (Abcam, ab176916; 1:500), H3K27me2 (Abcam, ab24684; 1:200) H3K27me3 (Cell Signaling Technology, C36B11 9733; 1:200), H3S28phos (Abcam, ab32388; 1:10,000), and H3S10T11phos (Abcam, ab32107; 1:100). Antibodies were optimized using the known positive and negative controls listed by the manufacturer. Scoring and interpretation of the TMAs was done using QuPath under the guidance of the study pathologist (JWB). Relative positivity values used in the comparison of primary tumors and their paired first events (metastasis or recurrence) were calculated using the default positive cell detection function. We used the highest interpretable cores to analyze our primary as well as their paired first event tumors. This follows the default scoring system for many studies involving TMAs . Staining in known normal tissues were excluded from analysis. Each antibody was tested against two cores from each tumor block to address tumor heterogeneity. The scoring system used a dichotomous scale of low (<25%) and high (≥25%) tumor nuclei expression . Dropped cores from the TMA or cores that did not include tumor were excluded from analysis. Outcomes including post-chemotherapeutic response at the time of definitive resection, represented by tumor necrosis percentage, were reformatted to binary values, being either a good (≥90% necrosis) or poor (<90% necrosis) response to treatment. TMA scoring categories were also formatted to binary values of low expression (<25%) and high expression (≥25%) . Pairwise deletion was used in these cases of missing data, whether it be unavailable due to missing information in the clinical charts, or randomly lost data (i.e. dropped TMA cores). Data were summarized through descriptive statistics including frequencies and percentages for categorical variables. χ 2 contingency analyses were performed and p-values of <0.05 were deemed to be statistically significant for this study. Fisher’s exact test was utilized in cases where the expected counts violated the assumptions of the analysis. The relationship between survival and histone modification was summarized Kaplan-Meier survival curves . A paired t-test was used for analysis between the primary and the recurrent or metastatic samples . Statistical analysis was performed using R statistical software (R Core team 2023, Vienna, Austria). 3.1 Patient Information All patients were aged 19 or younger at the time of their surgical procedure. There were 58 available diagnostic OS biopsies identified and 54 metastatic or recurrent samples from 20 of the original 58 diagnostic biopsy patients. Amongst those 20 patients with metastatic or recurrent tumors, the median number of samples per patient was 1.5 (range of 1 to 5). Overall survival was determined by ‘enrolment’ at the date of diagnosis up until either an event or the last known healthcare follow-up (censored). Among 58 included patients, 60% had localized disease only, 57% were male, and the median age at primary diagnosis was 13.83 years old (range of 4.43 to 17.84 years old), similar to published literature . Those with localized disease had superior overall survival than those with metastatic disease, consistent with that expected in the literature . The histologic morphologies seen in our patient cohort also follow the expected distribution with the osteoblastic morphology being the most common in 70.6% of patients . 3.2 Loss of H3K27me3 staining is significant in diagnostic tumors Amongst the samples in the diagnostic tumor TMA, we found that loss of H3K27me3 was associated with a poor neoadjuvant response (p = 0.0224) . Of the 26 poor responders, 22 had loss of the H3K27me3 nuclear stain within tumor cells in diagnostic biopsies. However, the pattern of H3K27me3 was not predictive of overall survival or development of metastatic or recurrent disease. 3.3 Loss of H3K4me3 and retention of H3K9me3 nuclear staining observed in all primary and subsequent event tumor nuclei Interestingly, H3K4me3 immunoexpression was found to be low to nearly absent (<25% of tumor cells) in all tumor cores upon TMA analysis . On the contrary, H3K9me3 nuclear expression (≥25% of tumor cells) was observed in almost all TMA cores at the time of analysis . We found that these patterns remained consistent in the paired metastases and recurrences on our metastatic and recurrent TMA sets. 3.4 H3K27me2 and H3S28phos immunoexpression is not associated with measured clinical outcomes in primary cases Observed counts of H3K27me2 and H3S28phos were not predictive of survival outcomes, post-therapy response, or EFS in our cohort. . 3.5 First event tumors have near complete reduction of H3S10T11phos expression relative to their primary tumors In comparing the expression of H3S10T11phos between primary and paired metastases and recurrence, we found that the latter have significantly lower expression (p = 0.011) . We noted other interesting patterns, such as reduced expression in H3K27me3 compared to primary tumors , but these results were not significant. Despite this pattern in comparing primary and first event tumors, H3S10T11phos expression on diagnostic biopsy was not found to be predictive of overall survival or necrosis response. Although not statistically significant (p = 0.096), a trend was seen for a favorable prognosis when expression levels were reduced . The results of global PTHM changes were not significant for H3K27me2 or H3S28phos . All patients were aged 19 or younger at the time of their surgical procedure. There were 58 available diagnostic OS biopsies identified and 54 metastatic or recurrent samples from 20 of the original 58 diagnostic biopsy patients. Amongst those 20 patients with metastatic or recurrent tumors, the median number of samples per patient was 1.5 (range of 1 to 5). Overall survival was determined by ‘enrolment’ at the date of diagnosis up until either an event or the last known healthcare follow-up (censored). Among 58 included patients, 60% had localized disease only, 57% were male, and the median age at primary diagnosis was 13.83 years old (range of 4.43 to 17.84 years old), similar to published literature . Those with localized disease had superior overall survival than those with metastatic disease, consistent with that expected in the literature . The histologic morphologies seen in our patient cohort also follow the expected distribution with the osteoblastic morphology being the most common in 70.6% of patients . Amongst the samples in the diagnostic tumor TMA, we found that loss of H3K27me3 was associated with a poor neoadjuvant response (p = 0.0224) . Of the 26 poor responders, 22 had loss of the H3K27me3 nuclear stain within tumor cells in diagnostic biopsies. However, the pattern of H3K27me3 was not predictive of overall survival or development of metastatic or recurrent disease. Interestingly, H3K4me3 immunoexpression was found to be low to nearly absent (<25% of tumor cells) in all tumor cores upon TMA analysis . On the contrary, H3K9me3 nuclear expression (≥25% of tumor cells) was observed in almost all TMA cores at the time of analysis . We found that these patterns remained consistent in the paired metastases and recurrences on our metastatic and recurrent TMA sets. Observed counts of H3K27me2 and H3S28phos were not predictive of survival outcomes, post-therapy response, or EFS in our cohort. . In comparing the expression of H3S10T11phos between primary and paired metastases and recurrence, we found that the latter have significantly lower expression (p = 0.011) . We noted other interesting patterns, such as reduced expression in H3K27me3 compared to primary tumors , but these results were not significant. Despite this pattern in comparing primary and first event tumors, H3S10T11phos expression on diagnostic biopsy was not found to be predictive of overall survival or necrosis response. Although not statistically significant (p = 0.096), a trend was seen for a favorable prognosis when expression levels were reduced . The results of global PTHM changes were not significant for H3K27me2 or H3S28phos . We report the first study that provides an immunohistochemical survey of epigenetic H3 alterations in diagnostic OS biopsies, with an enhanced analysis by evaluating the changes seen in available paired metastases or recurrences. In our study, we found that reduced expression of H3K27me3 is associated with poor response to neoadjuvant therapy at the time of definitive resection. Previous research has demonstrated a relationship between cisplatin and H3K27me3, suggesting that resistance to cisplatin was inversely related to H3K27me3 expression in in vitro and in vivo models . Our results corroborate these findings as the loss of H3K27me3 may predict poor response to neoadjuvant therapy, which routinely includes cisplatin. Therefore, further studies may seek to explore the possibility of including more intensive or novel therapies at the outset of treatment for localized OS patients who demonstrate loss of H3K27me3 on their diagnostic biopsies. Although the literature typically associates loss of H3K27me3 with a poor outcome, some emerging literature has found that this may not always be the case. Two studies in the CNS literature have identified that loss of H3K27me3 may actually portend a better outcome in a specific subset of gliomas . This alternative finding must be considered that other factors may be affecting outcomes in conjunction with this epigenetic pattern. Limited studies elsewhere have explored H3K27me3 in OS. One of these few studies found that 3 of 19 OS cases demonstrated loss of this marker by IHC, and did not show molecular or other features to suggest that these H3K27me3-negative OS cases were not MPNST with heterologous elements . The remaining studies of H3K27me3 in OS generally focus on therapeutic considerations, such as the role of methionine metabolism . Another relationship we identified was that H3S10T11phos exhibits low or absent IHC expression at the time of first relapse, often when positive in the preceding primary biopsy. This phenomenon is interesting as there was no statistically significant difference in survival for those with low or high expression of H3S10T11phos at the time of diagnostic biopsy. The expression of H3S10 phosphorylation has been shown to be associated with a more aggressive form of gastric cancer with loco-regional recurrence , and it is believe that the STAT3 and mitogen-and stress-activated protein kinase 1 (MSK1) are involved in phosphorylating H3S10 in these aggressive forms of gastric cancer . STAT3 has been implicated in a number of cancers, including OS , where overexpression is often seen in cases with worse outcomes, with an exception being breast cancer. MSK1 is part of the MAPK pathway, being regulated by p38 MAPK . However, the role of MSK1 is not well understood in OS in the context of overall MAPK/ERK pathway . Evaluation of these two H3S10 phosphorylation regulators may be high-value targets, biologically and therapeutically, to understand our observation of marked reduction in H3S10T11phos IHC expression in relapse specimens. Comparing pathway alterations at the transcriptome or proteome level between diagnostic and recurrence biopsies may further our understanding of developing metastatic potential in OS. H3S10phos has been observed to be co-expressed with H3K9me3, a marker of heterochromatin, and downregulates it prior to mitosis likely to allow for DNA replication . This is quite interesting, as we found that all cases that subsequently lost H3S10T11phos retained their H3K9me3 staining. On the other hand, H3T11phos has been shown to be a marker that optimizes demethylation of tri-methylated residues . For example, H3K27me3 is a marker of closed chromatin, so demethylation may contribute to increased gene expression of previously closed regions. The combination of the two markers, and perhaps overexpression, may contribute to a cascade where large regions of heterochromatic DNA are made accessible. This dysregulation of gene expression could be identified as a major factor in the proliferation of OS. However, much remains to be discussed, which cannot be done without more precise techniques that can highlight which regions of DNA are being affected. Moreover, these results also highlight the interplay and contextual factors that occur between PTHM markers which are yet to be explored in OS. Nonetheless, this evidence provides new information on the fluidity of the epigenetic landscape of OS with significant pattern changes between primary tumors and their metastases or recurrences. Limitations of this study are notable that this was a pilot study using a small and heterogenous cohort size of convenience. However, this method does allow for our study to identify potential candidate IHC markers of interest that may be better assessed in properly-powered studies. These future studies may also provide the opportunity to consider performing receiver operating characteristic curves to determine the optimal performance of these epigenetic markers to predict outcomes in OS. Given that OS samples are often hard and require decalcification, we were required to work up our antibodies under both non-decal and decal conditions. To do this, we utilized known normal and abnormal staining tissues as identified by the manufacturer, and tested these antibodies on these known tissues under both routine FFPE preparation and FFPE-decal preparations. Another limitation is that our study used a TMA-based assay, which is not able to assess for possible intra-tumoral heterogeneity. This may prove particularly true for phenomenon such as the loss of H3S10T11 in metastatic samples but no predictive power of metastasis in the primary biopsy. It is possible that the metastatic foci originate from foci outside of those sampled for the TMA. Epigenetic regulation is known have a spatial heterogeneity in many tumors , and this could be magnified as a source of sampling error in an OS TMA with known epigenetic and histologic heterogeneity . Given the findings in this study, future studies may benefit from both bulk and spatially resolved analytics, including transcriptomic and proteomic studies to determine if epigenetic regulation can be used in a prognostic or predictive fashion. In summary, we demonstrate that OS demonstrates intra and inter-tumoral heterogeneity by immunohistochemistry for a number of PTHM markers on diagnostic biopsies at the H3 residue. Of these, H3K27me3 shows promise as a marker of neoadjuvant response regardless of the histomorphology observed. On the other hand, we demonstrate that H3S10T11 appears to show a dramatic drop in expression level in metastatic and recurrent specimens which may indicate potential underlying biologic changes facilitate the development of metastatic or recurrent disease. S1 Fig Good responder histogram. Histogram showing the distribution of IHC expression for good responders (≥90% tumor necrosis on resection). (TIF) S2 Fig Poor responder histogram. Histogram showing the distribution of IHC expression for poor responders (<90% tumor necrosis on resection). (TIF) S3 Fig Deceased histogram. Histogram showing the distribution of IHC expression for patients and their primary tumor that died of their disease. (TIF) S4 Fig Paired event box. Box plot showing the expression of each IHC marker between the primary and 1 st relapse. (TIF) S1 File Minimal dataset. This file contains the minimum dataset, including the clinicopathologic data and computed scores for each IHC marker. (XLSX) |
Proteomic and phosphoproteomic profilings reveal distinct cellular responses during | 9faff89f-8d03-4c37-a874-53bd8a6fa45d | 11849505 | Biochemistry[mh] | Tilapinevirus tilapiae , or Tilapia lake virus (TiLV), a single-stranded RNA virus belonging to the Amnoonviridae family , poses a significant threat to the global tilapia aquaculture industry, as it affects 18 countries across four continents . The virus contains 10 genomic segments which encode 14 functional proteins . In tilapia, infection by TiLV causes high mortality rates of up to 90% across all life stages . Moribund fish commonly show external lesions, such as skin erosions and hemorrhage, anemia, and gill necrosis , as well as internal lesions, including enlargement of the hepatopancreas, spleen and kidney necrosis, gut dysbiosis, and neuroinflammation . Despite its remarkable economic impact, the molecular mechanisms underlying host–virus interactions during TiLV infection remain poorly understood. Current research on the pathogenesis of TiLV infection is primarily focused on the viral replication mechanisms, host immune responses, and subcellular processes that occur during viral infection . For example, previous studies demonstrated the presence of viral RNA in various tissues, which indicated the multi-tissue tropisms of TiLV. Likewise, transcriptomic analysis of the livers of TiLV-infected tilapia showed potential pathways involved in viral replication and the suppression of host immune responses . However, in-depth studies focusing on the mechanisms underlying protein and phosphoprotein alterations in virus-infected cells remain limited. Mass spectrometry-based proteomic and phosphoproteomic analyses have emerged as powerful tools for identifying and quantifying proteins and their phosphorylation states within infected hosts . Specifically, phosphoproteomic studies are valuable for understanding cellular responses to viral infections during the early stages , as these changes occur more rapidly than those in the proteome . Previously, both proteomic and phosphoproteomic techniques have been successfully applied to investigate host–pathogen interactions during emerging viral infections in animals and humans, such as the African swine fever virus and severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) . However, the application of these techniques in the context of TiLV infection has not been explored. In fish, proteomics has provided valuable information on fish immunity and host cellular responses to environmental stressors, such as osmotic stress , bacterial and viral infections . Specifically, a proteomic study of mandarin fish brain cells infected with infectious spleen and kidney necrosis virus demonstrated alterations in glucose metabolism and the promotion of apoptosis and autophagy pathways . Additionally, proteomic analysis has been instrumental in identifying novel biomarkers for cardiomyopathy syndrome and Aeromonas infection in Atlantic salmon . Based on these techniques and relevant information, in this study, we utilized proteomic and phosphoproteomic analyses to investigate the dynamics changes between the host and TiLV during the early stages of infection in two piscine cell lines, namely, E-11 and RHTiB, which has been widely used for studying TiLV entry, replication, subcellular functions, and intracellular signaling . The functional analyses yielded by the proteomic analysis offer comprehensive insights into these host–virus interactions and may contribute to a better understanding of the pathophysiology of this virus. Portions of this text were previously published as part of a preprint ( https://doi.org/10.21203/rs.3.rs-4730524/v1 ). Cell culture and TiLV infection protocol We selected two piscine cell lines, E-11 cells, derived from the snakehead fry (purchased from the European Collection Authenticated Cell Cultures), and RHTiB, brain cells derived from red hybrid tilapia ( Oreochromis spp.) which has been isolated and established previously , to investigate the proteomic and phosphoproteomic profiles following TiLV infection. Both cell lines (passage 30–35) were cultured in Leibovitz’s L-15 medium, which contained 2% fetal bovine serum (Cat No. F2442; FBS, Sigma-Aldrich, Saint-Louis, MO, USA) at 25 °C without CO 2 . When the cells reached 80% confluency, the TiLV strain VETKU-TV08, which had been isolated from moribund fish in Pathumthani province, was diluted in L-15 without FBS and inoculated into the cells at a multiplicity of infection (MOI) of 0.1 and incubated for 10, 30 min, 1, 3, 6, and 24 h, respectively ( n = 3 each). At each time point, the cytopathic effect was monitored using an inverted microscope (CKX53; Olympus, Tokyo, Japan). Cells were harvested at each time point using scrapers (Cat No. 179707PK; Thermo Fisher Scientific, Rochester, NY, USA), and 50 µL of lysates were preserved at −20 °C for viral quantification. The remainder cells were ground with an equal volume of sodium dodecyl sulfate (SDS) (Cat No. D05888; Sigma-Aldrich, Saint-Louis, MO, USA)-lysis buffer (4% w/v in 100 mM Tris/HCl, pH 8.2) containing cocktails of protease inhibitors (complete EDTA-free, Cat No. 11 873 580 001; Roche, Mannheim, Germany) and phosphatase inhibitors (Cat No. 04 906 837 001; PhosStop, Roche, Mannheim, Germany) to prevent protein degradation and dephosphorylation, respectively. The mixture was then collected and centrifuged at 10,000× g at 4 °C for 15 min. The supernatant was transferred into two new tubes for total protein and phosphoprotein analyses. Each fraction was mixed with two volumes of cold acetone and incubated overnight at −20 °C. The mixture was then centrifuged at 10,000× g for 15 min, and the supernatant was discarded. The resulting pellet was dried and stored at −80 °C until further use. TiLV RNA quantification in cells and tissues The E-11 and RHTiB cells were assessed for TiLV at 10 min post-infection (mpi), 30 mpi, 1 h post-infection (hpi), 3, 6, and 24 hpi using a quantitative reverse transcription polymerase chain reaction (RT-qPCR). Initially, total RNA extraction from the collected cell lysates was conducted as described by . Briefly, the cell lysates were mixed with GENEzol reagent (Cat No. GZR200; Geneaid Biotech, Taipei, Taiwan) and chloroform (Cat No. 288306; Sigma-Aldrich, Saint-Louis, MO, USA) and centrifuged at 15,000× g at 4 °C. The supernatant was collected and treated with DNase I (Cat No. AM2224; Thermo Fisher Scientific, Carlsbad, CA, USA), followed by 2-propanol (Cat No. 109634; Merck, Darmstadt, Germany), and stored at −20 °C for 2 h. After thawing and centrifugation, the pellets were washed with ethanol and air-dried. The collected RNA was then reconstituted with RNase-free water and converted to complementary deoxyribonucleic acid (cDNA) using the ReverTra Ace cDNA synthesis kit (Cat No. FSQ-201; Toyobo, Osaka, Japan). Briefly, one microgram of the RNA was added into 20 μL reaction mixture containing 2 μM Oligo d(T), 0.5 mM dNTPs mix and 100 U of reverse transcriptase, and the reactions were incubated as follow: 65 °C for 5 min, 42 °C for 60 min and 85 °C for 5 min in the T100 thermal cycler (Bio-Rad, Foster city, CA, USA). Finally, the TiLV viral copy number was assessed using a SYBR Green-based qPCR assay. The assay was carried out in a 20 μL reaction mixture containing 10 μL of iTaq universal SYBR green supermix (Cat No. 172–5125; Bio-Rad, Hercules, CA, USA), 0.3 μL of forward (CTGAGCTAAAGAGGCAATATGGATT) and reverse (CGTGCGTACTCGTTCAGTATAAGTTCT) primers, 4 μL of cDNA template and molecular-grade water to adjust the final volume. The cycling condition was set as follows: denaturation at 95 °C for 3 min, 40 cycles of 95 °C for 10 s, and 60 °C for 30 s . At the end of the qPCR cycle, the TiLV log copy number (PCR product size 112 bp) was retrieved from the standard curve of the melting temperature obtained using CFX Maestro Software (Bio-Rad, Chicago, IL, USA). Detection of TiLV in cell lines using an immunofluorescence assay In line with the protocol described by , an immunofluorescent (IFA) assay was performed to study the dynamics of TiLV infection in the RHTiB cells. Briefly, 1 × 10 5 RHTiB cells were seeded on a cell culture chamber slide (SPL Life Sciences, Gyeonggi-do, Korea) and cultured in an L-15 medium supplemented with 5% FBS at 25 °C until they reached 80–90% confluency. Subsequently, the cells were washed and inoculated with a control medium or 0.1 MOI of TiLV for 10, 30 min, 1, 3, or 24 h at 25 °C. Following incubation, the samples were fixed with ice-cold 100% methanol for 10 min, followed by permeabilization with 0.1% Triton X-100 (Cat No. T8532; Sigma-Aldrich, Saint-Louis, MO, USA) for 10 min. A blocking solution containing 2% bovine serum albumin (Cat No. 85040C; Sigma-Aldrich, Darmstadt, Germany) in phosphate buffer saline (PBS) was applied for 30 min to reduce non-specific binding. Subsequently, the cells were probed with a 1:100 dilution of rabbit polyclonal anti-TiLV IgG prepared in blocking solution overnight at 4 °C. The cells were then washed three times with PBS and incubated with a goat anti-rabbit IgG-Alexa Fluor 488 (Cat No. ab190195; Abcam, Carlsbad, CA, USA) diluted 1:1,000 in PBS for 1 h at room temperature. Finally, the cell nuclei were stained with 1 µg/mL of diaminophenylindole (Cat No. 10 236 276 001; DAPI, Sigma-Aldrich, St Louis, MO, USA) before visualization under a confocal microscope (Fluoview 3000; Olympus, Tokyo, Japan). The negative control cells probed with secondary IgG was conducted in parallel to confirm that non-specific signal had occurred in the experiment . Mass spectrometry-based proteomic and phosphoproteomic analyses Phosphoprotein enrichment and sample preparation The phosphoproteins in the samples were enriched using the immobilized metal affinity column resin charged with Ga 3+ ions as per the manufacturer’s instructions (Cat No. 90003; Pierce, Thermo Fisher Scientific, Rockford, IL, USA). Each sample was adjusted to a final concentration of 0.5 mg/mL using lysis/binding/wash buffer. The column was pre-equilibrated with lysis/binding/wash buffer containing 3-[(3-cholamidopropyl) dimethylammonio]-1-propanesulfonate (CHAPS; Cat No. 436550250; Thermo Fisher Scientific, Rockford, IL, USA) and centrifuged at 1,000× g for 1 min at 4 °C. The samples were applied to the column and incubated for 30 min at 4 °C. Following incubation, the column was centrifuged at 1,000× g for 1 min at 4 °C to collect the flow-through fraction. After three washes with lysis/binding/wash buffer containing CHAPS, the phosphoproteins were eluted from the column via incubation with 1 mL of elution buffer at room temperature for 3 min, then centrifuged at 1,000× g for 1 min at 4 °C. The elution process was repeated four times, and the collected fractions were pooled for liquid chromatography and mass spectrometry (LC-MS/MS) analysis. The protein concentrations in the samples were assessed using the modified Lowry technique . All the protein and phosphoprotein samples were prepared by adding 5 mM dithiothreitol (Cat No. 646563; Sigma-Aldrich, Saint Louis, MO USA) to 10 mM ammonium bicarbonate (NH 4 HCO 3 ) (Cat No. A6141; Sigma-Aldrich, Saint Louis, MO, USA) at 60 °C for 1 h to reduce the disulfide-containing compounds. Alkylation of the reduced cysteine residues was achieved by incubating with 15 mM iodoacetamide (Cat No. I6125; Sigma-Aldrich, Saint Louis, MO, USA) in 10 mM ammonium bicarbonate at 25 °C for 45 min. Subsequently, the proteins and phosphoproteins were digested with trypsin (Cat No. V5111; Promega, Madison, WI, USA) for 3 h at room temperature. Finally, the digested peptide samples were dissolved in 0.1% formic acid (FA) (Cat No. 270480100; Thermo Fisher Scientific, Waltham, MA, USA) and submitted for LC-MS/MS analysis. LC-MS/MS analysis The identification of the digested peptides was carried out using the Ultimate 3000 Nano/Capillary LC System (Thermo Fisher Scientific, Waltham, MA, USA) coupled to a ZenoTOF 7600 mass spectrometer (SCIEX, Framingham, MA, USA). The digested peptide samples were concentrated using Acclaim 5 µm PepMap 300 µ-Precolumns packed with C18 (300 µm × 5 mm, 5 µm, 100 Å, Thermo Fisher Scientific, Waltham, MA, USA) and subsequently separated using an Acclaim PepMap Rapid Separation Liquid Chromatography column (75 μm × 15 cm, 2 µm, 100 Å, Thermo Fisher Scientific, Waltham, MA, USA). The C18 column was enclosed within a temperature-controlled column oven set to 35 °C. For protein separation, the solvents A and B, which contained 0.1% FA in water and 0.1% FA in 80% acetonitrile, respectively, were utilized. A gradient of solvent B ranging from 5% to 55% over 30 min was employed at a flow rate of 0.30 μL/min. The source and gas parameters on the ZenoTOF 7600 system were configured as follows: ion source gas 1 at eight pounds per square inch (psi), curtain gas at 35 psi, CAD gas at 7 psi, source temperature at 200 °C, polarity set to positive, and spray voltage at 3,300 V. For data-dependent acquisition (DDA), the top 50 most abundant precursor ions per survey MS1 were selected for subsequent MS/MS analysis. These precursor ions met an intensity threshold exceeding 150 counts per second (cps). The precursor ions were dynamically excluded for 12 s after two incidences of MS/MS sampling (with dynamic collision energy enabled). The MS2 spectra were collected in the mass range of 100–1,800 m/z with a 50-ms accumulation time. The collision energy (CE) parameters consisted of an 80 V declustering potential (DP), no DP spread, and a CE spread of 0 V. The time bins, which incorporated all the channels, were summed using a Zeno trap threshold of 150,000 cps. The cycle time for the top 60 DDA method was 3.0 s. To ensure quality control throughout the analytical process, three replicates of the same sample were analyzed to monitor the reproducibility of the result. Additionally, the digestion of bovine serum albumin served as a quality control sample to assess the performance and reliability of the mass spectrometry instrument and the entire analytical workflow, as previously reported . Data processing and analysis For protein identification, the raw mass spectral data were processed using MaxQuant software (version 2.2.0.0) for peptide and protein identification. MS/MS searches were performed against the reference proteome database for Oreochromis niloticus , NCBI:txid8128 (downloaded from UniProt), which includes 76,021 entries, supplemented with a contaminant database. The significant threshold for protein identification was established with a p -value less than 0.05, and a false discovery rate (FDR) was set to 1% for both the peptides and proteins. The minimum peptide length was set to seven amino acid residues containing at least one unique peptide, as outlined in previous studies . Label-free quantification was performed using the MaxLFQ algorithm integrated into MaxQuant with a minimum ratio count of 2. For the phosphoproteomic analysis, phosphorylation of the serine, threonine, and tyrosine residues were included as variable modifications. The protein and phosphorylation site localizations were assessed using the MaxQuant software-supported post-translational modification localization probability algorithm. Bioinformatics analysis The protein and phosphoprotein abundance in the E-11 and RHTiB cells were compared across all time points using Metaboanalyst.ca (version 6.0), a web-based platform for statistical and pathway analysis of omics data. Analysis of variance (ANOVA) was utilized to identify statistically significant differences of proteins and phosphoproteins between the groups and among different time point of infection, while controlling FDRs and ensuring a significance level of <0.05 . Gene Ontology enrichment analysis was conducted to reveal the upregulation of the biological processes, molecular functions, and cellular components of the differentially expressed proteins and phosphoproteins. Cluster analyses and heatmap visualizations were employed through Metaboanalyst.ca to enhance the understanding of the protein expression and phosphorylation dynamics throughout the TiLV infection timeline. Protein–protein interaction (PPI) networks were constructed from the differential expressed proteins at each time point using the Search Tool for Interactions of Chemicals (STITCH database; http://stitch.embl.de ) to identify the cellular mechanisms involved in TiLV infection. Pathway analysis was performed using the Kyoto Encyclopedia of Genes and Genomes (KEGG) database , with confidence levels ranging from low edge (>0.15) to highest edge confidence scores (>0.900) to represent the key signal transduction pathways and metabolic processes impacted by TiLV infection in the host cells. The pathway analysis was validated using ShinyGO 0.80 tool ( bioinformatics.sdstate.edu/go/ ) to confirm the key pathways altered during TiLV infection . We selected two piscine cell lines, E-11 cells, derived from the snakehead fry (purchased from the European Collection Authenticated Cell Cultures), and RHTiB, brain cells derived from red hybrid tilapia ( Oreochromis spp.) which has been isolated and established previously , to investigate the proteomic and phosphoproteomic profiles following TiLV infection. Both cell lines (passage 30–35) were cultured in Leibovitz’s L-15 medium, which contained 2% fetal bovine serum (Cat No. F2442; FBS, Sigma-Aldrich, Saint-Louis, MO, USA) at 25 °C without CO 2 . When the cells reached 80% confluency, the TiLV strain VETKU-TV08, which had been isolated from moribund fish in Pathumthani province, was diluted in L-15 without FBS and inoculated into the cells at a multiplicity of infection (MOI) of 0.1 and incubated for 10, 30 min, 1, 3, 6, and 24 h, respectively ( n = 3 each). At each time point, the cytopathic effect was monitored using an inverted microscope (CKX53; Olympus, Tokyo, Japan). Cells were harvested at each time point using scrapers (Cat No. 179707PK; Thermo Fisher Scientific, Rochester, NY, USA), and 50 µL of lysates were preserved at −20 °C for viral quantification. The remainder cells were ground with an equal volume of sodium dodecyl sulfate (SDS) (Cat No. D05888; Sigma-Aldrich, Saint-Louis, MO, USA)-lysis buffer (4% w/v in 100 mM Tris/HCl, pH 8.2) containing cocktails of protease inhibitors (complete EDTA-free, Cat No. 11 873 580 001; Roche, Mannheim, Germany) and phosphatase inhibitors (Cat No. 04 906 837 001; PhosStop, Roche, Mannheim, Germany) to prevent protein degradation and dephosphorylation, respectively. The mixture was then collected and centrifuged at 10,000× g at 4 °C for 15 min. The supernatant was transferred into two new tubes for total protein and phosphoprotein analyses. Each fraction was mixed with two volumes of cold acetone and incubated overnight at −20 °C. The mixture was then centrifuged at 10,000× g for 15 min, and the supernatant was discarded. The resulting pellet was dried and stored at −80 °C until further use. The E-11 and RHTiB cells were assessed for TiLV at 10 min post-infection (mpi), 30 mpi, 1 h post-infection (hpi), 3, 6, and 24 hpi using a quantitative reverse transcription polymerase chain reaction (RT-qPCR). Initially, total RNA extraction from the collected cell lysates was conducted as described by . Briefly, the cell lysates were mixed with GENEzol reagent (Cat No. GZR200; Geneaid Biotech, Taipei, Taiwan) and chloroform (Cat No. 288306; Sigma-Aldrich, Saint-Louis, MO, USA) and centrifuged at 15,000× g at 4 °C. The supernatant was collected and treated with DNase I (Cat No. AM2224; Thermo Fisher Scientific, Carlsbad, CA, USA), followed by 2-propanol (Cat No. 109634; Merck, Darmstadt, Germany), and stored at −20 °C for 2 h. After thawing and centrifugation, the pellets were washed with ethanol and air-dried. The collected RNA was then reconstituted with RNase-free water and converted to complementary deoxyribonucleic acid (cDNA) using the ReverTra Ace cDNA synthesis kit (Cat No. FSQ-201; Toyobo, Osaka, Japan). Briefly, one microgram of the RNA was added into 20 μL reaction mixture containing 2 μM Oligo d(T), 0.5 mM dNTPs mix and 100 U of reverse transcriptase, and the reactions were incubated as follow: 65 °C for 5 min, 42 °C for 60 min and 85 °C for 5 min in the T100 thermal cycler (Bio-Rad, Foster city, CA, USA). Finally, the TiLV viral copy number was assessed using a SYBR Green-based qPCR assay. The assay was carried out in a 20 μL reaction mixture containing 10 μL of iTaq universal SYBR green supermix (Cat No. 172–5125; Bio-Rad, Hercules, CA, USA), 0.3 μL of forward (CTGAGCTAAAGAGGCAATATGGATT) and reverse (CGTGCGTACTCGTTCAGTATAAGTTCT) primers, 4 μL of cDNA template and molecular-grade water to adjust the final volume. The cycling condition was set as follows: denaturation at 95 °C for 3 min, 40 cycles of 95 °C for 10 s, and 60 °C for 30 s . At the end of the qPCR cycle, the TiLV log copy number (PCR product size 112 bp) was retrieved from the standard curve of the melting temperature obtained using CFX Maestro Software (Bio-Rad, Chicago, IL, USA). In line with the protocol described by , an immunofluorescent (IFA) assay was performed to study the dynamics of TiLV infection in the RHTiB cells. Briefly, 1 × 10 5 RHTiB cells were seeded on a cell culture chamber slide (SPL Life Sciences, Gyeonggi-do, Korea) and cultured in an L-15 medium supplemented with 5% FBS at 25 °C until they reached 80–90% confluency. Subsequently, the cells were washed and inoculated with a control medium or 0.1 MOI of TiLV for 10, 30 min, 1, 3, or 24 h at 25 °C. Following incubation, the samples were fixed with ice-cold 100% methanol for 10 min, followed by permeabilization with 0.1% Triton X-100 (Cat No. T8532; Sigma-Aldrich, Saint-Louis, MO, USA) for 10 min. A blocking solution containing 2% bovine serum albumin (Cat No. 85040C; Sigma-Aldrich, Darmstadt, Germany) in phosphate buffer saline (PBS) was applied for 30 min to reduce non-specific binding. Subsequently, the cells were probed with a 1:100 dilution of rabbit polyclonal anti-TiLV IgG prepared in blocking solution overnight at 4 °C. The cells were then washed three times with PBS and incubated with a goat anti-rabbit IgG-Alexa Fluor 488 (Cat No. ab190195; Abcam, Carlsbad, CA, USA) diluted 1:1,000 in PBS for 1 h at room temperature. Finally, the cell nuclei were stained with 1 µg/mL of diaminophenylindole (Cat No. 10 236 276 001; DAPI, Sigma-Aldrich, St Louis, MO, USA) before visualization under a confocal microscope (Fluoview 3000; Olympus, Tokyo, Japan). The negative control cells probed with secondary IgG was conducted in parallel to confirm that non-specific signal had occurred in the experiment . Phosphoprotein enrichment and sample preparation The phosphoproteins in the samples were enriched using the immobilized metal affinity column resin charged with Ga 3+ ions as per the manufacturer’s instructions (Cat No. 90003; Pierce, Thermo Fisher Scientific, Rockford, IL, USA). Each sample was adjusted to a final concentration of 0.5 mg/mL using lysis/binding/wash buffer. The column was pre-equilibrated with lysis/binding/wash buffer containing 3-[(3-cholamidopropyl) dimethylammonio]-1-propanesulfonate (CHAPS; Cat No. 436550250; Thermo Fisher Scientific, Rockford, IL, USA) and centrifuged at 1,000× g for 1 min at 4 °C. The samples were applied to the column and incubated for 30 min at 4 °C. Following incubation, the column was centrifuged at 1,000× g for 1 min at 4 °C to collect the flow-through fraction. After three washes with lysis/binding/wash buffer containing CHAPS, the phosphoproteins were eluted from the column via incubation with 1 mL of elution buffer at room temperature for 3 min, then centrifuged at 1,000× g for 1 min at 4 °C. The elution process was repeated four times, and the collected fractions were pooled for liquid chromatography and mass spectrometry (LC-MS/MS) analysis. The protein concentrations in the samples were assessed using the modified Lowry technique . All the protein and phosphoprotein samples were prepared by adding 5 mM dithiothreitol (Cat No. 646563; Sigma-Aldrich, Saint Louis, MO USA) to 10 mM ammonium bicarbonate (NH 4 HCO 3 ) (Cat No. A6141; Sigma-Aldrich, Saint Louis, MO, USA) at 60 °C for 1 h to reduce the disulfide-containing compounds. Alkylation of the reduced cysteine residues was achieved by incubating with 15 mM iodoacetamide (Cat No. I6125; Sigma-Aldrich, Saint Louis, MO, USA) in 10 mM ammonium bicarbonate at 25 °C for 45 min. Subsequently, the proteins and phosphoproteins were digested with trypsin (Cat No. V5111; Promega, Madison, WI, USA) for 3 h at room temperature. Finally, the digested peptide samples were dissolved in 0.1% formic acid (FA) (Cat No. 270480100; Thermo Fisher Scientific, Waltham, MA, USA) and submitted for LC-MS/MS analysis. LC-MS/MS analysis The identification of the digested peptides was carried out using the Ultimate 3000 Nano/Capillary LC System (Thermo Fisher Scientific, Waltham, MA, USA) coupled to a ZenoTOF 7600 mass spectrometer (SCIEX, Framingham, MA, USA). The digested peptide samples were concentrated using Acclaim 5 µm PepMap 300 µ-Precolumns packed with C18 (300 µm × 5 mm, 5 µm, 100 Å, Thermo Fisher Scientific, Waltham, MA, USA) and subsequently separated using an Acclaim PepMap Rapid Separation Liquid Chromatography column (75 μm × 15 cm, 2 µm, 100 Å, Thermo Fisher Scientific, Waltham, MA, USA). The C18 column was enclosed within a temperature-controlled column oven set to 35 °C. For protein separation, the solvents A and B, which contained 0.1% FA in water and 0.1% FA in 80% acetonitrile, respectively, were utilized. A gradient of solvent B ranging from 5% to 55% over 30 min was employed at a flow rate of 0.30 μL/min. The source and gas parameters on the ZenoTOF 7600 system were configured as follows: ion source gas 1 at eight pounds per square inch (psi), curtain gas at 35 psi, CAD gas at 7 psi, source temperature at 200 °C, polarity set to positive, and spray voltage at 3,300 V. For data-dependent acquisition (DDA), the top 50 most abundant precursor ions per survey MS1 were selected for subsequent MS/MS analysis. These precursor ions met an intensity threshold exceeding 150 counts per second (cps). The precursor ions were dynamically excluded for 12 s after two incidences of MS/MS sampling (with dynamic collision energy enabled). The MS2 spectra were collected in the mass range of 100–1,800 m/z with a 50-ms accumulation time. The collision energy (CE) parameters consisted of an 80 V declustering potential (DP), no DP spread, and a CE spread of 0 V. The time bins, which incorporated all the channels, were summed using a Zeno trap threshold of 150,000 cps. The cycle time for the top 60 DDA method was 3.0 s. To ensure quality control throughout the analytical process, three replicates of the same sample were analyzed to monitor the reproducibility of the result. Additionally, the digestion of bovine serum albumin served as a quality control sample to assess the performance and reliability of the mass spectrometry instrument and the entire analytical workflow, as previously reported . Data processing and analysis For protein identification, the raw mass spectral data were processed using MaxQuant software (version 2.2.0.0) for peptide and protein identification. MS/MS searches were performed against the reference proteome database for Oreochromis niloticus , NCBI:txid8128 (downloaded from UniProt), which includes 76,021 entries, supplemented with a contaminant database. The significant threshold for protein identification was established with a p -value less than 0.05, and a false discovery rate (FDR) was set to 1% for both the peptides and proteins. The minimum peptide length was set to seven amino acid residues containing at least one unique peptide, as outlined in previous studies . Label-free quantification was performed using the MaxLFQ algorithm integrated into MaxQuant with a minimum ratio count of 2. For the phosphoproteomic analysis, phosphorylation of the serine, threonine, and tyrosine residues were included as variable modifications. The protein and phosphorylation site localizations were assessed using the MaxQuant software-supported post-translational modification localization probability algorithm. Bioinformatics analysis The protein and phosphoprotein abundance in the E-11 and RHTiB cells were compared across all time points using Metaboanalyst.ca (version 6.0), a web-based platform for statistical and pathway analysis of omics data. Analysis of variance (ANOVA) was utilized to identify statistically significant differences of proteins and phosphoproteins between the groups and among different time point of infection, while controlling FDRs and ensuring a significance level of <0.05 . Gene Ontology enrichment analysis was conducted to reveal the upregulation of the biological processes, molecular functions, and cellular components of the differentially expressed proteins and phosphoproteins. Cluster analyses and heatmap visualizations were employed through Metaboanalyst.ca to enhance the understanding of the protein expression and phosphorylation dynamics throughout the TiLV infection timeline. Protein–protein interaction (PPI) networks were constructed from the differential expressed proteins at each time point using the Search Tool for Interactions of Chemicals (STITCH database; http://stitch.embl.de ) to identify the cellular mechanisms involved in TiLV infection. Pathway analysis was performed using the Kyoto Encyclopedia of Genes and Genomes (KEGG) database , with confidence levels ranging from low edge (>0.15) to highest edge confidence scores (>0.900) to represent the key signal transduction pathways and metabolic processes impacted by TiLV infection in the host cells. The pathway analysis was validated using ShinyGO 0.80 tool ( bioinformatics.sdstate.edu/go/ ) to confirm the key pathways altered during TiLV infection . The phosphoproteins in the samples were enriched using the immobilized metal affinity column resin charged with Ga 3+ ions as per the manufacturer’s instructions (Cat No. 90003; Pierce, Thermo Fisher Scientific, Rockford, IL, USA). Each sample was adjusted to a final concentration of 0.5 mg/mL using lysis/binding/wash buffer. The column was pre-equilibrated with lysis/binding/wash buffer containing 3-[(3-cholamidopropyl) dimethylammonio]-1-propanesulfonate (CHAPS; Cat No. 436550250; Thermo Fisher Scientific, Rockford, IL, USA) and centrifuged at 1,000× g for 1 min at 4 °C. The samples were applied to the column and incubated for 30 min at 4 °C. Following incubation, the column was centrifuged at 1,000× g for 1 min at 4 °C to collect the flow-through fraction. After three washes with lysis/binding/wash buffer containing CHAPS, the phosphoproteins were eluted from the column via incubation with 1 mL of elution buffer at room temperature for 3 min, then centrifuged at 1,000× g for 1 min at 4 °C. The elution process was repeated four times, and the collected fractions were pooled for liquid chromatography and mass spectrometry (LC-MS/MS) analysis. The protein concentrations in the samples were assessed using the modified Lowry technique . All the protein and phosphoprotein samples were prepared by adding 5 mM dithiothreitol (Cat No. 646563; Sigma-Aldrich, Saint Louis, MO USA) to 10 mM ammonium bicarbonate (NH 4 HCO 3 ) (Cat No. A6141; Sigma-Aldrich, Saint Louis, MO, USA) at 60 °C for 1 h to reduce the disulfide-containing compounds. Alkylation of the reduced cysteine residues was achieved by incubating with 15 mM iodoacetamide (Cat No. I6125; Sigma-Aldrich, Saint Louis, MO, USA) in 10 mM ammonium bicarbonate at 25 °C for 45 min. Subsequently, the proteins and phosphoproteins were digested with trypsin (Cat No. V5111; Promega, Madison, WI, USA) for 3 h at room temperature. Finally, the digested peptide samples were dissolved in 0.1% formic acid (FA) (Cat No. 270480100; Thermo Fisher Scientific, Waltham, MA, USA) and submitted for LC-MS/MS analysis. The identification of the digested peptides was carried out using the Ultimate 3000 Nano/Capillary LC System (Thermo Fisher Scientific, Waltham, MA, USA) coupled to a ZenoTOF 7600 mass spectrometer (SCIEX, Framingham, MA, USA). The digested peptide samples were concentrated using Acclaim 5 µm PepMap 300 µ-Precolumns packed with C18 (300 µm × 5 mm, 5 µm, 100 Å, Thermo Fisher Scientific, Waltham, MA, USA) and subsequently separated using an Acclaim PepMap Rapid Separation Liquid Chromatography column (75 μm × 15 cm, 2 µm, 100 Å, Thermo Fisher Scientific, Waltham, MA, USA). The C18 column was enclosed within a temperature-controlled column oven set to 35 °C. For protein separation, the solvents A and B, which contained 0.1% FA in water and 0.1% FA in 80% acetonitrile, respectively, were utilized. A gradient of solvent B ranging from 5% to 55% over 30 min was employed at a flow rate of 0.30 μL/min. The source and gas parameters on the ZenoTOF 7600 system were configured as follows: ion source gas 1 at eight pounds per square inch (psi), curtain gas at 35 psi, CAD gas at 7 psi, source temperature at 200 °C, polarity set to positive, and spray voltage at 3,300 V. For data-dependent acquisition (DDA), the top 50 most abundant precursor ions per survey MS1 were selected for subsequent MS/MS analysis. These precursor ions met an intensity threshold exceeding 150 counts per second (cps). The precursor ions were dynamically excluded for 12 s after two incidences of MS/MS sampling (with dynamic collision energy enabled). The MS2 spectra were collected in the mass range of 100–1,800 m/z with a 50-ms accumulation time. The collision energy (CE) parameters consisted of an 80 V declustering potential (DP), no DP spread, and a CE spread of 0 V. The time bins, which incorporated all the channels, were summed using a Zeno trap threshold of 150,000 cps. The cycle time for the top 60 DDA method was 3.0 s. To ensure quality control throughout the analytical process, three replicates of the same sample were analyzed to monitor the reproducibility of the result. Additionally, the digestion of bovine serum albumin served as a quality control sample to assess the performance and reliability of the mass spectrometry instrument and the entire analytical workflow, as previously reported . For protein identification, the raw mass spectral data were processed using MaxQuant software (version 2.2.0.0) for peptide and protein identification. MS/MS searches were performed against the reference proteome database for Oreochromis niloticus , NCBI:txid8128 (downloaded from UniProt), which includes 76,021 entries, supplemented with a contaminant database. The significant threshold for protein identification was established with a p -value less than 0.05, and a false discovery rate (FDR) was set to 1% for both the peptides and proteins. The minimum peptide length was set to seven amino acid residues containing at least one unique peptide, as outlined in previous studies . Label-free quantification was performed using the MaxLFQ algorithm integrated into MaxQuant with a minimum ratio count of 2. For the phosphoproteomic analysis, phosphorylation of the serine, threonine, and tyrosine residues were included as variable modifications. The protein and phosphorylation site localizations were assessed using the MaxQuant software-supported post-translational modification localization probability algorithm. The protein and phosphoprotein abundance in the E-11 and RHTiB cells were compared across all time points using Metaboanalyst.ca (version 6.0), a web-based platform for statistical and pathway analysis of omics data. Analysis of variance (ANOVA) was utilized to identify statistically significant differences of proteins and phosphoproteins between the groups and among different time point of infection, while controlling FDRs and ensuring a significance level of <0.05 . Gene Ontology enrichment analysis was conducted to reveal the upregulation of the biological processes, molecular functions, and cellular components of the differentially expressed proteins and phosphoproteins. Cluster analyses and heatmap visualizations were employed through Metaboanalyst.ca to enhance the understanding of the protein expression and phosphorylation dynamics throughout the TiLV infection timeline. Protein–protein interaction (PPI) networks were constructed from the differential expressed proteins at each time point using the Search Tool for Interactions of Chemicals (STITCH database; http://stitch.embl.de ) to identify the cellular mechanisms involved in TiLV infection. Pathway analysis was performed using the Kyoto Encyclopedia of Genes and Genomes (KEGG) database , with confidence levels ranging from low edge (>0.15) to highest edge confidence scores (>0.900) to represent the key signal transduction pathways and metabolic processes impacted by TiLV infection in the host cells. The pathway analysis was validated using ShinyGO 0.80 tool ( bioinformatics.sdstate.edu/go/ ) to confirm the key pathways altered during TiLV infection . Dynamics of TiLV in the E-11 and RHTiB cell lines We investigated the dynamics of TiLV in the E-11 and RHTiB cell lines using morphological observations, qRT-PCR for viral RNA quantification, and IFA . Up to 24 h post TiLV infection, neither cell line had exhibited cytopathic effects or morphological changes . Initially, TiLV RNA was detected at 10 mpi in both cell lines, with similar viral copy numbers of 6.71 ± 0.03 and 6.25 ± 0.18 log 10 copies/400 ng cDNA, respectively . Subsequently, the viral load increased gradually in both cell lines, peaking at 24 hpi with viral copy numbers of 8.22 ± 0.06 and 6.76 ± 0.04 log 10 copies/400 ng cDNA in the E-11 and RHTiB cells, respectively. Notably, the viral concentration in the E-11 cells was significantly higher than that in the RHTiB cell lines at 6 and 24 hpi ( p < 0.05). Additionally, TiLV was detected in the RHTiB cells throughout the studied period using IFA. Interestingly, an intracellular signal representing TiLV infection was initially observed at 10 mpi, while colocalization of TiLV with cell nuclei was first detected at 30 mpi . Proteomic and phosphoproteomic analyses of the E-11 and RHTiB cells The proteomic and phosphoproteomic patterns of the E-11 and RHTiB cells were analyzed using in-solution digestion and LC-MS/MS . Notably, we identified remarkable changes (2,057 proteins and 1,838 phosphoproteins) in the E-11 cells, while a total of 2,473 proteins and 2,157 phosphoproteins were identified in the RHTiB cells . Partial least squares discriminant analysis (PLS-DA) of the proteomic data revealed distinct protein profiles between the E-11 and RHTiB cells . In contrast, cluster analyses using PLS-DA of the phosphoproteins showed an overlap between the E-11 and RHTiB cells, which suggested some similar phosphorylation patterns between the two cell types . Additionally, the heat maps display the dynamic expression levels of protein and phosphoprotein in the E-11 and RHTiB cell lines throughout the study. Along the vertical axis, the hierarchical clustered by the dendrogram highlighted the distinct patterns of protein and phosphoprotein expression between the two cell lines following TiLV infection. Meanwhile, the horizontal axis illustrated the differential changes of protein and phosphoprotein patterns at each time point. ANOVA and enrichment pathway analyses of the TiLV-infected E-11 cells The dynamics of the protein and phosphoprotein alterations in the TiLV-infected E-11 cells were analyzed using ANOVA and PLS-DA to identify differentially expressed proteins and phosphoproteins . The analysis revealed significant changes in 53 proteins and 136 phosphoproteins throughout the study period . Notably, the control E-11 cells formed a separate cluster, which demonstrated a distinct protein profile compared to the infected cells . Conversely, PLS-DA analysis of the phosphoproteome data did not reveal distinct clusters among the E-11 cells collected at different time points . To explore the viral entry and replication mechanisms as well as the cellular response processes, PPI networks were constructed from differentially altered significant proteins and phosphoproteins at 10 and 24 mpi using the STITCH online software, and the enrichment pathways were analyzed against the KEGG database. At 10 mpi, when TiLV signals were initially detected in the E-11 cells, 24 proteins and 33 phosphoproteins were significantly decreased . The pathway analysis from the PPI of these proteins demonstrated relationships between basic helix-loop-helix family, member e41 (BHLHE41), calcium-binding and coiled-coil domain-containing protein 2 (CALCOCO2), transgelin (TAGLN), eukaryotic translation elongation factor 2 (EEF2), exosome component 3 (EXOSC3), prolyl 4-hydroxylase (P4HB), ADAM metallopeptidase domain 12 (ADAM12), and steroidogenic factor 1 (SF1), as well as the phosphoproteins NMT1, receptor-interacting protein kinase 1 (RIPK1), exportin-5 (XPO5), class II transactivator (CIITA), plakophilin-9 (PKP9), protein tyrosine phosphatase receptor type C (PTPRC), endonuclease G (ENDOG), Janus kinase-2 (JAK2), Ras-related nuclear protein (RAN), myosin phosphatase Rho interacting protein (MPRIP), Rho GTPase activating protein 21 (ARHGAP21), Rho guanine nucleotide exchange factor 12 (ARHGEF12), and caspase 1 (CASP1) . Specifically, the KEGG pathway analyses showed immediate suppression of several intracellular mechanisms, including RNA degradation, the JAK–signal transducer and activators of transcription (STAT) signaling pathway, and Fas-associated death domain protein (FADD)–tumor necrosis factor receptor associated factor (TRAF) . At 24 hpi, one protein and 51 phosphoproteins gradually increased in the E-11 cells . The constructed PPI network revealed relationships among the phosphoproteins, such as enolase 1 (ENO1), P4HB, muscle Ras oncogene (MRAS), Lamin-B2 (LMNB2), Ecto-NOX disulfide-thiol exchanger 1 (ENOX1), telomerase reverse transcriptase (TERT), thioredoxin interacting protein (TXNIP), tumor necrosis factor receptor-associated factor 2 (TRAF2) hydroxysteroid dehydrogenase like 1 (HSDL1), NMD3 ribosome export adaptor (NMD3), Epstein-Barr nuclear antigen 1 binding protein 2 (EBNA1BP2), SERTA domain-containing protein 2 (SERTAD2), caspase 9 (CASP9), intraflagellar transport 122 (IFT122), lysophosphatidylglycerol acyltransferase 1 (LPGAT1), golgin subfamily A member 4 (GOLGA4), histone proteins (H3F3B, HIST1H2BD), cluster of differentiation 97 (CD97), and calbindin 2 (CALB2) . These proteins participated in nucleotide oligomerization domain (NOD)-like receptor signaling and apoptotic pathways in the fish cells . ANOVA and enrichment pathway analyses of the TiLV-infected RHTiB cells The proteins and phosphoproteins obtained from the TiLV-infected RHTiB cells at each time point were analyzed using ANOVA and clustering methods . The ANOVA analysis identified differential expressions of 71 proteins and five phosphoproteins across the study . The PLS-DA plot of the proteomic data successfully separated the TiLV-infected RHTiB cells at each time point into distinct clusters . Similarly, the control RHTiB cells formed an isolated cluster, which resembled the control E-11 cells, and represented a unique protein profile compared to the infected cells. Conversely, the PLS-DA analysis of the phosphoproteomic profile did not reveal distinct clusters among the RHTiB groups at any time point . Based on the differential expression of the proteins and phosphoproteins in the RHTiB cells, we focused on the 10 mpi and 24 hpi groups to evaluate the possible mechanisms involved in the TiLV entry, replication, and host cellular responses. At 10 mpi, when the TiLV signals were first detected in the RHTiB cells, four proteins, namely, phosphatase 2 regulatory subunit B alpha (PPP2R2A), ATP-binding cassette sub-family C member 5 (ABCC5), methylenetetrahydrofolate dehydrogenase 2 (MTHFD2), and Tetraspanin-13 (TSPAN13), and one phosphoprotein, BCAS3, were significantly decreased compared to the control group . The enrichment analysis from these molecules suggested the suppression of adenosine monophosphate-activated protein kinase (AMPK) signaling in the RHTiB cells at this time point . At 24 hpi, we observed alterations in four proteins, that is, myosin heavy chain 9 (MYH9), pleckstrin homology domain-containing protein 2 (PLEKHH2), Tensin 3 (TNS3), and methylenetetrahydrofolate dehydrogenase 1 (MTHFD1), and two phosphoproteins, GTPase immunity-associated protein family member 7 (GIMAP7) and phosphorylated ribosomal protein L17 (RPL17). The relationships between these proteins indicated increased activity of the actin cytoskeleton in the TiLV-infected RHTiB cells . We investigated the dynamics of TiLV in the E-11 and RHTiB cell lines using morphological observations, qRT-PCR for viral RNA quantification, and IFA . Up to 24 h post TiLV infection, neither cell line had exhibited cytopathic effects or morphological changes . Initially, TiLV RNA was detected at 10 mpi in both cell lines, with similar viral copy numbers of 6.71 ± 0.03 and 6.25 ± 0.18 log 10 copies/400 ng cDNA, respectively . Subsequently, the viral load increased gradually in both cell lines, peaking at 24 hpi with viral copy numbers of 8.22 ± 0.06 and 6.76 ± 0.04 log 10 copies/400 ng cDNA in the E-11 and RHTiB cells, respectively. Notably, the viral concentration in the E-11 cells was significantly higher than that in the RHTiB cell lines at 6 and 24 hpi ( p < 0.05). Additionally, TiLV was detected in the RHTiB cells throughout the studied period using IFA. Interestingly, an intracellular signal representing TiLV infection was initially observed at 10 mpi, while colocalization of TiLV with cell nuclei was first detected at 30 mpi . The proteomic and phosphoproteomic patterns of the E-11 and RHTiB cells were analyzed using in-solution digestion and LC-MS/MS . Notably, we identified remarkable changes (2,057 proteins and 1,838 phosphoproteins) in the E-11 cells, while a total of 2,473 proteins and 2,157 phosphoproteins were identified in the RHTiB cells . Partial least squares discriminant analysis (PLS-DA) of the proteomic data revealed distinct protein profiles between the E-11 and RHTiB cells . In contrast, cluster analyses using PLS-DA of the phosphoproteins showed an overlap between the E-11 and RHTiB cells, which suggested some similar phosphorylation patterns between the two cell types . Additionally, the heat maps display the dynamic expression levels of protein and phosphoprotein in the E-11 and RHTiB cell lines throughout the study. Along the vertical axis, the hierarchical clustered by the dendrogram highlighted the distinct patterns of protein and phosphoprotein expression between the two cell lines following TiLV infection. Meanwhile, the horizontal axis illustrated the differential changes of protein and phosphoprotein patterns at each time point. The dynamics of the protein and phosphoprotein alterations in the TiLV-infected E-11 cells were analyzed using ANOVA and PLS-DA to identify differentially expressed proteins and phosphoproteins . The analysis revealed significant changes in 53 proteins and 136 phosphoproteins throughout the study period . Notably, the control E-11 cells formed a separate cluster, which demonstrated a distinct protein profile compared to the infected cells . Conversely, PLS-DA analysis of the phosphoproteome data did not reveal distinct clusters among the E-11 cells collected at different time points . To explore the viral entry and replication mechanisms as well as the cellular response processes, PPI networks were constructed from differentially altered significant proteins and phosphoproteins at 10 and 24 mpi using the STITCH online software, and the enrichment pathways were analyzed against the KEGG database. At 10 mpi, when TiLV signals were initially detected in the E-11 cells, 24 proteins and 33 phosphoproteins were significantly decreased . The pathway analysis from the PPI of these proteins demonstrated relationships between basic helix-loop-helix family, member e41 (BHLHE41), calcium-binding and coiled-coil domain-containing protein 2 (CALCOCO2), transgelin (TAGLN), eukaryotic translation elongation factor 2 (EEF2), exosome component 3 (EXOSC3), prolyl 4-hydroxylase (P4HB), ADAM metallopeptidase domain 12 (ADAM12), and steroidogenic factor 1 (SF1), as well as the phosphoproteins NMT1, receptor-interacting protein kinase 1 (RIPK1), exportin-5 (XPO5), class II transactivator (CIITA), plakophilin-9 (PKP9), protein tyrosine phosphatase receptor type C (PTPRC), endonuclease G (ENDOG), Janus kinase-2 (JAK2), Ras-related nuclear protein (RAN), myosin phosphatase Rho interacting protein (MPRIP), Rho GTPase activating protein 21 (ARHGAP21), Rho guanine nucleotide exchange factor 12 (ARHGEF12), and caspase 1 (CASP1) . Specifically, the KEGG pathway analyses showed immediate suppression of several intracellular mechanisms, including RNA degradation, the JAK–signal transducer and activators of transcription (STAT) signaling pathway, and Fas-associated death domain protein (FADD)–tumor necrosis factor receptor associated factor (TRAF) . At 24 hpi, one protein and 51 phosphoproteins gradually increased in the E-11 cells . The constructed PPI network revealed relationships among the phosphoproteins, such as enolase 1 (ENO1), P4HB, muscle Ras oncogene (MRAS), Lamin-B2 (LMNB2), Ecto-NOX disulfide-thiol exchanger 1 (ENOX1), telomerase reverse transcriptase (TERT), thioredoxin interacting protein (TXNIP), tumor necrosis factor receptor-associated factor 2 (TRAF2) hydroxysteroid dehydrogenase like 1 (HSDL1), NMD3 ribosome export adaptor (NMD3), Epstein-Barr nuclear antigen 1 binding protein 2 (EBNA1BP2), SERTA domain-containing protein 2 (SERTAD2), caspase 9 (CASP9), intraflagellar transport 122 (IFT122), lysophosphatidylglycerol acyltransferase 1 (LPGAT1), golgin subfamily A member 4 (GOLGA4), histone proteins (H3F3B, HIST1H2BD), cluster of differentiation 97 (CD97), and calbindin 2 (CALB2) . These proteins participated in nucleotide oligomerization domain (NOD)-like receptor signaling and apoptotic pathways in the fish cells . The proteins and phosphoproteins obtained from the TiLV-infected RHTiB cells at each time point were analyzed using ANOVA and clustering methods . The ANOVA analysis identified differential expressions of 71 proteins and five phosphoproteins across the study . The PLS-DA plot of the proteomic data successfully separated the TiLV-infected RHTiB cells at each time point into distinct clusters . Similarly, the control RHTiB cells formed an isolated cluster, which resembled the control E-11 cells, and represented a unique protein profile compared to the infected cells. Conversely, the PLS-DA analysis of the phosphoproteomic profile did not reveal distinct clusters among the RHTiB groups at any time point . Based on the differential expression of the proteins and phosphoproteins in the RHTiB cells, we focused on the 10 mpi and 24 hpi groups to evaluate the possible mechanisms involved in the TiLV entry, replication, and host cellular responses. At 10 mpi, when the TiLV signals were first detected in the RHTiB cells, four proteins, namely, phosphatase 2 regulatory subunit B alpha (PPP2R2A), ATP-binding cassette sub-family C member 5 (ABCC5), methylenetetrahydrofolate dehydrogenase 2 (MTHFD2), and Tetraspanin-13 (TSPAN13), and one phosphoprotein, BCAS3, were significantly decreased compared to the control group . The enrichment analysis from these molecules suggested the suppression of adenosine monophosphate-activated protein kinase (AMPK) signaling in the RHTiB cells at this time point . At 24 hpi, we observed alterations in four proteins, that is, myosin heavy chain 9 (MYH9), pleckstrin homology domain-containing protein 2 (PLEKHH2), Tensin 3 (TNS3), and methylenetetrahydrofolate dehydrogenase 1 (MTHFD1), and two phosphoproteins, GTPase immunity-associated protein family member 7 (GIMAP7) and phosphorylated ribosomal protein L17 (RPL17). The relationships between these proteins indicated increased activity of the actin cytoskeleton in the TiLV-infected RHTiB cells . Viruses are microorganisms that require host cells for their replication . Hence, studying virus–host cell interactions allow the impacts of viruses on host cellular regulation to be understood and provide insights into how host cells combat viral infections . Generally, host cells respond to viral infections through a variety of mechanisms, such as immune activation, metabolic alteration, and cell cycle arrest, which demonstrates the dynamic nature of virus–host interactions . TiLV, a novel RNA virus identified in tilapia, can infect multiple fish cell lines and induce cell death across a range of tissues, including tilapia brain-, heart-, and liver-derived cell lines and other piscine cell lines . Despite advances in identifying the broad tropisms of the virus, the mechanisms of TiLV entry, replication, and specific host cell responses pose significant research challenges. In this study, we applied proteomic and phosphoproteomic analyses to understand early host–virus interactions during TiLV infection in two cell lines, E-11 cells, derived from the snakehead fish, which are not the natural host of TiLV but has been extensively studied in previous research , and RHTiB cells, the primary brain cells from the red tilapia that can propagate the virus . Our findings using RT-qPCR, and an IFA assay confirmed the early detection of TiLV in RHTiB cells at 10 mpi. Moreover, the presence of TiLV in the cell nuclei at 30 mpi indicated successful viral entry and the utilization of the host machinery for replication. These observations are consistent with those of previous studies, which showed early TiLV detection in fish cells within 1 hpi using different methods . Similarly, our results support previous findings indicated that the virus can enter fish cells rapidly—within a few minutes—and potentially cause biological changes in the host cells. Additionally, these findings support the selected time frame for investigating cellular damage at multiple levels in infected cells. Interestingly, in addition to the observed cellular damage, our findings revealed variations in the infection dynamics and TiLV replication across different fish cells. Notably, the TiLV load in the E-11 cells was higher than that in the RHTiB cells at 24 hpi, which is consistent with the results of a previous study . We propose that while TiLV is capable of propagating in recently established RHTiB cells, these cells are less susceptible to TiLV compared to well-established E-11 cells. This reduced susceptibility could be attributed to variations in the availability of receptors or the cellular machinery necessary for TiLV replication . To support this hypothesis, the bioinformatic analyses presented in this study demonstrated distinct patterns in the protein and phosphoprotein responses between the E-11 and RHTiB cells, which suggests differences in the host response mechanisms. Comprehensive analysis of the important signaling pathways in both cell lines is essential to further elucidate the mechanisms of viral replication and host response. Bioinformatic analyses of the infected E-11 cells revealed unique shifts in the patterns of protein expression during TiLV infection. At 10 mpi, there was a significant suppression of the proteins associated with the JAK family and STAT pathways, such as Janus kinase 2 (JAK2), protein tyrosine phosphatase (PTPRC), and the major histocompatibility complex CIITA. Proteins associated with the FADD–TRAF pathways, including CASP1 and RIPK1, were also suppressed. Importantly, these pathways, particularly JAK–STAT and FADD–TRAF, are critical for cellular responses to inflammation and interferon production during viral infections . In fish, these pathways are known to facilitate interferon (IFN)-γ production through the activation of tyrosine phosphatases and mitogen-activated protein kinase (MAPK) signaling pathways, which lead to the release of pro-inflammatory cytokines and antiviral proteins . Therefore, the observed downregulation of the proteins in these pathways indicates that TiLV may interfere with interferon-mediated antiviral activity to facilitate viral entry and replication, as seen with other fish viruses . The suppression of RIPK1 suggests that TiLV also prevents nuclear factor kappa B pathway (NF-κB) activation and subsequently reduces MAPK signaling and IFN production and promotes cell death . These findings are supported by the increased levels of phosphoproteins involved in the apoptotic pathway (SERTAD, TXNIP, CASP9), NOD-like receptor signaling (TXNIP, TRAF2), and cell replication cycle (H3F3B, HIST1H2BD, TERT) observed in the TiLV-infected E-11 cells in our study at 24 hpi. Similarly, transcriptomic profiling of liver tissues from TiLV-infected fish revealed that TiLV may be recognized through the NOD-like receptor, leading to the upregulation of traf2 and the activation of the NF-κB pathway . Interestingly, other piscine viruses utilize histone proteins to manipulate host cellular functions and suppress antiviral mechanisms, thereby enhancing viral replication . Additionally, these viruses interact with TERT and telomeric functions, which induces cellular stress that leads to senescence and apoptosis in the infected cells . Although TiLV-induced apoptosis has not been well documented, a study demonstrated that infected E-11 cells displayed cytopathic effects (CPEs) together with mitochondrial damage and ATP depletion, which suggests similar interference with cellular integrity . Furthermore, while no CPEs were detected in E-11 cells after 24 hpi in this study, alterations in proteins and phosphoproteins involving in apoptotic pathway aligned with other reports which have highlighted remarkable CPEs in E-11 cells following 2 days post infection . Based on these findings, we hypothesize that the mechanism underlying TiLV infection in E-11 cells involves the suppression of the JAK–STAT and FADD–TRAF pathways. This suppression not only promotes viral replication but may facilitate the apoptotic process . Hence, targeting proteins associated with the JAK–STAT and cell replication pathways is crucial in controlling TiLV replication during the early stages of infection. Given the complex interplay of these pathways in viral pathogenesis, further studies on antiviral therapies are warranted. These studies should aim to explore and address the specific viral mechanisms that induce cell death, as has been previously outlined . Such research could provide significant insights into potential therapeutic interventions for TiLV and other similar viruses. In the RHTiB cells, infection by TiLV prompted a distinct cellular response compared to the E-11 cells. Specifically, at 10 mpi, there was a suppression of the protein PPP2R2A, an enzyme involved in the dephosphorylation of intracellular proteins , and MTHFD2, which plays a role in RNA metabolism and translation . Previous research has demonstrated that the downregulation of protein phosphatase leads to cell senescence in aging zebrafish neuronal cells , and the suppression of MTHFD2 has been linked to cell death in infectious hematopoietic necrosis virus (IHNV)-infected zebrafish larvae . Thus, the suppression of these proteins in infected RHTiB cells indicates that TiLV may inhibit host cell replication and metabolism early in the infection process while maintaining cellular protein phosphorylation, which facilitates viral entry and replication, as seen with other viruses, such as influenza A virus, hepatitis B, and Epstein–Barr virus (EBV) . Additionally, at 10 mpi, the TiLV-infected RHTiB cells showed inhibition of cytoskeletal proteins, including BCAS3, which controls the direction of cell migration . Similar suppression of cytoskeletal proteins associated with the RhoA GTPase pathway, including plakophilin-3 (PKP3), ARHGEF12, ARHGAP21, MPRIP, and TAGLN, has been observed in infected E-11 cells. Cytoskeletal proteins, particularly microtubules, are targeted by viruses to facilitate their entry mechanisms , such as endosomal formation and micropinocytosis . For instance, TiLV enters primary cells derived from the bulbus arteriosus of O. mossambicus through dynamin activity but not endosomal acidification . However, the downregulation of these cytoskeletal proteins in both E-11 and RHTiB cells suggests that the early TiLV entry mechanism may rely on other processes, such as receptor-mediated or transmembrane diffusion . In contrast, after 24 hpi, the infected RHTiB cells showed increased activity of cytoskeletal proteins such as MYH9 and PLEKHH2 . MYH9 has been identified as a receptor for the herpes simplex virus 1 and EBV and serves as a co-receptor in the entry process of SARS-CoV-2 . Additionally, MYH9 is essential for the cellular entry of infectious pancreatic necrosis virus and the replication of porcine reproductive and respiratory syndrome virus . Hence, the mechanism of TiLV proliferation in RHTiB cells may involve the cytoskeletal activity to facilitate viral entry into these susceptible cells. The discrepancy results between the cytoskeletal proteins at 10 mpi and 24 hpi in RHTiB-infected cells demonstrates that at different time course of infection, the virus may utilize different pathways and in depth-analysis should be warranted to fulfil the viral entry and replication mechanisms. In our study, we observed an increase in RPL17 and GIMAP7 in the infected RHTiB cells at 24 hpi. The phosphorylation of these proteins is associated with the concentration of infectious salmon anemia virus in the head kidney of infected salmon , whereas the phosphorylation of GIMAPs plays a crucial role in the antiviral response in zebrafish . These findings suggest that TiLV modulates the host immune response and promotes significant viral replication in RHTiB cells , which is consistent with the results of previous studies . Further investigations into immunomodulatory agents may therefore help reduce viral replication and help combat this widespread disease in tilapia. One of the primary limitations of our study was the lack of validation for protein expression, which could have further strengthened the reliability of this study. Indeed, many commercial antibodies are developed for mammalian species, often resulting in lower specificity and limited cross-reactivity with tilapia proteins . While developing antibodies specifically for tilapia proteins could be considered as an alternative approach, this process is resource-intensive and faces challenges related to sensitivity and consistency as described in comprehensive review . Other molecular techniques, such as targeting pro- and antiviral genes with siRNA and CRISPR/Cas9, could serve as complementary methods for validating proteomic results. Nonetheless, the use of specific antibodies may still be necessary for definitive confirmation of gene manipulation effects . To ensure the accuracy of our protein identifications, we conducted bioinformatics analyses using multiple tools, including STITCH and ShinyGO 0.80, which yielded consistent results. Furthermore, the biological and functional roles of these proteins, particularly those involved in apoptosis, were reported in the existing literature . Further in-depth studies using advanced tools such as kinase-substrate enrichment analysis (KSEA) or other omics approaches, including transcriptomics, should be employed to map the regulatory kinases involved and to elucidate broader signaling networks. These iterative approaches are expected to generate a more detailed timeline of virus-induced phosphorylation events and to identify potential host targets for therapeutic interventions. Understanding the early interactions and subsequent cellular changes between viruses and host cells is important for advancing the knowledge of viral pathogenesis. In this article, we have provided valuable insights into the distinct proteomic and phosphoproteomic alterations between two piscine cell lines, E-11 and RHTiB, during TiLV infection. Our results show that TiLV-infected E-11 cells exhibit significant changes in the proteins involved in the JAK–STAT and FADD–TRAF pathways during early infection, which further activate NOD-signaling and apoptotic pathways, leading to the viral replication and cell death. In contrast, in RHTiB cells, TiLV infection suppresses host cellular metabolism to facilitate viral entry during early infection, while later stages of viral replication require cytoskeletal proteins and promote host immune responses. Further experimental studies are essential to validate these hypotheses and elucidate the precise mechanisms by which these proteins influence TiLV infection in different cell lines. Additionally, integrating omics approaches, such as transcriptomics, proteomics, and functional assays, can provide a comprehensive understanding of the molecular interactions underlying TiLV pathogenesis and host–virus interactions and ultimately facilitate the development of effective antiviral strategies. 10.7717/peerj.18923/supp-1 Supplemental Information 1 Raw data of mass-spectrometry-based proteomic analysis of E-11 and RhTiB cells at different time points. 10.7717/peerj.18923/supp-2 Supplemental Information 2 Raw data of mass-spectrometry-based phosphoproteomic analysis of E-11 and RhTiB cells at different time points. 10.7717/peerj.18923/supp-3 Supplemental Information 3 Analysis of variance comparing differential protein expression in TiLV-infected E-11 cells across the study. 10.7717/peerj.18923/supp-4 Supplemental Information 4 Analysis of variance comparing differential phosphoprotein expression in TiLV-infected E-11 cells across the study. 10.7717/peerj.18923/supp-5 Supplemental Information 5 Altered protein and phosphoproteins in TiLV-infected E-11 cells at 10 mpi and 24 hpi. 10.7717/peerj.18923/supp-6 Supplemental Information 6 Pathway analysis from selected altered proteins and phosphoproteins in TiLV-infected E-11 cells. 10.7717/peerj.18923/supp-7 Supplemental Information 7 Analysis of variance comparing differential protein expression in TiLV-infected RHTiB cells across the study. 10.7717/peerj.18923/supp-8 Supplemental Information 8 Analysis of variance comparing differential phosphoprotein expression in TiLV-infected RHTiB cells across the study. 10.7717/peerj.18923/supp-9 Supplemental Information 9 Altered protein and phosphoproteins in TiLV-infected RHTiB cells at 10 mpi and 24 hpi. 10.7717/peerj.18923/supp-10 Supplemental Information 10 Pathway analysis from selected altered proteins and phosphoproteins in TiLV-infected RHTiB cells. 10.7717/peerj.18923/supp-11 Supplemental Information 11 Immunofluorescence micrograph of RHTiB cell stained with only goat anti-rabbit IgG-Alexa Fluor 488. The nuclei were counterstained with DAPI. |
Study protocol of a pilot randomised controlled trial assessing the feasibility and acceptability of RecoverEsupport: a digital health intervention to enhance recovery in women undergoing surgery for breast cancer | c74365f3-420a-4c74-8167-592184600d84 | 11808865 | Surgical Procedures, Operative[mh] | Background and rationale Worldwide, breast cancer is the second most diagnosed cancer with around 2.3 million people diagnosed each year and over 685 000 deaths annually. Globally, expenditure on breast cancer is estimated to exceed US$2 trillion by 2050. Most women with breast cancer will undergo surgery, with approximately 40% having a mastectomy requiring a hospital stay averaging between 1–2 days. Around 50% of those women will choose to have a breast reconstruction following a mastectomy with implant-based reconstructions requiring an additional average length of stay of 1–2 days. Complications occur in around 10% of breast cancer surgical patients and patients need to be physically and mentally recovered from their surgery, to return to their daily lives, and to prepare for the next phase of their treatment regime, usually chemotherapy and/or radiation therapy. As health services transition towards more patient-centred care models, it is essential to consider the role patients can play in their recovery. As such, best-practice and internationally endorsed Enhanced Recovery After Surgery (ERAS) guidelines include specific patient-managed recommendations in addition to clinician-managed recommendations, to optimise patient recovery outcomes. A 2019 systematic review of ERAS pathways in breast reconstruction suggests that the implementation of ERAS guidelines for patients reduces the length of hospital stay without increasing postoperative complications, decreases opioid use and can improve quality of life (QoL). These findings were also supported by a 2023 retrospective study which investigated outcomes for 92 patients having a single mastectomy. Findings for the 32 patients managed in the ERAS group suggested that implementing ERAS pathways for mastectomy patients (without immediate reconstruction) is associated with a shorter length of hospital stay and a reduction in postoperative complications. The patient-managed recommendations include: Pre-surgery: preadmission optimisation (smoking cessation, achieving a healthy weight, being physically active and alcohol reduction/cessation and provision of information and education). Post-surgery: early mobilisation, rapid resumption of oral feeding and drinking, opioid minimisation and physiotherapy exercises. Post-discharge: home support (supportive care for the management of drains and wounds) and physiotherapy exercises. Non-compliance with ERAS guidelines including early resumption of feeding and postoperative mobilisation, has been associated with higher rates of postoperative complications. Although reports vary, poor adherence to ERAS recommendations is well documented and to date, no studies of adherence to the patient-led ERAS guidelines within patients with breast cancer could be identified. However, a prospective study of 1391 patients undergoing colon surgery reported non-compliance with early feeding and mobilisation in up to 30% of patients which was associated with higher rates of postoperative morbidity. Resource constraints can impede patient adherence to ERAS recommendations through limited communication and collaboration between staff and patients, resistance to change from patients and staff and limited patient education preparing them to take an active role in their recovery. Digital health interventions (DHI) may provide a way of addressing these barriers. DHIs have been shown to be effective in producing health behaviour change in patients with cancer and offer advantages over more traditional forms of support in that they can be tailored, are highly scalable, are cost-effective, and can support patients within and beyond the hospital setting. Evidence also suggests that DHIs have been effective in changing behaviours and managing symptoms such as pain and anxiety in patients with cancer. As such, a DHI may support patients to adhere to the patient-led ERAS recommendations. Given the high prevalence of breast cancer surgery and the limited evidence regarding effective, cost-effective and scalable interventions to enhance patients’ recovery from surgery, a DHI has been developed to increase adherence to the patient-led ERAS recommendations. While previous literature shows that adhering to ERAS recommendations improves clinical outcomes, there is minimal research indicating how patients can be best supported to adhere to these ERAS recommendations. To date, there has been no randomised controlled trial (RCT) evaluating a behavioural intervention to support patients with breast cancer adhering to the comprehensive set of patient-managed ERAS recommendations across the perioperative period. Prior to conducting a fully powered RCT to evaluate the effectiveness of this intervention, this pilot RCT will be used to determine the acceptability of the DHI and the feasibility of components of the research design. Worldwide, breast cancer is the second most diagnosed cancer with around 2.3 million people diagnosed each year and over 685 000 deaths annually. Globally, expenditure on breast cancer is estimated to exceed US$2 trillion by 2050. Most women with breast cancer will undergo surgery, with approximately 40% having a mastectomy requiring a hospital stay averaging between 1–2 days. Around 50% of those women will choose to have a breast reconstruction following a mastectomy with implant-based reconstructions requiring an additional average length of stay of 1–2 days. Complications occur in around 10% of breast cancer surgical patients and patients need to be physically and mentally recovered from their surgery, to return to their daily lives, and to prepare for the next phase of their treatment regime, usually chemotherapy and/or radiation therapy. As health services transition towards more patient-centred care models, it is essential to consider the role patients can play in their recovery. As such, best-practice and internationally endorsed Enhanced Recovery After Surgery (ERAS) guidelines include specific patient-managed recommendations in addition to clinician-managed recommendations, to optimise patient recovery outcomes. A 2019 systematic review of ERAS pathways in breast reconstruction suggests that the implementation of ERAS guidelines for patients reduces the length of hospital stay without increasing postoperative complications, decreases opioid use and can improve quality of life (QoL). These findings were also supported by a 2023 retrospective study which investigated outcomes for 92 patients having a single mastectomy. Findings for the 32 patients managed in the ERAS group suggested that implementing ERAS pathways for mastectomy patients (without immediate reconstruction) is associated with a shorter length of hospital stay and a reduction in postoperative complications. The patient-managed recommendations include: Pre-surgery: preadmission optimisation (smoking cessation, achieving a healthy weight, being physically active and alcohol reduction/cessation and provision of information and education). Post-surgery: early mobilisation, rapid resumption of oral feeding and drinking, opioid minimisation and physiotherapy exercises. Post-discharge: home support (supportive care for the management of drains and wounds) and physiotherapy exercises. Non-compliance with ERAS guidelines including early resumption of feeding and postoperative mobilisation, has been associated with higher rates of postoperative complications. Although reports vary, poor adherence to ERAS recommendations is well documented and to date, no studies of adherence to the patient-led ERAS guidelines within patients with breast cancer could be identified. However, a prospective study of 1391 patients undergoing colon surgery reported non-compliance with early feeding and mobilisation in up to 30% of patients which was associated with higher rates of postoperative morbidity. Resource constraints can impede patient adherence to ERAS recommendations through limited communication and collaboration between staff and patients, resistance to change from patients and staff and limited patient education preparing them to take an active role in their recovery. Digital health interventions (DHI) may provide a way of addressing these barriers. DHIs have been shown to be effective in producing health behaviour change in patients with cancer and offer advantages over more traditional forms of support in that they can be tailored, are highly scalable, are cost-effective, and can support patients within and beyond the hospital setting. Evidence also suggests that DHIs have been effective in changing behaviours and managing symptoms such as pain and anxiety in patients with cancer. As such, a DHI may support patients to adhere to the patient-led ERAS recommendations. Given the high prevalence of breast cancer surgery and the limited evidence regarding effective, cost-effective and scalable interventions to enhance patients’ recovery from surgery, a DHI has been developed to increase adherence to the patient-led ERAS recommendations. While previous literature shows that adhering to ERAS recommendations improves clinical outcomes, there is minimal research indicating how patients can be best supported to adhere to these ERAS recommendations. To date, there has been no randomised controlled trial (RCT) evaluating a behavioural intervention to support patients with breast cancer adhering to the comprehensive set of patient-managed ERAS recommendations across the perioperative period. Prior to conducting a fully powered RCT to evaluate the effectiveness of this intervention, this pilot RCT will be used to determine the acceptability of the DHI and the feasibility of components of the research design. Objectives The primary aim of the trial is to assess the acceptability of the RecoverEsupport intervention to support patients in recovering from breast cancer surgery (mastectomy with or without reconstruction or reconstruction following a previous mastectomy) and to assess the feasibility of conducting a fully powered RCT. The secondary trial aims are to assess preliminary efficacy and cost-effectiveness, specifically looking at the estimate of the variability of the RecoverEsupport treatment effect on the length of hospital stay (assessed via medical records). Other outcomes to be assessed include patient behaviours related to patient-managed ERAS recommendations, quality of recovery, anxiety, QoL, health service utilisation post-discharge (assessed at 90 days postoperatively), hospital readmissions and emergency department presentations. Participants will also have the option to participate in an optional interview to identify relevant themes relating to their experience of RecoverEsupport. Trial design The trial design is a two-armed pilot RCT with participants randomly allocated to receive (1) control: usual perioperative care; or (2) intervention: usual care plus access to RecoverEsupport, an online programme to support patients recovering from their surgery. Outcome measures will be assessed in-hospital postsurgery and at 1 and 3 months post-surgery. This paper outlines the trial protocol based on the Standard Protocol Items: Recommendations for Interventional Trials (SPIRIT) recommendations. A completed SPIRIT checklist is available (see ). The trial was prospectively registered with the Australian New Zealand Clinical Trial Registry ACTRN12624000417583 and any future modifications will be added as approved. Study setting The study will take place in the perioperative clinic and breast surgery units at a major cancer hospital in a regional metropolitan area of New South Wales, Australia. The trial will run from July 2024 and will recruit for approximately 18 months. Eligibility criteria Patients are eligible if they are female, aged over 18 years, with either a planned mastectomy for breast cancer (with or without a reconstruction), or a delayed implant-based reconstruction following a previous mastectomy, are English speaking and free from cognitive and emotional impairment (if the recruiting nurse thinks receiving the information about the study at any time will cause the patient undue distress/anxiety), and have internet access and access to an email address. Patients unable to provide independent informed consent; patients having autologous reconstructions (deep inferior epigastric perforator flaps, transverse rectus abdominus myocutaneous flaps, latissimus doris flaps); those who have previously participated in this study; and those who need emergency surgery will be excluded. Recruitment Patients will be invited from the public and private perioperative clinics of the surgeons. Potential participants will be made aware of the study via a flyer. Hospital staff will identify eligible patients scheduled for surgery and will approach them via telephone or in-person (if attending the public clinic) using a recruitment script that was pre-approved by the ethics committee to ensure compliance with ethical standards. A recruitment pack (containing an invitation letter, participant information statement and a link to the online consent form and a reminder 5 days later will be emailed to all interested patients. Hospital staff will also seek patient permission for a member of the research team to contact them to answer any questions and confirm their interest in participating. Where provided, the age and reasons of non-consenting patients will be recorded to examine consent bias. Ethical considerations This study has been approved by the Human Research Ethics Committees of the Hunter New England Local Health District (2022/ETH02010) and the University of Newcastle (H-2023-0298) and the Calvary Mater Newcastle (2022/STE03757) and was prospectively registered. Each invited patient will receive a patient information sheet inviting them to participate in the pilot study and optional qualitative component, assuring them that their participation is entirely voluntary and that any information they provide will remain confidential. No incentive will be offered to patients to participate. All project staff are bound by confidentiality agreements. Collected data will be stored in a non-identifiable format. Randomisation At the completion of the baseline survey or if participants are within 5 days of surgery and have not yet completed the baseline survey, participants will be randomised via a randomisation module within REDCap in a 1:1 ratio in block sizes varying randomly from 4 to 6 to either the intervention or control group. Given the nature of this behavioural intervention, blinding participants to their group allocation will not be possible. However, neither participants nor clinicians will be aware of participants allocation at study enrolment. In addition, intervention participants will be placed into a subgroup analysis comparing two different schedules for receiving postdischarge exercise information: Group A—will have the option of setting their own schedule to receive postdischarge exercise reminders and Group B—will receive reminders according to a preset schedule to explore the impact of different schedules. In order to avoid a potential imbalance in the subgroup due to minimum block sizes and to keep the main study outcomes independent, the intervention subgroups were established via a different method. Inside REDCap, a pseudorandom number generator would allocate each intervention participant a value between 0 (inclusive) and 1 (exclusive). Intervention participants with a value≥0.5 were allocated to Group A and intervention participants with a number <0.5 were allocated to Group B. Usual care All patients will attend a presurgical and/or perioperative appointment (either face-to-face or via telephone) where they may meet with a breast care nurse and the surgeon and/or anaesthetist. The interventions provided postoperatively in the hospital under usual care include standard pain management with the use of medications, standard wound care, physiotherapy exercises and discharge planning and education. All patients who require drains will also receive standard postoperative information and care from a breast care nurse. Patients discharged with drains still in place will receive ongoing care and communication from a community nurse or hospital in the home until the drains are removed. Intervention development The intervention was developed based on an existing intervention for bowel cancer surgical patients and has been adapted for breast cancer surgical patients using iterative feedback from clinicians, consumers and previous breast cancer surgical patients. Theoretical basis for the intervention The COM-B (Capability, Opportunity, Motivation- Behaviour) framework has been used as the theoretical basis for intervention development, a behavioural theory used in both psychology and healthcare to help address barriers related to the capability, opportunity and motivation to engage in specific behaviours. outlines the behaviour change techniques used within the intervention and has been based on evidence-based behaviour change techniques taxonomy. Intervention description: RecoverEsupport Access to the online programme Immediately following randomisation, a personalised alert (SMS/email) and link to the ‘prescription’ letter from a surgeon, will be sent to all intervention participants. The letter will introduce the online programme and will outline the potential benefits of using the programme, provide instructions for access and include contact details of the research team. All participants will be sent two reminders to use the programme, approximately 3 and 7 days later. The online programme will be available for the participant to access on-demand until 3 months postsurgery. Participants will be encouraged to bring a digital device (laptop/smartphone/tablet) to the hospital to access the programme during their stay. Paper copies of the daily checklists will be provided in case participants cannot access the online programme during their hospital admission. The online programme The intervention consists of the following components: information, videos, quiz questions (with real-time feedback provided for incorrect responses), daily checklists (monitoring and feedback) and SMS/email alerts and reminders to improve patient knowledge and motivation to adhere to certain behaviours. The content will cover: Pre-operative support (preparation for surgery including recommended physical activity guidelines, alcohol and smoking reduction strategies, admission procedures), Post-operative support (eg, early mobilisation, early eating and drinking, pain relief, exercises and self-care/psychosocial care (the 5 ‘Recover-Es’) as well as discharge procedures) and Post-discharge support (eg, physiotherapy exercises, wound and drain management and managing follow-up appointments). Strategies to self-manage physical and psychological issues will also be included. The programme content is based on best practice guidelines, evidence from the literature and feedback from the clinical advisory group members, consumers and patients. Online programme delivery The online programme will be delivered via REDCap. REDCap allows participants to access multimedia content via their own devices (computer/smartphone/tablet). The programme will be used to: provide participants with information; collect information from participants; transfer information via the internet; and store information in a secure, central data collection repository. Ongoing participant access Participants will have access to the online programme presurgery and postsurgery, including during their hospital admission and up until 3 months post-surgery. They will receive a printed reminder in their hospital room to access the programme. Intervention participants will be asked to complete a brief online checklist each day (daily checklists) to assess the extent to which they are following recommended self-management strategies to aid recovery (eg, early mobilisation following surgery). Responses that indicate non-adherence to recommendations will be flagged with the breast cancer nurses who will follow up as required. Outcomes Participants will complete surveys at four time points: baseline, post-surgery (in hospital) and 1 and 3 months post-surgery. The baseline survey is sent following consent and if not completed, reminders will be sent approximately 3 and 6 days after the initial invitation. The same reminder procedure used at baseline will also be used for all subsequent surveys. The post-surgery survey is a brief online assessment that will be completed on Day 2 post-surgery. Printed copies will be made available for participants who cannot (or do not) access the in-hospital survey online. Participants will be sent SMS/email reminders to complete each survey within the next 48 hours (the next 24 hours for the in-hospital survey). Primary trial outcomes Feasibility of the RecoverEsupport intervention will be assessed based on whether the following prespecified targets are met: Participant recruitment (Target: n≥70 participants consent to the study). Retention rate (Target: ≥85% remain in the study at 1-month follow-up (ie, have not withdrawn). Data collection (Target: ≥85% of participants from the study have length of stay data collected). Adverse events assessed at 1-month follow-up (Target: no adverse events classified as grade 3 or above, based on items adapted from the Common Terminology Criteria for Adverse Events (V.5.0) including falls, muscle pain/discomfort and anxiety). This is a widely used measure for the reporting of adverse events whereby events are graded on a scale from 1 to 5 with higher grades indicating more severe adverse events. Acceptability of the RecoverEsupport intervention will be assessed among intervention participants only and will be based on the following: Intervention usability (Target: average score on the System Usability Scale>68, ie, ‘Okay’ or higher), measured at 1-month postsurgery. The System Usability Scale is a widely used standardised, brief 10-item scale to assess intervention usability that has good validity and reliability. Total scores range from 0 to 100 with scores>68 classed as satisfactory. Intervention engagement: use of RecoverEsupport, measured at 1 month and 3 months postsurgery. Use of the online programme will be monitored through analytics automatically recorded by REDCap. (Target: ≥75% of participants logged onto RecoverEsupport at least once). Intervention component acceptability: a series of questions using Likert scale response options will assess the acceptability of characteristics of the intervention components, measured at 1 month postsurgery. Questions (Likert scale) will assess the ease of use, relevance and quality of the support and information accessed. Participants will also be asked if they would recommend the programme to other people having surgery. (Target: ≥75% of participants would recommend RecoverEsupport). Secondary outcomes Preliminary efficacy Length of stay will be calculated as the date of discharge less the date of admission, based on information extracted from the patients’ medical records. Patient ERAS knowledge and behaviours (measured postoperatively in-hospital) Patient behaviours related to patient-managed ERAS recommendations (preadmission behaviours and mobilisation, oral diet, fluid intake, opioid minimisation and physio exercises) will be assessed via questions developed specifically for the study. Quality of Recovery (in-hospital, 1 month and 3 months follow-up) The QoR-15 is a validated and reliable tool that assesses the early post-operative health status of surgical patients. The sum of all scores ranges from 0 to 150 with higher scores indicating a better quality of recovery. Anxiety (baseline, 1 month and 3 months follow up) Assessed via the Hospital Anxiety and Depression Scale (HADS), a valid and reliable assessment that measures anxiety for patients in clinical settings. This is a widely used tool which has two scales, one for anxiety and one for depression. The Anxiety scale has 7 items which are summed up to form a score out of 21: 0–7 = normal, 8–10 borderline abnormal, 11–21 abnormal. QoL (baseline, 1 month and 3 months follow-up) Assessed using the European Organization for the Research and Treatment of Cancer Quality of Life Questionnaire (EORTC QLQ-C30 V.3.) The QLQ-C30 is a 30-item cancer-specific instrument that measures 5 functioning domains (physical, role, cognitive, emotional and social), 9 symptom scales (fatigue, nausea/vomiting, pain, dyspnoea, sleep disturbance, appetite loss, constipation, diarrhoea and financial impact), as well as global QoL and provides a summary score. Scores for each scale range from 0 to 100. Higher scores for global health status indicate a high level of functioning and a high QoL. The clinical validity of the scale is high and test–retest reliability is psychometrically credible. Health service utilisation Hospital readmissions, emergency department visits and health service utilisation postdischarge will be assessed using an adapted version of the Client Service Receipt Inventory at 3 months postsurgery. Cost data was collected throughout the duration of the intervention based on detailed project management records, assessed at the conclusion of the study. Nested trial outcomes The average number of completed exercise log entries (as recorded in REDcap) and the average number of completed exercise sessions (as reported by the participant in the 1-month follow-up) will be collected. Other data collected Demographic characteristics (baseline) Participant demographics including age, gender, education, country of birth, home postcode, language spoken at home, marital and employment status, health risk behaviours, internet use and surgical history will be obtained from the medical record, consent form and baseline survey. Medical record data (until 3 months postsurgery) Treatments received and length of stay data will be collected from a medical record audit. Analytics Use of the online programme will be automatically recorded by REDCap. Qualitative Interview At the end of the study, a research assistant will contact participants who indicated that they were willing to take part in an interview about their experience with RecoverEsupport. This is to ensure that data is collected about participants’ broad experiences using the intervention, to ensure that key aspects of their experience are not missed and to supplement the quantitative outcomes reported. This qualitative research component will be conducted using semistructured interviews to establish the acceptability of RecoverEsupport. The interview will be based on a predetermined discussion guide and will be conducted via phone call and recorded. The data collection schedule is outlined in . Sample size Given the nature of the trial, the sample size was determined based on the number of participants needed to assess the feasibility of the study protocol, obtain measures of study acceptability to inform the decision-making process to proceed to a larger RCT and provide relevant information on adverse events. A sample of 35 participants per experimental arm (approximately 70 in total) is the enrolment target for this study with similar studies reporting sample sizes between 30 and 36. Based on patient volume at the hospital, it is anticipated that recruitment will take 18 months. This study is not powered to detect significant differences in outcomes but to perform an exploratory analysis only. It is estimated that 10 participants will be required to identify relevant themes for the optional qualitative interviews. Analysis details the criteria for advancing to a larger trial. The criteria has been established based on recommendations for progression criteria for pilot trials and feasibility studies and will be used by the study team to evaluate whether a larger trial is warranted and to inform any required changes to the protocol. Demographic and treatment/disease characteristics : Baseline characteristics of the treatment groups will be described using descriptive statistics. Secondary outcomes To explore the preliminary efficacy of the intervention to inform a fully powered RCT, between group differences in length of stay will be examined. Analyses will be conducted on an intention-to-treat basis with the Bayesian framework using non-informative priors. A per-protocol analysis will also be conducted and exploratory subgroup analyses will be conducted to explore the influence of participant age. Other secondary outcomes will be modelled using regression with adjustment for baseline values and distribution where appropriate. Nested trial Among participants randomised to receive the RecoverEsupport intervention, regression analysis will be conducted to investigate the between-group difference (A vs B) in the average number of completed exercise log entries and the average number of completed exercise sessions. Economic analysis Subject to the assessment of feasibility, acceptability and preliminary efficacy, a trial-based economic evaluation involving costing, cost-consequence and cost-effectiveness analysis will be conducted. The analysis will compare the RecoverEsupport intervention against the control group (usual care) from a health service perspective. Resource use will be identified and measured for the intervention implementation. It will be assumed that the resource use in the usual care group is zero. The incremental cost-effectiveness ratio will be calculated as the between-group difference in mean total implementation cost divided by the observed between group difference in the length of hospital stay. Sensitivity and scenario analysis will be undertaken to test the impact of changing key design features of the intervention. Qualitative interviews Participant interviews will be transcribed and a thematic analysis conducted to identify emerging themes. Implications This study will gather evidence through the rigorous testing of the trial protocol, randomisation and recruitment processes, and the intervention itself, to identify any required amendments to inform the design of a future fully powered RCT. This will contribute to the evidence base for strategies to support patient self-management during the perioperative period for patients with breast cancer. The behavioural strategies used in the RecoverEsupport DHI are grounded in theory and experimental literature and are intended to support patients to take an active role in managing their preparation and recovery from surgery. The intervention has been designed to maximise its potential for adoption; and because key components of the intervention are online, it can be centrally managed and customised for different health services at relatively low cost. The use of digital technology has the potential to make cost-efficient use of scarce healthcare resources while providing personalised information and support for surgical patients. This technology can be readily integrated into routine care. While this study focuses on breast cancer surgery, the principles underpinning the intervention can be readily adapted to other types of surgery (both cancer and non-cancer). A limitation of this research is that the intervention has only been developed for English speakers. Should the intervention be established as feasible, acceptable and efficacious, it could be adapted for other patient groups including non-English speaking and hard-to-reach populations. With surgery being one of the most common treatment options for breast cancer, with between 80% and 96% of patients with breast cancer undergoing surgery, this intervention represents a potential opportunity to improve patient recovery outcomes while improving the efficiency of care. The primary aim of the trial is to assess the acceptability of the RecoverEsupport intervention to support patients in recovering from breast cancer surgery (mastectomy with or without reconstruction or reconstruction following a previous mastectomy) and to assess the feasibility of conducting a fully powered RCT. The secondary trial aims are to assess preliminary efficacy and cost-effectiveness, specifically looking at the estimate of the variability of the RecoverEsupport treatment effect on the length of hospital stay (assessed via medical records). Other outcomes to be assessed include patient behaviours related to patient-managed ERAS recommendations, quality of recovery, anxiety, QoL, health service utilisation post-discharge (assessed at 90 days postoperatively), hospital readmissions and emergency department presentations. Participants will also have the option to participate in an optional interview to identify relevant themes relating to their experience of RecoverEsupport. The trial design is a two-armed pilot RCT with participants randomly allocated to receive (1) control: usual perioperative care; or (2) intervention: usual care plus access to RecoverEsupport, an online programme to support patients recovering from their surgery. Outcome measures will be assessed in-hospital postsurgery and at 1 and 3 months post-surgery. This paper outlines the trial protocol based on the Standard Protocol Items: Recommendations for Interventional Trials (SPIRIT) recommendations. A completed SPIRIT checklist is available (see ). The trial was prospectively registered with the Australian New Zealand Clinical Trial Registry ACTRN12624000417583 and any future modifications will be added as approved. The study will take place in the perioperative clinic and breast surgery units at a major cancer hospital in a regional metropolitan area of New South Wales, Australia. The trial will run from July 2024 and will recruit for approximately 18 months. Patients are eligible if they are female, aged over 18 years, with either a planned mastectomy for breast cancer (with or without a reconstruction), or a delayed implant-based reconstruction following a previous mastectomy, are English speaking and free from cognitive and emotional impairment (if the recruiting nurse thinks receiving the information about the study at any time will cause the patient undue distress/anxiety), and have internet access and access to an email address. Patients unable to provide independent informed consent; patients having autologous reconstructions (deep inferior epigastric perforator flaps, transverse rectus abdominus myocutaneous flaps, latissimus doris flaps); those who have previously participated in this study; and those who need emergency surgery will be excluded. Patients will be invited from the public and private perioperative clinics of the surgeons. Potential participants will be made aware of the study via a flyer. Hospital staff will identify eligible patients scheduled for surgery and will approach them via telephone or in-person (if attending the public clinic) using a recruitment script that was pre-approved by the ethics committee to ensure compliance with ethical standards. A recruitment pack (containing an invitation letter, participant information statement and a link to the online consent form and a reminder 5 days later will be emailed to all interested patients. Hospital staff will also seek patient permission for a member of the research team to contact them to answer any questions and confirm their interest in participating. Where provided, the age and reasons of non-consenting patients will be recorded to examine consent bias. This study has been approved by the Human Research Ethics Committees of the Hunter New England Local Health District (2022/ETH02010) and the University of Newcastle (H-2023-0298) and the Calvary Mater Newcastle (2022/STE03757) and was prospectively registered. Each invited patient will receive a patient information sheet inviting them to participate in the pilot study and optional qualitative component, assuring them that their participation is entirely voluntary and that any information they provide will remain confidential. No incentive will be offered to patients to participate. All project staff are bound by confidentiality agreements. Collected data will be stored in a non-identifiable format. At the completion of the baseline survey or if participants are within 5 days of surgery and have not yet completed the baseline survey, participants will be randomised via a randomisation module within REDCap in a 1:1 ratio in block sizes varying randomly from 4 to 6 to either the intervention or control group. Given the nature of this behavioural intervention, blinding participants to their group allocation will not be possible. However, neither participants nor clinicians will be aware of participants allocation at study enrolment. In addition, intervention participants will be placed into a subgroup analysis comparing two different schedules for receiving postdischarge exercise information: Group A—will have the option of setting their own schedule to receive postdischarge exercise reminders and Group B—will receive reminders according to a preset schedule to explore the impact of different schedules. In order to avoid a potential imbalance in the subgroup due to minimum block sizes and to keep the main study outcomes independent, the intervention subgroups were established via a different method. Inside REDCap, a pseudorandom number generator would allocate each intervention participant a value between 0 (inclusive) and 1 (exclusive). Intervention participants with a value≥0.5 were allocated to Group A and intervention participants with a number <0.5 were allocated to Group B. All patients will attend a presurgical and/or perioperative appointment (either face-to-face or via telephone) where they may meet with a breast care nurse and the surgeon and/or anaesthetist. The interventions provided postoperatively in the hospital under usual care include standard pain management with the use of medications, standard wound care, physiotherapy exercises and discharge planning and education. All patients who require drains will also receive standard postoperative information and care from a breast care nurse. Patients discharged with drains still in place will receive ongoing care and communication from a community nurse or hospital in the home until the drains are removed. Intervention development The intervention was developed based on an existing intervention for bowel cancer surgical patients and has been adapted for breast cancer surgical patients using iterative feedback from clinicians, consumers and previous breast cancer surgical patients. Theoretical basis for the intervention The COM-B (Capability, Opportunity, Motivation- Behaviour) framework has been used as the theoretical basis for intervention development, a behavioural theory used in both psychology and healthcare to help address barriers related to the capability, opportunity and motivation to engage in specific behaviours. outlines the behaviour change techniques used within the intervention and has been based on evidence-based behaviour change techniques taxonomy. The intervention was developed based on an existing intervention for bowel cancer surgical patients and has been adapted for breast cancer surgical patients using iterative feedback from clinicians, consumers and previous breast cancer surgical patients. The COM-B (Capability, Opportunity, Motivation- Behaviour) framework has been used as the theoretical basis for intervention development, a behavioural theory used in both psychology and healthcare to help address barriers related to the capability, opportunity and motivation to engage in specific behaviours. outlines the behaviour change techniques used within the intervention and has been based on evidence-based behaviour change techniques taxonomy. Access to the online programme Immediately following randomisation, a personalised alert (SMS/email) and link to the ‘prescription’ letter from a surgeon, will be sent to all intervention participants. The letter will introduce the online programme and will outline the potential benefits of using the programme, provide instructions for access and include contact details of the research team. All participants will be sent two reminders to use the programme, approximately 3 and 7 days later. The online programme will be available for the participant to access on-demand until 3 months postsurgery. Participants will be encouraged to bring a digital device (laptop/smartphone/tablet) to the hospital to access the programme during their stay. Paper copies of the daily checklists will be provided in case participants cannot access the online programme during their hospital admission. The online programme The intervention consists of the following components: information, videos, quiz questions (with real-time feedback provided for incorrect responses), daily checklists (monitoring and feedback) and SMS/email alerts and reminders to improve patient knowledge and motivation to adhere to certain behaviours. The content will cover: Pre-operative support (preparation for surgery including recommended physical activity guidelines, alcohol and smoking reduction strategies, admission procedures), Post-operative support (eg, early mobilisation, early eating and drinking, pain relief, exercises and self-care/psychosocial care (the 5 ‘Recover-Es’) as well as discharge procedures) and Post-discharge support (eg, physiotherapy exercises, wound and drain management and managing follow-up appointments). Strategies to self-manage physical and psychological issues will also be included. The programme content is based on best practice guidelines, evidence from the literature and feedback from the clinical advisory group members, consumers and patients. Online programme delivery The online programme will be delivered via REDCap. REDCap allows participants to access multimedia content via their own devices (computer/smartphone/tablet). The programme will be used to: provide participants with information; collect information from participants; transfer information via the internet; and store information in a secure, central data collection repository. Ongoing participant access Participants will have access to the online programme presurgery and postsurgery, including during their hospital admission and up until 3 months post-surgery. They will receive a printed reminder in their hospital room to access the programme. Intervention participants will be asked to complete a brief online checklist each day (daily checklists) to assess the extent to which they are following recommended self-management strategies to aid recovery (eg, early mobilisation following surgery). Responses that indicate non-adherence to recommendations will be flagged with the breast cancer nurses who will follow up as required. Immediately following randomisation, a personalised alert (SMS/email) and link to the ‘prescription’ letter from a surgeon, will be sent to all intervention participants. The letter will introduce the online programme and will outline the potential benefits of using the programme, provide instructions for access and include contact details of the research team. All participants will be sent two reminders to use the programme, approximately 3 and 7 days later. The online programme will be available for the participant to access on-demand until 3 months postsurgery. Participants will be encouraged to bring a digital device (laptop/smartphone/tablet) to the hospital to access the programme during their stay. Paper copies of the daily checklists will be provided in case participants cannot access the online programme during their hospital admission. The intervention consists of the following components: information, videos, quiz questions (with real-time feedback provided for incorrect responses), daily checklists (monitoring and feedback) and SMS/email alerts and reminders to improve patient knowledge and motivation to adhere to certain behaviours. The content will cover: Pre-operative support (preparation for surgery including recommended physical activity guidelines, alcohol and smoking reduction strategies, admission procedures), Post-operative support (eg, early mobilisation, early eating and drinking, pain relief, exercises and self-care/psychosocial care (the 5 ‘Recover-Es’) as well as discharge procedures) and Post-discharge support (eg, physiotherapy exercises, wound and drain management and managing follow-up appointments). Strategies to self-manage physical and psychological issues will also be included. The programme content is based on best practice guidelines, evidence from the literature and feedback from the clinical advisory group members, consumers and patients. The online programme will be delivered via REDCap. REDCap allows participants to access multimedia content via their own devices (computer/smartphone/tablet). The programme will be used to: provide participants with information; collect information from participants; transfer information via the internet; and store information in a secure, central data collection repository. Participants will have access to the online programme presurgery and postsurgery, including during their hospital admission and up until 3 months post-surgery. They will receive a printed reminder in their hospital room to access the programme. Intervention participants will be asked to complete a brief online checklist each day (daily checklists) to assess the extent to which they are following recommended self-management strategies to aid recovery (eg, early mobilisation following surgery). Responses that indicate non-adherence to recommendations will be flagged with the breast cancer nurses who will follow up as required. Participants will complete surveys at four time points: baseline, post-surgery (in hospital) and 1 and 3 months post-surgery. The baseline survey is sent following consent and if not completed, reminders will be sent approximately 3 and 6 days after the initial invitation. The same reminder procedure used at baseline will also be used for all subsequent surveys. The post-surgery survey is a brief online assessment that will be completed on Day 2 post-surgery. Printed copies will be made available for participants who cannot (or do not) access the in-hospital survey online. Participants will be sent SMS/email reminders to complete each survey within the next 48 hours (the next 24 hours for the in-hospital survey). Feasibility of the RecoverEsupport intervention will be assessed based on whether the following prespecified targets are met: Participant recruitment (Target: n≥70 participants consent to the study). Retention rate (Target: ≥85% remain in the study at 1-month follow-up (ie, have not withdrawn). Data collection (Target: ≥85% of participants from the study have length of stay data collected). Adverse events assessed at 1-month follow-up (Target: no adverse events classified as grade 3 or above, based on items adapted from the Common Terminology Criteria for Adverse Events (V.5.0) including falls, muscle pain/discomfort and anxiety). This is a widely used measure for the reporting of adverse events whereby events are graded on a scale from 1 to 5 with higher grades indicating more severe adverse events. Acceptability of the RecoverEsupport intervention will be assessed among intervention participants only and will be based on the following: Intervention usability (Target: average score on the System Usability Scale>68, ie, ‘Okay’ or higher), measured at 1-month postsurgery. The System Usability Scale is a widely used standardised, brief 10-item scale to assess intervention usability that has good validity and reliability. Total scores range from 0 to 100 with scores>68 classed as satisfactory. Intervention engagement: use of RecoverEsupport, measured at 1 month and 3 months postsurgery. Use of the online programme will be monitored through analytics automatically recorded by REDCap. (Target: ≥75% of participants logged onto RecoverEsupport at least once). Intervention component acceptability: a series of questions using Likert scale response options will assess the acceptability of characteristics of the intervention components, measured at 1 month postsurgery. Questions (Likert scale) will assess the ease of use, relevance and quality of the support and information accessed. Participants will also be asked if they would recommend the programme to other people having surgery. (Target: ≥75% of participants would recommend RecoverEsupport). Preliminary efficacy Length of stay will be calculated as the date of discharge less the date of admission, based on information extracted from the patients’ medical records. Patient ERAS knowledge and behaviours (measured postoperatively in-hospital) Patient behaviours related to patient-managed ERAS recommendations (preadmission behaviours and mobilisation, oral diet, fluid intake, opioid minimisation and physio exercises) will be assessed via questions developed specifically for the study. Quality of Recovery (in-hospital, 1 month and 3 months follow-up) The QoR-15 is a validated and reliable tool that assesses the early post-operative health status of surgical patients. The sum of all scores ranges from 0 to 150 with higher scores indicating a better quality of recovery. Anxiety (baseline, 1 month and 3 months follow up) Assessed via the Hospital Anxiety and Depression Scale (HADS), a valid and reliable assessment that measures anxiety for patients in clinical settings. This is a widely used tool which has two scales, one for anxiety and one for depression. The Anxiety scale has 7 items which are summed up to form a score out of 21: 0–7 = normal, 8–10 borderline abnormal, 11–21 abnormal. QoL (baseline, 1 month and 3 months follow-up) Assessed using the European Organization for the Research and Treatment of Cancer Quality of Life Questionnaire (EORTC QLQ-C30 V.3.) The QLQ-C30 is a 30-item cancer-specific instrument that measures 5 functioning domains (physical, role, cognitive, emotional and social), 9 symptom scales (fatigue, nausea/vomiting, pain, dyspnoea, sleep disturbance, appetite loss, constipation, diarrhoea and financial impact), as well as global QoL and provides a summary score. Scores for each scale range from 0 to 100. Higher scores for global health status indicate a high level of functioning and a high QoL. The clinical validity of the scale is high and test–retest reliability is psychometrically credible. Health service utilisation Hospital readmissions, emergency department visits and health service utilisation postdischarge will be assessed using an adapted version of the Client Service Receipt Inventory at 3 months postsurgery. Cost data was collected throughout the duration of the intervention based on detailed project management records, assessed at the conclusion of the study. Nested trial outcomes The average number of completed exercise log entries (as recorded in REDcap) and the average number of completed exercise sessions (as reported by the participant in the 1-month follow-up) will be collected. Length of stay will be calculated as the date of discharge less the date of admission, based on information extracted from the patients’ medical records. Patient behaviours related to patient-managed ERAS recommendations (preadmission behaviours and mobilisation, oral diet, fluid intake, opioid minimisation and physio exercises) will be assessed via questions developed specifically for the study. The QoR-15 is a validated and reliable tool that assesses the early post-operative health status of surgical patients. The sum of all scores ranges from 0 to 150 with higher scores indicating a better quality of recovery. Assessed via the Hospital Anxiety and Depression Scale (HADS), a valid and reliable assessment that measures anxiety for patients in clinical settings. This is a widely used tool which has two scales, one for anxiety and one for depression. The Anxiety scale has 7 items which are summed up to form a score out of 21: 0–7 = normal, 8–10 borderline abnormal, 11–21 abnormal. Assessed using the European Organization for the Research and Treatment of Cancer Quality of Life Questionnaire (EORTC QLQ-C30 V.3.) The QLQ-C30 is a 30-item cancer-specific instrument that measures 5 functioning domains (physical, role, cognitive, emotional and social), 9 symptom scales (fatigue, nausea/vomiting, pain, dyspnoea, sleep disturbance, appetite loss, constipation, diarrhoea and financial impact), as well as global QoL and provides a summary score. Scores for each scale range from 0 to 100. Higher scores for global health status indicate a high level of functioning and a high QoL. The clinical validity of the scale is high and test–retest reliability is psychometrically credible. Hospital readmissions, emergency department visits and health service utilisation postdischarge will be assessed using an adapted version of the Client Service Receipt Inventory at 3 months postsurgery. Cost data was collected throughout the duration of the intervention based on detailed project management records, assessed at the conclusion of the study. Nested trial outcomes The average number of completed exercise log entries (as recorded in REDcap) and the average number of completed exercise sessions (as reported by the participant in the 1-month follow-up) will be collected. The average number of completed exercise log entries (as recorded in REDcap) and the average number of completed exercise sessions (as reported by the participant in the 1-month follow-up) will be collected. Demographic characteristics (baseline) Participant demographics including age, gender, education, country of birth, home postcode, language spoken at home, marital and employment status, health risk behaviours, internet use and surgical history will be obtained from the medical record, consent form and baseline survey. Medical record data (until 3 months postsurgery) Treatments received and length of stay data will be collected from a medical record audit. Analytics Use of the online programme will be automatically recorded by REDCap. Qualitative Interview At the end of the study, a research assistant will contact participants who indicated that they were willing to take part in an interview about their experience with RecoverEsupport. This is to ensure that data is collected about participants’ broad experiences using the intervention, to ensure that key aspects of their experience are not missed and to supplement the quantitative outcomes reported. This qualitative research component will be conducted using semistructured interviews to establish the acceptability of RecoverEsupport. The interview will be based on a predetermined discussion guide and will be conducted via phone call and recorded. The data collection schedule is outlined in . Participant demographics including age, gender, education, country of birth, home postcode, language spoken at home, marital and employment status, health risk behaviours, internet use and surgical history will be obtained from the medical record, consent form and baseline survey. Treatments received and length of stay data will be collected from a medical record audit. Use of the online programme will be automatically recorded by REDCap. At the end of the study, a research assistant will contact participants who indicated that they were willing to take part in an interview about their experience with RecoverEsupport. This is to ensure that data is collected about participants’ broad experiences using the intervention, to ensure that key aspects of their experience are not missed and to supplement the quantitative outcomes reported. This qualitative research component will be conducted using semistructured interviews to establish the acceptability of RecoverEsupport. The interview will be based on a predetermined discussion guide and will be conducted via phone call and recorded. The data collection schedule is outlined in . Given the nature of the trial, the sample size was determined based on the number of participants needed to assess the feasibility of the study protocol, obtain measures of study acceptability to inform the decision-making process to proceed to a larger RCT and provide relevant information on adverse events. A sample of 35 participants per experimental arm (approximately 70 in total) is the enrolment target for this study with similar studies reporting sample sizes between 30 and 36. Based on patient volume at the hospital, it is anticipated that recruitment will take 18 months. This study is not powered to detect significant differences in outcomes but to perform an exploratory analysis only. It is estimated that 10 participants will be required to identify relevant themes for the optional qualitative interviews. details the criteria for advancing to a larger trial. The criteria has been established based on recommendations for progression criteria for pilot trials and feasibility studies and will be used by the study team to evaluate whether a larger trial is warranted and to inform any required changes to the protocol. Demographic and treatment/disease characteristics : Baseline characteristics of the treatment groups will be described using descriptive statistics. To explore the preliminary efficacy of the intervention to inform a fully powered RCT, between group differences in length of stay will be examined. Analyses will be conducted on an intention-to-treat basis with the Bayesian framework using non-informative priors. A per-protocol analysis will also be conducted and exploratory subgroup analyses will be conducted to explore the influence of participant age. Other secondary outcomes will be modelled using regression with adjustment for baseline values and distribution where appropriate. Among participants randomised to receive the RecoverEsupport intervention, regression analysis will be conducted to investigate the between-group difference (A vs B) in the average number of completed exercise log entries and the average number of completed exercise sessions. Subject to the assessment of feasibility, acceptability and preliminary efficacy, a trial-based economic evaluation involving costing, cost-consequence and cost-effectiveness analysis will be conducted. The analysis will compare the RecoverEsupport intervention against the control group (usual care) from a health service perspective. Resource use will be identified and measured for the intervention implementation. It will be assumed that the resource use in the usual care group is zero. The incremental cost-effectiveness ratio will be calculated as the between-group difference in mean total implementation cost divided by the observed between group difference in the length of hospital stay. Sensitivity and scenario analysis will be undertaken to test the impact of changing key design features of the intervention. Participant interviews will be transcribed and a thematic analysis conducted to identify emerging themes. This study will gather evidence through the rigorous testing of the trial protocol, randomisation and recruitment processes, and the intervention itself, to identify any required amendments to inform the design of a future fully powered RCT. This will contribute to the evidence base for strategies to support patient self-management during the perioperative period for patients with breast cancer. The behavioural strategies used in the RecoverEsupport DHI are grounded in theory and experimental literature and are intended to support patients to take an active role in managing their preparation and recovery from surgery. The intervention has been designed to maximise its potential for adoption; and because key components of the intervention are online, it can be centrally managed and customised for different health services at relatively low cost. The use of digital technology has the potential to make cost-efficient use of scarce healthcare resources while providing personalised information and support for surgical patients. This technology can be readily integrated into routine care. While this study focuses on breast cancer surgery, the principles underpinning the intervention can be readily adapted to other types of surgery (both cancer and non-cancer). A limitation of this research is that the intervention has only been developed for English speakers. Should the intervention be established as feasible, acceptable and efficacious, it could be adapted for other patient groups including non-English speaking and hard-to-reach populations. With surgery being one of the most common treatment options for breast cancer, with between 80% and 96% of patients with breast cancer undergoing surgery, this intervention represents a potential opportunity to improve patient recovery outcomes while improving the efficiency of care. 10.1136/bmjopen-2024-093869 online supplemental file 1 10.1136/bmjopen-2024-093869 online supplemental file 2 |
Barriers and Facilitators to Ophthalmology Visit Adherence in an Urban Hospital Setting | cee4ddc4-8399-4992-af99-63747e487cb0 | 10587857 | Ophthalmology[mh] | More than 93 million people in the United States were at high risk for vision loss in 2017, a 43% increase since 2002. Although screening for refractive error and early eye disease could prevent unnecessary vision loss or blindness, , less than 60% of adults at high risk of vision loss reported receiving eye care in 2017. Minorities and individuals with limited socioeconomic resources have more barriers to care, resulting in the underuse of eye care and an increased risk of vision loss. – Because the major causes of visual impairment and blindness are treatable, adherence to eye care services that detect and manage vision loss should be a national health priority. – The University of Illinois’ Hospital and Health system serves residents of communities who primarily identify as racial and ethnic minorities. These communities have higher rates of unemployment, large numbers of uninsured individuals, and greater levels of poverty than in Illinois and the United States. We previously applied the 2018 Centers for Disease Control and Prevention Social Vulnerability Index (SVI) to examine the association between neighborhood-level social vulnerability and adherence to scheduled ophthalmology office visits within our diverse urban hospital setting. Briefly, the SVI has been used since 2000 and has demonstrated the effects of neighborhood-level social vulnerability on individual patient access and outcomes. – We previously reported that nonadherence to attending ophthalmology visits in our health system was associated with higher SVI scores after controlling for demographic variables, new or established appointment type, and distance from clinic. Although our previous work may assist in identifying patients at risk of nonadherence, it cannot explain why a patient cannot adhere to a scheduled appointment or what specific interventions might be most helpful in promoting adherence. Such information is needed to guide the design and evaluation of interventions to improve the accessibility of eye care. Qualitative studies provide the opportunity to gain insights into how patients interact with the health care system. The widely used Consolidated Framework for Implementation Research (CFIR) comprises 39 constructs within 5 major domains: outer setting (e.g., patient needs and external policies), inner setting (e.g., culture and implementation climate), characteristics of individuals (e.g., self-efficacy), implementation process (e.g., strategies for planning, execution, and evaluation), and intervention characteristics. The CFIR is also an adaptable framework and can be customized for use in individual situations . , The use of the CFIR in this work provides a heuristic to understand patient interactions in a more systematic way. Similarly, human-centered design (HCD) provides frameworks for characterizing elements of a health care delivery context. The objective of this study was, therefore, to use qualitative methods to engage patients from neighborhoods with high social vulnerability and their ophthalmology health care providers to identify barriers and facilitators to completing outpatient appointments for eye care and to situate these findings into CFIR, which would provide useful guidance for future interventions. This study had University of Illinois at Chicago Institutional Review Board approval (protocol 2022–0484) and adhered to the tenets of the Declaration of Helsinki. A research team with expertise in HCD (D.N., R.M.S., and H.M.), public health (D.N. and A.C.S.), and clinical care (A.C.S.) was established to conduct interviews and analysis. Setting Participants were recruited from the General Eye Clinic (GEC). The GEC serves a patient population that is 44% non-Hispanic Black, 29% Hispanic, and 11% non-Hispanic White, and more than one-half of patients have Medicaid insurance. This site was selected because it has the highest nonadherence rate in the department and most patients come from neighborhoods with high social vulnerability. Recruitment and Eligibility Patients Individuals 18 years and older with an SVI of greater than 0.61 who were English speaking were eligible to participate in this study. Briefly, the SVI is a composite measure representing neighborhood relative vulnerability compared to all other communities nationally by census tract. U.S. Census data are used to create a percentile rank using 15 social factors organized into 4 themes (socioeconomic status, household composition/disability, minority status/language, and housing type/transportation). Percentile ranking values range from 0 to 1, with 1 representing the 100th percentile for extreme social vulnerability. Residential addresses were geocoded using ArcGIS, a geographic information systems software, to append the publicly available neighborhood-level SVI to each individual patient. A classification tree using binary recursive partitioning was used to determine the optimal cutoff of an SVI of 0.61 using the Youden index to prioritize both specificity and sensitivity. This cutoff provides a misclassification rate of 28% with an area under the curve of 0.62. In this study, adherence was defined as completing the scheduled ophthalmology visit during the enrollment period. We aimed to recruit 50% of patients who adhered to a scheduled eye visit during the recruitment period and 50% who did not. We used purposeful sampling, which is widely used in qualitative research to identify and select “information-rich cases for the most effective use of limited resources.” , Subject eligibility was determined using multiple methods: (1) A report of patients (adherent and nonadherent) scheduled in the GEC between June 6 and 13, 2022, was pulled from the electronic health record. Eligible patients were mailed a letter informing them of the study. After 10 business days, a member of the study team called patients to determine interest, and interested participants were scheduled for an interview. (2) Additionally, providers recruited patients in the clinic. Eligibility was confirmed by the study team, and patients were called to schedule an interview. Providers GEC providers, including resident as well as attending physicians, were recruited via email. Those who expressed interest were scheduled for an interview. Individuals were recruited on a voluntary basis. Data Collection Patients Demographic characteristics pulled from the medical record included age, gender, race/ethnicity, patient status (new or established), and insurance status. Interviews with patients focused on social context, patient experience of eye-related health care appointments, and possible solutions to improve adherence (see for patient and provider interview guides). After the interview, participants voluntarily completed a demographic survey . Patient interviews were led by D.N. or R.M.S. A second investigator joined each interview to take field notes. Patients received a $40 gift card as compensation for their participation. A chart review of each patient was conducted after the interview to better understand the patient's medical history and journey to seeking care (i.e., disease, visit scheduling). Providers Interviews with providers asked open-ended questions about clinic roles, perceived patient barriers and facilitators to eye visit adherence, and design requirements for future interventions. Provider interviews were led by A.C.S. A second investigator joined each interview to take field notes. Providers were not compensated for their participation. Semistructured interviews were conducted in English in person, by phone, or video platform based on participant preference. Upon obtaining verbal consent, interviews were audio recorded. Informed consent and protected health information were documented in REDCap, a Health Insurance Portability and Accountability Act of 1996–compliant web-based system. Analysis Rapid qualitative analysis is often selected in implementation research to establish provisional assessments of contexts, primary drivers, or considerations of stakeholders. – HCD is an approach to understanding real-world context and behaviors of individuals, engage stakeholders, and rapidly prototype and test solutions. Although historically comprising two distinct fields, combining implementation science and HCD methodologies may have a synergistic effect when conducting qualitative research. We implemented two inductive HCD methods to identify patterns within interview data and group themes into distinct categories: (1) rapid affinity diagramming and (2) modified recursive abstraction. , To further characterize barriers and facilitators to adherence, as well as identify opportunities for intervention, interviews were then analyzed vis à vis CFIR. Affinity diagramming is a method frequently used within HCD in which researchers write down a single observation derived directly from the corresponding interview so that each observation may be considered on its own. Notes are then clustered based on their relatedness to one another and clusters are then reviewed and ascribed themes. A minimum of two team members conducted a debrief immediately after each interview during which they shared observations and discussed emergent findings. Each insight or quote was captured in Miro, a virtual white-boarding platform (Miro, 2022). Upon completion of all interviews, reflections that shared a theme were clustered into categories that researchers defined with a short title. Clusters were then arranged to visualize the salience of each theme and relationships between them. In parallel, interview transcripts were also analyzed using the following steps of a modified recursive abstraction method : (1) organization of all interview responses by question, (2) condensation of responses into synthesized themes, and (3) definition of each theme. In many respects, this process is very similar to other qualitative methods, such as grounded theory. Findings from each analysis method yielded similar themes. In the case of a discrepancy, researchers defaulted to findings from the recursive abstraction method, which could be more directly traced to interview transcripts. These findings were then situated within three domains of the CFIR framework: outer setting, inner setting, and characteristics of individual. Although CFIR is traditionally used to evaluate the readiness of an organization or environment for the implementation of a specific intervention, we used this framework to identify the setting in which an intervention to improve adherence would be most impactful. This novel use of CFIR necessitated slight modifications to the framework; therefore, the research team aligned upon a definition for each domain. Outer setting included economic, political, and social contexts, as well as patient needs and resources. The latter included social support for physical needs (e.g., a ride to appointment) and mental health (e.g., someone to talk to). Inner setting included both ophthalmology department and health system culture (i.e., norms, values, and basic assumptions of the given organization), , availability of resources (i.e., level of resources devoted to implementing an intervention such as money, training, education, physical space, and time), and readiness for implementation of interventions to support adherence. Finally, characteristics of individuals included patient-specific barriers and facilitators to accessing health care, knowledge, self-efficacy, and outcomes expectancy about eye care. Initial analysis of the first three interviews was performed independently by each research member (D.N., R.M.S., and A.C.S.) using the CFIR template to evaluate concordance. Once consensus was achieved, two researchers coded each transcript independently with discrepancies resolved through discussion among all three researchers. Participants were recruited from the General Eye Clinic (GEC). The GEC serves a patient population that is 44% non-Hispanic Black, 29% Hispanic, and 11% non-Hispanic White, and more than one-half of patients have Medicaid insurance. This site was selected because it has the highest nonadherence rate in the department and most patients come from neighborhoods with high social vulnerability. Patients Individuals 18 years and older with an SVI of greater than 0.61 who were English speaking were eligible to participate in this study. Briefly, the SVI is a composite measure representing neighborhood relative vulnerability compared to all other communities nationally by census tract. U.S. Census data are used to create a percentile rank using 15 social factors organized into 4 themes (socioeconomic status, household composition/disability, minority status/language, and housing type/transportation). Percentile ranking values range from 0 to 1, with 1 representing the 100th percentile for extreme social vulnerability. Residential addresses were geocoded using ArcGIS, a geographic information systems software, to append the publicly available neighborhood-level SVI to each individual patient. A classification tree using binary recursive partitioning was used to determine the optimal cutoff of an SVI of 0.61 using the Youden index to prioritize both specificity and sensitivity. This cutoff provides a misclassification rate of 28% with an area under the curve of 0.62. In this study, adherence was defined as completing the scheduled ophthalmology visit during the enrollment period. We aimed to recruit 50% of patients who adhered to a scheduled eye visit during the recruitment period and 50% who did not. We used purposeful sampling, which is widely used in qualitative research to identify and select “information-rich cases for the most effective use of limited resources.” , Subject eligibility was determined using multiple methods: (1) A report of patients (adherent and nonadherent) scheduled in the GEC between June 6 and 13, 2022, was pulled from the electronic health record. Eligible patients were mailed a letter informing them of the study. After 10 business days, a member of the study team called patients to determine interest, and interested participants were scheduled for an interview. (2) Additionally, providers recruited patients in the clinic. Eligibility was confirmed by the study team, and patients were called to schedule an interview. Providers GEC providers, including resident as well as attending physicians, were recruited via email. Those who expressed interest were scheduled for an interview. Individuals were recruited on a voluntary basis. Individuals 18 years and older with an SVI of greater than 0.61 who were English speaking were eligible to participate in this study. Briefly, the SVI is a composite measure representing neighborhood relative vulnerability compared to all other communities nationally by census tract. U.S. Census data are used to create a percentile rank using 15 social factors organized into 4 themes (socioeconomic status, household composition/disability, minority status/language, and housing type/transportation). Percentile ranking values range from 0 to 1, with 1 representing the 100th percentile for extreme social vulnerability. Residential addresses were geocoded using ArcGIS, a geographic information systems software, to append the publicly available neighborhood-level SVI to each individual patient. A classification tree using binary recursive partitioning was used to determine the optimal cutoff of an SVI of 0.61 using the Youden index to prioritize both specificity and sensitivity. This cutoff provides a misclassification rate of 28% with an area under the curve of 0.62. In this study, adherence was defined as completing the scheduled ophthalmology visit during the enrollment period. We aimed to recruit 50% of patients who adhered to a scheduled eye visit during the recruitment period and 50% who did not. We used purposeful sampling, which is widely used in qualitative research to identify and select “information-rich cases for the most effective use of limited resources.” , Subject eligibility was determined using multiple methods: (1) A report of patients (adherent and nonadherent) scheduled in the GEC between June 6 and 13, 2022, was pulled from the electronic health record. Eligible patients were mailed a letter informing them of the study. After 10 business days, a member of the study team called patients to determine interest, and interested participants were scheduled for an interview. (2) Additionally, providers recruited patients in the clinic. Eligibility was confirmed by the study team, and patients were called to schedule an interview. GEC providers, including resident as well as attending physicians, were recruited via email. Those who expressed interest were scheduled for an interview. Individuals were recruited on a voluntary basis. Patients Demographic characteristics pulled from the medical record included age, gender, race/ethnicity, patient status (new or established), and insurance status. Interviews with patients focused on social context, patient experience of eye-related health care appointments, and possible solutions to improve adherence (see for patient and provider interview guides). After the interview, participants voluntarily completed a demographic survey . Patient interviews were led by D.N. or R.M.S. A second investigator joined each interview to take field notes. Patients received a $40 gift card as compensation for their participation. A chart review of each patient was conducted after the interview to better understand the patient's medical history and journey to seeking care (i.e., disease, visit scheduling). Providers Interviews with providers asked open-ended questions about clinic roles, perceived patient barriers and facilitators to eye visit adherence, and design requirements for future interventions. Provider interviews were led by A.C.S. A second investigator joined each interview to take field notes. Providers were not compensated for their participation. Semistructured interviews were conducted in English in person, by phone, or video platform based on participant preference. Upon obtaining verbal consent, interviews were audio recorded. Informed consent and protected health information were documented in REDCap, a Health Insurance Portability and Accountability Act of 1996–compliant web-based system. Demographic characteristics pulled from the medical record included age, gender, race/ethnicity, patient status (new or established), and insurance status. Interviews with patients focused on social context, patient experience of eye-related health care appointments, and possible solutions to improve adherence (see for patient and provider interview guides). After the interview, participants voluntarily completed a demographic survey . Patient interviews were led by D.N. or R.M.S. A second investigator joined each interview to take field notes. Patients received a $40 gift card as compensation for their participation. A chart review of each patient was conducted after the interview to better understand the patient's medical history and journey to seeking care (i.e., disease, visit scheduling). Interviews with providers asked open-ended questions about clinic roles, perceived patient barriers and facilitators to eye visit adherence, and design requirements for future interventions. Provider interviews were led by A.C.S. A second investigator joined each interview to take field notes. Providers were not compensated for their participation. Semistructured interviews were conducted in English in person, by phone, or video platform based on participant preference. Upon obtaining verbal consent, interviews were audio recorded. Informed consent and protected health information were documented in REDCap, a Health Insurance Portability and Accountability Act of 1996–compliant web-based system. Rapid qualitative analysis is often selected in implementation research to establish provisional assessments of contexts, primary drivers, or considerations of stakeholders. – HCD is an approach to understanding real-world context and behaviors of individuals, engage stakeholders, and rapidly prototype and test solutions. Although historically comprising two distinct fields, combining implementation science and HCD methodologies may have a synergistic effect when conducting qualitative research. We implemented two inductive HCD methods to identify patterns within interview data and group themes into distinct categories: (1) rapid affinity diagramming and (2) modified recursive abstraction. , To further characterize barriers and facilitators to adherence, as well as identify opportunities for intervention, interviews were then analyzed vis à vis CFIR. Affinity diagramming is a method frequently used within HCD in which researchers write down a single observation derived directly from the corresponding interview so that each observation may be considered on its own. Notes are then clustered based on their relatedness to one another and clusters are then reviewed and ascribed themes. A minimum of two team members conducted a debrief immediately after each interview during which they shared observations and discussed emergent findings. Each insight or quote was captured in Miro, a virtual white-boarding platform (Miro, 2022). Upon completion of all interviews, reflections that shared a theme were clustered into categories that researchers defined with a short title. Clusters were then arranged to visualize the salience of each theme and relationships between them. In parallel, interview transcripts were also analyzed using the following steps of a modified recursive abstraction method : (1) organization of all interview responses by question, (2) condensation of responses into synthesized themes, and (3) definition of each theme. In many respects, this process is very similar to other qualitative methods, such as grounded theory. Findings from each analysis method yielded similar themes. In the case of a discrepancy, researchers defaulted to findings from the recursive abstraction method, which could be more directly traced to interview transcripts. These findings were then situated within three domains of the CFIR framework: outer setting, inner setting, and characteristics of individual. Although CFIR is traditionally used to evaluate the readiness of an organization or environment for the implementation of a specific intervention, we used this framework to identify the setting in which an intervention to improve adherence would be most impactful. This novel use of CFIR necessitated slight modifications to the framework; therefore, the research team aligned upon a definition for each domain. Outer setting included economic, political, and social contexts, as well as patient needs and resources. The latter included social support for physical needs (e.g., a ride to appointment) and mental health (e.g., someone to talk to). Inner setting included both ophthalmology department and health system culture (i.e., norms, values, and basic assumptions of the given organization), , availability of resources (i.e., level of resources devoted to implementing an intervention such as money, training, education, physical space, and time), and readiness for implementation of interventions to support adherence. Finally, characteristics of individuals included patient-specific barriers and facilitators to accessing health care, knowledge, self-efficacy, and outcomes expectancy about eye care. Initial analysis of the first three interviews was performed independently by each research member (D.N., R.M.S., and A.C.S.) using the CFIR template to evaluate concordance. Once consensus was achieved, two researchers coded each transcript independently with discrepancies resolved through discussion among all three researchers. A total of 17 of 85 patients and 8 of 12 providers invited to participate agreed to join the study. Patient sociodemographic characteristics are summarized in . Of the eight providers, four were attending physicians and four were residents with at least 2 years of experience working in the GEC. Interview length averaged 37 minutes (range, 23–56 minutes) for patients and 27 minutes (range, 21–36 minutes) for providers. There were four clear themes, which included transportation, time burden, economic situation, and social support . Transportation Patients and providers identified transportation as both a facilitator and barrier, fitting under both the outer setting and characteristics of individuals using the CFIR framework . Outer Setting In this population, many patients had the ability to schedule transportation through their health insurance carrier. Some transportation companies available to these patients offered to send a text message or call with reminders before pick up. Although this practice was seen as an overall facilitator to care, several barriers were found to exist. Patients reported that scheduling transportation through insurance carriers was difficult or inconvenient because paperwork was required to determine eligibility, and, for those with questions, few resources were available to help them navigate the paperwork. Additionally, insurance programs usually require at least 3 days advance notice, so patients whose primary mode of transportation fell through were unable to use this program as a backup option. Transportation services through insurance carriers may also be unreliable; multiple patients reported missing at least one visit because the service never arrived for the pickup. One patient stated the inconvenience of scheduling and poor reliability of the transportation service made public transportation his mode of choice. Although visually impaired and reliant on other passengers to inform him when he was at his stop, he elected to spend more than 4 hours round trip (i.e., 30-minute walk, 2 buses, 1 train) to avoid the transportation service provided by his insurance carrier. Although several patients reported living near public transportation, few reported using it as a primary method of getting to their appointments. Additionally, one patient reported avoiding the use of public transit owing to use of a walker. Providers also identified these barriers. Patients who described having a social support system, including friends or family, often relied on their “people” to give them a ride. A social support system in general was seen by patients as a facilitator, but could not solve all transportation-related concerns reliably, as noted further elsewhere in this article. Characteristics of Individual Patients who had the ability to drive themselves with a personal car often described this as a facilitator for adhering to visits. However, the cost of parking was mentioned as a barrier. For example, one patient reported having to refill her parking meter owing to the extended length of her visit. Providers described similar barriers and facilitators as noted elsewhere in this article. One additional barrier reported by providers that was not mentioned by patients was that temporary disability or visual impairment made travel to and from appointments more difficult. Time Burden Inner Setting Patients expressed significant dissatisfaction with wait times in the clinic. Although some patients reported getting in and out in less than 1 hour, others expressed frustration with wait times extending beyond 3 hours. Although this wait time was not a direct barrier, there was an interaction between the effects of wait time and other factors affecting the patient, such as using up all approved paid time off work or issues with transportation. Regarding transportation, some transportation services provided by the patient's insurance had an agreed upon window for pick-up, usually 3 hours, or limited hours of operation. Because transportation services had to be scheduled 3 days in advance, patients reported that if their visit time went over the scheduled pickup time, there was no way to reschedule their transportation. As a result, one patient even reported leaving before being seen by a provider to decrease the risk of being left without transportation after an appointment. Providers acknowledged an interaction between long wait times and other themes, including transportation and paid time off. The GEC is the only dedicated walk-in clinic for immediate eye care in Chicago. Providers offered that a reason for long wait time for scheduled visits was due to the need to, in some cases, prioritize walk-in patients who are frequently coming for urgent and emergent issues. Social Support Outer Setting (Patient Needs and Resources) Patients reported that having a social support system, such as family or friends, served as a facilitator. Specifically, as noted elsewhere in this article, patients relied on others to drive them to and/or attend appointments with them. Although this factor generally served as a facilitator, patients also stated that their ability to attend a visit depended on whether the appointment time and day also worked for the person driving them to the appointment. One patient shared that she missed an appointment owing to a family member's work schedule. Although providers agreed that reliance on another person for transportation to appointments could be a facilitator, they reiterated concerns that it could also be a barrier because it depended on fitting with two individuals’ schedules. Those who brought a companion to their visit also found it helpful for this other person to learn about their eye condition. Several providers described that patients coming with family or friends often served as a facilitator who could assist with reinforcing the care plan and education. However, providers found that caretakers who attended visits with patients coming from facilities such as nursing homes served mostly as transportation to the visit and were less frequently able to provide additional details about the patient's medical history. Inner Setting Some providers reported that early coronavirus disease 2019 policies, which limited visits to the patient only in many cases and have since been lifted, were a barrier to patient education and facilitation of follow-up visits. Several providers felt a responsibility to assist the patient with accessing resources. Examples included providing information on financial assistance or personally facilitating transportation. Another facilitator described by providers was their experience referring patients to social work for additional resources and a trained professional to assist with social support. Many felt that they had an overall positive experience with social work, but saved referrals for patients at highest risk of missing a visit, even though they felt most patients would benefit from this type of referral. One provider specifically noted that a majority of patients seen in the GEC would benefit from a social work referral, but that limited availability of social work caused the provider to have to strategically decide who to refer. The decision of who to refer to social work varied from provider to provider. Economic Situation Outer Setting Most patients interviewed had Medicaid insurance. Medicaid provides visit coverage and offers transportation through contracted services for patients with no other way to get to and from their visits. Although these factors serve as facilitators to care, patients did report that having this insurance type limited where they could get their care as well as their transportation options. Although not a direct barrier to adherence, this factor often affected the distance to a covered provider, creating additional patient barriers. Those reporting working full time discussed that their employer determines their ability to take time off for visits. Although paid time off was available for some patients, many discussed the increased wait time interacting with approved time off. Characteristics of Individual Transportation and economic situation were seen to overlap in several ways. For example, patients expressed cost of parking as a financial concern affecting their decision whether to drive to the clinic. Similarly, as noted in the outer setting, several patients reported their fixed income factoring into transportation options. Motivation Characteristics of Individual Providers often spoke of health literacy, described by providers as the “patient's understanding of diagnosis or disease,” as a motivating factor for attending ophthalmology visits. Most providers who participated in this study discussed an association between health literacy (or lack thereof) as a potential facilitator (and/or barrier) to attending visits; however, most patients who were interviewed did express a motivation to attend their visits for several reasons. Patients reported that their vision or eye health is important to them because it plays a major role in activities of daily living (e.g., taking medications, completing paperwork, caring for family member). Patients also reported that they felt they had a good understanding of the risk of inaction, either from the education they were provided during a visit or from seeing a friend or family member experiencing health complications or vision loss. Moreover, patients showed a high level of self-advocacy and even created workarounds to avoid barriers from the inner and outer settings. Examples included scheduling a later pickup time with the transportation service in case a visit ran late, scheduling visits early in the day to avoid the long wait times in the clinic, or creating a unique personal scheduling system. Two patients reported leaving an outside provider and working with a new insurance company to ensure they could be seen. One patient even reported paying for a copy of their medical records from an outside provider when transferring care to our health system. This self-advocacy was seen across all patients. Additional Themes Several additional factors within the inner setting were found to influence attitudes toward adherence, including scheduling, trust, and care provided. Scheduling (Inner Setting) Patients and providers both expressed concerns with appointment scheduling. Some patients stated that their appointment lead time, the time between when the appointment is scheduled and when it occurs, was short, less than 1 week. Others reported lead times between a few weeks and 5 months. In addition, those who cancelled found it difficult to reschedule a visit in a timely manner. Providers similarly described longer lead times, increasing the risk of no shows. Patients and providers agreed that leaving with a follow-up appointment in hand served as a great facilitator that helped to avoid having to call and schedule visits and decreased the risk of follow-up visits being scheduled incorrectly (e.g., missing additional linked visit for imaging). Patients and providers also reported that reminder calls and messages assisted with adherence; however, both noted that automated calls had less of an effect compared with having a person call, especially in the case of needing to reschedule. Providers also reported that reminder calls work only when the phone number in the system is correct, and they noted this was often not the case. Trust (Characteristics of Individual/Inner Setting) Regarding trust, patients often felt a loyalty to and reported getting all their care within the health system. Trust in the overall health system served as a facilitator and was not mentioned as a barrier. Care Provided (Inner Setting) Finally, patients expressed a very positive experience with the quality of care they received during their visits. Despite some patients reporting long wait times, many patients remarked that the staff were friendly and welcoming, the examination thorough, and doctors took the time to educate them on their condition. Suggested Interventions Providers were asked about potential interventions that could assist in improving adherence to visits. Recommendations varied and encompassed refining patient education (e.g., educational videos, handouts), reducing time burden (e.g., reducing patient volumes, improving efficiency), and addressing patient needs (e.g., counseling visits to address social determinants of health, adequate social work staffing, transportation assistance center). Patients and providers identified transportation as both a facilitator and barrier, fitting under both the outer setting and characteristics of individuals using the CFIR framework . Outer Setting In this population, many patients had the ability to schedule transportation through their health insurance carrier. Some transportation companies available to these patients offered to send a text message or call with reminders before pick up. Although this practice was seen as an overall facilitator to care, several barriers were found to exist. Patients reported that scheduling transportation through insurance carriers was difficult or inconvenient because paperwork was required to determine eligibility, and, for those with questions, few resources were available to help them navigate the paperwork. Additionally, insurance programs usually require at least 3 days advance notice, so patients whose primary mode of transportation fell through were unable to use this program as a backup option. Transportation services through insurance carriers may also be unreliable; multiple patients reported missing at least one visit because the service never arrived for the pickup. One patient stated the inconvenience of scheduling and poor reliability of the transportation service made public transportation his mode of choice. Although visually impaired and reliant on other passengers to inform him when he was at his stop, he elected to spend more than 4 hours round trip (i.e., 30-minute walk, 2 buses, 1 train) to avoid the transportation service provided by his insurance carrier. Although several patients reported living near public transportation, few reported using it as a primary method of getting to their appointments. Additionally, one patient reported avoiding the use of public transit owing to use of a walker. Providers also identified these barriers. Patients who described having a social support system, including friends or family, often relied on their “people” to give them a ride. A social support system in general was seen by patients as a facilitator, but could not solve all transportation-related concerns reliably, as noted further elsewhere in this article. Characteristics of Individual Patients who had the ability to drive themselves with a personal car often described this as a facilitator for adhering to visits. However, the cost of parking was mentioned as a barrier. For example, one patient reported having to refill her parking meter owing to the extended length of her visit. Providers described similar barriers and facilitators as noted elsewhere in this article. One additional barrier reported by providers that was not mentioned by patients was that temporary disability or visual impairment made travel to and from appointments more difficult. In this population, many patients had the ability to schedule transportation through their health insurance carrier. Some transportation companies available to these patients offered to send a text message or call with reminders before pick up. Although this practice was seen as an overall facilitator to care, several barriers were found to exist. Patients reported that scheduling transportation through insurance carriers was difficult or inconvenient because paperwork was required to determine eligibility, and, for those with questions, few resources were available to help them navigate the paperwork. Additionally, insurance programs usually require at least 3 days advance notice, so patients whose primary mode of transportation fell through were unable to use this program as a backup option. Transportation services through insurance carriers may also be unreliable; multiple patients reported missing at least one visit because the service never arrived for the pickup. One patient stated the inconvenience of scheduling and poor reliability of the transportation service made public transportation his mode of choice. Although visually impaired and reliant on other passengers to inform him when he was at his stop, he elected to spend more than 4 hours round trip (i.e., 30-minute walk, 2 buses, 1 train) to avoid the transportation service provided by his insurance carrier. Although several patients reported living near public transportation, few reported using it as a primary method of getting to their appointments. Additionally, one patient reported avoiding the use of public transit owing to use of a walker. Providers also identified these barriers. Patients who described having a social support system, including friends or family, often relied on their “people” to give them a ride. A social support system in general was seen by patients as a facilitator, but could not solve all transportation-related concerns reliably, as noted further elsewhere in this article. Patients who had the ability to drive themselves with a personal car often described this as a facilitator for adhering to visits. However, the cost of parking was mentioned as a barrier. For example, one patient reported having to refill her parking meter owing to the extended length of her visit. Providers described similar barriers and facilitators as noted elsewhere in this article. One additional barrier reported by providers that was not mentioned by patients was that temporary disability or visual impairment made travel to and from appointments more difficult. Inner Setting Patients expressed significant dissatisfaction with wait times in the clinic. Although some patients reported getting in and out in less than 1 hour, others expressed frustration with wait times extending beyond 3 hours. Although this wait time was not a direct barrier, there was an interaction between the effects of wait time and other factors affecting the patient, such as using up all approved paid time off work or issues with transportation. Regarding transportation, some transportation services provided by the patient's insurance had an agreed upon window for pick-up, usually 3 hours, or limited hours of operation. Because transportation services had to be scheduled 3 days in advance, patients reported that if their visit time went over the scheduled pickup time, there was no way to reschedule their transportation. As a result, one patient even reported leaving before being seen by a provider to decrease the risk of being left without transportation after an appointment. Providers acknowledged an interaction between long wait times and other themes, including transportation and paid time off. The GEC is the only dedicated walk-in clinic for immediate eye care in Chicago. Providers offered that a reason for long wait time for scheduled visits was due to the need to, in some cases, prioritize walk-in patients who are frequently coming for urgent and emergent issues. Patients expressed significant dissatisfaction with wait times in the clinic. Although some patients reported getting in and out in less than 1 hour, others expressed frustration with wait times extending beyond 3 hours. Although this wait time was not a direct barrier, there was an interaction between the effects of wait time and other factors affecting the patient, such as using up all approved paid time off work or issues with transportation. Regarding transportation, some transportation services provided by the patient's insurance had an agreed upon window for pick-up, usually 3 hours, or limited hours of operation. Because transportation services had to be scheduled 3 days in advance, patients reported that if their visit time went over the scheduled pickup time, there was no way to reschedule their transportation. As a result, one patient even reported leaving before being seen by a provider to decrease the risk of being left without transportation after an appointment. Providers acknowledged an interaction between long wait times and other themes, including transportation and paid time off. The GEC is the only dedicated walk-in clinic for immediate eye care in Chicago. Providers offered that a reason for long wait time for scheduled visits was due to the need to, in some cases, prioritize walk-in patients who are frequently coming for urgent and emergent issues. Outer Setting (Patient Needs and Resources) Patients reported that having a social support system, such as family or friends, served as a facilitator. Specifically, as noted elsewhere in this article, patients relied on others to drive them to and/or attend appointments with them. Although this factor generally served as a facilitator, patients also stated that their ability to attend a visit depended on whether the appointment time and day also worked for the person driving them to the appointment. One patient shared that she missed an appointment owing to a family member's work schedule. Although providers agreed that reliance on another person for transportation to appointments could be a facilitator, they reiterated concerns that it could also be a barrier because it depended on fitting with two individuals’ schedules. Those who brought a companion to their visit also found it helpful for this other person to learn about their eye condition. Several providers described that patients coming with family or friends often served as a facilitator who could assist with reinforcing the care plan and education. However, providers found that caretakers who attended visits with patients coming from facilities such as nursing homes served mostly as transportation to the visit and were less frequently able to provide additional details about the patient's medical history. Inner Setting Some providers reported that early coronavirus disease 2019 policies, which limited visits to the patient only in many cases and have since been lifted, were a barrier to patient education and facilitation of follow-up visits. Several providers felt a responsibility to assist the patient with accessing resources. Examples included providing information on financial assistance or personally facilitating transportation. Another facilitator described by providers was their experience referring patients to social work for additional resources and a trained professional to assist with social support. Many felt that they had an overall positive experience with social work, but saved referrals for patients at highest risk of missing a visit, even though they felt most patients would benefit from this type of referral. One provider specifically noted that a majority of patients seen in the GEC would benefit from a social work referral, but that limited availability of social work caused the provider to have to strategically decide who to refer. The decision of who to refer to social work varied from provider to provider. Patients reported that having a social support system, such as family or friends, served as a facilitator. Specifically, as noted elsewhere in this article, patients relied on others to drive them to and/or attend appointments with them. Although this factor generally served as a facilitator, patients also stated that their ability to attend a visit depended on whether the appointment time and day also worked for the person driving them to the appointment. One patient shared that she missed an appointment owing to a family member's work schedule. Although providers agreed that reliance on another person for transportation to appointments could be a facilitator, they reiterated concerns that it could also be a barrier because it depended on fitting with two individuals’ schedules. Those who brought a companion to their visit also found it helpful for this other person to learn about their eye condition. Several providers described that patients coming with family or friends often served as a facilitator who could assist with reinforcing the care plan and education. However, providers found that caretakers who attended visits with patients coming from facilities such as nursing homes served mostly as transportation to the visit and were less frequently able to provide additional details about the patient's medical history. Some providers reported that early coronavirus disease 2019 policies, which limited visits to the patient only in many cases and have since been lifted, were a barrier to patient education and facilitation of follow-up visits. Several providers felt a responsibility to assist the patient with accessing resources. Examples included providing information on financial assistance or personally facilitating transportation. Another facilitator described by providers was their experience referring patients to social work for additional resources and a trained professional to assist with social support. Many felt that they had an overall positive experience with social work, but saved referrals for patients at highest risk of missing a visit, even though they felt most patients would benefit from this type of referral. One provider specifically noted that a majority of patients seen in the GEC would benefit from a social work referral, but that limited availability of social work caused the provider to have to strategically decide who to refer. The decision of who to refer to social work varied from provider to provider. Outer Setting Most patients interviewed had Medicaid insurance. Medicaid provides visit coverage and offers transportation through contracted services for patients with no other way to get to and from their visits. Although these factors serve as facilitators to care, patients did report that having this insurance type limited where they could get their care as well as their transportation options. Although not a direct barrier to adherence, this factor often affected the distance to a covered provider, creating additional patient barriers. Those reporting working full time discussed that their employer determines their ability to take time off for visits. Although paid time off was available for some patients, many discussed the increased wait time interacting with approved time off. Characteristics of Individual Transportation and economic situation were seen to overlap in several ways. For example, patients expressed cost of parking as a financial concern affecting their decision whether to drive to the clinic. Similarly, as noted in the outer setting, several patients reported their fixed income factoring into transportation options. Most patients interviewed had Medicaid insurance. Medicaid provides visit coverage and offers transportation through contracted services for patients with no other way to get to and from their visits. Although these factors serve as facilitators to care, patients did report that having this insurance type limited where they could get their care as well as their transportation options. Although not a direct barrier to adherence, this factor often affected the distance to a covered provider, creating additional patient barriers. Those reporting working full time discussed that their employer determines their ability to take time off for visits. Although paid time off was available for some patients, many discussed the increased wait time interacting with approved time off. Transportation and economic situation were seen to overlap in several ways. For example, patients expressed cost of parking as a financial concern affecting their decision whether to drive to the clinic. Similarly, as noted in the outer setting, several patients reported their fixed income factoring into transportation options. Characteristics of Individual Providers often spoke of health literacy, described by providers as the “patient's understanding of diagnosis or disease,” as a motivating factor for attending ophthalmology visits. Most providers who participated in this study discussed an association between health literacy (or lack thereof) as a potential facilitator (and/or barrier) to attending visits; however, most patients who were interviewed did express a motivation to attend their visits for several reasons. Patients reported that their vision or eye health is important to them because it plays a major role in activities of daily living (e.g., taking medications, completing paperwork, caring for family member). Patients also reported that they felt they had a good understanding of the risk of inaction, either from the education they were provided during a visit or from seeing a friend or family member experiencing health complications or vision loss. Moreover, patients showed a high level of self-advocacy and even created workarounds to avoid barriers from the inner and outer settings. Examples included scheduling a later pickup time with the transportation service in case a visit ran late, scheduling visits early in the day to avoid the long wait times in the clinic, or creating a unique personal scheduling system. Two patients reported leaving an outside provider and working with a new insurance company to ensure they could be seen. One patient even reported paying for a copy of their medical records from an outside provider when transferring care to our health system. This self-advocacy was seen across all patients. Providers often spoke of health literacy, described by providers as the “patient's understanding of diagnosis or disease,” as a motivating factor for attending ophthalmology visits. Most providers who participated in this study discussed an association between health literacy (or lack thereof) as a potential facilitator (and/or barrier) to attending visits; however, most patients who were interviewed did express a motivation to attend their visits for several reasons. Patients reported that their vision or eye health is important to them because it plays a major role in activities of daily living (e.g., taking medications, completing paperwork, caring for family member). Patients also reported that they felt they had a good understanding of the risk of inaction, either from the education they were provided during a visit or from seeing a friend or family member experiencing health complications or vision loss. Moreover, patients showed a high level of self-advocacy and even created workarounds to avoid barriers from the inner and outer settings. Examples included scheduling a later pickup time with the transportation service in case a visit ran late, scheduling visits early in the day to avoid the long wait times in the clinic, or creating a unique personal scheduling system. Two patients reported leaving an outside provider and working with a new insurance company to ensure they could be seen. One patient even reported paying for a copy of their medical records from an outside provider when transferring care to our health system. This self-advocacy was seen across all patients. Several additional factors within the inner setting were found to influence attitudes toward adherence, including scheduling, trust, and care provided. Scheduling (Inner Setting) Patients and providers both expressed concerns with appointment scheduling. Some patients stated that their appointment lead time, the time between when the appointment is scheduled and when it occurs, was short, less than 1 week. Others reported lead times between a few weeks and 5 months. In addition, those who cancelled found it difficult to reschedule a visit in a timely manner. Providers similarly described longer lead times, increasing the risk of no shows. Patients and providers agreed that leaving with a follow-up appointment in hand served as a great facilitator that helped to avoid having to call and schedule visits and decreased the risk of follow-up visits being scheduled incorrectly (e.g., missing additional linked visit for imaging). Patients and providers also reported that reminder calls and messages assisted with adherence; however, both noted that automated calls had less of an effect compared with having a person call, especially in the case of needing to reschedule. Providers also reported that reminder calls work only when the phone number in the system is correct, and they noted this was often not the case. Trust (Characteristics of Individual/Inner Setting) Regarding trust, patients often felt a loyalty to and reported getting all their care within the health system. Trust in the overall health system served as a facilitator and was not mentioned as a barrier. Care Provided (Inner Setting) Finally, patients expressed a very positive experience with the quality of care they received during their visits. Despite some patients reporting long wait times, many patients remarked that the staff were friendly and welcoming, the examination thorough, and doctors took the time to educate them on their condition. Patients and providers both expressed concerns with appointment scheduling. Some patients stated that their appointment lead time, the time between when the appointment is scheduled and when it occurs, was short, less than 1 week. Others reported lead times between a few weeks and 5 months. In addition, those who cancelled found it difficult to reschedule a visit in a timely manner. Providers similarly described longer lead times, increasing the risk of no shows. Patients and providers agreed that leaving with a follow-up appointment in hand served as a great facilitator that helped to avoid having to call and schedule visits and decreased the risk of follow-up visits being scheduled incorrectly (e.g., missing additional linked visit for imaging). Patients and providers also reported that reminder calls and messages assisted with adherence; however, both noted that automated calls had less of an effect compared with having a person call, especially in the case of needing to reschedule. Providers also reported that reminder calls work only when the phone number in the system is correct, and they noted this was often not the case. Regarding trust, patients often felt a loyalty to and reported getting all their care within the health system. Trust in the overall health system served as a facilitator and was not mentioned as a barrier. Finally, patients expressed a very positive experience with the quality of care they received during their visits. Despite some patients reporting long wait times, many patients remarked that the staff were friendly and welcoming, the examination thorough, and doctors took the time to educate them on their condition. Providers were asked about potential interventions that could assist in improving adherence to visits. Recommendations varied and encompassed refining patient education (e.g., educational videos, handouts), reducing time burden (e.g., reducing patient volumes, improving efficiency), and addressing patient needs (e.g., counseling visits to address social determinants of health, adequate social work staffing, transportation assistance center). In this study, we engaged patients from neighborhoods with high social vulnerability and providers to evaluate barriers and facilitators to ophthalmology visit adherence. Key findings of this study are that (1) eye care providers and patients coming from neighborhoods with high socially vulnerability agreed on four key themes among barriers and facilitators to ophthalmology visit adherence—transportation (barrier and/or facilitator), time burden (barrier), economic situation (barrier and/or facilitator), and social support (barrier and/or facilitator). (2) Although providers perceived health literacy as a barrier affecting motivation, patients expressed a high motivation to attend visits and felt well-educated about their condition. (3) Combining implementation and HCD methods, such as CFIR and rapid affinity diagramming, respectively, helps to organize themes into possible intervention strategies. Four key themes emerged in interviews with patients and providers: transportation (barrier and facilitator), time burden (barrier), economic situation (barrier and facilitator), and social support (barrier and facilitator). Each of these themes are associated with social determinants of health, nonmedical factors in communities that affect health, functioning, and quality-of-life outcomes and risks. , Eye visit adherence is a challenge exhibited across the United States. , – Analogous research identifies similar themes in barriers and facilitators to eye visit adherence. For example, community interviews with providers and uninsured or underinsured patients in Michigan indicated priorities, knowledge, transportation, cost, and access as key themes. Similarly, focus groups among high-risk patients across multiple socioeconomic strata indicated cost, trust, communication, access, race, and the patient–doctor relationship as key themes. Other studies report similar barriers including lower income or cost of care, lack of insurance or public insurance, trust, communication, transportation, health literacy, patient or escort difficulty getting time off work for appointments, and access to care. – , , There was a noticeable difference in the perceptions providers and patients felt health literacy played in motivating patients to adhere to visits. Specifically, providers indicated they believed patients need more education about their condition to motivate them to attend ophthalmology visits. However, in this study, most patients expressed an understanding of the importance of eye care and a high desire to attend eye care appointments. This divergence—providers concerns regarding a potential lack of patient motivation owing to lack of understanding about their condition versus patients emphasizing their own highly motivated state—has been similarly identified in a Delphi study of patient and provider perspectives in glaucoma treatment adherence conducted in urban Alabama. This situation indicates a misunderstanding as to the motivation component underpinning patient adherence to eye care visits. The previously mentioned study, however, differed in that patients recognized sociobehavioral treatment facilitators such as social support, whereas providers focused on socioeconomic treatment barriers, such as cost or transportation. In contrast, patients and providers in our study aligned on the presence of ability-related facilitators (social support) and barriers (socioeconomic factors) to visit adherence. One must further consider that the term “adherence” naturally frames the situation as a behavioral (and motivational) problem; this health system perspective does not assess all the variables impacting difficult decisions made by patients to receive or not receive care. An interesting observation in this study was the interaction among themes as illustrated in . For example, transportation was seen as a theme in and of itself, but also interacted with time burden (e.g., visit wait times putting patients at risk of missing their scheduled transportation home), economic situation (e.g., cost of parking), and social support (e.g., reliance on family or friends for transportation). Although this finding in unsurprising, we must understand that, for patients coming from neighborhoods with high social vulnerability, these interactions among themes may be stronger and need to be considered when creating an intervention. Use of the CFIR framework helps to inform where across the health care and social ecosystem these barriers and facilitators reside and impact adherence to appointments. For example, potential interventions may be situated within the inner setting (e.g., adjustments to scheduling processes, parking fees, and social work support). However, it is also apparent that each of these four themes have impacts outside health care’s traditional scope (e.g., outer setting—paid time off, characteristics of individuals—visual impairment affects ability to drive to visit). For example, owing to a “lack of slack” in interventions such as insurance-based transportation, patients could not make a reliable backup plan should the initial intervention fail for whatever reason. An important implication is that if ability-related barriers to a behavior are sufficiently high, no amount of motivation will be able to successfully empower change. Therefore, although providers may believe patients require education-related interventions to increase motivation, and therefore improve adherence, socially vulnerable patients may lack the choice infrastructure necessary to allow high motivation to empower behavior change. A potential implication is that social determinants of health may impact the coding of a barrier or facilitator as a motivation-related or ability-related issue. Consider long wait times; although this factor may be seen as a motivation-related barrier to adherence (e.g., wishing to avoid long wait times, a patient decides to not complete an appointment), patient interviews indicate this to be an ability-related barrier. Long wait times disrupt transport plans, jeopardizing patients scheduled transportation service or family and friend pickups. This well-known issue emphasizes the importance of an ability-related framing when investigating eye visit adherence interventions for socially vulnerable populations. Transportation as a barrier to accessing eye care is well-documented. Although Medicaid offers access to transportation services, our study shows barriers that exist even within this offering. Ride share transportation services have been proven effective in improving adherence to visits and found to be economically feasible in cancer care. , This type of community-level intervention requires further evaluation within eye care. Our study has several limitations. Only 17 of 85 patients agreed to participate in this research study. Because we were looking to recruit patients with a history of nonadherence to visits, we expected challenges in recruitment and, therefore, reached out to a larger number of patients. Although this practice may have resulted in participation bias, we were able to interview a diverse population coming from neighborhoods with high social vulnerability, more than one-half (11 of 17 patients) of whom were nonadherent to previous visits and the specific population we aimed to interview. We also aimed to minimize response bias by assuring anonymity of responses and conducting interviews in private. Further research may be helpful in understanding effective strategies for recruiting historically marginalized communities for qualitative research. Although confirmation bias is always possible in qualitative research, we actively worked to decrease this by establishing concordance among the researchers. Finally, CFIR is a race-neutral tool that can reduce the ability to recognize how structural racism affects interventions designed to address health disparities. The CFIR framework was revised in October 2022, after study completion. The revised CFIR framework attempts to assess determinants related to equity in implementation and may be used for future research. Design and prototyping of implementation strategies must embody an antiracist perspective and actively engage representative stakeholders. Visit nonadherence is often attributed to a lack of health literacy affecting motivation to attend visits. – Our findings indicate that absence of ability, specifically owing to lack of resources, presents more significant barriers for patients from neighborhoods with high social vulnerability. Understanding factors within CFIR can provide useful guidance for where future interventions should be located, and looking at motivation versus ability can help to further refine the purpose of such interventions. Our research points to focusing future interventions on ability- or resource-related barriers away from patient education and in the outer setting, such as transportation. Such interventions may reside outside the four walls of the clinic; hence, new policies are needed to address these barriers and broaden the scope of assistance provided to vulnerable populations. Supplement 1 Supplement 2 |
Optimisation of the preparation phase for orthopaedic surgery: Study protocol for a student-led multimodal prehabilitation feasibility trial (BoneFit) | 0509e218-6dde-420e-ac1c-0dd3b5feeb13 | 11819578 | Surgical Procedures, Operative[mh] | In the United Kingdom (UK), musculoskeletal conditions account for >25% of surgical procedures undertaken by the National Health Service (NHS) . Total joint arthroplasty / replacement is the most common orthopaedic surgical procedure performed annually, particularly total hip replacement (THR), and total knee replacement (TKR) surgery . The Covid-19 pandemic had a significant deleterious impact on surgical interventions with the NHS pausing elective “non-urgent” surgery in April 2020 . Orthopaedic surgery was viewed as low priority during the pandemic leading to increased waiting times , with 7.2 million people listed for elective hospital treatment in January 2023, an increase of 58% since the start of the pandemic . The British Orthopaedic Association calculated that there were 24,000 people waiting in excess of one year for trauma and orthopaedic surgery across the UK, at the height of the pandemic. Orthopaedic surgery is associated with considerable morbidity, increased risk of complications, and excess mortality . From a patient perspective, surgery can lead to a reduction in physical function, a loss of independence due to continued inactivity, immobility and deconditioning. Increased pain and discomfort can lead to physical and mental complications including increased stress, anxiety and depression . These symptoms can lead to higher readmission rates and longer hospital stays , especially if individuals are waiting for over one year to receive surgery. In 2021, the Centre for Perioperative Care published a national position statement for preoperative assessment and optimisation for surgery . Clinical commissioners were urged to establish ‘prehabilitation’ services to support individuals requiring ‘optimisation’ of co-morbidities, nutritional status, psychological preparedness, and physical fitness, thus allowing patients to ‘wait well’ for surgery. Adoption and uptake of these guidelines have been patchy across the UK. Hospital trusts have been encouraged to reconsider the concept of ‘waiting’ lists, and instead consider the period between diagnosis and surgery as ‘preparation’ time . Implementing services allowing individuals to optimise their physical and mental wellbeing prior to surgery will likely lead to improved patient outcomes and could save the NHS money by reducing length of hospital stay, complications and readmission rates . In people requiring orthopaedic surgery, the evidence-base showing the positive impact of prehabilitation on surgical outcomes continues to grow. Recently, a large-scale systematic review and meta-analysis based on 48 unique trials involving 3,570 participants (62% female, mean age 64 years) reported level I moderate-certainty evidence supporting prehabilitation versus usual-care for improving pre-operative function and strength in people undergoing TKR surgery, and moderate-certainty evidence for increased health-related quality of life and muscle strength for individuals undergoing THR surgery . Early intervention is key, as ‘waiting’ list length is likely to be linked to a greater deterioration in function, and a greater challenge for an individual to start making positive lifestyle changes . Low mood and waning motivation can increase whilst ‘waiting’ for surgical intervention which can often exacerbate poor lifestyle choices e.g. increase tobacco use, poorer eating habits, wait gain, and reduced habitual physical activity . Our aim was to introduce the BoneFit trial, an open-label, non-randomised feasibility trial focused on determining the impact of a student-led multimodal prehabilitation intervention on physical and psychological function, quality of life, and clinical outcomes including length of study, complication and readmission rates, in people listed for TKR/THR surgery. We will also determine participant and clinician views of the intervention, and identify any challenges and enablers of inter-institutional partnership working. The BoneFit trial was ethically approved by the Health Research Authority (London Bridge Research Ethics Committee: REC reference: 24/PR/0092) in March 2024. The trial sponsor is the University of Hull. The trial was pre-registered at ClinicalTrials.gov (identifier: NCT06341920). Participants referred to the Department of Orthopaedics at the Hull University Teaching Hospitals NHS Trust (HUTHT) for TKR/THR surgery. BoneFit trial information will be provided to referrals by clinical staff or administrators. Interested parties will contact the Health, Injury and Performance Hub (Hip-Hub) clinic at the University of Hull ( https://hiphub.hull.ac.uk/ ) for an initial appointment. Participants All referrals are awaiting TKR/THR surgery at HUTHT. General inclusion and exclusion criteria are provided below: General inclusion criteria Age 18–75 years; Waiting for unilateral TKR/THR surgery; Able to provide informed consent; General exclusion criteria Previous TKR/THR surgery; Any medical conditions for which moderate to vigorous exercise is contraindicated; Patellar or hip joint instability; Any other disease/condition which severely effects functional performance e.g. stroke or Parkinson’s disease; Chronic depression or significant psychiatric disorder; Enrolled in another clinical trial (or recently completed one); Cognitive impairment which would affect compliance; Patients unable or unwilling to commit to required study follow-ups; Pregnancy; Randomisation and blinding This is an open-label trial as it would not be possible to mask group allocation (BoneFit intervention versus usual care ‘controls’) from participants or clinical staff administering the interventions. After the initial appointment and following the completion of baseline screening and assessments, participants will be allocated to the intervention or control group by a clinician who is independent to the BoneFit trial. All participants allocated to the BoneFit intervention will initially receive existing early recovery after surgery (ERAS) standard guidance regarding the development of healthier lifestyle choices in preparation for surgery (material delivered by post, or via app and website). Data collection and management Study data will be collected on a case report form by the research team at the point of consent and at each subsequent time-point. Each participant will be allocated a unique study ID number and will remain anonymised for the purposes of the trial. Data will be recorded using both an online system and hard copy data collection sheets and stored securely in the Hip-Hub clinic at the University of Hull. Sample size An a priori power calculation to determine sample size was not included as the BoneFit trial has been configured as a feasibility study. However, we did follow statistical guidance indicating that a minimum of 20 participants per group should be included . Therefore, to allow for a potential drop-out rate of approximately 20% (commonly reported in lifestyle interventions), we will attempt to recruit 25 participants to each group (intervention versus control), targeting 50 participants in total. Patient and public involvement and engagement (PPIE) Extensive PPIE work was conducted to inform the separate components of the BoneFit intervention. A patient advisory group including 12 patients aged between 60–80 years (75% male) were interviewed at HUTHT in 2023. We also interviewed clinicians (n = 4; dietician, physiotherapist, occupational therapist, clinical exercise practitioner) based at HUTHT whose roles were to support patients undergoing TKR/THR surgery. Their perspectives and input helped us to develop the selected interventions which have led to the development of BoneFit. Study procedures The Hip-Hub clinic offers student-led, patient centred care to local citizens living within the city of Hull. All students are enrolled on undergraduate or postgraduate health-related programmes at the University of Hull. All students work under the guidance of a qualified healthcare professional from their discipline area. To maximise patient engagement and adherence a person-centred approach to behaviour change support is employed . The approach is to focus on enhancing self-efficacy (confidence) in order to engage in new behaviours as well as developing strategies and action plans that meet their priorities and personal circumstances. Approaches will be individualised depending on the level of patient autonomy required to adhere to the PCPs. The schedule for enrolment, intervention and assessment is illustrated in . Study outcome measures The primary outcome measures are feasibility and acceptability of the BoneFit intervention. Feasibility will be assessed by determining the number of participants recruited, trained and retained at the end of the intervention, the proportion of sessions delivered and fidelity of delivery. Moreover, participant recruitment, retention and adherence to the intervention will be measured, as well as any adverse events. Secondary outcomes will attempt to identify a signal of efficacy for changes in physical health (exercise and nutrition), psychological wellbeing, and quality of life compared to usual care. Secondary outcomes for each of the three core components (exercise, nutrition and psychological outcomes) will be evaluated at 3 major time-points (baseline [2 months from surgery], immediately prior to surgery [2 to 10 days], and 3 months following surgery. The Duke Activity Status Index (DASI) will be completed remotely at two extra timepoints (from initial referral) to determine if functional capacity deteriorates before we intervene at the two month point from surgical intervention. We will also evaluate longer-term changes in functional capacity (via DASI) at 12 months following surgery. Clinical outcomes will mainly be assessed at one time-point (3 months following surgery). Controls will be assessed at 2 major time-points for comparative purposes (baseline [2 months from surgery], and 3 months following surgery. identifies which outcome measures will be recorded. A concurrent mixed methods process evaluation with explore safety, implementation, delivery, and acceptability of the intervention. We will use semi-structured interviews with participants and practitioners along with process data to determine the acceptability of the intervention and to explore barriers and enablers to the implementation of the intervention, interviews will be conducted amongst participants (n = 6), and clinical staff involved in referral and intervention delivery (n = 6). Themes which will be explored will include barriers to recruitment; acceptability and adherence to the intervention (dose received); intervention delivery (fidelity); how the intervention was embedded into clinical practice; safety outcomes: will include adverse events and serious adverse events; assess surgeons’ and surgical practitioners’ willingness to refer to BoneFit; assess participants experiences of the BoneFit intervention. Interventions We were guided by the 2019 NHS Long Term Plan , which advocated the development of personalised care plans (PCPs). Validated screening and assessment tools will enable assignment of participants to appropriate levels of support. Those with no increased risk factors and with no increased surgical risk will receive universal support. Further assessment will be undertaken for those requiring more than universal support and they will be allocated to targeted (intermediate risk/needs) or specialist (high risk/complex needs). Individuals may receive different levels of support for the different intervention components: exercise, nutrition and psychological support. Individuals assigned to specialist groups may be excluded from the intervention. Student-patient interactions will be delivered mainly in a face-to-face individual or group setting. However, in certain circumstances, virtual (one-to-one or group-based using Teams or equivalent) or via telephony may also be offered. Modes of engagement/support will be monitored and recorded as part of the evaluation process. Sessions will mainly be delivered live although some pre-recorded material may be used to supplement live sessions and an online resources library will be developed over time. Screening and assessment Following screening, if an individual (irrespective of allocation to intervention or control group) is identified as being in the specialist group (high risk) or nutrition or psychological support, they would be deemed unsuitable for BoneFit and re-referred for specialist care through usual professional service channels, and their GP would be informed. Physical function Students will use the Duke Activity Status Index (DASI) to screen for reduced functional capacity. Patients with a DASI score > 34 are at low risk and will be assigned to universal support, those with a DASI score <34 will be referred for an assessment. Assessment: An incremental shuttle walk test (ISWT) will be performed to assess patients’ functional capacity. Patients with ISWT distance of <475m will be assigned to targeted intervention. Patients with a ISWT distance of <400m or patients with a medical comorbidity that necessitates supervised exercise will be assigned to specialist intervention. If a patient is deemed unsuitable to complete the ISWT by the clinical supervisor due to functional limitations or progressive pain, we will ask them to just undertake a Timed Up and Go test and use this to screen into targeted (<18 seconds), or specialist groups (> = 18 seconds). Nutritional status We will use the Malnutrition Universal Screening Tool (MUST score) to screen for people at nutritional risk. If an individual scores <1 on MUST, they will be assigned to universal support. Assessment: Patients scoring >1 but <2 on MUST will be referred to a student nutritionist / dietitian for an assessment which will include using the patient-generated and professional component of the Patient-Generated Subjective Global Assessment (PG-SGA) . They will additionally perform a hand-grip strength test to enable a nutritional diagnosis and direct care in accordance with the Nutrition Care Process mode and will be allocated to the targeted group. If patients score ≥2 on MUST, they will be categorised as ‘specialist’ and be excluded from the BoneFit trial, and be referred to the relevant community dietetics department for further assessment, and their GP informed. Psychological health status Referrals will be screened by clinic staff using the General Anxiety Disorder Assessment (GAD-7) , the Patient Health Questionnaire 9 (PHQ-9) , and the ‘need for help’ emotions thermometer . Patients scoring <10 on the GAD-7 or ≤10 PHQ-9 will be assigned to universal support. Patients scoring 15+ on the GAD-7 or 20+ on the PHQ-9 will be categorised as ‘specialist’, and will be excluded from the BoneFit trial, and re-referred to the relevant community mental health services for further assessment, and their GP informed. If immediate risk is identified, ‘high risk’ patients will be referred to their local Crisis Service. Assessment: Patients will be eligible for the targeted intervention if they score 10–14 on the GAD-7 or 10–19 on the PHQ-9. This will include a psycho-education package with coping exercises and links to video/audio resources. They will also be able to opt-in for 1:1 sessions (maximum of 6 sessions) with trainee clinical psychologist to determine an appropriate intervention based on an individual basis. Other measures included health-related quality-of-life (HRQoL) assessed by the EuroQoL (EQ-5D-5L) , and either the short-form hip disability and osteoarthritis outcome score-12 (HOOS-12) , or short-form knee injury and osteoarthritis outcome score-12 (KOOS-12) . provides an outline of each intervention. Face-to-face interventions commence at 2 months from surgery, however, participants allocated to the BoneFit intervention arm will receive remote advice (signposting to lifestyle advice/mobile apps of exercise, nutrition and psychological wellbeing) from initial referral. Statistical analysis Feasibility outcomes will be reported as percentages and/or counts. Median and inter-quartile ranges will be used to describe the distribution of data. Data distribution assumptions will checked prior to analysis and an intention-to-treat analysis will be conducted with any missing data accounted for using appropriate techniques. Data will be analysed using SPSS (IBM, NY, USA). For quantitative data, normality testing will be conducted and appropriate parametric or non-parametric analysis will be conducted. Qualitative data from semi-structured interviews will be transcribed verbatim and NVivo software (Lumivero, USA) will be used to help explore themes which emanate from the discussions. Data monitoring, adverse events and auditing A trial management steering group made up from key collaborators will meet quarterly to discuss trial progression, data monitoring, trial conduct and safety considerations. Patient representation will also be included in the trial management steering group. Adverse events which may be attributable to the intervention will be monitored by the Hip-Hub manager (JSn) and reported to the management steering group and clinical lead for the trial (TS). Dissemination and impact Throughout the trial, media outlets (including social media) will be informed of progress, and the experiences gained will be presented at national conferences and non-academic outlets such as national governing body publications. On completion, the study results will be published in peer-reviewed journals and presented at scientific meetings. We hope that the BoneFit intervention will be impactful in a number of areas: 1) informing clinical guidelines and developing the evidence-base around multimodal prehabilitation for individuals requiring orthopaedic surgery; 2) improving local patient care and service delivery through enhancing equity of access to services and building on the principles required to deliver effective, safe services. Supporting people who would benefit from optimisation of co-morbidities and needs-based multimodal PCPs, thereby helping referrals “wait well” and “prepare” for surgery; 3) identifying and delivering education, training and advocacy for student healthcare professionals; 4) local workforce transformation through informing the development of new service pathways and strengthening inter-institutional working relationships. Further funding from national and local/regional sources will be sought if we can identify that the trial shows a signal for improving patient-focused and clinical outcomes. A fully-powered randomised controlled trial protocol would be developed under these circumstances. In conclusion, whilst waiting lists remain uncomfortably long in some surgical disciplines, as a legacy of the Covid-19 pandemic, healthcare providers can use the ‘waiting’ time as a period of preparation to allow individuals to “optimise’ their physical and psychological function so they are better prepared for the deleterious effects of major surgery. The BoneFit intervention has been designed with the input from patients and clinicians, combined with current NHS guidance. This feasibility study will determine the impact of inter-institutional partnership working, and a student-led clinic designed to improve physical and psychological outcomes, quality of life, and clinical outcomes in people listed for TKR/THR surgery. All referrals are awaiting TKR/THR surgery at HUTHT. General inclusion and exclusion criteria are provided below: Age 18–75 years; Waiting for unilateral TKR/THR surgery; Able to provide informed consent; Previous TKR/THR surgery; Any medical conditions for which moderate to vigorous exercise is contraindicated; Patellar or hip joint instability; Any other disease/condition which severely effects functional performance e.g. stroke or Parkinson’s disease; Chronic depression or significant psychiatric disorder; Enrolled in another clinical trial (or recently completed one); Cognitive impairment which would affect compliance; Patients unable or unwilling to commit to required study follow-ups; Pregnancy; This is an open-label trial as it would not be possible to mask group allocation (BoneFit intervention versus usual care ‘controls’) from participants or clinical staff administering the interventions. After the initial appointment and following the completion of baseline screening and assessments, participants will be allocated to the intervention or control group by a clinician who is independent to the BoneFit trial. All participants allocated to the BoneFit intervention will initially receive existing early recovery after surgery (ERAS) standard guidance regarding the development of healthier lifestyle choices in preparation for surgery (material delivered by post, or via app and website). Study data will be collected on a case report form by the research team at the point of consent and at each subsequent time-point. Each participant will be allocated a unique study ID number and will remain anonymised for the purposes of the trial. Data will be recorded using both an online system and hard copy data collection sheets and stored securely in the Hip-Hub clinic at the University of Hull. An a priori power calculation to determine sample size was not included as the BoneFit trial has been configured as a feasibility study. However, we did follow statistical guidance indicating that a minimum of 20 participants per group should be included . Therefore, to allow for a potential drop-out rate of approximately 20% (commonly reported in lifestyle interventions), we will attempt to recruit 25 participants to each group (intervention versus control), targeting 50 participants in total. Extensive PPIE work was conducted to inform the separate components of the BoneFit intervention. A patient advisory group including 12 patients aged between 60–80 years (75% male) were interviewed at HUTHT in 2023. We also interviewed clinicians (n = 4; dietician, physiotherapist, occupational therapist, clinical exercise practitioner) based at HUTHT whose roles were to support patients undergoing TKR/THR surgery. Their perspectives and input helped us to develop the selected interventions which have led to the development of BoneFit. The Hip-Hub clinic offers student-led, patient centred care to local citizens living within the city of Hull. All students are enrolled on undergraduate or postgraduate health-related programmes at the University of Hull. All students work under the guidance of a qualified healthcare professional from their discipline area. To maximise patient engagement and adherence a person-centred approach to behaviour change support is employed . The approach is to focus on enhancing self-efficacy (confidence) in order to engage in new behaviours as well as developing strategies and action plans that meet their priorities and personal circumstances. Approaches will be individualised depending on the level of patient autonomy required to adhere to the PCPs. The schedule for enrolment, intervention and assessment is illustrated in . The primary outcome measures are feasibility and acceptability of the BoneFit intervention. Feasibility will be assessed by determining the number of participants recruited, trained and retained at the end of the intervention, the proportion of sessions delivered and fidelity of delivery. Moreover, participant recruitment, retention and adherence to the intervention will be measured, as well as any adverse events. Secondary outcomes will attempt to identify a signal of efficacy for changes in physical health (exercise and nutrition), psychological wellbeing, and quality of life compared to usual care. Secondary outcomes for each of the three core components (exercise, nutrition and psychological outcomes) will be evaluated at 3 major time-points (baseline [2 months from surgery], immediately prior to surgery [2 to 10 days], and 3 months following surgery. The Duke Activity Status Index (DASI) will be completed remotely at two extra timepoints (from initial referral) to determine if functional capacity deteriorates before we intervene at the two month point from surgical intervention. We will also evaluate longer-term changes in functional capacity (via DASI) at 12 months following surgery. Clinical outcomes will mainly be assessed at one time-point (3 months following surgery). Controls will be assessed at 2 major time-points for comparative purposes (baseline [2 months from surgery], and 3 months following surgery. identifies which outcome measures will be recorded. A concurrent mixed methods process evaluation with explore safety, implementation, delivery, and acceptability of the intervention. We will use semi-structured interviews with participants and practitioners along with process data to determine the acceptability of the intervention and to explore barriers and enablers to the implementation of the intervention, interviews will be conducted amongst participants (n = 6), and clinical staff involved in referral and intervention delivery (n = 6). Themes which will be explored will include barriers to recruitment; acceptability and adherence to the intervention (dose received); intervention delivery (fidelity); how the intervention was embedded into clinical practice; safety outcomes: will include adverse events and serious adverse events; assess surgeons’ and surgical practitioners’ willingness to refer to BoneFit; assess participants experiences of the BoneFit intervention. We were guided by the 2019 NHS Long Term Plan , which advocated the development of personalised care plans (PCPs). Validated screening and assessment tools will enable assignment of participants to appropriate levels of support. Those with no increased risk factors and with no increased surgical risk will receive universal support. Further assessment will be undertaken for those requiring more than universal support and they will be allocated to targeted (intermediate risk/needs) or specialist (high risk/complex needs). Individuals may receive different levels of support for the different intervention components: exercise, nutrition and psychological support. Individuals assigned to specialist groups may be excluded from the intervention. Student-patient interactions will be delivered mainly in a face-to-face individual or group setting. However, in certain circumstances, virtual (one-to-one or group-based using Teams or equivalent) or via telephony may also be offered. Modes of engagement/support will be monitored and recorded as part of the evaluation process. Sessions will mainly be delivered live although some pre-recorded material may be used to supplement live sessions and an online resources library will be developed over time. Following screening, if an individual (irrespective of allocation to intervention or control group) is identified as being in the specialist group (high risk) or nutrition or psychological support, they would be deemed unsuitable for BoneFit and re-referred for specialist care through usual professional service channels, and their GP would be informed. Physical function Students will use the Duke Activity Status Index (DASI) to screen for reduced functional capacity. Patients with a DASI score > 34 are at low risk and will be assigned to universal support, those with a DASI score <34 will be referred for an assessment. Assessment: An incremental shuttle walk test (ISWT) will be performed to assess patients’ functional capacity. Patients with ISWT distance of <475m will be assigned to targeted intervention. Patients with a ISWT distance of <400m or patients with a medical comorbidity that necessitates supervised exercise will be assigned to specialist intervention. If a patient is deemed unsuitable to complete the ISWT by the clinical supervisor due to functional limitations or progressive pain, we will ask them to just undertake a Timed Up and Go test and use this to screen into targeted (<18 seconds), or specialist groups (> = 18 seconds). Nutritional status We will use the Malnutrition Universal Screening Tool (MUST score) to screen for people at nutritional risk. If an individual scores <1 on MUST, they will be assigned to universal support. Assessment: Patients scoring >1 but <2 on MUST will be referred to a student nutritionist / dietitian for an assessment which will include using the patient-generated and professional component of the Patient-Generated Subjective Global Assessment (PG-SGA) . They will additionally perform a hand-grip strength test to enable a nutritional diagnosis and direct care in accordance with the Nutrition Care Process mode and will be allocated to the targeted group. If patients score ≥2 on MUST, they will be categorised as ‘specialist’ and be excluded from the BoneFit trial, and be referred to the relevant community dietetics department for further assessment, and their GP informed. Psychological health status Referrals will be screened by clinic staff using the General Anxiety Disorder Assessment (GAD-7) , the Patient Health Questionnaire 9 (PHQ-9) , and the ‘need for help’ emotions thermometer . Patients scoring <10 on the GAD-7 or ≤10 PHQ-9 will be assigned to universal support. Patients scoring 15+ on the GAD-7 or 20+ on the PHQ-9 will be categorised as ‘specialist’, and will be excluded from the BoneFit trial, and re-referred to the relevant community mental health services for further assessment, and their GP informed. If immediate risk is identified, ‘high risk’ patients will be referred to their local Crisis Service. Assessment: Patients will be eligible for the targeted intervention if they score 10–14 on the GAD-7 or 10–19 on the PHQ-9. This will include a psycho-education package with coping exercises and links to video/audio resources. They will also be able to opt-in for 1:1 sessions (maximum of 6 sessions) with trainee clinical psychologist to determine an appropriate intervention based on an individual basis. Other measures included health-related quality-of-life (HRQoL) assessed by the EuroQoL (EQ-5D-5L) , and either the short-form hip disability and osteoarthritis outcome score-12 (HOOS-12) , or short-form knee injury and osteoarthritis outcome score-12 (KOOS-12) . provides an outline of each intervention. Face-to-face interventions commence at 2 months from surgery, however, participants allocated to the BoneFit intervention arm will receive remote advice (signposting to lifestyle advice/mobile apps of exercise, nutrition and psychological wellbeing) from initial referral. Students will use the Duke Activity Status Index (DASI) to screen for reduced functional capacity. Patients with a DASI score > 34 are at low risk and will be assigned to universal support, those with a DASI score <34 will be referred for an assessment. Assessment: An incremental shuttle walk test (ISWT) will be performed to assess patients’ functional capacity. Patients with ISWT distance of <475m will be assigned to targeted intervention. Patients with a ISWT distance of <400m or patients with a medical comorbidity that necessitates supervised exercise will be assigned to specialist intervention. If a patient is deemed unsuitable to complete the ISWT by the clinical supervisor due to functional limitations or progressive pain, we will ask them to just undertake a Timed Up and Go test and use this to screen into targeted (<18 seconds), or specialist groups (> = 18 seconds). We will use the Malnutrition Universal Screening Tool (MUST score) to screen for people at nutritional risk. If an individual scores <1 on MUST, they will be assigned to universal support. Assessment: Patients scoring >1 but <2 on MUST will be referred to a student nutritionist / dietitian for an assessment which will include using the patient-generated and professional component of the Patient-Generated Subjective Global Assessment (PG-SGA) . They will additionally perform a hand-grip strength test to enable a nutritional diagnosis and direct care in accordance with the Nutrition Care Process mode and will be allocated to the targeted group. If patients score ≥2 on MUST, they will be categorised as ‘specialist’ and be excluded from the BoneFit trial, and be referred to the relevant community dietetics department for further assessment, and their GP informed. Referrals will be screened by clinic staff using the General Anxiety Disorder Assessment (GAD-7) , the Patient Health Questionnaire 9 (PHQ-9) , and the ‘need for help’ emotions thermometer . Patients scoring <10 on the GAD-7 or ≤10 PHQ-9 will be assigned to universal support. Patients scoring 15+ on the GAD-7 or 20+ on the PHQ-9 will be categorised as ‘specialist’, and will be excluded from the BoneFit trial, and re-referred to the relevant community mental health services for further assessment, and their GP informed. If immediate risk is identified, ‘high risk’ patients will be referred to their local Crisis Service. Assessment: Patients will be eligible for the targeted intervention if they score 10–14 on the GAD-7 or 10–19 on the PHQ-9. This will include a psycho-education package with coping exercises and links to video/audio resources. They will also be able to opt-in for 1:1 sessions (maximum of 6 sessions) with trainee clinical psychologist to determine an appropriate intervention based on an individual basis. Other measures included health-related quality-of-life (HRQoL) assessed by the EuroQoL (EQ-5D-5L) , and either the short-form hip disability and osteoarthritis outcome score-12 (HOOS-12) , or short-form knee injury and osteoarthritis outcome score-12 (KOOS-12) . provides an outline of each intervention. Face-to-face interventions commence at 2 months from surgery, however, participants allocated to the BoneFit intervention arm will receive remote advice (signposting to lifestyle advice/mobile apps of exercise, nutrition and psychological wellbeing) from initial referral. Feasibility outcomes will be reported as percentages and/or counts. Median and inter-quartile ranges will be used to describe the distribution of data. Data distribution assumptions will checked prior to analysis and an intention-to-treat analysis will be conducted with any missing data accounted for using appropriate techniques. Data will be analysed using SPSS (IBM, NY, USA). For quantitative data, normality testing will be conducted and appropriate parametric or non-parametric analysis will be conducted. Qualitative data from semi-structured interviews will be transcribed verbatim and NVivo software (Lumivero, USA) will be used to help explore themes which emanate from the discussions. A trial management steering group made up from key collaborators will meet quarterly to discuss trial progression, data monitoring, trial conduct and safety considerations. Patient representation will also be included in the trial management steering group. Adverse events which may be attributable to the intervention will be monitored by the Hip-Hub manager (JSn) and reported to the management steering group and clinical lead for the trial (TS). Throughout the trial, media outlets (including social media) will be informed of progress, and the experiences gained will be presented at national conferences and non-academic outlets such as national governing body publications. On completion, the study results will be published in peer-reviewed journals and presented at scientific meetings. We hope that the BoneFit intervention will be impactful in a number of areas: 1) informing clinical guidelines and developing the evidence-base around multimodal prehabilitation for individuals requiring orthopaedic surgery; 2) improving local patient care and service delivery through enhancing equity of access to services and building on the principles required to deliver effective, safe services. Supporting people who would benefit from optimisation of co-morbidities and needs-based multimodal PCPs, thereby helping referrals “wait well” and “prepare” for surgery; 3) identifying and delivering education, training and advocacy for student healthcare professionals; 4) local workforce transformation through informing the development of new service pathways and strengthening inter-institutional working relationships. Further funding from national and local/regional sources will be sought if we can identify that the trial shows a signal for improving patient-focused and clinical outcomes. A fully-powered randomised controlled trial protocol would be developed under these circumstances. In conclusion, whilst waiting lists remain uncomfortably long in some surgical disciplines, as a legacy of the Covid-19 pandemic, healthcare providers can use the ‘waiting’ time as a period of preparation to allow individuals to “optimise’ their physical and psychological function so they are better prepared for the deleterious effects of major surgery. The BoneFit intervention has been designed with the input from patients and clinicians, combined with current NHS guidance. This feasibility study will determine the impact of inter-institutional partnership working, and a student-led clinic designed to improve physical and psychological outcomes, quality of life, and clinical outcomes in people listed for TKR/THR surgery. S1 Checklist SPIRIT 2013 checklist: Recommended items to address in a clinical trial protocol and related documents*. (DOCX) S1 File (PDF) |
Pleomorphic Liposarcoma Unraveled: Investigating Histopathological and Immunohistochemical Markers for Tailored Diagnosis and Therapeutic Innovations | 441bf293-24c8-4312-b848-2a11463bec61 | 11205576 | Anatomy[mh] | Liposarcomas are rare soft tissue tumors regarded as a heterogeneous group comprising entities with distinct histopathological, immunohistochemical, and molecular features . Liposarcomas are mainly subclassified into five subtypes with different morphologic and behavioral spectrum . Well-differentiated liposarcoma and dedifferentiated liposarcoma stand for the largest subgroup of liposarcomas and exhibit a propensity for local recurrence and metastases . Well-differentiated liposarcoma (atypical lipomatous tumor) is considered one of the most common soft tissue tumors, mainly affecting the extremities and retroperitoneum . Superficial lesions are usually located in the subcutaneous fat and are diagnosed as atypical lipomatous tumors . They exhibit locally aggressive behavior and a higher frequency of recurrence compared to conventional lipomas . Upon microscopic examination, WDL typically consists of mature adipocytes of variable size, encompassed by fibrous stroma exhibiting atypical spindle cells with hyperchromatic nuclei . Well-differentiated liposarcoma is related to dedifferentiated liposarcoma, a high-grade proliferation associated with similar genetic abnormalities, consisting of MDM2 (murine double minute 2) and CDK4 (cyclin-dependent kinase 4) amplification . Although the histopathological diagnosis of such lesions is usually straightforward, distinguishing them from other entities can be difficult, especially in tumors with peculiar locations . Immunohistochemical study and genetic analysis are the most useful methods of establishing the diagnosis . Well-differentiated liposarcoma usually shows MDM2 and CDK4 expression . However, some cases with atypical morphology and immunophenotype require FISH for identification of MDM2 amplifications, in order to avoid misdiagnosis . Dedifferentiated liposarcoma is an uncommon neoplasm arising predominantly in the retroperitoneum and deep soft tissue of the extremities . This malignancy is associated with an important risk of local recurrence, but its metastatic potential is related to the anatomic site of the tumor proliferation, as retroperitoneal masses have a worse prognosis . The histopathological aspect of dedifferentiated liposarcoma usually consists of a lipogenic tumor proliferation with well-differentiated liposarcoma features, exhibiting an abrupt transition towards a non-lipogenic high-grade sarcoma . The genetic background of dedifferentiated liposarcoma is characterized by MDM2 and CDK4 amplification; therefore, immunohistochemical expression of the corresponding markers is used for diagnosis . A mutant TP53 immunophenotype, with hyperexpression of p53, is also reported in some studies . Molecular analysis of this tumor is important, as new therapeutic targets have been identified . Treatment of dedifferentiated liposarcoma also consists of surgical resection; therefore, the location of the lesion and evaluation of surgical resection margins have prognostic significance . Tumors of the retroperitoneum are associated with a higher risk of recurrence, due to the difficulty of achieving complete surgical resection . Dedifferentiated liposarcoma of the extremities is an uncommon entity with distinct clinical–pathological behavior and incompletely understood pathogenesis . The most important entity that should be considered as a differential diagnosis for dedifferentiated liposarcoma of the extremities is myxoid liposarcoma . Myxoid liposarcoma is a distinct entity with unique genetic and molecular features consisting of FUS-DDIT3 or EWSR1-DDIT3 fusion, typically occurring in younger patients . Although myxoid liposarcoma displays a classic histopathological aspect, with lipogenic areas and basophilic stroma, tumors can exhibit unusual cellular features or metaplasia . Consequently, the differential diagnosis should imply an analysis of DDIT3 mutation, especially in high-grade lesions . This tumor presents variable histomorphology and peculiarities regarding therapeutic management and overall prognosis . The role of FUS-DDIT3 oncoprotein in adipocytic neoplasia has been investigated in the context of ATP-dependent chromatin remodeling and alteration of genomic architecture, resulting in upregulation of tumorigenic pathway in cell lines of myxoid liposarcoma . Researchers acknowledge that transcription factors, such as fusion oncoproteins, can activate oncogenic gene loci and bind to the BAF (barrier-to-autointegration factor) complex surface . Further studies on the FUS-DDIT3 fusion oncogene in myxoid liposarcoma suggest a link between this mutation and JAK-stat signaling in the stem cancerous cell . Currently, the most efficient treatment comprises surgical resection with preoperative radiotherapy frequently administered, whereas high-risk lesions of the limbs or trunk may receive chemotherapy . However, in the advanced setting, new strategies, including targeted therapy, have been developed . Considering the increasing number of studies on genetic alterations related to soft tissue sarcoma, histology-specific treatment protocols have been increasingly implemented and systemic treatment options apply for various liposarcoma subtypes . Pleomorphic liposarcoma is a rare aggressive subtype accounting for 10% of all liposarcomas and it is diagnosed upon detection of multivacuolated pleomorphic lipoblasts upon microscopic examination . This malignancy has no specific immunohistochemical or molecular features; therefore, diagnostic challenges include identifying lipoblasts that may be scarce and distinction from other pleomorphic sarcomas with special morphologic variants . Myxoid pleomorphic liposarcoma is a recently defined neoplasm affecting young patients, disclosing mixed histological features and complex chromosomal alterations . Although its distinctive clinical features support a separate classification of this tumor, studies suggest a link between it and conventional pleomorphic liposarcoma . Considering the rarity of pleomorphic liposarcoma and its mimickers, we have reviewed the literature to gain further knowledge about this malignancy, emphasizing its histopathological and molecular features in correlation with its clinical behavior . This is a narrative review of the scientific literature. We included complete-length papers, using the PubMed search engine, focusing on liposarcomas and their morphological and immunohistochemical features, in correlation with prognosis and treatment. In order to gain further knowledge about pleomorphic liposarcoma, we investigated all the articles published between 2018 and 2023. We interrogated all types of English-language articles, comprising original studies, case reports, and reviews. The associated papers were searched for additional useful references. The research keywords were liposarcoma, dedifferentiated liposarcoma, myxoid liposarcoma, pleomorphic liposarcoma, and immunohistochemistry. The research papers were provided by four reviewers (A.M.C., D.A.Ț., A.B., and A.M.). Three reviewers (A.M.C., A.V.D, and M.C.) analyzed the articles on pleomorphic liposarcoma for information concerning clinical, radiologic, histopathological, immunohistochemical, and molecular features. The whole process was supervised and validated by two reviewers (A.V.D and M.C.) Pleomorphic liposarcoma is the rarest liposarcoma variant and is diagnosed based on the detection of multivacuolated pleomorphic lipoblasts within the specimens, while molecular characteristics of this tumor have constantly been reviewed and discussed, especially considering the new trends in novel investigations and therapeutic strategies . Pleomorphic liposarcoma usually arises within deep soft tissue of the extremities of adult patients and is often a diagnostic challenge . Wakely et al. reviewed their experience with fine needle aspiration as a diagnostic method for this tumor, focusing on the recognition of pleomorphic lipoblasts, which may be found in variable numbers . The study included 20 patients with a 2.3/1 male/female ratio and a mean age of 58 and aspirates from the thigh, upper extremity, axilla, neck, and mediastinum were analyzed . In most of the cases, examination of the specimens noted pleomorphic, epithelioid, and bizarre cells, while pleomorphic lipoblasts were absent or rare in 45% of the cases . The researchers acknowledge that FNA biopsy may not be able to capture pleomorphic lipoblasts, due to the heterogeneous structure of this neoplasm . Additionally, diagnosing pleomorphic liposarcoma and distinguishing it from other sarcomas are difficult, as ancillary studies may be of limited use . The recent literature encompasses several studies on the genomics of each liposarcoma variant and suggests that pleomorphic liposarcoma differs from other subtypes, as it harbors chromosomal imbalances with large numbers of gains and deletions, but with no specific cytogenetic abnormality . Tyler et al. examined the P53 deletion rate, by performing microarray analysis, noting the present mutation within 60% of samples . Pleomorphic liposarcoma is associated with a poor prognosis and high recurrence, and its treatment is controversial . Wang et al. reported a series of six patients with confirmed pleomorphic liposarcoma who underwent surgical excision of the tumor . Researchers noted that five out of six patients developed local recurrences, with the shortest post-operative recurrence time of 4 months and the longest of 29 months . Pleomorphic liposarcoma with peculiar locations has also been discussed throughout scientific literature, to gain further knowledge about the therapeutic management of the malignancy . For instance, Agarwal et al. investigated the characteristics of pleomorphic liposarcoma of the head and neck and noted the importance of adjuvant therapy and negative resection margins in avoiding recurrent disease in a one-year follow-up . Halevi PD et al. reported an unfavorable outcome in the case of a primary pleomorphic liposarcoma arising within the thoracic epidural space in a 70-year-old male patient who presented with lower extremities weakness and back pain . The patient underwent surgical excision and received radiotherapy, but the lesion recurred 3 months after the surgery . Later on, he developed metastases and succumbed to the disease one year later . Researchers suggest that pleomorphic liposarcoma should be taken into consideration as a differential diagnosis for spinal tumor masses, although it is an exceptionally rare finding . As mentioned earlier, immunohistochemical study is of limited use in diagnosing mesenchymal neoplasms within the heterogeneous group of pleomorphic sarcomas . However, ancillary studies can be useful in distinguishing pleomorphic sarcomas from secondary tumors, as the lesions may be histologically similar to metastatic carcinoma or melanoma . In selected cases, the immunohistochemical staining panel includes pancytokeratin, SOX10, and MelanA, while the expression pattern of p53 and INI 1 can also be interrogated . In addition, immunohistochemical expression of MDM2 and CDK4 should be carried out, to rule out pleomorphic dedifferentiated liposarcoma, especially in tumors located within the retroperitoneum . Al-Attar et al. reported a case of pleomorphic liposarcoma of the gastrointestinal tract in a 71-year-old patient with a history of rectal adenocarcinoma . The lesion exhibited epithelioid cells with intracytoplasmic fatty droplets and high mitotic activity, which was initially interpreted as GIST, but expression of CD117 and DOG1 was absent and the lesion exhibited strong, diffuse positive p53 expression suggestive of a mutant phenotype . Considering the rarity of pleomorphic liposarcoma and the reported peculiar locations of this malignancy, researchers acknowledge the significance of judiciously investigating the histopathological and immunohistochemical features of the tumors . Myxoid pleomorphic liposarcoma is an uncommon, newly described liposarcoma variant, showing a propensity for mediastinum and usually affecting young patients and children . This soft tissue neoplasm with a troublesome location typically shows aggressive behavior and a tendency for local recurrence . Al-Kindi et al. reported a case of a giant mediastinal myxoid pleomorphic liposarcoma in an 18-year-old girl who underwent surgical excision followed by oncological treatment . Upon histopathological examination, the tumor showed myxoid stroma encompassing spindle and stellate cells, as well as pleomorphic lipoblasts . Immunohistochemical analysis revealed S100 positivity within the neoplastic cells . Studies suggest that myxoid pleomorphic liposarcoma should be taken into consideration in the differential diagnosis of soft tissue tumors of the mediastinum in children and even infants . As an example, Gami et al. reported the case of a 12-month-old infant presenting with respiratory distress who was finally diagnosed with myxoid pleomorphic liposarcoma . In this patient’s case, the CT scan revealed a large solid tumor mass occupying the left hemithorax, and surgical resection was carried out . The neoplastic proliferation disclosed pleomorphic multivacuolated lipoblasts and myxoid changes, as well as necrotic foci . The tumor cells were immunoreactive for S100, CD 34, and p16 and negative for CDK4, SMA, and desmin, and fluorescent in situ hybridization was used for confirmation . The patient later underwent chemotherapy to prevent disease recurrence . Considering the rarity of this malignancy, its pathogenesis and histopathological features have been observed and reviewed throughout several studies. Researchers have distinguished similar histopathological features in lesions with different locations, although cases of myxoid pleomorphic liposarcoma occurring in peculiar sites have been reported . As an example, Tan GZL et al. reported a case of myxoid pleomorphic liposarcoma of the orbit in a 12-year-old female patient with a gradually enlarging orbital mass causing proptosis of the globe . The lesion was surgically excised and the patient received chemotherapy, with no distant metastasis discovered at the time of diagnosis . Histopathological examination of the specimen revealed a liposarcoma with distinct myxoid and pleomorphic areas . The myxoid areas displayed moderate cellularity and abundant myxoid matrix with occasional pulmonary-edema-like microcystic spaces, while the pleomorphic areas included bizarre cells with multivacuolated cytoplasm, suggestive of pleomorphic lipoblasts . The neoplastic cells with adipocytic differentiation were highlighted using S100 and adipophilin immunohistochemical staining . Mutant expression of p53 with a null pattern was also noted, and INI 1 expression was retained . MDM2 expression was also interrogated, but the marker was negative within neoplastic cells, and in situ hybridization showed no evidence of DDIT3 rearrangement . To achieve a better understanding of myxoid pleomorphic liposarcoma pathogenesis, researchers investigated the association of this malignancy with specific genetic abnormalities . This pathological association is supported by the prevalence of pleomorphic liposarcoma in children and young patients . In this matter, Zare SY et al. reported a case of pleomorphic liposarcoma in a 34-year-old patient with germline TP53 gene mutation and Li–Fraumeni syndrome . The tumor presented as a well-demarcated, solid tumor mass located in the anterior chest wall, measuring 2.9 × 2.3 × 2 cm, and the histopathological aspect was characterized by low cellularity, myxoid stroma, and pleomorphic lipoblasts . In situ hybridization detected no DDIT3 rearrangement or MDM2 amplification . The authors acknowledge that this is the first case of myxoid pleomorphic liposarcoma associated with Li–Fraumeni syndrome and strongly recommend that further studies should be carried out regarding this condition . Two more cases of myxoid pleomorphic liposarcoma affecting patients with Li–Fraumeni syndrome have been reported in the scientific literature so far. Francom et al. reported the case of a myxoid pleomorphic liposarcoma of the head and neck diagnosed as a second primary tumor in an 11-year-old boy with Li–Fraumeni syndrome, who had recently undergone chemotherapy for medulloblastoma . Furthermore, Sinclair et al. reported the case of a 15-year-old female with Li–Fraumeni syndrome who developed a perineal mass, which was diagnosed as myxoid pleomorphic liposarcoma . A retrospective review of malignancies encountered in children with Li–Fraumeni children has been carried out by Rodriguez et al. and included 20 children within a cohort of patients diagnosed with this condition . Researchers reveal that 6 out of 27 malignancies reported in the analyzed group were sarcomas of the head and neck and were all demonstrated as rare entities—rhabdomyosarcoma, synovial sarcoma, and myxoid pleomorphic sarcoma . The study identifies a propensity for head and neck involvement of tumors associated with Li–Fraumeni syndrome and recommends a multidisciplinary care team surveilling the affected patients . Pleomorphic liposarcoma is the most uncommon liposarcoma of adult patients, and it can be associated with a severe prognosis and diagnostic difficulties . We discovered 36 cases of pleomorphic liposarcomas reported during the last five years, between 2018 and 2023. Most of the tumors were located within the abdominal cavity and 8 out of 36 presented as large masses identified within the retroperitoneum or pelvic region . Within the series reported by Wang L. et al., the group contained four men and two women, ranging from 46 to 82 years old, with a confirmed diagnosis of primary pleomorphic liposarcoma . The lesions presented as rapidly growing tumors within the retroperitoneum, abdominal, and pelvic cavities and were identified upon computed tomography, then underwent surgical resection . Histopathological examination of the specimens revealed the presence of bizarre mono or multinucleated giant lipoblasts, displaying heterotypic, hyperchromatic nuclei, and vacuolated cytoplasm . Five out of six tumors had no invasion of the surrounding tissue, and none of the proliferations were associated with invasion of the lymph nodes or peripheral blood vessels . However, in one case, the tumor mass infiltrated the fallopian tube, extending within its smooth muscle layer . Immunohistochemical analysis showed positive CD 34 staining within the tumor cells in four cases, while expression of S-100 protein and CD 68 was identified in three cases . Positive CD117 staining was noted in one of the lesions . All tumors were negative for SMA and AE1/AE3 . Local recurrence was reported in four out of the six patients . This study highlights the aggressive clinical behavior of liposarcoma developing within the abdominal cavity, as lesions can be diagnosed in a locally advanced stage, and surgical resection can be difficult. Studies also acknowledge that pleomorphic liposarcoma of the retroperitoneum can be associated with acute severe systemic complications, apart from its locally infiltrative and distant metastatic potential . As an example, Chen et al. reported a case of retroperitoneal pleomorphic liposarcoma associated with massive tumor embolism of the inferior vena cava and pulmonary arteries, discovered in a 54-year-old woman . Local extension of pleomorphic liposarcoma developing within the retroperitoneum is one of the most frequent and concerning aspects of this malignant tumor. Involvement of the kidney was reported in two of the cases reported within the scientific literature that we interrogated . El Haq et al. reported the case of an 84-year-old male patient who developed a pleomorphic liposarcoma with infiltration of the left flank and kidney, presenting with abdominal pain and hematuria . The second most common location of pleomorphic liposarcoma is the soft tissue, with a total of seven reported cases. Concerning this pathology, studies show that distant metastases are rarely associated with liposarcoma of the soft tissue . Therefore, special attention is required in such cases, as differential diagnosis may be necessary . Ciliberti et al. reported the case of a 51-year-old man with a history of lung carcinoma who developed a pleomorphic liposarcoma of the shoulder girdle deep soft tissue and acquired hepatic metastases . Considering the patient’s medical history, lung carcinoma metastasis was first taken into consideration, but a histopathological examination of the specimen revealed the presence of giant multinucleated lipoblasts with multivacuolated cytoplasm . Immunohistochemical analysis revealed the expression of vimentin within the tumor cells, while MDM2 and Hep-Par1 were negative . The authors underlined the importance of microscopic examination and ancillary studies for positive and differential diagnosis in patients with metastatic liposarcoma of the soft tissue . Throughout our research, we identified one case of pleomorphic liposarcoma of the soft tissue that was exposed to neoadjuvant treatment. The case was reported by Zhang et al. in a 59-year-old woman who developed an unresectable liposarcoma of the deep soft tissue of the abdominal wall and received a combination of radiotherapy and angiogenesis inhibitor anlotinib . The histopathological aspect of the tissue samples revealed the presence of pleomorphic epitheliod cells, occasionally associated with cytoplasmic vacuoles, but the immunohistochemical analysis was peculiar, because the neoplastic cells showed strong MDM2 expression . However, no MDM2 or CDK4 rearrangements were detected, and the lesion was diagnosed as pleomorphic liposarcoma . Examination of the surgical specimen also revealed significant changes related to neoadjuvant treatment, as the tumor volume was remarkedly reduced, and microscopic examination showed fibrotic areas and chronic inflammatory infiltrate, with no viable malignant cells . The study infers that pleomorphic liposarcoma is regarded as a radiotherapy-resistant tumor; therefore, targeted chemotherapy such as angiogenesis inhibitors is required to obtain a complete pathological response . Pleomorphic liposarcoma with a peculiar location was also investigated, as this lesion can reportedly occur within the viscera and bone. Liposarcomas of the testis and spermatic cord are exceptionally rare entities requiring special attention due to their high malignancy grade . We discovered two pleomorphic liposarcoma of the testis and paratesticular region. In both of the cases reported, the lesions presented as large tumor masses, which were surgically excised, and the histopathological examination revealed the typical aspect, implying pleomorphic lipoblasts with multivacuolated cytoplasm . According to the investigated articles, no immunohistochemical analysis was performed in addition to histopathological examination of the specimens, underlying the importance of a thorough microscopic evaluation of the samples. Surgical excision is regarded as the gold standard in the treatment of pleomorphic liposarcoma of the testis and paratesticular space . However, researchers suggest that pre-operative radiotherapy should be taken into consideration in these malignant lesions, in order to reduce the tumor volume and obtain surgical resection margins . Among pleomorphic liposarcomas with peculiar locations, malignant adipocytic tumors of the viscera require special attention . We identified two primary pleomorphic liposarcomas of the lung, reported by Li B. et al. and Dey T. et al., both of them identified in male patients with no history of smoking or alcohol abuse . The patients presented with shortness of breath and chest pain, and the tumors were described as large, hypodense masses identified using a CT scan . The two patients received chemotherapy, followed by surgical resection; however, one of them developed a retroperitoneal metastasis 7 months after treatment . Upon histopathological examination, both lesions were highly suggestive of pleomorphic liposarcoma . An ancillary study revealed intense S100 positivity and a mutant p53 immunophenotype within the tumor cells of both lesions described above . Primary pleomorphic liposarcoma of the lung is an exceedingly rare finding, accounting for less than 1% of pulmonary malignancies . The main challenge regarding differential diagnosis of these lesions refers to metastatic liposarcoma; therefore, a thorough clinical examination and history taking should be performed, in addition to imagistic investigations and histopathological and immunohistochemical examination. Within the scientific literature that we investigated, two cardiac liposarcomas were found. Both lesions reported displayed characteristic locations, involving the interventricular septum and associated pericardial effusion . Both lesions exhibited specific pleomorphic lipoblasts upon microscopic examination, while the ancillary studies performed revealed S100 positivity and a mutant p53 immunophenotype within the tumor cells . Tan NY et al. underlined the importance of preoperative chemotherapy in primary cardiac liposarcoma, while Burt J.R. et al. described the benefit of eribulin therapy in the case of patients developing large tumors with pericardial involvement, in which case, surgical resection is undesirable . In addition, a rare entity reported in the literature is pleomorphic liposarcoma of the breast. Throughout the investigated scientific literature, this proliferation has been reported in two cases, one female and one male patient, presenting as recurrent breast lump with locally aggressive behavior . In both cases, modified radical mastectomy was performed, and microscopic examination revealed the typical pleomorphic liposarcoma aspect, with no lymph node metastases identified . The authors underlined the importance of imaging studies in order to examine the tumor mass and assess its resectability . To do this, mammography, ultrasonography, CT scan, and MRI can be used . Although in both patients’ cases, the surgical resection margins were declared negative for tumor infiltration, both tumors recurred after surgery . The aggressive clinical course and propensity for local recurrence are regarded as features strongly associated with the pleomorphic liposarcoma subtype, although sarcoma of the breast is incompletely understood due to its remarkable uncommonness . Pleomorphic liposarcoma of the digestive tract is also a rare finding. However, we identified one liposarcoma of the pancreas and two liposarcoma involving the small intestine . All the investigated lesions were surgically excised, and histopathological and immunohistochemical studies were carried out . The tumor cells within the analyzed tumors showed positive S100 staining and were negative for CD34 and SMA . Al-Attar et al. noted a mutant TP53 immunophenotype with hyperexpression . None of the patients diagnosed with pleomorphic liposarcoma of the digestive tract reportedly developed metastases during the follow-up period . Although liposarcoma is an exceptional finding in the female reproductive tract, clinicians should be aware of pleomorphic liposarcoma of the uterine corpus . Valenciaga et al. reported a case of a 70-year-old patient who developed a liposarcoma of the uterus for which she received chemotherapy followed by a hysterectomy but was found with liver secondary determination 15 months after the surgery . During the study, the patient received entrectinib as part of a clinical trial implying a multicenter global phase II basket study of this treatment for patients with metastatic solid tumors associated with genetic mutations such as NTRK, ROS 1, or ALK gene rearrangements . Primary pleomorphic liposarcoma of the bone is also a rare clinical finding, with only two cases described within the past five years . Tiemeier et al. described an intramedullary adipose tissue tumor mass located within the proximal tibia discovered in an 18-year-old male . The lesion was treated with chemotherapy, comprising methotrexate, doxorubicin, and cisplatin, and later on, the tumor was surgically resected, and further reconstruction of the distal femur and proximal tibia was performed using endoprothesis . Immunohistochemical analysis of the tumor revealed expression of FABP4/aP2, a marker of adipocytic differentiation in the neoplastic cells, while S100, CD34, MDM2, and SMA were negative . The second case identified is that of a 20-year-old female who developed a pleomorphic liposarcoma of the femur, for which she underwent wide surgical resection and chemotherapy . Immunohistochemical analysis of the tumor showed positive 100 staining, while expression of CD34, MDM2, and SMA was absent in the neoplastic cells . No osteoblastic differentiation was disclosed, and the proliferation was negative for SATB2 and CD99 . The patient succumbed to her disease 8 months after the initial diagnosis, as she had developed local recurrence and liver metastases . Having considered all the reports discussed above, summarizes the possible primary locations of pleomorphic liposarcomas. Furthermore, we also summarized the immunohistochemical markers used for diagnosing pleomorphic liposarcomas . Pleomorphic liposarcoma and myxoid pleomorphic liposarcoma are the rarest and most aggressive malignant tumors with adipocytic differentiation. Pleomorphic liposarcoma usually occurs within the retroperitoneum and somatic soft tissue of the extremities, but lesions with peculiar locations can also be encountered. Pleomorphic liposarcoma of the viscera can be a diagnostic challenge, considering the histopathological aspect of the proliferation, often mimicking a metastatic tumor. Microscopic examination is a key aspect in determining the correct diagnosis, due to the specific presence of malignant pleomorphic lipoblasts. Ancillary studies may be of limited use with regard to pleomorphic liposarcoma, and it is often limited to ruling out other malignancies with a specific immunohistochemical background, such as dedifferentiated liposarcoma. Pathogenesis and histopathological aspects of this liposarcoma variant are still incompletely understood, considering the scarce data available on this malignancy, due to the limited number of cases reported so far. |
24-Month clinical evaluation of cervical restorations bonded using radio-opaque universal adhesive compared to conventional universal adhesive in carious cervical lesions: A randomized clinical trial | 30c0123c-4c0b-417d-8b37-cd0c519225e9 | 11828892 | Dentistry[mh] | Over recent decades, carious cervical lesions have been treated with various materials, such as glass-ionomer cements, resin-modified glass-ionomer cements, giomers and resin composites . Resin composites are frequently used for class V lesions due to their aesthetic potentials and capacity of bonding to both enamel and dentin . Despite developments in restorative materials and techniques, restoring cervical lesions remains challenging. The clinical success of cervical composite restorations can be affected by several factors, including resin composite type, adhesive technique, tooth type, and the operator’s experience and skills . Cervical cavities present a unique challenge due to their non-retentive form and margins that terminate on dentin or cementum, which are less ideal for bonding and located close to the gingival margin . One of the primary difficulties with cervical lesions is achieving a complete seal of the cavity, which can lead to microleakage. This microleakage may cause issues such as marginal discoloration and recurrent caries , the marginal failure of resin-based composites is often related to the quality of adhesion to tooth tissues . Adhesive systems play a crucial role in the success of resin composite restorations, prompting the development of various systems aimed at enhancing the bonding performance . The introduction of universal adhesives, the latest generation of adhesive systems, has simplified bonding procedures. Universal adhesives are adaptable systems that can be used in self-etch, selective-etch, or etch-and-rinse bonding strategies , allowing for reduced chair time and lower technique sensitivity, while maintaining reliable long-term bonding durability . Moreover, the inclusion of functional monomers, such as 10-methacryloyloxy-decyl-dihydrogen-phosphate (10-MDP), enhances bonding capabilities to various substrates, including resin composites, glass ceramics, zirconia, and metal alloys . Current research has largely focused on assessing the effectiveness and longevity of different materials to help practitioners select the most suitable options for clinical use. However, evidence remains inconclusive in endorsing a single material as the standard for restoring cervical lesions . Consequently, there is ongoing work to enhance the physical and adhesive properties of adhesive systems, driven by the dental market’s response to increased demand for adhesive advancements . A recent development in this field is the modified radio-opaque universal adhesive, Scotchbond™ Universal Plus Adhesive. This new formulation of universal adhesive has been modified by incorporating a new crosslinking resin, containing brominated dimethacrylate monomer providing dentin like radio-opacity, eliminating misinterpretation of MDP-based adhesive layer. Moreover, unlike other radio-opaque adhesive, it remains homogenous ensuring low viscosity and favourable handling while maintaining good mechanical properties. These benefit are accompanied with versatility, reduced technique sensitivity and enhanced durability present in the conventional universal adhesive . As with any new material, evidence-based data for this product remains limited, highlighting the need for clinical trials to establish consistent findings regarding its bonding performance in the oral environment after modifications in the chemical formula of the previous generation of universal adhesive. Class V clinical trials are considered the gold standard for assessing adhesive performance due to their unique ability to assess bonding effectiveness without macro-mechanical retention . The current clinical trial aimed to evaluate the performance of a universal adhesive containing a novel radiopaque crosslinking resin in comparison to a conventional universal adhesive over 24 months for restoring cervical carious lesions. The null hypothesis proposed was that there would be no difference in clinical performance between the radiopaque universal adhesive and the conventional universal adhesive in cervical restorations. Trial registration and study setting The protocol for the present study was submitted to the clinical trials registry NCT05509127 (19-08-2022) and received ethical approval from the Research Ethics Committee at the Faculty of Dentistry, Cairo University (18-11-22). The study was conducted within the Department of Conservative Dentistry. Trial design This study was a randomized clinical trial with a parallel, two-arm design, using a 1:1 allocation ratio and a superiority framework and reported according to the CONSORT guidelines . Recruitment Participants were recruited from the diagnostic center using convenient consecutive sampling based on the eligibility criteria between May 2022 and July 2022. Prior to enrollment, all candidates were informed about the study procedures and possible harms. Informed consent was obtained by having them sign an Arabic version of the consent form. Intervention was implemented between August 2022 and November 2022. Sample size calculation The sample size was calculated based on a previous study , in which success rate of resin composite cervical restorations using universal adhesive in total etch mode was 100% after 24 months. A two-tailed Z test was conducted to determine the difference between two independent proportions, with a 5% significance level (alpha) and 80% power. The minimum sample size required was 22 per group to detect a 30% difference. To account for potential dropouts, the sample size was increased by 15%, resulting in 25 teeth per group. Sample size calculation was performed using G*Power version 3.1.9.2 for Windows. Eligibility criteria Participants Inclusion criteria of participants in the study was to have carious cervical lesions in maxillary premolar teeth, age range between 20 and 40 years old, and exhibit only mild to moderate plaque accumulation according to Silness and Löe Plaque Index including score 0, 1 and 2. Exclusion criteria was cervical carious lesions in anterior teeth, molars, or mandibular teeth; non carious cervical lesions; any systemic conditions; allergies to resin; non-compliance; potential pregnancy; poor oral hygiene; heavy smoking; xerostomia; parafunctional habits or bruxism; and temporomandibular joint disorders. Teeth Teeth eligible for inclusion had small to moderate carious cervical lesions using visual-tactile examination (ICDAS scores 3 and 4) , vital maxillary premolars, and favorable occlusion with normal occlusal contact. Exclusion criteria of teeth included deep caries close to the pulp (less than 1 mm), irreversible pulpitis or pulp necrosis, dentin hypersensitivity, the likelihood of future prosthetic restoration on the teeth, and severe periodontal conditions. Sequence generation and allocation concealment Simple randomization method was used to generate numbers from 1 to 50 via the website ( https://www.random.org/sequences ) using random sequence generator and divided into two columns, designating either the intervention or control group. Allocation was concealed from the operating dentist, who selected a number from a sealed, opaque envelope. The current study was double blinded to both participants and outcome assessors who did not participate in any of the procedural steps in the present study. Implementation was done by a resident who was not involved in any of procedural steps or outcome assessment in the current trial. Intervention Tooth preparation Teeth to be prepared were anesthetized using buccal infiltration with a short needle and local anaesthetic solution (Articaine HCL 4% 1:100.000, Art Pharma Dent Pharmaceuticals, Giza, Egypt). Optimum shade of the restoration was selected before rubber dam isolation using direct mock-up technique using cured resin composite buttons, which was compared to the tooth and the closest shade to the tooth was determined .The teeth were then isolated with a rubber dam using the quadrant isolation technique, with a subgingival clamp employed for gingival retraction on the offending tooth. Class V cavity was prepared with a #330 or #245 bur using a high-speed contra-angled handpiece with oil-free air/water coolant. Mesio-distal width was limited before the labio-proximal line angles to avoid extending to the proximal surface, and occluso-gingival length was restricted to the cervical one-third. Carious tissue was eliminated till reaching firm affected dentin using low-speed large round carbide bur for hard carious dentin or sharp excavator for soft carious dentin, direction of excavation was from the periphery to the centre to avoid pulp exposure. The occlusal cavity margin was bevelled using a yellow-coded tapered finishing stone (TR-12). Each bur or diamond point was discarded after a maximum of five teeth . Adhesive procedures Enamel margins were etched for 15 s and dentin for 10 s with 3M™ Scotchbond™ Universal Etchant Gel. The surface was then rinsed for 15 s and dried with oil-free compressed air for an additional 15 s following the manufacturers’ recommendations. Afterwards Scotchbond™ Universal Plus Adhesive and Single Bond Universal Adhesive were applied according to the manufacturer’s instructions. Both adhesives were applied with a micro-brush, then agitated with the brush for 20 s, this was followed by gentle air-thinned for 5 s, and then light-curing with an LED curing unit (I-LED, Woodpecker, Guangxi, China) for 20 s. Restoration To standardize the restoration, a nano-filled resin composite (3M™ Filtek Z350 XT) was applied in both groups. Resin composite was placed in a maximum of 2 mm increments using dentin and enamel shades, dentin shade increment was cured for 40 s and enamel shade increment was cured for 20 s with an LED curing unit. Excess composite was removed using #12 blade, followed by a fine diamond bur (TR-12). Finishing and polishing were done with TOR VM discs on a low-speed handpiece with air/water coolant, following this sequence: coarse (70–90 μm), medium (40 μm), fine (24 μm), and super-fine (8 μm) aluminum oxide discs. Polishing was carried out using pre-impregnated rubber cups with intermittent water spray. (OneGloss PS, Shofu, California, USA). Table provides the names, descriptions, compositions, lot numbers, and manufacturers of the materials used. Figure shows restorative procedures in both groups. Outcome assessment The restorations were evaluated using modified USPHS criteria by two trained, calibrated, and blinded assessors with PhD degree and 15 years of experience in restorative dentistry, they are not involved in any procedural steps. Restorations were assessed at baseline, 12 and 24 months according to the outcome chart supplied (Table ). When assessors disagreed on a score, they discussed to reach a consensus. After training and calibration of assessors, interobserver agreement has shown Kappa coefficient of 0.92, which is nearly perfect agreement. Statistical analysis Data analysis was conducted using MedCalc software, version 22 for Windows (MedCalc Software Ltd, Ostend, Belgium). Categorical data were presented as frequencies and percentages. Intergroup comparisons between interventions at each follow-up were performed using the Chi-Square test with a significance level of ( p ≤ 0.05). Intragroup comparisons within each intervention between follow-up periods were conducted using Cochran’s Q test, with the significance level adjusted to ( p ≤ 0.016) after Bonferroni correction. Relative risk (RR) was calculated to assess clinical significance. The survival rate was analysed using the Kaplan-Meier method and the Log-rank test. The confidence interval was set at 95%, with 80% power, and all tests were two-tailed. The protocol for the present study was submitted to the clinical trials registry NCT05509127 (19-08-2022) and received ethical approval from the Research Ethics Committee at the Faculty of Dentistry, Cairo University (18-11-22). The study was conducted within the Department of Conservative Dentistry. Trial design This study was a randomized clinical trial with a parallel, two-arm design, using a 1:1 allocation ratio and a superiority framework and reported according to the CONSORT guidelines . Recruitment Participants were recruited from the diagnostic center using convenient consecutive sampling based on the eligibility criteria between May 2022 and July 2022. Prior to enrollment, all candidates were informed about the study procedures and possible harms. Informed consent was obtained by having them sign an Arabic version of the consent form. Intervention was implemented between August 2022 and November 2022. Sample size calculation The sample size was calculated based on a previous study , in which success rate of resin composite cervical restorations using universal adhesive in total etch mode was 100% after 24 months. A two-tailed Z test was conducted to determine the difference between two independent proportions, with a 5% significance level (alpha) and 80% power. The minimum sample size required was 22 per group to detect a 30% difference. To account for potential dropouts, the sample size was increased by 15%, resulting in 25 teeth per group. Sample size calculation was performed using G*Power version 3.1.9.2 for Windows. This study was a randomized clinical trial with a parallel, two-arm design, using a 1:1 allocation ratio and a superiority framework and reported according to the CONSORT guidelines . Participants were recruited from the diagnostic center using convenient consecutive sampling based on the eligibility criteria between May 2022 and July 2022. Prior to enrollment, all candidates were informed about the study procedures and possible harms. Informed consent was obtained by having them sign an Arabic version of the consent form. Intervention was implemented between August 2022 and November 2022. The sample size was calculated based on a previous study , in which success rate of resin composite cervical restorations using universal adhesive in total etch mode was 100% after 24 months. A two-tailed Z test was conducted to determine the difference between two independent proportions, with a 5% significance level (alpha) and 80% power. The minimum sample size required was 22 per group to detect a 30% difference. To account for potential dropouts, the sample size was increased by 15%, resulting in 25 teeth per group. Sample size calculation was performed using G*Power version 3.1.9.2 for Windows. Participants Inclusion criteria of participants in the study was to have carious cervical lesions in maxillary premolar teeth, age range between 20 and 40 years old, and exhibit only mild to moderate plaque accumulation according to Silness and Löe Plaque Index including score 0, 1 and 2. Exclusion criteria was cervical carious lesions in anterior teeth, molars, or mandibular teeth; non carious cervical lesions; any systemic conditions; allergies to resin; non-compliance; potential pregnancy; poor oral hygiene; heavy smoking; xerostomia; parafunctional habits or bruxism; and temporomandibular joint disorders. Teeth Teeth eligible for inclusion had small to moderate carious cervical lesions using visual-tactile examination (ICDAS scores 3 and 4) , vital maxillary premolars, and favorable occlusion with normal occlusal contact. Exclusion criteria of teeth included deep caries close to the pulp (less than 1 mm), irreversible pulpitis or pulp necrosis, dentin hypersensitivity, the likelihood of future prosthetic restoration on the teeth, and severe periodontal conditions. Sequence generation and allocation concealment Simple randomization method was used to generate numbers from 1 to 50 via the website ( https://www.random.org/sequences ) using random sequence generator and divided into two columns, designating either the intervention or control group. Allocation was concealed from the operating dentist, who selected a number from a sealed, opaque envelope. The current study was double blinded to both participants and outcome assessors who did not participate in any of the procedural steps in the present study. Implementation was done by a resident who was not involved in any of procedural steps or outcome assessment in the current trial. Intervention Tooth preparation Teeth to be prepared were anesthetized using buccal infiltration with a short needle and local anaesthetic solution (Articaine HCL 4% 1:100.000, Art Pharma Dent Pharmaceuticals, Giza, Egypt). Optimum shade of the restoration was selected before rubber dam isolation using direct mock-up technique using cured resin composite buttons, which was compared to the tooth and the closest shade to the tooth was determined .The teeth were then isolated with a rubber dam using the quadrant isolation technique, with a subgingival clamp employed for gingival retraction on the offending tooth. Class V cavity was prepared with a #330 or #245 bur using a high-speed contra-angled handpiece with oil-free air/water coolant. Mesio-distal width was limited before the labio-proximal line angles to avoid extending to the proximal surface, and occluso-gingival length was restricted to the cervical one-third. Carious tissue was eliminated till reaching firm affected dentin using low-speed large round carbide bur for hard carious dentin or sharp excavator for soft carious dentin, direction of excavation was from the periphery to the centre to avoid pulp exposure. The occlusal cavity margin was bevelled using a yellow-coded tapered finishing stone (TR-12). Each bur or diamond point was discarded after a maximum of five teeth . Adhesive procedures Enamel margins were etched for 15 s and dentin for 10 s with 3M™ Scotchbond™ Universal Etchant Gel. The surface was then rinsed for 15 s and dried with oil-free compressed air for an additional 15 s following the manufacturers’ recommendations. Afterwards Scotchbond™ Universal Plus Adhesive and Single Bond Universal Adhesive were applied according to the manufacturer’s instructions. Both adhesives were applied with a micro-brush, then agitated with the brush for 20 s, this was followed by gentle air-thinned for 5 s, and then light-curing with an LED curing unit (I-LED, Woodpecker, Guangxi, China) for 20 s. Restoration To standardize the restoration, a nano-filled resin composite (3M™ Filtek Z350 XT) was applied in both groups. Resin composite was placed in a maximum of 2 mm increments using dentin and enamel shades, dentin shade increment was cured for 40 s and enamel shade increment was cured for 20 s with an LED curing unit. Excess composite was removed using #12 blade, followed by a fine diamond bur (TR-12). Finishing and polishing were done with TOR VM discs on a low-speed handpiece with air/water coolant, following this sequence: coarse (70–90 μm), medium (40 μm), fine (24 μm), and super-fine (8 μm) aluminum oxide discs. Polishing was carried out using pre-impregnated rubber cups with intermittent water spray. (OneGloss PS, Shofu, California, USA). Table provides the names, descriptions, compositions, lot numbers, and manufacturers of the materials used. Figure shows restorative procedures in both groups. Outcome assessment The restorations were evaluated using modified USPHS criteria by two trained, calibrated, and blinded assessors with PhD degree and 15 years of experience in restorative dentistry, they are not involved in any procedural steps. Restorations were assessed at baseline, 12 and 24 months according to the outcome chart supplied (Table ). When assessors disagreed on a score, they discussed to reach a consensus. After training and calibration of assessors, interobserver agreement has shown Kappa coefficient of 0.92, which is nearly perfect agreement. Inclusion criteria of participants in the study was to have carious cervical lesions in maxillary premolar teeth, age range between 20 and 40 years old, and exhibit only mild to moderate plaque accumulation according to Silness and Löe Plaque Index including score 0, 1 and 2. Exclusion criteria was cervical carious lesions in anterior teeth, molars, or mandibular teeth; non carious cervical lesions; any systemic conditions; allergies to resin; non-compliance; potential pregnancy; poor oral hygiene; heavy smoking; xerostomia; parafunctional habits or bruxism; and temporomandibular joint disorders. Teeth eligible for inclusion had small to moderate carious cervical lesions using visual-tactile examination (ICDAS scores 3 and 4) , vital maxillary premolars, and favorable occlusion with normal occlusal contact. Exclusion criteria of teeth included deep caries close to the pulp (less than 1 mm), irreversible pulpitis or pulp necrosis, dentin hypersensitivity, the likelihood of future prosthetic restoration on the teeth, and severe periodontal conditions. Simple randomization method was used to generate numbers from 1 to 50 via the website ( https://www.random.org/sequences ) using random sequence generator and divided into two columns, designating either the intervention or control group. Allocation was concealed from the operating dentist, who selected a number from a sealed, opaque envelope. The current study was double blinded to both participants and outcome assessors who did not participate in any of the procedural steps in the present study. Implementation was done by a resident who was not involved in any of procedural steps or outcome assessment in the current trial. Tooth preparation Teeth to be prepared were anesthetized using buccal infiltration with a short needle and local anaesthetic solution (Articaine HCL 4% 1:100.000, Art Pharma Dent Pharmaceuticals, Giza, Egypt). Optimum shade of the restoration was selected before rubber dam isolation using direct mock-up technique using cured resin composite buttons, which was compared to the tooth and the closest shade to the tooth was determined .The teeth were then isolated with a rubber dam using the quadrant isolation technique, with a subgingival clamp employed for gingival retraction on the offending tooth. Class V cavity was prepared with a #330 or #245 bur using a high-speed contra-angled handpiece with oil-free air/water coolant. Mesio-distal width was limited before the labio-proximal line angles to avoid extending to the proximal surface, and occluso-gingival length was restricted to the cervical one-third. Carious tissue was eliminated till reaching firm affected dentin using low-speed large round carbide bur for hard carious dentin or sharp excavator for soft carious dentin, direction of excavation was from the periphery to the centre to avoid pulp exposure. The occlusal cavity margin was bevelled using a yellow-coded tapered finishing stone (TR-12). Each bur or diamond point was discarded after a maximum of five teeth . Adhesive procedures Enamel margins were etched for 15 s and dentin for 10 s with 3M™ Scotchbond™ Universal Etchant Gel. The surface was then rinsed for 15 s and dried with oil-free compressed air for an additional 15 s following the manufacturers’ recommendations. Afterwards Scotchbond™ Universal Plus Adhesive and Single Bond Universal Adhesive were applied according to the manufacturer’s instructions. Both adhesives were applied with a micro-brush, then agitated with the brush for 20 s, this was followed by gentle air-thinned for 5 s, and then light-curing with an LED curing unit (I-LED, Woodpecker, Guangxi, China) for 20 s. Restoration To standardize the restoration, a nano-filled resin composite (3M™ Filtek Z350 XT) was applied in both groups. Resin composite was placed in a maximum of 2 mm increments using dentin and enamel shades, dentin shade increment was cured for 40 s and enamel shade increment was cured for 20 s with an LED curing unit. Excess composite was removed using #12 blade, followed by a fine diamond bur (TR-12). Finishing and polishing were done with TOR VM discs on a low-speed handpiece with air/water coolant, following this sequence: coarse (70–90 μm), medium (40 μm), fine (24 μm), and super-fine (8 μm) aluminum oxide discs. Polishing was carried out using pre-impregnated rubber cups with intermittent water spray. (OneGloss PS, Shofu, California, USA). Table provides the names, descriptions, compositions, lot numbers, and manufacturers of the materials used. Figure shows restorative procedures in both groups. Teeth to be prepared were anesthetized using buccal infiltration with a short needle and local anaesthetic solution (Articaine HCL 4% 1:100.000, Art Pharma Dent Pharmaceuticals, Giza, Egypt). Optimum shade of the restoration was selected before rubber dam isolation using direct mock-up technique using cured resin composite buttons, which was compared to the tooth and the closest shade to the tooth was determined .The teeth were then isolated with a rubber dam using the quadrant isolation technique, with a subgingival clamp employed for gingival retraction on the offending tooth. Class V cavity was prepared with a #330 or #245 bur using a high-speed contra-angled handpiece with oil-free air/water coolant. Mesio-distal width was limited before the labio-proximal line angles to avoid extending to the proximal surface, and occluso-gingival length was restricted to the cervical one-third. Carious tissue was eliminated till reaching firm affected dentin using low-speed large round carbide bur for hard carious dentin or sharp excavator for soft carious dentin, direction of excavation was from the periphery to the centre to avoid pulp exposure. The occlusal cavity margin was bevelled using a yellow-coded tapered finishing stone (TR-12). Each bur or diamond point was discarded after a maximum of five teeth . Enamel margins were etched for 15 s and dentin for 10 s with 3M™ Scotchbond™ Universal Etchant Gel. The surface was then rinsed for 15 s and dried with oil-free compressed air for an additional 15 s following the manufacturers’ recommendations. Afterwards Scotchbond™ Universal Plus Adhesive and Single Bond Universal Adhesive were applied according to the manufacturer’s instructions. Both adhesives were applied with a micro-brush, then agitated with the brush for 20 s, this was followed by gentle air-thinned for 5 s, and then light-curing with an LED curing unit (I-LED, Woodpecker, Guangxi, China) for 20 s. To standardize the restoration, a nano-filled resin composite (3M™ Filtek Z350 XT) was applied in both groups. Resin composite was placed in a maximum of 2 mm increments using dentin and enamel shades, dentin shade increment was cured for 40 s and enamel shade increment was cured for 20 s with an LED curing unit. Excess composite was removed using #12 blade, followed by a fine diamond bur (TR-12). Finishing and polishing were done with TOR VM discs on a low-speed handpiece with air/water coolant, following this sequence: coarse (70–90 μm), medium (40 μm), fine (24 μm), and super-fine (8 μm) aluminum oxide discs. Polishing was carried out using pre-impregnated rubber cups with intermittent water spray. (OneGloss PS, Shofu, California, USA). Table provides the names, descriptions, compositions, lot numbers, and manufacturers of the materials used. Figure shows restorative procedures in both groups. The restorations were evaluated using modified USPHS criteria by two trained, calibrated, and blinded assessors with PhD degree and 15 years of experience in restorative dentistry, they are not involved in any procedural steps. Restorations were assessed at baseline, 12 and 24 months according to the outcome chart supplied (Table ). When assessors disagreed on a score, they discussed to reach a consensus. After training and calibration of assessors, interobserver agreement has shown Kappa coefficient of 0.92, which is nearly perfect agreement. Data analysis was conducted using MedCalc software, version 22 for Windows (MedCalc Software Ltd, Ostend, Belgium). Categorical data were presented as frequencies and percentages. Intergroup comparisons between interventions at each follow-up were performed using the Chi-Square test with a significance level of ( p ≤ 0.05). Intragroup comparisons within each intervention between follow-up periods were conducted using Cochran’s Q test, with the significance level adjusted to ( p ≤ 0.016) after Bonferroni correction. Relative risk (RR) was calculated to assess clinical significance. The survival rate was analysed using the Kaplan-Meier method and the Log-rank test. The confidence interval was set at 95%, with 80% power, and all tests were two-tailed. Demographic data This present study was conducted on 50 patients with 50 cervical carious lesions. After 24 months 44 restorations were assessed with 88% retention rate, six patients were dropped-out; four at 12 months and two at 24 months follow-up. CONSORT flow diagram illustrates participant flow through each stage of the trial (Fig. ). The mean age of the participants in the current trial was 28.5 ± 5.7 years, there was no statistically significant difference between both groups regarding age ( p = 0.511). Additionally, there was 12 males and 38 females in the current study, there was no statistically significant difference between both groups in gender distribution ( p = 0.5121). Clinical evaluation The comparison between adhesives showed no statistically significant differences within all follow-up periods regarding all tested parameters ( p > 0.05). Intragroup comparison between follow-up periods within both adhesives has shown no statistically significant change in scores for all tested outcomes ( p > 0.016) except for marginal adaptation within single bond universal, where there was statistically significant difference ( p = 0.005). At baseline, two restorations in the Scotchbond™ Universal Plus and four in the Single Bond Universal group exhibited postoperative sensitivity, which resolved in subsequent follow-ups. After 24 months, there was 87% less risk for score B in Scotchbond™ Universal Adhesive Plus when compared to Single Bond Universal (RR = 0.13, 95% CI 0.007161 to 2.3948, p = 0.1704) (Table ). Overall survival of Scotchbond™ Universal Plus and Single Bond Universal for carious cervical restorations was assessed after 24 months, no restorations scored B or C in Scotchbond™ Universal Adhesive Plus group after 24 months. However, in Single Bond Universal group three restorations scored B after 12 and 24 months in marginal adaptation and marginal discoloration. Kaplan-Meier analysis and Log-rank test showed no statistically significant difference between both materials ( p = 0.0769) (Fig. ). This present study was conducted on 50 patients with 50 cervical carious lesions. After 24 months 44 restorations were assessed with 88% retention rate, six patients were dropped-out; four at 12 months and two at 24 months follow-up. CONSORT flow diagram illustrates participant flow through each stage of the trial (Fig. ). The mean age of the participants in the current trial was 28.5 ± 5.7 years, there was no statistically significant difference between both groups regarding age ( p = 0.511). Additionally, there was 12 males and 38 females in the current study, there was no statistically significant difference between both groups in gender distribution ( p = 0.5121). The comparison between adhesives showed no statistically significant differences within all follow-up periods regarding all tested parameters ( p > 0.05). Intragroup comparison between follow-up periods within both adhesives has shown no statistically significant change in scores for all tested outcomes ( p > 0.016) except for marginal adaptation within single bond universal, where there was statistically significant difference ( p = 0.005). At baseline, two restorations in the Scotchbond™ Universal Plus and four in the Single Bond Universal group exhibited postoperative sensitivity, which resolved in subsequent follow-ups. After 24 months, there was 87% less risk for score B in Scotchbond™ Universal Adhesive Plus when compared to Single Bond Universal (RR = 0.13, 95% CI 0.007161 to 2.3948, p = 0.1704) (Table ). Overall survival of Scotchbond™ Universal Plus and Single Bond Universal for carious cervical restorations was assessed after 24 months, no restorations scored B or C in Scotchbond™ Universal Adhesive Plus group after 24 months. However, in Single Bond Universal group three restorations scored B after 12 and 24 months in marginal adaptation and marginal discoloration. Kaplan-Meier analysis and Log-rank test showed no statistically significant difference between both materials ( p = 0.0769) (Fig. ). After 24 months, cervical restorations in Scotchbond™ Universal Adhesive Plus showed 100% alpha score, while in Single Bond Universal group cervical restorations showed 85.7% alpha score, yet all restorations were clinically successful. There was no statistically significant difference between both adhesives for all tested criteria after 24 months ( p > 0.05), therefore, the null hypothesis cannot be rejected. The two adhesives showed similar clinical performance with regard to retention, postoperative sensitivity, and secondary caries. This may be due to similar composition of both adhesive mainly the functional monomers (10-MDP), Vitrebond co-polymer and HEMA, in addition to the application protocol and mild pH of 2.7 , . The MDP monomer enhances adhesion to tooth structure through a chemical bond with hydroxyapatite, a process referred to nano-layering. Additionally, the Vitrebond copolymer facilitates an ionic interaction between the carboxyl groups in polyalkenoic acid and hydroxyapatite in both enamel and dentin. This chemical reaction is considered fundamental to the bonding mechanism. The inclusion of the HEMA monomer makes the adhesive hydrophilic, which improves its wettability on the tooth surface . Current evidence suggests that for cervical lesions, universal adhesives should ideally be used with an etch-and-rinse protocol, which a previous systematic review concluded that it yields the best clinical outcomes . According to Hong et al. , the etch-and-rinse approach for universal adhesives provides superior clinical benefits, including improved retention, better marginal adaptation, and reduced marginal discoloration compared to the self-etch mode. However, Rodriguez et al. found no significant difference between the two adhesive strategies, indicating that the clinician’s preference and the specific clinical context are key factors in selecting the technique . Therefore, etch and rinse approach for universal adhesives was preferred in the present trial as it provides better retention of cervical restorations up to 36 months than self-etch approach . The bonding performance of adhesive agents is typically assessed by evaluating clinical performance in cervical restoration . To ensure standardized quality in clinical evaluations, various criteria have been developed, with the United States Public Health Service (USPHS) criteria being among the most widely adopted . Numerous clinical trials utilizing modified USPHS criteria support their validity and reliability . Regarding retention, no restoration loss occurred in either group during the trial, resulting in a 100% retention rate. This high retention has been similarly reported by previous trials for Single Bond Universal when using the etch-and-rinse technique , , . The excellent retention can be attributed to the chemical bonding from the 10-MDP monomer and Vitrebond copolymer, as previously discussed. This finding aligns with Carvalho et al. , who studied the bond durability of a mild two-step self-etch adhesive containing 10-MDP as a functional monomer and observed favorable results. Additionally, Alam et al. indicated that the survival rates of Scotchbond™ Universal Plus Adhesive and Scotchbond™ Universal Adhesive were comparable, regardless of the etching method, reflecting adequate bond strength and strong adhesive adaptation. The present study found no statistically significant difference in postoperative sensitivity between the two groups, aligning with previous clinical trials , – . At baseline, two cases in the intervention group and four in the control group exhibited postoperative sensitivity, which resolved in subsequent follow-ups. This sensitivity could have been related to factors other than the adhesive material, such as dentin etching, desiccation, gingival retraction that may expose the root surface immediately after restoration placement, finishing, polishing, or operational stress , . Concerning secondary caries, neither of the universal adhesives showed any reports of secondary caries and there were no statistically significant differences between the two materials at all follow-up intervals. Previous research has explored the effect of 10-MDP on caries inhibition potential, highlighting the significance of the acid-base resistant zone, which differs from the conventional hybrid layer and fluoride-releasing caries inhibition zone. This resistant zone is thought to play a crucial role in preventing secondary caries by sealing restoration margins and enhancing restoration longevity , . Marginal adaptation is a crucial indicator of the durability of dental restorations. Insufficient marginal integrity can lead to various complications, including gap formation, microleakage, recurrent caries, postoperative hypersensitivity, and ultimately pulp involvement . In the present study, Scotchbond™ Universal Plus showed 100% alpha score in marginal adaptation across various follow-up periods. However, Single Bond Universal showed statistically significant deterioration in marginal adaptation with 3 restorations scoring bravo after 24 months. The relatively satisfactory adaptation scores can be attributed to the etching protocol and the robust chemical reaction facilitated by the 10-MDP monomer and the Vitrebond copolymer, as previously suggested in research , , . Factors such as the etching protocol and agitation of the adhesive may have also improved bond strength and clinical performance , allowing for effective penetration of resin tags regardless of the adhesion protocol . The differences in adhesive thickness between Scotchbond™ Universal Plus and Single Bond Universal adhesives may have contributed to their varying behaviours in marginal quality. Alam et al. found a significant difference in viscosity, with Scotchbond™ Universal Plus having a mean viscosity of 50.2 ± 0.3 MPa compared to 115.5 ± 0.6 MPa for Single Bond Universal. The lower viscosity of Scotchbond™ Universal Plus can improve its wettability over the tooth surface, enhancing adaptation. While higher viscosity typically correlates with better mechanical properties, Scotchbond™ Universal Plus exhibited superior mechanical qualities despite its lower viscosity, likely due to its modified composition . Tsujimoto et al. found a positive correlation between the thickness of the adhesive layer and its bond strength. Scotchbond™ Universal Plus had a thinner adhesive layer in etch and rinse mode of 2.9 ± 0.2 μm, when compared to Single Bond Universal which showed an adhesive layer thickness of 6.1 ± 0.4 μm . Regarding marginal discoloration, no statistically significant differences were observed between the two groups at various intervals. Only one restoration in the control group received a Bravo score at the 24-month follow-up. The main factors to consider for this outcome are the use of the etch-and-rinse mode and the effect of 10-MDP and Vitrebond as previously mentioned. Higher rates of marginal discoloration have been noted in restorations with universal adhesives using the self-etch method, likely due to their reduced effectiveness in bonding to unetched enamel when compared to etched enamel . Marginal discoloration is often attributed to microleakage, allowing oral fluids and bacteria to penetrate . However, Kim et al. noted that marginal discoloration does not always indicate microleakage; only penetrating discoloration signifies its presence, while superficial discoloration may result from marginal chipping without evidence of microleakage. In such cases, repair or refurbishment could be a more conservative approach than complete restoration replacement . Few trials are available in the literature regarding restoration of carious cervical lesions and root caries. A previous clinical trial by Abdalla and Garcia-Godoy found 100% retention rate of cervical restorations using resin composite after two years using etch and rinse approach, selective enamel etching and self-etch approach . Another trial by Nassar et al. found 81.5% survival of resin composite restorations preceded by self-etch adhesive after one year . Moreover AlHumaid et al. found 100% retention rate of flowable composite preceded by etch and rinse approach in cervical lesions after 18 months . Vural et al. also found 85% success rate of root caries restorations after five years using resin composite preceded by self-etch adhesive . A second study by Vural et al. found 84.3% retention rate of cervical restorations using resin composite preceded by etch and rinse adhesive system after three years . The previous clinical trials support the finding of the present trial with success rate ranging from 80 to 100% after one to five years of clinical service. In the present study six patients were dropped out due to not responding to the follow up recalls, four patients were dropped out at 12 months follow-up and two patients were dropped out at 24 months follow-up with 88% retention rate after 24 months, the retention rate was 92% in the intervention group and 84% in the control group. The two restorations lost due to follow-up at 24 months in Single Bond Universal group scored alpha at 12 months follow-up in all assessed criteria, accordingly it was unlikely to deteriorate based on the performance of other restorations in this group. According to literature, there is 25–26% participants drop-out in clinical trials . In the present clinical trial during sample size calculation 15% was added to compensate for possible dropouts, however only 12% dropped out at the end of the trial in both group, which was less than previously mentioned average dropout rate, therefore the power of the sample size was not affected. To our knowledge, the current study was pioneer in assessing the clinical performance of Scotchbond™ Universal Adhesive Plus with its modified formula. According to ADA standards, for full acceptance of an adhesive restorations, clinical failures such as loss of restorations and microleakage should be limited to 10% after 18 months . There were no failures after 24 months, therefore, both adhesive materials can be recommended for restoration of carious cervical lesions. One of the limitations in the present study is the relatively small sample size. A sufficient sample size is recommended to detect any differences between both test groups with enhanced power and external validity. Moreover, extending the follow-up to at least three years is recommended in order to grant the full acceptance for adhesive materials . The newly introduced version of universal adhesive “Scotchbond™ Universal Plus Adhesive” showed satisfactory clinical performance, comparable to its predecessor Single bond Universal Adhesive after 24 months, despite its modified formula. The new formula enhanced the marginal quality of the upgraded version of universal adhesive. A larger sample size is needed to validate the findings of this study, along with extended follow-up periods to identify any long-term failures associated with both adhesives. |
Prescription Sequence Symmetry Analysis (PSSA) to assess prescribing cascades: a step-by-step guide | 8012eb0f-9198-4d61-b41e-f509b1ff5afa | 10782776 | Pharmacology[mh] | A prescribing cascade has been defined as a misinterpretation of an adverse drug reaction (ADR) as a new medical condition, which is subsequently treated with another medication . In some cases, treating or preventing an ADR with another medication is justified but when the ADR is not acknowledged as such, the resulting prescribing cascade is considered problematic. It is important to identify, manage or prevent such prescribing cascades because they can lead to polypharmacy, adverse outcomes and unnecessary healthcare costs . In recent years, the number of studies addressing prescribing cascades is increasing . To identify and quantify potential prescribing cascades prescription or administrative databases can be used. The sequence symmetry analysis (SSA) method has been used to assess the association between a medication and for example hospital diagnoses or aids (e.g. incontinence products). In the field of pharmacoepidemiology, the prescription sequence symmetry analysis (PSSA) method is used to assess the association between two medicines . PSSA quantifies prescribing cascades with the sequence ratio (SR) as a risk estimate. The crude SR (cSR) is calculated by the number of patients who initiated the initial medication (i.e. index medication) first and the medication to treat the ADR (i.e. marker medication) second divided by the number of patients who initiated the marker medication first and the index medication second . This cSR is sensitive to prescribing trends over time, e.g., due to expired patents or a change in treatment guideline recommendations. To correct for prescribing trends, the null-effect SR (SRnull) is calculated. The null-effect SR takes the prescribing trends in the background population into account, by computing an expected SR based on the probability of the sequence of initiation of the marker medication after the index medication in the absence of any causal association . By dividing the cSR by the SRnull, the adjusted SR (aSR) is calculated. If the aSR is more than 1.0, there is an increased probability that a prescribing cascade has occurred due to an ADR of the index medication . PSSA is a useful method to identify the occurrence of potential prescribing cascades in clinical practice when the ADR is known or strongly hypothesized. PSSA is easy to implement, requiring prescription data that include a patient identifier and prescription dates . Its conceptual framework but also statistical codes for calculating cSR, SRnull and aSR have been explained in previous papers . Previous studies, however, illustrate that different data collection methods, definitions and assumptions are used for PSSA . So far, the considerations for medication data collection and setting of time periods for relevant parameters have not been described or discussed in detail. This is needed to support future studies in providing meaningful information for identifying prescribing cascades or evaluating of the effects of interventions to reverse or prevent prescribing cascades. Therefore, our aim is 1) to provide a step-by-step guide for executing PSSA to assess prescribing cascades, and 2) to show the impact of changing assumptions on aSRs using two examples: angiotensin-converting enzyme inhibitors (ACEi)-induced cough followed by antitussives, and cardiovascular medication-induced erectile dysfunction followed by phosphodiesterase inhibitors. Previous papers have explained the PSSA method using a general pharmacoepidemiologic perspective . We briefly describe the basic principles of PSSA to assess prescribing cascades using the ACEi-antitussives prescribing cascade as example. When there would be no causal relationship between the use of an ACEi and the need for an antitussive, the probability that a patient is prescribed an antitussive before or after the start of an ACEi is expected to be equal. This would result in a symmetrical or random prescribing pattern of the initiation of the marker medication around the initiation of the index medication. In contrast, when the ACEi leads to an increased probability of cough that in turn is treated with the antitussive, it would result in an asymmetrical prescribing pattern of initial prescriptions (Fig. ). To quantify the occurrence of prescribing cascades, one first determines the cSR as described in the introduction. Next, to adjust for prescribing trends over time, the null-effect SR (SRnull) is calculated. The SRnull is calculated with the following two formulas: \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$Pa=\frac{\sum_{m=1}^{\mu }[{I}_{m}* \left(\Sigma\;patients\;starting\;marker\;after\;start\;date\;index\right)] }{\sum_{m=1}^{\mu }[{I}_{m}* \left(\left(\Sigma\;patients\;starting\;marker\;prior\;to\;start\;date\;index\right)+\left(\Sigma\;patients\;starting\;marker\;after\;start\;date\;index\right)\right)]}$$\end{document} P a = ∑ m = 1 μ [ I m ∗ Σ p a t i e n t s s t a r t i n g m a r k e r a f t e r s t a r t d a t e i n d e x ] ∑ m = 1 μ [ I m ∗ Σ p a t i e n t s s t a r t i n g m a r k e r p r i o r t o s t a r t d a t e i n d e x + Σ p a t i e n t s s t a r t i n g m a r k e r a f t e r s t a r t d a t e i n d e x ] Here, Pa stands for the overall probability that the marker medication will be prescribed after the index medication when the prescription pattern of the background population is taken into consideration; m indicates the consecutive day of the index medication of the study and µ indicates the last day of the study period. Im is the number of persons receiving their first index medication on the specific day and the start date index is the first day an index medication is prescribed . When the Pa is calculated, the null-effect SR can be calculated with the following formula: \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$Null-effect\;SR= \frac{Pa}{1-Pa}$$\end{document} N u l l - e f f e c t S R = Pa 1 - P a By dividing the cSR by the SRnull, the aSR is calculated. For the detailed calculations of the cSR, SRnull, aSR with 95% confidence interval (95% CI), we refer to Electronic Supplementary Material (ESM) . Executing PSSA for prescribing cascades Considerations for the data collection Although collecting data to conduct a PSSA for a prescribing cascade may seem straightforward, several decisions have to be made (Table ). First of all, sufficient history and a predefined follow-up period is needed. Since loss to follow-up could be non-random and the occurrence of true prescribing cascades may take some time, the effect of including patients with incomplete follow-up could bias the SR for the prescribing cascade. To define a patient as lost to follow-up, one should not limit this to only considering dispensings of the index or marker medication, but should also account for continuous enrollment when using claims datasets. To ensure continuous enrollment for patient eligibility, the presence of dispensings of any medication or other claims data during the period of interest can be used . Medication class level Interest in quantifying a prescribing cascade may start with a particular case, for example, lisinopril followed by codeine which may be indicative of treating the ADR cough. The first consideration is whether the ADR cough is a group effect of the index medication or an ADR caused by this individual medication. As cough is indeed a known group side effect of ACEi , assessing the prescribing cascade at the class level as index medication makes more sense than at the individual substance level lisinopril. In contrast, for amiodarone-induced hypothyroidism including the group of all antiarrhythmics would be inappropriate since hypothyroidism is a specific ADR of amiodarone . Similarly, for the selection of the marker medication, the medication class level needs to be considered. ACEi-induced cough can be treated with codeine but also with other antitussives . Moreover, focussing on antitussives only could fail to show the complete spectrum of prescribing cascades related to ACEi-induced cough. Cough has also been treated with salbutamol, antihistamines or antibiotics , so to quantify the ACEi-induced cough prescribing cascade, a range of marker medication classes could be included. However, including too many marker medications that are not very specific for treating cough could result in lower SRs in the PSSA, as the chances to identify asymmetry in prescribing sequences of the index and marker medication will decrease. Sensitivity analyses regarding the class level could be used to test the robustness of the PSSA results. Combination products Next, inclusion of combination products of the index as well as the marker medication should be considered. For example, when assessing the prescribing cascade of calcium channel blocker (CCB)-induced edema treated with a diuretic, CCBs can be defined as the index medication. The initiation of diuretics and/or combination products of CCBs with diuretics can be defined as the marker medication. Inclusion or exclusion of combination products of CCBs with diuretics as marker medication can change the aSR. In contrast, the combination of CCBs with a beta-blocker can be defined as index medication, since beta-blockers are not likely to be prescribed for treating edema. Dose dependence of ADR Some ADRs can be related to the actual dose of medication. If an ADR is dose-related, a false negative result could be found in a database with a high prevalence of patients using low doses of the index medication. If a dose-relationship is expected, it could be relevant to perform subanalyses examining aSRs for low and high dose prescriptions. Risk factors of ADR Certain comorbidities -but also sex or age- may influence the development of the ADR and thus the likelihood of a prescribing cascade. For example, for the prescribing cascade ACEi – antitussives, having obstructive airway disease can induce or contribute to the outcome of dry cough . Also, older patients are more likely to develop an ADR due to more comorbid conditions, polypharmacy and a higher sensitivity for medication effects . Amiodarone-induced hypothyroidism is more frequently reported in women . Therefore, the prescribing cascade could be more likely in women than in men. In such cases, it can be relevant to conduct subgroup analyses or select only the patients at risk of developing the ADR for identifying or quantifying the prescribing cascade. Co-medication It is not uncommon that there is co-medication that could result in the same ADR and thus the same prescribing cascade. For example, the ADR erectile dysfunction has been documented for a variety of cardiovascular medication groups, such as ACEi, betablockers and diuretics. If the prescribing cascade of interest is solely ACEi-induced erectile dysfunction, one could include patients who only use ACEi (i.e., sole users, patients 1, 2 and 4 in Fig. ). Patients who use other subgroups of cardiovascular medication could be excluded, as the sole effect of ACEi cannot be determined for these patients (Fig. , patients 3, 5 and 6, where the PSSA cannot differentiate between erectile dysfunction possibly caused by the ACEi versus the co-medication or due to a combination of both). Misclassifications can occur when co-medication is not adequately taken into account (Fig. , patients 2 and 3, where the sequence of phosphodiesterase (PDE) inhibitors followed by diuretics and betablockers respectively would be found when ACEi was disregarded). On the other hand, when the use of co-medication is unrelated to the index medication and its initiation is expected to be stable over the time period that is studied, it may not influence the asymmetry caused by the index medication of interest in the PSSA calculations. Alternatively, multiple subgroup analyses could be performed to study the impact of comedication, e.g., for patients using one versus two or three and more classes of cardiovascular medication causing the same ADR. However, including too many different classes for the index medication may lead to more risk of bias, such as confounding by indication. Parameters in PSSA analysis There are a number of time windows that should be imposed both for the sequence index-marker as well as for the sequence marker-index when conducting PSSA to assess prescribing cascades. Relevant parameters that are often considered for PSSA include the washout window to identify incident users and the exposure window for assessing the associations. In addition, a blackout period and continued exposure interval (CEI), which have been defined in other pharmacoepidemiological studies, can be relevant when using PSSA for assessing prescribing cascades . These parameters can best be imposed at medication episode level, for which a definition of medication discontinuation is needed. This could be equal to the period set for the washout window. Importantly, one should first establish continuous enrollment during the washout and the exposure window for patient eligibility (see also previous paragraph). Washout window The washout window (also called ‘waiting time period’ or ‘run-in period’) is the period that is imposed as a look-back period to ensure that the index or marker medication is indeed a first prescription (incidence) and exclude any prevalent users of the index or marker medication . By selecting incident users, the initiation of the marker medication after an index medication is more likely to indicate the treatment for a new ADR than treatment for an ongoing medical condition . Of note, dose-dependent ADRs may occur after dose increases in prevalent users and some ADRs only develop after continued exposure . The washout window should depend on the maximum period that medication is dispensed in a country to make sure that any prevalent users are excluded. In the Netherlands, chronic medication is generally dispensed for 3 months but this can be extended to a maximum of 6 months. A washout window of 12 months could be considered suitable taking into account any medication supply a patient could have in stock, and reducing the chance of including prevalent users. In previous PSSA studies, the washout window was set at 6 or 12 months . Figure shows examples of episodes that could be excluded because the criteria for the washout window are not met (examples A and B). Exposure window The exposure window (also called ‘exposure time window’ or ‘observation period’) is the defined follow-up period to capture the pairs of incident index and marker medication users and vice versa, so the maximum allowed time period between the start of the index and the start of the marker medication . The exposure window should depend on the expected time onset of the ADR caused by the index medication . For example, Pouwels et al. used an exposure window of 4 weeks in their study of ACEi-induced urinary tract infections, as reduced urine output and reduced glomerular filtration rate (GFR) after ACEi initiation have been reported in relatively short-term studies ranging from 7 days to 8 weeks . It should be noted that such a short period is likely to decrease the sample size and the precision of the SR calculations . Also, a short exposure window can result in missing ADRs that take longer to occur, e.g., amiodarone can induce hypothyroidism even at 39 months after amiodarone initiation . It should also be kept in mind that for some ADRs patients can have a delay in consulting a healthcare provider . The downside of a longer exposure window is that the association between the index and the marker medication may become weaker and there is a higher risk of time-varying confounding . Generally, a 12-month exposure window is used to reduce the impact of such confounding, which includes ageing, disease progression and other time-varying variables, such as change in diet and/or environment . An exposure window of 12 months was found optimal for achieving acceptable sensitivity (61%; 95% CI 0.46–0.74) and high specificity (93%; 95% CI 0.87–0.96) for 165 tested index-marker pairs . Choosing different time periods can provide insight regarding the onset time of an ADR. In Fig. , examples are shown of episodes that are included (example G, H, I, and J) and excluded (example D and E) when an exposure window of 12 months is used. Of note, in the final year of data collection, episodes of the marker or index medication are disregarded when the exposure window requirement can no longer be met (example E). Continued exposure interval (CEI) To define continued exposure to a drug, the continued exposure interval (CEI) (also called ‘maximum permissible length without medication supply’) needs to be defined. This is the gap between the expected end date of one prescription and the date of the next prescription. In many pharmacoepidemiological studies, this would refer to the exposure to one medication or medication class but it can also be applied when associations between pairs of medication are studied . When studying prescribing cascades, the CEI is the maximum acceptable period without medication supply for the first medication . This period is imposed to ensure patients are likely to be exposed to the first medication when initiating the other medication. The CEI should be based on the common dispensing period of medication and the expected medication taking behaviour and could be similar to what is used in studies that measure adherence using dispensing data . For example, after the first prescription, medication prescriptions are generally repeated every 3 months in the Netherlands. Taking into account any supply the patient still possesses from previous dispensings (stockpile), a CEI of 4 to 6 months could be used. In Fig. , examples D and F show episodes that are excluded when a CEI of 4 months is imposed. It should be kept in mind that in most prescription and administrative databases the expected end date is theoretical and often determined based on the amount dispensed and the dose instruction when available. For medication that is prescribed without a clear dose instruction (e.g., use as needed), determining the expected end date can be difficult. Alternatively, one can set a longer period after the start date of each prescription, similarly to the definition of a discontinuation as well as the washout period, to exclude cases where impact of the initial medication is no longer expected. When imposing the CEI it should be kept in mind that some medication can still result in ADRs after discontinuation. Blackout period The blackout period (also called ‘time-lag period’ and ‘lag time’) is the period immediately after the initiation of the first medication in which events, i.e. the start of the other medication, will not be taken into account. Such a period is relevant to ensure sufficient time for the development of the ADR induced by the index medication and subsequent treatment with the marker medication . Singh et al. studied the prescribing cascade of calcium channel blocker (CCB)-induced edema treated with diuretics . They imposed a blackout period of 7 days because the probability of developing edema within 7 days after initiating CCBs is low. Other studies mentioned that patients need to be excluded if they have their first prescription of the index and marker medication on the exact same date, as the sequence of the index and marker medication cannot be extracted (see Fig. , example C) . Formally, this is not defined as blackout period but is important to address as exclusion criterion. Although an ADR may occur immediately, it seems impossible that a prescribing cascade occurs on the day of the first prescription. Considerations for the data collection Although collecting data to conduct a PSSA for a prescribing cascade may seem straightforward, several decisions have to be made (Table ). First of all, sufficient history and a predefined follow-up period is needed. Since loss to follow-up could be non-random and the occurrence of true prescribing cascades may take some time, the effect of including patients with incomplete follow-up could bias the SR for the prescribing cascade. To define a patient as lost to follow-up, one should not limit this to only considering dispensings of the index or marker medication, but should also account for continuous enrollment when using claims datasets. To ensure continuous enrollment for patient eligibility, the presence of dispensings of any medication or other claims data during the period of interest can be used . Medication class level Interest in quantifying a prescribing cascade may start with a particular case, for example, lisinopril followed by codeine which may be indicative of treating the ADR cough. The first consideration is whether the ADR cough is a group effect of the index medication or an ADR caused by this individual medication. As cough is indeed a known group side effect of ACEi , assessing the prescribing cascade at the class level as index medication makes more sense than at the individual substance level lisinopril. In contrast, for amiodarone-induced hypothyroidism including the group of all antiarrhythmics would be inappropriate since hypothyroidism is a specific ADR of amiodarone . Similarly, for the selection of the marker medication, the medication class level needs to be considered. ACEi-induced cough can be treated with codeine but also with other antitussives . Moreover, focussing on antitussives only could fail to show the complete spectrum of prescribing cascades related to ACEi-induced cough. Cough has also been treated with salbutamol, antihistamines or antibiotics , so to quantify the ACEi-induced cough prescribing cascade, a range of marker medication classes could be included. However, including too many marker medications that are not very specific for treating cough could result in lower SRs in the PSSA, as the chances to identify asymmetry in prescribing sequences of the index and marker medication will decrease. Sensitivity analyses regarding the class level could be used to test the robustness of the PSSA results. Combination products Next, inclusion of combination products of the index as well as the marker medication should be considered. For example, when assessing the prescribing cascade of calcium channel blocker (CCB)-induced edema treated with a diuretic, CCBs can be defined as the index medication. The initiation of diuretics and/or combination products of CCBs with diuretics can be defined as the marker medication. Inclusion or exclusion of combination products of CCBs with diuretics as marker medication can change the aSR. In contrast, the combination of CCBs with a beta-blocker can be defined as index medication, since beta-blockers are not likely to be prescribed for treating edema. Dose dependence of ADR Some ADRs can be related to the actual dose of medication. If an ADR is dose-related, a false negative result could be found in a database with a high prevalence of patients using low doses of the index medication. If a dose-relationship is expected, it could be relevant to perform subanalyses examining aSRs for low and high dose prescriptions. Risk factors of ADR Certain comorbidities -but also sex or age- may influence the development of the ADR and thus the likelihood of a prescribing cascade. For example, for the prescribing cascade ACEi – antitussives, having obstructive airway disease can induce or contribute to the outcome of dry cough . Also, older patients are more likely to develop an ADR due to more comorbid conditions, polypharmacy and a higher sensitivity for medication effects . Amiodarone-induced hypothyroidism is more frequently reported in women . Therefore, the prescribing cascade could be more likely in women than in men. In such cases, it can be relevant to conduct subgroup analyses or select only the patients at risk of developing the ADR for identifying or quantifying the prescribing cascade. Co-medication It is not uncommon that there is co-medication that could result in the same ADR and thus the same prescribing cascade. For example, the ADR erectile dysfunction has been documented for a variety of cardiovascular medication groups, such as ACEi, betablockers and diuretics. If the prescribing cascade of interest is solely ACEi-induced erectile dysfunction, one could include patients who only use ACEi (i.e., sole users, patients 1, 2 and 4 in Fig. ). Patients who use other subgroups of cardiovascular medication could be excluded, as the sole effect of ACEi cannot be determined for these patients (Fig. , patients 3, 5 and 6, where the PSSA cannot differentiate between erectile dysfunction possibly caused by the ACEi versus the co-medication or due to a combination of both). Misclassifications can occur when co-medication is not adequately taken into account (Fig. , patients 2 and 3, where the sequence of phosphodiesterase (PDE) inhibitors followed by diuretics and betablockers respectively would be found when ACEi was disregarded). On the other hand, when the use of co-medication is unrelated to the index medication and its initiation is expected to be stable over the time period that is studied, it may not influence the asymmetry caused by the index medication of interest in the PSSA calculations. Alternatively, multiple subgroup analyses could be performed to study the impact of comedication, e.g., for patients using one versus two or three and more classes of cardiovascular medication causing the same ADR. However, including too many different classes for the index medication may lead to more risk of bias, such as confounding by indication. Parameters in PSSA analysis There are a number of time windows that should be imposed both for the sequence index-marker as well as for the sequence marker-index when conducting PSSA to assess prescribing cascades. Relevant parameters that are often considered for PSSA include the washout window to identify incident users and the exposure window for assessing the associations. In addition, a blackout period and continued exposure interval (CEI), which have been defined in other pharmacoepidemiological studies, can be relevant when using PSSA for assessing prescribing cascades . These parameters can best be imposed at medication episode level, for which a definition of medication discontinuation is needed. This could be equal to the period set for the washout window. Importantly, one should first establish continuous enrollment during the washout and the exposure window for patient eligibility (see also previous paragraph). Washout window The washout window (also called ‘waiting time period’ or ‘run-in period’) is the period that is imposed as a look-back period to ensure that the index or marker medication is indeed a first prescription (incidence) and exclude any prevalent users of the index or marker medication . By selecting incident users, the initiation of the marker medication after an index medication is more likely to indicate the treatment for a new ADR than treatment for an ongoing medical condition . Of note, dose-dependent ADRs may occur after dose increases in prevalent users and some ADRs only develop after continued exposure . The washout window should depend on the maximum period that medication is dispensed in a country to make sure that any prevalent users are excluded. In the Netherlands, chronic medication is generally dispensed for 3 months but this can be extended to a maximum of 6 months. A washout window of 12 months could be considered suitable taking into account any medication supply a patient could have in stock, and reducing the chance of including prevalent users. In previous PSSA studies, the washout window was set at 6 or 12 months . Figure shows examples of episodes that could be excluded because the criteria for the washout window are not met (examples A and B). Exposure window The exposure window (also called ‘exposure time window’ or ‘observation period’) is the defined follow-up period to capture the pairs of incident index and marker medication users and vice versa, so the maximum allowed time period between the start of the index and the start of the marker medication . The exposure window should depend on the expected time onset of the ADR caused by the index medication . For example, Pouwels et al. used an exposure window of 4 weeks in their study of ACEi-induced urinary tract infections, as reduced urine output and reduced glomerular filtration rate (GFR) after ACEi initiation have been reported in relatively short-term studies ranging from 7 days to 8 weeks . It should be noted that such a short period is likely to decrease the sample size and the precision of the SR calculations . Also, a short exposure window can result in missing ADRs that take longer to occur, e.g., amiodarone can induce hypothyroidism even at 39 months after amiodarone initiation . It should also be kept in mind that for some ADRs patients can have a delay in consulting a healthcare provider . The downside of a longer exposure window is that the association between the index and the marker medication may become weaker and there is a higher risk of time-varying confounding . Generally, a 12-month exposure window is used to reduce the impact of such confounding, which includes ageing, disease progression and other time-varying variables, such as change in diet and/or environment . An exposure window of 12 months was found optimal for achieving acceptable sensitivity (61%; 95% CI 0.46–0.74) and high specificity (93%; 95% CI 0.87–0.96) for 165 tested index-marker pairs . Choosing different time periods can provide insight regarding the onset time of an ADR. In Fig. , examples are shown of episodes that are included (example G, H, I, and J) and excluded (example D and E) when an exposure window of 12 months is used. Of note, in the final year of data collection, episodes of the marker or index medication are disregarded when the exposure window requirement can no longer be met (example E). Continued exposure interval (CEI) To define continued exposure to a drug, the continued exposure interval (CEI) (also called ‘maximum permissible length without medication supply’) needs to be defined. This is the gap between the expected end date of one prescription and the date of the next prescription. In many pharmacoepidemiological studies, this would refer to the exposure to one medication or medication class but it can also be applied when associations between pairs of medication are studied . When studying prescribing cascades, the CEI is the maximum acceptable period without medication supply for the first medication . This period is imposed to ensure patients are likely to be exposed to the first medication when initiating the other medication. The CEI should be based on the common dispensing period of medication and the expected medication taking behaviour and could be similar to what is used in studies that measure adherence using dispensing data . For example, after the first prescription, medication prescriptions are generally repeated every 3 months in the Netherlands. Taking into account any supply the patient still possesses from previous dispensings (stockpile), a CEI of 4 to 6 months could be used. In Fig. , examples D and F show episodes that are excluded when a CEI of 4 months is imposed. It should be kept in mind that in most prescription and administrative databases the expected end date is theoretical and often determined based on the amount dispensed and the dose instruction when available. For medication that is prescribed without a clear dose instruction (e.g., use as needed), determining the expected end date can be difficult. Alternatively, one can set a longer period after the start date of each prescription, similarly to the definition of a discontinuation as well as the washout period, to exclude cases where impact of the initial medication is no longer expected. When imposing the CEI it should be kept in mind that some medication can still result in ADRs after discontinuation. Blackout period The blackout period (also called ‘time-lag period’ and ‘lag time’) is the period immediately after the initiation of the first medication in which events, i.e. the start of the other medication, will not be taken into account. Such a period is relevant to ensure sufficient time for the development of the ADR induced by the index medication and subsequent treatment with the marker medication . Singh et al. studied the prescribing cascade of calcium channel blocker (CCB)-induced edema treated with diuretics . They imposed a blackout period of 7 days because the probability of developing edema within 7 days after initiating CCBs is low. Other studies mentioned that patients need to be excluded if they have their first prescription of the index and marker medication on the exact same date, as the sequence of the index and marker medication cannot be extracted (see Fig. , example C) . Formally, this is not defined as blackout period but is important to address as exclusion criterion. Although an ADR may occur immediately, it seems impossible that a prescribing cascade occurs on the day of the first prescription. Although collecting data to conduct a PSSA for a prescribing cascade may seem straightforward, several decisions have to be made (Table ). First of all, sufficient history and a predefined follow-up period is needed. Since loss to follow-up could be non-random and the occurrence of true prescribing cascades may take some time, the effect of including patients with incomplete follow-up could bias the SR for the prescribing cascade. To define a patient as lost to follow-up, one should not limit this to only considering dispensings of the index or marker medication, but should also account for continuous enrollment when using claims datasets. To ensure continuous enrollment for patient eligibility, the presence of dispensings of any medication or other claims data during the period of interest can be used . Medication class level Interest in quantifying a prescribing cascade may start with a particular case, for example, lisinopril followed by codeine which may be indicative of treating the ADR cough. The first consideration is whether the ADR cough is a group effect of the index medication or an ADR caused by this individual medication. As cough is indeed a known group side effect of ACEi , assessing the prescribing cascade at the class level as index medication makes more sense than at the individual substance level lisinopril. In contrast, for amiodarone-induced hypothyroidism including the group of all antiarrhythmics would be inappropriate since hypothyroidism is a specific ADR of amiodarone . Similarly, for the selection of the marker medication, the medication class level needs to be considered. ACEi-induced cough can be treated with codeine but also with other antitussives . Moreover, focussing on antitussives only could fail to show the complete spectrum of prescribing cascades related to ACEi-induced cough. Cough has also been treated with salbutamol, antihistamines or antibiotics , so to quantify the ACEi-induced cough prescribing cascade, a range of marker medication classes could be included. However, including too many marker medications that are not very specific for treating cough could result in lower SRs in the PSSA, as the chances to identify asymmetry in prescribing sequences of the index and marker medication will decrease. Sensitivity analyses regarding the class level could be used to test the robustness of the PSSA results. Combination products Next, inclusion of combination products of the index as well as the marker medication should be considered. For example, when assessing the prescribing cascade of calcium channel blocker (CCB)-induced edema treated with a diuretic, CCBs can be defined as the index medication. The initiation of diuretics and/or combination products of CCBs with diuretics can be defined as the marker medication. Inclusion or exclusion of combination products of CCBs with diuretics as marker medication can change the aSR. In contrast, the combination of CCBs with a beta-blocker can be defined as index medication, since beta-blockers are not likely to be prescribed for treating edema. Dose dependence of ADR Some ADRs can be related to the actual dose of medication. If an ADR is dose-related, a false negative result could be found in a database with a high prevalence of patients using low doses of the index medication. If a dose-relationship is expected, it could be relevant to perform subanalyses examining aSRs for low and high dose prescriptions. Risk factors of ADR Certain comorbidities -but also sex or age- may influence the development of the ADR and thus the likelihood of a prescribing cascade. For example, for the prescribing cascade ACEi – antitussives, having obstructive airway disease can induce or contribute to the outcome of dry cough . Also, older patients are more likely to develop an ADR due to more comorbid conditions, polypharmacy and a higher sensitivity for medication effects . Amiodarone-induced hypothyroidism is more frequently reported in women . Therefore, the prescribing cascade could be more likely in women than in men. In such cases, it can be relevant to conduct subgroup analyses or select only the patients at risk of developing the ADR for identifying or quantifying the prescribing cascade. Co-medication It is not uncommon that there is co-medication that could result in the same ADR and thus the same prescribing cascade. For example, the ADR erectile dysfunction has been documented for a variety of cardiovascular medication groups, such as ACEi, betablockers and diuretics. If the prescribing cascade of interest is solely ACEi-induced erectile dysfunction, one could include patients who only use ACEi (i.e., sole users, patients 1, 2 and 4 in Fig. ). Patients who use other subgroups of cardiovascular medication could be excluded, as the sole effect of ACEi cannot be determined for these patients (Fig. , patients 3, 5 and 6, where the PSSA cannot differentiate between erectile dysfunction possibly caused by the ACEi versus the co-medication or due to a combination of both). Misclassifications can occur when co-medication is not adequately taken into account (Fig. , patients 2 and 3, where the sequence of phosphodiesterase (PDE) inhibitors followed by diuretics and betablockers respectively would be found when ACEi was disregarded). On the other hand, when the use of co-medication is unrelated to the index medication and its initiation is expected to be stable over the time period that is studied, it may not influence the asymmetry caused by the index medication of interest in the PSSA calculations. Alternatively, multiple subgroup analyses could be performed to study the impact of comedication, e.g., for patients using one versus two or three and more classes of cardiovascular medication causing the same ADR. However, including too many different classes for the index medication may lead to more risk of bias, such as confounding by indication. Interest in quantifying a prescribing cascade may start with a particular case, for example, lisinopril followed by codeine which may be indicative of treating the ADR cough. The first consideration is whether the ADR cough is a group effect of the index medication or an ADR caused by this individual medication. As cough is indeed a known group side effect of ACEi , assessing the prescribing cascade at the class level as index medication makes more sense than at the individual substance level lisinopril. In contrast, for amiodarone-induced hypothyroidism including the group of all antiarrhythmics would be inappropriate since hypothyroidism is a specific ADR of amiodarone . Similarly, for the selection of the marker medication, the medication class level needs to be considered. ACEi-induced cough can be treated with codeine but also with other antitussives . Moreover, focussing on antitussives only could fail to show the complete spectrum of prescribing cascades related to ACEi-induced cough. Cough has also been treated with salbutamol, antihistamines or antibiotics , so to quantify the ACEi-induced cough prescribing cascade, a range of marker medication classes could be included. However, including too many marker medications that are not very specific for treating cough could result in lower SRs in the PSSA, as the chances to identify asymmetry in prescribing sequences of the index and marker medication will decrease. Sensitivity analyses regarding the class level could be used to test the robustness of the PSSA results. Next, inclusion of combination products of the index as well as the marker medication should be considered. For example, when assessing the prescribing cascade of calcium channel blocker (CCB)-induced edema treated with a diuretic, CCBs can be defined as the index medication. The initiation of diuretics and/or combination products of CCBs with diuretics can be defined as the marker medication. Inclusion or exclusion of combination products of CCBs with diuretics as marker medication can change the aSR. In contrast, the combination of CCBs with a beta-blocker can be defined as index medication, since beta-blockers are not likely to be prescribed for treating edema. Some ADRs can be related to the actual dose of medication. If an ADR is dose-related, a false negative result could be found in a database with a high prevalence of patients using low doses of the index medication. If a dose-relationship is expected, it could be relevant to perform subanalyses examining aSRs for low and high dose prescriptions. Certain comorbidities -but also sex or age- may influence the development of the ADR and thus the likelihood of a prescribing cascade. For example, for the prescribing cascade ACEi – antitussives, having obstructive airway disease can induce or contribute to the outcome of dry cough . Also, older patients are more likely to develop an ADR due to more comorbid conditions, polypharmacy and a higher sensitivity for medication effects . Amiodarone-induced hypothyroidism is more frequently reported in women . Therefore, the prescribing cascade could be more likely in women than in men. In such cases, it can be relevant to conduct subgroup analyses or select only the patients at risk of developing the ADR for identifying or quantifying the prescribing cascade. It is not uncommon that there is co-medication that could result in the same ADR and thus the same prescribing cascade. For example, the ADR erectile dysfunction has been documented for a variety of cardiovascular medication groups, such as ACEi, betablockers and diuretics. If the prescribing cascade of interest is solely ACEi-induced erectile dysfunction, one could include patients who only use ACEi (i.e., sole users, patients 1, 2 and 4 in Fig. ). Patients who use other subgroups of cardiovascular medication could be excluded, as the sole effect of ACEi cannot be determined for these patients (Fig. , patients 3, 5 and 6, where the PSSA cannot differentiate between erectile dysfunction possibly caused by the ACEi versus the co-medication or due to a combination of both). Misclassifications can occur when co-medication is not adequately taken into account (Fig. , patients 2 and 3, where the sequence of phosphodiesterase (PDE) inhibitors followed by diuretics and betablockers respectively would be found when ACEi was disregarded). On the other hand, when the use of co-medication is unrelated to the index medication and its initiation is expected to be stable over the time period that is studied, it may not influence the asymmetry caused by the index medication of interest in the PSSA calculations. Alternatively, multiple subgroup analyses could be performed to study the impact of comedication, e.g., for patients using one versus two or three and more classes of cardiovascular medication causing the same ADR. However, including too many different classes for the index medication may lead to more risk of bias, such as confounding by indication. There are a number of time windows that should be imposed both for the sequence index-marker as well as for the sequence marker-index when conducting PSSA to assess prescribing cascades. Relevant parameters that are often considered for PSSA include the washout window to identify incident users and the exposure window for assessing the associations. In addition, a blackout period and continued exposure interval (CEI), which have been defined in other pharmacoepidemiological studies, can be relevant when using PSSA for assessing prescribing cascades . These parameters can best be imposed at medication episode level, for which a definition of medication discontinuation is needed. This could be equal to the period set for the washout window. Importantly, one should first establish continuous enrollment during the washout and the exposure window for patient eligibility (see also previous paragraph). Washout window The washout window (also called ‘waiting time period’ or ‘run-in period’) is the period that is imposed as a look-back period to ensure that the index or marker medication is indeed a first prescription (incidence) and exclude any prevalent users of the index or marker medication . By selecting incident users, the initiation of the marker medication after an index medication is more likely to indicate the treatment for a new ADR than treatment for an ongoing medical condition . Of note, dose-dependent ADRs may occur after dose increases in prevalent users and some ADRs only develop after continued exposure . The washout window should depend on the maximum period that medication is dispensed in a country to make sure that any prevalent users are excluded. In the Netherlands, chronic medication is generally dispensed for 3 months but this can be extended to a maximum of 6 months. A washout window of 12 months could be considered suitable taking into account any medication supply a patient could have in stock, and reducing the chance of including prevalent users. In previous PSSA studies, the washout window was set at 6 or 12 months . Figure shows examples of episodes that could be excluded because the criteria for the washout window are not met (examples A and B). Exposure window The exposure window (also called ‘exposure time window’ or ‘observation period’) is the defined follow-up period to capture the pairs of incident index and marker medication users and vice versa, so the maximum allowed time period between the start of the index and the start of the marker medication . The exposure window should depend on the expected time onset of the ADR caused by the index medication . For example, Pouwels et al. used an exposure window of 4 weeks in their study of ACEi-induced urinary tract infections, as reduced urine output and reduced glomerular filtration rate (GFR) after ACEi initiation have been reported in relatively short-term studies ranging from 7 days to 8 weeks . It should be noted that such a short period is likely to decrease the sample size and the precision of the SR calculations . Also, a short exposure window can result in missing ADRs that take longer to occur, e.g., amiodarone can induce hypothyroidism even at 39 months after amiodarone initiation . It should also be kept in mind that for some ADRs patients can have a delay in consulting a healthcare provider . The downside of a longer exposure window is that the association between the index and the marker medication may become weaker and there is a higher risk of time-varying confounding . Generally, a 12-month exposure window is used to reduce the impact of such confounding, which includes ageing, disease progression and other time-varying variables, such as change in diet and/or environment . An exposure window of 12 months was found optimal for achieving acceptable sensitivity (61%; 95% CI 0.46–0.74) and high specificity (93%; 95% CI 0.87–0.96) for 165 tested index-marker pairs . Choosing different time periods can provide insight regarding the onset time of an ADR. In Fig. , examples are shown of episodes that are included (example G, H, I, and J) and excluded (example D and E) when an exposure window of 12 months is used. Of note, in the final year of data collection, episodes of the marker or index medication are disregarded when the exposure window requirement can no longer be met (example E). Continued exposure interval (CEI) To define continued exposure to a drug, the continued exposure interval (CEI) (also called ‘maximum permissible length without medication supply’) needs to be defined. This is the gap between the expected end date of one prescription and the date of the next prescription. In many pharmacoepidemiological studies, this would refer to the exposure to one medication or medication class but it can also be applied when associations between pairs of medication are studied . When studying prescribing cascades, the CEI is the maximum acceptable period without medication supply for the first medication . This period is imposed to ensure patients are likely to be exposed to the first medication when initiating the other medication. The CEI should be based on the common dispensing period of medication and the expected medication taking behaviour and could be similar to what is used in studies that measure adherence using dispensing data . For example, after the first prescription, medication prescriptions are generally repeated every 3 months in the Netherlands. Taking into account any supply the patient still possesses from previous dispensings (stockpile), a CEI of 4 to 6 months could be used. In Fig. , examples D and F show episodes that are excluded when a CEI of 4 months is imposed. It should be kept in mind that in most prescription and administrative databases the expected end date is theoretical and often determined based on the amount dispensed and the dose instruction when available. For medication that is prescribed without a clear dose instruction (e.g., use as needed), determining the expected end date can be difficult. Alternatively, one can set a longer period after the start date of each prescription, similarly to the definition of a discontinuation as well as the washout period, to exclude cases where impact of the initial medication is no longer expected. When imposing the CEI it should be kept in mind that some medication can still result in ADRs after discontinuation. Blackout period The blackout period (also called ‘time-lag period’ and ‘lag time’) is the period immediately after the initiation of the first medication in which events, i.e. the start of the other medication, will not be taken into account. Such a period is relevant to ensure sufficient time for the development of the ADR induced by the index medication and subsequent treatment with the marker medication . Singh et al. studied the prescribing cascade of calcium channel blocker (CCB)-induced edema treated with diuretics . They imposed a blackout period of 7 days because the probability of developing edema within 7 days after initiating CCBs is low. Other studies mentioned that patients need to be excluded if they have their first prescription of the index and marker medication on the exact same date, as the sequence of the index and marker medication cannot be extracted (see Fig. , example C) . Formally, this is not defined as blackout period but is important to address as exclusion criterion. Although an ADR may occur immediately, it seems impossible that a prescribing cascade occurs on the day of the first prescription. The washout window (also called ‘waiting time period’ or ‘run-in period’) is the period that is imposed as a look-back period to ensure that the index or marker medication is indeed a first prescription (incidence) and exclude any prevalent users of the index or marker medication . By selecting incident users, the initiation of the marker medication after an index medication is more likely to indicate the treatment for a new ADR than treatment for an ongoing medical condition . Of note, dose-dependent ADRs may occur after dose increases in prevalent users and some ADRs only develop after continued exposure . The washout window should depend on the maximum period that medication is dispensed in a country to make sure that any prevalent users are excluded. In the Netherlands, chronic medication is generally dispensed for 3 months but this can be extended to a maximum of 6 months. A washout window of 12 months could be considered suitable taking into account any medication supply a patient could have in stock, and reducing the chance of including prevalent users. In previous PSSA studies, the washout window was set at 6 or 12 months . Figure shows examples of episodes that could be excluded because the criteria for the washout window are not met (examples A and B). The exposure window (also called ‘exposure time window’ or ‘observation period’) is the defined follow-up period to capture the pairs of incident index and marker medication users and vice versa, so the maximum allowed time period between the start of the index and the start of the marker medication . The exposure window should depend on the expected time onset of the ADR caused by the index medication . For example, Pouwels et al. used an exposure window of 4 weeks in their study of ACEi-induced urinary tract infections, as reduced urine output and reduced glomerular filtration rate (GFR) after ACEi initiation have been reported in relatively short-term studies ranging from 7 days to 8 weeks . It should be noted that such a short period is likely to decrease the sample size and the precision of the SR calculations . Also, a short exposure window can result in missing ADRs that take longer to occur, e.g., amiodarone can induce hypothyroidism even at 39 months after amiodarone initiation . It should also be kept in mind that for some ADRs patients can have a delay in consulting a healthcare provider . The downside of a longer exposure window is that the association between the index and the marker medication may become weaker and there is a higher risk of time-varying confounding . Generally, a 12-month exposure window is used to reduce the impact of such confounding, which includes ageing, disease progression and other time-varying variables, such as change in diet and/or environment . An exposure window of 12 months was found optimal for achieving acceptable sensitivity (61%; 95% CI 0.46–0.74) and high specificity (93%; 95% CI 0.87–0.96) for 165 tested index-marker pairs . Choosing different time periods can provide insight regarding the onset time of an ADR. In Fig. , examples are shown of episodes that are included (example G, H, I, and J) and excluded (example D and E) when an exposure window of 12 months is used. Of note, in the final year of data collection, episodes of the marker or index medication are disregarded when the exposure window requirement can no longer be met (example E). To define continued exposure to a drug, the continued exposure interval (CEI) (also called ‘maximum permissible length without medication supply’) needs to be defined. This is the gap between the expected end date of one prescription and the date of the next prescription. In many pharmacoepidemiological studies, this would refer to the exposure to one medication or medication class but it can also be applied when associations between pairs of medication are studied . When studying prescribing cascades, the CEI is the maximum acceptable period without medication supply for the first medication . This period is imposed to ensure patients are likely to be exposed to the first medication when initiating the other medication. The CEI should be based on the common dispensing period of medication and the expected medication taking behaviour and could be similar to what is used in studies that measure adherence using dispensing data . For example, after the first prescription, medication prescriptions are generally repeated every 3 months in the Netherlands. Taking into account any supply the patient still possesses from previous dispensings (stockpile), a CEI of 4 to 6 months could be used. In Fig. , examples D and F show episodes that are excluded when a CEI of 4 months is imposed. It should be kept in mind that in most prescription and administrative databases the expected end date is theoretical and often determined based on the amount dispensed and the dose instruction when available. For medication that is prescribed without a clear dose instruction (e.g., use as needed), determining the expected end date can be difficult. Alternatively, one can set a longer period after the start date of each prescription, similarly to the definition of a discontinuation as well as the washout period, to exclude cases where impact of the initial medication is no longer expected. When imposing the CEI it should be kept in mind that some medication can still result in ADRs after discontinuation. The blackout period (also called ‘time-lag period’ and ‘lag time’) is the period immediately after the initiation of the first medication in which events, i.e. the start of the other medication, will not be taken into account. Such a period is relevant to ensure sufficient time for the development of the ADR induced by the index medication and subsequent treatment with the marker medication . Singh et al. studied the prescribing cascade of calcium channel blocker (CCB)-induced edema treated with diuretics . They imposed a blackout period of 7 days because the probability of developing edema within 7 days after initiating CCBs is low. Other studies mentioned that patients need to be excluded if they have their first prescription of the index and marker medication on the exact same date, as the sequence of the index and marker medication cannot be extracted (see Fig. , example C) . Formally, this is not defined as blackout period but is important to address as exclusion criterion. Although an ADR may occur immediately, it seems impossible that a prescribing cascade occurs on the day of the first prescription. To illustrate the impact of including co-medication and using different assumptions on the calculated aSRs, we used a dataset obtained from Ncontrol. Ncontrol holds data of dispensed prescriptions of more than 600 affiliated Dutch community pharmacies . Data were retrieved from January 1st 2015 until December 31st 2020. The prescribing cascade cardiovascular medication-induced erectile dysfunction was used to show the effect of including co-medication for three subgroups of cardiovascular medication. The prescribing cascade ACEi-antitussives was used to show the effect of different assumptions regarding the parameters needed for PSSA (i.e., washout window, exposure window, CEI and blackout period). The syntax for use in IBM SPSS Statistics version 27 (IBM Corporation, Armonk, New York, U.S.) is included in ESM and an additional file is necessary to make the syntax work (ESM ). The fictional sample dataset of ACEi-induced cough is included in ESM . Impact of co-medication on PSSA In Table the PSSA calculations are shown for sole users of respectively an ACEi, a beta-blocker or a high ceiling diuretic for the prescribing cascade erectile dysfunction. The aSR found for these sole users are somewhat lower than when any user of respectively an ACEi, a beta-blocker or a high ceiling diuretic is included in the PSSA calculations, with overlapping CIs. For example, including any ACEi user resulted in an aSR of 1.91 (95% CI: 1.84–1.99), while including sole ACEi users resulted in an aSR of 1.74 (95% CI: 1.62–1.86). When patients using two subgroups are included in the PSSA calculation the aSR increases to 2.18 (95% CI: 2.10–2.27) compared to sole users. For patients using a combination of three subgroups of cardiovascular medication, the increase is even higher with an aSR of 2.81 (95% CI: 2.69–2.93). These subgroup analyses confirm that this prescribing cascade is not specific for a particular medication group. Impact of changing assumptions on PSSA To illustrate the effect of different assumptions for the previously discussed parameters used in PSSA on the aSR, calculations have been made on the ACEi – antitussive medication prescribing cascade (Table ). In the primary analysis, the washout window and the exposure window were 12 months, the CEI was 4 months, and the blackout period was 7 days. The aSR of the primary analysis was 2.59 (95% CI: 2.56–2.62). In the second analysis, a washout window of 6 months was used instead of 12 months. This included more patients compared to the primary analysis and resulted in a higher aSR 2.73 (95% CI: 2.70–2.76). The higher aSR suggests a stronger association. In the third analysis, a 6-months exposure window was used instead of 12 months. This analysis included less patients as the follow-up period was shorter and resulted in a lower aSR of 1.80 (95% CI: 1.77–1.84). This may be the result of missing some of the more delayed prescribing cascades. In the fourth analysis, a 6-months CEI was used instead of 4 months. The aSR was lower 2.07 (95% CI: 2.04–2.10) than the first analysis, indicating that by choosing a longer period for the CEI, the association will be weaker as this analysis includes more patients initiating the marker medication for a non-related indication. In the final analysis, no blackout period was considered instead of a 7-days blackout period. This resulted in a similar aSR 2.52 (95% CI: 2.49–2.55) compared to the primary analysis. The chance that patients initiate antitussive medication within 7 days after initiating ACEi therapy (and vice versa) is relatively low, so the effect of implementing a short blackout period in this analysis was minimal. In Table the PSSA calculations are shown for sole users of respectively an ACEi, a beta-blocker or a high ceiling diuretic for the prescribing cascade erectile dysfunction. The aSR found for these sole users are somewhat lower than when any user of respectively an ACEi, a beta-blocker or a high ceiling diuretic is included in the PSSA calculations, with overlapping CIs. For example, including any ACEi user resulted in an aSR of 1.91 (95% CI: 1.84–1.99), while including sole ACEi users resulted in an aSR of 1.74 (95% CI: 1.62–1.86). When patients using two subgroups are included in the PSSA calculation the aSR increases to 2.18 (95% CI: 2.10–2.27) compared to sole users. For patients using a combination of three subgroups of cardiovascular medication, the increase is even higher with an aSR of 2.81 (95% CI: 2.69–2.93). These subgroup analyses confirm that this prescribing cascade is not specific for a particular medication group. To illustrate the effect of different assumptions for the previously discussed parameters used in PSSA on the aSR, calculations have been made on the ACEi – antitussive medication prescribing cascade (Table ). In the primary analysis, the washout window and the exposure window were 12 months, the CEI was 4 months, and the blackout period was 7 days. The aSR of the primary analysis was 2.59 (95% CI: 2.56–2.62). In the second analysis, a washout window of 6 months was used instead of 12 months. This included more patients compared to the primary analysis and resulted in a higher aSR 2.73 (95% CI: 2.70–2.76). The higher aSR suggests a stronger association. In the third analysis, a 6-months exposure window was used instead of 12 months. This analysis included less patients as the follow-up period was shorter and resulted in a lower aSR of 1.80 (95% CI: 1.77–1.84). This may be the result of missing some of the more delayed prescribing cascades. In the fourth analysis, a 6-months CEI was used instead of 4 months. The aSR was lower 2.07 (95% CI: 2.04–2.10) than the first analysis, indicating that by choosing a longer period for the CEI, the association will be weaker as this analysis includes more patients initiating the marker medication for a non-related indication. In the final analysis, no blackout period was considered instead of a 7-days blackout period. This resulted in a similar aSR 2.52 (95% CI: 2.49–2.55) compared to the primary analysis. The chance that patients initiate antitussive medication within 7 days after initiating ACEi therapy (and vice versa) is relatively low, so the effect of implementing a short blackout period in this analysis was minimal. PSSA has several strong points. First, the method has been validated in a study including 19 medications showed high specificity (93%) and moderate sensitivity (61%) for identifying ADRs . This indicates that PSSA is likely to identify many albeit not all prescribing cascades. In an overview of this method, Lai et al. stated that PSSA can capture prescribing cascades even when the ADR is rare . Second, PSSA is easy to implement using dispensing or prescription databases as little information is required for the basic calculations, that is, a patient identifier and the prescription dates for the index and marker medication. Third, PSSA is robust to patient confounders that are stable over time, such as sex and genetic factors, because PSSA is based on within-subject comparison . Several pitfalls of PSSA have been summarized which are typical of observational research, including time-varying confounding, protopathic bias or confounding by indication (i.e., the indication of the index medication leads to prescription of the marker medication) . Over the years, some approaches have been proposed to improve the application of the PSSA method, such as conducting sensitivity analyses and including positive controls (with known prescribing cascades) or negative controls (with unrelated marker medication) to test the robustness of the results . We add to this by providing a detailed description of considerations that are relevant for data collection and analysis when applying PSSA to assess prescribing cascades. Several considerations are common for the conduct of pharmacoepidemiological studies using prescription or dispensing databases, including setting a washout period to identify incident users and defining medication discontinuation to identify the end of a medication episode. When studying prescribing cascades, additional information about the ADR is needed for setting the time periods for the exposure window and blackout period. Of note, dose-dependent ADRs may occur after dose increases in prevalent users, requiring additional subgroup analyses. In addition, we have illustrated how comedication can influence the PSSA results when quantifying prescribing cascades. It might be relevant to study the prescribing cascade at a high medication class level or to stratify for combined medication use. In the examples provided in this study, we showed that changing the washout window, exposure window, CEI and blackout period can impact the strength of the associations observed. Although the overall direction of the association was not changed in our examples this could be different for other prescribing cascades, for example, when the aSR is close to one. This illustrates that conducting sensitivity analyses regarding all these parameters is relevant. Of note, when conducting sensitivity analyses for the exposure window, it is important to adjust the time periods for the null-effect SR as well. There are no rules for selecting the medication classes or setting the optimal time windows. We recommend that researchers clearly specify and explain all considerations regarding the index and marker medication included and the time windows set when studying prescribing cascades with PSSA. With this, information about prescribing cascades can be generated that is needed to address prescribing cascades in the future. Additional file 1. Additional file 2. Additional file 3. Additional file 4. |
This Month in | 6961afbc-e8cb-4226-87ca-efff683a98f0 | 8520165 | Pathology[mh] | Our understanding of COVID-19 pathophysiology is currently limited. Using rapid autopsy tissues from infected and uninfected individuals, Pujadas and Beaumont et al ( Am J Pathol 2021 , 2064–2071 ) studied the underlying mechanisms. Comparison of COVID-19 tissues with control tissues revealed four main regulatory pathways: blood vessel development, cytokine production, cell activation, and structure and degradation. Effectors within the identified pathways may be targeted to manage COVID-19. Aging affects the lacrimal glands and the microbiota; however, the underlying mechanisms are unclear. Using a mouse model, Jiao and Pei et al ( Am J Pathol 2021 , 2091–2116 ) studied these mechanisms. Mice were divided into three groups: young, old, and fecal microbiota transplant–treated old groups. Reconstitution of old mice with the microbiome of young mice over time shifted the microbial communities to young donors and significantly reduced the chronic inflammation, lipid deposition, and abnormal neural response of the aging lacrimal glands. Microbiome-based intervention may help reverse age-related dysfunction of the lacrimal glands. The role of casein kinase 1α (CK1α)—a widely expressed protein in the endometrium that regulates autophagy—in endometriosis is unclear. Using patient and control clinical samples and cultured cells, Zhou et al ( Am J Pathol 2021 , 2195–2202 ) studied this role. CK1α, phosphatase and tensin homolog (PTEN), and autophagy-related 7 (Atg7) were expressed at lower levels in the ectopic endometrium and a positive correlation was observed between CK1α and PTEN, CK1α and Atg7, and PTEN and Atg7. CK1α, PTEN, and autophagy-related markers were repressed in the endometrial stromal cells. CK1α regulates PTEN/Atg7-mediated autophagy in human endometrial stromal cells. The role of the arginine derivative nitric oxide (NO) in bladder cancer invasion is unclear. Using patient samples, cultured cells, and a zebrafish model, Sahu et al ( Am J Pathol 2021 , 2203–2218 ) studied this role. A stage-associated increase was observed in the expression of NO-generating enzymes, endothelial NOS (eNOS) and inducible NOS (iNOS), in human bladder cancer. Reducing NOS activity decreased cancer cell invasion; whereas, increasing NOS activity enhanced invasion. In vivo , reducing NOS activity decreased bladder cancer cell metastasis. The invasive tips of bladder cancer cells, invadopodia, were enriched in NOS proteins as well as mTORC2 activity, which in turn regulated invadopodia formation, eNOS and iNOS expression, and cyclicGMP production in the invadopodia. Blocking NO may help manage bladder cancer. The role of the cytokine B cell activating factor (BAFF) in aortic aneurysms is unclear. By performing prevention and intervention studies in a mouse model of abdominal aortic aneurysm (AAA), Spinosa et al ( Am J Pathol 2021 , 2231–2244 ) studied this role. The formation of AAA was attenuated by injecting BAFF antagonists before and after the induction of AAA in prevention and intervention studies, respectively. In the intervention group, BAFF antagonism enhanced resolution of inflammation in AAA. BAFF antagonism may deplete mature B cells, promote resolution of inflammation in aorta, and attenuate the growth of AAA. |
Primary adenocarcinoma of the spermatic cord: a case report and review of the literature | b2342ba0-c179-41af-b709-8c2334af2c8f | 11463123 | Anatomy[mh] | Tumors located in the spermatic cord are uncommon and mostly are sarcomas and metastatic tumors . The primary malignant tumors of the spermatic cord accounting for less than 30% of all spermatic cord tumors . Sarcomas are the most frequently encountered primary malignancy, whereas epithelial tumors, are exceptionally rare. To date, only one case of primary carcinoid tumor on the spermatic cord has been reported in the English literature . Adenocarcinoma of the spermatic cord (ACSC) is typically observed as a metastatic entity, commonly originating from gastrointestinal tract . Nevertheless, primary ACSC has not been reported. Here, we present a unique case of primary ACSC, along with a detailed describe of clinical, histopathological, and molecular characteristics. A 34-year-old male present with unilateral nodule of the spermatic cord with enlarged right scrotum and vague pain 1 month ago. Subsequently, enhanced computed tomography (CT) showed a 1.6 cm mass in the right testicular and spermatic cord regions (Fig. A and B). Consequently, the patient underwent tumor resection and right inguinal orchiectomy at a local hospital. Following the surgery, the right inguinal canal exhibited typical surgical changes, with no signs of new tumor lesions were detected on whole-body PET-CT scans (Fig. C). Gross pathological examination revealed a solid tumor, measuring 1.6 cm in greatest diameter. Given the tumor’s proximity to the epididymis, comprehensive sampling and detail histopathological examination were performed on the areas adjacent to the epididymis and the rete testis. No malignant cells were detected in these regions (Supplementary Fig. A, B). Histological examination confirmed the tumor’s localization within the spermatic cord, as evidenced by the surrounding tumor cells encircling residual vas deferens (Fig. A). Invasion into the adjacent fibroadipose tissue was observed (Fig. B). The tumor displayed a variety of growth patterns, including cord, cribriform, poorly differentiated solid and nest (Fig. C, D and E). Cytologically, epithelioid cells exhibited pale red cytoplasm and round or oval nuclei (Fig. F). Interstitial changes included focal necrosis and inflammatory cell infiltration. Immunohistochemical staining revealed positive expression for epithelial markers CK (AE1/AE3), CK19, CK8/18, MOC31 and Ber-EP4, while CK7 was negative (Fig. ). Mesothelioma markers, D2-40, WT1, and Mesothelial Cells (MC) were all negative (Fig. A and B). PAX-8 showed positive to the adjacent vas deferens epithelial, but negative to malignant cells (Supplementary Fig. C). Additionally, CEA, PSA, NKX3.1, TTF-1, Napsin A, SALL4, Inhibin a, Chromogranin A (CgA), Melan-A, and ALK (D5F3) were all negative. Interestingly, immunohistochemistry revealed complete loss of INI-1 expression and consistent BRG1 expression in all tumor cells (Fig. C and D). For genetic analysis, mutation profiling of DNA obtained from formalin-fixed paraffin-embedded (FFPE) tumor tissue sections was performed using next-generation sequencing (NGS). The NGS tests, which utilized the entire exonic regions of 310 genes and the hotspot mutation regions (exonic, intronic, or promoter regions) of 210 genes associated with PCa and were conducted at a centralized, CLIA-certified, and CAP-accredited clinical testing center (Nanjing Geneseeq Technology Inc., Nanjing, China). The sequencing platform used for the study was Illumina. The final report demonstrated that he had a nonsense mutation (c.842G > A) of the SMARCB1 gene, resulting in an amino acid change at p.W281, with a mutation frequency of 20.36%. Additionally, a missense mutation (c.154G > A) in KLHL6 gene was detected, leading to substitution of threonine for alanine at position 52 (p.A52T), with a mutation frequency of 7.00%. The tumor burden (TMB) was determined to be 1.99 mutations/Mb (TMB-L). Microsatellite stability is indicated by microsatellite stable (MSS), with no embryonic lineage mutations identified. Currently, the patient remains stable, no evidence of recurrence or distant metastasis observed ten months after radical resection and has not received any postoperative adjuvant therapy. Malignant neoplasms of the spermatic cord are rare, with 352 cases reported in PubMed from 1972 to 2024, including 87 reviews and 232 case reports (Table ). Sarcomas represent the predominant histological subtype, comprising 79% (278/352) of cases, while epithelial tumors occur less frequently, accounting for 16% (59/352) (Fig. A). Among sarcomas, liposarcomas constitute the majority at 52% (145/278), followed by leiomyosarcoma at 18% (50/278), with malignant fibrous histiocytomas ranking third at 12% (34/278) (Fig. B). In terms of epithelial neoplasm, metastatic carcinoma is the most prevalent type, primarily originating from the gastrointestinal system, representing 54% (32/59) of cases, followed by the kidney at 21% (12/59) and the pancreas at 12% (7/59), respectively (Fig. C). Notably, primary malignant epithelial neoplasm is rare, with only one reported case of a 52-year-old man presenting with a carcinoid tumor . To our knowledge, tumors of the spermatic cord initially presented with painful scrotal and inguinal masses, common symptoms of male genital metastasis . Some patients may exhibit hydrocele and scrotal swelling. While preoperative ultrasonography and CT imaging provide significant diagnostic assistance, the definitive diagnosis necessitates histopathological analysis. The pathological morphology of this ACSC resembles that of moderately to poorly differentiated adenocarcinoma in other organs. Primary adenocarcinoma of the spermatic cord is exceptionally rare and warrants differentiation from malignant mesothelioma, carcinoma with Müllerian differentiation, and metastatic adenocarcinoma. In this case, the absence of D2-40 and WT1 expression, along with positive MOC31 and Ber-EP4 expression, does not support the diagnosis of malignant mesothelioma. The negative expression of PAX-8 in malignant cells does not support a tumor of vas deferens or Müllerian differentiation. Furthermore, negative expression for WT1, NKX3.1, PSA, CEA, TTF-1, and NapsinA helps exclude certain metastatic adenocarcinomas. Additionally, extensive metastatic work-up, including abdomen, pelvis, and thoracic CT scans, as well as whole-body PET-CT imaging, revealed no evidence of new tumor lesions. Interestingly, the tumor cells exhibited complete loss of INI-1 expression. Molecular analysis further confirmed this as a SMARCB1 (INI-1) deficient adenocarcinoma. SMARCB1 is recognized as a tumor suppression gene . Loss of INI-1 expression as a result of SMARCB1 deletions/mutations has emerged as a defining diagnostic feature in a variety of neoplasms in children and adults, in particular malignant atypical teratoid/rhabdoid tumors of childhood, epithelioid sarcoma, and several epithelial tumor entities in adults and the elderly . SMARCB1-deficient tumors often present with a poor prognosis, characterized by widespread metastasis at the time of diagnosis . Considering the rarity of primary SMARCB1-deficient adenocarcinoma of the spermatic cord, there is currently no consensus on adjuvant treatment for this entity. It’s crucial to note the potential for a poor prognosis in such cases, even in the absence of present disease recurrence and progression. INI-1 (SMARCB1) deficiency is observed in approximately 90% of Epithelioid Sarcoma (ES) . EZH2 plays a role in DNA methylation and transcriptional repression, maintaining the epigenetic silencing of genes . When INI1 loses its regulatory function, EZH2 activity becomes deregulated, allowing EZH2 to play a driving, oncogenic role . With the application of the EZH2 inhibitor Tazemetostat in ES, it has a possibility of promising drug for SMARCB1-deficient tumors . Furthermore, recent data indicate that SMARCB1 loss may augment anti-tumor immunogenicity in specific subtypes of sarcomas and cancers . Hence, epigenetic modulators and immune checkpoint inhibitors could emerge as promising therapeutic modalities for SMARCB1-deficient tumors. In this study, we present the first case of primary adenocarcinoma of the spermatic cord with SMARCB1 (INI-1) deficiency. Our report encompasses a comprehensive description of its pathological morphology, immunohistology, and molecular characteristics. This case contributes to the expanding understanding of rare neoplasms and underscores the importance of further research into therapeutic strategies targeting SMARCB1-deficient tumors. Below is the link to the electronic supplementary material. Supplementary Material 1 Supplementary Material 2: Figure 1. (A) No malignant cells found in the epididymis. (B) No malignant cells found in the rete testis. (C) Normal vas deferens epithelium demonstrated positive expression of PAX-8, while the malignant cells were negative. |
Assessing the efficacy of a virtual reality lower leg fasciotomy surgery training model compared to cadaveric training | 34e601de-05ca-4f36-8a3f-421fd27c6bac | 11841149 | Surgical Procedures, Operative[mh] | Failure to promptly and adequately perform a lower extremity fasciotomy (LEF) is a well-recognized clinical problem with life-threatening consequences in civilian and military surgical domains . Fasciotomies are surgical procedures aimed at relieving increased pressure within inelastic fascia compartments - a condition is called compartment syndrome. In compartment syndrome, as pressure in the compartment becomes elevated, vascular flow is impeded leading to ischemia and rapidly increasing the need to perform an amputation . To aid in ability of surgeons to respond to this and other emergent time-critical trauma procedures, the Advanced Surgical Skills for Exposure in Trauma (ASSET) course teaches how to perform a fasciotomy . The ASSET course is cadaver-based training (CBT) often used to refresh skills as well as learn new procedures. Generally, CBT is 2–4 surgeons to a body which limits the amount of time, as well as number of procedures, that surgeons gain hands-on experience with. Courses with one trainee per cadaver can be found, though these are less common due to logistic of donor availability. A significant part of a typical CBT training can, thus, involve observation or assistance of fellow surgeons performing a procedure. Therefore, we aimed to explore the utility of other teaching methodologies where a procedure can be clearly observed as an equivalent method of educating/refreshing surgeons to improve surgical outcomes. Virtual reality (VR) holds an immense potential in surgical education and has only recently been actualized in a limited number of medical educational disciplines. Virtual reality (VR) training was proposed as an option since it is portable, relatively inexpensive and scalable - all features that make it suitable for the medical education setting as a primary or refresher learning tool. Procedural training utilizing cadaveric laboratory sessions in surgery is resource intensive, although ultimately it is the ‘gold -standard’ for surgical education and training refreshing. Cadaver based training occurs intermittently through a surgeon’s initial education and career. This is primarily as the resource is not continuously available or is excessively expensive to utilize at a higher frequency. Traditional supplements to cadaveric laboratory training include videos and books. However, the advent of VR and augmented reality (AR) provides the opportunity for a simulated environment that replicates many of the observational aspects of in person cadaveric training. Virtual reality training has been used in a number of clinical care settings. VR was found to be highly successful in the intensive care unit setting for training as compared to in person bed-side skills training . The results of that study showed simulator-trained residents scored significantly higher on the bedside skills assessment compared with traditionally trained residents. It has been suggested that VR environments may provide the most benefit to education when involving technical skills, team training and decision-making skills , consistent with the observed bedside skills improvement study . The use of virtual reality training in surgical education is continuing to grow. In one study, augmented reality was used to provide virtual surgical guidance on the operating field . Residents and medical students were evaluated on their performance of cadaveric leg fasciotomies with and without the use of augmented reality telemonitoring to provide guidance during the procedure. Participants who utilized the augmented reality telemonitoring performed fewer errors and scored higher on procedure performance. This study showed the benefit of using augmented reality at the same time the procedure was being performed. A systematic review concluded the use of virtual reality is positive due to the improved procedural times and cost-effectiveness . However, a separate systematic literature review into the use of virtual and augmented reality to improve trauma training for surgeons performing open emergency surgery concluded there was not enough evidence to suport replacement of cadavers/cadaveric training with computer interface technologies due to limited testing in surgeons and paucity of literature on the subject . Indeed, that very paucity of literature on the subject was a strong rationale for performing this study. There are several reasons why virtual reality could be beneficial for learners. Education theory suggests that retention is enhanced when delivered utilizing more than one sensory system . For example, a recent user study shows an 8.8% increase in recall when educated in VR immersive environments compared to 2D desktops . VR is able to incorporate both motor and sensory system modalities in an all-encompassing virtual environment through introduction of gameplay elements to the training. Conceptually, educational material utilizing VR should thus have benefit over traditional static didactic learning settings such as lectures assigned readings, or video. However, literature specific to VR training in trauma surgery is sparse. Our hypothesis was that medical students who undergo a virtual reality surgical education and training (VR SET) training module will demonstrate performance on a LEF similar to that obtained by completion of an in-person cadaveric training course. This study was performed from May 2021 -September 2021 at the University of Maryland Medical Center. Our pilot sample population was recruited from 4th year medical students at the University of Maryland School of Medicine and Johns Hopkins University School of Medicine. Inclusion criteria were participants aged 18 to 88 years old and medical students who have not completed the ASSET course. Exclusion criteria were participants 89 years or older as well as medical students who completed the ASSET course. 4th year medical students were chosen because we wanted to evaluate participants who are familiar with how to operate (i.e., the basic manual surgical skills), but not yet familiar with the specific procedures of a fasciotomy or other advanced trauma training received by residents. Our control group was a historical cohort of residents who completed pre-ASSET and post-ASSET evaluations in fasciotomy at the University of Maryland School of Medicine as part of a prior experiment . As part of ASSET training, the control group received a full day of training in the cadaver lab, performing a multitude of trauma surgical procedures, one of which was the lower extremity fasciotomy. The first session for the intervention group took place in an office environment (Fig. A), where participants undertook the VR-SET program using a head-mounted virtual reality display (Fig. B, HMD, Rift S, Oculus, Dallas, TX). The VR training module consists of step-by-step video demonstration of the lower extremity fasciotomy using a ‘surgeon perspective’ VR positioning in a 180-degree view. The team created a customized Unity VR application to display high-fidelity content acquired through a custom-built GoPro camera rig, and made alignments to address stereo separation, rotation, and camera distortion, ensuring that the visuals appeared natural to the viewer (Fig. C). This was achieved by instantiating two half-sphere projection screens to match the field of view, lens distortion, position, and rotation of the GoPro cameras used for data acquisition, one for each eye separately. Subsequently, placing these screens at horizontal separation matching the stereo separation of the human eye ensured that visual content was displayed identically to what the human eye would see naturally. The team also integrated advanced video decoding software to handle the dual 4 K videos required to play back the medical procedure. To enhance the interactivity and intuitiveness of the training module, we added hand-tracking capabilities using a Leap Motion hand tracker (Fig. D). Instead of relying on standard VR controllers, users can select from menus via the Leap Motion hand tracker and navigate through module sections using buttons projected onto their wrist (Fig. E) or gesture of their hands. We note that there is no commercially available hardware or software that provides similar levels of fidelity and detail. In the second session occurring 10–14 days after training, participants performed a lower extremity fasciotomy at the State Anatomy Board clinical training laboratories. During the procedure, participants were asked questions by the research team using a standardized script as an assessment of their anatomic and pathophysiology knowledge of compartment syndrome . After the standardized questions are completed, the participant is asked to complete the lower extremity fasciotomy. As the participant performs the procedure, two trained evaluators utilize a validated question and observation script called the Individual Procedure Score (IPS) to record the participants’ performance (Appendix ). This scoring system was used to evaluate both the intervention and control groups following the virtual reality or cadaver-based training, respectively. This scoring system has been validated as correlating with successful outcome of procedures using approximately 200 surgeons performing the procedure. A high IPS score correlates with improved performance. Successful decompression of the compartments was determined by trained evaluators who were familiar with the anatomy and skill of the procedure following the defined IPS assessment metrics. Successful decompression was determined as correctly accessing the compartment and opening it along the entire length of the compartment. The scores of the two trained evaluators are averaged to account for potential evaluator bias and prior studies show high inter-rater reliability using the IPS metric . Number of errors performed during the procedure were quantified through the script (Individual Procedure Score as highlighted in yellow in Appendix ). There were 9 maximum errors which could be made (e.g. inability to correctly identify the intermuscular septum, incomplete fascia opening of a compartment, etc.). At the conclusion of the assessment, the participant was debriefed on the procedure, but did not receive numerical results of their performance. A two-sample t-test was performed to test if VR-SET training is different to cadaver-based training with significance set at α = 0.05 comparing successful compartment decompressions and errors committed. Numerical values for the five IPS component scores were compared using ANOVA with significance set at α = 0.05. All data is represented as mean plus/minus standard deviation. VR-trained students successfully decompress the same number of compartments compared to resident controls Incomplete decompression in compartment syndrome is a significant problem that enhanced training may be able to ameliorate. VR-SET study participants successfully decompressed an average of 2.45 ± 1.09 (range 1 to 4) compartments (Fig. A). The control group decompressed an average of 2.06 ± 0.93 (range 0.5 to 4) compartments (Fig. A). The number of compartments VR-SET participants decompressed was statistically indistinguishable from control ( p = 0.35). VR-SET study participants committed an average of 3.4 ± 2.27 (range 1–6) errors (Fig. B) compared to the controls who committed an average of 4.45 ± 1.93 (range 0.5–7.5) errors (Fig. B) with no statistical difference between the two groups ( p = 0.18). VR-SET trainee individual procedure scores The individual procedure score metric is a validated measure of performance of the LEF consisting of five sub-components, anatomy, pathophysiology, patient management, procedure, and surgical technique. VR-SET performance did not statistically differ from controls in anatomy (VR-SET 0.67 ± 0.10, controls 0.60 ± 0.11; Fig. C), patient management (VR-SET 0.53 ± 0.30, controls 0.53 ± 0.21; Fig. C), or procedure (VR-SET 0.47 ± 0.18, controls 0.45 ± 0.15; Fig. C). Controls exhibited higher IPS scores in pathophysiology (VR-SET 0.36 ± 0.09, controls 0.5 ± 0.1; p = 0.0008; Fig. C) and surgical technique (VR-SET 0.59 ± 0.17, controls 0.76 ± 0.15; p = 0.009; Fig. C). Incomplete decompression in compartment syndrome is a significant problem that enhanced training may be able to ameliorate. VR-SET study participants successfully decompressed an average of 2.45 ± 1.09 (range 1 to 4) compartments (Fig. A). The control group decompressed an average of 2.06 ± 0.93 (range 0.5 to 4) compartments (Fig. A). The number of compartments VR-SET participants decompressed was statistically indistinguishable from control ( p = 0.35). VR-SET study participants committed an average of 3.4 ± 2.27 (range 1–6) errors (Fig. B) compared to the controls who committed an average of 4.45 ± 1.93 (range 0.5–7.5) errors (Fig. B) with no statistical difference between the two groups ( p = 0.18). The individual procedure score metric is a validated measure of performance of the LEF consisting of five sub-components, anatomy, pathophysiology, patient management, procedure, and surgical technique. VR-SET performance did not statistically differ from controls in anatomy (VR-SET 0.67 ± 0.10, controls 0.60 ± 0.11; Fig. C), patient management (VR-SET 0.53 ± 0.30, controls 0.53 ± 0.21; Fig. C), or procedure (VR-SET 0.47 ± 0.18, controls 0.45 ± 0.15; Fig. C). Controls exhibited higher IPS scores in pathophysiology (VR-SET 0.36 ± 0.09, controls 0.5 ± 0.1; p = 0.0008; Fig. C) and surgical technique (VR-SET 0.59 ± 0.17, controls 0.76 ± 0.15; p = 0.009; Fig. C). Our findings show that VR trained medical students decompress the same number of compartments as residents who have undertaken cadaver-based training and committed the same average number of errors. Our prior study looking at fasciotomy performance before and several weeks after in person ASSET training showed residents prior to ASSET training decompress 1.62 compartments and post-ASSET in person cadaver training decompress 2. 06 compartments . Our observation that VR-SET training resulted in 2.45 compartments is very close to that observed in residents with ASSET in person cadaver-based training. In the lower extremity fasciotomy, all 4 compartments should be decompressed to prevent ischemia and tissue death. The fact that both the control group and intervention group decompressed an average of only 2–3 compartments highlights the need for better training opportunities for this procedure. As with any surgical skill, success comes with repeated learning opportunities and VR training allows learners to continue their education at any point, in contrast to cadaver-based training which is intermittent at best once past initial training. VR may allow for a culture of continual and additional training for surgeons in procedures they perform infrequently. When comparing the knowledge of VR trained medical students to residents with cadaver-based training we observed indistinguishable levels of anatomical and procedural knowledge, both aspects emphasized in the VR-SET module and ASSET course. However, residents and fellows with cadaver-based training show a greater surgical technique and pathophysiological knowledge, likely due to their overall longer clinical experience in their fields compared to 4th year medical students, coupled with the VR-SET module addressing primarily anatomy and procedure, not pathophysiology or general technique. Surgical technique is gained with experience and as such we would expect our control group of residents to score better than the VR-SET group of 4th year medical students. The implication of these pilot results is that VR could become a suitable primary and/or refresher surgical education teaching tool, particularly in situations where the trainee already has strong general technique and requires only elaboration of the details in a specific procedure. For VR training to become a core part of surgical medical education, it will heavily depend on allocated funding so that suitably validated training materials and simulation environments can be developed. VR has the potential to be an improved educational environment, as it targets several modalities: auditory, visual and spatial along with an ability to individualize for different learner needs. Since learners have different experience levels, skills, and background, VR can provide a more adaptable training program for individualization as compared to a classroom setting, where dozens of individuals receive a homogenous training product. The environment that VR provides allows students to immerse themselves in the training without having to physically be in the same place or at the same time, thus allowing for customization of the process. Additionally, VR can present greater diversity in patient populations, including body habitus, ethnicity, and gender identity, areas in which donor programs struggle to have a large number of diverse donations. Military and rural medicine are both fields which can benefit significantly from virtual reality training and re-training opportunities. Literature on trauma training in military medicine demonstrates challenges to maintaining a competent corps ready for deployment. Edwards et al. discuss decreased surgical caseloads while deployed, which diminishes the time surgeons are clinically active. Surgeons deployed in support locations mission tasked to provide surgical capability within 60 min of an injury report performing less than one operative case in a month . The outcome of this is that military surgeons are asked to be continuously ready to perform trauma skills that are not part of their day-to-day practice, and they are typically unable to obtain gold-standard cadaveric refreshers while deployed in austere environments. Similarly, surgeons in rural environments are often expected to have trauma surgical competency, as they may be the first line of surgical stabilization before transfer to a trauma center and work in an area dominated by heavy industry such as agriculture with associated trauma risk. Glenn et al. discuss how rural surgeons greatly benefit from remote learning opportunities, as they usually practice alone and rarely have access to colleagues with a specialized skill set. Their research demonstrated that trauma consultation was one field in which rural surgeons would be interested in receiving telementor capabilities to provide additional assistance . Our study suggests that virtual reality trauma scenarios may be effective ways for rural or military surgeons to acquire on-demand refresher training in remote or austere environments. This remote learning capability could be actualized to allow medical providers with limited to no anatomic donor resources to receive practical refresher training before performing a critical procedure. Limitations VR training and our study have several limitations. At present VR training is unable to capture tactile sensation, and in that aspect, it may be lacking as compared to cadaveric training. For instance, during the lower extremity fasciotomy procedure, when operating on the lateral aspect of the lower extremity, tactile sensation of the durability of the intermuscular septum helps to confirm location and successful decompression of anterior and lateral compartments. Additionally, complete detachment of the soleus muscle from the tibia to access the deep posterior compartment is a highly tactile component involving finger assessment of adherent points for transection. This is particularly a concern if VR were used as primary training in general surgical technique. However, in-person technique training in medical school and residency combined with VR training in specific procedures could be effective. In the future, advances in haptic feedback gauntlets may allow for VR tactile sensation in VR scenarios. Currently, haptic feedback gauntlets are bulky and do not replicate the precision of sensation required for surgical performance. A limitation of our VR training program was that the environment had limited interactive components. Given the limitations of haptic systems, we propose that a simpler gesture-based aspect adding a gameplay component (e.g. zoom, replication of a surgical motion to advance the module, or anatomy/procedure games) may be effective additions to VR in the interim while full tactile haptic feedback system technology continues to mature. Participants highlighted a desire for a gesture-based gameplay within the module in post evaluation feedback. This reinforces the conceptual value gameplay may have in VR training and form the basis for a future study. A limitation of our study was that the level of experience between the control group and the intervention group differed. The control group was a group of early residents who had just completed the ASSET training course while our intervention group was a group of late 4th year medical students. Thus, there are 3–6 months of difference with the control group at a slightly higher training status. Additionally, the residents had a pre-ASSET assessment in addition to the post-ASSET assessment, which potentially could result in them gaining experience from the pre-training assessment itself. A pre-VRSET assessment was not performed on the 4th year students, as their training curriculum did not have any fasciotomy components and performance of a procedure where someone has no idea what they are supposed to do would be an unethical use of an anatomic donor for a situation not likely to yield meaningful data. Overall, by having a potentially slightly higher experience level for a control, outcomes are tilted against a finding of efficacy for the VR tool. Thus our finding a late 4th year can now perform at levels equivalent to a new resident after a VR only training implies VR is more effective than a few months of in person experience, however it is too early in overall VR training knowledge to make that conclusion, An additional limitation is that this was a pilot study with n = 10 in the VR intervention group, although we were able to robustly show that this group had no statistically significant difference to the control group (2.4 compartments with VR-SET vs. 2.0 in control with CBT). It is possible with a larger cohort that the small 0.4 compartment difference between the groups may become significant. A future direction for this study could evaluate this training tool as compared to other technologies such as augmented reality or animated tools such as Touch Surgery (Medtronic, 2024). VR offers the ability to have 3D information and high-resolution imagery for training. A significant limitation of current VR is the absence of tactile feedback, as current tactile systems are suboptimal for the precision of tactile sensation necessary for surgery. VR training and our study have several limitations. At present VR training is unable to capture tactile sensation, and in that aspect, it may be lacking as compared to cadaveric training. For instance, during the lower extremity fasciotomy procedure, when operating on the lateral aspect of the lower extremity, tactile sensation of the durability of the intermuscular septum helps to confirm location and successful decompression of anterior and lateral compartments. Additionally, complete detachment of the soleus muscle from the tibia to access the deep posterior compartment is a highly tactile component involving finger assessment of adherent points for transection. This is particularly a concern if VR were used as primary training in general surgical technique. However, in-person technique training in medical school and residency combined with VR training in specific procedures could be effective. In the future, advances in haptic feedback gauntlets may allow for VR tactile sensation in VR scenarios. Currently, haptic feedback gauntlets are bulky and do not replicate the precision of sensation required for surgical performance. A limitation of our VR training program was that the environment had limited interactive components. Given the limitations of haptic systems, we propose that a simpler gesture-based aspect adding a gameplay component (e.g. zoom, replication of a surgical motion to advance the module, or anatomy/procedure games) may be effective additions to VR in the interim while full tactile haptic feedback system technology continues to mature. Participants highlighted a desire for a gesture-based gameplay within the module in post evaluation feedback. This reinforces the conceptual value gameplay may have in VR training and form the basis for a future study. A limitation of our study was that the level of experience between the control group and the intervention group differed. The control group was a group of early residents who had just completed the ASSET training course while our intervention group was a group of late 4th year medical students. Thus, there are 3–6 months of difference with the control group at a slightly higher training status. Additionally, the residents had a pre-ASSET assessment in addition to the post-ASSET assessment, which potentially could result in them gaining experience from the pre-training assessment itself. A pre-VRSET assessment was not performed on the 4th year students, as their training curriculum did not have any fasciotomy components and performance of a procedure where someone has no idea what they are supposed to do would be an unethical use of an anatomic donor for a situation not likely to yield meaningful data. Overall, by having a potentially slightly higher experience level for a control, outcomes are tilted against a finding of efficacy for the VR tool. Thus our finding a late 4th year can now perform at levels equivalent to a new resident after a VR only training implies VR is more effective than a few months of in person experience, however it is too early in overall VR training knowledge to make that conclusion, An additional limitation is that this was a pilot study with n = 10 in the VR intervention group, although we were able to robustly show that this group had no statistically significant difference to the control group (2.4 compartments with VR-SET vs. 2.0 in control with CBT). It is possible with a larger cohort that the small 0.4 compartment difference between the groups may become significant. A future direction for this study could evaluate this training tool as compared to other technologies such as augmented reality or animated tools such as Touch Surgery (Medtronic, 2024). VR offers the ability to have 3D information and high-resolution imagery for training. A significant limitation of current VR is the absence of tactile feedback, as current tactile systems are suboptimal for the precision of tactile sensation necessary for surgery. This study suggests routine re-training could be achieved through a VR environment as a supplement to current in-person training programs. This could allow limited cadaveric resources to be focused on those most needing hands-on training and general surgical technique, while VR could serve as secondary training or supplemental training on uncommonly performed procedures. Below is the link to the electronic supplementary material. Supplementary Material 1 |
iTRAQ proteomic analysis of the anterior insula in morphine‐induced conditioned place preference rats with high‐frequency deep brain stimulation intervention | 2174aba2-d5b3-4433-987c-20f376d7c8e6 | 11747870 | Biochemistry[mh] | INTRODUCTION Opioid drugs are widely used as effective analgesics for chronic, moderate‐to‐severe pain relief. , , However, chronic opioid use often leads to misuse and addiction. , Opioid addiction has rapidly become a global public health and social concern. , Morphine, a representative opioid, is commonly prescribed in clinical settings and often results in morphine dependence or addiction. Therefore, understanding the molecular and functional mechanisms of morphine addiction is crucial for developing less addictive alternatives for morphine treatment. The concept of the ‘Morphinome’, as proposed by Bodzon‐Kulakowska et al. aims to explore the proteomic changes associated with morphine addiction and uncover its molecular mechanisms. In recent years, various research groups have identified proteins involved in morphine addiction. The Morphinome Database ( www.addictionproteomics.org ) was established to compile information on proteins regulated under the influence of morphine, aiding in the exploration of the molecular basis of morphine addiction. Data from 29 published articles until 2018 were included in this database. Many of these studies employed two‐dimensional electrophoresis (2DE) analysis to examine protein expression profiles. For instance, 2‐DE was used to assess the differential expression profile of prefrontal cortex (PFC) synapses in morphine‐induced conditioned place preference (CPP) rats. Different proteins were identified using the immobilized pH gradient 2‐DE in the hippocampal tissue of rats experiencing morphine addiction recurrence or during chronic morphine treatment and withdrawal. Proteomic analysis in other brain regions, including nucleus accumbens, , ventral tegmental area and amygdala, also utilized 2‐DE in different animal models or at different stages of addiction, but few studies have explored the insula's proteomics in morphine addiction animal models. The insula, a hub of interoception, plays a crucial role in drug addiction development and maintenance, contributing to all three addiction stages. , , , Investigating insula proteomics enhances our understanding of the structural and functional changes associated with morphine addiction at the molecular level. Current opioid addiction treatment includes μ‐opioid agonist, partial agonist and antagonist medications. However, these medications are not as widely as widely as needed. Deep brain stimulation (DBS), which modulates neuronal activity in specific brain circuits, is a reversible, adjustable, minimally invasive and safe neurosurgical intervention that has been applied successfully in treating neuropsychiatric disorders such as Parkinson's disease, , dystonia, obsessive‐compulsive disorder and severe depressive disorder. Preclinical and clinical studies suggest that DBS holds promise as an intervention for drug addiction, although its specific molecular mechanisms in drug addiction remain unclear. Therefore, proteomic analysis of the insula following DBS intervention in morphine‐addicted animals is expected to elucidate the molecular mechanisms involved in morphine addiction treatment and DBS intervention. This will contribute to a more comprehensive understanding of the biological processes underlying DBS in the treatment of neurological diseases. In this study, we further analysed the proteomic data of saline control, morphine and morphine with DBS groups from our recently published study and identified the signalling pathways involved in morphine addiction and DBS intervention. MATERIALS AND METHODS 2.1 Animals, DBS surgical implantation and morphine‐conditioned place preference (CPP) Male Sprague–Dawley rats, weighing 260–280 g (6‐8 weeks), were housed at the standard temperature of 23°C–29°C and 12‐h light/dark cycle (7:00 AM–7:00 PM) for free access to food pills and tap water. Experimental rats for measuring morphine preference were allocated to three different groups: (1) the saline group ( n = 15); (2) the morphine group that received alternate saline and morphine, without deep brain stimulation (DBS) apparatus implantation ( n = 15); and (3) the morphine–DBS group that received alternate saline and morphine, with DBS apparatus implantation and continuous electrical stimulation in every experiment phase ( n = 13). The rats were acclimated to the laboratory environment for 1 week. The behaviour experiment was carried out in the semidarkness condition. All procedures were approved by the Animal Ethics Committee of Ningxia Medical University. Morphine hydrochloride was purchased from Shenyang First Pharmaceutical Factory (10 mg/mL, Shenyang, China). Morphine‐conditioned place preference was established at a dose of 10 mg/kg (s.c) as in previous studies. , The surgical implantation of DBS apparatus and morphine‐conditioned place preference (CPP) were previously described. 2.2 iTRAQ proteome analysis 2.2.1 Protein preparation The animals (three rats from each saline, morphine and morphine–DBS groups) were anaesthetised by inhalation of isoflurane (3%–5%) and decapitated immediately after the last behavioural test, and the whole brain was removed. The meninges and residual blood were removed in ice physiological saline and then put into the brain slice mould of rats. One‐millimetre‐thick slices close to the electrode channel were obtained. According to the brain atlas of rats (reference), the left and right anterior insula were cut and placed in 1.5 mL EP tube, weighed and recorded, immediately put into the liquid nitrogen tank for 5 min, and quickly transferred to −80°C frozen storage. 2.2.2 Protein extraction The tissue samples were taken from the refrigerator at −80°C, ground into powder at low temperature, transferred to the centrifugal tube precooled with liquid nitrogen, lysed with a proper amount of protein cracking liquid (100 mM ammonium bicarbonate, 8 M urea, 0.2% SDS, pH = 8), shaken and mixed evenly and full cracked by ultrasound in an ice water bath. The lysate was centrifuged at 4°C and 12 000 g for 15 min, and the supernatant was added with 10mM DTT at 56°C for 1 h, followed by a sufficient amount of IAM at room temperature for 1 h in the dark. The samples were completely mixed with four times the volume of precooled acetone by vortexing and incubated at −20°C for at least 2 h, then centrifuged and the precipitation was collected. After washing twice with cold acetone, the pellet was dissolved by dissolution buffer which contained 0.1 M triethylammonium bicarbonate (TEAB, pH 8.5) and 6 M urea. 2.2.3 Protein quantification The protein precipitate was dissolved with a proper amount of protein solution (6 M urea, 100 mM TEAB, pH = 8.5), and the protein concentration was determined by Bradford protein quantitative kit. 2.2.4 iTRAQ labelling of peptides One hundred and twenty micrograms of each protein sample was taken, and the volume was made up to 100 μL with dissolution buffer. Then, 1.5 μg trypsin and 500 μL of 100 mM TEAB buffer were added, and the sample was mixed and digested at 37°C for 4 h. A total of 1.5 μg trypsin and CaCl 2 were added, and the sample was digested overnight. Formic acid was mixed with digested sample, adjusted pH under 3, and centrifuged at 12 000 g for 5 min at room temperature. The supernatant was slowly loaded to the C18 desalting column, washed with washing buffer (0.1% formic acid, 3% acetonitrile) three times and then eluted by some elution buffer (0.1% formic acid, 70% acetonitrile). The eluents of each sample were collected and lyophilized. Twenty microlitres of 1 M TEAB buffer was added to reconstitute, and enough iTRAQ labelling reagent (dissolved in isopropanol) was added; the sample was mixed with shaking for 2 h at room temperature. Then, the reaction was stopped by adding 100 μL of 50mM Tris–HCl (pH = 8). All labelling samples were mixed with equal volume, desalted and lyophilized. Separation of fractions of peptide mixtures labelled with iTRAQ was performed using mobile phases A (2% acetonitrile, adjusted pH to 10.0 using ammonium hydroxide) and B (98% acetonitrile) for gradient elution. The peptide mixture was dissolved in mobile phase A (2% acetonitrile, 98% water, adjusted to pH = 10) and centrifuged at 4°C for 20 min at 14 000 g. The sample was fractionated using a C18 column (Waters BEH C18 4.6 × 250 mm, 5 μm) on a Rigol L3000 HPLC system; the column oven was set to 50°C. Finally, 10 fractions were collected and dried under a vacuum. 2.2.5 LC–MS/MS analysis For transition library construction, shotgun proteomics analyses were performed using an EASY‐nLCTM 1200 UHPLC system (Thermo Fisher) coupled with an Q Exactive HF (X) mass spectrometer (Thermo Fisher) operating in the data‐dependent acquisition (DDA) mode. The sample was injected into a home‐made C18 Nano‐Trap column (2 cm × 75 μm, 3 μm). Mobile phase A (100% water, 0.1% formic acid) and B (80% acetonitrile, 0.1% formic acid) were prepared. Peptides were separated in a home‐made analytical column (15 cm × 150 μm, 1.9 μm), using a linear gradient elution. The separated peptides were analysed by Q Exactive HF (X) mass spectrometer (Thermo Fisher) with an ion source of Nanospray Flex™ (ESI, spray voltage of 2.5 kV and ion transport capillary temperature of 320°C. A full scan range from m/z 407 to 1500 was performed with a resolution of 60 000 (at m/z 200). The automatic gain control (AGC) target value was 3 × 10 6 , and a maximum ion injection time was 20 ms. The top 40 precursors of the highest abundant in the full scan were selected and fragmented by higher energy collisional dissociation (HCD) and analysed in MS/MS, where resolution was 15 000 (at m/z 200), the automatic gain control (AGC) target value was 5 × 10 4 , the maximum ion injection time was 45 ms, a normalized collision energy was set as 32%, an intensity threshold was 2.2 × 10 4 and the dynamic exclusion parameter was 20 s. The raw data of MS detection was named as ‘.raw’. 2.3 Data analysis 2.3.1 Identification and quantification of protein The resulting spectra from each run were searched separately according to the protein database by the search engines: Proteome Discoverer 2.2 (PD 2.2, Thermo). The searched parameters are set as follows: Mass tolerance for precursor ion was 10 ppm, and mass tolerance for product ion was 0.02 Da. Carbamidomethyl was specified as fixed modifications, oxidation of methionine (M) and iTRAQ plex were specified as dynamic modification and acetylation and iTRAQ plex were specified as N‐terminal modification in PD 2.2. A maximum of two miscleavage sites were allowed. In order to improve the quality of analysis results, the software PD 2.2 further filtered the retrieval results: peptide spectrum matches (PSMs) with a credibility of more than 99% were identified PSMs. The identified protein contains at least 1 unique peptide. The identified PSMs and protein were retained and performed with FDR no more than 1.0%. The protein quantitation results were statistically analysed by T ‐test. The proteins, whose quantitation was significantly different between experimental and control groups ( p < 0.05, FC > 1.2 or FC < 0.83 [fold change, FC]), were defined as differentially expressed proteins (DEP). 2.3.2 The functional analysis of proteins and DEPs Gene Ontology (GO) functional analysis was conducted using the InterProScan program against the non‐redundant protein database (including Pfam, PRINTS, ProDom, SMART, ProSite, PANTHER), and the databases of KEGG (Kyoto Encyclopedia of Genes and Genomes) were used to analyse the protein family and pathway. DPEs were used for Volcanic map analysis, cluster heat map analysis and enrichment analysis of GO, KEGG. The probable protein–protein interactions were predicted using the STRING‐db server ( http://string.embl.de/ ). 2.4 Validation of iTRAQ data for selected proteins by parallel reaction monitoring (PRM) assay PRM is a new development of targeted mass spectrometry, which has higher specificity and sensitivity than selected reaction monitoring and has been widely used in the quantification and detection of target proteins. In this study, the protein expression profile obtained by iTRAQ‐based proteomics analysis was confirmed by PRM‐MS analysis to quantify the expression levels of some selected proteins using additional samples. Two candidate proteins related to drug addiction were selected for PRM analysis. Targeted MS analysis using PRM was performed on a TripleTOF EASY‐nLC™ 1200 nano‐UHPLC system. Then, mobile phase A (100% water, 0.1% formic acid) and B (80% acetonitrile, 0.1% formic acid) were prepared. The same amount of trypsin treated‐peptide of each sample was taken and spiked with an equal amount of the labelled peptide DSPSAPVN V TVR (red bold V for heavy isotope labelling) as an internal standard. The UHPLC system was upgraded with Easy‐NLCTM 1200 nM, with a pre‐column of Acclaim Pepmap 100 C18 Nano‐Trap (2 cm × 75 μm, 3 μm) and a self‐made analysis column (15 cm × 150 μm, 1.9 μm). Q Exactive™ HF‐X mass spectrometer and NanoPray Flex™ (ESI) ion source were used. The ion spray voltage was set at 2.4 kV, and the temperature of the ion transport tube was set at 320°C. A full scan with a PRM scan was used for mass spectrometry. The resolution of full‐scan mass spectrometry was set as 60 000 (200 m/z), the maximum capacity of C‐TRAP was 3 × 10 6 and the maximum injection time of C‐TRAP was 20 ms. The PRM resolution was set as 30 000 (200 m/z), the maximum C‐TRAP capacity was 5 × 10 4 , the maximum C‐TRAP injection time was set as 80 ms, the peptide fragmentation collision energy was set as 27% and the raw data of mass spectrometry detection was generated (.RAW). The offline data was analysed by Skyline software, and the peak area was corrected using the internal standard peptide. Animals, DBS surgical implantation and morphine‐conditioned place preference (CPP) Male Sprague–Dawley rats, weighing 260–280 g (6‐8 weeks), were housed at the standard temperature of 23°C–29°C and 12‐h light/dark cycle (7:00 AM–7:00 PM) for free access to food pills and tap water. Experimental rats for measuring morphine preference were allocated to three different groups: (1) the saline group ( n = 15); (2) the morphine group that received alternate saline and morphine, without deep brain stimulation (DBS) apparatus implantation ( n = 15); and (3) the morphine–DBS group that received alternate saline and morphine, with DBS apparatus implantation and continuous electrical stimulation in every experiment phase ( n = 13). The rats were acclimated to the laboratory environment for 1 week. The behaviour experiment was carried out in the semidarkness condition. All procedures were approved by the Animal Ethics Committee of Ningxia Medical University. Morphine hydrochloride was purchased from Shenyang First Pharmaceutical Factory (10 mg/mL, Shenyang, China). Morphine‐conditioned place preference was established at a dose of 10 mg/kg (s.c) as in previous studies. , The surgical implantation of DBS apparatus and morphine‐conditioned place preference (CPP) were previously described. iTRAQ proteome analysis 2.2.1 Protein preparation The animals (three rats from each saline, morphine and morphine–DBS groups) were anaesthetised by inhalation of isoflurane (3%–5%) and decapitated immediately after the last behavioural test, and the whole brain was removed. The meninges and residual blood were removed in ice physiological saline and then put into the brain slice mould of rats. One‐millimetre‐thick slices close to the electrode channel were obtained. According to the brain atlas of rats (reference), the left and right anterior insula were cut and placed in 1.5 mL EP tube, weighed and recorded, immediately put into the liquid nitrogen tank for 5 min, and quickly transferred to −80°C frozen storage. 2.2.2 Protein extraction The tissue samples were taken from the refrigerator at −80°C, ground into powder at low temperature, transferred to the centrifugal tube precooled with liquid nitrogen, lysed with a proper amount of protein cracking liquid (100 mM ammonium bicarbonate, 8 M urea, 0.2% SDS, pH = 8), shaken and mixed evenly and full cracked by ultrasound in an ice water bath. The lysate was centrifuged at 4°C and 12 000 g for 15 min, and the supernatant was added with 10mM DTT at 56°C for 1 h, followed by a sufficient amount of IAM at room temperature for 1 h in the dark. The samples were completely mixed with four times the volume of precooled acetone by vortexing and incubated at −20°C for at least 2 h, then centrifuged and the precipitation was collected. After washing twice with cold acetone, the pellet was dissolved by dissolution buffer which contained 0.1 M triethylammonium bicarbonate (TEAB, pH 8.5) and 6 M urea. 2.2.3 Protein quantification The protein precipitate was dissolved with a proper amount of protein solution (6 M urea, 100 mM TEAB, pH = 8.5), and the protein concentration was determined by Bradford protein quantitative kit. 2.2.4 iTRAQ labelling of peptides One hundred and twenty micrograms of each protein sample was taken, and the volume was made up to 100 μL with dissolution buffer. Then, 1.5 μg trypsin and 500 μL of 100 mM TEAB buffer were added, and the sample was mixed and digested at 37°C for 4 h. A total of 1.5 μg trypsin and CaCl 2 were added, and the sample was digested overnight. Formic acid was mixed with digested sample, adjusted pH under 3, and centrifuged at 12 000 g for 5 min at room temperature. The supernatant was slowly loaded to the C18 desalting column, washed with washing buffer (0.1% formic acid, 3% acetonitrile) three times and then eluted by some elution buffer (0.1% formic acid, 70% acetonitrile). The eluents of each sample were collected and lyophilized. Twenty microlitres of 1 M TEAB buffer was added to reconstitute, and enough iTRAQ labelling reagent (dissolved in isopropanol) was added; the sample was mixed with shaking for 2 h at room temperature. Then, the reaction was stopped by adding 100 μL of 50mM Tris–HCl (pH = 8). All labelling samples were mixed with equal volume, desalted and lyophilized. Separation of fractions of peptide mixtures labelled with iTRAQ was performed using mobile phases A (2% acetonitrile, adjusted pH to 10.0 using ammonium hydroxide) and B (98% acetonitrile) for gradient elution. The peptide mixture was dissolved in mobile phase A (2% acetonitrile, 98% water, adjusted to pH = 10) and centrifuged at 4°C for 20 min at 14 000 g. The sample was fractionated using a C18 column (Waters BEH C18 4.6 × 250 mm, 5 μm) on a Rigol L3000 HPLC system; the column oven was set to 50°C. Finally, 10 fractions were collected and dried under a vacuum. 2.2.5 LC–MS/MS analysis For transition library construction, shotgun proteomics analyses were performed using an EASY‐nLCTM 1200 UHPLC system (Thermo Fisher) coupled with an Q Exactive HF (X) mass spectrometer (Thermo Fisher) operating in the data‐dependent acquisition (DDA) mode. The sample was injected into a home‐made C18 Nano‐Trap column (2 cm × 75 μm, 3 μm). Mobile phase A (100% water, 0.1% formic acid) and B (80% acetonitrile, 0.1% formic acid) were prepared. Peptides were separated in a home‐made analytical column (15 cm × 150 μm, 1.9 μm), using a linear gradient elution. The separated peptides were analysed by Q Exactive HF (X) mass spectrometer (Thermo Fisher) with an ion source of Nanospray Flex™ (ESI, spray voltage of 2.5 kV and ion transport capillary temperature of 320°C. A full scan range from m/z 407 to 1500 was performed with a resolution of 60 000 (at m/z 200). The automatic gain control (AGC) target value was 3 × 10 6 , and a maximum ion injection time was 20 ms. The top 40 precursors of the highest abundant in the full scan were selected and fragmented by higher energy collisional dissociation (HCD) and analysed in MS/MS, where resolution was 15 000 (at m/z 200), the automatic gain control (AGC) target value was 5 × 10 4 , the maximum ion injection time was 45 ms, a normalized collision energy was set as 32%, an intensity threshold was 2.2 × 10 4 and the dynamic exclusion parameter was 20 s. The raw data of MS detection was named as ‘.raw’. Protein preparation The animals (three rats from each saline, morphine and morphine–DBS groups) were anaesthetised by inhalation of isoflurane (3%–5%) and decapitated immediately after the last behavioural test, and the whole brain was removed. The meninges and residual blood were removed in ice physiological saline and then put into the brain slice mould of rats. One‐millimetre‐thick slices close to the electrode channel were obtained. According to the brain atlas of rats (reference), the left and right anterior insula were cut and placed in 1.5 mL EP tube, weighed and recorded, immediately put into the liquid nitrogen tank for 5 min, and quickly transferred to −80°C frozen storage. Protein extraction The tissue samples were taken from the refrigerator at −80°C, ground into powder at low temperature, transferred to the centrifugal tube precooled with liquid nitrogen, lysed with a proper amount of protein cracking liquid (100 mM ammonium bicarbonate, 8 M urea, 0.2% SDS, pH = 8), shaken and mixed evenly and full cracked by ultrasound in an ice water bath. The lysate was centrifuged at 4°C and 12 000 g for 15 min, and the supernatant was added with 10mM DTT at 56°C for 1 h, followed by a sufficient amount of IAM at room temperature for 1 h in the dark. The samples were completely mixed with four times the volume of precooled acetone by vortexing and incubated at −20°C for at least 2 h, then centrifuged and the precipitation was collected. After washing twice with cold acetone, the pellet was dissolved by dissolution buffer which contained 0.1 M triethylammonium bicarbonate (TEAB, pH 8.5) and 6 M urea. Protein quantification The protein precipitate was dissolved with a proper amount of protein solution (6 M urea, 100 mM TEAB, pH = 8.5), and the protein concentration was determined by Bradford protein quantitative kit. iTRAQ labelling of peptides One hundred and twenty micrograms of each protein sample was taken, and the volume was made up to 100 μL with dissolution buffer. Then, 1.5 μg trypsin and 500 μL of 100 mM TEAB buffer were added, and the sample was mixed and digested at 37°C for 4 h. A total of 1.5 μg trypsin and CaCl 2 were added, and the sample was digested overnight. Formic acid was mixed with digested sample, adjusted pH under 3, and centrifuged at 12 000 g for 5 min at room temperature. The supernatant was slowly loaded to the C18 desalting column, washed with washing buffer (0.1% formic acid, 3% acetonitrile) three times and then eluted by some elution buffer (0.1% formic acid, 70% acetonitrile). The eluents of each sample were collected and lyophilized. Twenty microlitres of 1 M TEAB buffer was added to reconstitute, and enough iTRAQ labelling reagent (dissolved in isopropanol) was added; the sample was mixed with shaking for 2 h at room temperature. Then, the reaction was stopped by adding 100 μL of 50mM Tris–HCl (pH = 8). All labelling samples were mixed with equal volume, desalted and lyophilized. Separation of fractions of peptide mixtures labelled with iTRAQ was performed using mobile phases A (2% acetonitrile, adjusted pH to 10.0 using ammonium hydroxide) and B (98% acetonitrile) for gradient elution. The peptide mixture was dissolved in mobile phase A (2% acetonitrile, 98% water, adjusted to pH = 10) and centrifuged at 4°C for 20 min at 14 000 g. The sample was fractionated using a C18 column (Waters BEH C18 4.6 × 250 mm, 5 μm) on a Rigol L3000 HPLC system; the column oven was set to 50°C. Finally, 10 fractions were collected and dried under a vacuum. LC–MS/MS analysis For transition library construction, shotgun proteomics analyses were performed using an EASY‐nLCTM 1200 UHPLC system (Thermo Fisher) coupled with an Q Exactive HF (X) mass spectrometer (Thermo Fisher) operating in the data‐dependent acquisition (DDA) mode. The sample was injected into a home‐made C18 Nano‐Trap column (2 cm × 75 μm, 3 μm). Mobile phase A (100% water, 0.1% formic acid) and B (80% acetonitrile, 0.1% formic acid) were prepared. Peptides were separated in a home‐made analytical column (15 cm × 150 μm, 1.9 μm), using a linear gradient elution. The separated peptides were analysed by Q Exactive HF (X) mass spectrometer (Thermo Fisher) with an ion source of Nanospray Flex™ (ESI, spray voltage of 2.5 kV and ion transport capillary temperature of 320°C. A full scan range from m/z 407 to 1500 was performed with a resolution of 60 000 (at m/z 200). The automatic gain control (AGC) target value was 3 × 10 6 , and a maximum ion injection time was 20 ms. The top 40 precursors of the highest abundant in the full scan were selected and fragmented by higher energy collisional dissociation (HCD) and analysed in MS/MS, where resolution was 15 000 (at m/z 200), the automatic gain control (AGC) target value was 5 × 10 4 , the maximum ion injection time was 45 ms, a normalized collision energy was set as 32%, an intensity threshold was 2.2 × 10 4 and the dynamic exclusion parameter was 20 s. The raw data of MS detection was named as ‘.raw’. Data analysis 2.3.1 Identification and quantification of protein The resulting spectra from each run were searched separately according to the protein database by the search engines: Proteome Discoverer 2.2 (PD 2.2, Thermo). The searched parameters are set as follows: Mass tolerance for precursor ion was 10 ppm, and mass tolerance for product ion was 0.02 Da. Carbamidomethyl was specified as fixed modifications, oxidation of methionine (M) and iTRAQ plex were specified as dynamic modification and acetylation and iTRAQ plex were specified as N‐terminal modification in PD 2.2. A maximum of two miscleavage sites were allowed. In order to improve the quality of analysis results, the software PD 2.2 further filtered the retrieval results: peptide spectrum matches (PSMs) with a credibility of more than 99% were identified PSMs. The identified protein contains at least 1 unique peptide. The identified PSMs and protein were retained and performed with FDR no more than 1.0%. The protein quantitation results were statistically analysed by T ‐test. The proteins, whose quantitation was significantly different between experimental and control groups ( p < 0.05, FC > 1.2 or FC < 0.83 [fold change, FC]), were defined as differentially expressed proteins (DEP). 2.3.2 The functional analysis of proteins and DEPs Gene Ontology (GO) functional analysis was conducted using the InterProScan program against the non‐redundant protein database (including Pfam, PRINTS, ProDom, SMART, ProSite, PANTHER), and the databases of KEGG (Kyoto Encyclopedia of Genes and Genomes) were used to analyse the protein family and pathway. DPEs were used for Volcanic map analysis, cluster heat map analysis and enrichment analysis of GO, KEGG. The probable protein–protein interactions were predicted using the STRING‐db server ( http://string.embl.de/ ). Identification and quantification of protein The resulting spectra from each run were searched separately according to the protein database by the search engines: Proteome Discoverer 2.2 (PD 2.2, Thermo). The searched parameters are set as follows: Mass tolerance for precursor ion was 10 ppm, and mass tolerance for product ion was 0.02 Da. Carbamidomethyl was specified as fixed modifications, oxidation of methionine (M) and iTRAQ plex were specified as dynamic modification and acetylation and iTRAQ plex were specified as N‐terminal modification in PD 2.2. A maximum of two miscleavage sites were allowed. In order to improve the quality of analysis results, the software PD 2.2 further filtered the retrieval results: peptide spectrum matches (PSMs) with a credibility of more than 99% were identified PSMs. The identified protein contains at least 1 unique peptide. The identified PSMs and protein were retained and performed with FDR no more than 1.0%. The protein quantitation results were statistically analysed by T ‐test. The proteins, whose quantitation was significantly different between experimental and control groups ( p < 0.05, FC > 1.2 or FC < 0.83 [fold change, FC]), were defined as differentially expressed proteins (DEP). The functional analysis of proteins and DEPs Gene Ontology (GO) functional analysis was conducted using the InterProScan program against the non‐redundant protein database (including Pfam, PRINTS, ProDom, SMART, ProSite, PANTHER), and the databases of KEGG (Kyoto Encyclopedia of Genes and Genomes) were used to analyse the protein family and pathway. DPEs were used for Volcanic map analysis, cluster heat map analysis and enrichment analysis of GO, KEGG. The probable protein–protein interactions were predicted using the STRING‐db server ( http://string.embl.de/ ). Validation of iTRAQ data for selected proteins by parallel reaction monitoring (PRM) assay PRM is a new development of targeted mass spectrometry, which has higher specificity and sensitivity than selected reaction monitoring and has been widely used in the quantification and detection of target proteins. In this study, the protein expression profile obtained by iTRAQ‐based proteomics analysis was confirmed by PRM‐MS analysis to quantify the expression levels of some selected proteins using additional samples. Two candidate proteins related to drug addiction were selected for PRM analysis. Targeted MS analysis using PRM was performed on a TripleTOF EASY‐nLC™ 1200 nano‐UHPLC system. Then, mobile phase A (100% water, 0.1% formic acid) and B (80% acetonitrile, 0.1% formic acid) were prepared. The same amount of trypsin treated‐peptide of each sample was taken and spiked with an equal amount of the labelled peptide DSPSAPVN V TVR (red bold V for heavy isotope labelling) as an internal standard. The UHPLC system was upgraded with Easy‐NLCTM 1200 nM, with a pre‐column of Acclaim Pepmap 100 C18 Nano‐Trap (2 cm × 75 μm, 3 μm) and a self‐made analysis column (15 cm × 150 μm, 1.9 μm). Q Exactive™ HF‐X mass spectrometer and NanoPray Flex™ (ESI) ion source were used. The ion spray voltage was set at 2.4 kV, and the temperature of the ion transport tube was set at 320°C. A full scan with a PRM scan was used for mass spectrometry. The resolution of full‐scan mass spectrometry was set as 60 000 (200 m/z), the maximum capacity of C‐TRAP was 3 × 10 6 and the maximum injection time of C‐TRAP was 20 ms. The PRM resolution was set as 30 000 (200 m/z), the maximum C‐TRAP capacity was 5 × 10 4 , the maximum C‐TRAP injection time was set as 80 ms, the peptide fragmentation collision energy was set as 27% and the raw data of mass spectrometry detection was generated (.RAW). The offline data was analysed by Skyline software, and the peak area was corrected using the internal standard peptide. RESULTS 3.1 Proteins identified based on iTRAQ The experimental design was illustrated in Figure . In order to understand the proteomic changes in morphine addiction withdrawal and DBS treatment, we used the quantitative proteomic approach based on iTRAQ coupled with 2D‐LC MS/MS (Figure ). A total of 4650 non‐redundant proteins were identified by global proteomic analysis with >99% confidence in correct sequence identification. There were 4224 proteins commonly identified among saline, morphine and morphine–DBS groups. There were 716 proteins detected only in morphine–DBS group (Figure ). PCA can reveal the underlying structure of the data in a way that best explains the variance in the original data set. In the present study, the samples are clustered respectively and separated from each other in the PCA score plot, which suggested that morphine or HF‐DBS intervention for the variations in the proteins identified (Figure ). The quality of the proteomic dataset was evaluated by the coefficient of variance (CV) analysis; the results suggested that there were no biases toward samples of three groups, demonstrating good reproducibility of biological replicates (Figure ). Protein with a significant quantitative difference ( p < 0.05, FC ≥ 1.2 or FC ≤ 0.83) was defined as differential expression protein (DEP). We also identified the significance and magnitude of change in the expression of DEPs across two comparisons, using a heatmap of a hierarchical clustering analysis (HCA) and showing a unique and diverse change pattern of protein abundance in each group (Figure ). Our results showed that there were 14 down‐regulated and three up‐regulated proteins in the morphine versus saline group, as well as 118 up‐regulated and 87 down‐regulated proteins in DBS versus morphine group respectively (Figure ). 3.2 Bioinformatics analysis In order to fully understand the above DEPs, GO analysis is carried out using InterProScan software. A total of 41 GO terms were obtained from differential proteins between morphine and saline groups, including eight biological process terms, three cellular component terms and 30 molecular function terms. The DEPs between DBS and morphine groups resulted in 221 GO terms, including 97 biological process terms, 24 cellular component terms and 100 molecular function terms. The common enriched GO terms from two sets of differential proteins under biological process included translational initiation, G‐protein coupled receptor signalling pathway, microtubule‐based movement (Figure and ). The common enriched GO terms under molecular function included symporter activity, carbonate dehydratase activity, signal transducer activity, G‐protein beta/gamma‐subunit complex binding, protein binding, translation initiation factor activity, microtubule motor activity, transmembrane transporter activity, hydrolase activity, protein binding and nucleoside–triphosphatase activity (Figure and ). DEPs were then mapped to the reference pathway in the KEGG database to determine the biological path involved in morphine addiction withdrawal and DBS therapy. Among these 17 identified DEPs from morphine versus saline, only seven DEPs (five down‐regulated and two up‐regulated) had a KEGG Orthology (KO) ID and were involved in 52 pathways. Among the 205 DEPs identified from DBS versus morphine, 94 DEPs (70 down‐regulated and 24 up‐regulated) were mapped to 106 pathways. Many enriched pathways are close to addiction including PI3K‐Akt signalling pathway, calcium signalling pathway, mTOR signalling pathway and nicotine addiction (Figure ). 3.3 Protein–protein interaction analysis of DEPs The protein–protein interaction (PPI) network of the DEPs identified in present study based on STRING action scores was analysed. The PPI of eight common DEPs is shown in Figure , revealing that guanine nucleotide‐binding protein G (olf) subunit alpha (Gnal), eukaryotic translation initiation factor 4E family member 2 (Eif4e2), kinesin‐like protein KIF1B (kif1b), amino acid transporter and translational activator of cytochrome c oxidase 1 (taco1) were present at hub positions of the networks (Figure ). 3.4 PRM verification of selected DEPs Two candidate proteins (eukaryotic translation initiation factor 4E family member 2, guanine nucleotide‐binding protein G (olf) subunit alpha) were of particular interest and selected because they were mapped to several pathways involved in addiction. As shown in Figure , the expression of proteins verified by PRM was in accordance with proteomic data. Proteins identified based on iTRAQ The experimental design was illustrated in Figure . In order to understand the proteomic changes in morphine addiction withdrawal and DBS treatment, we used the quantitative proteomic approach based on iTRAQ coupled with 2D‐LC MS/MS (Figure ). A total of 4650 non‐redundant proteins were identified by global proteomic analysis with >99% confidence in correct sequence identification. There were 4224 proteins commonly identified among saline, morphine and morphine–DBS groups. There were 716 proteins detected only in morphine–DBS group (Figure ). PCA can reveal the underlying structure of the data in a way that best explains the variance in the original data set. In the present study, the samples are clustered respectively and separated from each other in the PCA score plot, which suggested that morphine or HF‐DBS intervention for the variations in the proteins identified (Figure ). The quality of the proteomic dataset was evaluated by the coefficient of variance (CV) analysis; the results suggested that there were no biases toward samples of three groups, demonstrating good reproducibility of biological replicates (Figure ). Protein with a significant quantitative difference ( p < 0.05, FC ≥ 1.2 or FC ≤ 0.83) was defined as differential expression protein (DEP). We also identified the significance and magnitude of change in the expression of DEPs across two comparisons, using a heatmap of a hierarchical clustering analysis (HCA) and showing a unique and diverse change pattern of protein abundance in each group (Figure ). Our results showed that there were 14 down‐regulated and three up‐regulated proteins in the morphine versus saline group, as well as 118 up‐regulated and 87 down‐regulated proteins in DBS versus morphine group respectively (Figure ). Bioinformatics analysis In order to fully understand the above DEPs, GO analysis is carried out using InterProScan software. A total of 41 GO terms were obtained from differential proteins between morphine and saline groups, including eight biological process terms, three cellular component terms and 30 molecular function terms. The DEPs between DBS and morphine groups resulted in 221 GO terms, including 97 biological process terms, 24 cellular component terms and 100 molecular function terms. The common enriched GO terms from two sets of differential proteins under biological process included translational initiation, G‐protein coupled receptor signalling pathway, microtubule‐based movement (Figure and ). The common enriched GO terms under molecular function included symporter activity, carbonate dehydratase activity, signal transducer activity, G‐protein beta/gamma‐subunit complex binding, protein binding, translation initiation factor activity, microtubule motor activity, transmembrane transporter activity, hydrolase activity, protein binding and nucleoside–triphosphatase activity (Figure and ). DEPs were then mapped to the reference pathway in the KEGG database to determine the biological path involved in morphine addiction withdrawal and DBS therapy. Among these 17 identified DEPs from morphine versus saline, only seven DEPs (five down‐regulated and two up‐regulated) had a KEGG Orthology (KO) ID and were involved in 52 pathways. Among the 205 DEPs identified from DBS versus morphine, 94 DEPs (70 down‐regulated and 24 up‐regulated) were mapped to 106 pathways. Many enriched pathways are close to addiction including PI3K‐Akt signalling pathway, calcium signalling pathway, mTOR signalling pathway and nicotine addiction (Figure ). Protein–protein interaction analysis of DEPs The protein–protein interaction (PPI) network of the DEPs identified in present study based on STRING action scores was analysed. The PPI of eight common DEPs is shown in Figure , revealing that guanine nucleotide‐binding protein G (olf) subunit alpha (Gnal), eukaryotic translation initiation factor 4E family member 2 (Eif4e2), kinesin‐like protein KIF1B (kif1b), amino acid transporter and translational activator of cytochrome c oxidase 1 (taco1) were present at hub positions of the networks (Figure ). PRM verification of selected DEPs Two candidate proteins (eukaryotic translation initiation factor 4E family member 2, guanine nucleotide‐binding protein G (olf) subunit alpha) were of particular interest and selected because they were mapped to several pathways involved in addiction. As shown in Figure , the expression of proteins verified by PRM was in accordance with proteomic data. DISCUSSION In the current study, we utilized iTRAQ‐based proteomic analysis and determined the differential regulatory proteins in the insula during the morphine addiction withdrawal and DBS treatment in morphine addictive animal model. HF‐DBS has been shown to regulate neuronal activity in the stimulated area, alter the expression of neural activity genes and affect the occurrence and development of addiction, as confirmed in our previous work. Among them, there were 17 DEPs between morphine and saline groups and 205 when morphine–DBS compared to the morphine group (Data ). GO and KEGG analysis showed that DEPs were related to multiple signalling pathways including PI3K‐Akt pathway. Two DEPs involved in DBS therapy are of particular interest: Eif4e2 and Gnal, due to their important role in signalling pathways associated with addiction from relevant literature. PPI network analysis also revealed that Eif4e2 and Gnal were shown at the hub positions, which interacted with other important proteins. Eukaryotic translation initiation factor 4E (Eif4e) serves as a central player in the complex processes of eukaryotic translation initiation and regulation. Eif4e2 is recognized as one of its family members. Eif4e's specialized role involves binding to the 7‐methylguanosine triphosphate(m7GpppG) cap structure found at the 5′ end of mRNA. This binding profoundly affects mRNA metabolism, processing, transport and translation, making it a crucial contributor to the initial stage of protein synthesis. , , Recent research has shown that Eif4e‐dependent translation is important in nerve function, particularly in the context of neurodevelopment and neuropsychiatric diseases. , , Notably, Eif4e's activity is strictly regulated by various factors, such as hormones, growth factors, cytokines and extracellular stimuli. It is mainly involved in two key signalling pathways: MAPK/ERK and PI3K/mTOR. , Numerous studies have consistently demonstrated that u‐opioid receptor induces activation of PI3K/Akt/mTOR pathway. , , , In the context of morphine‐induced CPP, the PI3K/Akt–mTOR signalling pathway activated by a μ‐opioid receptor in the CA3 region of the hippocampus plays an important role in the acquisition of CPP. , We demonstrate that morphine‐regulated DEPs in the insular cortex are enriched in PI3K‐Akt pathway that is consistent with previous studies. These findings collectively highlight the pivotal role of the PI3K/Akt–mTOR pathway in the development of drug addiction. Furthermore, interventions targeting this pathway offer a promising and novel approach to address issues related to drug abuse. Guanine nucleotide‐binding protein G (olf) subunit α (Gαolf), encoded by GNAL gene located on chromosome 18p11, belongs to the family of GTP‐binding proteins (G protein). These proteins serve as intermediaries, connecting G protein‐coupled receptors (GPCRs) to adenylate cyclase. , The heterotrimeric G protein complex comprises three subunits: α, β and γ. In humans, there are 21 G21 Gα, 6Gβ and 12 Gγ units identified. G proteins can be categorized into four subfamilies (Gαs, Gαi/o, Gαq and Gα12) based on the sequence similarity of Gα subunits. Gα (olf), with an 88% homology to Gαs, is considered to be a member of the Gαs family. G‐proteins play an important role in transmitting external signals to the interior of cells, when activated by GPCRs localized on the cell membrane. They are involved in a wide array of physiological functions and are implicated in numerous pathological conditions. In our study, Gαolf was mapped to the dopaminergic synapse pathway (map 04728) using the Kyoto Encyclopedia of Genes and Genomes (KEGG) reference database ( http://www.genome.jp/kegg/pathway.html ). The dopaminergic pathways are associated with processes such as reward sensitivity, incentive motivation, conditioning, and control. Dysregulation of dopamine activity can lead to loss of control over intake and continued consumption despite adverse consequences, which is commonly observed in addiction and obesity. Gαolf functions as a modulator during neurotransmission and is closely associated with dopamine signalling. Dopamine primarily exerts its effects through D1‐like (D1, D5) and D2‐like (D2, D3, D4) receptors, which, in turn, regulate the activation or inhibition of cAMP accumulation, via Gs/olf or Gi/o proteins, respectively. Furthermore, Gαolf was also mapped to the calcium signalling pathway in our study. GPCRs and G‐protein effectors are intricately involved in calcium signalling. Various physiological stimuli, including odours, light, metal ions, peptide hormones and neurotransmitters, induce a conformational change in GPCRs, activating heterotrimeric G‐proteins and downstream effectors like adenylate cyclase and phospholipase C. These effectors generate diffusible second messengers that trigger intracellular changes including alterations in cytosolic calcium concentration. Calcium signalling is broadly implicated in regulating various aspects of cell function and can be influenced by tissue injury, resulting in alterations in cell functions. Dysregulation of calcium signalling and subsequent changes BDNF expression may be linked to drug addiction. In a proteomic analysis using iTRAQ, DEPs associated with calcium‐mediated signalling were identified in nicotine‐induced CPP rats. Notably, a pathway analysis conducted in a genome‐wide association study (GWAS) investigating opioid dependence in African‐American and European‐American populations highlighted calcium signalling as the most significant pathways, offering potential insights into novel therapeutic and preventive strategies for drug addiction. A gene network analysis employing the bioinformatics tool IPA, to identify commonly shared genes associated with alcohol, smoking and opioid addiction, revealed several top canonical pathways, including calcium signalling. GPCR signalling, cAMP‐mediated signalling, GABA receptor signalling and Gαi signalling. These pathways have been consistently linked to substance addiction in existing literature. Together with prior research, our study identifies Gαolf, an effector of GPCR, as a key player in morphine addiction mediated through the dopaminergic synapse pathway and calcium signalling pathways. Furthermore, our study demonstrates that DBS reverses the expression of Gαolf in morphine‐addicted animals, shedding light on the potential mechanism of DBS in treating drug addiction. In summary, our iTRAQ‐based proteomic analysis revealed 17 differential expressed proteins in the insula of morphine‐addicted rats following withdrawal. DBS treatment for drug addiction has been extensively analysed in our team's previous published article. The insula plays a crucial role in drug addiction, and compared to other brain regions or nuclei, the insula is larger, making it easier for high‐frequency DBS to receive clear stimulation interventions. Of note, eight of these proteins exhibited changes with DBS intervention, involving signal transduction, translation regulation and neurotransmitter transmission. This contributes to a deeper understanding of morphine addiction pathogenesis and the molecular mechanisms underlying DBS therapy. Additionally, the proteins identified in this study may serve as valuable therapeutic targets, contributing to the advancement of DBS‐based treatments for drug addiction. Haigang Chang performed the experiments, analysed the data and drafted the manuscript. Yaxiao Wang and Lei Hui conducted the study, including data collection and analysis. Yuling Diao and Pengju Ma contributed to the manuscript drafting. Feng Wang designed the experiments and contributed to manuscript drafting. Xiangsheng Li supervised the study. All authors critically reviewed the content and approved the final version for publication. The authors declare that there are no financial and personal relationships with other people or organizations that can inappropriately influence our work, and there is no professional or other personal interest of any nature or kind in any product, service and/or company that could be construed as influencing the position presented in, or the review of, the manuscript entitled. Data S1. Supporting Information. |
Investigation of the metabolic and endocrinological differences between daily and weekly growth hormone replacement therapy, somapacitan, in patients with adult growth hormone deficiency: A real-world pilot study | 586fc2f9-71b7-4b63-ae94-c992518e2440 | 10519569 | Physiology[mh] | Adult growth hormone deficiency (AGHD) is one of the anterior pituitary hormonal deficiencies. Growth hormone (GH) plays important roles not only in childhood but also in adulthood. Patients with severe AGHD have fatty liver/nonalcoholic steatohepatitis (NASH)/nonalcoholic fatty liver disease (NAFLD), increased visceral adiposity, osteoporosis, poor concentration/inattention, impaired quality of life, coronary artery disease, and heart failure. Moreover, Pappachan et al reported that AGHD can lead to increased mortality. Hence, GH replacement therapy is essential for patients with AGHD. For a long time, daily GH replacement therapy was the only treatment available for patients with AGHD. Recently, patients with AGHD have had the opportunity to receive weekly GH replacement therapy (long-acting GH: somapacitan). The efficacy and safety of somapacitan have been revealed in some phase 3 trials, however, no real-world study has been reported. Thus, we report the first investigation of the clinical, metabolic, and endocrinological differences between daily GH replacement therapy and weekly GH replacement therapy with somapacitan in patients with AGHD in the real world. 2.1. Ethical approval of the study protocol The study protocol was approved by the ethics review committees of Fukuoka University (Fukuoka, Japan). Written informed consent was obtained from all patients for participation in the study. The study was conducted in accordance with the principles of the Declaration of Helsinki. 2.2. Study participants We investigated 11 individuals with AGHD previously diagnosed with no or inadequate changes in GH levels after a GH-releasing peptide-2 test/insulin tolerance test/arginine test at Fukuoka University Chikushi Hospital and Nagasaki Prefecture Iki Hospital. All patients had received daily GH replacement therapy for over 2 years and switched from daily GH replacement therapy to weekly GH replacement therapy with somapacitan between January 2022 and June 2022. 2.3. Methods and disease definitions We administered and continued treatment with somapacitan for 6 months in patients with AGHD previously receiving daily GH replacement therapy. Starting doses of somapacitan were 1.5 mg/week for adults 18 to 60 years of age and 1.0 mg/week for patients aged >60 years, referring to the recommendation based on phase 3 trial in Japan (REAL Japan). Dose titration was performed according to the value. Dose titration of somapacitan was performed according to the value of insulin-like growth factor 1 (IGF1) which was checked monthly after switching to somapacitan. The IGF1 value was basically checked at the morning 7 days after the injection of somapacitan (the next somapacitan injection was performed at night on the day), when the IGF1 value was assumed to be bottom in the week. If the IGF1 value was lower than −1 standard deviation score (SDS), the dose of somapacitan was increased (+0.5 mg/week). If the IGF1 value was higher than +1 standard deviation score (SDS), the dose of somapacitan was increased (−0.5 mg/week). In addition, to avoid overdose, the IGF1 value was also sometimes checked in the morning 2 to 3 days after the injection of somapacitan, when the IGF1 value was assumed to peak in the week. The following variables were examined at switching and 6 months after switching to somapacitan: parameters of glucose control (glycated hemoglobin [HbA1c], fasting plasma glucose [FPG], homeostasis model assessment of insulin resistance [HOMA-IR], and homeostasis model assessment of β-cell function [HOMA-β]), markers of lipid metabolism (low-density lipoprotein-cholesterol, high-density lipoprotein cholesterol, and triglycerides), liver functions (aspartate transaminase [AST], alanine transaminase [ALT], and gamma-glutamyl transferase [γ-GTP]), estimated glomerular filtration rate, and body mass index (BMI). Blood samples were obtained after overnight fasting, and HOMA-IR was calculated using the following formula: HOMA-IR = FPG × fasting insulin/405 HOMA-β was calculated using the following formula: HOMA-β = 360 × fasting insulin/(FPG − 63) Endocrinologically, anterior pituitary hormones and related hormones (adrenocorticotropic hormone [ACTH], cortisol, TSH, free T4, luteinizing hormone [LH], follicle-stimulating hormone [FSH], and testosterone [male]/estradiol [female]) were measured at switching and 6 months after switching to somapacitan. ACTH deficiency was diagnosed by a combination of reduced ACTH and cortisol levels in the morning, and no or inadequate changes in ACTH or cortisol levels after a corticotropin-releasing hormone test. TSH deficiency was diagnosed based on a combination of reduced TSH levels, no or inadequate changes in TSH levels after a thyrotropin-releasing hormone test, and existing secondary hypothyroidism. Deficiency in LH or FSH was diagnosed by a combination of reduced LH or FSH levels, no or inadequate changes in LH or FSH levels after an LH-releasing hormone test, and existing secondary hypogonadism. Central diabetes insipidus was diagnosed by a combination of increased urinary volume; low urinary osmolarity; low antidiuretic hormone (ADH) levels compared with serum osmolarity; no or inadequate changes in ADH levels after a water restriction test/5% NaCl loading test; and increased ADH levels and decreased urinary volume after 1-desamino-8-D-arginine vasopressin administration. Medical treatment aside from GH replacement therapy did not change in any of the patients for the duration of this study. 2.4. Statistical analyses Data are shown as the mean ± standard deviation (SD). Statistical analyses were performed using Stata SE version 16 (StataCorp.2019. Stata Statistical Software: Release 16. College Station, TX: Stata Corp LLC.). The Student t test was used to assess the significance of differences between mean values. This relationship was examined using univariate regression analysis (Fisher test). P value < .05 was considered significant. The study protocol was approved by the ethics review committees of Fukuoka University (Fukuoka, Japan). Written informed consent was obtained from all patients for participation in the study. The study was conducted in accordance with the principles of the Declaration of Helsinki. We investigated 11 individuals with AGHD previously diagnosed with no or inadequate changes in GH levels after a GH-releasing peptide-2 test/insulin tolerance test/arginine test at Fukuoka University Chikushi Hospital and Nagasaki Prefecture Iki Hospital. All patients had received daily GH replacement therapy for over 2 years and switched from daily GH replacement therapy to weekly GH replacement therapy with somapacitan between January 2022 and June 2022. We administered and continued treatment with somapacitan for 6 months in patients with AGHD previously receiving daily GH replacement therapy. Starting doses of somapacitan were 1.5 mg/week for adults 18 to 60 years of age and 1.0 mg/week for patients aged >60 years, referring to the recommendation based on phase 3 trial in Japan (REAL Japan). Dose titration was performed according to the value. Dose titration of somapacitan was performed according to the value of insulin-like growth factor 1 (IGF1) which was checked monthly after switching to somapacitan. The IGF1 value was basically checked at the morning 7 days after the injection of somapacitan (the next somapacitan injection was performed at night on the day), when the IGF1 value was assumed to be bottom in the week. If the IGF1 value was lower than −1 standard deviation score (SDS), the dose of somapacitan was increased (+0.5 mg/week). If the IGF1 value was higher than +1 standard deviation score (SDS), the dose of somapacitan was increased (−0.5 mg/week). In addition, to avoid overdose, the IGF1 value was also sometimes checked in the morning 2 to 3 days after the injection of somapacitan, when the IGF1 value was assumed to peak in the week. The following variables were examined at switching and 6 months after switching to somapacitan: parameters of glucose control (glycated hemoglobin [HbA1c], fasting plasma glucose [FPG], homeostasis model assessment of insulin resistance [HOMA-IR], and homeostasis model assessment of β-cell function [HOMA-β]), markers of lipid metabolism (low-density lipoprotein-cholesterol, high-density lipoprotein cholesterol, and triglycerides), liver functions (aspartate transaminase [AST], alanine transaminase [ALT], and gamma-glutamyl transferase [γ-GTP]), estimated glomerular filtration rate, and body mass index (BMI). Blood samples were obtained after overnight fasting, and HOMA-IR was calculated using the following formula: HOMA-IR = FPG × fasting insulin/405 HOMA-β was calculated using the following formula: HOMA-β = 360 × fasting insulin/(FPG − 63) Endocrinologically, anterior pituitary hormones and related hormones (adrenocorticotropic hormone [ACTH], cortisol, TSH, free T4, luteinizing hormone [LH], follicle-stimulating hormone [FSH], and testosterone [male]/estradiol [female]) were measured at switching and 6 months after switching to somapacitan. ACTH deficiency was diagnosed by a combination of reduced ACTH and cortisol levels in the morning, and no or inadequate changes in ACTH or cortisol levels after a corticotropin-releasing hormone test. TSH deficiency was diagnosed based on a combination of reduced TSH levels, no or inadequate changes in TSH levels after a thyrotropin-releasing hormone test, and existing secondary hypothyroidism. Deficiency in LH or FSH was diagnosed by a combination of reduced LH or FSH levels, no or inadequate changes in LH or FSH levels after an LH-releasing hormone test, and existing secondary hypogonadism. Central diabetes insipidus was diagnosed by a combination of increased urinary volume; low urinary osmolarity; low antidiuretic hormone (ADH) levels compared with serum osmolarity; no or inadequate changes in ADH levels after a water restriction test/5% NaCl loading test; and increased ADH levels and decreased urinary volume after 1-desamino-8-D-arginine vasopressin administration. Medical treatment aside from GH replacement therapy did not change in any of the patients for the duration of this study. Data are shown as the mean ± standard deviation (SD). Statistical analyses were performed using Stata SE version 16 (StataCorp.2019. Stata Statistical Software: Release 16. College Station, TX: Stata Corp LLC.). The Student t test was used to assess the significance of differences between mean values. This relationship was examined using univariate regression analysis (Fisher test). P value < .05 was considered significant. Table presents the patient characteristics in our study. All patients underwent AGHD and daily GH replacement therapy. The mean age was 61.9 ± 18.9 years, and 9 patients were female. The BMI was 26.7 ± 5.2. Endocrinologically, 54.5, 54.5, 45.5, and 45.5% of the patients had ACTH, TSH, LH, and FSH deficiencies, respectively. A total of 9.1% of patients had central diabetes insipidus. In addition, hydrocortisone, levothyroxine, human chorionic gonadotrophin/human menopausal gonadotropin (HCG/HMG), and testosterone/estrogen, and desmopressin replacement therapy were performed in 54.5, 63.6, 0.0, 0.0, and 9.1% of patients (one patient did not have TSH deficiency but rather primary hypothyroidism, and was administered levothyroxine). As a result of titration, IGF1 values 6 months after switching to somapacitan were almost the same as those at switching (100.2 ± 44.3 vs 98.3 ± 40.5 ng/mL, P = .316). The IGF1 values of all patients at switching and 6 months after switching to somapacitan were between −1SDS and +1SDS, and the value of IGF1 at 2–3 days after the injection of somapacitan were almost the same as the 7 days after injection, which indicated larger amount of somapacitan must not be injected. The mean dose of daily GH replacement at switching was 0.20 ± 0.07 mg/day, and the dose of somapacitan 6 months after switching was 1.45 ± 0.35 mg/week (Table ). Table shows the changes in clinical, metabolic, and endocrinological parameters. In terms of glucose tolerance, HOMA-IR and FPG were significantly improved 6 months after switching compared with those at switching (HOMA-IR: 3.1 ± 1.6 vs 2.3 ± 1.3, P = .022; FPG: 104.5 ± 19.6 vs 99.3 ± 16.9 mg/dL, P = .044). Meanwhile, HbA1c and HOMA-β did not improve from at switching to 6 months after switching (HbA1c: 6.3 ± 0.5 vs 6.2 ± 0.5%, P = .174, HOMA-β: 148.0 ± 138.3 vs 110.5 ± 78.0, P = .140). The markers of lipid metabolism, estimated glomerular filtration rate, and electrolyte levels did not change significantly from switching to 6 months after switching. In contrast, all measured liver functions improved significantly from at switching to 6 months after switching (AST: 23.4 ± 3.4 vs 19.8 ± 5.1 U/L, P = .001; ALT: 19.6 ± 5.6 vs 15.0 ± 4.4 U/L, P = .004; γ-GTP: 25.2 ± 16.4 vs 20.8 ± 11.1 U/L, P = .018, respectively). Regarding clinical parameters, BMI improved significantly from at switching to 6 months after switching (26.7 ± 5.2 vs 26.1 ± 5.3, P = .007, respectively). In addition, systolic/diastolic blood pressure improved from at switching to 6 months after switching, but not significantly (142.2 ± 14.9 vs 139.4 ± 10.8/81.6 ± 10.9 vs 79.3 ± 8.4 mm Hg, P = .253/0.182). Regarding endocrinological parameters, there were no differences between the values at switching and those 6 months after switching for all anterior pituitary hormones and related hormones (Table ). In addition, Fisher test showed that age, sex, improvement in BMI or liver functions, presence of any hormonal deficiency, and the existence of any hormonal replacement therapy were not associated with improvement in HOMA-IR. However, daily GH replacement therapy periods were significantly positively associated with improvements in HOMA-IR ( P = .048) (Table ). AGHD is one of the pituitary hormonal deficiencies (anterior pituitary hormonal deficiencies and diabetes insipidus); however, replacement therapy could not be performed before 2006 in Japan. Furthermore, daily GH replacement therapy has been the only available treatment since 2006. It is well known that GH concentrations in the blood are normally high at midnight and low in the daytime, whereas those in patients with AGHD are extremely low during the entire day. Hence, patients with AGHD commonly self-injected GH on a nightly basis (7:00 pm–8.00 pm). However, GH concentrations in the blood of patients with AGHD during the daytime would remain severely low compared to that in normal subjects because the prolonged duration of daily GH formulation was <12 hours. Recently, patients with AGHD have been able to use somapacitan, the only weekly GH formulation in Japan. Nevertheless, the prolonged duration of somapacitan is more than 1 week, and the effect of this formulation continues throughout the day. Thus, these differences in duration could affect metabolic and endocrinological parameters. Patients with severe AGHD have many kinds of metabolic disorders. AGHD must be similar to metabolic syndrome at the point of the association with obesity, insulin resistance, visceral fat, lipid profile, NASH/NAFLD, and the risk of coronary heart disease. In our study, BMI was lower 6 months after switching to somapacitan than at switching. Since the prevalence of obesity in patients with AGHD is well known, and GH replacement therapy has been reported to improve obesity, the results of our study indicate that weekly GH replacement therapy with somapacitan could be better than daily GH replacement therapy. Nevertheless, in our study, the comparison between somapacitan and the other long-acting GH formulations was not performed. Future studies could prove the existence of the difference among long-acting GH formulations. Regarding liver dysfunction, AST/ALT/γ-GTP was significantly improved by switching from daily GH replacement therapy to weekly GH replacement therapy with somapacitan. AGHD is well-known to cause NASH/NAFLD. In phase 3 trials of somapacitan, the analysis of liver functions was not performed. The results of our study indicate that weekly GH replacement therapy with somapacitan could be more effective than daily GH replacement therapy, probably because continuous GH replacement therapy by somapacitan could improve liver dysfunction, considering that the values of IGF1 at switching and 6 months after switching were equivalent. Similarly, HOMA-IR and FPG levels improved after switching from daily GH replacement therapy to somapacitan. Considering these data and the lack of change in HOMA-β, the improvement in FPG might be due to the improvement in insulin resistance. Recently, Takahashi et al reported that there were no significant differences between daily GH formulations and somapacitan, summarizing the data of phase 3 clinical trials (REAL 1 [NCT02229851], REAL 2 [NCT02382939], REAL Japan [NCT03075644]). The differences could be caused that the no medical treatment of all patients in our study changed during this period, aside from GH formulations, while it is quite possible that medical treatment, aside from GH formulations, may have been changed in some patients in phase 3 trials of somapacitan. Thus, the results of insulin tolerance in our real-world study could be different from those in the previous report by Takahashi et al. Interestingly, in the same report, they also showed that the post hoc analysis of one of the phase 3 studies (REAL 1 [NCT02229851]) indicated the group receiving somapacitan had significantly lower HOMA-IR and FPG levels than those receiving daily GH formulation at 32 weeks after beginning. Moreover, the analysis also showed there were no significant differences in HbA1c values between the group receiving somapacitan and those receiving daily GH formulation. In our study, there were also no significant differences between the HbA1c values at switching and 6 months after switching to somapacitan. These results were consistent with those of the present study. Several factors regulate insulin resistance. In our study, BMI was lower 6 months after switching to somapacitan than at the time of switching, but with a very slight difference (26.7 ± 5.2 vs 26.1 ± 5.3 kg/m 2 ), which was difficult to explain the differences in HOMA-IR. In addition, a previous clinical trial showed no differences in body composition between the daily GH formulation and somapacitan. Our study demonstrated an improvement in liver functions after switching to somapacitan, which might lead to an improvement in hepatic insulin resistance. Nonetheless, patients with AGHD may have excessive hepatic insulin resistance caused by fatty liver/NASH/NAFLD. Hence, somapacitan could have more beneficial effects on insulin resistance as well as liver dysfunctions in AGHD than the daily GH formulation. On the other hand, a significant association between the improvement in HOMA-IR and liver functions was not revealed by Fisher test, probably because our study cohort was small. Thus, future studies with larger cohorts are expected to confirm the results of our study. Meanwhile, Fisher test revealed that improvement in HOMA-IR was significantly associated with the period of daily GH replacement therapy before switching to somapacitan, even though our cohort was small. These findings suggest that daily GH replacement formulation could be less efficient, at least for glucose intolerance, than somapacitan, considering that the values of IGF1 at switching and 6 months after switching were equivalent and switching from daily GH formulation to somapacitan should be considered as soon as possible if patients with AGHD were treated with daily GH replacement therapy. We also demonstrated that switching to GH replacement therapy did not affect endocrinological parameters. Previously, it was reported that GH could have antagonistic effects against 11β-HSD1, which could lead to lower cortisol values and higher ACTH values. Our study showed that ACTH and cortisol levels did not change after switching from daily GH replacement therapy to weekly GH replacement therapy with somapacitan. There were no changes in other endocrinological parameters. Hence, somapacitan can be used without impairing the endocrinological condition of patients. Our study had some limitations. First, the sample size was small because AGHD is a rare and intractable disease; this study was a real-world pilot study; and we excluded the patients whose medical treatment, aside from GH replacement therapy, was changed during the study period. In addition, the period of our study was not so long. Hence, future studies with larger cohorts are longer observation periods are required to confirm the results of our study. Second, we used HOMA-IR and HOMA-β as surrogate markers of insulin resistance, and insulin secretion as a substitute for an oral glucose tolerance test or hyperglycemic/hyperinsulinemic-euglycemic clamps. In addition, our study could not reveal the mechanism of improvement of HOMA-IR and FPG adequately. Future studies are also required on this point. Third, our study is a real-world pilot study, and thus could not check the changes in visceral fat content, bone density, bone metabolism markers, and myocardial zymogram, which were complications of AGHD. Therefore, future studies will investigate the differences between the group receiving somapacitan and those receiving daily GH formulation regarding the complications of AGHD. Our study is the first investigation of the effects of somapacitan on metabolic and endocrinological parameters in patients with AGHD who previously received daily GH replacement therapy in a real-world pilot study and reveals that weekly GH replacement therapy with somapacitan could have more beneficial effects on liver functions than daily GH replacement therapy. Furthermore, it is possible that weekly GH replacement therapy with somapacitan could improve glucose intolerance by reducing insulin resistance compared with daily GH replacement therapy. Future studies are required to confirm and develop our study. We thank Ms. Yumi Iriguchi for her assistance in conducting our study. Conceptualization: Ichiro Abe, Kaori Takeshita, Hideaki Shimada, Shigeaki Mukoubara, Kunihisa Kobayashi. Data curation: Ichiro Abe, Kaori Takeshita, Mai Nagata, Yuya Fujita, Kentaro Ochi, Midori Koga, Tadachika Kudo, Makiko Abe. Formal analysis: Ichiro Abe, Makiko Abe. Investigation: Ichiro Abe, Kaori Takeshita, Mai Nagata, Yuya Fujita, Kentaro Ochi, Midori Koga, Tadachika Kudo, Makiko Abe, Shigeaki Mukoubara, Kunihisa Kobayashi. Methodology: Ichiro Abe, Hideaki Shimada. Project administration: Ichiro Abe, Shigeaki Mukoubara, Kunihisa Kobayashi. Resources: Ichiro Abe. Supervision: Ichiro Abe, Shigeaki Mukoubara, Kunihisa Kobayashi. Validation: Ichiro Abe. Writing – original draft: Ichiro Abe. Writing – review & editing: Kaori Takeshita, Mai Nagata, Yuya Fujita, Kentaro Ochi, Midori Koga, Tadachika Kudo, Hideaki Shimada, Makiko Abe, Shigeaki Mukoubara, Kunihisa Kobayashi. |
How accurate is clinical prognostication by oncologists during routine practice in a general hospital and can it be improved by a specific prognosis training programme: a prospective interventional study | c28519f7-6653-469d-a397-58f8b53d0a62 | 11191806 | Internal Medicine[mh] | In order to deliver appropriate care to patients with cancer, an oncologist needs competence not only as a diagnostician or as a therapist but also as a prognosticator. The relevance of prognostication may not always be obvious, but it underlies virtually every aspect of patient care. For example, the same clinical symptom (severe respiratory insufficiency) may lead to very different courses of action (eg, transfer to the ICU vs best supportive care measures) depending on the estimate of the patient’s prognosis by the treating physician. The oncologist may have made these estimates consciously or unconsciously, but they nevertheless constitute the background in front of which everyday patient care takes place. Prognostication is an uncertain business and the ability of oncologists to estimate the future course of their patients is limited. Most explicit analyses of physicians’ abilities to prognosticate have been made in the palliative care setting on patients with very short life expectancies. In contrast, this paper evaluates the prognostic abilities of residents and experienced oncologists in a department of oncology and haematology in a major acute care teaching hospital, where patients had an average life expectancy of >2 years. This paper also investigates how far a training programme can improve prognostic performance. There are three major approaches to indicate the prognosis of a patient: (1) the temporal approach (also called ‘clinician prediction of survival’ (CPS)), (2) the surprise question (SQ) and (3) the probabilistic approach. Of these, the temporal approach is the most commonly used method within the literature. When using the temporal approach, the prognosticator is required to state a specific period in absolute numbers (days, months, years). Many studies consider the prognosis to be ‘accurate’ if the CPS is within the range 67%–133% in relation to the actual survival of the observed patient (which is set as 100%). The SQ, which was originally developed as a screening tool for the initiation of palliative care, requires physicians to answer whether they would be surprised if their patient had died within a certain time. There are only two options (yes/no) and the estimation period is defined (eg, 6 months). The SQ is accurate if the answer is ‘yes’ and the patient has survived or if the option is ‘no’ and the patient has died. The third option is the probabilistic approach. Here, the respondent must indicate the probability that a patient has or has not died at a given time. (Like in the SQ) the period of interest is defined (eg, 6 months), but the prognostic estimate is provided in percentage increments (usually 10%) and thus has a quantitative component. Prognostication per se (= estimating a prognosis) is not the same as conveying a prognosis to a patient (eg, ‘breaking bad news’). The latter ability (which is not the topic of this paper) is also of high relevance and is generally regarded to require considerable experience and clinical expertise (which is considered teachable). In contrast, the preceding process of estimating the prognosis—that is, prognostication as such—is described as ‘Medicine’s Lost Art’ and remains a black box in two ways: (1) The prognostic performance of clinicians—outside the inpatient palliative care setting—is often unknown. (2) Training of prognostication is often ‘implicit’—if it happens at all. This is in contrast to other fields outside clinical medicine where systematic training approaches to increase prognostic performance and accuracy do exist. For patients in routine oncological practice (and not in the palliative care setting)—no data on improving prognostication via a training programme exist to our knowledge. In this paper, we, therefore, evaluate not only the prognostic performance of residents and experienced oncologists in the acute care setting but also how far a structured training programme might have a positive impact on prognostication. Time period and general conditions The prospective study was conducted at the clinic for oncology and haematology at the Hospital Barmherzige Brüder in Regensburg, Germany. It consisted of 3 different phases with a duration of 1 month each, taking place in the 3 months of April, June and September of 2019. Data collection was performed as part of routine care on the oncology wards and in the oncology outpatient clinic. All physicians agreed to participate in the study and gave their informed consent. Since the study did not involve any intervention affecting the patient and the standard of care was not affected no informed consent was required for patients. The current study was carried out in accordance with the Declaration of Helsinki. Participants The participants of the study consisted of the whole medical team (= physicians) of the department of oncology and haematology (n=21), who (individually) generated the prognostic estimates on the patients (described below). This ‘whole group’ was subdivided into ‘residents’ (physicians in training, less than 1 year of training in clinical oncology = ‘inexperienced’) and into ‘oncologists’ (board-certified oncologists with a minimum experience in clinical oncology of 13+years = ‘experienced’). Procedure General procedure Patients who were discharged from the department of oncology and haematology during the 3 phases were included. All patients were required to have a malignant disease but could be in any stage of their disease. After each patient’s discharge, the resident physician who last cared for the patient and the supervising senior physician and/or chief physician independently evaluated the patient’s prognosis based on knowing the patient personally and based on the information from the discharge letter. To determine both a qualitative value and a quantitative value for the prognostic accuracy, questionnaires were used to elicit the physicians' responses to the following two questions: ‘Would you be surprised, if your patient had died in 6 months from today?’ (‘surprise question’ (SQ)). ‘If you were faced with a large group of patients with a very similar constellation of age, disease, comorbidities, prior therapies as in the patient whose discharge letter you have just completed, what median survival would you expect in this group?’ (‘clinician prediction of survival’ (CPS)). Sample size calculation/considerations We determined that each of the 3 phases was to consist of 4 consecutive weeks. Our calculation was that within this period of 4 weeks approximately 300 patients would leave the oncology department and could, therefore, be evaluated for prognosis by their treating physicians. Since the resident as well as one or two senior oncology consultants gave their prognosis estimates (PE) independently on the same patient a total number of approximately 600–900 prognoses per phase were expected (and de facto achieved—phase 1=925, phase 2=712, phase 3=790). We assumed that these rather high numbers (as compared with those within the literature of clinical prognostication) were needed to give an accurate descriptive picture of the status quo on ‘untrained’ clinical prognostication (in phase 1) and of prognostic performance after training (phases 2 and 3). Data collection Every physician’s estimate and every hospitalisation of a patient were considered. When there were multiple estimates for a patient (eg, by resident, senior physician and chief physician), all three estimates were recorded as distinct data points for this respective patient for this hospital stay. Patients, who received a PE on more than one occasion, were included in the analysis for each of their PE. For analysis all physicians together formed the ‘whole group’. In addition, according to the clinical experience, we defined the subgroups ‘residents’ and ‘oncologists’ (board-certified specialists in haematology and oncology). To avoid the possibility of non-representative values caused by outliers, only physicians who provided ≥10 PE per phase were included in the final analysis. Physicians, who did not work in the oncology wards in every phase (eg, due to rotation in residency training), were included only in the involved phase(s) if they had completed ≥10 PE. Real survival data (date of death or last recorded life status) were provided by the Tumor Center Regensburg in March 2023, with a follow-up of approximately 4 years. PE for patients, who did not have an oncological disease or who were not registered at the Tumor Center Regensburg, were not included in the analysis (because follow-up could not be ensured). Phase 1 versus phases 2 and 3/interventions A flow chart of the 3 phases with all participating physicians and patients is provided in . In phase 1, the PE was performed with no prior training and was thus based mainly on prior clinical experience and ‘gut feeling’ (‘thinking fast’). In contrast, during phase 2 ‘thinking slow’ (according to Kahneman ) was encouraged—that is to consciously take time for the PE. To enable that, a prognosis training was provided for the physicians prior to the start of the second phase and repeated before phase 3 (see ): 10.1136/bmjopen-2023-081661.supp1 Supplementary data 10.1136/bmjopen-2023-081661.supp2 Supplementary data The prognosis training consisted of a brochure designed to support the physicians with the aid of epidemiological data on the 21 most common oncological diseases of the clinic. For each disease, general information was provided on the median survival time across all groups as well as information for different patient groups (eg, according to patients’ age, stage or mutation status). Corresponding Kaplan-Meier curves were provided with information on absolute and relative survival for these subgroups. Sources of the information were derived mostly from the Munich Tumor Registry, other sources were, for example, the website UpToDate or clinical or epidemiological studies. These epidemiological data were intended to be used as numerical ‘anchors’ during the (conscious) process of prognostication. The brochure was introduced to the participating physicians within a teaching lecture during which a general approach to prognostication was provided (see ). In brief, this approach assumes that prognostic accuracy can be improved by first taking an ‘outside view’ on a situation—for example, by using anchor numbers derived from the brochure (‘actuarial’ approach). In subsequent steps, a more ‘personalised’ and multifaceted ‘inside view’ may follow in which relevant details of the individual patient can be used to adjust the prognostic estimate—using the principle of ‘Fermisation’ (referring to Enrico Fermi—see ). 10.1136/bmjopen-2023-081661.supp3 Supplementary data 10.1136/bmjopen-2023-081661.supp4 Supplementary data 10.1136/bmjopen-2023-081661.supp5 Supplementary data In addition, the SQ of the questionnaire was reworded in phase 2 to exclude the subjective factor of the term ‘surprise’. (‘ Do you think it is more likely (> 50%) that your patient will be alive in 6 months than that he will have died by that time?’ ) Prior to the start of phase 3, prognostic tools were provided for the most common oncological diseases of the clinic, if they were available either as a website or as an app (see ). In this phase, physicians had to provide two separate PE on the same patient—the first without knowledge of the PE made by the respective (electronic) prognostic tool and the second while being aware of the result. In phase 3, the questionnaire was complemented by another version of the SQ—that is, the ‘probabilistic SQ’: In addition to the answer option ‘yes/no’, the SQ was now to be answered probabilistically (‘ Please indicate the probability (in 10% increments from 0% to 100%) that your patient will be alive in 6 months.’ ). Statistical evaluations CPS was evaluated using the three categories ‘accurate’, ‘neutral’ and ‘inaccurate’. When the real OS time was set as 100%, we set the definition of ‘accurate’ estimates as those that ranged from 67% to 133% of this value. ‘Neutral’ estimates were in the range of 34%–67% and 133%–166% and ‘inaccurate’ estimates were beyond these limits, that is, <34% and >166% of the actual observed survival time. The SQ was considered ‘correct’ if the answer was ‘yes’ and the patient was alive at 6 months from the time of the PE or the answer was ‘no’ and the patient had died within 6 months. In the probabilistic form of the SQ, we defined estimates as ‘correct’ if the patient had died at 6 months when given answers ranging from 0% to 40% or lived at 6 months when given answers ranging from 50% to 100%. The rate of correctly answered SQ and the rate of ‘accurate’ CPS PE of the ‘whole group’ were collected as the primary endpoints of this study. As a secondary endpoint, we aimed to describe the potential impact of a ‘prognosis training’ on the prognostic performance of physicians. To analyse this intervention, performance in phase 1 (‘untrained’) was compared with the performance after training (both phases 2 and 3 after training—‘trained’) and tested for statistical significance by using a χ 2 test. These analyses were performed for the ‘whole group’ and for the two subgroups ‘residents’ and ‘oncologists’ for each of the 3 phases. In addition, ‘residents’ and ‘oncologists’ were tested against each other for differences in prognostic performance in each phase. For the probabilistic form of the SQ, we tested how far the estimated survival likelihood correlates with the actual survival (χ 2 test for linear trend). The basis of this analysis is the groups of patients who had in common their respective estimated survival likelihoods—that is, ‘0% group’, ‘10% group’, ‘20% group’, etc up to the ‘100% group’. For example, the ‘100% group’ comprises all patients, which had been given a survival likelihood of 100% at 6 months by their respective physicians. In this probabilistic setting we also calculated the Brier score for the overall group and the subgroups (oncologists and residents). The programme Microsoft Excel was used for descriptive statistics of the 3 phases as well as for documentation, analysis and creation of diagrams. The presentation of Kaplan-Meier curves was realised with the statistical program IBM SPSS Statistics 29.0. Patient and public involvement None. The prospective study was conducted at the clinic for oncology and haematology at the Hospital Barmherzige Brüder in Regensburg, Germany. It consisted of 3 different phases with a duration of 1 month each, taking place in the 3 months of April, June and September of 2019. Data collection was performed as part of routine care on the oncology wards and in the oncology outpatient clinic. All physicians agreed to participate in the study and gave their informed consent. Since the study did not involve any intervention affecting the patient and the standard of care was not affected no informed consent was required for patients. The current study was carried out in accordance with the Declaration of Helsinki. The participants of the study consisted of the whole medical team (= physicians) of the department of oncology and haematology (n=21), who (individually) generated the prognostic estimates on the patients (described below). This ‘whole group’ was subdivided into ‘residents’ (physicians in training, less than 1 year of training in clinical oncology = ‘inexperienced’) and into ‘oncologists’ (board-certified oncologists with a minimum experience in clinical oncology of 13+years = ‘experienced’). General procedure Patients who were discharged from the department of oncology and haematology during the 3 phases were included. All patients were required to have a malignant disease but could be in any stage of their disease. After each patient’s discharge, the resident physician who last cared for the patient and the supervising senior physician and/or chief physician independently evaluated the patient’s prognosis based on knowing the patient personally and based on the information from the discharge letter. To determine both a qualitative value and a quantitative value for the prognostic accuracy, questionnaires were used to elicit the physicians' responses to the following two questions: ‘Would you be surprised, if your patient had died in 6 months from today?’ (‘surprise question’ (SQ)). ‘If you were faced with a large group of patients with a very similar constellation of age, disease, comorbidities, prior therapies as in the patient whose discharge letter you have just completed, what median survival would you expect in this group?’ (‘clinician prediction of survival’ (CPS)). Sample size calculation/considerations We determined that each of the 3 phases was to consist of 4 consecutive weeks. Our calculation was that within this period of 4 weeks approximately 300 patients would leave the oncology department and could, therefore, be evaluated for prognosis by their treating physicians. Since the resident as well as one or two senior oncology consultants gave their prognosis estimates (PE) independently on the same patient a total number of approximately 600–900 prognoses per phase were expected (and de facto achieved—phase 1=925, phase 2=712, phase 3=790). We assumed that these rather high numbers (as compared with those within the literature of clinical prognostication) were needed to give an accurate descriptive picture of the status quo on ‘untrained’ clinical prognostication (in phase 1) and of prognostic performance after training (phases 2 and 3). Patients who were discharged from the department of oncology and haematology during the 3 phases were included. All patients were required to have a malignant disease but could be in any stage of their disease. After each patient’s discharge, the resident physician who last cared for the patient and the supervising senior physician and/or chief physician independently evaluated the patient’s prognosis based on knowing the patient personally and based on the information from the discharge letter. To determine both a qualitative value and a quantitative value for the prognostic accuracy, questionnaires were used to elicit the physicians' responses to the following two questions: ‘Would you be surprised, if your patient had died in 6 months from today?’ (‘surprise question’ (SQ)). ‘If you were faced with a large group of patients with a very similar constellation of age, disease, comorbidities, prior therapies as in the patient whose discharge letter you have just completed, what median survival would you expect in this group?’ (‘clinician prediction of survival’ (CPS)). We determined that each of the 3 phases was to consist of 4 consecutive weeks. Our calculation was that within this period of 4 weeks approximately 300 patients would leave the oncology department and could, therefore, be evaluated for prognosis by their treating physicians. Since the resident as well as one or two senior oncology consultants gave their prognosis estimates (PE) independently on the same patient a total number of approximately 600–900 prognoses per phase were expected (and de facto achieved—phase 1=925, phase 2=712, phase 3=790). We assumed that these rather high numbers (as compared with those within the literature of clinical prognostication) were needed to give an accurate descriptive picture of the status quo on ‘untrained’ clinical prognostication (in phase 1) and of prognostic performance after training (phases 2 and 3). Every physician’s estimate and every hospitalisation of a patient were considered. When there were multiple estimates for a patient (eg, by resident, senior physician and chief physician), all three estimates were recorded as distinct data points for this respective patient for this hospital stay. Patients, who received a PE on more than one occasion, were included in the analysis for each of their PE. For analysis all physicians together formed the ‘whole group’. In addition, according to the clinical experience, we defined the subgroups ‘residents’ and ‘oncologists’ (board-certified specialists in haematology and oncology). To avoid the possibility of non-representative values caused by outliers, only physicians who provided ≥10 PE per phase were included in the final analysis. Physicians, who did not work in the oncology wards in every phase (eg, due to rotation in residency training), were included only in the involved phase(s) if they had completed ≥10 PE. Real survival data (date of death or last recorded life status) were provided by the Tumor Center Regensburg in March 2023, with a follow-up of approximately 4 years. PE for patients, who did not have an oncological disease or who were not registered at the Tumor Center Regensburg, were not included in the analysis (because follow-up could not be ensured). A flow chart of the 3 phases with all participating physicians and patients is provided in . In phase 1, the PE was performed with no prior training and was thus based mainly on prior clinical experience and ‘gut feeling’ (‘thinking fast’). In contrast, during phase 2 ‘thinking slow’ (according to Kahneman ) was encouraged—that is to consciously take time for the PE. To enable that, a prognosis training was provided for the physicians prior to the start of the second phase and repeated before phase 3 (see ): 10.1136/bmjopen-2023-081661.supp1 Supplementary data 10.1136/bmjopen-2023-081661.supp2 Supplementary data The prognosis training consisted of a brochure designed to support the physicians with the aid of epidemiological data on the 21 most common oncological diseases of the clinic. For each disease, general information was provided on the median survival time across all groups as well as information for different patient groups (eg, according to patients’ age, stage or mutation status). Corresponding Kaplan-Meier curves were provided with information on absolute and relative survival for these subgroups. Sources of the information were derived mostly from the Munich Tumor Registry, other sources were, for example, the website UpToDate or clinical or epidemiological studies. These epidemiological data were intended to be used as numerical ‘anchors’ during the (conscious) process of prognostication. The brochure was introduced to the participating physicians within a teaching lecture during which a general approach to prognostication was provided (see ). In brief, this approach assumes that prognostic accuracy can be improved by first taking an ‘outside view’ on a situation—for example, by using anchor numbers derived from the brochure (‘actuarial’ approach). In subsequent steps, a more ‘personalised’ and multifaceted ‘inside view’ may follow in which relevant details of the individual patient can be used to adjust the prognostic estimate—using the principle of ‘Fermisation’ (referring to Enrico Fermi—see ). 10.1136/bmjopen-2023-081661.supp3 Supplementary data 10.1136/bmjopen-2023-081661.supp4 Supplementary data 10.1136/bmjopen-2023-081661.supp5 Supplementary data In addition, the SQ of the questionnaire was reworded in phase 2 to exclude the subjective factor of the term ‘surprise’. (‘ Do you think it is more likely (> 50%) that your patient will be alive in 6 months than that he will have died by that time?’ ) Prior to the start of phase 3, prognostic tools were provided for the most common oncological diseases of the clinic, if they were available either as a website or as an app (see ). In this phase, physicians had to provide two separate PE on the same patient—the first without knowledge of the PE made by the respective (electronic) prognostic tool and the second while being aware of the result. In phase 3, the questionnaire was complemented by another version of the SQ—that is, the ‘probabilistic SQ’: In addition to the answer option ‘yes/no’, the SQ was now to be answered probabilistically (‘ Please indicate the probability (in 10% increments from 0% to 100%) that your patient will be alive in 6 months.’ ). CPS was evaluated using the three categories ‘accurate’, ‘neutral’ and ‘inaccurate’. When the real OS time was set as 100%, we set the definition of ‘accurate’ estimates as those that ranged from 67% to 133% of this value. ‘Neutral’ estimates were in the range of 34%–67% and 133%–166% and ‘inaccurate’ estimates were beyond these limits, that is, <34% and >166% of the actual observed survival time. The SQ was considered ‘correct’ if the answer was ‘yes’ and the patient was alive at 6 months from the time of the PE or the answer was ‘no’ and the patient had died within 6 months. In the probabilistic form of the SQ, we defined estimates as ‘correct’ if the patient had died at 6 months when given answers ranging from 0% to 40% or lived at 6 months when given answers ranging from 50% to 100%. The rate of correctly answered SQ and the rate of ‘accurate’ CPS PE of the ‘whole group’ were collected as the primary endpoints of this study. As a secondary endpoint, we aimed to describe the potential impact of a ‘prognosis training’ on the prognostic performance of physicians. To analyse this intervention, performance in phase 1 (‘untrained’) was compared with the performance after training (both phases 2 and 3 after training—‘trained’) and tested for statistical significance by using a χ 2 test. These analyses were performed for the ‘whole group’ and for the two subgroups ‘residents’ and ‘oncologists’ for each of the 3 phases. In addition, ‘residents’ and ‘oncologists’ were tested against each other for differences in prognostic performance in each phase. For the probabilistic form of the SQ, we tested how far the estimated survival likelihood correlates with the actual survival (χ 2 test for linear trend). The basis of this analysis is the groups of patients who had in common their respective estimated survival likelihoods—that is, ‘0% group’, ‘10% group’, ‘20% group’, etc up to the ‘100% group’. For example, the ‘100% group’ comprises all patients, which had been given a survival likelihood of 100% at 6 months by their respective physicians. In this probabilistic setting we also calculated the Brier score for the overall group and the subgroups (oncologists and residents). The programme Microsoft Excel was used for descriptive statistics of the 3 phases as well as for documentation, analysis and creation of diagrams. The presentation of Kaplan-Meier curves was realised with the statistical program IBM SPSS Statistics 29.0. None. Test persons/assessing physicians 21 physicians participated by providing a total of 2486 prognostic estimates on 748 individual patients. 18 physicians could be included, 3 were excluded because they provided <10 PE. The whole group was subdivided into two groups. The ‘residents’ group was formed by rather inexperienced physicians and consisted of 11 different residents. The ‘oncologists’ group consisted of seven experienced board-qualified haematologists-oncologists. The professional experience of the residents averaged from entry level to 6 years of experience in internal medicine with a maximum experience in clinical oncology of 1 year. The oncologists had an experience in clinical oncology ranging from 13 to 34 years. Patients A total of 748 patients with a wide range of diseases were assessed—12 of which were excluded from the analysis because they either had no oncologic disease or no status on survival was available at the time of follow-up, so the final analysis consisted of 736 patients. The patients’ characteristics can be seen in . Of 736 patients, 410 (55.7%) had died by the time of the last follow-up in March 2023. The median overall survival was 2.5 years (see ). Clinician prediction of survival Comparison of phase 1 with phase 2 and 3 For CPS—that is, ‘long-term’ prognostication—1506 PE were available for evaluation. Overall, only 25.9% of them were ‘accurate’ according to the definition (±33% of actual observed survival). Accuracy varied over real survival as shown in —with the best results between 8 and 10 months. Over the 3 phases, there was a numerical trend towards a slightly better accuracy from 23.6% to 28.0%—which was also observed within the two subgroups (residents and oncologists)—which however lacked statistical significance in all subgroups (see ). When Kaplan-Meier curves of real and estimated overall survival were plotted against each other, the curves crossed at about 9 months for the whole group of patients. There is overestimation of OS in the ‘immediate’ future whereas prognosis in the more distant future tends to be underestimated. This finding is also backed by the dot plot which depicts the estimated survival over real survival. Corresponding analyses of the 3 phases can be found in . Up to a real survival of 9 months, we observed a tendency towards overestimation (671/853 (78.7%) above the diagonal of ‘perfect prediction’). After 9 months, there is a tendency towards underestimation (451/653 (69.0%)) (see ). Over time (from phases 1 to 3) the Kaplan-Meier curves tended to cross earlier—potentially indicating more conservative estimates after training. When comparing differences between the ‘estimated’ CPS and the ‘real’ OS curves there is a visual trend towards closer curves after training (phase 2 and 3) as compared with the ‘untrained’ phase 1 in the ‘immediate’ future (see ). Overestimation of survival—which was very evident in phase 1—seemed to be less pronounced after training (no test for significance). Surprise question Option ‘yes’/‘no’ In the evaluation of the SQ with the answer option ‘yes’/’no’, a total of 2427 answers were included. 76.8% of them were correct and 23.2% incorrect. Physicians achieved 72.6% accuracy in phase 1, 74.0% in phase 2 and 84.3% in phase 3. There was a highly significant (p<0.001) improvement in accuracy over time during the 3 phases for the whole group. Residents had an accuracy of 60.7% in phase 1, 63.2% in phase 2 and 81.1% in phase 3, which demonstrated a significant (p<0.01) improvement. The accuracy of the oncologists also increased significantly (p<0.05) from 75.7% in phase 1 to 77.0% in phase 2 to 85.6% in phase 3. In total, oncologists achieved a significantly (p<0.005) higher accuracy (79.1%) as compared with residents (69.5%). However, this was only significant in phase 1 and in phase 2, but not in phase 3 (see ). Probabilistic SQ 800 answers were available for the evaluation of the probabilistic SQ in phase 3 of which 83.1% were correct. The probabilistic SQ showed a high accuracy in the extremes of the spectrum, the answer ‘0%’ and ‘100%’ even showed an accuracy of 100.0% with 27/27 correct SQ (‘0%’) and 53/53 correct SQ (‘100%’). The closer the answer approached the mid-range values, the lower was the accuracy. These mid-range percentages (answers 40%–60%) only had a low accuracy of 54.5% which was significantly (p<0.0001) worse compared with the ‘extreme’ parts of the spectrum (0%–30% and 70%–100%) with a correct estimate in 89.9% of cases. This expression of mental indecisiveness is—of course—the equivalent of a coin flip. A detailed overview of the probabilistic SQ is shown in . The values for the Brier score were 0.193 for the whole group, 0.191 for the oncologists and 0.199 for the residents. An interesting finding for groups of patients was that a predicted higher likelihood of survival (eg, ‘group 90%’ vs ‘group 60%’) also correlated with a higher probability of real survival in these groups in a similar magnitude (p<0.0001) (see ). Therefore, the ‘coin flip’ estimate is not ‘worthless’ but carries information because the ‘group 50%’ comprises a group of patients of which roughly half will have died after 6 months. Accuracy of prognostic tools (SQ, CPS) During phase 3, established (electronic) prognostic tools were available and suitable for only 34.6% of patients and their respective clinical situation. When used alone for the SQ the accuracy of these prognostic tools was only 70.2%, which was significantly worse (p<0.0005) than the performance of the whole group (84.3%) and of the oncologists (85.6%). It was also numerically—if not statistically significant—worse than the performance of residents (81.1%). When prognostic tools were used for CPS estimates, an ‘accurate’ result was only achieved in 16.9% which was less than the rate of the whole group of physicians 25.9% (and of all subgroups). In a subgroup (n=443 patients) physicians made PE on CPS first without and then with the respective prognostic tool—however, accuracy was not significantly increased (28.0% to 29.4%). 21 physicians participated by providing a total of 2486 prognostic estimates on 748 individual patients. 18 physicians could be included, 3 were excluded because they provided <10 PE. The whole group was subdivided into two groups. The ‘residents’ group was formed by rather inexperienced physicians and consisted of 11 different residents. The ‘oncologists’ group consisted of seven experienced board-qualified haematologists-oncologists. The professional experience of the residents averaged from entry level to 6 years of experience in internal medicine with a maximum experience in clinical oncology of 1 year. The oncologists had an experience in clinical oncology ranging from 13 to 34 years. A total of 748 patients with a wide range of diseases were assessed—12 of which were excluded from the analysis because they either had no oncologic disease or no status on survival was available at the time of follow-up, so the final analysis consisted of 736 patients. The patients’ characteristics can be seen in . Of 736 patients, 410 (55.7%) had died by the time of the last follow-up in March 2023. The median overall survival was 2.5 years (see ). Comparison of phase 1 with phase 2 and 3 For CPS—that is, ‘long-term’ prognostication—1506 PE were available for evaluation. Overall, only 25.9% of them were ‘accurate’ according to the definition (±33% of actual observed survival). Accuracy varied over real survival as shown in —with the best results between 8 and 10 months. Over the 3 phases, there was a numerical trend towards a slightly better accuracy from 23.6% to 28.0%—which was also observed within the two subgroups (residents and oncologists)—which however lacked statistical significance in all subgroups (see ). When Kaplan-Meier curves of real and estimated overall survival were plotted against each other, the curves crossed at about 9 months for the whole group of patients. There is overestimation of OS in the ‘immediate’ future whereas prognosis in the more distant future tends to be underestimated. This finding is also backed by the dot plot which depicts the estimated survival over real survival. Corresponding analyses of the 3 phases can be found in . Up to a real survival of 9 months, we observed a tendency towards overestimation (671/853 (78.7%) above the diagonal of ‘perfect prediction’). After 9 months, there is a tendency towards underestimation (451/653 (69.0%)) (see ). Over time (from phases 1 to 3) the Kaplan-Meier curves tended to cross earlier—potentially indicating more conservative estimates after training. When comparing differences between the ‘estimated’ CPS and the ‘real’ OS curves there is a visual trend towards closer curves after training (phase 2 and 3) as compared with the ‘untrained’ phase 1 in the ‘immediate’ future (see ). Overestimation of survival—which was very evident in phase 1—seemed to be less pronounced after training (no test for significance). For CPS—that is, ‘long-term’ prognostication—1506 PE were available for evaluation. Overall, only 25.9% of them were ‘accurate’ according to the definition (±33% of actual observed survival). Accuracy varied over real survival as shown in —with the best results between 8 and 10 months. Over the 3 phases, there was a numerical trend towards a slightly better accuracy from 23.6% to 28.0%—which was also observed within the two subgroups (residents and oncologists)—which however lacked statistical significance in all subgroups (see ). When Kaplan-Meier curves of real and estimated overall survival were plotted against each other, the curves crossed at about 9 months for the whole group of patients. There is overestimation of OS in the ‘immediate’ future whereas prognosis in the more distant future tends to be underestimated. This finding is also backed by the dot plot which depicts the estimated survival over real survival. Corresponding analyses of the 3 phases can be found in . Up to a real survival of 9 months, we observed a tendency towards overestimation (671/853 (78.7%) above the diagonal of ‘perfect prediction’). After 9 months, there is a tendency towards underestimation (451/653 (69.0%)) (see ). Over time (from phases 1 to 3) the Kaplan-Meier curves tended to cross earlier—potentially indicating more conservative estimates after training. When comparing differences between the ‘estimated’ CPS and the ‘real’ OS curves there is a visual trend towards closer curves after training (phase 2 and 3) as compared with the ‘untrained’ phase 1 in the ‘immediate’ future (see ). Overestimation of survival—which was very evident in phase 1—seemed to be less pronounced after training (no test for significance). Option ‘yes’/‘no’ In the evaluation of the SQ with the answer option ‘yes’/’no’, a total of 2427 answers were included. 76.8% of them were correct and 23.2% incorrect. Physicians achieved 72.6% accuracy in phase 1, 74.0% in phase 2 and 84.3% in phase 3. There was a highly significant (p<0.001) improvement in accuracy over time during the 3 phases for the whole group. Residents had an accuracy of 60.7% in phase 1, 63.2% in phase 2 and 81.1% in phase 3, which demonstrated a significant (p<0.01) improvement. The accuracy of the oncologists also increased significantly (p<0.05) from 75.7% in phase 1 to 77.0% in phase 2 to 85.6% in phase 3. In total, oncologists achieved a significantly (p<0.005) higher accuracy (79.1%) as compared with residents (69.5%). However, this was only significant in phase 1 and in phase 2, but not in phase 3 (see ). Probabilistic SQ 800 answers were available for the evaluation of the probabilistic SQ in phase 3 of which 83.1% were correct. The probabilistic SQ showed a high accuracy in the extremes of the spectrum, the answer ‘0%’ and ‘100%’ even showed an accuracy of 100.0% with 27/27 correct SQ (‘0%’) and 53/53 correct SQ (‘100%’). The closer the answer approached the mid-range values, the lower was the accuracy. These mid-range percentages (answers 40%–60%) only had a low accuracy of 54.5% which was significantly (p<0.0001) worse compared with the ‘extreme’ parts of the spectrum (0%–30% and 70%–100%) with a correct estimate in 89.9% of cases. This expression of mental indecisiveness is—of course—the equivalent of a coin flip. A detailed overview of the probabilistic SQ is shown in . The values for the Brier score were 0.193 for the whole group, 0.191 for the oncologists and 0.199 for the residents. An interesting finding for groups of patients was that a predicted higher likelihood of survival (eg, ‘group 90%’ vs ‘group 60%’) also correlated with a higher probability of real survival in these groups in a similar magnitude (p<0.0001) (see ). Therefore, the ‘coin flip’ estimate is not ‘worthless’ but carries information because the ‘group 50%’ comprises a group of patients of which roughly half will have died after 6 months. Accuracy of prognostic tools (SQ, CPS) During phase 3, established (electronic) prognostic tools were available and suitable for only 34.6% of patients and their respective clinical situation. When used alone for the SQ the accuracy of these prognostic tools was only 70.2%, which was significantly worse (p<0.0005) than the performance of the whole group (84.3%) and of the oncologists (85.6%). It was also numerically—if not statistically significant—worse than the performance of residents (81.1%). When prognostic tools were used for CPS estimates, an ‘accurate’ result was only achieved in 16.9% which was less than the rate of the whole group of physicians 25.9% (and of all subgroups). In a subgroup (n=443 patients) physicians made PE on CPS first without and then with the respective prognostic tool—however, accuracy was not significantly increased (28.0% to 29.4%). In the evaluation of the SQ with the answer option ‘yes’/’no’, a total of 2427 answers were included. 76.8% of them were correct and 23.2% incorrect. Physicians achieved 72.6% accuracy in phase 1, 74.0% in phase 2 and 84.3% in phase 3. There was a highly significant (p<0.001) improvement in accuracy over time during the 3 phases for the whole group. Residents had an accuracy of 60.7% in phase 1, 63.2% in phase 2 and 81.1% in phase 3, which demonstrated a significant (p<0.01) improvement. The accuracy of the oncologists also increased significantly (p<0.05) from 75.7% in phase 1 to 77.0% in phase 2 to 85.6% in phase 3. In total, oncologists achieved a significantly (p<0.005) higher accuracy (79.1%) as compared with residents (69.5%). However, this was only significant in phase 1 and in phase 2, but not in phase 3 (see ). 800 answers were available for the evaluation of the probabilistic SQ in phase 3 of which 83.1% were correct. The probabilistic SQ showed a high accuracy in the extremes of the spectrum, the answer ‘0%’ and ‘100%’ even showed an accuracy of 100.0% with 27/27 correct SQ (‘0%’) and 53/53 correct SQ (‘100%’). The closer the answer approached the mid-range values, the lower was the accuracy. These mid-range percentages (answers 40%–60%) only had a low accuracy of 54.5% which was significantly (p<0.0001) worse compared with the ‘extreme’ parts of the spectrum (0%–30% and 70%–100%) with a correct estimate in 89.9% of cases. This expression of mental indecisiveness is—of course—the equivalent of a coin flip. A detailed overview of the probabilistic SQ is shown in . The values for the Brier score were 0.193 for the whole group, 0.191 for the oncologists and 0.199 for the residents. An interesting finding for groups of patients was that a predicted higher likelihood of survival (eg, ‘group 90%’ vs ‘group 60%’) also correlated with a higher probability of real survival in these groups in a similar magnitude (p<0.0001) (see ). Therefore, the ‘coin flip’ estimate is not ‘worthless’ but carries information because the ‘group 50%’ comprises a group of patients of which roughly half will have died after 6 months. During phase 3, established (electronic) prognostic tools were available and suitable for only 34.6% of patients and their respective clinical situation. When used alone for the SQ the accuracy of these prognostic tools was only 70.2%, which was significantly worse (p<0.0005) than the performance of the whole group (84.3%) and of the oncologists (85.6%). It was also numerically—if not statistically significant—worse than the performance of residents (81.1%). When prognostic tools were used for CPS estimates, an ‘accurate’ result was only achieved in 16.9% which was less than the rate of the whole group of physicians 25.9% (and of all subgroups). In a subgroup (n=443 patients) physicians made PE on CPS first without and then with the respective prognostic tool—however, accuracy was not significantly increased (28.0% to 29.4%). This prospective study examined physicians’ accuracy in prognostic estimates (PE) for oncological patients of all stages as well as a potential training effect. To do this, nearly 2500 PE on over 700 oncological patients were performed. We analysed three types of PE in our study: the (standard) SQ, the probabilistic SQ and CPS. The patients in our study had a median survival of 2.5 years, which is substantially longer than in most prior studies on prognostication in patients with cancer which have been performed on patients in very advanced stages or in a palliative setting. Main findings In our study, the whole group of participating physicians achieved an accuracy of 76.8% when using the SQ. This result is fairly comparable to the summarised accuracy of 78.6% for oncological patients calculated by a systematic review in 2017. It should be noted, however, that very different time intervals have been investigated so far when answering the SQ for oncological patients and that the patient cohorts differed considerably (see ). Quite evidently, the time interval chosen for the SQ has an impact on accuracy, since the longer the time interval the less certain the estimates will be and the accuracy will ‘suffer’. The selection of a time interval appropriate for the respective patient population is therefore essential. Interestingly, the prognostic accuracy of the whole group of participating physicians increased after training from an initial 72.6% to 84.3% in the last phase. This effect was especially pronounced in residents who started from a lower level (60.7% in phase 1) and then achieved 81.1% in the final phase (+20.4%). The experienced oncologists started from a significantly higher level of accuracy (75.7%) but also benefited from training (85.6% in the final phase)—however, to a less pronounced degree in absolute terms (+9.9%). Our data indicate that experienced physicians are better prognosticators than early career physicians which has so far been a controversial issue within the literature. Moreover, a prognosis-oriented training programme has a potential to improve the prognostic accuracy of both residents as well as of experienced oncologists. Such a focused training programme has been repeatedly recommended in the literature. While some guidelines and training concepts exist in the palliative setting. This is—to our knowledge—the first study to have implemented and systematically tested such an approach in a large group of oncological patients with a median survival over 2 years. (Our intention was to develop and test a short training programme and find out whether there is an effect. Our aim was not to specifically identify and isolate each reasoning principle within the programme and determine its incremental effectiveness.) Wider implications Prognostication using the SQ for a 6-month period can be considered ‘short-term prognostication’. In this situation, physicians’ accuracy was substantially better than chance and it was trainable. This finding contrasted with the situation of ‘long-term prognostication’ in which physicians were required to estimate overall survival (which most of the times exceeded 6 months—as demonstrated by a median OS of 2.5 years). The CPS was mostly ‘off-target’ with only 25.9% ‘accurate’ PE. Our results are in accordance with smaller studies which used the same definition for accuracy and demonstrated accuracy values between 20% and 35%. As mentioned above, previous studies assessed very advanced cancer patients with an OS of only a few months so that survival data were available for nearly all patients. Since only 60% of our patients had died at the time of analysis further follow-up of our patients might change the numerical value for accuracy of CPS. CPS and SQ at 6 months are different metrics and, therefore, the absolute values of these parameters cannot be directly compared. However, CPS can be ‘converted’ into the format of an SQ by assuming the following: The groups ‘low inaccurate’, ‘low moderate’ and ‘accurate’ will not have been ‘surprised’ that the patient has died at his/her specific time of death. To illustrate this, imagine the following thought experiment: If we (retrospectively) know that a patient has died 450 days after discharge from our care we can pose this constellation (discharge letter from that past discharge) to a group of physicians in the form of an individual SQ: ‘Would you be surprised if this patient had died within 450 days?’ A physician with a CPS of ‘200 days’ would not be ‘surprised’ as in our SQ definition and he would be ‘right’. In contrast, a physician stating a CPS of ‘700 days’ would be ‘surprised’ by the earlier demise of the patient and his estimate would have been ‘wrong’. This also illustrates that the SQ is the more ‘permissive’ metric because it considers estimates as correct (‘not surprised’) that are not in the ‘accurate’ category of the CPS definition—that is, ‘low inaccurate’ and ‘low moderate’. However—even when using the more permissive ‘individual SQ’ definition (=consider the sum of ‘low inaccurate’, ‘low moderate’ and ‘accurate’ as ‘not surprised’ and therefore as ‘correct’)—the value for the CPS for the whole group is only 53.8%—hardly better than chance. Interestingly, there is no (significant) trend towards improvement by training and also there seems to be no difference between residents and experienced oncologists. These findings can be used to estimate the ‘horizon’ of individual clinical prognostication—which is a controversial issue in the literature —in our study: The group of patients on which definite survival data were available (ie, had died at the time of analysis) and on which our CPS analysis was performed had a median (50%) survival of 9 months and 60% of patients had died within 1 year (data not shown). Assuming that prognostication over longer periods (than 1 year) will be even worse, we conclude that whereas individual prognostication within a time frame of 6 months by physicians is better than chance (and is trainable) this seems no longer to be the case when a time horizon of circa 1 year has been reached. It is relevant to differentiate between prognostication for an individual patient and prognostication for a group of patients. During phase 3, when we also analysed the ‘probabilistic SQ’. Its overall accuracy of 83.1% was nearly identical to that of the ‘standard SQ’ in that phase (84.3%) as expected. These are the numerical values that are relevant for the PE of individual patients. However, unlike the standard SQ, the probabilistic SQ also had a quantitative component (‘estimate the probability of survival at 6 months in 10% increments!’). This has two consequences: The forecasts in this format convey a sense of decisiveness and indecisiveness and thereby contain additional prognostic information/certitude: This is demonstrated by the high accuracy of the extreme ends of the spectrum (0%–30% = likely to die within 6 months; 70%–100% = likely to survive the next 6 months) which reached 89.9%. This was substantially higher than in the middle (‘indecisive’) part of the spectrum (40%–60%) where the accuracy was only 54.4%. This ‘indecisive’ accuracy of 54.4% in the mid-range is hardly better than chance and implies a ‘coin flip’ for the individual patient. However, this does not mean, that such a prognostic estimate contains no information at all and is basically worthless. In contrast, we could show that the collective prognostic estimates in 10% increments closely mirrored the de facto survival rates of groups at 6 months as demonstrated in . For example, the ‘group 50%’ comprises a group of patients (all considered by their treating physician to have a 50% likelihood of survival at 6 months) of which roughly half will in fact die within 6 months. Even though each physician performed discrete prognostic estimates on an individual patient the sum of all these forecasts nevertheless also contains prognostic meaning for groups. This ‘wisdom of the clinical crowd’ is also exemplified by the survival curve of our patient cohort which is fitted quite well (though not perfectly) by the multitude of forecasts done on these patients by the participating physicians. Another aspect that can be derived from these two Kaplan-Meier curves is that overestimation of prognosis tends to occur predominantly in the early phase (6–9 months) whereas underestimation tends to be predominant at the end of the first year of observation and thereafter. The two curves cross between 6 and 9 months after t 0 (= time of the prognostic estimate) with the crossing tending to be earlier in the later phases—which may have been due to more conservative forecasts after training. This finding is in accordance with Fairchild et al who also observed an overestimation for patients with a survival of less than 6 months and an underestimation for patients with a survival of more than 9 months. We have also visualised this finding in the dot plot and found a higher rate of overestimation for survival <9 months and a higher rate of underestimation for survival >9 months. Established prognostic tools were only available for a minority of patients in their specific clinical situation (34%) and—when used alone—performed worse (accuracy 70.2%) than physicians (84.3%) in phase 3. They also did not add to prognostic accuracy of physicians. This might have been due to the fact, that often these tools are optimised for only one specific situation (eg, primary diagnosis and start of curative therapy). In these situations, some of the tools worked very well. However, most tools were not or less helpful in the many other situations when a clinical encounter takes place (eg, 3 months into second-line therapy). Also most of the time these tools convey ‘collective’ prognoses and place a patient into a prognostic group—similar to our results of the probabilistic SQ. This can of course be helpful, but it is a more permissive/easier task than individual prognostication. Strengths and limitations This is the first study that presents a structured training programme for oncologists that has a positive impact on clinical prognostication. This prospective study included a large cohort of patients with different oncological and malignant haematological diseases at all disease stages. Prospective follow-up data of almost 4 years were collected, which allowed for the assessment of longer PE. The effect of training was not demonstrated by a randomised comparison but by comparing baseline (untrained) prognostication abilities with the performance after training in the same group of physicians. The study is a single-centre study—even though performed at a large teaching hospital. Conclusion A short and simple training programme was able to increase the prognostic accuracy of residents as well as experienced oncologists significantly. This programme conveyed the basic principles of prognostication, provided freely available epidemiological data and taught the use of simple algorithms. This teaching effect was apparent for prognostication of the intermediate future (up to 6 months) but is unlikely to be present beyond this time, especially for time horizons beyond 1 year. However, this does not come as a surprise because: ‘Medicine is a science of uncertainty and an art of probability’ (attributed to Sir William Osler). In our study, the whole group of participating physicians achieved an accuracy of 76.8% when using the SQ. This result is fairly comparable to the summarised accuracy of 78.6% for oncological patients calculated by a systematic review in 2017. It should be noted, however, that very different time intervals have been investigated so far when answering the SQ for oncological patients and that the patient cohorts differed considerably (see ). Quite evidently, the time interval chosen for the SQ has an impact on accuracy, since the longer the time interval the less certain the estimates will be and the accuracy will ‘suffer’. The selection of a time interval appropriate for the respective patient population is therefore essential. Interestingly, the prognostic accuracy of the whole group of participating physicians increased after training from an initial 72.6% to 84.3% in the last phase. This effect was especially pronounced in residents who started from a lower level (60.7% in phase 1) and then achieved 81.1% in the final phase (+20.4%). The experienced oncologists started from a significantly higher level of accuracy (75.7%) but also benefited from training (85.6% in the final phase)—however, to a less pronounced degree in absolute terms (+9.9%). Our data indicate that experienced physicians are better prognosticators than early career physicians which has so far been a controversial issue within the literature. Moreover, a prognosis-oriented training programme has a potential to improve the prognostic accuracy of both residents as well as of experienced oncologists. Such a focused training programme has been repeatedly recommended in the literature. While some guidelines and training concepts exist in the palliative setting. This is—to our knowledge—the first study to have implemented and systematically tested such an approach in a large group of oncological patients with a median survival over 2 years. (Our intention was to develop and test a short training programme and find out whether there is an effect. Our aim was not to specifically identify and isolate each reasoning principle within the programme and determine its incremental effectiveness.) Prognostication using the SQ for a 6-month period can be considered ‘short-term prognostication’. In this situation, physicians’ accuracy was substantially better than chance and it was trainable. This finding contrasted with the situation of ‘long-term prognostication’ in which physicians were required to estimate overall survival (which most of the times exceeded 6 months—as demonstrated by a median OS of 2.5 years). The CPS was mostly ‘off-target’ with only 25.9% ‘accurate’ PE. Our results are in accordance with smaller studies which used the same definition for accuracy and demonstrated accuracy values between 20% and 35%. As mentioned above, previous studies assessed very advanced cancer patients with an OS of only a few months so that survival data were available for nearly all patients. Since only 60% of our patients had died at the time of analysis further follow-up of our patients might change the numerical value for accuracy of CPS. CPS and SQ at 6 months are different metrics and, therefore, the absolute values of these parameters cannot be directly compared. However, CPS can be ‘converted’ into the format of an SQ by assuming the following: The groups ‘low inaccurate’, ‘low moderate’ and ‘accurate’ will not have been ‘surprised’ that the patient has died at his/her specific time of death. To illustrate this, imagine the following thought experiment: If we (retrospectively) know that a patient has died 450 days after discharge from our care we can pose this constellation (discharge letter from that past discharge) to a group of physicians in the form of an individual SQ: ‘Would you be surprised if this patient had died within 450 days?’ A physician with a CPS of ‘200 days’ would not be ‘surprised’ as in our SQ definition and he would be ‘right’. In contrast, a physician stating a CPS of ‘700 days’ would be ‘surprised’ by the earlier demise of the patient and his estimate would have been ‘wrong’. This also illustrates that the SQ is the more ‘permissive’ metric because it considers estimates as correct (‘not surprised’) that are not in the ‘accurate’ category of the CPS definition—that is, ‘low inaccurate’ and ‘low moderate’. However—even when using the more permissive ‘individual SQ’ definition (=consider the sum of ‘low inaccurate’, ‘low moderate’ and ‘accurate’ as ‘not surprised’ and therefore as ‘correct’)—the value for the CPS for the whole group is only 53.8%—hardly better than chance. Interestingly, there is no (significant) trend towards improvement by training and also there seems to be no difference between residents and experienced oncologists. These findings can be used to estimate the ‘horizon’ of individual clinical prognostication—which is a controversial issue in the literature —in our study: The group of patients on which definite survival data were available (ie, had died at the time of analysis) and on which our CPS analysis was performed had a median (50%) survival of 9 months and 60% of patients had died within 1 year (data not shown). Assuming that prognostication over longer periods (than 1 year) will be even worse, we conclude that whereas individual prognostication within a time frame of 6 months by physicians is better than chance (and is trainable) this seems no longer to be the case when a time horizon of circa 1 year has been reached. It is relevant to differentiate between prognostication for an individual patient and prognostication for a group of patients. During phase 3, when we also analysed the ‘probabilistic SQ’. Its overall accuracy of 83.1% was nearly identical to that of the ‘standard SQ’ in that phase (84.3%) as expected. These are the numerical values that are relevant for the PE of individual patients. However, unlike the standard SQ, the probabilistic SQ also had a quantitative component (‘estimate the probability of survival at 6 months in 10% increments!’). This has two consequences: The forecasts in this format convey a sense of decisiveness and indecisiveness and thereby contain additional prognostic information/certitude: This is demonstrated by the high accuracy of the extreme ends of the spectrum (0%–30% = likely to die within 6 months; 70%–100% = likely to survive the next 6 months) which reached 89.9%. This was substantially higher than in the middle (‘indecisive’) part of the spectrum (40%–60%) where the accuracy was only 54.4%. This ‘indecisive’ accuracy of 54.4% in the mid-range is hardly better than chance and implies a ‘coin flip’ for the individual patient. However, this does not mean, that such a prognostic estimate contains no information at all and is basically worthless. In contrast, we could show that the collective prognostic estimates in 10% increments closely mirrored the de facto survival rates of groups at 6 months as demonstrated in . For example, the ‘group 50%’ comprises a group of patients (all considered by their treating physician to have a 50% likelihood of survival at 6 months) of which roughly half will in fact die within 6 months. Even though each physician performed discrete prognostic estimates on an individual patient the sum of all these forecasts nevertheless also contains prognostic meaning for groups. This ‘wisdom of the clinical crowd’ is also exemplified by the survival curve of our patient cohort which is fitted quite well (though not perfectly) by the multitude of forecasts done on these patients by the participating physicians. Another aspect that can be derived from these two Kaplan-Meier curves is that overestimation of prognosis tends to occur predominantly in the early phase (6–9 months) whereas underestimation tends to be predominant at the end of the first year of observation and thereafter. The two curves cross between 6 and 9 months after t 0 (= time of the prognostic estimate) with the crossing tending to be earlier in the later phases—which may have been due to more conservative forecasts after training. This finding is in accordance with Fairchild et al who also observed an overestimation for patients with a survival of less than 6 months and an underestimation for patients with a survival of more than 9 months. We have also visualised this finding in the dot plot and found a higher rate of overestimation for survival <9 months and a higher rate of underestimation for survival >9 months. Established prognostic tools were only available for a minority of patients in their specific clinical situation (34%) and—when used alone—performed worse (accuracy 70.2%) than physicians (84.3%) in phase 3. They also did not add to prognostic accuracy of physicians. This might have been due to the fact, that often these tools are optimised for only one specific situation (eg, primary diagnosis and start of curative therapy). In these situations, some of the tools worked very well. However, most tools were not or less helpful in the many other situations when a clinical encounter takes place (eg, 3 months into second-line therapy). Also most of the time these tools convey ‘collective’ prognoses and place a patient into a prognostic group—similar to our results of the probabilistic SQ. This can of course be helpful, but it is a more permissive/easier task than individual prognostication. This is the first study that presents a structured training programme for oncologists that has a positive impact on clinical prognostication. This prospective study included a large cohort of patients with different oncological and malignant haematological diseases at all disease stages. Prospective follow-up data of almost 4 years were collected, which allowed for the assessment of longer PE. The effect of training was not demonstrated by a randomised comparison but by comparing baseline (untrained) prognostication abilities with the performance after training in the same group of physicians. The study is a single-centre study—even though performed at a large teaching hospital. A short and simple training programme was able to increase the prognostic accuracy of residents as well as experienced oncologists significantly. This programme conveyed the basic principles of prognostication, provided freely available epidemiological data and taught the use of simple algorithms. This teaching effect was apparent for prognostication of the intermediate future (up to 6 months) but is unlikely to be present beyond this time, especially for time horizons beyond 1 year. However, this does not come as a surprise because: ‘Medicine is a science of uncertainty and an art of probability’ (attributed to Sir William Osler). Reviewer comments Author's manuscript |
Intracranial Atherosclerotic Stenosis | 51f53e97-de59-4eb7-9562-77ffa888b013 | 11801852 | Surgical Procedures, Operative[mh] | Stroke is a leading cause of mortality and disability globally. Atherosclerosis in cerebral arteries is one of the major contributors of ischemic stroke, along with embolism originating from heart or small vessel disease. Atherosclerotic cerebral infarction is mainly attributed to lesions within extracranial arteries, but in Asians, intracranial atherosclerotic stenosis (ICAS) is a more important cause than extracranial stenosis. In addition to racial differences, intracranial atherosclerosis has a different mechanism for causing cerebral infarction compared to extracranial atherosclerosis and has a higher risk of stroke recurrence. Therefore, treatment strategies for preventing recurrence of cerebral infarction would be different. This chapter summarizes the characteristics and pathophysiology of intracranial atherosclerosis, and its management. The prevalence of ICAS in ischemic stroke patients is widely variable according to ethnicity. It is highest in Asian countries, such as India (50%) and China (47%) compared to the western countries (European countries and USA is 10–16%). Among Asians, although the prevalence of ICAS in Korea (25%) is lower than that in India or China, it is still higher than that in Western countries. Ischemic stroke patients with ICAS are known to be at higher risk of recurrent ischemic stroke compared to other causes. The recurrence rate varies from 4 to 19% annually, and it reaches up to 20% in 1 year in some populations . The 2-year risk of recurrent stroke in the territory of the stenotic artery is 38.2% . The development of ICAS is associated not only with conventional atherosclerotic risk factors such as older age, hypertension, diabetes mellitus (DM), hypercholesterolemia, smoking, sedentary lifestyle, and metabolic syndrome but also with racial factors, specifically among Black and Asian populations . The degree of stenosis is one of the best predictors of the rate of stroke occurrence in patients with ICAS . Lipoprotein biomarkers (high apolipoprotein B/A1, low serum adiponectin, increased lipoprotein (a)) could serve as useful markers for assessing the severity of intracranial atherosclerotic disease (ICAD) . However, the severity of vascular risk factors does not always reflect that of ICAS. For example, the severity of DM, as measured by HbA1c levels, may not be associated with that of ICAS. Nevertheless, some articles suggest role or effect of traditional risk factors for atherosclerotic disease will be different in ICAS compared to coronary artery disease or extracranial atherosclerosis regarding the following aspects . First, one of such difference may lie in the role of LDL or other atherogenic lipoprotein. It is considered that the effects of LDL or other atherogenic lipoprotein in ischemic stroke might be relatively lower in patients with ICAS than those in patients with extracranial atherosclerotic stenosis (ECAS) . In addition, in a multicenter study conducted in Korea, hyperlipidemia was more closely associated with ECAS than ICAS . Furthermore, vascular tortuosity was found to be associated with symptomatic ICAS according to a retrospective study suggesting its potential contribution to intracranial atherosclerosis . Moreover, recent studies have indicated that ICAS was strongly associated with the prevalence of dementia and Alzheimer’s disease has been suggested . In this study, it was proposed that ICAS in 2 or more vascular territories is associated with the risk of dementia and Alzheimer’s disease, and it could be contributor to cognitive impairment. The major pathophysiology underlying ICAD is atherosclerosis, caused by cholesterol deposition within the arterial wall. ICAS differs from ECAS including carotid artery, coronary artery or peripheral artery atherosclerosis in terms of releasing substance, structure and vessel wall composition, and the mechanisms resulting in ischemic stroke . In addition to atherosclerosis, non-atherosclerotic vasculopathies are frequently observed in intracranial artery stenosis, particularly in the Asian population. Compared to ECAS, ICAS is characterized by a higher prevalence of proliferative fibrosis with fewer complicated lesions, rather than lipid infiltration and inflammation of the intima or adventitia. It was reported that intracranial arteries are constituted with lower levels of LDL, oxidized LDL, and intimal macrophages and more enhanced activity of antioxidant enzymes (manganese superoxide dismutase, copper-zinc superoxide dismutase, catalase) compared with extracranial arteries . Due to the greater antioxidant response observed in intracranial arteries, it is speculated that the intracranial vessels are relatively resistant to the development of atherosclerosis, potentially leading to the onset of intracranial atherosclerosis approximately 20 years later than extracranial atherosclerosis . Conversely, this delayed onset might coincide with a more rapid acceleration of atherosclerosis in elderly subjects compared to extracranial arteries . The structural composition of the intracranial and extracranial arteries is also different due to their distinct embryonic origins of the arteries . Specifically, intracranial arteries are of mesodermal origin and exhibit a vessel wall comprising a thinner media and adventitia, and fewer elastic fibers compared to extracranial arteries which is ectodermal origin . In addition, it was observed that the vessel wall of intracranial arteries was constituted of denser internal elastic lamina and without external elastic lamina. In the adventitia of intracranial arteries, there was little vasa vasorum which is associated with vulnerable atherosclerotic plaque, and it might be because they are surrounded and supplied with nutritional support by cerebrospinal fluid. Metabolically, intracranial arteries differ from extracranial ones, characterized by lower contents of hexosamine, uronic acid and sulfur, lower proportion of hyaluronic acid, chondroitin sulfates, a lower ratio of ester to total cholesterol, and elevated antioxidant enzyme activity (manganese superoxide dismutase, copper-zinc superoxide dismutase, catalase). The structures surrounding matrix molecules (e.g., fibrillar collagen) are different, and it affects the inflammatory and fibroproliferative response . The main mechanisms of ischemic stroke related to atherosclerosis are perfusion failure resulting from in situ thrombosis due to plaque rupture of atheroma, or from branch occlusive disease caused by plaque extension over the ostia of small perforator arteries. Intracranial atherosclerosis is usually located distal to the circle of Willis and has less chance of collateral circulation compared to extracranial atherosclerosis which is located proximal to the circle of Willis. Additionally, ischemic stroke related to atherosclerosis in intracranial arteries is caused by artery-to-artery embolism from the stenotic segment. Microembolic signals were commonly detected in patients with symptomatic MCA stenosis during the acute phase of ischemic stroke . The combination of branch occlusive disease and artery-to-artery embolism is one of the main mechanisms of ischemic stroke of ICAS . Although most intracranial artery stenoses are considered atherosclerotic lesions, arterial dissection or ring finger protein 213 (RNF213) vasculopathy are also commonly found in the Asian population. It can also be developed by vasculitis or RCVS (reversible cerebral vasoconstriction syndrome) in some cases of intracranial artery stenosis. Therefore, a comprehensive evaluation of differential diagnosis, as shown in , is crucial. Several diagnostic methods used to identify ICAS, including conventional cerebral angiography, magnetic resonance angiography, computerized tomography angiography, transcranial Doppler, and high-resolution vessel wall imaging (HR-MRI). Although conventional cerebral angiography is a gold standard for differential diagnosis, it is not routinely conducted because it is relatively invasive and will not disclose changes in vessel wall. Initially, intracranial stenosis was usually identified with computerized tomography angiography or conventional magnetic resonance angiography. However, these diagnostic tools may not be enough to distinguish between atherosclerosis and other etiologies. Transfemoral cerebral angiography plays a key role in identifying fine collateral vessels such as Moyamoya vessels or a beaded appearance, and detecting distal stenosis suggesting vasculitis, helping in the differentiation of etiologies. In addition to neuroimaging, genetic screening may support the diagnosis of RNF vasculopathy. Especially, by using high-resolution MRI, containing T1-weighted images with or without contrast-enhancement, T2-weighted images, proton density images, and susceptibility-weighted images, it will be useful to evaluate the pathophysiology of intracranial arteries. These imaging modalities help differentiate between the various etiologies mentioned above based on the vessel wall characteristics of the stenotic lesion, such as patterns of intima thickening, plaque components, vessel wall remodeling, and the presence of wall enhancement or susceptibility lesions. The strategies to manage ICAS including medical and endovascular treatment have been suggested by several large randomized trials, such as WASID, SAMMPRIS, CATHARSIS, TOSS I&II, CSPS.com1&2, and VISSIT trials. Currently, the best medical treatment with aggressive risk factor control is recommended with asymptomatic and symptomatic ICAS. Endovascular or surgical treatment can be another option for severe ICAS, but they are considered only in limited cases. The clinical trials failed to show clinical benefit or superiority of endovascular treatment over the best medical treatment. Medical Treatment As a crucial medical treatment, antithrombotic treatment is used to reduce the risk of stroke in patients with ICAS. Antiplatelet agents help prevent activation and aggregation of platelets and anticoagulation interferes with the coagulation cascade. Though anticoagulation was also used traditionally as a choice of management for ICAS, it was suggested that using anticoagulation is not superior to antiplatelet therapy due to its bleeding risks through a large randomized study, WASID trial. Antiplatelet therapy is the main strategy for stroke prevention in ICAD. Aspirin is the most widely used antiplatelet therapy. Other antiplatelet agents would be a good option as an add-on therapy or replacement for aspirin, but there are not enough studies indicating the efficacy and safety of the other agents. It was identified that dual antiplatelet therapy including aspirin for up to 90 days in patients with symptomatic severe ICAS (70–99% stenosis) reduced the risk of stroke recurrence. Because of the advances in antiplatelet agents during recent years, antiplatelet therapy has resulted in a significant reduction in the risk of recurrent stroke or death in patients with symptomatic ICAS. There are various antiplatelet agents, but the standard regimen for short-term DAPT or long-term antiplatelet monotherapy for secondary stroke prevention has not been suggested. Clopidogrel has been widely used as first-line agents with aspirin for short-term DAPT. Recently, cilostazol, an inhibitor of PDE III, has been found to be an alternative for use in DAPT therapy for the secondary prevention of ischemic stroke by preventing the progression of ICAS, without increasing bleeding risk . Cilostazol inhibits platelet aggregation and vascular smooth muscle proliferation, has vasodilatory activity, and protects the vascular wall and endothelium . It could also be recommended for long-term use of a combination of cilostazol with aspirin or clopidogrel in non-cardioembolic ischemic stroke patients, particularly with ICAS of at least 50% in a major intracranial artery, based on recent studies, although the studies did not include the patients of acute phase of ischemic stroke . Especially, it was demonstrated that the combination of cilostazol and clopidogrel reduced the recurrence of ischemic stroke in patients at the chronic stage of high-risk, non-cardioembolic stroke, compared with clopidogrel alone . In addition, cilostazol was non-inferior to aspirin for the secondary prevention of ischemic stroke after acute phase in Asians . In addition to medication with antithrombotic agents, control of risk factors is also an important part of medical treatment. There are modifiable risk factors associated with ICAS and it is expected to lead improved outcome in case of best medical treatment by understanding and strictly controlling those risk factors. Traditionally known vascular risk factors such as hypertension, DM, dyslipidemia, and smoking are also associated with ICAS. Mean systolic blood pressure (BP) turned out to be associated with the progression of ICAS through post hoc analysis from the data of SAMMPRIS trial . The American Heart Association (AHA)/American Stroke Association (ASA) guidelines of 2014 recommended a BP target of less than 140 mm Hg for symptomatic 50% or higher ICAS of a major intracranial artery (class I; level of evidence B). Similarly, the European Society of Cardiology (ESC)/European Atherosclerosis Society (EAS) 2019 guidelines also recommend a BP target of less than 140/90 mm Hg. Meanwhile, it is still uncertain that more strict BP control would be better because a recent randomized trial failed to demonstrate the superiority of more strict BP control targeting systolic BP under 120 in patients with subacute ischemic stroke. Further studies are warranted to clarify the optimal approach to BP control in this context. Another risk factor, DM, is estimated by measuring HbA1c or serum glucose levels. The AHA/ASA 2019 guidelines have not proposed any specific target goals for HbA1c or serum glucose levels in patients with ICAS, but it would not be different in target goals for management of vascular risk factors in other atherosclerotic diseases recommended by ESC/EAS 2019 guidelines. Dyslipidemia is one of the important risk factors not only for ECAS, but also for ICAS. High-intensity statin therapy, regardless of the initial level of low-density lipoprotein cholesterol (LDL-C), is recommended to reduce the risk of vascular events for patients up to 75 years old. Lifestyle modification including smoking cessation and adequate physical exercise was also known to be associated with low risk of stroke recurrence. Especially, it was identified more strongly associated with ICAS than with ECAS. Aggressive risk factor control and lifestyle modification could be more important and effective factors for reducing stroke recurrence than use of antithrombotic agents. Surgical and Endovascular Options Endovascular or surgical treatment can be an alternative to medical treatment in severe symptomatic ICAS patients who are intractable to the best medical treatment or are hemodynamically unstable. The efficacy and safety of percutaneous transluminal angioplasty and stenting (PTAS) in symptomatic ICAS with 70–99% stenosis has been evaluated in at least three clinical trials with various stenting systems, including SAMMPRIS trial, which were failed to demonstrate clinical benefits or superiority of PTAS over the best medical treatment in reducing the risk of recurrent stroke. With advancements in technologies and tools, PTAS can be considered a treatment option in specific patients with 70–99% intracranial stenosis who are hemodynamically unstable or who suffer from recurrent ischemic events in spite of optimal medical treatment . Bypass surgery is considered in patients with severe ICAS when there is no other option for treatment. However, there is no evidence of reduction in the risk of stroke recurrence in clinical studies. It is usually recommended as a treatment for intracranial stenosis in Moyamoya disease rather than in other etiologies, including atherosclerotic stenosis. It may be considered in patients with severe multifocal atherosclerotic stenosis and no alternative treatments such as angioplasty or stenting. As a crucial medical treatment, antithrombotic treatment is used to reduce the risk of stroke in patients with ICAS. Antiplatelet agents help prevent activation and aggregation of platelets and anticoagulation interferes with the coagulation cascade. Though anticoagulation was also used traditionally as a choice of management for ICAS, it was suggested that using anticoagulation is not superior to antiplatelet therapy due to its bleeding risks through a large randomized study, WASID trial. Antiplatelet therapy is the main strategy for stroke prevention in ICAD. Aspirin is the most widely used antiplatelet therapy. Other antiplatelet agents would be a good option as an add-on therapy or replacement for aspirin, but there are not enough studies indicating the efficacy and safety of the other agents. It was identified that dual antiplatelet therapy including aspirin for up to 90 days in patients with symptomatic severe ICAS (70–99% stenosis) reduced the risk of stroke recurrence. Because of the advances in antiplatelet agents during recent years, antiplatelet therapy has resulted in a significant reduction in the risk of recurrent stroke or death in patients with symptomatic ICAS. There are various antiplatelet agents, but the standard regimen for short-term DAPT or long-term antiplatelet monotherapy for secondary stroke prevention has not been suggested. Clopidogrel has been widely used as first-line agents with aspirin for short-term DAPT. Recently, cilostazol, an inhibitor of PDE III, has been found to be an alternative for use in DAPT therapy for the secondary prevention of ischemic stroke by preventing the progression of ICAS, without increasing bleeding risk . Cilostazol inhibits platelet aggregation and vascular smooth muscle proliferation, has vasodilatory activity, and protects the vascular wall and endothelium . It could also be recommended for long-term use of a combination of cilostazol with aspirin or clopidogrel in non-cardioembolic ischemic stroke patients, particularly with ICAS of at least 50% in a major intracranial artery, based on recent studies, although the studies did not include the patients of acute phase of ischemic stroke . Especially, it was demonstrated that the combination of cilostazol and clopidogrel reduced the recurrence of ischemic stroke in patients at the chronic stage of high-risk, non-cardioembolic stroke, compared with clopidogrel alone . In addition, cilostazol was non-inferior to aspirin for the secondary prevention of ischemic stroke after acute phase in Asians . In addition to medication with antithrombotic agents, control of risk factors is also an important part of medical treatment. There are modifiable risk factors associated with ICAS and it is expected to lead improved outcome in case of best medical treatment by understanding and strictly controlling those risk factors. Traditionally known vascular risk factors such as hypertension, DM, dyslipidemia, and smoking are also associated with ICAS. Mean systolic blood pressure (BP) turned out to be associated with the progression of ICAS through post hoc analysis from the data of SAMMPRIS trial . The American Heart Association (AHA)/American Stroke Association (ASA) guidelines of 2014 recommended a BP target of less than 140 mm Hg for symptomatic 50% or higher ICAS of a major intracranial artery (class I; level of evidence B). Similarly, the European Society of Cardiology (ESC)/European Atherosclerosis Society (EAS) 2019 guidelines also recommend a BP target of less than 140/90 mm Hg. Meanwhile, it is still uncertain that more strict BP control would be better because a recent randomized trial failed to demonstrate the superiority of more strict BP control targeting systolic BP under 120 in patients with subacute ischemic stroke. Further studies are warranted to clarify the optimal approach to BP control in this context. Another risk factor, DM, is estimated by measuring HbA1c or serum glucose levels. The AHA/ASA 2019 guidelines have not proposed any specific target goals for HbA1c or serum glucose levels in patients with ICAS, but it would not be different in target goals for management of vascular risk factors in other atherosclerotic diseases recommended by ESC/EAS 2019 guidelines. Dyslipidemia is one of the important risk factors not only for ECAS, but also for ICAS. High-intensity statin therapy, regardless of the initial level of low-density lipoprotein cholesterol (LDL-C), is recommended to reduce the risk of vascular events for patients up to 75 years old. Lifestyle modification including smoking cessation and adequate physical exercise was also known to be associated with low risk of stroke recurrence. Especially, it was identified more strongly associated with ICAS than with ECAS. Aggressive risk factor control and lifestyle modification could be more important and effective factors for reducing stroke recurrence than use of antithrombotic agents. Endovascular or surgical treatment can be an alternative to medical treatment in severe symptomatic ICAS patients who are intractable to the best medical treatment or are hemodynamically unstable. The efficacy and safety of percutaneous transluminal angioplasty and stenting (PTAS) in symptomatic ICAS with 70–99% stenosis has been evaluated in at least three clinical trials with various stenting systems, including SAMMPRIS trial, which were failed to demonstrate clinical benefits or superiority of PTAS over the best medical treatment in reducing the risk of recurrent stroke. With advancements in technologies and tools, PTAS can be considered a treatment option in specific patients with 70–99% intracranial stenosis who are hemodynamically unstable or who suffer from recurrent ischemic events in spite of optimal medical treatment . Bypass surgery is considered in patients with severe ICAS when there is no other option for treatment. However, there is no evidence of reduction in the risk of stroke recurrence in clinical studies. It is usually recommended as a treatment for intracranial stenosis in Moyamoya disease rather than in other etiologies, including atherosclerotic stenosis. It may be considered in patients with severe multifocal atherosclerotic stenosis and no alternative treatments such as angioplasty or stenting. ICAD represents the most common cause of ischemic stroke in Asian populations, and it is also a leading cause of stroke recurrence. Further studies dedicated to appropriate management for ICAS are needed by focusing on its distinct pathophysiology. ICAS has been associated not only with older age and specific racial backgrounds but also with modifiable vascular risk factors such as hypertension, DM, and dyslipidemia. The pathophysiology underlying ICAS progression and the composition of intracranial vessel walls may be different from that of extracranial arteries. Although atherosclerosis remains the primary pathology in ICAS, other non-atherosclerotic arteriopathies such as RNF vasculopathy, arterial dissection and vasculitis, should also be considered in the differential diagnosis. Diagnostic tools like high-resolution vessel wall imaging or conventional angiography can be valuable in this regard. The best medical treatment indicating strict risk factor control and medication with optimal antiplatelet therapy is crucial for primary and secondary stroke prevention in both asymptomatic and symptomatic ICAS. Short-term dual antiplatelet therapy for 90 days is recommended in patients with symptomatic severe ICAS (70–99%). While clopidogrel is widely used as second-line therapy, cilostazol is also recommended not only for short-term dual antiplatelet regimens but also for long-term monotherapy regimens in ICAS patients at high risk of non-cardioembolic ischemic stroke after the acute phase. Endovascular or surgical treatment can also be considered an alternative to medical treatment in select cases with severe ICAS where patients are unresponsive to the best medical treatment or hemodynamically unstable. J.Y. Song and S.U. Kwon have no conflict of interest related with this article. The authors have no conflicts of interest to declare. This study was not supported by any sponsor or funder. J.Y.S. contributed to the manuscript draft. S.U.K. contributed to the conception and design of the work, manuscript revision, and supervision for critically important intellectual content. |
An international multicentre randomised controlled trial of | 0299c9f9-f5b1-4f90-812a-158f8c7fb792 | 11842884 | Surgical Procedures, Operative[mh] | High‐quality transurethral resection of bladder tumour (TURBT) is the main tool for the treatment of non‐muscle‐invasive bladder cancer (NMIBC) . Conventional TURBT (cTURBT) with fractioning of the tumour remains the standard surgical approach for NMIBC to date. However, it has come under criticism because it contradicts the ‘no touch’ principle of oncological surgery . Also, cTURBT causes cauterisation artefacts and thermal damage, which impairs the staging of bladder cancer . Thermal damage to tumour tissue can lead to difficulties in assessing the tumour stage and may result in under‐staging and under‐treatment of patients . Previous studies have demonstrated the importance of sub‐staging by dividing pT1 tumours into pT1a and pT1b , defined by lamina muscularis mucosae infiltration. Rouprêt et al. demonstrated the prognostic significance of T1 sub‐staging, in which patients with pT1b disease underwent radical cystectomy 6 months earlier than patients with pT1a disease because the disease progressed earlier. An en bloc resection of bladder tumour (ERBT) seeks to overcome this limitation. The ERBT is an emerging resection method for the treatment of NMIBC. In ERBT, the tumour is marked circumferentially before being incised to the detrusor muscle (DM) layer, and this depth is maintained until the tumour is resected en bloc . Previous studies on ERBT suggested more accurate staging, fewer complications, and lower recurrence rates compared to cTURBT; however, most studies comparing ERBT with cTURBT have been limited by either retrospective or single‐centre study design. We performed a single‐blinded randomised controlled trial (RCT) in seven European hospitals (Luebeck, Wolfsburg, Heilbronn, Berlin, Germany; Salzburg, Austria; Prague, Czech Republic; and Modena, Italy, German clinical trial register: DRKS00020738) from June 2019 until March 2022. Primary endpoints were non‐inferiority in recurrence rate and superiority in the presence of DM in the sample for ERBT. Secondary endpoints included progression rate, perioperative safety, and residual tumour rate at 2–6 weeks, tumour extraction method, long‐term recurrence at 24 months, and feasibility of histopathological staging (resection margins and T1 sub‐staging). Inclusion criteria were the initial diagnosis of NMIBC and tumour size >4.3 mm (which is the standard diameter of a Karl Storz resection sling to prevent en bloc resection during cTURBT). There were no restrictions on the number of tumours, tumour location, or maximum tumour diameter. All surgeons were required to perform a cystoscopy intraoperatively but before the resection procedure. If all tumours were potentially eligible for ERBT, patients were randomised intraoperatively in a 1:1 ratio to either the cTURBT or the ERBT group. Randomisation was performed with an on‐line randomisation tool ( https://www.randomizer.org ). Randomisation envelopes that contained the allocated group were concealed and placed in the operating room. If the surgeon determined that the tumour was not ERBT eligible, a screening failure was reported with justification (i.e., location, tumour size), and patients were not randomised (Fig. ). All visualisation methods and energy sources were eligible for this study. Exclusion criteria were recurrent NMIBC, solid, broad‐based tumours, as this study aimed to evaluate the ERBT of papillary urothelial carcinoma of the bladder, tumour diameter <4.3 mm, or singular carcinoma in situ . If a change in surgical method (‘switch’) to cTURBT was required, these cases were excluded from perioperative statistical analysis as per modified intention‐to‐treat approach to avoid confounding in statistical analysis. Study sites were provided with a guideline for the evaluation of en bloc tumours by the Institute of Pathology in Luebeck, Germany. Surgeons were free to obtain additional biopsies from the tumour ground/base of the tumour. The Institute of Biometry at Hannover Medical School (Germany) performed a statistical power analysis prior to the study. Sample size calculation for non‐inferiority in recurrence rate was based on a meta‐analysis by Sylvester et al. . As per inclusion criteria, we focused on prognosis scores 1–9. Assuming H0: odds ratio (OR) ≥ δ vs alternative hypothesis H1: OR < δ, with δ being defined as 1.5, significance level being α = 0.025 and statistical power of 0.9 we calculated a sample size of n = 177 patients per group. The statistical analysis was performed using the Statistical Package for the Social Sciences (SPSS), version 26 (IBM Corp., Armonk, NY, USA). As per study protocol we performed an interim analysis after 36 months being the planned time point for the final analysis, which was at ~27% of recruitment. Due to a significantly higher rate of bladder perforations in the intervention (ERBT) group, this study was terminated before the targeted patient number could be reached to ensure patient safety, as pre‐defined in study protocol. This study was approved by all ethics committees of the participating sites. A total of 134 patients were screened, of whom 97 were successfully enrolled (cTURBT n = 40, ERBT n = 57) from June 2019 to March 2022. There were 28 screening failures reported. Patient and Tumour Characteristics Tumours in the cTURBT group were 0.6 cm larger than ERBT tumours. In both groups, the tumours were mainly located on the lateral bladder walls (cTURBT 66.0%, ERBT 57.9%). In the ERBT group, 23.2% of tumours were located in the trigonal region and 2.9% on the anterior wall (Table ). In two patients, a switch from ERBT to cTURBT was required, with the reasons being tumour diameter (4.0 cm in one) and tumour extension into a bladder diverticulum (in one). No complications were reported in these cases. The mean diameter of ERBT tumours extracted after intravesical fragmentation was 3.3 cm. The largest tumour extracted en bloc measured 4.0 cm. An en bloc extraction was feasible in 83.6% of ERBT, while 16.4% of ERBT tumours were fragmented within the bladder and subsequently extracted ( P < 0.001). The mean (SD) drop of haemoglobin in the cTURBT group was 4.2 (0.1) g/L, while in the ERBT group it was 5.8 (0.102) g/L ( P = 0.488). The mean (SD) irrigation time was 21.2 (15.9) h in the cTURBT group and 21.6 (19.1) h in the ERBT group ( P = 0.898). The mean (SD) catheterisation time was 1.97 (0.87) days in the cTURBT group and 2.44 (2.29) days in the ERBT group ( P = 0.173). The overall intraoperative complication rate did not differ significantly between groups ( P = 0.142); however, more bladder perforations were observed in the ERBT group (none for cTURBT, seven [12.7%] for ERBT; P = 0.020). The postoperative complication rate did not differ significantly between groups (four [10.5%] in the cTURBT group and seven [12.7%] in the ERBT group, P = 1.000). Clavien–Dindo Grade I complications occurred in two patients in each group (5.3% in the cTURBT and 3.7% in the ERBT group). There were no Clavien–Dindo Grade II complications in any group. Clavien–Dindo Grade IIIa, IIIb, and IVa complications were observed (one patient each) in the ERBT group. No complications of Grade ≥III were observed in the cTURBT group. The Grade IVa complication was cardiac decompensation in an 81‐year‐old female patient (American Society of Anesthesiologists score of 1 , Eastern Cooperative of Oncology Group performance status of 1) after a holmium laser resection. Prolonged bladder irrigation was required in two (5.3%) of the cTURBT and three (5.6%) of the ERBT group ( P = 1.000). One patient (1.8%) in the ERBT required ureteric stenting and percutaneous nephrostomy after resection of a ureteric orifice ( P = 1.000). Two patients had postoperative bleeding, with one requiring transfusion after ERBT who had tumours located at the anterior wall. Bladder Perforations Bladder perforations occurred in seven patients who underwent ERBT in multiple participating centres (Tables and ). Bladder perforations were defined macroscopically by visible fat during resection, without mandatory radiographic measurement. Tumour size in these patients ranged from 1.0 to 6 cm, with tumours being mainly located at the left side wall and trigonal area (Table ). When compared to ERBT without perforation, bladder perforations did not lead to prolonged irrigation time ( P = 0.449) or hospitalisation time ( P = 0.993); however, there was evidence of longer catheterisation time, although this was not statistically significant (2.26 vs 3.71 days P = 0.058). Additionally, bladder perforations occurred in three out of 11 holmium cases (27.2%). There was no significant accumulation for study centres ( P = 0.721) or specific surgeons ( P = 0.504). Pathological and Oncological Outcomes Presence of DM was observed in 73.7% of cTURBT and 67.3% of ERBT specimens ( P = 0.69). R0 status was reported in 72.2% after ERBT vs 45% after cTURBT ( P = 0.006), whilst resection margins were not assessable (Rx) in 45% of cTURBT cases vs 25.5% after ERBT ( P = 0.047). R1 status was reported in four cTURBT cases (10%) and one ERBT case (1.8%) but showed no statistical significance ( P = 0.158; Table ). Second resection was less likely to be performed after ERBT ( P = 0.010). Multivariate regression showed only R0 status and ERBT to be independent predictors for no second resection after 6 weeks (Table ). There was no statistical difference in recurrence rates ( P = 0.625) after 6 months as shown in Table . Sub‐staging feasibility was reported in 100% of T1 tumours undergoing ERBT, compared to 38% of cTURBT cases ( P = 0.006). Tumours in the cTURBT group were 0.6 cm larger than ERBT tumours. In both groups, the tumours were mainly located on the lateral bladder walls (cTURBT 66.0%, ERBT 57.9%). In the ERBT group, 23.2% of tumours were located in the trigonal region and 2.9% on the anterior wall (Table ). In two patients, a switch from ERBT to cTURBT was required, with the reasons being tumour diameter (4.0 cm in one) and tumour extension into a bladder diverticulum (in one). No complications were reported in these cases. The mean diameter of ERBT tumours extracted after intravesical fragmentation was 3.3 cm. The largest tumour extracted en bloc measured 4.0 cm. An en bloc extraction was feasible in 83.6% of ERBT, while 16.4% of ERBT tumours were fragmented within the bladder and subsequently extracted ( P < 0.001). The mean (SD) drop of haemoglobin in the cTURBT group was 4.2 (0.1) g/L, while in the ERBT group it was 5.8 (0.102) g/L ( P = 0.488). The mean (SD) irrigation time was 21.2 (15.9) h in the cTURBT group and 21.6 (19.1) h in the ERBT group ( P = 0.898). The mean (SD) catheterisation time was 1.97 (0.87) days in the cTURBT group and 2.44 (2.29) days in the ERBT group ( P = 0.173). The overall intraoperative complication rate did not differ significantly between groups ( P = 0.142); however, more bladder perforations were observed in the ERBT group (none for cTURBT, seven [12.7%] for ERBT; P = 0.020). The postoperative complication rate did not differ significantly between groups (four [10.5%] in the cTURBT group and seven [12.7%] in the ERBT group, P = 1.000). Clavien–Dindo Grade I complications occurred in two patients in each group (5.3% in the cTURBT and 3.7% in the ERBT group). There were no Clavien–Dindo Grade II complications in any group. Clavien–Dindo Grade IIIa, IIIb, and IVa complications were observed (one patient each) in the ERBT group. No complications of Grade ≥III were observed in the cTURBT group. The Grade IVa complication was cardiac decompensation in an 81‐year‐old female patient (American Society of Anesthesiologists score of 1 , Eastern Cooperative of Oncology Group performance status of 1) after a holmium laser resection. Prolonged bladder irrigation was required in two (5.3%) of the cTURBT and three (5.6%) of the ERBT group ( P = 1.000). One patient (1.8%) in the ERBT required ureteric stenting and percutaneous nephrostomy after resection of a ureteric orifice ( P = 1.000). Two patients had postoperative bleeding, with one requiring transfusion after ERBT who had tumours located at the anterior wall. Bladder perforations occurred in seven patients who underwent ERBT in multiple participating centres (Tables and ). Bladder perforations were defined macroscopically by visible fat during resection, without mandatory radiographic measurement. Tumour size in these patients ranged from 1.0 to 6 cm, with tumours being mainly located at the left side wall and trigonal area (Table ). When compared to ERBT without perforation, bladder perforations did not lead to prolonged irrigation time ( P = 0.449) or hospitalisation time ( P = 0.993); however, there was evidence of longer catheterisation time, although this was not statistically significant (2.26 vs 3.71 days P = 0.058). Additionally, bladder perforations occurred in three out of 11 holmium cases (27.2%). There was no significant accumulation for study centres ( P = 0.721) or specific surgeons ( P = 0.504). Pathological and Oncological Outcomes Presence of DM was observed in 73.7% of cTURBT and 67.3% of ERBT specimens ( P = 0.69). R0 status was reported in 72.2% after ERBT vs 45% after cTURBT ( P = 0.006), whilst resection margins were not assessable (Rx) in 45% of cTURBT cases vs 25.5% after ERBT ( P = 0.047). R1 status was reported in four cTURBT cases (10%) and one ERBT case (1.8%) but showed no statistical significance ( P = 0.158; Table ). Second resection was less likely to be performed after ERBT ( P = 0.010). Multivariate regression showed only R0 status and ERBT to be independent predictors for no second resection after 6 weeks (Table ). There was no statistical difference in recurrence rates ( P = 0.625) after 6 months as shown in Table . Sub‐staging feasibility was reported in 100% of T1 tumours undergoing ERBT, compared to 38% of cTURBT cases ( P = 0.006). Presence of DM was observed in 73.7% of cTURBT and 67.3% of ERBT specimens ( P = 0.69). R0 status was reported in 72.2% after ERBT vs 45% after cTURBT ( P = 0.006), whilst resection margins were not assessable (Rx) in 45% of cTURBT cases vs 25.5% after ERBT ( P = 0.047). R1 status was reported in four cTURBT cases (10%) and one ERBT case (1.8%) but showed no statistical significance ( P = 0.158; Table ). Second resection was less likely to be performed after ERBT ( P = 0.010). Multivariate regression showed only R0 status and ERBT to be independent predictors for no second resection after 6 weeks (Table ). There was no statistical difference in recurrence rates ( P = 0.625) after 6 months as shown in Table . Sub‐staging feasibility was reported in 100% of T1 tumours undergoing ERBT, compared to 38% of cTURBT cases ( P = 0.006). This study was initially designed to evaluate 360 patients (180 per group). Due to the increased risk of bladder perforations and subsequent safety concerns for ERBT, this study had to be terminated after the enrolment and analysis of 97 patients. Thus, these data should be considered, knowing that statistical power is lower than initially planned. Of the 28 failed screenings, 15 patients (11.2% of screened patients) were explicitly excluded for ERBT based on the following criteria: localisation, tumour configuration, tumour size, solid tumours, or number of tumours, meaning that ERBT was potentially feasible in nearly 90% of the screened cases. In a 2017 meta‐analysis, our group reported an overall feasibility rate of 70% for ERBT . The HYBRIDBLUE study (German clinical trials register ID: DRKS00004414) by Gakis et al. showed an exclusion rate for en bloc hydrodissection of 69.06%, with more restrictive inclusion criteria. Our RCT seems to be in favour of ERBT compared with previous data. Recent RCTs comparing cTURBT and ERBT showed conversion rates of 3.4% (D'Andrea et al. ) up to 10% (4.3%, Gallioli et al. ). The reasons given for the conversions were suspected detrusor infiltration, resection of the ureteric meatus, laser failure, perforation, tumour size and localisation . In our study, two (3.5%) patients in the ERBT group required conversion to cTURBT, which is consistent with the data from D'Andrea et al. and lower than in the RCT by Teoh et al. . Hurle et al. conducted an observational study at a single centre and found a low conversion rate of 0.97%. While the selection for ERBT cases was more stringent in these trials than ours, excluding tumours >3 cm in diameter . Gallioli et al. also excluded multiple tumours. A systematic review by Naselli and Puppo found a conversion rate of 6.5%. Teoh et al. reported decreasing rates of successful ERBT with increasing tumour size, reaching 29.6% for tumours >3 cm. Our results lie between these findings and show that ERBT is feasible in 96.5% of cases without the need for intraoperative modification of the surgical method, after proper preoperative tumour assessment. Previous studies have shown that en bloc tumour extraction may be limited to ~3 cm . In our study, we did not set an upper limit for tumour size in the inclusion criteria. The average tumour size for tumours that were fragmented intravesically before extraction was 3.3 cm, whereas the largest tumour that was retrieved en bloc measured 4 cm. Figure shows the increasing rate of ERBT tumours that were fragmented intravesically, with an increase in the fragmented fraction at 3.5 cm. In only one case, tumour size was the reason for an intraoperative change of the resection method. Our findings confirm that tumour size is not a contraindication for successful ERBT, whereas larger tumours still pose a problem for intact recovery. The results of previous studies on operation time are inconsistent. The recent larger RCT of Gallioli et al. , D'Andrea et al. and Teoh et al. found no difference in median operative time, whilst a retrospective analysis from Li et al. showed a shorter duration for thulium laser ERBT (25.96 min) compared to cTURBT (37.18 min) ( P = 0.018). Meanwhile, Gakis et al. reported significantly longer operative time using a HybridKnife® (37.1 vs 22.4 min for cTURBT; P < 0.001). In our study and in congruence to the other reported RCTs, no significant difference in mean operative time was observed ( P = 0.450), thus, the operation time seems to be dependent on the energy source and the technique used for ERBT. In a previous observational study, ERBT was described to have an advantage over cTURBT concerning DM inclusion rate . In the RCT conducted by D'Andrea et al. a significantly higher rate of DM was reported in favour of the ERBT group (80.7% vs 71.1%). Other RCTs and non‐randomised prospective trials failed to prove significant differences for either one of the two methods regarding DM inclusion rate. Our results confirm these findings and found no benefit for either cTURBT or ERBT ( P = 0.69). In our study, ERBT showed a significantly higher rate of R0 assessability ( P = 0.006), and a significant reduction of necessary second resection ( P = 0.010). Multivariate regression showed ERBT and R0 status to be significant predictors for a second resection. This confirms the findings by Gakis et al. , where hydrodissection ERBT showed superiority in R0 status. Additionally, the RCTs by D'Andrea et al. and Teoh et al. reported higher margin assessability for the ERBT group . This seems to be a significant advantage for ERBT, especially when a second resection is performed due to uncertainty of complete resection. As reported in the RCT by Gallioli et al. , in our study sub‐staging was assessable in 100% of T1 cases ( P = 0.006) showing superiority in staging quality of ERBT. In this early data, there was no significant difference in recurrence rate after the 6‐month follow‐up period. Most previous RCTs showed no significant advantage for ERBT . However, Teoh et al. recently found a significant advantage for ERBT regarding the 1‐year recurrence rate, with a hazard ratio of 0.57, especially for patients with 1–3 cm tumours, a single tumour, Ta disease, or intermediate‐risk NMIBC. They did not detect any significant difference in the 1‐year progression rate between the ERBT and cTURBT groups ( P = 0.065). Our own long‐term follow up is still ongoing at this point. In a previous meta‐analysis, our group reported a significantly higher drop of haemoglobin with electrical ERBT compared to laser ERBT (4.6 vs 1.5 g/L, P = 0.0013) . The en bloc resection of urothelium carcinoma of the bladder (EBRUC) II study showed no statistical difference between the two groups ( P = 0.488). A mean haemoglobin drop averaged 4.2 g/L in the cTURBT group and 5.8 g/L in the ERBT group, and the difference can be considered clinically insignificant. In further studies, the use of haemoglobin levels as a standard value for comparison of TURBT should be evaluated critically. Mean irrigation time was not significantly different between groups ( P = 0.898). Our results confirm those of previous studies with an average irrigation time after ERBT of <24 h . Surprisingly, bladder perforations were observed more frequently after ERBT ( P = 0.02). Previous studies showed either no significant difference between the methods or an advantage for ERBT (5.6% vs 12%, D'Andrea et al. ), whilst perforation in the RCT performed by Gallioli et al. was defined as resection reaching the perivesical fat. In a single‐centre study by Hurle et al. , cystography was performed after macroscopic detection of bladder perforation. Other studies did not define bladder perforation or performed cystography routinely . Limitations Although the presented EBRUC II study was started by a pro‐ERBT group, the planned interim analysis showed a significantly increased perforation rate. It should be mentioned that in our study, bladder perforations were documented by the surgeon according to subjective assessment. The perforations did not occur in a single, but in multiple study sites. No objective measurements (i.e., cystography) were systematically implemented in the initial EBRUC II study protocol. Although the systematic use of cystograms was discussed during the initial development of the study protocol, it could not be implemented in a multicentre setting. Even if this limits the data quality, it was no longer possible to continue the study in a protocol‐compliant manner. As there are virtually no data on a comparison of perforation rates in the multicentre setting, we believe that these are worth reporting. In our view, this reflects the clinical reality in which bladder perforations are often defined macroscopically by visible fat during deep resection. This definition is also covered by the DEpth of Endoscopic Perforation (DEEP) scale reported by Breda et al. . Additionally, it is consistent with other trials. Although ERBT showed a perforation rate of 20% (28 cases) compared to 17% for cTURBT, no cystographic evidence was mentioned in the trial of Gallioli et al. (Table ) . Radiographic confirmation is left to the discretion of the surgeon and is not mandatory in the case of safe extraperitoneal perforations. They can usually be managed conservatively with longer catheter placement time and sparing bladder irrigation. A follow‐up study with the systematic use of cystograms would certainly be interesting for the final clarification of the perforation question. Fortunately, the perforations described in our trial occurred extraperitoneally, yet the increased risk of bladder perforation must be considered and evaluated when performing ERBT. In our case, it led to a prolongation of catheterisation time although not statistically significant. Additionally, early bladder instillations could not be administered. Multivariate analysis for intraoperative complication showed that patient age and duration of resection were predictors of intraoperative complication, but resection method was not a significant predictor (Table ). Bebane et al. showed a correlation between surgeon experience and complication rate when performing TURBT. Surgeon experience should be considered when conducting further studies on ERBT. This is a major limitation of this study, because ERBT is a relatively new procedure compared with cTURBT. Because surgeon identity was assessed without assessing the number of ERBTs conducted by each surgeon, intraoperative complications such as perforations and assessment of resection quality may be limited due to inexperienced surgeons. Bladder perforations occurred in three out of 11 holmium cases (27.2%), two on the left sidewall and one in the trigonal region. Gallioli et al. reported an overall perforation rate of 20% for ERBT vs 17% for cTURBT without statistical significance ( P = 0.9). They mentioned a 17% perforation rate for thulium laser ERBT, compared to 29% for bipolar and 14% for monopolar ERBT. Other trials did not show a higher perforation risk for laser ERBT . Laser resection techniques were originally developed for TURP. Whereas prostate resection ends at the compact and dense capsule where the adenoma can be cut off and torn out of the capsule, there is no such natural boundary in bladder tumour resection. Therefore, deep laser resection may carry a higher risk of bladder perforation, especially in inexperienced hands, although a lower rate of obturator nerve reflexes has been reported . The fact that the DM detection rates were comparable in both arms invalidates the assumption that cTURBT may not have been performed deeply enough or insufficiently (73.7% of cTURBT and 67.3% of ERBT specimens; P = 0.69). No standardised tumour ground biopsy was performed in the cTURBT arm, which limits the power to directly compare resection margin information. Nevertheless, resection margin information can be provided more frequently after ERBT if no additional biopsies are taken. However, this is the first multicentre RCT on ERBT in a real‐world setting without limitations regarding tumour localisation, tumour size or energy source. Although the presented EBRUC II study was started by a pro‐ERBT group, the planned interim analysis showed a significantly increased perforation rate. It should be mentioned that in our study, bladder perforations were documented by the surgeon according to subjective assessment. The perforations did not occur in a single, but in multiple study sites. No objective measurements (i.e., cystography) were systematically implemented in the initial EBRUC II study protocol. Although the systematic use of cystograms was discussed during the initial development of the study protocol, it could not be implemented in a multicentre setting. Even if this limits the data quality, it was no longer possible to continue the study in a protocol‐compliant manner. As there are virtually no data on a comparison of perforation rates in the multicentre setting, we believe that these are worth reporting. In our view, this reflects the clinical reality in which bladder perforations are often defined macroscopically by visible fat during deep resection. This definition is also covered by the DEpth of Endoscopic Perforation (DEEP) scale reported by Breda et al. . Additionally, it is consistent with other trials. Although ERBT showed a perforation rate of 20% (28 cases) compared to 17% for cTURBT, no cystographic evidence was mentioned in the trial of Gallioli et al. (Table ) . Radiographic confirmation is left to the discretion of the surgeon and is not mandatory in the case of safe extraperitoneal perforations. They can usually be managed conservatively with longer catheter placement time and sparing bladder irrigation. A follow‐up study with the systematic use of cystograms would certainly be interesting for the final clarification of the perforation question. Fortunately, the perforations described in our trial occurred extraperitoneally, yet the increased risk of bladder perforation must be considered and evaluated when performing ERBT. In our case, it led to a prolongation of catheterisation time although not statistically significant. Additionally, early bladder instillations could not be administered. Multivariate analysis for intraoperative complication showed that patient age and duration of resection were predictors of intraoperative complication, but resection method was not a significant predictor (Table ). Bebane et al. showed a correlation between surgeon experience and complication rate when performing TURBT. Surgeon experience should be considered when conducting further studies on ERBT. This is a major limitation of this study, because ERBT is a relatively new procedure compared with cTURBT. Because surgeon identity was assessed without assessing the number of ERBTs conducted by each surgeon, intraoperative complications such as perforations and assessment of resection quality may be limited due to inexperienced surgeons. Bladder perforations occurred in three out of 11 holmium cases (27.2%), two on the left sidewall and one in the trigonal region. Gallioli et al. reported an overall perforation rate of 20% for ERBT vs 17% for cTURBT without statistical significance ( P = 0.9). They mentioned a 17% perforation rate for thulium laser ERBT, compared to 29% for bipolar and 14% for monopolar ERBT. Other trials did not show a higher perforation risk for laser ERBT . Laser resection techniques were originally developed for TURP. Whereas prostate resection ends at the compact and dense capsule where the adenoma can be cut off and torn out of the capsule, there is no such natural boundary in bladder tumour resection. Therefore, deep laser resection may carry a higher risk of bladder perforation, especially in inexperienced hands, although a lower rate of obturator nerve reflexes has been reported . The fact that the DM detection rates were comparable in both arms invalidates the assumption that cTURBT may not have been performed deeply enough or insufficiently (73.7% of cTURBT and 67.3% of ERBT specimens; P = 0.69). No standardised tumour ground biopsy was performed in the cTURBT arm, which limits the power to directly compare resection margin information. Nevertheless, resection margin information can be provided more frequently after ERBT if no additional biopsies are taken. However, this is the first multicentre RCT on ERBT in a real‐world setting without limitations regarding tumour localisation, tumour size or energy source. This is the first European multicentre prospective RCT of ERBT compared to cTURBT in a real‐world setting, with no restrictions on tumour size, number of tumours, or energy sources. The feasibility of ERBT in our trial is higher than previously reported. Overall perioperative and safety parameters are comparable to cTURBT. Nevertheless, ERBT appears to carry an increased risk of bladder perforation, especially with laser and sidewall resections, which should be considered when performing ERBT. Preoperative assessment of the tumour size and location is essential for successful ERBT. Staging quality is superior in ERBT specimens resulting in lesser rates of second resections after initial ERBT. Recurrence rates after a 6‐month follow‐up are comparable. With longer follow‐up available, we intend to report on oncological outcomes focusing on in‐ and out‐field recurrence rates. None declared. Table S1. Surgical technique and main characteristics of patients who had a bladder perforation. |
Halophyte‐based crop managements induce biochemical, metabolomic and proteomic changes in tomato plants under saline conditions | 1f3b53b3-83fb-4d63-a67c-e1089c39e959 | 11739548 | Biochemistry[mh] | INTRODUCTION Salinity is one of the most significant environmental challenges limiting plant productivity. Around 830 million hectares are affected by salinity worldwide (source: FAO 2021), covering 20% of global cultivable land (Acosta‐Motos et al. ). The continuous growth of the population, especially in developing countries, and the need to increase food production are major challenges for the coming years. By the middle of this century, when the population will reach 9.7 billion, it is estimated that there will be less water resources available but also less arable land, mainly due to soil salinization and soil degradation (Ben Hamed et al. ). This situation is already forcing farmers to use groundwater, which contains salts, and inevitably to cultivate in areas affected by salinity. Thus, the use of salt‐tolerant plants, such as halophytes, may provide an important alternative for many developing countries in order to produce food and fodder (Ben Hamed et al. ; Custódio et al. ). Halophytes can complete their life cycle in the presence of high NaCl concentrations (300–500 mM) thanks to the development of adaptation mechanisms of morphological, anatomical and biochemical types (Acosta‐Motos et al. ; Ben Hamed et al. ), the latter including ion homeostasis, osmolarity regulation and up‐regulation of antioxidant defences (Ozgur et al. ; Bose et al. ). Different investigators have used halophyte plants to improve the production of plants of agronomic interest using two crop management strategies: intercropping (co‐cultivation of the halophyte with the crop plant of interest) and crop rotation (crop cultivation sequentially on the same plot where a halophyte has been previously grown). In that regard, intercropping has been successfully used to improve the production of watermelon, tomato and strawberry under moderate saline conditions (Albaho and Green ; Simpson et al. ; Karakas et al. ; Jurado‐Mañogil et al. ; Jurado et al. ). On the other hand, halophyte‐based crop rotation has been scarcely implemented (Rabhi et al. ; Barcia‐Piedras et al. ). In addition, the introduction of halophytes into agricultural systems must be economically profitable for farmers, with halophyte desalting ability and biomass valorization being of prime importance in this regard (Ben Hamed et al. ; Hasnain et al. ). Nowadays, different halophytes species are used as animal forage but also as food industry by‐products and sources for medicinal and nutraceutical compounds. Moreover, halophytes such as Salicornia spp., Arthrocaulon macrostachyum L., Cakile maritima , and Capparis spinosa L., among others, contain edible parts appreciated in gourmet cuisine (Barreira et al. ; Hasnain et al. ). In this sense, halophytes are a good source of polyunsaturated fatty acids, antioxidants and minerals (Barreira et al. ). Therefore, there is still much scope for improving knowledge on halophyte integration (cultivation practices and possible product transformation/use) into a sustainable production system. Tomato ( Solanum lycopersium Mill.) is one of the most cultivated vegetable crops in the Mediterranean area. The global tomato production in 2021 was 189 million metric tons, with Mediterranean countries accounting for 20% of the total (source: FAOSTAT 2021). Tomato is classified as moderately sensitive to salinity, which could result in reduced crop yields in soils with an electrical conductivity (EC) over 2.5 dS m −1 (Hanson and May ). In this work, we have applied two crop management strategies, intercropping and crop rotation, between the halophyte A. macrostachyum L., a salt accumulator C3 shrub, and tomato plants in moderately saline conditions, analysing their effect on soil salinity. In tomato plants, we investigated the effect of these crop managements at physiological and biochemical levels (including Na + and Cl − contents, chlorophyll fluorescence, antioxidant metabolism‐related parameters, and hormone profile), and at metabolomic and proteomic levels. Finally, in tomato plants, fruit production and quality were evaluated. MATERIALS AND METHODS 2.1 Plant material and sampling Field trials were conducted in the greenhouse facilities at the Agricultural Demonstration Center “La Pilica” (Aguilas, Murcia, Spain) (37.416253, −1.592437) from 21 st March 2022 to 20 th July 2022. The irrigation water obtained from the desalination plant of Águilas/Guadalentín (Murcia, Spain) had an EC of 0.3–0.5 dS m −1 . Additionally, starting on 4 th April, 4.5 L 30 mM NaCl per plant was added every 14 days. At the onset of trial implementation, concentrations in soil of Na + and Cl − were 2.94 and 1.64 g Kg −1 , respectively. Sixty‐five days old tomato plants (var. Scatolone 2), provided by the farm association “Coáguilas SCL” (Águilas, Murcia, Spain), and four months old A. macrostachyum L. plants, obtained from a local plant nursery (“Viveros Muzalé”, Abanilla, Murcia, Spain), were used. Plots were arranged in a randomised block design with three replicates. The experimental design included three types of plots: tomato in monoculture (T M ), tomato in mixed cultivation with halophyte (T H ), and tomato under crop rotation (T R ). Each plot consisted of a 10 m length row with 13 tomato plants. Additionally, 26 halophyte plants per plot were transplanted either simultaneously in T H – distributed at both sides of the tomato plant – or six months before tomato cultivation in T R . Two drippers per meter were arranged to provide ferti‐irrigation according to the commercial production practices of “Coáguilas SCL”. For the different analyses, fully expanded tomato leaves of the third and fourth nodes from the apex of the main stem were used. Samples were taken before fruit harvesting (June 2022, 68 days after NaCl application started); except for Na + and Cl − levels determination, these samples (leaves, roots and soil) were taken at the end of the experiment (121 days after planting, DAP). For additional analyses of antioxidant metabolism, hormones and ‐omics, plant samples were snap‐frozen in liquid nitrogen and stored at −80°C until use. Tomato fruits were harvested at 86, 105 and 121 DAP. 2.2 Na + and Cl − content Soil samples were taken at 20 to 30 cm depth and at a distance of 20 cm from the closest halophyte and tomato plants by using an auger (5 cm diameter) and dried at room temperature for 48 h. Plant material was properly washed with tap water, followed by three washes with distilled water. Then, plant material was dried at 60°C for 4 days, and the dry material was ground into powder using a mill. Leaf and root powder, as well as soil samples, were filtered through a sieve. Na + and Cl − contents were analyzed as previously described (Jurado et al. ). 2.3 Chlorophyll Fluorescence determination Chlorophyll fluorescence measurements were performed in dark‐adapted leaves between 9:00 and 11:00 hours (GMT), using a portable modulated chlorophyll fluorimeter (FMS2, Hansatech Instruments). The chlorophyll fluorescence parameters [maximum quantum efficiency of PSII photochemistry (Fv/Fm), quantum efficiency of PSII [Y(PSII)], photochemical quenching coefficient (qP), non‐photochemical quenching (NPQ) and its coefficient (qN), and electron transport rate (ETR)] were determined as described in Jurado et al. . 2.4 Antioxidant metabolism‐related parameters Lipid peroxidation was estimated by determining the concentration of thiobarbituric acid‐reactive substances (TBARS), as previously described (Cantabella et al. ). Superoxide anion radical (O 2 .‐) and hydrogen peroxide (H 2 O 2 ) accumulation were performed by incubating tomato leaves with 0.1 mg mL −1 nitroblue tetrazolium and 0.1 mg mL −1 3,3′‐diaminobenzidine, respectively (Hernandez et al. ). Then, chlorophyll was removed by incubating with 70% ethanol at 65°C and photographs were taken with an Olympus BX40 microscope (Olympus Medical Systems Corp.). Enzymatic antioxidants were determined in leaf samples as previously described (Cantabella et al. ; Jurado et al. ). 2.5 Hormone analysis Plant hormone analysis was performed using 30 mg of lyophilized leaf material at the Plant Hormones Quantification Platform (IBMCP, Valencia, Spain). The hormones extraction, analysis [using a Q‐Exactive mass spectrometer (Orbitrap detector; ThermoFisher Scientific)] and quantification were performed as previously described (Hernández et al. , ). 2.6 Metabolomic approach and data analysis Four samples of T M , five samples of T H and five samples of T R were analysed using a metabolomic approach. The sample extraction (70 mg of freeze‐dried material with 50% methanol in a ratio 1/20, w/v ) and non‐target metabolomics analyses (by an ultra‐performance liquid chromatography–quadrupole‐time‐of‐flight mass spectrometry) were performed as previously described (Jurado‐Mañogil et al. ; Barba‐Espín et al. ). Metabolomics data were analysed using the MetaboAnalyst 5.0 software ( https://www.metaboanalyst.ca ) with data subjected to the following normalization: logarithmic (base 10) transformation and Pareto scaling. Then, a Principal Component Analysis (PCA) and a clustering dendrogram analysis were conducted. Putative metabolite identification was done on the top 25 variable features ( m/z ) found by a heatmap analysis ( t ‐test/ANOVA; distance measure, Euclidean; clustering algorithm, Ward), with the identification based on their MS/MS spectra and using the Human Metabolome database ( http://www.hmdb.ca/ ). In addition, a pairwise comparison by Volcano Plot analysis [fold change ≥2 and p ‐value ≤0.05) was also performed. Finally, the Mummichog algorithm (Li et al. ) was used in order to decipher the biological meaning of pairwise metabolomic comparisons (Jurado‐Mañogil et al. ; Barba‐Espín et al. ). 2.7 Shotgun proteomics Twenty milligrams of freeze‐dried leaf sample were extracted in 1.5 mL acetone containing 10% trichloroacetic acid ( w/v ) using a Retch mill, followed by sonication for 10 min and incubation overnight at −20°C. After centrifugation (15,000 g , 5 min), the supernatant was discarded, and the resulting pellet dried under air before resuspension in 800 μL SDS buffer (2% sodium dodecyl sulfate, w/v ; 30% sucrose, w/v ; 5% beta‐mercaptoethanol, v/v ; 5 mM EDTA, 100 mM TRIS, pH 8) for 15 min at 25 °C using an orbital shaker (800 rpm). Upon addition of 400 μL phenol (TRIS saturated), the samples were vortexed and centrifuged (15,000 g , 10 min). The phenolic phase was transferred to a new tube and precipitated overnight with 100 mM ammonium acetate (1.6 mL) dissolved in methanol. The supernatant was removed following centrifugation (5 min, 15,000 g , 4 °C), and the pellet was washed with 80% acetone. The dried pellet was resuspended for 60 min in 200 μL 8 M urea dissolved in 100 mM NH 4 HCO 3 at 25°C using an orbital shaker (600 rpm). Total protein concentration was determined with Bradford assay, and aliquots containing 100 μg protein were transferred to Low Protein Binding tubes. Following cysteine alkylation, the samples were diluted with 50 mM NH 4 HCO 3 containing 2.5% acetonitrile and digested with 1 μg trypsin at 29°C overnight. The samples were desalted using C18 columns and aliquots corresponding to 2.5 μg total peptides were analysed by nanoflow C18 reverse‐phase liquid chromatography using a 15 cm column (Zorbax, Agilent Technologies), a Dionex Ultimate 3000 RSLC nano‐UPLC system (Thermo Fisher Scientific) and Orbitrap Fusion Lumos Tribrid Mass Spectrometer (Thermo Fisher Scientific) as described previously (Hallmark et al. ). The resulting spectra were searched against the reference S. lycopersicum (cv. Heinz 1706) database using Proteome Discoverer 2.4 with the Sequest HT (Thermo Fisher Scientific) and MS Amanda (Dorfer et al. ) search engines using the following parameters: max two missed cleavage sites; modifications‐carbamidomethyl (Cys) and up to three dynamic modifications including Met oxidation, Asn/Gln deamidation, N‐terminal acetylation; MS1 tolerance‐5 ppm (MS Amanda), 10 ppm (Sequest), MS2 tolerance—0.02 Da (MS Amanda), 0.1 Da (Sequest). Proteins with at least two unique peptides were considered for quantitative analysis. 2.8 Tomato fruit production and quality Mature fruits from ten plants per crop management were harvested, and the number of fruits and mean fruit weight per plant were determined. The data of the three harvests – 86, 105 and 121 DAP – were added to calculate the total production per plant. The juice of twenty‐five representative fruits from each crop management and harvest time point was analyzed in a PAL‐BX/ACID refractometer (Atago Co. Ltd.) in order to determine the total soluble solid (TSS, expressed as°Bx) and the acidity (expressed as citric acid equivalents). 2.9 Statistical analysis Except where indicated, at least five biological replicates per treatment were analyzed. Data were subjected to an analysis of variance (ANOVA), followed (when applicable) by Tukey's Multiple Range Test ( p ≤ 0.05), using SPSS® for Windows (IBM Statistics, version 27). Plant material and sampling Field trials were conducted in the greenhouse facilities at the Agricultural Demonstration Center “La Pilica” (Aguilas, Murcia, Spain) (37.416253, −1.592437) from 21 st March 2022 to 20 th July 2022. The irrigation water obtained from the desalination plant of Águilas/Guadalentín (Murcia, Spain) had an EC of 0.3–0.5 dS m −1 . Additionally, starting on 4 th April, 4.5 L 30 mM NaCl per plant was added every 14 days. At the onset of trial implementation, concentrations in soil of Na + and Cl − were 2.94 and 1.64 g Kg −1 , respectively. Sixty‐five days old tomato plants (var. Scatolone 2), provided by the farm association “Coáguilas SCL” (Águilas, Murcia, Spain), and four months old A. macrostachyum L. plants, obtained from a local plant nursery (“Viveros Muzalé”, Abanilla, Murcia, Spain), were used. Plots were arranged in a randomised block design with three replicates. The experimental design included three types of plots: tomato in monoculture (T M ), tomato in mixed cultivation with halophyte (T H ), and tomato under crop rotation (T R ). Each plot consisted of a 10 m length row with 13 tomato plants. Additionally, 26 halophyte plants per plot were transplanted either simultaneously in T H – distributed at both sides of the tomato plant – or six months before tomato cultivation in T R . Two drippers per meter were arranged to provide ferti‐irrigation according to the commercial production practices of “Coáguilas SCL”. For the different analyses, fully expanded tomato leaves of the third and fourth nodes from the apex of the main stem were used. Samples were taken before fruit harvesting (June 2022, 68 days after NaCl application started); except for Na + and Cl − levels determination, these samples (leaves, roots and soil) were taken at the end of the experiment (121 days after planting, DAP). For additional analyses of antioxidant metabolism, hormones and ‐omics, plant samples were snap‐frozen in liquid nitrogen and stored at −80°C until use. Tomato fruits were harvested at 86, 105 and 121 DAP. Na + and Cl − content Soil samples were taken at 20 to 30 cm depth and at a distance of 20 cm from the closest halophyte and tomato plants by using an auger (5 cm diameter) and dried at room temperature for 48 h. Plant material was properly washed with tap water, followed by three washes with distilled water. Then, plant material was dried at 60°C for 4 days, and the dry material was ground into powder using a mill. Leaf and root powder, as well as soil samples, were filtered through a sieve. Na + and Cl − contents were analyzed as previously described (Jurado et al. ). Chlorophyll Fluorescence determination Chlorophyll fluorescence measurements were performed in dark‐adapted leaves between 9:00 and 11:00 hours (GMT), using a portable modulated chlorophyll fluorimeter (FMS2, Hansatech Instruments). The chlorophyll fluorescence parameters [maximum quantum efficiency of PSII photochemistry (Fv/Fm), quantum efficiency of PSII [Y(PSII)], photochemical quenching coefficient (qP), non‐photochemical quenching (NPQ) and its coefficient (qN), and electron transport rate (ETR)] were determined as described in Jurado et al. . Antioxidant metabolism‐related parameters Lipid peroxidation was estimated by determining the concentration of thiobarbituric acid‐reactive substances (TBARS), as previously described (Cantabella et al. ). Superoxide anion radical (O 2 .‐) and hydrogen peroxide (H 2 O 2 ) accumulation were performed by incubating tomato leaves with 0.1 mg mL −1 nitroblue tetrazolium and 0.1 mg mL −1 3,3′‐diaminobenzidine, respectively (Hernandez et al. ). Then, chlorophyll was removed by incubating with 70% ethanol at 65°C and photographs were taken with an Olympus BX40 microscope (Olympus Medical Systems Corp.). Enzymatic antioxidants were determined in leaf samples as previously described (Cantabella et al. ; Jurado et al. ). Hormone analysis Plant hormone analysis was performed using 30 mg of lyophilized leaf material at the Plant Hormones Quantification Platform (IBMCP, Valencia, Spain). The hormones extraction, analysis [using a Q‐Exactive mass spectrometer (Orbitrap detector; ThermoFisher Scientific)] and quantification were performed as previously described (Hernández et al. , ). Metabolomic approach and data analysis Four samples of T M , five samples of T H and five samples of T R were analysed using a metabolomic approach. The sample extraction (70 mg of freeze‐dried material with 50% methanol in a ratio 1/20, w/v ) and non‐target metabolomics analyses (by an ultra‐performance liquid chromatography–quadrupole‐time‐of‐flight mass spectrometry) were performed as previously described (Jurado‐Mañogil et al. ; Barba‐Espín et al. ). Metabolomics data were analysed using the MetaboAnalyst 5.0 software ( https://www.metaboanalyst.ca ) with data subjected to the following normalization: logarithmic (base 10) transformation and Pareto scaling. Then, a Principal Component Analysis (PCA) and a clustering dendrogram analysis were conducted. Putative metabolite identification was done on the top 25 variable features ( m/z ) found by a heatmap analysis ( t ‐test/ANOVA; distance measure, Euclidean; clustering algorithm, Ward), with the identification based on their MS/MS spectra and using the Human Metabolome database ( http://www.hmdb.ca/ ). In addition, a pairwise comparison by Volcano Plot analysis [fold change ≥2 and p ‐value ≤0.05) was also performed. Finally, the Mummichog algorithm (Li et al. ) was used in order to decipher the biological meaning of pairwise metabolomic comparisons (Jurado‐Mañogil et al. ; Barba‐Espín et al. ). Shotgun proteomics Twenty milligrams of freeze‐dried leaf sample were extracted in 1.5 mL acetone containing 10% trichloroacetic acid ( w/v ) using a Retch mill, followed by sonication for 10 min and incubation overnight at −20°C. After centrifugation (15,000 g , 5 min), the supernatant was discarded, and the resulting pellet dried under air before resuspension in 800 μL SDS buffer (2% sodium dodecyl sulfate, w/v ; 30% sucrose, w/v ; 5% beta‐mercaptoethanol, v/v ; 5 mM EDTA, 100 mM TRIS, pH 8) for 15 min at 25 °C using an orbital shaker (800 rpm). Upon addition of 400 μL phenol (TRIS saturated), the samples were vortexed and centrifuged (15,000 g , 10 min). The phenolic phase was transferred to a new tube and precipitated overnight with 100 mM ammonium acetate (1.6 mL) dissolved in methanol. The supernatant was removed following centrifugation (5 min, 15,000 g , 4 °C), and the pellet was washed with 80% acetone. The dried pellet was resuspended for 60 min in 200 μL 8 M urea dissolved in 100 mM NH 4 HCO 3 at 25°C using an orbital shaker (600 rpm). Total protein concentration was determined with Bradford assay, and aliquots containing 100 μg protein were transferred to Low Protein Binding tubes. Following cysteine alkylation, the samples were diluted with 50 mM NH 4 HCO 3 containing 2.5% acetonitrile and digested with 1 μg trypsin at 29°C overnight. The samples were desalted using C18 columns and aliquots corresponding to 2.5 μg total peptides were analysed by nanoflow C18 reverse‐phase liquid chromatography using a 15 cm column (Zorbax, Agilent Technologies), a Dionex Ultimate 3000 RSLC nano‐UPLC system (Thermo Fisher Scientific) and Orbitrap Fusion Lumos Tribrid Mass Spectrometer (Thermo Fisher Scientific) as described previously (Hallmark et al. ). The resulting spectra were searched against the reference S. lycopersicum (cv. Heinz 1706) database using Proteome Discoverer 2.4 with the Sequest HT (Thermo Fisher Scientific) and MS Amanda (Dorfer et al. ) search engines using the following parameters: max two missed cleavage sites; modifications‐carbamidomethyl (Cys) and up to three dynamic modifications including Met oxidation, Asn/Gln deamidation, N‐terminal acetylation; MS1 tolerance‐5 ppm (MS Amanda), 10 ppm (Sequest), MS2 tolerance—0.02 Da (MS Amanda), 0.1 Da (Sequest). Proteins with at least two unique peptides were considered for quantitative analysis. Tomato fruit production and quality Mature fruits from ten plants per crop management were harvested, and the number of fruits and mean fruit weight per plant were determined. The data of the three harvests – 86, 105 and 121 DAP – were added to calculate the total production per plant. The juice of twenty‐five representative fruits from each crop management and harvest time point was analyzed in a PAL‐BX/ACID refractometer (Atago Co. Ltd.) in order to determine the total soluble solid (TSS, expressed as°Bx) and the acidity (expressed as citric acid equivalents). Statistical analysis Except where indicated, at least five biological replicates per treatment were analyzed. Data were subjected to an analysis of variance (ANOVA), followed (when applicable) by Tukey's Multiple Range Test ( p ≤ 0.05), using SPSS® for Windows (IBM Statistics, version 27). RESULTS AND DISCUSSION Salt‐tolerant plants, such as halophytes, can be a feasible alternative for agriculture production in soils affected by salinity and water scarcity and may also provide economic revenue to the farmers. Crop management approaches using halophytes, such as those proposed in this work, aim for salinity alleviation in soils where salt limits crop production (Simpson et al. ). In this work, we used A. macrostachyum , whose desalting capacity and nutritional benefits for human health have been proven (Barreira et al. ; Barcia‐Piedras et al. ). 3.1 Na + and Cl − content and EC analysis At the end of the experiment, samples of leaves and roots of tomato plants and soil were taken, and the levels of Na + and Cl − were determined. T H produced a significant decline in soil Na + content by 35%, whereas a significant decrease in Cl − occurred in T R (Table ). The observed changes in Na + and Cl − led to a decrease in the electrical conductivity of the soil, especially in T H samples (3.95, 2.90 and 3.10 dS m −1 for T M , T H and T R soils, respectively). This observed soil desalting capacity of A. macrostachyum is in agreement with that reported by other authors (Barcia‐Piedras et al. ). Regarding plant tissues, the analysis of variance indicates that crop management significantly affected Na + contents in roots and Cl − levels in leaves and roots. In this sense, the presence of the halophyte plants (T H ) provoked a 25% decline in tomato root Na + , whereas leaf Na + remained statistically unchanged (Table ); together with the lower soil Na + in T H , it can be suggested an enhanced Na + accumulation in the halophyte (data not shown). In tomato leaves, T R contained 44% more Cl − than the other crop managements, while both T H and T R increased root Cl − , especially in T R plants, with a content 62% higher than in T M roots (Table ). This behaviour for Na + and Cl − is in agreement with that observed in tomato var. “Sargento” when cultivated under intercropping with A. macrostachyum ; however, crop rotation led to contrasting results (Jurado et al. ). It is likely that the observed accumulation of Cl − in T H and T R plants contributes to cell osmotic adjustment, improving plant water uptake under saline conditions (Cui et al. ; Munns et al. ). 3.2 Chlorophyll fluorescence Fv/Fm value in dark‐adapted leaves relates to the potential of quantum PSII efficiency, which is being used as an indicator of proper photosynthesis performance. Higher plants present values around 0.80–0.83, whereas lower values can be associated with stress‐induced photo‐inhibition (Maxwell and Johnson ). Although in T R , a slight decrease in the Fv/Fm compared with the other crop managements was observed, under our experimental conditions, Fv/Fm values in the three crop managements were above 0.8 (Supplemental Figure ), which indicates that soil salinity is not affecting photosynthesis significantly. Increased photochemical quenching values and ETR are related to the proper performance of the photosynthetic machinery. On the other hand, non‐photochemical quenching is linked to the safe dissipation of excess light energy (Maxwell and Johnson ), which has also been associated with salt stress response (Acosta‐Motos et al. , ). In this work, crop rotation led to increases in the photochemical [qP, Y(II)] and non‐photochemical (qN, NPQ) quenching parameters, as well as in ETR values, with respect to T M . On the other hand, T H plants only showed significant changes with respect to ETR, the values being intermediate between those observed for T M and T R (Supplemental Figure ). These results are in contrast with those observed in the tomato var. “Sargento”, in which both intercropping and crop rotation with the halophyte increased photochemical and/or non‐photochemical parameters (Jurado‐Mañogil et al. ; Jurado et al. ). These contrasting results suggest that both the experimental conditions and the tomato variety used may influence photosynthesis performance. 3.3 Antioxidant metabolism The extent of lipid peroxidation, as well as the activity of some antioxidant enzymes, were analyzed in leaves. Both halophyte‐based crop managements increased monodehydroascorbate reductase (MDHAR) activity compared to T M plants, reaching this increase 2‐fold in T R plants (Supplemental Figure ). In this regard, overexpression of a MDHAR gene from the halophyte Avicennia marina led to increased MDHAR activity, which in turn enhanced salt tolerance in transgenic tobacco plants (Kavitha et al. ). According to the sum of the MDHAR and dehydroascorbate (DHAR) activities (data not shown), T R leaves presented a higher ascorbate‐recycling activity than the other crop management. Moreover, the contribution of the MDHAR activity to the sum MDHAR + DHAR was above 70% for both halophyte crop managements. This response implies that ascorbate is predominantly recycled by the MDHAR pathway, which is much more efficient energetically than the DHAR pathway. In this regard, it can be suggested that T R plants developed an improved ascorbate recycling ability, as reported recently on the tomato var. “Sargento” (Jurado et al. ). In addition, MDHAR activity has been linked to salinity tolerance in other plant species (Acosta‐Motos et al. , ; Cantabella et al. ). On the other hand, T R produced a remarkable increase in catalase (CAT) activity of about 66% in relation to the other crop managements (Supplemental Figure ), suggesting a more efficient elimination of the H 2 O 2 produced by photorespiration. The involvement of photorespiration in the salt‐tolerance response of higher plants has been suggested by different authors (Corpas et al. ; Acosta‐Motos et al. , ). In addition, a correlation between CAT activity and photosynthesis has been described since the increase in CAT activity reduced the photorespiratory loss of CO 2 (Brisson et al. ). Thus, the increase in CAT activity in T R plants could be related to the observed increase in the photochemical and non‐photochemical quenching parameters (Supplemental Figure ). For the rest of the antioxidant enzymes activities analyzed – ascorbate peroxidase (APX), glutathione reductase, superoxide dismutase (SOD) and peroxidase (POX) – no significant differences were found among crop managements, neither for the levels of lipid peroxidation (data not shown). Concerning histochemical staining (Figure ), O 2 .‐ and H 2 O 2 over‐accumulation in T H and T R leaf tissues suggests the establishment of mild oxidative stress in tomato plants induced by halophyte‐based crop management. The oxidative stress could favour adaptive responses in tomato plants under saline conditions. These accumulation patterns may correlate with the higher CAT and MDHAR activities observed in T R and in T H and T R , respectively, and may support the occurrence of a controlled mild oxidative stress in the halophyte‐mediated crop managements, as previously reported in the tomato var. “Sargento” (Jurado‐Mañogil et al. ; Jurado et al. ). In this sense, it may be suggested that the modulation of the tomato plant physiology can be mediated by allelopathic interactions between the tomato plant and the halophyte, driven by one or more biochemicals produced by the halophyte that influence the physiology of the tomato plant. In this regard, the allelopathic potential of A. macrostachyum has been reported (Mohamed et al. ). In crop rotation management, allelopathy has been widely documented in weed control, by which allelopathic crops release allelochemicals through roots or via the decomposition of crop residue to suppress weeds and other pests (Khamare et al. ). In addition, it can be argued that one of the defence mechanisms associated with salinity tolerance is improved antioxidant properties. Moreover, a strong correlation between the antioxidant activity of plants and their potential allelopathic activity has been found (Aniya et al. ). In this context, we may hypothesize that the interaction of halophyte/tomato is driven by the stimulation of the antioxidant system. 3.4 Hormone profile The introduction of halophytes in the crop system modified the levels of some plant hormones in tomato leaves. Regarding the stress hormones, intercropping slightly increased salicylic acid (SA) content, whereas crop rotation increased jasmonic acid (JA) levels by 50% and 43% compared to monoculture and intercropping, respectively (Figure ). It is well known that SA plays an important role in the response to abiotic stresses such as drought, chilling and saline stress (Takatsuji and Jiang ; Janda et al. ). The application of exogenous SA improved ion homeostasis and photosynthesis rate and decreased H 2 O 2 accumulation in leaves of salt‐stressed Egletes viscosa plants (Batista et al. ). However, the information about the effect of salinity on JA levels is scarce (Acosta‐Motos et al. ). The basal levels of JA in shoots and roots were found to be much higher in a salt‐tolerant tomato variety than in a salt‐sensitive variety, both in the absence and in the presence of NaCl stress (Pedranzani et al. ). In addition, the exogenous application of JA or methyljasmonate to barley plants before salinization improved growth and photosynthetic performance as well as the chlorophyll fluorescence parameters qP and Y(II) (Tsonev et al. ; Sirhindi et al. ). In the present work, we also observed a correlation between JA levels (Figure ) and photosynthesis performance (Figure ) in T R plants. On the other hand, the presence of the halophyte increased SA contents in T H plants regarding T R management (Figure ). It seems that SA and JA have antagonistic interactions, and the SA/JA ratio has been suggested as a salt stress marker (Gupta et al. ; Acosta‐Motos et al. ). A concomitant increase in SA/JA ratio along with salinity levels was reported in myrtle plants (Acosta‐Motos et al. ). In this work, crop rotation led to a lower SA/JA ratio in tomato leaves (data not shown), which could be related to a better plant performance in T R than in the other crop managements. On the other hand, abscisic acid (ABA) levels significantly decreased in T H leaves in relation to the other crop managements (Figure ). ABA is an important player in the response of the plant to stresses, acting as a signalling molecule allowing plants to cope with salinity (Keskin et al. ). The drop in ABA in T H plants was accompanied by an increase in SA. This opposite behaviour between ABA and SA has been previously reported in Brassica napus under stress conditions (Park et al. ). In addition, treatments with SA reduced ABA levels in salt‐stressed maize plants (Elhakem ). Regarding gibberellins, under our experimental conditions, we detected GA 19 , a GA 1 precursor, and GA 4 , the levels of the latter being much lower than those of GA 19 (Figure ). Compared with the stress hormones, the GA contents were very low (Figure ). The GA 19 content decreased in T H and T R in relation to T M , whereas no important changes were observed for GA 4 (Figure ). It has been reported that the Arabidopsis mutant line ga1‐3 , which shows low GA contents, displayed enhanced survival under salt stress (Achard et al. ). As a consequence, a correlation between lower GAs levels and better plant performance of tomato plants under saline conditions can be suggested. Taking together ABA and GAs levels, leaves of T R plants displayed a higher ABA/GAs ratio than the other crop management (data not shown). In this sense, it has been suggested that enhanced salinity tolerance is related to increased ABA/GAs ratio (Colebrook et al. ). 3.5 Metabolomic approach The metabolomic fingerprints in tomato leaf samples taken under different crop management were studied. PCA and clustering dendrogram analysis provided a clear segregation among group samples (T M , T H and T R ) (Figure ). The latter analysis revealed two main clusters, T M and T H samples, being grouped in one cluster (Figure ). These results were somewhat different to those previously reported in the tomato var. “Sargento” under intercropping conditions, in which no segregation was found between monoculture and intercropped tomato samples, although it must be taken into account that in that study, samples from crop rotation were not included (Jurado‐Mañogil et al. ). The MS/MS spectra identification of the top 25 variable mass features displayed by the heatmap analysis (Figure ) revealed that most of the putatively identified metabolites (a total of 13) were lipid and lipid‐related metabolites (Table ). In T H , the accumulation of several fatty acyl and acyl‐lipids was observed (Table ). In plants, acyl‐lipids act mainly as membrane components, carbon and energy storage forms, and signalling transduction players (Li‐Beisson et al. ; Cahoon and Li‐Beisson ). Likewise, as plant cuticle components, acyl‐lipids play important roles in plant water loss prevention and under environmental stress conditions, thus also being important components of plant adaptation mechanisms to the surrounding environment (Li‐Beisson et al. ; Cahoon and Li‐Beisson ). On the other hand, a metabolite identified as diacylglycerol, which is involved in plant lipid metabolism and membrane remodelling (Dong et al. ), was found to be downregulated by T H conditions (Table ). Previously, we described a decrease in diacylglycerol in A. macrostachyum plants grown under intercropping conditions but not in tomato plants (Jurado‐Mañogil et al. ). Intercropping‐induced changes in lipid composition differ depending on the tomato cultivar studied, which may affect lipid metabolism and lipid‐based signalling in different manners (Dong et al. ; Ruelland et al. ; Pohl and Jovanovic ); this may converge in reactive oxygen species (ROS)‐ and lipid‐based signalling pathways interplay (Ruelland et al. ), as suggested by the induction of moderate oxidative stress in the tomato var. “Sargento” (Jurado‐Mañogil et al. ) and in the present study. Among the lipid‐related metabolites, T H and T R leaves accumulated some terpene and terpenoid compounds (Table ). These classes of compounds are present in all living organisms, with plants exhibiting an unusually high number (Pichersky and Raguso ). In plants, terpenoids have myriad functions, playing key roles in biotic and abiotic interactions and in the modulation of ROS signalling (Pichersky and Raguso ; Boncan et al. ). In addition, terpenes and terpenoids have many implications for human health and in different industries (pharmaceutical, food, chemical and biofuel production). Skin UV protection, as well as anti‐tumour, ‐inflammatory and ‐microbial effects, are among the medicinal and nutraceutical properties attributed to terpenes/terpenoids (Jahangeer et al. ). It is well reported that moderate stresses that do not compromise plant cell integrity and functionality increase the content of plant specialized secondary metabolites (da Silva Magedans et al. ). In the present study, tomato plants were subjected to moderate salt stress, so the observed increase in terpenes/terpenoids may be related to salt‐mediated induction of the secondary metabolism rather than an effect of crop management. In contrast, in a previous study, the intercropping between the tomato var. “Sargento” and the halophyte led to a decrease in the content of terpenes/terpenoids (Jurado‐Mañogil et al. ). Moreover, an alteration of antioxidant enzyme activity levels (Figure ) and ROS accumulation (Figure ) by both halophyte‐based crop managements suggests the occurrence of a controlled moderate oxidative stress in tomato plants. Taking into account that, under stress conditions, terpenoids may contribute to ROS signalling by modulating membrane physicochemical properties (Blande et al. ; Sewelam et al. ; Boncan et al. ), a correlation between terpenes/terpenoids and oxidative signalling induced by the crop managements can be suggested. Kukoamine accumulated in T H plants. Kukoamines are dihydrocaffeic acid derivatives of polyamines (putrescine, spermidine and spermine). These specialized secondary metabolites have been found in plant species of the Solanaceae family, including tomato, and have attracted attention due to their bioactive properties (Li et al. ). Different metabolomics studies in plants have revealed that the biosynthesis of polyamines, as well as other compounds such as sugars, aminoacids, organic acids and phytohormones, are commonly altered (Rajkumari et al. ). Moreover, it has been reported that the maintenance of polyamines biosynthesis under salinity conditions contributes to increasing the antioxidant capability as well as to the protection of the photosynthetic machinery from oxidative damage (Ikbal et al. ). The Volcano Plot analysis showed a similar fluctuation in the number of significantly affected metabolites by both intercropping (34 downregulated and 44 upregulated) and crop rotation (33 downregulated and 41 upregulated) conditions compared with plants in monoculture (Figures ,b). However, a greater number of significant differential mass features were observed when T H and T R were compared, with 65 metabolites downregulated and 95 upregulated in the comparison intercropping/crop rotation (Figure ). The Mummichog algorithm (Li et al. ) allowed the identification of metabolic pathways altered by halophyte‐mediated crop management. Compared with control plants, in T H plants, different amino acids‐related pathways were affected (Figure and Table ), whereas the crop rotation led to an alteration of some amino acids metabolism pathways and to an increase in the abundance of different compounds related to sugar metabolism (Figure and Table ). The alteration of sugar metabolism in plants has been described as a common response under different stress conditions, affecting the production of secondary metabolites (Boonchaisri et al. ). Taking into account that photosynthesis and CO 2 fixation are closely related to sugar metabolism, the increase in some compounds related to sugar metabolism in T R may be linked to the increased levels of both photochemical [qP, Y(II)] and non‐photochemical (qN, NPQ) quenching parameters (Figure ). A similar response in terms of better photosynthesis performance and the induction of sugar metabolism was found in the tomato var. “Sargento” under intercropping conditions (Jurado‐Mañogil et al. ). Different sugars accumulate during salinity in both the early and late phases of the stress response, acting as osmolytes or as energy suppliers (Rajkumari et al. ). An increase in glucose, as well as its conversion to other organic compounds, has been described as an important trait in salt‐tolerant rice genotypes (Ma et al. ). Moreover, carbohydrate accumulation affected the activity of different H 2 O 2 ‐related antioxidant enzymes such as SOD, POX and CAT activities in rice (Rahman et al. ). In this sense, in T R plants, the increase in the abundance of sugar metabolism‐related compounds was accompanied by altered levels of antioxidant enzymes, including a strong increase in CAT activity (Figure ). In addition, in the tomato var. “Sargento”, both intercropping and crop rotation conditions increased the activity of the H 2 O 2 ‐scavenging enzyme APX (Jurado et al. ), also displaying in intercropped tomato plants a stimulation of sugar metabolism (Jurado‐Mañogil et al. ). Among the amino acid‐related pathways, “Tyrosine metabolism” was affected in both T H and T R plants (Figures and ; Tables and ). It is important to note that key compounds for plant survival (including tocopherols, plastoquinone, and ubiquinone), as well as a large variety of plant metabolites (including isoquinoline alkaloids), are tyrosine‐derived compounds (Xu et al. ). In transgenic tomato plants expressing a stilbene‐synthase gene, “Tyrosine metabolism” was also affected in both leaves and fruits (Barba‐Espín et al. ). Changes in amino acid accumulation have also been reported in different rice genotypes in response to salt stress, with increases in specific amino acids linked with both cellular signalling and structural processes (Rajkumari et al. ). In addition, the “alpha‐linolenic acid metabolism”, involved in the biosynthesis of the stress‐related phytohormone jasmonic acid, was also stimulated in T H plants (Figures and ; Tables and ), although an increase in JA was only observed in T R leaves (Figure ). In a salt‐tolerant rice variety, an accumulation of JA was also observed under salt stress conditions, this response being correlated with impaired root growth and reduced sodium translocation to shoots as well as with an increased antioxidant capacity (Rajkumari et al. ). The comparison between T H and T R plants allowed the identification of similar affected pathways to those observed in the crop management pairwise comparison with T M , since a stimulation of “alpha‐Linolenic acid metabolism” in T H and of sugar metabolism in T R was observed (Figure and Table ). 3.6 Proteomic approach We queried the effect of the halophyte‐based crop management systems on global protein abundances using a label‐free quantitative shotgun proteomics approach. More than 2,600 proteins characterized by at least two unique peptides were identified and subjected to statistical analysis. Both halophyte‐based crop management systems had a significant impact on the plant proteome in comparison to monoculture conditions (Figure ). While in the metabolomic analysis T M and T H samples match within the same cluster (Figure ), the proteomic analysis revealed that T H and T R proteome profiles were closely related, with similar patterns compared to T M (Figure ). In T H plants 65 proteins accumulated (FC >2, p ≤ 0.05), whereas the abundances of 27 proteins decreased (FC >2, p ≤ 0.05) compared to plants grown in monoculture. Crop rotation resulted in the over‐accumulation of 62 and a decrease of 55 proteins (FC >2, p ≤ 0.05). Most of them were similarly affected by both halophyte‐based crop management practices (Figure ). Among the proteins with increased abundance, Solyc09g083120.3.1 accumulated 6.2‐ and 9.2‐fold in T H and T R , respectively (Tables and ). Intriguingly, its Arabidopsis ortholog ACYLAMINO ACID‐RELEASING ENZYME (AARE) plays an important role in maintaining the cytoplasmic antioxidant system (Nakai et al. ). The accumulation of this protein could be related to the increase in MDHAR and the accumulation of ROS observed in T H and T R , as well as to the high CAT activity recorded in T R leaves (Figure ). Moreover, the suggested occurrence of controlled mild oxidative stress in tomato plants induced by halophyte‐based crop management may trigger the accumulation of AARE. Regarding proteins with decreased expression in both crop managements, we found the F‐box protein Solyc10g076290.2.1, whose levels in T H and T R were repressed 7.2 and 3.7‐fold, respectively (Table ; ). Their two Arabidopsis orthologs, RAE1 and RAH1, regulate the stability of the C2H2‐type zinc finger transcription factor STOP1 via ubiquitin‐directed degradation (Fang et al. ). STOP1 is an important regulator of proton and aluminium rhizotoxicity and is also required for root system architecture alterations in response to nitrate deficiency (Tian et al. ; Tokizawa et al. ). In Arabidopsis plants subjected to salt stress (200 mM NaCl), within the leaf transcriptome analysis, the STOP1 gene was upregulated (Alotaibi and Abulfaraj ); thus, the downregulation of Solyc10g076290.2.1 by T H and T R suggests an amelioration of salt stress by the halophyte‐based crop management. In contrast to proteins whose abundances either increased or decreased in both crop management practices, several proteins were only affected in one of the conditions relative to plants in monoculture . For example, the levels of cryptochrome 1b (Solyc12g057040.2.1) increased 2.2 times in T R compared to T M , whereas its abundance was not affected significantly under intercropping conditions . The orthologs of cryptochrome 1b in Arabidopsis are the blue light photoreceptors CRY1 and CRY2, which are crucial players in plant developmental and stress programs. In Arabidopsis, the overexpression of CRY genes confers hypersensitivity to ABA and salinity (Xu et al. ; Zhou et al. ), whereas the cry1 mutant displayed more salt‐stress tolerance than wild type plants (D'Amico‐Damião and Carvalho ). Similarly, the levels of the chloroplastic protease Do‐like 8 (Solyc02g067360.3.1) increased under crop rotation (FC 4.6), but its abundance in the presence of halophytes was only 1.4‐fold higher relative to T M . Its Arabidopsis ortholog DEG PROTEASE 8 is involved in photosystem II repair through cleavage of photo‐damaged D1 protein (Kato et al. ). The increase in those proteins in T R plants correlated with the recorded enhanced photosynthesis performance (Figure ). 3.7 Production and quality Concerning fruit production and quality, no major differences were observed in crop management. Tomato production was calculated as the aggregate of the three consecutive commercial harvests performed. The kg of tomatoes per plant was statistically equivalent for the different crop management (Figure ). However, the number of tomatoes per plant increased by over 15% in T R with respect to T M plants, whereas in T H plants, this variable remained comparable to T M and T R plants (Figure ). This is in contrast with previous studies in which intercropping with halophytes enhanced production in different crop species such as watermelon (Simpson et al. ), tomato (Zuccarini ; Jurado et al. ), pepper (Colla et al. ) and strawberry (Karakas et al. ). In this study, tomato, a moderately tolerant species to salinity, was cultivated under salinity conditions that did not have a major impact on production. Thus, it can be argued that the degree of soil salinity and the crop sensitivity to salinity are determinants of fruit yield of crops intercropped with halophytes. Finally, in relation to fruit quality, no changes in total soluble solids were registered, while a slight increase in acidity in fruits grown under crop rotation conditions was observed (Figure – D). Na + and Cl − content and EC analysis At the end of the experiment, samples of leaves and roots of tomato plants and soil were taken, and the levels of Na + and Cl − were determined. T H produced a significant decline in soil Na + content by 35%, whereas a significant decrease in Cl − occurred in T R (Table ). The observed changes in Na + and Cl − led to a decrease in the electrical conductivity of the soil, especially in T H samples (3.95, 2.90 and 3.10 dS m −1 for T M , T H and T R soils, respectively). This observed soil desalting capacity of A. macrostachyum is in agreement with that reported by other authors (Barcia‐Piedras et al. ). Regarding plant tissues, the analysis of variance indicates that crop management significantly affected Na + contents in roots and Cl − levels in leaves and roots. In this sense, the presence of the halophyte plants (T H ) provoked a 25% decline in tomato root Na + , whereas leaf Na + remained statistically unchanged (Table ); together with the lower soil Na + in T H , it can be suggested an enhanced Na + accumulation in the halophyte (data not shown). In tomato leaves, T R contained 44% more Cl − than the other crop managements, while both T H and T R increased root Cl − , especially in T R plants, with a content 62% higher than in T M roots (Table ). This behaviour for Na + and Cl − is in agreement with that observed in tomato var. “Sargento” when cultivated under intercropping with A. macrostachyum ; however, crop rotation led to contrasting results (Jurado et al. ). It is likely that the observed accumulation of Cl − in T H and T R plants contributes to cell osmotic adjustment, improving plant water uptake under saline conditions (Cui et al. ; Munns et al. ). Chlorophyll fluorescence Fv/Fm value in dark‐adapted leaves relates to the potential of quantum PSII efficiency, which is being used as an indicator of proper photosynthesis performance. Higher plants present values around 0.80–0.83, whereas lower values can be associated with stress‐induced photo‐inhibition (Maxwell and Johnson ). Although in T R , a slight decrease in the Fv/Fm compared with the other crop managements was observed, under our experimental conditions, Fv/Fm values in the three crop managements were above 0.8 (Supplemental Figure ), which indicates that soil salinity is not affecting photosynthesis significantly. Increased photochemical quenching values and ETR are related to the proper performance of the photosynthetic machinery. On the other hand, non‐photochemical quenching is linked to the safe dissipation of excess light energy (Maxwell and Johnson ), which has also been associated with salt stress response (Acosta‐Motos et al. , ). In this work, crop rotation led to increases in the photochemical [qP, Y(II)] and non‐photochemical (qN, NPQ) quenching parameters, as well as in ETR values, with respect to T M . On the other hand, T H plants only showed significant changes with respect to ETR, the values being intermediate between those observed for T M and T R (Supplemental Figure ). These results are in contrast with those observed in the tomato var. “Sargento”, in which both intercropping and crop rotation with the halophyte increased photochemical and/or non‐photochemical parameters (Jurado‐Mañogil et al. ; Jurado et al. ). These contrasting results suggest that both the experimental conditions and the tomato variety used may influence photosynthesis performance. Antioxidant metabolism The extent of lipid peroxidation, as well as the activity of some antioxidant enzymes, were analyzed in leaves. Both halophyte‐based crop managements increased monodehydroascorbate reductase (MDHAR) activity compared to T M plants, reaching this increase 2‐fold in T R plants (Supplemental Figure ). In this regard, overexpression of a MDHAR gene from the halophyte Avicennia marina led to increased MDHAR activity, which in turn enhanced salt tolerance in transgenic tobacco plants (Kavitha et al. ). According to the sum of the MDHAR and dehydroascorbate (DHAR) activities (data not shown), T R leaves presented a higher ascorbate‐recycling activity than the other crop management. Moreover, the contribution of the MDHAR activity to the sum MDHAR + DHAR was above 70% for both halophyte crop managements. This response implies that ascorbate is predominantly recycled by the MDHAR pathway, which is much more efficient energetically than the DHAR pathway. In this regard, it can be suggested that T R plants developed an improved ascorbate recycling ability, as reported recently on the tomato var. “Sargento” (Jurado et al. ). In addition, MDHAR activity has been linked to salinity tolerance in other plant species (Acosta‐Motos et al. , ; Cantabella et al. ). On the other hand, T R produced a remarkable increase in catalase (CAT) activity of about 66% in relation to the other crop managements (Supplemental Figure ), suggesting a more efficient elimination of the H 2 O 2 produced by photorespiration. The involvement of photorespiration in the salt‐tolerance response of higher plants has been suggested by different authors (Corpas et al. ; Acosta‐Motos et al. , ). In addition, a correlation between CAT activity and photosynthesis has been described since the increase in CAT activity reduced the photorespiratory loss of CO 2 (Brisson et al. ). Thus, the increase in CAT activity in T R plants could be related to the observed increase in the photochemical and non‐photochemical quenching parameters (Supplemental Figure ). For the rest of the antioxidant enzymes activities analyzed – ascorbate peroxidase (APX), glutathione reductase, superoxide dismutase (SOD) and peroxidase (POX) – no significant differences were found among crop managements, neither for the levels of lipid peroxidation (data not shown). Concerning histochemical staining (Figure ), O 2 .‐ and H 2 O 2 over‐accumulation in T H and T R leaf tissues suggests the establishment of mild oxidative stress in tomato plants induced by halophyte‐based crop management. The oxidative stress could favour adaptive responses in tomato plants under saline conditions. These accumulation patterns may correlate with the higher CAT and MDHAR activities observed in T R and in T H and T R , respectively, and may support the occurrence of a controlled mild oxidative stress in the halophyte‐mediated crop managements, as previously reported in the tomato var. “Sargento” (Jurado‐Mañogil et al. ; Jurado et al. ). In this sense, it may be suggested that the modulation of the tomato plant physiology can be mediated by allelopathic interactions between the tomato plant and the halophyte, driven by one or more biochemicals produced by the halophyte that influence the physiology of the tomato plant. In this regard, the allelopathic potential of A. macrostachyum has been reported (Mohamed et al. ). In crop rotation management, allelopathy has been widely documented in weed control, by which allelopathic crops release allelochemicals through roots or via the decomposition of crop residue to suppress weeds and other pests (Khamare et al. ). In addition, it can be argued that one of the defence mechanisms associated with salinity tolerance is improved antioxidant properties. Moreover, a strong correlation between the antioxidant activity of plants and their potential allelopathic activity has been found (Aniya et al. ). In this context, we may hypothesize that the interaction of halophyte/tomato is driven by the stimulation of the antioxidant system. Hormone profile The introduction of halophytes in the crop system modified the levels of some plant hormones in tomato leaves. Regarding the stress hormones, intercropping slightly increased salicylic acid (SA) content, whereas crop rotation increased jasmonic acid (JA) levels by 50% and 43% compared to monoculture and intercropping, respectively (Figure ). It is well known that SA plays an important role in the response to abiotic stresses such as drought, chilling and saline stress (Takatsuji and Jiang ; Janda et al. ). The application of exogenous SA improved ion homeostasis and photosynthesis rate and decreased H 2 O 2 accumulation in leaves of salt‐stressed Egletes viscosa plants (Batista et al. ). However, the information about the effect of salinity on JA levels is scarce (Acosta‐Motos et al. ). The basal levels of JA in shoots and roots were found to be much higher in a salt‐tolerant tomato variety than in a salt‐sensitive variety, both in the absence and in the presence of NaCl stress (Pedranzani et al. ). In addition, the exogenous application of JA or methyljasmonate to barley plants before salinization improved growth and photosynthetic performance as well as the chlorophyll fluorescence parameters qP and Y(II) (Tsonev et al. ; Sirhindi et al. ). In the present work, we also observed a correlation between JA levels (Figure ) and photosynthesis performance (Figure ) in T R plants. On the other hand, the presence of the halophyte increased SA contents in T H plants regarding T R management (Figure ). It seems that SA and JA have antagonistic interactions, and the SA/JA ratio has been suggested as a salt stress marker (Gupta et al. ; Acosta‐Motos et al. ). A concomitant increase in SA/JA ratio along with salinity levels was reported in myrtle plants (Acosta‐Motos et al. ). In this work, crop rotation led to a lower SA/JA ratio in tomato leaves (data not shown), which could be related to a better plant performance in T R than in the other crop managements. On the other hand, abscisic acid (ABA) levels significantly decreased in T H leaves in relation to the other crop managements (Figure ). ABA is an important player in the response of the plant to stresses, acting as a signalling molecule allowing plants to cope with salinity (Keskin et al. ). The drop in ABA in T H plants was accompanied by an increase in SA. This opposite behaviour between ABA and SA has been previously reported in Brassica napus under stress conditions (Park et al. ). In addition, treatments with SA reduced ABA levels in salt‐stressed maize plants (Elhakem ). Regarding gibberellins, under our experimental conditions, we detected GA 19 , a GA 1 precursor, and GA 4 , the levels of the latter being much lower than those of GA 19 (Figure ). Compared with the stress hormones, the GA contents were very low (Figure ). The GA 19 content decreased in T H and T R in relation to T M , whereas no important changes were observed for GA 4 (Figure ). It has been reported that the Arabidopsis mutant line ga1‐3 , which shows low GA contents, displayed enhanced survival under salt stress (Achard et al. ). As a consequence, a correlation between lower GAs levels and better plant performance of tomato plants under saline conditions can be suggested. Taking together ABA and GAs levels, leaves of T R plants displayed a higher ABA/GAs ratio than the other crop management (data not shown). In this sense, it has been suggested that enhanced salinity tolerance is related to increased ABA/GAs ratio (Colebrook et al. ). Metabolomic approach The metabolomic fingerprints in tomato leaf samples taken under different crop management were studied. PCA and clustering dendrogram analysis provided a clear segregation among group samples (T M , T H and T R ) (Figure ). The latter analysis revealed two main clusters, T M and T H samples, being grouped in one cluster (Figure ). These results were somewhat different to those previously reported in the tomato var. “Sargento” under intercropping conditions, in which no segregation was found between monoculture and intercropped tomato samples, although it must be taken into account that in that study, samples from crop rotation were not included (Jurado‐Mañogil et al. ). The MS/MS spectra identification of the top 25 variable mass features displayed by the heatmap analysis (Figure ) revealed that most of the putatively identified metabolites (a total of 13) were lipid and lipid‐related metabolites (Table ). In T H , the accumulation of several fatty acyl and acyl‐lipids was observed (Table ). In plants, acyl‐lipids act mainly as membrane components, carbon and energy storage forms, and signalling transduction players (Li‐Beisson et al. ; Cahoon and Li‐Beisson ). Likewise, as plant cuticle components, acyl‐lipids play important roles in plant water loss prevention and under environmental stress conditions, thus also being important components of plant adaptation mechanisms to the surrounding environment (Li‐Beisson et al. ; Cahoon and Li‐Beisson ). On the other hand, a metabolite identified as diacylglycerol, which is involved in plant lipid metabolism and membrane remodelling (Dong et al. ), was found to be downregulated by T H conditions (Table ). Previously, we described a decrease in diacylglycerol in A. macrostachyum plants grown under intercropping conditions but not in tomato plants (Jurado‐Mañogil et al. ). Intercropping‐induced changes in lipid composition differ depending on the tomato cultivar studied, which may affect lipid metabolism and lipid‐based signalling in different manners (Dong et al. ; Ruelland et al. ; Pohl and Jovanovic ); this may converge in reactive oxygen species (ROS)‐ and lipid‐based signalling pathways interplay (Ruelland et al. ), as suggested by the induction of moderate oxidative stress in the tomato var. “Sargento” (Jurado‐Mañogil et al. ) and in the present study. Among the lipid‐related metabolites, T H and T R leaves accumulated some terpene and terpenoid compounds (Table ). These classes of compounds are present in all living organisms, with plants exhibiting an unusually high number (Pichersky and Raguso ). In plants, terpenoids have myriad functions, playing key roles in biotic and abiotic interactions and in the modulation of ROS signalling (Pichersky and Raguso ; Boncan et al. ). In addition, terpenes and terpenoids have many implications for human health and in different industries (pharmaceutical, food, chemical and biofuel production). Skin UV protection, as well as anti‐tumour, ‐inflammatory and ‐microbial effects, are among the medicinal and nutraceutical properties attributed to terpenes/terpenoids (Jahangeer et al. ). It is well reported that moderate stresses that do not compromise plant cell integrity and functionality increase the content of plant specialized secondary metabolites (da Silva Magedans et al. ). In the present study, tomato plants were subjected to moderate salt stress, so the observed increase in terpenes/terpenoids may be related to salt‐mediated induction of the secondary metabolism rather than an effect of crop management. In contrast, in a previous study, the intercropping between the tomato var. “Sargento” and the halophyte led to a decrease in the content of terpenes/terpenoids (Jurado‐Mañogil et al. ). Moreover, an alteration of antioxidant enzyme activity levels (Figure ) and ROS accumulation (Figure ) by both halophyte‐based crop managements suggests the occurrence of a controlled moderate oxidative stress in tomato plants. Taking into account that, under stress conditions, terpenoids may contribute to ROS signalling by modulating membrane physicochemical properties (Blande et al. ; Sewelam et al. ; Boncan et al. ), a correlation between terpenes/terpenoids and oxidative signalling induced by the crop managements can be suggested. Kukoamine accumulated in T H plants. Kukoamines are dihydrocaffeic acid derivatives of polyamines (putrescine, spermidine and spermine). These specialized secondary metabolites have been found in plant species of the Solanaceae family, including tomato, and have attracted attention due to their bioactive properties (Li et al. ). Different metabolomics studies in plants have revealed that the biosynthesis of polyamines, as well as other compounds such as sugars, aminoacids, organic acids and phytohormones, are commonly altered (Rajkumari et al. ). Moreover, it has been reported that the maintenance of polyamines biosynthesis under salinity conditions contributes to increasing the antioxidant capability as well as to the protection of the photosynthetic machinery from oxidative damage (Ikbal et al. ). The Volcano Plot analysis showed a similar fluctuation in the number of significantly affected metabolites by both intercropping (34 downregulated and 44 upregulated) and crop rotation (33 downregulated and 41 upregulated) conditions compared with plants in monoculture (Figures ,b). However, a greater number of significant differential mass features were observed when T H and T R were compared, with 65 metabolites downregulated and 95 upregulated in the comparison intercropping/crop rotation (Figure ). The Mummichog algorithm (Li et al. ) allowed the identification of metabolic pathways altered by halophyte‐mediated crop management. Compared with control plants, in T H plants, different amino acids‐related pathways were affected (Figure and Table ), whereas the crop rotation led to an alteration of some amino acids metabolism pathways and to an increase in the abundance of different compounds related to sugar metabolism (Figure and Table ). The alteration of sugar metabolism in plants has been described as a common response under different stress conditions, affecting the production of secondary metabolites (Boonchaisri et al. ). Taking into account that photosynthesis and CO 2 fixation are closely related to sugar metabolism, the increase in some compounds related to sugar metabolism in T R may be linked to the increased levels of both photochemical [qP, Y(II)] and non‐photochemical (qN, NPQ) quenching parameters (Figure ). A similar response in terms of better photosynthesis performance and the induction of sugar metabolism was found in the tomato var. “Sargento” under intercropping conditions (Jurado‐Mañogil et al. ). Different sugars accumulate during salinity in both the early and late phases of the stress response, acting as osmolytes or as energy suppliers (Rajkumari et al. ). An increase in glucose, as well as its conversion to other organic compounds, has been described as an important trait in salt‐tolerant rice genotypes (Ma et al. ). Moreover, carbohydrate accumulation affected the activity of different H 2 O 2 ‐related antioxidant enzymes such as SOD, POX and CAT activities in rice (Rahman et al. ). In this sense, in T R plants, the increase in the abundance of sugar metabolism‐related compounds was accompanied by altered levels of antioxidant enzymes, including a strong increase in CAT activity (Figure ). In addition, in the tomato var. “Sargento”, both intercropping and crop rotation conditions increased the activity of the H 2 O 2 ‐scavenging enzyme APX (Jurado et al. ), also displaying in intercropped tomato plants a stimulation of sugar metabolism (Jurado‐Mañogil et al. ). Among the amino acid‐related pathways, “Tyrosine metabolism” was affected in both T H and T R plants (Figures and ; Tables and ). It is important to note that key compounds for plant survival (including tocopherols, plastoquinone, and ubiquinone), as well as a large variety of plant metabolites (including isoquinoline alkaloids), are tyrosine‐derived compounds (Xu et al. ). In transgenic tomato plants expressing a stilbene‐synthase gene, “Tyrosine metabolism” was also affected in both leaves and fruits (Barba‐Espín et al. ). Changes in amino acid accumulation have also been reported in different rice genotypes in response to salt stress, with increases in specific amino acids linked with both cellular signalling and structural processes (Rajkumari et al. ). In addition, the “alpha‐linolenic acid metabolism”, involved in the biosynthesis of the stress‐related phytohormone jasmonic acid, was also stimulated in T H plants (Figures and ; Tables and ), although an increase in JA was only observed in T R leaves (Figure ). In a salt‐tolerant rice variety, an accumulation of JA was also observed under salt stress conditions, this response being correlated with impaired root growth and reduced sodium translocation to shoots as well as with an increased antioxidant capacity (Rajkumari et al. ). The comparison between T H and T R plants allowed the identification of similar affected pathways to those observed in the crop management pairwise comparison with T M , since a stimulation of “alpha‐Linolenic acid metabolism” in T H and of sugar metabolism in T R was observed (Figure and Table ). Proteomic approach We queried the effect of the halophyte‐based crop management systems on global protein abundances using a label‐free quantitative shotgun proteomics approach. More than 2,600 proteins characterized by at least two unique peptides were identified and subjected to statistical analysis. Both halophyte‐based crop management systems had a significant impact on the plant proteome in comparison to monoculture conditions (Figure ). While in the metabolomic analysis T M and T H samples match within the same cluster (Figure ), the proteomic analysis revealed that T H and T R proteome profiles were closely related, with similar patterns compared to T M (Figure ). In T H plants 65 proteins accumulated (FC >2, p ≤ 0.05), whereas the abundances of 27 proteins decreased (FC >2, p ≤ 0.05) compared to plants grown in monoculture. Crop rotation resulted in the over‐accumulation of 62 and a decrease of 55 proteins (FC >2, p ≤ 0.05). Most of them were similarly affected by both halophyte‐based crop management practices (Figure ). Among the proteins with increased abundance, Solyc09g083120.3.1 accumulated 6.2‐ and 9.2‐fold in T H and T R , respectively (Tables and ). Intriguingly, its Arabidopsis ortholog ACYLAMINO ACID‐RELEASING ENZYME (AARE) plays an important role in maintaining the cytoplasmic antioxidant system (Nakai et al. ). The accumulation of this protein could be related to the increase in MDHAR and the accumulation of ROS observed in T H and T R , as well as to the high CAT activity recorded in T R leaves (Figure ). Moreover, the suggested occurrence of controlled mild oxidative stress in tomato plants induced by halophyte‐based crop management may trigger the accumulation of AARE. Regarding proteins with decreased expression in both crop managements, we found the F‐box protein Solyc10g076290.2.1, whose levels in T H and T R were repressed 7.2 and 3.7‐fold, respectively (Table ; ). Their two Arabidopsis orthologs, RAE1 and RAH1, regulate the stability of the C2H2‐type zinc finger transcription factor STOP1 via ubiquitin‐directed degradation (Fang et al. ). STOP1 is an important regulator of proton and aluminium rhizotoxicity and is also required for root system architecture alterations in response to nitrate deficiency (Tian et al. ; Tokizawa et al. ). In Arabidopsis plants subjected to salt stress (200 mM NaCl), within the leaf transcriptome analysis, the STOP1 gene was upregulated (Alotaibi and Abulfaraj ); thus, the downregulation of Solyc10g076290.2.1 by T H and T R suggests an amelioration of salt stress by the halophyte‐based crop management. In contrast to proteins whose abundances either increased or decreased in both crop management practices, several proteins were only affected in one of the conditions relative to plants in monoculture . For example, the levels of cryptochrome 1b (Solyc12g057040.2.1) increased 2.2 times in T R compared to T M , whereas its abundance was not affected significantly under intercropping conditions . The orthologs of cryptochrome 1b in Arabidopsis are the blue light photoreceptors CRY1 and CRY2, which are crucial players in plant developmental and stress programs. In Arabidopsis, the overexpression of CRY genes confers hypersensitivity to ABA and salinity (Xu et al. ; Zhou et al. ), whereas the cry1 mutant displayed more salt‐stress tolerance than wild type plants (D'Amico‐Damião and Carvalho ). Similarly, the levels of the chloroplastic protease Do‐like 8 (Solyc02g067360.3.1) increased under crop rotation (FC 4.6), but its abundance in the presence of halophytes was only 1.4‐fold higher relative to T M . Its Arabidopsis ortholog DEG PROTEASE 8 is involved in photosystem II repair through cleavage of photo‐damaged D1 protein (Kato et al. ). The increase in those proteins in T R plants correlated with the recorded enhanced photosynthesis performance (Figure ). Production and quality Concerning fruit production and quality, no major differences were observed in crop management. Tomato production was calculated as the aggregate of the three consecutive commercial harvests performed. The kg of tomatoes per plant was statistically equivalent for the different crop management (Figure ). However, the number of tomatoes per plant increased by over 15% in T R with respect to T M plants, whereas in T H plants, this variable remained comparable to T M and T R plants (Figure ). This is in contrast with previous studies in which intercropping with halophytes enhanced production in different crop species such as watermelon (Simpson et al. ), tomato (Zuccarini ; Jurado et al. ), pepper (Colla et al. ) and strawberry (Karakas et al. ). In this study, tomato, a moderately tolerant species to salinity, was cultivated under salinity conditions that did not have a major impact on production. Thus, it can be argued that the degree of soil salinity and the crop sensitivity to salinity are determinants of fruit yield of crops intercropped with halophytes. Finally, in relation to fruit quality, no changes in total soluble solids were registered, while a slight increase in acidity in fruits grown under crop rotation conditions was observed (Figure – D). CONCLUSIONS Halophyte‐based crop management may become a relevant component of farming systems, favouring cash crop productivity in saline environments while providing a high added‐value product. Figure summarizes the main changes observed in the tomato plants under the halophyte‐based crop management conditions in comparison to tomato plants grown in monoculture. Both halophyte‐based management strategies reduced the levels of Na + and Cl − in soil, whereas solely intercropping conditions led to a decline in tomato root Na + . Crop rotation enhanced photosynthesis and protective mechanisms at the photosynthetic level. In addition, both crop managements affected the hormone profile and the antioxidant capacity of tomato leaves, whereas a reactive oxygen species over‐accumulation in leaf tissues suggests the establishment of a controlled moderate oxidative stress in leaves. Metabolomic and proteomic analyses indicate intricate relationships occurring at the leaf level, influenced by the presence of the halophyte. These interactions suggest a connection between ROS/lipid‐based signalling pathways. Additionally, enhanced photosynthesis due to crop rotation was linked to the accumulation of metabolites associated with sugar metabolism and proteins related to photosynthesis. This work brings novelty to the interactions between halophyte and tomato in co‐cultivation strategies and supports the use of A. macrostachyum in farming systems. Conceptualization, G.B.‐E., J.A.H., and P.D.‐V.; methodology, G.B.‐E., C.J.‐M., Z.P., P.I.K. and P.D.‐V; formal analysis, G.B.‐E., C.J.‐M., J.A.H., and P.D.‐V.; investigation, G.B.‐E., J.A.H., and P.D.‐V.; data curation, G.B.‐E., P.I.K., P.D‐V.; writing—original draft preparation, G.B.‐E., C.J.‐M., P.I.K., J.A.H., and P.D.‐V.; writing—review and editing, G.B.‐E., J.A.H., and P.D.‐V. The authors declare no conflict of interest. Figure S1. Effect of halophyte‐based crop managements on chlorophyll fluorescence parameters. Maximum quantum efficiency of PSII photochemistry, Fv/Fm; quantum efficiency of PSII, Y(PSII); photochemical quenching coefficient, qP; non‐photochemical quenching parameters, qN and NPQ; electron transport rate, ETR. Different letters indicate significant differences according the Tukey's Multiple Range Test ( p ≤ 0.05). T M : tomato in monoculture; T H : tomato in intercropping; T R : tomato under crop rotation. Bars show the mean of at least 10 biological samples and the standard error. Figure S2 . Effect of halophyte‐based crop managements on catalase (CAT) and monodehydroascorbate reductase (MDHAR) activities in leaves of tomato plants. Different letters indicate significant differences according the Tukey's Multiple Range Test ( p ≤ 0.05). T M : tomato in monoculture; T H : tomato in intercropping; T R : tomato under crop rotation. Bars show the mean of at least 4 biological samples and the standard error. Figure S3 . Volcano Plot analysis (fold change 2, p ≤ 0.05) of tomato leaf samples. Pairwise comparisons: A) intercropping / monoculture; B) crop rotation / monoculture; C) intercropping / crop rotation). Blue, downregulated; grey, unchanged; red, upregulated. Figure S4 . Effect of the halophyte‐based crop management on the production and quality of tomato fruits. a) Kg of tomatoes per plant; b) number of tomatoes per plant; D) total soluble solids (TSS); E) acidity. Different letters indicate significant differences according the Tukey's Multiple Range Test ( p ≤ 0.05). T M : tomato in monoculture; T H : tomato in intercropping; T R : tomato under crop rotation. Data S1 : File S1. Relative abundance and fold change (FC) of differentially regulated proteins. |
Training in obstetrics and gynecology between reality and vision: results of a JAGO–NOGGO survey in 601 physicians (NOGGO—Monitor-12 trial) | a0f6df61-24ea-4748-9832-c5a070d32c67 | 11147899 | Gynaecology[mh] | Quality of training is the backbone for medical knowledge transfer and ongoing professional development in clinical routine and residency programs. However, medical education faces significant challenges due to the exponential growth of medical knowledge and the resulting complexity of surgical and medical procedures. These challenges are further compounded by the broad spectrum of Obstetrics and Gynecology (OBGYN) and its subdisciplines, making it difficult to provide in-depth training to all residents in each area. Given that surgeries are a major component of clinical practice, the acquisition of surgical skills is a fundamental aspect of OBGYN training. Currently, the standard training model for gynecologic surgery involves integrating residents into surgical teams and providing guidance by experienced colleagues. However, this educational system is constantly challenged by cost pressures, limited resources in healthcare systems, high workloads in clinical routine, and individual factors. To standardize training, in Germany, the training of OBGYN residents follows a 5-year curriculum that is comparable to international practice . This curriculum is structured by further training regulations from the German Medical Association (Bundesärztekammer), which includes a logbook to document particular surgical skills and a minimum number of surgical interventions. The current required numbers vary between the federal German states, but usually comprise about 100–200 minor gynecologic surgeries, 50–100 major gynecologic surgeries, and 25 operative interventions in obstetrics . However, despite its importance, there is a lack of actual data on the current quality of training in clinical routine . The primary objective of this study was to establish a benchmark by collecting baseline data on surgical education in OBGYN in Germany, including the factual number of operations performed. Secondary objectives were to identify challenges and opportunities for improvement in the field of surgical training in OBGYN and to gather diverse perspectives and appropriate suggestions from trainers and trainees in this area. Design and participants This national anonymous online survey was conducted from 24.01.2019 to 10.07.2019 by the Young Academy of Gynecologic Oncology (Junge Akademie Gynäkologische Onkologie—JAGO) of the North-Eastern German Society of Gynecologic Oncology (Die Nord-Ostdeutsche Gesellschaft für Gynäkologische Onkologie—NOGGO), with support from the Working Group of Gynecologic Oncology of the German Society for Gynecology and Obstetrics and the German Cancer Society (Arbeitsgemeinschaft Gynäkologische Onkologie—AGO). The study was approved by the local ethics committee at the Charité University Hospital in Berlin (local reference EA1/042/23). The survey was developed as part of a scientific and clinical fellowship program of the JAGO. During a five-part modular workshop, the fellows conducted a comprehensive literature review and developed a questionnaire under the advice of interprofessional and interdisciplinary experts. The survey was designed for both gynecologists in training and those responsible for training more junior colleagues. The questionnaire was distributed with the use of institutional mailing lists, such as those of the NOGGO and the newsletter of the Young Forum of the German Society for Gynecology and Obstetrics (DGGG). The data were collected via the web-based Surveymonkey © software. Questionnaire The questionnaire consisted of 30 items for trainees and 24 items for trainers, with the option to skip individual questions. It included 11 trainee-specific questions, 5 trainer-specific questions, and 19 questions that were common for both trainers and trainees. The quality and importance of surgical training were measured using rating scales, while the exact number of surgeries was collected through free-text responses. Dichotomous questions were used to determine whether there were difficulties with staffing or manipulation of logbook figures and whether annual training review meetings were held. Questions that allowed for multiple answers and free-text responses were used to evaluate current support, identify difficulties, and suggest improvements. Data analysis Categorical variables are presented as numbers and percentages. Unless otherwise specified, continuous variables are presented as median and interquartile range. Comparisons between trainees and trainers were performed using an unpaired t test. A subgroup analysis was conducted for independently performed surgeries by residents in their last year of training (5th year). P values less than 0.05 were considered significant. The analysis was performed using Microsoft Excel 2022® (Microsoft Corp./USA) and Prism 9.0® (GraphPad Software, Inc./USA). This national anonymous online survey was conducted from 24.01.2019 to 10.07.2019 by the Young Academy of Gynecologic Oncology (Junge Akademie Gynäkologische Onkologie—JAGO) of the North-Eastern German Society of Gynecologic Oncology (Die Nord-Ostdeutsche Gesellschaft für Gynäkologische Onkologie—NOGGO), with support from the Working Group of Gynecologic Oncology of the German Society for Gynecology and Obstetrics and the German Cancer Society (Arbeitsgemeinschaft Gynäkologische Onkologie—AGO). The study was approved by the local ethics committee at the Charité University Hospital in Berlin (local reference EA1/042/23). The survey was developed as part of a scientific and clinical fellowship program of the JAGO. During a five-part modular workshop, the fellows conducted a comprehensive literature review and developed a questionnaire under the advice of interprofessional and interdisciplinary experts. The survey was designed for both gynecologists in training and those responsible for training more junior colleagues. The questionnaire was distributed with the use of institutional mailing lists, such as those of the NOGGO and the newsletter of the Young Forum of the German Society for Gynecology and Obstetrics (DGGG). The data were collected via the web-based Surveymonkey © software. The questionnaire consisted of 30 items for trainees and 24 items for trainers, with the option to skip individual questions. It included 11 trainee-specific questions, 5 trainer-specific questions, and 19 questions that were common for both trainers and trainees. The quality and importance of surgical training were measured using rating scales, while the exact number of surgeries was collected through free-text responses. Dichotomous questions were used to determine whether there were difficulties with staffing or manipulation of logbook figures and whether annual training review meetings were held. Questions that allowed for multiple answers and free-text responses were used to evaluate current support, identify difficulties, and suggest improvements. Categorical variables are presented as numbers and percentages. Unless otherwise specified, continuous variables are presented as median and interquartile range. Comparisons between trainees and trainers were performed using an unpaired t test. A subgroup analysis was conducted for independently performed surgeries by residents in their last year of training (5th year). P values less than 0.05 were considered significant. The analysis was performed using Microsoft Excel 2022® (Microsoft Corp./USA) and Prism 9.0® (GraphPad Software, Inc./USA). Study population A total of 601 participants completed the survey, with 305 trainees and 296 trainers. Participants from all 16 German federal states were represented. General characteristics of the participants are presented in Table . The majority of participants worked in maximum care or university hospitals, that were authorized for full residency training. Most of institutions were certified as gynecologic and breast oncologic centers. Only a minority worked in offices (Table ). The majority of trainees were in their third to fifth year of training. Most of the trainers were consultants (38.18%, n = 105) or senior consultants (22.91%, n = 63), followed by chief physicians (20.73%, n = 57) and specialists (15.25%, n = 42). More than half of the trainers (58.03%, n = 172) had one or more subspecialties, including Gynecological Oncology (42.34%, n = 116), Fetal Medicine (23.72%, n = 65), or Gynecological Endocrinology and Reproductive Medicine (5.11%, n = 14). Surgery figures of trainees The trainees performed a median of 125 (IQR: 41–332) non-obstetric surgeries and 75 (IQR: 27–168) obstetric surgeries independently. Detailed figures of the surgeries performed are presented in Table . The majority of non-obstetric surgeries performed by the trainees were hysteroscopies and small vaginal procedures, such as dilation & curettage or conisations. Major procedures, such as type-III or type-IV laparoscopy or hysterectomy (open/vaginal), were carried out independently by the trainees to a very limited extent. Breast surgery was also rarely performed independently by trainees. The subgroup analysis of residents with four or more completed years of training showed a median of 277 (IQR: 120–453) performed non-obstetric surgeries and 148 (IQR: 90–243) obstetric procedures. Again, a high proportion of the non-obstetric surgeries were minor procedures such as hysteroscopies, small vaginal procedures, and non-complex laparoscopies. Complex laparoscopies and abdominal procedures are only performed to a very limited extent by last-year residents, as shown in Table . Logbook The majority of trainees (65.04%, n = 173) and about half of trainers (50.55%, n = 139) support that every gynecologic resident should perform 200 minor procedures and 100 major procedures during their residency, which were required in most federal states by the time of the survey. Considerably more trainers (42.91%, n = 118) than trainees (28.57%, n = 76) stated that the number of required surgeries is too high. Only a few trainees and trainers indicated that the number of required surgeries is too low (6.39%, n = 17 of trainees and 6.55%, n = 18 of trainers). A high proportion of trainees (64.79%, n = 173), but fewer trainers (39.27%, n = 108) reported that the required numbers in the logbook are being rounded up. More than half of the trainees (57.72%, n = 157) stated also that the obligatory annual training review meetings, which are documented in the training logbook, do not take place regularly. In contrast, only a minority of trainers (23.19%, n = 64) confirmed that annual training review meetings do not take place regularly. Evaluation of surgical training Most participants, 98.19% of trainers ( n = 270) and 93.38% ( n = 254) of trainees, stated that surgical training is important or very important. However, trainees rated the quality of surgical training significantly lower than trainers [on a scale from 1 (very good) to 6 (insufficient), mean ± standard deviation: trainees: 3.7 ± 1.4 vs. trainers: 2.7 ± 1.1, p < 0.0001]. Regarding the initiation of surgical training, the majority of trainers (65.58%, n = 181) reported that training starts at the resident level in their respective institutions, while 24.28% ( n = 67) reported that it starts in the last year of academic studies. Only a small percentage of trainers (10.15%, n = 28) reported that surgical training begins after residency. Trainees stated that they spend an average of 23.94 ± 14.65% of their working time in the operating theatre. With respect to the number of surgeries involving a trainee, 36.43% ( n = 98) of trainees and 27.53% ( n = 76) of trainers reported having a resident in the operating theatre in less than half of the surgeries performed. Trainers and trainees provided contradictory information when asked about the support for surgical training in their institutions. More than half of the trainees (53.16%, n = 143) reported that there is no explicit support for surgical training in their institution, but only a few trainers (7.25%, n = 20) confirmed the lack of explicit training support. Trainers reported more frequent support opportunities in surgical training, such as internal training courses, time off and budget for external trainings, availability of training models, and regular feedback. Table provides a detailed overview of the support for surgical skills improvement. Trainees identified a lack of structure in surgical training, a high proportion of administrative tasks, a lack of time for teaching in the operating theatre, and limited human resources as the main obstacles for surgical training (Table ). In contrast, limited time was seen as the main difficulty by trainers. Trainers also identified the complexity of procedures as a main problem for surgical training (Table ). Neither trainers nor trainees considered motivation a major issue in surgical training. 91.88% ( n = 249) of the trainees reported that good surgical training would be a reason to change their employer. Concurrently, 55.80% of trainers ( n = 154) stated that they have problems filling open positions for residents and specialists with surgical experience. Opportunities for improvement As an opportunity for improving surgical training, a majority of both trainees and trainers indicated that an external higher-level junior fellowship program (mentoring program) would be helpful (trainees: 58.96%, n = 158; trainers: 57.04%, n = 147). Furthermore, external fellowship programs for specific structured training in oncologic surgery or breast surgery were supported by both trainees (68.23%, n = 183) and trainers (67.65%, n = 184). A central and regular review of residency training, with potential consequences for the institution in case of non-compliance, was supported by 72.39% of trainees ( n = 194) and 42.28% of trainers ( n = 115). Regarding what a trainee can do on an individual level to perform more surgeries and receive better training, the most frequent answers from participating trainers were independent practice (60.36%, n = 166), actively asking to operate (58.18%, n = 160), increased motivation (53.45%, n = 147), accepting overtime (44.72%, n = 123), attending external trainings (34.55%, n = 95), remaining in the hospital after duty (24.00%, n = 66), and participating in internships at external clinics (21.45%, n = 59). A total of 601 participants completed the survey, with 305 trainees and 296 trainers. Participants from all 16 German federal states were represented. General characteristics of the participants are presented in Table . The majority of participants worked in maximum care or university hospitals, that were authorized for full residency training. Most of institutions were certified as gynecologic and breast oncologic centers. Only a minority worked in offices (Table ). The majority of trainees were in their third to fifth year of training. Most of the trainers were consultants (38.18%, n = 105) or senior consultants (22.91%, n = 63), followed by chief physicians (20.73%, n = 57) and specialists (15.25%, n = 42). More than half of the trainers (58.03%, n = 172) had one or more subspecialties, including Gynecological Oncology (42.34%, n = 116), Fetal Medicine (23.72%, n = 65), or Gynecological Endocrinology and Reproductive Medicine (5.11%, n = 14). The trainees performed a median of 125 (IQR: 41–332) non-obstetric surgeries and 75 (IQR: 27–168) obstetric surgeries independently. Detailed figures of the surgeries performed are presented in Table . The majority of non-obstetric surgeries performed by the trainees were hysteroscopies and small vaginal procedures, such as dilation & curettage or conisations. Major procedures, such as type-III or type-IV laparoscopy or hysterectomy (open/vaginal), were carried out independently by the trainees to a very limited extent. Breast surgery was also rarely performed independently by trainees. The subgroup analysis of residents with four or more completed years of training showed a median of 277 (IQR: 120–453) performed non-obstetric surgeries and 148 (IQR: 90–243) obstetric procedures. Again, a high proportion of the non-obstetric surgeries were minor procedures such as hysteroscopies, small vaginal procedures, and non-complex laparoscopies. Complex laparoscopies and abdominal procedures are only performed to a very limited extent by last-year residents, as shown in Table . The majority of trainees (65.04%, n = 173) and about half of trainers (50.55%, n = 139) support that every gynecologic resident should perform 200 minor procedures and 100 major procedures during their residency, which were required in most federal states by the time of the survey. Considerably more trainers (42.91%, n = 118) than trainees (28.57%, n = 76) stated that the number of required surgeries is too high. Only a few trainees and trainers indicated that the number of required surgeries is too low (6.39%, n = 17 of trainees and 6.55%, n = 18 of trainers). A high proportion of trainees (64.79%, n = 173), but fewer trainers (39.27%, n = 108) reported that the required numbers in the logbook are being rounded up. More than half of the trainees (57.72%, n = 157) stated also that the obligatory annual training review meetings, which are documented in the training logbook, do not take place regularly. In contrast, only a minority of trainers (23.19%, n = 64) confirmed that annual training review meetings do not take place regularly. Most participants, 98.19% of trainers ( n = 270) and 93.38% ( n = 254) of trainees, stated that surgical training is important or very important. However, trainees rated the quality of surgical training significantly lower than trainers [on a scale from 1 (very good) to 6 (insufficient), mean ± standard deviation: trainees: 3.7 ± 1.4 vs. trainers: 2.7 ± 1.1, p < 0.0001]. Regarding the initiation of surgical training, the majority of trainers (65.58%, n = 181) reported that training starts at the resident level in their respective institutions, while 24.28% ( n = 67) reported that it starts in the last year of academic studies. Only a small percentage of trainers (10.15%, n = 28) reported that surgical training begins after residency. Trainees stated that they spend an average of 23.94 ± 14.65% of their working time in the operating theatre. With respect to the number of surgeries involving a trainee, 36.43% ( n = 98) of trainees and 27.53% ( n = 76) of trainers reported having a resident in the operating theatre in less than half of the surgeries performed. Trainers and trainees provided contradictory information when asked about the support for surgical training in their institutions. More than half of the trainees (53.16%, n = 143) reported that there is no explicit support for surgical training in their institution, but only a few trainers (7.25%, n = 20) confirmed the lack of explicit training support. Trainers reported more frequent support opportunities in surgical training, such as internal training courses, time off and budget for external trainings, availability of training models, and regular feedback. Table provides a detailed overview of the support for surgical skills improvement. Trainees identified a lack of structure in surgical training, a high proportion of administrative tasks, a lack of time for teaching in the operating theatre, and limited human resources as the main obstacles for surgical training (Table ). In contrast, limited time was seen as the main difficulty by trainers. Trainers also identified the complexity of procedures as a main problem for surgical training (Table ). Neither trainers nor trainees considered motivation a major issue in surgical training. 91.88% ( n = 249) of the trainees reported that good surgical training would be a reason to change their employer. Concurrently, 55.80% of trainers ( n = 154) stated that they have problems filling open positions for residents and specialists with surgical experience. As an opportunity for improving surgical training, a majority of both trainees and trainers indicated that an external higher-level junior fellowship program (mentoring program) would be helpful (trainees: 58.96%, n = 158; trainers: 57.04%, n = 147). Furthermore, external fellowship programs for specific structured training in oncologic surgery or breast surgery were supported by both trainees (68.23%, n = 183) and trainers (67.65%, n = 184). A central and regular review of residency training, with potential consequences for the institution in case of non-compliance, was supported by 72.39% of trainees ( n = 194) and 42.28% of trainers ( n = 115). Regarding what a trainee can do on an individual level to perform more surgeries and receive better training, the most frequent answers from participating trainers were independent practice (60.36%, n = 166), actively asking to operate (58.18%, n = 160), increased motivation (53.45%, n = 147), accepting overtime (44.72%, n = 123), attending external trainings (34.55%, n = 95), remaining in the hospital after duty (24.00%, n = 66), and participating in internships at external clinics (21.45%, n = 59). Principal findings This comparable large survey, including both OBGYN trainees and trainers, highlights that surgical training in OBGYN is a crucial issue for most participants. However, the quality of surgical training is not rated well, especially by trainees due to the high proportion of administrative tasks and a lack of teaching time in the operating theater. Results from this nationwide survey conducted in Germany show that trainees, if at all, only achieve the required surgical figures in the training logbook for minor surgeries, but not for major surgeries. Trainees report also that the numbers required in training logbooks are often rounded up, and the required annual training reviews do not occur. Results in the context of literature A recent survey of German OBGYN residents revealed a high value placed on learning laparoscopic surgical techniques . However, our study showed that the quality of surgical training in OBGYN was only rated as satisfactory to sufficient, with trainers giving an average rating of 2.7 and trainees giving a rating of 3.7 on a scale from 1 to 6. This finding is in line with previous studies that have evaluated training in OBGYN or in senology and confirms that trainees generally evaluate surgical training more negatively than trainers . Like many other European countries, Germany maintains a logbook to document performed surgical procedures during OBGYN training . Trainers are required to verify the factual numbers of surgeries completed by trainees, but they are not asked to assess the trainees’ surgical skills. At the time of the survey, most German federal states required 100–200 minor surgeries and 50–100 major surgeries during residency training . In our survey, trainees were able to meet the targeted numbers for minor surgical procedures outlined in the logbook. However, there is growing international consensus that more complex surgeries, such as hysterectomies, should also be included in core gynecologic training . Our survey revealed that a large proportion of trainees in their final year of training did not achieve the required numbers for these major procedures. This finding is consistent with a German 2011 survey that highlighted a mismatch between actual surgical training and logbook requirements . Notably, international gynecological (oncology) trainees have emphasized the importance of hands-on surgical skills, particularly for complex surgeries and laparoscopies . Our survey supports this view, as trainees reported a preferred minimum of >30 performed laparoscopic surgeries, which is in line with the previous reports . Only half of the trainees in this cohort reported receiving explicit support for their surgical training, highlighting a significant lack of structured educational guidance. A substantial portion of administrative tasks undertaken by trainees and a lack of structure in surgical training were thereby identified as primary obstacles to surgical training, a well-documented issue within the broader context of residency training . Furthermore, trainers underscored the increasing challenge of limited teaching time in the operating room. Considering the inadequate training situation in structured surgical training for OBGYN residents, discussions about alternative training concepts are warranted. In addition to live training in the operating theatre, simulation-based training has demonstrated enhanced performance among participating residents . These simulations serve as potential additional tools for delivering standardized, in-depth training and should be broadly available in residency training. Nevertheless, one-on-one mentoring by a supervisor remains the core component of surgical training. Only a few study participants in our survey reported having a dedicated mentor or specific supervisor for their surgical training. The lack of mentoring reported by trainees in the survey can be addressed through external mentoring programs, which are widely supported by residents and educators in our study. Recently, fellowship programs established by major German gynecologic societies, such as the Young Academy of Gynecologic Oncology (JAGO) of the North-Eastern German Society of Gynecologic Oncology (NOGGO), the Young Talents of the Working Group Gynecologic Oncology (AGO), and the Junior Academy of the Working Group Urogynecology and Pelvicfloor Reconstruction (AGUB), have been established to provide structured training and mentorship for residents and could address the lack of surgical training during residency, especially for those residents interested in a more comprehensive surgical career in the respective subspecialities. In addition, trainees in this survey demanded a central and regular review of residency training, with potential consequences for the respective institution in case of non-compliance. The “visiting system” established for this purpose by the European Board and College of Obstetrics and Gynecology (EBCOG) and the European Union of Medical Specialists (UEMS) could serve as an example . However, this accreditation program is based on voluntariness and does not include a formal European exam. This comparable large survey, including both OBGYN trainees and trainers, highlights that surgical training in OBGYN is a crucial issue for most participants. However, the quality of surgical training is not rated well, especially by trainees due to the high proportion of administrative tasks and a lack of teaching time in the operating theater. Results from this nationwide survey conducted in Germany show that trainees, if at all, only achieve the required surgical figures in the training logbook for minor surgeries, but not for major surgeries. Trainees report also that the numbers required in training logbooks are often rounded up, and the required annual training reviews do not occur. A recent survey of German OBGYN residents revealed a high value placed on learning laparoscopic surgical techniques . However, our study showed that the quality of surgical training in OBGYN was only rated as satisfactory to sufficient, with trainers giving an average rating of 2.7 and trainees giving a rating of 3.7 on a scale from 1 to 6. This finding is in line with previous studies that have evaluated training in OBGYN or in senology and confirms that trainees generally evaluate surgical training more negatively than trainers . Like many other European countries, Germany maintains a logbook to document performed surgical procedures during OBGYN training . Trainers are required to verify the factual numbers of surgeries completed by trainees, but they are not asked to assess the trainees’ surgical skills. At the time of the survey, most German federal states required 100–200 minor surgeries and 50–100 major surgeries during residency training . In our survey, trainees were able to meet the targeted numbers for minor surgical procedures outlined in the logbook. However, there is growing international consensus that more complex surgeries, such as hysterectomies, should also be included in core gynecologic training . Our survey revealed that a large proportion of trainees in their final year of training did not achieve the required numbers for these major procedures. This finding is consistent with a German 2011 survey that highlighted a mismatch between actual surgical training and logbook requirements . Notably, international gynecological (oncology) trainees have emphasized the importance of hands-on surgical skills, particularly for complex surgeries and laparoscopies . Our survey supports this view, as trainees reported a preferred minimum of >30 performed laparoscopic surgeries, which is in line with the previous reports . Only half of the trainees in this cohort reported receiving explicit support for their surgical training, highlighting a significant lack of structured educational guidance. A substantial portion of administrative tasks undertaken by trainees and a lack of structure in surgical training were thereby identified as primary obstacles to surgical training, a well-documented issue within the broader context of residency training . Furthermore, trainers underscored the increasing challenge of limited teaching time in the operating room. Considering the inadequate training situation in structured surgical training for OBGYN residents, discussions about alternative training concepts are warranted. In addition to live training in the operating theatre, simulation-based training has demonstrated enhanced performance among participating residents . These simulations serve as potential additional tools for delivering standardized, in-depth training and should be broadly available in residency training. Nevertheless, one-on-one mentoring by a supervisor remains the core component of surgical training. Only a few study participants in our survey reported having a dedicated mentor or specific supervisor for their surgical training. The lack of mentoring reported by trainees in the survey can be addressed through external mentoring programs, which are widely supported by residents and educators in our study. Recently, fellowship programs established by major German gynecologic societies, such as the Young Academy of Gynecologic Oncology (JAGO) of the North-Eastern German Society of Gynecologic Oncology (NOGGO), the Young Talents of the Working Group Gynecologic Oncology (AGO), and the Junior Academy of the Working Group Urogynecology and Pelvicfloor Reconstruction (AGUB), have been established to provide structured training and mentorship for residents and could address the lack of surgical training during residency, especially for those residents interested in a more comprehensive surgical career in the respective subspecialities. In addition, trainees in this survey demanded a central and regular review of residency training, with potential consequences for the respective institution in case of non-compliance. The “visiting system” established for this purpose by the European Board and College of Obstetrics and Gynecology (EBCOG) and the European Union of Medical Specialists (UEMS) could serve as an example . However, this accreditation program is based on voluntariness and does not include a formal European exam. The study’s main limitations are possible selection bias, as it is underrepresented by physicians working in local surgeries and overrepresented by those working in maximum care hospitals. Furthermore, the majority of participants showed a special interest in gynecologic oncology. This selection bias may be attributed to the distribution channels used to disseminate the survey. It is also possible that dissatisfied physicians were overrepresented in the survey, leading to potential biases such as falsely low surgery figures. Another limitation lies in the subjective evaluation of training quality. Incorporating objective metrics or performance outcomes could provide a more comprehensive evaluation of training quality. Additionally, the present study offers a static view of the training situation and no longitudinal data. Longitudinal studies could aid in determining whether the quality of surgical training is progressing, regressing, or staying constant over time… Nonetheless, with 601 participants, this survey is one of the largest and provides the first nationwide data on surgical OBGYN training in Germany, including specific surgery figures for residents. More female participants than male participants are represented, particularly among trainees, which reflects the gender distribution of OBGYN trainee physicians in Germany. In summary, this survey provides an overview of the status of surgical gynecologic training in Germany and emphasizes the need for improvement. To address the unsatisfactory surgical training situation of gynecologic residents, adjustments to the residency curriculum are urgently needed. A structured training program is necessary to provide in-depth surgical training to all residents, especially final-year residents who perform much fewer major procedures than required by the logbook. Independent evaluation mechanisms to track the training progress of each resident, teacher and trainee supervision, and mentoring concepts can aid in implementation. Additionally, fellowship or cross-institutional programs can address the lack of surgical training and provide further opportunities for residents to gain valuable experience. Future studies should address the question of how far structured training programs such as fellowship or mentoring programs can really improve the quality of training in OBGYN. |
Posterior Limbal Mesenchymal Stromal Cells Promote Proliferation and Stemness of Transition Zone Cells: A Novel Insight Into Corneal Endothelial Rejuvenation | 9fd40a79-0da4-4ae8-991b-312ef1085908 | 11753478 | Anatomy[mh] | Human and Mouse Corneal Tissues Human donor corneoscleral rims were provided by the New Zealand National Eye Bank, after the central corneal tissues had been utilized for corneal transplantation. A total of 12 donor corneoscleral rims were collected for primary culture. The research was approved by the “Northern A” Health and Disability Ethics Committee of New Zealand (number NTX/07/08/080/AM04) and handled in compliance with the Declaration of Helsinki. Mouse corneas were collected from euthanized mice (8-week-old; male; CD1 strain) under SOP836 by staff at the Vernon Jenson Unit of the Faculty of Medical and Health Sciences, University of Auckland. The ethical approval was waived by the Ethics Committee of University of Auckland in view of the routine colony maintenance of excess stock of animals. A total of 24 mouse corneas were used for organ culture. All procedures adhered to the ARVO statement for the Use of Animals in Ophthalmic and Vision Research. Cell Culture The primary culture of human TZ cells was established according to our previously published protocol. Briefly, TZ tissue including the peripheral endothelium, Schwalbe’s line, and the insert portion of the trabecular meshwork was peeled from the posterior side of the human corneoscleral rim. It was further cut into 12 segments and seeded on plates pre-coated with 10 µg/mL fibronectin (341631; Sigma-Aldrich, Allentown, PA, USA) and 35 µg/mL collagen I (A1048301; Thermo Fisher Scientific, Branford, CT, USA). The TZ explants were cultivated in Opti-MEM (31985070; Thermo Fisher Scientific) supplemented with 8% fetal bovine serum (FBS; 16000044, Thermo Fisher Scientific), 5 ng/mL epidermal growth factor (PHG0313; Thermo Fisher Scientific), 40 ng/mL fibroblast growth factor (FGF; 13256-029, Thermo Fisher Scientific), 20 ng/mL nerve growth factor (PHG0126; Thermo Fisher Scientific), 20 µg/mL L-ascorbic acid 2-phosphate (A8960; Sigma-Aldrich), 200 µg/mL calcium chloride (C7902; Sigma-Aldrich), 0.04% chondroitin sulphate (C6737; Sigma-Aldrich), 50 µg/mL gentamicin (15710064; Thermo Fisher Scientific), and 1% antibiotics-antimycotics (15240062; Thermo Fisher Scientific). The culture medium was refreshed every 2 days. Once the cells reached confluency, they were passaged using TrypLE Express (12604013; Thermo Fisher Scientific) at a ratio of 1:4 and seeded according to the downstream experiment. Human A-LMSC and P-LMSC cultures were established by modifying previous protocols. , Specifically, after TZ tissue was peeled from the human corneoscleral rim, the remaining rim was incised at 1 mm inside and outside the anatomic limbus, so that excess cornea, sclera, and conjunctiva were removed. The limbus was digested in 1.2 U/mL Dispase (17105041; Thermo Fisher Scientific) for 40 minutes on a shaker at 37°C to loosen the basement membrane, and the epithelium was scraped off using a scalpel. The remaining limbal stroma was cut into anterior and posterior portions for A-LMSC and P-LMSC explant culture in Dulbecco's Modified Eagle Medium: Nutrient Mixture F-12 (DMEM/F12 1:1, 11320033; Thermo Fisher Scientific) added with 10% FBS and 1% antibiotics-antimycotics. The conditioned medium derived from A-LMSCs and P-LMSCs was obtained from the cultured cells at 60% to 80% confluence every 24 hours. The conditioned medium was then centrifuged at 2000 revolutions per minute (rpm) for 10 minutes, and the supernatant was collected for organ culture of mouse corneas. The immortalized human corneal endothelial cell line HCEC-B4G12 (from Dr. Li Wen, Sydney Eye Hospital) was used as an experimental control for droplet digital polymerase chain reaction (ddPCR) and Western blotting. B4G12 cells were seeded onto plates pre-coated with 10 µg/mL laminin (L2020; Sigma-Aldrich) and 10 mg/mL chondroitin sulfate, and cultured in Human Endothelial Serum Free Medium (11111044; Thermo Fisher Scientific) supplemented with 2% FBS, 10 ng/mL FGF, and 1% antibiotics-antimycotics. Transwell Coculture TZ cells were cocultured with A-LMSCs or P-LMSCs in a Transwell system, separated by a permeable membrane. The A-LMSCs or P-LMSCs at passage 1 (P1) were first seeded in the 12-well Transwell inserts (3460; Corning, Corning, NY, USA) at a density of 1 × 10 4 /cm 2 . After the A-LMSCs or P-LMSCs reached confluency, P1 TZ cells were seeded into the bottom of the wells, some with coverslips, coated with fibronectin and collagen at a density of 0.2 × 10 4 /cm 2 . The cocultures were maintained in TZ medium for 2 weeks. The 5-ethynyl-2′-deoxyuridine (EdU) assay and scratch wound assay were conducted during the coculture process. The cocultured TZ cells after 2 weeks were further harvested for colony forming assay, ddPCR, Western blotting, and immunocytochemistry. EdU Assay The proliferation of TZ cells was evaluated by the EdU Imaging Kit (C10639; Thermo Fisher Scientific). The P1 TZ cells were seeded on fibronectin and collagen coated coverslips and cocultured with A-LMSCs or P-LMSCs for 3 days. TZ cells without coculture were used as the control group ( n = 3 wells in each group). The TZ cells were then cultured in 10 µM EdU labeling solution for 6 hours and fixed with 4% paraformaldehyde (PFA) for 20 minutes. Subsequently, the cells were incubated with 0.5% Triton X-100 for 20 minutes to permeabilize, Click-iT reaction cocktail for 30 minutes to detect EdU, and 1 µg/mL 4′,6-diamidino-2-phenylindole (DAPI; D9542; Sigma-Aldrich) for 10 minutes to label the nuclei. The coverslips were further mounted on glass slides and imaged under a Zeiss Colibri 7 fluorescence microscope. The percentages of proliferating TZ cells in the three groups were assessed using ImageJ software (version 1.45b, National Institutes of Health, Bethesda, MD, USA) and quantified by dividing the number of EdU positive cells by the total cell number as indicated by DAPI staining. Scratch Wound Assay The wound healing ability of TZ cells was measured with the scratch wound assay. Once the TZ cells met confluency after coculturing with A-LMSCs or P-LMSCs in the Transwell system, a horizontal scratch using a sterile 1 mL pipette tip was performed in each well, with non-cocultured TZ cells as the control group ( n = 3 wells in each group). Then, the detached cells were rinsed off using phosphate buffered saline (PBS). The remaining TZ cells were continuously cocultured with A-LMSCs or P-LMSCs in TZ medium. The scratching in each well was marked with a marker pen at the bottom of the plate to ensure the same area was imaged at all time points. The TZ cells continued to be cocultured with A-LMSCs or P-LMSCs and imaged at 0, 4, 8, 16, 20, 24, 30, and 44 hours after scratching. The wound area was assessed with ImageJ software. The relative wound area was quantified by dividing the remaining wound area by the initial wound area. Colony Forming Assay After coculturing with A-LMSCs or P-LMSCs for 2 weeks, TZ cells were passaged onto a fibronectin and collagen-coated 6-well at 1000 cells per well and cultured for 12 days to form colonies. TZ cells without coculture were processed in the same procedure ( n = 3 wells in each group). The colonies were fixed with 4% PFA for 20 minutes and stained with 1% cresyl violet (C5042; Sigma-Aldrich) for 1 hour. The colony forming efficiency was determined by calculating the ratio of the number of colonies to the seeded number of TZ cells. Droplet Digital Polymerase Chain Reaction TZ cells cocultured with A-LMSCs or P-LMSCs were collected for ddPCR to compare the gene profile of stem cell and differentiated markers, with non-cocultured TZ cells and B4G12 cells as the control groups ( n = 3 in B4G12, TZ, and P-TZ and n = 2 in A-TZ). The RNA was isolated by the Purelink RNA mini kit (12183020; Thermo Fisher Scientific). The extracted RNA was tested using the Tape Station (Agilent, Santa Clara, CA, USA) to evaluate the quantity and quality of isolated RNA, followed by the SPUD assay to examine the presence of PCR inhibitors. After passing quality control, RNA was reverse-transcribed into cDNA using the SuperScript VILO cDNA Synthesis Kit (11754050; Thermo Fisher Scientific). The successful synthesis of cDNA was verified via PCR amplification of β-actin and agarose gel electrophoresis of the amplified products. Then, the cDNA samples were analyzed using ddPCR. Specifically, each PCR reaction was set up containing 1.1 µL of PrimeTime pre-designed gene expression assay (IDT, Coralville, IA, USA), 11 µL of ddPCR supermix for probes (no dUTP; Bio-Rad, Hercules, CA, USA), and 1.1 µL of cDNA (2 ng/µL) in a total volume of 22 µL. A “no template” control was included alongside the experimental samples for each target/reference gene. Droplet generation involved adding 20 µL of PCR reaction and 70 µL of droplet oil to the DG8 cartridges (Bio-Rad), with droplets formed using QX200 droplet generator (Bio-Rad). Then, 40 µL of the generated droplets were subjected to a C1000 Touch Thermal Cycler (Bio-Rad) for amplification: initial pre-denaturation at 95°C for 10 minutes, followed by 40 cycles of denaturation at 94°C for 30 seconds and annealing at 60°C for 60 seconds, extension at 98°C for 10 minutes, and cooling at 12°C for temporary storage. After amplification, the droplets were individually detected according to the fluorescence signal from each droplet using the QX200 droplet reader (Bio-Rad), and the concentration of positive droplets was analyzed using QuantaSoft analysis software (Bio-Rad). If the concentration numbers were higher than 5000, the cDNA sample was diluted and ddPCR was repeated. In cases where expression was undetectable, increased volumes of cDNA (up to 9.9 µL) were used for repeated reactions. The expression for each gene was normalized to the geometric mean of two most stable reference genes, selected from a set of seven, using the NormFinder algorithm. B4G12 cells were used as positive controls for CEC gene expression, but this group was not included in the 1-way ANOVA analysis. Detailed information on PrimeTime assays of target and reference genes is provided in . Western Blotting TZ cells in different treatment groups and B4G12 cells ( n = 3 in each group) were lysed by radioimmunoprecipitation assay (RIPA) buffer added with protease inhibitor cocktail (04693159001; Roche, Basel, Switzerland) for protein extraction. Protein concentration was measured with the detergent compatible protein assay (5000116; Bio-Rad). A total of 30 µg of protein were then separated via electrophoresis on precast polyacrylamide stain-free gels (Bio-Rad) and subsequently transferred onto polyvinylidene difluoride membranes using a Bio-Rad Trans-Blot Turbo transfer system. The membranes were blocked with 5% trim milk for 1 hour and incubated with primary antibodies overnight at 4°C and secondary antibodies at room temperature for 2 hours. The target protein was detected using Pierce enhanced chemiluminescence (ECL) Plus Substrate (32132; Thermo Fisher Scientific), and the film was scanned by ChemiDoc MP imaging system (Bio-Rad). Band densitometry was analyzed using Image Lab software (Bio-Rad, version 6.1). Relative protein level was calculated as the integrated density of the protein band divided by the integrated density of housekeeping marker α-tubulin on the same blot. The full-length Western blots of target and housekeeping proteins with the ladder are provided in to . Immunofluorescence Staining Immunofluorescence staining was performed to characterize P-LMSCs at P1 and compare the stem cell protein expression in TZ cells. Specifically, the cells were first fixed with 4% PFA for 15 minutes, permeabilized with 2% Triton X-100 and 10% goat/horse serum in PBS for 60 minutes, and incubated with primary antibodies overnight at 4°C and secondary antibodies for 2 hours at room temperature, with 3 times of washing using PBS between each step. The nuclei were labeled by DAPI, and the slides were mounted with Citifluor antifadent solution (Electron Microscopy Sciences, Hatfield, PA, USA). Images were taken under a Zeiss Colibri 7 fluorescence microscope. The details of antibodies are supplied in . Organ Culture of Wounded Mouse Corneas The mouse corneal tissues, with a small rim of sclera attached, were collected after the removal of connective tissue and conjunctiva from the eyeball. The wounding of mouse corneas was made by carefully scraping the entire endothelium using a silicone tube sheathed on an anterior chamber syringe. The corneas were stained with trypan blue (15250061; Thermo Fisher Scientific) for 2 minutes, rinsed in PBS, and imaged using a Zeiss Discovery V20 stereomicroscope. The wounded corneas were cultured in 3 different media: (1) medium control group: basal medium made of DMEM/F12 added with 10% FBS and 1% antibiotics-antimycotics; (2) A-LMSC CM group: conditioned medium from A-LMSCs mixed with the basal medium at a ratio of 1:1; and (3) P-LMSC CM group: conditioned medium from P-LMSCs mixed with the basal medium at a ratio of 1:1. The corneas were cultured for 2 weeks with media changed every 2 to 3 days. Trypan blue staining was repeated on days 4, 7, 10, and 14. The trypan blue staining area was quantified using ImageJ software. The relative wound area was determined by dividing the trypan blue staining area by the total area of the mouse cornea. On day 1 and day 6, the corneas were incubated in the media containing 10 µM EdU for 24 hours and the detection of EdU was performed as described previously. After fixing in 4% PFA for 1 hour, the corneal epithelium was scraped using a scalpel to avoid imaging the EdU positive epithelium in the transparent corneal tissue. All effort was made to remove the epithelium as thoroughly as possible, but some still remained, and they could be identified in images by being out-of-focus. On day 6, the corneal tissues with incorporated EdU were further incubated in ZO-1 and the corresponding secondary antibody, as mentioned above. The nuclei were stained with DAPI and the corneas were flat-mounted in Citifluor antifadent solution. Images were taken under a Zeiss Colibri 7 fluorescence microscope. Statistics All data were shown as mean ± standard error (SE) and analyzed using 1-way ANOVA in SPSS Statistics 29.0 (IBM, Armonk, NY, USA). Homogeneity of variances was first tested between all the groups. When the data met the assumption of homogeneity of variances, Fisher's least significant difference (LSD) test was applied for post hoc pairwise comparisons, otherwise, Games-Howell test was used. P < 0.05 was considered statistically significant. Human donor corneoscleral rims were provided by the New Zealand National Eye Bank, after the central corneal tissues had been utilized for corneal transplantation. A total of 12 donor corneoscleral rims were collected for primary culture. The research was approved by the “Northern A” Health and Disability Ethics Committee of New Zealand (number NTX/07/08/080/AM04) and handled in compliance with the Declaration of Helsinki. Mouse corneas were collected from euthanized mice (8-week-old; male; CD1 strain) under SOP836 by staff at the Vernon Jenson Unit of the Faculty of Medical and Health Sciences, University of Auckland. The ethical approval was waived by the Ethics Committee of University of Auckland in view of the routine colony maintenance of excess stock of animals. A total of 24 mouse corneas were used for organ culture. All procedures adhered to the ARVO statement for the Use of Animals in Ophthalmic and Vision Research. The primary culture of human TZ cells was established according to our previously published protocol. Briefly, TZ tissue including the peripheral endothelium, Schwalbe’s line, and the insert portion of the trabecular meshwork was peeled from the posterior side of the human corneoscleral rim. It was further cut into 12 segments and seeded on plates pre-coated with 10 µg/mL fibronectin (341631; Sigma-Aldrich, Allentown, PA, USA) and 35 µg/mL collagen I (A1048301; Thermo Fisher Scientific, Branford, CT, USA). The TZ explants were cultivated in Opti-MEM (31985070; Thermo Fisher Scientific) supplemented with 8% fetal bovine serum (FBS; 16000044, Thermo Fisher Scientific), 5 ng/mL epidermal growth factor (PHG0313; Thermo Fisher Scientific), 40 ng/mL fibroblast growth factor (FGF; 13256-029, Thermo Fisher Scientific), 20 ng/mL nerve growth factor (PHG0126; Thermo Fisher Scientific), 20 µg/mL L-ascorbic acid 2-phosphate (A8960; Sigma-Aldrich), 200 µg/mL calcium chloride (C7902; Sigma-Aldrich), 0.04% chondroitin sulphate (C6737; Sigma-Aldrich), 50 µg/mL gentamicin (15710064; Thermo Fisher Scientific), and 1% antibiotics-antimycotics (15240062; Thermo Fisher Scientific). The culture medium was refreshed every 2 days. Once the cells reached confluency, they were passaged using TrypLE Express (12604013; Thermo Fisher Scientific) at a ratio of 1:4 and seeded according to the downstream experiment. Human A-LMSC and P-LMSC cultures were established by modifying previous protocols. , Specifically, after TZ tissue was peeled from the human corneoscleral rim, the remaining rim was incised at 1 mm inside and outside the anatomic limbus, so that excess cornea, sclera, and conjunctiva were removed. The limbus was digested in 1.2 U/mL Dispase (17105041; Thermo Fisher Scientific) for 40 minutes on a shaker at 37°C to loosen the basement membrane, and the epithelium was scraped off using a scalpel. The remaining limbal stroma was cut into anterior and posterior portions for A-LMSC and P-LMSC explant culture in Dulbecco's Modified Eagle Medium: Nutrient Mixture F-12 (DMEM/F12 1:1, 11320033; Thermo Fisher Scientific) added with 10% FBS and 1% antibiotics-antimycotics. The conditioned medium derived from A-LMSCs and P-LMSCs was obtained from the cultured cells at 60% to 80% confluence every 24 hours. The conditioned medium was then centrifuged at 2000 revolutions per minute (rpm) for 10 minutes, and the supernatant was collected for organ culture of mouse corneas. The immortalized human corneal endothelial cell line HCEC-B4G12 (from Dr. Li Wen, Sydney Eye Hospital) was used as an experimental control for droplet digital polymerase chain reaction (ddPCR) and Western blotting. B4G12 cells were seeded onto plates pre-coated with 10 µg/mL laminin (L2020; Sigma-Aldrich) and 10 mg/mL chondroitin sulfate, and cultured in Human Endothelial Serum Free Medium (11111044; Thermo Fisher Scientific) supplemented with 2% FBS, 10 ng/mL FGF, and 1% antibiotics-antimycotics. TZ cells were cocultured with A-LMSCs or P-LMSCs in a Transwell system, separated by a permeable membrane. The A-LMSCs or P-LMSCs at passage 1 (P1) were first seeded in the 12-well Transwell inserts (3460; Corning, Corning, NY, USA) at a density of 1 × 10 4 /cm 2 . After the A-LMSCs or P-LMSCs reached confluency, P1 TZ cells were seeded into the bottom of the wells, some with coverslips, coated with fibronectin and collagen at a density of 0.2 × 10 4 /cm 2 . The cocultures were maintained in TZ medium for 2 weeks. The 5-ethynyl-2′-deoxyuridine (EdU) assay and scratch wound assay were conducted during the coculture process. The cocultured TZ cells after 2 weeks were further harvested for colony forming assay, ddPCR, Western blotting, and immunocytochemistry. The proliferation of TZ cells was evaluated by the EdU Imaging Kit (C10639; Thermo Fisher Scientific). The P1 TZ cells were seeded on fibronectin and collagen coated coverslips and cocultured with A-LMSCs or P-LMSCs for 3 days. TZ cells without coculture were used as the control group ( n = 3 wells in each group). The TZ cells were then cultured in 10 µM EdU labeling solution for 6 hours and fixed with 4% paraformaldehyde (PFA) for 20 minutes. Subsequently, the cells were incubated with 0.5% Triton X-100 for 20 minutes to permeabilize, Click-iT reaction cocktail for 30 minutes to detect EdU, and 1 µg/mL 4′,6-diamidino-2-phenylindole (DAPI; D9542; Sigma-Aldrich) for 10 minutes to label the nuclei. The coverslips were further mounted on glass slides and imaged under a Zeiss Colibri 7 fluorescence microscope. The percentages of proliferating TZ cells in the three groups were assessed using ImageJ software (version 1.45b, National Institutes of Health, Bethesda, MD, USA) and quantified by dividing the number of EdU positive cells by the total cell number as indicated by DAPI staining. The wound healing ability of TZ cells was measured with the scratch wound assay. Once the TZ cells met confluency after coculturing with A-LMSCs or P-LMSCs in the Transwell system, a horizontal scratch using a sterile 1 mL pipette tip was performed in each well, with non-cocultured TZ cells as the control group ( n = 3 wells in each group). Then, the detached cells were rinsed off using phosphate buffered saline (PBS). The remaining TZ cells were continuously cocultured with A-LMSCs or P-LMSCs in TZ medium. The scratching in each well was marked with a marker pen at the bottom of the plate to ensure the same area was imaged at all time points. The TZ cells continued to be cocultured with A-LMSCs or P-LMSCs and imaged at 0, 4, 8, 16, 20, 24, 30, and 44 hours after scratching. The wound area was assessed with ImageJ software. The relative wound area was quantified by dividing the remaining wound area by the initial wound area. After coculturing with A-LMSCs or P-LMSCs for 2 weeks, TZ cells were passaged onto a fibronectin and collagen-coated 6-well at 1000 cells per well and cultured for 12 days to form colonies. TZ cells without coculture were processed in the same procedure ( n = 3 wells in each group). The colonies were fixed with 4% PFA for 20 minutes and stained with 1% cresyl violet (C5042; Sigma-Aldrich) for 1 hour. The colony forming efficiency was determined by calculating the ratio of the number of colonies to the seeded number of TZ cells. TZ cells cocultured with A-LMSCs or P-LMSCs were collected for ddPCR to compare the gene profile of stem cell and differentiated markers, with non-cocultured TZ cells and B4G12 cells as the control groups ( n = 3 in B4G12, TZ, and P-TZ and n = 2 in A-TZ). The RNA was isolated by the Purelink RNA mini kit (12183020; Thermo Fisher Scientific). The extracted RNA was tested using the Tape Station (Agilent, Santa Clara, CA, USA) to evaluate the quantity and quality of isolated RNA, followed by the SPUD assay to examine the presence of PCR inhibitors. After passing quality control, RNA was reverse-transcribed into cDNA using the SuperScript VILO cDNA Synthesis Kit (11754050; Thermo Fisher Scientific). The successful synthesis of cDNA was verified via PCR amplification of β-actin and agarose gel electrophoresis of the amplified products. Then, the cDNA samples were analyzed using ddPCR. Specifically, each PCR reaction was set up containing 1.1 µL of PrimeTime pre-designed gene expression assay (IDT, Coralville, IA, USA), 11 µL of ddPCR supermix for probes (no dUTP; Bio-Rad, Hercules, CA, USA), and 1.1 µL of cDNA (2 ng/µL) in a total volume of 22 µL. A “no template” control was included alongside the experimental samples for each target/reference gene. Droplet generation involved adding 20 µL of PCR reaction and 70 µL of droplet oil to the DG8 cartridges (Bio-Rad), with droplets formed using QX200 droplet generator (Bio-Rad). Then, 40 µL of the generated droplets were subjected to a C1000 Touch Thermal Cycler (Bio-Rad) for amplification: initial pre-denaturation at 95°C for 10 minutes, followed by 40 cycles of denaturation at 94°C for 30 seconds and annealing at 60°C for 60 seconds, extension at 98°C for 10 minutes, and cooling at 12°C for temporary storage. After amplification, the droplets were individually detected according to the fluorescence signal from each droplet using the QX200 droplet reader (Bio-Rad), and the concentration of positive droplets was analyzed using QuantaSoft analysis software (Bio-Rad). If the concentration numbers were higher than 5000, the cDNA sample was diluted and ddPCR was repeated. In cases where expression was undetectable, increased volumes of cDNA (up to 9.9 µL) were used for repeated reactions. The expression for each gene was normalized to the geometric mean of two most stable reference genes, selected from a set of seven, using the NormFinder algorithm. B4G12 cells were used as positive controls for CEC gene expression, but this group was not included in the 1-way ANOVA analysis. Detailed information on PrimeTime assays of target and reference genes is provided in . TZ cells in different treatment groups and B4G12 cells ( n = 3 in each group) were lysed by radioimmunoprecipitation assay (RIPA) buffer added with protease inhibitor cocktail (04693159001; Roche, Basel, Switzerland) for protein extraction. Protein concentration was measured with the detergent compatible protein assay (5000116; Bio-Rad). A total of 30 µg of protein were then separated via electrophoresis on precast polyacrylamide stain-free gels (Bio-Rad) and subsequently transferred onto polyvinylidene difluoride membranes using a Bio-Rad Trans-Blot Turbo transfer system. The membranes were blocked with 5% trim milk for 1 hour and incubated with primary antibodies overnight at 4°C and secondary antibodies at room temperature for 2 hours. The target protein was detected using Pierce enhanced chemiluminescence (ECL) Plus Substrate (32132; Thermo Fisher Scientific), and the film was scanned by ChemiDoc MP imaging system (Bio-Rad). Band densitometry was analyzed using Image Lab software (Bio-Rad, version 6.1). Relative protein level was calculated as the integrated density of the protein band divided by the integrated density of housekeeping marker α-tubulin on the same blot. The full-length Western blots of target and housekeeping proteins with the ladder are provided in to . Immunofluorescence staining was performed to characterize P-LMSCs at P1 and compare the stem cell protein expression in TZ cells. Specifically, the cells were first fixed with 4% PFA for 15 minutes, permeabilized with 2% Triton X-100 and 10% goat/horse serum in PBS for 60 minutes, and incubated with primary antibodies overnight at 4°C and secondary antibodies for 2 hours at room temperature, with 3 times of washing using PBS between each step. The nuclei were labeled by DAPI, and the slides were mounted with Citifluor antifadent solution (Electron Microscopy Sciences, Hatfield, PA, USA). Images were taken under a Zeiss Colibri 7 fluorescence microscope. The details of antibodies are supplied in . The mouse corneal tissues, with a small rim of sclera attached, were collected after the removal of connective tissue and conjunctiva from the eyeball. The wounding of mouse corneas was made by carefully scraping the entire endothelium using a silicone tube sheathed on an anterior chamber syringe. The corneas were stained with trypan blue (15250061; Thermo Fisher Scientific) for 2 minutes, rinsed in PBS, and imaged using a Zeiss Discovery V20 stereomicroscope. The wounded corneas were cultured in 3 different media: (1) medium control group: basal medium made of DMEM/F12 added with 10% FBS and 1% antibiotics-antimycotics; (2) A-LMSC CM group: conditioned medium from A-LMSCs mixed with the basal medium at a ratio of 1:1; and (3) P-LMSC CM group: conditioned medium from P-LMSCs mixed with the basal medium at a ratio of 1:1. The corneas were cultured for 2 weeks with media changed every 2 to 3 days. Trypan blue staining was repeated on days 4, 7, 10, and 14. The trypan blue staining area was quantified using ImageJ software. The relative wound area was determined by dividing the trypan blue staining area by the total area of the mouse cornea. On day 1 and day 6, the corneas were incubated in the media containing 10 µM EdU for 24 hours and the detection of EdU was performed as described previously. After fixing in 4% PFA for 1 hour, the corneal epithelium was scraped using a scalpel to avoid imaging the EdU positive epithelium in the transparent corneal tissue. All effort was made to remove the epithelium as thoroughly as possible, but some still remained, and they could be identified in images by being out-of-focus. On day 6, the corneal tissues with incorporated EdU were further incubated in ZO-1 and the corresponding secondary antibody, as mentioned above. The nuclei were stained with DAPI and the corneas were flat-mounted in Citifluor antifadent solution. Images were taken under a Zeiss Colibri 7 fluorescence microscope. All data were shown as mean ± standard error (SE) and analyzed using 1-way ANOVA in SPSS Statistics 29.0 (IBM, Armonk, NY, USA). Homogeneity of variances was first tested between all the groups. When the data met the assumption of homogeneity of variances, Fisher's least significant difference (LSD) test was applied for post hoc pairwise comparisons, otherwise, Games-Howell test was used. P < 0.05 was considered statistically significant. Characterization of P-LMSCs To characterize the phenotype of P-LMSCs, we carried out explant culture, and assessed their morphology and protein expression in comparison to A-LMSCs. Results showed that A-LMSC and P-LMSC cultures could be established from anterior and posterior limbal stromal explants, respectively, and showed similar fibroblastic morphology for several passages . Immunocytochemistry results demonstrated that P-LMSCs were positive for mesenchymal marker vimentin and stem cell markers Nestin, TRA-1-60, and Oct3/4, all recognized markers for the identification of A-LMSCs . Morphology of TZ Cells Cocultured With P-LMSCs To investigate the effect of P-LMSCs on TZ cells, we performed P-LMSCs and TZ cell coculture (P-TZ) in a Transwell system, with A-LMSCs and TZ cell coculture (A-TZ) as a comparison group and TZ cells alone as a control group. A showed TZ tissue dissected from the human donor corneal rim after corneal transplantation. The TZ explant included the peripheral endothelium, Schwalbe’s line, and the insert portion of the trabecular meshwork. Primary cultured TZ cells reached confluency in 20 days (see A). P1 TZ cells demonstrated similar elongated and fibroblastic morphology when coculturing with P-LMSCs, as well as in A-LMSC coculture and without coculture. After coculturing for 7 days, TZ cells reached confluency in P-LMSC and A-LMSC coculture groups, whereas there were some vacant areas in the TZ-only control group ( B). Proliferation of TZ Cells Cocultured With P-LMSCs To compare the proliferation of TZ cells cocultured with P-LMSCs, A-LMSCs, and without coculture, EdU assay was performed in the three groups. Results showed that a significantly higher proportion of nuclei was positive for EdU labeling in TZ cells after coculturing with P-LMSCs (44.38% ± 0.59%, P < 0.001) and A-LMSCs (36.02% ± 2.23%, P = 0.047) compared to TZ only control (24.11% ± 0.72%; ). Wound Healing Capacity of TZ Cells Cocultured With P-LMSCs A scratch wound assay was performed to compare the wound healing ability of TZ cells. TZ cells cocultured with P-LMSCs demonstrated the highest healing speed after wounding compared to the other two groups ( A). Thirty hours after scratching, the average relative wound area was 4.39% ± 0.60% in the P-TZ group ( P = 0.002 versus control) and 21.12% ± 3.03% in the A-TZ group ( P > 0.05 versus control), compared to 32.36% ± 5.51% in the TZ only control group ( B). Stemness Properties of TZ Cells Cocultured With P-LMSCs To compare the stemness of TZ cells cocultured with P-LMSCs, A-LMSCs, and without coculture, the colony forming assay was performed. The results of cresyl violet colony staining showed that TZ cells cocultured with P-LMSCs yielded more colonies compared to those without coculture . Statistical analysis demonstrated a significantly higher colony forming efficiency in the P-TZ group (3.57% ± 0.24%) than the TZ control group (2.20% ± 0.31%, P = 0.007). The colony forming efficiency of the A-TZ group (3.30% ± 0.15%, P = 0.018) was also statistically higher than the control group. Gene Expression of TZ Cells Cocultured With P-LMSCs Normfinder analysis identified that HPRT1 and POLR2A were the most stable reference gene pair in this dataset, therefore the expressions of target genes were normalized to the geometric mean of HPRT1 and POLR2A . The results of ddPCR demonstrated that TZ cells cocultured with P-LMSCs expressed significantly higher levels of pluripotency gene NANOG (0.0016 ± 0.0003 vs. 0.0007 ± 0.0002, P = 0.034), neural crest genes SOX9 (0.2326 ± 0.0048 vs. 0.1863 ± 0.0080, P = 0.019), and TFAP2A (0.7271 ± 0.0467 vs. 0.5601 ± 0.0509, P = 0.048), and periocular mesenchyme gene PITX2 (1.9485 ± 0.0350 vs. 1.5235 ± 0.1224, P = 0.019) than the TZ only control . Moreover, the expression of neural crest gene NESTIN in TZ cells cocultured with P-LMSCs was significantly higher than those cocultured with A-LMSCs (1.4644 ± 0.0714 vs. 1.1010 ± 0.0347, P = 0.022) and showed an increasing trend compared to TZ cells without coculture (1.2286 ± 0.0829, P = 0.064). The corneal endothelial gene SLC4A11 in TZ cells was downregulated after coculturing with P-LMSCs (0.0220 ± 0.0003 vs. TZ control 0.0298 ± 0.0009, P = 0.015). However, the corneal endothelial gene AQP1 was upregulated after coculturing with A-LMSCs (1.3046 ± 0.0849 vs. TZ control 0.4168 ± 0.0512, P = 0.004) and P-LMSCs (0.9759 ± 0.1566, P = 0.015). Other corneal endothelial genes, such as COL8A1 , TJP1 , and ATP1A1 , did not show significant differences among TZ cells with and without coculture. Overall, pluripotency genes exhibited very low expressions compared to neural crest, periocular mesenchyme, and CEC genes. Protein Expression of TZ Cells Cocultured With P-LMSCs Western blot results demonstrated that TZ cells cocultured with P-LMSCs showed significantly higher expressions of neural crest markers Nestin ( P < 0.001), Sox9 ( P = 0.033), AP-2α ( P = 0.035), and periocular mesenchyme markers FoxC1 ( P = 0.031), Pitx2 ( P = 0.039), and an increasing trend of neural crest marker Sox10 ( P = 0.055) than TZ cells alone. Expression levels of Nestin ( P < 0.001), Sox9 ( P = 0.026), AP-2α ( P = 0.032), and Sox10 ( P = 0.041) were statistically higher in TZ cells cocultured with P-LMSCs compared to TZ cocultured with A-LMSCs ( A, B). Moreover, CEC markers Na + /K + ATPase, ZO-1, and Col8A1 exhibited similar expressions among TZ cells with different treatments and B4G12 cells ( C). Immunocytochemistry confirmed higher levels of Nestin, Sox9, and Pitx2 in TZ cells cocultured with P-LMSCs than TZ cells in the other two groups . P-LMSC Conditioned Medium Promotes the Regeneration of TZ Cells in Organ-Cultured Mouse Corneas after Endothelial Scraping To test whether P-LMSCs could promote the regeneration of TZ cells in wounded cornea tissues, conditioned medium (CM) from cultured P-LMSCs was added onto organ-cultured mouse corneas after entire endothelial scraping. On day 7, the relative wound area was significantly decreased in the P-LMSC CM group (average relative wound area 59.40% ± 2.75%), compared to the A-LMSC CM (75.97% ± 1.19%, P < 0.001) and basal medium control (84.82% ± 1.96%, P < 0.001). Over 14 days of culture, the wounded corneas in P-LMSC CM (relative wound area 31.1% ± 5.49%, P < 0.001, vs. control) and A-LMSC CM (41.79% ± 8.78%, P = 0.005, vs. control) demonstrated superior regeneration of corneal endothelium compared to the basal medium control (72.18% ± 4.66%). Generally, the regeneration of the corneal endothelium originated from the peripheral endothelium - TZ region . Regeneration of corneal endothelium from the TZ region via proliferation was further confirmed by EdU and ZO-1 labeling. After entire endothelial scraping at 2 days, the EdU incorporation was sparsely detected in the TZ region of the control corneas, but obviously observed in A-LMSC CM and P-LMSC CM groups , supporting the contribution of proliferation for the corneal endothelial recovery. On day 7, EdU was substantially incorporated in the TZ region and periphery endothelium of the corneas in P-LMSC CM compared to those in A-LMSC CM and basal medium control ( A). Furthermore, ZO-1 staining demonstrated the reformation of a contiguous sheet of CECs with tight junctions in periphery endothelium of the corneas cultured in P-LMSC CM, whereas corneas in A-LMSC CM and basal medium control only showed sparse expression of ZO-1 in TZ region (see A, B). To characterize the phenotype of P-LMSCs, we carried out explant culture, and assessed their morphology and protein expression in comparison to A-LMSCs. Results showed that A-LMSC and P-LMSC cultures could be established from anterior and posterior limbal stromal explants, respectively, and showed similar fibroblastic morphology for several passages . Immunocytochemistry results demonstrated that P-LMSCs were positive for mesenchymal marker vimentin and stem cell markers Nestin, TRA-1-60, and Oct3/4, all recognized markers for the identification of A-LMSCs . To investigate the effect of P-LMSCs on TZ cells, we performed P-LMSCs and TZ cell coculture (P-TZ) in a Transwell system, with A-LMSCs and TZ cell coculture (A-TZ) as a comparison group and TZ cells alone as a control group. A showed TZ tissue dissected from the human donor corneal rim after corneal transplantation. The TZ explant included the peripheral endothelium, Schwalbe’s line, and the insert portion of the trabecular meshwork. Primary cultured TZ cells reached confluency in 20 days (see A). P1 TZ cells demonstrated similar elongated and fibroblastic morphology when coculturing with P-LMSCs, as well as in A-LMSC coculture and without coculture. After coculturing for 7 days, TZ cells reached confluency in P-LMSC and A-LMSC coculture groups, whereas there were some vacant areas in the TZ-only control group ( B). To compare the proliferation of TZ cells cocultured with P-LMSCs, A-LMSCs, and without coculture, EdU assay was performed in the three groups. Results showed that a significantly higher proportion of nuclei was positive for EdU labeling in TZ cells after coculturing with P-LMSCs (44.38% ± 0.59%, P < 0.001) and A-LMSCs (36.02% ± 2.23%, P = 0.047) compared to TZ only control (24.11% ± 0.72%; ). A scratch wound assay was performed to compare the wound healing ability of TZ cells. TZ cells cocultured with P-LMSCs demonstrated the highest healing speed after wounding compared to the other two groups ( A). Thirty hours after scratching, the average relative wound area was 4.39% ± 0.60% in the P-TZ group ( P = 0.002 versus control) and 21.12% ± 3.03% in the A-TZ group ( P > 0.05 versus control), compared to 32.36% ± 5.51% in the TZ only control group ( B). To compare the stemness of TZ cells cocultured with P-LMSCs, A-LMSCs, and without coculture, the colony forming assay was performed. The results of cresyl violet colony staining showed that TZ cells cocultured with P-LMSCs yielded more colonies compared to those without coculture . Statistical analysis demonstrated a significantly higher colony forming efficiency in the P-TZ group (3.57% ± 0.24%) than the TZ control group (2.20% ± 0.31%, P = 0.007). The colony forming efficiency of the A-TZ group (3.30% ± 0.15%, P = 0.018) was also statistically higher than the control group. Normfinder analysis identified that HPRT1 and POLR2A were the most stable reference gene pair in this dataset, therefore the expressions of target genes were normalized to the geometric mean of HPRT1 and POLR2A . The results of ddPCR demonstrated that TZ cells cocultured with P-LMSCs expressed significantly higher levels of pluripotency gene NANOG (0.0016 ± 0.0003 vs. 0.0007 ± 0.0002, P = 0.034), neural crest genes SOX9 (0.2326 ± 0.0048 vs. 0.1863 ± 0.0080, P = 0.019), and TFAP2A (0.7271 ± 0.0467 vs. 0.5601 ± 0.0509, P = 0.048), and periocular mesenchyme gene PITX2 (1.9485 ± 0.0350 vs. 1.5235 ± 0.1224, P = 0.019) than the TZ only control . Moreover, the expression of neural crest gene NESTIN in TZ cells cocultured with P-LMSCs was significantly higher than those cocultured with A-LMSCs (1.4644 ± 0.0714 vs. 1.1010 ± 0.0347, P = 0.022) and showed an increasing trend compared to TZ cells without coculture (1.2286 ± 0.0829, P = 0.064). The corneal endothelial gene SLC4A11 in TZ cells was downregulated after coculturing with P-LMSCs (0.0220 ± 0.0003 vs. TZ control 0.0298 ± 0.0009, P = 0.015). However, the corneal endothelial gene AQP1 was upregulated after coculturing with A-LMSCs (1.3046 ± 0.0849 vs. TZ control 0.4168 ± 0.0512, P = 0.004) and P-LMSCs (0.9759 ± 0.1566, P = 0.015). Other corneal endothelial genes, such as COL8A1 , TJP1 , and ATP1A1 , did not show significant differences among TZ cells with and without coculture. Overall, pluripotency genes exhibited very low expressions compared to neural crest, periocular mesenchyme, and CEC genes. Western blot results demonstrated that TZ cells cocultured with P-LMSCs showed significantly higher expressions of neural crest markers Nestin ( P < 0.001), Sox9 ( P = 0.033), AP-2α ( P = 0.035), and periocular mesenchyme markers FoxC1 ( P = 0.031), Pitx2 ( P = 0.039), and an increasing trend of neural crest marker Sox10 ( P = 0.055) than TZ cells alone. Expression levels of Nestin ( P < 0.001), Sox9 ( P = 0.026), AP-2α ( P = 0.032), and Sox10 ( P = 0.041) were statistically higher in TZ cells cocultured with P-LMSCs compared to TZ cocultured with A-LMSCs ( A, B). Moreover, CEC markers Na + /K + ATPase, ZO-1, and Col8A1 exhibited similar expressions among TZ cells with different treatments and B4G12 cells ( C). Immunocytochemistry confirmed higher levels of Nestin, Sox9, and Pitx2 in TZ cells cocultured with P-LMSCs than TZ cells in the other two groups . To test whether P-LMSCs could promote the regeneration of TZ cells in wounded cornea tissues, conditioned medium (CM) from cultured P-LMSCs was added onto organ-cultured mouse corneas after entire endothelial scraping. On day 7, the relative wound area was significantly decreased in the P-LMSC CM group (average relative wound area 59.40% ± 2.75%), compared to the A-LMSC CM (75.97% ± 1.19%, P < 0.001) and basal medium control (84.82% ± 1.96%, P < 0.001). Over 14 days of culture, the wounded corneas in P-LMSC CM (relative wound area 31.1% ± 5.49%, P < 0.001, vs. control) and A-LMSC CM (41.79% ± 8.78%, P = 0.005, vs. control) demonstrated superior regeneration of corneal endothelium compared to the basal medium control (72.18% ± 4.66%). Generally, the regeneration of the corneal endothelium originated from the peripheral endothelium - TZ region . Regeneration of corneal endothelium from the TZ region via proliferation was further confirmed by EdU and ZO-1 labeling. After entire endothelial scraping at 2 days, the EdU incorporation was sparsely detected in the TZ region of the control corneas, but obviously observed in A-LMSC CM and P-LMSC CM groups , supporting the contribution of proliferation for the corneal endothelial recovery. On day 7, EdU was substantially incorporated in the TZ region and periphery endothelium of the corneas in P-LMSC CM compared to those in A-LMSC CM and basal medium control ( A). Furthermore, ZO-1 staining demonstrated the reformation of a contiguous sheet of CECs with tight junctions in periphery endothelium of the corneas cultured in P-LMSC CM, whereas corneas in A-LMSC CM and basal medium control only showed sparse expression of ZO-1 in TZ region (see A, B). This study revealed that P-LMSCs expressed similar proteins as A-LMSCs and demonstrated significantly superior stimulation on the proliferation and stemness of TZ cells in both cell and organ culture models. P-LMSCs expressed the same recognized markers for A-LMSCs, including mesenchymal marker vimentin and stem cell markers Nestin, TRA-1-60, and Oct3/4. , Expression of these markers in A-LMSCs has been demonstrated to be essential for them to prevent differentiation and maintain stemness of limbal epithelial stem cells. This suggests that P-LMSCs might have similar supporting potential for the surrounding stem cells. TZ cells cocultured with P-LMSCs demonstrated more EdU incorporation and higher wound healing speed. These results show that P-LMSCs can support the proliferation and wound healing ability of TZ cells, both critical properties for endogenous corneal endothelial regeneration under pathologic conditions and in response to injury. Stemness, another essential property for tissue repair and regeneration, was compared in TZ cells with different treatments. Our results demonstrated higher colony forming efficiency in TZ cells supported by P-LMSCs than in TZ cells only. The upregulation of pluripotency gene NANOG , neural crest genes SOX9 and TFAP2A , and periocular mesenchyme gene PITX2 in TZ cells after coculturing with P-LMSCs suggest that the expression of stem cell genes was promoted by P-LMSCs. Western blotting and immunocytochemistry results confirmed increased protein levels of neural crest markers Nestin, Sox9, and AP-2α, and periocular mesenchyme markers FoxC1 and Pitx2 in TZ cells with the support of P-LMSCs. These results collectively suggest that P-LMSCs can stimulate the stemness of TZ cells. Previous studies demonstrated the support effect on corneal endothelium by conditioned media from multiple types of MSCs, – and MSCs from sources closer to the CECs may be superior to those farther away. Our findings support that P-LMSCs, rather than A-LMSCs, seemed to be more potent in supporting the stemness of TZ cells, perhaps due to their close anatomical association with TZ cells. The results suggest that P-LMSCs might compose a more favorable microenvironment for the self-renewal and stemness of TZ cells. The differentiated properties of TZ cells appeared to be maintained by LMSCs. Herein, we observed similar gene and protein levels of Na + /K + ATPase, ZO-1, and Col8A1, suggesting little alteration of corneal endothelial marker expression, and perhaps function, in TZ cells supported by P-LMSCs and A-LMSCs. Interestingly, the gene expression of AQP1 in TZ cells was upregulated by both A-LMSC and P-LMSCs, indicating the support of fluid transport function by LMSCs. Although SLC4A11 , encoding solute carrier family 4 member 11, was downregulated in TZ cells after coculturing with P-LMSCs, the expression level of this gene was relatively low compared to other corneal endothelial genes, which might limit its influence on corneal endothelial function. By using an immortalized human corneal endothelial cell line, B4G12, as a positive control for corneal endothelial markers, we demonstrated great potential applications of TZ cells in treating corneal endothelial diseases. We observed similar protein expressions of Na + /K + ATPase, ZO-1, and Col8A1 between B4G12 cells and all TZ cells, suggesting that TZ cells have the capacity to function as normal CECs. We also aimed to use B4G12 cells as a negative control for neural crest and periocular mesenchyme markers. Unexpectedly, B4G12 cells showed protein expressions for AP-2α, Sox10, and Pitx2. This might be attributed to the common problem of genetic or phenotype drift for immortalized cell lines. , Further experiments, such as trans-endothelial electrical resistance measurement and in vivo transplantation, are needed to test the corneal endothelial function of TZ cells. In this study, we provided further evidence supporting TZ cells are the stem cell/progenitor source of corneal endothelium. We observed the regeneration of corneal endothelium from the TZ region after scraping the entire endothelium in mouse corneas, remarkably enhanced by conditioned medium from P-LMSCs. The contribution of proliferation in this process was confirmed by EdU incorporation. Moreover, the reformation of ZO-1 tight junctions within the regenerated endothelium indicated that the intact barrier function was restored by the secretome released by P-LMSCs. These findings were consistent with previous studies which showed that stem cells residing in the TZ region responded to corneal injury to initiate an endothelial repair process. , Our results further suggest that, with the stimulation of P-LMSC secretome, TZ cells demonstrate stronger regeneration capacity to repair the wounded corneal endothelium and restore natural function. The process of corneal endothelial regeneration has not been fully understood. A previous study demonstrated dynamic corneal endothelial–posterior stromal communication traversing Descemet's membrane exclusively in the peripheral area of mature murine and human corneas, indicating that CECs might be directly influenced by mesenchymal cells in the posterior limbus. A recent study identified elastic fibers, composed of bundles of microfibrils with or without an elastin core, exclusively within the posterior peripheral corneal stroma of human early embryonic corneas, and absent throughout the rest of the cornea. These fibers were found directly anterior to the corneal endothelium, and CEC projections were also found branching toward the mesenchymal cells, supporting the existence of local communication between CECs and adjacent posterior mesenchymal cells. Moreover, a prior study showed increased density and proliferation capacity of human corneal endothelium after stimulating stromal secretion of interleukin-1β. Likewise, another study showed that mechanical injury to the corneal endothelium triggered apoptosis of 25% of the posterior stromal cells overlying the site of endothelial injury. These studies collectively suggest interactions between the corneal endothelium and the posterior stroma, and potential niche support from P-LMSCs to TZ cells. However, to delineate the stem cell niche in the posterior limbus, further studies identifying the secreted factors, cell adhesion, and the precise cellular organization between TZ cells and P-LMSCs are needed. Expanding the sample size in future studies could strengthen the robustness of our findings. Nevertheless, our research points to a potential stem cell environment in the posterior limbus, which opens up opportunities for further investigation, including the molecular pathways involved in this crosstalk. Our results suggest that P-LMSCs can stimulate the proliferation, wound healing ability, and stemness of TZ cells in both cell and organ culture models. This supports our hypothesis that TZ cells derive critical support from the posterior limbus, where P-LMSCs play an essential role in these cellular interactions. It will help unravel the underlying mechanisms that govern the cell fate of progenitors for corneal endothelium, in order to explore pharmacological interventions towards cell fate manipulation to enhance endogenous regeneration of corneal endothelium. Supplement 1 |
The Role of 3D Printing and Augmented Reality in the Management of Hepatic Malignancies | 4f630a71-1b3f-446f-807b-9e8ed31921b7 | 11843687 | Surgical Procedures, Operative[mh] | What Technology Is for Medicine and the Necessity to Insert Technology to Medicine Medicine and technology are two communicating vessels, since technology creates and applies theoretical knowledge to aid medicine and medical problems are initiating ideas for a new innovative technology. There are currently, many technological advancements that are aiding medicine or trying to do so. Some of these applications could be, machine learning in fields of medicine, artificial organs, new interactive platforms for medicine education, drug delivery, , robotic surgery, 3-dimensional (3D) printing and many more. Introduction to 3D Printing 3D printing is straightforward in its concept, it is the procedure where “an object is created by starting with nothing and adding material a layer at a time until you have a completed object” or “creating an object by building it up layer by layer, rather than machining it away, the way you would by making something from a block of wood, or squirting something into a mold, as you would for injection-molded plastic parts”. Even though the 3D printing concept is straightforward in simple geometries the difficulty on printing complex geometries rises abruptly. Common problems with printing warping, stringing, over-extrusion and more. These problems need to be addressed by experienced personnel and the many times a series of efforts and valuable time are necessary. It is important in 3D printing to gain the best quality of a 3D printed structure. Some quality control measures include the calibration of the printer, the constant and high-quality material selection, the temperature control during printing, the cooling, the support structures, the optimal extrusion speed and printing speed among others. Moreover, a typical layer height for liver for liver 3D printing using fuzed deposition modeling is 0.2 mm. Another important factor is the printing time which in the articles found in this study range from 8-72 h. 3D printing is something with four decades of life. The first working robot 3D printed is credited to Charles W. Hull created in 1984. Then this printed was commercialized in 1989. Then the rest is history with many types of 3D printers having been designed till this day. These include digital light projection (DLP), fuzed deposition modeling (FDM), stereolithography (SLA), selective laser sintering (SLS), and others. Introduction to Augmented Reality Augmented reality (AR) is a concept related to virtual reality (VR). The AR was invented by Thomas Preston Caudell in 1992. He developed an AR application to view some assembly diagrams. Using elements from virtual reality, augmented reality superimposes them on the actual world through a live video, which plays on an electronic device's screen. In simpler and medical terms AR integrated information regarding the surgical field into surgeons’ mind, so as to aid in the procedure, and simultaneously the surgeon is allowed to maintain direct contact with the environment. As a result, since AR allows for overlaying digital information onto the physical real world, AR is a valuable tool for surgical education and surgical navigation. With the constant development of technology, AR systems have gotten more realistic and portable. Scope of This Review The scope of this review is to find the cases of use of 3D-printing and augmented reality for the use in hepatic malignancies (see ), present them and provide a critical discussion regarding the potential of this technologies and the necessary steps for utilizing these technologies in various fields regarding hepatic malignancies. Medicine and technology are two communicating vessels, since technology creates and applies theoretical knowledge to aid medicine and medical problems are initiating ideas for a new innovative technology. There are currently, many technological advancements that are aiding medicine or trying to do so. Some of these applications could be, machine learning in fields of medicine, artificial organs, new interactive platforms for medicine education, drug delivery, , robotic surgery, 3-dimensional (3D) printing and many more. 3D printing is straightforward in its concept, it is the procedure where “an object is created by starting with nothing and adding material a layer at a time until you have a completed object” or “creating an object by building it up layer by layer, rather than machining it away, the way you would by making something from a block of wood, or squirting something into a mold, as you would for injection-molded plastic parts”. Even though the 3D printing concept is straightforward in simple geometries the difficulty on printing complex geometries rises abruptly. Common problems with printing warping, stringing, over-extrusion and more. These problems need to be addressed by experienced personnel and the many times a series of efforts and valuable time are necessary. It is important in 3D printing to gain the best quality of a 3D printed structure. Some quality control measures include the calibration of the printer, the constant and high-quality material selection, the temperature control during printing, the cooling, the support structures, the optimal extrusion speed and printing speed among others. Moreover, a typical layer height for liver for liver 3D printing using fuzed deposition modeling is 0.2 mm. Another important factor is the printing time which in the articles found in this study range from 8-72 h. 3D printing is something with four decades of life. The first working robot 3D printed is credited to Charles W. Hull created in 1984. Then this printed was commercialized in 1989. Then the rest is history with many types of 3D printers having been designed till this day. These include digital light projection (DLP), fuzed deposition modeling (FDM), stereolithography (SLA), selective laser sintering (SLS), and others. Augmented reality (AR) is a concept related to virtual reality (VR). The AR was invented by Thomas Preston Caudell in 1992. He developed an AR application to view some assembly diagrams. Using elements from virtual reality, augmented reality superimposes them on the actual world through a live video, which plays on an electronic device's screen. In simpler and medical terms AR integrated information regarding the surgical field into surgeons’ mind, so as to aid in the procedure, and simultaneously the surgeon is allowed to maintain direct contact with the environment. As a result, since AR allows for overlaying digital information onto the physical real world, AR is a valuable tool for surgical education and surgical navigation. With the constant development of technology, AR systems have gotten more realistic and portable. The scope of this review is to find the cases of use of 3D-printing and augmented reality for the use in hepatic malignancies (see ), present them and provide a critical discussion regarding the potential of this technologies and the necessary steps for utilizing these technologies in various fields regarding hepatic malignancies. A comprehensive literature search was conducted on two databases (ie, Scopus and Pubmed). The latest search was conducted on September 5, 2024. Various combinations of the terms “3D printing”, “3-dimensional printing”, “Augmented reality”, “Hepatic tumors”, “Hepatic cancer”, “Hepatic malignancies” were used in both databases. One specific search in Pubmed and in Scopus as an example is the following “((3d OR 3-dimentional) AND (printing OR bioprinting)) AND (augmented AND reality) AND ((hepatic OR liver) AND (malignancies OR tumor OR cancer))”. Note that in Scopus the search was within article title-article abstract-keywords. In the review research studies only were included that contained 3D printing and/or AR technology for hepatic malignancies. Pre-Operative Planning and Surgical Education Valls-Esteve et al, presented a novel and low-cost approach to produce patient-specific 3D anatomical models. These were used for hands-on simulation and training. As a result, three hepatic surgeries, where all three livers presented complex hepatic tumors, were planned, with 3D simulators built using 3D printing and silicone molding techniques. Specifically, the tumors were in case #1 biliary tract rhabdomyosarcoma, in case #2 PreText II hepatoblastoma and in case #3 mesenchymal hamartoma. The 3D physical models demonstrated remarkably accurate replications of the actual condition and proved to be more cost-effective than other models. Finally, their constructs allowed for proper simulation training and pre-surgical planning. In the multicenter study called LIV3DPRINT, accurate 3D printing of models from original image sources for use in hepatobiliary surgical planning on complex hepatobiliary tumors, patient communication and teaching was validated. In total thirty-five patients from eight centers were included in the study. The centers were from Spain and Germany. The findings of the study, claims that 3D-printing does not affect the surgical outcome necessarily. In addition, the study claim that 3D-printing hepatic models present a good correlation compared with CT/MRI and surgical pathology and they are useful for education, understanding, and surgical planning. In a recent study of Joo et al, the authors aimed to investigate the usefulness of a personalized, 3D-printed, liver model (ie, transparent) with focal liver lesions for lesion-by-lesion imaging-pathologic matching. In this study 20 patients participated. The authors concluded that these personalized, 3D-printed liver model with focal liver lesions may improve the lesion-by-lesion imaging-pathologic matching for small focal liver lesions. This outcome results in accurate pathologic tumor staging and the obtaining of a trustworthy reference for imaging-detected focal liver lesions. Oshiro et al, in their study, developed a novel structure to solve two problems of the creation of a 3D-printed liver model. The one being the cost and the other being the slightly hindered visibility of the inner structures. In their structure the authors did not use transparent loading material to reduce the overall cost and also having the ability to see inside of the structure. The authors performed a hepatectomy using this 3D-printed structure of a liver model. Specifically, they were able to clearly simulate the resection line and safely perform the surgery. The performed hepatectomy using this 3D-printed liver model took place in a patient with liver metastasis after sigmoid colon cancer resection. Overall, their procedure of 3D liver production was cost effective, fast to produce and improved the visibility of the inside of the 3D-printed structures. Another effort (ie, case report) is from Souzaki et al, where the authors conducted an extended left lobectomy to an underage patient diagnosed with PRETEXT IV hepatoblastoma. Even though the tumor was decreased after the neoadjuvant chemotherapy, the hepatoblastoma was still located at the porta hepatis. The lobectomy was undergone after surgical simulation using 3D printing liver model based on pre-operative computed tomography (CT). According to the authors the positional relationships of liver structures, such as blood vessels, and the tumor in the liver were almost completely matched to the real anatomy of the patient's liver. Finally, the tumor was successfully resected completely. Patient's Education Giehl-Brown et al, conducted a prospective, randomized pilot study. In that study they compared regular patient education to 3D liver model-enhanced (3D-LiMo) surgical education during pre-operative consultation. In most of the cases (ie, 97.5%) the underlying disease, needed hepatobiliary surgery, was a malignancy. The results showed that Patient satisfaction with surgical education is improved by individual 3D-printed liver models as soon as patient education is completed. Furthermore, liver models involve an educational benefit for patients and generally strengthen their participation in the decision-making process. In a technical note published by some of the authors in Tooulias et al, the researchers tried to make the initial accurate representation of the liver (including parenchyma, vessels, tumors) as a digital 3D model. In addition, created a 1:1 scale 3D print of it, depicting the entire size of the liver with the tumor(s). The university's surgical department has used one model of this type to plan complex hepatobiliary surgeries, improve the training of medical students and resident surgeons and fellows and finally, provide more accurate information to the families of the patients and the patients themselves. Valls-Esteve et al, presented a novel and low-cost approach to produce patient-specific 3D anatomical models. These were used for hands-on simulation and training. As a result, three hepatic surgeries, where all three livers presented complex hepatic tumors, were planned, with 3D simulators built using 3D printing and silicone molding techniques. Specifically, the tumors were in case #1 biliary tract rhabdomyosarcoma, in case #2 PreText II hepatoblastoma and in case #3 mesenchymal hamartoma. The 3D physical models demonstrated remarkably accurate replications of the actual condition and proved to be more cost-effective than other models. Finally, their constructs allowed for proper simulation training and pre-surgical planning. In the multicenter study called LIV3DPRINT, accurate 3D printing of models from original image sources for use in hepatobiliary surgical planning on complex hepatobiliary tumors, patient communication and teaching was validated. In total thirty-five patients from eight centers were included in the study. The centers were from Spain and Germany. The findings of the study, claims that 3D-printing does not affect the surgical outcome necessarily. In addition, the study claim that 3D-printing hepatic models present a good correlation compared with CT/MRI and surgical pathology and they are useful for education, understanding, and surgical planning. In a recent study of Joo et al, the authors aimed to investigate the usefulness of a personalized, 3D-printed, liver model (ie, transparent) with focal liver lesions for lesion-by-lesion imaging-pathologic matching. In this study 20 patients participated. The authors concluded that these personalized, 3D-printed liver model with focal liver lesions may improve the lesion-by-lesion imaging-pathologic matching for small focal liver lesions. This outcome results in accurate pathologic tumor staging and the obtaining of a trustworthy reference for imaging-detected focal liver lesions. Oshiro et al, in their study, developed a novel structure to solve two problems of the creation of a 3D-printed liver model. The one being the cost and the other being the slightly hindered visibility of the inner structures. In their structure the authors did not use transparent loading material to reduce the overall cost and also having the ability to see inside of the structure. The authors performed a hepatectomy using this 3D-printed structure of a liver model. Specifically, they were able to clearly simulate the resection line and safely perform the surgery. The performed hepatectomy using this 3D-printed liver model took place in a patient with liver metastasis after sigmoid colon cancer resection. Overall, their procedure of 3D liver production was cost effective, fast to produce and improved the visibility of the inside of the 3D-printed structures. Another effort (ie, case report) is from Souzaki et al, where the authors conducted an extended left lobectomy to an underage patient diagnosed with PRETEXT IV hepatoblastoma. Even though the tumor was decreased after the neoadjuvant chemotherapy, the hepatoblastoma was still located at the porta hepatis. The lobectomy was undergone after surgical simulation using 3D printing liver model based on pre-operative computed tomography (CT). According to the authors the positional relationships of liver structures, such as blood vessels, and the tumor in the liver were almost completely matched to the real anatomy of the patient's liver. Finally, the tumor was successfully resected completely. Giehl-Brown et al, conducted a prospective, randomized pilot study. In that study they compared regular patient education to 3D liver model-enhanced (3D-LiMo) surgical education during pre-operative consultation. In most of the cases (ie, 97.5%) the underlying disease, needed hepatobiliary surgery, was a malignancy. The results showed that Patient satisfaction with surgical education is improved by individual 3D-printed liver models as soon as patient education is completed. Furthermore, liver models involve an educational benefit for patients and generally strengthen their participation in the decision-making process. In a technical note published by some of the authors in Tooulias et al, the researchers tried to make the initial accurate representation of the liver (including parenchyma, vessels, tumors) as a digital 3D model. In addition, created a 1:1 scale 3D print of it, depicting the entire size of the liver with the tumor(s). The university's surgical department has used one model of this type to plan complex hepatobiliary surgeries, improve the training of medical students and resident surgeons and fellows and finally, provide more accurate information to the families of the patients and the patients themselves. In this case report Bonomi et al, present a case of a 65 y.o. male patient where they mention the practical use of a custom-made 3D model, for a case of bilateral colorectal liver metastases following neoadjuvant chemotherapy. It is mentioned that pre-operative visualization of 3D reconstructions changed significantly the pre-operative surgical plan. In the supplementary material, the authors present how they used the 3D model in an AR setup to aid them in the completion of the surgery and mention that the availability of the 3D model in the operating room was crucial in the surgical field to guide safe surgical pathways. The patient was discharged in the 12 th postoperative day. They report that the pre-operative surgical strategy was drastically altered by the pre-operative visualization of 3D reconstructions. A team located in France has recently developed Hepataug, an AR software. It has the ability to project the invisible intrahepatic tumors onto the laparoscopic images and also allows the surgeon to localize them precisely. Recently, this team aimed to measure the 3D tumor prediction error of Hepataug. This took place with eight 3D virtual livers. The livers were created from the CT scan of a healthy human liver. The virtual livers were then deformed, and 3D printed to form 3D liver phantoms and these were placed inside a pelvitrainer. Finally, the surgeons had to point the center of eight virtual tumors per liver with a pointing tool. As a result of this work, despite a lower precision of AR for the tumors that were located in the posterior part of the liver, it could allow the surgeons to access these lesions without completely mobilizing the liver. This results in decreasing the surgical trauma. In a clinical case report from Tang et al, the authors utilized AR display technology based on videos to help with surgical resection of hilar cholangiocarcinoma and concomitant left hemihepatectomy. They used enhanced CT and MRCP data to generate 3D images of the patient's hepatic hilar structures. The AR technology was used for the intra-operative navigation during open tumor resection and hemihepatectomy and the pre-operative surgical planning. Furthermore, pre-operatively, a 3D-printed model of the patient's biliary tree and surrounding vasculature was made using the reconstructed 3D images. The patient over a 9-month follow-up was disease-free and complication-free. In a recent study of Wang et al, 11 patients underwent laparoscopic right hemi-hepatectomy plus total caudate lobectomy with AR navigation technology and the anterior approach being utilized in this operation. The patients had type II or IIIa perihilar cholangiocarcinoma. As a result, the right hemi-hepatectomy plus total caudate lobectomy successfully in all 11 patients. Huber et al, conducted a prospective randomized-controlled pilot trial. The researchers aimed with this trial to evaluate computer-assisted 3D-navigation for liver surgery. In the study patients were randomized in two groups (ie, in non-navigated or navigated group) and 20 liver tumors from 16 patients were used for the study to perform open liver resections or primary laparoscopic resections. The navigation system was used for intra-operative computer-assisted 3D-navigation. They came to the conclusion that although surgical accuracy is not yet better than the existing level of intra-operative orientation, intra-operative navigation is a technology that can be employed safely during liver resection. Specifically, they did not find any differences between the navigated and the non-navigated group regarding morbidity, duration of surgery nor length of hospital stay among others. In the current literature review, there are many limitations that can be identified. To begin with, this review is a simple review not a systematic review that follows PRISMA protocol or even a systematized review following partially , PRISMA protocol. In addition, 3D printing technology and the AR technology are presented and acknowledged but close/related technologies such as virtual reality (VR) or 3D bioprinting are not presented. The presented cases were different, and they had a lack of homogeneity in the application of 3D printing and AR technologies thus the presented elements from each study were different. In the current literature search, there were twelve works presented here. Note that these works present mostly the feasibility in the use of 3D printing and AR. Meaning that the true potential and the benefit or not of these applications have not been tested as a whole. For instance, an idea for a future study would be to use 3D printing and AR technologies combined together or not to see how the learning curve of a team of young surgeons in hepatic surgeries will be using these technologies in comparison to another team of young surgeons not using them. In addition, more qualitative and quantitative elements regarding a surgeon will have to be measured. For instance, the heart rate as an indication for stress should be measured during surgery using these technologies and not using them to see whether they aid the surgeon in his stress management and overall confidence. Additionally, future studies should try to find ways to reduce the overall printing cost and printing time. It is of high importance also for the researchers to include in their research the effort to find one material or a group of materials with more realistic representation of a tissue. This tissue-mimicking material could be a hydrogel. An example of a hydrogel-based 3D printed material can be seen in Liu et al,. Artificial intelligence could also be used as a tool in 3D printing and AR to aid in the procedure with a fast segmentation of the scanned body. Moreover, researchers should try to increase the overall anatomic accuracy and simplify the procedure from the beginning till the production of the AR image or the printed model. Furthermore, technologies such as digital twins could be utilized in AR technology in order to have real time organs while performing a procedure such as hepatectomy, which would be used to aid the surgeon and also it could be used to educate and train resident doctors and students how a surgical procedure is in real time. Last but not least, the standardization of the procedures is a key element to make this kind of research comparable from surgical center to surgical center around the world. Note also that the current understanding and practice of 3D printing in hepatic malignancies has limitations. First of all, the cost of each structure can be high and for some health systems this is a major restriction for the use and spreading of 3D printing. Secondly, specialized equipment and personnel is necessary for the production of the structures. In addition, in urgent surgeries time is limited so the time for the production of a structure with good quality is sometimes larger that the time the surgeons have to start the operation. In this case it is better for the surgical team to opt for the AR technologies. Finally, to those who want to take a deeper understanding in 3D printing and AR application in hepatic, and not only, surgery there are many works that can be used by the readers to comprehend the use of those two technologies. , - Specifically, in the work of Kasai et al, the role of 3D printing and virtual reality in liver surgery are discussed. In a recent systematic review in pediatric patients, the enhancement of surgical planning with the aid of 3D visualization techniques is discussed. In addition, a bibliometric analysis from 2023 discusses the role of 3D technology in liver cancer resection. Finally, in Gavriilidis et al, the state of the art in navigated liver surgery is discussed and in Kościuszko et al, the pre-operative planning in pediatric liver tumor surgery is discussed. In addition, another concept of 3D printing the so-called 3D bioprinting, where cells and pharmaceutical ingredients are printed in 3D structures, shows an upgrade of 3D printing that can be used for creating more detailed, closer to reality and educational if not functioning structures for liver malignant cases. The readers can find the in vivo applications of 3D bioprinting in the works of Hong et al. In conclusion, the two technologies mentioned in the article (ie, 3D printing technology and AR) can be used and are being in use alone or in combination to contribute in the treatment and management of hepatic malignancies. Till this day there are encouraging results (eg, efforts to reduce cost of 3D printing, improve surgical pre-planning, usefulness in training of medical personnel and in education of patients) from the use of these technologies, which show that the medical staff (eg, resident surgeons, surgeons) can help patients and improve their part of the healthcare system. It is also true that it is necessary for more works to be published to see how helpful or not these two technologies are in the management of the hepatic malignancies and also to provide easier-to-implement technologies with low-cost solutions and with little trained personnel. |
Review of genetic and pharmacogenetic differences in cytotoxic and targeted therapies for pancreatic cancer in African Americans | 7ecfc064-410a-4287-97cb-26853bbcf9ca | 10639003 | Pharmacology[mh] | In the United States, pancreatic ductal adenocarcinoma (PDAC) is the third and soon to be second leading cause of cancer death; approximately 4% survive five years given 85% of patients have advanced unresectable disease at diagnosis. , Africa has the lowest age-standardized incidence rate (2.2/100,000) while Europe (7.7/100,000) and the Americas (7.6/100,000) have the highest rates worldwide. However, in the United States, African Americans have a 50% – 90% higher incidence of PDAC and have a poorer prognosis compared to other racial groups. , These disparities are multifactorial and may reflect underlying differences in socioeconomic status and access to healthcare. Furthermore, chemotherapeutics such as gemcitabine and paclitaxel have been approved for PDAC treatment, yet responses are less than ideal. This is particularly evident for African Americans, who show worse outcomes compared to Caucasians. , We propose that disparities in PDAC may also be a consequence of genetic variation resulting in variable (1) cancer therapeutic response (pharmacogenetics), (2) cancer predisposition, and (3) the unknown somatic mutational landscape of PDAC in African American limiting the benefit of druggable genes (precision oncology) in the group. According to Dere and Suto, pharmacogenetics is the study of individual genetic influence on drug response and pharmacogenomics studies the genetic influence of multiple mutations that concurrently influence a patient’s therapeutic response. This review will focus on improving the understanding of how genetics impacts PDAC drug metabolism, the efficacy of therapeutically targeted germline and tumor mutations with consequential outcomes resulting in disparities. This narrative review included articles published from 1995 to 2022. The primary method was to find genes that interfere with drug metabolism of FDA-approved drugs. A PubMed search included the keywords: pharmacogenetics and pancreatic cancer. Next, searches of the name of each FDA-approved drug, toxicity, and African or Black was performed. Gene names found in studies were searched on PubMed using either “race” or “ethnicity” or “African” or “Black” as keywords. Some studies were found from the cited by or cited section from PubMed. The University of Alabama at Birmingham Cancer data analysis portal (UACLAN) was then used to identify whether genes found both in literature and database were significant for overall survival (OS) from PDAC. Data availability statement The data generated in this study are publicly available in UALCAN database [ http://ualcan.path.uab.edu/analysis.html ] and PubMed [ https://pubmed.ncbi.nlm.nih.gov/ ] The data generated in this study are publicly available in UALCAN database [ http://ualcan.path.uab.edu/analysis.html ] and PubMed [ https://pubmed.ncbi.nlm.nih.gov/ ] Our review consisted of 51 peer reviewed studies that investigated the pharmacogenetics of cytotoxic therapies, therapeutics targeting cancer predisposition and DNA repair deficiency, and therapeutics targeting somatic mutations. The prevalence of these genes in African and European ancestral populations are outlined in . Pharmacogenetics of cytotoxic therapies Fluoropyrimidines: fluorouracil (5-FU and Capecitabine). 5-Fluorouracil is a cytostatic antimetabolite drug utilized to treat various solid tumors. Capecitabine is an inactive prodrug of 5-FU that requires a 3-step conversion to 5-FU by carboxylesterase (CES), cytidine deaminase (CDA), and thymidine phosphorylase (TYMP) to become activated. , The University of Alabama at Birmingham Cancer data analysis portal (UACLAN) showed that CDA expression is not statistically significant for OS. , Allelic frequencies of CDA associated with decreased enzymatic activity are found in . Dihydropyrimidine dehydrogenase (DPD) gene is encoded by DPYD which serves as the rate- limiting enzyme for metabolizing fluoropyrimidines. Complete (homozygous) DPD deficiency is rare and can result in significant toxicities including myelosuppression, diarrhea, and mucositis. , Clinical Pharmacogenetics Implementation Consortium (CPIC) guidelines categorizes these patients as poor metabolizers having two nonfunctional alleles or one nonfunctional allele plus one allele with decreased function. Partial DPYD deficient patients are intermediate metabolizers (one normal function plus either one nonfunctional or decreased function allele, or two alleles with decreased function), according to CPIC guidelines. Approximately 3-5% of patients have partial DPYD deficiency and overdose can still occur in these patients. , The prevalence of DPD deficiency is more common in African Americans ranging from 4 to 12% compared to 3 to 5% in European Americans. , – Specific variants of DPYD vary between African and European Americans. According to CPIC guidelines, the DPD variant HapB3 with c.1129–5923C>G is found in 4.7% of Europeans and is the most common variant for decreased function among Europeans. Offer et al. assessed circulating mononuclear-cell DPD enzyme activity in African American ( n = 94) and European-American ( n = 81) participants. The DPYD-Y186C variant was only identified in African ancestral populations and showed 46% lower DPD activity in carriers as compared with noncarriers. Genetic polymorphisms of thymidylate synthase (TYMS), are also associated with 5-FU toxicity. TYMS is involved in DNA synthesis and is inhibited by fluoropyrimidines. Inhibition of DNA synthesis eventually leads to cell death. Patients with a lower expression of TYMS mRNA (2R/2R or 2R/3R polymorphisms) experience more severe side effects because they are less able to inhibit the effects of 5-fu. Patients with higher TMYS expression (3R/3R) genotype experience less toxicity. The prevalence of 2R is high in both European and African ancestral populations. Khushman et al compared genotypic differences of TYMS and discovered 28% of African Americans had the 2R/2R genotype which was marginally higher than European Americans at 24%. Data from UALCAN demonstrated that TYMS expression level was significantly associated with survival for PDAC. , Irinotecan. Irinotecan is a prodrug that kills cancer cells by inhibiting DNA topoisomerase 1. Common toxicities after administration of irinotecan include neutropenia occurring in 20-54% of patients and diarrhea occurring in 11-23%. – Uridine diphosphate (UDP) glucuronosyltransferase (UGT) facilitates the glucuronidation of many drugs, including the active SN-38 (7-ethyl-10-hydroxycamptothecin), which subsequently increase water solubility. This increase in water solubility allows for the elimination of bilirubin and urine. Therefore, a decrease in the biologic activity of UGT can lead to irinotecan toxicity due to the accumulation of SN-38. , The UGTA1 allele is a subfamily of UGT with varying numbers of thymine adenine (TA) repeats on the promoter region. The wild type allele UGT1A1*1 has 6 TA repeats and is associated with normal function of the gene. Alleles with TA repeats higher than the wild type allele UGT1A1*1, are typically associated with decreased transcription levels, and subsequent lower activity as seen in UGT1A1*28 and UGT1A1*37 alleles with 7 and 8 TA repeats on their promotor region, respectively. , These genes with lower activity have greater risk for dose-limiting toxicities. There are 3 UGT1A1 polymorphisms that have been widely studied and associated with toxicity: UGT1A1*28, UGT1A1*93, and UGT1A1*6. The prevalence of UGT1A1*28 expression in African and European populations is 43% and 39% respectively, indicating lower gene activity in more patients with African ancestry. An additional polymorphism significantly associated with toxicity include UGT1A1*93 found in 34% and 27% African and European populations, respectively. UGT1A*6 is commonly seen in the Asian population (15%) and less commonly in African (0.1%) and European (1%) ancestral populations. Decreased enzymatic activity of UGT1A1 is the hallmark of Gilbert syndrome, causing mild unconjugated hyperbilirubinemia. According to CPIC guidelines, UGT1A1*28/*28 and UGT1A1*6/*6 are the most common genotypes associated with Gilbert syndrome. Package insert for irinotecan includes a recommendation for UGT1A1 testing. Despite an association with increased toxicity, expression level for UGT1A1 was not significant for PDAC survival according to the UALCAN database. , Gemcitabine. Gemcitabine (2’-deoxy-2’,2’-difluorocytidine, dFdC) has variable responses ranging from lack of efficacy to severe cytotoxicity that may be attributed to variability in drug exposure and metabolism. Several variants in genes directly involved in gemcitabine metabolism have been reported to impact gemcitabine response (examples are deoxycytidine kinase, DCK; cytidine deaminase, CDA; and transporters: SLC28A1, SLC28A2, SLC28A3, SLC29A1 (hENT1 expression), SLC29A2 (hENT2 expression), ABCB1, ABCC2, and ABCC10. – Fukunaga et al. evaluated the allelic frequencies of polymorphisms involved in gemcitabine metabolism. From the 14 polymorphisms studied, 12 were seen in Africans and Europeans and 9 statistically significantly or highly statistically significantly varied between both groups. Statistically significant polymorphisms include: DCK 2190A>T, POLA2 2089G>A, SLC28A1 1543G>A and the highly significant polymorphs were- CDA 79A>C, CDA 208G>A, DCTD 315T>C, SLC28A1 1576T>C, SLC28A2 283A>C, TYMS 1494del. Wong et al reviewed genetic polymorphisms with clinical relevance for cancer patients on gemcitabine therapy. In this study, the prevalence of CDA 79A>C (30-36% in Europeans and 4-10.8% in Africans) and CDA 435 C>T (30-32.5% in Europeans and 36% in Africans) were linked to lower progression free survival for patients receiving gemcitabine. The CDA 208 G>A was linked to increased neutropenia and decreased clearance of the drug. Mohelnikova-Duchonova et al collected tissue samples from patients with PDAC who received surgical resection and found that higher expression of SLC281 was associated with worse OS. Although POLA2 is primarily involved in DNA repair, it’s knockdown increased the chemoresistance to Gemcitabine for patients with lung cancer and it’s expression level was significant for OS among patients with PDAC. , UALCAN data report overall expression levels of DCK, ABCB1, ABCC2, ABCC10, SLC28A1, SLC28A2, SLC28A3, SLC29A1, SLC29A2, CDA, and POLE are not significant for OS and race. , Nab-paclitaxel. The combination of nab-paclitaxel with gemcitabine (GemNab) is a recommended first-line treatment option for patients with advanced or metastatic PDAC. Neutropenia, thrombocytopenia, and diarrhea are toxicities related to GemNab. Several studies evaluated the presence of single nucleotide polymorphisms (SNPs) in the ATP Binding Cassette Subfamily B Member (ABCB transporters), and in the CDA genes in patients treated with gemcitabine or nab-paclitaxel chemotherapy and developing severe adverse effects. For instance, the association of hematological toxicity in patients with the CDA 79 A>C mutation. Nab-paclitaxel can inactivate CDA, which results in inhibition of gemcitabine catabolism, leading to higher levels of gemcitabine and a higher response rate in the genetically engineered mouse models known as KPC models. Prevalence of CDA polymorphisms in African and European ancestral populations of CDA 79 A>C (Lys27Gln), CDA 208 G>A (Ala70Thr), CDA 435 C>T (Thr145Thr) are outlined in . Polymorphisms in ABCB gene have been reviewed and correlated with diverse expression of efflux pumps in several tissue compartments and, as a result, modified drug levels. In addition, ABCB1 polymorphisms have been related to hematological adverse effects in cancer patients receiving nab-paclitaxel. Genes that code for solute carriers (SLCs) are linked to paclitaxel-induced cytotoxicity. A subset of genes - SLC31A2, SLC43A1, SLC35A5, and SLC41A2 were shown to be associated with paclitaxel sensitivity and to regulate SNPs that were also linked to paclitaxel-induced cytotoxicity. A population with Northern and Western European heritage from Utah, a Yoruba community in Ibadan, Nigeria, and an African American population from the Southwest of the United States were used in the study. Increased expression of these three SLC genes, SLC31A2, SLC41A2, and SLC35A5, was linked to paclitaxel resistance in lymphoblastoid cell lines in this study. The same study discovered a link between higher SLC43A1 expression and increased drug sensitivity. The UALCAN database was utilized to evaluate the association between the expression of SLC31A2, SLC43A1, SLC35A5, and SLC41A2 in PDAC patients, and survival among different races. , Decreased expression of SLC31A2, SLC43A1, SLC35A5 among Europeans and African Americans corresponded with increases in survival. For SLC41A2 gene however, an increased expression in Europeans led to a slightly shorter survival time whilst a decreased expression led to a much higher survival rate. African Americans, on the other hand, showed increased survival with increased expression and a decreased survival with a decreased expression. Europeans generally had a higher survival rate in comparison with African Americans across all four genes, highlighting disparities arising from genetic polymorphisms. The analysis of this data, however, showed no statistical significance. , CYP2C8 is involved in paclitaxel metabolism, and UALCAN data showed overall expression was significantly associated with survival for PDAC. , Compared to the wild type CYP2C8*1, CYP2C8*3, was linked to neuropathy as a result of clearance reduction. There is lower CYP2C8*3 allelic frequency in African Americans than European Americans. , The CYP2C8*2 frequency was expressed in 18% African Americans and no Caucasians and is also associated with lower paclitaxel clearance. Therapeutics targeting DNA repair deficiency genes Platinums (oxaliplatin/cisplatin). Platinum agents such as oxaliplatin and cisplatin are cytotoxic chemotherapies used to treat a variety of cancers including lung, PDAC, and colorectal cancers. Platinum agents form covalent cross-links of platinum-DNA between the bases of damaged DNA. Once crosslinks are formed, DNA repair is prohibited eventually leading to cell death. Efficacy of platinum agents are shown to be highly reliant on the inability of tumor cells to repair damaged DNA. Therefore, they are sensitive to cells with homologous repair deficiencies (HRD), including ATM, PALB2 and BRCA mutations. – The NCCN recommends combination regimens with platinum agents for those with HRD due to inherent oxaliplatin and cisplatin sensitivity. HRDs are found in 5-9% of PDAC. African Americans have significantly higher number of HRDs across multiple tumor types compared to other racial groups. An analysis by Hsiao et al., showed the genes TP53, R151, and SMG are most strongly associated with HRD predisposition and is common among African Americans, Caucasians, and Asians. Repair genes such as nucleotide excision repair (NER), base excision repair (BER), ECCR1, and ECCR2 are important biomarkers that influence the efficacy of platinum treatments. High expression of these repair genes can also increase cisplatin and oxaliplatin resistance as they interfere with the DNA damaging mechanism of the treatments. , Based on UALCAN data, ERCC1 and ECCR2 genes are not significantly associated with OS. , The difficulty in treatment with platinum agents is achieving little toxicity with an effective regimen and personalization of treatment. O’Donnell et al shown that Asian derived genes were the most sensitive while African derived genes were the most resistant to cytotoxicity from platinum agents. A study by Gao et al., investigated whole blood samples of 320 males to access racial differences in the expression of these biomarkers. Results demonstrated that the CC genotype of ERCC1 N118N (500C>T) was seen more frequently in African Americans at 76% compared to European Americans at 21% and TT genotype was seen more in European Americans at 30% compared to African Americans at 3%. The study references that the TT genotype may reduce expression of ERCC1 and subsequently increase sensitivity to cisplatin, though a review by Amable (2016) reports this to be conflicting. Further exploration on the connection between these repair genes and racial difference in treatment response is needed. PARP inhibitors. Poly (ADP-ribose) polymerase PARP inhibitors use multiple mechanisms such as trapping PARP-1 and PARP-2 on DNA at single stranded break sites to hinder appropriate repair of the damaged DNA. The detained repair mechanism will eventually kill tumor cells due to further compiling of damaged DNA. – Tumors that harbor a defect in HRD are particularly vulnerable to PARP inhibitors such as tumors that harbor BRCA-1 and BRCA-2 mutations. In PDAC, BRCA-1 mutations have a 1% prevalence and BRCA-2 have a 5-10% prevalence. Based on UALCAN data, expression levels of BRCA2 is statistically significant for OS, but not for BRCA1. Olaparib is a PARP inhibitor approved for maintenance treatment of adult patients with deleterious germline BRCA mutated metastatic PDAC whose disease has not progressed on at least 16 weeks of a first-line platinum-based chemotherapy regimen. In a randomized, double-blind, placebo-controlled, phase 3 trial (POLO trial), maintenance Olaparib provided significantly longer progression free survival (7.4 months vs. 3.8 months) but not overall survival. Subsequently, national guidelines recommend germline testing for all patients after a confirmed diagnosis of PDAC. , A retrospective analysis by Golan et al. examined the prevalence of BRCA mutations among African Americans with PDAC. This analysis geographically examined the first 2,206 patients with metastatic PDAC screened to enter the phase 3 POLO trial. African Americans had higher rates of newly identified germline BRCA mutations (10.7%) in addition to the highest prevalence in total population (13.8%) when compared to other racial groups. Investigators noted potential disparities in genetic testing amongst racial groups. These results suggest further evaluation is needed with larger sample sizes. Other targeted therapeutics Pembrolizumab. Pembrolizumab is a PD-1 inhibitor which targets immune checkpoint proteins and has transformed the care of metastatic melanoma, non-small lung cancer, and many malignancies. However, results of clinical trials involving immunotherapy in PDAC have been disappointing. The identification of subsets of patients who will positively respond to immunotherapies continues to be investigated. Pembrolizumab is FDA approved for patients with microsatellite instability-high (MSI-H) tumors. Patients with mismatch repair deficiency (dMMR)/ MSI-H are most likely to have sustained clinical responses to immunotherapy. dMMR and MSI-H mutations are rarely occurring in <5% of all diagnosed cancers and is the hallmark of autosomal dominant hereditary condition Lynch syndrome (LS). In addition to being at high risk for colorectal and endometrial cancer, patients with LS have an 8.6-fold increase in developing PDAC. Rosenblum et al. detected 1 in 200 African ancestral populations to harbor LS variants compared with 1 in 518 European ancestral populations. NTRK inhibitors. Approximately 1% of solid tumor malignancies harbor NTRK fusion genes. They are extremely rare in PDAC and its incidence is < 1% in African and European ancestral patients with PDAC. 63 The NTRK1, NTRK2, and NTRK3 genes encode the receptors (proteins) TRKA, TRKB, and TRKC which are drivers of oncogenesis. NTRK fusion genes can be detected by DNA sequencing, RNA sequencing and plasma cell-free DNA profiling. Expression level of these genes are not significantly correlated with OS in PDAC based on data from UALCAN. , Entrectinib and larotrectinib are NTRK fusion protein inhibitors that are FDA approved with a tumor agnostic indication for metastatic cancers or unresectable cancers with NTRK gene fusions that have progressed or have no other alternative treatment options. In clinical trial cohorts (ALKA-372-001, STARTRK-1, STARTRK-2), entrectinib demonstrated an objective response rate (ORR) of 57% (95% CI, 43.2–70.8) and median duration of response (DOR) of 10.4 months (95% CI, 7.1–not evaluable). A pooled analysis of three phase 1/2 clinical trials with larotrectinib in 153 evaluable patients demonstrated an ORR of 79% (95% CI, 72.0–85.0), 16% complete response (CR), and a median DOR of 35.2 months (95% CI, 21.2–not evaluable). There were few patients with PDAC that were studied in these pivotal clinical trials. Two of three patients with PDAC treated with entrectinib had a PR, and one patient with PDAC treated with larotrectinib had a PR. Although rare, testing for NTRK fusion genes should be performed on all patients with unresectable or metastatic PDAC as these agents have demonstrated promising response rate. Fluoropyrimidines: fluorouracil (5-FU and Capecitabine). 5-Fluorouracil is a cytostatic antimetabolite drug utilized to treat various solid tumors. Capecitabine is an inactive prodrug of 5-FU that requires a 3-step conversion to 5-FU by carboxylesterase (CES), cytidine deaminase (CDA), and thymidine phosphorylase (TYMP) to become activated. , The University of Alabama at Birmingham Cancer data analysis portal (UACLAN) showed that CDA expression is not statistically significant for OS. , Allelic frequencies of CDA associated with decreased enzymatic activity are found in . Dihydropyrimidine dehydrogenase (DPD) gene is encoded by DPYD which serves as the rate- limiting enzyme for metabolizing fluoropyrimidines. Complete (homozygous) DPD deficiency is rare and can result in significant toxicities including myelosuppression, diarrhea, and mucositis. , Clinical Pharmacogenetics Implementation Consortium (CPIC) guidelines categorizes these patients as poor metabolizers having two nonfunctional alleles or one nonfunctional allele plus one allele with decreased function. Partial DPYD deficient patients are intermediate metabolizers (one normal function plus either one nonfunctional or decreased function allele, or two alleles with decreased function), according to CPIC guidelines. Approximately 3-5% of patients have partial DPYD deficiency and overdose can still occur in these patients. , The prevalence of DPD deficiency is more common in African Americans ranging from 4 to 12% compared to 3 to 5% in European Americans. , – Specific variants of DPYD vary between African and European Americans. According to CPIC guidelines, the DPD variant HapB3 with c.1129–5923C>G is found in 4.7% of Europeans and is the most common variant for decreased function among Europeans. Offer et al. assessed circulating mononuclear-cell DPD enzyme activity in African American ( n = 94) and European-American ( n = 81) participants. The DPYD-Y186C variant was only identified in African ancestral populations and showed 46% lower DPD activity in carriers as compared with noncarriers. Genetic polymorphisms of thymidylate synthase (TYMS), are also associated with 5-FU toxicity. TYMS is involved in DNA synthesis and is inhibited by fluoropyrimidines. Inhibition of DNA synthesis eventually leads to cell death. Patients with a lower expression of TYMS mRNA (2R/2R or 2R/3R polymorphisms) experience more severe side effects because they are less able to inhibit the effects of 5-fu. Patients with higher TMYS expression (3R/3R) genotype experience less toxicity. The prevalence of 2R is high in both European and African ancestral populations. Khushman et al compared genotypic differences of TYMS and discovered 28% of African Americans had the 2R/2R genotype which was marginally higher than European Americans at 24%. Data from UALCAN demonstrated that TYMS expression level was significantly associated with survival for PDAC. , Irinotecan. Irinotecan is a prodrug that kills cancer cells by inhibiting DNA topoisomerase 1. Common toxicities after administration of irinotecan include neutropenia occurring in 20-54% of patients and diarrhea occurring in 11-23%. – Uridine diphosphate (UDP) glucuronosyltransferase (UGT) facilitates the glucuronidation of many drugs, including the active SN-38 (7-ethyl-10-hydroxycamptothecin), which subsequently increase water solubility. This increase in water solubility allows for the elimination of bilirubin and urine. Therefore, a decrease in the biologic activity of UGT can lead to irinotecan toxicity due to the accumulation of SN-38. , The UGTA1 allele is a subfamily of UGT with varying numbers of thymine adenine (TA) repeats on the promoter region. The wild type allele UGT1A1*1 has 6 TA repeats and is associated with normal function of the gene. Alleles with TA repeats higher than the wild type allele UGT1A1*1, are typically associated with decreased transcription levels, and subsequent lower activity as seen in UGT1A1*28 and UGT1A1*37 alleles with 7 and 8 TA repeats on their promotor region, respectively. , These genes with lower activity have greater risk for dose-limiting toxicities. There are 3 UGT1A1 polymorphisms that have been widely studied and associated with toxicity: UGT1A1*28, UGT1A1*93, and UGT1A1*6. The prevalence of UGT1A1*28 expression in African and European populations is 43% and 39% respectively, indicating lower gene activity in more patients with African ancestry. An additional polymorphism significantly associated with toxicity include UGT1A1*93 found in 34% and 27% African and European populations, respectively. UGT1A*6 is commonly seen in the Asian population (15%) and less commonly in African (0.1%) and European (1%) ancestral populations. Decreased enzymatic activity of UGT1A1 is the hallmark of Gilbert syndrome, causing mild unconjugated hyperbilirubinemia. According to CPIC guidelines, UGT1A1*28/*28 and UGT1A1*6/*6 are the most common genotypes associated with Gilbert syndrome. Package insert for irinotecan includes a recommendation for UGT1A1 testing. Despite an association with increased toxicity, expression level for UGT1A1 was not significant for PDAC survival according to the UALCAN database. , Gemcitabine. Gemcitabine (2’-deoxy-2’,2’-difluorocytidine, dFdC) has variable responses ranging from lack of efficacy to severe cytotoxicity that may be attributed to variability in drug exposure and metabolism. Several variants in genes directly involved in gemcitabine metabolism have been reported to impact gemcitabine response (examples are deoxycytidine kinase, DCK; cytidine deaminase, CDA; and transporters: SLC28A1, SLC28A2, SLC28A3, SLC29A1 (hENT1 expression), SLC29A2 (hENT2 expression), ABCB1, ABCC2, and ABCC10. – Fukunaga et al. evaluated the allelic frequencies of polymorphisms involved in gemcitabine metabolism. From the 14 polymorphisms studied, 12 were seen in Africans and Europeans and 9 statistically significantly or highly statistically significantly varied between both groups. Statistically significant polymorphisms include: DCK 2190A>T, POLA2 2089G>A, SLC28A1 1543G>A and the highly significant polymorphs were- CDA 79A>C, CDA 208G>A, DCTD 315T>C, SLC28A1 1576T>C, SLC28A2 283A>C, TYMS 1494del. Wong et al reviewed genetic polymorphisms with clinical relevance for cancer patients on gemcitabine therapy. In this study, the prevalence of CDA 79A>C (30-36% in Europeans and 4-10.8% in Africans) and CDA 435 C>T (30-32.5% in Europeans and 36% in Africans) were linked to lower progression free survival for patients receiving gemcitabine. The CDA 208 G>A was linked to increased neutropenia and decreased clearance of the drug. Mohelnikova-Duchonova et al collected tissue samples from patients with PDAC who received surgical resection and found that higher expression of SLC281 was associated with worse OS. Although POLA2 is primarily involved in DNA repair, it’s knockdown increased the chemoresistance to Gemcitabine for patients with lung cancer and it’s expression level was significant for OS among patients with PDAC. , UALCAN data report overall expression levels of DCK, ABCB1, ABCC2, ABCC10, SLC28A1, SLC28A2, SLC28A3, SLC29A1, SLC29A2, CDA, and POLE are not significant for OS and race. , Nab-paclitaxel. The combination of nab-paclitaxel with gemcitabine (GemNab) is a recommended first-line treatment option for patients with advanced or metastatic PDAC. Neutropenia, thrombocytopenia, and diarrhea are toxicities related to GemNab. Several studies evaluated the presence of single nucleotide polymorphisms (SNPs) in the ATP Binding Cassette Subfamily B Member (ABCB transporters), and in the CDA genes in patients treated with gemcitabine or nab-paclitaxel chemotherapy and developing severe adverse effects. For instance, the association of hematological toxicity in patients with the CDA 79 A>C mutation. Nab-paclitaxel can inactivate CDA, which results in inhibition of gemcitabine catabolism, leading to higher levels of gemcitabine and a higher response rate in the genetically engineered mouse models known as KPC models. Prevalence of CDA polymorphisms in African and European ancestral populations of CDA 79 A>C (Lys27Gln), CDA 208 G>A (Ala70Thr), CDA 435 C>T (Thr145Thr) are outlined in . Polymorphisms in ABCB gene have been reviewed and correlated with diverse expression of efflux pumps in several tissue compartments and, as a result, modified drug levels. In addition, ABCB1 polymorphisms have been related to hematological adverse effects in cancer patients receiving nab-paclitaxel. Genes that code for solute carriers (SLCs) are linked to paclitaxel-induced cytotoxicity. A subset of genes - SLC31A2, SLC43A1, SLC35A5, and SLC41A2 were shown to be associated with paclitaxel sensitivity and to regulate SNPs that were also linked to paclitaxel-induced cytotoxicity. A population with Northern and Western European heritage from Utah, a Yoruba community in Ibadan, Nigeria, and an African American population from the Southwest of the United States were used in the study. Increased expression of these three SLC genes, SLC31A2, SLC41A2, and SLC35A5, was linked to paclitaxel resistance in lymphoblastoid cell lines in this study. The same study discovered a link between higher SLC43A1 expression and increased drug sensitivity. The UALCAN database was utilized to evaluate the association between the expression of SLC31A2, SLC43A1, SLC35A5, and SLC41A2 in PDAC patients, and survival among different races. , Decreased expression of SLC31A2, SLC43A1, SLC35A5 among Europeans and African Americans corresponded with increases in survival. For SLC41A2 gene however, an increased expression in Europeans led to a slightly shorter survival time whilst a decreased expression led to a much higher survival rate. African Americans, on the other hand, showed increased survival with increased expression and a decreased survival with a decreased expression. Europeans generally had a higher survival rate in comparison with African Americans across all four genes, highlighting disparities arising from genetic polymorphisms. The analysis of this data, however, showed no statistical significance. , CYP2C8 is involved in paclitaxel metabolism, and UALCAN data showed overall expression was significantly associated with survival for PDAC. , Compared to the wild type CYP2C8*1, CYP2C8*3, was linked to neuropathy as a result of clearance reduction. There is lower CYP2C8*3 allelic frequency in African Americans than European Americans. , The CYP2C8*2 frequency was expressed in 18% African Americans and no Caucasians and is also associated with lower paclitaxel clearance. 5-Fluorouracil is a cytostatic antimetabolite drug utilized to treat various solid tumors. Capecitabine is an inactive prodrug of 5-FU that requires a 3-step conversion to 5-FU by carboxylesterase (CES), cytidine deaminase (CDA), and thymidine phosphorylase (TYMP) to become activated. , The University of Alabama at Birmingham Cancer data analysis portal (UACLAN) showed that CDA expression is not statistically significant for OS. , Allelic frequencies of CDA associated with decreased enzymatic activity are found in . Dihydropyrimidine dehydrogenase (DPD) gene is encoded by DPYD which serves as the rate- limiting enzyme for metabolizing fluoropyrimidines. Complete (homozygous) DPD deficiency is rare and can result in significant toxicities including myelosuppression, diarrhea, and mucositis. , Clinical Pharmacogenetics Implementation Consortium (CPIC) guidelines categorizes these patients as poor metabolizers having two nonfunctional alleles or one nonfunctional allele plus one allele with decreased function. Partial DPYD deficient patients are intermediate metabolizers (one normal function plus either one nonfunctional or decreased function allele, or two alleles with decreased function), according to CPIC guidelines. Approximately 3-5% of patients have partial DPYD deficiency and overdose can still occur in these patients. , The prevalence of DPD deficiency is more common in African Americans ranging from 4 to 12% compared to 3 to 5% in European Americans. , – Specific variants of DPYD vary between African and European Americans. According to CPIC guidelines, the DPD variant HapB3 with c.1129–5923C>G is found in 4.7% of Europeans and is the most common variant for decreased function among Europeans. Offer et al. assessed circulating mononuclear-cell DPD enzyme activity in African American ( n = 94) and European-American ( n = 81) participants. The DPYD-Y186C variant was only identified in African ancestral populations and showed 46% lower DPD activity in carriers as compared with noncarriers. Genetic polymorphisms of thymidylate synthase (TYMS), are also associated with 5-FU toxicity. TYMS is involved in DNA synthesis and is inhibited by fluoropyrimidines. Inhibition of DNA synthesis eventually leads to cell death. Patients with a lower expression of TYMS mRNA (2R/2R or 2R/3R polymorphisms) experience more severe side effects because they are less able to inhibit the effects of 5-fu. Patients with higher TMYS expression (3R/3R) genotype experience less toxicity. The prevalence of 2R is high in both European and African ancestral populations. Khushman et al compared genotypic differences of TYMS and discovered 28% of African Americans had the 2R/2R genotype which was marginally higher than European Americans at 24%. Data from UALCAN demonstrated that TYMS expression level was significantly associated with survival for PDAC. , Irinotecan is a prodrug that kills cancer cells by inhibiting DNA topoisomerase 1. Common toxicities after administration of irinotecan include neutropenia occurring in 20-54% of patients and diarrhea occurring in 11-23%. – Uridine diphosphate (UDP) glucuronosyltransferase (UGT) facilitates the glucuronidation of many drugs, including the active SN-38 (7-ethyl-10-hydroxycamptothecin), which subsequently increase water solubility. This increase in water solubility allows for the elimination of bilirubin and urine. Therefore, a decrease in the biologic activity of UGT can lead to irinotecan toxicity due to the accumulation of SN-38. , The UGTA1 allele is a subfamily of UGT with varying numbers of thymine adenine (TA) repeats on the promoter region. The wild type allele UGT1A1*1 has 6 TA repeats and is associated with normal function of the gene. Alleles with TA repeats higher than the wild type allele UGT1A1*1, are typically associated with decreased transcription levels, and subsequent lower activity as seen in UGT1A1*28 and UGT1A1*37 alleles with 7 and 8 TA repeats on their promotor region, respectively. , These genes with lower activity have greater risk for dose-limiting toxicities. There are 3 UGT1A1 polymorphisms that have been widely studied and associated with toxicity: UGT1A1*28, UGT1A1*93, and UGT1A1*6. The prevalence of UGT1A1*28 expression in African and European populations is 43% and 39% respectively, indicating lower gene activity in more patients with African ancestry. An additional polymorphism significantly associated with toxicity include UGT1A1*93 found in 34% and 27% African and European populations, respectively. UGT1A*6 is commonly seen in the Asian population (15%) and less commonly in African (0.1%) and European (1%) ancestral populations. Decreased enzymatic activity of UGT1A1 is the hallmark of Gilbert syndrome, causing mild unconjugated hyperbilirubinemia. According to CPIC guidelines, UGT1A1*28/*28 and UGT1A1*6/*6 are the most common genotypes associated with Gilbert syndrome. Package insert for irinotecan includes a recommendation for UGT1A1 testing. Despite an association with increased toxicity, expression level for UGT1A1 was not significant for PDAC survival according to the UALCAN database. , Gemcitabine (2’-deoxy-2’,2’-difluorocytidine, dFdC) has variable responses ranging from lack of efficacy to severe cytotoxicity that may be attributed to variability in drug exposure and metabolism. Several variants in genes directly involved in gemcitabine metabolism have been reported to impact gemcitabine response (examples are deoxycytidine kinase, DCK; cytidine deaminase, CDA; and transporters: SLC28A1, SLC28A2, SLC28A3, SLC29A1 (hENT1 expression), SLC29A2 (hENT2 expression), ABCB1, ABCC2, and ABCC10. – Fukunaga et al. evaluated the allelic frequencies of polymorphisms involved in gemcitabine metabolism. From the 14 polymorphisms studied, 12 were seen in Africans and Europeans and 9 statistically significantly or highly statistically significantly varied between both groups. Statistically significant polymorphisms include: DCK 2190A>T, POLA2 2089G>A, SLC28A1 1543G>A and the highly significant polymorphs were- CDA 79A>C, CDA 208G>A, DCTD 315T>C, SLC28A1 1576T>C, SLC28A2 283A>C, TYMS 1494del. Wong et al reviewed genetic polymorphisms with clinical relevance for cancer patients on gemcitabine therapy. In this study, the prevalence of CDA 79A>C (30-36% in Europeans and 4-10.8% in Africans) and CDA 435 C>T (30-32.5% in Europeans and 36% in Africans) were linked to lower progression free survival for patients receiving gemcitabine. The CDA 208 G>A was linked to increased neutropenia and decreased clearance of the drug. Mohelnikova-Duchonova et al collected tissue samples from patients with PDAC who received surgical resection and found that higher expression of SLC281 was associated with worse OS. Although POLA2 is primarily involved in DNA repair, it’s knockdown increased the chemoresistance to Gemcitabine for patients with lung cancer and it’s expression level was significant for OS among patients with PDAC. , UALCAN data report overall expression levels of DCK, ABCB1, ABCC2, ABCC10, SLC28A1, SLC28A2, SLC28A3, SLC29A1, SLC29A2, CDA, and POLE are not significant for OS and race. , The combination of nab-paclitaxel with gemcitabine (GemNab) is a recommended first-line treatment option for patients with advanced or metastatic PDAC. Neutropenia, thrombocytopenia, and diarrhea are toxicities related to GemNab. Several studies evaluated the presence of single nucleotide polymorphisms (SNPs) in the ATP Binding Cassette Subfamily B Member (ABCB transporters), and in the CDA genes in patients treated with gemcitabine or nab-paclitaxel chemotherapy and developing severe adverse effects. For instance, the association of hematological toxicity in patients with the CDA 79 A>C mutation. Nab-paclitaxel can inactivate CDA, which results in inhibition of gemcitabine catabolism, leading to higher levels of gemcitabine and a higher response rate in the genetically engineered mouse models known as KPC models. Prevalence of CDA polymorphisms in African and European ancestral populations of CDA 79 A>C (Lys27Gln), CDA 208 G>A (Ala70Thr), CDA 435 C>T (Thr145Thr) are outlined in . Polymorphisms in ABCB gene have been reviewed and correlated with diverse expression of efflux pumps in several tissue compartments and, as a result, modified drug levels. In addition, ABCB1 polymorphisms have been related to hematological adverse effects in cancer patients receiving nab-paclitaxel. Genes that code for solute carriers (SLCs) are linked to paclitaxel-induced cytotoxicity. A subset of genes - SLC31A2, SLC43A1, SLC35A5, and SLC41A2 were shown to be associated with paclitaxel sensitivity and to regulate SNPs that were also linked to paclitaxel-induced cytotoxicity. A population with Northern and Western European heritage from Utah, a Yoruba community in Ibadan, Nigeria, and an African American population from the Southwest of the United States were used in the study. Increased expression of these three SLC genes, SLC31A2, SLC41A2, and SLC35A5, was linked to paclitaxel resistance in lymphoblastoid cell lines in this study. The same study discovered a link between higher SLC43A1 expression and increased drug sensitivity. The UALCAN database was utilized to evaluate the association between the expression of SLC31A2, SLC43A1, SLC35A5, and SLC41A2 in PDAC patients, and survival among different races. , Decreased expression of SLC31A2, SLC43A1, SLC35A5 among Europeans and African Americans corresponded with increases in survival. For SLC41A2 gene however, an increased expression in Europeans led to a slightly shorter survival time whilst a decreased expression led to a much higher survival rate. African Americans, on the other hand, showed increased survival with increased expression and a decreased survival with a decreased expression. Europeans generally had a higher survival rate in comparison with African Americans across all four genes, highlighting disparities arising from genetic polymorphisms. The analysis of this data, however, showed no statistical significance. , CYP2C8 is involved in paclitaxel metabolism, and UALCAN data showed overall expression was significantly associated with survival for PDAC. , Compared to the wild type CYP2C8*1, CYP2C8*3, was linked to neuropathy as a result of clearance reduction. There is lower CYP2C8*3 allelic frequency in African Americans than European Americans. , The CYP2C8*2 frequency was expressed in 18% African Americans and no Caucasians and is also associated with lower paclitaxel clearance. Platinums (oxaliplatin/cisplatin). Platinum agents such as oxaliplatin and cisplatin are cytotoxic chemotherapies used to treat a variety of cancers including lung, PDAC, and colorectal cancers. Platinum agents form covalent cross-links of platinum-DNA between the bases of damaged DNA. Once crosslinks are formed, DNA repair is prohibited eventually leading to cell death. Efficacy of platinum agents are shown to be highly reliant on the inability of tumor cells to repair damaged DNA. Therefore, they are sensitive to cells with homologous repair deficiencies (HRD), including ATM, PALB2 and BRCA mutations. – The NCCN recommends combination regimens with platinum agents for those with HRD due to inherent oxaliplatin and cisplatin sensitivity. HRDs are found in 5-9% of PDAC. African Americans have significantly higher number of HRDs across multiple tumor types compared to other racial groups. An analysis by Hsiao et al., showed the genes TP53, R151, and SMG are most strongly associated with HRD predisposition and is common among African Americans, Caucasians, and Asians. Repair genes such as nucleotide excision repair (NER), base excision repair (BER), ECCR1, and ECCR2 are important biomarkers that influence the efficacy of platinum treatments. High expression of these repair genes can also increase cisplatin and oxaliplatin resistance as they interfere with the DNA damaging mechanism of the treatments. , Based on UALCAN data, ERCC1 and ECCR2 genes are not significantly associated with OS. , The difficulty in treatment with platinum agents is achieving little toxicity with an effective regimen and personalization of treatment. O’Donnell et al shown that Asian derived genes were the most sensitive while African derived genes were the most resistant to cytotoxicity from platinum agents. A study by Gao et al., investigated whole blood samples of 320 males to access racial differences in the expression of these biomarkers. Results demonstrated that the CC genotype of ERCC1 N118N (500C>T) was seen more frequently in African Americans at 76% compared to European Americans at 21% and TT genotype was seen more in European Americans at 30% compared to African Americans at 3%. The study references that the TT genotype may reduce expression of ERCC1 and subsequently increase sensitivity to cisplatin, though a review by Amable (2016) reports this to be conflicting. Further exploration on the connection between these repair genes and racial difference in treatment response is needed. PARP inhibitors. Poly (ADP-ribose) polymerase PARP inhibitors use multiple mechanisms such as trapping PARP-1 and PARP-2 on DNA at single stranded break sites to hinder appropriate repair of the damaged DNA. The detained repair mechanism will eventually kill tumor cells due to further compiling of damaged DNA. – Tumors that harbor a defect in HRD are particularly vulnerable to PARP inhibitors such as tumors that harbor BRCA-1 and BRCA-2 mutations. In PDAC, BRCA-1 mutations have a 1% prevalence and BRCA-2 have a 5-10% prevalence. Based on UALCAN data, expression levels of BRCA2 is statistically significant for OS, but not for BRCA1. Olaparib is a PARP inhibitor approved for maintenance treatment of adult patients with deleterious germline BRCA mutated metastatic PDAC whose disease has not progressed on at least 16 weeks of a first-line platinum-based chemotherapy regimen. In a randomized, double-blind, placebo-controlled, phase 3 trial (POLO trial), maintenance Olaparib provided significantly longer progression free survival (7.4 months vs. 3.8 months) but not overall survival. Subsequently, national guidelines recommend germline testing for all patients after a confirmed diagnosis of PDAC. , A retrospective analysis by Golan et al. examined the prevalence of BRCA mutations among African Americans with PDAC. This analysis geographically examined the first 2,206 patients with metastatic PDAC screened to enter the phase 3 POLO trial. African Americans had higher rates of newly identified germline BRCA mutations (10.7%) in addition to the highest prevalence in total population (13.8%) when compared to other racial groups. Investigators noted potential disparities in genetic testing amongst racial groups. These results suggest further evaluation is needed with larger sample sizes. Platinum agents such as oxaliplatin and cisplatin are cytotoxic chemotherapies used to treat a variety of cancers including lung, PDAC, and colorectal cancers. Platinum agents form covalent cross-links of platinum-DNA between the bases of damaged DNA. Once crosslinks are formed, DNA repair is prohibited eventually leading to cell death. Efficacy of platinum agents are shown to be highly reliant on the inability of tumor cells to repair damaged DNA. Therefore, they are sensitive to cells with homologous repair deficiencies (HRD), including ATM, PALB2 and BRCA mutations. – The NCCN recommends combination regimens with platinum agents for those with HRD due to inherent oxaliplatin and cisplatin sensitivity. HRDs are found in 5-9% of PDAC. African Americans have significantly higher number of HRDs across multiple tumor types compared to other racial groups. An analysis by Hsiao et al., showed the genes TP53, R151, and SMG are most strongly associated with HRD predisposition and is common among African Americans, Caucasians, and Asians. Repair genes such as nucleotide excision repair (NER), base excision repair (BER), ECCR1, and ECCR2 are important biomarkers that influence the efficacy of platinum treatments. High expression of these repair genes can also increase cisplatin and oxaliplatin resistance as they interfere with the DNA damaging mechanism of the treatments. , Based on UALCAN data, ERCC1 and ECCR2 genes are not significantly associated with OS. , The difficulty in treatment with platinum agents is achieving little toxicity with an effective regimen and personalization of treatment. O’Donnell et al shown that Asian derived genes were the most sensitive while African derived genes were the most resistant to cytotoxicity from platinum agents. A study by Gao et al., investigated whole blood samples of 320 males to access racial differences in the expression of these biomarkers. Results demonstrated that the CC genotype of ERCC1 N118N (500C>T) was seen more frequently in African Americans at 76% compared to European Americans at 21% and TT genotype was seen more in European Americans at 30% compared to African Americans at 3%. The study references that the TT genotype may reduce expression of ERCC1 and subsequently increase sensitivity to cisplatin, though a review by Amable (2016) reports this to be conflicting. Further exploration on the connection between these repair genes and racial difference in treatment response is needed. Poly (ADP-ribose) polymerase PARP inhibitors use multiple mechanisms such as trapping PARP-1 and PARP-2 on DNA at single stranded break sites to hinder appropriate repair of the damaged DNA. The detained repair mechanism will eventually kill tumor cells due to further compiling of damaged DNA. – Tumors that harbor a defect in HRD are particularly vulnerable to PARP inhibitors such as tumors that harbor BRCA-1 and BRCA-2 mutations. In PDAC, BRCA-1 mutations have a 1% prevalence and BRCA-2 have a 5-10% prevalence. Based on UALCAN data, expression levels of BRCA2 is statistically significant for OS, but not for BRCA1. Olaparib is a PARP inhibitor approved for maintenance treatment of adult patients with deleterious germline BRCA mutated metastatic PDAC whose disease has not progressed on at least 16 weeks of a first-line platinum-based chemotherapy regimen. In a randomized, double-blind, placebo-controlled, phase 3 trial (POLO trial), maintenance Olaparib provided significantly longer progression free survival (7.4 months vs. 3.8 months) but not overall survival. Subsequently, national guidelines recommend germline testing for all patients after a confirmed diagnosis of PDAC. , A retrospective analysis by Golan et al. examined the prevalence of BRCA mutations among African Americans with PDAC. This analysis geographically examined the first 2,206 patients with metastatic PDAC screened to enter the phase 3 POLO trial. African Americans had higher rates of newly identified germline BRCA mutations (10.7%) in addition to the highest prevalence in total population (13.8%) when compared to other racial groups. Investigators noted potential disparities in genetic testing amongst racial groups. These results suggest further evaluation is needed with larger sample sizes. Pembrolizumab. Pembrolizumab is a PD-1 inhibitor which targets immune checkpoint proteins and has transformed the care of metastatic melanoma, non-small lung cancer, and many malignancies. However, results of clinical trials involving immunotherapy in PDAC have been disappointing. The identification of subsets of patients who will positively respond to immunotherapies continues to be investigated. Pembrolizumab is FDA approved for patients with microsatellite instability-high (MSI-H) tumors. Patients with mismatch repair deficiency (dMMR)/ MSI-H are most likely to have sustained clinical responses to immunotherapy. dMMR and MSI-H mutations are rarely occurring in <5% of all diagnosed cancers and is the hallmark of autosomal dominant hereditary condition Lynch syndrome (LS). In addition to being at high risk for colorectal and endometrial cancer, patients with LS have an 8.6-fold increase in developing PDAC. Rosenblum et al. detected 1 in 200 African ancestral populations to harbor LS variants compared with 1 in 518 European ancestral populations. NTRK inhibitors. Approximately 1% of solid tumor malignancies harbor NTRK fusion genes. They are extremely rare in PDAC and its incidence is < 1% in African and European ancestral patients with PDAC. 63 The NTRK1, NTRK2, and NTRK3 genes encode the receptors (proteins) TRKA, TRKB, and TRKC which are drivers of oncogenesis. NTRK fusion genes can be detected by DNA sequencing, RNA sequencing and plasma cell-free DNA profiling. Expression level of these genes are not significantly correlated with OS in PDAC based on data from UALCAN. , Entrectinib and larotrectinib are NTRK fusion protein inhibitors that are FDA approved with a tumor agnostic indication for metastatic cancers or unresectable cancers with NTRK gene fusions that have progressed or have no other alternative treatment options. In clinical trial cohorts (ALKA-372-001, STARTRK-1, STARTRK-2), entrectinib demonstrated an objective response rate (ORR) of 57% (95% CI, 43.2–70.8) and median duration of response (DOR) of 10.4 months (95% CI, 7.1–not evaluable). A pooled analysis of three phase 1/2 clinical trials with larotrectinib in 153 evaluable patients demonstrated an ORR of 79% (95% CI, 72.0–85.0), 16% complete response (CR), and a median DOR of 35.2 months (95% CI, 21.2–not evaluable). There were few patients with PDAC that were studied in these pivotal clinical trials. Two of three patients with PDAC treated with entrectinib had a PR, and one patient with PDAC treated with larotrectinib had a PR. Although rare, testing for NTRK fusion genes should be performed on all patients with unresectable or metastatic PDAC as these agents have demonstrated promising response rate. Pembrolizumab is a PD-1 inhibitor which targets immune checkpoint proteins and has transformed the care of metastatic melanoma, non-small lung cancer, and many malignancies. However, results of clinical trials involving immunotherapy in PDAC have been disappointing. The identification of subsets of patients who will positively respond to immunotherapies continues to be investigated. Pembrolizumab is FDA approved for patients with microsatellite instability-high (MSI-H) tumors. Patients with mismatch repair deficiency (dMMR)/ MSI-H are most likely to have sustained clinical responses to immunotherapy. dMMR and MSI-H mutations are rarely occurring in <5% of all diagnosed cancers and is the hallmark of autosomal dominant hereditary condition Lynch syndrome (LS). In addition to being at high risk for colorectal and endometrial cancer, patients with LS have an 8.6-fold increase in developing PDAC. Rosenblum et al. detected 1 in 200 African ancestral populations to harbor LS variants compared with 1 in 518 European ancestral populations. Approximately 1% of solid tumor malignancies harbor NTRK fusion genes. They are extremely rare in PDAC and its incidence is < 1% in African and European ancestral patients with PDAC. 63 The NTRK1, NTRK2, and NTRK3 genes encode the receptors (proteins) TRKA, TRKB, and TRKC which are drivers of oncogenesis. NTRK fusion genes can be detected by DNA sequencing, RNA sequencing and plasma cell-free DNA profiling. Expression level of these genes are not significantly correlated with OS in PDAC based on data from UALCAN. , Entrectinib and larotrectinib are NTRK fusion protein inhibitors that are FDA approved with a tumor agnostic indication for metastatic cancers or unresectable cancers with NTRK gene fusions that have progressed or have no other alternative treatment options. In clinical trial cohorts (ALKA-372-001, STARTRK-1, STARTRK-2), entrectinib demonstrated an objective response rate (ORR) of 57% (95% CI, 43.2–70.8) and median duration of response (DOR) of 10.4 months (95% CI, 7.1–not evaluable). A pooled analysis of three phase 1/2 clinical trials with larotrectinib in 153 evaluable patients demonstrated an ORR of 79% (95% CI, 72.0–85.0), 16% complete response (CR), and a median DOR of 35.2 months (95% CI, 21.2–not evaluable). There were few patients with PDAC that were studied in these pivotal clinical trials. Two of three patients with PDAC treated with entrectinib had a PR, and one patient with PDAC treated with larotrectinib had a PR. Although rare, testing for NTRK fusion genes should be performed on all patients with unresectable or metastatic PDAC as these agents have demonstrated promising response rate. This review identified genetic mutations in African Americans that may affect toxicity and therapeutic response of cytotoxic and targeted therapies that are FDA approved for the treatment of PDAC. The UALCAN database was used to ascertain if these genetic mutations were significant for OS. , This database is publicly available and provides cancer genomics data to analyze genes of interest based on projects from The Cancer Genome Atlas (TCGA) and other projects. Genes associated with OS include TYMS ( p = 0.0052), POLA2 ( p = 0.022), CYP2C8 ( p = 0.027), BRCA2 ( p = 0.027), PALB2, and RRM1. Nonetheless, the low sample size of African Americans ( n = 6) compared with Caucasians ( n = 155) posed a limitation for properly stratifying racial differences in OS using this database. Furthermore, projects such as TCGA are important for developing therapeutics geared towards precision medicine. However, the small sizes of non-European samples limit generalizability of precision therapeutics conceptualized from the TCGA projects. Additionally, such small sample sizes also overlook diversities occurring within African ancestral populations and other racial groups. Categorizing all Black persons as a homogeneous group when collating cancer mortality data may mask significant differences linked to their ancestry as well as undermine the environmental and epigenetic risk factors associated with the disease. For instance, a study by Pinheiro et al., 2016, demonstrated that Black people born in the United States had higher PDAC mortality rates as compared to Black people born outside the United States. A subsequent study showed that PDAC mortality rates between 2012 and 2017 among African American males and females in Florida, Minnesota, California, and New York were higher than those of Afro- Caribbeans and Africans. Afro- Caribbeans have PDAC mortality rates in between that of African Americans and Africans. Africans have the lowest PDAC mortality in comparison with other Black populations. Additionally, Africa has the widest genetic diversity. For instance, de Rocha et al., reported the allelic distribution of the DPD variant rs2297595-C to vary across Africa in 1% of west Africans, 6-10% East Africans and 12% South African Zulus. Cancer research geared towards comparing mortality rates among Black people of different ancestry in the United States, could reveal significant variations that could lead to better prevention, control, and treatment options. Medical mistrust among African Americans plays a role in low participation in omics-based cancer research. Significant contributors to medical mistrust include the belief that the study will be financially profitable for researchers with little advancement to medicine, the possibility of negative side effects, and the uncertainty of who can access their personal information. Added barriers include difficulties commuting to study location , and establishing consistent communication with participants. Study recruiters have approached these barriers with improved inclusivity efforts such as including the utilization of diverse health navigators and community health workers representing those communities being asked to participate, develop culturally competent clinical trial education, frequent appointment reminders, offer peer support, in addition to connecting the participants with other helpful resources. , Furthermore, incorporating travel reimbursements, monetary incentives, culture competency training of medical staff, along with the positive expectation of improving cancer treatments increase the likelihood of study participation. With the continued proper intentionality during recruitment, the racial variability in genome-wide association studies is promising. Park et al., identified over 700 loci in genome-wide association studies that increase cancer risk, and 21 loci increase pancreatic cancer risk. Less than 1% of the 700 loci have been identified in African populations while more than 80% identified in European populations. Improving the sample sizes of African Americans and other minorities in biorepositories can lead to a greater understanding of pharmacogenetic differences and achieve an equitable distribution of care and outcomes for patients with PDAC. |
Detection of hypophosphatasia in hospitalised adults in rheumatology and internal medicine departments: a multicentre study over 10 years | 7494f3dd-155f-4453-ae88-c70c8aa07cc9 | 11002352 | Internal Medicine[mh] | Hypophosphatasia is a rare and often undiagnosed disorder. Low alkaline phosphatase (ALP) values are overlooked by a majority of clinicians. In this multicentre study, low ALPs are poorly recognised by clinicians. 70.8% of patients treated with bisphosphonates never underwent ALP measurement before treatment initiation. Using a combination of multiple evocative symptoms to select patients for genetic testing seems interesting as a means of increasing the diagnosis rate and control healthcare costs. Mild to moderate adult hypophosphatasia may be more frequent than previously thought. Sensitisation of clinicians to ALP values is needed. ALP measurement should be mandatory in the secondary osteoporosis investigations before bisphosphonate treatment initiation. Hypophosphatasia (HPP) is a rare genetic skeletal disease due to an inherited metabolic disorder caused by mutations of the ALPL gene coding for tissue non-specific alkaline phosphatase (TNSALP). Prevalence of severe forms is estimated as ranging from 1/100 000 to 1/300 00, while prevalence of mild HPP was estimated at 1/6370 in Europe. Six forms of the disease have been defined: perinatal severe HPP, perinatal benign HPP, infantile HPP, childhood HPP, adult HPP and odontohypophosphatasia. In adults, clinical manifestations are dominated by fractures and joint disease. The most evocative fractures are localised at the metatarsals. These fractures are usually recurrent, with delayed consolidation potentially leading to pseudarthrosis. Other typical fractures affect the femoral diaphysis and occur mainly in the lateral cortex of the subtrochanteric region. Joint disease is represented mainly by calcium pyrophosphate deposition disease. While elevated ALP is usually taken into account by clinicians, low ALP levels are easily overlooked. A monocentre study in a tertiary care hospital in France found that notification was given in only 3% of cases. The aetiologies of low ALP are multiple and differ according to hypophosphatasaemia temporality. Furthermore, these causes are often unknown by clinicians. The aims of this study were to estimate the recognition of hypophosphatasaemia in rheumatology and internal medicine departments, to analyse the characteristics of the population presenting persistently low ALP measurements and to estimate the number of patients highly suspected of adult HPP. Secondary analyses were performed to compare patients with persistently low ALP measurements while using or not using bisphosphonates, and those with or without an identified cause of persistently low ALP levels. Study design This retrospective, descriptive and multicentre study included patients from the University hospitals of Poitiers, Nantes, Rennes, Brest, Angers and Tours. It consisted of the detection of low ALP measurement at least twice among patients hospitalised in the departments of Rheumatology and Internal Medicine between 1 July 2007 and 1 July 2017. No limit in duration between two measurements was necessary. We followed STrengthening the Reporting of OBservational studies in Epidemiology (STROBE) instructions throughout this work. In France, internal medicine departments are departments encompassing a combination of geriatrics, clinical immunology, infectious diseases, oncohaematological diseases and rheumatology subspecialties dealing with systemic autoimmune and autoinflammatory disorders. Patients The listing of patients was established from records of the Biochemistry Department of several French university hospitals by laboratory database request on the criteria of low ALP values ≤35 U/L (normal range: 40–120 IU/L). Low cut-off values were identical for men and women in the laboratories that performed the analysis. A minimum of 2 low ALP values (≤35 IU/L) was required to minimise the likelihood of an analytical error; 35 U/L was defined as it is an average between the lower bounds of adult normal values and less exclusive than previous studies have set the limit at 30 UI/L. Patients who had previously denied or restricted access to their record for research purposes and aged less than 18 years were excluded from the study. Once authorisations were given, paper and electronic medical records were used to search patient history, symptoms, laboratory results (basic tests, calcium phosphate metabolism and specialised blood tests (bone ALP)), bone densitometry, X-rays, CT- scan and MRI results. If done, the genetic test was notified. Bone demineralisation was defined as a T-score <−2.5 SD. Chondrocalcinosis and hydroxyapatite deposition disease were diagnosed based on the aspect of the calcifications visualised on X-rays. Only non-traumatic fractures were considered in the analyses. Scoliosis was considered as present if described in the radiologist reports of spine imaging. Aetiologies for low ALP were defined as follows: Corticosteroids were considered as a possible cause of hypophosphatasaemia if patients received at least a very high dose of corticosteroids (>100 mg per day) using the standardised nomenclature for glucocorticoid dosages by Buttgereit et al . Severe anaemia was defined as haemoglobin (Hb) <60 g/L. Pernicious anaemia was considered if patients were currently not substituted. Hypothyroidism was considered if patients were not substituted and thyroid-stimulating hormone (TSH) was higher than 4 mUI/L. Hepatic insufficiency was considered if prothrombin time was lower than 50%. Hypervitaminosis D was considered if 25-OH vitamin D was higher than 150 ng/L. Hypomagnesaemia was considered if magnesium level was lower than 0.7 mmol/L. Vitamin C insufficiency was considered for levels lower than 2.5 mg/L. Zinc insufficiency was considered for levels lower than 9 µmol/L. Cushing disease, coeliac disease and Wilson disease were considered only if they were not currently being treated or at equilibrium. Intensive care stay was considered if it was an actual stay or less than 1 month before. Ongoing oncohaematological disease, bisphosphonate treatment, denosumab treatment, septicaemia, inflammatory disease flare and intravenous immunoglobulins were considered as potential causes of low ALP. To determine the number of patients for whom low ALP ≤35 U/L was recognised and noted in their records, the discharge summary, the diagnosis written in the letter and/or the ICD-10 (International Classification of Diseases 10th Revision) code were used. Patients were defined as possible HPP if they exhibited at least three symptoms evocative of HPP in addition to persistent low ALP (arthralgia, fractures, stress fractures, low bone density, dental abnormalities, chondrocalcinosis, scoliosis, high B 6 levels or high urinary phosphoethanolamine). Biochemical assays Both of the instruments measure ALP activity by a kinetic rate method in which a colourless organic phosphate ester substrate (nitrophenylphosphate) is hydrolysed by ALP to the yellow-coloured product p-nitrophenol and phosphate at pH of 10.3, thereby explaining the term ‘alkaline’. Changes in absorbance at 410 nm are directly proportional to the enzymatic activity of ALP. A requirement of two low ALP values (≤35 U/L) was set so as to minimise the likelihood of low ALP results due to analytic error. For each selected patient, all previous ALP values against time were visually examined to determine the temporal pattern of the qualifying serum ALP values and to separate two groups of patients. When the temporal pattern of ALP values indicated a precipitous fall from usually normal values, the patient was considered to have acute hypophosphatasaemia. Diagnostic conditions and circumstances associated with acute hypophosphatasaemia were analysed. Laboratory used Glims, JMP or DXLab software. When the temporal pattern of ALP values indicated a persistently low ALP or only 2 values, both of them under 35 U/L, the patient was considered to have persistent hypophosphatasaemia. More precise analysis was carried out to identify detailed patient history, symptoms, laboratory results (basic tests including calcium phosphate metabolism exploration and specialised blood tests such as bone ALP), bone densitometry, X-rays, CT scan and MRI results and genetic test when they had been made. To determine the number of patients in whom persistently low ALP ≤35 U/L was recognised, the discharge summary, the written diagnosis and/or the ICD-10 code (E833) were used. Statistical methodology Qualitative data were expressed as percentages and quantitative data as means±SD. Analysis was conducted using the Student’s t-test (or Wilcoxon, as appropriate) for quantitative data and χ² (or Fisher’s exact test, as appropriate) for qualitative data. A p value of 0.05 was considered as significant. Statistical analysis was performed by using SAS software, V.9.1 (SAS Institute) and GraphPad Prism (GraphPad Software, California). Patient and public involvement Patients or members of the public were not involved in the design, or conduct, or reporting, or dissemination plans of the research. This retrospective, descriptive and multicentre study included patients from the University hospitals of Poitiers, Nantes, Rennes, Brest, Angers and Tours. It consisted of the detection of low ALP measurement at least twice among patients hospitalised in the departments of Rheumatology and Internal Medicine between 1 July 2007 and 1 July 2017. No limit in duration between two measurements was necessary. We followed STrengthening the Reporting of OBservational studies in Epidemiology (STROBE) instructions throughout this work. In France, internal medicine departments are departments encompassing a combination of geriatrics, clinical immunology, infectious diseases, oncohaematological diseases and rheumatology subspecialties dealing with systemic autoimmune and autoinflammatory disorders. The listing of patients was established from records of the Biochemistry Department of several French university hospitals by laboratory database request on the criteria of low ALP values ≤35 U/L (normal range: 40–120 IU/L). Low cut-off values were identical for men and women in the laboratories that performed the analysis. A minimum of 2 low ALP values (≤35 IU/L) was required to minimise the likelihood of an analytical error; 35 U/L was defined as it is an average between the lower bounds of adult normal values and less exclusive than previous studies have set the limit at 30 UI/L. Patients who had previously denied or restricted access to their record for research purposes and aged less than 18 years were excluded from the study. Once authorisations were given, paper and electronic medical records were used to search patient history, symptoms, laboratory results (basic tests, calcium phosphate metabolism and specialised blood tests (bone ALP)), bone densitometry, X-rays, CT- scan and MRI results. If done, the genetic test was notified. Bone demineralisation was defined as a T-score <−2.5 SD. Chondrocalcinosis and hydroxyapatite deposition disease were diagnosed based on the aspect of the calcifications visualised on X-rays. Only non-traumatic fractures were considered in the analyses. Scoliosis was considered as present if described in the radiologist reports of spine imaging. Aetiologies for low ALP were defined as follows: Corticosteroids were considered as a possible cause of hypophosphatasaemia if patients received at least a very high dose of corticosteroids (>100 mg per day) using the standardised nomenclature for glucocorticoid dosages by Buttgereit et al . Severe anaemia was defined as haemoglobin (Hb) <60 g/L. Pernicious anaemia was considered if patients were currently not substituted. Hypothyroidism was considered if patients were not substituted and thyroid-stimulating hormone (TSH) was higher than 4 mUI/L. Hepatic insufficiency was considered if prothrombin time was lower than 50%. Hypervitaminosis D was considered if 25-OH vitamin D was higher than 150 ng/L. Hypomagnesaemia was considered if magnesium level was lower than 0.7 mmol/L. Vitamin C insufficiency was considered for levels lower than 2.5 mg/L. Zinc insufficiency was considered for levels lower than 9 µmol/L. Cushing disease, coeliac disease and Wilson disease were considered only if they were not currently being treated or at equilibrium. Intensive care stay was considered if it was an actual stay or less than 1 month before. Ongoing oncohaematological disease, bisphosphonate treatment, denosumab treatment, septicaemia, inflammatory disease flare and intravenous immunoglobulins were considered as potential causes of low ALP. To determine the number of patients for whom low ALP ≤35 U/L was recognised and noted in their records, the discharge summary, the diagnosis written in the letter and/or the ICD-10 (International Classification of Diseases 10th Revision) code were used. Patients were defined as possible HPP if they exhibited at least three symptoms evocative of HPP in addition to persistent low ALP (arthralgia, fractures, stress fractures, low bone density, dental abnormalities, chondrocalcinosis, scoliosis, high B 6 levels or high urinary phosphoethanolamine). Both of the instruments measure ALP activity by a kinetic rate method in which a colourless organic phosphate ester substrate (nitrophenylphosphate) is hydrolysed by ALP to the yellow-coloured product p-nitrophenol and phosphate at pH of 10.3, thereby explaining the term ‘alkaline’. Changes in absorbance at 410 nm are directly proportional to the enzymatic activity of ALP. A requirement of two low ALP values (≤35 U/L) was set so as to minimise the likelihood of low ALP results due to analytic error. For each selected patient, all previous ALP values against time were visually examined to determine the temporal pattern of the qualifying serum ALP values and to separate two groups of patients. When the temporal pattern of ALP values indicated a precipitous fall from usually normal values, the patient was considered to have acute hypophosphatasaemia. Diagnostic conditions and circumstances associated with acute hypophosphatasaemia were analysed. Laboratory used Glims, JMP or DXLab software. When the temporal pattern of ALP values indicated a persistently low ALP or only 2 values, both of them under 35 U/L, the patient was considered to have persistent hypophosphatasaemia. More precise analysis was carried out to identify detailed patient history, symptoms, laboratory results (basic tests including calcium phosphate metabolism exploration and specialised blood tests such as bone ALP), bone densitometry, X-rays, CT scan and MRI results and genetic test when they had been made. To determine the number of patients in whom persistently low ALP ≤35 U/L was recognised, the discharge summary, the written diagnosis and/or the ICD-10 code (E833) were used. Qualitative data were expressed as percentages and quantitative data as means±SD. Analysis was conducted using the Student’s t-test (or Wilcoxon, as appropriate) for quantitative data and χ² (or Fisher’s exact test, as appropriate) for qualitative data. A p value of 0.05 was considered as significant. Statistical analysis was performed by using SAS software, V.9.1 (SAS Institute) and GraphPad Prism (GraphPad Software, California). Patients or members of the public were not involved in the design, or conduct, or reporting, or dissemination plans of the research. Population characteristics Between 1 July 2007 and 1 July 2017, 144 242 ALP measurements were performed; 56 321 hospitalised patients had at least 2 serum ALP measurements. Inclusion period differed according to the centres, with mean inclusion time of 8.42 years (±2.478). A total of 664 patients hospitalised in the rheumatology and internal medicine departments of the University Hospitals of Poitiers, Nantes, Rennes, Brest, Angers and Tours had at least two ALP values below or equal to 35 IU/L ( and ). There was a difference in the sex ratio with 57.8% of female patients (208/360) in internal medicine departments vs 66.8% in rheumatology departments (203/304) (p=0.017). Prevalence of all-cause hypophosphatasaemia was 1.18%. Among the patients, 182 (27.4 %) had persistently low serum ALP levels, representing a general prevalence of 0.32% for persistent hypophosphatasaemia (182/56321), while 482 patients (72,6%) had fluctuating serum ALP values, at least two of which were below or equal 35 IU/L, which representing prevalence of 0.86%. All in all, 38.1% of patients were male. In only 24 cases (3.61%) was hypophosphatasaemia reported in the patient’s records. Reasons for hospitalisations were various. In rheumatology departments, the top 10 reasons for hospitalisations were lumbosacral radiculopathy, haemopathy, osteoporosis, arthritis, fractures, rheumatoid arthritis, polyarthralgia, low back pain, spondyloarthritis and suspicion of rheumatic disease. In internal medicine departments, the top 10 reasons for hospitalisations were infectious disease, chronic myeloid leukaemia or lymphoma or multiple myeloma, polyarthralgia, severe anaemia/haemorrhage, autoimmune cytopenia, vasculitis, inflammatory myositis, intravenous immunoglobulin infusions, systemic lupus erythematosus, undernutrition or severe anorexia or hydro electrolytic disorders. Initial comparisons of characteristics of patients with transient versus persistent hypophosphatasaemia Clinical characteristics of the patients with transient and persistent hypophosphatasaemia were compared . Patients with persistent hypophosphatasaemia were younger (53.36 vs 62.93 years/old), were less heavy (64.01 vs 68.82 kg) and were more frequently treated in the rheumatology department (74.2% vs 35.1%). Their mean ALP values were significantly lower than in the transient group (28.0 vs 30.1 UI/L respectively). In terms of recognition, persistent hypophosphatasaemia was more frequently identified than transient hypophosphatasaemia (12.6% vs 0.2%). Among the 182 patients with persistent hypophosphatasaemia, 70 patients (38.4%) had no joint imaging and 49 patients (26.9%) had no spinal imaging. Only 12 patients in the transient hypophosphatasaemia group had peripheral joint X-rays and nine had spinal X-rays . Patients with persistent hypophosphatasaemia experienced pain more frequently (90.1% vs 22.8%). Stress fractures were present only in patients with persistent hypophosphatasaemia. As regards medical history, chondrocalcinosis, hydroxyapatite deposition disease, dental abnormalities, early tooth loss, childhood rickets, family HPP and convulsion were found only in patients with persistent hypophosphatasaemia. A total of 48 patients with persistent hypophosphatasaemia had a bone mineral density (BMD) measurement. However, values were not always available and sometimes only the conclusion appeared. Osteopenia was diagnosed in 20 patients, and osteoporosis in 15, while 13 patients presented with normal BMD. Details are found in . Only four patients in the transient hypophosphatasaemia group had a BMD measurement, with three normal BMD and one osteopenia. 10.1136/rmdopen-2024-004316.supp1 Supplementary data 10.1136/rmdopen-2024-004316.supp2 Supplementary data Further comparisons of aetiologies in patients with transient versus persistent hypophosphatasaemia Potential aetiologies of hypophosphatasaemia were compared . Considering possible causes of hypophosphatasaemia, severe anaemia, intensive care unit stay, active oncohaematological disease, ongoing bisphosphonate treatment, sepsis, inflammatory disease flare and intravenous immunoglobulin treatment were more frequently found in transient hypophosphatasaemia, while corticosteroid intake was more frequent in persistent hypophosphatasaemia. In those patients, HPP was possible in 69 patients, in 37 of whom there was no identified cause. Documentation of low ALP values in patients with persistent hypophosphatasaemia with bisphosphonate treatment Since bisphosphonate treatment is contraindicated in HPP, we analysed whether patients with low ALP measurements were tested before bisphosphonate initiation. Among the 24 patients treated with bisphosphonates, 19 (79.2%) had never undergone ALP measurement before treatment, while in 5 patients (20.8%), this treatment had been initiated despite an abnormal decrease of ALP. Details for those patients are in . Comparisons of clinical and radiological features of patients with persistent hypophosphatasaemia with and without identified cause Out of the 182 ‘persistent’ patients, 84 cases had an identified cause and 98 did not . There were no differences in ALP measurements between groups. Patients with unidentified cause of hypophosphatasaemia were more likely to have mechanical pain (70.5% vs 44.7%), diffuse pain (26.9% vs 15.3%) and knee chondrocalcinosis history (66.7% vs 11.1%), while they less frequently had pain in the limbs (28.2% vs 47.1%), fracture history (16.7% vs 29.9%), mixed pattern pain (10.3% vs 28.2%), low BMD (10.7% vs 37.1%) and radiographic vertebral fractures (10.7% vs 31.2%). HPP among patients with persistent hypophosphatasaemia Among all patients with persistent hypophosphatasaemia, 69 presented at least three symptoms evocative of HPP in addition to persistent low ALP and were classified as possible HPP . Among them, 18 underwent genetic analysis in search of ALPL gene mutation, and 11 patients presented with genetically proven HPP (61.1%). The diagnosis of genetic HPP was thereby confirmed in at least 1.7% of our total population (11/664). Among those 11 patients, 3 had another potential cause of low ALP (2 had taken corticosteroids, and 1 had a vitamin C deficiency). Selection of patients with persistently decreased ALP rendered genetic analysis more cost-effective, with a positive diagnosis ranging from at least 1.7% (11/664) to at least 6% (11/182), and even higher than 15.9% (11/69) if they were classified as possible HPP. Pyridoxal phosphate (PLP) measurements had been performed in only 6 patients out of the 664 patients included (mean±SD: 58.33±18.26 nmol/L (normal range: 30–100 nmol/L)). All of them had persistent low ALP: five were genetically tested, among whom three were positive Between 1 July 2007 and 1 July 2017, 144 242 ALP measurements were performed; 56 321 hospitalised patients had at least 2 serum ALP measurements. Inclusion period differed according to the centres, with mean inclusion time of 8.42 years (±2.478). A total of 664 patients hospitalised in the rheumatology and internal medicine departments of the University Hospitals of Poitiers, Nantes, Rennes, Brest, Angers and Tours had at least two ALP values below or equal to 35 IU/L ( and ). There was a difference in the sex ratio with 57.8% of female patients (208/360) in internal medicine departments vs 66.8% in rheumatology departments (203/304) (p=0.017). Prevalence of all-cause hypophosphatasaemia was 1.18%. Among the patients, 182 (27.4 %) had persistently low serum ALP levels, representing a general prevalence of 0.32% for persistent hypophosphatasaemia (182/56321), while 482 patients (72,6%) had fluctuating serum ALP values, at least two of which were below or equal 35 IU/L, which representing prevalence of 0.86%. All in all, 38.1% of patients were male. In only 24 cases (3.61%) was hypophosphatasaemia reported in the patient’s records. Reasons for hospitalisations were various. In rheumatology departments, the top 10 reasons for hospitalisations were lumbosacral radiculopathy, haemopathy, osteoporosis, arthritis, fractures, rheumatoid arthritis, polyarthralgia, low back pain, spondyloarthritis and suspicion of rheumatic disease. In internal medicine departments, the top 10 reasons for hospitalisations were infectious disease, chronic myeloid leukaemia or lymphoma or multiple myeloma, polyarthralgia, severe anaemia/haemorrhage, autoimmune cytopenia, vasculitis, inflammatory myositis, intravenous immunoglobulin infusions, systemic lupus erythematosus, undernutrition or severe anorexia or hydro electrolytic disorders. Clinical characteristics of the patients with transient and persistent hypophosphatasaemia were compared . Patients with persistent hypophosphatasaemia were younger (53.36 vs 62.93 years/old), were less heavy (64.01 vs 68.82 kg) and were more frequently treated in the rheumatology department (74.2% vs 35.1%). Their mean ALP values were significantly lower than in the transient group (28.0 vs 30.1 UI/L respectively). In terms of recognition, persistent hypophosphatasaemia was more frequently identified than transient hypophosphatasaemia (12.6% vs 0.2%). Among the 182 patients with persistent hypophosphatasaemia, 70 patients (38.4%) had no joint imaging and 49 patients (26.9%) had no spinal imaging. Only 12 patients in the transient hypophosphatasaemia group had peripheral joint X-rays and nine had spinal X-rays . Patients with persistent hypophosphatasaemia experienced pain more frequently (90.1% vs 22.8%). Stress fractures were present only in patients with persistent hypophosphatasaemia. As regards medical history, chondrocalcinosis, hydroxyapatite deposition disease, dental abnormalities, early tooth loss, childhood rickets, family HPP and convulsion were found only in patients with persistent hypophosphatasaemia. A total of 48 patients with persistent hypophosphatasaemia had a bone mineral density (BMD) measurement. However, values were not always available and sometimes only the conclusion appeared. Osteopenia was diagnosed in 20 patients, and osteoporosis in 15, while 13 patients presented with normal BMD. Details are found in . Only four patients in the transient hypophosphatasaemia group had a BMD measurement, with three normal BMD and one osteopenia. 10.1136/rmdopen-2024-004316.supp1 Supplementary data 10.1136/rmdopen-2024-004316.supp2 Supplementary data Potential aetiologies of hypophosphatasaemia were compared . Considering possible causes of hypophosphatasaemia, severe anaemia, intensive care unit stay, active oncohaematological disease, ongoing bisphosphonate treatment, sepsis, inflammatory disease flare and intravenous immunoglobulin treatment were more frequently found in transient hypophosphatasaemia, while corticosteroid intake was more frequent in persistent hypophosphatasaemia. In those patients, HPP was possible in 69 patients, in 37 of whom there was no identified cause. Since bisphosphonate treatment is contraindicated in HPP, we analysed whether patients with low ALP measurements were tested before bisphosphonate initiation. Among the 24 patients treated with bisphosphonates, 19 (79.2%) had never undergone ALP measurement before treatment, while in 5 patients (20.8%), this treatment had been initiated despite an abnormal decrease of ALP. Details for those patients are in . Out of the 182 ‘persistent’ patients, 84 cases had an identified cause and 98 did not . There were no differences in ALP measurements between groups. Patients with unidentified cause of hypophosphatasaemia were more likely to have mechanical pain (70.5% vs 44.7%), diffuse pain (26.9% vs 15.3%) and knee chondrocalcinosis history (66.7% vs 11.1%), while they less frequently had pain in the limbs (28.2% vs 47.1%), fracture history (16.7% vs 29.9%), mixed pattern pain (10.3% vs 28.2%), low BMD (10.7% vs 37.1%) and radiographic vertebral fractures (10.7% vs 31.2%). Among all patients with persistent hypophosphatasaemia, 69 presented at least three symptoms evocative of HPP in addition to persistent low ALP and were classified as possible HPP . Among them, 18 underwent genetic analysis in search of ALPL gene mutation, and 11 patients presented with genetically proven HPP (61.1%). The diagnosis of genetic HPP was thereby confirmed in at least 1.7% of our total population (11/664). Among those 11 patients, 3 had another potential cause of low ALP (2 had taken corticosteroids, and 1 had a vitamin C deficiency). Selection of patients with persistently decreased ALP rendered genetic analysis more cost-effective, with a positive diagnosis ranging from at least 1.7% (11/664) to at least 6% (11/182), and even higher than 15.9% (11/69) if they were classified as possible HPP. Pyridoxal phosphate (PLP) measurements had been performed in only 6 patients out of the 664 patients included (mean±SD: 58.33±18.26 nmol/L (normal range: 30–100 nmol/L)). All of them had persistent low ALP: five were genetically tested, among whom three were positive In our study, the prevalence of all-cause hypophosphatasaemia among patients hospitalised in the internal medicine and rheumatology departments was 1.18% while that of persistent hypophosphatasaemia was 0.32%. This proportion was higher than in the study by Maman et al in which 0.13% of hospitalised patients (every department except the emergency department) had persistently low values with a less stringent threshold of 40 UI/L, as in the study of Hepp et al , in which prevalence of 0.20% was found in adults admitted to an endocrinological outpatient clinic in Denmark, or in the study by García-Fontana et al with prevalence of 0.12% in a Spanish university hospital. A German study retrospectively analysing 6 918 126 subjects with a measurement of ALP between 2011 and 2016 in a single laboratory identified prevalence of ALP values below 30 of 8.46% and 9.47% between 30 and 40 UI/L, respectively, thereby underscoring the need to focus on persistent low ALP levels since transient hypophosphatasaemia is quite common. McKiernan et al identified 1.1% of patients with at least two values under 40 UI/L among consultants in a multidisciplinary centre, and 0.06% of ALP level persistently below 30 UI/L. This is concordant with a German study, which found 1.31% of patients treated in rheumatology at the University Hospital of Bonn from 2017 to 2019 showed persistently low serum ALP levels (<35 UI/L). As regards the proportion of patients with persistent hypophosphatasaemia, it was 33.3% in the study by McKiernan et al , and 39% in a study by Vieira et al , which is concordant with our result of 27.6%. Similarly, Feurstein et al found 5.5% of patients with at least one low ALP value under 40 UI/L, with only 13.9% of patients presenting persistent low ALP levels and musculoskeletal symptoms; they represented 0.8% of the whole population from a rheumatology outpatient clinic in Vienna specialised in rheumatology and rare bone diseases. In terms of notification, reporting of low ALP values was found in 3.61% in our population, which is close to the 3% noted by Maman et al . Low ALP is clearly not sufficiently recognised, even if rheumatologists seem to better identify this abnormality with a reported 6.91% vs 0.83% in internal medicine. As a result, adult HPP is highly underdiagnosed. A few years ago, some laboratories only indicated the ‘high’ cut-off and, in the absence of personal knowledge of the lower normal cut-off, the ALP drop was not always noticed, and therefore, easily overlooked. In our study, many patients had not been explored, and the final report never mentioned low ALP. Indeed, it ‘normal liver test’ was often noted without details, even though the ALP levels were lower than 35 IU/L. That is why hypophosphatasaemia was not coded and did not result in further explorations. The difficulty of diagnosing HPP led several teams to propose algorithms to enhance the rate of diagnosis. The first strategy is based on the adjunction of PLP measurement to ALP so as to better stratify the likelihood of HPP diagnosis with high PLP and low ALP as features of HPP. Another team added BMD measurement by Dual-energy X-ray absorptiometry to generate a strategy of rationalised mutational analysis in resource-limiting conditions. While these approaches are interesting, PLP is not performed in daily practice, thereby limiting its usefulness if low ALP has not been previously identified. In our study, PLP measurements had been performed in only 6 out of the 664 patients included. All of them had persistent low ALP; five were genetically tested, among whom three were positive. This lack of data is not surprising since low ALP is poorly recognised in daily practice. Therefore, physicians would not dose PLP since they did not take low ALP into account. Moreover, the algorithm mentioning PLP measurement in order to better screen patients was published after our inclusion period, which may be another, though less important, explanation. Another approach is to focus on populations with highly suggestive features of HPP. Tsiantouli et al analysed ALP values in a population of 72 patients with atypical femur fractures (AFF) with at least 1 ALP value available. There was no difference in the median value of ALP compared with the control group with hip fracture, and no difference in the titre of ALP if they were treated with antiresorptive agent. Moreover, none of the patients with AFF without antiresorptive drugs in this single-centre study presented with low ALP levels. Similarly, Marini et al performed ALPL genotyping in patients with AFF or other biochemical or clinical signs of adult HPP. This led them to identify three rare variants of ALPL (2.8%) in this population. Monozygotic ALPL common variants were found in 11.3% of the patients, with a higher proportion of 22% of patients with normal ALP values, 30.8% of patients with AFF, 16.7% of patients with normal ALP and high PLP levels and also, unfortunately, in 13.5% of non-HPP controls. Those results should draw the attention of clinicians to the need to carefully consider the possibility that some variants have no detrimental effect on the ALP protein and that different kinds of disease severity or carrying a nonpathogenic mutation can be encountered. Since metatarsal fractures are suggestive of HPP, Koehler et al focused on this population and found 0.12% prevalence of pathogenic ALPL variants in a population of 1611 metatarsal fractures, a proportion that rose to 15% when low ALP measurement was associated. In our study, the same approach using clinical, biological and imaging features identified 69 patients with at least three evocative symptoms of HPP in addition to persistently low ALP values (possible HPP) among which 11 were found to have genetically proven HPP, representing a diagnosis rate of at least 15.9%. This value is probably underestimated since only 18 patients benefited from genetic testing, corresponding to a diagnosis rate of 61.1% of the tested patients. The combination of at least three signs in addition to persistent hypophosphatasaemia should, therefore, be tested in a larger population to evaluate its cost-effectiveness. As expected, possible causes of hypophosphatasaemia were more frequently found in transient cases. Interestingly, corticosteroids were more frequently found in persistent hypophosphatasaemia. Since patients with persistent hypophosphatasaemia more frequently had crystal arthropathy history as well as pain, we may hypothesise that this difference is the result of its use to treat arthritis flares or pain as well as long-term treatments in systemic inflammatory disease. Indeed, patients with chronic low dose corticosteroid use were frequently treated with bisphosphonates in order to prevent corticosteroid-induced osteoporosis. Pain in itself is also an important point to consider insofar as more than 90% of patients with persistent hypophosphatasaemia in our study reported pain. Pain also represents the greatest burden in HPP patients, as shown in the global HPP registry. Moreover, there is multiple evidence in the literature that TNSALP exerts a role in the biosynthesis of adenosine, a key molecule with antinociceptive effect with TNSALP, prostatic acid phosphatase and ecto-5’-nucleotidase playing crucial roles in determining the overall sensitivity of the nociceptive circuits, as reviewed extensively by Street and Sowa. The relationship between tissue-nonspecific ALP and inflammation is an increasing source of interest. A recent review article by Graser et al affirms that TNSALP deficiency contributes to inflammatory reactions. TNSALP is implicated in the balance between proinflammatory ATP effects and anti-inflammatory effects of adenosine. Moreover, TNSALP’s ectophosphatase activity is involved in the modulation of TLR ligands like LPS and double-stranded RNA mimic poly-inosine:cytosine. TNSALP is also a T-cell activity modulator. In synthesis, TNSALP is now known to exert an anti-inflammatory effect. Furthermore, ALP levels are higher in case of systemic inflammation. In our study, the two main reasons in which inflammatory disease flare were associated with low ALP were corticosteroids, intravenous immunoglobulin treatments or bisphosphonate therapy for prevention of corticosteroid-induced osteoporosis. In Internal Medicine departments, patients were often hospitalised for sepsis or active oncohaematological disease, which can also explain low ALP. Concerning the patients with persistent hypophosphatasaemia, patients under bisphosphonate treatment were analysed separately to identify differentiating features. The observed differences all seem to be related to osteoporosis with frequent history of fractures, bone deformities, bone demineralisation and vertebral fractures. In terms of bone frailty, vertebral fractures were numerically less frequent in patients with persistent hypophosphatasaemia without identified cause. In the study by Hepp et al , none of the HPP patients had vertebral fractures. The study by Genest et al found a significant correlation between low ALP levels and high spine BMD in a cohort of HPP patients. In the literature review by Sadhukhan et al , vertebral fractures were not observed in HPP patients and high lumbar spine BMD was more likely. This study has some limitations. First of all, this study has a retrospective design which induced differences in the number of observations of the different parameters, and therefore, may wane some of our conclusions. There was a bias regarding the variability of the number of patients included per centre, and it did not allow us to be completely exhaustive. Indeed, 10 years of inclusion was not possible in all centres. The listing established through the cooperation with the laboratory technicians of each centre was not always complete and the extraction of data by the resident at each site was time-consuming. This difference can come from the diversity of software laboratories, which did not always lead to online results for a number of years. This required a longer duration of analysis with a risk of error and limitation of the data to a more restricted period. In addition, some centres underwent software changes during the inclusion period, generating difficulty in returning to previous data. Moreover, the time interval between the two measurements was variable, depending on each patient. The need for two measurements to limit the risk of analytical error may also be at the origin of a bias since body weight could be different at each time point. Since this condition is poorly recognised, genetic testing was performed in only a few cases in which physicians suspected HPP. In this study, long-term low-dose glucocorticoid treatment was not taken into account as a cause of low ALP levels which is another limitation. However, literature is scarce and conflicting about this point which led us not to consider it as a possible cause of low ALP levels. Indeed, in a study by LoCascio et al about 23 patients treated with 10–25 mg a day of glucocorticoid for various immune diseases, no significant decline of ALP levels were found after 1–2 months, 5–7 months or 12 months glucocorticoid treatment. Moreover, in 13 patients treated for chronic glomerulonephritis with a mean dose of 43.8 mg a day of GC progressively tapered, Sasaki et al demonstrated that ALP levels decreased significantly at 1, 3 and 6 months endpoints compared with baseline and bone-specific ALP levels decreased significantly only at 3 and 6 months of follow-up but each measurement remained in the normal range. Pearce et al showed that GC doses of 10 mg and less for polymyalgia rheumatica resulted in higher bone specific ALP levels during a 27-month follow-up. Finally, Korczowska et al showed that ALP levels increased significantly at 12 months after glucocorticoid initiation in patients with rheumatoid arthritis, while there was no difference in patients already treated before the beginning of the follow-up. Although this study could not be exhaustive (missing data, impossibility to carry out genetic research or to perform PLP measurements in all the patients suspected of HPP in a retrospective study), it has the advantage of inventorying our practice in view of improvement, and the multicentre character over 10 years reinforces our conclusions. The results suggest that moderate forms of genetic HPP in adults are certainly more frequent than previously thought and highlight the need for special attention to the value of ALP. The situation is further complicated by the hypothesis that some variants could lead to low density osteopathy, without HPP-disease criteria. Indeed, the presence of heterozygosity in some patients with suggestive symptoms suggests that other mechanisms are involved in the phenotypic expression of adult HPP. For very mild adult forms and exclusive dental forms, mutations may be heterozygous. In their proposal of genetic-based nosology of HPP, Mornet et al described a mild HPP form with adult onset of unspecific symptoms caused by an autosomal dominant haploinsufficiency with prevalence of 1/508. However, the existence of such an entity is still controversial. As a conclusion, hypophosphatasaemia was recognised only in 3.61% of the patients presenting this biological abnormality and hospitalised in rheumatology and internal medicine departments. At least 15.9% of patients with three or more evocative symptoms of HPP in addition to persistent hypophosphatasaemia had HPP. This multicentre retrospective study shows that adult HPP remains underdiagnosed. The prevalence of moderate forms of adult HPP appears to be higher than previously thought and highlights the need for according special attention to ALP values. |
Use of Intravenous Albumin | b0a80794-fdf5-4808-986f-85541ffbfc43 | 11317816 | Internal Medicine[mh] | Intravenous albumin is a human-derived blood product manufactured from donated human plasma. It is used broadly in hospitalized patients, as well as in outpatients with complications of cirrhosis. Intravenous albumin has been studied in numerous, large, well-designed, randomized controlled clinical trials in multiple patient populations; the data show few applications of albumin that improve patient outcomes. Albumin is more expensive to manufacture and to provide to patients, when compared with crystalloids. The International Collaboration for Transfusion Medicine Guidelines undertook this guideline development process to provide clinicians with actionable recommendations for appropriate use of intravenous albumin. 1. In critically ill adult patients (excluding patients with thermal injuries and ARDS), intravenous albumin is not suggested for first-line volume replacement or to increase serum albumin levels (Conditional Recommendation, Moderate Certainty of Evidence of Effect). 2. In critically ill adult patients with thermal injuries or ARDS, intravenous albumin is not suggested for volume replacement or to increase serum albumin level (Conditional Recommendation, Very Low Certainty of Evidence of Effect). 3. In critically ill adult patients, intravenous albumin in conjunction with diuretics is not suggested for removal of extravascular fluid (Conditional Recommendation, Very Low Certainty of Evidence of Effect). 4. In pediatric patients with infection and hypoperfusion, intravenous albumin is not recommended to reduce mortality (Strong Recommendation, Low Certainty of Evidence of Effect). 5. In preterm neonates (≤ 36 weeks) with low serum albumin levels and respiratory distress, intravenous albumin is not suggested to improve respiratory function (Conditional Recommendation, Very Low Certainty of Evidence of Effect). 6. In preterm neonates (≤ 32 weeks or ≤ 1,500 g) with or without hypoperfusion, intravenous albumin is not suggested for volume replacement (Conditional Recommendation, Very Low Certainty of Evidence of Effect). 7. In patients undergoing kidney replacement therapy, intravenous albumin is not suggested for prevention or treatment of intradialytic hypotension or for improving ultrafiltration (Conditional Recommendation, Very Low Certainty of Evidence of Effect). 8. In adult patients undergoing cardiovascular surgery, intravenous albumin is not suggested for priming the cardiovascular bypass circuit or volume replacement (Conditional Recommendation, Moderate Certainty of Evidence of Effect). 9. In pediatric patients undergoing cardiovascular surgery, intravenous albumin is not suggested for priming the cardiovascular bypass circuit or volume replacement (Conditional Recommendation, Very Low Certainty of Evidence of Effect). 10. In patients with cirrhosis and ascites undergoing large - volume paracentesis (> 5 L ), intravenous albumin is suggested to prevent paracentesis-induced circulatory dysfunction (Conditional Recommendation, Very Low Certainty of Evidence of Effect). 11. In patients with cirrhosis and spontaneous bacterial peritonitis, intravenous albumin is suggested to reduce mortality (Conditional Recommendation, Low Certainty of Evidence of Effect). 12. In patients with cirrhosis and extraperitoneal infections, intravenous albumin is not suggested to reduce mortality or kidney failure (Conditional Recommendation, Low Certainty of Evidence of Effect). 13. In hospitalized patients with decompensated cirrhosis with hypoalbuminemia (< 30 g/L), repeated intravenous albumin to increase albumin levels to > 30 g/L is not suggested to reduce infection, kidney dysfunction , or death (Conditional Recommendation, Low Certainty of Evidence of Effect). 14. In outpatients with cirrhosis and uncomplicated ascites despite diuretic therapy, intravenous albumin is not routinely suggested to reduce complications associated with cirrhosis (Conditional Recommendation, Low Certainty of Evidence of Effect). Albumin is administered in a wide spectrum of clinical scenarios including complications of cirrhosis, intradialytic hypotension, volume resuscitation, and priming of cardiopulmonary bypass circuit. Iso-oncotic albumin often is used to maintain intravascular volume in patients with hypovolemia, assuming that crystalloid resuscitation will be ineffective given its shorter intravascular half-life. Hyperoncotic albumin is used to correct low serum albumin levels or to mobilize extravascular fluid. Hypoalbuminemia is common in acute and chronic illness. Hospitalized patients with hypoalbuminemia have been described as having greater morbidity compared with patients with preserved albumin levels, promoting the use of IV albumin. , In the postoperative period, serum albumin levels decreases precipitously by 10 to 15 g/L ; hypoalbuminemia is thought to be the result of suppressed synthesis by inflammatory cytokines and transcapillary loss. In addition to its use in patients with hypoalbuminemia, edema, or both, albumin also is used for the prevention and treatment of hypovolemia, particular after administration of large volumes of IV crystalloid solutions. Practice audits describing the use of albumin show highly variable practice among regions. , Albumin is manufactured from large volumes of plasma and is expensive (approximately $130/25 g United States dollars; warehouse acquisition cost of albumin), with the acquisition cost likely a fraction of the total health care expenditure. Albumin also can be associated with adverse consequences, including fluid overload, , hypotension, hemodilution requiring RBC transfusion, anaphylaxis, and peripheral gangrene from dilution of natural anticoagulants. Because potential benefits and risks are associated with its use, a multidisciplinary, international guideline panel was convened to develop evidence-based recommendations for the use of albumin in patient populations where it is prescribed commonly. These guidelines are designed to assist clinicians in their decisions on the use of albumin for its most common uses. Guidelines Focus These recommendations apply to patients receiving albumin in critical care settings with hypovolemia, sepsis, hypoalbuminemia, thermal injuries, and ARDS; cirrhosis; intradialytic hypotension; and cardiovascular surgery. These settings were included based on common uses of albumin, the systematic review of the published randomized controlled trials (RCTs), and with input from the panel. We included studies that compared the use of albumin with that of other resuscitation fluids, other pharmaceutical treatments, or standard of care. Target Population These guidelines provide actionable recommendations for the most common indications for the use of albumin. The use of albumin for therapeutic apheresis was excluded because recent guidelines were published. Guidelines Development Process Panel Composition This guideline development process was funded by the Ontario Regional Blood Coordinating Network (Ontario, Canada) and the International Collaboration for Transfusion Medicine Guidelines (ICTMG; funded by Canadian Blood Services). Neither entity had any input on recommendations or guidelines content. An international panel of neonatal, pediatric, and adult specialists with expertise in the use of albumin developed the recommendations. This panel included 20 members with expertise in intensive care, hepatology, gastroenterology, nephrology, hematology, pathology, neonatology, transfusion medicine, cardiothoracic anesthesiology, internal medicine, and methodology and a patient representative. A framework and related clinical questions were developed according to the United States Preventative Services Task Force Criteria. Disclosures were ascertained yearly from all members. Systematic Review of the Evidence A systematic search for articles published between inception and November 23, 2022, in MEDLINE, EMBASE, Cochrane, the National Health Service Economic Evaluation Database Cochrane Database of Systematic Reviews, Cochrane Central Register of Controlled Trials, Ovid MEDLINE, Ovid MEDLINE epub ahead of print and in-process, and other nonindexed citations was completed with the assistance of an information specialist. The Preferred Reporting Items for Systematic Reviews and Meta-Analyses flow diagram for this review is presented in . The guideline development group conducted two systematic reviews: one for patients with critical illness or cirrhosis or requiring kidney replacement therapy (International Prospective Register of Systematic Reviews Identifier: CRD42019145152) and the other for patients undergoing cardiovascular surgery (International Prospective Register of Systematic Reviews Identifier: CRD42020171876). Manually searched references of primary articles, relevant reviews, and additional articles identified by panel members were included. The search strategy is detailed in . Study inclusion criteria were: (1) original peer-reviewed published RCTs comparing albumin with an alternative strategy, (2) systematic reviews and meta-analyses reporting on RCTs, or both, (3) including at least one of the following outcomes of interest: mortality, multisystem organ failure, need for kidney replacement therapy or kidney failure, need for vasoactive medications, need for mechanical ventilation, hypotension, hemodynamic metrics, length of stay (hospital and intensive care), quality of life, health care use, and albumin levels; and (4) published in English. InsightScope screened publications for eligibility and extracted characteristics, outcomes, and risk of bias for all indications, with the exception of studies published between November 2018 and November 2022 and the systematic review for cardiovascular surgery. Quality and risk-of-bias assessment were conducted using the established criteria, , presented in detail for all systematic reviews in . Discrepancies were resolved by a third reviewer. With the exception of cardiovascular surgery, comprehensive systematic reviews were available for all other settings that were used to develop recommendations. For cardiovascular surgery, where no systematic review had been performed, a systematic review and meta-analysis was conducted. Evidence tables for all indications are presented in . Grading of the Evidence Recommendations were formulated on the basis of the Grading of Recommendations, Assessment, Development, and Evaluation (GRADE) (GRADEpro GDT; McMaster University). The evidence certainty was graded as high, moderate, low, or very low certainty based on GRADE criteria. The panel ranked clinical outcomes (electronic survey) relevant for the development of recommendations according to GRADE criteria. Outcomes were ranked on a nine-part Likert scale for all relevant clinical outcomes identified by panel members (1-3 = low importance, 4-6 = important but not critical, and 7-9 = critical) . Recommendation strength was evaluated as strong or conditional. A strong recommendation was made according to GRADE if the panel was “confident that the desirable effects outweighed the undesirable effects.” A conditional recommendation was made if the panel concluded that the “desirable effects probably outweigh the undesirable effects,” but the trade-offs were not well defined and the recommendation may not be applicable to all patients. The terms recommend and suggest were used to reflect strong and conditional recommendations, respectively. Virtual conferences and electronic correspondence were used to discuss the clinical questions and to formulate recommendations. Electronic surveys were sent to all members to assess agreement with recommendations. Disagreements were resolved by discussion. If disagreements could not be resolved, a recommendation was accepted if most members (50% or more of the panel) agreed. Members recorded their disclosures, but none were excluded from voting . The final guidance document was disseminated widely for public consultation to numerous medical societies . The reviewers from these societies were sent a survey consisting of open-ended and closed-ended questions to determine agreement with each recommendation and to identify facilitators and barriers to guideline implementation. Comments from reviewers subsequently were discussed by panel members and addressed. The recommendations in this guidance document will be reviewed every 3 years. If a study is published that may impact the recommendations critically before that time, a comment will be added on the ICTMG website. Recommendations are intended for critical care physicians, nephrologists, hepatologists, gastroenterologists, anesthesiologists, cardiovascular surgeons, general internists, hospitalists, hematologists, pathologists, pharmacists, laboratory technologists, and transfusion medicine physicians. The ICTMG website ( https://www.ictmg.org ) will be used to post implementation tools (eg, podcasts, order sets). The guideline process adhered to the 2011 Institute of Medicine (United States) Committee on Standards for Developing Trustworthy Clinical Practice Guidelines. These recommendations apply to patients receiving albumin in critical care settings with hypovolemia, sepsis, hypoalbuminemia, thermal injuries, and ARDS; cirrhosis; intradialytic hypotension; and cardiovascular surgery. These settings were included based on common uses of albumin, the systematic review of the published randomized controlled trials (RCTs), and with input from the panel. We included studies that compared the use of albumin with that of other resuscitation fluids, other pharmaceutical treatments, or standard of care. These guidelines provide actionable recommendations for the most common indications for the use of albumin. The use of albumin for therapeutic apheresis was excluded because recent guidelines were published. Panel Composition This guideline development process was funded by the Ontario Regional Blood Coordinating Network (Ontario, Canada) and the International Collaboration for Transfusion Medicine Guidelines (ICTMG; funded by Canadian Blood Services). Neither entity had any input on recommendations or guidelines content. An international panel of neonatal, pediatric, and adult specialists with expertise in the use of albumin developed the recommendations. This panel included 20 members with expertise in intensive care, hepatology, gastroenterology, nephrology, hematology, pathology, neonatology, transfusion medicine, cardiothoracic anesthesiology, internal medicine, and methodology and a patient representative. A framework and related clinical questions were developed according to the United States Preventative Services Task Force Criteria. Disclosures were ascertained yearly from all members. Systematic Review of the Evidence A systematic search for articles published between inception and November 23, 2022, in MEDLINE, EMBASE, Cochrane, the National Health Service Economic Evaluation Database Cochrane Database of Systematic Reviews, Cochrane Central Register of Controlled Trials, Ovid MEDLINE, Ovid MEDLINE epub ahead of print and in-process, and other nonindexed citations was completed with the assistance of an information specialist. The Preferred Reporting Items for Systematic Reviews and Meta-Analyses flow diagram for this review is presented in . The guideline development group conducted two systematic reviews: one for patients with critical illness or cirrhosis or requiring kidney replacement therapy (International Prospective Register of Systematic Reviews Identifier: CRD42019145152) and the other for patients undergoing cardiovascular surgery (International Prospective Register of Systematic Reviews Identifier: CRD42020171876). Manually searched references of primary articles, relevant reviews, and additional articles identified by panel members were included. The search strategy is detailed in . Study inclusion criteria were: (1) original peer-reviewed published RCTs comparing albumin with an alternative strategy, (2) systematic reviews and meta-analyses reporting on RCTs, or both, (3) including at least one of the following outcomes of interest: mortality, multisystem organ failure, need for kidney replacement therapy or kidney failure, need for vasoactive medications, need for mechanical ventilation, hypotension, hemodynamic metrics, length of stay (hospital and intensive care), quality of life, health care use, and albumin levels; and (4) published in English. InsightScope screened publications for eligibility and extracted characteristics, outcomes, and risk of bias for all indications, with the exception of studies published between November 2018 and November 2022 and the systematic review for cardiovascular surgery. Quality and risk-of-bias assessment were conducted using the established criteria, , presented in detail for all systematic reviews in . Discrepancies were resolved by a third reviewer. With the exception of cardiovascular surgery, comprehensive systematic reviews were available for all other settings that were used to develop recommendations. For cardiovascular surgery, where no systematic review had been performed, a systematic review and meta-analysis was conducted. Evidence tables for all indications are presented in . Grading of the Evidence Recommendations were formulated on the basis of the Grading of Recommendations, Assessment, Development, and Evaluation (GRADE) (GRADEpro GDT; McMaster University). The evidence certainty was graded as high, moderate, low, or very low certainty based on GRADE criteria. The panel ranked clinical outcomes (electronic survey) relevant for the development of recommendations according to GRADE criteria. Outcomes were ranked on a nine-part Likert scale for all relevant clinical outcomes identified by panel members (1-3 = low importance, 4-6 = important but not critical, and 7-9 = critical) . Recommendation strength was evaluated as strong or conditional. A strong recommendation was made according to GRADE if the panel was “confident that the desirable effects outweighed the undesirable effects.” A conditional recommendation was made if the panel concluded that the “desirable effects probably outweigh the undesirable effects,” but the trade-offs were not well defined and the recommendation may not be applicable to all patients. The terms recommend and suggest were used to reflect strong and conditional recommendations, respectively. Virtual conferences and electronic correspondence were used to discuss the clinical questions and to formulate recommendations. Electronic surveys were sent to all members to assess agreement with recommendations. Disagreements were resolved by discussion. If disagreements could not be resolved, a recommendation was accepted if most members (50% or more of the panel) agreed. Members recorded their disclosures, but none were excluded from voting . The final guidance document was disseminated widely for public consultation to numerous medical societies . The reviewers from these societies were sent a survey consisting of open-ended and closed-ended questions to determine agreement with each recommendation and to identify facilitators and barriers to guideline implementation. Comments from reviewers subsequently were discussed by panel members and addressed. The recommendations in this guidance document will be reviewed every 3 years. If a study is published that may impact the recommendations critically before that time, a comment will be added on the ICTMG website. Recommendations are intended for critical care physicians, nephrologists, hepatologists, gastroenterologists, anesthesiologists, cardiovascular surgeons, general internists, hospitalists, hematologists, pathologists, pharmacists, laboratory technologists, and transfusion medicine physicians. The ICTMG website ( https://www.ictmg.org ) will be used to post implementation tools (eg, podcasts, order sets). The guideline process adhered to the 2011 Institute of Medicine (United States) Committee on Standards for Developing Trustworthy Clinical Practice Guidelines. This guideline development process was funded by the Ontario Regional Blood Coordinating Network (Ontario, Canada) and the International Collaboration for Transfusion Medicine Guidelines (ICTMG; funded by Canadian Blood Services). Neither entity had any input on recommendations or guidelines content. An international panel of neonatal, pediatric, and adult specialists with expertise in the use of albumin developed the recommendations. This panel included 20 members with expertise in intensive care, hepatology, gastroenterology, nephrology, hematology, pathology, neonatology, transfusion medicine, cardiothoracic anesthesiology, internal medicine, and methodology and a patient representative. A framework and related clinical questions were developed according to the United States Preventative Services Task Force Criteria. Disclosures were ascertained yearly from all members. A systematic search for articles published between inception and November 23, 2022, in MEDLINE, EMBASE, Cochrane, the National Health Service Economic Evaluation Database Cochrane Database of Systematic Reviews, Cochrane Central Register of Controlled Trials, Ovid MEDLINE, Ovid MEDLINE epub ahead of print and in-process, and other nonindexed citations was completed with the assistance of an information specialist. The Preferred Reporting Items for Systematic Reviews and Meta-Analyses flow diagram for this review is presented in . The guideline development group conducted two systematic reviews: one for patients with critical illness or cirrhosis or requiring kidney replacement therapy (International Prospective Register of Systematic Reviews Identifier: CRD42019145152) and the other for patients undergoing cardiovascular surgery (International Prospective Register of Systematic Reviews Identifier: CRD42020171876). Manually searched references of primary articles, relevant reviews, and additional articles identified by panel members were included. The search strategy is detailed in . Study inclusion criteria were: (1) original peer-reviewed published RCTs comparing albumin with an alternative strategy, (2) systematic reviews and meta-analyses reporting on RCTs, or both, (3) including at least one of the following outcomes of interest: mortality, multisystem organ failure, need for kidney replacement therapy or kidney failure, need for vasoactive medications, need for mechanical ventilation, hypotension, hemodynamic metrics, length of stay (hospital and intensive care), quality of life, health care use, and albumin levels; and (4) published in English. InsightScope screened publications for eligibility and extracted characteristics, outcomes, and risk of bias for all indications, with the exception of studies published between November 2018 and November 2022 and the systematic review for cardiovascular surgery. Quality and risk-of-bias assessment were conducted using the established criteria, , presented in detail for all systematic reviews in . Discrepancies were resolved by a third reviewer. With the exception of cardiovascular surgery, comprehensive systematic reviews were available for all other settings that were used to develop recommendations. For cardiovascular surgery, where no systematic review had been performed, a systematic review and meta-analysis was conducted. Evidence tables for all indications are presented in . Recommendations were formulated on the basis of the Grading of Recommendations, Assessment, Development, and Evaluation (GRADE) (GRADEpro GDT; McMaster University). The evidence certainty was graded as high, moderate, low, or very low certainty based on GRADE criteria. The panel ranked clinical outcomes (electronic survey) relevant for the development of recommendations according to GRADE criteria. Outcomes were ranked on a nine-part Likert scale for all relevant clinical outcomes identified by panel members (1-3 = low importance, 4-6 = important but not critical, and 7-9 = critical) . Recommendation strength was evaluated as strong or conditional. A strong recommendation was made according to GRADE if the panel was “confident that the desirable effects outweighed the undesirable effects.” A conditional recommendation was made if the panel concluded that the “desirable effects probably outweigh the undesirable effects,” but the trade-offs were not well defined and the recommendation may not be applicable to all patients. The terms recommend and suggest were used to reflect strong and conditional recommendations, respectively. Virtual conferences and electronic correspondence were used to discuss the clinical questions and to formulate recommendations. Electronic surveys were sent to all members to assess agreement with recommendations. Disagreements were resolved by discussion. If disagreements could not be resolved, a recommendation was accepted if most members (50% or more of the panel) agreed. Members recorded their disclosures, but none were excluded from voting . The final guidance document was disseminated widely for public consultation to numerous medical societies . The reviewers from these societies were sent a survey consisting of open-ended and closed-ended questions to determine agreement with each recommendation and to identify facilitators and barriers to guideline implementation. Comments from reviewers subsequently were discussed by panel members and addressed. The recommendations in this guidance document will be reviewed every 3 years. If a study is published that may impact the recommendations critically before that time, a comment will be added on the ICTMG website. Recommendations are intended for critical care physicians, nephrologists, hepatologists, gastroenterologists, anesthesiologists, cardiovascular surgeons, general internists, hospitalists, hematologists, pathologists, pharmacists, laboratory technologists, and transfusion medicine physicians. The ICTMG website ( https://www.ictmg.org ) will be used to post implementation tools (eg, podcasts, order sets). The guideline process adhered to the 2011 Institute of Medicine (United States) Committee on Standards for Developing Trustworthy Clinical Practice Guidelines. Recommendations are outlined in . Clinical Setting 1: Critically Ill Adult Patients Recommendations Recommendation 1: In critically ill adult patients (excluding patients with thermal injuries and ARDS), intravenous albumin is not suggested for first-line volume replacement or to increase serum albumin levels (Conditional Recommendation, Moderate Certainty of Evidence of Effect). Recommendation 2: In critically ill adult patients with thermal injuries or ARDS, intravenous albumin is not suggested for volume replacement or to increase serum albumin level (Conditional Recommendation, Very Low Certainty of Evidence of Effect). Recommendation 3: In critically ill adult patients, intravenous albumin in conjunction with diuretics is not suggested for removal of extravascular fluid (Conditional Recommendation, Very Low Certainty of Evidence of Effect). Evidence Summary Sixteen , , , , , , , , , , , , , , , of 19 systematic reviews were retrieved and included. These reports included a broad critical care patient population, including patients with critical illness, sepsis, thermal injuries, and ARDS. Three of the 19 systematic reviews were excluded because they assessed the impact of albumin only on fluid balance, gelatin vs colloids, or all colloids compared with crystalloids (without reporting albumin vs other fluids). A systematic review from 2019 identified 55 RCTs comparing crystalloid with colloids in critical care. Data on mortality were available for 26,329 patients from 46 studies. No mortality benefit was found when crystalloid was compared with albumin (relative risk [RR] 1.02; 95% CI, 0.96-1.10). Crystalloids were less effective than colloids in hemodynamic resuscitation end points (eg, mean arterial pressure) but this did not translate into improvements in patient outcomes. After this systematic review, one RCT was identified that examined 360 patients with sepsis with an underlying diagnosis of cancer (albumin was compared with Ringer’s lactate); no differences in mortality or other outcomes were found. A systematic review from 2018 conducted by the Cochrane collaboration found no difference in mortality in patients in the ICU (20 studies; N = 13,047) when patients managed with crystalloids were compared with those managed with albumin at the end of follow-up (RR, 0.98; 95% CI, 0.92-1.06), at 30 days (RR, 0.99; 95% CI, 0.93-1.06), or at 90 days (RR, 0.98; 95% CI, 0.92-1.04) or who needed kidney replacement therapy (RR, 1.11; 95% CI, 0.96-1.27). The largest randomized trial is the Saline Versus Albumin Fluid Evaluation trial published in 2004, which enrolled 6,997 patients receiving critical care (including a mix of medical and surgical patients) and compared 4% albumin with 0.9% normal saline. No differences were found in outcomes, including 28-day mortality (RR, 0.99; 95% CI, 0.91-1.09). A 2015 systematic review evaluated the administration of albumin in critical care patients with traumatic injury; the review included five trials comparing albumin with crystalloid and found a higher mortality in albumin-treated patients (RR, 1.35; 95% CI, 1.03-1.77). This systematic review was dominated by the Saline Versus Albumin Fluid Evaluation trial (57% of patients). The Saline Versus Albumin Fluid Evaluation trial subgroup analysis found that patients with traumatic brain injury showed a higher mortality rate (RR, 1.62; 95% CI, 1.12-2.34), but those without traumatic brain injury did not (RR, 1.00; 95% CI, 0.56-1.79). Hence, it is uncertain whether albumin may be unsafe only in patients with traumatic brain injury as compared with the wider trauma population. A 2020 systematic review and sequential network analysis of RCTs in the setting of sepsis included 23 randomized trials (N = 14,659); the vast majority of the trials used a physiologic target for volume resuscitation or at the discretion of the clinician, rather than a target albumin level. The review found albumin not to be superior to crystalloids for mortality or acute kidney injury. A 2014 systematic review included 16 randomized trials (N = 4,190) comparing crystalloid or albumin and found no difference in mortality (RR, 0.94; 95% CI, 0.87-1.01). Two network meta-analyses have been performed and reported no mortality benefit from albumin. , The largest randomized trial in sepsis was the Albumin Italian Outcome Sepsis trial, which randomized 1,818 patients with sepsis at 100 sites to 20% albumin (targeting plasma albumin level of ≥ 30 g/L) vs crystalloid. The Albumin Italian Outcome Sepsis trial did not observe improvements in mortality at 28 days (RR, 1.00; 95% CI, 0.87-1.14) or other important outcomes. Three systematic reviews found no impact of albumin in critically ill adults on the need for kidney replacement therapy, including two network meta-analyses , and the 2018 Cochrane review. A systematic review evaluated the impact of albumin on patient outcomes after thermal injuries. The report identified four RCTs and found no impact on the incidence of kidney failure or mortality (RR, 1.41; 95% CI, 0.27-7.38). In a 2022 systematic review evaluating the impact of albumin and diuretics, as compared with diuretics alone, in mechanically ventilated patients (three trials; N = 129), albumin reduced hypotensive episodes, but did not shorten the duration of mechanical ventilation or improve the mortality rate. A 2014 systematic review evaluated the impact of albumin, as compared with crystalloid, in patients with ARDS. Three RCTs (N = 204) were included; no difference in mortality was found (RR, 0.89; 95% CI, 0.62-1.28). Similarly, a 2014 systematic review that included two small RCTs (N = 70) found no difference in ventilator-free days or mortality when albumin with diuretics, as compared with diuretics alone, were used to improve respiratory status in critically ill patients. A 2014 systematic review evaluated the impact of albumin with furosemide, compared with furosemide alone, to facilitate fluid removal in patients with hypoalbuminemia and hypervolemia. The systematic review identified 10 studies (N = 343). Although urine output was higher at 6 h in the patients receiving albumin-furosemide, no difference was found in urine output at 24 h. One RCT of 49 patients with edema receiving critical care was identified subsequent to this systematic review that compared albumin and furosemide with furosemide alone; no difference in urine output at 8 h was found. Rationale for Recommendations A substantial amount of evidence from RCTs in critically ill adult patients across a wide range of patient subgroups provides little supportive evidence for the use of albumin as fluid replacement to reduce mortality, the need for kidney replacement therapy, or other outcomes considered important or critical for decision-making by the panel. Given the wide CIs for the estimates from the systematic reviews, all recommendations were considered conditional because of the residual uncertainty. In systematic reviews evaluating the role of albumin in patients with sepsis, the use of albumin has not been found to be associated with improved outcomes, although a benefit has not been excluded because of the wide CI in the most recent systematic review. The Surviving Sepsis Campaign guidelines published in 2021 recommend albumin in addition to crystalloids when patients require large volumes of crystalloids (Weak Recommendation, Moderate-Quality Evidence). Specific formulations of albumin (4%-5% or 20%-25%), volumes or doses, serum albumin targets, or a combination thereof were not described. The guidelines state, “The lack of proven benefit and higher cost of albumin compared to crystalloid contributed to our strong recommendation for the use of crystalloids as first-line fluid for resuscitation in sepsis and septic shock.” More studies will be needed to evaluate the role and timing of albumin as a rescue fluid in patients with sepsis failing front-line crystalloid resuscitation, particularly given the considerably higher cost of albumin compared with crystalloids, the risks of albumin, and the lack of benefit shown in RCTs. Clinical Setting 2: Critically Ill Pediatric Patients With Severe Infection Recommendation Recommendation 4: In pediatric patients with infection and hypoperfusion, intravenous albumin is not recommended to reduce mortality (Strong Recommendation, Low Certainty of Evidence of Effect). Evidence Summary A single systematic review identified three RCTs that compared albumin with crystalloid in critically ill children. , , All RCTs enrolled children primarily in African countries with either severe malaria or febrile illness with impaired perfusion. The first trial enrolled 61 children with severe malaria and found no difference in mortality when albumin was compared with crystalloid. The second trial enrolled 150 children with severe malaria and found an improvement in the mortality with albumin as compared with crystalloid. A mortality difference was not found in a large, well-designed RCT (Fluid Expansion as Supportive Therapy; N = 3,141) that included children with severe febrile illness with impaired perfusion (60% had malaria). This RCT had three arms comparing saline bolus, 5% albumin bolus, and no bolus. The trial was terminated by the independent data safety monitoring committee at the fifth interim analysis based on data from 2,995 children and after 3,141 of 3,600 planned patients were enrolled because of excess mortality in the patients treated with both the albumin bolus (RR, 1.45; 95% CI, 1.10-1.92) and the saline bolus (RR, 1.44; 95% CI, 1.09-1.90) compared with children who received no bolus. No mortality difference was found when the albumin bolus arm was compared with the crystalloid bolus arm (RR, 1.00; 95% CI, 0.78-1.29) at 48 h. Similar differences in mortality were observed between groups at 28 days, again with an excess mortality in the albumin and saline bolus groups compared with the no bolus group (RR, 1.40 [95% CI, 1.08-1.80] and 1.38 [95% CI, 1.07-1.78]). Children treated with both saline and albumin boluses showed higher rates of respiratory and neurologic dysfunction and of hyperchloremic acidosis and a greater reduction in hemoglobin levels. Rationale for Recommendations The systematic review of the literature for pediatric patients receiving critical care found fewer RCTs as compared with studies of adult patients. Among them, a very large trial of children with febrile illness and hypoperfusion found excess mortality when either an albumin bolus or a crystalloid bolus strategy was compared with a no bolus strategy. Given the extensive, albeit indirect, literature base in adult critical care showing no improvement in mortality or other important outcomes and the above large trial in children suggesting excess mortality with a front-line albumin bolus strategy, pediatric intensivists probably should not use albumin as a first-line treatment outside of a clinical trial for severe infections in critically ill children. Because most children enrolled in these RCTs had malaria, it is uncertain if the results are applicable to all critically ill children with infections or the broader pediatric critical care population. In addition, the increased mortality in the Fluid Expansion as Supportive Therapy trial may be the result of the bolus administration, rather than the type of fluid, with substudies of the Fluid Expansion as Supportive Therapy trial showing that the bolus of either fluid type was associated with higher rates of cardiovascular collapse. Clinical Setting 3: Neonates in Critical Care Recommendation 5: In preterm neonates (≤ 36 weeks) with low serum albumin levels and respiratory distress, intravenous albumin is not suggested to improve respiratory function (Conditional Recommendation, Very Low Certainty of Evidence of Effect). Recommendation 6: In preterm neonates (≤ 32 weeks or ≤ 1,500 g) with or without hypoperfusion, intravenous albumin is not suggested for volume replacement (Conditional Recommendation, Very Low Certainty of Evidence of Effect). Evidence Summary A Cochrane systematic review evaluated the use of albumin in preterm neonates (≤ 36 weeks’ gestation at birth) with hypoalbuminemia (two RCTs enrolling 64 preterm neonates). Only one study reported mortality rates and no difference was found. No other important differences in outcomes were observed. A Cochrane systematic review of RCTs of early volume expansion compared normal saline, plasma, albumin, plasma substitutes, or blood with no treatment or another fluid treatment in preterm neonates (≤ 32 weeks or ≤ 1,500 g). Early volume expansion was defined as > 10 mL/kg of body weight in the first 3 days. The studies included variable indications for the administration of IV fluids. Eight studies were identified, with four studies evaluating albumin with a comparative arm (two vs normal saline, one vs plasma, and one vs no treatment). The two studies (N = 102 and N = 63) comparing 5% albumin with normal saline in hypotensive infants found no difference in mortality (RR, 1.02; 95% CI, 0.50-2.06) or any other patient-important outcomes. The one study (N = 25) comparing 20% albumin with no treatment in normotensive infants also found no difference in mortality (RR, 0.92; 95% CI, 0.23-3.72). Finally, one trial (N = 20) in hypotensive infants compared plasma with albumin and found no difference in duration of ventilation (mortality not reported). Since the publication of these two Cochrane reviews, a single RCT (N = 33) was identified comparing 5% albumin with normal saline (both 10 mL/kg) for term infants with dehydration, metabolic acidosis, and diarrhea and found no differences in outcomes. Rationale for Recommendations Few RCTs have evaluated the impact of albumin compared with other alternative fluids in preterm or term neonates with either hypoalbuminemia or hypovolemia. Very little evidence is available in the literature to guide the use of albumin in term neonates. All trials in the two systematic reviews included small numbers of neonates, preventing any definitive conclusions. Indirect evidence from the adult and pediatric literature, the costs of albumin, and the lack of trials assessing the potential harms of albumin should be considered when including albumin in neonatal fluid protocols. Clinical Setting 4: Patients Undergoing Kidney Replacement Therapy Recommendation Recommendation 7: In patients undergoing kidney replacement therapy, intravenous albumin is not suggested for prevention or treatment of intradialytic hypotension or for improving ultrafiltration (Conditional Recommendation, Very Low Certainty of Evidence of Effect). Evidence Summary A single Cochrane systematic review was identified evaluating the use of albumin, compared with an alternative strategy, for the treatment of intradialytic hypotension. The review identified a single (N = 45) randomized crossover trial of 5% albumin compared with normal saline and did not find a difference in the primary outcome (percentage target ultrafiltration achieved) or other clinical outcomes. Two small crossover trials identified in this review evaluated 20% albumin as compared with gelatin (N = 10) and a three-arm study compared 20% albumin with both saline and hydroxyethyl starch (N = 10). , These RCTs suggested that BP was maintained better with albumin vs other fluid, but found no improvements in other outcomes, including improving ultrafiltration. Finally, a 2021 randomized crossover trial involving 65 hospitalized patients requiring hemodialysis with serum albumin levels of < 30 g/L found that hypotension, lowest intradialytic systolic BP, and ultrafiltration rate were improved with 25% albumin compared with saline. Rationale for Recommendation Intradialytic hypotension and fluid overload are experienced commonly during kidney replacement therapy. , Patients with intradialytic hypotension are at greater risk of morbidity and mortality. Given the costs of albumin, the need for thrice weekly treatment for patients receiving maintenance hemodialysis, and the lack of evidence to support superiority over less costly fluid alternatives, alternative fluids or treatments need to be considered. The annual cost of 25 g of albumin given with thrice-weekly maintenance dialysis is estimated at $20,000 per patient (United States dollars). Midodrine (an oral vasopressor) given alone or in combination with use of a high dialysate calcium concentration and lower dialysate temperature has been explored as a therapeutic option to mitigate intradialytic hypotension. , , In patients prescribed kidney replacement therapy, higher dialysate calcium, lower dialysate temperature, individualized ultrafiltration rates, or a combination of these strategies may mitigate intradialytic hypotension. , , Further studies are needed to understand the pathophysiology of intradialytic hypotension to determine if albumin prevents intradialytic hypotension or improves ultrafiltration, mitigates associated symptoms, or improves patient-important outcomes. Clinical Setting 5: Patients Undergoing Cardiac or Vascular Surgery Recommendations Recommendation 8: In adult patients undergoing cardiovascular surgery, intravenous albumin is not suggested for priming the cardiovascular bypass circuit or volume replacement (Conditional Recommendation, Moderate Certainty of Evidence of Effect). Recommendation 9: In pediatric patients undergoing cardiovascular surgery, intravenous albumin is not suggested for priming the cardiovascular bypass circuit or volume replacement (Conditional Recommendation, Very Low Certainty of Evidence of Effect). Evidence Summary A systematic review and meta-analysis of RCTs in pediatric and adult patients undergoing cardiovascular surgery was performed. We identified 43 randomized studies (N = 3,862), comparing albumin with gelatin, starches, or crystalloid solutions for priming the cardiopulmonary bypass circuit, volume expansion, or both. The majority of the trials were conducted in patients undergoing on-pump cardiac surgery, with the exception of two RCTs conducted in patients undergoing off-pump cardiac surgery. , Albumin infusion did not result in a lower mortality rate when compared with other fluids (risk difference, 0.00; 95% CI, –0.01 to 0.01; N = 2,711). No differences were found for the rates of kidney failure (risk difference, 0.01; 95% CI, –0.01 to 0.03; N = 1,703), blood loss (mean difference [MD], –0.04 L; 95% CI, –0.04 to 0.01 L), ICU length of stay (MD, –0.12 days; 95% CI, –0.31 to 0.06 days; N = 1,371), hospital length of stay (MD, 0.02 days; 95% CI, –0.95 to 1.00 days; N = 870), blood component use (MD, 0.03 L; 95% CI, –0.03 to 0.08 L; N = 1,547), or cardiac index (MD, 0.07 L/min/m 2 ; 95% CI, –0.10 to 0.25 L/min/m 2 ; N = 499). Fluid balance was lower with albumin compared with alternative solutions (MD, –0.55 L; 95% CI, –1.06 to –0.40 L; N = 450). The largest trial enrolled 1,386 patients and compared 4% albumin (20% albumin diluted in Ringer’s lactate) with Ringer’s lactate for both the pump prime and for fluid resuscitation ; albumin-treated patients showed higher rates of bleeding, resternotomy, and infection. Rationale for Recommendations Despite the common use of albumin during cardiovascular surgery, little evidence supports the use of albumin to improve patient outcomes. The largest study to date performed in 1,386 patients at a single center, Albumin in Cardiac Surgery, found increased morbidity when albumin was compared with Ringer’s lactate. Albumin in Cardiac Surgery was performed predominantly in low-risk cardiac surgery, and therefore, its role in improving outcomes in high-risk cardiac surgery has yet to be studied (a 590-patient RCT is underway, Albumin in Cardiac Surgery Australia; Identifier, ACTRN12619001355167). Clinical Setting 6: Patients With Cirrhosis Recommendations Recommendation 10: In patients with cirrhosis and ascites undergoing large - volume paracentesis (> 5 L ), intravenous albumin is suggested to prevent paracentesis-induced circulatory dysfunction (Conditional Recommendation, Very Low Certainty of Evidence of Effect). Recommendation 11: In patients with cirrhosis and spontaneous bacterial peritonitis, intravenous albumin is suggested to reduce mortality (Conditional Recommendation, Low Certainty of Evidence of Effect). Recommendation 12: In patients with cirrhosis and extraperitoneal infections, intravenous albumin is not suggested to reduce mortality or kidney failure (Conditional Recommendation, Low Certainty of Evidence of Effect). Recommendation 13: In hospitalized patients with decompensated cirrhosis with hypoalbuminemia (< 30 g/L), repeated intravenous albumin to increase albumin levels to > 30 g/L is not suggested to reduce infection, kidney dysfunction , or death (Conditional Recommendation, Low Certainty of Evidence of Effect). Recommendation 14: In outpatients with cirrhosis and uncomplicated ascites despite diuretic therapy, intravenous albumin is not routinely suggested to reduce complications associated with cirrhosis (Conditional Recommendation, Low Certainty of Evidence of Effect). Evidence Summary We identified a 2019 Cochrane systematic review including 27 RCTs (N = 1,592) examining the use of any plasma volume expanders in patients with cirrhosis undergoing paracentesis. In general, enrolled patients were undergoing large-volume paracentesis (> 5 L), and the most commonly used albumin doses were either 6 to 8 g of albumin per 1 L of fluid removed or a standard dose of 20 to 40 g. Compared with no plasma expander, no statistically significant effect of using hyperoncotic (20%-25%) albumin on mortality (RR, 0.52; 95% CI, 0.06-4.83), kidney impairment (RR, 0.32; 95% CI, 0.02-5.88), or recurrence of ascites (RR, 1.3; 95% CI, 0.49-3.42) was found. Compared with hyperoncotic albumin, use of other fluids showed uncertain effects on mortality (RR, 1.03; 95% CI, 0.82-1.30), kidney impairment (RR, 1.17; 95% CI, 0.71-1.91), and recurrence of ascites (RR, 1.14; 95% CI, 0.96-1.36). Paracentesis-induced circulatory dysfunction was more frequent with nonalbumin plasma expanders (RR, 1.98; 95% CI, 1.31-2.99) compared with albumin. A 2020 systematic review focused on the impact of different therapies (albumin, other fluids, vasoactive drugs) on the rate of postparacentesis circulatory dysfunction and identified nine RCTs (N = 620). Albumin at a dose of 8 g/L was found to be superior to other volume expanders for the prevention of postparacentesis circulatory dysfunction (rise in plasma renin activity by ≥ 50% of baseline). Similar to the Cochrane review, uncertainty regarding the role of albumin as compared with alternative treatments was noted for the prevention of complications after paracentesis. RCTs comparing high-dose albumin (6-8 g/L of ascitic fluid removed) with low-dose albumin (2-4 g/L of ascitic fluid removed) found no difference in the rate of paracentesis associated circulatory dysfunction, although uncertainty exists regarding the risk to benefit profile of the two doses, given the small sample size (two studies [N = 120]; RR, 1.00; 95% CI, 0.22-4.49). Two systematic reviews (in 2013 and 2020) identified five open-label RCTs in patients with spontaneous bacterial peritonitis both using variable doses and duration of hyperoncotic albumin (eg, 0.5-1.0 g/kg every 3 days for a maximum of 21 days; 1.5 g/kg on day 1 and 1.0 g/kg on day 3). , Albumin reduced the rate of kidney impairment (OR, 0.21; 95% CI, 0.11-0.42) and mortality (OR, 0.34; 95% CI, 0.19-0.60). The largest randomized trial randomized 126 patients to albumin (plus antibiotics) or antibiotics alone (without an explicit fluid resuscitation for the control arm). Patients treated with albumin showed lower rates of kidney impairment (10% vs 33%; P = .002) and in-hospital mortality (10% vs 29%; P = .01). The second largest trial randomized 118 patients to albumin (plus antibiotics) or antibiotics alone (without an explicit fluid resuscitation for the control arm). The primary end point of in-hospital mortality was not different (13% albumin vs 11% antibiotics alone; P = .66). A 2020 systematic review and meta-analysis of RCTs comparing albumin plus antibiotics with antibiotics alone in patients with cirrhosis and extraperitoneal infections found no effect on mortality or kidney impairment, but observed higher rates of pulmonary edema with albumin (three studies [N = 406]; OR, 5.17; 95% CI, 1.62-16.47). A 2019 systematic review in the same population also found no improvements in outcomes when albumin with antibiotics was compared with antibiotics alone. Subsequent to this 2020 systematic review, two randomized trials have been published (308 and 100 patients, respectively) comparing albumin with crystalloid in patients with cirrhosis and hypotension resulting from sepsis. , Both trials included patients with sepsis from all causes, including a small proportion (20%-25%) with spontaneous bacterial peritonitis. In the larger trial, survival at 7 days was not improved in the albumin-treated patients (saline, 39.0% vs albumin, 43.5%; P = .42, Fisher exact test); longer-term outcomes were not reported. In the second, smaller trial, albumin was superior to crystalloid for reversal of hypotension without initiation of vasopressors at 3 h (22% vs 62%; P < .001), but this improvement in hemodynamics did not reduce the rate of dialysis, length of stay, or mortality at 28 days. In the latter trial, patients randomized to albumin vs crystalloid showed higher rates of circulatory overload. We identified one RCT, Albumin to Prevent Infection in Chronic Liver Failure (N = 777), that evaluated the role of hyperoncotic albumin to target an albumin level of > 30 g/L (median, 200 g albumin over 14 days) as compared with no albumin in hospitalized patients with decompensated cirrhosis and hypoalbuminemia (< 30 g/L). No difference was found in the primary end point (composite of new infections, kidney dysfunction, or death between days 3 and 15) between groups (OR, 0.98; 95% CI, 0.71-1.33). More severe or life-threatening serious adverse events were reported in the albumin-treated patients, primarily a numerical increase in pulmonary edema. A 2021 systematic review was identified that evaluated albumin in patients with hepatic encephalopathy. The review identified two RCTs (N = 176). Albumin resulted in a reduction in hepatic encephalopathy (RR, 0.60; 95% CI, 0.38-0.95) and mortality (RR, 0.54; 95% CI, 0.33-0.90). The first open-label trial randomized 120 patients to albumin (1.5 g/kg/d for up to 10 days and lactulose) vs lactulose alone. Complete resolution of hepatic encephalopathy by day 10 was seen in 75% of the albumin-lactulose group vs 53% of the lactulose alone group ( P = .03). Mortality was 18% in the albumin-lactulose group vs 32% in the lactulose alone group at day 10 ( P = .04). The second masked RCT of albumin (1.5 g/kg on day 1 and 1.0 g/kg on day 3) vs normal saline enrolled 56 patients. No difference was found in the rate of resolution of hepatic encephalopathy at day 4 (albumin, 58% vs saline, 53%; P = .7). The mortality rate was lower in albumin-treated patients at 90 days (23% vs 47%) and transplant-free survival was improved ( P = .02, Kaplan-Meier estimate). A 2021 systematic review of RCTs and cohort studies evaluating the role of albumin in prevention and treatment suggested that albumin may assist with the resolution or prevention of hepatic encephalopathy and may reduce mortality ; only the two RCTs identified in the aforementioned systematic review were identified for the treatment of hepatic encephalopathy. In the subsequent large Albumin to Prevent Infection in Chronic Liver Failure trial, the subgroup (N = 149) of patients admitted with hepatic encephalopathy randomized to albumin as compared with placebo did not show an improvement in the composite end point of new infections, kidney dysfunction, or death between days 3 and 15 (adjusted OR, 0.91; 95% CI, 0.44-1.86). Subsequent to the two systematic reviews, a single RCT was identified that randomized 48 outpatients with hepatic encephalopathy to weekly hyperoncotic albumin for 5 weeks as compared with saline and found improvements in cognitive function with albumin. A 2021 systematic review of RCTs evaluating outpatient hyperoncotic albumin for patients with cirrhosis and ascites identified five trials (N = 716). The systematic review found no difference in mortality at 12 to 36 months (RR, 0.88; 95% CI, 0.67-1.14) or any other outcome, with the exception of reducing the need for paracentesis (RR, 0.56; 95% CI, 0.48-0.67). Two large randomized trials were included in the review. , The first unmasked trial randomized 440 patients with cirrhosis and uncomplicated, persistent ascites despite diuretic therapy to albumin (40 g twice weekly for 2 weeks and then 40 g weekly for up to 18 months) or no albumin. Patients randomized to albumin experienced longer time to first paracentesis; required fewer paracenteses; were less likely to demonstrate hepatic encephalopathy, hepatorenal syndrome, spontaneous bacterial peritonitis, nonperitonitis infections, hyponatremia, or episodes of kidney dysfunction; experienced fewer days in hospital; and showed lower all-cause mortality (77% vs 66% survival at 18 months; hazard ratio, 0.62; 95% CI, 0.40-0.95). The most important limitation of this study is that the albumin-treated patients underwent weekly health care interactions and the control group did not, raising the concern that the observed differences may have been the result of increased health care exposure. The second trial randomized 196 outpatients with cirrhosis and ascites awaiting liver transplantation to oral midodrine and albumin as compared with placebo tablets and a 0.9% saline infusion and found no difference in patient outcomes. The dose of albumin given as part of the intervention was lower (40 g every 15 days). The study improved on the methodology of the first trial by achieving masking to treatment assignment and ensuring the same health care exposure in both study groups. Rationale for Recommendations Approximately one-third of albumin is used for patients with cirrhosis, and although this practice is exceedingly common, the certainty of evidence supporting this therapy in this population is insufficient to allow for strong recommendations. Although the use of albumin for large-volume paracentesis is a commonly accepted clinical practice and is endorsed by guidelines, , , the reported trials have important limitations that affect the certainty in outcomes. These trials included a small number of patients and findings for most patient-important outcomes (mortality, kidney dysfunction) were imprecise, leaving residual uncertainty regarding true clinical benefits and harms. Albumin, as compared with other fluid expanders, may be superior for the prevention of paracentesis-induced circulatory dysfunction (rise in serum renin level on the sixth day after paracentesis), but whether this translates to improvement in patient-important outcomes is less certain. Plasma renin levels are predictive of greater morbidity in patients with cirrhosis. , , The panel suggested continuing this commonly accepted practice for patients undergoing large-volume paracentesis, but believed the data supported only a conditional recommendation based on low-quality evidence. Further trials are needed urgently to clarify if albumin improves patient important outcomes, to elucidate the optimal dosing strategy, to further the understanding of the safety profile of the treatment, and to evaluate alternative fluids and therapies. It is unclear if improving laboratory measures of paracentesis-induced circulatory dysfunction will translate into reductions in renal failure, hospital admission, or other patient-important outcomes. The panel also highlighted the need to personalize the use of albumin, the dose after paracentesis, or both, considering the patient’s baseline creatinine, volume of ascites removed, and history of hypotensive symptoms after prior procedures. Similarly, the role of albumin for improving outcomes in patients with spontaneous bacterial peritonitis is unclear. The trial data specific to this patient population are limited. , The two largest RCTs failed to provide an explicit fluid resuscitation protocol for the patients randomized to no albumin, raising the concern for underresuscitation in the control arms of both studies. When similar albumin dosing strategies were used in trials examining patients with cirrhosis and extraperitoneal infections, no benefit was seen and concern for harm was expressed. The panel suggested the use of albumin for spontaneous bacterial peritonitis (conditional recommendation), but raised concerns regarding the dosing protocol used in two of the four trials and the risk of fluid overload (1.5 g/kg on day 1 and 1.0 g/kg on day 3) and the lack of data suggesting this specific regimen is beneficial compared with alternative dosing (eg, lower dose daily for 3 days). The panel also considered the lack of clarity on whether albumin is necessary for all patients with spontaneous bacterial peritonitis or whether it could be used selectively (ie, patients at high risk of kidney failure or death: serum bilirubin > 4 mg/dL or serum creatinine >1 mg/dL). Additional studies are necessary to address dosing, to address the benefit for patients with and without kidney impairment, and to clarify the risks of adverse events. The panel also noted that not all physicians currently adhere to the trial dosing strategy, , although it continues to be recommended in current guidelines. , A careful assessment of the patient’s volume status, cardiovascular status, and degree of kidney impairment before transfusion is advised and the dose, frequency, or both being modified accordingly. In contrast, the RCTs find no support for the use of albumin in patients with cirrhosis and extraperitoneal infections. In the setting of patients admitted with decompensated cirrhosis and hypoalbuminemia, this guideline is informed by an RCT involving 777 patients that found no improvement in patient important outcomes and a concern for increased adverse events. This led the panel to suggest conditionally against the use of albumin in this setting. Although a 2021 systematic review of two small RCTs suggested a benefit for facilitating resolution of hepatic encephalopathy and reducing mortality, the subgroup of patients in the Albumin to Prevent Infection in Chronic Liver Failure study admitted with hepatic encephalopathy did not show improvements in mortality. The panel had uncertainty regarding the benefit of albumin in this patient population and few data on the risks of the treatment, and therefore abstained from making a statement on the role of albumin in this setting until further adequately powered RCTs are conducted. In nonhospitalized patients with cirrhosis and persistent ascites despite optimized medical management, the role of weekly or biweekly albumin infusions remains unclear. One unmasked study of weekly albumin infusions found improvements in outcomes, but this was not replicated in a placebo-controlled trial that examined biweekly albumin infusions. The latter trial enrolled a smaller number of patients and used a lower dose, and therefore may have failed to detect a difference in outcomes. The panel reported residual uncertainty regarding the benefit of this treatment and given this, suggested against its routine use until additional RCTs have been conducted. The use of weekly albumin in this patient population would have considerable impacts on patients, would require chronic IV access, would have considerable impacts on outpatient infusion clinics, and would require a dependable supply of albumin. Although the unmasked trial reported cost-effectiveness, additional masked trials with cost-effectiveness analyses are necessary to improve precision and generalizability and to inform future guidelines. A 2020 and a 2019 systematic review on the treatment of hepatorenal syndrome did not identify any randomized trials examining albumin for these patients as compared with placebo or no treatment. Rather, all trials examining this patient population uniformly have administered albumin in both treatment and control arms and have compared vasoconstrictor agents (eg, terlipressin, midodrine) with placebo infusions. Hence, no recommendations regarding the use of albumin for patients with cirrhosis and hepatorenal syndrome could be made. Recommendations Recommendation 1: In critically ill adult patients (excluding patients with thermal injuries and ARDS), intravenous albumin is not suggested for first-line volume replacement or to increase serum albumin levels (Conditional Recommendation, Moderate Certainty of Evidence of Effect). Recommendation 2: In critically ill adult patients with thermal injuries or ARDS, intravenous albumin is not suggested for volume replacement or to increase serum albumin level (Conditional Recommendation, Very Low Certainty of Evidence of Effect). Recommendation 3: In critically ill adult patients, intravenous albumin in conjunction with diuretics is not suggested for removal of extravascular fluid (Conditional Recommendation, Very Low Certainty of Evidence of Effect). Evidence Summary Sixteen , , , , , , , , , , , , , , , of 19 systematic reviews were retrieved and included. These reports included a broad critical care patient population, including patients with critical illness, sepsis, thermal injuries, and ARDS. Three of the 19 systematic reviews were excluded because they assessed the impact of albumin only on fluid balance, gelatin vs colloids, or all colloids compared with crystalloids (without reporting albumin vs other fluids). A systematic review from 2019 identified 55 RCTs comparing crystalloid with colloids in critical care. Data on mortality were available for 26,329 patients from 46 studies. No mortality benefit was found when crystalloid was compared with albumin (relative risk [RR] 1.02; 95% CI, 0.96-1.10). Crystalloids were less effective than colloids in hemodynamic resuscitation end points (eg, mean arterial pressure) but this did not translate into improvements in patient outcomes. After this systematic review, one RCT was identified that examined 360 patients with sepsis with an underlying diagnosis of cancer (albumin was compared with Ringer’s lactate); no differences in mortality or other outcomes were found. A systematic review from 2018 conducted by the Cochrane collaboration found no difference in mortality in patients in the ICU (20 studies; N = 13,047) when patients managed with crystalloids were compared with those managed with albumin at the end of follow-up (RR, 0.98; 95% CI, 0.92-1.06), at 30 days (RR, 0.99; 95% CI, 0.93-1.06), or at 90 days (RR, 0.98; 95% CI, 0.92-1.04) or who needed kidney replacement therapy (RR, 1.11; 95% CI, 0.96-1.27). The largest randomized trial is the Saline Versus Albumin Fluid Evaluation trial published in 2004, which enrolled 6,997 patients receiving critical care (including a mix of medical and surgical patients) and compared 4% albumin with 0.9% normal saline. No differences were found in outcomes, including 28-day mortality (RR, 0.99; 95% CI, 0.91-1.09). A 2015 systematic review evaluated the administration of albumin in critical care patients with traumatic injury; the review included five trials comparing albumin with crystalloid and found a higher mortality in albumin-treated patients (RR, 1.35; 95% CI, 1.03-1.77). This systematic review was dominated by the Saline Versus Albumin Fluid Evaluation trial (57% of patients). The Saline Versus Albumin Fluid Evaluation trial subgroup analysis found that patients with traumatic brain injury showed a higher mortality rate (RR, 1.62; 95% CI, 1.12-2.34), but those without traumatic brain injury did not (RR, 1.00; 95% CI, 0.56-1.79). Hence, it is uncertain whether albumin may be unsafe only in patients with traumatic brain injury as compared with the wider trauma population. A 2020 systematic review and sequential network analysis of RCTs in the setting of sepsis included 23 randomized trials (N = 14,659); the vast majority of the trials used a physiologic target for volume resuscitation or at the discretion of the clinician, rather than a target albumin level. The review found albumin not to be superior to crystalloids for mortality or acute kidney injury. A 2014 systematic review included 16 randomized trials (N = 4,190) comparing crystalloid or albumin and found no difference in mortality (RR, 0.94; 95% CI, 0.87-1.01). Two network meta-analyses have been performed and reported no mortality benefit from albumin. , The largest randomized trial in sepsis was the Albumin Italian Outcome Sepsis trial, which randomized 1,818 patients with sepsis at 100 sites to 20% albumin (targeting plasma albumin level of ≥ 30 g/L) vs crystalloid. The Albumin Italian Outcome Sepsis trial did not observe improvements in mortality at 28 days (RR, 1.00; 95% CI, 0.87-1.14) or other important outcomes. Three systematic reviews found no impact of albumin in critically ill adults on the need for kidney replacement therapy, including two network meta-analyses , and the 2018 Cochrane review. A systematic review evaluated the impact of albumin on patient outcomes after thermal injuries. The report identified four RCTs and found no impact on the incidence of kidney failure or mortality (RR, 1.41; 95% CI, 0.27-7.38). In a 2022 systematic review evaluating the impact of albumin and diuretics, as compared with diuretics alone, in mechanically ventilated patients (three trials; N = 129), albumin reduced hypotensive episodes, but did not shorten the duration of mechanical ventilation or improve the mortality rate. A 2014 systematic review evaluated the impact of albumin, as compared with crystalloid, in patients with ARDS. Three RCTs (N = 204) were included; no difference in mortality was found (RR, 0.89; 95% CI, 0.62-1.28). Similarly, a 2014 systematic review that included two small RCTs (N = 70) found no difference in ventilator-free days or mortality when albumin with diuretics, as compared with diuretics alone, were used to improve respiratory status in critically ill patients. A 2014 systematic review evaluated the impact of albumin with furosemide, compared with furosemide alone, to facilitate fluid removal in patients with hypoalbuminemia and hypervolemia. The systematic review identified 10 studies (N = 343). Although urine output was higher at 6 h in the patients receiving albumin-furosemide, no difference was found in urine output at 24 h. One RCT of 49 patients with edema receiving critical care was identified subsequent to this systematic review that compared albumin and furosemide with furosemide alone; no difference in urine output at 8 h was found. Rationale for Recommendations A substantial amount of evidence from RCTs in critically ill adult patients across a wide range of patient subgroups provides little supportive evidence for the use of albumin as fluid replacement to reduce mortality, the need for kidney replacement therapy, or other outcomes considered important or critical for decision-making by the panel. Given the wide CIs for the estimates from the systematic reviews, all recommendations were considered conditional because of the residual uncertainty. In systematic reviews evaluating the role of albumin in patients with sepsis, the use of albumin has not been found to be associated with improved outcomes, although a benefit has not been excluded because of the wide CI in the most recent systematic review. The Surviving Sepsis Campaign guidelines published in 2021 recommend albumin in addition to crystalloids when patients require large volumes of crystalloids (Weak Recommendation, Moderate-Quality Evidence). Specific formulations of albumin (4%-5% or 20%-25%), volumes or doses, serum albumin targets, or a combination thereof were not described. The guidelines state, “The lack of proven benefit and higher cost of albumin compared to crystalloid contributed to our strong recommendation for the use of crystalloids as first-line fluid for resuscitation in sepsis and septic shock.” More studies will be needed to evaluate the role and timing of albumin as a rescue fluid in patients with sepsis failing front-line crystalloid resuscitation, particularly given the considerably higher cost of albumin compared with crystalloids, the risks of albumin, and the lack of benefit shown in RCTs. Recommendation 1: In critically ill adult patients (excluding patients with thermal injuries and ARDS), intravenous albumin is not suggested for first-line volume replacement or to increase serum albumin levels (Conditional Recommendation, Moderate Certainty of Evidence of Effect). Recommendation 2: In critically ill adult patients with thermal injuries or ARDS, intravenous albumin is not suggested for volume replacement or to increase serum albumin level (Conditional Recommendation, Very Low Certainty of Evidence of Effect). Recommendation 3: In critically ill adult patients, intravenous albumin in conjunction with diuretics is not suggested for removal of extravascular fluid (Conditional Recommendation, Very Low Certainty of Evidence of Effect). Sixteen , , , , , , , , , , , , , , , of 19 systematic reviews were retrieved and included. These reports included a broad critical care patient population, including patients with critical illness, sepsis, thermal injuries, and ARDS. Three of the 19 systematic reviews were excluded because they assessed the impact of albumin only on fluid balance, gelatin vs colloids, or all colloids compared with crystalloids (without reporting albumin vs other fluids). A systematic review from 2019 identified 55 RCTs comparing crystalloid with colloids in critical care. Data on mortality were available for 26,329 patients from 46 studies. No mortality benefit was found when crystalloid was compared with albumin (relative risk [RR] 1.02; 95% CI, 0.96-1.10). Crystalloids were less effective than colloids in hemodynamic resuscitation end points (eg, mean arterial pressure) but this did not translate into improvements in patient outcomes. After this systematic review, one RCT was identified that examined 360 patients with sepsis with an underlying diagnosis of cancer (albumin was compared with Ringer’s lactate); no differences in mortality or other outcomes were found. A systematic review from 2018 conducted by the Cochrane collaboration found no difference in mortality in patients in the ICU (20 studies; N = 13,047) when patients managed with crystalloids were compared with those managed with albumin at the end of follow-up (RR, 0.98; 95% CI, 0.92-1.06), at 30 days (RR, 0.99; 95% CI, 0.93-1.06), or at 90 days (RR, 0.98; 95% CI, 0.92-1.04) or who needed kidney replacement therapy (RR, 1.11; 95% CI, 0.96-1.27). The largest randomized trial is the Saline Versus Albumin Fluid Evaluation trial published in 2004, which enrolled 6,997 patients receiving critical care (including a mix of medical and surgical patients) and compared 4% albumin with 0.9% normal saline. No differences were found in outcomes, including 28-day mortality (RR, 0.99; 95% CI, 0.91-1.09). A 2015 systematic review evaluated the administration of albumin in critical care patients with traumatic injury; the review included five trials comparing albumin with crystalloid and found a higher mortality in albumin-treated patients (RR, 1.35; 95% CI, 1.03-1.77). This systematic review was dominated by the Saline Versus Albumin Fluid Evaluation trial (57% of patients). The Saline Versus Albumin Fluid Evaluation trial subgroup analysis found that patients with traumatic brain injury showed a higher mortality rate (RR, 1.62; 95% CI, 1.12-2.34), but those without traumatic brain injury did not (RR, 1.00; 95% CI, 0.56-1.79). Hence, it is uncertain whether albumin may be unsafe only in patients with traumatic brain injury as compared with the wider trauma population. A 2020 systematic review and sequential network analysis of RCTs in the setting of sepsis included 23 randomized trials (N = 14,659); the vast majority of the trials used a physiologic target for volume resuscitation or at the discretion of the clinician, rather than a target albumin level. The review found albumin not to be superior to crystalloids for mortality or acute kidney injury. A 2014 systematic review included 16 randomized trials (N = 4,190) comparing crystalloid or albumin and found no difference in mortality (RR, 0.94; 95% CI, 0.87-1.01). Two network meta-analyses have been performed and reported no mortality benefit from albumin. , The largest randomized trial in sepsis was the Albumin Italian Outcome Sepsis trial, which randomized 1,818 patients with sepsis at 100 sites to 20% albumin (targeting plasma albumin level of ≥ 30 g/L) vs crystalloid. The Albumin Italian Outcome Sepsis trial did not observe improvements in mortality at 28 days (RR, 1.00; 95% CI, 0.87-1.14) or other important outcomes. Three systematic reviews found no impact of albumin in critically ill adults on the need for kidney replacement therapy, including two network meta-analyses , and the 2018 Cochrane review. A systematic review evaluated the impact of albumin on patient outcomes after thermal injuries. The report identified four RCTs and found no impact on the incidence of kidney failure or mortality (RR, 1.41; 95% CI, 0.27-7.38). In a 2022 systematic review evaluating the impact of albumin and diuretics, as compared with diuretics alone, in mechanically ventilated patients (three trials; N = 129), albumin reduced hypotensive episodes, but did not shorten the duration of mechanical ventilation or improve the mortality rate. A 2014 systematic review evaluated the impact of albumin, as compared with crystalloid, in patients with ARDS. Three RCTs (N = 204) were included; no difference in mortality was found (RR, 0.89; 95% CI, 0.62-1.28). Similarly, a 2014 systematic review that included two small RCTs (N = 70) found no difference in ventilator-free days or mortality when albumin with diuretics, as compared with diuretics alone, were used to improve respiratory status in critically ill patients. A 2014 systematic review evaluated the impact of albumin with furosemide, compared with furosemide alone, to facilitate fluid removal in patients with hypoalbuminemia and hypervolemia. The systematic review identified 10 studies (N = 343). Although urine output was higher at 6 h in the patients receiving albumin-furosemide, no difference was found in urine output at 24 h. One RCT of 49 patients with edema receiving critical care was identified subsequent to this systematic review that compared albumin and furosemide with furosemide alone; no difference in urine output at 8 h was found. A substantial amount of evidence from RCTs in critically ill adult patients across a wide range of patient subgroups provides little supportive evidence for the use of albumin as fluid replacement to reduce mortality, the need for kidney replacement therapy, or other outcomes considered important or critical for decision-making by the panel. Given the wide CIs for the estimates from the systematic reviews, all recommendations were considered conditional because of the residual uncertainty. In systematic reviews evaluating the role of albumin in patients with sepsis, the use of albumin has not been found to be associated with improved outcomes, although a benefit has not been excluded because of the wide CI in the most recent systematic review. The Surviving Sepsis Campaign guidelines published in 2021 recommend albumin in addition to crystalloids when patients require large volumes of crystalloids (Weak Recommendation, Moderate-Quality Evidence). Specific formulations of albumin (4%-5% or 20%-25%), volumes or doses, serum albumin targets, or a combination thereof were not described. The guidelines state, “The lack of proven benefit and higher cost of albumin compared to crystalloid contributed to our strong recommendation for the use of crystalloids as first-line fluid for resuscitation in sepsis and septic shock.” More studies will be needed to evaluate the role and timing of albumin as a rescue fluid in patients with sepsis failing front-line crystalloid resuscitation, particularly given the considerably higher cost of albumin compared with crystalloids, the risks of albumin, and the lack of benefit shown in RCTs. Recommendation Recommendation 4: In pediatric patients with infection and hypoperfusion, intravenous albumin is not recommended to reduce mortality (Strong Recommendation, Low Certainty of Evidence of Effect). Evidence Summary A single systematic review identified three RCTs that compared albumin with crystalloid in critically ill children. , , All RCTs enrolled children primarily in African countries with either severe malaria or febrile illness with impaired perfusion. The first trial enrolled 61 children with severe malaria and found no difference in mortality when albumin was compared with crystalloid. The second trial enrolled 150 children with severe malaria and found an improvement in the mortality with albumin as compared with crystalloid. A mortality difference was not found in a large, well-designed RCT (Fluid Expansion as Supportive Therapy; N = 3,141) that included children with severe febrile illness with impaired perfusion (60% had malaria). This RCT had three arms comparing saline bolus, 5% albumin bolus, and no bolus. The trial was terminated by the independent data safety monitoring committee at the fifth interim analysis based on data from 2,995 children and after 3,141 of 3,600 planned patients were enrolled because of excess mortality in the patients treated with both the albumin bolus (RR, 1.45; 95% CI, 1.10-1.92) and the saline bolus (RR, 1.44; 95% CI, 1.09-1.90) compared with children who received no bolus. No mortality difference was found when the albumin bolus arm was compared with the crystalloid bolus arm (RR, 1.00; 95% CI, 0.78-1.29) at 48 h. Similar differences in mortality were observed between groups at 28 days, again with an excess mortality in the albumin and saline bolus groups compared with the no bolus group (RR, 1.40 [95% CI, 1.08-1.80] and 1.38 [95% CI, 1.07-1.78]). Children treated with both saline and albumin boluses showed higher rates of respiratory and neurologic dysfunction and of hyperchloremic acidosis and a greater reduction in hemoglobin levels. Rationale for Recommendations The systematic review of the literature for pediatric patients receiving critical care found fewer RCTs as compared with studies of adult patients. Among them, a very large trial of children with febrile illness and hypoperfusion found excess mortality when either an albumin bolus or a crystalloid bolus strategy was compared with a no bolus strategy. Given the extensive, albeit indirect, literature base in adult critical care showing no improvement in mortality or other important outcomes and the above large trial in children suggesting excess mortality with a front-line albumin bolus strategy, pediatric intensivists probably should not use albumin as a first-line treatment outside of a clinical trial for severe infections in critically ill children. Because most children enrolled in these RCTs had malaria, it is uncertain if the results are applicable to all critically ill children with infections or the broader pediatric critical care population. In addition, the increased mortality in the Fluid Expansion as Supportive Therapy trial may be the result of the bolus administration, rather than the type of fluid, with substudies of the Fluid Expansion as Supportive Therapy trial showing that the bolus of either fluid type was associated with higher rates of cardiovascular collapse. Recommendation 4: In pediatric patients with infection and hypoperfusion, intravenous albumin is not recommended to reduce mortality (Strong Recommendation, Low Certainty of Evidence of Effect). A single systematic review identified three RCTs that compared albumin with crystalloid in critically ill children. , , All RCTs enrolled children primarily in African countries with either severe malaria or febrile illness with impaired perfusion. The first trial enrolled 61 children with severe malaria and found no difference in mortality when albumin was compared with crystalloid. The second trial enrolled 150 children with severe malaria and found an improvement in the mortality with albumin as compared with crystalloid. A mortality difference was not found in a large, well-designed RCT (Fluid Expansion as Supportive Therapy; N = 3,141) that included children with severe febrile illness with impaired perfusion (60% had malaria). This RCT had three arms comparing saline bolus, 5% albumin bolus, and no bolus. The trial was terminated by the independent data safety monitoring committee at the fifth interim analysis based on data from 2,995 children and after 3,141 of 3,600 planned patients were enrolled because of excess mortality in the patients treated with both the albumin bolus (RR, 1.45; 95% CI, 1.10-1.92) and the saline bolus (RR, 1.44; 95% CI, 1.09-1.90) compared with children who received no bolus. No mortality difference was found when the albumin bolus arm was compared with the crystalloid bolus arm (RR, 1.00; 95% CI, 0.78-1.29) at 48 h. Similar differences in mortality were observed between groups at 28 days, again with an excess mortality in the albumin and saline bolus groups compared with the no bolus group (RR, 1.40 [95% CI, 1.08-1.80] and 1.38 [95% CI, 1.07-1.78]). Children treated with both saline and albumin boluses showed higher rates of respiratory and neurologic dysfunction and of hyperchloremic acidosis and a greater reduction in hemoglobin levels. The systematic review of the literature for pediatric patients receiving critical care found fewer RCTs as compared with studies of adult patients. Among them, a very large trial of children with febrile illness and hypoperfusion found excess mortality when either an albumin bolus or a crystalloid bolus strategy was compared with a no bolus strategy. Given the extensive, albeit indirect, literature base in adult critical care showing no improvement in mortality or other important outcomes and the above large trial in children suggesting excess mortality with a front-line albumin bolus strategy, pediatric intensivists probably should not use albumin as a first-line treatment outside of a clinical trial for severe infections in critically ill children. Because most children enrolled in these RCTs had malaria, it is uncertain if the results are applicable to all critically ill children with infections or the broader pediatric critical care population. In addition, the increased mortality in the Fluid Expansion as Supportive Therapy trial may be the result of the bolus administration, rather than the type of fluid, with substudies of the Fluid Expansion as Supportive Therapy trial showing that the bolus of either fluid type was associated with higher rates of cardiovascular collapse. Recommendation 5: In preterm neonates (≤ 36 weeks) with low serum albumin levels and respiratory distress, intravenous albumin is not suggested to improve respiratory function (Conditional Recommendation, Very Low Certainty of Evidence of Effect). Recommendation 6: In preterm neonates (≤ 32 weeks or ≤ 1,500 g) with or without hypoperfusion, intravenous albumin is not suggested for volume replacement (Conditional Recommendation, Very Low Certainty of Evidence of Effect). Evidence Summary A Cochrane systematic review evaluated the use of albumin in preterm neonates (≤ 36 weeks’ gestation at birth) with hypoalbuminemia (two RCTs enrolling 64 preterm neonates). Only one study reported mortality rates and no difference was found. No other important differences in outcomes were observed. A Cochrane systematic review of RCTs of early volume expansion compared normal saline, plasma, albumin, plasma substitutes, or blood with no treatment or another fluid treatment in preterm neonates (≤ 32 weeks or ≤ 1,500 g). Early volume expansion was defined as > 10 mL/kg of body weight in the first 3 days. The studies included variable indications for the administration of IV fluids. Eight studies were identified, with four studies evaluating albumin with a comparative arm (two vs normal saline, one vs plasma, and one vs no treatment). The two studies (N = 102 and N = 63) comparing 5% albumin with normal saline in hypotensive infants found no difference in mortality (RR, 1.02; 95% CI, 0.50-2.06) or any other patient-important outcomes. The one study (N = 25) comparing 20% albumin with no treatment in normotensive infants also found no difference in mortality (RR, 0.92; 95% CI, 0.23-3.72). Finally, one trial (N = 20) in hypotensive infants compared plasma with albumin and found no difference in duration of ventilation (mortality not reported). Since the publication of these two Cochrane reviews, a single RCT (N = 33) was identified comparing 5% albumin with normal saline (both 10 mL/kg) for term infants with dehydration, metabolic acidosis, and diarrhea and found no differences in outcomes. Rationale for Recommendations Few RCTs have evaluated the impact of albumin compared with other alternative fluids in preterm or term neonates with either hypoalbuminemia or hypovolemia. Very little evidence is available in the literature to guide the use of albumin in term neonates. All trials in the two systematic reviews included small numbers of neonates, preventing any definitive conclusions. Indirect evidence from the adult and pediatric literature, the costs of albumin, and the lack of trials assessing the potential harms of albumin should be considered when including albumin in neonatal fluid protocols. A Cochrane systematic review evaluated the use of albumin in preterm neonates (≤ 36 weeks’ gestation at birth) with hypoalbuminemia (two RCTs enrolling 64 preterm neonates). Only one study reported mortality rates and no difference was found. No other important differences in outcomes were observed. A Cochrane systematic review of RCTs of early volume expansion compared normal saline, plasma, albumin, plasma substitutes, or blood with no treatment or another fluid treatment in preterm neonates (≤ 32 weeks or ≤ 1,500 g). Early volume expansion was defined as > 10 mL/kg of body weight in the first 3 days. The studies included variable indications for the administration of IV fluids. Eight studies were identified, with four studies evaluating albumin with a comparative arm (two vs normal saline, one vs plasma, and one vs no treatment). The two studies (N = 102 and N = 63) comparing 5% albumin with normal saline in hypotensive infants found no difference in mortality (RR, 1.02; 95% CI, 0.50-2.06) or any other patient-important outcomes. The one study (N = 25) comparing 20% albumin with no treatment in normotensive infants also found no difference in mortality (RR, 0.92; 95% CI, 0.23-3.72). Finally, one trial (N = 20) in hypotensive infants compared plasma with albumin and found no difference in duration of ventilation (mortality not reported). Since the publication of these two Cochrane reviews, a single RCT (N = 33) was identified comparing 5% albumin with normal saline (both 10 mL/kg) for term infants with dehydration, metabolic acidosis, and diarrhea and found no differences in outcomes. Few RCTs have evaluated the impact of albumin compared with other alternative fluids in preterm or term neonates with either hypoalbuminemia or hypovolemia. Very little evidence is available in the literature to guide the use of albumin in term neonates. All trials in the two systematic reviews included small numbers of neonates, preventing any definitive conclusions. Indirect evidence from the adult and pediatric literature, the costs of albumin, and the lack of trials assessing the potential harms of albumin should be considered when including albumin in neonatal fluid protocols. Recommendation Recommendation 7: In patients undergoing kidney replacement therapy, intravenous albumin is not suggested for prevention or treatment of intradialytic hypotension or for improving ultrafiltration (Conditional Recommendation, Very Low Certainty of Evidence of Effect). Evidence Summary A single Cochrane systematic review was identified evaluating the use of albumin, compared with an alternative strategy, for the treatment of intradialytic hypotension. The review identified a single (N = 45) randomized crossover trial of 5% albumin compared with normal saline and did not find a difference in the primary outcome (percentage target ultrafiltration achieved) or other clinical outcomes. Two small crossover trials identified in this review evaluated 20% albumin as compared with gelatin (N = 10) and a three-arm study compared 20% albumin with both saline and hydroxyethyl starch (N = 10). , These RCTs suggested that BP was maintained better with albumin vs other fluid, but found no improvements in other outcomes, including improving ultrafiltration. Finally, a 2021 randomized crossover trial involving 65 hospitalized patients requiring hemodialysis with serum albumin levels of < 30 g/L found that hypotension, lowest intradialytic systolic BP, and ultrafiltration rate were improved with 25% albumin compared with saline. Rationale for Recommendation Intradialytic hypotension and fluid overload are experienced commonly during kidney replacement therapy. , Patients with intradialytic hypotension are at greater risk of morbidity and mortality. Given the costs of albumin, the need for thrice weekly treatment for patients receiving maintenance hemodialysis, and the lack of evidence to support superiority over less costly fluid alternatives, alternative fluids or treatments need to be considered. The annual cost of 25 g of albumin given with thrice-weekly maintenance dialysis is estimated at $20,000 per patient (United States dollars). Midodrine (an oral vasopressor) given alone or in combination with use of a high dialysate calcium concentration and lower dialysate temperature has been explored as a therapeutic option to mitigate intradialytic hypotension. , , In patients prescribed kidney replacement therapy, higher dialysate calcium, lower dialysate temperature, individualized ultrafiltration rates, or a combination of these strategies may mitigate intradialytic hypotension. , , Further studies are needed to understand the pathophysiology of intradialytic hypotension to determine if albumin prevents intradialytic hypotension or improves ultrafiltration, mitigates associated symptoms, or improves patient-important outcomes. Recommendation 7: In patients undergoing kidney replacement therapy, intravenous albumin is not suggested for prevention or treatment of intradialytic hypotension or for improving ultrafiltration (Conditional Recommendation, Very Low Certainty of Evidence of Effect). A single Cochrane systematic review was identified evaluating the use of albumin, compared with an alternative strategy, for the treatment of intradialytic hypotension. The review identified a single (N = 45) randomized crossover trial of 5% albumin compared with normal saline and did not find a difference in the primary outcome (percentage target ultrafiltration achieved) or other clinical outcomes. Two small crossover trials identified in this review evaluated 20% albumin as compared with gelatin (N = 10) and a three-arm study compared 20% albumin with both saline and hydroxyethyl starch (N = 10). , These RCTs suggested that BP was maintained better with albumin vs other fluid, but found no improvements in other outcomes, including improving ultrafiltration. Finally, a 2021 randomized crossover trial involving 65 hospitalized patients requiring hemodialysis with serum albumin levels of < 30 g/L found that hypotension, lowest intradialytic systolic BP, and ultrafiltration rate were improved with 25% albumin compared with saline. Intradialytic hypotension and fluid overload are experienced commonly during kidney replacement therapy. , Patients with intradialytic hypotension are at greater risk of morbidity and mortality. Given the costs of albumin, the need for thrice weekly treatment for patients receiving maintenance hemodialysis, and the lack of evidence to support superiority over less costly fluid alternatives, alternative fluids or treatments need to be considered. The annual cost of 25 g of albumin given with thrice-weekly maintenance dialysis is estimated at $20,000 per patient (United States dollars). Midodrine (an oral vasopressor) given alone or in combination with use of a high dialysate calcium concentration and lower dialysate temperature has been explored as a therapeutic option to mitigate intradialytic hypotension. , , In patients prescribed kidney replacement therapy, higher dialysate calcium, lower dialysate temperature, individualized ultrafiltration rates, or a combination of these strategies may mitigate intradialytic hypotension. , , Further studies are needed to understand the pathophysiology of intradialytic hypotension to determine if albumin prevents intradialytic hypotension or improves ultrafiltration, mitigates associated symptoms, or improves patient-important outcomes. Recommendations Recommendation 8: In adult patients undergoing cardiovascular surgery, intravenous albumin is not suggested for priming the cardiovascular bypass circuit or volume replacement (Conditional Recommendation, Moderate Certainty of Evidence of Effect). Recommendation 9: In pediatric patients undergoing cardiovascular surgery, intravenous albumin is not suggested for priming the cardiovascular bypass circuit or volume replacement (Conditional Recommendation, Very Low Certainty of Evidence of Effect). Evidence Summary A systematic review and meta-analysis of RCTs in pediatric and adult patients undergoing cardiovascular surgery was performed. We identified 43 randomized studies (N = 3,862), comparing albumin with gelatin, starches, or crystalloid solutions for priming the cardiopulmonary bypass circuit, volume expansion, or both. The majority of the trials were conducted in patients undergoing on-pump cardiac surgery, with the exception of two RCTs conducted in patients undergoing off-pump cardiac surgery. , Albumin infusion did not result in a lower mortality rate when compared with other fluids (risk difference, 0.00; 95% CI, –0.01 to 0.01; N = 2,711). No differences were found for the rates of kidney failure (risk difference, 0.01; 95% CI, –0.01 to 0.03; N = 1,703), blood loss (mean difference [MD], –0.04 L; 95% CI, –0.04 to 0.01 L), ICU length of stay (MD, –0.12 days; 95% CI, –0.31 to 0.06 days; N = 1,371), hospital length of stay (MD, 0.02 days; 95% CI, –0.95 to 1.00 days; N = 870), blood component use (MD, 0.03 L; 95% CI, –0.03 to 0.08 L; N = 1,547), or cardiac index (MD, 0.07 L/min/m 2 ; 95% CI, –0.10 to 0.25 L/min/m 2 ; N = 499). Fluid balance was lower with albumin compared with alternative solutions (MD, –0.55 L; 95% CI, –1.06 to –0.40 L; N = 450). The largest trial enrolled 1,386 patients and compared 4% albumin (20% albumin diluted in Ringer’s lactate) with Ringer’s lactate for both the pump prime and for fluid resuscitation ; albumin-treated patients showed higher rates of bleeding, resternotomy, and infection. Rationale for Recommendations Despite the common use of albumin during cardiovascular surgery, little evidence supports the use of albumin to improve patient outcomes. The largest study to date performed in 1,386 patients at a single center, Albumin in Cardiac Surgery, found increased morbidity when albumin was compared with Ringer’s lactate. Albumin in Cardiac Surgery was performed predominantly in low-risk cardiac surgery, and therefore, its role in improving outcomes in high-risk cardiac surgery has yet to be studied (a 590-patient RCT is underway, Albumin in Cardiac Surgery Australia; Identifier, ACTRN12619001355167). Recommendation 8: In adult patients undergoing cardiovascular surgery, intravenous albumin is not suggested for priming the cardiovascular bypass circuit or volume replacement (Conditional Recommendation, Moderate Certainty of Evidence of Effect). Recommendation 9: In pediatric patients undergoing cardiovascular surgery, intravenous albumin is not suggested for priming the cardiovascular bypass circuit or volume replacement (Conditional Recommendation, Very Low Certainty of Evidence of Effect). A systematic review and meta-analysis of RCTs in pediatric and adult patients undergoing cardiovascular surgery was performed. We identified 43 randomized studies (N = 3,862), comparing albumin with gelatin, starches, or crystalloid solutions for priming the cardiopulmonary bypass circuit, volume expansion, or both. The majority of the trials were conducted in patients undergoing on-pump cardiac surgery, with the exception of two RCTs conducted in patients undergoing off-pump cardiac surgery. , Albumin infusion did not result in a lower mortality rate when compared with other fluids (risk difference, 0.00; 95% CI, –0.01 to 0.01; N = 2,711). No differences were found for the rates of kidney failure (risk difference, 0.01; 95% CI, –0.01 to 0.03; N = 1,703), blood loss (mean difference [MD], –0.04 L; 95% CI, –0.04 to 0.01 L), ICU length of stay (MD, –0.12 days; 95% CI, –0.31 to 0.06 days; N = 1,371), hospital length of stay (MD, 0.02 days; 95% CI, –0.95 to 1.00 days; N = 870), blood component use (MD, 0.03 L; 95% CI, –0.03 to 0.08 L; N = 1,547), or cardiac index (MD, 0.07 L/min/m 2 ; 95% CI, –0.10 to 0.25 L/min/m 2 ; N = 499). Fluid balance was lower with albumin compared with alternative solutions (MD, –0.55 L; 95% CI, –1.06 to –0.40 L; N = 450). The largest trial enrolled 1,386 patients and compared 4% albumin (20% albumin diluted in Ringer’s lactate) with Ringer’s lactate for both the pump prime and for fluid resuscitation ; albumin-treated patients showed higher rates of bleeding, resternotomy, and infection. Despite the common use of albumin during cardiovascular surgery, little evidence supports the use of albumin to improve patient outcomes. The largest study to date performed in 1,386 patients at a single center, Albumin in Cardiac Surgery, found increased morbidity when albumin was compared with Ringer’s lactate. Albumin in Cardiac Surgery was performed predominantly in low-risk cardiac surgery, and therefore, its role in improving outcomes in high-risk cardiac surgery has yet to be studied (a 590-patient RCT is underway, Albumin in Cardiac Surgery Australia; Identifier, ACTRN12619001355167). Recommendations Recommendation 10: In patients with cirrhosis and ascites undergoing large - volume paracentesis (> 5 L ), intravenous albumin is suggested to prevent paracentesis-induced circulatory dysfunction (Conditional Recommendation, Very Low Certainty of Evidence of Effect). Recommendation 11: In patients with cirrhosis and spontaneous bacterial peritonitis, intravenous albumin is suggested to reduce mortality (Conditional Recommendation, Low Certainty of Evidence of Effect). Recommendation 12: In patients with cirrhosis and extraperitoneal infections, intravenous albumin is not suggested to reduce mortality or kidney failure (Conditional Recommendation, Low Certainty of Evidence of Effect). Recommendation 13: In hospitalized patients with decompensated cirrhosis with hypoalbuminemia (< 30 g/L), repeated intravenous albumin to increase albumin levels to > 30 g/L is not suggested to reduce infection, kidney dysfunction , or death (Conditional Recommendation, Low Certainty of Evidence of Effect). Recommendation 14: In outpatients with cirrhosis and uncomplicated ascites despite diuretic therapy, intravenous albumin is not routinely suggested to reduce complications associated with cirrhosis (Conditional Recommendation, Low Certainty of Evidence of Effect). Evidence Summary We identified a 2019 Cochrane systematic review including 27 RCTs (N = 1,592) examining the use of any plasma volume expanders in patients with cirrhosis undergoing paracentesis. In general, enrolled patients were undergoing large-volume paracentesis (> 5 L), and the most commonly used albumin doses were either 6 to 8 g of albumin per 1 L of fluid removed or a standard dose of 20 to 40 g. Compared with no plasma expander, no statistically significant effect of using hyperoncotic (20%-25%) albumin on mortality (RR, 0.52; 95% CI, 0.06-4.83), kidney impairment (RR, 0.32; 95% CI, 0.02-5.88), or recurrence of ascites (RR, 1.3; 95% CI, 0.49-3.42) was found. Compared with hyperoncotic albumin, use of other fluids showed uncertain effects on mortality (RR, 1.03; 95% CI, 0.82-1.30), kidney impairment (RR, 1.17; 95% CI, 0.71-1.91), and recurrence of ascites (RR, 1.14; 95% CI, 0.96-1.36). Paracentesis-induced circulatory dysfunction was more frequent with nonalbumin plasma expanders (RR, 1.98; 95% CI, 1.31-2.99) compared with albumin. A 2020 systematic review focused on the impact of different therapies (albumin, other fluids, vasoactive drugs) on the rate of postparacentesis circulatory dysfunction and identified nine RCTs (N = 620). Albumin at a dose of 8 g/L was found to be superior to other volume expanders for the prevention of postparacentesis circulatory dysfunction (rise in plasma renin activity by ≥ 50% of baseline). Similar to the Cochrane review, uncertainty regarding the role of albumin as compared with alternative treatments was noted for the prevention of complications after paracentesis. RCTs comparing high-dose albumin (6-8 g/L of ascitic fluid removed) with low-dose albumin (2-4 g/L of ascitic fluid removed) found no difference in the rate of paracentesis associated circulatory dysfunction, although uncertainty exists regarding the risk to benefit profile of the two doses, given the small sample size (two studies [N = 120]; RR, 1.00; 95% CI, 0.22-4.49). Two systematic reviews (in 2013 and 2020) identified five open-label RCTs in patients with spontaneous bacterial peritonitis both using variable doses and duration of hyperoncotic albumin (eg, 0.5-1.0 g/kg every 3 days for a maximum of 21 days; 1.5 g/kg on day 1 and 1.0 g/kg on day 3). , Albumin reduced the rate of kidney impairment (OR, 0.21; 95% CI, 0.11-0.42) and mortality (OR, 0.34; 95% CI, 0.19-0.60). The largest randomized trial randomized 126 patients to albumin (plus antibiotics) or antibiotics alone (without an explicit fluid resuscitation for the control arm). Patients treated with albumin showed lower rates of kidney impairment (10% vs 33%; P = .002) and in-hospital mortality (10% vs 29%; P = .01). The second largest trial randomized 118 patients to albumin (plus antibiotics) or antibiotics alone (without an explicit fluid resuscitation for the control arm). The primary end point of in-hospital mortality was not different (13% albumin vs 11% antibiotics alone; P = .66). A 2020 systematic review and meta-analysis of RCTs comparing albumin plus antibiotics with antibiotics alone in patients with cirrhosis and extraperitoneal infections found no effect on mortality or kidney impairment, but observed higher rates of pulmonary edema with albumin (three studies [N = 406]; OR, 5.17; 95% CI, 1.62-16.47). A 2019 systematic review in the same population also found no improvements in outcomes when albumin with antibiotics was compared with antibiotics alone. Subsequent to this 2020 systematic review, two randomized trials have been published (308 and 100 patients, respectively) comparing albumin with crystalloid in patients with cirrhosis and hypotension resulting from sepsis. , Both trials included patients with sepsis from all causes, including a small proportion (20%-25%) with spontaneous bacterial peritonitis. In the larger trial, survival at 7 days was not improved in the albumin-treated patients (saline, 39.0% vs albumin, 43.5%; P = .42, Fisher exact test); longer-term outcomes were not reported. In the second, smaller trial, albumin was superior to crystalloid for reversal of hypotension without initiation of vasopressors at 3 h (22% vs 62%; P < .001), but this improvement in hemodynamics did not reduce the rate of dialysis, length of stay, or mortality at 28 days. In the latter trial, patients randomized to albumin vs crystalloid showed higher rates of circulatory overload. We identified one RCT, Albumin to Prevent Infection in Chronic Liver Failure (N = 777), that evaluated the role of hyperoncotic albumin to target an albumin level of > 30 g/L (median, 200 g albumin over 14 days) as compared with no albumin in hospitalized patients with decompensated cirrhosis and hypoalbuminemia (< 30 g/L). No difference was found in the primary end point (composite of new infections, kidney dysfunction, or death between days 3 and 15) between groups (OR, 0.98; 95% CI, 0.71-1.33). More severe or life-threatening serious adverse events were reported in the albumin-treated patients, primarily a numerical increase in pulmonary edema. A 2021 systematic review was identified that evaluated albumin in patients with hepatic encephalopathy. The review identified two RCTs (N = 176). Albumin resulted in a reduction in hepatic encephalopathy (RR, 0.60; 95% CI, 0.38-0.95) and mortality (RR, 0.54; 95% CI, 0.33-0.90). The first open-label trial randomized 120 patients to albumin (1.5 g/kg/d for up to 10 days and lactulose) vs lactulose alone. Complete resolution of hepatic encephalopathy by day 10 was seen in 75% of the albumin-lactulose group vs 53% of the lactulose alone group ( P = .03). Mortality was 18% in the albumin-lactulose group vs 32% in the lactulose alone group at day 10 ( P = .04). The second masked RCT of albumin (1.5 g/kg on day 1 and 1.0 g/kg on day 3) vs normal saline enrolled 56 patients. No difference was found in the rate of resolution of hepatic encephalopathy at day 4 (albumin, 58% vs saline, 53%; P = .7). The mortality rate was lower in albumin-treated patients at 90 days (23% vs 47%) and transplant-free survival was improved ( P = .02, Kaplan-Meier estimate). A 2021 systematic review of RCTs and cohort studies evaluating the role of albumin in prevention and treatment suggested that albumin may assist with the resolution or prevention of hepatic encephalopathy and may reduce mortality ; only the two RCTs identified in the aforementioned systematic review were identified for the treatment of hepatic encephalopathy. In the subsequent large Albumin to Prevent Infection in Chronic Liver Failure trial, the subgroup (N = 149) of patients admitted with hepatic encephalopathy randomized to albumin as compared with placebo did not show an improvement in the composite end point of new infections, kidney dysfunction, or death between days 3 and 15 (adjusted OR, 0.91; 95% CI, 0.44-1.86). Subsequent to the two systematic reviews, a single RCT was identified that randomized 48 outpatients with hepatic encephalopathy to weekly hyperoncotic albumin for 5 weeks as compared with saline and found improvements in cognitive function with albumin. A 2021 systematic review of RCTs evaluating outpatient hyperoncotic albumin for patients with cirrhosis and ascites identified five trials (N = 716). The systematic review found no difference in mortality at 12 to 36 months (RR, 0.88; 95% CI, 0.67-1.14) or any other outcome, with the exception of reducing the need for paracentesis (RR, 0.56; 95% CI, 0.48-0.67). Two large randomized trials were included in the review. , The first unmasked trial randomized 440 patients with cirrhosis and uncomplicated, persistent ascites despite diuretic therapy to albumin (40 g twice weekly for 2 weeks and then 40 g weekly for up to 18 months) or no albumin. Patients randomized to albumin experienced longer time to first paracentesis; required fewer paracenteses; were less likely to demonstrate hepatic encephalopathy, hepatorenal syndrome, spontaneous bacterial peritonitis, nonperitonitis infections, hyponatremia, or episodes of kidney dysfunction; experienced fewer days in hospital; and showed lower all-cause mortality (77% vs 66% survival at 18 months; hazard ratio, 0.62; 95% CI, 0.40-0.95). The most important limitation of this study is that the albumin-treated patients underwent weekly health care interactions and the control group did not, raising the concern that the observed differences may have been the result of increased health care exposure. The second trial randomized 196 outpatients with cirrhosis and ascites awaiting liver transplantation to oral midodrine and albumin as compared with placebo tablets and a 0.9% saline infusion and found no difference in patient outcomes. The dose of albumin given as part of the intervention was lower (40 g every 15 days). The study improved on the methodology of the first trial by achieving masking to treatment assignment and ensuring the same health care exposure in both study groups. Rationale for Recommendations Approximately one-third of albumin is used for patients with cirrhosis, and although this practice is exceedingly common, the certainty of evidence supporting this therapy in this population is insufficient to allow for strong recommendations. Although the use of albumin for large-volume paracentesis is a commonly accepted clinical practice and is endorsed by guidelines, , , the reported trials have important limitations that affect the certainty in outcomes. These trials included a small number of patients and findings for most patient-important outcomes (mortality, kidney dysfunction) were imprecise, leaving residual uncertainty regarding true clinical benefits and harms. Albumin, as compared with other fluid expanders, may be superior for the prevention of paracentesis-induced circulatory dysfunction (rise in serum renin level on the sixth day after paracentesis), but whether this translates to improvement in patient-important outcomes is less certain. Plasma renin levels are predictive of greater morbidity in patients with cirrhosis. , , The panel suggested continuing this commonly accepted practice for patients undergoing large-volume paracentesis, but believed the data supported only a conditional recommendation based on low-quality evidence. Further trials are needed urgently to clarify if albumin improves patient important outcomes, to elucidate the optimal dosing strategy, to further the understanding of the safety profile of the treatment, and to evaluate alternative fluids and therapies. It is unclear if improving laboratory measures of paracentesis-induced circulatory dysfunction will translate into reductions in renal failure, hospital admission, or other patient-important outcomes. The panel also highlighted the need to personalize the use of albumin, the dose after paracentesis, or both, considering the patient’s baseline creatinine, volume of ascites removed, and history of hypotensive symptoms after prior procedures. Similarly, the role of albumin for improving outcomes in patients with spontaneous bacterial peritonitis is unclear. The trial data specific to this patient population are limited. , The two largest RCTs failed to provide an explicit fluid resuscitation protocol for the patients randomized to no albumin, raising the concern for underresuscitation in the control arms of both studies. When similar albumin dosing strategies were used in trials examining patients with cirrhosis and extraperitoneal infections, no benefit was seen and concern for harm was expressed. The panel suggested the use of albumin for spontaneous bacterial peritonitis (conditional recommendation), but raised concerns regarding the dosing protocol used in two of the four trials and the risk of fluid overload (1.5 g/kg on day 1 and 1.0 g/kg on day 3) and the lack of data suggesting this specific regimen is beneficial compared with alternative dosing (eg, lower dose daily for 3 days). The panel also considered the lack of clarity on whether albumin is necessary for all patients with spontaneous bacterial peritonitis or whether it could be used selectively (ie, patients at high risk of kidney failure or death: serum bilirubin > 4 mg/dL or serum creatinine >1 mg/dL). Additional studies are necessary to address dosing, to address the benefit for patients with and without kidney impairment, and to clarify the risks of adverse events. The panel also noted that not all physicians currently adhere to the trial dosing strategy, , although it continues to be recommended in current guidelines. , A careful assessment of the patient’s volume status, cardiovascular status, and degree of kidney impairment before transfusion is advised and the dose, frequency, or both being modified accordingly. In contrast, the RCTs find no support for the use of albumin in patients with cirrhosis and extraperitoneal infections. In the setting of patients admitted with decompensated cirrhosis and hypoalbuminemia, this guideline is informed by an RCT involving 777 patients that found no improvement in patient important outcomes and a concern for increased adverse events. This led the panel to suggest conditionally against the use of albumin in this setting. Although a 2021 systematic review of two small RCTs suggested a benefit for facilitating resolution of hepatic encephalopathy and reducing mortality, the subgroup of patients in the Albumin to Prevent Infection in Chronic Liver Failure study admitted with hepatic encephalopathy did not show improvements in mortality. The panel had uncertainty regarding the benefit of albumin in this patient population and few data on the risks of the treatment, and therefore abstained from making a statement on the role of albumin in this setting until further adequately powered RCTs are conducted. In nonhospitalized patients with cirrhosis and persistent ascites despite optimized medical management, the role of weekly or biweekly albumin infusions remains unclear. One unmasked study of weekly albumin infusions found improvements in outcomes, but this was not replicated in a placebo-controlled trial that examined biweekly albumin infusions. The latter trial enrolled a smaller number of patients and used a lower dose, and therefore may have failed to detect a difference in outcomes. The panel reported residual uncertainty regarding the benefit of this treatment and given this, suggested against its routine use until additional RCTs have been conducted. The use of weekly albumin in this patient population would have considerable impacts on patients, would require chronic IV access, would have considerable impacts on outpatient infusion clinics, and would require a dependable supply of albumin. Although the unmasked trial reported cost-effectiveness, additional masked trials with cost-effectiveness analyses are necessary to improve precision and generalizability and to inform future guidelines. A 2020 and a 2019 systematic review on the treatment of hepatorenal syndrome did not identify any randomized trials examining albumin for these patients as compared with placebo or no treatment. Rather, all trials examining this patient population uniformly have administered albumin in both treatment and control arms and have compared vasoconstrictor agents (eg, terlipressin, midodrine) with placebo infusions. Hence, no recommendations regarding the use of albumin for patients with cirrhosis and hepatorenal syndrome could be made. Recommendation 10: In patients with cirrhosis and ascites undergoing large - volume paracentesis (> 5 L ), intravenous albumin is suggested to prevent paracentesis-induced circulatory dysfunction (Conditional Recommendation, Very Low Certainty of Evidence of Effect). Recommendation 11: In patients with cirrhosis and spontaneous bacterial peritonitis, intravenous albumin is suggested to reduce mortality (Conditional Recommendation, Low Certainty of Evidence of Effect). Recommendation 12: In patients with cirrhosis and extraperitoneal infections, intravenous albumin is not suggested to reduce mortality or kidney failure (Conditional Recommendation, Low Certainty of Evidence of Effect). Recommendation 13: In hospitalized patients with decompensated cirrhosis with hypoalbuminemia (< 30 g/L), repeated intravenous albumin to increase albumin levels to > 30 g/L is not suggested to reduce infection, kidney dysfunction , or death (Conditional Recommendation, Low Certainty of Evidence of Effect). Recommendation 14: In outpatients with cirrhosis and uncomplicated ascites despite diuretic therapy, intravenous albumin is not routinely suggested to reduce complications associated with cirrhosis (Conditional Recommendation, Low Certainty of Evidence of Effect). We identified a 2019 Cochrane systematic review including 27 RCTs (N = 1,592) examining the use of any plasma volume expanders in patients with cirrhosis undergoing paracentesis. In general, enrolled patients were undergoing large-volume paracentesis (> 5 L), and the most commonly used albumin doses were either 6 to 8 g of albumin per 1 L of fluid removed or a standard dose of 20 to 40 g. Compared with no plasma expander, no statistically significant effect of using hyperoncotic (20%-25%) albumin on mortality (RR, 0.52; 95% CI, 0.06-4.83), kidney impairment (RR, 0.32; 95% CI, 0.02-5.88), or recurrence of ascites (RR, 1.3; 95% CI, 0.49-3.42) was found. Compared with hyperoncotic albumin, use of other fluids showed uncertain effects on mortality (RR, 1.03; 95% CI, 0.82-1.30), kidney impairment (RR, 1.17; 95% CI, 0.71-1.91), and recurrence of ascites (RR, 1.14; 95% CI, 0.96-1.36). Paracentesis-induced circulatory dysfunction was more frequent with nonalbumin plasma expanders (RR, 1.98; 95% CI, 1.31-2.99) compared with albumin. A 2020 systematic review focused on the impact of different therapies (albumin, other fluids, vasoactive drugs) on the rate of postparacentesis circulatory dysfunction and identified nine RCTs (N = 620). Albumin at a dose of 8 g/L was found to be superior to other volume expanders for the prevention of postparacentesis circulatory dysfunction (rise in plasma renin activity by ≥ 50% of baseline). Similar to the Cochrane review, uncertainty regarding the role of albumin as compared with alternative treatments was noted for the prevention of complications after paracentesis. RCTs comparing high-dose albumin (6-8 g/L of ascitic fluid removed) with low-dose albumin (2-4 g/L of ascitic fluid removed) found no difference in the rate of paracentesis associated circulatory dysfunction, although uncertainty exists regarding the risk to benefit profile of the two doses, given the small sample size (two studies [N = 120]; RR, 1.00; 95% CI, 0.22-4.49). Two systematic reviews (in 2013 and 2020) identified five open-label RCTs in patients with spontaneous bacterial peritonitis both using variable doses and duration of hyperoncotic albumin (eg, 0.5-1.0 g/kg every 3 days for a maximum of 21 days; 1.5 g/kg on day 1 and 1.0 g/kg on day 3). , Albumin reduced the rate of kidney impairment (OR, 0.21; 95% CI, 0.11-0.42) and mortality (OR, 0.34; 95% CI, 0.19-0.60). The largest randomized trial randomized 126 patients to albumin (plus antibiotics) or antibiotics alone (without an explicit fluid resuscitation for the control arm). Patients treated with albumin showed lower rates of kidney impairment (10% vs 33%; P = .002) and in-hospital mortality (10% vs 29%; P = .01). The second largest trial randomized 118 patients to albumin (plus antibiotics) or antibiotics alone (without an explicit fluid resuscitation for the control arm). The primary end point of in-hospital mortality was not different (13% albumin vs 11% antibiotics alone; P = .66). A 2020 systematic review and meta-analysis of RCTs comparing albumin plus antibiotics with antibiotics alone in patients with cirrhosis and extraperitoneal infections found no effect on mortality or kidney impairment, but observed higher rates of pulmonary edema with albumin (three studies [N = 406]; OR, 5.17; 95% CI, 1.62-16.47). A 2019 systematic review in the same population also found no improvements in outcomes when albumin with antibiotics was compared with antibiotics alone. Subsequent to this 2020 systematic review, two randomized trials have been published (308 and 100 patients, respectively) comparing albumin with crystalloid in patients with cirrhosis and hypotension resulting from sepsis. , Both trials included patients with sepsis from all causes, including a small proportion (20%-25%) with spontaneous bacterial peritonitis. In the larger trial, survival at 7 days was not improved in the albumin-treated patients (saline, 39.0% vs albumin, 43.5%; P = .42, Fisher exact test); longer-term outcomes were not reported. In the second, smaller trial, albumin was superior to crystalloid for reversal of hypotension without initiation of vasopressors at 3 h (22% vs 62%; P < .001), but this improvement in hemodynamics did not reduce the rate of dialysis, length of stay, or mortality at 28 days. In the latter trial, patients randomized to albumin vs crystalloid showed higher rates of circulatory overload. We identified one RCT, Albumin to Prevent Infection in Chronic Liver Failure (N = 777), that evaluated the role of hyperoncotic albumin to target an albumin level of > 30 g/L (median, 200 g albumin over 14 days) as compared with no albumin in hospitalized patients with decompensated cirrhosis and hypoalbuminemia (< 30 g/L). No difference was found in the primary end point (composite of new infections, kidney dysfunction, or death between days 3 and 15) between groups (OR, 0.98; 95% CI, 0.71-1.33). More severe or life-threatening serious adverse events were reported in the albumin-treated patients, primarily a numerical increase in pulmonary edema. A 2021 systematic review was identified that evaluated albumin in patients with hepatic encephalopathy. The review identified two RCTs (N = 176). Albumin resulted in a reduction in hepatic encephalopathy (RR, 0.60; 95% CI, 0.38-0.95) and mortality (RR, 0.54; 95% CI, 0.33-0.90). The first open-label trial randomized 120 patients to albumin (1.5 g/kg/d for up to 10 days and lactulose) vs lactulose alone. Complete resolution of hepatic encephalopathy by day 10 was seen in 75% of the albumin-lactulose group vs 53% of the lactulose alone group ( P = .03). Mortality was 18% in the albumin-lactulose group vs 32% in the lactulose alone group at day 10 ( P = .04). The second masked RCT of albumin (1.5 g/kg on day 1 and 1.0 g/kg on day 3) vs normal saline enrolled 56 patients. No difference was found in the rate of resolution of hepatic encephalopathy at day 4 (albumin, 58% vs saline, 53%; P = .7). The mortality rate was lower in albumin-treated patients at 90 days (23% vs 47%) and transplant-free survival was improved ( P = .02, Kaplan-Meier estimate). A 2021 systematic review of RCTs and cohort studies evaluating the role of albumin in prevention and treatment suggested that albumin may assist with the resolution or prevention of hepatic encephalopathy and may reduce mortality ; only the two RCTs identified in the aforementioned systematic review were identified for the treatment of hepatic encephalopathy. In the subsequent large Albumin to Prevent Infection in Chronic Liver Failure trial, the subgroup (N = 149) of patients admitted with hepatic encephalopathy randomized to albumin as compared with placebo did not show an improvement in the composite end point of new infections, kidney dysfunction, or death between days 3 and 15 (adjusted OR, 0.91; 95% CI, 0.44-1.86). Subsequent to the two systematic reviews, a single RCT was identified that randomized 48 outpatients with hepatic encephalopathy to weekly hyperoncotic albumin for 5 weeks as compared with saline and found improvements in cognitive function with albumin. A 2021 systematic review of RCTs evaluating outpatient hyperoncotic albumin for patients with cirrhosis and ascites identified five trials (N = 716). The systematic review found no difference in mortality at 12 to 36 months (RR, 0.88; 95% CI, 0.67-1.14) or any other outcome, with the exception of reducing the need for paracentesis (RR, 0.56; 95% CI, 0.48-0.67). Two large randomized trials were included in the review. , The first unmasked trial randomized 440 patients with cirrhosis and uncomplicated, persistent ascites despite diuretic therapy to albumin (40 g twice weekly for 2 weeks and then 40 g weekly for up to 18 months) or no albumin. Patients randomized to albumin experienced longer time to first paracentesis; required fewer paracenteses; were less likely to demonstrate hepatic encephalopathy, hepatorenal syndrome, spontaneous bacterial peritonitis, nonperitonitis infections, hyponatremia, or episodes of kidney dysfunction; experienced fewer days in hospital; and showed lower all-cause mortality (77% vs 66% survival at 18 months; hazard ratio, 0.62; 95% CI, 0.40-0.95). The most important limitation of this study is that the albumin-treated patients underwent weekly health care interactions and the control group did not, raising the concern that the observed differences may have been the result of increased health care exposure. The second trial randomized 196 outpatients with cirrhosis and ascites awaiting liver transplantation to oral midodrine and albumin as compared with placebo tablets and a 0.9% saline infusion and found no difference in patient outcomes. The dose of albumin given as part of the intervention was lower (40 g every 15 days). The study improved on the methodology of the first trial by achieving masking to treatment assignment and ensuring the same health care exposure in both study groups. Approximately one-third of albumin is used for patients with cirrhosis, and although this practice is exceedingly common, the certainty of evidence supporting this therapy in this population is insufficient to allow for strong recommendations. Although the use of albumin for large-volume paracentesis is a commonly accepted clinical practice and is endorsed by guidelines, , , the reported trials have important limitations that affect the certainty in outcomes. These trials included a small number of patients and findings for most patient-important outcomes (mortality, kidney dysfunction) were imprecise, leaving residual uncertainty regarding true clinical benefits and harms. Albumin, as compared with other fluid expanders, may be superior for the prevention of paracentesis-induced circulatory dysfunction (rise in serum renin level on the sixth day after paracentesis), but whether this translates to improvement in patient-important outcomes is less certain. Plasma renin levels are predictive of greater morbidity in patients with cirrhosis. , , The panel suggested continuing this commonly accepted practice for patients undergoing large-volume paracentesis, but believed the data supported only a conditional recommendation based on low-quality evidence. Further trials are needed urgently to clarify if albumin improves patient important outcomes, to elucidate the optimal dosing strategy, to further the understanding of the safety profile of the treatment, and to evaluate alternative fluids and therapies. It is unclear if improving laboratory measures of paracentesis-induced circulatory dysfunction will translate into reductions in renal failure, hospital admission, or other patient-important outcomes. The panel also highlighted the need to personalize the use of albumin, the dose after paracentesis, or both, considering the patient’s baseline creatinine, volume of ascites removed, and history of hypotensive symptoms after prior procedures. Similarly, the role of albumin for improving outcomes in patients with spontaneous bacterial peritonitis is unclear. The trial data specific to this patient population are limited. , The two largest RCTs failed to provide an explicit fluid resuscitation protocol for the patients randomized to no albumin, raising the concern for underresuscitation in the control arms of both studies. When similar albumin dosing strategies were used in trials examining patients with cirrhosis and extraperitoneal infections, no benefit was seen and concern for harm was expressed. The panel suggested the use of albumin for spontaneous bacterial peritonitis (conditional recommendation), but raised concerns regarding the dosing protocol used in two of the four trials and the risk of fluid overload (1.5 g/kg on day 1 and 1.0 g/kg on day 3) and the lack of data suggesting this specific regimen is beneficial compared with alternative dosing (eg, lower dose daily for 3 days). The panel also considered the lack of clarity on whether albumin is necessary for all patients with spontaneous bacterial peritonitis or whether it could be used selectively (ie, patients at high risk of kidney failure or death: serum bilirubin > 4 mg/dL or serum creatinine >1 mg/dL). Additional studies are necessary to address dosing, to address the benefit for patients with and without kidney impairment, and to clarify the risks of adverse events. The panel also noted that not all physicians currently adhere to the trial dosing strategy, , although it continues to be recommended in current guidelines. , A careful assessment of the patient’s volume status, cardiovascular status, and degree of kidney impairment before transfusion is advised and the dose, frequency, or both being modified accordingly. In contrast, the RCTs find no support for the use of albumin in patients with cirrhosis and extraperitoneal infections. In the setting of patients admitted with decompensated cirrhosis and hypoalbuminemia, this guideline is informed by an RCT involving 777 patients that found no improvement in patient important outcomes and a concern for increased adverse events. This led the panel to suggest conditionally against the use of albumin in this setting. Although a 2021 systematic review of two small RCTs suggested a benefit for facilitating resolution of hepatic encephalopathy and reducing mortality, the subgroup of patients in the Albumin to Prevent Infection in Chronic Liver Failure study admitted with hepatic encephalopathy did not show improvements in mortality. The panel had uncertainty regarding the benefit of albumin in this patient population and few data on the risks of the treatment, and therefore abstained from making a statement on the role of albumin in this setting until further adequately powered RCTs are conducted. In nonhospitalized patients with cirrhosis and persistent ascites despite optimized medical management, the role of weekly or biweekly albumin infusions remains unclear. One unmasked study of weekly albumin infusions found improvements in outcomes, but this was not replicated in a placebo-controlled trial that examined biweekly albumin infusions. The latter trial enrolled a smaller number of patients and used a lower dose, and therefore may have failed to detect a difference in outcomes. The panel reported residual uncertainty regarding the benefit of this treatment and given this, suggested against its routine use until additional RCTs have been conducted. The use of weekly albumin in this patient population would have considerable impacts on patients, would require chronic IV access, would have considerable impacts on outpatient infusion clinics, and would require a dependable supply of albumin. Although the unmasked trial reported cost-effectiveness, additional masked trials with cost-effectiveness analyses are necessary to improve precision and generalizability and to inform future guidelines. A 2020 and a 2019 systematic review on the treatment of hepatorenal syndrome did not identify any randomized trials examining albumin for these patients as compared with placebo or no treatment. Rather, all trials examining this patient population uniformly have administered albumin in both treatment and control arms and have compared vasoconstrictor agents (eg, terlipressin, midodrine) with placebo infusions. Hence, no recommendations regarding the use of albumin for patients with cirrhosis and hepatorenal syndrome could be made. The evidence-base guiding the use of intravenous albumin was largely instigated by the Cochrane Injuries Group Albumin systematic review in 1998, which raised the concern for harm from albumin. Subsequent to this publication, RCTs comparing albumin with other fluid treatments in multiple patient populations were completed. These trials failed to confirm the concerns for higher mortality rates in albumin-treated patients. The ICTMG undertook these evidence-based albumin guidelines because no comprehensive evidence-based guidelines had been published yet. The goal of the guidelines is to provide clinicians with recommendations and evidence summaries for common indications for albumin, information on ongoing clinical trials, and areas in need of additional research. The ICTMG guidelines group suggested that albumin should not be used routinely for neonatal, pediatric, and adult patients in critical care; for patients experiencing intradialytic hypotension; for patients undergoing cardiovascular surgery; for admitted patients with cirrhosis for treatment (or correction) of hypoalbuminemia or extraperitoneal infections; or for outpatients with ascites. The ICTMG guidelines conditionally recommended the use of albumin for patients with cirrhosis undergoing large-volume paracentesis or with spontaneous bacterial peritonitis. One of 14 recommendations was a strong recommendation based on more definitive clinical trial evidence, but most of the recommendations were conditional based on low- or very low-quality evidence because of the paucity or conflicting RCT evidence, highlighting the need for ongoing research. The implementation of the guidelines will help to reduce the unnecessary transfusion of albumin and the variability between hospitals. Guidelines for select patient populations have been published in some jurisdictions, particularly in patients with cirrhosis. The British Society for Gastroenterology published guidelines on the management of patients with cirrhosis and ascites. They recommend albumin for patients undergoing large-volume paracentesis or with spontaneous bacterial peritonitis. The French Society of Anesthesiology and Critical Care Medicine and the French Association for the Study of the Liver jointly released guidelines for the management of liver failure in critical care. They recommend the use of albumin for hepatorenal syndrome (with terlipressin), large-volume paracentesis (> 5 L), and spontaneous bacterial peritonitis. The American Association for the Study of Liver Disease guidelines from 2021 recommend the use of albumin for large-volume paracentesis, severe muscle cramps, severe hyponatremia (sodium < 120 mEq/L), spontaneous bacterial peritonitis, and hepatorenal syndrome. The Italian Association for the Study of Liver Disease and the Italian Society for Transfusion Medicine and Immunohematology guidelines update from 2020 include the use of albumin for ascites requiring moderate doses of diuretics as an outpatient treatment. This was an update from their 2016 guidelines that also recommended the use of albumin in patients requiring large-volume paracentesis, with spontaneous bacterial peritonitis, or with hepatorenal syndrome. Similarly, the European Association for the Study of the Liver 2018 guidelines detailing the management of patients with decompensated cirrhosis recommended albumin for patients undergoing large-volume paracentesis, with spontaneous bacterial peritonitis, with acute kidney injury without known cause, or with hepatorenal syndrome. The ICTMG guidelines are concordant with these guidelines for recommending albumin for large-volume paracentesis and spontaneous bacterial peritonitis, but report insufficient evidence to support its use in other settings. The use of albumin for hepatorenal syndrome, in conjunction with terlipressin, was recommended commonly in prior guidelines, likely based on both expert opinion and the fact that randomized trials used albumin in both treatment arms (albumin vs albumin plus terlipressin). We elected to refrain from making a recommendation without clinical trial evidence to support its use and highlight that this indication needs further study. Guidelines from the Association of the Scientific Medical Societies in Germany published perioperative fluid guidelines for children in 2017. They recommended that colloids, including albumin, be used during surgery where crystalloids alone are not sufficiently effective and blood products are not indicated. In 2021, the Surviving Sepsis Campaign guidelines recommended the use of albumin in the fluid resuscitation of severe sepsis and septic shock when patients required large volumes of crystalloids. Five RCTs that will enroll an additional 4,864 patients are ongoing and are expected to provide additional clarity on the role of albumin . These trials will add clarity to the ICTMG recommendations for intensive care patients with infection, high-risk adult cardiac surgery, patients with acute kidney injury receiving kidney replacement therapy, and outpatients with decompensated cirrhosis. This guideline is limited by the uncertainty in the evidence identified in the literature search for many different patient populations and the limitation of the search to the English language. The lack of comparative dosing strategies leaves uncertainty on the choice between 4% to 5% and 20% to 25% albumin formulations, the dose for each indication, the risk of fluid overload, and the dosing schedules. The guidelines are limited to common uses of albumin and cannot address every possible patient scenario where albumin has been used in RCTs. The published studies often did not collect or did report adverse reactions from IV albumin, or both, limiting the conclusions regarding the potential risks of albumin. These guidelines improve on those previously published because of the rigorous methodology, broad scope of the recommendations, inclusion of a patient representative in the guideline process, and broad community consultation process. The guidelines will be supported by tools developed by the ICTMG Dissemination and Implementation Committee to assist hospitals with aligning practice with the evidence. Future research is needed in multiple clinical settings including: (1) the role and timing of albumin in patients with sepsis or other conditions with insufficient response to crystalloids, (2) the role of albumin in patients undergoing surgery, (3) the role of albumin for intradialytic hypotension, and (4) the role of albumin in all indications for patients with cirrhosis. Research also is needed to understand therapeutic targets of albumin resuscitation (hemodynamic, urinary output, laboratory), the optimal formulation, and the dosing strategy. The risk of IV albumin infusions needs further investigation to allow clinicians to weigh the risk to benefit profile appropriately. Studies should include patient-important outcomes, rather than focusing on short-term physiologic outcomes. Funded by the Ontario Regional Blood Coordinating Network and the International Collaboration for Transfusion Medicine Guidelines . The ICTMG receives funding from Canadian Blood Services (funded by the federal government [Health Canada] and the provincial and territorial ministries of health). The authors have reported to CHEST the following: J. C. receives research support from Canadian Blood Services and Octapharma and serves on the board of directors of the Canadian Hematology Society. N. J. S. is a director of the National Board of Echocardiography and receives royalties from Wolters Kluwer. A. B. is an employee of Canadian Blood Services. H. K. is an employee of Canadian Blood Services. E. G. C. receives research funding (related to albumin) from Department of Medicine, The Ottawa Hospital and University of Ottawa, The Ottawa Hospital Academic Medical Organization, Kidney Foundation of Canada, and Physician Services Incorporated Foundation; is an editorial board member of the Canadian Journal of Kidney Health and Disease ; and is a member of the Contrast-Associated Acute Kidney Injury guideline panel for the Canadian Association of Radiologists. B. R. is a guideline methodologist for American Thoracic Society, the Society of Critical Care Medicine, and Canadian Blood Services; is the Knowledge Translation director for Canadian Critical Care Society; is the grants and manuscripts chair for Canadian Critical Care Trials Group, and in a guideline group member for multiple guidelines. S. R. B. is the chair of the Clinical Pharmacy and Pharmacology section for the Society of Critical Care Medicine (not albumin use related), is a paid consultant for Wolters Kluwer (Lexicomp), is a Society of Critical Care Medicine Social Media Committee member, is a Surviving Sepsis Campaign Research Committee member, and has received a research grant from the National Institute of General Medicine Sciences. L. C. is a guideline group member for the British Society of Gastroenterology (management of ascites in liver cirrhosis), is involved in peer-reviewed publications (multiple topics including relevant to albumin use), received lecturer honoraria for the Canadian Liver Conference 2022, is a hepatology consultant for the Royal Free Hospital London, and is a Liver Committee member of the British Society of Gastroenterology. M. F. receives consultant fees from Cerus Corporation and Biocogniv, Inc.; has received honoraria from Grifols (none were albumin related); is a board member for Project Santa Fe Foundation and the American Board of Pathology; is the Histocompatibility and Identity Testing Committee Chair for College of American Pathologists; is co-team leader for the Biomedical Excellence for Safer Transfusion (BEST)Collaborative; and is the Editorial Committee co-chair for the ICTMG. R. J. has received fellowship funding from Canadian Blood Services, is an employee of William Osler Health System and the University of Cincinnati Medical Center, and is a panel member for ICTMG Platelet Utilization guideline development group. K. P. serves on the board of directors in North America for International Society for Blood Transfusion (ISBT), is a 2023 Association for the Advancement of Blood and Biotherapies (AABB) Red Blood Cells (RBC) guideline panel member, and is a member of the National Advisory Committee of Blood and Blood Products. P. S. S. is director of the Canadian Neonatal Network, director of the Canadian Preterm Birth Network, director of the International Network to Evaluate Outcomes of Neonates, and an external advisory board member for the Canadian Perinatal Surveillance system (none related to albumin manufacturers). H. S. is a consultant for Terumo and Cerus (not albumin related). Z. M. S. is a consultant and advisory board member of Grifols, Fresenius Kabi, and Novartis; receives research funding from Erydel and Fresenius Kabi; serves on the board of directors for the BEST Collaborative and International Council for Commonality in Blood Banking Automation (ICCBBA), Inc.; is the AABB Committee Chair; is vice chair, treasurer, and committee chair for ICCBBA, Inc.; is treasurer for BEST Collaborative; and has a family member (child) who is a summer intern with Grifols, Inc. T. T. is a paid consultant for Inter-View Partners France, A+A, Bayer HealthCare SAS, BVA, Axess Research, and All Global; has received honoraria from AbbeVie, Gilead Sciences, Advanz Pharma France, and Ipsen Pharma; in the principal investigator of randomized controlled trial Albumin Administration in Cirrhotic Patients With Bacterial Infection and a Systemic Inflammatory Response Syndrome Unrelated to Spontaneous Bacterial Peritonitis (ALB-CIRINF) (ClinicalTrials.gov Identifier, NCT01359813) published in 2015; and is a member of the Liver Cirrhosis-related Complications (LCC)-International Special Interest Group. B. W. is a resident physician at Loma Linda University Medical Center. S. S. is chair of the ICTMG and is an employee of National Health Service Blood and Transplant (NHSBT), a blood service operator in England. However, NHSBT is not a manufacturer of the intervention. N. S. is an employee of Canadian Blood Services; receives research funding from the Canadian Institutes for Health Research (Transfusion Requirements in Younger Patients Undergoing Cardiac Surgery [TRICS-IV] RBC transfusion in young cardiac patients; not related to albumin); is an advisory board member for Fresenius Kabius and Janssen; has received honoraria from the International Financial Corporation of the World Bank, Canadian Blood Services, and Ferring; and serves on the PKD guideline panel and ICTMG guideline panels (Fetal Neonatal Alloimmune Thrombocytopenia [FNAIT], Hemolytic Disease of the Newborn [HDN], platelet transfusion, RBC specifications). None declared (D. F., S. A., M. N., C. P., S. R.). See for the ICTMG Conflict of Interest Policy. |
Clinical Characteristics and Pathological Features of “Crawling-type” Early Gastric Carcinoma: A Retrospective Series of Eight Cases | 5a0b17df-798c-4e9f-8d63-bdfc0b662782 | 11826854 | Surgical Procedures, Operative[mh] | Gastric cancer (GC) is a malignant neoplasm originating from the epithelium of the gastric mucosa. Although its incidence has declined in recent years, it remains one of the most common malignant tumors globally. In China, GC was the third most important cancer in terms of incidence and mortality in 2022. GC is a heterogeneous disease, and clinicians may occasionally overlook rare variants of GC due to their rarity. “Crawling-type” early gastric carcinoma (EGC) is one such rare variant of GC and is characterized by irregular glandular fusion, low-grade cellular atypia, and a tendency for lateral diffusion within the mucosa. , The microscopic examination of crawling-type EGC typically reveals intestinal metaplasia cells, irregular glandular structures, occasional signet ring cells, “handshake structures,” or “Whyx-type” structures. , Despite minimal atypia and extension into the area of epithelial hyperplasia, the mucosal surface of crawling-type EGC generally appears normal, posing a diagnostic challenge, particularly in small, limited biopsy specimens. Cytological features lacking distinct characteristics are often misdiagnosed as ambiguous tumor formation or reactive intestinal metaplasia, leading to an underestimation of their malignant potential. , Moreover, the boundaries of crawling-type EGC within the mucosa are frequently indistinct due to the lack of comparison with surrounding non-neoplastic mucosa, making complete removal through endoscopic submucosal dissection (ESD) difficult. Studies on the clinicopathological and molecular characteristics of crawling-type EGC are limited, complicating the accurate diagnosis of this condition. - In addition, there has been insufficient focus on the prognosis after surgical intervention. Therefore, this study aimed to report the clinical characteristics, diagnostic methods, and short-term prognosis of patients with crawling-type EGC after ESD or gastrectomy. This case series study retrospectively included consecutive patients diagnosed with crawling-type EGC who underwent ESD or gastrectomy at the East Hospital Affiliated to Tongji University between January 2019 and March 2022. The inclusion criteria were (1) patients diagnosed with crawling-type EGC for the first time and (2) who underwent ESD or gastrectomy. The exclusion criteria were (1) received cancer treatments before surgery for crawling-type EGC or (2) incomplete clinical data. This study was approved by the Ethics Committee of the East Hospital Affiliated to Tongji University (2024YS-270), and the requirement for individual informed consent was exempted due to the retrospective nature of the study. The reporting of this study conforms to STROBE guidelines. Data Collection Data were collected from the original patient records. All surgical specimens were examined by experienced pathologists. Variables such as preoperative and postoperative pathological results, postoperative lesion/tumor size and shape, depth of invasion, atypia, 52-week prognosis, TNM staging, and type of surgery were collected. All patients were routinely followed up at 6 and 12 months after surgery using gastroscopy to observe eventual changes in gastric and scar mucosa and using computed tomography (CT) to observe lymph nodes and nearby organs. All patient details were de-identified. Statistical Analysis Only descriptive analysis was performed. Categorical data were expressed as n . Data were collected from the original patient records. All surgical specimens were examined by experienced pathologists. Variables such as preoperative and postoperative pathological results, postoperative lesion/tumor size and shape, depth of invasion, atypia, 52-week prognosis, TNM staging, and type of surgery were collected. All patients were routinely followed up at 6 and 12 months after surgery using gastroscopy to observe eventual changes in gastric and scar mucosa and using computed tomography (CT) to observe lymph nodes and nearby organs. All patient details were de-identified. Only descriptive analysis was performed. Categorical data were expressed as n . Eight patients (5 males and 3 females) were included in this study. The mean age was 63.5 ± 7.8 years. The tumors were primarily localized in the gastric cardia and epigastric fundus gland regions in 4 patients , while the remaining 4 patients had tumors in the gastric antrum. Preoperative pathological biopsies revealed one patient with crawling-type carcinoma, 3 with high-grade intraepithelial neoplasia, and 4 with high-grade intraepithelial tumor, with only one patient showing submucosal invasion. Four patients underwent ESD, and 4 underwent partial gastrectomy. Postoperative pathology classified 7 patients as gastrointestinal mixed type, with only one case being pure intestinal type. The mucinous subtype was observed in 7 patients, and signet ring cells were noted in 5 patients. Regarding gross morphology, 5 tumors were flat, 2 were concave, and one was convex. Ulcers were found in 2 patients . Curative resection was achieved in all patients. Preoperative gastroscopy revealed atrophic gastritis and intestinal metaplasia in all 8 patients. At 6 and 12 months postoperatively, no recurrence events were observed in any patient during gastroscopy and CT examinations . Typical Cases Patient 4: A 64-year-old man presented with abdominal distension and belching for more than 1 month. Gastroscopic examination revealed a light red lesion classified as 0-IIb + IIc under white light and Indigo blush dyeing endoscopy . Magnifying narrow-band imaging showed irregular microsurface and microvascular patterns . The preoperative pathological examination suggested chronic inflammation of gastric mucosa in the minor curvature of the gastric antrum and high-grade intraepithelial neoplasia of focal glands. The patient underwent ESD . The postoperative pathological examination revealed highly to moderately differentiated tubular adenocarcinoma/tub2 > tub1 (in part, crawling type morphological changes as the major differentiation). The mucinous type was classified as the gastrointestinal mixed type, and the visual classification was superficial uplift and concave type (0-IIb + IIc). The tumor size was approximately 20 mm × 11 mm under the microscope. The carcinoma tissue infiltrated into mucous lamina propria (LPM/pT1a). There was no ulcer formation (UL0). The postoperative diagnosis was crawling-type EGC . There was no recurrence after 1 year of follow-up. Patient 4: A 64-year-old man presented with abdominal distension and belching for more than 1 month. Gastroscopic examination revealed a light red lesion classified as 0-IIb + IIc under white light and Indigo blush dyeing endoscopy . Magnifying narrow-band imaging showed irregular microsurface and microvascular patterns . The preoperative pathological examination suggested chronic inflammation of gastric mucosa in the minor curvature of the gastric antrum and high-grade intraepithelial neoplasia of focal glands. The patient underwent ESD . The postoperative pathological examination revealed highly to moderately differentiated tubular adenocarcinoma/tub2 > tub1 (in part, crawling type morphological changes as the major differentiation). The mucinous type was classified as the gastrointestinal mixed type, and the visual classification was superficial uplift and concave type (0-IIb + IIc). The tumor size was approximately 20 mm × 11 mm under the microscope. The carcinoma tissue infiltrated into mucous lamina propria (LPM/pT1a). There was no ulcer formation (UL0). The postoperative diagnosis was crawling-type EGC . There was no recurrence after 1 year of follow-up. This study suggests that crawling-type EGC may exhibit distinct clinical characteristics and pathological features compared with classical GC. The rate of curative resection post-surgery appears satisfactory, and the short-term prognosis following surgical treatment may be favorable. These findings may provide valuable insights for diagnosing and treating this GC subtype. In this case series, the mean age of the patients was 63.5 ± 13.45, slightly higher than previously reported by Woo et al. The present study found that crawling-type EGC was mainly located in the upper part of the stomach, from the cardia to the fundus gland region and the gastric antrum. This finding differs from Haruta et al earlier, who demonstrated that the middle third of the stomach was the preferential site for crawling-type EGC. The EGCs observed in the present study were mainly characterized by a flat morphology, whereas previous reports showed that more than 70% of crawling-type EGCs exhibited a depressed structure. , , According to the Lauren classification, most of these lesions were of the gastrointestinal mixed type. This result aligns with the classification of crawling-type EGC as a very well-differentiated GC of the intestinal type. , , Notably, most of the lesions were limited to the mucosa, with only one tumor penetrating the submucosal layer, a finding supported by previous research. As the literature describes, , - crawling-type EGC is characterized by unique histological features. In agreement with these reports, , - the cases reported here displayed heterotypic structures, crawling-type cells, and irregular glandular formations. These features are the most significant pathological characteristics of crawling-type EGC. , - Among the 5 tumors containing signet ring cells, 4 had a small amount of signet ring cell carcinoma, and one case had a large area of signet ring cell carcinoma. Nevertheless, postoperative pathology still suggested curative resection. It is worth considering whether there is an association with older patient age, as most previous reports of signet ring cell carcinoma were in younger patients with poorer prognosis. This point warrants further exploration. The finding indicates that the presence of signet ring cells could be a helpful indicator for diagnosing crawling-type EGC, but it will have to be validated in future studies. In addition, the present case series showed large tumors, with 6 being larger than 2 cm. The borders of the lesions were frequently poorly defined due to a lack of contrast with the surrounding non-neoplastic mucosa. These features may pose difficulties for the complete resection of lesions by ESD, as the absolute indications for ESD include highly or moderately differentiated intramucosal adenocarcinoma less than 2.0 cm in diameter without ulcerative changes. Fortunately, no recurrence events were found in any patient. In this case series, 4 patients underwent ESD, and 4 underwent partial gastrectomy. Although the long-term outcomes remain to be determined, treatment outcomes for all patients were optimal after 12 months of follow-up. ESD for crawling-type EGC can be difficult because of the similar appearance of the lesion to the adjacent tissue, but it can still be possible in selected cases. Nevertheless, according to the “ Guidelines for endoscopic submucosal dissection and endoscopic mucosal resection for early gastric cancer ,” ESD treatment for crawling-type EGC should preferably meet the following indications: (1) UL0 cT1a differentiated-type carcinomas with a long diameter greater than 2 cm; (2) UL1 cT1a differentiated-type carcinomas with a long diameter measuring 3 cm or less; (3) UL0 cT1a undifferentiated-type carcinomas with a long diameter of 2 cm or less. Of course, the resection margins must be carefully evaluated, and the patient and surgeons must be prepared for an eventual gastrectomy if ESD has positive margins. The present study had several limitations. Firstly, the sample size was small, precluding the observation of certain features. Secondly, the postoperative follow-up was short, impeding an assessment of the long-term efficacy of surgical resection. In addition, the study did not investigate the expression profiles of crawling-type EGC, a factor that could offer supplementary information for diagnosing this condition. Crawling-type EGC may exhibit distinct clinical characteristics and pathological features from classical GC. Curative resection was achieved in all patients, and the short-term prognosis of surgical treatment may be favorable. Preoperative gastroscopy may potentially misdiagnose crawling-type EGC. The exact prognosis of crawling-type EGC remains unknown, and the selection of adjuvant treatments should be made based on the final pathological results. |
Are palpation-guided interventional procedures on the adductor longus muscle safe? A cadaveric and sonographic investigation | 0a476a6b-3405-4f38-ba38-194e4e4bce65 | 11805763 | Surgical Procedures, Operative[mh] | The adductor longus muscle is one of the most injured muscles in the hip and therefore is the target of different interventional musculoskeletal procedures, such as dry needling, percutaneous needle electrolysis, percutaneous needle neuromodulation, platelet-rich plasma injection, obturator nerve block and surgery , with what this implies in terms of possible compromise of the relevant neurovascular bundles in the area. Before reaching the medial muscular compartment of the thigh, where the adductor longus is located, needles go through the skin, the tela subcutanea - superficial fascia -, the fascia lata -the investing fascia of the thigh- and, once into the medial compartment, the adductor longus. On the way to the adductor longus, the needle may reach the great saphenous vein, its tributaries, the external pudendal arteries, the obturator nerve, and even more importantly, all their likely anatomic variants . In the case of the great saphenous vein, three different anatomical variations were described by Dwight et al. in 1907 . One of these is that the great saphenous vein may perforate the fascia lata some distance below the saphenous opening, the second is that it can be occasionally double, and the third is that it can be replaced by a network of veins in which a main trunk is not identified. However, when this region was described with duplex ultrasound by Cavezzi et al. , five different variations were described. In this case, the first was that the great saphenous vein ascends in the saphenous compartment (splitting of the fascia lata in which the vein is housed) without large parallel tributaries; the second is that the great saphenous vein comprises two parallel veins in the saphenous compartment; the third is that a single great saphenous vein is located in the saphenous compartment with a large subcutaneous tributary that attaches to it at a variable level in the thigh; the forth is that the great saphenous vein and an anterior accessory saphenous vein join just before the femoral saphenous junction; and the fifth one is that there is a single great saphenous vein in the proximal thigh, in the saphenous compartment, and a large subcutaneous tributary joins it at a variable point along the thigh. Bergman et al. identified the variants in the tributaries of the great saphenous vein in 1984 describing that in 37% of the cases, the superficial iliac circumflex and the superficial epigastric veins form a common trunk to drain into the great saphenous vein, while the superficial external pudendal vein drains independently before the great saphenous vein enters the saphenous hiatus. In 9% of cases, the anterior accessory saphenous, the superficial iliac circumflex, and the superficial epigastric veins form a common trunk that drains into the saphenous, while also in 9% of the cases, the anterior accessory saphenous and the superficial circumflex iliac veins form a common trunk draining into the saphenous hiatus. Regarding the external pudendal arteries, these can be present in numbers between one to three, the majority being in the presence of two . Another important structure that is relevant when performing interventional procedures is the obturator nerve, which is located deeper into the aforementioned blood vessels. In a recent anatomical study, it was observed that in all the dissected cadavers the anterior branch of the obturator nerve ran anteriorly to the adductor brevis muscle and the posterior branch ran posteriorly to it. On the one hand, the anterior branch runs between the adductor brevis muscle on its ventral side and the pectineus and adductor longus muscles on its dorsal sides, covered throughout its entire length by the fascia of the adductor brevis, and providing numerous muscular branches for the pectineus muscle, the adductor longus, and the adductor brevis. On the other hand, the posterior branch, after emerging from the obturator canal, goes rapidly under the fascia of the adductor magnus muscle and travels in contact with this muscle, giving branches for the adductor brevis and adductor magnus muscles. However, there is a high anatomical variability in the divisions and subdivisions of the obturator nerve, which explains the difficulty frequently found in the application of different techniques, such as regional anesthetics . However, to date, no studies have compared the cadaveric and ultrasound features of the adductor longus and its relationships with neurovascular bundles in the areas most frequently treated with interventional musculoskeletal procedures. The study aimed to investigate if it is possible to define a “safe” entry route (approach window) that avoids needling the superficial vessels as well as determining the average distance between the dermis and the relevant structures in the territory of the adductor longus, describing the anatomy of the adductor longus through cadaveric and ultrasound study and its relationships with the areas most frequently treated within interventional musculoskeletal procedures. Study design An observational study was carried out in the period between March and May 2023. In the first stage, a cadaver study was carried out for the recognition of the relevant anatomical structures in the territory of the adductor longus muscle. Subsequently, in a sample of healthy subjects, an ultrasound evaluation was carried out to identify the anatomical structures to study their features and establish possible correlations. Study on cadaver The anatomic study was carried out on cadavers belonging to the Body Donation Center and Dissecting Rooms of the Complutense University of Madrid. The anterior and medial compartments of the thigh from six embalmed lower limbs (4 females and 2 males -ages ranging from 70 to 83-, 3 right and 3 left) were dissected to identify the anatomical structures of interest, to analyze the risks in their approach, and to define the anatomical landmarks that will facilitate its exploration by ultrasound. Corpses were embalmed with Cambridge mixture following the protocol described in Valderrama et al. . Ultrasound study 26 subjects ( n = 52; right and left side) without previous pathology, 12 women and 14 men, aged 33.6 ± 10.4 years, participated voluntarily in the present study, after signing an informed consent. A Logiq E R8 ultrasound machine with a 12L-RS linear probe was used, operating at a frequency of 10 MHz. A standardized protocol was defined in the collection of ultrasound images. A total of 52 limbs were evaluated, in a supine position on a table with the backrest positioned at about 45 degrees and with the hips in a comfortable position with slight abduction and external rotation. This position is similar to the FABER maneuver but with less external rotation to increase the tolerance of the participants . The ultrasound probe was placed in a transversal section to the adductor longus muscle to obtain a good visualization of the proximal myotendinous junction at a variable distance between 8 and 12 cm from the insertion of the adductor longus into the body of the pubic ramus. Additionally, the potential impact of probe pressure on the assessment of vascular structures was carefully considered. All examinations were performed with minimal contact, avoiding any exerted pressure beyond what was necessary to maintain proper imaging. The experienced examiner standardized and consistently followed this approach to ensure reliable and accurate evaluations of vascular structures. The study parameters were standardized in B-mode and Power Doppler Imaging (PDI) mode. Specifically, the PDI mode evaluation was performed with a frequency of 6.3 MHz, gain set at 20, and pulse repetition frequency (PRF) of 0.8 kHz. In B-mode, the adductor longus muscle of each lower limb was divided into two sections, with one being the image of the lateral cut (Fig. A) and the other the image of the medial section for B-mode (Fig. B), thus obtaining two cuts in B-mode of each adductor longus. This circumstance was motivated since it was not possible to visualize in a single window the amplitude of the adductor longus muscle. In PDI mode (Fig. ), for the visualization of blood vessels in each of the sections, the size of the Doppler box was adapted to the middle of the ultrasound screen, with the upper and lower limits allowing the maximum of the screen and the width of the box delimited by the central notch of the ultrasound screen that marks the center of the image. In this way, two images have been taken in PDI mode in each of the cuts in B-mode, always placing the Doppler box in the same sequence. To determine the optimal Doppler box size, preliminary tests were conducted comparing the sensitivity for detecting vascular structures with the dimensions used in the study and a 50% reduction of the box size. No differences were observed between these configurations. Therefore, the chosen box size was standardized to ensure consistency across subjects and avoid potential measurement variations. To name each image the following cipher has been used: right or left adductor longus, cut “1” or “2” where “1” is lateral and “2” medial, subject number and Doppler with the same sequence in the placement of the box, to the left of the screen first corresponding to lateral of the image in the lower right member and to the right of the screen then corresponding to medial in this case, while in the lower left limb, the Doppler box on the left would correspond to medial and on the right to lateral of the member. Therefore, there were obtained six images of each long adductor, two in B-mode and four in PDI mode. Based on the anatomical study of cadavers and the bibliographic references, the following outcomes were defined for the ultrasound study: Number of saphenous vein/s; Location of the saphenous vein/s concerning the proximal myotendinous junction: medial, anterior, anterolateral, or possible combinations (the way the images were divided is explained in Supplementary Information); Number of vessels in the thickness of the muscle or superficial to it taking the same references with the proximal myotendinous junction (medial, anterior, and anterolateral); Distance from the dermis to the lower limit of the adductor longus muscle was measured to establish the boundary with the anterior branch of the obturator nerve, which courses between the adductor longus and brevis muscles. Measurements were taken in the medial, anterior, and anterolateral zones. Statistical analysis Statistical analysis was performed with SPSS version 25. For each of the variables mentioned above, basic descriptive statistics were carried out (frequency, percentage, valid percentage, and cumulative percentage). Concerning the measurements from the surface of the dermis to the depth of the adductor longus muscle, the mean and standard deviation, as well as the minimum and maximum, were calculated. An observational study was carried out in the period between March and May 2023. In the first stage, a cadaver study was carried out for the recognition of the relevant anatomical structures in the territory of the adductor longus muscle. Subsequently, in a sample of healthy subjects, an ultrasound evaluation was carried out to identify the anatomical structures to study their features and establish possible correlations. The anatomic study was carried out on cadavers belonging to the Body Donation Center and Dissecting Rooms of the Complutense University of Madrid. The anterior and medial compartments of the thigh from six embalmed lower limbs (4 females and 2 males -ages ranging from 70 to 83-, 3 right and 3 left) were dissected to identify the anatomical structures of interest, to analyze the risks in their approach, and to define the anatomical landmarks that will facilitate its exploration by ultrasound. Corpses were embalmed with Cambridge mixture following the protocol described in Valderrama et al. . 26 subjects ( n = 52; right and left side) without previous pathology, 12 women and 14 men, aged 33.6 ± 10.4 years, participated voluntarily in the present study, after signing an informed consent. A Logiq E R8 ultrasound machine with a 12L-RS linear probe was used, operating at a frequency of 10 MHz. A standardized protocol was defined in the collection of ultrasound images. A total of 52 limbs were evaluated, in a supine position on a table with the backrest positioned at about 45 degrees and with the hips in a comfortable position with slight abduction and external rotation. This position is similar to the FABER maneuver but with less external rotation to increase the tolerance of the participants . The ultrasound probe was placed in a transversal section to the adductor longus muscle to obtain a good visualization of the proximal myotendinous junction at a variable distance between 8 and 12 cm from the insertion of the adductor longus into the body of the pubic ramus. Additionally, the potential impact of probe pressure on the assessment of vascular structures was carefully considered. All examinations were performed with minimal contact, avoiding any exerted pressure beyond what was necessary to maintain proper imaging. The experienced examiner standardized and consistently followed this approach to ensure reliable and accurate evaluations of vascular structures. The study parameters were standardized in B-mode and Power Doppler Imaging (PDI) mode. Specifically, the PDI mode evaluation was performed with a frequency of 6.3 MHz, gain set at 20, and pulse repetition frequency (PRF) of 0.8 kHz. In B-mode, the adductor longus muscle of each lower limb was divided into two sections, with one being the image of the lateral cut (Fig. A) and the other the image of the medial section for B-mode (Fig. B), thus obtaining two cuts in B-mode of each adductor longus. This circumstance was motivated since it was not possible to visualize in a single window the amplitude of the adductor longus muscle. In PDI mode (Fig. ), for the visualization of blood vessels in each of the sections, the size of the Doppler box was adapted to the middle of the ultrasound screen, with the upper and lower limits allowing the maximum of the screen and the width of the box delimited by the central notch of the ultrasound screen that marks the center of the image. In this way, two images have been taken in PDI mode in each of the cuts in B-mode, always placing the Doppler box in the same sequence. To determine the optimal Doppler box size, preliminary tests were conducted comparing the sensitivity for detecting vascular structures with the dimensions used in the study and a 50% reduction of the box size. No differences were observed between these configurations. Therefore, the chosen box size was standardized to ensure consistency across subjects and avoid potential measurement variations. To name each image the following cipher has been used: right or left adductor longus, cut “1” or “2” where “1” is lateral and “2” medial, subject number and Doppler with the same sequence in the placement of the box, to the left of the screen first corresponding to lateral of the image in the lower right member and to the right of the screen then corresponding to medial in this case, while in the lower left limb, the Doppler box on the left would correspond to medial and on the right to lateral of the member. Therefore, there were obtained six images of each long adductor, two in B-mode and four in PDI mode. Based on the anatomical study of cadavers and the bibliographic references, the following outcomes were defined for the ultrasound study: Number of saphenous vein/s; Location of the saphenous vein/s concerning the proximal myotendinous junction: medial, anterior, anterolateral, or possible combinations (the way the images were divided is explained in Supplementary Information); Number of vessels in the thickness of the muscle or superficial to it taking the same references with the proximal myotendinous junction (medial, anterior, and anterolateral); Distance from the dermis to the lower limit of the adductor longus muscle was measured to establish the boundary with the anterior branch of the obturator nerve, which courses between the adductor longus and brevis muscles. Measurements were taken in the medial, anterior, and anterolateral zones. Statistical analysis was performed with SPSS version 25. For each of the variables mentioned above, basic descriptive statistics were carried out (frequency, percentage, valid percentage, and cumulative percentage). Concerning the measurements from the surface of the dermis to the depth of the adductor longus muscle, the mean and standard deviation, as well as the minimum and maximum, were calculated. From the anatomical sample, in the subcutaneous layer, the most conspicuous structure was the great saphenous vein that rests on the adductor longus muscle (Fig. A and C). At the deep level, the proximal myotendinous junction of the muscle and the anterior branch of the obturator nerve that runs in its deepest superficial plane to the adductor brevis muscle were identified, as well as a network of vessels between the dermis and the thickness of the adductor longus muscle (Fig. D). The main potential risks identified on cadavers with interventional procedures at this level were to damage the great saphenous vein, the anterior branch of the obturator nerve, and the vascular network that crosses the adductor longus muscle (Fig. A and D). It should be noted that it was not possible to define any anatomical window that guarantees the palpation-guided approach without risk. The ultrasound study included 52 thighs (26 right and 26 left), from 26 healthy subjects aged 33.6 ± 10.4 years (12 women). In the total of 52 cases studied, 75% had one saphenous vein, 17.3% had two and in 7.7% the saphenous vein was not visualized at the height of the muscle. In 20.8% the location of the saphenous vein was medial to the proximal myotendinous junction of the adductor longus, in 43.8% it was anterior, and in 16.7% it was anterolateral. In the cases in which two saphenous veins were seen, there were different locations for the proximal myotendinous junction (medial plus anterior, medial plus anterolateral, or anterior plus anterolateral) (Table ). In addition to the saphenous vein, different vessels have also been counted. Between the dermis and the deepest limit of the muscle, two anterolateral vessels were found in 5.8% of cases, one anterolateral vessel in 42.3% of cases, and no anterolateral vessel in 51.9% of cases. In the case of the anterior vessels, there was a great variety, with three vessels (1.9%), two vessels (7.7%), one vessel (61.5%) and no vessel (28.8%). Finally, in the case of medial vessels, there were two (3.8%), one (44.2%), and none (51.9%). Regarding the total number of vessels in their anterolateral, anterior, and medial locations, it should be noted that 91.4% had at least one vessel and that 59.6% had between two and five vessels in the region (Table ). Regarding the distance between the surface of the dermis to the depth of the adductor longus muscle in its anterolateral, anterior, and medial zone with the anterior branch of the obturator nerve, the mean distance ranged from 3.63 to 3.93 cm (Table ). The cadaveric and ultrasound evaluation of the different neurovascular bundles around the area of the adductor longus muscle shows that 91.4% of the sample (47 cases) presented at least a vessel in the transversal section at the height of the proximal myotendinous junction, where acute and chronic muscle injuries are frequent and therefore a vessel can be potentially reached with when an interventional procedure is performed at this level. The variability observed in the number and distribution of blood vessels in the area of the adductor longus (almost 60% had between two and five vessels in the thickness of the adductor longus) prevents the establishment of a “safe” window for the palpation-guided approach with anatomical references. In this sense, it is essential for safety and to avoid adverse effects such as hematoma, that interventional musculoskeletal procedures are performed ultrasound-guided. Regarding the obturator nerve, it is important to note that at a mean depth of 3.63–3.93 cm., just before the thickness of the adductor longus, the anterior branch of this nerve is located, and it could also be damaged when performing interventional musculoskeletal procedures at this level. Although there is no extensive literature about interventional musculoskeletal procedures, it is frequent that the obturator nerve is treated with an injection of local anesthetics and/or percutaneous needle neuromodulation, which are recommended to be performed ultrasound-guided to locate the nerve precisely and to reduce the rates of vascular puncture . This nerve has been the subject of several cadaveric studies because of its varied anatomic locations and the presence of accessory branches . The first study in which ultrasound visualization of this structure was carried out to improve regional block in clinical anesthesia was that of Soong et al. in 2007 . In this study, the mean distance from the ultrasound transducer to the pubic tubercle when visualizing the anterior and posterior obturator nerve branches in parallel was 2.1 ± 1.2 cm laterally and 2.1 ± 2.0 cm distally. The common obturator nerve and the posterior division were found deeper (25.9 ± 7.6 and 29.3 ± 7.9 mm, respectively) than the anterior division (15.5 ± 3.9 mm). There is now growing interest in the analysis of adverse effects found after interventional musculoskeletal procedures. The study of Brady et al. about deep dry needling reports a rate of 19.8% of adverse effects. The common ones described were hematoma (7.55%), bleeding (4.65%), pain during treatment (3.01%), and post-needling pain (2.19%). The high variability of the vascular distribution around the adductor longus muscle is a potential factor of adverse effects if the approach is not ultrasound-guided. Although nerves do not present as much anatomical variability as arteries and veins, it is possible to damage a nerve during an interventional musculoskeletal procedure . One of the study’s strengths is its clinical orientation, as the territory of the proximal myotendinous junction of the adductor longus muscle was selected since, according to the authors ´ clinical experience, it is the most prevalent injured area. We have not identified any studies that have described such a circumstance. However, the study also presents some limitations that must be considered when interpreting the findings. First, the ultrasound section performed exclusively at the proximal myotendinous junction makes it impossible to describe the tributary veins and the external pudendal artery defined in the scientific literature [3,5,6,17]. To overcome this limitation, a study including a sweep from proximal to distal should be performed to fully describe the region. Furthermore, the dissection images do not aim to establish a direct correlation with the findings from cadaver ultrasound scans, as this would require a larger anatomical sample or an expanded study design that integrates additional dissection. Additionally, the use of large Doppler boxes may have underestimated the number of vessels and their precise location. While this approach ensures the capture of all potential vessels along the needle trajectory for clinical safety during interventional musculoskeletal procedures, adapting the Doppler boxes more specifically to the structures of interest could enhance anatomical precision. Lastly, another limitation was that the fatty volume of the area analyzed hindered the visualization of the neurovascular bundles. Therefore, recommendations for future studies include the definition of different subgroups based on body mass index or body fat percentage to overcome this limitation and using different cuts along the muscle to see the different routes of the vessels, as well as taking advantage of the ultrasound to perform a sweep that allows to see where the vessels run throughout their journey, thus allowing to describe approach windows at the level of the proximal and distal myotendinous junction, as well as in the muscular belly. There was a high variability in the number and location of vascular bundles (superficial and deep) in the territory of the adductor longus, which impedes defining a “safe” approach window for palpation-guided interventional musculoskeletal procedures only with anatomical references. In terms of nerve structures, there does not seem to be as much variability and the risk appears after exceeding a depth of more than four centimeters. Therefore, the ultrasound-guided approach would be the only safe way to perform interventional techniques at this level. Below is the link to the electronic supplementary material. Supplementary Material 1 |
Gut microbiome shifts in adolescents after sleeve gastrectomy with increased oral-associated taxa and pro-inflammatory potential | 0e185ed3-48c6-4b71-8816-1903bf6637a1 | 11845021 | Surgical Procedures, Operative[mh] | The epidemic of childhood obesity continues unabated, with 19.3% of children and adolescents having obesity and 6.1% having severe obesity in the USA. Multiple comorbidities are associated with childhood obesity including type 2 diabetes mellitus (T2DM), metabolic dysfunction-associated steatotic liver disease (MASLD), dyslipidemia, and continuation of obesity into adulthood. , If weight loss can be achieved prior to entering adulthood, the risk of these conditions is mitigated. Therefore, childhood obesity is a key target for intervention. Bariatric surgery achieves significant weight loss and reduces or resolves associated comorbidities in children with severe obesity. Clinical guidelines recommend bariatric surgery as a potential intervention for children and adolescents with class II obesity and comorbidities or class III obesity with or without comorbidities. , Laparoscopic vertical sleeve gastrectomy (VSG), with removal of 80–90% of the greater curvature of the stomach, is the most common pediatric bariatric procedure and the only type performed on those <13 years of age. Although VSG in children and adolescents is highly effective, the biological mechanisms underlying the weight loss and metabolic improvement are not fully elucidated. Restriction of food intake by reduction of stomach capacity plays an important role but the degree of enhanced metabolism post-VSG cannot be explained by caloric restriction alone. VSG also leads to weight loss through altered neurohormonal feedback mechanisms including increases in glucagon-like peptide-1 (GLP-1) and peptide YY, which reduces appetite. There has been growing interest in the role of the gut microbiome in the mechanisms behind VSG. Many studies have shown a difference in the gut microbiome between individuals with and without obesity. Moreover, murine models with fecal transplant (FT) from humans with obesity to germ-free mice have shown transfer of the obesity phenotype indicating a causal role for the gut microbiome. Studies examining gut microbiome changes after bariatric surgery in adults suggest an increase in microbiota diversity and decrease in Firmicutes/Bacteroidetes ratio, although specific changes differ between surgery type and among studies. , To our knowledge, the role of the gut microbiome in adolescents undergoing bariatric surgery has not yet been examined. This is uniquely important to study as the microbiome of children and adolescents differs from adults. The developing microbiome in childhood also plays a clear role in establishing metabolic and inflammatory pathways that could impact energy regulation in obesity. , Additionally, there is mounting evidence suggesting that the age of onset of obesity significantly impacts overall cardiometabolic risk, with childhood obesity possibly representing a more virulent form of the disease as both MASLD and T2DM progress more rapidly in children than adults. Thus, studying microbiome changes in children and adolescents may be of particular importance. Furthermore, many of the studies in adults undergoing bariatric surgery focused on Roux-en-Y gastric bypass, not VSG, used 16S rRNA gene microbiome sequencing, and did not examine microbiome function and metabolites. Also, there are limited studies examining whether the microbiota changes found with bariatric surgery have a causal role in the beneficial phenotype changes seen post-surgery. Therefore, we aimed to 1) deeply examine gut microbiome structure and function changes with VSG in adolescents using shotgun metagenomic sequencing and wide metabolomic analysis and 2) assess if the microbiome/metabolome differences seen have a causal role in the phenotypic changes observed with VSG by performing FT of stool from adolescents pre-VSG and post-VSG into germ-free mice. Experimental model and subject details Human methods and analysis Subjects and samples Children and adolescents undergoing VSG between January 2021 and February 2023 were enrolled in an Institutional Review Board (IRB) approved longitudinal cohort study “Biorepository of Specimens for Pediatric Obesity Research Use” (Children’s National Hospital IRB protocol #Pro00015976) with signed consent and assent (latter for those >11 years). Inclusion criteria for this sub-study were children and adolescents (<19 years) undergoing vertical sleeve gastrectomy at the Children’s National Hospital in Washington DC. To be eligible for surgery, children required a BMI ≥ 120th percentile of 95th percentile for age and sex with an obesity-related comorbidity and/or BMI ≥ 140th percentile. Participants were enrolled up to 2 months prior to their planned VSG. After consent was obtained, a stool and urine collection container were mailed to participants. For stool, participants mailed back a sample aliquoted in an OMNIgene-gut tube and OMNImet-gut tube for DNA and metabolite preservation respectively (DNA Genotek Inc). Wherever possible participants also returned whole stool aliquoted into tubes with glycerol (for bacterial preservation) and whole stool without preservatives. Participants gave a clean catch urine sample mixed with AssayAssure ® Genelock (Sierra Molecular) for protein preservation. All samples were mailed back overnight on icepacks and stored at −80°C until analysis. The same procedure was followed for post-VSG samples. Clinical data was obtained via extraction of medical records from clinical appointments pre-VSG and post-VSG and patient collected data accompanying the samples and was stored in a secure REDCap database. For clinical data analysis, BMI, weight, HbA1c%, LDL, HDL, triglycerides, and ALT passed Shapiro-Wilks normality and were compared pre-VSG and post-VSG via paired t-tests. T2DM, dyslipidemia, and hypertension were compared via McNemar exact tests. Patient involvement Patients were involved in the design and conduct of the trial. We received input from patients in prior trials for the design of convenient sample collection methods that would minimize inconvenience and were suitable for the planned downstream assays. We intend to disseminate the main results to study participants. Microbiome sequencing DNA Extraction : DNA was extracted from human and murine fecal samples in two stages. First, approximately 50 mg of fecal material and 650 μL MBL lysis buffer from the PowerMicrobiome DNA/RNA EP Kit (Qiagen) were added to Lysis Matrix E (LME) tubes (MP Biomedicals). LME tubes were transferred to a Precellys 24 Tissue Homogenizer (Bertin Technologies) and fecal samples were homogenized, centrifuged, with the resultant supernatant transferred to a deep-well 96-well plate. The second stage consisted of DNA isolation from the above supernatant using the MagAttract PowerMicrobiome DNA/RNA EP Kit (Qiagen) on an automated liquid handling system as detailed by the manufacturer (Eppendorf). Shotgun Metagenomic Sequencing . Total gene content of the microbiome was assessed through shotgun metagenomic sequencing. Metagenomic libraries were constructed from 100 ng of DNA as starting material using the Illumina DNA Prep kit. Illumina DNA/RNA UD Indexes were used to add sample-specific sequencing indices to both ends of the libraries. An Agilent 4200 TapeStation system with High Sensitivity D5000 ScreenTape (Agilent Technologies, Inc) was used to verify quality and assess final library size. A positive control (MSA-2002 20 Strain Even Mix Whole-Cell Material (ATCC) and a buffer extraction-negative control were included. Metagenomic libraries were normalized and pooled at an equimolar concentration. Final pools were diluted to 750 pM and sequenced on a NextSeq2000 sequencer using a paired-end (100×100) NextSeq 1000/2000 P2 (200 cycles) kit (Illumina, Inc). Microbiome analysis Data Processing : The quality of raw paired-end sequence reads was assessed with FastQC and MutiQC. Adapters revealed by FastQC were removed using bbtools’ bbduk software. Reads then underwent the Whole-Genome Sequence Assembly 2 (WGSA2) protocol using the Nephele platform (version 2.24.2). In brief, reads were processed with fastp and minimal trimming and filtering by ensuring an average read quality of 10, a trim of the 3’ end of the read at a quality of 15 and trimming of the 5’ end at a Q score of 20, with additional filtering of reads if they were less than 60 bases after trimming. Decontamination was undertaken using Kraken2 with a database containing the human and mouse genome. After adapter trimming and filtering, samples contained between 15 M and 30 M paired-end reads, of which between 8.7 M and 22 M were classified to the bacterial kingdom. Taxonomic identification was performed on the trimmed, error-corrected, and decontaminated reads in Kraken2 with the default RefSeq database. Assembly and Gene Annotation . Within the WGSA2 pipeline, the trimmed, error-corrected and decontaminated reads were assembled into contiguous sequences, or contigs, using metaSPAdes. Reads were recruited back to contigs using bowtie29 and SAMtools to produce information on scaffold coverage and quality. Protein coding regions (CDS) were predicted from assembled scaffolds using prodigal. Predicted CDS regions were processed by EggNog-mapper2 to identify and annotate genes with KEGG Orthology (KO), Enzyme Commission (EC) and Clusters of Orthologous Genes (COG) identifiers. Abundances were calculated using verse to obtain Transcripts Per Million (TPM) at the CDS level and summed to obtain TPM by gene. CAZymes . Carbohydrate Active Enzymes (CAZymes) were annotated from assembled metagenomic scaffolds using the dbCAN software in meta mode with default settings and default databases to obtain eCAMI, HMMER and DIAMOND-based annotations. Annotations were provided at the gene level and merged with predicted gene abundances. Genes were also annotated with taxonomy using Kraken2, creating a table of CAZyme abundances both stratified by taxonomy and unstratified. Gene abundances were normalized to copies per million before analysis. The eCAMI-based CAZyme identification was used in analysis and focused specifically on CAZymes supported by DIAMOND or HMMER when available. Antibiotic Resistance Genes . Antibiotic resistance genes were identified using the Comprehensive Antibiotic Resistance Database with the Resistance Gene Identifier tool v6.0.1, nudging loose hits to strict and including low-quality assemblies with prediction of partial genes. Resulting annotations were processed as for CAZymes. Virome . After assembly with WGSA2, assembled scaffolds and binary alignment maps (BAM) were processed for the presence of viral diversity (ssDNA, dsDNA phage, and giant DNA viruses) using Nephele’s DiscoVir pipeline ( https://nephele.niaid.nih.gov/pipeline_details/discovir/ ). Briefly, geNomad predicted viral genomes and fragments using default confidence parameters as defined by the tool, and VERSE calculated read counts of viral genomes based on BAM files. , Next, CheckV assessed quality and all scaffolds identified as viral were retained for downstream processing and analysis. Viral genomes and fragments greater than 1000 basepairs were clustered with bbtools and MMseqs2 to generated vOTUs which were functionally annotated with DRAM-v with KOfam, Pfam, and Viral Orthologous Group (VOG) databases. Auxiliary metabolic genes were identified with VirSorter2.0 and DRAM-v. Phage hosts were predicted using iPHoP. Statistical analysis . All statistical analysis of the microbiome was performed using R 4.3.0. For taxonomic analysis, reads were filtered to those that aligned to the bacterial kingdom and normalized with rarefaction to 8 million reads per sample. Several measures of alpha diversity were calculated, including Chao1 Richness, Observed Taxa, Evenness, Inverse Simpson, and Shannon Diversity using estimate_richness from phyloseq. For all data types, Bray Curtis and Canberra distance matrices were calculated using phyloseq’s distance function. Alpha diversity was compared between pre-VSG and post-VSG samples using linear mixed-effects models within the lmerTest package. The composition of the microbiome was compared against study covariates with Permutational Analysis of Variance (PERMANOVA) using the adonis2 function from the vegan package. Significant relationships were visualized using Principal Coordinates Analysis (PCoA) ordination in phyloseq and ggplot2. All differential abundance analyses were undertaken with Maaslin2, wherein data were analyzed with linear models after log-transformation. Features were filtered if they exhibited a minimum prevalence of less than 10% and a minimum variance of 0.01. When paired samples were included, Subject ID was provided as a random effect. An FDR-corrected p-value less than 0.2 was considered significant. Co-associated networks of taxa or EC abundances were generated using Weighted Gene Co-Association Network Analysis. In brief, this tool reduces multi-dimensional data to co-associated networks or modules. For taxa-based modules, taxa were included that were found to change nominally after surgery at a false-discovery-corrected p-value of less than 0.5. Blockwise modules were generated using a soft power threshold of 14, an unsigned topology overlap matrix, minimum module size of 20, merge cut height of 0.15, and deepSplit of 3. This process generated eight modules of taxa that were annotated manually by combining the taxonomic information, the average abundance of the taxon across samples, and the module membership, calculated as the correlation of the taxon abundance with the module eigengene. WGCNA modules were also created for EC abundances using similar settings. ECs were included if they changed significantly due to surgery at an FDR P-value of 0.4. Modules were constructed using the same parameters as above except with a soft power threshold of 7. This resulted in four distinct modules of ECs. These were described using the pathway enrichment tool OmePath, where scores used were based on module membership and calculated as described previously. Relationships between module eigengenes and metabolites were identified using linear mixed effects models. Results were visualized using the igraph package 21 if they were significant at an FDR p-value less than 0.05. Directionality of the results was represented by edge color and data type was indicated through color of nodes. This work utilized the computational resources of the NIH HPC Biowulf cluster ( http://hpc.nih.gov .) and the NIAID Locus cluster ( http://locus.niaid.nih.gov ). Metabolomics Metabolite and Lipid Sample Preparation : For all liquid chromatography mass spectrometry (LCMS) methods, LCMS-grade solvents were used. For bile acid analysis 400 µL of homogenized feces was taken from the fecal collection tubes and added to 500 µL of ice-cold methanol. To each sample 5 µL of the Bile Acid SPLASH® (Avanti Polar Lipids Inc.) and 2 µg of butylated hydroxytoluene was added. Samples were agitated via shaking at 4°C for 20 min and then centrifuged at 16k ×g for 20 min. An aliquot of the supernatant was taken directly for liquid LCMS analysis. For short-chain fatty acid (SCFA) and polar metabolomics, a separate 400 µL aliquot of homogenized feces was added to 400 µL of water. Following mixing, 400 µL of chloroform was added. Samples were shaken for 30 min at 4°C and subsequently centrifuged at 16k ×g for 20 min. About 400 µL of the top (aqueous) layer was collected. The aqueous layer was sub-aliquoted for SCFA derivatization or diluted 5× in 50% methanol in water and prepared for LCMS injection. SCFA derivatization . To preserve SCFAs for analysis an aliquot of the aqueous fraction was derivatized with O-benzylhydroxylamine (O-BHA) as previously described with modifications. , The reaction buffer contained 1 M pyridine and 0.5 M hydrochloric acid in water. To 35 µL of sample, 10 µL of 1 M O-BHA in reaction buffer and 10 µL of 1 M 1-Ethyl-3-(3-dimethylaminopropyl) carbodiimide in reaction buffer were added. Samples were derivatized for 2 h at room temperature with constant agitation. Addition of 50 µL of 0.1% formic acid was used to quench the reaction, which eliminated the potential for formate to be measured in these samples. To extract derivatized molecules 400 µL of ethyl acetate was added. The samples were centrifuged at 16k ×g and 4°C for 5 min to induce layering. The upper (organic) layer was collected and dried under vacuum. Samples were resuspended in 300 µL of water for LCMS injection. Liquid chromatography mass spectrometry . Tributylamine and all synthetic molecular references were purchased from Millipore Sigma. LCMS grade water, methanol, isopropanol and acetic acid were purchased through Fisher Scientific. All samples were separated using a Sciex ExionLC™ AC system and measured using a Sciex 5500 QTRAP® or Sciex 6500+ QTRAP® mass spectrometer. Polar metabolites were analyzed as previously described. For all metabolomics analysis, quality control samples, consisting of a mixture of the analyzed samples, were injected after every 10 injections to control for signal stability. Samples were analyzed via separate negative ionization and positive ionization methods. For negative mode analysis, a Waters Atlantis T3 column (100 Å, 3 μm, 3 mm × 100 mm) with a gradient from 5 mm tributylamine, 5 mm acetic acid in 2% isopropanol, 5% methanol, 93% water (v/v) to 100% isopropanol over 15 min was used. For positive mode analysis, samples were separated on a Phenomenex Kinetex F5 column (100 Å, 2.6 μm, 2.1 mm × 100 mm) column with a gradient from 100% water with 0.1% formic acid to 95% acetonitrile with 0.1% formic acid over 5 min. Each metabolite was measured using at two distinct multiple-reaction monitoring (MRM) signals and a defined retention time. For SCFA analysis samples were separated on a Waters™ Atlantis dC18 column (100Å, 3 µm, 3 mm × 100 mm) with a 6 min gradient from 5% to 80% B with buffer A as 0.1% formic acid in water and B as 0.1% formic acid in methanol. All SCFA were measured using positive ionization using MRMs that featured the characteristic 91 daughter ion from O-BHA derivatization. Identity was confirmed via comparison to previous standards. Bile acid samples were separated on a Phenomenex Kinetex® Polar C18 (100Å, 2.6 µm, 3 mm × 100 mm) using a binary gradient of A: 0.01% acetic acid in water and B: 0.01% acetic acid in methanol. A 20 min gradient from 40% to 100% B was utilized for separation. Samples were detected in negative MRM mode using previously validated MRMs. Internal bile acid standard signals were used to confirm signal identities and retention times. Metabolomic analysis . All signals were integrated using SciexOS 3.1 (AB Sciex Pte. Ltd.). Signals with greater than 50% missing values were discarded and the remaining missing values were replaced with the lowest registered signal value. All signals with a QC coefficient of variance greater than 30% were discarded. Metabolites with multiple MRMs were quantified with the higher signal-to-noise MRM. Filtered datasets were total sum normalized prior to analysis. The SCFA dataset and the two polar metabolomics datasets were scaled and combined using common signal for serine and succinate for the positive mode metabolite method and the SCFA method respectively. A paired t-test was used for all bile acid and metabolite statistics and a Benjamini–Hochberg method for correction for multiple comparisons was imposed where indicated. Fecal Calprotectin (FC) FC was assessed using the Buhlmann Fecal Calprotectin ELISA kit (BÜHLMANN fCAL® ELISA, https://buhlmannlabs.com/buhlmann-fcal-elisa/ ) following the manufacturer’s guidelines. The FCs included all subjects who did not pass Shapiro-Wilks normality, so pre- and post-VSG were compared via a Wilcoxon test. FCs excluding outliers passed normality and were compared pre- and post-VSG via a paired t-test. Urine proteomics Urine samples were analyzed using the SomaScan V4.1 Assay, an aptamer-based quantitative proteomic biomarker discovery platform (SomaLogic; Boulder, CO). The assay was run according to manufacturer’s specifications which includes pH adjustment and buffer exchange by gel filtration prior to normalizing total protein concentration of urine samples to a standard input concentration ( https://somalogic.com/wp-content/uploads/2023/09/D0005009_Rev1_2023-09_SomaScan-7K-v4.1-UrinePre-processing-User-Manual.pdf ). Data was then subjected to the manufacturer’s standard normalization methods, including adaptive normalization by maximum likelihood by SomaLogic. Identified enriched gene sets were determined utilizing the pre-ranked gene-set enrichment analysis (GSEA) algorithm, as implemented in the FGSEA R package. Genes were prioritized based on moderated T statistics derived from the limma model’s relevant coefficient, and enrichment analysis was conducted using the Reactome database, with correction of p values applied for multiple sampling. This analysis can be used to identify significant enrichment of a set of foreground genes or proteins, in predefined gene sets, compared against a reference set. Quality control and initial data processing were performed using an R package ( https://github.com/foocheung/sqs ) and Shiny app. Human methods and analysis Subjects and samples Children and adolescents undergoing VSG between January 2021 and February 2023 were enrolled in an Institutional Review Board (IRB) approved longitudinal cohort study “Biorepository of Specimens for Pediatric Obesity Research Use” (Children’s National Hospital IRB protocol #Pro00015976) with signed consent and assent (latter for those >11 years). Inclusion criteria for this sub-study were children and adolescents (<19 years) undergoing vertical sleeve gastrectomy at the Children’s National Hospital in Washington DC. To be eligible for surgery, children required a BMI ≥ 120th percentile of 95th percentile for age and sex with an obesity-related comorbidity and/or BMI ≥ 140th percentile. Participants were enrolled up to 2 months prior to their planned VSG. After consent was obtained, a stool and urine collection container were mailed to participants. For stool, participants mailed back a sample aliquoted in an OMNIgene-gut tube and OMNImet-gut tube for DNA and metabolite preservation respectively (DNA Genotek Inc). Wherever possible participants also returned whole stool aliquoted into tubes with glycerol (for bacterial preservation) and whole stool without preservatives. Participants gave a clean catch urine sample mixed with AssayAssure ® Genelock (Sierra Molecular) for protein preservation. All samples were mailed back overnight on icepacks and stored at −80°C until analysis. The same procedure was followed for post-VSG samples. Clinical data was obtained via extraction of medical records from clinical appointments pre-VSG and post-VSG and patient collected data accompanying the samples and was stored in a secure REDCap database. For clinical data analysis, BMI, weight, HbA1c%, LDL, HDL, triglycerides, and ALT passed Shapiro-Wilks normality and were compared pre-VSG and post-VSG via paired t-tests. T2DM, dyslipidemia, and hypertension were compared via McNemar exact tests. Patient involvement Patients were involved in the design and conduct of the trial. We received input from patients in prior trials for the design of convenient sample collection methods that would minimize inconvenience and were suitable for the planned downstream assays. We intend to disseminate the main results to study participants. Microbiome sequencing DNA Extraction : DNA was extracted from human and murine fecal samples in two stages. First, approximately 50 mg of fecal material and 650 μL MBL lysis buffer from the PowerMicrobiome DNA/RNA EP Kit (Qiagen) were added to Lysis Matrix E (LME) tubes (MP Biomedicals). LME tubes were transferred to a Precellys 24 Tissue Homogenizer (Bertin Technologies) and fecal samples were homogenized, centrifuged, with the resultant supernatant transferred to a deep-well 96-well plate. The second stage consisted of DNA isolation from the above supernatant using the MagAttract PowerMicrobiome DNA/RNA EP Kit (Qiagen) on an automated liquid handling system as detailed by the manufacturer (Eppendorf). Shotgun Metagenomic Sequencing . Total gene content of the microbiome was assessed through shotgun metagenomic sequencing. Metagenomic libraries were constructed from 100 ng of DNA as starting material using the Illumina DNA Prep kit. Illumina DNA/RNA UD Indexes were used to add sample-specific sequencing indices to both ends of the libraries. An Agilent 4200 TapeStation system with High Sensitivity D5000 ScreenTape (Agilent Technologies, Inc) was used to verify quality and assess final library size. A positive control (MSA-2002 20 Strain Even Mix Whole-Cell Material (ATCC) and a buffer extraction-negative control were included. Metagenomic libraries were normalized and pooled at an equimolar concentration. Final pools were diluted to 750 pM and sequenced on a NextSeq2000 sequencer using a paired-end (100×100) NextSeq 1000/2000 P2 (200 cycles) kit (Illumina, Inc). Microbiome analysis Data Processing : The quality of raw paired-end sequence reads was assessed with FastQC and MutiQC. Adapters revealed by FastQC were removed using bbtools’ bbduk software. Reads then underwent the Whole-Genome Sequence Assembly 2 (WGSA2) protocol using the Nephele platform (version 2.24.2). In brief, reads were processed with fastp and minimal trimming and filtering by ensuring an average read quality of 10, a trim of the 3’ end of the read at a quality of 15 and trimming of the 5’ end at a Q score of 20, with additional filtering of reads if they were less than 60 bases after trimming. Decontamination was undertaken using Kraken2 with a database containing the human and mouse genome. After adapter trimming and filtering, samples contained between 15 M and 30 M paired-end reads, of which between 8.7 M and 22 M were classified to the bacterial kingdom. Taxonomic identification was performed on the trimmed, error-corrected, and decontaminated reads in Kraken2 with the default RefSeq database. Assembly and Gene Annotation . Within the WGSA2 pipeline, the trimmed, error-corrected and decontaminated reads were assembled into contiguous sequences, or contigs, using metaSPAdes. Reads were recruited back to contigs using bowtie29 and SAMtools to produce information on scaffold coverage and quality. Protein coding regions (CDS) were predicted from assembled scaffolds using prodigal. Predicted CDS regions were processed by EggNog-mapper2 to identify and annotate genes with KEGG Orthology (KO), Enzyme Commission (EC) and Clusters of Orthologous Genes (COG) identifiers. Abundances were calculated using verse to obtain Transcripts Per Million (TPM) at the CDS level and summed to obtain TPM by gene. CAZymes . Carbohydrate Active Enzymes (CAZymes) were annotated from assembled metagenomic scaffolds using the dbCAN software in meta mode with default settings and default databases to obtain eCAMI, HMMER and DIAMOND-based annotations. Annotations were provided at the gene level and merged with predicted gene abundances. Genes were also annotated with taxonomy using Kraken2, creating a table of CAZyme abundances both stratified by taxonomy and unstratified. Gene abundances were normalized to copies per million before analysis. The eCAMI-based CAZyme identification was used in analysis and focused specifically on CAZymes supported by DIAMOND or HMMER when available. Antibiotic Resistance Genes . Antibiotic resistance genes were identified using the Comprehensive Antibiotic Resistance Database with the Resistance Gene Identifier tool v6.0.1, nudging loose hits to strict and including low-quality assemblies with prediction of partial genes. Resulting annotations were processed as for CAZymes. Virome . After assembly with WGSA2, assembled scaffolds and binary alignment maps (BAM) were processed for the presence of viral diversity (ssDNA, dsDNA phage, and giant DNA viruses) using Nephele’s DiscoVir pipeline ( https://nephele.niaid.nih.gov/pipeline_details/discovir/ ). Briefly, geNomad predicted viral genomes and fragments using default confidence parameters as defined by the tool, and VERSE calculated read counts of viral genomes based on BAM files. , Next, CheckV assessed quality and all scaffolds identified as viral were retained for downstream processing and analysis. Viral genomes and fragments greater than 1000 basepairs were clustered with bbtools and MMseqs2 to generated vOTUs which were functionally annotated with DRAM-v with KOfam, Pfam, and Viral Orthologous Group (VOG) databases. Auxiliary metabolic genes were identified with VirSorter2.0 and DRAM-v. Phage hosts were predicted using iPHoP. Statistical analysis . All statistical analysis of the microbiome was performed using R 4.3.0. For taxonomic analysis, reads were filtered to those that aligned to the bacterial kingdom and normalized with rarefaction to 8 million reads per sample. Several measures of alpha diversity were calculated, including Chao1 Richness, Observed Taxa, Evenness, Inverse Simpson, and Shannon Diversity using estimate_richness from phyloseq. For all data types, Bray Curtis and Canberra distance matrices were calculated using phyloseq’s distance function. Alpha diversity was compared between pre-VSG and post-VSG samples using linear mixed-effects models within the lmerTest package. The composition of the microbiome was compared against study covariates with Permutational Analysis of Variance (PERMANOVA) using the adonis2 function from the vegan package. Significant relationships were visualized using Principal Coordinates Analysis (PCoA) ordination in phyloseq and ggplot2. All differential abundance analyses were undertaken with Maaslin2, wherein data were analyzed with linear models after log-transformation. Features were filtered if they exhibited a minimum prevalence of less than 10% and a minimum variance of 0.01. When paired samples were included, Subject ID was provided as a random effect. An FDR-corrected p-value less than 0.2 was considered significant. Co-associated networks of taxa or EC abundances were generated using Weighted Gene Co-Association Network Analysis. In brief, this tool reduces multi-dimensional data to co-associated networks or modules. For taxa-based modules, taxa were included that were found to change nominally after surgery at a false-discovery-corrected p-value of less than 0.5. Blockwise modules were generated using a soft power threshold of 14, an unsigned topology overlap matrix, minimum module size of 20, merge cut height of 0.15, and deepSplit of 3. This process generated eight modules of taxa that were annotated manually by combining the taxonomic information, the average abundance of the taxon across samples, and the module membership, calculated as the correlation of the taxon abundance with the module eigengene. WGCNA modules were also created for EC abundances using similar settings. ECs were included if they changed significantly due to surgery at an FDR P-value of 0.4. Modules were constructed using the same parameters as above except with a soft power threshold of 7. This resulted in four distinct modules of ECs. These were described using the pathway enrichment tool OmePath, where scores used were based on module membership and calculated as described previously. Relationships between module eigengenes and metabolites were identified using linear mixed effects models. Results were visualized using the igraph package 21 if they were significant at an FDR p-value less than 0.05. Directionality of the results was represented by edge color and data type was indicated through color of nodes. This work utilized the computational resources of the NIH HPC Biowulf cluster ( http://hpc.nih.gov .) and the NIAID Locus cluster ( http://locus.niaid.nih.gov ). Metabolomics Metabolite and Lipid Sample Preparation : For all liquid chromatography mass spectrometry (LCMS) methods, LCMS-grade solvents were used. For bile acid analysis 400 µL of homogenized feces was taken from the fecal collection tubes and added to 500 µL of ice-cold methanol. To each sample 5 µL of the Bile Acid SPLASH® (Avanti Polar Lipids Inc.) and 2 µg of butylated hydroxytoluene was added. Samples were agitated via shaking at 4°C for 20 min and then centrifuged at 16k ×g for 20 min. An aliquot of the supernatant was taken directly for liquid LCMS analysis. For short-chain fatty acid (SCFA) and polar metabolomics, a separate 400 µL aliquot of homogenized feces was added to 400 µL of water. Following mixing, 400 µL of chloroform was added. Samples were shaken for 30 min at 4°C and subsequently centrifuged at 16k ×g for 20 min. About 400 µL of the top (aqueous) layer was collected. The aqueous layer was sub-aliquoted for SCFA derivatization or diluted 5× in 50% methanol in water and prepared for LCMS injection. SCFA derivatization . To preserve SCFAs for analysis an aliquot of the aqueous fraction was derivatized with O-benzylhydroxylamine (O-BHA) as previously described with modifications. , The reaction buffer contained 1 M pyridine and 0.5 M hydrochloric acid in water. To 35 µL of sample, 10 µL of 1 M O-BHA in reaction buffer and 10 µL of 1 M 1-Ethyl-3-(3-dimethylaminopropyl) carbodiimide in reaction buffer were added. Samples were derivatized for 2 h at room temperature with constant agitation. Addition of 50 µL of 0.1% formic acid was used to quench the reaction, which eliminated the potential for formate to be measured in these samples. To extract derivatized molecules 400 µL of ethyl acetate was added. The samples were centrifuged at 16k ×g and 4°C for 5 min to induce layering. The upper (organic) layer was collected and dried under vacuum. Samples were resuspended in 300 µL of water for LCMS injection. Liquid chromatography mass spectrometry . Tributylamine and all synthetic molecular references were purchased from Millipore Sigma. LCMS grade water, methanol, isopropanol and acetic acid were purchased through Fisher Scientific. All samples were separated using a Sciex ExionLC™ AC system and measured using a Sciex 5500 QTRAP® or Sciex 6500+ QTRAP® mass spectrometer. Polar metabolites were analyzed as previously described. For all metabolomics analysis, quality control samples, consisting of a mixture of the analyzed samples, were injected after every 10 injections to control for signal stability. Samples were analyzed via separate negative ionization and positive ionization methods. For negative mode analysis, a Waters Atlantis T3 column (100 Å, 3 μm, 3 mm × 100 mm) with a gradient from 5 mm tributylamine, 5 mm acetic acid in 2% isopropanol, 5% methanol, 93% water (v/v) to 100% isopropanol over 15 min was used. For positive mode analysis, samples were separated on a Phenomenex Kinetex F5 column (100 Å, 2.6 μm, 2.1 mm × 100 mm) column with a gradient from 100% water with 0.1% formic acid to 95% acetonitrile with 0.1% formic acid over 5 min. Each metabolite was measured using at two distinct multiple-reaction monitoring (MRM) signals and a defined retention time. For SCFA analysis samples were separated on a Waters™ Atlantis dC18 column (100Å, 3 µm, 3 mm × 100 mm) with a 6 min gradient from 5% to 80% B with buffer A as 0.1% formic acid in water and B as 0.1% formic acid in methanol. All SCFA were measured using positive ionization using MRMs that featured the characteristic 91 daughter ion from O-BHA derivatization. Identity was confirmed via comparison to previous standards. Bile acid samples were separated on a Phenomenex Kinetex® Polar C18 (100Å, 2.6 µm, 3 mm × 100 mm) using a binary gradient of A: 0.01% acetic acid in water and B: 0.01% acetic acid in methanol. A 20 min gradient from 40% to 100% B was utilized for separation. Samples were detected in negative MRM mode using previously validated MRMs. Internal bile acid standard signals were used to confirm signal identities and retention times. Metabolomic analysis . All signals were integrated using SciexOS 3.1 (AB Sciex Pte. Ltd.). Signals with greater than 50% missing values were discarded and the remaining missing values were replaced with the lowest registered signal value. All signals with a QC coefficient of variance greater than 30% were discarded. Metabolites with multiple MRMs were quantified with the higher signal-to-noise MRM. Filtered datasets were total sum normalized prior to analysis. The SCFA dataset and the two polar metabolomics datasets were scaled and combined using common signal for serine and succinate for the positive mode metabolite method and the SCFA method respectively. A paired t-test was used for all bile acid and metabolite statistics and a Benjamini–Hochberg method for correction for multiple comparisons was imposed where indicated. Fecal Calprotectin (FC) FC was assessed using the Buhlmann Fecal Calprotectin ELISA kit (BÜHLMANN fCAL® ELISA, https://buhlmannlabs.com/buhlmann-fcal-elisa/ ) following the manufacturer’s guidelines. The FCs included all subjects who did not pass Shapiro-Wilks normality, so pre- and post-VSG were compared via a Wilcoxon test. FCs excluding outliers passed normality and were compared pre- and post-VSG via a paired t-test. Urine proteomics Urine samples were analyzed using the SomaScan V4.1 Assay, an aptamer-based quantitative proteomic biomarker discovery platform (SomaLogic; Boulder, CO). The assay was run according to manufacturer’s specifications which includes pH adjustment and buffer exchange by gel filtration prior to normalizing total protein concentration of urine samples to a standard input concentration ( https://somalogic.com/wp-content/uploads/2023/09/D0005009_Rev1_2023-09_SomaScan-7K-v4.1-UrinePre-processing-User-Manual.pdf ). Data was then subjected to the manufacturer’s standard normalization methods, including adaptive normalization by maximum likelihood by SomaLogic. Identified enriched gene sets were determined utilizing the pre-ranked gene-set enrichment analysis (GSEA) algorithm, as implemented in the FGSEA R package. Genes were prioritized based on moderated T statistics derived from the limma model’s relevant coefficient, and enrichment analysis was conducted using the Reactome database, with correction of p values applied for multiple sampling. This analysis can be used to identify significant enrichment of a set of foreground genes or proteins, in predefined gene sets, compared against a reference set. Quality control and initial data processing were performed using an R package ( https://github.com/foocheung/sqs ) and Shiny app. Subjects and samples Children and adolescents undergoing VSG between January 2021 and February 2023 were enrolled in an Institutional Review Board (IRB) approved longitudinal cohort study “Biorepository of Specimens for Pediatric Obesity Research Use” (Children’s National Hospital IRB protocol #Pro00015976) with signed consent and assent (latter for those >11 years). Inclusion criteria for this sub-study were children and adolescents (<19 years) undergoing vertical sleeve gastrectomy at the Children’s National Hospital in Washington DC. To be eligible for surgery, children required a BMI ≥ 120th percentile of 95th percentile for age and sex with an obesity-related comorbidity and/or BMI ≥ 140th percentile. Participants were enrolled up to 2 months prior to their planned VSG. After consent was obtained, a stool and urine collection container were mailed to participants. For stool, participants mailed back a sample aliquoted in an OMNIgene-gut tube and OMNImet-gut tube for DNA and metabolite preservation respectively (DNA Genotek Inc). Wherever possible participants also returned whole stool aliquoted into tubes with glycerol (for bacterial preservation) and whole stool without preservatives. Participants gave a clean catch urine sample mixed with AssayAssure ® Genelock (Sierra Molecular) for protein preservation. All samples were mailed back overnight on icepacks and stored at −80°C until analysis. The same procedure was followed for post-VSG samples. Clinical data was obtained via extraction of medical records from clinical appointments pre-VSG and post-VSG and patient collected data accompanying the samples and was stored in a secure REDCap database. For clinical data analysis, BMI, weight, HbA1c%, LDL, HDL, triglycerides, and ALT passed Shapiro-Wilks normality and were compared pre-VSG and post-VSG via paired t-tests. T2DM, dyslipidemia, and hypertension were compared via McNemar exact tests. Patient involvement Patients were involved in the design and conduct of the trial. We received input from patients in prior trials for the design of convenient sample collection methods that would minimize inconvenience and were suitable for the planned downstream assays. We intend to disseminate the main results to study participants. Microbiome sequencing DNA Extraction : DNA was extracted from human and murine fecal samples in two stages. First, approximately 50 mg of fecal material and 650 μL MBL lysis buffer from the PowerMicrobiome DNA/RNA EP Kit (Qiagen) were added to Lysis Matrix E (LME) tubes (MP Biomedicals). LME tubes were transferred to a Precellys 24 Tissue Homogenizer (Bertin Technologies) and fecal samples were homogenized, centrifuged, with the resultant supernatant transferred to a deep-well 96-well plate. The second stage consisted of DNA isolation from the above supernatant using the MagAttract PowerMicrobiome DNA/RNA EP Kit (Qiagen) on an automated liquid handling system as detailed by the manufacturer (Eppendorf). Shotgun Metagenomic Sequencing . Total gene content of the microbiome was assessed through shotgun metagenomic sequencing. Metagenomic libraries were constructed from 100 ng of DNA as starting material using the Illumina DNA Prep kit. Illumina DNA/RNA UD Indexes were used to add sample-specific sequencing indices to both ends of the libraries. An Agilent 4200 TapeStation system with High Sensitivity D5000 ScreenTape (Agilent Technologies, Inc) was used to verify quality and assess final library size. A positive control (MSA-2002 20 Strain Even Mix Whole-Cell Material (ATCC) and a buffer extraction-negative control were included. Metagenomic libraries were normalized and pooled at an equimolar concentration. Final pools were diluted to 750 pM and sequenced on a NextSeq2000 sequencer using a paired-end (100×100) NextSeq 1000/2000 P2 (200 cycles) kit (Illumina, Inc). Microbiome analysis Data Processing : The quality of raw paired-end sequence reads was assessed with FastQC and MutiQC. Adapters revealed by FastQC were removed using bbtools’ bbduk software. Reads then underwent the Whole-Genome Sequence Assembly 2 (WGSA2) protocol using the Nephele platform (version 2.24.2). In brief, reads were processed with fastp and minimal trimming and filtering by ensuring an average read quality of 10, a trim of the 3’ end of the read at a quality of 15 and trimming of the 5’ end at a Q score of 20, with additional filtering of reads if they were less than 60 bases after trimming. Decontamination was undertaken using Kraken2 with a database containing the human and mouse genome. After adapter trimming and filtering, samples contained between 15 M and 30 M paired-end reads, of which between 8.7 M and 22 M were classified to the bacterial kingdom. Taxonomic identification was performed on the trimmed, error-corrected, and decontaminated reads in Kraken2 with the default RefSeq database. Assembly and Gene Annotation . Within the WGSA2 pipeline, the trimmed, error-corrected and decontaminated reads were assembled into contiguous sequences, or contigs, using metaSPAdes. Reads were recruited back to contigs using bowtie29 and SAMtools to produce information on scaffold coverage and quality. Protein coding regions (CDS) were predicted from assembled scaffolds using prodigal. Predicted CDS regions were processed by EggNog-mapper2 to identify and annotate genes with KEGG Orthology (KO), Enzyme Commission (EC) and Clusters of Orthologous Genes (COG) identifiers. Abundances were calculated using verse to obtain Transcripts Per Million (TPM) at the CDS level and summed to obtain TPM by gene. CAZymes . Carbohydrate Active Enzymes (CAZymes) were annotated from assembled metagenomic scaffolds using the dbCAN software in meta mode with default settings and default databases to obtain eCAMI, HMMER and DIAMOND-based annotations. Annotations were provided at the gene level and merged with predicted gene abundances. Genes were also annotated with taxonomy using Kraken2, creating a table of CAZyme abundances both stratified by taxonomy and unstratified. Gene abundances were normalized to copies per million before analysis. The eCAMI-based CAZyme identification was used in analysis and focused specifically on CAZymes supported by DIAMOND or HMMER when available. Antibiotic Resistance Genes . Antibiotic resistance genes were identified using the Comprehensive Antibiotic Resistance Database with the Resistance Gene Identifier tool v6.0.1, nudging loose hits to strict and including low-quality assemblies with prediction of partial genes. Resulting annotations were processed as for CAZymes. Virome . After assembly with WGSA2, assembled scaffolds and binary alignment maps (BAM) were processed for the presence of viral diversity (ssDNA, dsDNA phage, and giant DNA viruses) using Nephele’s DiscoVir pipeline ( https://nephele.niaid.nih.gov/pipeline_details/discovir/ ). Briefly, geNomad predicted viral genomes and fragments using default confidence parameters as defined by the tool, and VERSE calculated read counts of viral genomes based on BAM files. , Next, CheckV assessed quality and all scaffolds identified as viral were retained for downstream processing and analysis. Viral genomes and fragments greater than 1000 basepairs were clustered with bbtools and MMseqs2 to generated vOTUs which were functionally annotated with DRAM-v with KOfam, Pfam, and Viral Orthologous Group (VOG) databases. Auxiliary metabolic genes were identified with VirSorter2.0 and DRAM-v. Phage hosts were predicted using iPHoP. Statistical analysis . All statistical analysis of the microbiome was performed using R 4.3.0. For taxonomic analysis, reads were filtered to those that aligned to the bacterial kingdom and normalized with rarefaction to 8 million reads per sample. Several measures of alpha diversity were calculated, including Chao1 Richness, Observed Taxa, Evenness, Inverse Simpson, and Shannon Diversity using estimate_richness from phyloseq. For all data types, Bray Curtis and Canberra distance matrices were calculated using phyloseq’s distance function. Alpha diversity was compared between pre-VSG and post-VSG samples using linear mixed-effects models within the lmerTest package. The composition of the microbiome was compared against study covariates with Permutational Analysis of Variance (PERMANOVA) using the adonis2 function from the vegan package. Significant relationships were visualized using Principal Coordinates Analysis (PCoA) ordination in phyloseq and ggplot2. All differential abundance analyses were undertaken with Maaslin2, wherein data were analyzed with linear models after log-transformation. Features were filtered if they exhibited a minimum prevalence of less than 10% and a minimum variance of 0.01. When paired samples were included, Subject ID was provided as a random effect. An FDR-corrected p-value less than 0.2 was considered significant. Co-associated networks of taxa or EC abundances were generated using Weighted Gene Co-Association Network Analysis. In brief, this tool reduces multi-dimensional data to co-associated networks or modules. For taxa-based modules, taxa were included that were found to change nominally after surgery at a false-discovery-corrected p-value of less than 0.5. Blockwise modules were generated using a soft power threshold of 14, an unsigned topology overlap matrix, minimum module size of 20, merge cut height of 0.15, and deepSplit of 3. This process generated eight modules of taxa that were annotated manually by combining the taxonomic information, the average abundance of the taxon across samples, and the module membership, calculated as the correlation of the taxon abundance with the module eigengene. WGCNA modules were also created for EC abundances using similar settings. ECs were included if they changed significantly due to surgery at an FDR P-value of 0.4. Modules were constructed using the same parameters as above except with a soft power threshold of 7. This resulted in four distinct modules of ECs. These were described using the pathway enrichment tool OmePath, where scores used were based on module membership and calculated as described previously. Relationships between module eigengenes and metabolites were identified using linear mixed effects models. Results were visualized using the igraph package 21 if they were significant at an FDR p-value less than 0.05. Directionality of the results was represented by edge color and data type was indicated through color of nodes. This work utilized the computational resources of the NIH HPC Biowulf cluster ( http://hpc.nih.gov .) and the NIAID Locus cluster ( http://locus.niaid.nih.gov ). Metabolomics Metabolite and Lipid Sample Preparation : For all liquid chromatography mass spectrometry (LCMS) methods, LCMS-grade solvents were used. For bile acid analysis 400 µL of homogenized feces was taken from the fecal collection tubes and added to 500 µL of ice-cold methanol. To each sample 5 µL of the Bile Acid SPLASH® (Avanti Polar Lipids Inc.) and 2 µg of butylated hydroxytoluene was added. Samples were agitated via shaking at 4°C for 20 min and then centrifuged at 16k ×g for 20 min. An aliquot of the supernatant was taken directly for liquid LCMS analysis. For short-chain fatty acid (SCFA) and polar metabolomics, a separate 400 µL aliquot of homogenized feces was added to 400 µL of water. Following mixing, 400 µL of chloroform was added. Samples were shaken for 30 min at 4°C and subsequently centrifuged at 16k ×g for 20 min. About 400 µL of the top (aqueous) layer was collected. The aqueous layer was sub-aliquoted for SCFA derivatization or diluted 5× in 50% methanol in water and prepared for LCMS injection. SCFA derivatization . To preserve SCFAs for analysis an aliquot of the aqueous fraction was derivatized with O-benzylhydroxylamine (O-BHA) as previously described with modifications. , The reaction buffer contained 1 M pyridine and 0.5 M hydrochloric acid in water. To 35 µL of sample, 10 µL of 1 M O-BHA in reaction buffer and 10 µL of 1 M 1-Ethyl-3-(3-dimethylaminopropyl) carbodiimide in reaction buffer were added. Samples were derivatized for 2 h at room temperature with constant agitation. Addition of 50 µL of 0.1% formic acid was used to quench the reaction, which eliminated the potential for formate to be measured in these samples. To extract derivatized molecules 400 µL of ethyl acetate was added. The samples were centrifuged at 16k ×g and 4°C for 5 min to induce layering. The upper (organic) layer was collected and dried under vacuum. Samples were resuspended in 300 µL of water for LCMS injection. Liquid chromatography mass spectrometry . Tributylamine and all synthetic molecular references were purchased from Millipore Sigma. LCMS grade water, methanol, isopropanol and acetic acid were purchased through Fisher Scientific. All samples were separated using a Sciex ExionLC™ AC system and measured using a Sciex 5500 QTRAP® or Sciex 6500+ QTRAP® mass spectrometer. Polar metabolites were analyzed as previously described. For all metabolomics analysis, quality control samples, consisting of a mixture of the analyzed samples, were injected after every 10 injections to control for signal stability. Samples were analyzed via separate negative ionization and positive ionization methods. For negative mode analysis, a Waters Atlantis T3 column (100 Å, 3 μm, 3 mm × 100 mm) with a gradient from 5 mm tributylamine, 5 mm acetic acid in 2% isopropanol, 5% methanol, 93% water (v/v) to 100% isopropanol over 15 min was used. For positive mode analysis, samples were separated on a Phenomenex Kinetex F5 column (100 Å, 2.6 μm, 2.1 mm × 100 mm) column with a gradient from 100% water with 0.1% formic acid to 95% acetonitrile with 0.1% formic acid over 5 min. Each metabolite was measured using at two distinct multiple-reaction monitoring (MRM) signals and a defined retention time. For SCFA analysis samples were separated on a Waters™ Atlantis dC18 column (100Å, 3 µm, 3 mm × 100 mm) with a 6 min gradient from 5% to 80% B with buffer A as 0.1% formic acid in water and B as 0.1% formic acid in methanol. All SCFA were measured using positive ionization using MRMs that featured the characteristic 91 daughter ion from O-BHA derivatization. Identity was confirmed via comparison to previous standards. Bile acid samples were separated on a Phenomenex Kinetex® Polar C18 (100Å, 2.6 µm, 3 mm × 100 mm) using a binary gradient of A: 0.01% acetic acid in water and B: 0.01% acetic acid in methanol. A 20 min gradient from 40% to 100% B was utilized for separation. Samples were detected in negative MRM mode using previously validated MRMs. Internal bile acid standard signals were used to confirm signal identities and retention times. Metabolomic analysis . All signals were integrated using SciexOS 3.1 (AB Sciex Pte. Ltd.). Signals with greater than 50% missing values were discarded and the remaining missing values were replaced with the lowest registered signal value. All signals with a QC coefficient of variance greater than 30% were discarded. Metabolites with multiple MRMs were quantified with the higher signal-to-noise MRM. Filtered datasets were total sum normalized prior to analysis. The SCFA dataset and the two polar metabolomics datasets were scaled and combined using common signal for serine and succinate for the positive mode metabolite method and the SCFA method respectively. A paired t-test was used for all bile acid and metabolite statistics and a Benjamini–Hochberg method for correction for multiple comparisons was imposed where indicated. Fecal Calprotectin (FC) FC was assessed using the Buhlmann Fecal Calprotectin ELISA kit (BÜHLMANN fCAL® ELISA, https://buhlmannlabs.com/buhlmann-fcal-elisa/ ) following the manufacturer’s guidelines. The FCs included all subjects who did not pass Shapiro-Wilks normality, so pre- and post-VSG were compared via a Wilcoxon test. FCs excluding outliers passed normality and were compared pre- and post-VSG via a paired t-test. Urine proteomics Urine samples were analyzed using the SomaScan V4.1 Assay, an aptamer-based quantitative proteomic biomarker discovery platform (SomaLogic; Boulder, CO). The assay was run according to manufacturer’s specifications which includes pH adjustment and buffer exchange by gel filtration prior to normalizing total protein concentration of urine samples to a standard input concentration ( https://somalogic.com/wp-content/uploads/2023/09/D0005009_Rev1_2023-09_SomaScan-7K-v4.1-UrinePre-processing-User-Manual.pdf ). Data was then subjected to the manufacturer’s standard normalization methods, including adaptive normalization by maximum likelihood by SomaLogic. Identified enriched gene sets were determined utilizing the pre-ranked gene-set enrichment analysis (GSEA) algorithm, as implemented in the FGSEA R package. Genes were prioritized based on moderated T statistics derived from the limma model’s relevant coefficient, and enrichment analysis was conducted using the Reactome database, with correction of p values applied for multiple sampling. This analysis can be used to identify significant enrichment of a set of foreground genes or proteins, in predefined gene sets, compared against a reference set. Quality control and initial data processing were performed using an R package ( https://github.com/foocheung/sqs ) and Shiny app. Children and adolescents undergoing VSG between January 2021 and February 2023 were enrolled in an Institutional Review Board (IRB) approved longitudinal cohort study “Biorepository of Specimens for Pediatric Obesity Research Use” (Children’s National Hospital IRB protocol #Pro00015976) with signed consent and assent (latter for those >11 years). Inclusion criteria for this sub-study were children and adolescents (<19 years) undergoing vertical sleeve gastrectomy at the Children’s National Hospital in Washington DC. To be eligible for surgery, children required a BMI ≥ 120th percentile of 95th percentile for age and sex with an obesity-related comorbidity and/or BMI ≥ 140th percentile. Participants were enrolled up to 2 months prior to their planned VSG. After consent was obtained, a stool and urine collection container were mailed to participants. For stool, participants mailed back a sample aliquoted in an OMNIgene-gut tube and OMNImet-gut tube for DNA and metabolite preservation respectively (DNA Genotek Inc). Wherever possible participants also returned whole stool aliquoted into tubes with glycerol (for bacterial preservation) and whole stool without preservatives. Participants gave a clean catch urine sample mixed with AssayAssure ® Genelock (Sierra Molecular) for protein preservation. All samples were mailed back overnight on icepacks and stored at −80°C until analysis. The same procedure was followed for post-VSG samples. Clinical data was obtained via extraction of medical records from clinical appointments pre-VSG and post-VSG and patient collected data accompanying the samples and was stored in a secure REDCap database. For clinical data analysis, BMI, weight, HbA1c%, LDL, HDL, triglycerides, and ALT passed Shapiro-Wilks normality and were compared pre-VSG and post-VSG via paired t-tests. T2DM, dyslipidemia, and hypertension were compared via McNemar exact tests. Patients were involved in the design and conduct of the trial. We received input from patients in prior trials for the design of convenient sample collection methods that would minimize inconvenience and were suitable for the planned downstream assays. We intend to disseminate the main results to study participants. DNA Extraction : DNA was extracted from human and murine fecal samples in two stages. First, approximately 50 mg of fecal material and 650 μL MBL lysis buffer from the PowerMicrobiome DNA/RNA EP Kit (Qiagen) were added to Lysis Matrix E (LME) tubes (MP Biomedicals). LME tubes were transferred to a Precellys 24 Tissue Homogenizer (Bertin Technologies) and fecal samples were homogenized, centrifuged, with the resultant supernatant transferred to a deep-well 96-well plate. The second stage consisted of DNA isolation from the above supernatant using the MagAttract PowerMicrobiome DNA/RNA EP Kit (Qiagen) on an automated liquid handling system as detailed by the manufacturer (Eppendorf). Shotgun Metagenomic Sequencing . Total gene content of the microbiome was assessed through shotgun metagenomic sequencing. Metagenomic libraries were constructed from 100 ng of DNA as starting material using the Illumina DNA Prep kit. Illumina DNA/RNA UD Indexes were used to add sample-specific sequencing indices to both ends of the libraries. An Agilent 4200 TapeStation system with High Sensitivity D5000 ScreenTape (Agilent Technologies, Inc) was used to verify quality and assess final library size. A positive control (MSA-2002 20 Strain Even Mix Whole-Cell Material (ATCC) and a buffer extraction-negative control were included. Metagenomic libraries were normalized and pooled at an equimolar concentration. Final pools were diluted to 750 pM and sequenced on a NextSeq2000 sequencer using a paired-end (100×100) NextSeq 1000/2000 P2 (200 cycles) kit (Illumina, Inc). Data Processing : The quality of raw paired-end sequence reads was assessed with FastQC and MutiQC. Adapters revealed by FastQC were removed using bbtools’ bbduk software. Reads then underwent the Whole-Genome Sequence Assembly 2 (WGSA2) protocol using the Nephele platform (version 2.24.2). In brief, reads were processed with fastp and minimal trimming and filtering by ensuring an average read quality of 10, a trim of the 3’ end of the read at a quality of 15 and trimming of the 5’ end at a Q score of 20, with additional filtering of reads if they were less than 60 bases after trimming. Decontamination was undertaken using Kraken2 with a database containing the human and mouse genome. After adapter trimming and filtering, samples contained between 15 M and 30 M paired-end reads, of which between 8.7 M and 22 M were classified to the bacterial kingdom. Taxonomic identification was performed on the trimmed, error-corrected, and decontaminated reads in Kraken2 with the default RefSeq database. Assembly and Gene Annotation . Within the WGSA2 pipeline, the trimmed, error-corrected and decontaminated reads were assembled into contiguous sequences, or contigs, using metaSPAdes. Reads were recruited back to contigs using bowtie29 and SAMtools to produce information on scaffold coverage and quality. Protein coding regions (CDS) were predicted from assembled scaffolds using prodigal. Predicted CDS regions were processed by EggNog-mapper2 to identify and annotate genes with KEGG Orthology (KO), Enzyme Commission (EC) and Clusters of Orthologous Genes (COG) identifiers. Abundances were calculated using verse to obtain Transcripts Per Million (TPM) at the CDS level and summed to obtain TPM by gene. CAZymes . Carbohydrate Active Enzymes (CAZymes) were annotated from assembled metagenomic scaffolds using the dbCAN software in meta mode with default settings and default databases to obtain eCAMI, HMMER and DIAMOND-based annotations. Annotations were provided at the gene level and merged with predicted gene abundances. Genes were also annotated with taxonomy using Kraken2, creating a table of CAZyme abundances both stratified by taxonomy and unstratified. Gene abundances were normalized to copies per million before analysis. The eCAMI-based CAZyme identification was used in analysis and focused specifically on CAZymes supported by DIAMOND or HMMER when available. Antibiotic Resistance Genes . Antibiotic resistance genes were identified using the Comprehensive Antibiotic Resistance Database with the Resistance Gene Identifier tool v6.0.1, nudging loose hits to strict and including low-quality assemblies with prediction of partial genes. Resulting annotations were processed as for CAZymes. Virome . After assembly with WGSA2, assembled scaffolds and binary alignment maps (BAM) were processed for the presence of viral diversity (ssDNA, dsDNA phage, and giant DNA viruses) using Nephele’s DiscoVir pipeline ( https://nephele.niaid.nih.gov/pipeline_details/discovir/ ). Briefly, geNomad predicted viral genomes and fragments using default confidence parameters as defined by the tool, and VERSE calculated read counts of viral genomes based on BAM files. , Next, CheckV assessed quality and all scaffolds identified as viral were retained for downstream processing and analysis. Viral genomes and fragments greater than 1000 basepairs were clustered with bbtools and MMseqs2 to generated vOTUs which were functionally annotated with DRAM-v with KOfam, Pfam, and Viral Orthologous Group (VOG) databases. Auxiliary metabolic genes were identified with VirSorter2.0 and DRAM-v. Phage hosts were predicted using iPHoP. Statistical analysis . All statistical analysis of the microbiome was performed using R 4.3.0. For taxonomic analysis, reads were filtered to those that aligned to the bacterial kingdom and normalized with rarefaction to 8 million reads per sample. Several measures of alpha diversity were calculated, including Chao1 Richness, Observed Taxa, Evenness, Inverse Simpson, and Shannon Diversity using estimate_richness from phyloseq. For all data types, Bray Curtis and Canberra distance matrices were calculated using phyloseq’s distance function. Alpha diversity was compared between pre-VSG and post-VSG samples using linear mixed-effects models within the lmerTest package. The composition of the microbiome was compared against study covariates with Permutational Analysis of Variance (PERMANOVA) using the adonis2 function from the vegan package. Significant relationships were visualized using Principal Coordinates Analysis (PCoA) ordination in phyloseq and ggplot2. All differential abundance analyses were undertaken with Maaslin2, wherein data were analyzed with linear models after log-transformation. Features were filtered if they exhibited a minimum prevalence of less than 10% and a minimum variance of 0.01. When paired samples were included, Subject ID was provided as a random effect. An FDR-corrected p-value less than 0.2 was considered significant. Co-associated networks of taxa or EC abundances were generated using Weighted Gene Co-Association Network Analysis. In brief, this tool reduces multi-dimensional data to co-associated networks or modules. For taxa-based modules, taxa were included that were found to change nominally after surgery at a false-discovery-corrected p-value of less than 0.5. Blockwise modules were generated using a soft power threshold of 14, an unsigned topology overlap matrix, minimum module size of 20, merge cut height of 0.15, and deepSplit of 3. This process generated eight modules of taxa that were annotated manually by combining the taxonomic information, the average abundance of the taxon across samples, and the module membership, calculated as the correlation of the taxon abundance with the module eigengene. WGCNA modules were also created for EC abundances using similar settings. ECs were included if they changed significantly due to surgery at an FDR P-value of 0.4. Modules were constructed using the same parameters as above except with a soft power threshold of 7. This resulted in four distinct modules of ECs. These were described using the pathway enrichment tool OmePath, where scores used were based on module membership and calculated as described previously. Relationships between module eigengenes and metabolites were identified using linear mixed effects models. Results were visualized using the igraph package 21 if they were significant at an FDR p-value less than 0.05. Directionality of the results was represented by edge color and data type was indicated through color of nodes. This work utilized the computational resources of the NIH HPC Biowulf cluster ( http://hpc.nih.gov .) and the NIAID Locus cluster ( http://locus.niaid.nih.gov ). Metabolite and Lipid Sample Preparation : For all liquid chromatography mass spectrometry (LCMS) methods, LCMS-grade solvents were used. For bile acid analysis 400 µL of homogenized feces was taken from the fecal collection tubes and added to 500 µL of ice-cold methanol. To each sample 5 µL of the Bile Acid SPLASH® (Avanti Polar Lipids Inc.) and 2 µg of butylated hydroxytoluene was added. Samples were agitated via shaking at 4°C for 20 min and then centrifuged at 16k ×g for 20 min. An aliquot of the supernatant was taken directly for liquid LCMS analysis. For short-chain fatty acid (SCFA) and polar metabolomics, a separate 400 µL aliquot of homogenized feces was added to 400 µL of water. Following mixing, 400 µL of chloroform was added. Samples were shaken for 30 min at 4°C and subsequently centrifuged at 16k ×g for 20 min. About 400 µL of the top (aqueous) layer was collected. The aqueous layer was sub-aliquoted for SCFA derivatization or diluted 5× in 50% methanol in water and prepared for LCMS injection. SCFA derivatization . To preserve SCFAs for analysis an aliquot of the aqueous fraction was derivatized with O-benzylhydroxylamine (O-BHA) as previously described with modifications. , The reaction buffer contained 1 M pyridine and 0.5 M hydrochloric acid in water. To 35 µL of sample, 10 µL of 1 M O-BHA in reaction buffer and 10 µL of 1 M 1-Ethyl-3-(3-dimethylaminopropyl) carbodiimide in reaction buffer were added. Samples were derivatized for 2 h at room temperature with constant agitation. Addition of 50 µL of 0.1% formic acid was used to quench the reaction, which eliminated the potential for formate to be measured in these samples. To extract derivatized molecules 400 µL of ethyl acetate was added. The samples were centrifuged at 16k ×g and 4°C for 5 min to induce layering. The upper (organic) layer was collected and dried under vacuum. Samples were resuspended in 300 µL of water for LCMS injection. Liquid chromatography mass spectrometry . Tributylamine and all synthetic molecular references were purchased from Millipore Sigma. LCMS grade water, methanol, isopropanol and acetic acid were purchased through Fisher Scientific. All samples were separated using a Sciex ExionLC™ AC system and measured using a Sciex 5500 QTRAP® or Sciex 6500+ QTRAP® mass spectrometer. Polar metabolites were analyzed as previously described. For all metabolomics analysis, quality control samples, consisting of a mixture of the analyzed samples, were injected after every 10 injections to control for signal stability. Samples were analyzed via separate negative ionization and positive ionization methods. For negative mode analysis, a Waters Atlantis T3 column (100 Å, 3 μm, 3 mm × 100 mm) with a gradient from 5 mm tributylamine, 5 mm acetic acid in 2% isopropanol, 5% methanol, 93% water (v/v) to 100% isopropanol over 15 min was used. For positive mode analysis, samples were separated on a Phenomenex Kinetex F5 column (100 Å, 2.6 μm, 2.1 mm × 100 mm) column with a gradient from 100% water with 0.1% formic acid to 95% acetonitrile with 0.1% formic acid over 5 min. Each metabolite was measured using at two distinct multiple-reaction monitoring (MRM) signals and a defined retention time. For SCFA analysis samples were separated on a Waters™ Atlantis dC18 column (100Å, 3 µm, 3 mm × 100 mm) with a 6 min gradient from 5% to 80% B with buffer A as 0.1% formic acid in water and B as 0.1% formic acid in methanol. All SCFA were measured using positive ionization using MRMs that featured the characteristic 91 daughter ion from O-BHA derivatization. Identity was confirmed via comparison to previous standards. Bile acid samples were separated on a Phenomenex Kinetex® Polar C18 (100Å, 2.6 µm, 3 mm × 100 mm) using a binary gradient of A: 0.01% acetic acid in water and B: 0.01% acetic acid in methanol. A 20 min gradient from 40% to 100% B was utilized for separation. Samples were detected in negative MRM mode using previously validated MRMs. Internal bile acid standard signals were used to confirm signal identities and retention times. Metabolomic analysis . All signals were integrated using SciexOS 3.1 (AB Sciex Pte. Ltd.). Signals with greater than 50% missing values were discarded and the remaining missing values were replaced with the lowest registered signal value. All signals with a QC coefficient of variance greater than 30% were discarded. Metabolites with multiple MRMs were quantified with the higher signal-to-noise MRM. Filtered datasets were total sum normalized prior to analysis. The SCFA dataset and the two polar metabolomics datasets were scaled and combined using common signal for serine and succinate for the positive mode metabolite method and the SCFA method respectively. A paired t-test was used for all bile acid and metabolite statistics and a Benjamini–Hochberg method for correction for multiple comparisons was imposed where indicated. FC was assessed using the Buhlmann Fecal Calprotectin ELISA kit (BÜHLMANN fCAL® ELISA, https://buhlmannlabs.com/buhlmann-fcal-elisa/ ) following the manufacturer’s guidelines. The FCs included all subjects who did not pass Shapiro-Wilks normality, so pre- and post-VSG were compared via a Wilcoxon test. FCs excluding outliers passed normality and were compared pre- and post-VSG via a paired t-test. Urine samples were analyzed using the SomaScan V4.1 Assay, an aptamer-based quantitative proteomic biomarker discovery platform (SomaLogic; Boulder, CO). The assay was run according to manufacturer’s specifications which includes pH adjustment and buffer exchange by gel filtration prior to normalizing total protein concentration of urine samples to a standard input concentration ( https://somalogic.com/wp-content/uploads/2023/09/D0005009_Rev1_2023-09_SomaScan-7K-v4.1-UrinePre-processing-User-Manual.pdf ). Data was then subjected to the manufacturer’s standard normalization methods, including adaptive normalization by maximum likelihood by SomaLogic. Identified enriched gene sets were determined utilizing the pre-ranked gene-set enrichment analysis (GSEA) algorithm, as implemented in the FGSEA R package. Genes were prioritized based on moderated T statistics derived from the limma model’s relevant coefficient, and enrichment analysis was conducted using the Reactome database, with correction of p values applied for multiple sampling. This analysis can be used to identify significant enrichment of a set of foreground genes or proteins, in predefined gene sets, compared against a reference set. Quality control and initial data processing were performed using an R package ( https://github.com/foocheung/sqs ) and Shiny app. Mice and ft Germ-free C57BL/6NTac were bred and maintained at the NCI or NIAID Gnotobiotic animal facility. Mice were screened by microbiological culture and 16S rRNA PCR to ensure their germ-free status. Mice were fed a standard irradiated 5KAI diet (LabDiet). Pre-VSG and post-VSG fecal samples from two human subjects stored in 20% glycerol at −80°C were used for fecal microbiota transplant (FMT) into 5–6-week-old male mice. Each mouse was orally inoculated with 200 μl of 35–60 mg stool slurry resuspended in sterile pre-reduced PBS in the anaerobic chamber. Stool suspension was stored in a Hungate tube. Each mouse was housed in a separate cage in biocontainment racks in the same facility following inoculation. Body weights and feed consumption were measured weekly and stool samples were collected for sequencing. Experiments were performed independently for each human subject and 4–6 mice were inoculated with either pre-VSG or post-VSG stool samples for each subject. All procedures were performed in accordance with approved animal study proposals by NCI or NIAID Animal Care and Use Committees. Glucose tolerance test Mice were fasted overnight for 12–14 h. Glucose (2 g/kg body weight) was administered intraperitoneally, and glucose measurements were performed via tail bleeds using a glucometer (DSS Precision Xtra) before and at 15, 30, 60, 90, and 120 mins after glucose administration. Micro-computed tomography (Micro-ct) QuantumGX scanner was used to obtain in-vivo high-resolution micro-CT imaging. Mice were anesthetized using 2.5% isoflurane, eyes protected with Artificial Tears ointment and transferred onto the imaging bed and maintained on 2% isoflurane during the imaging process. Images were obtained in 3 slices at 70 mm magnification and reconstructed. Adipose tissue for each mouse was quantified and analyzed using Analyze14 software. Intra-abdominal and subcutaneous fat were isolated using a threshold range of approximately −300 to −50 hounsfield units. Tissue harvest and isolation of cells from large intestine and mesenteric lymph nodes Immediately after micro-CT imaging animals were euthanized with CO 2 and blood was collected by cardiac exsanguination. Subcutaneous fat from the abdominal region, epididymal fat pads, and liver were harvested and weighed. Mesenteric lymph nodes were harvested, and cells were isolated by passing through a 70-μm cell strainer, centrifuged at 1500 rpm for 5 mins for flow cytometry analysis. Large intestines (LI, cecum and colon) were collected and placed on ice-cold complete media (RPMI 1640 supplemented with 20 mm HEPES, 2 mm L-glutamine, 1 mm sodium pyruvate, 1 mm nonessential amino acids, 50 mm β-mercaptoethanol, 100 U/ml penicillin and 100 mg/ml streptomycin) + 3% fetal bovine serum (FBS). The tissue was then opened longitudinally, fecal contents removed and cut into 1–2 cm pieces, and incubated in 20 ml of complete media + 3% FBS +5 mm EDTA +0.145 mg/ml DL-dithiothreitol for 20 mins at 37°C and 5% CO 2 with shaking. To remove epithelial cells, LI were strained and vigorously shaken in a 50 ml tube containing 10 ml complete media +2 mm EDTA, thrice. LI were then finely chopped and digested with 10 ml of digestion media (complete media +0.1 mg/ml Liberase TL, Roche + 0.05% DNase I, Sigma) for 25 mins at 37°C and 5% CO 2 with shaking. The digestion was stopped by adding 10 ml complete media + 3% FBS. Digested tissues were then passed through 100-μm cell strainers, centrifuged at 1500 rpm for 5 mins, followed by straining through 40-μm cell strainers and centrifugation at 1500 rpm for 5 mins. Isolated lamina propria cells were then resuspended in complete media + 10% FBS for flow cytometry analysis. Spectral flow cytometry Isolated single-cell suspensions were assessed for cytokine production potential by stimulation with Cell Stimulation cocktail 500X (Thermo Fisher Scientific) prepared in complete media + 10% FBS for 2.5 hours at 37°C. To assess lymphoid cells, cells were incubated with Zombie NIR Fixable Viability Dye for 15 mins at room temperature. Cells were then washed and incubated with a cocktail of fluorescently labeled antibodies prepared in PBS + 1% FBS + 10% Brilliant Stain Buffer + 10% TruStain FcX for 25 mins in the dark at 4°C: anti-CD45, anti-TCRβ, anti-CD44, anti-CD90.2, anti-CD8β, anti-CD4, anti-TCRγδ and anti-NK1.1. Cells were then fixed/permeabilized in eBioscience FoxP3 Fixation/Permeabilization Solution kit overnight at 4°C and stained with a cocktail of fluorescently labeled antibodies against intracellular antigens prepared in eBioscience intracellular staining buffer: anti-FoxP3, anti-GATA3, anti-RORγt, anti-Tbet, anti-IFNγ, anti-IL17A and anti-IL22. All samples were collected on an Aurora spectral cytometer (Cytek). Spectral unmixing was performed using single-control controls using cells from corresponding tissues or UltraComp eBeads (Invitrogen) and data were analyzed using FlowJo version 10. Serum metabolic hormones measurement Mouse serum concentrations of metabolic hormones were assessed using the Hormone Exp Panel kit (Millipore, MMHE-44K-06) according to the manufacturer’s protocol and measured using Luminex MAGPIX Instrument (Bio-Rad). Data were analyzed using GraphPad Prism 9.0 (Graph Pad software, La Jolla, CA, USA) with unpaired t-tests. Differences were considered to be statistically significant when p < 0.05. Microbiome sequencing:16S ribosomal RNA Gene Sequencing DNA extraction was as per human samples. Microbiome composition was assessed via dual-index amplification of the V4 region of the 16S ribosomal RNA gene (16S rRNA). This method used the V4 16S rRNA 515F and 806 R primers with individual sample-specific indices and Illumina sequencing adapters appended as previously described. The V4 region was amplified using: 5 μM of F/R primers, 1X Phusion High-Fidelity DNA Polymerase (New England Biolabs) and 100 ng of DNA as starting material. PCR conditions for amplification were as follows: initial template denaturation at 98°C for 60 s; 25 cycles of denaturation at 98°C for 10 s, primer annealing 55°C for 30 s, and template extension at 72°C 60 s; with a final template extension at 72°C for 5 min. AMPure XP beads (Beckman-Coulter) at a 1:1 ratio with the above PCR reaction were used to isolate final PCR products. Final 16S rRNA V4 libraries were quantified using the KAPA qPCR Library Quantification Kit (Kapa Biosystems) and pooled at an equimolar concentration. Pools were normalized to 8 pM, spiked-in with 15% phiX control library (Illumina) to add sequence diversity, and sequenced on the Illumina MiSeq instrument utilizing the 600 cycle Paired-End (250×250) Reagent Plate with the addition of 16S V4 rRNA-specific sequencing primers. Microbiome analysis Samples collected from the murine study at 6 weeks as well as the human inoculum (pre-VSG and post-VSG samples from two subjects) underwent 16S rRNA sequencing on the Illumina MiSeq. The resulting sequences were reviewed for quality using FastQC and multiQC through Nephele’s microbiome analysis platform. Sequences were trimmed to 210 bases on the forward reads and 180 bases on the reverse read, and those with a maximum expected error greater than two were filtered out, reads then underwent Divisive Amplicon Denoising Algorithm 2 (DADA2) utilizing Nephele’s microbiome analysis platform. After the identification of sequence variants through denoising, they were checked for chimeras and assigned taxonomy up to the species level using the SILVA database. If a sequence variant aligned to multiple species with 100% identity, all species were listed. Once reads were agglomerated into a table of counts, reads from negative controls were subtracted from samples to be conservative about potential sources of contamination. The sequence variant table was rarefied to 70,000 reads per sample after reviewing rarefaction curves and sequences available per sample. Alpha diversity metrics, including Chao1 richness, Evenness, Inverse Simpson and Shannon diversity values, as well as a Bray Curtis distance matrix, were calculated using phyloseq. Linear models determined differences in alpha diversity, and vegan’s PERMANOVA utilizing marginal adjustment for the subject that served as the inoculum source identified differences in community composition between the gut microbiome of pre-VSG and post-VSG mice. Multi-Omic analysis DIABLO from the mixOmics package implemented sparse Partial Least Squares (sPLS) to perform discriminant analysis of multi-omic data. Extensive k-fold cross validation and leave-one-out analysis were performed but did not provide stable estimates to obtain an optimal number of features. Therefore, the top 10 most discriminant features from each data type and human inoculum sources that differentiated pre-VSG mice from post-VSG mice were selected for inclusion in the combined visualization of microbiota and flow data sources. Germ-free C57BL/6NTac were bred and maintained at the NCI or NIAID Gnotobiotic animal facility. Mice were screened by microbiological culture and 16S rRNA PCR to ensure their germ-free status. Mice were fed a standard irradiated 5KAI diet (LabDiet). Pre-VSG and post-VSG fecal samples from two human subjects stored in 20% glycerol at −80°C were used for fecal microbiota transplant (FMT) into 5–6-week-old male mice. Each mouse was orally inoculated with 200 μl of 35–60 mg stool slurry resuspended in sterile pre-reduced PBS in the anaerobic chamber. Stool suspension was stored in a Hungate tube. Each mouse was housed in a separate cage in biocontainment racks in the same facility following inoculation. Body weights and feed consumption were measured weekly and stool samples were collected for sequencing. Experiments were performed independently for each human subject and 4–6 mice were inoculated with either pre-VSG or post-VSG stool samples for each subject. All procedures were performed in accordance with approved animal study proposals by NCI or NIAID Animal Care and Use Committees. Mice were fasted overnight for 12–14 h. Glucose (2 g/kg body weight) was administered intraperitoneally, and glucose measurements were performed via tail bleeds using a glucometer (DSS Precision Xtra) before and at 15, 30, 60, 90, and 120 mins after glucose administration. QuantumGX scanner was used to obtain in-vivo high-resolution micro-CT imaging. Mice were anesthetized using 2.5% isoflurane, eyes protected with Artificial Tears ointment and transferred onto the imaging bed and maintained on 2% isoflurane during the imaging process. Images were obtained in 3 slices at 70 mm magnification and reconstructed. Adipose tissue for each mouse was quantified and analyzed using Analyze14 software. Intra-abdominal and subcutaneous fat were isolated using a threshold range of approximately −300 to −50 hounsfield units. Immediately after micro-CT imaging animals were euthanized with CO 2 and blood was collected by cardiac exsanguination. Subcutaneous fat from the abdominal region, epididymal fat pads, and liver were harvested and weighed. Mesenteric lymph nodes were harvested, and cells were isolated by passing through a 70-μm cell strainer, centrifuged at 1500 rpm for 5 mins for flow cytometry analysis. Large intestines (LI, cecum and colon) were collected and placed on ice-cold complete media (RPMI 1640 supplemented with 20 mm HEPES, 2 mm L-glutamine, 1 mm sodium pyruvate, 1 mm nonessential amino acids, 50 mm β-mercaptoethanol, 100 U/ml penicillin and 100 mg/ml streptomycin) + 3% fetal bovine serum (FBS). The tissue was then opened longitudinally, fecal contents removed and cut into 1–2 cm pieces, and incubated in 20 ml of complete media + 3% FBS +5 mm EDTA +0.145 mg/ml DL-dithiothreitol for 20 mins at 37°C and 5% CO 2 with shaking. To remove epithelial cells, LI were strained and vigorously shaken in a 50 ml tube containing 10 ml complete media +2 mm EDTA, thrice. LI were then finely chopped and digested with 10 ml of digestion media (complete media +0.1 mg/ml Liberase TL, Roche + 0.05% DNase I, Sigma) for 25 mins at 37°C and 5% CO 2 with shaking. The digestion was stopped by adding 10 ml complete media + 3% FBS. Digested tissues were then passed through 100-μm cell strainers, centrifuged at 1500 rpm for 5 mins, followed by straining through 40-μm cell strainers and centrifugation at 1500 rpm for 5 mins. Isolated lamina propria cells were then resuspended in complete media + 10% FBS for flow cytometry analysis. Isolated single-cell suspensions were assessed for cytokine production potential by stimulation with Cell Stimulation cocktail 500X (Thermo Fisher Scientific) prepared in complete media + 10% FBS for 2.5 hours at 37°C. To assess lymphoid cells, cells were incubated with Zombie NIR Fixable Viability Dye for 15 mins at room temperature. Cells were then washed and incubated with a cocktail of fluorescently labeled antibodies prepared in PBS + 1% FBS + 10% Brilliant Stain Buffer + 10% TruStain FcX for 25 mins in the dark at 4°C: anti-CD45, anti-TCRβ, anti-CD44, anti-CD90.2, anti-CD8β, anti-CD4, anti-TCRγδ and anti-NK1.1. Cells were then fixed/permeabilized in eBioscience FoxP3 Fixation/Permeabilization Solution kit overnight at 4°C and stained with a cocktail of fluorescently labeled antibodies against intracellular antigens prepared in eBioscience intracellular staining buffer: anti-FoxP3, anti-GATA3, anti-RORγt, anti-Tbet, anti-IFNγ, anti-IL17A and anti-IL22. All samples were collected on an Aurora spectral cytometer (Cytek). Spectral unmixing was performed using single-control controls using cells from corresponding tissues or UltraComp eBeads (Invitrogen) and data were analyzed using FlowJo version 10. Mouse serum concentrations of metabolic hormones were assessed using the Hormone Exp Panel kit (Millipore, MMHE-44K-06) according to the manufacturer’s protocol and measured using Luminex MAGPIX Instrument (Bio-Rad). Data were analyzed using GraphPad Prism 9.0 (Graph Pad software, La Jolla, CA, USA) with unpaired t-tests. Differences were considered to be statistically significant when p < 0.05. DNA extraction was as per human samples. Microbiome composition was assessed via dual-index amplification of the V4 region of the 16S ribosomal RNA gene (16S rRNA). This method used the V4 16S rRNA 515F and 806 R primers with individual sample-specific indices and Illumina sequencing adapters appended as previously described. The V4 region was amplified using: 5 μM of F/R primers, 1X Phusion High-Fidelity DNA Polymerase (New England Biolabs) and 100 ng of DNA as starting material. PCR conditions for amplification were as follows: initial template denaturation at 98°C for 60 s; 25 cycles of denaturation at 98°C for 10 s, primer annealing 55°C for 30 s, and template extension at 72°C 60 s; with a final template extension at 72°C for 5 min. AMPure XP beads (Beckman-Coulter) at a 1:1 ratio with the above PCR reaction were used to isolate final PCR products. Final 16S rRNA V4 libraries were quantified using the KAPA qPCR Library Quantification Kit (Kapa Biosystems) and pooled at an equimolar concentration. Pools were normalized to 8 pM, spiked-in with 15% phiX control library (Illumina) to add sequence diversity, and sequenced on the Illumina MiSeq instrument utilizing the 600 cycle Paired-End (250×250) Reagent Plate with the addition of 16S V4 rRNA-specific sequencing primers. Samples collected from the murine study at 6 weeks as well as the human inoculum (pre-VSG and post-VSG samples from two subjects) underwent 16S rRNA sequencing on the Illumina MiSeq. The resulting sequences were reviewed for quality using FastQC and multiQC through Nephele’s microbiome analysis platform. Sequences were trimmed to 210 bases on the forward reads and 180 bases on the reverse read, and those with a maximum expected error greater than two were filtered out, reads then underwent Divisive Amplicon Denoising Algorithm 2 (DADA2) utilizing Nephele’s microbiome analysis platform. After the identification of sequence variants through denoising, they were checked for chimeras and assigned taxonomy up to the species level using the SILVA database. If a sequence variant aligned to multiple species with 100% identity, all species were listed. Once reads were agglomerated into a table of counts, reads from negative controls were subtracted from samples to be conservative about potential sources of contamination. The sequence variant table was rarefied to 70,000 reads per sample after reviewing rarefaction curves and sequences available per sample. Alpha diversity metrics, including Chao1 richness, Evenness, Inverse Simpson and Shannon diversity values, as well as a Bray Curtis distance matrix, were calculated using phyloseq. Linear models determined differences in alpha diversity, and vegan’s PERMANOVA utilizing marginal adjustment for the subject that served as the inoculum source identified differences in community composition between the gut microbiome of pre-VSG and post-VSG mice. DIABLO from the mixOmics package implemented sparse Partial Least Squares (sPLS) to perform discriminant analysis of multi-omic data. Extensive k-fold cross validation and leave-one-out analysis were performed but did not provide stable estimates to obtain an optimal number of features. Therefore, the top 10 most discriminant features from each data type and human inoculum sources that differentiated pre-VSG mice from post-VSG mice were selected for inclusion in the combined visualization of microbiota and flow data sources. Participants and clinical data Twelve participants provided paired stool samples within the 8 weeks (mean 2 weeks) prior to VSG (pre-VSG) and follow-up stool samples 3–7 months (mean 5 months) after VSG (post-VSG). The mean age at VSG was 15 years (range 10–18 years), 8/12 participants were female and 9/12 Black or African American . Notably, 2 participants were identical twins. At VSG, subjects had a mean body mass index (BMI) of 48.7 kg/m 2 which decreased to 39.9 kg/m 2 ( p < 0.0001) post-VSG . Total body weight loss (TBWL) averaged 17.8% (range 5.9%−32.9%). 8/12 participants had T2DM or prediabetes pre-VSG with a reduction to 0/12 post-VSG ( p = 0.0078). 7/12 participants had elevated alanine aminotransferase (ALT) pre-VSG, indicating a high likelihood of MASLD; only 1/12 participants had a liver biopsy which showed metabolic dysfunction-associated steatohepatitis. Overall, mean ALT decreased from 28.2 U/L to 15.7 U/L ( p = 0.0025) with VSG. 4/12 participants had dyslipidemia with elevated low-density lipoprotein (LDL) pre-VSG compared to 1/12 post-VSG, with an increase in high-density lipoprotein (HDL) cholesterol post-VSG from an average of 41.3 mg/dL to 49.0 mg/dL ( p = 0.0317). Of note, every participant was prescribed daily omeprazole for 2 weeks post-VSG and one participant remained on omeprazole at the time of the follow-up stool collection. Stool microbiome changes with VSG Alpha and beta diversity Pre-VSG and post-VSG stool samples underwent shotgun metagenomic sequencing. Bacterial diversity increased post-VSG (Shannon p = 0.047, Inv Simpson p = 0.04, Evenness ( p = 0.042) ). Significant changes in microbiome composition (beta diversity) were seen using the Canberra distance, which places more weight on lower abundance species ( p = 0.015, ) but not Bray Curtis distance. Pre-VSG, alpha diversity was significantly lower in those with diabetes compared with pre-diabetes and no diabetes (Supplementary Figure 1a) and microbiome composition also differed (Supplementary Figure 1b). Post-VSG, there were no differences in alpha or beta diversity between those who previously had diabetes and the other subjects. There were no significant differences between changes in alpha and beta diversity and other clinical parameters. Taxonomic shifts There was a significant enrichment of 76 bacterial taxa post-VSG. Notably, the top 18 species enriched post-VSG were from the Streptococcus and Actinomyces genera ( . Supplementary Table 1). This included an enrichment of Streptococcus salivarius, Streptococcus vestibularis, Streptococcus parasanguinis, Actinomyces oris and Actinomyces oral taxon , all of which are commonly associated with the oral cavity. No individual taxa are significantly correlated with clinical characteristics. Carbohydrate-Active Enzymes (CAZymes) CAZymes were examined due to their role in influencing host metabolism. While the overall composition of CAZymes only showed a moderate change with VSG (Canberra PERMANOVA R2 = 0.049, p = 0.056, ), five specific CAZymes exhibited significant enrichment post-VSG ( , Supplementary Table 2). Many of the CAZyme enrichments post-VSG were associated with Streptococcus species . In addition, Glycoside Hydrolase 13 (GH13)+Carbohydrate Binding Module 20 (CBM 20) significantly associated with increased TBWL (q = 0.002, Supplementary Table 2). Antibiotic resistance genes (ARGs) ARGs were examined using the RGI CARD database. Four ARGs were enriched post-VSG: qacJ, tetA (46), tetB(46) and vanY in vanF cluster . These resistance genes, especially tetAB(46), were primarily found in contigs belonging to several Streptococcus species enriched post-VSG . Conserved functional enrichments There were no functional pathway-level differences with VSG. Therefore, reads were aligned with Enzyme Commission (EC) gene annotations to identify more refined gene function differences. There was a moderate change in lower-abundance genes post-VSG (Canberra R2 = 0.05, p = 0.059, Supplementary Figure 2a), most notably with enrichment of EC 1.1.1.105, an all-trans retinol dehydrogenase gene, from the oxidoreductases class (Supplementary Figure 2b, Supplementary Table 3). In addition, these enriched ECs formed three distinct modules of co-associated genes (blue, brown and turquoise) that increased significantly post-VSG (Supplementary Figure 2c-d), driven by different bacterial genera (Supplementary Figure 2e). ECs in the blue module were primarily from mevalonate, hemiterpene biosynthesis and heme biosynthesis pathways; those in the brown module focused on sugar acid degradation, and the turquoise module contained several tRNA synthetases (Supplementary Table 3). Bacteriophage and viral composition When looking at the DNA virome, there were no significant differences observed with VSG in taxonomy, host taxonomy, taxonomic diversity, taxonomic composition, viral protein family diversity or composition (Supplementary Figure 3a-f). As most DNA viruses in the gut microbiome are phages, this suggests phage-containing bacteria did not significantly change with VSG despite large overall bacterial changes. Stool metabolome changes with VSG Post-VSG stool displayed higher ratios of secondary to primary bile acids compared to pre-VSG stool . Notably, this pattern was broad and included all ratios for which the corresponding precursor primary bile acid and product secondary bile acid were detected. Amongst polar metabolites, only the elevation of citrulline post-VSG passed a false discovery rate cutoff of 10% . However, several trends known to be important to gut health were observed . These included a decrease in acyl-carnitines, an increase in neurotransmitters known to be directly microbially produced (GABA, dopamine, and histamine), and increases in redox cofactor metabolites. No changes in SCFAs were observed. Polar metabolites were related to a range of demographic and clinical factors (Supplementary Figure 4, Supplementary Table 4). Several amino acid metabolites (purple) positively associated with HDL, and negatively associated with hemoglobin A1C (HbA1c) pre-VSG. In addition, several metabolites correlated negatively with BMI pre-VSG, including those in Redox and Co-Enzyme metabolite (CoA, FMN, NADH), Glycolysis (G1P, Glycerol-3P), and Nucleotide (UMP, AMP) classes. Post-VSG, several metabolites associated with the degree of TBWL, including neurotransmitters such as GABA and Dopamine, which also correlated negatively with triglyceride levels. Citrulline did not correlate with any factors pre-VSG or post-VSG. Stool microbiome and metabolite correlations with VSG To reduce the dimensionality of the microbiota, taxa were agglomerated into seven co-associated networks. Each increased in abundance post-VSG except for the red network, which exhibited a trending decrease ( p = 0.087; Supplementary Figure 5(a,b), Supplementary Table 5). Changes in these networks were correlated with changes in polar metabolites which revealed that the Bacteroides and the Alistipes/Akkermansia/Actinomyces prominent group both increased in concert with several SCFAs including butyrate, isovalerate, and isobutyrate (Supplementary Figure 5c, Supplementary Table 5). Conversely, the Streptococcus prominent group had the fewest correlations, only exhibiting positive correlations with isocitrate and cytidine and a negative correlation with urate. FT into germ-free mice and microbiome diversity Next, to assess whether the microbiome/metabolome changes seen with VSG had a phenotypic effect, germ-free mice were inoculated with pre-VSG or post-VSG stool from the same participant . Two human participants were chosen who had both pre-VSG and post-VSG stool stored adequately in glycerol to preserve bacteria viability. Of note, one subject had the second-to-greatest TBWL and the second had prediabetes pre-VSG which resolved post-VSG. Using 16S rRNA gene sequencing, 6 weeks post-FT, mice transplanted with post-VSG stool had a higher alpha diversity and different bacterial composition (PERMANOVA Bray Curtis R2 = 0.114, p = 0.006, adjusted for human inoculum source, ) compared to mice that received pre-VSG stool, which was similar to the human stool microbiota findings. The mouse microbiome samples did also show significant separation based on the human subject used as the inoculum (PERMANOVA Bray Curtis R2 = 0.495, p = 0.001, adjusted for FT timepoint). Phenotype changes in germ-free mice Phenotypic differences resulting from FT with pre-VSG and post-VSG stools were compared. Six weeks post-FT, there was no difference in body weight, food consumption, intraperitoneal glucose tolerance tests, nor tissue weights (epididymal fat, subcutaneous fat and liver weights) between mice that received pre-VSG versus post-VSG stool ( respectively). Consistently, micro-computed tomography revealed no differences in subcutaneous or intraabdominal adipose tissue volume between the two groups . There was a significant increase in serum resistin in mice that received post-VSG stool compared to pre-VSG stool ( p = 0.047), but no differences in other metabolic hormones (GLP-1, insulin, leptin, gastric inhibitory polypeptide, ). Immune and inflammatory changes in germ-free mice Immune changes in the murine models were assessed in the large intestine lamina propria and mesenteric lymph nodes using flow cytometry (Supplementary Figure 6a). Of note, mesenteric lymph node data were only available from one set of mice pre- and post-VSG FT. In the large intestine, there was a significant decrease in γδ T cells, a non-significant increase in CD4 + T cells and decrease in CD8 + T and NK cells in post-VSG mice (Supplementary Figure 6b). More importantly, there was a significant increase in the number and proportion of Th17 cells, along with a significant decrease in GATA3 + regulatory T cell (Treg) proportion in the large intestine and mesenteric lymph nodes of post-VSG mice ( , Supplementary Figure 6c), with one participant (pat_10) displaying a more pronounced phenotype of the two. Th17 cytokines (IL-17A and IL-22) were also increased in the mesenteric lymph node of the post-VSG mice of this participant (Supplementary Figure 6d) but did not reach significance in the colon. While Th17 cells and cytokines have been associated with both determinantal and beneficial roles, an increase in Th17 cells and a corresponding decrease in regulatory T cells suggests an inflammatory potential in the individuals studied, that warrants further investigation. Sparse Partial Least Squares analysis of the most discriminatory immune and metabolic readouts and microbial taxa showed a distinct separation of pre- and post-VSG parameters indicating a direct or indirect association between the microbiota and observed immune milieu. Notably, the increased Th17 phenotype observed in the post-VSG mice clustered with increases in relative abundances of microbial taxa belonging to Ruminococcaceae, Erysipelatoclostridiaceae, and Monoglobus. Conversely, the increased Treg populations in the pre-VSG mice were associated with increased levels of the taxa Parabacteroides (Supplementary Figure 6e). Inflammatory changes in humans with VSG Given the inflammatory phenotype post-VSG observed in the murine model, colonic inflammatory changes were assessed in available human stool samples by fecal calprotectin (FC). 8/12 participants had stool frozen without preservatives to allow for FC assessment. 6/8 individuals had an increase in FC to a clinically elevated level post-VSG (>120ug/g, ). Of these 6 participants, which included the 2 participants used for mouse studies, the mean FC pre-VSG increased from 47ug/g to 497ug/g post-VSG ( p = 0.0016, ). 2/8 patients with an elevated FC level pre-VSG had a decrease post-VSG. There was no correlation between length of time from surgery and FC levels. Subjects with raised FC post-VSG did not report typical clinical symptoms associated with high levels such as diarrhea. Additionally, the microbiome of subjects with increased FC also exhibited increased post-VSG microbiota similarity compared to pre-VSG where samples were more distinct (p for interaction = 0.016, Supplementary Figure 7a-b). No individual taxa correlated with FC increases. To assess systemic immunity/inflammation changes with VSG, urine samples were utilized due to the unavailability of blood samples. Proteomic analysis was performed using the SomaScan V4.1 Assay, an aptamer-based assay optimized for peripheral blood but which has been used with urine by prior studies in which a subset of the proteins found to be of interest were validated by orthogonal assays. From the overall 7000 analyte panel, a subset of 1856 proteins were identified which showed (Supplementary Figure 8a). The use of SomaLogic to assay urine was further supported by detection of differential expression for proteins previously implicated in inflammatory bowel disease (IBD) when comparing cases to controls (Supplementary Figure 8b), and observation of correlation between protein levels detected in urine and peripheral blood for most of the proteins detected to be differentially expressed in IBD (Supplementary Figure 8c). When the 1856 proteins which could be assayed in urine were compared between paired pre-VSG and post-VSG urine samples for 6 subjects, no individual proteins differed significantly. However, 17 pathways exhibited significant positive enrichment post-VSG (FDR < 0.05, ). The top enriched pathways included four pathways involved in immune/inflammation regulation: IRAK2 mediated activation of TAK1 complex; TICAM1, TRAF6-dependent induction of TAK1 complex; IRAK1 recruits IKK complex; and IRAK1 recruits IKK complex upon TLR7/8 or 9 stimulation. Enrichment in all four pathways was driven by the same set of three leading edge proteins related to ubiquitin, and the increased expression observed post-VSG was conserved across participants with one exception . Twelve participants provided paired stool samples within the 8 weeks (mean 2 weeks) prior to VSG (pre-VSG) and follow-up stool samples 3–7 months (mean 5 months) after VSG (post-VSG). The mean age at VSG was 15 years (range 10–18 years), 8/12 participants were female and 9/12 Black or African American . Notably, 2 participants were identical twins. At VSG, subjects had a mean body mass index (BMI) of 48.7 kg/m 2 which decreased to 39.9 kg/m 2 ( p < 0.0001) post-VSG . Total body weight loss (TBWL) averaged 17.8% (range 5.9%−32.9%). 8/12 participants had T2DM or prediabetes pre-VSG with a reduction to 0/12 post-VSG ( p = 0.0078). 7/12 participants had elevated alanine aminotransferase (ALT) pre-VSG, indicating a high likelihood of MASLD; only 1/12 participants had a liver biopsy which showed metabolic dysfunction-associated steatohepatitis. Overall, mean ALT decreased from 28.2 U/L to 15.7 U/L ( p = 0.0025) with VSG. 4/12 participants had dyslipidemia with elevated low-density lipoprotein (LDL) pre-VSG compared to 1/12 post-VSG, with an increase in high-density lipoprotein (HDL) cholesterol post-VSG from an average of 41.3 mg/dL to 49.0 mg/dL ( p = 0.0317). Of note, every participant was prescribed daily omeprazole for 2 weeks post-VSG and one participant remained on omeprazole at the time of the follow-up stool collection. Alpha and beta diversity Pre-VSG and post-VSG stool samples underwent shotgun metagenomic sequencing. Bacterial diversity increased post-VSG (Shannon p = 0.047, Inv Simpson p = 0.04, Evenness ( p = 0.042) ). Significant changes in microbiome composition (beta diversity) were seen using the Canberra distance, which places more weight on lower abundance species ( p = 0.015, ) but not Bray Curtis distance. Pre-VSG, alpha diversity was significantly lower in those with diabetes compared with pre-diabetes and no diabetes (Supplementary Figure 1a) and microbiome composition also differed (Supplementary Figure 1b). Post-VSG, there were no differences in alpha or beta diversity between those who previously had diabetes and the other subjects. There were no significant differences between changes in alpha and beta diversity and other clinical parameters. Taxonomic shifts There was a significant enrichment of 76 bacterial taxa post-VSG. Notably, the top 18 species enriched post-VSG were from the Streptococcus and Actinomyces genera ( . Supplementary Table 1). This included an enrichment of Streptococcus salivarius, Streptococcus vestibularis, Streptococcus parasanguinis, Actinomyces oris and Actinomyces oral taxon , all of which are commonly associated with the oral cavity. No individual taxa are significantly correlated with clinical characteristics. Carbohydrate-Active Enzymes (CAZymes) CAZymes were examined due to their role in influencing host metabolism. While the overall composition of CAZymes only showed a moderate change with VSG (Canberra PERMANOVA R2 = 0.049, p = 0.056, ), five specific CAZymes exhibited significant enrichment post-VSG ( , Supplementary Table 2). Many of the CAZyme enrichments post-VSG were associated with Streptococcus species . In addition, Glycoside Hydrolase 13 (GH13)+Carbohydrate Binding Module 20 (CBM 20) significantly associated with increased TBWL (q = 0.002, Supplementary Table 2). Antibiotic resistance genes (ARGs) ARGs were examined using the RGI CARD database. Four ARGs were enriched post-VSG: qacJ, tetA (46), tetB(46) and vanY in vanF cluster . These resistance genes, especially tetAB(46), were primarily found in contigs belonging to several Streptococcus species enriched post-VSG . Conserved functional enrichments There were no functional pathway-level differences with VSG. Therefore, reads were aligned with Enzyme Commission (EC) gene annotations to identify more refined gene function differences. There was a moderate change in lower-abundance genes post-VSG (Canberra R2 = 0.05, p = 0.059, Supplementary Figure 2a), most notably with enrichment of EC 1.1.1.105, an all-trans retinol dehydrogenase gene, from the oxidoreductases class (Supplementary Figure 2b, Supplementary Table 3). In addition, these enriched ECs formed three distinct modules of co-associated genes (blue, brown and turquoise) that increased significantly post-VSG (Supplementary Figure 2c-d), driven by different bacterial genera (Supplementary Figure 2e). ECs in the blue module were primarily from mevalonate, hemiterpene biosynthesis and heme biosynthesis pathways; those in the brown module focused on sugar acid degradation, and the turquoise module contained several tRNA synthetases (Supplementary Table 3). Bacteriophage and viral composition When looking at the DNA virome, there were no significant differences observed with VSG in taxonomy, host taxonomy, taxonomic diversity, taxonomic composition, viral protein family diversity or composition (Supplementary Figure 3a-f). As most DNA viruses in the gut microbiome are phages, this suggests phage-containing bacteria did not significantly change with VSG despite large overall bacterial changes. Pre-VSG and post-VSG stool samples underwent shotgun metagenomic sequencing. Bacterial diversity increased post-VSG (Shannon p = 0.047, Inv Simpson p = 0.04, Evenness ( p = 0.042) ). Significant changes in microbiome composition (beta diversity) were seen using the Canberra distance, which places more weight on lower abundance species ( p = 0.015, ) but not Bray Curtis distance. Pre-VSG, alpha diversity was significantly lower in those with diabetes compared with pre-diabetes and no diabetes (Supplementary Figure 1a) and microbiome composition also differed (Supplementary Figure 1b). Post-VSG, there were no differences in alpha or beta diversity between those who previously had diabetes and the other subjects. There were no significant differences between changes in alpha and beta diversity and other clinical parameters. There was a significant enrichment of 76 bacterial taxa post-VSG. Notably, the top 18 species enriched post-VSG were from the Streptococcus and Actinomyces genera ( . Supplementary Table 1). This included an enrichment of Streptococcus salivarius, Streptococcus vestibularis, Streptococcus parasanguinis, Actinomyces oris and Actinomyces oral taxon , all of which are commonly associated with the oral cavity. No individual taxa are significantly correlated with clinical characteristics. CAZymes were examined due to their role in influencing host metabolism. While the overall composition of CAZymes only showed a moderate change with VSG (Canberra PERMANOVA R2 = 0.049, p = 0.056, ), five specific CAZymes exhibited significant enrichment post-VSG ( , Supplementary Table 2). Many of the CAZyme enrichments post-VSG were associated with Streptococcus species . In addition, Glycoside Hydrolase 13 (GH13)+Carbohydrate Binding Module 20 (CBM 20) significantly associated with increased TBWL (q = 0.002, Supplementary Table 2). ARGs were examined using the RGI CARD database. Four ARGs were enriched post-VSG: qacJ, tetA (46), tetB(46) and vanY in vanF cluster . These resistance genes, especially tetAB(46), were primarily found in contigs belonging to several Streptococcus species enriched post-VSG . There were no functional pathway-level differences with VSG. Therefore, reads were aligned with Enzyme Commission (EC) gene annotations to identify more refined gene function differences. There was a moderate change in lower-abundance genes post-VSG (Canberra R2 = 0.05, p = 0.059, Supplementary Figure 2a), most notably with enrichment of EC 1.1.1.105, an all-trans retinol dehydrogenase gene, from the oxidoreductases class (Supplementary Figure 2b, Supplementary Table 3). In addition, these enriched ECs formed three distinct modules of co-associated genes (blue, brown and turquoise) that increased significantly post-VSG (Supplementary Figure 2c-d), driven by different bacterial genera (Supplementary Figure 2e). ECs in the blue module were primarily from mevalonate, hemiterpene biosynthesis and heme biosynthesis pathways; those in the brown module focused on sugar acid degradation, and the turquoise module contained several tRNA synthetases (Supplementary Table 3). When looking at the DNA virome, there were no significant differences observed with VSG in taxonomy, host taxonomy, taxonomic diversity, taxonomic composition, viral protein family diversity or composition (Supplementary Figure 3a-f). As most DNA viruses in the gut microbiome are phages, this suggests phage-containing bacteria did not significantly change with VSG despite large overall bacterial changes. Post-VSG stool displayed higher ratios of secondary to primary bile acids compared to pre-VSG stool . Notably, this pattern was broad and included all ratios for which the corresponding precursor primary bile acid and product secondary bile acid were detected. Amongst polar metabolites, only the elevation of citrulline post-VSG passed a false discovery rate cutoff of 10% . However, several trends known to be important to gut health were observed . These included a decrease in acyl-carnitines, an increase in neurotransmitters known to be directly microbially produced (GABA, dopamine, and histamine), and increases in redox cofactor metabolites. No changes in SCFAs were observed. Polar metabolites were related to a range of demographic and clinical factors (Supplementary Figure 4, Supplementary Table 4). Several amino acid metabolites (purple) positively associated with HDL, and negatively associated with hemoglobin A1C (HbA1c) pre-VSG. In addition, several metabolites correlated negatively with BMI pre-VSG, including those in Redox and Co-Enzyme metabolite (CoA, FMN, NADH), Glycolysis (G1P, Glycerol-3P), and Nucleotide (UMP, AMP) classes. Post-VSG, several metabolites associated with the degree of TBWL, including neurotransmitters such as GABA and Dopamine, which also correlated negatively with triglyceride levels. Citrulline did not correlate with any factors pre-VSG or post-VSG. To reduce the dimensionality of the microbiota, taxa were agglomerated into seven co-associated networks. Each increased in abundance post-VSG except for the red network, which exhibited a trending decrease ( p = 0.087; Supplementary Figure 5(a,b), Supplementary Table 5). Changes in these networks were correlated with changes in polar metabolites which revealed that the Bacteroides and the Alistipes/Akkermansia/Actinomyces prominent group both increased in concert with several SCFAs including butyrate, isovalerate, and isobutyrate (Supplementary Figure 5c, Supplementary Table 5). Conversely, the Streptococcus prominent group had the fewest correlations, only exhibiting positive correlations with isocitrate and cytidine and a negative correlation with urate. Next, to assess whether the microbiome/metabolome changes seen with VSG had a phenotypic effect, germ-free mice were inoculated with pre-VSG or post-VSG stool from the same participant . Two human participants were chosen who had both pre-VSG and post-VSG stool stored adequately in glycerol to preserve bacteria viability. Of note, one subject had the second-to-greatest TBWL and the second had prediabetes pre-VSG which resolved post-VSG. Using 16S rRNA gene sequencing, 6 weeks post-FT, mice transplanted with post-VSG stool had a higher alpha diversity and different bacterial composition (PERMANOVA Bray Curtis R2 = 0.114, p = 0.006, adjusted for human inoculum source, ) compared to mice that received pre-VSG stool, which was similar to the human stool microbiota findings. The mouse microbiome samples did also show significant separation based on the human subject used as the inoculum (PERMANOVA Bray Curtis R2 = 0.495, p = 0.001, adjusted for FT timepoint). Phenotypic differences resulting from FT with pre-VSG and post-VSG stools were compared. Six weeks post-FT, there was no difference in body weight, food consumption, intraperitoneal glucose tolerance tests, nor tissue weights (epididymal fat, subcutaneous fat and liver weights) between mice that received pre-VSG versus post-VSG stool ( respectively). Consistently, micro-computed tomography revealed no differences in subcutaneous or intraabdominal adipose tissue volume between the two groups . There was a significant increase in serum resistin in mice that received post-VSG stool compared to pre-VSG stool ( p = 0.047), but no differences in other metabolic hormones (GLP-1, insulin, leptin, gastric inhibitory polypeptide, ). Immune changes in the murine models were assessed in the large intestine lamina propria and mesenteric lymph nodes using flow cytometry (Supplementary Figure 6a). Of note, mesenteric lymph node data were only available from one set of mice pre- and post-VSG FT. In the large intestine, there was a significant decrease in γδ T cells, a non-significant increase in CD4 + T cells and decrease in CD8 + T and NK cells in post-VSG mice (Supplementary Figure 6b). More importantly, there was a significant increase in the number and proportion of Th17 cells, along with a significant decrease in GATA3 + regulatory T cell (Treg) proportion in the large intestine and mesenteric lymph nodes of post-VSG mice ( , Supplementary Figure 6c), with one participant (pat_10) displaying a more pronounced phenotype of the two. Th17 cytokines (IL-17A and IL-22) were also increased in the mesenteric lymph node of the post-VSG mice of this participant (Supplementary Figure 6d) but did not reach significance in the colon. While Th17 cells and cytokines have been associated with both determinantal and beneficial roles, an increase in Th17 cells and a corresponding decrease in regulatory T cells suggests an inflammatory potential in the individuals studied, that warrants further investigation. Sparse Partial Least Squares analysis of the most discriminatory immune and metabolic readouts and microbial taxa showed a distinct separation of pre- and post-VSG parameters indicating a direct or indirect association between the microbiota and observed immune milieu. Notably, the increased Th17 phenotype observed in the post-VSG mice clustered with increases in relative abundances of microbial taxa belonging to Ruminococcaceae, Erysipelatoclostridiaceae, and Monoglobus. Conversely, the increased Treg populations in the pre-VSG mice were associated with increased levels of the taxa Parabacteroides (Supplementary Figure 6e). Given the inflammatory phenotype post-VSG observed in the murine model, colonic inflammatory changes were assessed in available human stool samples by fecal calprotectin (FC). 8/12 participants had stool frozen without preservatives to allow for FC assessment. 6/8 individuals had an increase in FC to a clinically elevated level post-VSG (>120ug/g, ). Of these 6 participants, which included the 2 participants used for mouse studies, the mean FC pre-VSG increased from 47ug/g to 497ug/g post-VSG ( p = 0.0016, ). 2/8 patients with an elevated FC level pre-VSG had a decrease post-VSG. There was no correlation between length of time from surgery and FC levels. Subjects with raised FC post-VSG did not report typical clinical symptoms associated with high levels such as diarrhea. Additionally, the microbiome of subjects with increased FC also exhibited increased post-VSG microbiota similarity compared to pre-VSG where samples were more distinct (p for interaction = 0.016, Supplementary Figure 7a-b). No individual taxa correlated with FC increases. To assess systemic immunity/inflammation changes with VSG, urine samples were utilized due to the unavailability of blood samples. Proteomic analysis was performed using the SomaScan V4.1 Assay, an aptamer-based assay optimized for peripheral blood but which has been used with urine by prior studies in which a subset of the proteins found to be of interest were validated by orthogonal assays. From the overall 7000 analyte panel, a subset of 1856 proteins were identified which showed (Supplementary Figure 8a). The use of SomaLogic to assay urine was further supported by detection of differential expression for proteins previously implicated in inflammatory bowel disease (IBD) when comparing cases to controls (Supplementary Figure 8b), and observation of correlation between protein levels detected in urine and peripheral blood for most of the proteins detected to be differentially expressed in IBD (Supplementary Figure 8c). When the 1856 proteins which could be assayed in urine were compared between paired pre-VSG and post-VSG urine samples for 6 subjects, no individual proteins differed significantly. However, 17 pathways exhibited significant positive enrichment post-VSG (FDR < 0.05, ). The top enriched pathways included four pathways involved in immune/inflammation regulation: IRAK2 mediated activation of TAK1 complex; TICAM1, TRAF6-dependent induction of TAK1 complex; IRAK1 recruits IKK complex; and IRAK1 recruits IKK complex upon TLR7/8 or 9 stimulation. Enrichment in all four pathways was driven by the same set of three leading edge proteins related to ubiquitin, and the increased expression observed post-VSG was conserved across participants with one exception . We comprehensively examined stool microbiome and metabolome changes pre- and post-VSG in adolescents and assessed if these changes were causal in VSG-associated phenotypes using a murine FT model. Our major findings were 1) increased microbiome diversity post-VSG, with enrichment of several taxa, most notably in those usually associated with the oral cavity; 2) increased ratios of secondary to primary bile acids post-VSG; 3) no differences in metabolic phenotypes in germ-free mice transplanted with pre-VSG and post-VSG stool; 4) an inflammatory phenotype in germ-free mice transplanted with post-VSG stool compared to pre-VSG stool defined by an increase in Th17 cells and decrease in GATA3 + Tregs and γδ T cells in the gut and a systemic increase in resistin and 5) a corresponding inflammatory phenotype in a subset of adolescents post-VSG with an increase in FC. In our cohort, the gut microbiome increased in diversity and changed in composition with VSG. Generally, an increase in alpha diversity is seen with all forms of bariatric surgery in adults, but specific compositional and taxa changes vary by surgery type and study. , In our study, there was enrichment of taxa post-VSG, most notably in species commonly found in the oral cavity from the Streptococcus and Actinomyces genera. Some other studies examining microbiome changes post-VSG in adults have also seen an increase in these genera, although species could not be accurately identified due to the use of 16S rRNA sequencing. One study using FT in germ-free mice described a Th1-induced inflammation driven by the oral bacteria, Klebsiella, isolated from a human saliva sample. This inflammatory phenotype is in line with the inflammatory potential observed in our murine FT study, discussed below, although we observe an increase in Th17 instead of Th1. This increase in oral-associated taxa post-VSG is likely through a variety of mechanisms, including removing some of the physical barrier with VSG that usually prevents the passage of oral taxa and also an increase in stomach pH post-VSG. These changes are very likely due to VSG itself as they are rarely described in weight loss alone. It is also possible that 1) the diet recommended post-VSG (high in lean protein, low in fat and sugar) and 2) proton pump inhibitor use post-VSG, may contribute somewhat to the microbiome changes seen. These changes may have potential microbial and immunological consequences. Specifically, post-VSG within the enriched Streptococcus species, there was CAZyme enrichment of GH13+CBM 20, a combination shown in fungi to enhance complex starch degradation. Moreover, several ARGs enriched post-VSG were within the enriched Streptococcus species. These findings collectively suggest that the Streptococcus species that become enriched in the gut microbiome post-VSG are augmented for functions that may enhance their virulence. Overall, the increase in oral-associated taxa may be clinically relevant as there has been increasing recognition that enrichment of oral taxa in the gut maybe associated with several adverse inflammatory patient outcomes, including inflammatory bowel disease (IBD) and subclinical coronary atherosclerosis. There were significant changes in the stool metabolome and microbiome function post-VSG, many of which are considered beneficial. There was a broad increase in the microbiota-driven conversion of host-derived primary bile acids into secondary bile acids post-VSG, suggesting bolstering of the host-to-microbiome communication cycle post-VSG. Further, citrulline, which is understood as both a marker of gut health and to have a role in obesity, was increased post-VSG. , This molecule participates in the regulation of numerous pathways relevant to obesity and a therapeutic effect of citrulline supplementation in the regulation of obesity associated metabolic imbalances has also been suggested Moreover, microbially metabolized neurotransmitters including GABA, dopamine, and histamine also trended upward post-VSG in support of a gut-neural homeostasis improvement. Lastly, there was post-VSG enrichment of EC 1.1.1.105, an all-trans retinol dehydrogenase gene, with Vitamin A signaling and homeostasis reported to play a role in mitigating obesity. Overall, it is difficult to assess whether these potentially beneficial trends in fecal metabolites are a result of VSG, post-VSG weight loss or both. The absence of SCFAs changes post-VSG was notable, as changes have previously been reported in an adult study. Large inter-sample variance may have obscured trends in our cohort. An assessment for a causal role of the microbiome/metabolome changes with VSG performed via FT into germ-free mice did not identify favorable phenotypes such as decreased adiposity or improved glucose tolerance in mice that received post-VSG stool. These findings contrast with similar studies performing human to mouse FTs reporting improvements in body fat, glycemic control and energy expenditure However, those studies involved stool samples from older adults who were prediabetic or diabetic and employed mouse models that differed in age and sex of mice, fecal inoculation, experimental timelines, and diet administered. We may not have seen similar changes as the mice were harvested at 6 weeks following FT, which may not have been long enough to allow phenotypic changes to develop, and only one of the human participants used for FT had prediabetes pre-VSG. An intriguing finding in our mouse model was the inflammatory phenotype in those receiving post-VSG stools. Given that obesity is considered an inflammatory condition and subjects used for FT experienced weight loss, decreased inflammation in mice receiving post-VSG FT was expected. Indeed, a decrease in different measures of inflammation has been seen following both weight loss of patients with obesity and in one report of decreased gut and systemic inflammation in adults post-VSG. , Conversely, we observed an increase in the systemic levels of resistin in post-VSG mice, an adipokine increasingly being recognized as playing a role in inflammation. , Moreover, Th17 cells were increased and GATA3+ Tregs were decreased in the large intestine and mesenteric lymph nodes of post-VSG mice. However, this was particularly pronounced in one of the two participants used for murine FT studies and so may not be generalizable across the cohort. Numerous studies have described the role of Th17 and imbalance of Th17/Treg ratio in inflammatory bowel disease. , With relevance to VSG, one study of adults undergoing VSG reported decreases in peripheral Tregs. Further, Erysipelatoclostridiceae that positively correlated with Th17 in our study has been previously associated with inflammation in mouse models , In contrast, a recent study in mice demonstrated the beneficial role of microbiota-induced Th17 in protection against diet-induced obesity and metabolic syndrome, a finding not observed in our model. Collectively, these data indicate that stool microbiome/metabolome changes from VSG in humans may have gut and systemic inflammatory potential when transplanted to germ-mice, although the functional implications of this increased Th17 phenotype and underlying mechanisms remain to be elucidated. When inflammation was assessed in our pediatric participants, FC, a gut-specific marker of inflammation, was found to be raised in a subset of adolescents post-VSG, to levels well above the normal reference range. Two studies in adults report persistently raised FC levels after bariatric surgery, with one study showing a persistent increase above baseline 1-year post-VSG. , Interestingly in all studies including ours, raised FC levels did not correlate with adverse gastrointestinal symptoms or TBWL. In our study, systemically, four pathways involved in immune/inflammation regulation were enriched post-VSG, driven by proteins related to ubiquitin. Ubiquitination of the IKK complex mediated via proteins IRAK and TRAF-6 is an important step in the activation of the NF-kB pathway that plays a critical role in the immune response to microbes including the transcription of pro-inflammatory genes such as IL-2, IL-6, IL-12, TNFα, activation of macrophages and differentiation of CD4 + T cells. Increased activation of this pathway has been demonstrated as an important driver of IBD. Furthermore, the gut microbiota has been shown to modulate the expression of this signaling cascade and thus directly influence gut inflammation. Elevated FC levels and enrichment of pathways related to pro-inflammatory responses in humans post-VSG and an increased proportion of intestinal Th17 in comparison to Tregs, with increased resistin levels in mice colonized with post-VSG microbiota, collectively suggest a potential inflammatory state of subjects post-VSG. This may be clinically relevant given studies, including a large case series and two national database studies, showing an increased incidence of IBD after bariatric surgery. One study reported an increased incidence of ulcerative colitis, but not Crohn’s disease, with VSG. It is widely accepted that IBD is at least in part a gut microbiome-mediated disease in susceptible individuals. Therefore, we postulate that the large shifts in the gut microbiome seen post-VSG can have inflammatory potential, and in certain individuals, this may increase their risk of IBD or other inflammatory diseases. It could be argued that increases in FC may be related to the actual VSG rather than gut microbiome changes resulting from VSG. However, against this, is the inflammatory phenotype seen in mice transplanted with post-VSG stool who did not receive surgery themselves and the level of FC post-VSG not correlating with the length of time post-surgery. Our findings by no means negate the many beneficial effects of VSG in adolescents with obesity. However, they do highlight a new finding that microbiome changes post-VSG may be inflammatory in some adolescents. This paves the way for further research to gain insight into potential inflammatory mechanisms with gut microbiota changes post-VSG. Additionally, larger studies of children and adolescents undergoing VSG followed for a longer period are needed to assess whether microbiome changes and inflammation are persistent, and if those with persistent inflammatory changes are at increased risk for adverse outcomes such as IBD. Despite this being the first study to our knowledge to comprehensively examine microbiome and metabolome changes post-VSG in adolescents and show that these changes with VSG can be potentially inflammatory, the study has many limitations. The sample size is limited, partly due to the limited number of adolescents undergoing VSG. Additionally, longitudinal samples after 7 months post-VSG were unavailable. While a large increase in oral-associated taxa was seen in the stool post-VSG, no oral samples were available to confirm whether the same taxa from the oral cavity were engrafting in the gut within a specific participant. Additionally, if these oral taxa contribute to the inflammatory phenotype remains to be determined. A complete assessment of inflammation in adolescents was limited as only stool and urine were available. Limited samples were available for the FT murine models, with both human participants used having an increase in FC post-VSG, and one participant used driving the potential inflammatory phenotype in mice more than the other, thus limiting the ability to assess whether the findings were generalizable across the participant cohort. Further, the murine models did not assess measures of intestinal inflammation, such as intestinal histology and rectal bleeding, or determine the functional capability of the increased Th17 cells to identify Th17 subsets previously implicated with IBD. In conclusion, large changes in the stool microbiome and metabolome were seen in adolescents post-VSG. There was a notable enrichment of oral-associated taxa in the gut post-VSG. The post-VSG changes in the gut microbiome/metabolome were shown to have inflammatory potential when transferred to a germ-free mouse model. Furthermore, raised FC and inflammatory systemic pathways were seen in a subset of adolescents post-VSG. While VSG is highly effective for weight loss and reduction of comorbidities in adolescents with obesity, we show the novel finding of potential inflammatory microbiome changes post-VSG. This may be of importance given the growing recognition of an increased incidence of IBD after bariatric surgery and warrants further investigation. Supplemental Material |
Exploring the contribution of integrated healthcare practices to malaria control in Ghana: perspectives of medical herbalists | efa2a28a-3bf1-4506-a7ab-2c2ee68ecbea | 11734445 | Pharmacology[mh] | Malaria remains a major public health concern, particularly in Sub-Saharan Africa, where it accounts for millions of deaths annually . Approximately 70% of the global malaria burden is concentrated in 11 African countries, including Ghana . Despite Ghana’s progress in reducing severe malaria cases and achieving a 13% decline in childhood mortality , the disease still exerts a substantial strain on orthodox healthcare facilities. Considerable international investment from the Global Fund, World Bank, and bilateral donors has targeted malaria control, primarily promoting the use of orthodox anti-malarial drugs . However, the high cost of these medications has hindered effective malaria control efforts in Ghana . In contrast, the Ghanaian population has a documented history of utilising traditional herbal medicine (THM) for malaria treatment . Factors influencing this choice include dissatisfaction, perceived ineffectiveness, or adverse effects associated with orthodox treatments , as well as alignment with cultural beliefs and health philosophies . Thus, THM plays a vital role in the Ghanaian healthcare system, particularly for malaria. Recognising the significance of THM, the Ghanaian government has integrated it into the formal healthcare system through policy formulation, the establishment of THM clinics within select public hospitals, and the creation of a THM Department at Kwame Nkrumah University of Science and Technology (KNUST) . This allows individuals to access certified THM services and products (herbal products manufactured by Centre for Scientific Research into Plant Medicine [CSRPM] and approved by the Food and Drug Authority [FDA]) within public health facilities (integrated hospitals and private THM clinics) for malaria treatment . Integrated hospitals as used in this study refer to government-owned hospitals with THM clinics while private THM clinics are owned and operated by individuals. Despite this integration, no research to date has assessed its contribution to malaria control. This study aims to explore the perceived effects of integrated healthcare practices on malaria control in Ghana, specifically through the experiences of medical herbalists in the country’s coastal, forest, and savannah regions. We aim to: Investigate the positive contributions of THM integration to healthcare service delivery for malaria control as perceived by medical herbalists. Examine the challenges associated with the practice of integration and its impact on malaria control. Theoretical framework Donabedian’s framework for evaluating healthcare quality guided this research . This framework identifies three key elements - structure, process, and outcome (Fig. ) , and emphasises their interrelationship in assessing healthcare quality . Structure refers to the physical settings in which health services are delivered, including factors such as facility availability, healthcare personnel, and equipment. Process encompasses the delivery of healthcare services, such as referrals and patient follow-up. Finally, outcome reflects the impact of healthcare on patient populations, with positive outcomes reflecting reduced malaria prevalence/cases . While limited research has employed Donabedian’s framework in the context of malaria control interventions, its application here is intended to identify and understand how structural and process factors influence malaria control outcomes (positive or negative) within the integrated THM and orthodox medicine system in Ghana. Donabedian’s framework for evaluating healthcare quality guided this research . This framework identifies three key elements - structure, process, and outcome (Fig. ) , and emphasises their interrelationship in assessing healthcare quality . Structure refers to the physical settings in which health services are delivered, including factors such as facility availability, healthcare personnel, and equipment. Process encompasses the delivery of healthcare services, such as referrals and patient follow-up. Finally, outcome reflects the impact of healthcare on patient populations, with positive outcomes reflecting reduced malaria prevalence/cases . While limited research has employed Donabedian’s framework in the context of malaria control interventions, its application here is intended to identify and understand how structural and process factors influence malaria control outcomes (positive or negative) within the integrated THM and orthodox medicine system in Ghana. Study setting Ghana’s diverse ecological landscape can be broadly categorised into coastal, forest, and savannah belts (Fig. ) . To represent these variations, the Central (coastal), Ashanti (forest), and Upper West (savannah) regions were selected. These regions experience high malaria prevalence (15–20% in Ashanti, 20–25% in Upper West, and 25–30% in Central) . The presence of integrated healthcare facilities and private THM clinics in these regions makes them suitable for exploring the effects of integration on malaria management. Within each region, the respective capitals (Kumasi, Cape Coast, and Wa) were chosen as specific study sites (Fig. ). These capitals are endowed with public hospitals with THM units as well as privately owned THM clinics . Design This qualitative study employed a phenomenological design to delve into the lived experiences of medical herbalists regarding integrated healthcare practices and malaria control. The design is well-suited for understanding lived experiences and allowed us to explore the “what” and “how” of the participants’ experiences with integration’s contribution to malaria control, ultimately providing a collective understanding of their encounters . We extracted this paper from a larger study that explored the contribution of integrated healthcare to malaria control by gathering the experiences of health practitioners (medical doctors, pharmacists and medical herbalists) in the coastal, forest and savannah regions of Ghana. Other aspects of the research have been submitted for publication elsewhere. Study population The study population consisted of experienced medical herbalists aged 18 or older who were practicing in integrated health facilities or private THM clinics in the Kumasi, Cape Coast metropolis and Wa municipality. Participants were required to be at least 18 years of age to ensure their capacity for informed consent, consistent with Ghanaian legal and ethical standards . Medical herbalists were chosen as key stakeholders in the Ghanaian healthcare system, playing a crucial role in providing alternative healthcare services for malaria. Sample size and sampling technique Purposive sampling was employed to recruit participants from 16 healthcare facilities within the designated regions. These facilities included six integrated hospitals and ten private THM clinics. Participant selection was primarily guided by specific criteria, such as the age, type of facility (integrated hospital or private THM clinic) and geographic location. This approach aimed to enhance the credibility of findings by ensuring variation in the sample. The recruitment process utilised strategies such as direct contact and leveraging existing networks. All medical herbalists approached, agreed to participate in the study, providing detailed information that facilitated a comprehensive understanding of the research topics. Data saturation guided the sample size, with recruitment ending at 19 participants when no further new information or themes were observed in the data. Data collection period Data collection spanned three months (July - October 2023). Three research assistants (two males and one female), holding master’s degrees in public health and with expertise in qualitative research, were recruited from the University of Cape Coast (UCC) and University of Development Studies (UDS) to assist with data collection. A training session (both online and in-person at UCC campus) was conducted in the fourth week of June 2023 to ensure they understood the study objectives and interview guide. Each session lasted four hours. Data collection procedure Before actual data collection, three pilot interviews were conducted by the assistants and reviewed by IGA to ensure clarity of the questions and precision of data. Face-to-face, in-depth interviews served as the primary data collection method, utilising a semi-structured interview guide developed based on Donabedian’s framework for healthcare quality evaluation (See Supplementary file 1, Study Instrument). English was the primary language used, and interviews were conducted in participant-preferred, comfortable environments. All interviews were audio-recorded and lasted 45–60 min. Following participant identification, information sheets explaining the study’s objectives, benefits, and ethical considerations were provided. Informed consent was obtained from each participant before the interview commenced. To mitigate bias and ensure interviewers adhered to the interview guide, a training manual was developed to guide question phrasing. The first author (IGA) observed the first three interviews to ensure consistency in the interview procedure. To uphold anonymity, participants were assigned codes instead of identifiable information. Basic demographics (sex, age, specialty, and facility type) were collected at the beginning of interviews. The interview guide covered topics centred on participants’ perceptions and experiences regarding malaria management through THM integration in Ghana. Specifically, discussions focused on the perceived contributions of integration to malaria control and challenges associated with it. Data saturation was achieved when no new information or themes emerged after the 16th interview. However, to avoid inadvertent exclusion of new perspectives, the assistants interviewed three further interested participants who had previously expressed interest. Additionally, the assistants documented their observations during data collection. While repeat interviews were not necessary, interpretations were sought from some participants after the interviews had been transcribed. Data analysis An experienced transcriber transcribed the audio-recorded interviews, with IGA reviewing them for accuracy. Transcribed data were then shared with some participants during follow-up meetings for verification and potential modifications. A framework analytical approach, involving both inductive and deductive techniques, was employed to analyse the transcribed data in NVivo software (version 12) . Two authors (IGA and GAA) independently conducted data analysis. The transcribed data were thoroughly read by both researchers to become familiar with the content. Following this, a thematic framework outlining core themes and concepts was developed based on notes taken during familiarisation. This stage primarily employed inductive analysis, allowing themes to emerge directly from the data. Indexing involved systematically marking relevant data segments under identified themes. Here, an inductive approach was again utilised to ensure themes remained grounded in the data. Next, charting involved organising the coded data into charts according to the identified themes. Finally, the mapping and interpretation stage involved arranging the charted data to illustrate the participants’ experiences with THM integration and its impact on malaria management in Ghana. Deductive analysis was utilised during mapping and interpretation, grouping themes under the components of Donabedian’s framework for healthcare quality evaluation. To further ensure rigorous data analysis, introductory coding and theme generation were initially performed independently by IGA and GAA. Discrepancies in coding were resolved through discussion and consensus during a dedicated meeting. Additionally, one author (TIE) reviewed the finalised themes and quotes to enhance the trustworthiness of the research findings. The trustworthiness of the research was further bolstered through various strategies, including: Supervisor/peer debriefing: Regular discussions with the supervisor (TIE) and peer researchers ensured a balanced perspective and identification of potential biases. Member checking: Sharing transcribed interviews with some participants for verification and potential modifications allowed for participant validation of the interpretations. Researcher/assistant triangulation: Consistency in data collection and interpretation was maintained through discussions between the lead author (IGA) and research assistants. Thorough description of study setting and methods: Extensive detail regarding the study context and methodological approach provides transparency and facilitates replicability . The identified themes are presented in the results section, accompanied by illustrative quotes and participant location/setting (e.g., Participant 1, Cape Coast). The Consolidated Criteria for Reporting Qualitative Studies (COREQ) checklist was used to appraise the final version of the manuscript (See Supplementary file 2, COREQ Checklist). Ghana’s diverse ecological landscape can be broadly categorised into coastal, forest, and savannah belts (Fig. ) . To represent these variations, the Central (coastal), Ashanti (forest), and Upper West (savannah) regions were selected. These regions experience high malaria prevalence (15–20% in Ashanti, 20–25% in Upper West, and 25–30% in Central) . The presence of integrated healthcare facilities and private THM clinics in these regions makes them suitable for exploring the effects of integration on malaria management. Within each region, the respective capitals (Kumasi, Cape Coast, and Wa) were chosen as specific study sites (Fig. ). These capitals are endowed with public hospitals with THM units as well as privately owned THM clinics . This qualitative study employed a phenomenological design to delve into the lived experiences of medical herbalists regarding integrated healthcare practices and malaria control. The design is well-suited for understanding lived experiences and allowed us to explore the “what” and “how” of the participants’ experiences with integration’s contribution to malaria control, ultimately providing a collective understanding of their encounters . We extracted this paper from a larger study that explored the contribution of integrated healthcare to malaria control by gathering the experiences of health practitioners (medical doctors, pharmacists and medical herbalists) in the coastal, forest and savannah regions of Ghana. Other aspects of the research have been submitted for publication elsewhere. The study population consisted of experienced medical herbalists aged 18 or older who were practicing in integrated health facilities or private THM clinics in the Kumasi, Cape Coast metropolis and Wa municipality. Participants were required to be at least 18 years of age to ensure their capacity for informed consent, consistent with Ghanaian legal and ethical standards . Medical herbalists were chosen as key stakeholders in the Ghanaian healthcare system, playing a crucial role in providing alternative healthcare services for malaria. Purposive sampling was employed to recruit participants from 16 healthcare facilities within the designated regions. These facilities included six integrated hospitals and ten private THM clinics. Participant selection was primarily guided by specific criteria, such as the age, type of facility (integrated hospital or private THM clinic) and geographic location. This approach aimed to enhance the credibility of findings by ensuring variation in the sample. The recruitment process utilised strategies such as direct contact and leveraging existing networks. All medical herbalists approached, agreed to participate in the study, providing detailed information that facilitated a comprehensive understanding of the research topics. Data saturation guided the sample size, with recruitment ending at 19 participants when no further new information or themes were observed in the data. Data collection spanned three months (July - October 2023). Three research assistants (two males and one female), holding master’s degrees in public health and with expertise in qualitative research, were recruited from the University of Cape Coast (UCC) and University of Development Studies (UDS) to assist with data collection. A training session (both online and in-person at UCC campus) was conducted in the fourth week of June 2023 to ensure they understood the study objectives and interview guide. Each session lasted four hours. Before actual data collection, three pilot interviews were conducted by the assistants and reviewed by IGA to ensure clarity of the questions and precision of data. Face-to-face, in-depth interviews served as the primary data collection method, utilising a semi-structured interview guide developed based on Donabedian’s framework for healthcare quality evaluation (See Supplementary file 1, Study Instrument). English was the primary language used, and interviews were conducted in participant-preferred, comfortable environments. All interviews were audio-recorded and lasted 45–60 min. Following participant identification, information sheets explaining the study’s objectives, benefits, and ethical considerations were provided. Informed consent was obtained from each participant before the interview commenced. To mitigate bias and ensure interviewers adhered to the interview guide, a training manual was developed to guide question phrasing. The first author (IGA) observed the first three interviews to ensure consistency in the interview procedure. To uphold anonymity, participants were assigned codes instead of identifiable information. Basic demographics (sex, age, specialty, and facility type) were collected at the beginning of interviews. The interview guide covered topics centred on participants’ perceptions and experiences regarding malaria management through THM integration in Ghana. Specifically, discussions focused on the perceived contributions of integration to malaria control and challenges associated with it. Data saturation was achieved when no new information or themes emerged after the 16th interview. However, to avoid inadvertent exclusion of new perspectives, the assistants interviewed three further interested participants who had previously expressed interest. Additionally, the assistants documented their observations during data collection. While repeat interviews were not necessary, interpretations were sought from some participants after the interviews had been transcribed. An experienced transcriber transcribed the audio-recorded interviews, with IGA reviewing them for accuracy. Transcribed data were then shared with some participants during follow-up meetings for verification and potential modifications. A framework analytical approach, involving both inductive and deductive techniques, was employed to analyse the transcribed data in NVivo software (version 12) . Two authors (IGA and GAA) independently conducted data analysis. The transcribed data were thoroughly read by both researchers to become familiar with the content. Following this, a thematic framework outlining core themes and concepts was developed based on notes taken during familiarisation. This stage primarily employed inductive analysis, allowing themes to emerge directly from the data. Indexing involved systematically marking relevant data segments under identified themes. Here, an inductive approach was again utilised to ensure themes remained grounded in the data. Next, charting involved organising the coded data into charts according to the identified themes. Finally, the mapping and interpretation stage involved arranging the charted data to illustrate the participants’ experiences with THM integration and its impact on malaria management in Ghana. Deductive analysis was utilised during mapping and interpretation, grouping themes under the components of Donabedian’s framework for healthcare quality evaluation. To further ensure rigorous data analysis, introductory coding and theme generation were initially performed independently by IGA and GAA. Discrepancies in coding were resolved through discussion and consensus during a dedicated meeting. Additionally, one author (TIE) reviewed the finalised themes and quotes to enhance the trustworthiness of the research findings. The trustworthiness of the research was further bolstered through various strategies, including: Supervisor/peer debriefing: Regular discussions with the supervisor (TIE) and peer researchers ensured a balanced perspective and identification of potential biases. Member checking: Sharing transcribed interviews with some participants for verification and potential modifications allowed for participant validation of the interpretations. Researcher/assistant triangulation: Consistency in data collection and interpretation was maintained through discussions between the lead author (IGA) and research assistants. Thorough description of study setting and methods: Extensive detail regarding the study context and methodological approach provides transparency and facilitates replicability . The identified themes are presented in the results section, accompanied by illustrative quotes and participant location/setting (e.g., Participant 1, Cape Coast). The Consolidated Criteria for Reporting Qualitative Studies (COREQ) checklist was used to appraise the final version of the manuscript (See Supplementary file 2, COREQ Checklist). Nineteen medical herbalists participated in the study, with ages ranging from 24 to 57 years (mean = 35 years). The majority ( n = 16, 84.2%) were males. The sex imbalance is consistent with the lower number of female medical herbalists practicing in Ghana compared to males. Nearly half ( n = 9, 47.4%) practiced in integrated healthcare facilities, while the remainder operated in private THM clinics. All participants were general practitioners, treating various ailments including malaria. The coastal belt (Cape Coast) had the highest representation ( n = 9, 47.4%) among participants (Table ). Themes Thematic analysis of participant narratives revealed five key themes (Table ): Benefits of THM Practice: This theme explored the perceived advantages of THM use. Knowledge/Understanding of THM Integration: This theme examined participants’ understanding of the integrated healthcare model. Structure/Infrastructure for Integrated Care Delivery: This theme addressed the physical resources and settings associated with integrated healthcare delivery. Processes/Activities in Integrated Healthcare Delivery: This theme described the activities and workflows involved in the integrated care model. Outcomes/Consequences of THM Integration on Malaria Control: This theme investigated the perceived impact of THM integration on malaria management in Ghana. Benefits of THM practice Participants highlighted various advantages of THM practice, including its perceived efficacy, minimal side effects, and job creation potential. Participants believed THM offers faster malaria cures due to the body’s familiarity with such remedies. Others acknowledged that although all medications have side effects, those associated with THM were considered negligible. Additionally, some participants viewed the traditional health system as a source of employment, with private THM clinics creating job opportunities for the populace. The quotes that follow expatiate on the views expressed by participants regarding the merits of THM practice in Ghana: Efficacy of THM I can say herbal medicine cures malaria better than orthodox medicine. Our human system knows herbal medicine. It is like introducing food to your baby…. Our system accepts herbal medicine well [Participant 12, Cape Coast]. Minimal side effects All medications have side effects. Once it is an external thing that you are introducing into your body, there will be some side effects. But it is up to the degree of the side effects. When you compare the side effects, it is less for herbal medicines than the orthodox medications [Participant 11, Cape Coast]. Employment creation …THM has created jobs in Ghana. For example, when you go to Kumasi, there are a lot of herbal clinics/outlets that have been established and these clinics have and keep employing people [Participant 1, Cape Coast]. Knowledge/understanding of the practice of integrated healthcare Participants generally demonstrated awareness of THM integration into the Ghanaian healthcare system. They specifically mentioned the establishment of a Traditional Herbal Medicine program at KNUST as evidence of this integration. I am aware of the integration. Perhaps, that is why they have started teaching THM at KNUST. We have students from ‘Tech’ that are studying THM as a course to become professionals [Participant 5, Cape Coast]. This suggests that participants perceive the inclusion of THM within the formal education system as a significant step towards its legitimisation and integration within mainstream healthcare delivery. Structure/infrastructure associated with integrated healthcare delivery This theme addressed the physical resources and settings associated with integrated healthcare delivery. Four key sub-themes emerged: Availability of integrated health facilities Participants acknowledged the presence of THM units within some government hospitals. However, they expressed concern about the limited number of these facilities. They attributed this inadequacy to factors such as low public awareness, insufficient government support, and consequently, low patient utilisation of integrated healthcare services. …we have 55 piloted facilities. That number is small compared to the number of orthodox healthcare facilities in the country. That also accounts for the unpopularity of the THM unit in the country because the more there are herbal medicine facilities, the more people will get to know and patronise such facilities [Participant 11, Cape Coast]. I am aware of the availability of integrated facilities. We have Tafo government hospital, Juaben government hospital, Bekwai, and in Accra we have LEKMA hospital, Police hospital, Tema general hospital, and then the Tamale teaching hospital, Ho government hospital, and Wawra government hospital in the Oti region. So, it is all over the country [Participant 14, Kumasi]. In the Upper West region, there is one there. That is, the Wa municipal hospital. There are about two medical herbalists at the Wa Municipal hospital [Participant 4, Wa]. Quality assurance Participants practicing within integrated health facilities highlighted their commitment to quality healthcare delivery. They described adhering to established medical principles and protocols for patient care which included prescribing only herbal products listed on the Ministry of Health’s recommended herbal medicine list. Additionally, some participants viewed regular inspections by the FDA as a means to ensure the quality, efficacy, and safety of the herbal products and services they provide. We have the recommended herbal medicine list. So, the recommendation should be based on the list from the Ministry of Health which has been certified so we stick to defined medical principles [Participant 2, Wa]. These days, once you are a certified THM provider, the FDA will come for inspection and check that your facility is following scientific methods to ensure quality, effectiveness, and safety to those who would use it [Participant 6, Cape Coast]. Inadequate health personnel Participants highlighted a shortage of qualified healthcare personnel within integrated healthcare facilities. They described performing multiple roles beyond their core duties, including procurement, reporting, medication request follow-up, and drug dispensing. This role overload was attributed to the limited number of staff and a perceived lack of personnel with expertise in herbal medicine. Furthermore, some participants suggested that existing medical herbalists might be drawn to other career paths, such as academia or positions with the Ghana FDA, potentially exacerbating the staffing shortage. In our case, we must initiate, follow up with memo to request for medications. They say they don’t know much about herbal medicines, so I must take charge and lead the procurement. So, you become part of pharmacy because you have to monitor and make sure they are dispensing the right drugs. So you move up and down; it is a lot of work because one person you are involved in reporting, payment process, monitoring, and everything [Participant 2, Wa]. …comparing to the country’s population, I don’t think we have enough qualified medical herbalists. The whole of Cape Coast metropolis, we are just five. Can you imagine! [Participant 8, Cape Coast]. The numbers are not enough…. we started herbal medicine in KNUST in 2001 and most of us don’t practice. They enter a different sector like FDA. Those who really practice is very few. They just branch into a different field. Some are in academia and the FDA [Participant 17, Kumasi]. Insufficient medicines and equipment This sub-theme emerged primarily among participants from the forest belt. They reported experiencing disruptions in service delivery due to a shortage of essential herbal medicines. They attributed this shortage to cumbersome procurement processes perceived as overly bureaucratic. Additionally, delays in deliveries from approved manufacturing centres were identified as contributing to stockouts. These shortages reportedly hampered their ability to provide quality care. In contrast, participants from the coastal belt highlighted the issue of inadequate equipment, particularly a lack of computers. This lack of technology was perceived as hindering their productivity and overall job satisfaction. They associated this equipment deficiency with limited or absent funding for integrated healthcare facilities. We sometimes run out of drugs because to order the drugs must pass through so many bureaucratic lines. When I talk about the bureaucratic system, I mean that it must go through many offices before it gets approved. Over here, they will bring the pro forma invoice then we work on it and get the voucher from the account. Then the signatory must come from different people. So, if one signatory has travelled, it becomes a challenge to order the drugs, hence limited or no drugs [Participant 16, Kumasi]. When we order our medicines from Akuapem Mampong, it takes a long time. It delays and that doesn’t help because the client needs it, which disturbs you too. Why should you delay in giving treatment to someone suffering from malaria? [Participant 15, Kumasi]. … We don’t have computers to assist with our work. And today, everything is e-health and so, if I do not come with my personal computer, then it means that I will be unable to work. Computer is a basic equipment that we think if we had, it would improve our healthcare delivery, but we don’t have them due to the lack of funds. It doesn’t make the work enjoyable [Participant 8, Cape Coast]. Processes/activities related to the delivery of integrated health services The main issues identified under this theme were categorised into five sub-themes; health system-based challenges, national/general challenges, patient-centred care, follow ups, and THM training and research. Health system-based challenges This section explores challenges identified by participants within the processes and activities related to integrated healthcare delivery for malaria control. Two issues emerged: First, disapproval from orthodox medicine providers: Participants expressed concerns regarding negative perceptions and attitudes towards THM integration from some orthodox medicine providers. These negative views were perceived to hinder effective collaboration and patient referrals. One example cited by participants was the misconception that THM can cause kidney and lung problems. They reported efforts to address these misconceptions through clinical meetings, but with limited success. Another challenge we are facing, which we are trying to solve has to do with the attitudes and perceptions of the medical doctors and other orthodox practitioners. They have the perception that when you take herbal medicine, it will affect your kidneys and lung. it is a major challenge to the integration. We organise clinical meetings to explain to them, yet they are not convinced [Participant 8, Cape Coast]. ……for the medical doctors, they don’t want to hear about THM. They think that they go to school to learn about scientific medicine. But they don’t know that currently THM providers also go through scientific training. Their opposition is one of our challenges [Participant 3, Wa]. Secondly, participants, particularly those from the coastal and savannah regions, highlighted challenges associated with referral practices within the integrated healthcare system. They reported that orthodox medical doctors often exhibited reluctance to formally refer patients to medical herbalists. Instead, some doctors resorted to informal methods of recommending THM, blurring the lines between a formal referral and casual advice. Participants attributed this reluctance to several factors: Perceived Superiority: Some medical herbalists believed that orthodox doctors held a sense of superiority, viewing them as apprentices rather than qualified healthcare providers. Patient Affordability Concerns: Participants suggested that doctors might hesitate due to concerns about patients’ ability to afford certified herbal medicines offered at integrated facilities, as these often require out-of-pocket payments. It is difficult for the orthodox medicine provider to accept and refer patients to us. For those who refer, they do it ‘backdoor’, in the form of an advice rather than as a formal referral [Participant 6, Cape Coast]. …the medical doctors prefer to treat malaria cases rather than referring to us…like I said, the herbal medications are cash-and-carry. So, the doctors are concerned about the cash. They are not certain that the patient will pay for the medication, and so, they prefer to treat the patient with their medication rather than referring them for herbal treatment [Participant 8, Cape Coast]. For my practice over ten years, no medical officer has referred a patient to me. We have a referral form that we fill. The challenge is that they frown on our referral forms. So, what we do is to refer the person verbally…. we do it in an informal setting. we don’t have a problem referring patients to the orthodox. But they will not do that. Why will they [orthodox] do that? They see themselves as superior, so why would they want to refer to us? They [orthodox providers] see us as apprentice. [Participant 19, Wa]. National/general challenges This theme highlighted issues at the national level that hindered effective management of malaria through the practice of integrated healthcare. Major national/general challenges that emerged were non-exhaustive NHIS coverage and inadequate promotional activities on THM integration. Participants narrated that the non-comprehensive nature of the NHIS (exclusion of herbal anti-malarial drugs) accounts for low patronage of services at THM clinics at integrated hospitals because clients always opt for free orthodox anti-malaria medicines rather than paying full cost for the herbal ones. …. the malaria medication for orthodox is on the health insurance but when you opt for the herbal medicine, then you will be paying about ghs80 or more. So, free or ghs80? The cost involved is what is accounting for the low patronage of herbal medicines for malaria in our THM clinics [Participant 18, Kumasi]. Promotional activities play a vital role in the successful implementation of interventions. When the providers were asked to share their experiences concerning publicity of integration and its implication on malaria control, they gave an account, which suggests inadequacy of promotional activities regarding the integration programme. … One major challenge we are facing with this integration is the lack of publicity or awareness. Most people don’t know about this integration. The Ministry of Health should have projected the integration right from the onset. Now, when you visit these selected hospitals (integrated facilities), there are herbal practitioners available, but people do not know about that, so they do not patronise our services [Participant 8, Cape Coast]. Patient-centred care When discussing patient care, participants recounted that attending to their clients on time and spending ample time with them is something they were proud of. They emphasised that the patients really appreciated that kind of care unlike the orthodox unit, where less time is spent. The ensuing quote represent this finding: It is something we are very proud of at our unit. We spend a lot of time, about an hour with one patient.…. They really do appreciate that unlike the orthodox side that within 2–3 minutes they are done and don’t have time for them. So, when it comes to patient-provider relationship, it is the best in our unit [Participant 11, Cape Coast]. Follow ups We found that follow ups, which is a feature of functional health systems was well implemented by the participants. They narrated that their desire to promote the welfare of clients motivated them to follow up and conduct medical reviews to avert undesirable treatment outcomes. If you are interested in a case, then you follow up! There are others (clients) that I had to give my contact for them to call back. I do this to avoid any adverse effect of the treatment or medication [Participant 2, Wa]. … Some of the clients come for review. Through the follow ups we schedule reviews. After taking the THM, we don’t let them stay home. We schedule reviews and when they come, we test and make sure they are fine [Participant 15, Kumasi]. We also do a lot of follow up on our patients. We call to check up on them and find out whether they have had any side effects [Participant 8, Cape Coast]. THM training and research Participants recognised that the practice of integration has boosted research on herbal remedies for treating malaria. They perceived the herbal anti-malarial medications prescribed and utilised within the integrated system to be safe and effective because they go through scientific scrutiny at the Centre for Research into Plant Medicine and/or the Nogouchi Memorial Institute for Medical Research. The quotes below reiterate this finding: I think it has boosted research on THM for malaria control because herbal anti-malaria drugs must be tested either at Nogouchi Memorial Institute for Medical Research or Centre for Research into Plant Medicine to verify whether it is safe for human consumption [Participant 1, Cape Coast]. Our THM comes from the Centre for Research into Plant Medicine; a lot of research has gone into these products, and they are safe for use. They are now safe and efficacious for treating malaria [Participant 15, Kumasi]. In addition, participants within the forest belt reported that they usually undergo continuous professional development training. According to them, such trainings were organised by the traditional and alternative medicine directorate and involved the interpretation of scientific medical activities such as x-ray reading, understanding laboratory results and clinical emergencies. They believed that this knowledge enabled them to interpret and confirm diagnosis of diseases, especially malaria. The quotations below summarise this report: …. as medical herbalists we go for continuous professional development training. The training is mostly dependent on current situations. So, if there is a new herbal medication or new ways of doing things, we do that to improve upon the knowledge that we have. If there are research works going on, they update us on findings. We also do research because science is always updating. The training is mostly done by the Traditional and Alternative Medicine Directorate. They organise and get resource people to take us through it [Participant 14, Kumasi]. We have been going for continuous professional development (CPD) training. We do it yearly to upgrade our practice and the training largely depends on the topics that we deal with in that year. We have received training on the interpretation of x-rays, lab results and clinical emergencies…. the knowledge acquired enable us to request for lab testing and interpret the results to confirm our diagnosis for malaria [Participant 15, Kumasi]. Interestingly, participants at the savannah belt believed that THM training focuses on exposing medical herbalists to both modern/orthodox and traditional medicines. However, the same could not be said for orthodox medicine providers where training is more focused on orthodox approaches. This unbalanced training was perceived to have created knowledge gaps among mainstream healthcare providers, leading to a lack of understanding and appreciation of the contribution of THM to healthcare delivery in Ghana. Some of the orthodox practitioners do not understand herbal medicine. We are trained to understand both orthodox and herbal medicines, but they are not trained that way. We have an appreciation of both sides but for them, they weren’t trained like that. They were trained one-sided and so they don’t understand herbal medicine, creating opposition from them [Participant 2, Wa]. Outcome/consequence of the practice of THM integration on malaria control A key aspect of the study was to explore the perceived impact of THM integration on malaria control. Participants identified two main impacts - reduced pressure/burden on orthodox medicine providers and evidence-based THM practice leading to appropriate management of malaria. The participants perceived that their presence at public hospitals reduced pressure on orthodox medical doctors because they mostly treat malaria cases at their units and that the majority of the clients prefer herbal anti-malaria medication such as the ‘Mebeema’. When you take the cases we treat in our unit, malaria is among the top five. So, if the orthodox doctor was treating about 1 , 000 cases, the numbers have reduced now because most people want to use the herbal anti malaria medicine, that is, Mebeema [Participant 11, Cape Coast]. …. even in our facility, we have about three different drugs for malaria. The most potent one is Mebeema from Akuapem Mampong. It has been used in the system for long and the response has been very good. Lots of people who come to the unit opt for it. Clearly, our presence here at the hospital has reduced pressure on the orthodox providers [Participant 16, Kumasi]. Besides, reducing pressure on orthodox medical doctors, participants irrespective of location, reported that the integration of THM has been beneficial because malaria cases are now appropriately treated through the evidence-based practice of their field. They believed the application of scientific clinical procedures such as correct diagnosis through laboratory testing/prescription, health education, and FDA approval of herbal anti malarial medications have led to effective malaria treatment, hence reducing prevalence. The following quotes elaborate on the views expressed by the study participants. I will say that about 50% of our clients come for malaria treatment. Just as they do for orthodox medicine, we also make sure that the client is taken through education, then we ask them to get tested through the lab and confirm they have malaria before we allow them to take our medication. Evidently, we apply scientific clinical procedures in treating our clients, and it has proven effective because the cases are reducing, and people are getting healed quickly [Participant 9, Cape Coast]. … integration has helped. …. At first, they will go to the market and get anything. But now, they (patients) come to the hospital and go through the process before receiving treatment. Malaria cases have gone down in recent times. When they come in for the medication, we also advise them to keep their environment clean and sleep under treated bed net. So, it is difficult for you to see someone come here with malaria, get treated and return to the hospital again because of malaria [Participant 14, Kumasi]. …. integration has helped a lot in controlling malaria in the country. I say that because the medicine has gone through FDA to be tested for its effectiveness and safety. If the patient takes the medications according to the prescription, you realise that most of them recover fully due to our medication. Many times, we are able to even treat severe malaria, so the cases are reducing [Participant 13, Wa]. Figure summarises the study findings guided by Donabedian framework for evaluating quality healthcare. Findings relating to infrastructure have been placed under the ‘structure’ component of the framework, while issues/activities that involved direct health service delivery have been grouped under the ‘process’ component. The effect of structure and process are presented under ‘outcome’ (Fig. ). Thematic analysis of participant narratives revealed five key themes (Table ): Benefits of THM Practice: This theme explored the perceived advantages of THM use. Knowledge/Understanding of THM Integration: This theme examined participants’ understanding of the integrated healthcare model. Structure/Infrastructure for Integrated Care Delivery: This theme addressed the physical resources and settings associated with integrated healthcare delivery. Processes/Activities in Integrated Healthcare Delivery: This theme described the activities and workflows involved in the integrated care model. Outcomes/Consequences of THM Integration on Malaria Control: This theme investigated the perceived impact of THM integration on malaria management in Ghana. Participants highlighted various advantages of THM practice, including its perceived efficacy, minimal side effects, and job creation potential. Participants believed THM offers faster malaria cures due to the body’s familiarity with such remedies. Others acknowledged that although all medications have side effects, those associated with THM were considered negligible. Additionally, some participants viewed the traditional health system as a source of employment, with private THM clinics creating job opportunities for the populace. The quotes that follow expatiate on the views expressed by participants regarding the merits of THM practice in Ghana: I can say herbal medicine cures malaria better than orthodox medicine. Our human system knows herbal medicine. It is like introducing food to your baby…. Our system accepts herbal medicine well [Participant 12, Cape Coast]. All medications have side effects. Once it is an external thing that you are introducing into your body, there will be some side effects. But it is up to the degree of the side effects. When you compare the side effects, it is less for herbal medicines than the orthodox medications [Participant 11, Cape Coast]. …THM has created jobs in Ghana. For example, when you go to Kumasi, there are a lot of herbal clinics/outlets that have been established and these clinics have and keep employing people [Participant 1, Cape Coast]. Participants generally demonstrated awareness of THM integration into the Ghanaian healthcare system. They specifically mentioned the establishment of a Traditional Herbal Medicine program at KNUST as evidence of this integration. I am aware of the integration. Perhaps, that is why they have started teaching THM at KNUST. We have students from ‘Tech’ that are studying THM as a course to become professionals [Participant 5, Cape Coast]. This suggests that participants perceive the inclusion of THM within the formal education system as a significant step towards its legitimisation and integration within mainstream healthcare delivery. This theme addressed the physical resources and settings associated with integrated healthcare delivery. Four key sub-themes emerged: Availability of integrated health facilities Participants acknowledged the presence of THM units within some government hospitals. However, they expressed concern about the limited number of these facilities. They attributed this inadequacy to factors such as low public awareness, insufficient government support, and consequently, low patient utilisation of integrated healthcare services. …we have 55 piloted facilities. That number is small compared to the number of orthodox healthcare facilities in the country. That also accounts for the unpopularity of the THM unit in the country because the more there are herbal medicine facilities, the more people will get to know and patronise such facilities [Participant 11, Cape Coast]. I am aware of the availability of integrated facilities. We have Tafo government hospital, Juaben government hospital, Bekwai, and in Accra we have LEKMA hospital, Police hospital, Tema general hospital, and then the Tamale teaching hospital, Ho government hospital, and Wawra government hospital in the Oti region. So, it is all over the country [Participant 14, Kumasi]. In the Upper West region, there is one there. That is, the Wa municipal hospital. There are about two medical herbalists at the Wa Municipal hospital [Participant 4, Wa]. Participants acknowledged the presence of THM units within some government hospitals. However, they expressed concern about the limited number of these facilities. They attributed this inadequacy to factors such as low public awareness, insufficient government support, and consequently, low patient utilisation of integrated healthcare services. …we have 55 piloted facilities. That number is small compared to the number of orthodox healthcare facilities in the country. That also accounts for the unpopularity of the THM unit in the country because the more there are herbal medicine facilities, the more people will get to know and patronise such facilities [Participant 11, Cape Coast]. I am aware of the availability of integrated facilities. We have Tafo government hospital, Juaben government hospital, Bekwai, and in Accra we have LEKMA hospital, Police hospital, Tema general hospital, and then the Tamale teaching hospital, Ho government hospital, and Wawra government hospital in the Oti region. So, it is all over the country [Participant 14, Kumasi]. In the Upper West region, there is one there. That is, the Wa municipal hospital. There are about two medical herbalists at the Wa Municipal hospital [Participant 4, Wa]. Participants practicing within integrated health facilities highlighted their commitment to quality healthcare delivery. They described adhering to established medical principles and protocols for patient care which included prescribing only herbal products listed on the Ministry of Health’s recommended herbal medicine list. Additionally, some participants viewed regular inspections by the FDA as a means to ensure the quality, efficacy, and safety of the herbal products and services they provide. We have the recommended herbal medicine list. So, the recommendation should be based on the list from the Ministry of Health which has been certified so we stick to defined medical principles [Participant 2, Wa]. These days, once you are a certified THM provider, the FDA will come for inspection and check that your facility is following scientific methods to ensure quality, effectiveness, and safety to those who would use it [Participant 6, Cape Coast]. Participants highlighted a shortage of qualified healthcare personnel within integrated healthcare facilities. They described performing multiple roles beyond their core duties, including procurement, reporting, medication request follow-up, and drug dispensing. This role overload was attributed to the limited number of staff and a perceived lack of personnel with expertise in herbal medicine. Furthermore, some participants suggested that existing medical herbalists might be drawn to other career paths, such as academia or positions with the Ghana FDA, potentially exacerbating the staffing shortage. In our case, we must initiate, follow up with memo to request for medications. They say they don’t know much about herbal medicines, so I must take charge and lead the procurement. So, you become part of pharmacy because you have to monitor and make sure they are dispensing the right drugs. So you move up and down; it is a lot of work because one person you are involved in reporting, payment process, monitoring, and everything [Participant 2, Wa]. …comparing to the country’s population, I don’t think we have enough qualified medical herbalists. The whole of Cape Coast metropolis, we are just five. Can you imagine! [Participant 8, Cape Coast]. The numbers are not enough…. we started herbal medicine in KNUST in 2001 and most of us don’t practice. They enter a different sector like FDA. Those who really practice is very few. They just branch into a different field. Some are in academia and the FDA [Participant 17, Kumasi]. This sub-theme emerged primarily among participants from the forest belt. They reported experiencing disruptions in service delivery due to a shortage of essential herbal medicines. They attributed this shortage to cumbersome procurement processes perceived as overly bureaucratic. Additionally, delays in deliveries from approved manufacturing centres were identified as contributing to stockouts. These shortages reportedly hampered their ability to provide quality care. In contrast, participants from the coastal belt highlighted the issue of inadequate equipment, particularly a lack of computers. This lack of technology was perceived as hindering their productivity and overall job satisfaction. They associated this equipment deficiency with limited or absent funding for integrated healthcare facilities. We sometimes run out of drugs because to order the drugs must pass through so many bureaucratic lines. When I talk about the bureaucratic system, I mean that it must go through many offices before it gets approved. Over here, they will bring the pro forma invoice then we work on it and get the voucher from the account. Then the signatory must come from different people. So, if one signatory has travelled, it becomes a challenge to order the drugs, hence limited or no drugs [Participant 16, Kumasi]. When we order our medicines from Akuapem Mampong, it takes a long time. It delays and that doesn’t help because the client needs it, which disturbs you too. Why should you delay in giving treatment to someone suffering from malaria? [Participant 15, Kumasi]. … We don’t have computers to assist with our work. And today, everything is e-health and so, if I do not come with my personal computer, then it means that I will be unable to work. Computer is a basic equipment that we think if we had, it would improve our healthcare delivery, but we don’t have them due to the lack of funds. It doesn’t make the work enjoyable [Participant 8, Cape Coast]. The main issues identified under this theme were categorised into five sub-themes; health system-based challenges, national/general challenges, patient-centred care, follow ups, and THM training and research. Health system-based challenges This section explores challenges identified by participants within the processes and activities related to integrated healthcare delivery for malaria control. Two issues emerged: First, disapproval from orthodox medicine providers: Participants expressed concerns regarding negative perceptions and attitudes towards THM integration from some orthodox medicine providers. These negative views were perceived to hinder effective collaboration and patient referrals. One example cited by participants was the misconception that THM can cause kidney and lung problems. They reported efforts to address these misconceptions through clinical meetings, but with limited success. Another challenge we are facing, which we are trying to solve has to do with the attitudes and perceptions of the medical doctors and other orthodox practitioners. They have the perception that when you take herbal medicine, it will affect your kidneys and lung. it is a major challenge to the integration. We organise clinical meetings to explain to them, yet they are not convinced [Participant 8, Cape Coast]. ……for the medical doctors, they don’t want to hear about THM. They think that they go to school to learn about scientific medicine. But they don’t know that currently THM providers also go through scientific training. Their opposition is one of our challenges [Participant 3, Wa]. Secondly, participants, particularly those from the coastal and savannah regions, highlighted challenges associated with referral practices within the integrated healthcare system. They reported that orthodox medical doctors often exhibited reluctance to formally refer patients to medical herbalists. Instead, some doctors resorted to informal methods of recommending THM, blurring the lines between a formal referral and casual advice. Participants attributed this reluctance to several factors: Perceived Superiority: Some medical herbalists believed that orthodox doctors held a sense of superiority, viewing them as apprentices rather than qualified healthcare providers. Patient Affordability Concerns: Participants suggested that doctors might hesitate due to concerns about patients’ ability to afford certified herbal medicines offered at integrated facilities, as these often require out-of-pocket payments. It is difficult for the orthodox medicine provider to accept and refer patients to us. For those who refer, they do it ‘backdoor’, in the form of an advice rather than as a formal referral [Participant 6, Cape Coast]. …the medical doctors prefer to treat malaria cases rather than referring to us…like I said, the herbal medications are cash-and-carry. So, the doctors are concerned about the cash. They are not certain that the patient will pay for the medication, and so, they prefer to treat the patient with their medication rather than referring them for herbal treatment [Participant 8, Cape Coast]. For my practice over ten years, no medical officer has referred a patient to me. We have a referral form that we fill. The challenge is that they frown on our referral forms. So, what we do is to refer the person verbally…. we do it in an informal setting. we don’t have a problem referring patients to the orthodox. But they will not do that. Why will they [orthodox] do that? They see themselves as superior, so why would they want to refer to us? They [orthodox providers] see us as apprentice. [Participant 19, Wa]. National/general challenges This theme highlighted issues at the national level that hindered effective management of malaria through the practice of integrated healthcare. Major national/general challenges that emerged were non-exhaustive NHIS coverage and inadequate promotional activities on THM integration. Participants narrated that the non-comprehensive nature of the NHIS (exclusion of herbal anti-malarial drugs) accounts for low patronage of services at THM clinics at integrated hospitals because clients always opt for free orthodox anti-malaria medicines rather than paying full cost for the herbal ones. …. the malaria medication for orthodox is on the health insurance but when you opt for the herbal medicine, then you will be paying about ghs80 or more. So, free or ghs80? The cost involved is what is accounting for the low patronage of herbal medicines for malaria in our THM clinics [Participant 18, Kumasi]. Promotional activities play a vital role in the successful implementation of interventions. When the providers were asked to share their experiences concerning publicity of integration and its implication on malaria control, they gave an account, which suggests inadequacy of promotional activities regarding the integration programme. … One major challenge we are facing with this integration is the lack of publicity or awareness. Most people don’t know about this integration. The Ministry of Health should have projected the integration right from the onset. Now, when you visit these selected hospitals (integrated facilities), there are herbal practitioners available, but people do not know about that, so they do not patronise our services [Participant 8, Cape Coast]. Patient-centred care When discussing patient care, participants recounted that attending to their clients on time and spending ample time with them is something they were proud of. They emphasised that the patients really appreciated that kind of care unlike the orthodox unit, where less time is spent. The ensuing quote represent this finding: It is something we are very proud of at our unit. We spend a lot of time, about an hour with one patient.…. They really do appreciate that unlike the orthodox side that within 2–3 minutes they are done and don’t have time for them. So, when it comes to patient-provider relationship, it is the best in our unit [Participant 11, Cape Coast]. Follow ups We found that follow ups, which is a feature of functional health systems was well implemented by the participants. They narrated that their desire to promote the welfare of clients motivated them to follow up and conduct medical reviews to avert undesirable treatment outcomes. If you are interested in a case, then you follow up! There are others (clients) that I had to give my contact for them to call back. I do this to avoid any adverse effect of the treatment or medication [Participant 2, Wa]. … Some of the clients come for review. Through the follow ups we schedule reviews. After taking the THM, we don’t let them stay home. We schedule reviews and when they come, we test and make sure they are fine [Participant 15, Kumasi]. We also do a lot of follow up on our patients. We call to check up on them and find out whether they have had any side effects [Participant 8, Cape Coast]. This section explores challenges identified by participants within the processes and activities related to integrated healthcare delivery for malaria control. Two issues emerged: First, disapproval from orthodox medicine providers: Participants expressed concerns regarding negative perceptions and attitudes towards THM integration from some orthodox medicine providers. These negative views were perceived to hinder effective collaboration and patient referrals. One example cited by participants was the misconception that THM can cause kidney and lung problems. They reported efforts to address these misconceptions through clinical meetings, but with limited success. Another challenge we are facing, which we are trying to solve has to do with the attitudes and perceptions of the medical doctors and other orthodox practitioners. They have the perception that when you take herbal medicine, it will affect your kidneys and lung. it is a major challenge to the integration. We organise clinical meetings to explain to them, yet they are not convinced [Participant 8, Cape Coast]. ……for the medical doctors, they don’t want to hear about THM. They think that they go to school to learn about scientific medicine. But they don’t know that currently THM providers also go through scientific training. Their opposition is one of our challenges [Participant 3, Wa]. Secondly, participants, particularly those from the coastal and savannah regions, highlighted challenges associated with referral practices within the integrated healthcare system. They reported that orthodox medical doctors often exhibited reluctance to formally refer patients to medical herbalists. Instead, some doctors resorted to informal methods of recommending THM, blurring the lines between a formal referral and casual advice. Participants attributed this reluctance to several factors: Perceived Superiority: Some medical herbalists believed that orthodox doctors held a sense of superiority, viewing them as apprentices rather than qualified healthcare providers. Patient Affordability Concerns: Participants suggested that doctors might hesitate due to concerns about patients’ ability to afford certified herbal medicines offered at integrated facilities, as these often require out-of-pocket payments. It is difficult for the orthodox medicine provider to accept and refer patients to us. For those who refer, they do it ‘backdoor’, in the form of an advice rather than as a formal referral [Participant 6, Cape Coast]. …the medical doctors prefer to treat malaria cases rather than referring to us…like I said, the herbal medications are cash-and-carry. So, the doctors are concerned about the cash. They are not certain that the patient will pay for the medication, and so, they prefer to treat the patient with their medication rather than referring them for herbal treatment [Participant 8, Cape Coast]. For my practice over ten years, no medical officer has referred a patient to me. We have a referral form that we fill. The challenge is that they frown on our referral forms. So, what we do is to refer the person verbally…. we do it in an informal setting. we don’t have a problem referring patients to the orthodox. But they will not do that. Why will they [orthodox] do that? They see themselves as superior, so why would they want to refer to us? They [orthodox providers] see us as apprentice. [Participant 19, Wa]. This theme highlighted issues at the national level that hindered effective management of malaria through the practice of integrated healthcare. Major national/general challenges that emerged were non-exhaustive NHIS coverage and inadequate promotional activities on THM integration. Participants narrated that the non-comprehensive nature of the NHIS (exclusion of herbal anti-malarial drugs) accounts for low patronage of services at THM clinics at integrated hospitals because clients always opt for free orthodox anti-malaria medicines rather than paying full cost for the herbal ones. …. the malaria medication for orthodox is on the health insurance but when you opt for the herbal medicine, then you will be paying about ghs80 or more. So, free or ghs80? The cost involved is what is accounting for the low patronage of herbal medicines for malaria in our THM clinics [Participant 18, Kumasi]. Promotional activities play a vital role in the successful implementation of interventions. When the providers were asked to share their experiences concerning publicity of integration and its implication on malaria control, they gave an account, which suggests inadequacy of promotional activities regarding the integration programme. … One major challenge we are facing with this integration is the lack of publicity or awareness. Most people don’t know about this integration. The Ministry of Health should have projected the integration right from the onset. Now, when you visit these selected hospitals (integrated facilities), there are herbal practitioners available, but people do not know about that, so they do not patronise our services [Participant 8, Cape Coast]. When discussing patient care, participants recounted that attending to their clients on time and spending ample time with them is something they were proud of. They emphasised that the patients really appreciated that kind of care unlike the orthodox unit, where less time is spent. The ensuing quote represent this finding: It is something we are very proud of at our unit. We spend a lot of time, about an hour with one patient.…. They really do appreciate that unlike the orthodox side that within 2–3 minutes they are done and don’t have time for them. So, when it comes to patient-provider relationship, it is the best in our unit [Participant 11, Cape Coast]. We found that follow ups, which is a feature of functional health systems was well implemented by the participants. They narrated that their desire to promote the welfare of clients motivated them to follow up and conduct medical reviews to avert undesirable treatment outcomes. If you are interested in a case, then you follow up! There are others (clients) that I had to give my contact for them to call back. I do this to avoid any adverse effect of the treatment or medication [Participant 2, Wa]. … Some of the clients come for review. Through the follow ups we schedule reviews. After taking the THM, we don’t let them stay home. We schedule reviews and when they come, we test and make sure they are fine [Participant 15, Kumasi]. We also do a lot of follow up on our patients. We call to check up on them and find out whether they have had any side effects [Participant 8, Cape Coast]. Participants recognised that the practice of integration has boosted research on herbal remedies for treating malaria. They perceived the herbal anti-malarial medications prescribed and utilised within the integrated system to be safe and effective because they go through scientific scrutiny at the Centre for Research into Plant Medicine and/or the Nogouchi Memorial Institute for Medical Research. The quotes below reiterate this finding: I think it has boosted research on THM for malaria control because herbal anti-malaria drugs must be tested either at Nogouchi Memorial Institute for Medical Research or Centre for Research into Plant Medicine to verify whether it is safe for human consumption [Participant 1, Cape Coast]. Our THM comes from the Centre for Research into Plant Medicine; a lot of research has gone into these products, and they are safe for use. They are now safe and efficacious for treating malaria [Participant 15, Kumasi]. In addition, participants within the forest belt reported that they usually undergo continuous professional development training. According to them, such trainings were organised by the traditional and alternative medicine directorate and involved the interpretation of scientific medical activities such as x-ray reading, understanding laboratory results and clinical emergencies. They believed that this knowledge enabled them to interpret and confirm diagnosis of diseases, especially malaria. The quotations below summarise this report: …. as medical herbalists we go for continuous professional development training. The training is mostly dependent on current situations. So, if there is a new herbal medication or new ways of doing things, we do that to improve upon the knowledge that we have. If there are research works going on, they update us on findings. We also do research because science is always updating. The training is mostly done by the Traditional and Alternative Medicine Directorate. They organise and get resource people to take us through it [Participant 14, Kumasi]. We have been going for continuous professional development (CPD) training. We do it yearly to upgrade our practice and the training largely depends on the topics that we deal with in that year. We have received training on the interpretation of x-rays, lab results and clinical emergencies…. the knowledge acquired enable us to request for lab testing and interpret the results to confirm our diagnosis for malaria [Participant 15, Kumasi]. Interestingly, participants at the savannah belt believed that THM training focuses on exposing medical herbalists to both modern/orthodox and traditional medicines. However, the same could not be said for orthodox medicine providers where training is more focused on orthodox approaches. This unbalanced training was perceived to have created knowledge gaps among mainstream healthcare providers, leading to a lack of understanding and appreciation of the contribution of THM to healthcare delivery in Ghana. Some of the orthodox practitioners do not understand herbal medicine. We are trained to understand both orthodox and herbal medicines, but they are not trained that way. We have an appreciation of both sides but for them, they weren’t trained like that. They were trained one-sided and so they don’t understand herbal medicine, creating opposition from them [Participant 2, Wa]. A key aspect of the study was to explore the perceived impact of THM integration on malaria control. Participants identified two main impacts - reduced pressure/burden on orthodox medicine providers and evidence-based THM practice leading to appropriate management of malaria. The participants perceived that their presence at public hospitals reduced pressure on orthodox medical doctors because they mostly treat malaria cases at their units and that the majority of the clients prefer herbal anti-malaria medication such as the ‘Mebeema’. When you take the cases we treat in our unit, malaria is among the top five. So, if the orthodox doctor was treating about 1 , 000 cases, the numbers have reduced now because most people want to use the herbal anti malaria medicine, that is, Mebeema [Participant 11, Cape Coast]. …. even in our facility, we have about three different drugs for malaria. The most potent one is Mebeema from Akuapem Mampong. It has been used in the system for long and the response has been very good. Lots of people who come to the unit opt for it. Clearly, our presence here at the hospital has reduced pressure on the orthodox providers [Participant 16, Kumasi]. Besides, reducing pressure on orthodox medical doctors, participants irrespective of location, reported that the integration of THM has been beneficial because malaria cases are now appropriately treated through the evidence-based practice of their field. They believed the application of scientific clinical procedures such as correct diagnosis through laboratory testing/prescription, health education, and FDA approval of herbal anti malarial medications have led to effective malaria treatment, hence reducing prevalence. The following quotes elaborate on the views expressed by the study participants. I will say that about 50% of our clients come for malaria treatment. Just as they do for orthodox medicine, we also make sure that the client is taken through education, then we ask them to get tested through the lab and confirm they have malaria before we allow them to take our medication. Evidently, we apply scientific clinical procedures in treating our clients, and it has proven effective because the cases are reducing, and people are getting healed quickly [Participant 9, Cape Coast]. … integration has helped. …. At first, they will go to the market and get anything. But now, they (patients) come to the hospital and go through the process before receiving treatment. Malaria cases have gone down in recent times. When they come in for the medication, we also advise them to keep their environment clean and sleep under treated bed net. So, it is difficult for you to see someone come here with malaria, get treated and return to the hospital again because of malaria [Participant 14, Kumasi]. …. integration has helped a lot in controlling malaria in the country. I say that because the medicine has gone through FDA to be tested for its effectiveness and safety. If the patient takes the medications according to the prescription, you realise that most of them recover fully due to our medication. Many times, we are able to even treat severe malaria, so the cases are reducing [Participant 13, Wa]. Figure summarises the study findings guided by Donabedian framework for evaluating quality healthcare. Findings relating to infrastructure have been placed under the ‘structure’ component of the framework, while issues/activities that involved direct health service delivery have been grouped under the ‘process’ component. The effect of structure and process are presented under ‘outcome’ (Fig. ). This qualitative study employed Donabedian’s framework to explore the effects of integrated healthcare on malaria control in Ghana. Thematic analysis revealed structural and process-related factors influencing the integration practice and its impact on malaria management. These factors can operate independently or interact with underlying issues. Structural considerations Quality assurance, availability of integrated facilities, staffing levels, and medication/equipment availability emerged as key structural elements. The study suggests that integration promotes quality care for malaria patients. Medical herbalists, both in private clinics and integrated facilities, reported adhering to acceptable medical practices and undergoing regular FDA inspections, ensuring the quality of their services. This indicates that the traditional herbal system could be a viable alternative to the mainstream system if providers consistently follow appropriate procedures. The study findings align with those of Ampomah, Malau-Aduli , since the participants indicated their awareness of the practice and existence of integrated hospitals in Ghana. However, in the current study, the participants felt that the number of integrated hospitals was insufficient compared to the national population. They attributed this inadequacy to low governmental support for integration and low community utilisation of these services. With a growing population and a stagnant number of integrated facilities, access for many Ghanaians, particularly in rural areas, would be limited. The results further showed gender discrepancy among medical herbalists in Ghana, where there are more males than female providers; however, this might not serve as a hindrance to access (not prevent patients especially females from seeking care at integrated hospitals) because malaria is not an intimate health problem. Challenges with personnel, equipment, and procurement Findings revealed that medical herbalists handle a wide range of administrative and dispensary tasks to maintain service delivery within the integrated system. This is likely due to insufficient number of registered medical herbalists and limited support from orthodox healthcare providers, who may lack expertise in THM. This highlights a critical gap in personnel recruitment and deployment for successful integration implementation. These findings resonate with previous studies emphasising the importance of training/professional development for both providers to aid quality service delivery . Another key issue was inadequate equipment and medication supply within THM units of integrated facilities. While participants reported access to some laboratory services, those in the coastal region lacked essential computers and software for e-health practices. Donabedian’s framework emphasises the importance of equipment/drug supply and its availability for quality healthcare . Participants in the forest zone reported disruptions due to insufficient herbal anti-malarial medications, attributing it to bureaucratic procurement processes. The issue of shortage of certified herbal medications aligns with findings from an earlier study and highlights a problem prevalent in low- and middle-income countries . A re-evaluation of procurement procedures within the healthcare system is necessary to eliminate bureaucratic obstacles hindering appropriate treatment delivery. Deliberations on process Participants identified challenges related to healthcare system processes. A prominent concern was the disapproval of orthodox healthcare providers towards THM due to negative perceptions. This has also been reported in previous studies and hinders collaboration in managing malaria cases. Addressing these “erroneous notions” requires increased education for both healthcare provider groups. Another challenge was the prevalence of informal referrals instead of formal ones, echoing findings from Ashanti , Brong Ahafo , and Northern parts of Ghana . This potentially leads patients to seek care from uncertified THM providers, jeopardising their health. To improve access to safe herbal anti-malarial treatment and enhance user well-being, it is crucial to revise or develop integrated policies to revamp referral structures. The limited scope of the NHIS and inadequate promotion of integrated healthcare were perceived by participants in this study as national challenges. The NHIS, designed to address healthcare access issues , was found to be insufficient for covering herbal anti-malarial therapies within integrated facilities. This aligns with previous studies among health service users , providers, and hospital administrators , where NHIS limitations hindered patients’ access to THM clinics, consequently undermining the program’s effectiveness. Policy reform towards a more patient-centred healthcare system is needed to ensure equitable access. The Donabedian framework emphasises patient-provider respect and trust . This was evident in our study, as participants described their patient-centred approach, including extended consultations and follow-up care to promote overall well-being and minimise adverse treatment effects. This finding resonates with research in Ghana’s Ashanti region where users and providers reported a compassionate approach to healthcare delivery. Such an attitude might encourage users to continue seeking care from THM providers. The study confirmed that medical herbalists across facilities receive formal training. Participants in the forest zone attributed their ability to interpret lab results and diagnose effectively to ongoing professional development programs. However, those in the savannah belt felt hampered in collaborating with orthodox counterparts due to imbalanced training. Medical herbalists receive training on both traditional and modern healthcare, while their orthodox counterparts lack exposure to THM. This perceived knowledge gap hinders appreciation for traditional therapies and their role in healthcare delivery. Studies , have shown similar findings, where orthodox practitioners acknowledge their lack of THM knowledge as a barrier to collaboration. While trained medical herbalists can expand Ghana’s healthcare workforce, unbalanced training may disrupt service delivery, especially for malaria patients . Outcome – effects of structure and process on malaria control in Ghana The Donabedian framework suggests that good structures and processes should lead to positive health outcomes . While some structural and process factors were not ideal, most participants believed that integrated healthcare has reduced pressure on orthodox providers, who frequently manage malaria cases. The availability of medical herbalists in public hospitals was seen to contribute to decreased malaria prevalence in Ghana through evidence based THM practices. This finding offers a novel contribution to literature. Practice implications The integration of herbal therapies into mainstream healthcare reflects an attempt to regulate and manage healthcare users, providers, and herbal medication use. Patients are directed towards trained medical herbalists who follow established medical standards, including scientific manufacturing, packaging, and administration under medical facility supervision. This aligns with a Vietnamese study which described herbal medicine modernisation as a means to safeguard public health. While integration primarily aims to protect public health and well-being, it also serves to regulate herbal remedy usage within a population. As this research demonstrates, the Ghanaian government aims to provide patients with a single point of access for malaria treatment through integration. However, achieving this health goal is hindered by national and health system-based challenges. Strengths and limitations This study is one of the few to explore the contribution of integrated healthcare to malaria control in Ghana. A key strength lies in the inclusion of medical herbalists from diverse regions (coastal, forest, and savannah), providing insights from a crucial stakeholder group directly involved in healthcare delivery. The study offers valuable knowledge by examining the contribution of integrated medicine to malaria control. However, limitations exist. The small sample size, chosen through non-probabilistic sampling, reduces the findings’ representativeness and generalisability. While the study successfully recruited medical herbalists as appropriate participants, it is important to acknowledge the potential influence of personal biases on their perspectives. This could have led to overestimation or exaggeration of certain narratives. Furthermore, the exclusive use of face-to-face individual in-depth interviews as the data collection method may have limited the generalisability and reliability of the study findings. Quality assurance, availability of integrated facilities, staffing levels, and medication/equipment availability emerged as key structural elements. The study suggests that integration promotes quality care for malaria patients. Medical herbalists, both in private clinics and integrated facilities, reported adhering to acceptable medical practices and undergoing regular FDA inspections, ensuring the quality of their services. This indicates that the traditional herbal system could be a viable alternative to the mainstream system if providers consistently follow appropriate procedures. The study findings align with those of Ampomah, Malau-Aduli , since the participants indicated their awareness of the practice and existence of integrated hospitals in Ghana. However, in the current study, the participants felt that the number of integrated hospitals was insufficient compared to the national population. They attributed this inadequacy to low governmental support for integration and low community utilisation of these services. With a growing population and a stagnant number of integrated facilities, access for many Ghanaians, particularly in rural areas, would be limited. The results further showed gender discrepancy among medical herbalists in Ghana, where there are more males than female providers; however, this might not serve as a hindrance to access (not prevent patients especially females from seeking care at integrated hospitals) because malaria is not an intimate health problem. Findings revealed that medical herbalists handle a wide range of administrative and dispensary tasks to maintain service delivery within the integrated system. This is likely due to insufficient number of registered medical herbalists and limited support from orthodox healthcare providers, who may lack expertise in THM. This highlights a critical gap in personnel recruitment and deployment for successful integration implementation. These findings resonate with previous studies emphasising the importance of training/professional development for both providers to aid quality service delivery . Another key issue was inadequate equipment and medication supply within THM units of integrated facilities. While participants reported access to some laboratory services, those in the coastal region lacked essential computers and software for e-health practices. Donabedian’s framework emphasises the importance of equipment/drug supply and its availability for quality healthcare . Participants in the forest zone reported disruptions due to insufficient herbal anti-malarial medications, attributing it to bureaucratic procurement processes. The issue of shortage of certified herbal medications aligns with findings from an earlier study and highlights a problem prevalent in low- and middle-income countries . A re-evaluation of procurement procedures within the healthcare system is necessary to eliminate bureaucratic obstacles hindering appropriate treatment delivery. Deliberations on process Participants identified challenges related to healthcare system processes. A prominent concern was the disapproval of orthodox healthcare providers towards THM due to negative perceptions. This has also been reported in previous studies and hinders collaboration in managing malaria cases. Addressing these “erroneous notions” requires increased education for both healthcare provider groups. Another challenge was the prevalence of informal referrals instead of formal ones, echoing findings from Ashanti , Brong Ahafo , and Northern parts of Ghana . This potentially leads patients to seek care from uncertified THM providers, jeopardising their health. To improve access to safe herbal anti-malarial treatment and enhance user well-being, it is crucial to revise or develop integrated policies to revamp referral structures. The limited scope of the NHIS and inadequate promotion of integrated healthcare were perceived by participants in this study as national challenges. The NHIS, designed to address healthcare access issues , was found to be insufficient for covering herbal anti-malarial therapies within integrated facilities. This aligns with previous studies among health service users , providers, and hospital administrators , where NHIS limitations hindered patients’ access to THM clinics, consequently undermining the program’s effectiveness. Policy reform towards a more patient-centred healthcare system is needed to ensure equitable access. The Donabedian framework emphasises patient-provider respect and trust . This was evident in our study, as participants described their patient-centred approach, including extended consultations and follow-up care to promote overall well-being and minimise adverse treatment effects. This finding resonates with research in Ghana’s Ashanti region where users and providers reported a compassionate approach to healthcare delivery. Such an attitude might encourage users to continue seeking care from THM providers. The study confirmed that medical herbalists across facilities receive formal training. Participants in the forest zone attributed their ability to interpret lab results and diagnose effectively to ongoing professional development programs. However, those in the savannah belt felt hampered in collaborating with orthodox counterparts due to imbalanced training. Medical herbalists receive training on both traditional and modern healthcare, while their orthodox counterparts lack exposure to THM. This perceived knowledge gap hinders appreciation for traditional therapies and their role in healthcare delivery. Studies , have shown similar findings, where orthodox practitioners acknowledge their lack of THM knowledge as a barrier to collaboration. While trained medical herbalists can expand Ghana’s healthcare workforce, unbalanced training may disrupt service delivery, especially for malaria patients . Outcome – effects of structure and process on malaria control in Ghana The Donabedian framework suggests that good structures and processes should lead to positive health outcomes . While some structural and process factors were not ideal, most participants believed that integrated healthcare has reduced pressure on orthodox providers, who frequently manage malaria cases. The availability of medical herbalists in public hospitals was seen to contribute to decreased malaria prevalence in Ghana through evidence based THM practices. This finding offers a novel contribution to literature. Practice implications The integration of herbal therapies into mainstream healthcare reflects an attempt to regulate and manage healthcare users, providers, and herbal medication use. Patients are directed towards trained medical herbalists who follow established medical standards, including scientific manufacturing, packaging, and administration under medical facility supervision. This aligns with a Vietnamese study which described herbal medicine modernisation as a means to safeguard public health. While integration primarily aims to protect public health and well-being, it also serves to regulate herbal remedy usage within a population. As this research demonstrates, the Ghanaian government aims to provide patients with a single point of access for malaria treatment through integration. However, achieving this health goal is hindered by national and health system-based challenges. Strengths and limitations This study is one of the few to explore the contribution of integrated healthcare to malaria control in Ghana. A key strength lies in the inclusion of medical herbalists from diverse regions (coastal, forest, and savannah), providing insights from a crucial stakeholder group directly involved in healthcare delivery. The study offers valuable knowledge by examining the contribution of integrated medicine to malaria control. However, limitations exist. The small sample size, chosen through non-probabilistic sampling, reduces the findings’ representativeness and generalisability. While the study successfully recruited medical herbalists as appropriate participants, it is important to acknowledge the potential influence of personal biases on their perspectives. This could have led to overestimation or exaggeration of certain narratives. Furthermore, the exclusive use of face-to-face individual in-depth interviews as the data collection method may have limited the generalisability and reliability of the study findings. Participants identified challenges related to healthcare system processes. A prominent concern was the disapproval of orthodox healthcare providers towards THM due to negative perceptions. This has also been reported in previous studies and hinders collaboration in managing malaria cases. Addressing these “erroneous notions” requires increased education for both healthcare provider groups. Another challenge was the prevalence of informal referrals instead of formal ones, echoing findings from Ashanti , Brong Ahafo , and Northern parts of Ghana . This potentially leads patients to seek care from uncertified THM providers, jeopardising their health. To improve access to safe herbal anti-malarial treatment and enhance user well-being, it is crucial to revise or develop integrated policies to revamp referral structures. The limited scope of the NHIS and inadequate promotion of integrated healthcare were perceived by participants in this study as national challenges. The NHIS, designed to address healthcare access issues , was found to be insufficient for covering herbal anti-malarial therapies within integrated facilities. This aligns with previous studies among health service users , providers, and hospital administrators , where NHIS limitations hindered patients’ access to THM clinics, consequently undermining the program’s effectiveness. Policy reform towards a more patient-centred healthcare system is needed to ensure equitable access. The Donabedian framework emphasises patient-provider respect and trust . This was evident in our study, as participants described their patient-centred approach, including extended consultations and follow-up care to promote overall well-being and minimise adverse treatment effects. This finding resonates with research in Ghana’s Ashanti region where users and providers reported a compassionate approach to healthcare delivery. Such an attitude might encourage users to continue seeking care from THM providers. The study confirmed that medical herbalists across facilities receive formal training. Participants in the forest zone attributed their ability to interpret lab results and diagnose effectively to ongoing professional development programs. However, those in the savannah belt felt hampered in collaborating with orthodox counterparts due to imbalanced training. Medical herbalists receive training on both traditional and modern healthcare, while their orthodox counterparts lack exposure to THM. This perceived knowledge gap hinders appreciation for traditional therapies and their role in healthcare delivery. Studies , have shown similar findings, where orthodox practitioners acknowledge their lack of THM knowledge as a barrier to collaboration. While trained medical herbalists can expand Ghana’s healthcare workforce, unbalanced training may disrupt service delivery, especially for malaria patients . The Donabedian framework suggests that good structures and processes should lead to positive health outcomes . While some structural and process factors were not ideal, most participants believed that integrated healthcare has reduced pressure on orthodox providers, who frequently manage malaria cases. The availability of medical herbalists in public hospitals was seen to contribute to decreased malaria prevalence in Ghana through evidence based THM practices. This finding offers a novel contribution to literature. The integration of herbal therapies into mainstream healthcare reflects an attempt to regulate and manage healthcare users, providers, and herbal medication use. Patients are directed towards trained medical herbalists who follow established medical standards, including scientific manufacturing, packaging, and administration under medical facility supervision. This aligns with a Vietnamese study which described herbal medicine modernisation as a means to safeguard public health. While integration primarily aims to protect public health and well-being, it also serves to regulate herbal remedy usage within a population. As this research demonstrates, the Ghanaian government aims to provide patients with a single point of access for malaria treatment through integration. However, achieving this health goal is hindered by national and health system-based challenges. This study is one of the few to explore the contribution of integrated healthcare to malaria control in Ghana. A key strength lies in the inclusion of medical herbalists from diverse regions (coastal, forest, and savannah), providing insights from a crucial stakeholder group directly involved in healthcare delivery. The study offers valuable knowledge by examining the contribution of integrated medicine to malaria control. However, limitations exist. The small sample size, chosen through non-probabilistic sampling, reduces the findings’ representativeness and generalisability. While the study successfully recruited medical herbalists as appropriate participants, it is important to acknowledge the potential influence of personal biases on their perspectives. This could have led to overestimation or exaggeration of certain narratives. Furthermore, the exclusive use of face-to-face individual in-depth interviews as the data collection method may have limited the generalisability and reliability of the study findings. Key findings indicate high awareness of integration among Ghanaian medical herbalists. Participants believe that quality assurance, designated facilities, patient-centred care, follow-up practices, and continuous professional training for medical herbalists contribute to reduced burden on orthodox healthcare providers and evidence-based THM practices leading to improved malaria management. However, structural barriers (inadequate personnel, medication, and equipment) and process challenges (limited NHIS coverage, inadequate promotion, ineffective referrals) hinder integration’s contribution to malaria control. These findings, therefore, offer the model or baseline information and a platform for further discussion towards improving the Ghanaian integrated health system to serve as a tool for achieving a malaria free country. Eradicating malaria through integrated healthcare requires policy modifications and improved implementation. Hence, we offer the following recommendations: the creation of herbal department at KNUST is impressive, however, the educational policies need to be revised to expand herbal medicine training to other institutions and incorporate herbal medicine into the curriculum of biomedical healthcare providers. This might enhance the conventional healthcare practitioners’ understanding on herbal medicine and improve collaborations between medical herbalists and their biomedical colleagues. The government needs to formulate policies that would increase public awareness about the practice of integrated healthcare (availability of medical herbalists and certified herbal anti-malarial medications in government hospitals) among health service users. Increased awareness could lead to improved access to such services. Future studies could focus on developing healthcare frameworks tailored to eliminate infectious diseases, particularly malaria, that prioritise quality service delivery within an integrated system. Below is the link to the electronic supplementary material. Supplementary Material 1 Supplementary Material 2 |
Obtaining person-related information from employees with chronic health problems: a focus group study | 2990c643-b3bb-44cd-9616-937ecd0691fd | 6768897 | Preventive Medicine[mh] | Having a chronic disease can negatively impact participation in work (De Boer et al. ; Jetha et al. ). Occupational physicians (OPs) and insurance physicians (IPs) can play an important role in increasing work participation and limiting sickness absence under employees with a chronic disease, by intervening on factors which influence work participation (Dekkers-Sánchez et al. ; Verbeek ). Certain perceptions and cognitions—such as motivation, self-efficacy, and expectations regarding recovery or return to work (RTW)—are important person-related factors that influence work participation (Hallegraeff et al. ; Boyle et al. ; Besen et al. ). A systematic review by De Wit et al. demonstrated an association between work participation and ten person-related factors: expectations regarding recovery or RTW, optimism/pessimism, self-efficacy, motivation, feelings of control, perceived health, coping strategies, fear-avoidance beliefs, perceived work-relatedness and catastrophizing. For example, catastrophizing and fear-avoidance beliefs were associated with an increased time until RTW, whereas having positive expectations concerning RTW or recovery was a predictor of a shorter time until RTW (De Wit et al. ). Previous qualitative research has shown that both employees and physicians view person-related factors as important in work participation (Dekkers-Sánchez et al. ; Matérne et al. ; Van Muijen et al. ; Dunn et al. ; Wilbanks and Ivankova ), making these factors key targets for interventions to increase work participation. To intervene effectively on these factors, it is imperative that OPs and IPs are able to obtain information concerning those person-related factors that encourage or hinder work participation in employees. This can be achieved through physician–patient interaction during consultations. However, to obtain information concerning these factors, it is crucial that employees disclose information about these factors. Physician use of specific communication skills, such as asking open-ended questions and active listening, can encourage patients to share information about themselves (Ha and Longnecker ; Lewis et al. ; Matusitz and Spear ). It is possible that these techniques may also encourage employees to disclose more information concerning person-related factors during consultations. This is, however, dependent on the communication skills of the individual physician. Physicians and patients can differ in their interpretation of physician communication skills; physicians who think they are communicating well may not always be perceived as good communicators by their patients (Kenny et al. ). These discrepancies can further limit the disclosure of important patient information, such as that concerning person-related factors. To enhance physician–patient communication and facilitate the disclosure of information regarding relevant person-related factors, it is important to evaluate patients’ opinions concerning how these factors should be discussed. The opinion of employees regarding how physicians should obtain person-related information is, however, yet unstudied. This study, therefore, poses the research question: what is the most effective way for OPs and IPs to obtain information concerning person-related factors, in the opinion of employees with chronic health problems? This qualitative study utilizes three focus group discussions (FGDs). We chose this study method because FGDs allow for the collation of a diverse range of participants and opinions: for example, through the inclusion of employees with different disabilities and different experiences with OPs and IPs. The moderator of a FGD can respond to questions from participants about complex or academic subjects (e.g. person-related factors) and can request more detailed responses from participants when clarification of their responses is needed (Wong ). The consolidated criteria for reporting qualitative research (COREQ) were used to comprehensively report the focus group process (Tong et al. ). Participants FGD participants were recruited via a panel of more than 23,500 patients from the Patient Federation in the Netherlands, an association representing 170 patient and consumer organisations. In February 2018, members of the panel were invited by email to participate in one of the focus groups. In addition, four consumer organisations affiliated with the Patient Federation (Lung Foundation Netherlands, Heart Council, Kidney Patients Association Netherlands and Care Importance Brabant) were approached and agreed to send invitations to their members. Individuals were eligible to participate if they were employees who had experienced limitations during paid work due to chronic health problems, spoke Dutch fluently and were between 18 and 67 years of age. Employees who expressed interest in participating received information by email detailing the purpose of the FGDs, the person-related factors that would be discussed, the professional background of the interviewers, and possible dates for the FGDs. Thirty employees agreed to participate in the study. Participants were assigned to one of the three focus groups, with the aim of achieving an equal spread of gender, age and disabilities over the groups. Three of the 30 employees who agreed to take part in the study were unable to participate due to other appointments or due to health problems. Four employees did not attend for reasons unknown. In total, 23 employees participated in the study, divided between the three focus groups (focus group A and B both had seven employees, and focus group C consisted of nine employees). Demographics of the participants are presented in Table . Procedure The three FGDs were conducted between March and April 2018 at the Amsterdam UMC, location Academic Medical Center in Amsterdam. The moderator for each FGD was one of two male authors (CH or HW), respectively OP and IP. Both are employed at the Coronel Institute of Occupational Health, have a Doctorate of Medicine and of Philosophy and have previous experience in qualitative research and conducting FGDs. The discussions were recorded with an audio recorder, and field notes were taken by another author (MdW). The authors did not know the participants before the FGDs. Apart from the researchers and participants, no-one else was present during the FGDs. Before the start of each 2-h FGD—all of which were conducted in Dutch—each participant signed an informed consent form. The FGDs started with an explanation of the purpose of the discussion, a brief introduction of the participants and an explanation of the structure of the FGD by the moderator. During the discussion that followed, the primary question addressed was: what is the most effective way for OPs and IPs to obtain information concerning person-related factors? The person-related factors defined were ten factors identified in a preceding systematic review (De Wit et al. ). The person-related factors were explained through ten case descriptions, presenting fictional situations in which the factor in question influenced the work participation of an employee with chronic health problems. During the discussion, the participants were encouraged to speak openly about their views and thoughts. When needed, the moderator asked the participants to clarify their answers. At the end of each FGD, participants received a travel allowance and a gift card of 25 euros in return for their participation. Data analysis The recordings of the discussions were transcribed verbatim and anonymized. We did not send the transcripts back to the participants for comments or correction, and we did not ask for feedback on the findings. For data analysis purposes, we used qualitative content analysis (Mayring ). The transcripts from the FGDs were coded using MAXQDA 12 Software (VERBI Software ). Codes were assigned by one author (MdW) to segments of the transcript of the first two FGDs. These were then checked by a second author (HW). Disagreements about the coding were resolved by discussion. A coding framework consisting of main themes and subthemes was built by categorizing the codes. The main themes and subthemes were discussed between all authors until a consensus about the framework was reached. Following author consensus regarding the codes and coding framework, the transcript of the third FGD was coded using the coding framework by one author (MdW). The different themes of the coding framework are described in the “ ” section. To illustrate our findings, we have included quotations of participant discussions from the focus groups. A native English speaker translated these from Dutch into English. FGD participants were recruited via a panel of more than 23,500 patients from the Patient Federation in the Netherlands, an association representing 170 patient and consumer organisations. In February 2018, members of the panel were invited by email to participate in one of the focus groups. In addition, four consumer organisations affiliated with the Patient Federation (Lung Foundation Netherlands, Heart Council, Kidney Patients Association Netherlands and Care Importance Brabant) were approached and agreed to send invitations to their members. Individuals were eligible to participate if they were employees who had experienced limitations during paid work due to chronic health problems, spoke Dutch fluently and were between 18 and 67 years of age. Employees who expressed interest in participating received information by email detailing the purpose of the FGDs, the person-related factors that would be discussed, the professional background of the interviewers, and possible dates for the FGDs. Thirty employees agreed to participate in the study. Participants were assigned to one of the three focus groups, with the aim of achieving an equal spread of gender, age and disabilities over the groups. Three of the 30 employees who agreed to take part in the study were unable to participate due to other appointments or due to health problems. Four employees did not attend for reasons unknown. In total, 23 employees participated in the study, divided between the three focus groups (focus group A and B both had seven employees, and focus group C consisted of nine employees). Demographics of the participants are presented in Table . The three FGDs were conducted between March and April 2018 at the Amsterdam UMC, location Academic Medical Center in Amsterdam. The moderator for each FGD was one of two male authors (CH or HW), respectively OP and IP. Both are employed at the Coronel Institute of Occupational Health, have a Doctorate of Medicine and of Philosophy and have previous experience in qualitative research and conducting FGDs. The discussions were recorded with an audio recorder, and field notes were taken by another author (MdW). The authors did not know the participants before the FGDs. Apart from the researchers and participants, no-one else was present during the FGDs. Before the start of each 2-h FGD—all of which were conducted in Dutch—each participant signed an informed consent form. The FGDs started with an explanation of the purpose of the discussion, a brief introduction of the participants and an explanation of the structure of the FGD by the moderator. During the discussion that followed, the primary question addressed was: what is the most effective way for OPs and IPs to obtain information concerning person-related factors? The person-related factors defined were ten factors identified in a preceding systematic review (De Wit et al. ). The person-related factors were explained through ten case descriptions, presenting fictional situations in which the factor in question influenced the work participation of an employee with chronic health problems. During the discussion, the participants were encouraged to speak openly about their views and thoughts. When needed, the moderator asked the participants to clarify their answers. At the end of each FGD, participants received a travel allowance and a gift card of 25 euros in return for their participation. The recordings of the discussions were transcribed verbatim and anonymized. We did not send the transcripts back to the participants for comments or correction, and we did not ask for feedback on the findings. For data analysis purposes, we used qualitative content analysis (Mayring ). The transcripts from the FGDs were coded using MAXQDA 12 Software (VERBI Software ). Codes were assigned by one author (MdW) to segments of the transcript of the first two FGDs. These were then checked by a second author (HW). Disagreements about the coding were resolved by discussion. A coding framework consisting of main themes and subthemes was built by categorizing the codes. The main themes and subthemes were discussed between all authors until a consensus about the framework was reached. Following author consensus regarding the codes and coding framework, the transcript of the third FGD was coded using the coding framework by one author (MdW). The different themes of the coding framework are described in the “ ” section. To illustrate our findings, we have included quotations of participant discussions from the focus groups. A native English speaker translated these from Dutch into English. Coding framework Four primary themes of discussion were identified from the FGD transcripts. They were defined as the main categories for the coding framework: (1) methods to obtain information concerning person-related factors, (2) prerequisites for talking about person-related factors during consultations, (3) positive influences on conversations concerning person-related factors, and (4) negative influences on conversations concerning person-related factors. Methods to obtain information concerning person-related factors Participants largely acknowledged the importance of obtaining person-related information and talked about three different ways to do this. In Table , the methods identified with the corresponding quotations of participants are presented. One method for the physician to obtain information, according to the FGD participants, is to ask the employee to keep a diary and to discuss this during consultations. Employees may thereby record information such as their activities or feelings. In the opinion of some of the participants, discussing this diary with employees can help physicians to gain insight into the limitations the patient faces during the day and into the patient’s cognitions and perceptions around this. A second method described by participants was the use of a checklist or questionnaire. But participants expressed skepticism about using this method. They voiced concern that using a standardised preformat or checklist may limit the comprehensiveness of the answers an employee provides. Some participants felt that employees may not always give honest answers due to a fear that other people than the physician may read their answers. Partially due to these limitations of checklists and questionnaires, most employees preferred to discuss the factors directly during their consultations with the physician. In contrast to keeping a diary and completing questionnaires, all participants had experience with consultations; this method, therefore, provided the bulk of discussion during the FGDs. Different factors were identified that could influence the effectiveness and development of conversations pertaining to person-related factors. Prerequisites for obtaining information during consultations Before effective questions can be asked by physicians about person-related factors during consultations, FGD participants defined a set of prerequisites they felt to be of importance. Table shows these identified prerequisites, with corresponding quotations from the participants. The most important prerequisite was a mutual trust between the employee and the physician. Trust is an important factor that can facilitate the disclosure of information. All participants agreed that without this trust, a meaningful conversation about person-related factors was not possible. A second prerequisite was that the physician shows interest or demonstrates involvement with the employee. Participants agreed that it is important that employees feel they are being heard by the physician, and that, subsequently, obtaining information about person-related factors would be facilitated during the conversation when the physician shows a genuine interest in their situation and makes the employees feel like an individual. The last described prerequisite was the understanding of the physician. Participants felt that it was important that the physician understands the employee’s feelings and cognitions and acknowledges that these are not unusual. Positive influences on the development of conversations concerning person-related factors Over the FGDs, it became apparent that a number of factors can positively influence the instigation and development of a conversation about person-related factors. These factors can be broadly divided into three different subthemes: (1) communication skills of the physician, (2) context of the conversation, and (3) knowledge of the physician, and are detailed in Table along with corresponding quotations from participants. Communication skills of the physician Participants viewed it as very important that physicians listened carefully to employee responses, to prevent misinterpreting information about certain person-related factors. Furthermore, physicians should avoid closed questions and ask open questions to facilitate discussion around person-related factors during consultations. Such open questions may be focused on a variety of topics. Important themes to ask about included the work of the employee (e.g. “What adaptations have already been made?”), the employee’s private situation (e.g. “What do you do on a day?”), the future of the employee (e.g. “How do you think you will continue in the future?”), the employee’s complaints or concerns and what had been done to address them (e.g. “What are you struggling with?” and “What process have you started to recover?”) and how the physician could help the employee (e.g. “What do you need to be able to resume part of your work?”). Some participants felt that it was important to end the conversation with a question about how the employee experienced the current consultation with the physician (e.g. “How did you find this consultation?”), in order for the physician to be able to improve future conversations concerning person-related factors with employees. It is crucial that the physician makes the employee aware of what improvements are realistic and defines boundaries for the activities of the employee. The physician should focus on regaining health rather than returning to work. The consultation was felt to run more smoothly when the physician adopted the role of a coach. The physician should give tips for the employee to improve their current situation, should set small goals for the employee and should show appreciation when small goals are reached, or progress is made. Context of the conversation FGD participants emphasized the value of leaving enough time in consultations to discuss person-related factors and structuring successive consultations accordingly. Some participants felt that physicians should not address these factors immediately but should wait until sufficient rapport is established between physician and employee to allow the employee to feel comfortable to discuss them. Some employees even thought that a physician should not begin to address the factors until the second or third consultation. It is essential that the overall atmosphere of the conversation is pleasant before the physician starts to talk about the factors. Knowledge of the physician Participants agreed that a physician would obtain more information about person-related factors if they developed greater personal knowledge of the employee. Physicians need to be aware of the intellectual level of the employee, therefore, they can adapt their way of talking accordingly. Also paramount was that the physician had sufficient information about the disease or disorder of the employee and the (invisible) impairments that might exist as a result of this. The physician needs to be aware that the employee complaints and corresponding cognitions and perceptions may differ between individuals and can change over time. In addition to this, discussions around person-related factors were described to be more effective when the physician knew something of the company, the employer and the corporate culture in which the employee works. Negative influences on the development of conversations concerning person-related factors Aside from positive factors, participants also discussed issues that negatively influenced the instigation and development of a conversation. These negative influences described were diverse, but can be broadly divided into four different subthemes: (1) negative influences of the occupational health and social security systems, (2) negative influences of the physician, (3) negative influences of the employee, and (4) negative influences of the employer. Table summarises the different negative influences and provides some corresponding quotations from participants. Negative influences of occupational health and social security systems A significant negative influence on conversations described by participants was a low frequency of contact between employees and physicians. Physicians were often not accessible and getting in touch with them could prove very difficult. FGD participants sometimes did not have any direct contact with OPs, and only had contact with a designated case manager. This makes discussing person-related factors with OPs impossible. In contrast, other participants stated that discussions around person-related factors could be impeded by continually changing the physician they had contact with, and so, despite multiple consultations, they would never see the same physician twice. Another factor described as negatively influencing employee–physician conversations was that participants felt that social security organisations and employers were often focused on financial issues, rather than the wellbeing of employees. Participants stated that sometimes economic interests would seem to be more important than human interests. Other participants felt that the physician’s role was merely to limit the costs of the employer, instead of helping employees to get better. Despite this perceived overemphasis regarding money, many participants felt that physicians did not always take the reduced income of the employee into account. Feelings such as this lead to distrust towards the physician and this can disrupt and impede conversations about person-related factors. A final negative influence of the occupational health and social security systems is that employees often have little knowledge of the working practices of OPs and IPs, and about the disability assessment. Participants described that it is not always clear when they need to talk to physicians and where employees should go to get more information regarding this. This lack of adequate information can lead to uncertainty and anxiety in employees, which in turn can have negative consequences in developing conversations concerning person-related factors. Negative influences of the physician Participants also described that the physician could exert a negative influence on conversations pertaining to person-related factors. A lack of time on the part of the physician—specifically not taking the time to ask about person-related factors—will limit the possibility of obtaining person-related information. Some participants felt that physicians sometimes put too much pressure on employees to return to work, which may, in turn, have a negative influence on the development of the conversation. Negative influences of the employee Almost all FGD participants agreed that employee feelings of anxiety could negatively impact conversations concerning person-related factors. Most of this anxiety appeared to be centered around the disability assessment by the IPs, with employees reticent to disclose too much information for fear of negative consequences for the disability assessment. Other participants described anxiety around disclosing too much or too little information towards colleagues and employers concerning their health problems. Negative influences of the employer The employer can also have a negative impact on the conversation between employee and physician. Owing to the communication between the employer and physician, FGD participants felt that the confidentiality usually afforded to doctor-patient interactions was not present, leading employees to lack the feeling of trust needed to open up in conversations. These feelings of distrust can be increased when there are conflicts between the employee and employer. Four primary themes of discussion were identified from the FGD transcripts. They were defined as the main categories for the coding framework: (1) methods to obtain information concerning person-related factors, (2) prerequisites for talking about person-related factors during consultations, (3) positive influences on conversations concerning person-related factors, and (4) negative influences on conversations concerning person-related factors. Participants largely acknowledged the importance of obtaining person-related information and talked about three different ways to do this. In Table , the methods identified with the corresponding quotations of participants are presented. One method for the physician to obtain information, according to the FGD participants, is to ask the employee to keep a diary and to discuss this during consultations. Employees may thereby record information such as their activities or feelings. In the opinion of some of the participants, discussing this diary with employees can help physicians to gain insight into the limitations the patient faces during the day and into the patient’s cognitions and perceptions around this. A second method described by participants was the use of a checklist or questionnaire. But participants expressed skepticism about using this method. They voiced concern that using a standardised preformat or checklist may limit the comprehensiveness of the answers an employee provides. Some participants felt that employees may not always give honest answers due to a fear that other people than the physician may read their answers. Partially due to these limitations of checklists and questionnaires, most employees preferred to discuss the factors directly during their consultations with the physician. In contrast to keeping a diary and completing questionnaires, all participants had experience with consultations; this method, therefore, provided the bulk of discussion during the FGDs. Different factors were identified that could influence the effectiveness and development of conversations pertaining to person-related factors. Before effective questions can be asked by physicians about person-related factors during consultations, FGD participants defined a set of prerequisites they felt to be of importance. Table shows these identified prerequisites, with corresponding quotations from the participants. The most important prerequisite was a mutual trust between the employee and the physician. Trust is an important factor that can facilitate the disclosure of information. All participants agreed that without this trust, a meaningful conversation about person-related factors was not possible. A second prerequisite was that the physician shows interest or demonstrates involvement with the employee. Participants agreed that it is important that employees feel they are being heard by the physician, and that, subsequently, obtaining information about person-related factors would be facilitated during the conversation when the physician shows a genuine interest in their situation and makes the employees feel like an individual. The last described prerequisite was the understanding of the physician. Participants felt that it was important that the physician understands the employee’s feelings and cognitions and acknowledges that these are not unusual. Over the FGDs, it became apparent that a number of factors can positively influence the instigation and development of a conversation about person-related factors. These factors can be broadly divided into three different subthemes: (1) communication skills of the physician, (2) context of the conversation, and (3) knowledge of the physician, and are detailed in Table along with corresponding quotations from participants. Communication skills of the physician Participants viewed it as very important that physicians listened carefully to employee responses, to prevent misinterpreting information about certain person-related factors. Furthermore, physicians should avoid closed questions and ask open questions to facilitate discussion around person-related factors during consultations. Such open questions may be focused on a variety of topics. Important themes to ask about included the work of the employee (e.g. “What adaptations have already been made?”), the employee’s private situation (e.g. “What do you do on a day?”), the future of the employee (e.g. “How do you think you will continue in the future?”), the employee’s complaints or concerns and what had been done to address them (e.g. “What are you struggling with?” and “What process have you started to recover?”) and how the physician could help the employee (e.g. “What do you need to be able to resume part of your work?”). Some participants felt that it was important to end the conversation with a question about how the employee experienced the current consultation with the physician (e.g. “How did you find this consultation?”), in order for the physician to be able to improve future conversations concerning person-related factors with employees. It is crucial that the physician makes the employee aware of what improvements are realistic and defines boundaries for the activities of the employee. The physician should focus on regaining health rather than returning to work. The consultation was felt to run more smoothly when the physician adopted the role of a coach. The physician should give tips for the employee to improve their current situation, should set small goals for the employee and should show appreciation when small goals are reached, or progress is made. Context of the conversation FGD participants emphasized the value of leaving enough time in consultations to discuss person-related factors and structuring successive consultations accordingly. Some participants felt that physicians should not address these factors immediately but should wait until sufficient rapport is established between physician and employee to allow the employee to feel comfortable to discuss them. Some employees even thought that a physician should not begin to address the factors until the second or third consultation. It is essential that the overall atmosphere of the conversation is pleasant before the physician starts to talk about the factors. Knowledge of the physician Participants agreed that a physician would obtain more information about person-related factors if they developed greater personal knowledge of the employee. Physicians need to be aware of the intellectual level of the employee, therefore, they can adapt their way of talking accordingly. Also paramount was that the physician had sufficient information about the disease or disorder of the employee and the (invisible) impairments that might exist as a result of this. The physician needs to be aware that the employee complaints and corresponding cognitions and perceptions may differ between individuals and can change over time. In addition to this, discussions around person-related factors were described to be more effective when the physician knew something of the company, the employer and the corporate culture in which the employee works. Participants viewed it as very important that physicians listened carefully to employee responses, to prevent misinterpreting information about certain person-related factors. Furthermore, physicians should avoid closed questions and ask open questions to facilitate discussion around person-related factors during consultations. Such open questions may be focused on a variety of topics. Important themes to ask about included the work of the employee (e.g. “What adaptations have already been made?”), the employee’s private situation (e.g. “What do you do on a day?”), the future of the employee (e.g. “How do you think you will continue in the future?”), the employee’s complaints or concerns and what had been done to address them (e.g. “What are you struggling with?” and “What process have you started to recover?”) and how the physician could help the employee (e.g. “What do you need to be able to resume part of your work?”). Some participants felt that it was important to end the conversation with a question about how the employee experienced the current consultation with the physician (e.g. “How did you find this consultation?”), in order for the physician to be able to improve future conversations concerning person-related factors with employees. It is crucial that the physician makes the employee aware of what improvements are realistic and defines boundaries for the activities of the employee. The physician should focus on regaining health rather than returning to work. The consultation was felt to run more smoothly when the physician adopted the role of a coach. The physician should give tips for the employee to improve their current situation, should set small goals for the employee and should show appreciation when small goals are reached, or progress is made. FGD participants emphasized the value of leaving enough time in consultations to discuss person-related factors and structuring successive consultations accordingly. Some participants felt that physicians should not address these factors immediately but should wait until sufficient rapport is established between physician and employee to allow the employee to feel comfortable to discuss them. Some employees even thought that a physician should not begin to address the factors until the second or third consultation. It is essential that the overall atmosphere of the conversation is pleasant before the physician starts to talk about the factors. Participants agreed that a physician would obtain more information about person-related factors if they developed greater personal knowledge of the employee. Physicians need to be aware of the intellectual level of the employee, therefore, they can adapt their way of talking accordingly. Also paramount was that the physician had sufficient information about the disease or disorder of the employee and the (invisible) impairments that might exist as a result of this. The physician needs to be aware that the employee complaints and corresponding cognitions and perceptions may differ between individuals and can change over time. In addition to this, discussions around person-related factors were described to be more effective when the physician knew something of the company, the employer and the corporate culture in which the employee works. Aside from positive factors, participants also discussed issues that negatively influenced the instigation and development of a conversation. These negative influences described were diverse, but can be broadly divided into four different subthemes: (1) negative influences of the occupational health and social security systems, (2) negative influences of the physician, (3) negative influences of the employee, and (4) negative influences of the employer. Table summarises the different negative influences and provides some corresponding quotations from participants. Negative influences of occupational health and social security systems A significant negative influence on conversations described by participants was a low frequency of contact between employees and physicians. Physicians were often not accessible and getting in touch with them could prove very difficult. FGD participants sometimes did not have any direct contact with OPs, and only had contact with a designated case manager. This makes discussing person-related factors with OPs impossible. In contrast, other participants stated that discussions around person-related factors could be impeded by continually changing the physician they had contact with, and so, despite multiple consultations, they would never see the same physician twice. Another factor described as negatively influencing employee–physician conversations was that participants felt that social security organisations and employers were often focused on financial issues, rather than the wellbeing of employees. Participants stated that sometimes economic interests would seem to be more important than human interests. Other participants felt that the physician’s role was merely to limit the costs of the employer, instead of helping employees to get better. Despite this perceived overemphasis regarding money, many participants felt that physicians did not always take the reduced income of the employee into account. Feelings such as this lead to distrust towards the physician and this can disrupt and impede conversations about person-related factors. A final negative influence of the occupational health and social security systems is that employees often have little knowledge of the working practices of OPs and IPs, and about the disability assessment. Participants described that it is not always clear when they need to talk to physicians and where employees should go to get more information regarding this. This lack of adequate information can lead to uncertainty and anxiety in employees, which in turn can have negative consequences in developing conversations concerning person-related factors. Negative influences of the physician Participants also described that the physician could exert a negative influence on conversations pertaining to person-related factors. A lack of time on the part of the physician—specifically not taking the time to ask about person-related factors—will limit the possibility of obtaining person-related information. Some participants felt that physicians sometimes put too much pressure on employees to return to work, which may, in turn, have a negative influence on the development of the conversation. Negative influences of the employee Almost all FGD participants agreed that employee feelings of anxiety could negatively impact conversations concerning person-related factors. Most of this anxiety appeared to be centered around the disability assessment by the IPs, with employees reticent to disclose too much information for fear of negative consequences for the disability assessment. Other participants described anxiety around disclosing too much or too little information towards colleagues and employers concerning their health problems. Negative influences of the employer The employer can also have a negative impact on the conversation between employee and physician. Owing to the communication between the employer and physician, FGD participants felt that the confidentiality usually afforded to doctor-patient interactions was not present, leading employees to lack the feeling of trust needed to open up in conversations. These feelings of distrust can be increased when there are conflicts between the employee and employer. A significant negative influence on conversations described by participants was a low frequency of contact between employees and physicians. Physicians were often not accessible and getting in touch with them could prove very difficult. FGD participants sometimes did not have any direct contact with OPs, and only had contact with a designated case manager. This makes discussing person-related factors with OPs impossible. In contrast, other participants stated that discussions around person-related factors could be impeded by continually changing the physician they had contact with, and so, despite multiple consultations, they would never see the same physician twice. Another factor described as negatively influencing employee–physician conversations was that participants felt that social security organisations and employers were often focused on financial issues, rather than the wellbeing of employees. Participants stated that sometimes economic interests would seem to be more important than human interests. Other participants felt that the physician’s role was merely to limit the costs of the employer, instead of helping employees to get better. Despite this perceived overemphasis regarding money, many participants felt that physicians did not always take the reduced income of the employee into account. Feelings such as this lead to distrust towards the physician and this can disrupt and impede conversations about person-related factors. A final negative influence of the occupational health and social security systems is that employees often have little knowledge of the working practices of OPs and IPs, and about the disability assessment. Participants described that it is not always clear when they need to talk to physicians and where employees should go to get more information regarding this. This lack of adequate information can lead to uncertainty and anxiety in employees, which in turn can have negative consequences in developing conversations concerning person-related factors. Participants also described that the physician could exert a negative influence on conversations pertaining to person-related factors. A lack of time on the part of the physician—specifically not taking the time to ask about person-related factors—will limit the possibility of obtaining person-related information. Some participants felt that physicians sometimes put too much pressure on employees to return to work, which may, in turn, have a negative influence on the development of the conversation. Almost all FGD participants agreed that employee feelings of anxiety could negatively impact conversations concerning person-related factors. Most of this anxiety appeared to be centered around the disability assessment by the IPs, with employees reticent to disclose too much information for fear of negative consequences for the disability assessment. Other participants described anxiety around disclosing too much or too little information towards colleagues and employers concerning their health problems. The employer can also have a negative impact on the conversation between employee and physician. Owing to the communication between the employer and physician, FGD participants felt that the confidentiality usually afforded to doctor-patient interactions was not present, leading employees to lack the feeling of trust needed to open up in conversations. These feelings of distrust can be increased when there are conflicts between the employee and employer. Key findings Employees with work limitations due to chronic health problems acknowledge the importance of person-related factors in their management and are most comfortable sharing these factors with OPs and IPs directly in consultations. Trust, understanding and interest were considered essential to allow effective discussion or conversations concerning person-related factors. Aside from these prerequisites, issues pertaining to the communication skills of the physician, the knowledge of the physician, and the context of the consultation were identified being able to impact the development of the conversation positively. Employees identified issues related to occupational health and social security systems, the physician, the employer and the employee which can negatively influence the instigation and development of such conversations. Trust between employee and physician was perceived as the most important prerequisite for obtaining person-related information during consultations. This is in accordance with previous studies that describe the importance of trust for patients in disclosing information during conversations about medical issues (Julliard et al. ; Main et al. ; Kelak et al. ). An interview study by Julliard et al. identifies trust, compassion and respect, as prerequisites for patients sharing health information with their physician. Studies by Main et al. and Kelak et al. also emphasize the importance of trust for disclosing information during consultations. According to Ridd et al. and Skirbekk et al. , trust arises when patients and physicians spend more time with each other in consultations. This is consistent with our findings that employees valued physicians taking time to develop a mutual trust before addressing person-related factors. This association between spent time in consultations and trust could also help to explain why a lack of contact with the physician and limited accessibility were perceived as negative influences on the development of conversations concerning person-related information. In addition, employees described the negative influence of seeing different physicians each time. All of these factors limit the time that employees spend with the same physician, potentially disrupting the process of building trust (Ridd et al. ; Skirbekk et al. ). Appropriate timing of conversations about person-related factors—as well as taking enough time to discuss them—are essential for obtaining reliable person-related information during consultations. Other prerequisites for obtaining information about person-related factors involved the physician showing interest, being involved and understanding. This is consistent with results of a review by Ridd et al. showing that patients value doctors who appear interested during consultations, and results of a study by Kelak et al. in which involvement of the physician was identified as a critical component for patients to disclose information. The results are also supported by a study by Mazzi et al. , in which taking the patient seriously and treating the patient as a person were identified as two of the five most important recommendations from patients for physicians to make consultations more effective. Participants of the FGDs identified, in addition, a number of different factors that may influence the development of the conversation about person-related factors. Several factors, such as listening, asking open questions, and having knowledge about the patient’s complaints have also been identified in other studies as important factors for the development of medical consultations (Main et al. ; Ha and Longnecker ; Julliard et al. ; Ranjan et al. ; Mazzi et al. ; Kelak et al. ). Other studies also identified factors which were important for the development of the consultation, that were not mentioned by our participants, such as the importance of non-verbal signals from physicians, like keeping eye contact with the patient (Main et al. ; Bensing et al. ; Deledda et al. ). Strengths and limitations A strength of our study is that the focus groups consisted of participants with different types of disabilities, making the findings generalizable to employees with various health problems. Another strength is that the experiences of the patients with physicians diverged from positive to very negative, providing information about both facilitators and barriers to obtaining information about person-related factors. A limitation of this study is that participants had difficulty answering some of the questions asked during the FGDs. Instead of talking about how to obtain information about cognitions and perceptions, participants had the tendency to talk about different ways to change the cognitions and perceptions of the employee. Although this information can be useful in future research, it was not included in this study because it did not help us in answering our research question. Implications for practice and future research We recommend that physicians consider person-related factors during their consultations to increase work participation in employees with health problems. Physicians should be especially aware that trust, understanding and showing interest are essential in order for an employee to feel comfortable to disclose person-related information during these conversations. Physicians need to be accessible for employees and need to be aware that time frames are crucial when talking about person-related factors. During the conversation, we recommend that physicians listen to the employee and ask open questions regarding different subjects, such as the employee’s work, thoughts about the future, complaints, and about possible ways to help the employee. This increases the knowledge of the physician about the employee and the employee’s situation and can prove to be beneficial in the development of conversations addressing person-related factors. This study indicated that—from employees perspective—the most crucial prerequisite for discussing person-related factors during consultations is trust. Therefore, it is important that future research examines how mutual trust between physician and employee can arise, be maintained, or be increased. However, numerous factors were identified which can negatively influence the conversation about person-related factors, making discussing these factors a complex process. This might be one of the reasons why some physicians, according to the participants, do not always ask about all these person-related factors. Future research might be needed to examine the reasons why physicians do not always discuss all person-related factors, or to study the factors that make discussing these factors difficult from the perspective of physicians. Despite the complexity of conversations concerning person-related factors, as far as we know, there is no tool or training available to help OPs and IPs structure these conversations. We recommend that researchers use the information from this study to develop such a tool or training program. Additionally, considering all person-related factors during consultations is time-consuming for the physician. Therefore, it is also of importance that future researchers determine whether considering person-related factors during consultations really improves the practices of OPs and IPs to increase work participation of employees with health problems. Employees with work limitations due to chronic health problems acknowledge the importance of person-related factors in their management and are most comfortable sharing these factors with OPs and IPs directly in consultations. Trust, understanding and interest were considered essential to allow effective discussion or conversations concerning person-related factors. Aside from these prerequisites, issues pertaining to the communication skills of the physician, the knowledge of the physician, and the context of the consultation were identified being able to impact the development of the conversation positively. Employees identified issues related to occupational health and social security systems, the physician, the employer and the employee which can negatively influence the instigation and development of such conversations. Trust between employee and physician was perceived as the most important prerequisite for obtaining person-related information during consultations. This is in accordance with previous studies that describe the importance of trust for patients in disclosing information during conversations about medical issues (Julliard et al. ; Main et al. ; Kelak et al. ). An interview study by Julliard et al. identifies trust, compassion and respect, as prerequisites for patients sharing health information with their physician. Studies by Main et al. and Kelak et al. also emphasize the importance of trust for disclosing information during consultations. According to Ridd et al. and Skirbekk et al. , trust arises when patients and physicians spend more time with each other in consultations. This is consistent with our findings that employees valued physicians taking time to develop a mutual trust before addressing person-related factors. This association between spent time in consultations and trust could also help to explain why a lack of contact with the physician and limited accessibility were perceived as negative influences on the development of conversations concerning person-related information. In addition, employees described the negative influence of seeing different physicians each time. All of these factors limit the time that employees spend with the same physician, potentially disrupting the process of building trust (Ridd et al. ; Skirbekk et al. ). Appropriate timing of conversations about person-related factors—as well as taking enough time to discuss them—are essential for obtaining reliable person-related information during consultations. Other prerequisites for obtaining information about person-related factors involved the physician showing interest, being involved and understanding. This is consistent with results of a review by Ridd et al. showing that patients value doctors who appear interested during consultations, and results of a study by Kelak et al. in which involvement of the physician was identified as a critical component for patients to disclose information. The results are also supported by a study by Mazzi et al. , in which taking the patient seriously and treating the patient as a person were identified as two of the five most important recommendations from patients for physicians to make consultations more effective. Participants of the FGDs identified, in addition, a number of different factors that may influence the development of the conversation about person-related factors. Several factors, such as listening, asking open questions, and having knowledge about the patient’s complaints have also been identified in other studies as important factors for the development of medical consultations (Main et al. ; Ha and Longnecker ; Julliard et al. ; Ranjan et al. ; Mazzi et al. ; Kelak et al. ). Other studies also identified factors which were important for the development of the consultation, that were not mentioned by our participants, such as the importance of non-verbal signals from physicians, like keeping eye contact with the patient (Main et al. ; Bensing et al. ; Deledda et al. ). A strength of our study is that the focus groups consisted of participants with different types of disabilities, making the findings generalizable to employees with various health problems. Another strength is that the experiences of the patients with physicians diverged from positive to very negative, providing information about both facilitators and barriers to obtaining information about person-related factors. A limitation of this study is that participants had difficulty answering some of the questions asked during the FGDs. Instead of talking about how to obtain information about cognitions and perceptions, participants had the tendency to talk about different ways to change the cognitions and perceptions of the employee. Although this information can be useful in future research, it was not included in this study because it did not help us in answering our research question. We recommend that physicians consider person-related factors during their consultations to increase work participation in employees with health problems. Physicians should be especially aware that trust, understanding and showing interest are essential in order for an employee to feel comfortable to disclose person-related information during these conversations. Physicians need to be accessible for employees and need to be aware that time frames are crucial when talking about person-related factors. During the conversation, we recommend that physicians listen to the employee and ask open questions regarding different subjects, such as the employee’s work, thoughts about the future, complaints, and about possible ways to help the employee. This increases the knowledge of the physician about the employee and the employee’s situation and can prove to be beneficial in the development of conversations addressing person-related factors. This study indicated that—from employees perspective—the most crucial prerequisite for discussing person-related factors during consultations is trust. Therefore, it is important that future research examines how mutual trust between physician and employee can arise, be maintained, or be increased. However, numerous factors were identified which can negatively influence the conversation about person-related factors, making discussing these factors a complex process. This might be one of the reasons why some physicians, according to the participants, do not always ask about all these person-related factors. Future research might be needed to examine the reasons why physicians do not always discuss all person-related factors, or to study the factors that make discussing these factors difficult from the perspective of physicians. Despite the complexity of conversations concerning person-related factors, as far as we know, there is no tool or training available to help OPs and IPs structure these conversations. We recommend that researchers use the information from this study to develop such a tool or training program. Additionally, considering all person-related factors during consultations is time-consuming for the physician. Therefore, it is also of importance that future researchers determine whether considering person-related factors during consultations really improves the practices of OPs and IPs to increase work participation of employees with health problems. |
Implementation of cardiovascular disease prevention in primary health care: enhancing understanding using normalisation process theory | 2813b484-18b7-4db4-904b-bdb1be2c2eaf | 5324228 | Preventive Medicine[mh] | Cardiovascular disease (CVD) was the leading cause of death in Australia in 2011 . Primary health care has an important role in supporting CVD prevention, however prevention-orientated activities are not routinely undertaken in Australian general practice . While the need for more heath promoting health systems has long been recognised, attempts to reorientate health services towards prevention have proven highly resistant to change . The need for greater expertise in how to implement CVD prevention strategies in practice has been identified as key to addressing the CVD burden worldwide . The Model for Prevention study (MoFoP) is a case study exploration of a whole-of-system CVD prevention intervention framed by the Expanded Chronic Care Model (ECCM) . The ECCM provides an evidence-based approach to health system redesign for prevention and management of chronic disease integrating the Chronic Care Model and the five action areas from the Ottawa Charter for Health Promotion . The intervention was of 12 months duration, with strategies including improvement of clinical and community information systems, support for health practitioner decision making for CVD risk management, provision of a health coaching service to support patients to develop lifestyle modification skills and health system redesign to provide greater health behaviour change support across the general practice setting. Patients from six general practices in the Australian Capital Territory (ACT) identified via existing clinical data as being at high risk for CVD disease were provided with enhanced risk management support, including access to a Lifestyle Advisor service (health coach) for up to 12 months (average of four sessions). The intervention also included strategies to build the capacity of community-based lifestyle modification services to support patients in the community setting. Interventions to address chronic disease are complex and, while guidelines-based care has been shown to be effective in very controlled situations, translating these outcomes into real world practice has proven difficult to sustain. In their review of Chronic Care Model-framed interventions which aim to improve chronic disease outcomes, Kadu and Stolee found the need for more research focused on understanding the inner settings of organisations, including the characteristics of the work of individual practitioners, in order to better understand how to achieve and sustain positive outcomes . One of the main aims of the MoFoP study was to focus on the feasibility of embedding the intervention approach into real world practice, both in the general practice and community setting. The community setting aspects of the study have been reported elsewhere . This paper focuses on the aspects of the intervention that occurred within general practice involving general practitioners (GP), practice nurses (PN), practice management (PM), lifestyle advisors (LA) and patients who participated in the intervention (P). To make sense of the social and organisational aspects of the intervention Normalisation Process Theory (NPT) was chosen as a tool to frame the analysis. NPT is a mid-level theory developed to understand and evaluate the processes by which complex interventions are embedded into routine practice . The theory takes a whole-of-system perspective, which aligned well with the intervention design. NPT has also been used as a tool for assessing the suitability of trial approaches or providing information to optimise trials . Data collection The study used a qualitative design employing semi-structure interviews. The interviews were conducted with staff of all the six practices, who delivered the CVD prevention intervention. All staff members were sent individual invitations by the research team to participate in the interviews. Additionally, all patients who participated in the intervention ( n = 30) and both Lifestyle Advisors were sent invitations to be interviewed. Semi-structured interviews were used instead of focus groups due to the nature of the general practice environment, which makes it difficult for groups to get together at one time. Interviews also provide a degree of anonymity for staff members who were often employees or junior staff of other interviewees. Topic guides for practice staff and LAs were informed by NPT constructs. These constructs help to explain the work involved in embedding interventions into routine practice. This includes making meaning and sense of the intervention (coherence), committing to and engaging with the intervention (cognitive participation), delivering the intervention (collective action) and reflecting and appraising the intervention approach (reflective monitoring). Topic guides for the interviews with patients were developed in very early stages of the research and contained evaluation questions not directly informed by NPT. The decision was made to include patient data in this analysis given the identified need to include service user evaluation in the NPT literature . All topic guides were pilot tested before data collection commenced. Interview questions for all stakeholders are provided in Additional file . Purposive sampling was used to ensure representation from a range of practice staff across all of the intervention general practices. Practice staff were invited to participate via a letter from the researcher, which was distributed by practice management. Patients and LAs were also invited by letter to participate after the completion of the intervention. Interviews with general practice staff were conducted mostly on site at practices and lasted between 20 and 60 min. Interviews were conducted individually rather than in a group due to sensitivity to employee/employer relationships which may have impacted on the ability of some staff to answer questions honestly. Patients were interviewed face to face at their general practice or by phone (for their convenience) and lasted between 30 and 60 min. LAs were interviewed face to face at the office of the interviewer and these interviews lasted around 60 min. Interviews were conducted between December 2013 and July 2014. All interviews were conducted by the first author, audio taped and transcribed. Ethical approval for the study was obtained from University of Canberra Human Research Ethics Committee (Project number 11–141). A description of each of the stakeholder groups is outlined in Table . Data analysis The study used NPT constructs to frame the data analysis. May and colleagues proposed that NPT could structure the way that qualitative data is coded, analysed and understood . In this instance, the four constructs (and components) were used to code the data in line with Strauss and Corbin’s single coding approach . Drawing on the work of Murray, Blakeman, and Gallacher, a NPT informed coding framework was developed and interview transcripts were coded against the framework . The framework is described at Table . A subset (20%) of transcripts were coded independently by a second researcher and then compared and discussed to ensure consistent coding against NPT constructs. The data generally aligned with the constructs and, where data did not fit, it was coded as “other”. Data handling was facilitated using NVivo 10 software. After coding, narratives were developed under each of the constructs. The study used a qualitative design employing semi-structure interviews. The interviews were conducted with staff of all the six practices, who delivered the CVD prevention intervention. All staff members were sent individual invitations by the research team to participate in the interviews. Additionally, all patients who participated in the intervention ( n = 30) and both Lifestyle Advisors were sent invitations to be interviewed. Semi-structured interviews were used instead of focus groups due to the nature of the general practice environment, which makes it difficult for groups to get together at one time. Interviews also provide a degree of anonymity for staff members who were often employees or junior staff of other interviewees. Topic guides for practice staff and LAs were informed by NPT constructs. These constructs help to explain the work involved in embedding interventions into routine practice. This includes making meaning and sense of the intervention (coherence), committing to and engaging with the intervention (cognitive participation), delivering the intervention (collective action) and reflecting and appraising the intervention approach (reflective monitoring). Topic guides for the interviews with patients were developed in very early stages of the research and contained evaluation questions not directly informed by NPT. The decision was made to include patient data in this analysis given the identified need to include service user evaluation in the NPT literature . All topic guides were pilot tested before data collection commenced. Interview questions for all stakeholders are provided in Additional file . Purposive sampling was used to ensure representation from a range of practice staff across all of the intervention general practices. Practice staff were invited to participate via a letter from the researcher, which was distributed by practice management. Patients and LAs were also invited by letter to participate after the completion of the intervention. Interviews with general practice staff were conducted mostly on site at practices and lasted between 20 and 60 min. Interviews were conducted individually rather than in a group due to sensitivity to employee/employer relationships which may have impacted on the ability of some staff to answer questions honestly. Patients were interviewed face to face at their general practice or by phone (for their convenience) and lasted between 30 and 60 min. LAs were interviewed face to face at the office of the interviewer and these interviews lasted around 60 min. Interviews were conducted between December 2013 and July 2014. All interviews were conducted by the first author, audio taped and transcribed. Ethical approval for the study was obtained from University of Canberra Human Research Ethics Committee (Project number 11–141). A description of each of the stakeholder groups is outlined in Table . The study used NPT constructs to frame the data analysis. May and colleagues proposed that NPT could structure the way that qualitative data is coded, analysed and understood . In this instance, the four constructs (and components) were used to code the data in line with Strauss and Corbin’s single coding approach . Drawing on the work of Murray, Blakeman, and Gallacher, a NPT informed coding framework was developed and interview transcripts were coded against the framework . The framework is described at Table . A subset (20%) of transcripts were coded independently by a second researcher and then compared and discussed to ensure consistent coding against NPT constructs. The data generally aligned with the constructs and, where data did not fit, it was coded as “other”. Data handling was facilitated using NVivo 10 software. After coding, narratives were developed under each of the constructs. There were 40 face to face interviews conducted with participants in all six general practices involved in the MoFoP study. Interviews were held with 11 General Practitioners (GPs) (26% of GP participants), 12 Practice Nurses (PNs) (75% of PN participants), six Practice Managers (PMs) (100% of PM participants), two LAs (100% of LA participants) and nine patients (30% of patient participants). Characteristics of interview participants are detailed in Table . While male and female practice staff members were interviewed, gender is not reported for each category to maintain confidentiality of individuals. All patients interviewed were male, consistent with the total MoFoP intervention population, where all but one participating patient was male. The stakeholder interviews provided a rich description of the processes of implementation of the MoFoP intervention. After the data were analysed using the NPT coding framework, narratives were developed for each of the four NPT constructs. These narratives were summarised and are presented below with illustrative quotes. While not all elements within each construct had the same density of coded information, overall most stakeholders could comment across most of the NPT constructs. The quotes are attributed to participants by professional grouping, years working in that role, and age for patients. Making sense of the intervention For the general practices and their patients, the intervention was the first time they had taken a systematic approach to identification and management of CVD risk. The prevention-orientated service was considered by practice staff to be important for the health of their patients and an important function of general practice, as one GP stated: I feel this is what general practice is about. We are here to prevent health problems and manage patients. (GP11, 25 years as a GP). While practice staff saw the intervention approach to be consistent with the goals of general practice, most agreed it was different from their everyday experience, which focused on responding to acute illness. As one PN put it, “ Normally it is about treating; this was about prevention ” (PN7, 1 year as a PN). The patients who participated in the intervention also agreed that preventing CVD was important and they expected their general practice to have a role in supporting them to reduce their risk. Most reported that they had existing risk factors for CVD and felt the intervention was therefore relevant to them. However, not all participants were convinced that they were at high risk and for one patient the recall letter was totally unexpected. He said, “ I think the word I would use is bemused. I didn’t know, or didn’t really suspect, that I might have been at high risk ” (P4, 66 years old). Stakeholder investments in the intervention While the six general practices volunteered to be involved in the intervention, the degree of commitment and engagement varied across the practices, between practice staff groups and for patients in each of the different practices. One PN noted that in their practice “ there was great support from management, but I think the clinicians (GPs) had varying levels of commitment to the project. ” (PN 2, 6 years as a PN). In another practice a strong interest in CVD prevention by one particular staff member led to a high level of engagement with the intervention. A GP from this practice noted, “ Having a nurse with a background in cardiovascular disease meant that we weren’t going to let anything cardiovascular pass us by ” (GP 1, 20 years as a GP). While many practice staff reported that they participated in the intervention in the interests of their patients, others undertook activities because they were expected to as part of their employment. As one GP said “ I did what I was told ” (GP3, 25 years as a GP). Both LAs also felt that there were varying levels of support for the intervention across practices with one LA stating that , “ I would probably say one of the four (GPs) was committed ” (LA2, 27 years in lifestyle modification). Communication with practice staff regarding the delivery of the intervention was generally good. However, staff turnover in all practices meant that as the intervention progressed some practitioners were not informed of their required activities. One GP noted that “ I think the doctors who came along after the intervention had started did not know what was involved…they did not go to the talks ” (GP1, 20 years as a GP). The patients interviewed had all made the effort to participate in the intervention by reading the provided information on CVD risk, making an appointment to see their GP, and in some cases, having a blood test to update lipid levels. Most patients considered the intervention to be a normal extension of care from their general practice. They had established relationships with the practice and its staff and assumed that participation would be beneficial. As one patient said “ It seemed ordinary, in a sense that this practice does look after me ” (P5, 69 years old). Stakeholders work to enact the intervention While all practices reported following the intervention protocol, the size of the practice, skills and preferences of the workforce, and the business model led to the work of the intervention being configured in a range of ways. These ranged from a nurse led approach to the intervention, with limited input by the GP, to non-participation where individual GPs in a practice would not allow their patients to be recalled because they did not receive a direct incentive payment. One GP explained: It depends on how the practice is set up and what your nursing staff are like. I suppose also how you like to do things as a GP as well. If you like to maintain that relationship and do the education yourself rather than pass the buck to whoever it may be. (GP9, 2 years as a GP) Successful implementation of the intervention relied on administration staff undertaking a wide range of tasks, including many considered central to achieving a systems approach. While clinicians were responsible for recording most of the clinical and demographic data, administration staff took the lead in most practices in educating and reminding staff about the need to enter the key demographic and CVD risk factor data and how to enter this data correctly. They used their usual practice communication strategies such as staff meetings and some novel approaches, for example, messages on the back of staff toilet doors. One PM recounted her approach, “We were sticking little stickers [Post It notes] all over the GP’s computers saying, “please do this or please do that”. (PM4, 8 years as a PM). Other tasks that were central to successful implementation were related to navigating patients through the intervention processes, such as responding appropriately to phone calls about the recall visits. When administration staff failed to perform these tasks the practice experienced difficulties communicating with eligible patients. One patient recalled his experience of contacting his practice and the receptionist not being fully aware of her role in the intervention. I had to ring up the practice and tell them that I wanted an appointment for this program. So I did that and they made me feel that I was stupid…why did I need this appointment she said. (P8, 70 years old). All GPs reported that the intervention did not add a significant workload and they were happy to refer patients to the LA service. They frequently commented that it was beneficial to give patients access to a broader range of team members who could spend more time with them on lifestyle modification issues. As one GP said: I can only do so much for this patient because I have 15 minutes … so that team-based model… I think the program got that team approach. (GP1, 20 years as a GP). For the PNs the intervention activities led to a greater consultation role, which most of them enjoyed. This work differed from their usual task-orientation and provided opportunities to build deeper relationships with patients. One practice nurse said, “ There is not a lot of relationship building (usually). When talking about prevention, we need to build relationships, and be in it for the long haul ” (PN2, 6 years as a PN). The LAs were highly motivated to develop good working relationships with the GPs from the beginning of the intervention. While they reported to be disappointed that the contact was generally restricted to written communication, they did feel that being present at the practice made their service more credible with the clinical staff. One LA said “ they had more confidence in what we do because we were actually in the medical centre. ” (LA1, 27 years working in lifestyle modification). While the intervention introduced an LA as a new health care provider to the usual practice based care team, the patients did not feel that their relationship with their GP or other practice staff was diminished. As one participant stated, “ I don’t think he thought I was being poached away. ” (P 3, 66 years). Most found that the health coaching approach used by the LAs was non-judgemental, allowing them to be honest about areas they were willing to change and supported in achieving the goals they set for themselves. One patient outlined how the process worked for them: Basically what happened is I told her what I wanted to do, she listened and I took over. She steered me in another direction sometimes or pointed something out I might need. (P7, 65 years old). Stakeholder appraisal of the intervention All those interviewed considered the intervention to be worthwhile at some level. Most patients made positive changes to their lifestyle as a result of the intervention. One patient had good outcomes for issues that had previously been resistant to change. He said: I have certainly reduced smoking and alcohol. I have been able to put alcohol free days into my program which I found difficult before (P1, 69 years old). Practice staff and LAs recalled positive health and wellbeing outcomes for patients and the practice and systems level changes achieved. These changes included both practice and opportunistic risk factor data captured for enhanced CVD risk assessment and management. As explained by one of the PNs interviewed: Since the program was introduced I think it improved our collection of data, when patients come now all patients who come through the treatment room get height, weight, blood pressure, allergies and smoking assessment (PN12, 5 years as a PN). The key role of administrative staff improving the systems response to CVD prevention was also recognised during the study. One experienced PM noted that “ Since my involvement with this project I train my staff that are sending out the recall letter… so when a patient calls and makes an appointment they know exactly what to tell them ” (PM6, 18 years as a PM). Some staff believed the intervention resulted in a greater awareness by GPs of the importance of managing CVD risk and an increased number of CVD risk assessments being conducted. One PN reported that in her practice, CVD risk assessment had increased, stating that “ Doctors are doing a lot more cardiovascular assessments…they are actually doing them! ” (PN2, 6 years as a PN). There were staff in every practice that identified barriers to them reorganising practice to align with the intervention approach. While practice nurses were interested in an expanded role, many felt underprepared to address lifestyle modification and felt they needed more training in the area. This was highlighted by one nurse who said, “ I haven’t been trained for any of this, it has all been a learning curve for me ” (PN6, 7 years as a PN). Current financing models, time pressures and practical issues such as poorly integrated clinical software were also considered issues to be addressed if they were to become a more prevention-orientated practice. One GP was very clear about this, stating that: Anything that puts more work on the staff or the doctors is unrealistic… If there’s no incentive for us, at least make it that simple that I don’t have to invest extra work into it. (GP10, 20 years as a GP). When reflecting back on the intervention, almost all practice staff identified the difficulty of achieving sustained lifestyle modification itself as a major impediment to providing prevention-orientated services. They also considered these challenges would be exacerbated if there were out of pocket costs for patients. It’s a hard sell and that is why it should be free. (PN7, 1 year as a PN). For the general practices and their patients, the intervention was the first time they had taken a systematic approach to identification and management of CVD risk. The prevention-orientated service was considered by practice staff to be important for the health of their patients and an important function of general practice, as one GP stated: I feel this is what general practice is about. We are here to prevent health problems and manage patients. (GP11, 25 years as a GP). While practice staff saw the intervention approach to be consistent with the goals of general practice, most agreed it was different from their everyday experience, which focused on responding to acute illness. As one PN put it, “ Normally it is about treating; this was about prevention ” (PN7, 1 year as a PN). The patients who participated in the intervention also agreed that preventing CVD was important and they expected their general practice to have a role in supporting them to reduce their risk. Most reported that they had existing risk factors for CVD and felt the intervention was therefore relevant to them. However, not all participants were convinced that they were at high risk and for one patient the recall letter was totally unexpected. He said, “ I think the word I would use is bemused. I didn’t know, or didn’t really suspect, that I might have been at high risk ” (P4, 66 years old). While the six general practices volunteered to be involved in the intervention, the degree of commitment and engagement varied across the practices, between practice staff groups and for patients in each of the different practices. One PN noted that in their practice “ there was great support from management, but I think the clinicians (GPs) had varying levels of commitment to the project. ” (PN 2, 6 years as a PN). In another practice a strong interest in CVD prevention by one particular staff member led to a high level of engagement with the intervention. A GP from this practice noted, “ Having a nurse with a background in cardiovascular disease meant that we weren’t going to let anything cardiovascular pass us by ” (GP 1, 20 years as a GP). While many practice staff reported that they participated in the intervention in the interests of their patients, others undertook activities because they were expected to as part of their employment. As one GP said “ I did what I was told ” (GP3, 25 years as a GP). Both LAs also felt that there were varying levels of support for the intervention across practices with one LA stating that , “ I would probably say one of the four (GPs) was committed ” (LA2, 27 years in lifestyle modification). Communication with practice staff regarding the delivery of the intervention was generally good. However, staff turnover in all practices meant that as the intervention progressed some practitioners were not informed of their required activities. One GP noted that “ I think the doctors who came along after the intervention had started did not know what was involved…they did not go to the talks ” (GP1, 20 years as a GP). The patients interviewed had all made the effort to participate in the intervention by reading the provided information on CVD risk, making an appointment to see their GP, and in some cases, having a blood test to update lipid levels. Most patients considered the intervention to be a normal extension of care from their general practice. They had established relationships with the practice and its staff and assumed that participation would be beneficial. As one patient said “ It seemed ordinary, in a sense that this practice does look after me ” (P5, 69 years old). While all practices reported following the intervention protocol, the size of the practice, skills and preferences of the workforce, and the business model led to the work of the intervention being configured in a range of ways. These ranged from a nurse led approach to the intervention, with limited input by the GP, to non-participation where individual GPs in a practice would not allow their patients to be recalled because they did not receive a direct incentive payment. One GP explained: It depends on how the practice is set up and what your nursing staff are like. I suppose also how you like to do things as a GP as well. If you like to maintain that relationship and do the education yourself rather than pass the buck to whoever it may be. (GP9, 2 years as a GP) Successful implementation of the intervention relied on administration staff undertaking a wide range of tasks, including many considered central to achieving a systems approach. While clinicians were responsible for recording most of the clinical and demographic data, administration staff took the lead in most practices in educating and reminding staff about the need to enter the key demographic and CVD risk factor data and how to enter this data correctly. They used their usual practice communication strategies such as staff meetings and some novel approaches, for example, messages on the back of staff toilet doors. One PM recounted her approach, “We were sticking little stickers [Post It notes] all over the GP’s computers saying, “please do this or please do that”. (PM4, 8 years as a PM). Other tasks that were central to successful implementation were related to navigating patients through the intervention processes, such as responding appropriately to phone calls about the recall visits. When administration staff failed to perform these tasks the practice experienced difficulties communicating with eligible patients. One patient recalled his experience of contacting his practice and the receptionist not being fully aware of her role in the intervention. I had to ring up the practice and tell them that I wanted an appointment for this program. So I did that and they made me feel that I was stupid…why did I need this appointment she said. (P8, 70 years old). All GPs reported that the intervention did not add a significant workload and they were happy to refer patients to the LA service. They frequently commented that it was beneficial to give patients access to a broader range of team members who could spend more time with them on lifestyle modification issues. As one GP said: I can only do so much for this patient because I have 15 minutes … so that team-based model… I think the program got that team approach. (GP1, 20 years as a GP). For the PNs the intervention activities led to a greater consultation role, which most of them enjoyed. This work differed from their usual task-orientation and provided opportunities to build deeper relationships with patients. One practice nurse said, “ There is not a lot of relationship building (usually). When talking about prevention, we need to build relationships, and be in it for the long haul ” (PN2, 6 years as a PN). The LAs were highly motivated to develop good working relationships with the GPs from the beginning of the intervention. While they reported to be disappointed that the contact was generally restricted to written communication, they did feel that being present at the practice made their service more credible with the clinical staff. One LA said “ they had more confidence in what we do because we were actually in the medical centre. ” (LA1, 27 years working in lifestyle modification). While the intervention introduced an LA as a new health care provider to the usual practice based care team, the patients did not feel that their relationship with their GP or other practice staff was diminished. As one participant stated, “ I don’t think he thought I was being poached away. ” (P 3, 66 years). Most found that the health coaching approach used by the LAs was non-judgemental, allowing them to be honest about areas they were willing to change and supported in achieving the goals they set for themselves. One patient outlined how the process worked for them: Basically what happened is I told her what I wanted to do, she listened and I took over. She steered me in another direction sometimes or pointed something out I might need. (P7, 65 years old). All those interviewed considered the intervention to be worthwhile at some level. Most patients made positive changes to their lifestyle as a result of the intervention. One patient had good outcomes for issues that had previously been resistant to change. He said: I have certainly reduced smoking and alcohol. I have been able to put alcohol free days into my program which I found difficult before (P1, 69 years old). Practice staff and LAs recalled positive health and wellbeing outcomes for patients and the practice and systems level changes achieved. These changes included both practice and opportunistic risk factor data captured for enhanced CVD risk assessment and management. As explained by one of the PNs interviewed: Since the program was introduced I think it improved our collection of data, when patients come now all patients who come through the treatment room get height, weight, blood pressure, allergies and smoking assessment (PN12, 5 years as a PN). The key role of administrative staff improving the systems response to CVD prevention was also recognised during the study. One experienced PM noted that “ Since my involvement with this project I train my staff that are sending out the recall letter… so when a patient calls and makes an appointment they know exactly what to tell them ” (PM6, 18 years as a PM). Some staff believed the intervention resulted in a greater awareness by GPs of the importance of managing CVD risk and an increased number of CVD risk assessments being conducted. One PN reported that in her practice, CVD risk assessment had increased, stating that “ Doctors are doing a lot more cardiovascular assessments…they are actually doing them! ” (PN2, 6 years as a PN). There were staff in every practice that identified barriers to them reorganising practice to align with the intervention approach. While practice nurses were interested in an expanded role, many felt underprepared to address lifestyle modification and felt they needed more training in the area. This was highlighted by one nurse who said, “ I haven’t been trained for any of this, it has all been a learning curve for me ” (PN6, 7 years as a PN). Current financing models, time pressures and practical issues such as poorly integrated clinical software were also considered issues to be addressed if they were to become a more prevention-orientated practice. One GP was very clear about this, stating that: Anything that puts more work on the staff or the doctors is unrealistic… If there’s no incentive for us, at least make it that simple that I don’t have to invest extra work into it. (GP10, 20 years as a GP). When reflecting back on the intervention, almost all practice staff identified the difficulty of achieving sustained lifestyle modification itself as a major impediment to providing prevention-orientated services. They also considered these challenges would be exacerbated if there were out of pocket costs for patients. It’s a hard sell and that is why it should be free. (PN7, 1 year as a PN). The study examined the work required by general practice staff, Lifestyle Advisers and patients to implement CVD prevention in primary care consistent with existing evidence-based practice guidelines . The findings provide insight into the feasibility of intervention approach being embedded into real world practice. There were many aspects of the existing general practice system that could be supported, some with very small investment, to achieve and sustain more health promoting general practice setting. However, significant barriers to system change did exist, such as the design of current funding and the challenging nature of lifestyle modification, making it difficult for practitioners, and their patients, to move away from their usual practice. While usual practice did not normally include prevention-orientated activities, practice staff felt that prevention-orientated care was part of their role and that of general practice. Patients agreed that CVD prevention was important, important to their own health and they expected their General Practice to support them to reduce their risk of CVD. To enhance the likelihood of the intervention approach being adopted, the study outcomes highlight the need for an increased focus on strategies that build relationships across general practice. Mazza and colleagues found that for patients, having trust and a good rapport with a GP encouraged them to participate in prevention-orientated care . Consistent with this finding, most patients in the intervention identified strongly with the General Practice or their particular GP. The study outcomes also emphasised the value of nurturing relationships between patients and other members of the practice team. Practice Nurses were open to building these relationships and developing more comprehensive clinical roles. The administration staff also had an important role in engaging patients into the intervention particularly in areas related to coherence, supporting patients to see their general practice as interested in prevention and cognitive participation, getting the multiple internal and external relationships to work for the intervention and making it easy for patients to engage in the new activities. Greater recognition of the importance of this workforce and the development more specific strategies to support their role, should be a feature of future interventions. Redistribution of the work undertaken by general practice staff to support patients in health behaviour change was an important element of the intervention approach. This included the adoption of new or enhanced tasks related to health information quality, CVD risk assessment and behaviour change support for the practice staff. There was also the integration of a new primary health care workforce in the Lifestyle Advisors. The GPs were generally happy to undertake small additions to practice to deliver the intervention and to transfer more lengthy discussions on lifestyle modification to the LA. Most PNs welcomed the opportunity to expand their role, but found lifestyle modification a challenging area of practice because they had had little previous training or experience in the area. Limitations in health behaviour change skills of nurses working in primary health care and has been previously identified as a barrier to PN’s more active engagement in chronic disease prevention . Ongoing professional development, particularly using an academic detailing approach that offers tailored on-site support for the different needs of individual staff and practices, is needed to build the competence and confidence of this workforce in this new role. The positioning of a health coaching service in a primary care practice is a relatively new innovation. While most patients and practice staff reported positive outcomes from the LAs service, the LAs themselves perceived varying levels of interest, or as they put it “curiosity” about their service, particularly by practice staff. Liddy and colleagues found that it does take time for practices to adjust to having this new role in place and that a poor understanding of health coaching by other practice staff can limit its effectiveness . Therefore the addition of this new workforce has potential to make prevention-orientated services easier to deliver in the general practice setting, but it will need to be supported over an extended period to allow the potential benefits to be fully recognised. To continue this role would be an area requiring a new source of recurrent funding, as in the current intervention these positions were fully supported with short term project funds. The need for more recurrent funding for prevention activities and the challenging nature of lifestyle modification were identified as fundamental barriers to the intervention approach being adopted as usual practice. The lack of adequate funding, funding along with limited time and other competing priorities have been found by other researchers to be barriers to lifestyle modification in general practice . This ambivalence remains a fundamental barrier to prevention-orientated general practice, with the benefits of moving to delivery of more prevention services over remaining with the status quo, not yet compelling for the practices involved in the study. However, if positive patient health and practice system outcomes could be maintained, even at some level, over time the experience of CVD prevention could improve and help to build the value proposition for prevention-orientated services. The need to build the value proposition for prevention was also emphasised by the community-based lifestyle modification providers in their attempt to operate viable businesses . Until prevention-orientated services hold greater value for everyone and attract the collective action and subsequent advocacy for policy and funding reform, it is unlikely there will be the community and political will required to achieve change. The study demonstrated the utility of exploring implementation processes and the work required at an individual and organisational level to see translation of evidence-based CVD prevention practice into everyday general practice. It also showed the benefit of using NPT as a tool for examining implementation of a prevention-orientated activity. It allowed the researchers to ‘think through’ the data in a structured way and highlighted the work required by all stakeholders to implement such a complex intervention. Much of this information was unlikely to have been captured systematically in other ways. In particular, it helped to expose the ‘hidden work’ that needs to occur to create health promoting systems. The limitations of the study relate to the sample and the timing of the interviews. The study took place in a single city with a generally high standard of living. The sample of practice staff included only those willing to be interviewed, who may have been more interested in the intervention outcomes. Not all staff interviewed worked at one of the six practices over the entire period of the intervention, which limited their capacity to comment on the earlier stages of implementation. Data was collected at the completion of the pilot study, those interviewed may have forgotten aspects relevant to the early stages of the intervention. While consideration was given to collecting baseline information from stakeholders about what they expected from enhanced CVD prevention, given the novel nature of the intervention for many of the staff and patients, it was decided that this step would provide little additional information to justify the increased burden on participants. However, the study team did use the NPT constructs in the development of the strategies for the intervention. The study was strengthened by inclusion of all key workforce groups in general practice, and in particular, by including the patients who were the users of the service. The practices in the study operated using a range of business models and serviced a range of demographic groups. While some of the issues raised were directly related to the predominately fee-for-service funding model of general practice in Australia, many of the findings are relevant to the challenges faced by health systems globally in attempts to improve prevention-orientated primary health care. Finally, use of the NPT provided a rich and detailed framework for analysis and a strong theoretical grounding to the study. Despite widespread agreement that increasing CVD prevention activities in primary care is important, progress in reorientating health systems towards prevention-orientated practice has been limited. This study highlights the value of examining the implementation processes at a detailed level, including the experience of all stakeholders, in understanding of the feasibility of a complex intervention being embedded into usual practice. While many barriers will continue to impede the translation of the intervention approach, the study was able to highlight parts of the system that are highly influential to improving CVD prevention and are ready to change. Over time, supporting these areas will increase the capacity of general practice to become a health promoting setting and a true primary health care setting, which is a goal definitely worth pursuing. |
Histopathological and immunohistochemical characteristics of adult renal tumors: a five-year retrospective study in Mureş County, Romania | 4bde7d10-4832-450b-b612-5f68745026c7 | 11657323 | Anatomy[mh] | Renal tumors account for approximately 3% of all adult neoplasms . The occurrence is higher in countries with high incomes compared to middle- and low-income ones, presently standing as the seventh most prevalent cancer among men and the tenth most prevalent among women, with an increasing prevalence. It is commonly seen in patients in their sixth and seventh decade of life, with a median age of 64 years . Apart from smoking, obesity, and hypertension, genetic factors have been linked in the development of renal cell carcinoma (RCC). Therefore, RCC can occur as a sporadic form (~96%) or can be associated with familial syndromes [4% – a small percentage being associated with von Hippel–Lindau (VHL) disease] . The 2022 World Health Organization (WHO) Classification of renal tumors includes a wide spectrum of tumors ranging from benign to malignant, with several new entities defined on cytological, architectural, immunohistochemical (IHC), and cytogenetic characteristics. Clear cell renal cell carcinoma (CCRCC) still remains the most frequent histopathological subtype, followed by papillary renal cell carcinoma (PRCC), and chromophobe renal cell carcinoma (ChRCC), while renal oncocytoma (RO) is the most common benign tumor . Histology currently represents the “gold standard” of the diagnosis not only for establishing the tumor type, but particularly for identifying those factors which are directly correlated with the severity and prognosis of RCC : tumor stage and grade, presence of sarcomatoid/rhabdoid features and necrosis. Immunohistochemistry is an additional tool for establishing the histological subtype in cases where the histology is not conclusive . Unfortunately, most of the cases are discovered incidentally, in advanced stages, following ultrasonography (US), magnetic resonance imaging (MRI) or computed tomography (CT) examination for various medical conditions . Despite the progress made in diagnosing and treating RCC over the past 20 years, it remains the deadliest urological cancer, with a mortality rate of approximately 40% . Diagnosis in early stages would be an important step in order to improve the patients’ chance of survival . Aim The aim of our study was to assess the prevalence trend, as well as demographic and pathological characteristics of renal tumors over the last five years in Mureş County Clinical Hospital, Romania, and border areas. The aim of our study was to assess the prevalence trend, as well as demographic and pathological characteristics of renal tumors over the last five years in Mureş County Clinical Hospital, Romania, and border areas. Case selection We performed a retrospective study between January 2018 to December 2022 at the Pathology Department of Mureş County Clinical Hospital, including all patients who underwent a total or partial nephrectomy for renal tumor. All data were processed with the informed consent of the patients. This study was approved by the Ethics Committee of Mureş County Clinical Hospital (Approval No. 6817/2022). Clinical and pathological data Patient demographics (age and gender group) were collected from the Medical Records of Urology and Pathology Departments, respectively. All cases were reviewed by two pathologists specialized in uropathology (AB, AL). The histopathological features [tumor subtype, tumor stage, vascular invasion, necrosis, lymph node metastasis (LNM) and distant metastasis, when it was included in the resection specimen] were re-assessed based on Hematoxylin–Eosin (HE)-stained slides according to latest 2022 WHO Classification of Urinary and Male Genital Tumours (5th edition). We considered CCRCC as a tumor with optically clear cytoplasm cells, arranged in nests, tubes, or alveoli with a distinctive vascular pattern as a diagnostic indicator to distinguish it from other tumors . All cases of papillary carcinoma, regardless of whether they were type 1 or 2, were included in a single category according to the latest 2022 WHO Classification of Urinary and Male Genital Tumours (5th edition). Thus, a diagnosis of PRCC was made when the tumor had papillary or tubulo-papillary architecture with foamy macrophages in the axis of the papillae, lined by cuboidal cells with pale basophilic cytoplasm and uniform nuclei. Other morphological patterns included a solid growth, or thinly branching papillae lined by a pseudostratified layer of tumor cells, with abundant eosinophilic cytoplasm and nuclear atypia . All tumors with a solid architecture and with two types of cells (one with finely reticulated and translucent appearance, of the cytoplasm, often exhibiting a clear to foamy aspect and another type with densely eosinophilic and granular cytoplasm) were diagnosed as ChRCC. The nuclei were hyperchromatic, with irregular and wrinkled (raisinoid) appearance, with the presence of perinuclear haloes as a distinguishing characteristic, frequently with binucleation . Sarcomatoid and rhabdoid differentiation were assessed in all histological subtypes. Sarcomatoid differentiation was considered when spindle cells organized in sheets and fascicles with cellular atypia and high nuclear pleomorphism were identified. Rhabdoid differentiation was considered when either cohesive groups of cells or individual round to polygonal cells with a loose arrangement and eosinophilic inclusions within their globular cytoplasm and large, irregular nuclei situated off-center were present . A diagnosis of RO was established when the tumor was well-circumscribed with a solid, microcystic, macrocystic or tubular architecture, and a characteristic central scar. The tumor cells were round to polygonal, displaying densely granular eosinophilic cytoplasm with round, uniform nuclei and a small nucleolus at the center, embedded in myxoid or hypocellular hyalinized stroma . Other rare histological subtypes were defined by a set of histopathological criteria in accordance with latest 2022 WHO Classification of Urinary and Male Genital Tumours (5th edition). Tumors included in the study were graded according to Fuhrman’s and WHO/International Society of Urological Pathology (ISUP) grading systems. All tumors graded before 2019 according to Fuhrman’s grading system were re-assessed in line with the WHO/ISUP grading system (4th edition), as follows: grade 1 – absent or not easily noticeable nucleoli at high magnification (40×); grade 2 – not prominently visible nucleoli at a low magnification (10×), but visible, eosinophilic nucleoli at high magnification (40×); grade 3 – clearly visible, reddish nucleoli at low magnification (10×); grade 4 – nuclei with extreme variation in shape, presence of multinucleated cells, displaying features of rhabdoid and/or sarcomatoid differentiation. The pathological stage was assessed according to the latest 2022 WHO Classification of Urinary and Male Genital Tumours (5th edition) . In cases with misleading morphology, a panel of five antibodies was used: anti-paired-box 8 (PAX8; mouse monoclonal primary antibody, clone MRQ-50, Cat# 760-4618), anti-cluster of differentiation 10 (CD10; rabbit monoclonal antibody, clone SP-67, Cat# 790-4506), anti-alpha-methylacyl-coenzyme A (CoA) racemase (AMACR; mouse monoclonal antibody, clone SP116, Cat# 760-501), anti-cytokeratin 7 (CK7; rabbit monoclonal primary antibody, clone SP52, Cat# 790-4462), anti-cluster of differentiation 117 [CD117 (c-Kit); monoclonal primary antibody, clone 9.7, Cat# 790-2951]. Staining procedures were performed on the Ventana BenchMark ULTRA (Ventana Medical Systems, Inc., Tucson, AZ, USA) automated stainer following the manufacturer’s protocols. The IHC reactions were considered positive as follows: nuclear staining for PAX8, membranous staining for CD10 and CD117, cytoplasmic staining for AMACR, cytoplasmic and membranous staining for CK7. Statistical analysis Statistical analysis was performed using Microsoft Excel and Statistical Package for the Social Sciences (SPSS; IBM v29.0) program for macOS. Descriptive statistics were applied to analyze the data. We performed a retrospective study between January 2018 to December 2022 at the Pathology Department of Mureş County Clinical Hospital, including all patients who underwent a total or partial nephrectomy for renal tumor. All data were processed with the informed consent of the patients. This study was approved by the Ethics Committee of Mureş County Clinical Hospital (Approval No. 6817/2022). Patient demographics (age and gender group) were collected from the Medical Records of Urology and Pathology Departments, respectively. All cases were reviewed by two pathologists specialized in uropathology (AB, AL). The histopathological features [tumor subtype, tumor stage, vascular invasion, necrosis, lymph node metastasis (LNM) and distant metastasis, when it was included in the resection specimen] were re-assessed based on Hematoxylin–Eosin (HE)-stained slides according to latest 2022 WHO Classification of Urinary and Male Genital Tumours (5th edition). We considered CCRCC as a tumor with optically clear cytoplasm cells, arranged in nests, tubes, or alveoli with a distinctive vascular pattern as a diagnostic indicator to distinguish it from other tumors . All cases of papillary carcinoma, regardless of whether they were type 1 or 2, were included in a single category according to the latest 2022 WHO Classification of Urinary and Male Genital Tumours (5th edition). Thus, a diagnosis of PRCC was made when the tumor had papillary or tubulo-papillary architecture with foamy macrophages in the axis of the papillae, lined by cuboidal cells with pale basophilic cytoplasm and uniform nuclei. Other morphological patterns included a solid growth, or thinly branching papillae lined by a pseudostratified layer of tumor cells, with abundant eosinophilic cytoplasm and nuclear atypia . All tumors with a solid architecture and with two types of cells (one with finely reticulated and translucent appearance, of the cytoplasm, often exhibiting a clear to foamy aspect and another type with densely eosinophilic and granular cytoplasm) were diagnosed as ChRCC. The nuclei were hyperchromatic, with irregular and wrinkled (raisinoid) appearance, with the presence of perinuclear haloes as a distinguishing characteristic, frequently with binucleation . Sarcomatoid and rhabdoid differentiation were assessed in all histological subtypes. Sarcomatoid differentiation was considered when spindle cells organized in sheets and fascicles with cellular atypia and high nuclear pleomorphism were identified. Rhabdoid differentiation was considered when either cohesive groups of cells or individual round to polygonal cells with a loose arrangement and eosinophilic inclusions within their globular cytoplasm and large, irregular nuclei situated off-center were present . A diagnosis of RO was established when the tumor was well-circumscribed with a solid, microcystic, macrocystic or tubular architecture, and a characteristic central scar. The tumor cells were round to polygonal, displaying densely granular eosinophilic cytoplasm with round, uniform nuclei and a small nucleolus at the center, embedded in myxoid or hypocellular hyalinized stroma . Other rare histological subtypes were defined by a set of histopathological criteria in accordance with latest 2022 WHO Classification of Urinary and Male Genital Tumours (5th edition). Tumors included in the study were graded according to Fuhrman’s and WHO/International Society of Urological Pathology (ISUP) grading systems. All tumors graded before 2019 according to Fuhrman’s grading system were re-assessed in line with the WHO/ISUP grading system (4th edition), as follows: grade 1 – absent or not easily noticeable nucleoli at high magnification (40×); grade 2 – not prominently visible nucleoli at a low magnification (10×), but visible, eosinophilic nucleoli at high magnification (40×); grade 3 – clearly visible, reddish nucleoli at low magnification (10×); grade 4 – nuclei with extreme variation in shape, presence of multinucleated cells, displaying features of rhabdoid and/or sarcomatoid differentiation. The pathological stage was assessed according to the latest 2022 WHO Classification of Urinary and Male Genital Tumours (5th edition) . In cases with misleading morphology, a panel of five antibodies was used: anti-paired-box 8 (PAX8; mouse monoclonal primary antibody, clone MRQ-50, Cat# 760-4618), anti-cluster of differentiation 10 (CD10; rabbit monoclonal antibody, clone SP-67, Cat# 790-4506), anti-alpha-methylacyl-coenzyme A (CoA) racemase (AMACR; mouse monoclonal antibody, clone SP116, Cat# 760-501), anti-cytokeratin 7 (CK7; rabbit monoclonal primary antibody, clone SP52, Cat# 790-4462), anti-cluster of differentiation 117 [CD117 (c-Kit); monoclonal primary antibody, clone 9.7, Cat# 790-2951]. Staining procedures were performed on the Ventana BenchMark ULTRA (Ventana Medical Systems, Inc., Tucson, AZ, USA) automated stainer following the manufacturer’s protocols. The IHC reactions were considered positive as follows: nuclear staining for PAX8, membranous staining for CD10 and CD117, cytoplasmic staining for AMACR, cytoplasmic and membranous staining for CK7. Statistical analysis was performed using Microsoft Excel and Statistical Package for the Social Sciences (SPSS; IBM v29.0) program for macOS. Descriptive statistics were applied to analyze the data. Two hundred twenty cases of total/partial nephrectomies were included in our study. The distribution of cases during the study period is shown in Figure . Notably, there was a decrease of cases during the initial phase of the coronavirus disease 2019 (COVID-19) pandemic (2020), followed by a significant increase in 2021. Table (gender, site of the tumor, histological subtype) summarizes the demographic and pathological data for the study cases. Among the 220 patients included in this study, the majority (n=138, 62.7%) were men, less than half (n=82, 37.2%) being women. Over half of the patients (n=146, 63.63%) were ≥60 years old, 32.27% (n=71) were between 40–59 years old, while only 1.36% (n=3) were less than 40 years old. The mean age was 62.72 years (range: 33–84 years). We found a predominance of right kidney tumor masses, with 52.2% (n=115), compared to 47.7% in the left kidney. As shown in Table , the most common histological subtype was CCRCC (n=181, 82.27%), some of the cases with sarcomatoid/rhabdoid differentiation (n=24, 10.9%) (Figures , ), followed by RO (n=16, 7.27%), PRCC (n=15, 6.81%) (Figure ), ChRCC (n=6, 2.72%) (Figure ), and rare subtypes [Xp11.2 translocation RCC, n=1, 0.45%; thyroid-like follicular carcinoma (TLFC), n=1, 0.45%]. IHC was performed for cases with atypical or overlapping morphology (n=41, 18.63%). IHC staining with PAX8, CD10 and AMACR confirmed the CCRCC diagnosis in 18 out of 181 cases, while AMACR and CK7 were performed for PRCC in 10 out of 15 cases. CD117 and CK7 were used to differentiate ChRCC from RO [n=6 (2.7%) and n=16 (7.2%), respectively]. Table documents the demographic and pathological characteristics of the renal malignant tumors according to the histological subtype. We observed a minimal variation in the distribution of cases among different stages of RCC, therefore pT3a was the most frequent stage (n=64, 29.09%), followed closely by pT1b (n=58, 26.4%) and pT1a (n=57, 25.9%), while the least common was pT4 stage (n=3, 1.36%). We examined each histological subtype according to different pathological characteristics and found that most CCRCC cases were pT1b (n=51, 23.2%), closely followed by pT1a (n=48, 21.8%) and pT3a (n=41, 18.6%). All pT3a cases were accompanied by vascular invasion (n=30, 13.6%), tumor necrosis (n=38, 17.3%), sinus fat invasion (n=20, 9.0%), perinephric fat invasion (n=11, 5.0%) and distant metastasis (n=1, 0.9%), showing a more aggressive behavior. Moreover, CCRCC with sarcomatoid/rhabdoid differentiation was more frequently identified in advanced stages such as pT3a (n=18, 8.2%), all of them associated with LNM, vascular invasion, tumor necrosis, sinus fat invasion and perinephric fat invasion, without distant metastasis. Out of 15 analyzed cases of PRCC, pT1a stage (n=6, 2.7%) was most commonly identified, followed closely by pT3a stage (n=4, 1.8%) associated with LNM, tumor necrosis and vascular invasions, and pT2a stage (n=3, 1.4%) associated with tumor necrosis. ChRCC was one of the least common histological subtypes, identified in pT1b (n=3, 1.4%) and pT2a (n=2, 0.9%) stages, with no LNM, distant metastases or vascular invasions showing a less aggressive profile compared to other types of renal carcinoma. Although sinus fat invasion was observed in 35 cases, primarily of CCRCC with and without sarcomatoid/rhabdoid differentiation (n=32, 91.4%), perinephric fat invasion was noted in 21 (9.54%) cases of CCRCC, including 10 (4.54%) cases with sarcomatoid/rhabdoid features, one (0.45%) case of ChRCC, and one (0.45%) case of Xp11.2 translocation. Necrosis was identified in tumors with a higher stage, over half of them in pT3a and pT4 stages. There were 55 cases of CCRCC of which 17 cases with sarcomatoid/rhabdoid features, five cases of PRCC, one case of ChRCC and one case of Xp11.2 translocation. In terms of distant metastasis, one case of CCRCC had ipsilateral adrenal gland metastasis (found at the time of surgery). Histopathological analysis of the 181 cases of CCRCC and 15 of PRCC showed that that more than half (n=122, 62.24%) of the cases were low-grade tumors (WHO/ISUP grade 2) including pT1b (n=42), pT1a (n=39) and pT3a (n=17) stages for CCRCC and pT1a (n=4), pT2a (n=3) and pT3a (n=2) stages for PRCC, with WHO/ISUP grade 3 as the second most common finding. Most of the high-grade tumors (WHO/ISUP grade 3 and WHO/ISUP grade 4) were observed in a higher stage such as pT3a. All cases with sarcomatoid/rhabdoid features were graded as high-grade tumors (WHO/ISUP grade 4) (Table ). The results of our study provide a comprehensive overview of the demographic (gender, age) and pathological characteristics (RCC subtypes, tumor stage, vascular invasion, necrosis, LNM and distant metastasis) of renal tumors over a five-year period in our region and border areas. The incidence rates of RCC show notable variations across different regions worldwide. Globally, North America stands out with the highest estimated incidence (12/100 000), indicating cumulative risks of 1.8% for males and 0.9% for females. In Western Europe, the highest incidence rates (9.8%) are for both genders. In contrast, Africa exhibits the lowest incidence and mortality rates, with cumulative risks less than 0.2% for both genders . The impact of COVID-19 has been widely documented in the literature. This epidemiological emergency has led to a decline in clinical and surgical activities, affecting patients’ access to high-quality medical services . In our department, we observed a decrease of renal tumors cases during the initial phase of the COVID-19 pandemic (year 2020), followed by a significant increase in 2021. This trend was also reported by Mariotto et al., with a decrease of cases by 10% in 2020 compared to 2019, due to delays in cancer diagnosis and screening caused by the pandemic. However, they observed, that the incidence rates increased in 2021 by 9.4% compared to 2020, indicating a partial recovery in cancer detection and treatment services . Our study found a male predominance (62.7%) compared to female patients (37.2%), with a male-to-female (M:F) ratio of 1.6:1, similar to the literature where there is described a higher occurrence of renal tumors in males compared to females with a M:F ratio of 1.5:1 . Tang et al. reported in their study that the highest incidence of RCC was found in the sixth or seventh decades of life, with a mean age of 61 years . The mean age in our study was 62.6 years, with more than half of the patients being over 60 years old (range 33–84 years). In a study conducted by Padala et al., there was a higher prevalence of right-sided kidney tumors in 68.6% of cases which corresponds with our results where over half of the cases had right kidney renal tumors . According to the data reported in the literature, the most common histological subtypes of RCC are CCRCC, PRCC and ChRCC. Regarding benign tumors, RO is the most common benign epithelial tumor . In the study reported by Sheenu et al., there was a lower prevalence of RO, with only one (1.9%) case identified . However, the exact percentages can vary across different studies . The findings of this present study aligned with the literature, showing that CCRCC is the most common histological subtype (82.3%), with a significant number of cases with sarcomatoid/rhabdoid features (13.3%), followed by PRCC (6.8%), ChRCC (2.7%) and other rare subtypes [Xp11.2 translocation RCC (0.5%) and TLFC of the kidney (TLFC-K) (0.5%)]. RO accounted for only 7.2% of our cases. Mohamed et al. reported frequencies of 67.8% for CCRCC, 16.6% for PRCC, and 7.1% for ChRCC, respectively, while Ali et al. found 71.1% for CCRCC, 13.6% for PRCC, and 11% for ChRCC. These findings underscore a lower prevalence of the chromophobe subtype and the highest prevalence of the clear cell subtype. In rare cases, morphology and immunohistochemistry are not specific and molecular biology can help in establishing the histological subtype. In one case of CCRCC, although the microscopic appearance (tubular, microcystic and focally papillary architecture) and an atypical IHC profile (CK7+, CD10–) were not specific for CCRCC, the molecular study that was conducted revealed a VHL gene mutation. These findings underscore that CCRCC can have an unusual microscopic appearance. Ancillary tests are essential for tumors with misleading morphology, because each subtype has a different outcome. In one case, based on a unique histological pattern, exhibiting a combination of clear cells and papillary architecture a suspicion of RCC with Xp11.2 translocation was raised, and the case was referred to a specialized center, Hospital Lyon Sud, France. Ancillary advanced tests confirmed the diagnosis. Xp11.2 translocation-associated RCC has a prognosis that is relatively less understood due to the rarity of reported cases of this neoplasm. The final diagnosis requires confirmation of the Xp11.2 translocation through fluorescence in situ hybridization (FISH) . Another rare subtype of RCC was TLFC-K. Due to the rarity of this entity (<50 cases described in the literature), it was removed from the classification of the latest 2022 WHO Classification of Urinary and Male Genital Tumours (5th edition) . In a study performed by Kuthi et al., the WHO/ISUP grading system for CCRCC indicated that most of the cases were low-grade tumors classified as WHO/ISUP grade 2 (n=364, 47.02%) and WHO/ISUP grade 1 (n=157, 20.28%) . These findings are similar to our results, where the majority of cases were classified as WHO/ISUP grade 2 (n=112, 61.87%). In contrary, WHO/ISUP grade 3 (n=38, 21%) was the second most common finding in our CCRCC cases. Khor et al. showed no significant difference in outcomes between grades 1, 2, and 3, but the difference was significant for grade 4, with a poorer prognosis . In a study conducted by Qian et al., the WHO/ISUP grading system for PRCC indicated that 56 (64.4%) individuals were classified as WHO/ISUP grade 1, 17 (19.5%) as WHO/ISUP grade 2, seven (8.05%) as WHO/ISUP grade 3, and seven (8.05%) as WHO/ISUP grade 4 . In our study, the patients were classified mostly as WHO/ISUP grade 2 (n=10, 66.7%) and WHO/ISUP grade 3 (n=5, 33.3%). Regarding the tumor stage, in Mohamed et al. study, tumor pT2 stage was observed in 50% of the patients, while pT1 stage accounted for 21.4%, pT4 stage for 15.5%, and pT3 stage for 13.1% . These findings are different from the results reported in the majority of other studies . Salako et al. (similar to many other studies in the literature), indicated that most of the cases were in tumor pT3 stage (n=18, 35.3%), followed by pT2 stage (n=14, 27.5%) and pT1 stage (n=10, 19.6%), with the least common pT4 stage (n=9, 17.6%) and sustain our results. Our study shows that pT3 stage was predominant (66.3%), followed by pT1b stage (28.9%) and pT1a stage (27.9%), with the least common pT4 stage (1.5%). Over 50% of RCCs are detected incidentally due to routine imaging for various medical conditions. Only 30% of RCC patients are diagnosed based on symptoms because small masses often remain asymptomatic. Therefore, up to 20–30% of patients have metastasis at diagnosis, while the classic triad of flank pain, hematuria, and abdominal mass is rare, presenting in only 4–17% of cases. As a result, the tumors are most commonly identified at an advanced stage such as pT3, with a higher nuclear grade as our study also demonstrated (WHO/ISUP grade 3 was the second most common finding). Early diagnosis would significantly improve prognosis . Sinus fat invasion, perinephric fat invasion or invasion into the renal vein are included in pT3a stage, independently of tumor size. In our study, sinus fat invasion was identified in 35 (15.9%) cases: 32 CCRCC cases of which 12 with sarcomatoid/rhabdoid features, two of PRCC and one of Xp11.2 translocation. Twenty-three cases had perinephric fat invasion: 21 cases of CCRCC, one case of ChRCC and one case of Xp11.2 translocation. Some studies, as the one conducted by Stühler et al. revealed that tumors with sinus fat invasion are more aggressive, with a poorer prognosis, compared to those with perinephric fat invasion . These findings emphasize the importance of a well done macroscopy, with multiple sections sampled from the hilum, in order to identify sinus fat invasion. LNM was found in five cases of CCRCC with sarcomatoid features and in two cases of PRCC. All tumors with LNM were of pT3a stage. The presence of LNM is correlated with a higher tumor stage , as our findings also demonstrate. There is also a higher probability of distant metastases for tumors in advanced stages . In our study, we found only one case of CCRCC with distant metastasis. In a comprehensive retrospective study conducted across seven Latin American countries and Spain, it was described that the five-year survival rate for surgically resected RCC reached 86.1%. Several factors identified as significantly influencing survival outcomes leading to a poor prognostic were age, LNM, fat and/or vascular invasion, tumor necrosis, and tumor size exceeding 7 cm . The presence of sarcomatoid/rhabdoid features and vascular invasion were identified as morphological aspects with prognostic value, with the former being associated with a higher grade of the disease and a poor long-term prognosis . In our study, vascular invasion was found in 46 out of 220 (20.9%) cases, most of them in CCRCC (n=43, 19.5%), including some with sarcomatoid/rhabdoid features (n=13, 5.9%), indicating a more aggressive behavior that leads to an unfavorable prognosis. Tumor necrosis is another important histological feature with prognostic value and can be associated with larger tumor size and a higher grade, leading to aggressive tumor behavior. It has been documented in 21–32% of CCRCCs. Several studies have also associated the presence of tumor necrosis with poorer survival outcomes. Researchers found that analyzing both WHO/ISUP grading system and tumor necrosis changed the prognostic compared to WHO/ISUP grading alone. They observed that tumor necrosis significantly influenced prognosis, particularly in WHO/ISUP grade 3 tumors . In our study, tumor necrosis was documented in 28.18% (n=62) of cases, most of them in CCRCC (n=55, 25%) of which 7.7% (n=17) with sarcomatoid/rhabdoid features. Over half of cases were identified in advanced stages such as pT3a, closely followed by pT2a and pT1b stages, and were associated with higher nuclear grades, as demonstrated by our findings with WHO/ISUP grade 3 being the second most common identified, suggesting an aggressive behavior and a poorer prognosis. The spectrum of adult renal tumors in our region (Mureş County, Romania) aligns with the data reported in the literature. CCRCC was the most frequent histological subtype and pT3a the most common stage identified. Even if in most cases the diagnosis is based on the morphological aspect and IHC profile, in rare cases molecular biology and genetics can help in establishing an accurate diagnosis. Our study contributes to the understanding of renal tumors characteristics, identifying potential factors with impact in the progression and prognosis of the disease. The authors declare that they have no conflict of interests. |
Physician-to-Physician eConsultations to Ophthalmologists at an Academic Medical Center | 5db88e6c-c9e1-4e89-91fe-44b6b963a74a | 11463702 | Ophthalmology[mh] | Electronic consultations (eConsults) are asynchronous, provider-to-provider exchanges that occur within shared electronic medical record (EMR) systems or secure online platforms. In eConsults, referring providers submit a clinical question via an EMR, and consulting providers review that question, along with available information in the patient's medical record, and document clinical impressions and recommendations in a clinical note; there is no direct consultant/patient interaction. This asynchronous method of telecommunication confers a range of advantages to patients, providers, and healthcare systems, including increased timely access to specialist consultation, decreased resource waste, improved care coordination, greater patient and provider satisfaction, and cost savings. – Through eConsults, providers can determine whether patients require in-person evaluation, which can decrease the number of unnecessary referrals. When patient referral is deemed beneficial, eConsults can streamline the process by identifying the appropriate subspecialty for referral. , Furthermore, eConsults can determine necessary pre-visit diagnostic workup or imaging, thus optimizing clinical evaluation during the visit and reducing further delays in patient care. , In addition to care coordination, eConsults can ensure that subspecialty evaluation is achieved. The eConsults can bypass many of the hurdles associated with completing specialist referrals, increase physician follow-up and awareness of referral completion, and remove the burden of seeking specialty care from patients. By improving the efficiency and effectiveness of interprofessional clinical exchange, eConsults can also improve access to and quality of care. , However, eConsults also have drawbacks. Namely, there are risks of diagnostic errors and concerns for patient safety given the lack of an in-person examination. Ophthalmology, like many other medical specialties, has increased use of telemedical modalities including eConsults during the COVID-19 pandemic as a means of continuing access to specialty care while limiting in-person exposures. , Many subspecialties of medicine have reported on the benefits of eConsults. , Despite the many known advantages of eConsults, the role of eConsults in ophthalmology has yet to be explored. In this study, we describe the use of ophthalmology eConsults in an academic medical center. We report specifically on the diagnostic accuracy, outcomes, and response timeliness of eConsults, as well as the different ophthalmic conditions inquired about through eConsults. In doing so, we characterize the types of clinical questions asked and identify the learning needs of nonophthalmic providers. Finally, we assess the utility of eConsults to safely and comprehensively manage nonurgent ophthalmic conditions remotely. Unlike other literature published on various modalities of ophthalmic telehealth, this study is the first to describe the feasibility and potential impact of eConsults in ophthalmology. Study Design We conducted a retrospective cohort study of all eConsults submitted to the Ophthalmology Department of Massachusetts Eye and Ear (MEE) from February 11, 2019 through August 18, 2021. The eConsult Program Massachusetts General Brigham (MGB) is a tertiary academic health care system with multiple inpatient and ambulatory health centers throughout the state of Massachusetts. The MGB eConsult program was established in 2013 and expanded to include ophthalmology eConsults in 2019. This program enables providers across all specialties within the MGB healthcare system to submit a clinical question to an ophthalmologist through a shared EMR system, and the ophthalmologist can respond via the EMR without evaluating the patient in person. The eConsult is requested through an EMR order and consists of three fields for referring providers to complete: (1) the reason for eConsult, (2) specific patient care questions, and (3) any additional comments, pertinent patient history, and attachments. Clinical images taken by the referring provider can be uploaded into the EMR. In this study, the term “referring provider” refers to a non-ophthalmology healthcare provider who submits an eConsult to an ophthalmologist on behalf of a patient. The terms “consulting provider” and “eConsultant” are used interchangeably and refer to the ophthalmologist receiving and reviewing the eConsult. The term “referral request” refers to the act of placing an electronic order in the EMR to refer a patient for an in-person evaluation by an ophthalmologist. All ophthalmology eConsults were assigned to and addressed by a comprehensive ophthalmologist, who additionally reviewed the patient's medical history, ophthalmic history, recent progress notes, associated images, and diagnostic tests. The general ophthalmologist asked for clinical advice from appropriate ophthalmology subspecialists to answer eConsult questions as needed. Clinical recommendations were returned to the referring provider through an “eConsult note” within the EMR. The referring provider then determined next steps, including relaying the eConsultant's recommendations to the patient, scheduling appropriate follow-up, and determining which aspects of the eConsultant recommendations act upon. A workflow diagram of the eConsult process is included in . There was no direct ophthalmologist-to-patient exchange, and all eConsults included standardized language indicating that ongoing management of the patient's clinical problem is the responsibility of the referring provider and other members of the patient's care team, that eConsults are conducted in the absence of an examination or direct conversation with the patient, and that recommendations made are solely based on the information provided by the referring provider in the eConsult and information available in the EMR. The eConsult program is funded internally by MGB and does not involve charges to patient insurance; rather, MGB pays eConsultants a standard fixed fee for their consulting service. Data Extraction and Analysis All MEE ophthalmology eConsults were retrospectively reviewed through manual review of eConsult encounters in the EMR system, including the reason for eConsult, the date and time an eConsult order was placed and subsequently responded to, the presence of external images of patient eyes, results of diagnostic and screening tests (e.g., fundus photography, newborn vision screens, computed tomography, magnetic resonance imaging, radiography), eConsultant diagnosis, and clinical recommendation . Patient demographic data, including patient age at time of eConsult, sex, race, ethnicity, and insurance coverage status, were collected. Demographic data of referring providers (medical specialty, academic degree, and location) were also collected . Whether patients had known past ocular history or an established ophthalmologist, defined as any visit with an ophthalmologist in the five years before eConsult, was also recorded. Each eConsult was classified into clinically meaningful diagnostic categories based on the literature. The eConsult questions underwent thematic review to identify the types of clinical questions asked by providers and each question was assigned a category type. Seven question types were identified, including issues of diagnosis, treatment, management, triage/referral, workup, risk assessment and timing of routine screenings, and medication management . The eConsults were assessed for whether follow-up with an ophthalmologist was recommended, which party (eConsultant or referring provider) was responsible for placing the referral, whether the referral was placed, and whether a follow-up appointment occurred. Patient charts were reviewed up to six-months following the initial eConsult to assess for subsequent emergency department (ED) encounters associated with the eConsult concern, evidence of eConsult recommendations being relayed to the patient, and patient presentation to an ophthalmologist for in-person evaluation within the MGB system. Clinical details of subsequent in-person ophthalmology visits were assessed, including the diagnoses made during in-person ophthalmology visits which were assessed for concordance with any diagnoses made during eConsults. Institutional level data on ICD-10 codes for patient visits to the MEE ophthalmology-specific ED during the period of this study were obtained through automated EMR reports. This data was used to identify the most common eye diagnoses presenting to the MEE ED, which were used as a comparison for the eConsult eye concerns. Descriptive statistics and chi-square analyses were executed using R version 4.1.0 (R Foundation for Statistical Computing). A P value of α <0.05 was considered to be statistically significant. Institutional Review Board This study protocol was reviewed by the MGB Institutional Review Board and determined to be a Quality Improvement study. Therefore approval was not required. Data Availability The data used in this study is available on request. We conducted a retrospective cohort study of all eConsults submitted to the Ophthalmology Department of Massachusetts Eye and Ear (MEE) from February 11, 2019 through August 18, 2021. Massachusetts General Brigham (MGB) is a tertiary academic health care system with multiple inpatient and ambulatory health centers throughout the state of Massachusetts. The MGB eConsult program was established in 2013 and expanded to include ophthalmology eConsults in 2019. This program enables providers across all specialties within the MGB healthcare system to submit a clinical question to an ophthalmologist through a shared EMR system, and the ophthalmologist can respond via the EMR without evaluating the patient in person. The eConsult is requested through an EMR order and consists of three fields for referring providers to complete: (1) the reason for eConsult, (2) specific patient care questions, and (3) any additional comments, pertinent patient history, and attachments. Clinical images taken by the referring provider can be uploaded into the EMR. In this study, the term “referring provider” refers to a non-ophthalmology healthcare provider who submits an eConsult to an ophthalmologist on behalf of a patient. The terms “consulting provider” and “eConsultant” are used interchangeably and refer to the ophthalmologist receiving and reviewing the eConsult. The term “referral request” refers to the act of placing an electronic order in the EMR to refer a patient for an in-person evaluation by an ophthalmologist. All ophthalmology eConsults were assigned to and addressed by a comprehensive ophthalmologist, who additionally reviewed the patient's medical history, ophthalmic history, recent progress notes, associated images, and diagnostic tests. The general ophthalmologist asked for clinical advice from appropriate ophthalmology subspecialists to answer eConsult questions as needed. Clinical recommendations were returned to the referring provider through an “eConsult note” within the EMR. The referring provider then determined next steps, including relaying the eConsultant's recommendations to the patient, scheduling appropriate follow-up, and determining which aspects of the eConsultant recommendations act upon. A workflow diagram of the eConsult process is included in . There was no direct ophthalmologist-to-patient exchange, and all eConsults included standardized language indicating that ongoing management of the patient's clinical problem is the responsibility of the referring provider and other members of the patient's care team, that eConsults are conducted in the absence of an examination or direct conversation with the patient, and that recommendations made are solely based on the information provided by the referring provider in the eConsult and information available in the EMR. The eConsult program is funded internally by MGB and does not involve charges to patient insurance; rather, MGB pays eConsultants a standard fixed fee for their consulting service. All MEE ophthalmology eConsults were retrospectively reviewed through manual review of eConsult encounters in the EMR system, including the reason for eConsult, the date and time an eConsult order was placed and subsequently responded to, the presence of external images of patient eyes, results of diagnostic and screening tests (e.g., fundus photography, newborn vision screens, computed tomography, magnetic resonance imaging, radiography), eConsultant diagnosis, and clinical recommendation . Patient demographic data, including patient age at time of eConsult, sex, race, ethnicity, and insurance coverage status, were collected. Demographic data of referring providers (medical specialty, academic degree, and location) were also collected . Whether patients had known past ocular history or an established ophthalmologist, defined as any visit with an ophthalmologist in the five years before eConsult, was also recorded. Each eConsult was classified into clinically meaningful diagnostic categories based on the literature. The eConsult questions underwent thematic review to identify the types of clinical questions asked by providers and each question was assigned a category type. Seven question types were identified, including issues of diagnosis, treatment, management, triage/referral, workup, risk assessment and timing of routine screenings, and medication management . The eConsults were assessed for whether follow-up with an ophthalmologist was recommended, which party (eConsultant or referring provider) was responsible for placing the referral, whether the referral was placed, and whether a follow-up appointment occurred. Patient charts were reviewed up to six-months following the initial eConsult to assess for subsequent emergency department (ED) encounters associated with the eConsult concern, evidence of eConsult recommendations being relayed to the patient, and patient presentation to an ophthalmologist for in-person evaluation within the MGB system. Clinical details of subsequent in-person ophthalmology visits were assessed, including the diagnoses made during in-person ophthalmology visits which were assessed for concordance with any diagnoses made during eConsults. Institutional level data on ICD-10 codes for patient visits to the MEE ophthalmology-specific ED during the period of this study were obtained through automated EMR reports. This data was used to identify the most common eye diagnoses presenting to the MEE ED, which were used as a comparison for the eConsult eye concerns. Descriptive statistics and chi-square analyses were executed using R version 4.1.0 (R Foundation for Statistical Computing). A P value of α <0.05 was considered to be statistically significant. This study protocol was reviewed by the MGB Institutional Review Board and determined to be a Quality Improvement study. Therefore approval was not required. The data used in this study is available on request. A total of 100 pediatric and adult ophthalmic eConsults were ordered and completed between February 11, 2019, and August 18, 2021. Ophthalmic eConsults were most frequently ordered by internal medicine providers (67%), followed by family medicine providers (13%), pediatricians (9%), medicine-pediatrics providers (5%), and six other types of specialists . Nearly one quarter (24%) of eConsults asked two or more clinical questions. Regarding ophthalmic diagnoses, hordeola and chalazia accounted for 14% (14) of eConsults, with a total of 45 different ophthalmic diagnoses being asked about . All eConsults were considered nonurgent. Of the 100 eConsults ordered, 52% included a picture, and 8% included diagnostic imaging from previous visits. The average response time and standard deviation (SD) from when the eConsult order was placed to the time the consultation was completed was 1.6 days (SD ±1.9). The average number of days from eConsult response to in-person follow-up was 28.9 (SD ± 27.4) days. Of the 100 eConsults, in-person evaluation was recommended in 62% (n = 62) of cases; roughly 13% of these were due to medication follow-up (e.g., initiation of ophthalmic steroid drops). For patients requiring referral for in person evaluation by an ophthalmic subspecialty, MEE staff volunteered to submit the referral request for 41.9% (n = 26) of patients to ensure rapid referral placement and patient follow-up; referring providers were responsible for submitting a referral request for the remaining 58.1% (n = 36) of patients. However, the rate of actually submitting those referrals varied where MEE staff submitted the referral request 76.9% (n = 20/26) of the time whereas the referring provider submitted the referral request 66.7% (n = 24/36) of the time. No statistically significant difference was found between referring providers and eConsultants in the likelihood of placing the referral request ( P = 0.3799). Of the 62 patients recommended for in-person evaluation, 48.4% (n = 30) presented to an ophthalmologist for evaluation. Agreement in diagnostic concordance between the eConsultant clinical diagnosis and in-person diagnosis occurred in 93.3% (n = 28) of cases. Two cases of non-concordant diagnoses were documented (6.9%). The first was a case of light sensitivity and blurred vision secondary to central retinal vein occlusion initially attributed to post-concussion syndrome; the second was a case of orbital fat prolapse initially thought to be a hordeola. Referring providers proceeded to follow eConsult recommendations in 91% (n = 91) of cases. Evidence that the referring provider relayed the ophthalmologist plan to the patient was documented in 75% (n = 75) of cases based on documentation anywhere in the EMR (e.g., visit note, telephone encounter). Only 5% (n = 5) of patients presented to any MGB ED for issues related to the eConsult after the eConsult was placed. Seventy-nine percent of patients presented to any MGB clinic within the six months after the eConsult request, which indicated they were not lost to follow-up in the system. Before the time of eConsult placement, the vast majority of patients (73%, n = 73) had never seen an ophthalmologist or had any documented visit from an ophthalmologist accessible in the EMR and had no known ocular history (62%, n = 62) . Patient demographics including age, race/ethnicity, and sex are presented in . In this study we found that eConsults in ophthalmology provide for high diagnostic accuracy, are useful across a range of clinical diagnoses and concerns, and result in timely responses. Additionally, this study thematically describes eConsult questions and topic areas. Although eConsults have not previously been reported within the field of ophthalmology, this study supports the efficacy and feasibility of eConsults to provide asynchronous specialty advice to non-ophthalmic providers. , Clinical Outcomes The results of this study demonstrate the feasibility of eConsults to accurately diagnose nonurgent ophthalmic conditions. Of the patients recommended for an in-person evaluation with an ophthalmologist and subsequently followed up, concordance between the clinical assessment of the eConsultant and in-person ophthalmologist occurred in 93.1% of cases, with only two cases of missed diagnoses identified (central retinal vein occlusion and orbital fat prolapse as described in the results section). The first of the two cases of missed diagnosis deserves special attention. This case was a patient who presented two months after a motor vehicle accident that resulted in a concussion with light sensitivity and blurred vision. The patient’s symptoms were initially thought to be related to post-concussion syndrome; however, the eConsultant recommended the patient be evaluated to rule out other serious and vision-threatening etiologies. This patient was given an in-person appointment within three weeks of the eConsult response, during which this patient was found to have multiple peripheral retinal hemorrhages and elevated intraocular pressure and was referred for evaluation of a central retinal vein occlusion by the retina service. This scenario highlights the importance of referral for in-person evaluation to fully evaluate concerning clinical presentations and to rule out serious pathology. The growing practice of eConsultations will help build a body of evidence that can be leveraged to provide guidance in the future on such cases. The rate of diagnostic accuracy found in this study is similar to prior reports of high diagnostic accuracy when eConsults were used in other subspecialties such as dermatology. This is encouraging because access to ophthalmic care has become increasingly challenging given the growing shortage of ophthalmologists, persistent geographic barriers to ophthalmic care, and widening socioeconomic inequities in accessing specialty care. – Ophthalmology Referrals to Subspecialty Services The eConsults allow prompt ophthalmology evaluation to be achieved for patients presenting with ocular concerns in non-ophthalmic ambulatory care settings. Rates of completion for referrals to in-person specialty care vary tremendously from 30% to 80% across specialties, making eConsults an attractive alternative. , Among ophthalmology referrals for in-person evaluation specifically, reported rates of completion are similarly variable, ranging from 5% to 75% within the recommended referral period. , Patients also report that the cost of visits, insurance status, distance to clinics, lack of transportation, language barriers and limited health literacy, and work schedule conflicts lead to delayed or complete inaccessibility to ophthalmic care. , , The eConsults address many of these barriers through the elimination of time, cost, or travel to achieve an initial subspecialty evaluation. Additionally, eConsults create a direct line of communication among all members of the patient’s care team, delivering well-coordinated, high-quality patient-centered care. In the present study, 100% of submitted eConsults were completed, and 75% of cases had evidence that the ophthalmologist's recommendations were communicated to the patient. This reflects the suitability of eConsults as a mechanism to ensure patient access to ophthalmic care. Timing of eConsults and Subsequent Evaluation Importantly, eConsults have a demonstrated benefit of decreasing excessive wait times to reach specialty care. , , In the literature, the estimated median time from when a specialist referral is placed to the completion of an in-person visit is 7.5 to 8.7 weeks. , In our study, the average response time to complete an eConsult was 1.6 days. This is consistent with the results of other eConsult programs, including dermatology, allergy/immunology, endocrinology, and rheumatology, where eConsult were completed within three days. , – This is an especially notable benefit of eConsults because lengthy waiting periods are a common deterrent to seeking follow-up care, and the wait times for specialty appointments are steadily increasing. , The eConsults also improve referral quality, ensuring that patients requiring in-person evaluation are scheduled with the appropriate subspecialist. , In our study, 50% of subsequent referrals were made to ophthalmology subspecialties, including retina, pediatrics, neuro-ophthalmology, and oculoplastics. The eConsult program thus bypassed the standard practice of an initial in-person evaluation by a comprehensive ophthalmologist prior to receiving a subspecialist appointment. Of the many benefits that can be inferred from this, key among them is expedition of time to care, decreased unnecessary patient visits, and associated patient cost savings. The ability of eConsults to decrease unnecessary specialist referrals and potential economic advantages to the healthcare system merits further investigation in future studies. Advantages of Ophthalmology eConsults The eConsults also confer potential advantages across other care settings. An average of two million eye-related ED visits occur annually, with nearly half of those visits being for nonurgent conditions that can safely be managed in the outpatient setting. , EDs across the country are poorly positioned to provide optimal ophthalmic care because of inadequate or nonexistent ophthalmology service coverage and insufficient ophthalmic training for ED providers—limitations that are exacerbated in rural and underserved populations. , These limitations have also been associated with ED provider discomfort and inaccuracy evaluating ocular concerns. At the national level, hordeola are the second-most common cause of nonurgent ocular presentation to an ED. At the institutional level, 6% of all ophthalmic related presentations to the MEE ED (approximately 935 visits annually) from May 2019 to June 2021 had a primary diagnosis of hordeola and chalazia. In our study, hordeola and chalazia were the most common ophthalmic conditions inquired about in eConsults, accounting for 14% of the submitted eConsults. All eConsults related to hordeola and chalazia were managed conservatively with symptom improvement or resolution documented in patients’ charts, except in the case of a missed diagnosis described above. None of these eConsults were followed by an ED presentation, which suggests that nonurgent ocular conditions can safely and accurately be diagnosed and managed remotely and are a potential source of avoidable ED visits. Evidence suggests that the cost of managing nonurgent ophthalmic conditions in the ED can be anywhere from two to four times higher than in an ambulatory care setting. , Ophthalmic eConsults can help patients and healthcare systems avoid unnecessary costs for nonurgent, non-vision-threatening ophthalmic issues. Additionally, reduction in nonurgent ocular ED visits can help decrease overall ED crowding and allow resources to be directed toward patients with urgent ophthalmic and medical conditions. The eConsult offers educational advantages to providers and medical educators. – By communicating a plan of care back to the referring provider, eConsults create an opportunity for immediate feedback and education of the referring provider as it relates to ophthalmic issues. , This is in contrast to the traditional referral process where providers may not be aware of the specialist recommendation or plan of care for prolonged periods of time. Furthermore, providers will be able to apply information learned in the eConsult to similar cases in the future. Understanding the types of clinical problems and content areas providers most commonly ask questions about is necessary to better guide educational efforts. For example, this study suggests that further education on hordeolum/chalazion identification and management may be beneficial to physicians in ambulatory care settings. Limitations There are a few notable limitations in the use of eConsults in our study. First, the present study included only 100 eConsults over a 30-month period submitted by 74 individual providers, which suggests that many of the nearly 1400 primary care providers in the MGB system were unaware of or did not use the service. Understanding attitudes toward ophthalmic eConsult practices and seeking a broader sample size would be advantageous to inform widespread implementation and adoption of such programs. Second, the role of eConsults in decreasing or exacerbating disparities in ophthalmic care has yet to be established. An understanding of how eConsults can be used to decrease disparities in the provision of healthcare would be useful. Another important limitation is the retrospective nature of this study, because follow-up of patient outcomes was limited by chart review and documentation. Additionally, the percentage of patients who presented to an ophthalmologist after receiving a referral for an in-person evaluation was low, thus limiting our ability to assess the true rates of diagnostic concordance. Furthermore, all eConsults were answered by a single ophthalmologist at a single academic institution. Patients who sought subsequent eye care outside of this institution were unidentifiable in this study, because they would not be identifiable in the institutional EMR. Future research should evaluate the use of eConsults in various practice sites and locations, which would improve the generalizability of the findings. This study did not collect survey data on patient perspectives regarding confidence in, comfort with, or perceived benefits of eConsults. As a result, this study is unable to comment on patient perspectives of the ophthalmology eConsult program and the potential impact it could have on seeking care elsewhere or delaying care if symptoms progress. Future studies should consider a patient survey component. Last, this study focuses solely on eConsults and is therefore unable to compare the advantages and limitations of eConsults to other synchronous and asynchronous forms of telehealth. In this study, ophthalmic eConsults were associated with timely response back to the referring provider, and with a high rate of diagnostic accuracy among a subset of patients subsequently seen for an in-person ophthalmology visit. Our results support the use of eConsults as an effective telehealth modality to obtain timely diagnosis, access, and management of nonurgent eye conditions. Our study also demonstrates that eConsults enhance patient quality of care by soliciting specialist input, ensuring that timely ophthalmology evaluation is achieved, coordinating appropriate referral management and triage, and collaborating across interdisciplinary care teams. The results of this study demonstrate the feasibility of eConsults to accurately diagnose nonurgent ophthalmic conditions. Of the patients recommended for an in-person evaluation with an ophthalmologist and subsequently followed up, concordance between the clinical assessment of the eConsultant and in-person ophthalmologist occurred in 93.1% of cases, with only two cases of missed diagnoses identified (central retinal vein occlusion and orbital fat prolapse as described in the results section). The first of the two cases of missed diagnosis deserves special attention. This case was a patient who presented two months after a motor vehicle accident that resulted in a concussion with light sensitivity and blurred vision. The patient’s symptoms were initially thought to be related to post-concussion syndrome; however, the eConsultant recommended the patient be evaluated to rule out other serious and vision-threatening etiologies. This patient was given an in-person appointment within three weeks of the eConsult response, during which this patient was found to have multiple peripheral retinal hemorrhages and elevated intraocular pressure and was referred for evaluation of a central retinal vein occlusion by the retina service. This scenario highlights the importance of referral for in-person evaluation to fully evaluate concerning clinical presentations and to rule out serious pathology. The growing practice of eConsultations will help build a body of evidence that can be leveraged to provide guidance in the future on such cases. The rate of diagnostic accuracy found in this study is similar to prior reports of high diagnostic accuracy when eConsults were used in other subspecialties such as dermatology. This is encouraging because access to ophthalmic care has become increasingly challenging given the growing shortage of ophthalmologists, persistent geographic barriers to ophthalmic care, and widening socioeconomic inequities in accessing specialty care. – The eConsults allow prompt ophthalmology evaluation to be achieved for patients presenting with ocular concerns in non-ophthalmic ambulatory care settings. Rates of completion for referrals to in-person specialty care vary tremendously from 30% to 80% across specialties, making eConsults an attractive alternative. , Among ophthalmology referrals for in-person evaluation specifically, reported rates of completion are similarly variable, ranging from 5% to 75% within the recommended referral period. , Patients also report that the cost of visits, insurance status, distance to clinics, lack of transportation, language barriers and limited health literacy, and work schedule conflicts lead to delayed or complete inaccessibility to ophthalmic care. , , The eConsults address many of these barriers through the elimination of time, cost, or travel to achieve an initial subspecialty evaluation. Additionally, eConsults create a direct line of communication among all members of the patient’s care team, delivering well-coordinated, high-quality patient-centered care. In the present study, 100% of submitted eConsults were completed, and 75% of cases had evidence that the ophthalmologist's recommendations were communicated to the patient. This reflects the suitability of eConsults as a mechanism to ensure patient access to ophthalmic care. Importantly, eConsults have a demonstrated benefit of decreasing excessive wait times to reach specialty care. , , In the literature, the estimated median time from when a specialist referral is placed to the completion of an in-person visit is 7.5 to 8.7 weeks. , In our study, the average response time to complete an eConsult was 1.6 days. This is consistent with the results of other eConsult programs, including dermatology, allergy/immunology, endocrinology, and rheumatology, where eConsult were completed within three days. , – This is an especially notable benefit of eConsults because lengthy waiting periods are a common deterrent to seeking follow-up care, and the wait times for specialty appointments are steadily increasing. , The eConsults also improve referral quality, ensuring that patients requiring in-person evaluation are scheduled with the appropriate subspecialist. , In our study, 50% of subsequent referrals were made to ophthalmology subspecialties, including retina, pediatrics, neuro-ophthalmology, and oculoplastics. The eConsult program thus bypassed the standard practice of an initial in-person evaluation by a comprehensive ophthalmologist prior to receiving a subspecialist appointment. Of the many benefits that can be inferred from this, key among them is expedition of time to care, decreased unnecessary patient visits, and associated patient cost savings. The ability of eConsults to decrease unnecessary specialist referrals and potential economic advantages to the healthcare system merits further investigation in future studies. The eConsults also confer potential advantages across other care settings. An average of two million eye-related ED visits occur annually, with nearly half of those visits being for nonurgent conditions that can safely be managed in the outpatient setting. , EDs across the country are poorly positioned to provide optimal ophthalmic care because of inadequate or nonexistent ophthalmology service coverage and insufficient ophthalmic training for ED providers—limitations that are exacerbated in rural and underserved populations. , These limitations have also been associated with ED provider discomfort and inaccuracy evaluating ocular concerns. At the national level, hordeola are the second-most common cause of nonurgent ocular presentation to an ED. At the institutional level, 6% of all ophthalmic related presentations to the MEE ED (approximately 935 visits annually) from May 2019 to June 2021 had a primary diagnosis of hordeola and chalazia. In our study, hordeola and chalazia were the most common ophthalmic conditions inquired about in eConsults, accounting for 14% of the submitted eConsults. All eConsults related to hordeola and chalazia were managed conservatively with symptom improvement or resolution documented in patients’ charts, except in the case of a missed diagnosis described above. None of these eConsults were followed by an ED presentation, which suggests that nonurgent ocular conditions can safely and accurately be diagnosed and managed remotely and are a potential source of avoidable ED visits. Evidence suggests that the cost of managing nonurgent ophthalmic conditions in the ED can be anywhere from two to four times higher than in an ambulatory care setting. , Ophthalmic eConsults can help patients and healthcare systems avoid unnecessary costs for nonurgent, non-vision-threatening ophthalmic issues. Additionally, reduction in nonurgent ocular ED visits can help decrease overall ED crowding and allow resources to be directed toward patients with urgent ophthalmic and medical conditions. The eConsult offers educational advantages to providers and medical educators. – By communicating a plan of care back to the referring provider, eConsults create an opportunity for immediate feedback and education of the referring provider as it relates to ophthalmic issues. , This is in contrast to the traditional referral process where providers may not be aware of the specialist recommendation or plan of care for prolonged periods of time. Furthermore, providers will be able to apply information learned in the eConsult to similar cases in the future. Understanding the types of clinical problems and content areas providers most commonly ask questions about is necessary to better guide educational efforts. For example, this study suggests that further education on hordeolum/chalazion identification and management may be beneficial to physicians in ambulatory care settings. There are a few notable limitations in the use of eConsults in our study. First, the present study included only 100 eConsults over a 30-month period submitted by 74 individual providers, which suggests that many of the nearly 1400 primary care providers in the MGB system were unaware of or did not use the service. Understanding attitudes toward ophthalmic eConsult practices and seeking a broader sample size would be advantageous to inform widespread implementation and adoption of such programs. Second, the role of eConsults in decreasing or exacerbating disparities in ophthalmic care has yet to be established. An understanding of how eConsults can be used to decrease disparities in the provision of healthcare would be useful. Another important limitation is the retrospective nature of this study, because follow-up of patient outcomes was limited by chart review and documentation. Additionally, the percentage of patients who presented to an ophthalmologist after receiving a referral for an in-person evaluation was low, thus limiting our ability to assess the true rates of diagnostic concordance. Furthermore, all eConsults were answered by a single ophthalmologist at a single academic institution. Patients who sought subsequent eye care outside of this institution were unidentifiable in this study, because they would not be identifiable in the institutional EMR. Future research should evaluate the use of eConsults in various practice sites and locations, which would improve the generalizability of the findings. This study did not collect survey data on patient perspectives regarding confidence in, comfort with, or perceived benefits of eConsults. As a result, this study is unable to comment on patient perspectives of the ophthalmology eConsult program and the potential impact it could have on seeking care elsewhere or delaying care if symptoms progress. Future studies should consider a patient survey component. Last, this study focuses solely on eConsults and is therefore unable to compare the advantages and limitations of eConsults to other synchronous and asynchronous forms of telehealth. In this study, ophthalmic eConsults were associated with timely response back to the referring provider, and with a high rate of diagnostic accuracy among a subset of patients subsequently seen for an in-person ophthalmology visit. Our results support the use of eConsults as an effective telehealth modality to obtain timely diagnosis, access, and management of nonurgent eye conditions. Our study also demonstrates that eConsults enhance patient quality of care by soliciting specialist input, ensuring that timely ophthalmology evaluation is achieved, coordinating appropriate referral management and triage, and collaborating across interdisciplinary care teams. |
How to maintain equilibrium between the quantum and quality of cataract surgery training and patient safety measures | b6b9cc10-6837-46b0-8e9c-607451f07612 | 10841802 | Ophthalmology[mh] | |
Isolation of Actinobacteria from Date Palm Rhizosphere with Enzymatic, Antimicrobial, Antioxidant, and Protein Denaturation Inhibitory Activities | d8d79a92-e382-46ce-9bb4-8ab1edd05bfc | 11764267 | Microbiology[mh] | To meet the growing demand for new compounds in various industrial and agricultural sectors, as well as to counteract antibiotic-resistant pathogens, research and industries are exploring new microorganisms in diverse and understudied environments . Among the microbes of interest, actinobacteria are great producers of various biomolecules . Actinobacteria, also known as actinomycetes, are Gram-positive filamentous bacteria, sporigenous, with high G + C content in their DNA (around 70%) . Bioactive metabolites derived from actinobacteria account for approximately 70% of natural compounds currently used clinically . With these metabolites, antiviral, antifungal, antimalarial, antibacterial, immunosuppressive, antitumor, enzyme inhibitor, antioxidant, anti-inflammatory, and cytotoxic drugs have been described . Additionally, actinobacteria secrete a wide range of enzymes with significant industrial relevance . Amylase, for example, is used in the food, textile, and paper industries . Cellulase, on the other hand, is employed in the production of biofuels, textiles, paper pulp, and detergents, as well as in the food and animal feed industries . Lipase is essential in the detergent, food, and pharmaceutical industries . Lastly, proteases are widely used in the food, pharmaceutical, leather, detergent, and photography sectors . Actinobacteria are present in both terrestrial and aquatic habitats. This lineage is widely found in nutrient-rich rhizospheres, accounting for high percentages of microbial communities and where they produce various agricultural bioactive compounds with plant growth-promoting properties . Given their cellular resistance and wide-range metabolic adaptation, they can also colonize extreme environments , where they can biosynthesize a wide variety of novel natural bioactive compounds . Among the extreme environments, the Algerian Sahara is part of the world’s largest hot desert, occupying almost 90% of the country’s surface . Its climate is characterized by low and irregular precipitation, high temperatures, intense sunlight, and high evaporation . Only limited research has been focused on date palms’ rhizosphere actinobacteria in these desertic soils. Some studies have explored the biodiversity and antimicrobial activity of actinobacteria in Tamenrasset , as well as their distribution in the rhizosphere of date palms sensitive and resistant to Fusarium wilt in Adrar . Thus, the main objective of this research is to isolate bacteria belonging to the Actinobacteria phylum from the rhizosphere of date palms ( Phoenix dactylifera L.) in the Ghardïa desert of Algeria (an ecosystem that, to the best of our knowledge, has not been previously exploited) and to evaluate their potential to produce biologically active substances. Based on the rhizosphere containing organic matter from an arid climate, which has not been the subject of microbiological and biotechnological studies, we hypothesized that this environment is rich in microorganisms producing new bioactive compounds. To test this hypothesis, we collected rhizosphere samples from the desert date palm in Ghardaia, from which we highlighted the ability of culturable actinobacteria to produce enzymes and antibacterial and antifungal agents. Based on different tests, a single isolate was selected to extract its secondary metabolites and study their antioxidant and protein denaturation inhibitory activities. The molecular identification of this isolate was also carried out based on the phylogenetic analysis of 16S rDNA . 2.1. Area Study, Sample Collection, and Soil Physicochemical Characterization A rhizosphere sample from date palm ( Phoenix dactylifera L.) was obtained from the Noumérat region of Ghardaïa city in the Sahara Desert, located in north-central Algeria (32°29′27″ N; 3°40′24″ E). Sampling was carried out on three different points. Rhizosphere soil was collected by gently shaking loose soil from the roots and scraping firmly, adhering soil from the root surface into sterile containers . The collected samples were kept refrigerated. The samples for the isolation of actinobacteria were processed as soon as they were brought to the laboratory. The electrical conductivity, pH, organic matter, and moisture content of the soil samples were measured according to established methods . 2.2. Isolation of Cultivable Actinobacteria The soil was dried at room temperature for one week. Filamentous actinobacteria were isolated using the suspension-dilution method. One gram of soil was suspended in 9 mL of sterile physiological water (0.9%) and vigorously shaken using a vortex. A series of decimal dilutions was performed up to a dilution of 10 −5 . Inoculation was carried out by spreading 0.1 mL of each dilution onto the surface of two culture media, SCA (Starch Casein Agar) and ISP 2 (International Streptomyces Project 2) . To prevent the development of other bacteria and fungi, the culture media were supplemented with nalidixic acid (75 µg mL −1 ) and amphotericin B (25 µg mL −1 ). The plates were incubated at 30 °C and observed daily for a period of 7 to 21 days. The total number of colonies exhibiting the morphological characteristics of mycelial actinobacteria (presence of substrate mycelium and, very often, aerial mycelium) was expressed in CFU g −1 . These colonies were purified and preserved on the same isolation media in slant tubes at 4 °C for further studies. 2.3. Enzymatic Activities of Actinobacteria Various enzymatic tests on agar media were performed to detect the production of extracellular hydrolases by the actinobacteria isolates. The different tests are described as follows: Cellulolytic and amylolytic activities: Actionobacteria isolates were plated on ISP 2 agar medium containing 1% Carboxy Methyl Cellulose (CMC) and on nutrient agar supplemented with 1% soluble starch. After incubation at 30 °C for 7 days, the cultures were covered with Lugol solution. The presence of clear halo around the colonies indicated cellulase and amylase activity . Esterase (lipase) activity: Actinobacteria were grown on Sierra’s medium supplemented with tween 80 (1%). After 7 days of incubation at 30 °C, the appearance of an opaque halo around the colonies indicated esterase activity. Lecithinase and lipoproteinase: Isolates were streaked on 10% egg yolk agar and incubated at 30 °C for 7 days and two enzymes were tested: (1) lecithinase, with the appearance of an opaque, yellowish-white pearly halo with a clear edge under the colony or at its limit; (2) lipoproteinase, with the appearance of a clear zone around the culture . Caseinase: Protease activity was detected by inoculating actinobacteria onto milk nutrient agar containing 5% skimmed milk. After incubation for 7 days at 30 °C, a clear halo around the colonies indicated caseinase activity . Gelatinase: Actinobacteria were inoculated into tubes containing nutrient gelatin and incubated at 30 °C for 21 days. The tubes were then refrigerated for one hour. Solidification of the gelatin indicated that it had not been hydrolyzed, while liquid gelatin indicated the presence of gelatinase . Peptonization and coagulation of milk: Isolates were inoculated into tubes containing sterile skimmed milk and incubated at 30 °C. Regular observations over 14 days were performed to record the coagulation and peptonization of milk induced by each isolate . Catalase: A colony of each isolate was placed on a slide containing a drop of 10% hydrogen peroxide. The presence of catalase was indicated by the appearance of air bubbles due to the release of O 2 gas . 2.4. Antimicrobial Activity The antimicrobial activity of actinobacteria was tested against a panel of pathogenic microorganisms, including three Gram-positive ( Enterococcus faecalis ATCC 29212, Bacillus spizizenii ATCC 6633, and Staphylococcus aureus ATCC 25923) and four Gram-negative bacteria ( Pseudomonas aeruginosa ATCC 27853, Escherichia coli ATCC 25922, Klebsiella pneumoniae ATCC 700603, and Salmonella typhimurium ATCC 14028), one yeast ( Candida albicans ATCC 10231), and four filamentous fungi ( Fusarium oxysporum CIP 62572, Aspergillus niger MH109542, Penicillium sp., and Alternaria sp.). Suspensions of these microorganisms were prepared according to the protocols of Bramki et al. and Nemouchi et al. . Actinobacteria strains were inoculated on ISP 2 medium and incubated for 14 days. Agar cylinders (6 mm diameter) from well-developed and well-sporulated cultures were cut out and placed on the culture media (Mueller Hinton for bacteria and Sabouraud for fungi) previously inoculated with the test germs. Petri dishes were kept at 4 °C for 4 h to enable proper diffusion of bioactive metabolites . Then, inhibition diameters were measured after 24 to 48 h of incubation at 37 °C for bacteria and yeast and for 48 to 72 h at 28 °C for filamentous fungi. 2.5. Selected Isolate Bioactive Molecules Extraction Well-developed cultures of the SGI16 isolate (10 days old) were fragmented into small pieces and completely covered with ethyl acetate. Maceration was carried out twice in order to recover the maximum of produced bioactive molecules. The crude extracts were filtered through Whatman N° 01 paper (11 µm) and then concentrated to dryness using a rotary evaporator (Heidolph, Germany) . 2.5.1. Evaluation of Antioxidant Activity The antioxidant effect of the SGI16 isolate extract was assessed by measuring the DPPH (1,1-diphenyl-2-picryhydrazyl, Sigma-Aldrich, St. Louis, MO, USA) (radical scavenging capacity. A volume of 2 mL of each extract concentration was mixed with 1.6 mL of 0.002% methanolic DPPH solution. The mixture was incubated at room temperature in the dark for 30 min. Ascorbic acid was used as a positive control and the absorbance was measured against a blank at a wavelength of 517 nm. The percentage DPPH free radical inhibition was calculated according to the equation: D P P H s c a v e n g i n g e f f e c t % = A b s o r b a n c e C o n t r o l − A b s o r b a n c e s a m p l e A b s o r b a n c e C o n t r o l × 100 The results were expressed as IC 50 values (μg mL −1 ), representing the concentration required to achieve 50% inhibition . 2.5.2. Evaluation of Protein Denaturation Inhibitory Capacity The ability of the extract to inhibit protein denaturation was determined using the method of Kar et al. with slight modification. In a 96-well microplate, 100 µL of extract was added to 100 µL of 0.2% BSA (Bovine Serum Albumin, Sigma-Aldrich, St. Louis, MO, USA), and the mixture was then incubated at 72 °C for 20 min. After cooling, turbidity was measured at 660 nm against a blank prepared from 100 µL of Tris-HCl buffer (0.05 M, pH 6.6) and 100 µL of the SGI16 isolate extract. A negative control was prepared with 100 µL of BSA and 100 µL of ultrapure water, while diclofenac was used as a positive control. The inhibition percentage was calculated using the following formula: % I n h i b i t i o n = A b s o r b a n c e C o n t r o l − A b s o r b a n c e s a m p l e A b s o r b a n c e C o n t r o l × 100 2.5.3. Molecular Identification and Phylogenetic Analysis of the SGI16 Isolate The 16S rRNA gene barcoding was performed on the most promising isolate, SGI16, by The Environmental Sciences Section, Department MeSVA, University of L’Aquila and BMR Genomics, Padua, Italy. DNA was amplified by direct PCR using universal bacterial primers (27F/1492R) and then sequenced. The obtained sequence was analyzed by Finch TV software version 1.4.0 (Geospiza, Inc.; Seattle, WA, USA; https://www.softpedia.com/get/Science-CAD/FinchTV.shtml , accessed on 1 September 2024) and compared with those present in the EZBioCloud database ( https://www.ezbiocloud.net/ , accessed on 2 September 2024). The sequences were aligned using the Clustal X 2.0.12 program . A rooted phylogenetic tree was constructed with MEGA version 11 . The distance matrix was computed using the neighbor-joining method , which is based on the Kimura two-parameter model . Tree topology was assessed using bootstrap analysis with 1000 replicates . This analysis involved 12 nucleotide sequences. All positions containing gaps and missing data were eliminated (complete deletion option). There were a total of 1172 positions in the final dataset. The SGI16 sequence is available in GenBank under accession number PQ240116. 2.6. Statistical Analysis All analyses were carried out in triplicate and the experimental data were reported as means ± standard deviation. Statistical analysis was carried out using XLSTAT software 2014.5.03 (Addinsoft, New York, NY, USA). Significant differences were determined at p ≤ 0.05 by one-way analysis of variance (ANOVA) followed by Tukey’s HSD post hoc test. A rhizosphere sample from date palm ( Phoenix dactylifera L.) was obtained from the Noumérat region of Ghardaïa city in the Sahara Desert, located in north-central Algeria (32°29′27″ N; 3°40′24″ E). Sampling was carried out on three different points. Rhizosphere soil was collected by gently shaking loose soil from the roots and scraping firmly, adhering soil from the root surface into sterile containers . The collected samples were kept refrigerated. The samples for the isolation of actinobacteria were processed as soon as they were brought to the laboratory. The electrical conductivity, pH, organic matter, and moisture content of the soil samples were measured according to established methods . The soil was dried at room temperature for one week. Filamentous actinobacteria were isolated using the suspension-dilution method. One gram of soil was suspended in 9 mL of sterile physiological water (0.9%) and vigorously shaken using a vortex. A series of decimal dilutions was performed up to a dilution of 10 −5 . Inoculation was carried out by spreading 0.1 mL of each dilution onto the surface of two culture media, SCA (Starch Casein Agar) and ISP 2 (International Streptomyces Project 2) . To prevent the development of other bacteria and fungi, the culture media were supplemented with nalidixic acid (75 µg mL −1 ) and amphotericin B (25 µg mL −1 ). The plates were incubated at 30 °C and observed daily for a period of 7 to 21 days. The total number of colonies exhibiting the morphological characteristics of mycelial actinobacteria (presence of substrate mycelium and, very often, aerial mycelium) was expressed in CFU g −1 . These colonies were purified and preserved on the same isolation media in slant tubes at 4 °C for further studies. Various enzymatic tests on agar media were performed to detect the production of extracellular hydrolases by the actinobacteria isolates. The different tests are described as follows: Cellulolytic and amylolytic activities: Actionobacteria isolates were plated on ISP 2 agar medium containing 1% Carboxy Methyl Cellulose (CMC) and on nutrient agar supplemented with 1% soluble starch. After incubation at 30 °C for 7 days, the cultures were covered with Lugol solution. The presence of clear halo around the colonies indicated cellulase and amylase activity . Esterase (lipase) activity: Actinobacteria were grown on Sierra’s medium supplemented with tween 80 (1%). After 7 days of incubation at 30 °C, the appearance of an opaque halo around the colonies indicated esterase activity. Lecithinase and lipoproteinase: Isolates were streaked on 10% egg yolk agar and incubated at 30 °C for 7 days and two enzymes were tested: (1) lecithinase, with the appearance of an opaque, yellowish-white pearly halo with a clear edge under the colony or at its limit; (2) lipoproteinase, with the appearance of a clear zone around the culture . Caseinase: Protease activity was detected by inoculating actinobacteria onto milk nutrient agar containing 5% skimmed milk. After incubation for 7 days at 30 °C, a clear halo around the colonies indicated caseinase activity . Gelatinase: Actinobacteria were inoculated into tubes containing nutrient gelatin and incubated at 30 °C for 21 days. The tubes were then refrigerated for one hour. Solidification of the gelatin indicated that it had not been hydrolyzed, while liquid gelatin indicated the presence of gelatinase . Peptonization and coagulation of milk: Isolates were inoculated into tubes containing sterile skimmed milk and incubated at 30 °C. Regular observations over 14 days were performed to record the coagulation and peptonization of milk induced by each isolate . Catalase: A colony of each isolate was placed on a slide containing a drop of 10% hydrogen peroxide. The presence of catalase was indicated by the appearance of air bubbles due to the release of O 2 gas . The antimicrobial activity of actinobacteria was tested against a panel of pathogenic microorganisms, including three Gram-positive ( Enterococcus faecalis ATCC 29212, Bacillus spizizenii ATCC 6633, and Staphylococcus aureus ATCC 25923) and four Gram-negative bacteria ( Pseudomonas aeruginosa ATCC 27853, Escherichia coli ATCC 25922, Klebsiella pneumoniae ATCC 700603, and Salmonella typhimurium ATCC 14028), one yeast ( Candida albicans ATCC 10231), and four filamentous fungi ( Fusarium oxysporum CIP 62572, Aspergillus niger MH109542, Penicillium sp., and Alternaria sp.). Suspensions of these microorganisms were prepared according to the protocols of Bramki et al. and Nemouchi et al. . Actinobacteria strains were inoculated on ISP 2 medium and incubated for 14 days. Agar cylinders (6 mm diameter) from well-developed and well-sporulated cultures were cut out and placed on the culture media (Mueller Hinton for bacteria and Sabouraud for fungi) previously inoculated with the test germs. Petri dishes were kept at 4 °C for 4 h to enable proper diffusion of bioactive metabolites . Then, inhibition diameters were measured after 24 to 48 h of incubation at 37 °C for bacteria and yeast and for 48 to 72 h at 28 °C for filamentous fungi. Well-developed cultures of the SGI16 isolate (10 days old) were fragmented into small pieces and completely covered with ethyl acetate. Maceration was carried out twice in order to recover the maximum of produced bioactive molecules. The crude extracts were filtered through Whatman N° 01 paper (11 µm) and then concentrated to dryness using a rotary evaporator (Heidolph, Germany) . 2.5.1. Evaluation of Antioxidant Activity The antioxidant effect of the SGI16 isolate extract was assessed by measuring the DPPH (1,1-diphenyl-2-picryhydrazyl, Sigma-Aldrich, St. Louis, MO, USA) (radical scavenging capacity. A volume of 2 mL of each extract concentration was mixed with 1.6 mL of 0.002% methanolic DPPH solution. The mixture was incubated at room temperature in the dark for 30 min. Ascorbic acid was used as a positive control and the absorbance was measured against a blank at a wavelength of 517 nm. The percentage DPPH free radical inhibition was calculated according to the equation: D P P H s c a v e n g i n g e f f e c t % = A b s o r b a n c e C o n t r o l − A b s o r b a n c e s a m p l e A b s o r b a n c e C o n t r o l × 100 The results were expressed as IC 50 values (μg mL −1 ), representing the concentration required to achieve 50% inhibition . 2.5.2. Evaluation of Protein Denaturation Inhibitory Capacity The ability of the extract to inhibit protein denaturation was determined using the method of Kar et al. with slight modification. In a 96-well microplate, 100 µL of extract was added to 100 µL of 0.2% BSA (Bovine Serum Albumin, Sigma-Aldrich, St. Louis, MO, USA), and the mixture was then incubated at 72 °C for 20 min. After cooling, turbidity was measured at 660 nm against a blank prepared from 100 µL of Tris-HCl buffer (0.05 M, pH 6.6) and 100 µL of the SGI16 isolate extract. A negative control was prepared with 100 µL of BSA and 100 µL of ultrapure water, while diclofenac was used as a positive control. The inhibition percentage was calculated using the following formula: % I n h i b i t i o n = A b s o r b a n c e C o n t r o l − A b s o r b a n c e s a m p l e A b s o r b a n c e C o n t r o l × 100 2.5.3. Molecular Identification and Phylogenetic Analysis of the SGI16 Isolate The 16S rRNA gene barcoding was performed on the most promising isolate, SGI16, by The Environmental Sciences Section, Department MeSVA, University of L’Aquila and BMR Genomics, Padua, Italy. DNA was amplified by direct PCR using universal bacterial primers (27F/1492R) and then sequenced. The obtained sequence was analyzed by Finch TV software version 1.4.0 (Geospiza, Inc.; Seattle, WA, USA; https://www.softpedia.com/get/Science-CAD/FinchTV.shtml , accessed on 1 September 2024) and compared with those present in the EZBioCloud database ( https://www.ezbiocloud.net/ , accessed on 2 September 2024). The sequences were aligned using the Clustal X 2.0.12 program . A rooted phylogenetic tree was constructed with MEGA version 11 . The distance matrix was computed using the neighbor-joining method , which is based on the Kimura two-parameter model . Tree topology was assessed using bootstrap analysis with 1000 replicates . This analysis involved 12 nucleotide sequences. All positions containing gaps and missing data were eliminated (complete deletion option). There were a total of 1172 positions in the final dataset. The SGI16 sequence is available in GenBank under accession number PQ240116. The antioxidant effect of the SGI16 isolate extract was assessed by measuring the DPPH (1,1-diphenyl-2-picryhydrazyl, Sigma-Aldrich, St. Louis, MO, USA) (radical scavenging capacity. A volume of 2 mL of each extract concentration was mixed with 1.6 mL of 0.002% methanolic DPPH solution. The mixture was incubated at room temperature in the dark for 30 min. Ascorbic acid was used as a positive control and the absorbance was measured against a blank at a wavelength of 517 nm. The percentage DPPH free radical inhibition was calculated according to the equation: D P P H s c a v e n g i n g e f f e c t % = A b s o r b a n c e C o n t r o l − A b s o r b a n c e s a m p l e A b s o r b a n c e C o n t r o l × 100 The results were expressed as IC 50 values (μg mL −1 ), representing the concentration required to achieve 50% inhibition . The ability of the extract to inhibit protein denaturation was determined using the method of Kar et al. with slight modification. In a 96-well microplate, 100 µL of extract was added to 100 µL of 0.2% BSA (Bovine Serum Albumin, Sigma-Aldrich, St. Louis, MO, USA), and the mixture was then incubated at 72 °C for 20 min. After cooling, turbidity was measured at 660 nm against a blank prepared from 100 µL of Tris-HCl buffer (0.05 M, pH 6.6) and 100 µL of the SGI16 isolate extract. A negative control was prepared with 100 µL of BSA and 100 µL of ultrapure water, while diclofenac was used as a positive control. The inhibition percentage was calculated using the following formula: % I n h i b i t i o n = A b s o r b a n c e C o n t r o l − A b s o r b a n c e s a m p l e A b s o r b a n c e C o n t r o l × 100 The 16S rRNA gene barcoding was performed on the most promising isolate, SGI16, by The Environmental Sciences Section, Department MeSVA, University of L’Aquila and BMR Genomics, Padua, Italy. DNA was amplified by direct PCR using universal bacterial primers (27F/1492R) and then sequenced. The obtained sequence was analyzed by Finch TV software version 1.4.0 (Geospiza, Inc.; Seattle, WA, USA; https://www.softpedia.com/get/Science-CAD/FinchTV.shtml , accessed on 1 September 2024) and compared with those present in the EZBioCloud database ( https://www.ezbiocloud.net/ , accessed on 2 September 2024). The sequences were aligned using the Clustal X 2.0.12 program . A rooted phylogenetic tree was constructed with MEGA version 11 . The distance matrix was computed using the neighbor-joining method , which is based on the Kimura two-parameter model . Tree topology was assessed using bootstrap analysis with 1000 replicates . This analysis involved 12 nucleotide sequences. All positions containing gaps and missing data were eliminated (complete deletion option). There were a total of 1172 positions in the final dataset. The SGI16 sequence is available in GenBank under accession number PQ240116. All analyses were carried out in triplicate and the experimental data were reported as means ± standard deviation. Statistical analysis was carried out using XLSTAT software 2014.5.03 (Addinsoft, New York, NY, USA). Significant differences were determined at p ≤ 0.05 by one-way analysis of variance (ANOVA) followed by Tukey’s HSD post hoc test. 3.1. Soil Physicochemical Characterization and Isolation of Actinobacteria The physicochemical characteristics of the rhizosphere are listed in . The filamentous colonies of actinobacteria ranging in size from 1 to 10 mm, with a powdery, rough, rugged, colored, or uncolored appearance, and often embedded in the agar, with or without aerial mycelium, were counted. According to the obtained results, the total number of actinomycetes isolated from the sample on the SCA medium (7.47 × 10 5 ± 1.02 CFU g −1 ) was significantly higher than that obtained on the ISP 2 medium (3.15 × 10 5 ± 0.26 CFU g −1 ). Therefore, one gram of rhizosphere contains approximately 7.47 × 10 5 filamentous actinobacteria. After purification, 18 morphologically different actinomycete isolates were selected. The isolates were coded based to their origin and the isolation media as follows: SGA4, SGA6, SGA7, SGA8, SGA9, SGA10, SGI2, SGI5, SGI6, SGI7, SGI8, SGI13, SGI16, SGI18, SGI19, SGI22, SGI24, and SGI25. 3.2. Enzymatic Activities The different enzymatic activities of the obtained actinomycetes isolates are presented in . All the obtained isolates showed positive cellulolytic activity. Most of the isolates (94%) exhibited amylase production and possessed enzymes capable of degrading lipids, such as esterase (83%), lecithinase, and lipoproteinase (78%). Regarding gelatin hydrolysis, 72% of actinobacteria were positive. Slightly over half (56%) of these bacteria showed peptonization of skimmed milk, while the remaining isolates (28%) caused coagulation. 3.3. Antimicrobial Activity Most actinomycetes showed broad-spectrum antimicrobial activities against different pathogenic microorganisms. The positive results of the antibacterial activities are reported in . Thirteen actinomycete isolates (72.22%) of the eighteen studied produced metabolites with antibacterial activity against at least one of the seven test bacteria. Significant antibacterial activity was observed against B. spizizenii (28 ± 1.00 mm), followed by S. aureus (19 ± 1.00 mm) and E. faecalis (17 ± 1.00 mm). The isolates SGI13, SGI16, and SGA9 exhibited significant activity against S. typhimurium , with inhibition zones ranging from 10 to 15 mm. In addition, isolates SGI13, SGI16, SGA10, SGA7, and SGA8 demonstrated significant activity against K. pneumoniae , with inhibition zones ranging from 8 to 13 mm. Inhibition zones of 11 mm and 12 mm were observed for isolates SGI16 and SGA9 against E. coli . In contrast, no activity was detected against P. aeruginosa . The positive results of the antifungal activities are reported in . Ten out of eighteen actinobacterial isolates (55.55%) exhibited antifungal activity against at least one of the five tested phytopathogenic or clinical strains. Among them, only one isolate (SGI16) inhibited A. niger , while two others (SGI16 and SGA9) were active against F. oxysporium. Ten isolates (55.55%) showed significant activity against Penicillium sp., and three isolates (SGI1, SGA9, and SGI19) were effective against Alternaria sp. Only one actinobacterium (SGA9) was significantly active against C. albicans . The maximum inhibition zones of 30.33 ± 0.57 and 31 ± 1.00 mm in diameter were produced by isolates SGA6 and SGI16 against Penicillium sp. and Alternaria sp., respectively. Furthermore, isolate SGI16 showed significant activity against F. oxysporium , with an inhibition zone of 25 ± 1.00 mm, compared to other isolates. 3.4. Selected Isolate’s Bioactive Molecule Extraction The isolate SGI16, identified as the best producer of enzymes and antimicrobial agents, was selected to study its antioxidant and protein denaturation inhibitory activities. Indeed, in addition to its ability to produce seven enzymes (cellulase, amylase, catalase, esterase, lipoproteinase, caseinase, and gelatinase), it also exhibits activity against Gram-positive bacteria such as S. aureus and E. faecalis , as well as the Gram-negative S. typhimurium , E. coli , and K. pneumoniae , with inhibition zone diameters ranging from 10.66 ± 0.57 mm to 13.33 ± 0.57 mm. Moreover, it effectively acts against all tested molds, including A. niger , F. oxysporum , Penicillium sp., and Alternaria sp., with inhibition zone diameters ranging from 12 ± 0.00 mm to 31 ± 1.00 mm. This isolate was cultivated on a solid medium to extract secondary metabolites. The biological activity of the isolate is presented in . The antioxidant activity of the SGI16 crude extract was evaluated using the DPPH test. The results showed that this extract exhibited significant antioxidant power. Indeed, ascorbic acid demonstrated an IC 50 value of 2.68 ± 0.01 μg/mL, which is close to that of the SGI16 isolate extract, which was 7.24 ± 0.21 μg/mL. The inhibitory activity of the SGI16 extract on BSA denaturation was assessed as a preliminary indicator of its biological activity. The results indicate that both the SGI16 extract and diclofenac inhibited BSA denaturation. Although the inhibitory effect of diclofenac was stronger (IC 50 = 128.83 ± 0.08 μg mL −1 ) compared to the tested extract (IC 50 = 492.41 ± 0.47 μg mL −1 ), the extract still showed significant protein denaturation inhibitory activity. 3.5. Molecular Characterization of Selected Isolate The characterization of the actinobacterium SGI16 using 16S rDNA barcoding identified it as being closely related to the species Streptomyces yangpuensis, with the highest similarity percentage of 99.08%, and Streptomyces flavotricini and Streptomyces amritsarensis, with the same similarity percentage of 98.99. However, in the phylogenetic tree shown in , the affiliation of SGI16 is not clearly defined in relation to the closest species: S. yangpuensis and S. amritsarensis . Therefore, in order to clarify the taxonomic position of the strain SGI16 at the species level, additional tests such as DNA-DNA hybridization and whole genome sequencing should be performed. The physicochemical characteristics of the rhizosphere are listed in . The filamentous colonies of actinobacteria ranging in size from 1 to 10 mm, with a powdery, rough, rugged, colored, or uncolored appearance, and often embedded in the agar, with or without aerial mycelium, were counted. According to the obtained results, the total number of actinomycetes isolated from the sample on the SCA medium (7.47 × 10 5 ± 1.02 CFU g −1 ) was significantly higher than that obtained on the ISP 2 medium (3.15 × 10 5 ± 0.26 CFU g −1 ). Therefore, one gram of rhizosphere contains approximately 7.47 × 10 5 filamentous actinobacteria. After purification, 18 morphologically different actinomycete isolates were selected. The isolates were coded based to their origin and the isolation media as follows: SGA4, SGA6, SGA7, SGA8, SGA9, SGA10, SGI2, SGI5, SGI6, SGI7, SGI8, SGI13, SGI16, SGI18, SGI19, SGI22, SGI24, and SGI25. The different enzymatic activities of the obtained actinomycetes isolates are presented in . All the obtained isolates showed positive cellulolytic activity. Most of the isolates (94%) exhibited amylase production and possessed enzymes capable of degrading lipids, such as esterase (83%), lecithinase, and lipoproteinase (78%). Regarding gelatin hydrolysis, 72% of actinobacteria were positive. Slightly over half (56%) of these bacteria showed peptonization of skimmed milk, while the remaining isolates (28%) caused coagulation. Most actinomycetes showed broad-spectrum antimicrobial activities against different pathogenic microorganisms. The positive results of the antibacterial activities are reported in . Thirteen actinomycete isolates (72.22%) of the eighteen studied produced metabolites with antibacterial activity against at least one of the seven test bacteria. Significant antibacterial activity was observed against B. spizizenii (28 ± 1.00 mm), followed by S. aureus (19 ± 1.00 mm) and E. faecalis (17 ± 1.00 mm). The isolates SGI13, SGI16, and SGA9 exhibited significant activity against S. typhimurium , with inhibition zones ranging from 10 to 15 mm. In addition, isolates SGI13, SGI16, SGA10, SGA7, and SGA8 demonstrated significant activity against K. pneumoniae , with inhibition zones ranging from 8 to 13 mm. Inhibition zones of 11 mm and 12 mm were observed for isolates SGI16 and SGA9 against E. coli . In contrast, no activity was detected against P. aeruginosa . The positive results of the antifungal activities are reported in . Ten out of eighteen actinobacterial isolates (55.55%) exhibited antifungal activity against at least one of the five tested phytopathogenic or clinical strains. Among them, only one isolate (SGI16) inhibited A. niger , while two others (SGI16 and SGA9) were active against F. oxysporium. Ten isolates (55.55%) showed significant activity against Penicillium sp., and three isolates (SGI1, SGA9, and SGI19) were effective against Alternaria sp. Only one actinobacterium (SGA9) was significantly active against C. albicans . The maximum inhibition zones of 30.33 ± 0.57 and 31 ± 1.00 mm in diameter were produced by isolates SGA6 and SGI16 against Penicillium sp. and Alternaria sp., respectively. Furthermore, isolate SGI16 showed significant activity against F. oxysporium , with an inhibition zone of 25 ± 1.00 mm, compared to other isolates. The isolate SGI16, identified as the best producer of enzymes and antimicrobial agents, was selected to study its antioxidant and protein denaturation inhibitory activities. Indeed, in addition to its ability to produce seven enzymes (cellulase, amylase, catalase, esterase, lipoproteinase, caseinase, and gelatinase), it also exhibits activity against Gram-positive bacteria such as S. aureus and E. faecalis , as well as the Gram-negative S. typhimurium , E. coli , and K. pneumoniae , with inhibition zone diameters ranging from 10.66 ± 0.57 mm to 13.33 ± 0.57 mm. Moreover, it effectively acts against all tested molds, including A. niger , F. oxysporum , Penicillium sp., and Alternaria sp., with inhibition zone diameters ranging from 12 ± 0.00 mm to 31 ± 1.00 mm. This isolate was cultivated on a solid medium to extract secondary metabolites. The biological activity of the isolate is presented in . The antioxidant activity of the SGI16 crude extract was evaluated using the DPPH test. The results showed that this extract exhibited significant antioxidant power. Indeed, ascorbic acid demonstrated an IC 50 value of 2.68 ± 0.01 μg/mL, which is close to that of the SGI16 isolate extract, which was 7.24 ± 0.21 μg/mL. The inhibitory activity of the SGI16 extract on BSA denaturation was assessed as a preliminary indicator of its biological activity. The results indicate that both the SGI16 extract and diclofenac inhibited BSA denaturation. Although the inhibitory effect of diclofenac was stronger (IC 50 = 128.83 ± 0.08 μg mL −1 ) compared to the tested extract (IC 50 = 492.41 ± 0.47 μg mL −1 ), the extract still showed significant protein denaturation inhibitory activity. The characterization of the actinobacterium SGI16 using 16S rDNA barcoding identified it as being closely related to the species Streptomyces yangpuensis, with the highest similarity percentage of 99.08%, and Streptomyces flavotricini and Streptomyces amritsarensis, with the same similarity percentage of 98.99. However, in the phylogenetic tree shown in , the affiliation of SGI16 is not clearly defined in relation to the closest species: S. yangpuensis and S. amritsarensis . Therefore, in order to clarify the taxonomic position of the strain SGI16 at the species level, additional tests such as DNA-DNA hybridization and whole genome sequencing should be performed. Arid and semi-arid regions, characterized by high evaporation rates, tend to accumulate salts in the soil as water evaporates, leaving the dissolved salts behind . Based on the criteria established by the Soil Science Division Staff, the soil in this study is classified as alkaline (pH = 8.07) . This value is close to that reported by previous studies for desert soils . Referring to the salinity scale, which is directly related to electrical conductivity (E.C) as defined by Richards, the studied rhizosphere is not slightly saline (EC = 0.73 dS/m) . Indeed, the scarcity of rainfall is one of the main factors contributing to salinization. During rainfall, water dissolves the salts in the soil and transports them deeper into the ground. Meanwhile, arid and semi-arid regions are characterized by high evaporation rates. When water evaporates from the soil, the salts dissolved in the water remain in the soil . According to the classification by Lee and Hwang, this soil is characterized by low moisture (5.61%) and organic matter content (6.52%) . The rhizosphere is a preferred environment for developing microorganisms due to the availability of nutrients and organic matter from root exudates . In addition, soil characteristics such as pH, salinity, soil water content, soil fertility, etc., are the main factors affecting the distribution and composition of soil microbial communities . The higher counting obtained for SCA medium than ISP 2 can be attributed to the SCA’s higher concentration of carbon-rich (starch) and nitrogen-rich (casein) substrates as well as the presence of CaCO 3 , which promotes sporulation and consequently increases the number of sporulating actinobacteria. This composition supports the growth of actinobacteria and facilitates their preferential isolation over other bacteria . These researchers confirmed the effectiveness of SCA medium for the selective isolation of actinomycetes from various ecosystems. Moreover, the rhizosphere in this study exhibited a higher abundance of actinomycetes compared to several plant soils in Korea, where the number ranged from 1.17 × 10 6 to 4.20 × 10 6 CFU g −1 of dry soil . Regarding the enzymatic activities, our results are in line with similar studies. Ranjani et al. reported that several actinomycetes, such as Streptomyces rubber , Thermobifida halotolerant, and Thermomonospora sp., are potent sources of cellulase enzyme . In contrast, Mansour et al. found that none of the actinobacteria they isolated from soil were capable of producing this enzyme . Janatiningrum and Lestari reported that all actinobacteria isolates from the rhizosphere soil of a medicinal Fiscus deltordea Jacq. produced amylase . Given that actinobacteria are aerobic bacteria, catalase activity is expected . Catalase released by rhizobacteria can mitigate the effects of reactive oxygen species (ROS) under abiotic stress conditions in plants . Moreover, most isolates from two types of environments (sebkha and palm) in the Algerian Sahara produce lipid-degrading enzymes . Actinobacteria from various ecosystems (rhizosphere, compost, and soils), especially Streptomyces , are known to secrete multiple proteases, such as caseinase, gelatinase, peptonization, and milk coagulation . Additionally, actinobacterial enzymes are important biocatalysts with various applications in industries, including pharmaceuticals, food, paper, textiles, biorefineries, and detergents . The antimicrobial activity that the isolates showed can be attributed to bioactive molecules that inhibit the growth of various test bacteria, mainly through the lysis of their cell walls . Gram-positive tested bacteria were more sensitive to the antibacterial substances produced by the actinobacteria than the Gram-negative ones. This sensitivity has already been noted in similar studies . This is often due to their cell membranes’ structural and compositional differences. Indeed, the Gram-negative membrane is composed of phospholipids and glycoproteins, making it much less permeable to lipophilic solutes like antibiotics. However, the Gram-positive membrane consists of an outer peptidoglycan layer that facilitates the penetration of molecules . Recent studies have also reported that actinobacteria from the rhizosphere of various other plants, such as olive, Junierus excelsa tree, and bamboo, have been reported for their good antibacterial activity against S. aureus , E. faecalis , and B. subtilis . The actinomycetes, particularly Streptomyces sp., remain the microbes offering the greatest economic and biotechnological benefits, producing the majority of antibiotics used in medicine . Actinomycetes isolated from the rhizosphere of several plants have already shown broad-spectrum antifungal activity against A. niger , Penicillium notatum , F. oxysporum , Alternaria alternata , and C. alibicans . As previously described, the lytic enzymes produced by most of our actinobacteria, such as cellulase, amylase, lipases, and proteases, exert a direct inhibitory effect on fungal pathogens by degrading their cell walls . For many actinomycetes, the production of bioactive molecules is generally more abundant and of better quality when carried out on a solid medium rather than in a liquid one. There are even microorganisms that lose their production capacity when grown in a liquid medium . This difference is attributed to the growth of Streptomyces in liquid media, where their hyphae fragment, thereby reducing their ability to produce several bioactive molecules . For the antioxidant activity, the IC 50 values obtained for SG16 were lower than those found in the literature, indicating that a low amount of the extract is necessary to inhibit the radical (i.e., higher antioxidant activity). For example, Streptomyces antioxidans MUSC 164T extracts with a concentration of 2000 µg mL −1 showed a DPPH free radical scavenging activity of 18% (which corresponds approximatively to an IC 50 of 5555 µg mL −1 ) . For Streptomyces sp. V002, the DPPH IC 50 value was 834 µg mL −1 . The protein denaturation inhibition results are also consistent with previous findings. Hegazy et al. demonstrated that the pigment extract of Streptomyces tunisiensis W4MT573222 effectively inhibited protein denaturation. Their results showed that the extract prevented BSA denaturation, with the inhibitory effect increasing as the pigment concentration increased. At a concentration of 200 µg/mL, the extract achieved an inhibition percentage of 85.9%. Molecular analysis showed that the SG16 strain belongs to the Streptomyces genus. This taxon is commonly found abundant in the rhizosphere of various plants , notably in the rhizosphere of several date palm cultivars from the Algerian Sahara and Saudi Arabia . Streptomyces occupies a prominent position in the field of biotechnology, owing to its rich secondary metabolism, which enables the production of a multitude of bioactive compounds, including several clinically relevant drugs . Indeed, this genus is a prolific source of secondary metabolites such as antibiotics (e.g., streptomycin, gentamicin, tetracycline, chloramphenicol, and erythromycin) , antifungals (e.g., piericidin-A1 and nigericin) , anticancer agents (e.g., doxorubicin and bleomycin) , immunosuppressants (e.g., rapamycin) , and antivirals (e.g., virantmycin B1) . More than 74% of the antibiotics currently available have been synthesized by this genus . The biomolecules of Streptomyces also play a crucial role in agriculture, where they are used as biopesticides and biocontrol agents (such as blasticidin-S and validamycin) . The present investigation demonstrated that actinobacteria isolated from the rhizosphere of Phoenix dactylifera L. in the Algerian Ghardaïa region produce bioactive compounds of biotechnological interest. The findings revealed a wide variety of enzymes and significant antimicrobial, antioxidant, and protein denaturation inhibition activities of the isolates, specifically the isolate Streptomyces sp. SGI16. The results offer a promising application in the industrial, pharmaceutical, and agricultural sectors, particularly in the development of new solutions for human and plant health. Additional investigations are needed to further investigate the strain and its extracts. Future research can be directed toward the purification and chemical characterization of the bioactive molecules that can be retrieved from the strains. Biological characterization can be further studied using in vivo models to support their biotechnological use. |
Precise multi-factor immediate implant placement decision models based on machine learning | e067b4d6-293e-43e0-9dbc-4c895125fc4e | 11814095 | Dentistry[mh] | Immediate implant placement at the time of tooth extraction is increasing in popularity for shortening the treatment period, reducing the number of surgeries interventions and trauma. Studies have validated the feasibility of this technique. Survival rates are similar between immediate and delayed placement, if enough primary stability of the implant was obtained in the residual bone of the tooth extraction socket . However, bone defects and rugged bone morphology after tooth extraction make it more difficult for implants to achieve enough primary stability than delayed implant placement . Decision-making regarding patient eligibility for immediate implant placement, choice of surgical technique, and selection of implant type largely relies on the doctor’s clinical experience. Preoperative misjudgment may result in suboptimal clinical outcomes. Consequently, developing a comprehensive and scientifically rigorous assessment method is crucial. Such a method would enable a more accurate analysis and prediction of immediate implant success rates for patients, aid clinicians in clinical decision-making, and enhance communication between doctors and patients. Primary stability refers to the initial mechanical fixation of an implant within the bone immediately following placement. It is instrumental in achieving osseointegration by minimizing micromotion and averting fibrous encapsulation. This primary or mechanical stability transitions into secondary or biologic stability through the process of osseointegration, which involves bone remodeling around the implant . Insertion torque was used to evaluates the primary stability of implants by recording the final torque value during implant insertion using torque wrench or surgical motor . Optimal insertion torque is considered to facilitate bone cell differentiation. Low insertion torque may result in micro-movements, fibrous tissue formation, and premature implant failure. While high insertion torque values could potentially induce excessive peri-implant bone remodeling, buccal soft tissue recession, and increased bone resorption . Thus, the prediction of insertion torque serves as a valuable metric for assessing the success rate of implant treatment. Previous studies have identified factors that influence the insertion torque of implant, including the macro-design of implant, surgical technique of osteotomy preparation, bone quality of implant site, etc. – . It is important to note that the majority of these studies are conducted within the context of delayed implant placement. However, achieving the required bone volume for full-length implant placement is often impractical in clinical settings due to anatomical constraints and patient-specific factors. Therefore, it is essential to simulate the conditions of immediate implant placement and subsequently reassess the factors that affect the insertion torque. Current studies on insertion torque typically isolate single factors, which, while informative, do not fully represent the combined effect of multiple interrelated factors in clinical settings. Effective clinical decision-making requires a holistic consideration of these factors for planning surgeries, selecting techniques, and choosing implants. Single-factor assessments are often insufficient for accurate preoperative decisions, especially for inexperienced doctors, and may increase patient risks . Hence, there’s an urgent need for an advanced decision-making model that considers a comprehensive analysis of multiple factors. Creating a decision model based on multi-factor analysis is challenging due to the complex interactions and individual influences of these factors on outcomes. This process requires extensive sample data and sophisticated statistical methods. Machine learning-based artificial intelligence, particularly through in vitro research, can facilitate the development of such a model. Hence, the focus of this research is to investigate the effects and co-effect of various implant apex designs, bone densities, implant intraosseous depths, and osteotomy preparation protocols on the implant insertion torque required for prior to making decisions of immediate implant surgery, and to construct machine learning preoperative prediction model for immediate implant insertion torque. Implants and polyurethane foam blocks Three different Nobel implant systems characterized with different apex designs were used in the present in vitro study, including Nobel Active, Nobel Parallel CC and Nobel Replace CC. The study evaluated the performance of three distinct dental implant apex designs: the Nobel Active implant has tapered apex with deep thread and long cutting edge. Nobel Parallel CC has tapered apex with shallow thread and long cutting edge. Nobel Replace CC has tapered apex with shallow thread and no cutting edge (Table ). The polyurethane block (Sawbones Europe AB, Malmö, Sweden) represents an alternative to animal or corpse bone and exhibits common mechanical properties according to standards defined by the manufacturer (ASTM F-1839-08), which reduces the variables, alterations, and deformations found using cadaver bone or in animal bone. In this study, polyurethane blocks with three densities (15, 20, 30 pounds per cubic feet (pcf)) were used to simulate soft, medium, hard bone respectively , , . Experimental procedure Six implant replicas of each of the three implant systems (Nobel Active, Nobel Parallel CC and Nobel Replace CC) were placed in each of the three densities of polyurethane foam block (15, 20 and 30pcf) with two osteotomy preparation protocols (normal preparation and undersized preparation) at three implant intraosseous depths (3 mm, 5 mm and 7 mm) (Fig. ). The procedure of normal preparation was performed according to the manufacturer, while the procedure of undersized preparation reduced the use of the terminal drill on the basis of normal preparation procedure. Drill stop ring was used to control the preparation depth at 3 mm, 5 mm and 7 mm. All implants were inserted self-tapping with an implant drill unit set on a maximum torque of 70 N·cm (iChiropo™, Bien-Air Dental SA, Bienne, Switzerland). A time-torque curve was registered for each implant site, which was used to extract the maximal insertion torque (Max IT ) in N·cm for each site (Fig. ). All surgical procedures were performed by the same experienced surgeon who had inserted more than 1500 dental implants. Ethics approval was not required for this in vitro study. Statistical analysis Mean Max IT ± standard deviations (SD) were calculated. Due to the Max ITs were not normally distributed, the raw data of Max ITs were transformed with Box-cox transformation for subsequent analysis. Box-cox (Max IT)=(Max IT × λ + 1) 1/λ , λ = 0.3.Then the data were further processed by statistical using SPSS 29.0. The sample size were calculated using G*Power software (version 3.1.9.7) and leveraging a linear model calculation method. The parameters employed consisted of a power level of 0.95, an alpha level of 0.05, an effect size of d = 0.15, and 4 predictors. These parameters pointed to a need for a minimum of 169 samples. A total of 324 samples ( n = 6) were adopted in this study. The effect of implant apex design, intraosseous depth, bone quality and osteotomy preparation on the Max IT was analyzed using one-way ANOVA and four-way ANOVA with post hoc Bonferroni analysis. Then the significant factors were analyzed by multiple linear regression analysis to establish a prediction model for insertion torque. A decision tree regression model was constructed for predicting implant insertion torque based on machine learning methods using SPSSPRO. 70% of the raw data were utilized as the training set for model construction, while the remaining 30% of the raw data served as the validation set to assess the model’s performance. Compare the predictive indicators of the two models, including MSE (Mean Squared Error), RMSE (Root Mean Squared Error), MAE (Mean Absolute Error), MAPE (Mean Absolute Percentage Error), and R² (Coefficient of Determination), to evaluate the predictive power of both models. Statistical significance was analyzed with p < 0.05 considered significant. Three different Nobel implant systems characterized with different apex designs were used in the present in vitro study, including Nobel Active, Nobel Parallel CC and Nobel Replace CC. The study evaluated the performance of three distinct dental implant apex designs: the Nobel Active implant has tapered apex with deep thread and long cutting edge. Nobel Parallel CC has tapered apex with shallow thread and long cutting edge. Nobel Replace CC has tapered apex with shallow thread and no cutting edge (Table ). The polyurethane block (Sawbones Europe AB, Malmö, Sweden) represents an alternative to animal or corpse bone and exhibits common mechanical properties according to standards defined by the manufacturer (ASTM F-1839-08), which reduces the variables, alterations, and deformations found using cadaver bone or in animal bone. In this study, polyurethane blocks with three densities (15, 20, 30 pounds per cubic feet (pcf)) were used to simulate soft, medium, hard bone respectively , , . Six implant replicas of each of the three implant systems (Nobel Active, Nobel Parallel CC and Nobel Replace CC) were placed in each of the three densities of polyurethane foam block (15, 20 and 30pcf) with two osteotomy preparation protocols (normal preparation and undersized preparation) at three implant intraosseous depths (3 mm, 5 mm and 7 mm) (Fig. ). The procedure of normal preparation was performed according to the manufacturer, while the procedure of undersized preparation reduced the use of the terminal drill on the basis of normal preparation procedure. Drill stop ring was used to control the preparation depth at 3 mm, 5 mm and 7 mm. All implants were inserted self-tapping with an implant drill unit set on a maximum torque of 70 N·cm (iChiropo™, Bien-Air Dental SA, Bienne, Switzerland). A time-torque curve was registered for each implant site, which was used to extract the maximal insertion torque (Max IT ) in N·cm for each site (Fig. ). All surgical procedures were performed by the same experienced surgeon who had inserted more than 1500 dental implants. Ethics approval was not required for this in vitro study. Mean Max IT ± standard deviations (SD) were calculated. Due to the Max ITs were not normally distributed, the raw data of Max ITs were transformed with Box-cox transformation for subsequent analysis. Box-cox (Max IT)=(Max IT × λ + 1) 1/λ , λ = 0.3.Then the data were further processed by statistical using SPSS 29.0. The sample size were calculated using G*Power software (version 3.1.9.7) and leveraging a linear model calculation method. The parameters employed consisted of a power level of 0.95, an alpha level of 0.05, an effect size of d = 0.15, and 4 predictors. These parameters pointed to a need for a minimum of 169 samples. A total of 324 samples ( n = 6) were adopted in this study. The effect of implant apex design, intraosseous depth, bone quality and osteotomy preparation on the Max IT was analyzed using one-way ANOVA and four-way ANOVA with post hoc Bonferroni analysis. Then the significant factors were analyzed by multiple linear regression analysis to establish a prediction model for insertion torque. A decision tree regression model was constructed for predicting implant insertion torque based on machine learning methods using SPSSPRO. 70% of the raw data were utilized as the training set for model construction, while the remaining 30% of the raw data served as the validation set to assess the model’s performance. Compare the predictive indicators of the two models, including MSE (Mean Squared Error), RMSE (Root Mean Squared Error), MAE (Mean Absolute Error), MAPE (Mean Absolute Percentage Error), and R² (Coefficient of Determination), to evaluate the predictive power of both models. Statistical significance was analyzed with p < 0.05 considered significant. Table presents the means and means and standard deviations of Max IT for simulated immediate implant placement situation. The ANOVA analysis revealed that all these factors (implant apex design, intraosseous depth, bone quality and osteotomy preparation) significantly affected Max IT values (Table ). The interaction between osteotomy preparation protocol and intraosseous depth ( p = 0.629), among implant apex design, intraosseous depth and bone quality ( p = 0.430), and among osteotomy preparation protocol, intraosseous depth and bone quality ( p = 0.328) was not significant, while other interactions were significant ( p < 0.05) (Table ). From the analysis of the results of the F-test, it can be concluded that the P-value of significance is 0.000*** and identified all the variables (Implant Apex Design, Osteotomy preparation protocol, Intraosseous depth and Bone quality) as predictors of Max IT (Table ); For the collinearity of variables, all VIFs are less than 10, so the model does not have multi collinearity issues. The multiple linear regression models showed fair fitting with R 2 (R 2 = 0.839) greater than 0.80, indicating the model is good fit. Based on the multiple linear regression equation, Box-cox (Max IT) = -1.631 + 2.629×Bone Quality (Hard) + 0.6519×Bone Quality (Medium) + 0.7089×Implant Apex Design (Nobel Active) + 0.2699×Implant Apex Design (Nobel Parallel CC) + 1.3499×Osteotomy preparation protocol (Undersized) + 0.75×Intraosseous Depth. The analysis was conducted using the SPSSPRO software, where the data was divided into a training set and a validation set in a 7:3 ratio, and the predictive evaluation metrics of the decision tree model were calculated. Training set: MSE = 10.937, RMSE = 3.307, MAE = 2.364, MAPE = 18.197, R 2 = 0.959. Validation set: MSE = 13.309, RMSE = 3.648, MAE = 2.915, MAPE = 24.732, R 2 = 0.951. Following the same method, we calculated other evaluation metrics for the multiple linear regression model: MSE = 35.361, RMSE = 5.946, MAE = 4.447, MAPE = 33.639, R 2 = 0.839 (Table ). The true values of the validation set, predictions from the MLR (Multiple Linear Regression) model, and predictions from the Decision Tree Regression model were included in a line chart with the highest implant torque as the vertical axis to compare their fit, as shown in Fig. . The predictions from the Decision Tree Regression model fit the true values more closely, indicating that the predictive power of the Decision Tree Regression model is superior to that of the MLR model. Table presents “insertion torque performance” of Nobel implant system under different conditions based on decision tree model established in this study. Practicality and advancement of the decision model base on multi-factor analysis Preoperative prediction of insertion torque is essential for clinical decision-making. Currently, preoperative assessment of insertion torque in immediate implant placement remains a relatively understudied area. Some studies have established predictive models based on single factors, such as implant macrogeometries , bone density , drilling protocols , etc. However, the reference value for complex clinical situations is limited by the approach of considering only single factors. In this study, four factors including implant apex design, osteotomy preparation, intraosseous depth, and bone quality were simultaneously considered, and a predictive model was established using multiple linear regression analysis, which yielded a high R-value of 0.839. The model established in this study exhibits greater scientific rigor and advanced characteristics, providing enhanced guidance for clinical practice. In this study, we initiated the evaluation from an intraosseous depth of 3 mm, as our preliminary experiments revealed that depths of 1–2 mm were insufficient to achieve an insertion torque of over 10 N·cm. To better study the effect of intraosseous depth on insertion torque, intraosseous depths of 5 mm and 7 mm were set to simulate actual application scenarios. Our results indicated that when the intraosseous depth is 3 mm, employing a combination of deep-threaded implants featuring long cutting edges, along with undersized preparation, becomes essential to ensure optimal insertion torque. Nevertheless, it is important to recognize that with medium- to low-density bone, the method struggles to achieve an optimal insertion torque of over 25 N·cm. In the literature, an implant torque of at least 25 N·cm is widely accepted as the threshold criterion for immediate loading . Our findings indicated that the use of undersized preparation significantly enhances the insertion torque of implants, regardless of bone quality, implant design, or depth of insertion. Literature suggests that an insertion torque exceeding 50 N·cm is considered excessive and may heighten the risk of thermal bone damage and resorption , . Notably, at an intraosseous depth of 7 mm in hard bone conditions, the insertion torque achieved with undersized preparation surpassed 50 N·cm, irrespective of implant apex design, thus it is necessary to adopt the conventional osteotomy preparation protoco rather than the undersized preparation technique. For immediate implant placement, only the intraosseous portion of the implant can affect the insertion torque. Therefore, the impplant apex design should be also considered as a significant factor for immediate implant placement. Studies have demonstrated that implant apex with deep and dense threads are more beneficial for increasing bone-implant contact and obtaining high insertion torque. In addition, implant apex with cutting edge have good self-tapping ability, allowing them to be more easily inserted in undersized implant beds , – . The results of this study are generally consistent with the aforementioned research. However, it is worth noting that the implants with Tapered/Deep thread/Long cutting edge apex designs demonstrate their advantages primarily in hard bone and when using undersized preparation technique. The current consensus is that there is a positive correlation between bone quality and insertion torque . The results of this study also support this consensus that bone quality is the most significant influencing factor, compared to osteotomy preparation protocols, intraosseous depth and implant apex design. In addition, this study found that with an intraosseous depth of 3 mm, achieving an insertion torque of 25 N·m is only possible in cases of hard bone when using an implant with deep threads and cutting edges at the apex, and employing the undersized preparation technique. While in soft bone condition, even with an intraosseous depth of 7 mm, employing the undersized preparation technique along with an implant featuring deep threads and cutting edges at the apex, it was not possible to achieve a torque greater than 25 N·m, which is the minimum implant torque requirement for immediate loading . Based on our research findings, these four factors are crucial in influencing implant torque, while the clinical outcomes reflect the combined effects of these factors. Isolating the analysis of each factor independently would lead to misjudgments of the results. However, the establishment of a comprehensive analysis model incorporating multiple factors undoubtedly presents significant challenges. In addition to traditional statistical methods, artificial intelligence and machine learning algorithms, such as decision tree regression, can also be utilized in predictive model construction. Decision tree regression models are notable for their capacity to manage small sample sizes effectively. Moreover, they offer excellent visualization and transparent decision-making processes, distinguishing them from other ‘black-box’ models . In this study, we developed a decision tree regression model for predictive purposes, achieving an R-value of 0.951, which exceeds that of the multiple linear regression model. This suggests that the decision tree regression model offers superior predictive accuracy for the insertion torque in immediate implant placement. With the rapid advancement of artificial intelligence, the medical field has emerged as a prominent domain for AI applications. Leveraging extensive medical data, AI technology holds the promise to aid clinicians in making more precise decisions regarding disease diagnosis, treatment planning, and patient risk assessment, thereby enhancing the quality and efficiency of healthcare services. A more sophisticated decision model for clinical application guidance If postoperative implant insertion torque is inadequate, it is necessary to forgo immediate loading and consider submerged healing instead. However, for immediate implant placement, insufficient soft tissue volume can complicate achieving complete wound closure required for submerged healing. Hence, preoperative prediction of insertion torque is essential for clinical decision-making.In actual clinical scenarios, the factors mentioned in this study often cannot be simultaneously optimized. However, we can optimize some adjustable factors according to the actual situation to accurately control the insertion torque and maximize the success rate of the implant. Assuming we only have Nobel Replace CC implants available with tapered/ shallow thread/ no cutting edge apex design, and the patient suffers from osteoporosis, but the mandibular molar to be extracted has a bone height greater than 7 mm from the inferior alveolar nerve canal, we can adopt an undersized osteotomy technique and ensure that the intraosseous depth of implant is at least 7 mm. By doing so, we can achieve an insertion torque close to 25 N·cm (Table ). Furthemore, if the insertion torque still cannot be increased to a safe range within the scope of various factors that can be optimized, then we should not recommend the patient to adopt the immediate implant placement. This is of significant reference value for optimizing clinical decision-making, improving the success rate of implants, and enhancing the efficiency of doctor-patient communication. A more comprehensive and standardized evaluation system can be established Currently, many implant manufacturers claim that their products can achieve good primary stability (or insertion torque), but this is not rigorous. Based on the insertion torque prediction model established in this study, except for the intrinsic factors of the implant, it is also necessary to consider the actual clinical application scenarios, such as intraosseous depth of the implant, bone quality of the implant site, and the osteotomy preparation protocol. Based on the results of this study, we propose the concept of “implant insertion torque performance”. Manufacturers can calculate the insertion torque of the implant under different clinical scenarios using the prediction model of this study, and list specific values and reasonable osteotomy preparation protocol in the implant instruction manual for reference by clinical doctors. As shown in Table , taking the Nobel implant system as an example, we have sorted out its “insertion torque performance” table according to the prediction model established in this study, which can greatly improve the safety and success rate of implant surgery and increase the decision-making efficiency of doctors. At the same time, in this way, the primary stability (or insertion torque performance) provided by different brands of implants will be more comparable. The present study has certain limitations. Firstly, it is an in vitro investigation utilizing a standardized model, which may not entirely replicate the complex bone conditions encountered in clinical settings . Secondly, the diameter of the implants was not included in the study, despite being reported in numerous studies as one of the factors influencing the insertion torque of implants . Thirdly, surgeon’s inexperience may be another elements for low insertion torque . Preoperative prediction of insertion torque is essential for clinical decision-making. Currently, preoperative assessment of insertion torque in immediate implant placement remains a relatively understudied area. Some studies have established predictive models based on single factors, such as implant macrogeometries , bone density , drilling protocols , etc. However, the reference value for complex clinical situations is limited by the approach of considering only single factors. In this study, four factors including implant apex design, osteotomy preparation, intraosseous depth, and bone quality were simultaneously considered, and a predictive model was established using multiple linear regression analysis, which yielded a high R-value of 0.839. The model established in this study exhibits greater scientific rigor and advanced characteristics, providing enhanced guidance for clinical practice. In this study, we initiated the evaluation from an intraosseous depth of 3 mm, as our preliminary experiments revealed that depths of 1–2 mm were insufficient to achieve an insertion torque of over 10 N·cm. To better study the effect of intraosseous depth on insertion torque, intraosseous depths of 5 mm and 7 mm were set to simulate actual application scenarios. Our results indicated that when the intraosseous depth is 3 mm, employing a combination of deep-threaded implants featuring long cutting edges, along with undersized preparation, becomes essential to ensure optimal insertion torque. Nevertheless, it is important to recognize that with medium- to low-density bone, the method struggles to achieve an optimal insertion torque of over 25 N·cm. In the literature, an implant torque of at least 25 N·cm is widely accepted as the threshold criterion for immediate loading . Our findings indicated that the use of undersized preparation significantly enhances the insertion torque of implants, regardless of bone quality, implant design, or depth of insertion. Literature suggests that an insertion torque exceeding 50 N·cm is considered excessive and may heighten the risk of thermal bone damage and resorption , . Notably, at an intraosseous depth of 7 mm in hard bone conditions, the insertion torque achieved with undersized preparation surpassed 50 N·cm, irrespective of implant apex design, thus it is necessary to adopt the conventional osteotomy preparation protoco rather than the undersized preparation technique. For immediate implant placement, only the intraosseous portion of the implant can affect the insertion torque. Therefore, the impplant apex design should be also considered as a significant factor for immediate implant placement. Studies have demonstrated that implant apex with deep and dense threads are more beneficial for increasing bone-implant contact and obtaining high insertion torque. In addition, implant apex with cutting edge have good self-tapping ability, allowing them to be more easily inserted in undersized implant beds , – . The results of this study are generally consistent with the aforementioned research. However, it is worth noting that the implants with Tapered/Deep thread/Long cutting edge apex designs demonstrate their advantages primarily in hard bone and when using undersized preparation technique. The current consensus is that there is a positive correlation between bone quality and insertion torque . The results of this study also support this consensus that bone quality is the most significant influencing factor, compared to osteotomy preparation protocols, intraosseous depth and implant apex design. In addition, this study found that with an intraosseous depth of 3 mm, achieving an insertion torque of 25 N·m is only possible in cases of hard bone when using an implant with deep threads and cutting edges at the apex, and employing the undersized preparation technique. While in soft bone condition, even with an intraosseous depth of 7 mm, employing the undersized preparation technique along with an implant featuring deep threads and cutting edges at the apex, it was not possible to achieve a torque greater than 25 N·m, which is the minimum implant torque requirement for immediate loading . Based on our research findings, these four factors are crucial in influencing implant torque, while the clinical outcomes reflect the combined effects of these factors. Isolating the analysis of each factor independently would lead to misjudgments of the results. However, the establishment of a comprehensive analysis model incorporating multiple factors undoubtedly presents significant challenges. In addition to traditional statistical methods, artificial intelligence and machine learning algorithms, such as decision tree regression, can also be utilized in predictive model construction. Decision tree regression models are notable for their capacity to manage small sample sizes effectively. Moreover, they offer excellent visualization and transparent decision-making processes, distinguishing them from other ‘black-box’ models . In this study, we developed a decision tree regression model for predictive purposes, achieving an R-value of 0.951, which exceeds that of the multiple linear regression model. This suggests that the decision tree regression model offers superior predictive accuracy for the insertion torque in immediate implant placement. With the rapid advancement of artificial intelligence, the medical field has emerged as a prominent domain for AI applications. Leveraging extensive medical data, AI technology holds the promise to aid clinicians in making more precise decisions regarding disease diagnosis, treatment planning, and patient risk assessment, thereby enhancing the quality and efficiency of healthcare services. If postoperative implant insertion torque is inadequate, it is necessary to forgo immediate loading and consider submerged healing instead. However, for immediate implant placement, insufficient soft tissue volume can complicate achieving complete wound closure required for submerged healing. Hence, preoperative prediction of insertion torque is essential for clinical decision-making.In actual clinical scenarios, the factors mentioned in this study often cannot be simultaneously optimized. However, we can optimize some adjustable factors according to the actual situation to accurately control the insertion torque and maximize the success rate of the implant. Assuming we only have Nobel Replace CC implants available with tapered/ shallow thread/ no cutting edge apex design, and the patient suffers from osteoporosis, but the mandibular molar to be extracted has a bone height greater than 7 mm from the inferior alveolar nerve canal, we can adopt an undersized osteotomy technique and ensure that the intraosseous depth of implant is at least 7 mm. By doing so, we can achieve an insertion torque close to 25 N·cm (Table ). Furthemore, if the insertion torque still cannot be increased to a safe range within the scope of various factors that can be optimized, then we should not recommend the patient to adopt the immediate implant placement. This is of significant reference value for optimizing clinical decision-making, improving the success rate of implants, and enhancing the efficiency of doctor-patient communication. Currently, many implant manufacturers claim that their products can achieve good primary stability (or insertion torque), but this is not rigorous. Based on the insertion torque prediction model established in this study, except for the intrinsic factors of the implant, it is also necessary to consider the actual clinical application scenarios, such as intraosseous depth of the implant, bone quality of the implant site, and the osteotomy preparation protocol. Based on the results of this study, we propose the concept of “implant insertion torque performance”. Manufacturers can calculate the insertion torque of the implant under different clinical scenarios using the prediction model of this study, and list specific values and reasonable osteotomy preparation protocol in the implant instruction manual for reference by clinical doctors. As shown in Table , taking the Nobel implant system as an example, we have sorted out its “insertion torque performance” table according to the prediction model established in this study, which can greatly improve the safety and success rate of implant surgery and increase the decision-making efficiency of doctors. At the same time, in this way, the primary stability (or insertion torque performance) provided by different brands of implants will be more comparable. The present study has certain limitations. Firstly, it is an in vitro investigation utilizing a standardized model, which may not entirely replicate the complex bone conditions encountered in clinical settings . Secondly, the diameter of the implants was not included in the study, despite being reported in numerous studies as one of the factors influencing the insertion torque of implants . Thirdly, surgeon’s inexperience may be another elements for low insertion torque . The following conclusions are drawn from the results of this in vitro experimental study: In the context of immediate implant placement, the factors influencing implant insertion torque are ranked in the following order: bone quality, intraosseous depth, osteotomy preparation protocol, and implant apex design. Both traditional on multiple linear regression and novel machine learning models have demonstrated the capability to construct highly accurate predictive models for preoperative decision-making regarding the feasibility of immediate implant surgery. The methodologies employed in this study standardize the assessment of ‘insertion torque performance’ across different dental implant systems, thereby facilitating more precise decision-making. |
Multi-omics approaches for understanding gene-environment interactions in noncommunicable diseases: techniques, translation, and equity issues | 947c51e5-83ac-4ae9-bd62-46d266c8e00e | 11786457 | Biochemistry[mh] | Non-communicable diseases (NCDs), such as cardiovascular diseases, cancers, chronic respiratory diseases, diabetes, mental health disorders, and other complex diseases, pose a significant and growing global health challenge . Annually, NCDs account for 41 million deaths, constituting 60% of Disability Adjusted Life Years (DALYs), 81% of Years Lived with Disability (YLDs), and 74% of all global fatalities , making them the primary cause of disease burden and death worldwide . For example, cardiovascular diseases alone claim 17.9 million lives each year, followed by cancers (9.3 million), chronic respiratory diseases (4.1 million), and diabetes-related conditions (2.0 million), together accounting for over 80% of all premature NCD deaths . Economically, the cumulative global burden of NCDs from 2010 to 2030 is estimated to exceed USD 47 trillion—a figure that represents 75% of the global GDP in 2010 . This rise in NCDs can largely be attributed to rapid unplanned urbanization, the globalization of unhealthy lifestyles, and an aging population . NCDs affect individuals across all demographics and countries, with a disproportionately severe impact on low- and middle-income countries, where over three-quarters of global NCD-related deaths, approximately 31.4 million, occur annually . These diseases arise from complex interactions between genetic and environmental— including physical inactivity, unhealthy diets, obesity, and the use of tobacco or alcohol . Although NCDs typically manifest in adulthood, their roots are often traced back to behaviors and conditions established during childhood and adolescence . The burden of NCDs alongside existing infectious diseases poses significant economic stability and development challenges, exacerbating poverty and straining health systems, reducing resilience to emergencies such as infectious disease outbreaks and natural disasters . Furthermore, the high burden of NCDs is a major obstacle to progress towards the 2030 Agenda for Sustainable Development, specifically the target to reduce premature mortality from the four principal NCDs (cancers, cardiovascular diseases, chronic respiratory diseases, and diabetes) by one-third by 2030 . Family- and population-based studies have revealed that most NCDs possess substantial genetic components, with diseases such as coronary artery disease (CAD) and autism spectrum disorder (ASD) demonstrating high heritability, estimated at approximately 50% and 80%, respectively. Most NCDs are predominantly polygenic, involving numerous genetic variants that each contribute subtly to overall disease risk [ –. Advances in omics technologies, particularly genome-wide association studies (GWAS), have successfully identified many genetic variants linked to NCDs [ –. However, our understanding of the genetic etiology of NCDs remains incomplete . There are several challenges, including the 'missing heritability problem,' where known genetic variants associated with a disease/trait account for only a small fraction of the expected heritability [ –. Additionally, pinpointing the true causal variants that contribute to disease mechanisms has proven difficult due to the complex linkage disequilibrium (LD) structure of GWAS nominated variants, which limits thier clinical utility [ –. Recent advances in whole genome sequencing (WGS) studies have begun to elucidate the role of rare genetic variants in NCDs, while also offering additional insights into the contribution of common variants through improved resolution and comprehensive genomic coverage. Despite these insights, rare variants do not fully explain the missing heritability in NCDs, underscoring their complexity and multi-factorial nature . A key factor that may explain this missing heritability is the complex interplay between genetic variants and environmental factors—often called gene-environment (GxE) interactions [ , –. In this context, an ‘environment’ could be any endogenous or exogenous non-genetic factor that influences the risk of developing NCDs . To fully understand the complex GxE interactions that underpin the biological basis of NCDs, it is essential to integrate information across multiple levels [ –. This integration encompasses molecular profiles from the genome, epigenome, transcriptome, proteome, metabolome, lipidome, and microbiome—collectively referred to as multi-omics—along with environmental exposures known as the exposome. Rapid advancements in computational methodology have made the integration of multi-omics including high-throughput sequencing, mass spectrometry, smart wearable devices, and expanded electronic health records (EHRs) data increasingly feasible . Such omics integration generates comprehensive data at an unprecedented speed and scale, enhancing our understanding of disease mechanisms and revolutionizing precision medicine by enabling targeted prevention, precise diagnostics, personalized treatments, and accurate prognosis . The multi-omics approach requires innovative integration methods that combine information from diverse omics data sources . These methods facilitate the assessment of information flow from one omics layer to another and help elucidate the intricate interplay between various molecular profiles. Recently, advances in statistical modeling and machine learning have enabled more effective integration of omics data, which is crucial for tackling the complexity of NCDs . Despite this potential, several significant challenges remain. A primary obstacle is deciding which omics layer to prioritize in multi-omics studies . While many researchers adopt a genome-first approach, the optimal strategy may vary depending on the specific disease and available data. Another challenge is the lack of genetic diversity in most multi-omics datasets, as most of these datasets have predominantly been based on samples of European genetic ancestry [ –. Studies have shown that results derived from predominantly European datasets often do not translate well to individuals of non-European ancestry, potentially exacerbating health disparities by limiting research benefits to certain groups . Furthermore, the heterogeneity and massive scale of multi-omics datasets pose substantial challenges in data integration, requiring significant computational resources, skills, and advanced analytical techniques . In this scoping review, we examine the multi-omics literature comprehensively, specifically focusing on NCDs and omics (multi-omics) diversity. Our primary goal is to assess the current landscape of global multi-omics data as it relates to NCDs, summarizing key advances in data integration techniques that enable a deeper understanding of the intricate GxE interactions at play. We will delve into the significant role of multi-omics research in elucidating the complex pathways influencing the development, progression, and response to treatment in NCDs. Next, we illustrate practical translational applications and point out critical limitations currently facing the field. Additionally, we discuss the transformative potential of global multi-omics research initiatives in advancing precision medicine, specifically in tailoring prevention, diagnosis, and treatment strategies to individual genetic and environmental profiles. Lastly, we propose directions to address existing challenges of multi-omics research to enhance our understanding of the biological mechanisms of NCDs and development of effective interventions. Research has consistently demonstrated that the risk of developing most NCDs and the effectiveness of treatments are influenced by both the independent effects of an individual's genetic makeup and various environmental exposures, as well as by the potential synergistic or antagonistic interactions between these two factors . One type of GxE interaction occurs when an individual’s genotype modulates the effect of environmental exposure on disease risk. For example, certain genetic variants may alter the risk of developing Parkinson's disease in individuals exposed to organophosphate pesticides . Conversely, another form of GxE interaction happens when the influence of a genotype on disease risk changes with different environmental exposures . A notable case is how the impact of the FTO gene on body mass index (BMI) can significantly vary depending on lifestyle factors such as physical activity, diet, alcohol consumption, and sleep duration . These examples underscore the dynamic interplay between our largely static genetic code and the responsive molecular layers of the genome and epigenome. These layers dynamically respond to environmental changes, affecting gene expression and cellular functions, representing key mechanisms through which GxE interactions manifest. Omics technologies— powered by advances in high-throughput sequencing technologies such as next-generation sequencing (NGS) and rapidly expanding electronic data (exosomes), enable a comprehensive analysis of various biological systems . Each technology focuses on a different aspect: genomics and epigenomics explore genetic and epigenetic variations; transcriptomics examines gene expression dynamics; proteomics investigates protein functions and interactions; metabolomics assesses metabolic responses; exposomics evaluates lifelong environmental exposures. While each technology excels at quantifying specific types of biomolecules, the complete picture of disease mechanisms often involves intricate molecular machinery such as transcriptional and translational regulation, RNA and peptide degradation, posttranslational modifications, and molecular transport . Thus, focusing solely on one type of omics data can overlook critical interactions between these processes. Genomics Genomics, the most established omics technologies, has profoundly enhanced our understanding of NCDs through extensive profiling of genetic variants such as SNPs, insertions-deletions, and structural variants [ , , , –. Pioneering advancements in NGS technologies have been crucial, providing extensive genome-wide coverage that is faster and more cost-effective than ever before . Significant and fast reduction in sequencing costs has spurred substantial growth in genomic and multi-omics research, making large-scale studies more feasible and affordable (Fig. ). So far, over 6000 GWASs have been conducted for more than 3000 traits, yielding thousands of associated genetic variants . This represents a substantial advance over the pre-GWAS era when only a handful of genetic associations were robustly identified . For instance, a GWAS of Crohn's disease implicated the IL-12/IL-23 pathway in the development of the disease, which subsequently informed clinical trials of drugs that targeted these pathways . Furthermore, polygenic scores (PGSs) that aggregate genetic risk information across the genome are increasingly used to predict an individual's risk of developing NCDs and other diseases . A recent clinical study demonstrated the effectiveness of PGS-based risk assessments for 10 NCDs, including coronary artery disease, atrial fibrillation, type 2 diabetes, chronic kidney disease, and breast cancer. Notably, this study returned genome-informed risk assessment results to patients, marking a significant milestone in clinical genetics . Additionally, in psychiatric genomics, PGSs have shown promise in predicting treatment outcomes for mental health disorders, including treatment response, resistance, side effects, and hospitalization rates [ –. Although GWASs have successfully identified replicable genetic variants associated with many NCDs and other traits, there are significant methodological and ethical challenges that must be addressed before these findings can be fully translated into preventive and clinical treatments. One major limitation is the poor transferability of findings across different genetic ancestries . This discrepancy largely stems from the underrepresentation of non-European ancestry in GWAS cohorts (only 14%) . This lack of diversity not only impedes the clinical application of PGSs but also exacerbates health disparities, because the benefits of ancestry-biased genetic research cannot equitably be distributed across populations [ , , –. A critical aspect often overlooked is that Africa harbors the greatest human genetic diversity in the world, which offers unique opportunities for understanding genetic susceptibility to NCDs and other complex traits [ –. The African Genome Variation Project (AGVP), for example, uncovered over 8 million novel variants, with a substantial proportion identified in Ethiopian and Zulu populations . Moreover, African populations possess shorter haplotype blocks and complex population substructures, which allow for more precise fine mapping of disease susceptibility alleles . This diversity, combined with the unique genetic adaptations in response to diverse climates, diets, and infectious diseases, underscores the necessity of expanding large-scale sequencing efforts in African populations. Incorporating these genomes will not only advance our understanding of NCDs but also ensure that the benefits of genomic medicine are equitably distributed across all populations . Despite the slow progress, there are promising global efforts aimed at tackling the significant lack of diversity in genomic research. The Human Heredity and Health in Africa (H3Africa) initiative, the largest genomic research consortium in Africa, is spearheading this effort with a 10-year project aimed at studying the genetic basis of disease among African populations and establishing sustainable genomics research across the continent . This initiative includes the creation of three biorepositories in Uganda, South Africa, and Nigeria, and the development of the Pan African Bioinformatics Network (H3ABioNet), supporting advances in handling biological data . In Latin America, the Latin American Genomics Consortium is harmonizing data from existing cohorts and planning new recruitments to build a substantial biobank, addressing the underrepresentation of admixed populations . In the United States, the All of Us Research Program aims to mirror the country's diversity by collecting data from over one million participants, half of whom are of non-European genetic ancestry. This program has identified over 275 million previously unreported genetic variants, with 77% of its participants coming from historically underrepresented communities in biomedical research . Despite these encouraging efforts, the progress is far from sufficient. There is a substantial disparity within continents, particularly in Africa, Latin America, South Asia, and West Asia, where only a few countries have well-established biobanks . This highlights the ongoing need for more comprehensive initiatives to ensure that genetic research benefits all global populations equitably. Another limitation of standard population-based GWAS is the bias arising from population stratification and assortative mating, which can distort the estimated effects of variants on phenotypes [ –. Standard-GWAS results are influenced by several factors, including the direct effects of alleles carried by an individual on their phenotype; the indirect effects of alleles carried by relative(s) through environmental influences (genetic nurture); and confounding due to population stratification and assortative mating. Although methods such as principal-component (PC) analysis and linear mixed models (LMMs) are used to adjust for population stratification, residual confounding often persists in GWAS summary statistics [ –. These biases are particularly pronounced in polygenic scores (PGSs), which aggregate genetic risk information from thousands of variants . Additionally, such biases can also impact post-GWAS analyses, including biological annotation, heritability estimation, genetic correlations, Mendelian Randomization (MR), and GxE interaction . While family-based GWASs typically have lower power than population-based GWASs due to smaller sample sizes, they have been shown to mitigate biases from population stratification effectively . A recent within-family GWAS, conducted on a large sample of siblings, has demonstrated that within-family association estimates are significantly attenuated compared to standard GWAS estimates for traits such as depressive symptoms, height, and smoking . The increasing availability of family-based data offers great potential for disentangling direct and indirect genetic effects affecting NCDs, thereby aiding in unraveling complex GxE interactions. Transcriptomics Transcriptomics, through RNA sequencing (RNA-seq) technologies, has become instrumental in elucidating cellular pathways critical to the pathophysiology of many NCDs . By analyzing all RNA transcripts, including coding and non-coding types, RNA-seq provides comprehensive insights into mRNA abundance, alternative splicing, nucleotide variations, and structural alterations . By revealing how gene expression is regulated and altered under various conditions, transcriptomics plays a pivotal role in bridging genotypic variations with phenotypic manifestations. For instance, a study by Romanoski et al. (2010) integrated transcriptomics and genomics to examine human aortic endothelial cells' response to oxidized phospholipids, a key factor in atherosclerosis—a major cause of heart disease . They treated cells with oxidized phospholipids, known to induce vascular inflammation, and simultaneously performed analysis to identify expression quantitative trait loci (eQTLs) influencing gene expression changes. This approach revealed that approximately one-third of the highly regulated transcripts exhibited gene-environment (GxE) interactions, often influenced by distal, trans-acting effects. Some notable interactions were further validated through small interfering RNA (siRNA) knockdown experiments, confirming the significant role of specific genetic loci in modulating gene expression responses to environmental stimuli . This and other related studies [ – illustrate how integrating genomic and transcriptomic data can uncover complex GxE interactions, enhancing our understanding of the genetic and environmental underpinnings of cardiovascular diseases and other NCDs. Building on the capabilities of RNA sequencing technologies, the surge in transcriptomic data has prompted the establishment of comprehensive consortia tasked with managing, curating, and distributing these resources to the broader scientific community. Among the most noteworthy is The Cancer Genome Atlas (TCGA), which provides a rich repository of cancer-related transcriptomic data. Similarly, the Allen Human Brain Atlas offers specialized RNAseq databases focusing on brain diseases, encompassing studies on aging, dementia, and traumatic brain injury. Additionally, repositories such as the Gene Expression Omnibus (GEO), Encyclopedia of DNA Elements (ENCODE) , and the Genotype-Tissue Expression (GTEx ) Project significantly contribute to the availability of transcriptomic data across various tissues and conditions. By providing access to extensive transcriptomic data, these consortia support the ongoing exploration of how gene expression is intricately regulated and modified, thus continuing to bridge genotypic variations with phenotypic manifestations in complex disease research. Epigenomics Epigenomics, which examines the full spectrum of epigenetic modifications such as DNA methylation and histone modification, plays a crucial role in understanding how environmental factors and genetic predispositions interact to influence the development of diseases . These modifications regulate gene expression without altering the DNA sequence and are involved in critical processes like cellular differentiation and tumorigenesis . The epigenome's responsiveness to various environmental exposures—such as metals, air pollution, electromagnetic radiation—and lifestyle factors like diet, smoking, and physical activity, as well as the natural aging process, underscores its dynamic nature . For example, chronic exposure to arsenic and lead is associated with DNA methylation changes that heighten the risk of various cancers . Similarly, prenatal dietary factors like folate intake can alter the epigenome, influencing fetal development and disease susceptibility later in life [ –. Additionally, medications such as sodium valproate (VPA), used for treating epilepsy and bipolar disorder, demonstrate the complexity of interactions between pharmacological treatments and epigenetic regulation by affecting gene expression through their histone deacetylase (HDAC) inhibitor properties . Epigenomics can provide a molecular framework to understand how GxE effects manifest in NCDs. Through studying epigenetic modifications, researchers can discover novel genes and pathways influenced by genetic factors and environmental exposures. The epigenome is partly regulated by the genome, with genetic variation influencing the establishment of DNA methylation marks [ –, while also being highly responsive to environmental factors [ –. This dual regulation highlights the complexity of gene-environment interactions. For instance, the expression of certain NCD risk variants may depend on specific DNA methylation states, which environmental factors can alter. Alternatively, genetic variations might predispose certain epigenomic profiles to respond differently to environmental exposures, thus influencing disease risk . Recent studies, such as those by Teh et al. (2014), have shown that a significant proportion of variably methylated regions, areas where methylation levels vary substantially among individuals, can be attributed to GxE interactions, revealing the intricate molecular mechanisms at play . Advances in next-generation sequencing technologies have significantly enhanced the precision and scope of epigenic profiling . While techniques like bisulfite conversion have been widely used to map methylation, they come with challenges, such as incomplete conversion and DNA degradation. The advent of long-read sequencing technologies, such as PacBio’s HiFi sequencing, has addressed some of these limitations. HiFi sequencing can directly detect 5mC methylation without the need for bisulfite conversion, offering both high accuracy and the ability to resolve methylation profiles alongside phased haplotyping in a single run . This capability significantly improves our understanding of epigenetic modifications linked to genetic variants and environmental factors. On the other hand, major collaborative efforts like the NIH Roadmap Epigenomics project, the International Human Epigenome Consortium (IHEC), and ENCODE project have provided comprehensive maps of the human epigenome. By linking epigenetic changes to functional outcomes, these consortia enhance our understanding of the complex interactions that define health and disease, paving the way for advances in precision medicine. Proteomics Proteomics explores the entire array of proteins produced or modified by an organism, offering crucial insights into the development and progression of NCDs. The proteome is highly dynamic, exhibiting considerable variability due to processes like alternative splicing, protein modifications, and the complex assembly of proteins into signaling networks . These processes, regulated spatially and temporally, allow proteomics to measure critical changes in amino acid mutations, peptide isoforms, and posttranslational modifications (PTMs) . PTMs like phosphorylation, acetylation, and glycosylation are especially significant, as their dysregulation is often implicated in cancer, cardiovascular diseases, and neurodegenerative disorders . Proteomic profiles also capture responses to environmental stimuli, such as diet , chemical exposure , and smoking , highlighting their value in understanding complex gene-environment or proteome-environment interactions and refining the selection of target genes for further investigation . A hallmark example of proteomics integration with genomic and phenotypic data is the UK Biobank Pharma Proteomics Project (UKB-PPP), a public–private partnership that profiled over 2,900 proteins in plasma samples from over 54,000 participants . This initiative identified over 14,000 protein quantitative trait loci (pQTLs), with 81% being novel. By comparing results from different platforms like Olink and SomaScan and across diverse ancestries, the project underscored the power of multi-level data integration in revealing protein-level differences that influence disease studies, enhancing our understanding of genomic associations and disease mechanisms across populations . Among the most notable associations with NCDs include a strong link between natriuretic peptide B (BNP) and heart failure and inflammatory bowel disease (IBD) associated with higher plasma levels of prostaglandin-H2 D-isomerase . While most analyses focused on participants of European genetic ancestry (n = 34,557), ancestry-specific pQTL studies in African (n = 934), Central/South Asian (n = 920), and other non-European groups revealed unique variants, many of which were absent or rare in Europeans . These findings underscore the importance of expanding proteomic studies to diverse ancestries to capture population-specific genetic and proteomic interactions, addressing disparities in disease risk and treatment. Another UK Biobank study identified over 5,000 associations between rare protein-coding variants and plasma protein abundances, significantly expanding our understanding of how rare variations influence proteomic profiles and highlighting their potential in identifying new therapeutic targets and biomarkers . Despite its promise, proteomics faces challenges in scalability, cost and analytical complexity. High-throughput platforms such as mass spectrometry (MS) and proximity extension assays (PEA) enable precise protein profiling from minimal amounts of biological samples, but they remain costly, limiting their application in large-scale studies . The high dynamic range of protein expression and complexity of many PTMs and sequence variations pose further technical hurdles . Addressing these challenges is crucial for fully leveraging the potential of proteomics for novel biomarker discovery, targeted drug development, and understanding NCD mechanisms. Public–private collaborations like the UKB-PPP highlight the transformative potential of proteomics to bridge the gap between genetics and phenotypes in multi-omics research. By integrating proteomics into population-scale biobanks, researchers can enhance causal gene identification, refine patient stratification, and accelerate therapeutic discovery. However, ensuring equitable applications requires broadening the representation of underrepresented populations and addressing cost barriers. These advancements will enable proteomics to significantly contribute to precision medicine and effective management of NCDs globally. Metabolomics Metabolomics focuses on small-molecule metabolites—such as hormones, amino acids, and lipids—that serve as substrates, intermediates, and products of metabolism, offering a direct window into the biochemical pathways driving complex diseases, including NCDs . Closer to the actual phenotype than mRNA or protein, metabolite levels provide a particularly valuable physiological readout because they integrate environmental and multiple regulatory inputs . Each tissue or cell type has a unique metabolic signature, allowing metabolomics to highlight organ or tissue-specific changes linked to disease . The dynamic nature of the metabolome, highly responsive to factors such as diet and chemical exposure, makes it indispensable for studying gene-environment interactions in NCDs . Integrating metabolomics data with other omics layers, such as genomics and proteomics, enhances our ability to map metabolic pathways, predict metabolite abundances, and identify novel biomarkers and therapeutic targets across diverse populations. Metabolomics enables the quantification of both endogenous metabolites and xenobiotics—foreign substances like environmental chemicals, pollutants, and drugs—offering a comprehensive view of how external exposures impact biological systems . By analyzing these external compounds alongside changes in the endogenous metabolome, metabolomics reveals critical insights into the biological effects of environmental exposures. For example, a study on occupational exposure to trichloroethylene (TCE) identified TCE metabolites in human plasma and linked them to changes in endogenous metabolites associated with immunosuppression, hepatotoxicity, and nephrotoxicity, highlighting the toxic effects of TCE . Similarly, the EXPOsOMICS project explored biofluids and exhaled breath for disinfection by-products (DBPs) from swimming pools, uncovering potential disruptions to metabolites in the tryptophan pathway . In another study, researchers examined the relationship between SNPs in the methionine salvage enzyme APIP and mortality risk in sepsis triggered by infections like Salmonella. By analyzing plasma metabolomic profiles from about 1,000 patients, the study showed that sepsis survivors had significantly lower levels of the enzyme’s substrate, methylthioadenosine, than nonsurvivors, illustrating how genetic variation and metabolite levels jointly influence sepsis outcomes . These examples demonstrate metabolomics’ capacity to unravel gene-environment interactions and the biological consequences of external exposures. A key challenge in metabolomics is the identification and measurement of metabolites, but recent advancements have significantly eased this bottleneck . Technological advances in nuclear magnetic resonance (NMR) and mass spectrometry (MS)-based methods, such as GC–MS and LC–MS, have also improved the precision and range of metabolite quantification. Expanded metabolite databases, such as The Human Metabolome Database (HMDB) and XCMS-METLIN, now contain tens of thousands of metabolites, including xenobiotics from environmental sources. Additionally, various bioinformatics tools now enable more robust analysis, linking metabolic signatures to disease states and outcomes, thus enhancing the potential of metabolomics in NCD research . Exposomics Exposomics is a burgeoning field that explores the comprehensive impact of environmental factors on human health over an individual's lifetime . This discipline considers various exposures—from chemical and biological agents to psychosocial factors, socioeconomic status and interpersonal relationships. These factors can trigger various biological responses, including changes in gene and protein expression, which in turn may influence the microbiome and epigenome . This complex interplay underscores how environmental factors, intertwined with genetic predispositions, contribute to the development of NCDs (Fig. ). Examples of replicated gene-environment interactions include BRCA-1 associated protein-1 (BAP1) mutations and asbestos exposure for mesothelioma , chromodomain helicase DNA-binding protein 8 ( CHD8 ) and pesticide exposure for autism spectrum disorder , the fat mass and obesity-associated gene ( FTO ) and physical activity for obesity, and dopamine receptor D4 ( DRD4 ) and parenting style for attention-deficit/hyperactivity disorder (ADHD) . These interactions highlight how specific genetic susceptibilities can be activated or exacerbated by environmental factors, demonstrating the crucial role of exposomics in understanding NCDs. A subset of notable GxE interactions implicated in common NCDs is shown in Fig. . While most of these GxE examples have focused on single environmental variables, the broader and more systematic measurement of environmental factors—such as those captured through exposomics—holds tremendous potential for deepening our understanding of complex diseases. Although still in its early stages, a few studies have ventured into multi-exposure genome-wide interaction analysis, jointly modeling the effects of the genome and the environment using methods like StructLMM (structured linear mixed model) , GxEMMs (GxE Mixed Model) , and IGE (integrative analysis of genomic and exposomic data) . These approaches account for genome-exposome correlations and the interrelationships among exposome variables, offering a more holistic view of gene-environment interactions . However, such methods are often computationally intensive and challenging to interpret, underscoring the complexity and potential of exposomics in advancing our understanding of NCDs. Recent technological advancements, especially in high-resolution mass spectrometry (HRMS) and wearable devices, have significantly improved the ability to measure the exposome with precision and individual specificity . HRMS enables the detailed detection of small molecules in biological samples like plasma and urine, allowing for an in-depth analysis of exposures to pharmaceuticals, pollutants, and nutrients . Complementing this, wearable technologies such as silicone wristbands and other personal passive samplers have emerged as powerful tools for capturing personal exposure data. These devices can monitor environmental exposures in real-time across different settings and critical life stages, such as during pregnancy or early childhood, thus providing a dynamic and personalized exposome profile . For instance, studies employing wristband samplers in China have successfully profiled personal chemical exposures, demonstrating the diversity and complexity of environmental interactions individuals face daily . Another innovative approach uses a miniaturized wearable device that samples air to capture particulates, further analyzed using HRMS to identify both hydrophobic and hydrophilic chemical compounds . These studies exemplify how wearables can offer insights into the spatiotemporal dynamics of personal exposures and their potential health impacts. By integrating data from these wearables with systems biology approaches, researchers can now begin to unravel the intricate gene-environment interactions that significantly influence the pathogenesis of NCDs, paving the way for targeted prevention and therapeutic strategies. Furthermore, new initiatives are expanding the scope of exposomics research, leveraging large-scale resources to deepen our understanding of environmental contributions to health. The All of Us Research Program is increasingly integrating environmental exposure data with genomic, clinical, and demographic information from its diverse cohort of participants. By linking geospatial data with exposure estimates from tools like the Environmental Justice Index, the program aims to examine how environmental factors influence disease susceptibility. Similarly, the European Human Exposome Network (EHEN), the world’s largest network of exposome-focused projects, is advancing research on the health impacts of air pollution, noise, chemicals, and urbanization . Together, these efforts are equipping researchers with unparalleled data resources to elucidate the complex interplay between genetic and environmental factors in NCDs (Fig. ). Genomics, the most established omics technologies, has profoundly enhanced our understanding of NCDs through extensive profiling of genetic variants such as SNPs, insertions-deletions, and structural variants [ , , , –. Pioneering advancements in NGS technologies have been crucial, providing extensive genome-wide coverage that is faster and more cost-effective than ever before . Significant and fast reduction in sequencing costs has spurred substantial growth in genomic and multi-omics research, making large-scale studies more feasible and affordable (Fig. ). So far, over 6000 GWASs have been conducted for more than 3000 traits, yielding thousands of associated genetic variants . This represents a substantial advance over the pre-GWAS era when only a handful of genetic associations were robustly identified . For instance, a GWAS of Crohn's disease implicated the IL-12/IL-23 pathway in the development of the disease, which subsequently informed clinical trials of drugs that targeted these pathways . Furthermore, polygenic scores (PGSs) that aggregate genetic risk information across the genome are increasingly used to predict an individual's risk of developing NCDs and other diseases . A recent clinical study demonstrated the effectiveness of PGS-based risk assessments for 10 NCDs, including coronary artery disease, atrial fibrillation, type 2 diabetes, chronic kidney disease, and breast cancer. Notably, this study returned genome-informed risk assessment results to patients, marking a significant milestone in clinical genetics . Additionally, in psychiatric genomics, PGSs have shown promise in predicting treatment outcomes for mental health disorders, including treatment response, resistance, side effects, and hospitalization rates [ –. Although GWASs have successfully identified replicable genetic variants associated with many NCDs and other traits, there are significant methodological and ethical challenges that must be addressed before these findings can be fully translated into preventive and clinical treatments. One major limitation is the poor transferability of findings across different genetic ancestries . This discrepancy largely stems from the underrepresentation of non-European ancestry in GWAS cohorts (only 14%) . This lack of diversity not only impedes the clinical application of PGSs but also exacerbates health disparities, because the benefits of ancestry-biased genetic research cannot equitably be distributed across populations [ , , –. A critical aspect often overlooked is that Africa harbors the greatest human genetic diversity in the world, which offers unique opportunities for understanding genetic susceptibility to NCDs and other complex traits [ –. The African Genome Variation Project (AGVP), for example, uncovered over 8 million novel variants, with a substantial proportion identified in Ethiopian and Zulu populations . Moreover, African populations possess shorter haplotype blocks and complex population substructures, which allow for more precise fine mapping of disease susceptibility alleles . This diversity, combined with the unique genetic adaptations in response to diverse climates, diets, and infectious diseases, underscores the necessity of expanding large-scale sequencing efforts in African populations. Incorporating these genomes will not only advance our understanding of NCDs but also ensure that the benefits of genomic medicine are equitably distributed across all populations . Despite the slow progress, there are promising global efforts aimed at tackling the significant lack of diversity in genomic research. The Human Heredity and Health in Africa (H3Africa) initiative, the largest genomic research consortium in Africa, is spearheading this effort with a 10-year project aimed at studying the genetic basis of disease among African populations and establishing sustainable genomics research across the continent . This initiative includes the creation of three biorepositories in Uganda, South Africa, and Nigeria, and the development of the Pan African Bioinformatics Network (H3ABioNet), supporting advances in handling biological data . In Latin America, the Latin American Genomics Consortium is harmonizing data from existing cohorts and planning new recruitments to build a substantial biobank, addressing the underrepresentation of admixed populations . In the United States, the All of Us Research Program aims to mirror the country's diversity by collecting data from over one million participants, half of whom are of non-European genetic ancestry. This program has identified over 275 million previously unreported genetic variants, with 77% of its participants coming from historically underrepresented communities in biomedical research . Despite these encouraging efforts, the progress is far from sufficient. There is a substantial disparity within continents, particularly in Africa, Latin America, South Asia, and West Asia, where only a few countries have well-established biobanks . This highlights the ongoing need for more comprehensive initiatives to ensure that genetic research benefits all global populations equitably. Another limitation of standard population-based GWAS is the bias arising from population stratification and assortative mating, which can distort the estimated effects of variants on phenotypes [ –. Standard-GWAS results are influenced by several factors, including the direct effects of alleles carried by an individual on their phenotype; the indirect effects of alleles carried by relative(s) through environmental influences (genetic nurture); and confounding due to population stratification and assortative mating. Although methods such as principal-component (PC) analysis and linear mixed models (LMMs) are used to adjust for population stratification, residual confounding often persists in GWAS summary statistics [ –. These biases are particularly pronounced in polygenic scores (PGSs), which aggregate genetic risk information from thousands of variants . Additionally, such biases can also impact post-GWAS analyses, including biological annotation, heritability estimation, genetic correlations, Mendelian Randomization (MR), and GxE interaction . While family-based GWASs typically have lower power than population-based GWASs due to smaller sample sizes, they have been shown to mitigate biases from population stratification effectively . A recent within-family GWAS, conducted on a large sample of siblings, has demonstrated that within-family association estimates are significantly attenuated compared to standard GWAS estimates for traits such as depressive symptoms, height, and smoking . The increasing availability of family-based data offers great potential for disentangling direct and indirect genetic effects affecting NCDs, thereby aiding in unraveling complex GxE interactions. Transcriptomics, through RNA sequencing (RNA-seq) technologies, has become instrumental in elucidating cellular pathways critical to the pathophysiology of many NCDs . By analyzing all RNA transcripts, including coding and non-coding types, RNA-seq provides comprehensive insights into mRNA abundance, alternative splicing, nucleotide variations, and structural alterations . By revealing how gene expression is regulated and altered under various conditions, transcriptomics plays a pivotal role in bridging genotypic variations with phenotypic manifestations. For instance, a study by Romanoski et al. (2010) integrated transcriptomics and genomics to examine human aortic endothelial cells' response to oxidized phospholipids, a key factor in atherosclerosis—a major cause of heart disease . They treated cells with oxidized phospholipids, known to induce vascular inflammation, and simultaneously performed analysis to identify expression quantitative trait loci (eQTLs) influencing gene expression changes. This approach revealed that approximately one-third of the highly regulated transcripts exhibited gene-environment (GxE) interactions, often influenced by distal, trans-acting effects. Some notable interactions were further validated through small interfering RNA (siRNA) knockdown experiments, confirming the significant role of specific genetic loci in modulating gene expression responses to environmental stimuli . This and other related studies [ – illustrate how integrating genomic and transcriptomic data can uncover complex GxE interactions, enhancing our understanding of the genetic and environmental underpinnings of cardiovascular diseases and other NCDs. Building on the capabilities of RNA sequencing technologies, the surge in transcriptomic data has prompted the establishment of comprehensive consortia tasked with managing, curating, and distributing these resources to the broader scientific community. Among the most noteworthy is The Cancer Genome Atlas (TCGA), which provides a rich repository of cancer-related transcriptomic data. Similarly, the Allen Human Brain Atlas offers specialized RNAseq databases focusing on brain diseases, encompassing studies on aging, dementia, and traumatic brain injury. Additionally, repositories such as the Gene Expression Omnibus (GEO), Encyclopedia of DNA Elements (ENCODE) , and the Genotype-Tissue Expression (GTEx ) Project significantly contribute to the availability of transcriptomic data across various tissues and conditions. By providing access to extensive transcriptomic data, these consortia support the ongoing exploration of how gene expression is intricately regulated and modified, thus continuing to bridge genotypic variations with phenotypic manifestations in complex disease research. Epigenomics, which examines the full spectrum of epigenetic modifications such as DNA methylation and histone modification, plays a crucial role in understanding how environmental factors and genetic predispositions interact to influence the development of diseases . These modifications regulate gene expression without altering the DNA sequence and are involved in critical processes like cellular differentiation and tumorigenesis . The epigenome's responsiveness to various environmental exposures—such as metals, air pollution, electromagnetic radiation—and lifestyle factors like diet, smoking, and physical activity, as well as the natural aging process, underscores its dynamic nature . For example, chronic exposure to arsenic and lead is associated with DNA methylation changes that heighten the risk of various cancers . Similarly, prenatal dietary factors like folate intake can alter the epigenome, influencing fetal development and disease susceptibility later in life [ –. Additionally, medications such as sodium valproate (VPA), used for treating epilepsy and bipolar disorder, demonstrate the complexity of interactions between pharmacological treatments and epigenetic regulation by affecting gene expression through their histone deacetylase (HDAC) inhibitor properties . Epigenomics can provide a molecular framework to understand how GxE effects manifest in NCDs. Through studying epigenetic modifications, researchers can discover novel genes and pathways influenced by genetic factors and environmental exposures. The epigenome is partly regulated by the genome, with genetic variation influencing the establishment of DNA methylation marks [ –, while also being highly responsive to environmental factors [ –. This dual regulation highlights the complexity of gene-environment interactions. For instance, the expression of certain NCD risk variants may depend on specific DNA methylation states, which environmental factors can alter. Alternatively, genetic variations might predispose certain epigenomic profiles to respond differently to environmental exposures, thus influencing disease risk . Recent studies, such as those by Teh et al. (2014), have shown that a significant proportion of variably methylated regions, areas where methylation levels vary substantially among individuals, can be attributed to GxE interactions, revealing the intricate molecular mechanisms at play . Advances in next-generation sequencing technologies have significantly enhanced the precision and scope of epigenic profiling . While techniques like bisulfite conversion have been widely used to map methylation, they come with challenges, such as incomplete conversion and DNA degradation. The advent of long-read sequencing technologies, such as PacBio’s HiFi sequencing, has addressed some of these limitations. HiFi sequencing can directly detect 5mC methylation without the need for bisulfite conversion, offering both high accuracy and the ability to resolve methylation profiles alongside phased haplotyping in a single run . This capability significantly improves our understanding of epigenetic modifications linked to genetic variants and environmental factors. On the other hand, major collaborative efforts like the NIH Roadmap Epigenomics project, the International Human Epigenome Consortium (IHEC), and ENCODE project have provided comprehensive maps of the human epigenome. By linking epigenetic changes to functional outcomes, these consortia enhance our understanding of the complex interactions that define health and disease, paving the way for advances in precision medicine. Proteomics explores the entire array of proteins produced or modified by an organism, offering crucial insights into the development and progression of NCDs. The proteome is highly dynamic, exhibiting considerable variability due to processes like alternative splicing, protein modifications, and the complex assembly of proteins into signaling networks . These processes, regulated spatially and temporally, allow proteomics to measure critical changes in amino acid mutations, peptide isoforms, and posttranslational modifications (PTMs) . PTMs like phosphorylation, acetylation, and glycosylation are especially significant, as their dysregulation is often implicated in cancer, cardiovascular diseases, and neurodegenerative disorders . Proteomic profiles also capture responses to environmental stimuli, such as diet , chemical exposure , and smoking , highlighting their value in understanding complex gene-environment or proteome-environment interactions and refining the selection of target genes for further investigation . A hallmark example of proteomics integration with genomic and phenotypic data is the UK Biobank Pharma Proteomics Project (UKB-PPP), a public–private partnership that profiled over 2,900 proteins in plasma samples from over 54,000 participants . This initiative identified over 14,000 protein quantitative trait loci (pQTLs), with 81% being novel. By comparing results from different platforms like Olink and SomaScan and across diverse ancestries, the project underscored the power of multi-level data integration in revealing protein-level differences that influence disease studies, enhancing our understanding of genomic associations and disease mechanisms across populations . Among the most notable associations with NCDs include a strong link between natriuretic peptide B (BNP) and heart failure and inflammatory bowel disease (IBD) associated with higher plasma levels of prostaglandin-H2 D-isomerase . While most analyses focused on participants of European genetic ancestry (n = 34,557), ancestry-specific pQTL studies in African (n = 934), Central/South Asian (n = 920), and other non-European groups revealed unique variants, many of which were absent or rare in Europeans . These findings underscore the importance of expanding proteomic studies to diverse ancestries to capture population-specific genetic and proteomic interactions, addressing disparities in disease risk and treatment. Another UK Biobank study identified over 5,000 associations between rare protein-coding variants and plasma protein abundances, significantly expanding our understanding of how rare variations influence proteomic profiles and highlighting their potential in identifying new therapeutic targets and biomarkers . Despite its promise, proteomics faces challenges in scalability, cost and analytical complexity. High-throughput platforms such as mass spectrometry (MS) and proximity extension assays (PEA) enable precise protein profiling from minimal amounts of biological samples, but they remain costly, limiting their application in large-scale studies . The high dynamic range of protein expression and complexity of many PTMs and sequence variations pose further technical hurdles . Addressing these challenges is crucial for fully leveraging the potential of proteomics for novel biomarker discovery, targeted drug development, and understanding NCD mechanisms. Public–private collaborations like the UKB-PPP highlight the transformative potential of proteomics to bridge the gap between genetics and phenotypes in multi-omics research. By integrating proteomics into population-scale biobanks, researchers can enhance causal gene identification, refine patient stratification, and accelerate therapeutic discovery. However, ensuring equitable applications requires broadening the representation of underrepresented populations and addressing cost barriers. These advancements will enable proteomics to significantly contribute to precision medicine and effective management of NCDs globally. Metabolomics focuses on small-molecule metabolites—such as hormones, amino acids, and lipids—that serve as substrates, intermediates, and products of metabolism, offering a direct window into the biochemical pathways driving complex diseases, including NCDs . Closer to the actual phenotype than mRNA or protein, metabolite levels provide a particularly valuable physiological readout because they integrate environmental and multiple regulatory inputs . Each tissue or cell type has a unique metabolic signature, allowing metabolomics to highlight organ or tissue-specific changes linked to disease . The dynamic nature of the metabolome, highly responsive to factors such as diet and chemical exposure, makes it indispensable for studying gene-environment interactions in NCDs . Integrating metabolomics data with other omics layers, such as genomics and proteomics, enhances our ability to map metabolic pathways, predict metabolite abundances, and identify novel biomarkers and therapeutic targets across diverse populations. Metabolomics enables the quantification of both endogenous metabolites and xenobiotics—foreign substances like environmental chemicals, pollutants, and drugs—offering a comprehensive view of how external exposures impact biological systems . By analyzing these external compounds alongside changes in the endogenous metabolome, metabolomics reveals critical insights into the biological effects of environmental exposures. For example, a study on occupational exposure to trichloroethylene (TCE) identified TCE metabolites in human plasma and linked them to changes in endogenous metabolites associated with immunosuppression, hepatotoxicity, and nephrotoxicity, highlighting the toxic effects of TCE . Similarly, the EXPOsOMICS project explored biofluids and exhaled breath for disinfection by-products (DBPs) from swimming pools, uncovering potential disruptions to metabolites in the tryptophan pathway . In another study, researchers examined the relationship between SNPs in the methionine salvage enzyme APIP and mortality risk in sepsis triggered by infections like Salmonella. By analyzing plasma metabolomic profiles from about 1,000 patients, the study showed that sepsis survivors had significantly lower levels of the enzyme’s substrate, methylthioadenosine, than nonsurvivors, illustrating how genetic variation and metabolite levels jointly influence sepsis outcomes . These examples demonstrate metabolomics’ capacity to unravel gene-environment interactions and the biological consequences of external exposures. A key challenge in metabolomics is the identification and measurement of metabolites, but recent advancements have significantly eased this bottleneck . Technological advances in nuclear magnetic resonance (NMR) and mass spectrometry (MS)-based methods, such as GC–MS and LC–MS, have also improved the precision and range of metabolite quantification. Expanded metabolite databases, such as The Human Metabolome Database (HMDB) and XCMS-METLIN, now contain tens of thousands of metabolites, including xenobiotics from environmental sources. Additionally, various bioinformatics tools now enable more robust analysis, linking metabolic signatures to disease states and outcomes, thus enhancing the potential of metabolomics in NCD research . Exposomics is a burgeoning field that explores the comprehensive impact of environmental factors on human health over an individual's lifetime . This discipline considers various exposures—from chemical and biological agents to psychosocial factors, socioeconomic status and interpersonal relationships. These factors can trigger various biological responses, including changes in gene and protein expression, which in turn may influence the microbiome and epigenome . This complex interplay underscores how environmental factors, intertwined with genetic predispositions, contribute to the development of NCDs (Fig. ). Examples of replicated gene-environment interactions include BRCA-1 associated protein-1 (BAP1) mutations and asbestos exposure for mesothelioma , chromodomain helicase DNA-binding protein 8 ( CHD8 ) and pesticide exposure for autism spectrum disorder , the fat mass and obesity-associated gene ( FTO ) and physical activity for obesity, and dopamine receptor D4 ( DRD4 ) and parenting style for attention-deficit/hyperactivity disorder (ADHD) . These interactions highlight how specific genetic susceptibilities can be activated or exacerbated by environmental factors, demonstrating the crucial role of exposomics in understanding NCDs. A subset of notable GxE interactions implicated in common NCDs is shown in Fig. . While most of these GxE examples have focused on single environmental variables, the broader and more systematic measurement of environmental factors—such as those captured through exposomics—holds tremendous potential for deepening our understanding of complex diseases. Although still in its early stages, a few studies have ventured into multi-exposure genome-wide interaction analysis, jointly modeling the effects of the genome and the environment using methods like StructLMM (structured linear mixed model) , GxEMMs (GxE Mixed Model) , and IGE (integrative analysis of genomic and exposomic data) . These approaches account for genome-exposome correlations and the interrelationships among exposome variables, offering a more holistic view of gene-environment interactions . However, such methods are often computationally intensive and challenging to interpret, underscoring the complexity and potential of exposomics in advancing our understanding of NCDs. Recent technological advancements, especially in high-resolution mass spectrometry (HRMS) and wearable devices, have significantly improved the ability to measure the exposome with precision and individual specificity . HRMS enables the detailed detection of small molecules in biological samples like plasma and urine, allowing for an in-depth analysis of exposures to pharmaceuticals, pollutants, and nutrients . Complementing this, wearable technologies such as silicone wristbands and other personal passive samplers have emerged as powerful tools for capturing personal exposure data. These devices can monitor environmental exposures in real-time across different settings and critical life stages, such as during pregnancy or early childhood, thus providing a dynamic and personalized exposome profile . For instance, studies employing wristband samplers in China have successfully profiled personal chemical exposures, demonstrating the diversity and complexity of environmental interactions individuals face daily . Another innovative approach uses a miniaturized wearable device that samples air to capture particulates, further analyzed using HRMS to identify both hydrophobic and hydrophilic chemical compounds . These studies exemplify how wearables can offer insights into the spatiotemporal dynamics of personal exposures and their potential health impacts. By integrating data from these wearables with systems biology approaches, researchers can now begin to unravel the intricate gene-environment interactions that significantly influence the pathogenesis of NCDs, paving the way for targeted prevention and therapeutic strategies. Furthermore, new initiatives are expanding the scope of exposomics research, leveraging large-scale resources to deepen our understanding of environmental contributions to health. The All of Us Research Program is increasingly integrating environmental exposure data with genomic, clinical, and demographic information from its diverse cohort of participants. By linking geospatial data with exposure estimates from tools like the Environmental Justice Index, the program aims to examine how environmental factors influence disease susceptibility. Similarly, the European Human Exposome Network (EHEN), the world’s largest network of exposome-focused projects, is advancing research on the health impacts of air pollution, noise, chemicals, and urbanization . Together, these efforts are equipping researchers with unparalleled data resources to elucidate the complex interplay between genetic and environmental factors in NCDs (Fig. ). EHRs are digital versions of patients' medical histories, encompassing a broad spectrum of data, including demographics, medical histories, vital signs, laboratory test results, radiology images, diagnoses, treatment procedures, and medications . Longitudinal data available in practice-based EHRs, such as those from chronic disease management clinics, enable researchers to characterize genetic factors with small but reproducible effects on drug outcomes. For instance, the electronic Medical Records and Genomics (eMERGE) network, supported by EHRs, presents a novel opportunity to coordinate such investigative efforts across multiple institutions, facilitating the dissection of GxE interactions. With advancements in healthcare technology, EHRs have expanded to include Personal Health Records (PHRs), which capture out-of-clinic data such as daily behaviors and physiological measurements collected by smart wearable devices. In the realm of precision medicine, EHRs serve as crucial repositories that connect detailed clinical data with genetic profiles from multi-omics studies. This integration offers a holistic view of a patient's health landscape, combining structured data like lab results and diagnosis codes with unstructured data, such as free-text clinical notes. Although rich with information, EHRs present challenges in data heterogeneity, quality, and management, especially given their mix of unstructured and structured formats . These challenges complicate the extraction and analysis of data but are essential to address for leveraging EHRs in enhancing our understanding of NCDs through multi-omics integration. The field of multi-omics integration has rapidly evolved to enhance our understanding of the GxE interactions that underlie NCDs and other complex diseases . Individual omics approaches, such as GWAS, have successfully identified numerous SNPs associated with various NCDs [ –. However, the challenge of uncovering the functional roles of these SNPs, especially those located in non-coding regions, necessitates the integration of genomic data with transcriptomic, proteomic, metabolomic, epigenomic, and other omics datasets . This comprehensive integration is crucial for mapping the flow of genomic information and elucidating the interactive networks essential for the onset and progression of NCDs. New advances in analytical methods and software have significantly improved our ability to integrate data across multiple omics layers, offering a deeper understanding of how genetic variations interact with environmental factors to influence biological pathways and disease outcomes . By synthesizing data from various domains, multi-omics approaches provide powerful tools for elucidating the intricate dynamics of GxE interactions in NCD research, paving the way for targeted interventions and personalized medicine. These integration techniques utilize approaches that often fall under two broad categories: a suite of post-GWAS analyses and machine learning-based methods . Post-GWAS analysis enhances the interpretation of GWAS results by incorporating additional omics data, enriching our understanding of how identified genetic variants influence disease phenotypes. Conversely, machine learning methods utilize algorithms to model complex interactions across different biological layers, offering robust tools for deciphering the intricate dynamics of GxE interactions and advancing NCD research. Post-GWAS multi-omics integration approaches Enrichment-based methods Enrichment-based methods provide a powerful way to integrate GWAS data with additional omics layers, thereby enhancing the understanding of complex GxE interactions that underpin NCDs . These approaches utilize overlap, correlation, or association analysis techniques to identify quantitative trait loci (QTLs) that are significantly associated with molecular features such as gene expression (eQTLs), methylation intensity (meQTLs), and protein levels (pQTLs) . For example, the GTEx , ENCODE , and Roadmap Epigenomics projects have systematically cataloged associations between SNPs and various molecular features, creating valuable research resources. Integration of GWAS significant variants and QTLs is achieved through overlapping or positional mapping with functional annotations, confirmed by statistical tests to ensure enrichment is significant and not due to random chance. For instance, a study on atrial fibrillation utilized an integrative multi-omics approach combining genomics, transcriptomics, and proteomics from human atrial tissues. This cross-sectional study identified the widespread effects of genetic variants on both mRNA and protein expression, pinpointing transcription factor NKX2-5 as a crucial link between a GWAS SNP and atrial fibrillation . Similarly, in schizophrenia, enrichment-based methods revealed that risk loci were associated with meQTLs in fetal brain tissue (most notable associations include, rs2535627-cg11645453 and rs4648845-cg02275930), suggesting that altered DNA methylation may play a role in the disease's pathogenesis . These and numerous other examples illustrate how enrichment methods can reveal the cellular origins and molecular networks of disease mechanisms. However, these enrichment estimates can be biased by factors like linkage disequilibrium and the presence of multiple functional variants . Advanced statistical methods, such as hierarchical Bayesian modeling and permutation tests , are employed to mitigate these biases. These strategies not only aid the functional annotation of genetic data but also the discovery of novel biomarkers, offering insights into the tissues and mechanistic pathways involved in NCDs. Statistical fine-mapping methods Statistical fine-mapping methods crucially enhance the integration of GWAS with various quantitative trait loci (QTLs), aiding in identifying causal variants that may influence NCDs and other complex conditions . These approaches, such as colocalization and Mendelian randomization (MR), are pivotal in determining the specific genetic variants that could contribute to both observed molecular traits and disease phenotypes. Colocalization analysis, often conducted using Bayesian statistical methods among others, evaluates whether a genetic variant (s) can be linked to both a GWAS trait and a molecular QTL. This can highlight potential causal genes and pathways implicated in diseases, exemplified by research in Alzheimer’s disease that linked genetic risk variants with eQTLs affecting novel and known genes . Mendelian randomization uses genetic variants as instrumental variables to explore causal relationships between modifiable molecular traits and disease outcomes, functioning under the strong assumption that the variant influences the disease solely through its effect on an intermediary molecular trait—a premise that is challenging to validate [ –. For example, a study on depression linked genetically regulated brain protein levels to the disease, suggesting causality through MR analysis . Furthermore, the comprehensive review by Markozannes et al. (2022) on cancer risk used MR to validate causal associations, such as the effects of BMI on kidney and endometrial cancers and circulating sex hormones on breast cancer . These robust associations highlight the utility of MR in confirming causal pathways, providing a basis for targeted preventive strategies and therapies. Both colocalization and MR provide insights into potential causal mechanisms, with significant colocalization often implying a causal pathway that might be validated through MR. These methods not only pinpoint underlying genetic interactions but also guide the development of targeted therapies and preventive measures across a spectrum of complex diseases. Recent methodological advances in multi-ancestry fine-mapping strategies have significantly enhanced the ability to identify causal genes, offering insights beyond those provided by single-ancestry approaches . MA-FOCUS (multi-ancestry fine-mapping of causal gene sets), for example, integrates GWAS, eQTL, and LD data from multiple ancestries without assuming shared eQTL architecture . It focuses on consistency in causal genes across populations, improving accuracy in identifying disease-relevant genes for traits like hematopoietic and cardiovascular diseases. Similarly, SuSiEx, building on the single-population framework of Sum of Single Effects (SuSiE), offers a powerful cross-population fine-mapping tool. It integrates data across ancestries, models population-specific allele frequencies and LD patterns, and handles multiple causal variants within genomic regions using GWAS summary statistics . In evaluations involving traits from the UK Biobank and Taiwan Biobank, and a schizophrenia GWAS across East Asian and European ancestries, SuSiEx fine-mapped more association signals, produced smaller credible sets, and achieved higher posterior inclusion probability for causal variants, even capturing population-specific causal variants . Both MA-FOCUS and SuSiEx highlight the critical importance of including genetic data from diverse ancestries to improve the resolution of genetic studies and to uncover more precise therapeutic targets. Imputation-based methods Imputation-based methods are a powerful tool for integrating genomic and multi-omics data, utilizing extensive datasets from sources such as GTEx and ENCODE . These methods depend on a reference panel built from robust genetic prediction models derived from genotype data and molecular measurements (e.g., gene or protein levels) of healthy individuals . These models are crafted using statistical techniques, including LASSO, ridge regression, and elastic net . Through this approach, imputation-based methods impute molecular features within GWAS datasets, enabling the identification of associations between genetically predicted molecular features and various NCDs. Key findings from this method often reveal molecular features that are differentially expressed between cases and controls, thus highlighting potential pathways of disease manifestation. Transcriptome-wide association studies (TWAS) are a common application, successfully identifying molecular features linked to various traits and conditions . Beyond gene expression, imputation-based integration has been adapted to explore other molecular features, such as DNA methylation and protein levels, though these remain less commonly applied compared to TWAS . Furthermore, multi-omics integration approaches often extend beyond the use of GWAS data alone, employing previously described enrichment-based methods to merge findings from different omic layers—such as transcriptomics and epigenomics—through overlap and correlation analyses. These integrated analyses provide deeper insight into the complex interactions during the pathogenesis of NCDs. For instance, in one integrative study, researchers analyzed the relationships among intestinal microbiota, serum metabolome, and inflammatory cytokines in groups with and without schizophrenia . Utilizing weighted gene co-expression network analysis , they identified significant co-abundance clusters of metabolites and gut bacteria, which correlated with cytokine levels. This suggests that specific bacteria could influence inflammatory responses through metabolic modulation . Such integrative studies underscore the potential of using multi-omics data to uncover biological networks involved in NCDs, particularly highlighting pathways such as the gut-brain and gut-immune axes, which are crucial for understanding complex diseases (Additional file 1). AI/machine-learning-based method Machine learning (ML) methods are increasingly used to understand the complex GxE interactions in various NCDs . These methods adeptly handle the integration of noisy, high-dimensional multi-omic datasets, essential for elucidating the multifaceted biological processes underlying NCDs . Several integration strategies have been developed, each tailored to optimize the handling of these complex datasets in different scenarios . Early integration, for example, concatenates datasets sample-wise, creating a comprehensive input matrix for ML models . However, the sheer size and complexity of such integrated data can be challenging for many ML algorithms, especially with smaller sample sizes. To mitigate these challenges, other strategies such as mixed integration —which reduces dataset complexity individually—and intermediate integration—which reduces complexity jointly—are utilized [ –. Late integration, conversely, analyzes each omics dataset independently before aggregating the outputs for a final decision . In contrast, hierarchical integration systematically incorporates known biological regulatory frameworks into the analysis, reflecting the sequences of molecular interactions . The versatility of ML methods facilitates a broad spectrum of applications in multi-omics data integration for NCDs, from diagnostic classification to prognosis prediction and evaluating treatment responses. The forthcoming sections will explore these applications in detail, presenting examples of specific ML frameworks that have shown promise in enhancing our understanding of many NCDs and other complex diseases. By leveraging these advanced ML approaches, researchers can pinpoint potential biomarkers, unravel disease mechanisms, and enhance the personalization of healthcare—key components in advancing the field of precision medicine for chronic diseases. In Additional file 2: Table S2, we provide a non-exhaustive list of multi-omics integration and GxE interaction analyses approaches. Diagnostic classification Diagnostic classification through ML involves accurately grouping patients into predefined classes representing specific disease diagnoses, a process crucial for managing NCDs . In cardiovascular disease (CVD), ML classifiers have been utilized to predict CVD and related risk factors from omics data, illustrating the application's potential. A study by Drouard et al. (2024) compared various ML strategies using blood-derived metabolomics, epigenetics, and transcriptomics data to predict CVD risk factors . The findings revealed that multi-omics predictions generally outperformed single-omics predictions, particularly in distinguishing individuals with extreme levels of CVD risk factors. Techniques like semi-supervised autoencoders, which refine feature representation before classification, demonstrated improved predictive accuracy over unsupervised methods, highlighting the capabilities of ML to enhance diagnostic precision in complex disease settings . In the realm of cancer diagnostics, ML has shown significant promise. For instance, a study by Khadirnaikar et al. (2023) on non-small cell lung cancer (NSCLC) employed ML to identify novel subtypes, enhancing prognostic accuracy and treatment personalization . By applying consensus K-means clustering to multi-omics data, the study identified five distinct NSCLC clusters with varying survival outcomes and genetic characteristics, demonstrating the superior performance of multi-omics over single-omics models. Similarly, a novel approach by Abassi et al. (2024) utilized a combination of ML and deep learning (DL) techniques to improve diagnostic accuracy for leukemia . Employing various ML algorithms and deep learning networks like recurrent neural networks (RNNs), they achieved up to 98% accuracy in predicting leukemia from multi-omics data. This approach not only emphasizes the potential of integrating various omics data for cancer diagnostics but also showcases the efficiency of ML and DL in refining diagnostic classifications across different cancer types (Additional file 3). Similarly, in psychiatric disorders, which share overlapping genetic, environmental risk factors, and symptomatology, ML tools are invaluable for refining diagnosis. Xie et al. (2021) demonstrated this by using gene expression and DNA methylation data to construct models that effectively distinguished patients with major depressive disorder (MDD) from healthy controls . Their approach identified genes that were either upregulated and hypomethylated or downregulated and hypermethylated in MDD patients. Although the gene expression classifier exhibited superior predictive power compared to the DNA methylation classifier, both models underscore the potential of ML in enhancing diagnostic accuracy. Clinical outcome (Risk) prediction Risk prediction is a vital ML application in the multi-omics analysis of NCDs . This approach leverages ML to identify and prioritize molecular features that may forecast an increased risk of diseases. Typically, these features are unearthed through detailed single-omics analyses or ML-based feature selection within advanced integration strategies. Once identified, these molecular characteristics inform the development of statistical models designed to predict individual disease risks. A notable method employed in risk prediction is the generation of polygenic risk scores (PGS), which calculate an individual's disease susceptibility based on quantitative trait loci (QTLs) and other regulatory genetic variants . These scores sum up an individual's risk alleles, each weighted by its effect size derived from GWAS. Techniques such as penalized regression—LASSO, elastic net, ridge regression—and Bayesian methods refine these risk scores, enhancing their predictive accuracy. An innovative adaptation in this domain involves integrating PGS with other omics data, which allows for a more nuanced interpretation of genetic contributions to disease risk . For instance, in a study by Wang et al. (2022), researchers explored the integration of multi-omics data for predicting clinical outcomes in neuroblastoma, a complex cancer . They employed network-based methods, constructing Patient Similarity Networks (PSN) by assessing distances among patients using omics-derived features. Two distinct integration strategies were tested: network-level fusion, using the Similarity Network Fusion algorithm to merge PSNs across various omics types, and feature-level fusion, combining network features from individual PSNs . Their findings highlighted that network-level fusion provided superior performance in integrating diverse omics data, demonstrating the potential of ML to enhance outcome predictions in NCDs through multi-omics integration techniques. Despite these advances, the clinical adoption of such models remains modest, underscoring the ongoing challenges in model validation and generalizability within healthcare settings. Treatment response prediction Predicting treatment responses is another critical application of ML in the context of multi-omics for NCDs . This ML application spans various treatment-response assessment regimes, including pharmacotherapy, psychotherapy, and more, aiming to forecast outcomes such as prognosis, relapse, or therapeutic efficacy . Particularly in chronic diseases where treatment paths can vary significantly among individuals, leveraging multi-omics data can markedly enhance the precision of these predictions. In cancer, where genetic heterogeneity strongly influences treatment outcomes, ML models have shown promise in predicting responses to anticancer drugs. A study by Wang et al. (2022) illustrates this with a deep neural network that integrates multi-omics data—including gene expressions, copy number variations, gene mutations, protein expressions, and metabolomics—from cancer cell lines . The model features innovative components such as a graph embedding layer to incorporate interactome data and an attention layer to prioritize relevant omics features. This approach achieved an impressive R 2 value of 0.90, outperforming standard neural networks in predicting drug responses using data from the Cancer Cell Line Encyclopedia (CCLE) and the Genomics of Drug Sensitivity in Cancer (GDSC). This example underscores the power of ML in harnessing multi-omics data to enhance the personalization of cancer treatments. Another study by Joyce et al. (2021) explored the predictive power of combining genomics and plasma metabolomics to determine the effectiveness of combination pharmacotherapy in treating major depressive disorder (MDD) . They developed two models: one using only metabolomics and another incorporating both metabolomics and genomics. The latter, a multi-omics approach, utilized penalized linear regression and XGBoost algorithms, demonstrating superior predictive performance as evidenced by a higher area under the curve (AUC) compared to the metabolomics-only model. This study underscores the added value of integrating multiple types of omics data to enhance the accuracy of predicting treatment responses. Estimating GxE interactions Unraveling gene-environment (GxE) interactions is crucial for understanding the complex etiology of NCDs. However, analytical tools for GxE interaction analysis remain limited due to challenges posed by high data dimensionality, significant noise, and heterogeneity in genetic and environmental factors across populations, which can obscure true interactions and hinder replicability. Traditional GxE interaction analyses often rely on regression techniques [ –, linking response variables to main genetic and environmental effects and their interactions. These methods face limitations such as stringent requirements to maintain a "main effects, interactions" hierarchy . This hierarchy demands that if an interaction effect is identified, its corresponding main effects must also be considered in the model, which complicates the analysis by imposing additional constraints on variable selection . Moreover, high dimensionality demands multiple comparison adjustments, increasing the risk of Type II errors (failing to detect true effects), and many studies lack sufficient power due to small effect sizes and limited sample sizes . The emergence of large-scale biobanks and observational studies like the UK Biobank , the All of Us Research Program , FinnGen , and initiatives supported by the Barcelona Institute for Global Health (ISGlobal) are helping address sample size limitations by providing extensive genetic and environmental data across diverse populations. Leveraging these rich datasets, researchers have turned to machine learning (ML) and artificial intelligence (AI) approaches to enhance GxE interaction analysis . For instance, Wu et al. (2023) recently developed a novel methodology that leverages deep learning to enhance GxE interaction analysis . This approach integrates deep neural networks with penalization strategies to simultaneously estimate and select significant GxE interactions and corresponding main effects while respecting the required hierarchical structure. Demonstrations through simulation studies and applications in NCD contexts, such as lung adenocarcinoma and skin cutaneous melanoma, show that this method not only manages the complexity of the data but also surpasses traditional regression methods in predictive accuracy and feature selection . Madhukar et al. (2019) also introduced BANDIT, a Bayesian machine-learning approach for drug target identification using diverse data types . BANDIT integrates over 20 million data points from six distinct data types – including drug efficacies, transcriptional responses, drug structures, adverse effects, bioassay results, and known targets – to predict drug-target interactions. Benchmarking showed approximately 90 percent accuracy in correctly identifying known drug targets across over 2,000 small molecules. Applied to compounds without known targets, BANDIT generated novel molecule-target predictions that were experimentally validated, including identifying new microtubule inhibitors effective against resistant cancer cells . Although primarily focused on drug discovery, BANDIT exemplifies how integrating heterogeneous omics data through machine learning can elucidate complex biological interactions, including GxE interactions relevant to NCDs. Similarly, other ML-based methods have shown promise in addressing the complexities of GxE interactions . Zou et al. (2010) introduced a nonparametric Bayesian approach for mapping quantitative trait loci (QTL) that captures both main effects and higher-order interactions, including gene-environment interactions, without requiring explicit specification of interaction terms . This method employs a Gaussian process prior combined with variable selection to identify important genetic and environmental factors. By modeling all potential interactions in a single framework, it avoids the computational and multiple-testing challenges associated with parametric approaches. Applied to the polygenic mouse model of obesity, the method identified key quantitative trait loci (QTLs) influencing fat pad weight and highlighted how nonparametric Bayesian variable selection could improve the detection of GxE interactions in complex traits. Spanbauer et al. (2020) employed a non-parametric machine learning approach using Bayesian additive regression trees with mixed models (mixedBART) for precision medicine. This method adeptly identifies patient characteristics associated with treatment effect heterogeneity in clinical trials . In a study focusing on type II diabetes mellitus among African-American patients, mixedBART predicted individualized treatment effects based on demographic and health measures. While additional analyses showed insufficient evidence for treatment effects, mixedBART facilitated the multi- exploration of treatment heterogeneity, underscoring its potential in GxE interaction studies . In addition, the advent of multimodal medical large language models (LLMs) offers promising avenues for future GxE interaction studies in NCDs. Building on established medical LLMs , several multimodal models such as LLaVA-Med (Large Language and Vision Assistant for BioMedicine) have been proposed . These models are designed to process medical images and generate text-based interpretations, demonstrating medical image understanding and diagnosis capabilities. While current multimodal LLMs primarily handle modalities like text and imaging data, there is growing interest in extending these models to incorporate molecular-level omics data, including genomics. For instance, preliminary efforts like MedGPT have explored analyzing genomic data using LLMs, although they remain at a proof-of-concept stage with preliminary results . As these models evolve and integrate more diverse datasets, they have the potential to enhance our ability to interpret complex biological interactions, including GxE interactions relevant to NCDs. However, significant challenges remain, and more research is needed to fully realize the integration of multimodal omics data in LLMs. In summary, these advancements illustrate the growing role of ML/AI tools in addressing the challenges of GxE interaction analysis in NCDs and other complex diseases. By harnessing large and diverse datasets and employing sophisticated analytical methods, researchers can better understand the complex interplay between multi-omic factors and the exposome. However, applying AI/ML methods in this context also presents challenges. Bias remains a significant concern, as algorithms trained on datasets that underrepresent certain demographic groups can yield skewed predictions, potentially exacerbating existing health disparities among populations affected by NCDs . The “black box” nature of many AI/ML models, particularly deep learning approaches, poses another hurdle, as the lack of interpretability may undermine clinical decision-making and trust, especially when transparent reasoning is crucial for evaluating risk factors or treatment options . Furthermore, the use of sensitive patient data in NCD research heightens the risk of privacy breaches, raising complex ethical and legal challenges in data governance . Overcoming these challenges requires diverse and representative training datasets, the development of interpretable AI models tailored to NCD applications, and robust privacy protections to ensure ethical and equitable use of AI/ML in advancing GxE research and clinical practice. Together, these efforts not only enhance our understanding of disease mechanisms but also contribute to the development of personalized interventions and treatments (Fig. ). Current challenges and opportunities Diversity of omics and multi-omics datasets Despite efforts to diversify genomic datasets, the vast majority of GWAS, about 85% as of 2023, predominantly feature individuals of European genetic ancestry . Progress toward including under-represented populations has been slow, with the share of studies involving these groups either stagnating or even declining in recent years . Although there has been a modest rise in the representation of Asian ancestries, African, Latin American, and Indigenous populations remain markedly underrepresented . This imbalance is compounded by the over-reliance on easily accessible and homogeneous resources like the UK Biobank, which primarily comprises individuals of European ancestry, whereas other ancestry groups often have limited data repositories available . Figure presents the global distribution of total GWAS sample sizes by country, underscoring significant regional disparities. This lack of diversity leads to a substantial problem: PGSs derived from predominantly European datasets show dramatically reduced predictive accuracy when applied to non-European populations . For instance, Martin et al. (2019) reported a decline in PGS accuracy of about 37%, 50%, and 78% for individuals of South Asian, East Asian, and African ancestries, respectively . Further studies, such as those by Privé et al. (2021) and Ding et al. (2023), confirm that PGS accuracy not only diminishes across different ancestries but also varies significantly within them depending on the genetic distance from the European training populations . The limited generalizability of these genetic insights could potentially exacerbate health disparities, underscoring the urgent need to broaden the genetic diversity in omics research to ensure that genomic advancements benefit all populations equitably . Furthermore, increasing the diversity of genomic data not only mitigates disparities but also significantly enhances the fine-mapping of GWAS signals and the identification of target genes . This is crucial for uncovering the genetic mechanisms influencing the development of NCDs and other complex conditions. Underrepresented groups, such as those of African and South Asian ancestries, exhibit higher genetic diversity, which translates into substantial gains in genomic research . Studies incorporating these populations have unearthed population-enriched clinically important variants that were previously undiscovered in predominantly European datasets. For example, research into African genetic ancestry has led to critical insights, including the link between APOL1 variants and chronic kidney disease , the identification of G6PD variants that refine diabetes diagnostics , and loss of function variants in PCSK9 that contribute to lower low-density lipoprotein cholesterol levels—this latter discovery has spurred the development of PCSK9 inhibitor drugs . These findings underscore the value of including diverse genetic backgrounds in research to achieve a comprehensive understanding of genetic factors across all populations, enhancing the overall impact of genomic studies on global health. The lack of genetic diversity is a pervasive issue across various omics datasets, not just genomics . For instance, bulk and single-cell transcriptomic analyses are beginning to uncover significant heterogeneity in gene expression across different cell types and even within the same type. This diversity is especially pronounced across different genetic ancestries, shaped by distinct environmental and genetic interactions. Major research efforts, such as single-cell consortia including KPMP, LungMAP, HTCA, GTEx , HuBMAP, Azimuth, HCA, and the Allen Brain Atlas, have predominantly focused on populations of European genetic ancestry, resulting in the underrepresentation of other groups. For example, of the 4,723 samples analyzed across these consortia, the majority are from individuals of European descent, starkly contrasted with the minimal representation from African, Hispanic, and East Asian ancestries. Addressing this imbalance is critical for enhancing our understanding of context-specific cellular mechanisms and improving the detection and treatment of diseases that vary regionally due to factors like genetic drift and migration. This understanding is particularly vital in pharmacogenomics, where knowing context-specific gene regulation can significantly advance personalized medicine. In a significant step toward addressing this imbalance, the Chan Zuckerberg Initiative has recently funded the Ancestry Networks for the Human Cell Atlas (HCA) with a $28 million grant, supporting the inclusion of ancestrally diverse tissue samples to ensure a broader representation and deeper insights into the genetic underpinnings of health and disease across populations. Similarly, the representation of genetic diversity in epigenomic data is markedly limited , as demonstrated by a study by Breeze et al. (2022). This study revealed that among the 5,048 epigenetic experiments from the US-based ENCODE data and the International Human Epigenome Consortium (IHEC), 87.1% (n = 4,397) predominantly featured samples of European genetic ancestry, with other ancestries severely underrepresented. Such disparities underscore a significant bias in the samples analyzed, with only a fraction representing African, Asian, and other ancestries. This lack of diversity impedes our ability to fully understand and interpret disease-associated genomic regions across populations. Epigenomic markers such as promoters, enhancers, and repressors are crucial for annotating non-coding regions identified by GWAS, which often have unclear functional implications. Broadening the scope of epigenomic data to include diverse populations could enhance the interpretation of GWAS loci, offering vital insights into the regulatory mechanisms affecting diseases that disproportionately impact non-European populations, like prostate cancer, hypertension, and chronic kidney disease. Measuring exposomes Measuring exposomes in multi-omics research on NCDs involves significant challenges due to the complexity and diversity of environmental exposures. Exposomes encompass a range of external factors, like pollution and radiation, alongside internal factors, such as microbiome interactions and metabolic processes. Technologies like mass spectrometry (MS) and geographic information systems (GIS) are essential for quantifying these exposures. MS, particularly untargeted MS, excels in detecting a broad spectrum of small molecules in biological samples, providing a comprehensive snapshot of chemical exposures. However, the vast amount of data generated requires advanced bioinformatics for accurate analysis, and detection sensitivity varies significantly among different chemical classes. GIS tools assess environmental exposure by integrating diverse data sources to model spatial and temporal distribution patterns of factors like air and water quality. This modeling is crucial for evaluating health risks linked to environmental factors. Additionally, wearable sensor technologies revolutionize exposure monitoring by providing real-time, individual exposure data to elements such as air quality and UV radiation, offering granular insights into daily exposure patterns. Despite these advancements, the dynamic nature of environmental exposures and the heterogeneity in measurement techniques pose substantial challenges. These include the need for standardized data collection methods and the development of structured data sharing protocols to facilitate comparisons and enhance the accuracy of exposome research in understanding NCDs. Establishing and maintaining biobanks Establishing and maintaining biobanks is a critical yet challenging endeavor in omics and multi-omics research, particularly in low and middle-income countries (LMICs) . While most biobanks are found in high-income countries, equipped with advanced infrastructure and technical capacity, LMICs face substantial barriers such as inadequate funding, limited institutional capacity, and a shortage of skilled professionals. This disparity is especially pronounced in Africa and South Asia, which are severely underrepresented in genomic research . Most genomic studies in LMICs rely on funding from high-income countries through collaborative efforts, often resulting in research agendas set by external priorities rather than local needs . Significant and sustained investment in biobanking infrastructure in under-represented regions is crucial to address the lack of diversity in omics research. Initiatives like the China Kadoorie Biobank and the South African Human Genome Project provide hopeful examples of how national governments are recognizing the value of omics studies . In addition, further improvement in this field could be achieved by global consortia directing technical and financial resources to build local biobanking capacities in LMICs. This approach not only helps in establishing the necessary infrastructure for sample processing, genotyping, sequencing, and computational analysis, but also facilitates equitable data, ultimately benefiting the global scientific community. For instance, the Africa Wits-INDEPTH Partnership for Genomic Research (AWI-Gen) exemplifies a strategic regional collaboration funded by the NIH . This project has established a cross-sectional population cohort of about 12,000 adults across four African countries, leveraging existing Health and Demographic Surveillance System centers and community engagement to span a wide representation of social and genetic variability . Similarly, initiatives such as the H3Africa, H3Africa Bioinformatics Network(H3AfricaBioNet), and the Data Science for Health Discovery and Innovation in Africa, strategic funding commitments by the NIH, exemplify efforts to bolster genetic research capacity in Africa . However, it is important to note that future funding commitments in genomics will benefit from expansion to broader continental regions to address health problems and capacity-building needs of countries with no pre-existing omics research infrastructure. Another significant hurdle is the lack of expertise for addressing the ethical, legal, and social implications (ELSIs) of multi-omics research, which hinders the conduct of research and efficient sharing of data . To address this, it is essential to create national and local opportunities for advanced training, foster continuous professional development, and develop comprehensive ELSI guidelines that can be integrated into study designs. These measures will ensure that multi-omics research is conducted responsibly and its benefits are equitably shared, maintaining the integrity and relevance of the research. Additionally, promoting workforce diversity in omics research is crucial for building trust and fostering engagement among underrepresented groups. Diverse research teams are more likely to focus on health issues pertinent to their communities, which in turn encourages broader participation and consent in biobank studies. This not only strengthens the relationship between researchers and participants but also enhances the quality of research data, making genomic studies more impactful and relevant across populations . Multi-omics data and integration methods Integrating multi-omics data to unravel complex GxE interactions in NCDs is complicated by diverse data formats and significant preprocessing requirements . The lack of standardized methods for preprocessing and integrating data from various omics platforms often compromises the effectiveness of analyses . Additionally, the integration process is challenged by the "curse of dimensionality." This term describes issues that arise in high-dimensional datasets, where the volume of variables far exceeds the number of samples, leading to data sparsity and inconsistency across samples . This makes it difficult to draw reliable conclusions from the data, emphasizing the need for robust analytical tools and methods that can handle and integrate vast and varied omics data effectively. On the other hand, tissue and cell-type heterogeneity present another significant challenge in multi-omics integration, particularly relevant to studying complex diseases . Different cell types within a single tissue sample can exhibit unique omics profiles, influenced by the tissue's specific section or the physiological condition of an individual . These variations can skew biomarker levels and lead to misleading associations that reflect cellular differences rather than the disease itself. Although statistical methods have been developed to adjust for cell-type heterogeneity, they may not fully account for the true biological variations or might even over-correct them. Ideally, single-cell omics would provide a clearer picture by isolating the profiles of each cell type, but this approach is often impractical due to high costs and material requirements . The challenges of sample heterogeneity and technical artifacts, such as batch effects during sequencing, underscore the complexity of data preprocessing in multi-omics studies. Ensuring consistent data processing and leveraging appropriate statistical controls are crucial for mitigating these issues and enhancing the reliability of multi-omics analyses. Furthermore, while NGS technologies have made sequencing faster and more affordable, they have also introduced challenges such as increased costs for participant recruitment and sample processing, and complexities in data management and storage . Privacy concerns frequently limit data sharing between institutions, sometimes leading to the withdrawal of large datasets from public access due to potential identification risks . Moreover, proprietary standards for biomedical devices and health IT systems hinder seamless data integration across different sources . Addressing these issues requires comprehensive efforts to harmonize data across various healthcare providers and omics modalities, necessitating a collaborative approach from all stakeholders in healthcare and research to enhance real-world evidence-based practices and improve healthcare outcomes. Another multi-omics integration challenge, particularly when applying enrichment-based methods to uncover gene-environment interactions in NCDs, is the potential bias introduced by linkage disequilibrium, colocalization of multiple functional variants, and unaccounted confounders . Fine-mapping and imputation-based methods, which are crucial for developing biomarkers and understanding molecular mechanisms, depend heavily on the accuracy of population-specific linkage disequilibrium matrices . These methods also rely on robust genetic reference models for molecular features such as gene expression or methylation, which are difficult to obtain for features other than gene expression . The variability of QTL architecture across different tissues further complicates these analyses, necessitating careful consideration of tissue relevance to the disease mechanisms under study . Researchers must ensure they are well-versed in the biological assumptions, statistical constraints, and computational demands of the integration tools they choose to employ to enhance the reliability and applicability of their findings in NCD research. Validation of GxE interactions and translational applications Validating GxE interactions identified in human research and translating them into actionable insights remains a critical challenge. Translational studies using model organisms bridge observational findings with mechanistic understanding, allowing researchers to explore how genetic and environmental factors interplay in the development of NCDs . Model organisms such as mice , rats , Drosophila melanogaster , and Caenorhabditis elegans offer controlled environments where genetic and environmental variables can be precisely manipulated. This control facilitates the dissection of complex biological processes that are challenging to study directly in humans due to ethical and practical constraints. Moreover, hypotheses generated from these studies can be tested using human genetic data, improving detection power and enabling a more detailed analysis of subpopulations to understand GxE interactions better. Incorporating functional annotations from resources such as ENCODE, GTEx, and Roadmap Epigenomics further enhances this process by prioritizing candidate variants and regulatory regions for GxE studies, particularly those in non-coding regions often affected by environmental exposures . For example, genetically diverse rodent models like the Collaborative Cross and Diversity Outbred lines , which have high sequence homology with humans, have been instrumental in identifying QTLs and candidate genes involved in GxE interactions relevant to human NCDs. A notable case involves mutations in the tumor suppressor gene BAP1 , which have been linked with increased susceptibility to mesothelioma following asbestos exposure . Exploring how BAP1 mutations interact with asbestos exposure could elucidate key molecular pathways in carcinogenesis, with the potential to inform targeted screening, prevention strategies, and therapies tailored to the underlying mechanisms. Similarly, studies in Drosophila and C. elegans have facilitated high-throughput screening of genetic variants and environmental exposures, uncovering genetic pathways that modulate responses to environmental stressors and offering translational insights about human health . However, the translation of findings from model systems to human populations is not without challenges. While model organisms provide controlled environments, they cannot fully replicate the genetic complexity, environmental diversity, or numerous confounding factors that influence human health. For instance, gene synteny between humans and model organisms often diverges, particularly for non-coding and regulatory regions, limiting the applicability of some findings. Studies such as Seok et al. (2013) have demonstrated that genomic responses in mouse models often poorly mimic human inflammatory diseases, reflecting the inherent differences in gene regulatory networks and physiological responses. Furthermore, humans are exposed to a far more diverse range of environmental factors—such as diet, pollution, and stress—than those typically replicated in model organism studies, which limits the generalizability of findings (Table ). Functional annotations and perturbation studies, conducted in both in vitro and in vivo settings, hold promise for unraveling the complexities of GxE interactions in NCDs and other complex diseases . Functional annotations derived from large-scale projects, such as ENCODE and GTEx, systematically map regulatory elements and link genetic variants to potential functional effects, guiding the identification of candidate variants and regulatory regions . Perturbation studies, including CRISPR-Cas9-based approaches, enable direct testing of causal hypotheses . For example, CRISPR interference (CRISPRi) and activation (CRISPRa) screens in human induced pluripotent stem cell (hiPSC)-derived neurons have identified essential genes for neuronal survival under chronic oxidative stress–a key environmental factor relevant to neurodegenerative diseases–revealing critical mediators like GPX4 and other selenoprotein synthesis genes . Translational applications of GxE analysis have direct implications for personalized medicine and public health interventions . In precision environmental health, identifying how specific genetic variations influence susceptibility to environmental exposures enables the development of tailored interventions. For instance, genetic variation in CYP2D6 may influence susceptibility to Parkinson’s disease through pesticide exposure, with poor metabolizers potentially at greater risk . These findings could inform strategies to reduce exposure in vulnerable populations, though further research is needed for confirmation. The ALDH2*2 variant, common in certain populations, impairs acetaldehyde metabolism and may increase the risk of esophageal cancer with alcohol intake, suggesting the potential for personalized dietary recommendations and targeted prevention strategies in affected populations . In pharmacogenomics, GxE interactions can guide personalized medication regimens to optimize efficacy and minimize adverse effects. Personalized warfarin dosing based on variations in genes like VKOR1 and CYP2C9 has been shown to improve therapeutic outcomes and reduce the risk of bleeding complications . Variants in the TPMT gene necessitate dose adjustments of thiopurine drugs to prevent toxicity in treating conditions like leukemia and autoimmune diseases . CYP2D6 gene variants inform the selection and dosing of antidepressants, enhancing treatment response and reducing side effects . In oncology, identifying BRCA1/2 mutations allows for the use of Poly(ADP-ribose) polymerase (PARP) inhibitors in targeted cancer therapy, while HER2 expression guides the use of trastuzumab in breast cancer treatment, exemplifying how GxE insights contribute to precision medicine . Approaches that integrate biological pathways and regulatory annotations can further enhance the discovery and application of such GxE findings. These translational applications underscore the importance of validating GxE interactions through model organisms and advanced experimental systems. However, significant challenges persist. Limited experimental validation of GxE findings in model organisms and translational settings undermines confidence in the biological mechanisms underlying these interactions, slowing their application to precision medicine and public health interventions . High costs and the technical complexity of integrating environmental monitoring data with omics insights further impede progress. Additionally, the lack of standardized protocols for validating GxE findings, combined with the scarcity of diverse model systems, restricts the development of tailored therapies and prevention strategies . These issues collectively limit the potential of GxE research to address global health disparities effectively, particularly in low-resource settings where both environmental and omics data are underrepresented. Enrichment-based methods Enrichment-based methods provide a powerful way to integrate GWAS data with additional omics layers, thereby enhancing the understanding of complex GxE interactions that underpin NCDs . These approaches utilize overlap, correlation, or association analysis techniques to identify quantitative trait loci (QTLs) that are significantly associated with molecular features such as gene expression (eQTLs), methylation intensity (meQTLs), and protein levels (pQTLs) . For example, the GTEx , ENCODE , and Roadmap Epigenomics projects have systematically cataloged associations between SNPs and various molecular features, creating valuable research resources. Integration of GWAS significant variants and QTLs is achieved through overlapping or positional mapping with functional annotations, confirmed by statistical tests to ensure enrichment is significant and not due to random chance. For instance, a study on atrial fibrillation utilized an integrative multi-omics approach combining genomics, transcriptomics, and proteomics from human atrial tissues. This cross-sectional study identified the widespread effects of genetic variants on both mRNA and protein expression, pinpointing transcription factor NKX2-5 as a crucial link between a GWAS SNP and atrial fibrillation . Similarly, in schizophrenia, enrichment-based methods revealed that risk loci were associated with meQTLs in fetal brain tissue (most notable associations include, rs2535627-cg11645453 and rs4648845-cg02275930), suggesting that altered DNA methylation may play a role in the disease's pathogenesis . These and numerous other examples illustrate how enrichment methods can reveal the cellular origins and molecular networks of disease mechanisms. However, these enrichment estimates can be biased by factors like linkage disequilibrium and the presence of multiple functional variants . Advanced statistical methods, such as hierarchical Bayesian modeling and permutation tests , are employed to mitigate these biases. These strategies not only aid the functional annotation of genetic data but also the discovery of novel biomarkers, offering insights into the tissues and mechanistic pathways involved in NCDs. Statistical fine-mapping methods Statistical fine-mapping methods crucially enhance the integration of GWAS with various quantitative trait loci (QTLs), aiding in identifying causal variants that may influence NCDs and other complex conditions . These approaches, such as colocalization and Mendelian randomization (MR), are pivotal in determining the specific genetic variants that could contribute to both observed molecular traits and disease phenotypes. Colocalization analysis, often conducted using Bayesian statistical methods among others, evaluates whether a genetic variant (s) can be linked to both a GWAS trait and a molecular QTL. This can highlight potential causal genes and pathways implicated in diseases, exemplified by research in Alzheimer’s disease that linked genetic risk variants with eQTLs affecting novel and known genes . Mendelian randomization uses genetic variants as instrumental variables to explore causal relationships between modifiable molecular traits and disease outcomes, functioning under the strong assumption that the variant influences the disease solely through its effect on an intermediary molecular trait—a premise that is challenging to validate [ –. For example, a study on depression linked genetically regulated brain protein levels to the disease, suggesting causality through MR analysis . Furthermore, the comprehensive review by Markozannes et al. (2022) on cancer risk used MR to validate causal associations, such as the effects of BMI on kidney and endometrial cancers and circulating sex hormones on breast cancer . These robust associations highlight the utility of MR in confirming causal pathways, providing a basis for targeted preventive strategies and therapies. Both colocalization and MR provide insights into potential causal mechanisms, with significant colocalization often implying a causal pathway that might be validated through MR. These methods not only pinpoint underlying genetic interactions but also guide the development of targeted therapies and preventive measures across a spectrum of complex diseases. Recent methodological advances in multi-ancestry fine-mapping strategies have significantly enhanced the ability to identify causal genes, offering insights beyond those provided by single-ancestry approaches . MA-FOCUS (multi-ancestry fine-mapping of causal gene sets), for example, integrates GWAS, eQTL, and LD data from multiple ancestries without assuming shared eQTL architecture . It focuses on consistency in causal genes across populations, improving accuracy in identifying disease-relevant genes for traits like hematopoietic and cardiovascular diseases. Similarly, SuSiEx, building on the single-population framework of Sum of Single Effects (SuSiE), offers a powerful cross-population fine-mapping tool. It integrates data across ancestries, models population-specific allele frequencies and LD patterns, and handles multiple causal variants within genomic regions using GWAS summary statistics . In evaluations involving traits from the UK Biobank and Taiwan Biobank, and a schizophrenia GWAS across East Asian and European ancestries, SuSiEx fine-mapped more association signals, produced smaller credible sets, and achieved higher posterior inclusion probability for causal variants, even capturing population-specific causal variants . Both MA-FOCUS and SuSiEx highlight the critical importance of including genetic data from diverse ancestries to improve the resolution of genetic studies and to uncover more precise therapeutic targets. Imputation-based methods Imputation-based methods are a powerful tool for integrating genomic and multi-omics data, utilizing extensive datasets from sources such as GTEx and ENCODE . These methods depend on a reference panel built from robust genetic prediction models derived from genotype data and molecular measurements (e.g., gene or protein levels) of healthy individuals . These models are crafted using statistical techniques, including LASSO, ridge regression, and elastic net . Through this approach, imputation-based methods impute molecular features within GWAS datasets, enabling the identification of associations between genetically predicted molecular features and various NCDs. Key findings from this method often reveal molecular features that are differentially expressed between cases and controls, thus highlighting potential pathways of disease manifestation. Transcriptome-wide association studies (TWAS) are a common application, successfully identifying molecular features linked to various traits and conditions . Beyond gene expression, imputation-based integration has been adapted to explore other molecular features, such as DNA methylation and protein levels, though these remain less commonly applied compared to TWAS . Furthermore, multi-omics integration approaches often extend beyond the use of GWAS data alone, employing previously described enrichment-based methods to merge findings from different omic layers—such as transcriptomics and epigenomics—through overlap and correlation analyses. These integrated analyses provide deeper insight into the complex interactions during the pathogenesis of NCDs. For instance, in one integrative study, researchers analyzed the relationships among intestinal microbiota, serum metabolome, and inflammatory cytokines in groups with and without schizophrenia . Utilizing weighted gene co-expression network analysis , they identified significant co-abundance clusters of metabolites and gut bacteria, which correlated with cytokine levels. This suggests that specific bacteria could influence inflammatory responses through metabolic modulation . Such integrative studies underscore the potential of using multi-omics data to uncover biological networks involved in NCDs, particularly highlighting pathways such as the gut-brain and gut-immune axes, which are crucial for understanding complex diseases (Additional file 1). Enrichment-based methods provide a powerful way to integrate GWAS data with additional omics layers, thereby enhancing the understanding of complex GxE interactions that underpin NCDs . These approaches utilize overlap, correlation, or association analysis techniques to identify quantitative trait loci (QTLs) that are significantly associated with molecular features such as gene expression (eQTLs), methylation intensity (meQTLs), and protein levels (pQTLs) . For example, the GTEx , ENCODE , and Roadmap Epigenomics projects have systematically cataloged associations between SNPs and various molecular features, creating valuable research resources. Integration of GWAS significant variants and QTLs is achieved through overlapping or positional mapping with functional annotations, confirmed by statistical tests to ensure enrichment is significant and not due to random chance. For instance, a study on atrial fibrillation utilized an integrative multi-omics approach combining genomics, transcriptomics, and proteomics from human atrial tissues. This cross-sectional study identified the widespread effects of genetic variants on both mRNA and protein expression, pinpointing transcription factor NKX2-5 as a crucial link between a GWAS SNP and atrial fibrillation . Similarly, in schizophrenia, enrichment-based methods revealed that risk loci were associated with meQTLs in fetal brain tissue (most notable associations include, rs2535627-cg11645453 and rs4648845-cg02275930), suggesting that altered DNA methylation may play a role in the disease's pathogenesis . These and numerous other examples illustrate how enrichment methods can reveal the cellular origins and molecular networks of disease mechanisms. However, these enrichment estimates can be biased by factors like linkage disequilibrium and the presence of multiple functional variants . Advanced statistical methods, such as hierarchical Bayesian modeling and permutation tests , are employed to mitigate these biases. These strategies not only aid the functional annotation of genetic data but also the discovery of novel biomarkers, offering insights into the tissues and mechanistic pathways involved in NCDs. Statistical fine-mapping methods crucially enhance the integration of GWAS with various quantitative trait loci (QTLs), aiding in identifying causal variants that may influence NCDs and other complex conditions . These approaches, such as colocalization and Mendelian randomization (MR), are pivotal in determining the specific genetic variants that could contribute to both observed molecular traits and disease phenotypes. Colocalization analysis, often conducted using Bayesian statistical methods among others, evaluates whether a genetic variant (s) can be linked to both a GWAS trait and a molecular QTL. This can highlight potential causal genes and pathways implicated in diseases, exemplified by research in Alzheimer’s disease that linked genetic risk variants with eQTLs affecting novel and known genes . Mendelian randomization uses genetic variants as instrumental variables to explore causal relationships between modifiable molecular traits and disease outcomes, functioning under the strong assumption that the variant influences the disease solely through its effect on an intermediary molecular trait—a premise that is challenging to validate [ –. For example, a study on depression linked genetically regulated brain protein levels to the disease, suggesting causality through MR analysis . Furthermore, the comprehensive review by Markozannes et al. (2022) on cancer risk used MR to validate causal associations, such as the effects of BMI on kidney and endometrial cancers and circulating sex hormones on breast cancer . These robust associations highlight the utility of MR in confirming causal pathways, providing a basis for targeted preventive strategies and therapies. Both colocalization and MR provide insights into potential causal mechanisms, with significant colocalization often implying a causal pathway that might be validated through MR. These methods not only pinpoint underlying genetic interactions but also guide the development of targeted therapies and preventive measures across a spectrum of complex diseases. Recent methodological advances in multi-ancestry fine-mapping strategies have significantly enhanced the ability to identify causal genes, offering insights beyond those provided by single-ancestry approaches . MA-FOCUS (multi-ancestry fine-mapping of causal gene sets), for example, integrates GWAS, eQTL, and LD data from multiple ancestries without assuming shared eQTL architecture . It focuses on consistency in causal genes across populations, improving accuracy in identifying disease-relevant genes for traits like hematopoietic and cardiovascular diseases. Similarly, SuSiEx, building on the single-population framework of Sum of Single Effects (SuSiE), offers a powerful cross-population fine-mapping tool. It integrates data across ancestries, models population-specific allele frequencies and LD patterns, and handles multiple causal variants within genomic regions using GWAS summary statistics . In evaluations involving traits from the UK Biobank and Taiwan Biobank, and a schizophrenia GWAS across East Asian and European ancestries, SuSiEx fine-mapped more association signals, produced smaller credible sets, and achieved higher posterior inclusion probability for causal variants, even capturing population-specific causal variants . Both MA-FOCUS and SuSiEx highlight the critical importance of including genetic data from diverse ancestries to improve the resolution of genetic studies and to uncover more precise therapeutic targets. Imputation-based methods are a powerful tool for integrating genomic and multi-omics data, utilizing extensive datasets from sources such as GTEx and ENCODE . These methods depend on a reference panel built from robust genetic prediction models derived from genotype data and molecular measurements (e.g., gene or protein levels) of healthy individuals . These models are crafted using statistical techniques, including LASSO, ridge regression, and elastic net . Through this approach, imputation-based methods impute molecular features within GWAS datasets, enabling the identification of associations between genetically predicted molecular features and various NCDs. Key findings from this method often reveal molecular features that are differentially expressed between cases and controls, thus highlighting potential pathways of disease manifestation. Transcriptome-wide association studies (TWAS) are a common application, successfully identifying molecular features linked to various traits and conditions . Beyond gene expression, imputation-based integration has been adapted to explore other molecular features, such as DNA methylation and protein levels, though these remain less commonly applied compared to TWAS . Furthermore, multi-omics integration approaches often extend beyond the use of GWAS data alone, employing previously described enrichment-based methods to merge findings from different omic layers—such as transcriptomics and epigenomics—through overlap and correlation analyses. These integrated analyses provide deeper insight into the complex interactions during the pathogenesis of NCDs. For instance, in one integrative study, researchers analyzed the relationships among intestinal microbiota, serum metabolome, and inflammatory cytokines in groups with and without schizophrenia . Utilizing weighted gene co-expression network analysis , they identified significant co-abundance clusters of metabolites and gut bacteria, which correlated with cytokine levels. This suggests that specific bacteria could influence inflammatory responses through metabolic modulation . Such integrative studies underscore the potential of using multi-omics data to uncover biological networks involved in NCDs, particularly highlighting pathways such as the gut-brain and gut-immune axes, which are crucial for understanding complex diseases (Additional file 1). Machine learning (ML) methods are increasingly used to understand the complex GxE interactions in various NCDs . These methods adeptly handle the integration of noisy, high-dimensional multi-omic datasets, essential for elucidating the multifaceted biological processes underlying NCDs . Several integration strategies have been developed, each tailored to optimize the handling of these complex datasets in different scenarios . Early integration, for example, concatenates datasets sample-wise, creating a comprehensive input matrix for ML models . However, the sheer size and complexity of such integrated data can be challenging for many ML algorithms, especially with smaller sample sizes. To mitigate these challenges, other strategies such as mixed integration —which reduces dataset complexity individually—and intermediate integration—which reduces complexity jointly—are utilized [ –. Late integration, conversely, analyzes each omics dataset independently before aggregating the outputs for a final decision . In contrast, hierarchical integration systematically incorporates known biological regulatory frameworks into the analysis, reflecting the sequences of molecular interactions . The versatility of ML methods facilitates a broad spectrum of applications in multi-omics data integration for NCDs, from diagnostic classification to prognosis prediction and evaluating treatment responses. The forthcoming sections will explore these applications in detail, presenting examples of specific ML frameworks that have shown promise in enhancing our understanding of many NCDs and other complex diseases. By leveraging these advanced ML approaches, researchers can pinpoint potential biomarkers, unravel disease mechanisms, and enhance the personalization of healthcare—key components in advancing the field of precision medicine for chronic diseases. In Additional file 2: Table S2, we provide a non-exhaustive list of multi-omics integration and GxE interaction analyses approaches. Diagnostic classification Diagnostic classification through ML involves accurately grouping patients into predefined classes representing specific disease diagnoses, a process crucial for managing NCDs . In cardiovascular disease (CVD), ML classifiers have been utilized to predict CVD and related risk factors from omics data, illustrating the application's potential. A study by Drouard et al. (2024) compared various ML strategies using blood-derived metabolomics, epigenetics, and transcriptomics data to predict CVD risk factors . The findings revealed that multi-omics predictions generally outperformed single-omics predictions, particularly in distinguishing individuals with extreme levels of CVD risk factors. Techniques like semi-supervised autoencoders, which refine feature representation before classification, demonstrated improved predictive accuracy over unsupervised methods, highlighting the capabilities of ML to enhance diagnostic precision in complex disease settings . In the realm of cancer diagnostics, ML has shown significant promise. For instance, a study by Khadirnaikar et al. (2023) on non-small cell lung cancer (NSCLC) employed ML to identify novel subtypes, enhancing prognostic accuracy and treatment personalization . By applying consensus K-means clustering to multi-omics data, the study identified five distinct NSCLC clusters with varying survival outcomes and genetic characteristics, demonstrating the superior performance of multi-omics over single-omics models. Similarly, a novel approach by Abassi et al. (2024) utilized a combination of ML and deep learning (DL) techniques to improve diagnostic accuracy for leukemia . Employing various ML algorithms and deep learning networks like recurrent neural networks (RNNs), they achieved up to 98% accuracy in predicting leukemia from multi-omics data. This approach not only emphasizes the potential of integrating various omics data for cancer diagnostics but also showcases the efficiency of ML and DL in refining diagnostic classifications across different cancer types (Additional file 3). Similarly, in psychiatric disorders, which share overlapping genetic, environmental risk factors, and symptomatology, ML tools are invaluable for refining diagnosis. Xie et al. (2021) demonstrated this by using gene expression and DNA methylation data to construct models that effectively distinguished patients with major depressive disorder (MDD) from healthy controls . Their approach identified genes that were either upregulated and hypomethylated or downregulated and hypermethylated in MDD patients. Although the gene expression classifier exhibited superior predictive power compared to the DNA methylation classifier, both models underscore the potential of ML in enhancing diagnostic accuracy. Clinical outcome (Risk) prediction Risk prediction is a vital ML application in the multi-omics analysis of NCDs . This approach leverages ML to identify and prioritize molecular features that may forecast an increased risk of diseases. Typically, these features are unearthed through detailed single-omics analyses or ML-based feature selection within advanced integration strategies. Once identified, these molecular characteristics inform the development of statistical models designed to predict individual disease risks. A notable method employed in risk prediction is the generation of polygenic risk scores (PGS), which calculate an individual's disease susceptibility based on quantitative trait loci (QTLs) and other regulatory genetic variants . These scores sum up an individual's risk alleles, each weighted by its effect size derived from GWAS. Techniques such as penalized regression—LASSO, elastic net, ridge regression—and Bayesian methods refine these risk scores, enhancing their predictive accuracy. An innovative adaptation in this domain involves integrating PGS with other omics data, which allows for a more nuanced interpretation of genetic contributions to disease risk . For instance, in a study by Wang et al. (2022), researchers explored the integration of multi-omics data for predicting clinical outcomes in neuroblastoma, a complex cancer . They employed network-based methods, constructing Patient Similarity Networks (PSN) by assessing distances among patients using omics-derived features. Two distinct integration strategies were tested: network-level fusion, using the Similarity Network Fusion algorithm to merge PSNs across various omics types, and feature-level fusion, combining network features from individual PSNs . Their findings highlighted that network-level fusion provided superior performance in integrating diverse omics data, demonstrating the potential of ML to enhance outcome predictions in NCDs through multi-omics integration techniques. Despite these advances, the clinical adoption of such models remains modest, underscoring the ongoing challenges in model validation and generalizability within healthcare settings. Treatment response prediction Predicting treatment responses is another critical application of ML in the context of multi-omics for NCDs . This ML application spans various treatment-response assessment regimes, including pharmacotherapy, psychotherapy, and more, aiming to forecast outcomes such as prognosis, relapse, or therapeutic efficacy . Particularly in chronic diseases where treatment paths can vary significantly among individuals, leveraging multi-omics data can markedly enhance the precision of these predictions. In cancer, where genetic heterogeneity strongly influences treatment outcomes, ML models have shown promise in predicting responses to anticancer drugs. A study by Wang et al. (2022) illustrates this with a deep neural network that integrates multi-omics data—including gene expressions, copy number variations, gene mutations, protein expressions, and metabolomics—from cancer cell lines . The model features innovative components such as a graph embedding layer to incorporate interactome data and an attention layer to prioritize relevant omics features. This approach achieved an impressive R 2 value of 0.90, outperforming standard neural networks in predicting drug responses using data from the Cancer Cell Line Encyclopedia (CCLE) and the Genomics of Drug Sensitivity in Cancer (GDSC). This example underscores the power of ML in harnessing multi-omics data to enhance the personalization of cancer treatments. Another study by Joyce et al. (2021) explored the predictive power of combining genomics and plasma metabolomics to determine the effectiveness of combination pharmacotherapy in treating major depressive disorder (MDD) . They developed two models: one using only metabolomics and another incorporating both metabolomics and genomics. The latter, a multi-omics approach, utilized penalized linear regression and XGBoost algorithms, demonstrating superior predictive performance as evidenced by a higher area under the curve (AUC) compared to the metabolomics-only model. This study underscores the added value of integrating multiple types of omics data to enhance the accuracy of predicting treatment responses. Estimating GxE interactions Unraveling gene-environment (GxE) interactions is crucial for understanding the complex etiology of NCDs. However, analytical tools for GxE interaction analysis remain limited due to challenges posed by high data dimensionality, significant noise, and heterogeneity in genetic and environmental factors across populations, which can obscure true interactions and hinder replicability. Traditional GxE interaction analyses often rely on regression techniques [ –, linking response variables to main genetic and environmental effects and their interactions. These methods face limitations such as stringent requirements to maintain a "main effects, interactions" hierarchy . This hierarchy demands that if an interaction effect is identified, its corresponding main effects must also be considered in the model, which complicates the analysis by imposing additional constraints on variable selection . Moreover, high dimensionality demands multiple comparison adjustments, increasing the risk of Type II errors (failing to detect true effects), and many studies lack sufficient power due to small effect sizes and limited sample sizes . The emergence of large-scale biobanks and observational studies like the UK Biobank , the All of Us Research Program , FinnGen , and initiatives supported by the Barcelona Institute for Global Health (ISGlobal) are helping address sample size limitations by providing extensive genetic and environmental data across diverse populations. Leveraging these rich datasets, researchers have turned to machine learning (ML) and artificial intelligence (AI) approaches to enhance GxE interaction analysis . For instance, Wu et al. (2023) recently developed a novel methodology that leverages deep learning to enhance GxE interaction analysis . This approach integrates deep neural networks with penalization strategies to simultaneously estimate and select significant GxE interactions and corresponding main effects while respecting the required hierarchical structure. Demonstrations through simulation studies and applications in NCD contexts, such as lung adenocarcinoma and skin cutaneous melanoma, show that this method not only manages the complexity of the data but also surpasses traditional regression methods in predictive accuracy and feature selection . Madhukar et al. (2019) also introduced BANDIT, a Bayesian machine-learning approach for drug target identification using diverse data types . BANDIT integrates over 20 million data points from six distinct data types – including drug efficacies, transcriptional responses, drug structures, adverse effects, bioassay results, and known targets – to predict drug-target interactions. Benchmarking showed approximately 90 percent accuracy in correctly identifying known drug targets across over 2,000 small molecules. Applied to compounds without known targets, BANDIT generated novel molecule-target predictions that were experimentally validated, including identifying new microtubule inhibitors effective against resistant cancer cells . Although primarily focused on drug discovery, BANDIT exemplifies how integrating heterogeneous omics data through machine learning can elucidate complex biological interactions, including GxE interactions relevant to NCDs. Similarly, other ML-based methods have shown promise in addressing the complexities of GxE interactions . Zou et al. (2010) introduced a nonparametric Bayesian approach for mapping quantitative trait loci (QTL) that captures both main effects and higher-order interactions, including gene-environment interactions, without requiring explicit specification of interaction terms . This method employs a Gaussian process prior combined with variable selection to identify important genetic and environmental factors. By modeling all potential interactions in a single framework, it avoids the computational and multiple-testing challenges associated with parametric approaches. Applied to the polygenic mouse model of obesity, the method identified key quantitative trait loci (QTLs) influencing fat pad weight and highlighted how nonparametric Bayesian variable selection could improve the detection of GxE interactions in complex traits. Spanbauer et al. (2020) employed a non-parametric machine learning approach using Bayesian additive regression trees with mixed models (mixedBART) for precision medicine. This method adeptly identifies patient characteristics associated with treatment effect heterogeneity in clinical trials . In a study focusing on type II diabetes mellitus among African-American patients, mixedBART predicted individualized treatment effects based on demographic and health measures. While additional analyses showed insufficient evidence for treatment effects, mixedBART facilitated the multi- exploration of treatment heterogeneity, underscoring its potential in GxE interaction studies . In addition, the advent of multimodal medical large language models (LLMs) offers promising avenues for future GxE interaction studies in NCDs. Building on established medical LLMs , several multimodal models such as LLaVA-Med (Large Language and Vision Assistant for BioMedicine) have been proposed . These models are designed to process medical images and generate text-based interpretations, demonstrating medical image understanding and diagnosis capabilities. While current multimodal LLMs primarily handle modalities like text and imaging data, there is growing interest in extending these models to incorporate molecular-level omics data, including genomics. For instance, preliminary efforts like MedGPT have explored analyzing genomic data using LLMs, although they remain at a proof-of-concept stage with preliminary results . As these models evolve and integrate more diverse datasets, they have the potential to enhance our ability to interpret complex biological interactions, including GxE interactions relevant to NCDs. However, significant challenges remain, and more research is needed to fully realize the integration of multimodal omics data in LLMs. In summary, these advancements illustrate the growing role of ML/AI tools in addressing the challenges of GxE interaction analysis in NCDs and other complex diseases. By harnessing large and diverse datasets and employing sophisticated analytical methods, researchers can better understand the complex interplay between multi-omic factors and the exposome. However, applying AI/ML methods in this context also presents challenges. Bias remains a significant concern, as algorithms trained on datasets that underrepresent certain demographic groups can yield skewed predictions, potentially exacerbating existing health disparities among populations affected by NCDs . The “black box” nature of many AI/ML models, particularly deep learning approaches, poses another hurdle, as the lack of interpretability may undermine clinical decision-making and trust, especially when transparent reasoning is crucial for evaluating risk factors or treatment options . Furthermore, the use of sensitive patient data in NCD research heightens the risk of privacy breaches, raising complex ethical and legal challenges in data governance . Overcoming these challenges requires diverse and representative training datasets, the development of interpretable AI models tailored to NCD applications, and robust privacy protections to ensure ethical and equitable use of AI/ML in advancing GxE research and clinical practice. Together, these efforts not only enhance our understanding of disease mechanisms but also contribute to the development of personalized interventions and treatments (Fig. ). Diagnostic classification through ML involves accurately grouping patients into predefined classes representing specific disease diagnoses, a process crucial for managing NCDs . In cardiovascular disease (CVD), ML classifiers have been utilized to predict CVD and related risk factors from omics data, illustrating the application's potential. A study by Drouard et al. (2024) compared various ML strategies using blood-derived metabolomics, epigenetics, and transcriptomics data to predict CVD risk factors . The findings revealed that multi-omics predictions generally outperformed single-omics predictions, particularly in distinguishing individuals with extreme levels of CVD risk factors. Techniques like semi-supervised autoencoders, which refine feature representation before classification, demonstrated improved predictive accuracy over unsupervised methods, highlighting the capabilities of ML to enhance diagnostic precision in complex disease settings . In the realm of cancer diagnostics, ML has shown significant promise. For instance, a study by Khadirnaikar et al. (2023) on non-small cell lung cancer (NSCLC) employed ML to identify novel subtypes, enhancing prognostic accuracy and treatment personalization . By applying consensus K-means clustering to multi-omics data, the study identified five distinct NSCLC clusters with varying survival outcomes and genetic characteristics, demonstrating the superior performance of multi-omics over single-omics models. Similarly, a novel approach by Abassi et al. (2024) utilized a combination of ML and deep learning (DL) techniques to improve diagnostic accuracy for leukemia . Employing various ML algorithms and deep learning networks like recurrent neural networks (RNNs), they achieved up to 98% accuracy in predicting leukemia from multi-omics data. This approach not only emphasizes the potential of integrating various omics data for cancer diagnostics but also showcases the efficiency of ML and DL in refining diagnostic classifications across different cancer types (Additional file 3). Similarly, in psychiatric disorders, which share overlapping genetic, environmental risk factors, and symptomatology, ML tools are invaluable for refining diagnosis. Xie et al. (2021) demonstrated this by using gene expression and DNA methylation data to construct models that effectively distinguished patients with major depressive disorder (MDD) from healthy controls . Their approach identified genes that were either upregulated and hypomethylated or downregulated and hypermethylated in MDD patients. Although the gene expression classifier exhibited superior predictive power compared to the DNA methylation classifier, both models underscore the potential of ML in enhancing diagnostic accuracy. Risk prediction is a vital ML application in the multi-omics analysis of NCDs . This approach leverages ML to identify and prioritize molecular features that may forecast an increased risk of diseases. Typically, these features are unearthed through detailed single-omics analyses or ML-based feature selection within advanced integration strategies. Once identified, these molecular characteristics inform the development of statistical models designed to predict individual disease risks. A notable method employed in risk prediction is the generation of polygenic risk scores (PGS), which calculate an individual's disease susceptibility based on quantitative trait loci (QTLs) and other regulatory genetic variants . These scores sum up an individual's risk alleles, each weighted by its effect size derived from GWAS. Techniques such as penalized regression—LASSO, elastic net, ridge regression—and Bayesian methods refine these risk scores, enhancing their predictive accuracy. An innovative adaptation in this domain involves integrating PGS with other omics data, which allows for a more nuanced interpretation of genetic contributions to disease risk . For instance, in a study by Wang et al. (2022), researchers explored the integration of multi-omics data for predicting clinical outcomes in neuroblastoma, a complex cancer . They employed network-based methods, constructing Patient Similarity Networks (PSN) by assessing distances among patients using omics-derived features. Two distinct integration strategies were tested: network-level fusion, using the Similarity Network Fusion algorithm to merge PSNs across various omics types, and feature-level fusion, combining network features from individual PSNs . Their findings highlighted that network-level fusion provided superior performance in integrating diverse omics data, demonstrating the potential of ML to enhance outcome predictions in NCDs through multi-omics integration techniques. Despite these advances, the clinical adoption of such models remains modest, underscoring the ongoing challenges in model validation and generalizability within healthcare settings. Predicting treatment responses is another critical application of ML in the context of multi-omics for NCDs . This ML application spans various treatment-response assessment regimes, including pharmacotherapy, psychotherapy, and more, aiming to forecast outcomes such as prognosis, relapse, or therapeutic efficacy . Particularly in chronic diseases where treatment paths can vary significantly among individuals, leveraging multi-omics data can markedly enhance the precision of these predictions. In cancer, where genetic heterogeneity strongly influences treatment outcomes, ML models have shown promise in predicting responses to anticancer drugs. A study by Wang et al. (2022) illustrates this with a deep neural network that integrates multi-omics data—including gene expressions, copy number variations, gene mutations, protein expressions, and metabolomics—from cancer cell lines . The model features innovative components such as a graph embedding layer to incorporate interactome data and an attention layer to prioritize relevant omics features. This approach achieved an impressive R 2 value of 0.90, outperforming standard neural networks in predicting drug responses using data from the Cancer Cell Line Encyclopedia (CCLE) and the Genomics of Drug Sensitivity in Cancer (GDSC). This example underscores the power of ML in harnessing multi-omics data to enhance the personalization of cancer treatments. Another study by Joyce et al. (2021) explored the predictive power of combining genomics and plasma metabolomics to determine the effectiveness of combination pharmacotherapy in treating major depressive disorder (MDD) . They developed two models: one using only metabolomics and another incorporating both metabolomics and genomics. The latter, a multi-omics approach, utilized penalized linear regression and XGBoost algorithms, demonstrating superior predictive performance as evidenced by a higher area under the curve (AUC) compared to the metabolomics-only model. This study underscores the added value of integrating multiple types of omics data to enhance the accuracy of predicting treatment responses. Unraveling gene-environment (GxE) interactions is crucial for understanding the complex etiology of NCDs. However, analytical tools for GxE interaction analysis remain limited due to challenges posed by high data dimensionality, significant noise, and heterogeneity in genetic and environmental factors across populations, which can obscure true interactions and hinder replicability. Traditional GxE interaction analyses often rely on regression techniques [ –, linking response variables to main genetic and environmental effects and their interactions. These methods face limitations such as stringent requirements to maintain a "main effects, interactions" hierarchy . This hierarchy demands that if an interaction effect is identified, its corresponding main effects must also be considered in the model, which complicates the analysis by imposing additional constraints on variable selection . Moreover, high dimensionality demands multiple comparison adjustments, increasing the risk of Type II errors (failing to detect true effects), and many studies lack sufficient power due to small effect sizes and limited sample sizes . The emergence of large-scale biobanks and observational studies like the UK Biobank , the All of Us Research Program , FinnGen , and initiatives supported by the Barcelona Institute for Global Health (ISGlobal) are helping address sample size limitations by providing extensive genetic and environmental data across diverse populations. Leveraging these rich datasets, researchers have turned to machine learning (ML) and artificial intelligence (AI) approaches to enhance GxE interaction analysis . For instance, Wu et al. (2023) recently developed a novel methodology that leverages deep learning to enhance GxE interaction analysis . This approach integrates deep neural networks with penalization strategies to simultaneously estimate and select significant GxE interactions and corresponding main effects while respecting the required hierarchical structure. Demonstrations through simulation studies and applications in NCD contexts, such as lung adenocarcinoma and skin cutaneous melanoma, show that this method not only manages the complexity of the data but also surpasses traditional regression methods in predictive accuracy and feature selection . Madhukar et al. (2019) also introduced BANDIT, a Bayesian machine-learning approach for drug target identification using diverse data types . BANDIT integrates over 20 million data points from six distinct data types – including drug efficacies, transcriptional responses, drug structures, adverse effects, bioassay results, and known targets – to predict drug-target interactions. Benchmarking showed approximately 90 percent accuracy in correctly identifying known drug targets across over 2,000 small molecules. Applied to compounds without known targets, BANDIT generated novel molecule-target predictions that were experimentally validated, including identifying new microtubule inhibitors effective against resistant cancer cells . Although primarily focused on drug discovery, BANDIT exemplifies how integrating heterogeneous omics data through machine learning can elucidate complex biological interactions, including GxE interactions relevant to NCDs. Similarly, other ML-based methods have shown promise in addressing the complexities of GxE interactions . Zou et al. (2010) introduced a nonparametric Bayesian approach for mapping quantitative trait loci (QTL) that captures both main effects and higher-order interactions, including gene-environment interactions, without requiring explicit specification of interaction terms . This method employs a Gaussian process prior combined with variable selection to identify important genetic and environmental factors. By modeling all potential interactions in a single framework, it avoids the computational and multiple-testing challenges associated with parametric approaches. Applied to the polygenic mouse model of obesity, the method identified key quantitative trait loci (QTLs) influencing fat pad weight and highlighted how nonparametric Bayesian variable selection could improve the detection of GxE interactions in complex traits. Spanbauer et al. (2020) employed a non-parametric machine learning approach using Bayesian additive regression trees with mixed models (mixedBART) for precision medicine. This method adeptly identifies patient characteristics associated with treatment effect heterogeneity in clinical trials . In a study focusing on type II diabetes mellitus among African-American patients, mixedBART predicted individualized treatment effects based on demographic and health measures. While additional analyses showed insufficient evidence for treatment effects, mixedBART facilitated the multi- exploration of treatment heterogeneity, underscoring its potential in GxE interaction studies . In addition, the advent of multimodal medical large language models (LLMs) offers promising avenues for future GxE interaction studies in NCDs. Building on established medical LLMs , several multimodal models such as LLaVA-Med (Large Language and Vision Assistant for BioMedicine) have been proposed . These models are designed to process medical images and generate text-based interpretations, demonstrating medical image understanding and diagnosis capabilities. While current multimodal LLMs primarily handle modalities like text and imaging data, there is growing interest in extending these models to incorporate molecular-level omics data, including genomics. For instance, preliminary efforts like MedGPT have explored analyzing genomic data using LLMs, although they remain at a proof-of-concept stage with preliminary results . As these models evolve and integrate more diverse datasets, they have the potential to enhance our ability to interpret complex biological interactions, including GxE interactions relevant to NCDs. However, significant challenges remain, and more research is needed to fully realize the integration of multimodal omics data in LLMs. In summary, these advancements illustrate the growing role of ML/AI tools in addressing the challenges of GxE interaction analysis in NCDs and other complex diseases. By harnessing large and diverse datasets and employing sophisticated analytical methods, researchers can better understand the complex interplay between multi-omic factors and the exposome. However, applying AI/ML methods in this context also presents challenges. Bias remains a significant concern, as algorithms trained on datasets that underrepresent certain demographic groups can yield skewed predictions, potentially exacerbating existing health disparities among populations affected by NCDs . The “black box” nature of many AI/ML models, particularly deep learning approaches, poses another hurdle, as the lack of interpretability may undermine clinical decision-making and trust, especially when transparent reasoning is crucial for evaluating risk factors or treatment options . Furthermore, the use of sensitive patient data in NCD research heightens the risk of privacy breaches, raising complex ethical and legal challenges in data governance . Overcoming these challenges requires diverse and representative training datasets, the development of interpretable AI models tailored to NCD applications, and robust privacy protections to ensure ethical and equitable use of AI/ML in advancing GxE research and clinical practice. Together, these efforts not only enhance our understanding of disease mechanisms but also contribute to the development of personalized interventions and treatments (Fig. ). Diversity of omics and multi-omics datasets Despite efforts to diversify genomic datasets, the vast majority of GWAS, about 85% as of 2023, predominantly feature individuals of European genetic ancestry . Progress toward including under-represented populations has been slow, with the share of studies involving these groups either stagnating or even declining in recent years . Although there has been a modest rise in the representation of Asian ancestries, African, Latin American, and Indigenous populations remain markedly underrepresented . This imbalance is compounded by the over-reliance on easily accessible and homogeneous resources like the UK Biobank, which primarily comprises individuals of European ancestry, whereas other ancestry groups often have limited data repositories available . Figure presents the global distribution of total GWAS sample sizes by country, underscoring significant regional disparities. This lack of diversity leads to a substantial problem: PGSs derived from predominantly European datasets show dramatically reduced predictive accuracy when applied to non-European populations . For instance, Martin et al. (2019) reported a decline in PGS accuracy of about 37%, 50%, and 78% for individuals of South Asian, East Asian, and African ancestries, respectively . Further studies, such as those by Privé et al. (2021) and Ding et al. (2023), confirm that PGS accuracy not only diminishes across different ancestries but also varies significantly within them depending on the genetic distance from the European training populations . The limited generalizability of these genetic insights could potentially exacerbate health disparities, underscoring the urgent need to broaden the genetic diversity in omics research to ensure that genomic advancements benefit all populations equitably . Furthermore, increasing the diversity of genomic data not only mitigates disparities but also significantly enhances the fine-mapping of GWAS signals and the identification of target genes . This is crucial for uncovering the genetic mechanisms influencing the development of NCDs and other complex conditions. Underrepresented groups, such as those of African and South Asian ancestries, exhibit higher genetic diversity, which translates into substantial gains in genomic research . Studies incorporating these populations have unearthed population-enriched clinically important variants that were previously undiscovered in predominantly European datasets. For example, research into African genetic ancestry has led to critical insights, including the link between APOL1 variants and chronic kidney disease , the identification of G6PD variants that refine diabetes diagnostics , and loss of function variants in PCSK9 that contribute to lower low-density lipoprotein cholesterol levels—this latter discovery has spurred the development of PCSK9 inhibitor drugs . These findings underscore the value of including diverse genetic backgrounds in research to achieve a comprehensive understanding of genetic factors across all populations, enhancing the overall impact of genomic studies on global health. The lack of genetic diversity is a pervasive issue across various omics datasets, not just genomics . For instance, bulk and single-cell transcriptomic analyses are beginning to uncover significant heterogeneity in gene expression across different cell types and even within the same type. This diversity is especially pronounced across different genetic ancestries, shaped by distinct environmental and genetic interactions. Major research efforts, such as single-cell consortia including KPMP, LungMAP, HTCA, GTEx , HuBMAP, Azimuth, HCA, and the Allen Brain Atlas, have predominantly focused on populations of European genetic ancestry, resulting in the underrepresentation of other groups. For example, of the 4,723 samples analyzed across these consortia, the majority are from individuals of European descent, starkly contrasted with the minimal representation from African, Hispanic, and East Asian ancestries. Addressing this imbalance is critical for enhancing our understanding of context-specific cellular mechanisms and improving the detection and treatment of diseases that vary regionally due to factors like genetic drift and migration. This understanding is particularly vital in pharmacogenomics, where knowing context-specific gene regulation can significantly advance personalized medicine. In a significant step toward addressing this imbalance, the Chan Zuckerberg Initiative has recently funded the Ancestry Networks for the Human Cell Atlas (HCA) with a $28 million grant, supporting the inclusion of ancestrally diverse tissue samples to ensure a broader representation and deeper insights into the genetic underpinnings of health and disease across populations. Similarly, the representation of genetic diversity in epigenomic data is markedly limited , as demonstrated by a study by Breeze et al. (2022). This study revealed that among the 5,048 epigenetic experiments from the US-based ENCODE data and the International Human Epigenome Consortium (IHEC), 87.1% (n = 4,397) predominantly featured samples of European genetic ancestry, with other ancestries severely underrepresented. Such disparities underscore a significant bias in the samples analyzed, with only a fraction representing African, Asian, and other ancestries. This lack of diversity impedes our ability to fully understand and interpret disease-associated genomic regions across populations. Epigenomic markers such as promoters, enhancers, and repressors are crucial for annotating non-coding regions identified by GWAS, which often have unclear functional implications. Broadening the scope of epigenomic data to include diverse populations could enhance the interpretation of GWAS loci, offering vital insights into the regulatory mechanisms affecting diseases that disproportionately impact non-European populations, like prostate cancer, hypertension, and chronic kidney disease. Measuring exposomes Measuring exposomes in multi-omics research on NCDs involves significant challenges due to the complexity and diversity of environmental exposures. Exposomes encompass a range of external factors, like pollution and radiation, alongside internal factors, such as microbiome interactions and metabolic processes. Technologies like mass spectrometry (MS) and geographic information systems (GIS) are essential for quantifying these exposures. MS, particularly untargeted MS, excels in detecting a broad spectrum of small molecules in biological samples, providing a comprehensive snapshot of chemical exposures. However, the vast amount of data generated requires advanced bioinformatics for accurate analysis, and detection sensitivity varies significantly among different chemical classes. GIS tools assess environmental exposure by integrating diverse data sources to model spatial and temporal distribution patterns of factors like air and water quality. This modeling is crucial for evaluating health risks linked to environmental factors. Additionally, wearable sensor technologies revolutionize exposure monitoring by providing real-time, individual exposure data to elements such as air quality and UV radiation, offering granular insights into daily exposure patterns. Despite these advancements, the dynamic nature of environmental exposures and the heterogeneity in measurement techniques pose substantial challenges. These include the need for standardized data collection methods and the development of structured data sharing protocols to facilitate comparisons and enhance the accuracy of exposome research in understanding NCDs. Establishing and maintaining biobanks Establishing and maintaining biobanks is a critical yet challenging endeavor in omics and multi-omics research, particularly in low and middle-income countries (LMICs) . While most biobanks are found in high-income countries, equipped with advanced infrastructure and technical capacity, LMICs face substantial barriers such as inadequate funding, limited institutional capacity, and a shortage of skilled professionals. This disparity is especially pronounced in Africa and South Asia, which are severely underrepresented in genomic research . Most genomic studies in LMICs rely on funding from high-income countries through collaborative efforts, often resulting in research agendas set by external priorities rather than local needs . Significant and sustained investment in biobanking infrastructure in under-represented regions is crucial to address the lack of diversity in omics research. Initiatives like the China Kadoorie Biobank and the South African Human Genome Project provide hopeful examples of how national governments are recognizing the value of omics studies . In addition, further improvement in this field could be achieved by global consortia directing technical and financial resources to build local biobanking capacities in LMICs. This approach not only helps in establishing the necessary infrastructure for sample processing, genotyping, sequencing, and computational analysis, but also facilitates equitable data, ultimately benefiting the global scientific community. For instance, the Africa Wits-INDEPTH Partnership for Genomic Research (AWI-Gen) exemplifies a strategic regional collaboration funded by the NIH . This project has established a cross-sectional population cohort of about 12,000 adults across four African countries, leveraging existing Health and Demographic Surveillance System centers and community engagement to span a wide representation of social and genetic variability . Similarly, initiatives such as the H3Africa, H3Africa Bioinformatics Network(H3AfricaBioNet), and the Data Science for Health Discovery and Innovation in Africa, strategic funding commitments by the NIH, exemplify efforts to bolster genetic research capacity in Africa . However, it is important to note that future funding commitments in genomics will benefit from expansion to broader continental regions to address health problems and capacity-building needs of countries with no pre-existing omics research infrastructure. Another significant hurdle is the lack of expertise for addressing the ethical, legal, and social implications (ELSIs) of multi-omics research, which hinders the conduct of research and efficient sharing of data . To address this, it is essential to create national and local opportunities for advanced training, foster continuous professional development, and develop comprehensive ELSI guidelines that can be integrated into study designs. These measures will ensure that multi-omics research is conducted responsibly and its benefits are equitably shared, maintaining the integrity and relevance of the research. Additionally, promoting workforce diversity in omics research is crucial for building trust and fostering engagement among underrepresented groups. Diverse research teams are more likely to focus on health issues pertinent to their communities, which in turn encourages broader participation and consent in biobank studies. This not only strengthens the relationship between researchers and participants but also enhances the quality of research data, making genomic studies more impactful and relevant across populations . Multi-omics data and integration methods Integrating multi-omics data to unravel complex GxE interactions in NCDs is complicated by diverse data formats and significant preprocessing requirements . The lack of standardized methods for preprocessing and integrating data from various omics platforms often compromises the effectiveness of analyses . Additionally, the integration process is challenged by the "curse of dimensionality." This term describes issues that arise in high-dimensional datasets, where the volume of variables far exceeds the number of samples, leading to data sparsity and inconsistency across samples . This makes it difficult to draw reliable conclusions from the data, emphasizing the need for robust analytical tools and methods that can handle and integrate vast and varied omics data effectively. On the other hand, tissue and cell-type heterogeneity present another significant challenge in multi-omics integration, particularly relevant to studying complex diseases . Different cell types within a single tissue sample can exhibit unique omics profiles, influenced by the tissue's specific section or the physiological condition of an individual . These variations can skew biomarker levels and lead to misleading associations that reflect cellular differences rather than the disease itself. Although statistical methods have been developed to adjust for cell-type heterogeneity, they may not fully account for the true biological variations or might even over-correct them. Ideally, single-cell omics would provide a clearer picture by isolating the profiles of each cell type, but this approach is often impractical due to high costs and material requirements . The challenges of sample heterogeneity and technical artifacts, such as batch effects during sequencing, underscore the complexity of data preprocessing in multi-omics studies. Ensuring consistent data processing and leveraging appropriate statistical controls are crucial for mitigating these issues and enhancing the reliability of multi-omics analyses. Furthermore, while NGS technologies have made sequencing faster and more affordable, they have also introduced challenges such as increased costs for participant recruitment and sample processing, and complexities in data management and storage . Privacy concerns frequently limit data sharing between institutions, sometimes leading to the withdrawal of large datasets from public access due to potential identification risks . Moreover, proprietary standards for biomedical devices and health IT systems hinder seamless data integration across different sources . Addressing these issues requires comprehensive efforts to harmonize data across various healthcare providers and omics modalities, necessitating a collaborative approach from all stakeholders in healthcare and research to enhance real-world evidence-based practices and improve healthcare outcomes. Another multi-omics integration challenge, particularly when applying enrichment-based methods to uncover gene-environment interactions in NCDs, is the potential bias introduced by linkage disequilibrium, colocalization of multiple functional variants, and unaccounted confounders . Fine-mapping and imputation-based methods, which are crucial for developing biomarkers and understanding molecular mechanisms, depend heavily on the accuracy of population-specific linkage disequilibrium matrices . These methods also rely on robust genetic reference models for molecular features such as gene expression or methylation, which are difficult to obtain for features other than gene expression . The variability of QTL architecture across different tissues further complicates these analyses, necessitating careful consideration of tissue relevance to the disease mechanisms under study . Researchers must ensure they are well-versed in the biological assumptions, statistical constraints, and computational demands of the integration tools they choose to employ to enhance the reliability and applicability of their findings in NCD research. Validation of GxE interactions and translational applications Validating GxE interactions identified in human research and translating them into actionable insights remains a critical challenge. Translational studies using model organisms bridge observational findings with mechanistic understanding, allowing researchers to explore how genetic and environmental factors interplay in the development of NCDs . Model organisms such as mice , rats , Drosophila melanogaster , and Caenorhabditis elegans offer controlled environments where genetic and environmental variables can be precisely manipulated. This control facilitates the dissection of complex biological processes that are challenging to study directly in humans due to ethical and practical constraints. Moreover, hypotheses generated from these studies can be tested using human genetic data, improving detection power and enabling a more detailed analysis of subpopulations to understand GxE interactions better. Incorporating functional annotations from resources such as ENCODE, GTEx, and Roadmap Epigenomics further enhances this process by prioritizing candidate variants and regulatory regions for GxE studies, particularly those in non-coding regions often affected by environmental exposures . For example, genetically diverse rodent models like the Collaborative Cross and Diversity Outbred lines , which have high sequence homology with humans, have been instrumental in identifying QTLs and candidate genes involved in GxE interactions relevant to human NCDs. A notable case involves mutations in the tumor suppressor gene BAP1 , which have been linked with increased susceptibility to mesothelioma following asbestos exposure . Exploring how BAP1 mutations interact with asbestos exposure could elucidate key molecular pathways in carcinogenesis, with the potential to inform targeted screening, prevention strategies, and therapies tailored to the underlying mechanisms. Similarly, studies in Drosophila and C. elegans have facilitated high-throughput screening of genetic variants and environmental exposures, uncovering genetic pathways that modulate responses to environmental stressors and offering translational insights about human health . However, the translation of findings from model systems to human populations is not without challenges. While model organisms provide controlled environments, they cannot fully replicate the genetic complexity, environmental diversity, or numerous confounding factors that influence human health. For instance, gene synteny between humans and model organisms often diverges, particularly for non-coding and regulatory regions, limiting the applicability of some findings. Studies such as Seok et al. (2013) have demonstrated that genomic responses in mouse models often poorly mimic human inflammatory diseases, reflecting the inherent differences in gene regulatory networks and physiological responses. Furthermore, humans are exposed to a far more diverse range of environmental factors—such as diet, pollution, and stress—than those typically replicated in model organism studies, which limits the generalizability of findings (Table ). Functional annotations and perturbation studies, conducted in both in vitro and in vivo settings, hold promise for unraveling the complexities of GxE interactions in NCDs and other complex diseases . Functional annotations derived from large-scale projects, such as ENCODE and GTEx, systematically map regulatory elements and link genetic variants to potential functional effects, guiding the identification of candidate variants and regulatory regions . Perturbation studies, including CRISPR-Cas9-based approaches, enable direct testing of causal hypotheses . For example, CRISPR interference (CRISPRi) and activation (CRISPRa) screens in human induced pluripotent stem cell (hiPSC)-derived neurons have identified essential genes for neuronal survival under chronic oxidative stress–a key environmental factor relevant to neurodegenerative diseases–revealing critical mediators like GPX4 and other selenoprotein synthesis genes . Translational applications of GxE analysis have direct implications for personalized medicine and public health interventions . In precision environmental health, identifying how specific genetic variations influence susceptibility to environmental exposures enables the development of tailored interventions. For instance, genetic variation in CYP2D6 may influence susceptibility to Parkinson’s disease through pesticide exposure, with poor metabolizers potentially at greater risk . These findings could inform strategies to reduce exposure in vulnerable populations, though further research is needed for confirmation. The ALDH2*2 variant, common in certain populations, impairs acetaldehyde metabolism and may increase the risk of esophageal cancer with alcohol intake, suggesting the potential for personalized dietary recommendations and targeted prevention strategies in affected populations . In pharmacogenomics, GxE interactions can guide personalized medication regimens to optimize efficacy and minimize adverse effects. Personalized warfarin dosing based on variations in genes like VKOR1 and CYP2C9 has been shown to improve therapeutic outcomes and reduce the risk of bleeding complications . Variants in the TPMT gene necessitate dose adjustments of thiopurine drugs to prevent toxicity in treating conditions like leukemia and autoimmune diseases . CYP2D6 gene variants inform the selection and dosing of antidepressants, enhancing treatment response and reducing side effects . In oncology, identifying BRCA1/2 mutations allows for the use of Poly(ADP-ribose) polymerase (PARP) inhibitors in targeted cancer therapy, while HER2 expression guides the use of trastuzumab in breast cancer treatment, exemplifying how GxE insights contribute to precision medicine . Approaches that integrate biological pathways and regulatory annotations can further enhance the discovery and application of such GxE findings. These translational applications underscore the importance of validating GxE interactions through model organisms and advanced experimental systems. However, significant challenges persist. Limited experimental validation of GxE findings in model organisms and translational settings undermines confidence in the biological mechanisms underlying these interactions, slowing their application to precision medicine and public health interventions . High costs and the technical complexity of integrating environmental monitoring data with omics insights further impede progress. Additionally, the lack of standardized protocols for validating GxE findings, combined with the scarcity of diverse model systems, restricts the development of tailored therapies and prevention strategies . These issues collectively limit the potential of GxE research to address global health disparities effectively, particularly in low-resource settings where both environmental and omics data are underrepresented. Despite efforts to diversify genomic datasets, the vast majority of GWAS, about 85% as of 2023, predominantly feature individuals of European genetic ancestry . Progress toward including under-represented populations has been slow, with the share of studies involving these groups either stagnating or even declining in recent years . Although there has been a modest rise in the representation of Asian ancestries, African, Latin American, and Indigenous populations remain markedly underrepresented . This imbalance is compounded by the over-reliance on easily accessible and homogeneous resources like the UK Biobank, which primarily comprises individuals of European ancestry, whereas other ancestry groups often have limited data repositories available . Figure presents the global distribution of total GWAS sample sizes by country, underscoring significant regional disparities. This lack of diversity leads to a substantial problem: PGSs derived from predominantly European datasets show dramatically reduced predictive accuracy when applied to non-European populations . For instance, Martin et al. (2019) reported a decline in PGS accuracy of about 37%, 50%, and 78% for individuals of South Asian, East Asian, and African ancestries, respectively . Further studies, such as those by Privé et al. (2021) and Ding et al. (2023), confirm that PGS accuracy not only diminishes across different ancestries but also varies significantly within them depending on the genetic distance from the European training populations . The limited generalizability of these genetic insights could potentially exacerbate health disparities, underscoring the urgent need to broaden the genetic diversity in omics research to ensure that genomic advancements benefit all populations equitably . Furthermore, increasing the diversity of genomic data not only mitigates disparities but also significantly enhances the fine-mapping of GWAS signals and the identification of target genes . This is crucial for uncovering the genetic mechanisms influencing the development of NCDs and other complex conditions. Underrepresented groups, such as those of African and South Asian ancestries, exhibit higher genetic diversity, which translates into substantial gains in genomic research . Studies incorporating these populations have unearthed population-enriched clinically important variants that were previously undiscovered in predominantly European datasets. For example, research into African genetic ancestry has led to critical insights, including the link between APOL1 variants and chronic kidney disease , the identification of G6PD variants that refine diabetes diagnostics , and loss of function variants in PCSK9 that contribute to lower low-density lipoprotein cholesterol levels—this latter discovery has spurred the development of PCSK9 inhibitor drugs . These findings underscore the value of including diverse genetic backgrounds in research to achieve a comprehensive understanding of genetic factors across all populations, enhancing the overall impact of genomic studies on global health. The lack of genetic diversity is a pervasive issue across various omics datasets, not just genomics . For instance, bulk and single-cell transcriptomic analyses are beginning to uncover significant heterogeneity in gene expression across different cell types and even within the same type. This diversity is especially pronounced across different genetic ancestries, shaped by distinct environmental and genetic interactions. Major research efforts, such as single-cell consortia including KPMP, LungMAP, HTCA, GTEx , HuBMAP, Azimuth, HCA, and the Allen Brain Atlas, have predominantly focused on populations of European genetic ancestry, resulting in the underrepresentation of other groups. For example, of the 4,723 samples analyzed across these consortia, the majority are from individuals of European descent, starkly contrasted with the minimal representation from African, Hispanic, and East Asian ancestries. Addressing this imbalance is critical for enhancing our understanding of context-specific cellular mechanisms and improving the detection and treatment of diseases that vary regionally due to factors like genetic drift and migration. This understanding is particularly vital in pharmacogenomics, where knowing context-specific gene regulation can significantly advance personalized medicine. In a significant step toward addressing this imbalance, the Chan Zuckerberg Initiative has recently funded the Ancestry Networks for the Human Cell Atlas (HCA) with a $28 million grant, supporting the inclusion of ancestrally diverse tissue samples to ensure a broader representation and deeper insights into the genetic underpinnings of health and disease across populations. Similarly, the representation of genetic diversity in epigenomic data is markedly limited , as demonstrated by a study by Breeze et al. (2022). This study revealed that among the 5,048 epigenetic experiments from the US-based ENCODE data and the International Human Epigenome Consortium (IHEC), 87.1% (n = 4,397) predominantly featured samples of European genetic ancestry, with other ancestries severely underrepresented. Such disparities underscore a significant bias in the samples analyzed, with only a fraction representing African, Asian, and other ancestries. This lack of diversity impedes our ability to fully understand and interpret disease-associated genomic regions across populations. Epigenomic markers such as promoters, enhancers, and repressors are crucial for annotating non-coding regions identified by GWAS, which often have unclear functional implications. Broadening the scope of epigenomic data to include diverse populations could enhance the interpretation of GWAS loci, offering vital insights into the regulatory mechanisms affecting diseases that disproportionately impact non-European populations, like prostate cancer, hypertension, and chronic kidney disease. Measuring exposomes in multi-omics research on NCDs involves significant challenges due to the complexity and diversity of environmental exposures. Exposomes encompass a range of external factors, like pollution and radiation, alongside internal factors, such as microbiome interactions and metabolic processes. Technologies like mass spectrometry (MS) and geographic information systems (GIS) are essential for quantifying these exposures. MS, particularly untargeted MS, excels in detecting a broad spectrum of small molecules in biological samples, providing a comprehensive snapshot of chemical exposures. However, the vast amount of data generated requires advanced bioinformatics for accurate analysis, and detection sensitivity varies significantly among different chemical classes. GIS tools assess environmental exposure by integrating diverse data sources to model spatial and temporal distribution patterns of factors like air and water quality. This modeling is crucial for evaluating health risks linked to environmental factors. Additionally, wearable sensor technologies revolutionize exposure monitoring by providing real-time, individual exposure data to elements such as air quality and UV radiation, offering granular insights into daily exposure patterns. Despite these advancements, the dynamic nature of environmental exposures and the heterogeneity in measurement techniques pose substantial challenges. These include the need for standardized data collection methods and the development of structured data sharing protocols to facilitate comparisons and enhance the accuracy of exposome research in understanding NCDs. Establishing and maintaining biobanks is a critical yet challenging endeavor in omics and multi-omics research, particularly in low and middle-income countries (LMICs) . While most biobanks are found in high-income countries, equipped with advanced infrastructure and technical capacity, LMICs face substantial barriers such as inadequate funding, limited institutional capacity, and a shortage of skilled professionals. This disparity is especially pronounced in Africa and South Asia, which are severely underrepresented in genomic research . Most genomic studies in LMICs rely on funding from high-income countries through collaborative efforts, often resulting in research agendas set by external priorities rather than local needs . Significant and sustained investment in biobanking infrastructure in under-represented regions is crucial to address the lack of diversity in omics research. Initiatives like the China Kadoorie Biobank and the South African Human Genome Project provide hopeful examples of how national governments are recognizing the value of omics studies . In addition, further improvement in this field could be achieved by global consortia directing technical and financial resources to build local biobanking capacities in LMICs. This approach not only helps in establishing the necessary infrastructure for sample processing, genotyping, sequencing, and computational analysis, but also facilitates equitable data, ultimately benefiting the global scientific community. For instance, the Africa Wits-INDEPTH Partnership for Genomic Research (AWI-Gen) exemplifies a strategic regional collaboration funded by the NIH . This project has established a cross-sectional population cohort of about 12,000 adults across four African countries, leveraging existing Health and Demographic Surveillance System centers and community engagement to span a wide representation of social and genetic variability . Similarly, initiatives such as the H3Africa, H3Africa Bioinformatics Network(H3AfricaBioNet), and the Data Science for Health Discovery and Innovation in Africa, strategic funding commitments by the NIH, exemplify efforts to bolster genetic research capacity in Africa . However, it is important to note that future funding commitments in genomics will benefit from expansion to broader continental regions to address health problems and capacity-building needs of countries with no pre-existing omics research infrastructure. Another significant hurdle is the lack of expertise for addressing the ethical, legal, and social implications (ELSIs) of multi-omics research, which hinders the conduct of research and efficient sharing of data . To address this, it is essential to create national and local opportunities for advanced training, foster continuous professional development, and develop comprehensive ELSI guidelines that can be integrated into study designs. These measures will ensure that multi-omics research is conducted responsibly and its benefits are equitably shared, maintaining the integrity and relevance of the research. Additionally, promoting workforce diversity in omics research is crucial for building trust and fostering engagement among underrepresented groups. Diverse research teams are more likely to focus on health issues pertinent to their communities, which in turn encourages broader participation and consent in biobank studies. This not only strengthens the relationship between researchers and participants but also enhances the quality of research data, making genomic studies more impactful and relevant across populations . Integrating multi-omics data to unravel complex GxE interactions in NCDs is complicated by diverse data formats and significant preprocessing requirements . The lack of standardized methods for preprocessing and integrating data from various omics platforms often compromises the effectiveness of analyses . Additionally, the integration process is challenged by the "curse of dimensionality." This term describes issues that arise in high-dimensional datasets, where the volume of variables far exceeds the number of samples, leading to data sparsity and inconsistency across samples . This makes it difficult to draw reliable conclusions from the data, emphasizing the need for robust analytical tools and methods that can handle and integrate vast and varied omics data effectively. On the other hand, tissue and cell-type heterogeneity present another significant challenge in multi-omics integration, particularly relevant to studying complex diseases . Different cell types within a single tissue sample can exhibit unique omics profiles, influenced by the tissue's specific section or the physiological condition of an individual . These variations can skew biomarker levels and lead to misleading associations that reflect cellular differences rather than the disease itself. Although statistical methods have been developed to adjust for cell-type heterogeneity, they may not fully account for the true biological variations or might even over-correct them. Ideally, single-cell omics would provide a clearer picture by isolating the profiles of each cell type, but this approach is often impractical due to high costs and material requirements . The challenges of sample heterogeneity and technical artifacts, such as batch effects during sequencing, underscore the complexity of data preprocessing in multi-omics studies. Ensuring consistent data processing and leveraging appropriate statistical controls are crucial for mitigating these issues and enhancing the reliability of multi-omics analyses. Furthermore, while NGS technologies have made sequencing faster and more affordable, they have also introduced challenges such as increased costs for participant recruitment and sample processing, and complexities in data management and storage . Privacy concerns frequently limit data sharing between institutions, sometimes leading to the withdrawal of large datasets from public access due to potential identification risks . Moreover, proprietary standards for biomedical devices and health IT systems hinder seamless data integration across different sources . Addressing these issues requires comprehensive efforts to harmonize data across various healthcare providers and omics modalities, necessitating a collaborative approach from all stakeholders in healthcare and research to enhance real-world evidence-based practices and improve healthcare outcomes. Another multi-omics integration challenge, particularly when applying enrichment-based methods to uncover gene-environment interactions in NCDs, is the potential bias introduced by linkage disequilibrium, colocalization of multiple functional variants, and unaccounted confounders . Fine-mapping and imputation-based methods, which are crucial for developing biomarkers and understanding molecular mechanisms, depend heavily on the accuracy of population-specific linkage disequilibrium matrices . These methods also rely on robust genetic reference models for molecular features such as gene expression or methylation, which are difficult to obtain for features other than gene expression . The variability of QTL architecture across different tissues further complicates these analyses, necessitating careful consideration of tissue relevance to the disease mechanisms under study . Researchers must ensure they are well-versed in the biological assumptions, statistical constraints, and computational demands of the integration tools they choose to employ to enhance the reliability and applicability of their findings in NCD research. Validating GxE interactions identified in human research and translating them into actionable insights remains a critical challenge. Translational studies using model organisms bridge observational findings with mechanistic understanding, allowing researchers to explore how genetic and environmental factors interplay in the development of NCDs . Model organisms such as mice , rats , Drosophila melanogaster , and Caenorhabditis elegans offer controlled environments where genetic and environmental variables can be precisely manipulated. This control facilitates the dissection of complex biological processes that are challenging to study directly in humans due to ethical and practical constraints. Moreover, hypotheses generated from these studies can be tested using human genetic data, improving detection power and enabling a more detailed analysis of subpopulations to understand GxE interactions better. Incorporating functional annotations from resources such as ENCODE, GTEx, and Roadmap Epigenomics further enhances this process by prioritizing candidate variants and regulatory regions for GxE studies, particularly those in non-coding regions often affected by environmental exposures . For example, genetically diverse rodent models like the Collaborative Cross and Diversity Outbred lines , which have high sequence homology with humans, have been instrumental in identifying QTLs and candidate genes involved in GxE interactions relevant to human NCDs. A notable case involves mutations in the tumor suppressor gene BAP1 , which have been linked with increased susceptibility to mesothelioma following asbestos exposure . Exploring how BAP1 mutations interact with asbestos exposure could elucidate key molecular pathways in carcinogenesis, with the potential to inform targeted screening, prevention strategies, and therapies tailored to the underlying mechanisms. Similarly, studies in Drosophila and C. elegans have facilitated high-throughput screening of genetic variants and environmental exposures, uncovering genetic pathways that modulate responses to environmental stressors and offering translational insights about human health . However, the translation of findings from model systems to human populations is not without challenges. While model organisms provide controlled environments, they cannot fully replicate the genetic complexity, environmental diversity, or numerous confounding factors that influence human health. For instance, gene synteny between humans and model organisms often diverges, particularly for non-coding and regulatory regions, limiting the applicability of some findings. Studies such as Seok et al. (2013) have demonstrated that genomic responses in mouse models often poorly mimic human inflammatory diseases, reflecting the inherent differences in gene regulatory networks and physiological responses. Furthermore, humans are exposed to a far more diverse range of environmental factors—such as diet, pollution, and stress—than those typically replicated in model organism studies, which limits the generalizability of findings (Table ). Functional annotations and perturbation studies, conducted in both in vitro and in vivo settings, hold promise for unraveling the complexities of GxE interactions in NCDs and other complex diseases . Functional annotations derived from large-scale projects, such as ENCODE and GTEx, systematically map regulatory elements and link genetic variants to potential functional effects, guiding the identification of candidate variants and regulatory regions . Perturbation studies, including CRISPR-Cas9-based approaches, enable direct testing of causal hypotheses . For example, CRISPR interference (CRISPRi) and activation (CRISPRa) screens in human induced pluripotent stem cell (hiPSC)-derived neurons have identified essential genes for neuronal survival under chronic oxidative stress–a key environmental factor relevant to neurodegenerative diseases–revealing critical mediators like GPX4 and other selenoprotein synthesis genes . Translational applications of GxE analysis have direct implications for personalized medicine and public health interventions . In precision environmental health, identifying how specific genetic variations influence susceptibility to environmental exposures enables the development of tailored interventions. For instance, genetic variation in CYP2D6 may influence susceptibility to Parkinson’s disease through pesticide exposure, with poor metabolizers potentially at greater risk . These findings could inform strategies to reduce exposure in vulnerable populations, though further research is needed for confirmation. The ALDH2*2 variant, common in certain populations, impairs acetaldehyde metabolism and may increase the risk of esophageal cancer with alcohol intake, suggesting the potential for personalized dietary recommendations and targeted prevention strategies in affected populations . In pharmacogenomics, GxE interactions can guide personalized medication regimens to optimize efficacy and minimize adverse effects. Personalized warfarin dosing based on variations in genes like VKOR1 and CYP2C9 has been shown to improve therapeutic outcomes and reduce the risk of bleeding complications . Variants in the TPMT gene necessitate dose adjustments of thiopurine drugs to prevent toxicity in treating conditions like leukemia and autoimmune diseases . CYP2D6 gene variants inform the selection and dosing of antidepressants, enhancing treatment response and reducing side effects . In oncology, identifying BRCA1/2 mutations allows for the use of Poly(ADP-ribose) polymerase (PARP) inhibitors in targeted cancer therapy, while HER2 expression guides the use of trastuzumab in breast cancer treatment, exemplifying how GxE insights contribute to precision medicine . Approaches that integrate biological pathways and regulatory annotations can further enhance the discovery and application of such GxE findings. These translational applications underscore the importance of validating GxE interactions through model organisms and advanced experimental systems. However, significant challenges persist. Limited experimental validation of GxE findings in model organisms and translational settings undermines confidence in the biological mechanisms underlying these interactions, slowing their application to precision medicine and public health interventions . High costs and the technical complexity of integrating environmental monitoring data with omics insights further impede progress. Additionally, the lack of standardized protocols for validating GxE findings, combined with the scarcity of diverse model systems, restricts the development of tailored therapies and prevention strategies . These issues collectively limit the potential of GxE research to address global health disparities effectively, particularly in low-resource settings where both environmental and omics data are underrepresented. This scoping review highlighted that NCDs, such as cardiovascular diseases, cancers, chronic respiratory diseases, and diabetes, result from the complex interaction of gene and environmental factors, such as diet, physical inactivity, and tobacco use. To unravel the complexity of GxE interactions and gain an understanding of the multiple factors underlying many NCDs, a multi-omics approach is indeed essential. By employing multi-omics and data integration techniques, it can be possible to fully understand how the interaction of genetic and environmental factors influences NCD development, progression, and treatment response. This involves exploring a range of omics disciplines—genomics, transcriptomics, epigenomics, proteomics, and exposomics and understanding how they individually and collectively influence the risk to NCDs. Despite the transformative potential of global multi-omics research in advancing precision medicine, there are significant challenges and opportunities related to its practical translational applications. Our review highlighted that the genome can not be viewed as a static entity, as both genetic and environmental factors dynamically influence disease onset and progression. We provide several examples of how different modalities complement genomic data by revealing dynamic changes in gene expression, pathways, and networks due to environmental exposures. Thus, comprehensive omics integration is essential for identifying novel biomarkers and therapeutic targets and enhancing diagnostic, prognostic, and treatment strategies. Integrative analyses ideally would involve multiple omics data from the same individuals, although practical challenges such as cost and tissue accessibility may often limit this ideal. International consortia and national biobanks have been established to address these challenges, collecting detailed phenotypic and, increasingly, omics biomarker data to fill major gaps in NCD research. Moreover, we underscored a pervasive issue of limited diversity across omics and multi-omics datasets, affecting the transferability of research findings and tools across different genetic ancestries. Currently, most genomic and omics studies predominantly feature individuals of European descent, significantly underrepresenting African, Latin American, and Indigenous populations. This underrepresentation compromises the predictive accuracy of polygenic scores and other genomic tools when applied to non-European groups. Furthermore, including diverse genetic ancestries is crucial not only for enhancing the precision of GWAS signal mapping but also for discovering clinically relevant genetic variants that remain unidentified in predominantly European datasets. For example, studies involving African ancestry have led to key discoveries in chronic kidney disease and diabetes management. Addressing this lack of diversity is essential not only to improve the scientific robustness of omics research but also to mitigate health disparities, ensuring that the benefits of genomic advances reach all populations equitably. To address fairness in multi-omics for equitable health advancements, concerted efforts should focus on increasing the diversity of both omics and multi-omics datasets. Future research should also aim to develop equity-centered genomics medicine advanced computational methods and tools that efficiently integrate various omics datasets. These tools must generate biomarkers or risk predictors that are broadly transferable across genetic ancestries, with a particular emphasis on accounting for known confounders such as gene-environment correlations, including population stratification and assortative mating. By doing so, we can improve the scientific robustness of omics research and ensure that the benefits of genomic advances reach all populations equitably, thereby helping to mitigate health disparities globally. Lastly, our exploration of multi-omics integration methods has illuminated the intricate challenges of combining diverse datasets to decode complex GxE interactions in NCDs. The heterogeneity of tissue and cell types significantly compounds these challenges, with variations in omics profiles within a single tissue potentially misleading biomarker identification. Statistical adjustments for cell-type heterogeneity aim to correct these variations, yet there is a risk of over-correction, obscuring genuine biological differences. High costs and logistical constraints often thwart the ideal solution of employing single-cell omics to circumvent these issues. Moreover, while advancements like NGS have reduced costs and increased the speed of data acquisition, they introduce new difficulties in data management, participant recruitment, and inter-institutional data sharing due to privacy concerns. This requires a coordinated effort to harmonize multi-omics data across different healthcare settings, requiring a collaborative approach among all stakeholders to leverage real-world evidence for improving health outcomes effectively. Additional file 1 . Additional file 2 . Additional file 3 . |
Serum tsRNA as a novel molecular diagnostic biomarker for lupus nephritis | 668add56-e08b-4f6b-a387-8fc5e8534bd8 | 9121311 | Pathology[mh] | This study has been supported by the National Natural Science Foundation of China (No. 61971216) and the Key Research and Development Project of Jiangsu Province (No. BE2019603, BE2020768). All authors affirm no conflict of interest. Supplementary information Click here for additional data file. |
Long term effectiveness of intraoperative radiotherapy given as a boost in adjuvant treatment for oral cavity cancers | 32c2203d-44d8-4ef9-bc46-4463d806d4ff | 11807087 | Surgical Procedures, Operative[mh] | Various radiotherapy methods of ‘boost dose’ (e.g. concomitant boost brachytherapy, chemo-acceleration) have been developed and validated in the clinic to improve local tumour outcomes, especially in case of high risk of local recurrences following conventional irradiation. Intraoperative radiotherapy (IORT) is one of the interesting solutions. Over the past three decades, interest in IORT has gone and is now coming back again. New enthusiasm for IORT has arisen due to the development of compact and portable electron or X-ray beam facilities (INTRABEAM™ machines). IORT remains a unique and attractive treatment method for selected patients to improve local tumour control of locally advanced cancers – . It has several advantages, including immediate start of radiation at the time of surgery, strict cooperation with the surgeon, and a conformity of a boost dose. The IORT has often been used for patients with breast cancer and other deep-seated tumours such as gastric, pancreatic, and rectal cancers, and is regarded as an intraoperative boost dose prior to postoperative fractionated radiotherapy. The effective use of IORT requires a multidisciplinary approach, including close cooperation between surgeons, radiation oncologists, clinical oncologists and physicists. However, despite promising clinical data, there is a lack of large prospective trials, except for breast cancer. There are several publications that present the use of IORT in oral cavity squamous cell cancer to treat primary or recurrent tumours – . This study aims to present our experience and evaluate the effectiveness of the IORT for oral cavity patients. All consecutive patients who received IORT in our cancer centre during the treatment of oral cancer were included into analysis. The main inclusion criteria for IORT was the diagnosis of oral cavity cancer preoperatively assessed as a potentially microscopically doubt by the surgeon. Lack of informed consent, large tumour volume, no operating conditions during surgery, and gross microscopic margin involvement were the main exclusion criteria. From 2003 to 2019 a total number of 23 patients with pathologically proven oral cavity squamous cell cancer (14 localized on the mobile tongue, 9 in the floor of the month) have been treated with surgery, IORT and postoperative radiotherapy. According to TNM, 13 tumours were in stage T 1 , nine in T 2 and one in T 3 stage. There were 18 cases with N 0 , two with N 1 , two with N 2 and one case with N 3 stage. Tumour volume ranged from 0.03 cc to 67.78 cc (median 1.26 cc). Stage M 0 was proven in all cases based on X-ray and ultrasound (USG) images. The patient’s age raged from 38 to 84 years, with a median of 63 (IQR 14). The male to female ratio was 1.6:1. No adjuvant treatment was administered. Such slow recruitment of patient was an effect of progressive changes in surgery and the implementation of reconstructive procedures that modified the indications for radiotherapy. Even though X-ray therapy was a well-established treatment in Poland in the early 2000s, during screening each patient was informed about the specificity of the method, its expected benefits and side effects. All patients also had the opportunity to get answers to any questions. In case of patient doubts or lack of informed consent, only postoperative EBRT boost was used. The study was approved by the ethical committee at the MSC National Research Institute of Oncology in Gliwice (number: KB/KB/430-70/23) and performed according to the Helsinki Declaration and treatment protocol of Maria Sklodowska-Curie National Research Institute of Oncology, branch Gliwice. The protocol of the pilot study was approved by the institutional ethics committee in 2003. Treatment All patients underwent tumorectomy, which was combined in nine cases with some type of reconstruction or plastic surgery to restore topography and function of surrounding healthy tissues. In four cases free cutaneous transfers were used, pedicle submental flap in three cases, free fibular flap in one case, and transverse tongue flap in one case. The type of surgery and reconstruction depended on the extent of the disease and the size of the resection, and the surgical team’s decision. Based on the initial clinical stage and the results of intraoperative lymph node biopsies, lymph node dissection was performed in 20 cases. In each case, margin status was determined during surgery by frozen section. Decision of surgical extension or IORT implementation was making immediate by the surgeon and radiation oncologist based on the intraoperative report. Patients were considered eligible for IORT due to their narrow surgical margins, which suggested nonradical microscopically tumour dissection, or intraoperative biopsies that demonstrated histologically positive margins. Both of these factors are risk factors for local recurrence. The decision of combined treatment (high dose external beam radiotherapy (EBRT) + IORT) was preferred in these cases over mutilating surgery and approved by the surgeon. Definitive margins were confirmed later in the final pathology. IORT When pathological or surgical risk margins were defined and the diameter of the tumour bed was measured, an adequate spherical applicator in the size range of 2 - 4 cm,was put in the tumour bed area. . The applicator’s diameter was precisely determined to encompass the entire high-risk area. The target for the IORT boost was defined as the tumour bed. Due to the specificity of the technique and the device, it was not possible to vary the dose in different directions depending on the size of the margins. The dose selection process considered margin status (main factor increasing dose), volume of tumour bed (applicator size), proximity to critical structures. Due to lack of final pathology, decision-making of dose was based on frozen section probe: for positive or close (< 1 mm in frozen section) margin dose 7 to 7.5 Gy was given, for negative, 5 Gy as standard dose was delivered. In result, there was no need to reduce dose because of the tumour bed size and nearby critical structures. A single dose of 5–7.5 Gy was delivered with a 50 kV X-ray beam generated by INTRABEAM™, measured at a distance of 5 mm from the surface of applicator. It is difficult to compare the clinical effectiveness of a single X-ray dose with a RBE of 1.4 with a fractionated dose of photon with a RBE of 1.0 , . For this purpose, the IORT dose was normalized (EQD 2 ) if it was given in 2.0 Gy fractions using the α/β value of 10 Gy using the formula: \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\text{EQD}}_{{2\,{\text{IORT}}}} = {\text{RBE}} \times {\text{TD}}_{{{\text{IORT}}}} {{\left( {\upalpha /\upbeta + {\text{TD}}_{{{\text{IORT}}}} } \right)} \mathord{\left/ {\vphantom {{\left( {\upalpha /\upbeta + {\text{TD}}_{{{\text{IORT}}}} } \right)} {\left( {\alpha /\beta + 2.0\,{\text{Gy}}} \right)}}} \right. \kern-\nulldelimiterspace} {\left( {\alpha /\beta + 2.0\,{\text{Gy}}} \right)}}$$\end{document} The calculated biological dose of IORT was on average 11.95 izoGy 2.0 (8.7 izoGy 2.0 − 15.2 izoGy 2.0 ). The median treatment time was 13 minutes (range 6.5–19.5 min). The IORT procedure was described in our previous articles and has not changed since 2003 , . External beam radiotherapy (EBRT) Postoperative 3D radiotherapy usually used a total dose of 50 Gy delivered in 25 fractions. In most cases (17 patients) the dynamic techniques (intensity modulated radiotherapy—IMRT) were used, 3D conformal techniques (3D CRT) were used rarely (6 patients) (mainly for patients treated in the first years of the study when it was a standard technique). Clinical target volume included the tumour bed and lymph nodes areas. The total dose (TD) was escalated to 60 Gy in 30 fractions for patients with extracapsular invasion and narrow or positive surgical margins. Only in one case, EBRT was not used (patient with a small tumour and no other risk factors in the final pathology). The median time between IORT—surgery and EBRT was 53 days (range 30–134, IQR—17 days. Finally, the total normalized dose of both radiation treatments was between 62.95 and 72.95 izoGy 2.0 (58.7 izoGy – 75.2 izoGy 2.0 ). Details of the clinical characteristics of the patients and parameters of radiotherapy are shown in Table . Study end points In all patients, early and late toxicity was evaluated. Two scales were used: Dische and RTOG. The early mucosal side effects were scored using the Dische scale (14 points on the Dische correspond to confluent mucositis—CM). The RTOG scale was used to evaluate late effects. Mild and clinically irrelevant symptoms were accounted for. During the follow-up period, patients had check-up visits and regular examinations were performed, including physical examination and computed tomography (CT) or magnetic resonance imaging (MRI) if necessary. Local recurrences within the irradiated area were confirmed trough biopsy and histopathological examination. Periodically, chest radiographs and abdominal sonography were used to exclude distant metastases. Local tumour control (LTC) was defined as the time without local recurrence within the irradiated tumour bed area and calculated from the final day of EBRT. Disease-free survival (DFS) was calculated from the last day of EBRT to any (local or distant) relapse or death from any case. Overall Survival (OS) was calculated from the last day of EBRT to the day of death or loss from the observation (censored data) or the date of the last follow-up visit. LTC, DFS and OS estimates were calculated using the Kaplan Maier method. Univariate analysis to test the significance of the results was performed using log rank test and Chi-square tests. All patients underwent tumorectomy, which was combined in nine cases with some type of reconstruction or plastic surgery to restore topography and function of surrounding healthy tissues. In four cases free cutaneous transfers were used, pedicle submental flap in three cases, free fibular flap in one case, and transverse tongue flap in one case. The type of surgery and reconstruction depended on the extent of the disease and the size of the resection, and the surgical team’s decision. Based on the initial clinical stage and the results of intraoperative lymph node biopsies, lymph node dissection was performed in 20 cases. In each case, margin status was determined during surgery by frozen section. Decision of surgical extension or IORT implementation was making immediate by the surgeon and radiation oncologist based on the intraoperative report. Patients were considered eligible for IORT due to their narrow surgical margins, which suggested nonradical microscopically tumour dissection, or intraoperative biopsies that demonstrated histologically positive margins. Both of these factors are risk factors for local recurrence. The decision of combined treatment (high dose external beam radiotherapy (EBRT) + IORT) was preferred in these cases over mutilating surgery and approved by the surgeon. Definitive margins were confirmed later in the final pathology. When pathological or surgical risk margins were defined and the diameter of the tumour bed was measured, an adequate spherical applicator in the size range of 2 - 4 cm,was put in the tumour bed area. . The applicator’s diameter was precisely determined to encompass the entire high-risk area. The target for the IORT boost was defined as the tumour bed. Due to the specificity of the technique and the device, it was not possible to vary the dose in different directions depending on the size of the margins. The dose selection process considered margin status (main factor increasing dose), volume of tumour bed (applicator size), proximity to critical structures. Due to lack of final pathology, decision-making of dose was based on frozen section probe: for positive or close (< 1 mm in frozen section) margin dose 7 to 7.5 Gy was given, for negative, 5 Gy as standard dose was delivered. In result, there was no need to reduce dose because of the tumour bed size and nearby critical structures. A single dose of 5–7.5 Gy was delivered with a 50 kV X-ray beam generated by INTRABEAM™, measured at a distance of 5 mm from the surface of applicator. It is difficult to compare the clinical effectiveness of a single X-ray dose with a RBE of 1.4 with a fractionated dose of photon with a RBE of 1.0 , . For this purpose, the IORT dose was normalized (EQD 2 ) if it was given in 2.0 Gy fractions using the α/β value of 10 Gy using the formula: \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\text{EQD}}_{{2\,{\text{IORT}}}} = {\text{RBE}} \times {\text{TD}}_{{{\text{IORT}}}} {{\left( {\upalpha /\upbeta + {\text{TD}}_{{{\text{IORT}}}} } \right)} \mathord{\left/ {\vphantom {{\left( {\upalpha /\upbeta + {\text{TD}}_{{{\text{IORT}}}} } \right)} {\left( {\alpha /\beta + 2.0\,{\text{Gy}}} \right)}}} \right. \kern-\nulldelimiterspace} {\left( {\alpha /\beta + 2.0\,{\text{Gy}}} \right)}}$$\end{document} The calculated biological dose of IORT was on average 11.95 izoGy 2.0 (8.7 izoGy 2.0 − 15.2 izoGy 2.0 ). The median treatment time was 13 minutes (range 6.5–19.5 min). The IORT procedure was described in our previous articles and has not changed since 2003 , . Postoperative 3D radiotherapy usually used a total dose of 50 Gy delivered in 25 fractions. In most cases (17 patients) the dynamic techniques (intensity modulated radiotherapy—IMRT) were used, 3D conformal techniques (3D CRT) were used rarely (6 patients) (mainly for patients treated in the first years of the study when it was a standard technique). Clinical target volume included the tumour bed and lymph nodes areas. The total dose (TD) was escalated to 60 Gy in 30 fractions for patients with extracapsular invasion and narrow or positive surgical margins. Only in one case, EBRT was not used (patient with a small tumour and no other risk factors in the final pathology). The median time between IORT—surgery and EBRT was 53 days (range 30–134, IQR—17 days. Finally, the total normalized dose of both radiation treatments was between 62.95 and 72.95 izoGy 2.0 (58.7 izoGy – 75.2 izoGy 2.0 ). Details of the clinical characteristics of the patients and parameters of radiotherapy are shown in Table . In all patients, early and late toxicity was evaluated. Two scales were used: Dische and RTOG. The early mucosal side effects were scored using the Dische scale (14 points on the Dische correspond to confluent mucositis—CM). The RTOG scale was used to evaluate late effects. Mild and clinically irrelevant symptoms were accounted for. During the follow-up period, patients had check-up visits and regular examinations were performed, including physical examination and computed tomography (CT) or magnetic resonance imaging (MRI) if necessary. Local recurrences within the irradiated area were confirmed trough biopsy and histopathological examination. Periodically, chest radiographs and abdominal sonography were used to exclude distant metastases. Local tumour control (LTC) was defined as the time without local recurrence within the irradiated tumour bed area and calculated from the final day of EBRT. Disease-free survival (DFS) was calculated from the last day of EBRT to any (local or distant) relapse or death from any case. Overall Survival (OS) was calculated from the last day of EBRT to the day of death or loss from the observation (censored data) or the date of the last follow-up visit. LTC, DFS and OS estimates were calculated using the Kaplan Maier method. Univariate analysis to test the significance of the results was performed using log rank test and Chi-square tests. The median follow-up was 64 months (range 8–222 months, IQR 150). Only one patient was withdrawn from follow up due to interrupted treatment. All patients underwent complete gross tumour resection, with 95% achieving negative surgical margins. None of the patients had reoperation. The early tolerance of IORT was good as only 5 patients experienced superficial mucosal erosion due to the small distance between the mucosa and the applicator surface (where the contact dose is higher than the prescribed dose). This lesions were self-healing and did not required any surgical procedures. The early mucosal reactions during EBRT were mild, with a median Dische score of 11 (range 1–15, IQR 4). These reactions generally started to heal at the end of combined treatment. No significant serious late effects (grade 3 or 4 on the RTOG scale) were observed. During the follow-up one patient experienced local recurrence (4.5%), two had nodal metastases (9%) and one patient (4.5%) had both local and nodal recurrence. Distant metastases were diagnosed in two patients, in one in the lung, and the other in the bones. The overall 5-year LTC was 92% and decreased to 82.5% during the next 5 years (Fig. ). In the statistical analysis, we did not find significant differences regarding sex, stage T, tumour grade, tumour localization and IORT boost dose. The pathological status (positive or negative) and the time interval between surgery and EBRT have no impact on the risk of recurrence. However, looking at the total normalized dose of both radiation treatments delivered, statistically significant impact on DFS was found ( p = 0.028, HR 0.97) with no impact on OS or LTC (p = ns). Five-year overall survival was 56% and 46% of 10-year OS, respectively. 5 and 10-year DFS was 52% and 40%, respectively (Fig. ). Although head and neck cancers are not the leading tumours in terms of morbidity, they continue to be a subject of intensive and challenging studies of various radiation therapy regimens with the aim of improving treatment efficacy. Cancers of the tongue or floor of the mouth, even in their early stages, are associated with a relatively high risk of local and/or nodal recurrence. That risk is significantly increased by the presence of positive surgical margins . Aggressive radical surgery can likely minimise that risk but, it has a significant detrimental impact on speech and swallowing dysfunctions and, consequently, on the physical and psychological quality of a patient’s life. Therefore, surgery is often combined with different types of radiotherapy and chemotherapy in order to minimize the risk of recurrence , , , – . Various strategies have been employed to escalate the radiation dose without extending the overall time of radiation, focusing on one extra fraction dose during the last two and half weeks of radiation (concomitant boost) or brachytherapy, chemo-acceleration directly after completing the EBRT – . Intraoperative radiotherapy was introduced nearly 50 years ago as a promising technique for boosting irradiation of tumours in various localizations. This method has been well documented as being highly effective – , , , , . However, it was not used as frequently for patients with head and neck cancers as for other tumour localizations, and even less frequently for patients with tongue and floor of mouth cancers , . Despite that, it seems that IORT is an attractive modality of radiation for both surgeons and radiotherapists who cooperate in the head and neck cancer team for a few reasons . First, in the case of inadequate surgical margins confirmed intraoperatively, this allows for intensifying the treatment and sterilizing the tumour bed from cancer cells. Secondly, the timing and precision of radiation delivery during surgery to the microscopic disease area indicated by the surgeon. An additional favourable aspect of IORT is the possibility of defining the high-risk area and minimising geographic errors, which are crucial after reconstructive surgery. Another rationale for using IORT is the reduction in the treatment time. Protracted time of adjuvant radiotherapy, especially by the boost dose administered in the last days of treatment, is fundamental to increasing the probability of failure due to accelerated tumour cells repopulation. IORT reduces the duration of radiotherapy, and in most cases diminishes the gap between surgery and adjuvant RT – , , . Due to its conformity, IORT also limits the volume of healthy tissues exposed to radiation compared to conventional EBRT , . The main disadvantage of single IORT doses is that they are less effective at killing the cells in the radioresistant phase of the cell cycle , . However, during fractionated radiotherapy, these cells can transfer into the radiosensitive phase and be reoxygeneted as well. Combinations of IORT followed by EBRT, as demonstrated in our study, are radiobiologically reasonable . A significant advantage of the low dose X-ray IORT over other IORT delivery methods such as electron beam or brachytherapy is its theoretically superior biological effectiveness. This is a consequence of the dosimetric characteristics of the INTRABEAM™ system applicators, the biological effects of a large single dose, and the higher RBE. Due to the dose fall off, the tissue contact dose at the applicator surface is significantly higher than the prescribed (at 5 mm), in fact. The smaller the applicator, the higher the dose on its surface. Therefore, if a standard dose of 5 Gy is prescribed at a specified 5 mm, it delivers a dose of 7–12 Gy on the surface. Given the increased surface dose and high RBE of X-ray radiation, the biological dose within the tumour bed is significantly higher than prescribed and is comparable to those given in high dose rate brachytherapy and radiosurgery boosts with all its consequences. It is therefore not surprising that the present study pertains to patients who were treated with IORT for a duration of 16 years and who were individually qualified for that method of boost. For the treatment of our patients, we used an INTRABEAM™ system that generate 50 kV X-rays. In contrast to electron beam accelerators, INTRABEAM™, does not require expensive radioprotective adaptations in the operating theatre. It is also mobile and can be used in multiple rooms. Moreover, the X-ray beam characterizes a higher RBE (ranging from 1.4 to 3.5) compared to the electron beam , . For the present group of patients, the biologically normalized median dose IORT as if it were given in 2.0 Gy fractions was 11.95 izoGy 2.0 (range 8.7 izoGy 2.0 − 15.2 izoGy 2.0 ). The duration of the IORT combined with postoperative EBRT has been found to be an important prognostic factor for therapeutic efficacy. Therefore the duration of interval between IORT therapy and EBRT plays a significant role. A study conducted on the IORT technique applied to early breast cancers patients has demonstrated that in patients at high risk for local or distant failures, IORT surgery and EBRT intervals shorter than 60 days resulted in no local recurrences. In contrast, for longer intervals, the rate of local recurrences increased to approximately 12% . Among the present group of patients with cancer of the tongue and floor of the mouth, the median time interval was approximately 50 days. The 5-year local tumour control of 82% in the present study (Fig. ) is satisfactory, and it is higher than reported in other IORT studies , . It is possible that this is due to a relatively homogeneous group of patients (all were treated primarily and curatively), and despite the long period of 16 years during which patients were recruited in the protocol, the methods of therapeutic modalities were maintained unchanged. This study has some limitations. Due to the small sample size of the present study, it may rise some questions regarding its significance. However, the method was applied to highly selected patients, which was reflected by a small study group during the 16-year recruitment period. Even with the conditions that were present at the beginning of our study (early 2000s), it would be extremely difficult to increase the number of the study group. Moreover, the standards of surgery (advances in reconstructive surgery) and adjuvant treatment have changed (for instance radiochemotherapy in N+ patients, competing high-precision EBRT). The low 5-year OS of 56% may be surprising; however during the initial years of the IORT applications, the recruited patients were aged close to or exceeding 75 years. This may explain why the OS decreased by 44% in the first five years of follow-up and only by 10% in the next five years (Fig. ). The number of all cases is too small to exclude patients, let say patients over 60 years old, because a much smaller set of cases would definitely raise justified uncertainty. The five-year DFS was also unexpectedly low (52%), but it was defined as any failure or death for any reason, and the main cause of this is mentioned patients’ advanced age, which is why deaths from natural causes influenced the decrease of DFS, as well as OS. In general, it is difficult to compare these results to other studies because IORT is not commonly used in the case of patients with head and neck cancer, and there are no randomized studies evaluating clearly the effectiveness of such treatment. Regrettably, also there are no published studies comparing face-to-face IORT with other boost modalities for patients with oral cancer. Head and neck cancers are a heterogeneous group, and it is hard to compare them if they do not involve the same organ. Postoperative treatment encompasses additional parameters beyond radiation therapy alone or radiation therapy combined with chemotherapy or other biological agents, but without surgical intervention. An additional difficulty is the fact that in the available literature, even in the case of head and neck cancer patients, IORT was used in various modifications and localizations. In some studies, it was delivered as an electron beam, while in others as HDR brachytherapy. In less frequent instances, it was performed using low-energy kV, such as in our centre , . Considering the aim of treatment, in contrast to our study, IORT was used much more often as part of salvage or palliative treatment in the management of tumour recurrence after definitive therapy . Differences also concern the anatomical sites of the tumour, and the wide range of doses administered , . Although the analysed group was not treated under a rigid prospective study protocol, it was quite homogeneous in terms of localization, histopathology, treatment regimen and doses used, in contrast to others. The differentiating feature is the usage of orthovoltage radiation. In this study, 5-year local tumour control was comparable to results observed by other authors using different boost modalities. Blažek et al. used stereotactic boost (2 × 5Gy) and reported 5-year LTC 62% . Also Yamazaki et al. performed Cyber Knife based stereotactic boost and observed 5-year LTC of 71%, but the subgroup of oral cancer patients was relatively small (3 patients) . These encouraging results and continuous technological progress, make stereotactic techniques a very promising option for irradiation boost. Hsieh et al. collected outcomes of 196 oral cancer patients irradiated postoperatively using sequential or simultaneous integrated (SIB) external beam boost and showed 5 year LTC of 90% vs. 74% and the difference was statistically significant . Interstitial brachytherapy is another boost delivering technique, often being used for oral cavity cancers. Lapeyre et al. published results of 46 oral cavity cancer patients treated with EBRT and brachytherapy boost and reported 92% LTC at 5 years . In addition, Beitler et al. demonstrated a series of 29 patients treated with brachytherapy boost. Local tumour control was 93% . Although BT is very effective and gives high local control rates, it usually involves the need for another ‘small surgery’ procedure with general anaesthesia and the risk of surgical complications. Among those who experienced acute side effects, all were scored as very mild according to the Dische scale. None of the patients experienced serious late complications such as osteonecrosis of the mandible or rupture of the artery. According to the literature, a dose of IORT below 15 Gy is safe in the majority of cases, which is higher than the dose given to our patients . Furthermore, in the analysed group, the locations of tumour beds were distant from the carotid and usually separated enough from bone to receive high dose. The good treatment tolerance demonstrated that low-energy X-ray IORT with a dose of 5–7.5 Gy could be considered a safe method for boost therapy. Over the past decade, stereotactic hypofractionated radiotherapy (SHRT) has been demonstrated to be highly effective and has emerged as a compelling alternative to IORT. Despite that, both therapeutic modalities are restricted to relatively small local tumours . However, it does not eliminate the use of IORT, but the time between IORT surgery and EBRT should be as short as possible. If the duration is very prolonged, the biological efficacy of the IORT dose will be diminished, thereby increasing the likelihood of local recurrence and distant failure. However, due to the small number of cases included in the present study, it is not possible to evaluate the effect of the IORT dose. Considering the encouraging results of our study and its comparable efficacy with other boost modalities reported by the cited authors, it seems justified to confirm the efficacy of IORT boost in a prospective multicentre clinical trial comparing it with SBRT or BT boost. Low-energy X-ray IORT applied as a boost to conventionally fractionated radiotherapy is a safe and effective therapeutic modality for patients with oral cavity cancer. IORT likely reduces the risk of local recurrences, but the time interval between surgery and EBRT should be as short as possible and therefore should be planned, before the start of combined therapy. To clearly confirm the effectiveness of the IORT boost for oral cavity cancer patients, a further prospective multicentre clinical trial is needed. |
The Relationship between Workplace Environment and Metabolic Syndrome | 842dff46-30a5-4139-af10-12aab8664302 | 6466990 | Preventive Medicine[mh] | The workplace environment is associated with metabolic syndrome. The prevalence of metabolic syndrome was 19.8% among a group of Korean blue-collar male workers. Those chronically exposed to metalworking fluid is more likely to develop metabolic syndrome compared with nonexposed workers (OR 1.79, 95% CI 1.06 to 3.01). The prevalence of metabolic syndrome, a risk factor for cardiovascular disease and diabetes, varies from 11.9% to 37.1% in most Asian countries, and show a continuously increasing trend in some countries. In Korea, metabolic syndrome is recognized as a major health problem. The prevalence of metabolic syndrome in Korean adults is reported to be 28.9% in 2013. Lifestyle is an important factor in the development of metabolic syndrome; so do sex and age. Smoking, alcohol consumption and physical activity were common lifestyle factors associated with metabolic syndrome. , Type of occupation is also important in development of metabolic syndrome. For example, the incidence of metabolic syndrome in the white collar workers is higher than other male workers. Those with sedentary or shift work carry a higher risk of metabolic syndrome. , The incidence of metabolic syndrome is 2.3-fold higher in those working for 10 or more hours per day. Workplace environment may also affect the occurrence of metabolic syndrome. In animal experiments, it was confirmed that chronic noise increases blood glucose. According to the Occupational Safety and Health Act in South Korea, there are three occupational health services provided to workers, including periodic occupational medical examinations, workplace environment measurements, and health management services. Korean workers who particularly exposed to 177 hazardous substances and environments must receive certain health examinations. However, studies investigating the relationship between work environment and metabolic syndrome are scarce. We therefore, conducted this study to determine the relationship between work environment and metabolic syndrome among a group of Korean blue-collar male workers. Using medical records, this cross-sectional study was conducted on 1334 Korean blue-collar workers who received special health checkup at a health center affiliated to a university hospital in Gyeongju, South Korea, from March 1 to July 31, 2016. According to the Occupational Safety and Health Act of South Korea, we classified the workplace environment into six categories based on exposure to organic compounds, metals, acids and bases, metalworking fluid, dust, and noise. The categorization was based on the 177 occupational hazardous substances using the individual worker's reports for workplace environment measurement in the same workplace (Appendix). We also identified shift workers. Data collected included demographic variables and results of blood tests. In this study, we used the diagnostic criteria of the National Cholesterol Education Program's Adults Treatment Panel (NCEP ATP) III to identify metabolic syndrome. Alcohol consumption was defined as “drinking more than 14 glasses over a week regardless of the main alcoholic beverage.” For assessment of physical activity, short form of the Korean version of the International Physical Activity Questionnaire (IPAQ) was used. The reliability and validity of the questionnaire were approved earlier. The questionnaire consists of seven questions on vigorous activities ( eg , aerobics, fasting bicycling), moderate activities ( eg , double tennis, bicycling at a regular pace), walking activities and inactivity for last seven days. The total physical activities converted to total METs using the following equations: Total METs (week) = (3.3×Walking minutes×days) + (4.0×Moderate activities minutes×days) + (8.0×Vigorous activities) where, “Walking” is equivalent to 3.3 METs, “Moderate activity” is 4.0 METs, and “Vigorous activities” is 8.0 METs. A physical activity <600 METs was considered “lack of physical activity.” Ethics All studied participants gave written informed consent. The study protocol was approved by the same hospital Institutional Review Border. Statistical Analysis SPSS ver 20.0 (SPSS Inc, USA) was used for data analyses. Logistic regression analysis was used to determine the independent work environmental factors affecting the development of metabolic syndrome after adjusting for demographic and lifestyle factors. A p value <0.05 was considered statistically significant. All studied participants gave written informed consent. The study protocol was approved by the same hospital Institutional Review Border. SPSS ver 20.0 (SPSS Inc, USA) was used for data analyses. Logistic regression analysis was used to determine the independent work environmental factors affecting the development of metabolic syndrome after adjusting for demographic and lifestyle factors. A p value <0.05 was considered statistically significant. Of 1337 workers, 1297 (97.0%) were included in the final analysis, excluding women and those with missing data. Characteristics of studied workers are presented in . Two-hundred and fourteen (16.5%) workers were taking medication for hypertension or diabetes. The prevalence of metabolic syndrome was 19.8% (95% CI 17.6% to 22.0%). The independent predictors of metabolic syndrome were age, body mass index (BMI), smoking, alcohol consumption, and exposure to metalworking fluid . Although exposure to dust was found a significant risk factor of metabolic syndrome in univariate analysis, the significance was abolished after adjusting for confounders . The prevalence of metabolic syndrome in Korean male adults is 30.8%. The prevalence observed in this study was 19.8%. The observed difference may be attributed to the age difference of the studied subjects. Furthermore, we studied blue collar workers rather than office workers. Metalworking fluid, commonly used in work for machining, is a liquid used to reduce heat and friction during metal work. We found that those handling metalworking fluid were more likely to develop metabolic syndrome. This was still significant after adjusting for age, BMI, smoking, and alcohol consumption, which are known to be associated with metabolic syndrome. Because metalworking fluid exists in the form of mist, some are absorbed by inhalation and some through the skin. Existing medical conditions related to metalworking fluid are respiratory diseases such as work-related asthma and hypersensitivity pneumonitis. A recent observational study about exposure to metalworking fluid and disease outbreaks for 10 years revealed a the exposure causes skin diseases more often than respiratory illnesses. In addition, some studies report the association between development of certain cancers and exposure to metalworking fluid. The mechanism through which metalworking fluid causes respiratory disease is believed to be mainly inhalation of mist contaminated with micro-organisms and various chemical additives in the fluid that are known to be stimulants of the respiratory tract. On the other hand, animal experiments show that some components of the metalworking fluid would cause acute inflammation and oxidative stress in the body. In fact, the low-level persistent inflammatory response exists in those chronically exposed to meta working fluid, is recognized as an important factor in the pathogenesis of metabolic syndrome. This mechanism may explain the association between exposure to metalworking fluid and metabolic syndrome observed in our study. This study had some limitations. First, according to the law, workers in this study were wearing personal protective equipment and we could not assess the exposure dosage of the studied environmental factors. Second, because the studied workers in this study worked in only one workplace, the result cannot be generalized. In conclusion, the workplace environment is associated with metabolic syndrome. It seems that workers who are handling metalworking fluid would be particularly at higher risk of developing metabolic syndrome. This work was supported by the Dongguk University Research Fund, 2014. None declared. None. |
Level and comfort of caregiver–young adolescent communication on sexual and reproductive health: a cross-sectional survey in south-western Uganda | f61f823b-6357-4fd7-b09c-b7f0b4c75a4b | 9675188 | Health Communication[mh] | The sexual and reproductive health (SRH) of young adolescents (YAs: 10–14 years) is an emerging public health priority in developing countries. YAs comprise about half of the 1.2 billion adolescents aged 10–19 years globally . Young adolescence is often regarded as a relatively healthy phase compared to other age groups . Nevertheless, it is a period of profound changes characterized by the onset of puberty, which comes with physical, emotional, social and cognitive changes that affect their well-being, as well as their sense of self and self-esteem, and the ability to assess risks and consequences . Previous research indicates that puberty accelerates risk-taking among YAs . At this stage of their life, YAs are initiating intimate relationships and acts such as kissing, hugging and fondling . Studies have also shown that they are already engaging in sexual activities, including sexual intercourse . YAs in developing countries are disproportionately affected by SRH challenges, including coerced or forced sex, early marriage and gender-based violence . These often culminate in early, unintended and unwanted pregnancies and sexually transmitted infections (STIs), including HIV . YAs also lack information, knowledge, skills and cognitive readiness to make informed decisions related to their SRH, including consensual sex, and condom and contraceptive use . Furthermore, gender norms that depict boys as virile and girls as weak and vulnerable often intensify these risks . Caregivers play a significant role in socializing and shaping the attitudes of YAs at an early age that are critical to laying the foundations for positive and safe SRH behaviours. This is through practices such as gender socialization and communication about sexuality in general . Blum’s conceptual framework on early adolescence underscores the significant role of caregivers as part of the micro-environment that influences positive SRH outcomes for YAs . Studies have found a strong association between caregiver SRH communication and reduced sexual risk-taking behaviours among adolescents , including delayed sexual initiation and safe SRH practices . Other studies further point to the need to start SRH discussions at an early age, and to provide accurate SRH information . There is considerable research into communication between caregivers and children about SRH in sub-Saharan Africa that emphasizes the existence of such communication despite traditional perspectives . This SRH communication is often punitive , limited in breadth to comfortable topics such as abstinence , and less about broader SRH topics such as prevention of pregnancy through contraception and the use of condoms . Moreover, many caregivers do not approve of YAs engaging in sexual and romantic relationships, since they are deemed too young and, therefore, not ready to receive SRH information . Caregivers are also not in the position to decipher SRH topics, due to cultural and religious dispositions that inhibit explicit discussions about sex . Low self-efficacy of caregivers, as well as uncertainties about the appropriate timing of sexuality communication, impede sexuality communication . Other structural factors such as caregiver–child connectedness and socio-economic factors may influence communication between caregivers and children. Studies also report a substantial variation in caregiver–child communication by gender, with more pronounced communication between mothers and daughters . Current research specifically into caregiver–child communication on SRH emphasizes older adolescents and barely addresses YAs . Moreover, several studies on caregiver–child communication on SRH present evidence on the level and frequency of SRH communication, but hardly any on the level of comfort with discussing SRH with YAs. There is barely any research assessing correlates of SRH communication and comfort with discussing SRH with YAs in settings where sexuality communication is a cultural taboo. Our research presents data derived from a baseline household survey of caregivers and their YAs (10–14 years) in a community-based participatory research project in rural south-western Uganda. The project aims to improve caregiver communication with YAs through a culturally sensitive intervention targeting caregivers. This paper has two objectives: to describe the current level of, and comfort with, caregiver SRH communication with their children, and to identify their correlates. Study design and setting A cross-sectional household survey was conducted in January and February 2020 among caregiver–YA dyads in six villages in Rwebishekye parish, Rwanyamahembe sub-county, Kashari county in rural Mbarara district of south-western Uganda. The study community comprised approximately 1,520 households, of which 29% headed by women, and an estimated population of 6,061 people . The community comprised a relatively homogenous and stable population with one main linguistic group, the Banyankore-Bakiga. The community is served by one public health facility (Bwizibwera Health Centre IV), located about 5 km from the furthest village. Study population and sample selection The study sampling frame comprised all households in the study community with YAs (10–14 years) and their caregivers. A community household profiling exercise was conducted at the start of the study and established an estimate of 300 households comprising YAs. The final sample comprised 218 caregiver–YA dyads (436 study participants overall). The sample size was calculated for the effectiveness study for an intervention to improve SRH communication between caregivers and YAs. It allows a moderate change (effect size 0.2) to be measured in good caregiver–adolescent communication between pre- and post-intervention measurements with a power of 0.8 and alpha of 0.05. This required a total sample size of 277 respondents. Accounting for design effect (× 1.3) and drop-out between waves (× 0.2), the required sample size amounted to 432 participants or 216 dyads. We used consecutive sampling, based on whether a household contained a YA and whether both caregiver and YA were present simultaneously at the time of the survey. For households comprising more than one YA, we considered the oldest. Caregivers were either biological or non-biological. Within the sample, caregivers included biological parents, step-parents, foster parents or relatives, including older siblings entrusted with the greatest responsibility for the daily care and rearing of the child. Eligibility for caregivers included being 18 years or older, consenting to participate in the study, living in the community for the past six months and living with a YA in their household for whom they were the caregiver for the past six months. Data collection The survey was conducted by trained research assistants who were fluent in English and native speakers of Runyankore-Rukiga. Data were collected using a structured, pre-tested questionnaire administered by an interviewer. The surveys were computer-assisted using Kobo Collect software. The interviews with the caregiver and YA were conducted simultaneously but separately in convenient locations to avoid overhearing and to ensure open and truthful responses. The time for completion of the survey varied from one hour to one hour and 15 min. The survey team was coordinated by two team leaders and community leaders, who assisted in identifying the preselected households. There were also three monitors to check the data for consistency and completeness. Community advisory board Given the participatory and sensitive nature of this research project, the survey questionnaire was reviewed by a multidisciplinary team of researchers and the Community Advisory Board (CAB) in December 2019. The CAB comprised community representatives, including caregivers, young people, teachers, community leaders and influential members of the community, as well as religious leaders from the four majority faiths: Catholic, Anglican, Muslim and Pentecostal. The CAB also included representatives of different government entities, including the Ministry of Health, the Ministry of Gender, Labour and Social Development, and the Ministry of Education and Sports. These stakeholders reviewed the data collection tools and provided feedback on the pertinence and clarity of the survey questions. Measures Dependent variables SRH communication Caregiver–YA SRH communication was explored using 10 SRH-related topics indicated in Table . The scale is adapted from the parent-adolescent communication scale (PACS) . However this was adapted based on recommendations from the community advisory board (CAB). Caregivers were asked if they had ever had a discussion on any of the 10 topics. They were presented with the statement ‘Have you ever talked to your child about general health and bodily hygiene?’ The response options were ‘Yes’ and ‘No’. The number of topics discussed was summed, frequencies were run for each response, and the level of communication was stratified by dyad. P-values were based on Fisher’s Exact Test, due to the small number of participants in each dyad. Comfort with SRH discussions Caregiver comfort with SRH communication was explored using nine SRH topics. Caregivers were asked how comfortable they were discussing any of the SRH topics with their YA children. The topics are indicated in Additional file : Appendix A. Caregivers were presented with statements such as ‘How comfortable do you feel discussing general health and bodily hygiene with your YA child?’ The response options for caregivers were ‘very comfortable’, ‘somewhat comfortable’, ‘somewhat uncomfortable’ and ‘very uncomfortable.’ The summated composite score for comfort was calculated with a minimum score of 10 and maximum score of 27. The scores were classified based on Bloom’s criteria . These were organized into 3 groups; scores 22–27 (80–100%) were reported as high comfort; scores 16–21.99 (60–79%) were reported as moderate comfort while scores < 16 (< 60%) were reported as low comfort with SRH discussions. This scale had a Cronbach alpha of 0.73. Independent variables Background characteristics Information on socio-demographic variables of caregivers, including age, sex, marital status and religious affiliation, was obtained . The questionnaire included questions on the number of YAs living in the household at the time of the survey, dyad type and the parenting structure of the household (single-parent or two-parent household). Household socio-economic status (SES) This was measured using variables from the Uganda Bureau of Statistics socio-economic survey . Parameters such as water source (location and the time it takes to reach it), housing characteristics and asset ownership were used to measure SES. They were combined into a proxy indicator—wealth index—using principal component analysis . SES was transformed into an overall variable and recoded as low, medium or high. Connectedness between the caregiver and the YA Connectedness was measured using three subscales. The parent involvement subscale comprised 10 items, and the positive parenting scale comprised 6 items. Both scales were drawn from the Alabama Parenting Questionnaire, whose target audience is caregivers of children aged 6–18 years . This questionnaire measures five dimensions of parenting that are relevant to the etiology and treatment of children’s externalizing problems . Five Likert-type items were used to assess parental involvement and positive parenting—for example, ‘You have a friendly talk with your child.’ The scores were 5 = always, 4 = often, 3 = sometimes, 2 = almost never and 1 = never. The parental expertise and accessibility scale comprised nine items which assessed both the caregivers’ and the adolescents’ perceptions of the caregivers’ expertise, trustworthiness and accessibility. It is intended for early adolescents (11–14 years) but was adapted for male caregivers in this study. Five Likert-type items were used to assess this scale—for example, ‘My child thinks I give good advice.’ The scores were 1 = strongly agree, 2 = moderately agree, 3 = neither agree nor disagree, 4 = moderately disagree and 5 = strongly disagree . Attitudes towards SRH issues of YAs Attitudes were measured using an eight-item scale on a five-point Likert scale. Caregivers were presented with statements such as ‘You approve of your child having a boyfriend or girlfriend.’ The scores were strongly agree, moderately agree, neither agree nor disagree, moderately disagree and strongly disagree. The scale was scored based on the highest and the lowest scores, with a high score indicating a positive attitude and a low score indicating a negative attitude. The scores were reversed to allow a high score to be indicated as a positive attitude. The summated composite score for attitude was calculated with a minimum score of 8 and maximum score of 24. The scores were classified based on Bloom’s criteria . These were organized into 3 groups; scores 19–24 (80–100%) were reported as positive attitude; scores 14–18.99 (60–79%) were reported as neutral while scores < 14 (< 60%) were reported as negative attitude. The scale had a Cronbach alpha of 0.56. SRH knowledge This was measured through 27 items to assess knowledge on three main sub-topics: puberty (7 questions), HIV/AIDS (13 questions) and pregnancy prevention (7 questions). A summary score was computed, with the highest score indicating a high level of knowledge and the lowest score indicating a low level of knowledge.. The summated composite score for knowledge was calculated with a minimum score of 49 and maximum score of 81. The scores were classified based on Bloom’s criteria . These were organized into 3 groups; scores 65–81 (80–100%) were reported as high knowledge; scores 49–64.99 (60–79%) were reported as moderate knowledge while scores < 49 (< 60%) were reported as low knowledge. Data analysis Data analysis was performed using STATA 14 (College Station, Texas, USA). Descriptive statistics were used to describe numbers and percentages for the dependent and independent variables. The prevalence of discussion for each of the 10 SRH topics was presented by dyad type. Fischer’s Exact Tests were used to test for the level of significance of the difference in SRH communication across the dyad type for each of the 10 SRH topics (a 5% level of significance was set). The mean score for the number of topics discussed across the dyads was presented. Bivariate analysis was performed between the dependent variables (level of SRH communication and caregivers’ comfort with SRH discussions) and independent variables. The dependent variables were treated as linear variables (they were normally distributed). We conducted hierarchical linear regression analyses to examine the relationship between the dependent variables (level of SRH communication and caregivers’ comfort with SRH discussions) and the independent variables (demographic characteristics of caregivers, household characteristics, level of comfort, attitudes towards SRH, knowledge of SRH and level of connectedness). Separate linear regression models for number of SRH topics discussed and caregivers’ comfort with SRH discussions were run using a manual backward stepwise selection method. Multi-collinearity was tested using variance inflation factors; none of the variables were affected. Results from the bivariate and multivariate linear regression model for predictors of caregiver and YA communication and comfort with SRH communication are reported in Tables , , and , respectively. Results from the bivariate analysis informed which variables to include in the multivariate linear regression model. A cross-sectional household survey was conducted in January and February 2020 among caregiver–YA dyads in six villages in Rwebishekye parish, Rwanyamahembe sub-county, Kashari county in rural Mbarara district of south-western Uganda. The study community comprised approximately 1,520 households, of which 29% headed by women, and an estimated population of 6,061 people . The community comprised a relatively homogenous and stable population with one main linguistic group, the Banyankore-Bakiga. The community is served by one public health facility (Bwizibwera Health Centre IV), located about 5 km from the furthest village. The study sampling frame comprised all households in the study community with YAs (10–14 years) and their caregivers. A community household profiling exercise was conducted at the start of the study and established an estimate of 300 households comprising YAs. The final sample comprised 218 caregiver–YA dyads (436 study participants overall). The sample size was calculated for the effectiveness study for an intervention to improve SRH communication between caregivers and YAs. It allows a moderate change (effect size 0.2) to be measured in good caregiver–adolescent communication between pre- and post-intervention measurements with a power of 0.8 and alpha of 0.05. This required a total sample size of 277 respondents. Accounting for design effect (× 1.3) and drop-out between waves (× 0.2), the required sample size amounted to 432 participants or 216 dyads. We used consecutive sampling, based on whether a household contained a YA and whether both caregiver and YA were present simultaneously at the time of the survey. For households comprising more than one YA, we considered the oldest. Caregivers were either biological or non-biological. Within the sample, caregivers included biological parents, step-parents, foster parents or relatives, including older siblings entrusted with the greatest responsibility for the daily care and rearing of the child. Eligibility for caregivers included being 18 years or older, consenting to participate in the study, living in the community for the past six months and living with a YA in their household for whom they were the caregiver for the past six months. The survey was conducted by trained research assistants who were fluent in English and native speakers of Runyankore-Rukiga. Data were collected using a structured, pre-tested questionnaire administered by an interviewer. The surveys were computer-assisted using Kobo Collect software. The interviews with the caregiver and YA were conducted simultaneously but separately in convenient locations to avoid overhearing and to ensure open and truthful responses. The time for completion of the survey varied from one hour to one hour and 15 min. The survey team was coordinated by two team leaders and community leaders, who assisted in identifying the preselected households. There were also three monitors to check the data for consistency and completeness. Given the participatory and sensitive nature of this research project, the survey questionnaire was reviewed by a multidisciplinary team of researchers and the Community Advisory Board (CAB) in December 2019. The CAB comprised community representatives, including caregivers, young people, teachers, community leaders and influential members of the community, as well as religious leaders from the four majority faiths: Catholic, Anglican, Muslim and Pentecostal. The CAB also included representatives of different government entities, including the Ministry of Health, the Ministry of Gender, Labour and Social Development, and the Ministry of Education and Sports. These stakeholders reviewed the data collection tools and provided feedback on the pertinence and clarity of the survey questions. Dependent variables SRH communication Caregiver–YA SRH communication was explored using 10 SRH-related topics indicated in Table . The scale is adapted from the parent-adolescent communication scale (PACS) . However this was adapted based on recommendations from the community advisory board (CAB). Caregivers were asked if they had ever had a discussion on any of the 10 topics. They were presented with the statement ‘Have you ever talked to your child about general health and bodily hygiene?’ The response options were ‘Yes’ and ‘No’. The number of topics discussed was summed, frequencies were run for each response, and the level of communication was stratified by dyad. P-values were based on Fisher’s Exact Test, due to the small number of participants in each dyad. Comfort with SRH discussions Caregiver comfort with SRH communication was explored using nine SRH topics. Caregivers were asked how comfortable they were discussing any of the SRH topics with their YA children. The topics are indicated in Additional file : Appendix A. Caregivers were presented with statements such as ‘How comfortable do you feel discussing general health and bodily hygiene with your YA child?’ The response options for caregivers were ‘very comfortable’, ‘somewhat comfortable’, ‘somewhat uncomfortable’ and ‘very uncomfortable.’ The summated composite score for comfort was calculated with a minimum score of 10 and maximum score of 27. The scores were classified based on Bloom’s criteria . These were organized into 3 groups; scores 22–27 (80–100%) were reported as high comfort; scores 16–21.99 (60–79%) were reported as moderate comfort while scores < 16 (< 60%) were reported as low comfort with SRH discussions. This scale had a Cronbach alpha of 0.73. Independent variables Background characteristics Information on socio-demographic variables of caregivers, including age, sex, marital status and religious affiliation, was obtained . The questionnaire included questions on the number of YAs living in the household at the time of the survey, dyad type and the parenting structure of the household (single-parent or two-parent household). Household socio-economic status (SES) This was measured using variables from the Uganda Bureau of Statistics socio-economic survey . Parameters such as water source (location and the time it takes to reach it), housing characteristics and asset ownership were used to measure SES. They were combined into a proxy indicator—wealth index—using principal component analysis . SES was transformed into an overall variable and recoded as low, medium or high. Connectedness between the caregiver and the YA Connectedness was measured using three subscales. The parent involvement subscale comprised 10 items, and the positive parenting scale comprised 6 items. Both scales were drawn from the Alabama Parenting Questionnaire, whose target audience is caregivers of children aged 6–18 years . This questionnaire measures five dimensions of parenting that are relevant to the etiology and treatment of children’s externalizing problems . Five Likert-type items were used to assess parental involvement and positive parenting—for example, ‘You have a friendly talk with your child.’ The scores were 5 = always, 4 = often, 3 = sometimes, 2 = almost never and 1 = never. The parental expertise and accessibility scale comprised nine items which assessed both the caregivers’ and the adolescents’ perceptions of the caregivers’ expertise, trustworthiness and accessibility. It is intended for early adolescents (11–14 years) but was adapted for male caregivers in this study. Five Likert-type items were used to assess this scale—for example, ‘My child thinks I give good advice.’ The scores were 1 = strongly agree, 2 = moderately agree, 3 = neither agree nor disagree, 4 = moderately disagree and 5 = strongly disagree . Attitudes towards SRH issues of YAs Attitudes were measured using an eight-item scale on a five-point Likert scale. Caregivers were presented with statements such as ‘You approve of your child having a boyfriend or girlfriend.’ The scores were strongly agree, moderately agree, neither agree nor disagree, moderately disagree and strongly disagree. The scale was scored based on the highest and the lowest scores, with a high score indicating a positive attitude and a low score indicating a negative attitude. The scores were reversed to allow a high score to be indicated as a positive attitude. The summated composite score for attitude was calculated with a minimum score of 8 and maximum score of 24. The scores were classified based on Bloom’s criteria . These were organized into 3 groups; scores 19–24 (80–100%) were reported as positive attitude; scores 14–18.99 (60–79%) were reported as neutral while scores < 14 (< 60%) were reported as negative attitude. The scale had a Cronbach alpha of 0.56. SRH knowledge This was measured through 27 items to assess knowledge on three main sub-topics: puberty (7 questions), HIV/AIDS (13 questions) and pregnancy prevention (7 questions). A summary score was computed, with the highest score indicating a high level of knowledge and the lowest score indicating a low level of knowledge.. The summated composite score for knowledge was calculated with a minimum score of 49 and maximum score of 81. The scores were classified based on Bloom’s criteria . These were organized into 3 groups; scores 65–81 (80–100%) were reported as high knowledge; scores 49–64.99 (60–79%) were reported as moderate knowledge while scores < 49 (< 60%) were reported as low knowledge. Data analysis Data analysis was performed using STATA 14 (College Station, Texas, USA). Descriptive statistics were used to describe numbers and percentages for the dependent and independent variables. The prevalence of discussion for each of the 10 SRH topics was presented by dyad type. Fischer’s Exact Tests were used to test for the level of significance of the difference in SRH communication across the dyad type for each of the 10 SRH topics (a 5% level of significance was set). The mean score for the number of topics discussed across the dyads was presented. Bivariate analysis was performed between the dependent variables (level of SRH communication and caregivers’ comfort with SRH discussions) and independent variables. The dependent variables were treated as linear variables (they were normally distributed). We conducted hierarchical linear regression analyses to examine the relationship between the dependent variables (level of SRH communication and caregivers’ comfort with SRH discussions) and the independent variables (demographic characteristics of caregivers, household characteristics, level of comfort, attitudes towards SRH, knowledge of SRH and level of connectedness). Separate linear regression models for number of SRH topics discussed and caregivers’ comfort with SRH discussions were run using a manual backward stepwise selection method. Multi-collinearity was tested using variance inflation factors; none of the variables were affected. Results from the bivariate and multivariate linear regression model for predictors of caregiver and YA communication and comfort with SRH communication are reported in Tables , , and , respectively. Results from the bivariate analysis informed which variables to include in the multivariate linear regression model. SRH communication Caregiver–YA SRH communication was explored using 10 SRH-related topics indicated in Table . The scale is adapted from the parent-adolescent communication scale (PACS) . However this was adapted based on recommendations from the community advisory board (CAB). Caregivers were asked if they had ever had a discussion on any of the 10 topics. They were presented with the statement ‘Have you ever talked to your child about general health and bodily hygiene?’ The response options were ‘Yes’ and ‘No’. The number of topics discussed was summed, frequencies were run for each response, and the level of communication was stratified by dyad. P-values were based on Fisher’s Exact Test, due to the small number of participants in each dyad. Caregiver–YA SRH communication was explored using 10 SRH-related topics indicated in Table . The scale is adapted from the parent-adolescent communication scale (PACS) . However this was adapted based on recommendations from the community advisory board (CAB). Caregivers were asked if they had ever had a discussion on any of the 10 topics. They were presented with the statement ‘Have you ever talked to your child about general health and bodily hygiene?’ The response options were ‘Yes’ and ‘No’. The number of topics discussed was summed, frequencies were run for each response, and the level of communication was stratified by dyad. P-values were based on Fisher’s Exact Test, due to the small number of participants in each dyad. Caregiver comfort with SRH communication was explored using nine SRH topics. Caregivers were asked how comfortable they were discussing any of the SRH topics with their YA children. The topics are indicated in Additional file : Appendix A. Caregivers were presented with statements such as ‘How comfortable do you feel discussing general health and bodily hygiene with your YA child?’ The response options for caregivers were ‘very comfortable’, ‘somewhat comfortable’, ‘somewhat uncomfortable’ and ‘very uncomfortable.’ The summated composite score for comfort was calculated with a minimum score of 10 and maximum score of 27. The scores were classified based on Bloom’s criteria . These were organized into 3 groups; scores 22–27 (80–100%) were reported as high comfort; scores 16–21.99 (60–79%) were reported as moderate comfort while scores < 16 (< 60%) were reported as low comfort with SRH discussions. This scale had a Cronbach alpha of 0.73. Background characteristics Information on socio-demographic variables of caregivers, including age, sex, marital status and religious affiliation, was obtained . The questionnaire included questions on the number of YAs living in the household at the time of the survey, dyad type and the parenting structure of the household (single-parent or two-parent household). Household socio-economic status (SES) This was measured using variables from the Uganda Bureau of Statistics socio-economic survey . Parameters such as water source (location and the time it takes to reach it), housing characteristics and asset ownership were used to measure SES. They were combined into a proxy indicator—wealth index—using principal component analysis . SES was transformed into an overall variable and recoded as low, medium or high. Connectedness between the caregiver and the YA Connectedness was measured using three subscales. The parent involvement subscale comprised 10 items, and the positive parenting scale comprised 6 items. Both scales were drawn from the Alabama Parenting Questionnaire, whose target audience is caregivers of children aged 6–18 years . This questionnaire measures five dimensions of parenting that are relevant to the etiology and treatment of children’s externalizing problems . Five Likert-type items were used to assess parental involvement and positive parenting—for example, ‘You have a friendly talk with your child.’ The scores were 5 = always, 4 = often, 3 = sometimes, 2 = almost never and 1 = never. The parental expertise and accessibility scale comprised nine items which assessed both the caregivers’ and the adolescents’ perceptions of the caregivers’ expertise, trustworthiness and accessibility. It is intended for early adolescents (11–14 years) but was adapted for male caregivers in this study. Five Likert-type items were used to assess this scale—for example, ‘My child thinks I give good advice.’ The scores were 1 = strongly agree, 2 = moderately agree, 3 = neither agree nor disagree, 4 = moderately disagree and 5 = strongly disagree . Attitudes towards SRH issues of YAs Attitudes were measured using an eight-item scale on a five-point Likert scale. Caregivers were presented with statements such as ‘You approve of your child having a boyfriend or girlfriend.’ The scores were strongly agree, moderately agree, neither agree nor disagree, moderately disagree and strongly disagree. The scale was scored based on the highest and the lowest scores, with a high score indicating a positive attitude and a low score indicating a negative attitude. The scores were reversed to allow a high score to be indicated as a positive attitude. The summated composite score for attitude was calculated with a minimum score of 8 and maximum score of 24. The scores were classified based on Bloom’s criteria . These were organized into 3 groups; scores 19–24 (80–100%) were reported as positive attitude; scores 14–18.99 (60–79%) were reported as neutral while scores < 14 (< 60%) were reported as negative attitude. The scale had a Cronbach alpha of 0.56. SRH knowledge This was measured through 27 items to assess knowledge on three main sub-topics: puberty (7 questions), HIV/AIDS (13 questions) and pregnancy prevention (7 questions). A summary score was computed, with the highest score indicating a high level of knowledge and the lowest score indicating a low level of knowledge.. The summated composite score for knowledge was calculated with a minimum score of 49 and maximum score of 81. The scores were classified based on Bloom’s criteria . These were organized into 3 groups; scores 65–81 (80–100%) were reported as high knowledge; scores 49–64.99 (60–79%) were reported as moderate knowledge while scores < 49 (< 60%) were reported as low knowledge. Information on socio-demographic variables of caregivers, including age, sex, marital status and religious affiliation, was obtained . The questionnaire included questions on the number of YAs living in the household at the time of the survey, dyad type and the parenting structure of the household (single-parent or two-parent household). This was measured using variables from the Uganda Bureau of Statistics socio-economic survey . Parameters such as water source (location and the time it takes to reach it), housing characteristics and asset ownership were used to measure SES. They were combined into a proxy indicator—wealth index—using principal component analysis . SES was transformed into an overall variable and recoded as low, medium or high. Connectedness was measured using three subscales. The parent involvement subscale comprised 10 items, and the positive parenting scale comprised 6 items. Both scales were drawn from the Alabama Parenting Questionnaire, whose target audience is caregivers of children aged 6–18 years . This questionnaire measures five dimensions of parenting that are relevant to the etiology and treatment of children’s externalizing problems . Five Likert-type items were used to assess parental involvement and positive parenting—for example, ‘You have a friendly talk with your child.’ The scores were 5 = always, 4 = often, 3 = sometimes, 2 = almost never and 1 = never. The parental expertise and accessibility scale comprised nine items which assessed both the caregivers’ and the adolescents’ perceptions of the caregivers’ expertise, trustworthiness and accessibility. It is intended for early adolescents (11–14 years) but was adapted for male caregivers in this study. Five Likert-type items were used to assess this scale—for example, ‘My child thinks I give good advice.’ The scores were 1 = strongly agree, 2 = moderately agree, 3 = neither agree nor disagree, 4 = moderately disagree and 5 = strongly disagree . Attitudes were measured using an eight-item scale on a five-point Likert scale. Caregivers were presented with statements such as ‘You approve of your child having a boyfriend or girlfriend.’ The scores were strongly agree, moderately agree, neither agree nor disagree, moderately disagree and strongly disagree. The scale was scored based on the highest and the lowest scores, with a high score indicating a positive attitude and a low score indicating a negative attitude. The scores were reversed to allow a high score to be indicated as a positive attitude. The summated composite score for attitude was calculated with a minimum score of 8 and maximum score of 24. The scores were classified based on Bloom’s criteria . These were organized into 3 groups; scores 19–24 (80–100%) were reported as positive attitude; scores 14–18.99 (60–79%) were reported as neutral while scores < 14 (< 60%) were reported as negative attitude. The scale had a Cronbach alpha of 0.56. This was measured through 27 items to assess knowledge on three main sub-topics: puberty (7 questions), HIV/AIDS (13 questions) and pregnancy prevention (7 questions). A summary score was computed, with the highest score indicating a high level of knowledge and the lowest score indicating a low level of knowledge.. The summated composite score for knowledge was calculated with a minimum score of 49 and maximum score of 81. The scores were classified based on Bloom’s criteria . These were organized into 3 groups; scores 65–81 (80–100%) were reported as high knowledge; scores 49–64.99 (60–79%) were reported as moderate knowledge while scores < 49 (< 60%) were reported as low knowledge. Data analysis was performed using STATA 14 (College Station, Texas, USA). Descriptive statistics were used to describe numbers and percentages for the dependent and independent variables. The prevalence of discussion for each of the 10 SRH topics was presented by dyad type. Fischer’s Exact Tests were used to test for the level of significance of the difference in SRH communication across the dyad type for each of the 10 SRH topics (a 5% level of significance was set). The mean score for the number of topics discussed across the dyads was presented. Bivariate analysis was performed between the dependent variables (level of SRH communication and caregivers’ comfort with SRH discussions) and independent variables. The dependent variables were treated as linear variables (they were normally distributed). We conducted hierarchical linear regression analyses to examine the relationship between the dependent variables (level of SRH communication and caregivers’ comfort with SRH discussions) and the independent variables (demographic characteristics of caregivers, household characteristics, level of comfort, attitudes towards SRH, knowledge of SRH and level of connectedness). Separate linear regression models for number of SRH topics discussed and caregivers’ comfort with SRH discussions were run using a manual backward stepwise selection method. Multi-collinearity was tested using variance inflation factors; none of the variables were affected. Results from the bivariate and multivariate linear regression model for predictors of caregiver and YA communication and comfort with SRH communication are reported in Tables , , and , respectively. Results from the bivariate analysis informed which variables to include in the multivariate linear regression model. Participant characteristics A total of 218 caregivers were enrolled in the study, of which 76% were women. The mean age of the caregivers was 44.9 years (SD = 12.61). Seventy-three per cent were biological caregivers, and the majority (68.3%) had attained primary education, while 11.9% had not received any formal education. The caregiver–YA dyads comprised 96 with female caregiver and daughter (44.0%), 68 male caregiver and daughter (31.2%), 30 female caregiver and son (13.8%) and 24 male caregiver and son 24 (11%). The majority (76.6%) of the households were two-caregiver households, with an average of two YAs (Table ). Descriptive analysis Level of SRH communication Ten SRH topics were explored in the study. None of the respondents discussed all 10 topics. Two per cent of the caregivers reported ever discussing nine SRH topics with the YAs, while 3.5% reported never discussing any topics. The mean number of topics ever discussed was 3.9. Twenty-two per cent of the caregivers reported discussing at least three of the topics, and 7% reported discussing at least one of the topics. Overall, general health and bodily hygiene was discussed by majority of the dyads (89.9%), followed by HIV/AIDS and other STIs (77.5%). In contrast, only 4.3% of the dyads discussed night emissions in boys. There was no significant difference in communication of SRH topics across the dyads, except for HIV/AIDS and other STIs, which were more likely to be discussed in dyads with female caregivers ( p < 0.05) (Table ). Caregivers’ comfort with SRH discussions with young adolescents The majority of the caregivers (63.4%) reported a high level of comfort with SRH discussions with YAs, 31.7% were moderately comfortable, and 4.9% reported a low level of comfort. There was a higher level of comfort among female caregivers (63%) than among male caregivers, but the difference was not statistically significant ( p > 0.05). General health and bodily hygiene was the most comfortable topic, followed by HIV/AIDS and other STIs. Having babies, birth control and night emissions in boys were the least comfortable topics (Additional file : Appendix A). Attitudes towards SRH issues of young adolescents The majority of the caregivers (84.8%) had a negative attitude towards SRH issues of YAs, while only about 0.7% of the caregivers had a positive attitude, and 14.5% had a moderate attitude. Female caregivers (85%) reported a significantly higher negative attitude towards SRH issues of YAs compared to male caregivers ( p = 0.05). The median score for attitude was 11 out of a maximum of 12. SRH knowledge of caregivers Fifteen per cent of the caregivers reported a high level of knowledge of SRH, while 3 per cent reported a low level of knowledge. The vast majority (84%) of the caregivers reported a moderate level of SRH knowledge. Female caregivers had greater knowledge than male caregivers, though this was not statistically significant. Connectedness between caregivers and young adolescents Sixteen per cent of the caregivers reported a high level of connectedness with YAs, while 34% reported a low level of connectedness. Around half (49%) reported a medium level of connectedness. Female caregivers reported a higher level of connectedness than male caregivers. Connectedness was measured through three subscales: caregiver involvement, positive parenting, and parental expertise and accessibility. A third (33%) of the caregivers reported a high level of involvement, with the majority of these being female. Over a third (39%) of the caregivers reported a high level of positive parenting, while 32% reported a high level of parental expertise and accessibility. Correlates of level of SRH communication between caregivers and young adolescents Bivariate linear regression was carried out to investigate the relationship between socio-demographic characteristics of the caregivers, comfort with SRH communication, attitudes towards SRH, level of connectedness and knowledge of SRH with SRH communication (Table ). The analysis indicated a significant ( p < 0.001) positive linear relationship between comfort with SRH discussions and level of SRH communication. A unit of increase in comfort with SRH discussions increases SRH communication by 0.25 units (SE = 0.04). On the other hand, the level of SRH communication reduced with an increase in the number of YAs in a household (-0.45, SE = 0.19; p < 0.05). In the multivariate linear regression analysis, we ran three models using the manual backward stepwise approach to identify the variables significantly predicting SRH communication. We considered variables that were statistically significant in the bivariate analysis (level of comfort with SRH communication and number of YAs in a household), those with a borderline p-value (caregiving structure) and those that indicate biological plausibility (based on previous findings on predictors of caregiver–child communication on SRH) (Table ). These included caregiver involvement, sex and relationship type. The overall regression was statistically significant ( R 2 = 0.23, F (7,183) = 7.61; p < 0.001. It was found that the level of comfort with SRH communication significantly predicted the level of SRH communication (0.22 (0.04); p > 0.001). Correlates of level of comfort with SRH communication between caregivers and young adolescents The bivariate analysis for level of comfort with SRH communication indicated that the number of YAs in a household (-0.98, SE = 0.34) significantly predicted comfort with SRH communication, although it had a negative correlation (Table ). The more the YAs in a given households, the less comfortable a caregiver felt discussing SRH. We ran three models using the manual backward stepwise approach. In the final model, we considered the variables that were statistically significant in the bivariate analysis and also included those that were biologically plausible, as well as borderline p-value (religion and sex) (Table ). The overall regression was not statistically significant ( R 2 = 0.09, F (9,169) = 1.84; p = 0.06). However, it was found that the number of YAs in a household significantly predicted the level of SRH communication ( p < 0.05). A total of 218 caregivers were enrolled in the study, of which 76% were women. The mean age of the caregivers was 44.9 years (SD = 12.61). Seventy-three per cent were biological caregivers, and the majority (68.3%) had attained primary education, while 11.9% had not received any formal education. The caregiver–YA dyads comprised 96 with female caregiver and daughter (44.0%), 68 male caregiver and daughter (31.2%), 30 female caregiver and son (13.8%) and 24 male caregiver and son 24 (11%). The majority (76.6%) of the households were two-caregiver households, with an average of two YAs (Table ). Level of SRH communication Ten SRH topics were explored in the study. None of the respondents discussed all 10 topics. Two per cent of the caregivers reported ever discussing nine SRH topics with the YAs, while 3.5% reported never discussing any topics. The mean number of topics ever discussed was 3.9. Twenty-two per cent of the caregivers reported discussing at least three of the topics, and 7% reported discussing at least one of the topics. Overall, general health and bodily hygiene was discussed by majority of the dyads (89.9%), followed by HIV/AIDS and other STIs (77.5%). In contrast, only 4.3% of the dyads discussed night emissions in boys. There was no significant difference in communication of SRH topics across the dyads, except for HIV/AIDS and other STIs, which were more likely to be discussed in dyads with female caregivers ( p < 0.05) (Table ). Caregivers’ comfort with SRH discussions with young adolescents The majority of the caregivers (63.4%) reported a high level of comfort with SRH discussions with YAs, 31.7% were moderately comfortable, and 4.9% reported a low level of comfort. There was a higher level of comfort among female caregivers (63%) than among male caregivers, but the difference was not statistically significant ( p > 0.05). General health and bodily hygiene was the most comfortable topic, followed by HIV/AIDS and other STIs. Having babies, birth control and night emissions in boys were the least comfortable topics (Additional file : Appendix A). Attitudes towards SRH issues of young adolescents The majority of the caregivers (84.8%) had a negative attitude towards SRH issues of YAs, while only about 0.7% of the caregivers had a positive attitude, and 14.5% had a moderate attitude. Female caregivers (85%) reported a significantly higher negative attitude towards SRH issues of YAs compared to male caregivers ( p = 0.05). The median score for attitude was 11 out of a maximum of 12. SRH knowledge of caregivers Fifteen per cent of the caregivers reported a high level of knowledge of SRH, while 3 per cent reported a low level of knowledge. The vast majority (84%) of the caregivers reported a moderate level of SRH knowledge. Female caregivers had greater knowledge than male caregivers, though this was not statistically significant. Connectedness between caregivers and young adolescents Sixteen per cent of the caregivers reported a high level of connectedness with YAs, while 34% reported a low level of connectedness. Around half (49%) reported a medium level of connectedness. Female caregivers reported a higher level of connectedness than male caregivers. Connectedness was measured through three subscales: caregiver involvement, positive parenting, and parental expertise and accessibility. A third (33%) of the caregivers reported a high level of involvement, with the majority of these being female. Over a third (39%) of the caregivers reported a high level of positive parenting, while 32% reported a high level of parental expertise and accessibility. Correlates of level of SRH communication between caregivers and young adolescents Bivariate linear regression was carried out to investigate the relationship between socio-demographic characteristics of the caregivers, comfort with SRH communication, attitudes towards SRH, level of connectedness and knowledge of SRH with SRH communication (Table ). The analysis indicated a significant ( p < 0.001) positive linear relationship between comfort with SRH discussions and level of SRH communication. A unit of increase in comfort with SRH discussions increases SRH communication by 0.25 units (SE = 0.04). On the other hand, the level of SRH communication reduced with an increase in the number of YAs in a household (-0.45, SE = 0.19; p < 0.05). In the multivariate linear regression analysis, we ran three models using the manual backward stepwise approach to identify the variables significantly predicting SRH communication. We considered variables that were statistically significant in the bivariate analysis (level of comfort with SRH communication and number of YAs in a household), those with a borderline p-value (caregiving structure) and those that indicate biological plausibility (based on previous findings on predictors of caregiver–child communication on SRH) (Table ). These included caregiver involvement, sex and relationship type. The overall regression was statistically significant ( R 2 = 0.23, F (7,183) = 7.61; p < 0.001. It was found that the level of comfort with SRH communication significantly predicted the level of SRH communication (0.22 (0.04); p > 0.001). Correlates of level of comfort with SRH communication between caregivers and young adolescents The bivariate analysis for level of comfort with SRH communication indicated that the number of YAs in a household (-0.98, SE = 0.34) significantly predicted comfort with SRH communication, although it had a negative correlation (Table ). The more the YAs in a given households, the less comfortable a caregiver felt discussing SRH. We ran three models using the manual backward stepwise approach. In the final model, we considered the variables that were statistically significant in the bivariate analysis and also included those that were biologically plausible, as well as borderline p-value (religion and sex) (Table ). The overall regression was not statistically significant ( R 2 = 0.09, F (9,169) = 1.84; p = 0.06). However, it was found that the number of YAs in a household significantly predicted the level of SRH communication ( p < 0.05). Ten SRH topics were explored in the study. None of the respondents discussed all 10 topics. Two per cent of the caregivers reported ever discussing nine SRH topics with the YAs, while 3.5% reported never discussing any topics. The mean number of topics ever discussed was 3.9. Twenty-two per cent of the caregivers reported discussing at least three of the topics, and 7% reported discussing at least one of the topics. Overall, general health and bodily hygiene was discussed by majority of the dyads (89.9%), followed by HIV/AIDS and other STIs (77.5%). In contrast, only 4.3% of the dyads discussed night emissions in boys. There was no significant difference in communication of SRH topics across the dyads, except for HIV/AIDS and other STIs, which were more likely to be discussed in dyads with female caregivers ( p < 0.05) (Table ). The majority of the caregivers (63.4%) reported a high level of comfort with SRH discussions with YAs, 31.7% were moderately comfortable, and 4.9% reported a low level of comfort. There was a higher level of comfort among female caregivers (63%) than among male caregivers, but the difference was not statistically significant ( p > 0.05). General health and bodily hygiene was the most comfortable topic, followed by HIV/AIDS and other STIs. Having babies, birth control and night emissions in boys were the least comfortable topics (Additional file : Appendix A). The majority of the caregivers (84.8%) had a negative attitude towards SRH issues of YAs, while only about 0.7% of the caregivers had a positive attitude, and 14.5% had a moderate attitude. Female caregivers (85%) reported a significantly higher negative attitude towards SRH issues of YAs compared to male caregivers ( p = 0.05). The median score for attitude was 11 out of a maximum of 12. Fifteen per cent of the caregivers reported a high level of knowledge of SRH, while 3 per cent reported a low level of knowledge. The vast majority (84%) of the caregivers reported a moderate level of SRH knowledge. Female caregivers had greater knowledge than male caregivers, though this was not statistically significant. Sixteen per cent of the caregivers reported a high level of connectedness with YAs, while 34% reported a low level of connectedness. Around half (49%) reported a medium level of connectedness. Female caregivers reported a higher level of connectedness than male caregivers. Connectedness was measured through three subscales: caregiver involvement, positive parenting, and parental expertise and accessibility. A third (33%) of the caregivers reported a high level of involvement, with the majority of these being female. Over a third (39%) of the caregivers reported a high level of positive parenting, while 32% reported a high level of parental expertise and accessibility. Bivariate linear regression was carried out to investigate the relationship between socio-demographic characteristics of the caregivers, comfort with SRH communication, attitudes towards SRH, level of connectedness and knowledge of SRH with SRH communication (Table ). The analysis indicated a significant ( p < 0.001) positive linear relationship between comfort with SRH discussions and level of SRH communication. A unit of increase in comfort with SRH discussions increases SRH communication by 0.25 units (SE = 0.04). On the other hand, the level of SRH communication reduced with an increase in the number of YAs in a household (-0.45, SE = 0.19; p < 0.05). In the multivariate linear regression analysis, we ran three models using the manual backward stepwise approach to identify the variables significantly predicting SRH communication. We considered variables that were statistically significant in the bivariate analysis (level of comfort with SRH communication and number of YAs in a household), those with a borderline p-value (caregiving structure) and those that indicate biological plausibility (based on previous findings on predictors of caregiver–child communication on SRH) (Table ). These included caregiver involvement, sex and relationship type. The overall regression was statistically significant ( R 2 = 0.23, F (7,183) = 7.61; p < 0.001. It was found that the level of comfort with SRH communication significantly predicted the level of SRH communication (0.22 (0.04); p > 0.001). The bivariate analysis for level of comfort with SRH communication indicated that the number of YAs in a household (-0.98, SE = 0.34) significantly predicted comfort with SRH communication, although it had a negative correlation (Table ). The more the YAs in a given households, the less comfortable a caregiver felt discussing SRH. We ran three models using the manual backward stepwise approach. In the final model, we considered the variables that were statistically significant in the bivariate analysis and also included those that were biologically plausible, as well as borderline p-value (religion and sex) (Table ). The overall regression was not statistically significant ( R 2 = 0.09, F (9,169) = 1.84; p = 0.06). However, it was found that the number of YAs in a household significantly predicted the level of SRH communication ( p < 0.05). This study sought to assess the current level of communication between caregivers and YAs about SRH, and caregivers’ comfort with such discussions, and identify their correlates. The study was conducted in a rural community in south-western Uganda, where an intervention to improve communication between caregivers and YAs on SRH would be tested. Unlike many studies which focus on SRH communication with older adolescents, this study focuses on YAs aged 10–14 years. This approach is driven by the notion that young adolescence is a stage of transition from childhood to adulthood where critical changes occur, especially in terms of sexual development . Addressing SRH issues during this transitional phase is considered to have more positive outcomes than dealing with them later in life. However, there is also building evidence the risk-taking behaviours is already occurring at this age . Overall, our findings indicate that communication about SRH does take place between caregivers and their YAs. However, this was relatively rare and varied according to the topics discussed. On average, 21.6% of caregivers in the study reported ever discussing an average of 3.9 of the 10 SRH topics listed in the questionnaire. This finding falls in tandem with several other studies in similar settings—for example, a study conducted in Korogocho settlement in western Kenya indicated that communication between caregivers and very young adolescents was rare . Similar findings are reported in a study conducted in Zanzibar, where only 40% of caregivers had ever communicated with their children about SRH . However, the latter study reports communication about SRH with older adolescents (aged 15–19). A considerable number of caregivers reported discussing general health and bodily hygiene, and HIV/AIDS and other STIs. Indeed, on the comfort scale, caregivers reported high levels of comfort discussing HIV/AIDS and other STIs, as well as general health and bodily hygiene. Notwithstanding is the major finding of this study that the number of SRH topics discussed increased with an increasing level of comfort with SRH discussions. The probable reason for high reports of discussions on general health and bodily hygiene is that these topics can be discussed with minimal embarrassment. As far as HIV/AIDS is concerned, excess messaging around HIV/AIDS in the media, coupled with the high risk perception of HIV infection in many communities, may have triggered a lot of discussion around this topic. Topics deemed to be sensitive—such as night emissions in boys, condoms, birth control and sexual conduct—were discussed the least. Low or moderate levels of knowledge and a high proportion of caregivers reporting a negative attitude towards SRH in our findings could account for the low level of discussion of these latter topics. Additionally, evidence also shows that parents associate discussions with adolescents about condoms and birth control with being comparable to encouraging them to engage in sexual intercourse . The relationship between the SRH topics most and least commonly discussed and their perceived sensitivity strongly justifies the finding that the level of SRH communication increases with increasing level of comfort. This interrelates with the notion that open discussions about sexual issues are a taboo in many African settings, and the fact that many caregivers believed that it was too early to begin initiating discussions about sex . These factors, though not addressed in this study, serve as proxies for SRH communication by influencing how comfortable caregivers feel discussing these SRH topics with YAS. Our findings specifically reveal that religion and the number of adolescents in a household influence caregivers’ comfort with SRH communication. In their review, Abdallah et al. (2017) report religion as one of the factors influencing SRH communication in East Africa . Although there was no significant difference in SRH communication across the dyad types for each of the SRH topics except for HIV/AIDS and other STIs, mother–daughter dyads were reported to have the highest mean number of topics discussed, while mother–son dyads were reported to have the lowest mean number of topics. Many studies implicate the influence of gender on caregiver–child communication, with mothers communicating more than fathers, and girls receiving more communication than boys. Girls are disproportionately vulnerable to SRH risks than boys, and mothers spend more time with children than fathers . Moreover, evidence indicates that mothers are the preferred partners for socializing their daughters about sexuality . We found that the level of comfort with SRH communication reduced with an increase in the number of YAs in a household. It is possible that caregivers may find it uncomfortable having SRH discussions with more children than it would be if they were fewer. Previous studies have particularly investigated the effect of family size on the level of SRH communication. In this study, the number of YAs in a given household could serve as an indicator for family size. Studies in Bangladesh and Ethiopia indicate that the bigger the family size, the lower the level of SRH communication . Another study in Ethiopia reports no association between family size and the level of SRH communication . Zakaria et al. attributes their findings to presence of older siblings in the household that the adolescents would most likely prefer to talk to about their SRH issues rather than their parents . Muhwezi et al. reveals that adolescents preferred to talk to their siblings about SRH than their parents because their parents were not comfortable about these SRH discussions . The other reason given for not discussing SRH issues in larger families could be due to parents feeling overburdened by the number of children to speak to and that parents are less concerned for SRH communication as the family size increases . However, there is need for further research to explore the association between comfort with SRH communication and family size. The results of this study suggest that SRH communication between caregivers and YAs was low. SRH communication was also found to increase with increase in comfort with SRH communication. We also found that the more the YAs in a household, the lower the level of SRH communication. Comfort of SRH communication was found to reduce with an increase in number of YAs in a household. These findings provide a basis for interventions to improve communication between caregivers and children. First, training on value clarification and communication skills that enable caregivers to discuss SRH topics with less embarrassment and create a predisposition towards a positive attitude towards YA SRH is important. Topics focusing on general parenting skills—particularly the quality of their relationships—with the assumption that it inculcates positive caregiver–child relationships, would facilitate and increase the level of comfort with SRH discussions. There is a need for qualitative studies to gain a deeper understanding of determinants of comfort with discussing SRH with YAs. Study limitations Limitations of our study include a relatively small sample size, which affects the power of the study. This affects comparison of SRH communication by gender, yet evidence highlights its important influence on SRH communication . The samples for males are generally too small to make a substantial comparison. Limitations of our study include a relatively small sample size, which affects the power of the study. This affects comparison of SRH communication by gender, yet evidence highlights its important influence on SRH communication . The samples for males are generally too small to make a substantial comparison. Additional file 1: Appendix A. Level of comfort of SRH communication. Appendix B. SRH communication by number of topics. Appendix C. Caregiver attitudes towards SRH of Adolescents. Appendix D. Comfort of SRH Discussions by dyad type. Appendix E. Distribution of independent variables by sex of the caregiver. |
Technological dental sealants: in vitro evaluation of material properties and antibiofilm potential | c22637c3-6718-4f66-983f-93aeba8cabac | 11786408 | Dentistry[mh] | Dental caries remains a pervasive global health issue, affecting approximately 35% of individuals across all age groups . Notably, a significant proportion (50%) of caries cases occur in occlusal pits and fissures, which constitute only 15% of the total tooth surface area . Permanent first and second molars are particularly susceptible to caries initiation . The complex morphology of occlusal surfaces predisposes them to bacterial plaque accumulation and biofilm formation . The oral microbiome, comprising over 700 bacterial species, including anaerobic and aerobic bacteria, forms a diverse microbial community. Streptococcus mutans is a key cariogenic bacterium that metabolizes sucrose, produces acid, and facilitates biofilm development, leading to tooth demineralization . Effective caries prevention strategies include water fluoridation, dietary sugar control, oral hygiene practices, and professional fluoride applications . Antimicrobial peptides (AMPs) have emerged as promising agents for dental applications, exhibiting potential to inhibit bacterial adhesion and disrupt biofilm formation . However, fissure sealants remain a well-established preventive measure, demonstrating a significant reduction in caries incidence on occlusal surfaces . Sealants are resinous materials designed to seal pits and fissures, preventing bacterial colonization and subsequent carbohydrate fermentation. Additionally, fluoride-releasing sealants can promote remineralization and inhibit bacterial growth . Several factors influence sealant longevity, including microhardness, surface roughness, and retention . A well-sealed surface can effectively resist wear and tear, minimize bacterial adhesion, and maintain a long-lasting protective barrier . The incorporation of antimicrobial agents, such as those found in glass ionomer-based sealants, can further enhance their efficacy . Despite advancements in sealant technology, a comprehensive comparison of different materials, particularly those incorporating novel technologies, is lacking. This study aims to address this gap by evaluating the physical, mechanical, and antimicrobial properties of various sealants. The null hypothesis is that the tested materials have no significant differences, regardless of their composition. By investigating these parameters, this study seeks to provide valuable insights into the performance of contemporary sealants and aid clinicians in making informed decisions for optimal patient care. Tested materials and sample size calculation To evaluate the performance of four dental sealant materials, an in vitro study was conducted. Sample sizes for each material group (Table ) were determined using G*Power, considering a significance level of α = 0.05, power of 80%, and an effect size of 0.8. A 10–20% increase in sample size was incorporated to account for potential losses. Consequently, quantitative tests required 5–10 specimens per group, while qualitative tests necessitated one specimen per group. Specimen preparation and experimental design Cylindrical specimens (6 mm diameter × 2 mm depth) were prepared using an acrylic matrix and handled according to the manufacturer's instructions . The materials were applied in a single step, covered with a polyester strip, and light-cured using an LED curing light (CV-218, wavelength: 430–485 nm, intensity ≥ 1800 mW/cm 2 ) for the recommended duration. After 24 h of storage in distilled water at room temperature, the specimens were polished sequentially with sandpaper discs (#400, #600, #1200, #1500, and #2000) and cleaned with an air/water spray. The specimens were randomly assigned to four groups: Self-etching: Beautisealant® (Shofu, Kyoto, Japan) Control: Fluroshield® (Dentsply, Bogotá, Colombia) Self-adhesive and self-etching: Constic® (DMG, Hamburg, Germany) Conventional: Beautiful Flow Plus® F03 (Shofu, Kyoto, Japan) with 37% phosphoric acid etching Each group was subjected to a series of tests, submitted to mechanical (Surface Roughness—RS and Vickers Microhardness—VM), compositional (Energy Dispersive Spectroscopy—EDS), qualitative tests (Scanning Electron Microscopy Analysis SEM), and microbiological analysis. Mechanical analysis Surface roughness measurement The surface roughness measurement was performed using a rugosimeter (Mitutoyo Corporation, Japan) following the ISO 1997 standard. Each sample ( N = 8) was carefully dried with absorbent paper before readings. The value of the initial reading (Ra; µm) was obtained through the arithmetic mean of 5 consecutive readings in each specimen in different regions, thus obtaining the mean and standard deviation as well . Surface microhardness measurement Surface microhardness was evaluated using a digital microhardness meter (FM-700, Poland) coupled with software standardized to a Vickers-type pyramidal diamond indenter (Vickers Microhardness – VM). The measurement was performed using five readings for each specimen ( N = 8) in different regions with an analysis force of 100 gF/mm 2 for 15 s. The values were obtained in gF/mm 2 . For this measurement, we used the 6507 ISO. Compositional analysis Energy Dispersive Spectroscopy Analysis (EDS) To analyze the chemical composition, the samples ( N = 8) were metalized with gold alloys and subjected to EDS (Oxford INCA X-ACT, 51-ADD0048, Abingdon-on-Thames, UK) with measurements at the center of each sample. The samples were fixed in stubs, metalized with gold (MED 010, Balzers, USA). Qualitative analysis Scanning Electron Microscopy Analysis (SEM) The sample ( N = 1) was fixed in stubs, metalized with gold (MED 010, Balzers, USA), and analyzed in a scanning electron microscope (SEM, JEOL-JMS-T33A Scanning Microscope, JEOL – USA Inc., Peabody, MA, USA). Analysis of the surface of the samples by scanning electron microscopy (SEM) was performed under a microscope with a readout using a qualitative surface analysis method of resin dental materials developed by ZHANG et al. . The presence of inorganic nanoparticles dispersed by larger fillers and irregularly shaped fillers distributed in the resin matrix were analyzed. Microbiological analysis The bacterial inhibition capacity of specimens made in BHI agar culture medium and the ability of bacterial adhesion using Streptococcus mutans biofilm were tested. For that, was used for the test in triplicate. Group 2: Control (Fluroshield® – Dentsply, Bogotá, Colombia) was used as a control group in the microbiological tests due to the manufacturer's claims regarding its anti-cariogenic capacity and fluoride release. Inhibition halo analysis Streptococcus mutans strain #25,175 was reactivated on BHI agar plates under microaerophilic conditions at 37 °C for 48 h. Bacterial cells were Gram-stained to confirm their identity. A bacterial suspension was prepared in phosphate-buffered saline (PBS) to a concentration of 1 × 10^8 CFU/mL. BHI agar plates were inoculated with the S. mutans suspension and samples ( N = 5) incubated under microaerophilic conditions at 37 °C for 48 h. Control plates containing only BHI medium and S. mutans inoculum were included to monitor sterility and bacterial growth . Superficial bacterial adherence analysis Streptococcus mutans strain #25,175 was reactivated and cultured as described previously. A portion of the culture was cryopreserved in BHI broth containing 10% DMSO at −20 °C. A bacterial suspension was prepared to a concentration of 1 × 10^8 CFU/mL. Sterilized specimens ( N = 4) were inoculated with S. mutans and incubated in BHI broth supplemented with 5% sucrose for 24 h to allow initial biofilm formation. This process was repeated daily for five days, with fresh medium and inoculum added each day. Control groups included BHI broth alone and BHI broth with S. mutans . After five days, the biofilm was collected by scraping the specimen surface with a sterile loop. The biofilm suspension was serially diluted and plated on BHI agar plates. Colony-forming units (CFU) were counted after 48 h of incubation at 37 °C. Microshear bond strength analysis Specimen preparation Twenty-five healthy bovine incisors were selected for the study. The teeth were cleaned with a pumice stone slurry and water using a low-speed motor and then stored in distilled water. To ensure sterilization without compromising enamel properties, the teeth were disinfected with a 0.1% thymol solution for five days. The teeth were divided into five groups: Self-etching: Beautisealant® (Shofu, Kyoto, Japan) Control: 37% phosphoric acid etching + Fluroshield® (Dentsply, Bogotá, Colombia) Self-adhesive and self-etching: Constic® (DMG, Hamburg, Germany) Conventional: 37% phosphoric acid etching + Single Bond Universal® + Beautiful Flow Plus® F03 (Shofu, Kyoto, Japan) Conventional: 37% phosphoric acid etching + Single Bond Universal® + FluroShield® (Dentsply) All materials were applied according to the manufacturer's instructions. The roots of the teeth were sectioned 1 mm below the cementoenamel junction using a low-speed diamond saw under water coolant. The crown portions were cleaned with a pumice stone slurry and water using a low-speed micromotor. The teeth were then embedded in acrylic resin blocks, exposing the buccal surface. The embedded teeth were sectioned using a low-speed diamond saw to create standardized specimens measuring 10 mm in length and 6 mm in width. After cleaning and ultrasonic debridement, the specimens were stored in distilled water at 37 °C for 24 h. Two 2-mm diameter cavities were prepared on the exposed enamel surface using a rubber dam and a standardized bur. For groups 2, 4, and 5, the enamel surface was etched with 37% phosphoric acid (Condac®, FGM, Joinville, SC, Brazil) for 30 s, followed by a 30-s water rinse and 15-s air-drying. Groups 4 and 5 received an additional step of adhesive application. 3 M™ Single Bond Universal® adhesive was applied with a microbrush for 10 s, air-dried for 5 s, and light-cured for 20 s using a Bluephase® curing unit (Ivoclar, Switzerland, wavelength: 380–515 nm, intensity: 1200 mW/cm 2 ). After material application, the specimens ( N = 10) were embedded in a testing device. A universal testing machine (Triax Digital 50, Controls, Milan, Italy) was used to apply a shear load at a crosshead speed of 0.5 mm/min until failure. The shear bond strength (MPa) was calculated based on the load at failure and the diameter of the detached composite cylinder. Statistical analysis Before the inferential analysis of the data, the normality of data distribution was verified using the statistical program (SPSS), version 20.0, which was performed for all variables (Roughness, Microhardness, EDS, Inhibition Halo, Superficial Bacterial Adherence, and Microshear Bond Strength). After verification of normality, the ANOVA parametric test (one way) was used to compare all variables followed by the Tukey, considering the value of p < 0.05. To evaluate the performance of four dental sealant materials, an in vitro study was conducted. Sample sizes for each material group (Table ) were determined using G*Power, considering a significance level of α = 0.05, power of 80%, and an effect size of 0.8. A 10–20% increase in sample size was incorporated to account for potential losses. Consequently, quantitative tests required 5–10 specimens per group, while qualitative tests necessitated one specimen per group. Cylindrical specimens (6 mm diameter × 2 mm depth) were prepared using an acrylic matrix and handled according to the manufacturer's instructions . The materials were applied in a single step, covered with a polyester strip, and light-cured using an LED curing light (CV-218, wavelength: 430–485 nm, intensity ≥ 1800 mW/cm 2 ) for the recommended duration. After 24 h of storage in distilled water at room temperature, the specimens were polished sequentially with sandpaper discs (#400, #600, #1200, #1500, and #2000) and cleaned with an air/water spray. The specimens were randomly assigned to four groups: Self-etching: Beautisealant® (Shofu, Kyoto, Japan) Control: Fluroshield® (Dentsply, Bogotá, Colombia) Self-adhesive and self-etching: Constic® (DMG, Hamburg, Germany) Conventional: Beautiful Flow Plus® F03 (Shofu, Kyoto, Japan) with 37% phosphoric acid etching Each group was subjected to a series of tests, submitted to mechanical (Surface Roughness—RS and Vickers Microhardness—VM), compositional (Energy Dispersive Spectroscopy—EDS), qualitative tests (Scanning Electron Microscopy Analysis SEM), and microbiological analysis. Surface roughness measurement The surface roughness measurement was performed using a rugosimeter (Mitutoyo Corporation, Japan) following the ISO 1997 standard. Each sample ( N = 8) was carefully dried with absorbent paper before readings. The value of the initial reading (Ra; µm) was obtained through the arithmetic mean of 5 consecutive readings in each specimen in different regions, thus obtaining the mean and standard deviation as well . Surface microhardness measurement Surface microhardness was evaluated using a digital microhardness meter (FM-700, Poland) coupled with software standardized to a Vickers-type pyramidal diamond indenter (Vickers Microhardness – VM). The measurement was performed using five readings for each specimen ( N = 8) in different regions with an analysis force of 100 gF/mm 2 for 15 s. The values were obtained in gF/mm 2 . For this measurement, we used the 6507 ISO. The surface roughness measurement was performed using a rugosimeter (Mitutoyo Corporation, Japan) following the ISO 1997 standard. Each sample ( N = 8) was carefully dried with absorbent paper before readings. The value of the initial reading (Ra; µm) was obtained through the arithmetic mean of 5 consecutive readings in each specimen in different regions, thus obtaining the mean and standard deviation as well . Surface microhardness was evaluated using a digital microhardness meter (FM-700, Poland) coupled with software standardized to a Vickers-type pyramidal diamond indenter (Vickers Microhardness – VM). The measurement was performed using five readings for each specimen ( N = 8) in different regions with an analysis force of 100 gF/mm 2 for 15 s. The values were obtained in gF/mm 2 . For this measurement, we used the 6507 ISO. Energy Dispersive Spectroscopy Analysis (EDS) To analyze the chemical composition, the samples ( N = 8) were metalized with gold alloys and subjected to EDS (Oxford INCA X-ACT, 51-ADD0048, Abingdon-on-Thames, UK) with measurements at the center of each sample. The samples were fixed in stubs, metalized with gold (MED 010, Balzers, USA). To analyze the chemical composition, the samples ( N = 8) were metalized with gold alloys and subjected to EDS (Oxford INCA X-ACT, 51-ADD0048, Abingdon-on-Thames, UK) with measurements at the center of each sample. The samples were fixed in stubs, metalized with gold (MED 010, Balzers, USA). Scanning Electron Microscopy Analysis (SEM) The sample ( N = 1) was fixed in stubs, metalized with gold (MED 010, Balzers, USA), and analyzed in a scanning electron microscope (SEM, JEOL-JMS-T33A Scanning Microscope, JEOL – USA Inc., Peabody, MA, USA). Analysis of the surface of the samples by scanning electron microscopy (SEM) was performed under a microscope with a readout using a qualitative surface analysis method of resin dental materials developed by ZHANG et al. . The presence of inorganic nanoparticles dispersed by larger fillers and irregularly shaped fillers distributed in the resin matrix were analyzed. Microbiological analysis The bacterial inhibition capacity of specimens made in BHI agar culture medium and the ability of bacterial adhesion using Streptococcus mutans biofilm were tested. For that, was used for the test in triplicate. Group 2: Control (Fluroshield® – Dentsply, Bogotá, Colombia) was used as a control group in the microbiological tests due to the manufacturer's claims regarding its anti-cariogenic capacity and fluoride release. Inhibition halo analysis Streptococcus mutans strain #25,175 was reactivated on BHI agar plates under microaerophilic conditions at 37 °C for 48 h. Bacterial cells were Gram-stained to confirm their identity. A bacterial suspension was prepared in phosphate-buffered saline (PBS) to a concentration of 1 × 10^8 CFU/mL. BHI agar plates were inoculated with the S. mutans suspension and samples ( N = 5) incubated under microaerophilic conditions at 37 °C for 48 h. Control plates containing only BHI medium and S. mutans inoculum were included to monitor sterility and bacterial growth . Superficial bacterial adherence analysis Streptococcus mutans strain #25,175 was reactivated and cultured as described previously. A portion of the culture was cryopreserved in BHI broth containing 10% DMSO at −20 °C. A bacterial suspension was prepared to a concentration of 1 × 10^8 CFU/mL. Sterilized specimens ( N = 4) were inoculated with S. mutans and incubated in BHI broth supplemented with 5% sucrose for 24 h to allow initial biofilm formation. This process was repeated daily for five days, with fresh medium and inoculum added each day. Control groups included BHI broth alone and BHI broth with S. mutans . After five days, the biofilm was collected by scraping the specimen surface with a sterile loop. The biofilm suspension was serially diluted and plated on BHI agar plates. Colony-forming units (CFU) were counted after 48 h of incubation at 37 °C. The sample ( N = 1) was fixed in stubs, metalized with gold (MED 010, Balzers, USA), and analyzed in a scanning electron microscope (SEM, JEOL-JMS-T33A Scanning Microscope, JEOL – USA Inc., Peabody, MA, USA). Analysis of the surface of the samples by scanning electron microscopy (SEM) was performed under a microscope with a readout using a qualitative surface analysis method of resin dental materials developed by ZHANG et al. . The presence of inorganic nanoparticles dispersed by larger fillers and irregularly shaped fillers distributed in the resin matrix were analyzed. The bacterial inhibition capacity of specimens made in BHI agar culture medium and the ability of bacterial adhesion using Streptococcus mutans biofilm were tested. For that, was used for the test in triplicate. Group 2: Control (Fluroshield® – Dentsply, Bogotá, Colombia) was used as a control group in the microbiological tests due to the manufacturer's claims regarding its anti-cariogenic capacity and fluoride release. Streptococcus mutans strain #25,175 was reactivated on BHI agar plates under microaerophilic conditions at 37 °C for 48 h. Bacterial cells were Gram-stained to confirm their identity. A bacterial suspension was prepared in phosphate-buffered saline (PBS) to a concentration of 1 × 10^8 CFU/mL. BHI agar plates were inoculated with the S. mutans suspension and samples ( N = 5) incubated under microaerophilic conditions at 37 °C for 48 h. Control plates containing only BHI medium and S. mutans inoculum were included to monitor sterility and bacterial growth . Streptococcus mutans strain #25,175 was reactivated and cultured as described previously. A portion of the culture was cryopreserved in BHI broth containing 10% DMSO at −20 °C. A bacterial suspension was prepared to a concentration of 1 × 10^8 CFU/mL. Sterilized specimens ( N = 4) were inoculated with S. mutans and incubated in BHI broth supplemented with 5% sucrose for 24 h to allow initial biofilm formation. This process was repeated daily for five days, with fresh medium and inoculum added each day. Control groups included BHI broth alone and BHI broth with S. mutans . After five days, the biofilm was collected by scraping the specimen surface with a sterile loop. The biofilm suspension was serially diluted and plated on BHI agar plates. Colony-forming units (CFU) were counted after 48 h of incubation at 37 °C. Specimen preparation Twenty-five healthy bovine incisors were selected for the study. The teeth were cleaned with a pumice stone slurry and water using a low-speed motor and then stored in distilled water. To ensure sterilization without compromising enamel properties, the teeth were disinfected with a 0.1% thymol solution for five days. The teeth were divided into five groups: Self-etching: Beautisealant® (Shofu, Kyoto, Japan) Control: 37% phosphoric acid etching + Fluroshield® (Dentsply, Bogotá, Colombia) Self-adhesive and self-etching: Constic® (DMG, Hamburg, Germany) Conventional: 37% phosphoric acid etching + Single Bond Universal® + Beautiful Flow Plus® F03 (Shofu, Kyoto, Japan) Conventional: 37% phosphoric acid etching + Single Bond Universal® + FluroShield® (Dentsply) All materials were applied according to the manufacturer's instructions. The roots of the teeth were sectioned 1 mm below the cementoenamel junction using a low-speed diamond saw under water coolant. The crown portions were cleaned with a pumice stone slurry and water using a low-speed micromotor. The teeth were then embedded in acrylic resin blocks, exposing the buccal surface. The embedded teeth were sectioned using a low-speed diamond saw to create standardized specimens measuring 10 mm in length and 6 mm in width. After cleaning and ultrasonic debridement, the specimens were stored in distilled water at 37 °C for 24 h. Two 2-mm diameter cavities were prepared on the exposed enamel surface using a rubber dam and a standardized bur. For groups 2, 4, and 5, the enamel surface was etched with 37% phosphoric acid (Condac®, FGM, Joinville, SC, Brazil) for 30 s, followed by a 30-s water rinse and 15-s air-drying. Groups 4 and 5 received an additional step of adhesive application. 3 M™ Single Bond Universal® adhesive was applied with a microbrush for 10 s, air-dried for 5 s, and light-cured for 20 s using a Bluephase® curing unit (Ivoclar, Switzerland, wavelength: 380–515 nm, intensity: 1200 mW/cm 2 ). After material application, the specimens ( N = 10) were embedded in a testing device. A universal testing machine (Triax Digital 50, Controls, Milan, Italy) was used to apply a shear load at a crosshead speed of 0.5 mm/min until failure. The shear bond strength (MPa) was calculated based on the load at failure and the diameter of the detached composite cylinder. Statistical analysis Before the inferential analysis of the data, the normality of data distribution was verified using the statistical program (SPSS), version 20.0, which was performed for all variables (Roughness, Microhardness, EDS, Inhibition Halo, Superficial Bacterial Adherence, and Microshear Bond Strength). After verification of normality, the ANOVA parametric test (one way) was used to compare all variables followed by the Tukey, considering the value of p < 0.05. Twenty-five healthy bovine incisors were selected for the study. The teeth were cleaned with a pumice stone slurry and water using a low-speed motor and then stored in distilled water. To ensure sterilization without compromising enamel properties, the teeth were disinfected with a 0.1% thymol solution for five days. The teeth were divided into five groups: Self-etching: Beautisealant® (Shofu, Kyoto, Japan) Control: 37% phosphoric acid etching + Fluroshield® (Dentsply, Bogotá, Colombia) Self-adhesive and self-etching: Constic® (DMG, Hamburg, Germany) Conventional: 37% phosphoric acid etching + Single Bond Universal® + Beautiful Flow Plus® F03 (Shofu, Kyoto, Japan) Conventional: 37% phosphoric acid etching + Single Bond Universal® + FluroShield® (Dentsply) All materials were applied according to the manufacturer's instructions. The roots of the teeth were sectioned 1 mm below the cementoenamel junction using a low-speed diamond saw under water coolant. The crown portions were cleaned with a pumice stone slurry and water using a low-speed micromotor. The teeth were then embedded in acrylic resin blocks, exposing the buccal surface. The embedded teeth were sectioned using a low-speed diamond saw to create standardized specimens measuring 10 mm in length and 6 mm in width. After cleaning and ultrasonic debridement, the specimens were stored in distilled water at 37 °C for 24 h. Two 2-mm diameter cavities were prepared on the exposed enamel surface using a rubber dam and a standardized bur. For groups 2, 4, and 5, the enamel surface was etched with 37% phosphoric acid (Condac®, FGM, Joinville, SC, Brazil) for 30 s, followed by a 30-s water rinse and 15-s air-drying. Groups 4 and 5 received an additional step of adhesive application. 3 M™ Single Bond Universal® adhesive was applied with a microbrush for 10 s, air-dried for 5 s, and light-cured for 20 s using a Bluephase® curing unit (Ivoclar, Switzerland, wavelength: 380–515 nm, intensity: 1200 mW/cm 2 ). After material application, the specimens ( N = 10) were embedded in a testing device. A universal testing machine (Triax Digital 50, Controls, Milan, Italy) was used to apply a shear load at a crosshead speed of 0.5 mm/min until failure. The shear bond strength (MPa) was calculated based on the load at failure and the diameter of the detached composite cylinder. Before the inferential analysis of the data, the normality of data distribution was verified using the statistical program (SPSS), version 20.0, which was performed for all variables (Roughness, Microhardness, EDS, Inhibition Halo, Superficial Bacterial Adherence, and Microshear Bond Strength). After verification of normality, the ANOVA parametric test (one way) was used to compare all variables followed by the Tukey, considering the value of p < 0.05. The results revealed no statistically significant difference between groups ( p > 0.05) regarding surface roughness, with all groups showing values smaller than 0,2 µm. The one-way ANOVA revealed a statistically significant difference among the groups ( p < 0,001) on the microhardness measurement. The Tukey test revealed no difference between G1 and G2 ( p = 0.99). However, G3 and G4 showed the highest values, differing from the other groups and between each other ( p < 0.01), with Beautiful Flow Plus F03® showing the highest value (Fig. ). Carbon (C) and oxygen (O), the primary components of the organic matrix, were detected in all materials. Sodium (Na) was present in BeautiSealant® (G1) and Beautifil Flow Plus F03® (G4). Fluoride (F) was detected only in Beautifil Flow Plus F03® (G4), absent from biointeractive materials like BeautiSealant® (G1) and FluroShield® (G2). Aluminum (Al) was present in all materials. Silicon (Si) was detected in all groups, as expected due to its role as a nucleating agent in the inorganic matrix. Strontium (Sr) was present in groups G1 and G4, both containing GIOMER technology. Tungsten (W) was detected only in BeautiSealant® (G1). Barium (Ba) was found in FluroShield® (G2), Constic® (G3), and trace amounts in Beautifil Flow Plus F03® (G4) (Table and Fig. ). Plows and furrows are more clearly visible in G1—BeautiSealant® than in the G4—Beautifil Flow Plus F03®. However, G1 – BeautiSealant® and G4 – Beautifil Flow Plus F03® are materials that present agglomerates of well-dispersed filler particles on their surface. This organization is also seen in the G3 – Constic® group. However, it is not observed in the G2 – Fluroshield®. The G2 – FluroShield® group has a qualitatively more significant number of grooves/pits compared to the other compounds studied. Even though each group presents singularities on its surface, surface roughness rates were similar across all materials (Fig. ). All groups, including Constic® (G3) which lacks antibacterial properties, exhibited an inhibition halo. However, no statistically significant differences in inhibition halo size were observed among the groups ( p > 0.05). Similarly, no significant differences in superficial bacterial adhesion were found among the groups, despite variations in material composition and the presence of antibacterial agents (Fig. ). However, Group G4 (Beautiful Flow Plus® F03) showed the lowest level of bacterial adhesion (5.06 ± 0.24 CFU/mL). BeautiSealant® (G1) and Constic® (G3) exhibited the lowest microshear bond strength, significantly lower than all other groups ( p < 0.05). However, no significant difference was observed between these two groups (Tukey's test). The conventional sealant FluroShield® (G2) demonstrated similar microshear bond strength to the traditional resin Beautifil Flow Plus® F03 (G4), with no significant difference ( p > 0.05). Additionally, the conventional sealant FluroShield® + Adhesive (G5) exhibited the highest microshear bond strength, but no significant difference was observed compared to FluroShield® (G2) ( p > 0.05) (Fig. ). Pit and fissure sealants have been shown to effectively prevent caries lesions on occlusal surfaces . Newer materials, such as self-adhesive and self-etching sealants, offer potential advantages in clinical practice, especially for pediatric patients, due to their reduced sensitivity to moisture contamination and easier application . This laboratory study aimed to investigate the physical, compositional, and antibacterial properties of various dental sealants. Our findings demonstrate significant differences among the tested materials, rejecting the null hypothesis. All groups exhibited comparable surface roughness values ( p = 0.61), likely due to the similarity in particle size distribution among the materials. This suggests that the consistent particle size distribution among the materials contributed to the similar surface roughness. Similar findings were reported by Hernández-mendieta et al . for BeautiSealant® and Beautiful Flow Plus®. Regarding FluroShield® surface roughness, Kantovitz et al. reported a value of 0.15 ± 0.02, consistent with our findings. Leal et al . found a similar value for Constic® (0.11 ± 0.019). It's important to note that surface roughness can be influenced by factors such as the degree of polymerization, material hardness, and filler composition . Beautifil Flow Plus® exhibited the highest microhardness (37.9 ± 4.87), significantly differing from other groups. This higher microhardness may be attributed to its higher inorganic filler content and potential water sorption resistance. Hernández-mendieta et al . reported similar findings, with Beautifil Flow Plus® showing higher microhardness than BeautiSealant®. However, our study revealed a more significant difference. This discrepancy may be due to variations in testing conditions or material batches. FluroShield® exhibited a lower microhardness (16.28 ± 4.91), comparable to findings by Alexandre et al . . Constic® showed intermediate microhardness (26.11 ± 3.46). The differing microhardness values can be explained by variations in filler type, concentration, and the nature of the organic matrix. For instance, GIOMER materials, containing UDMA, TEGDMA, and Bis-GMA, may be more susceptible to water sorption and degradation. In contrast, materials with higher inorganic filler content, like Beautiful Flow Plus®, may exhibit improved mechanical properties. Still, FluroShield® and Constic® incorporate Bis-GMA and Silicon Dioxide, while Beautiful Flow Plus® has a higher inorganic filler content, contributing to its superior microhardness. The higher silicate filler concentration in Constic® may enhance its bond strength and resistance to wear. SEM analysis revealed distinct surface morphologies. BeautiSealant® and Beautiful Flow Plus® exhibited smooth surfaces with dispersed nanoparticle fillers, aligning with previous studies . In contrast, FluroShield® showed larger surface grooves, which could potentially serve as niches for bacterial growth . This observation is consistent with earlier research by Cooley et al . , who noted the presence of voids and air bubbles within FluroShield®. These findings highlight the importance of evenly distributed nanoparticle clusters for optimal bond strength and surface properties. The materials analyzed contained carbon and oxygen, forming the organic matrix. Additionally, filler elements such as calcium, aluminum, strontium, and fluoride were incorporated to enhance the material's properties. A higher inorganic filler content, as seen in Beautiful Flow Plus®, is associated with improved mechanical properties like microhardness . While fluoride, a key remineralizing agent, was detected in Beautiful Flow Plus®, it was absent in BeautiSealant® and FluroShield®. This suggests that fluoride release may occur during storage in aqueous environments . Aluminum, present in all materials, can contribute to desensitizing effects and remineralization . Silicon, a significant component of Constic®, acts as a nucleating agent, potentially contributing to its higher microhardness. Strontium, found in GIOMER materials, can release Sr 2 ⁺ ions to form strontium apatite, supporting remineralization . Barium, present in FluroShield®, complements the inorganic matrix, while boron, a potential antibacterial agent, may be released from GIOMER materials . All materials exhibited similar halo inhibition zones, suggesting that the inherent antimicrobial properties of the resin-based materials, such as the presence of TEGDMA and Bis-GMA, may have contributed to bacterial growth inhibition . While bioactive ions released from materials like GIOMERS can exhibit antimicrobial effects, the static nature of the halo inhibition test may not have fully captured these properties . Factors such as pH, temperature, and the presence of organic matter can influence ion release and subsequent antimicrobial activity. Like the halo inhibition results, no significant differences were observed in biofilm formation among the materials. While GIOMERS have been shown to exhibit lower bacterial adhesion , the similar surface roughness of the tested materials may have mitigated this effect . The limited antibiofilm activity of the S-PRG particles may be attributed to the static nature of the biofilm model used in this study. Dynamic biofilm models, incorporating factors like sucrose challenge and pH fluctuations, may better simulate oral conditions and reveal differences in antimicrobial activity . Clinical studies have demonstrated similar results for bioactive sealants, with BeautiSealant® showing lower retention and marginal adaptation compared to FluroShield® . Further research is needed to optimize the formulation and application of S-PRG sealants to improve their clinical performance. Recent advancements in dental materials have introduced self-etching and self-adhesive technologies, aiming to simplify clinical procedures. However, our findings suggest that conventional etching techniques may still offer superior bond strength. Self-etching materials, such as BeautiSealant® and Constic®, demonstrated lower microshear bond strength compared to conventional techniques. This may be attributed to their limited ability to demineralize enamel and create optimal micromechanical retention . A systematic review by Botton et al . further supports this, indicating that self-etch systems may have lower retention rates over time. In contrast, conventional techniques, including the use of adhesive systems with enamel etching, exhibited higher bond strengths. The combination of phosphoric acid etching and a universal adhesive, such as Single Bond Universal, can enhance micromechanical retention and chemical bonding . The compatibility between FluroShield® and Single Bond Universal, facilitated by the presence of MDP and HEMA, may contribute to the improved bond strength . Clinical studies have consistently demonstrated the superior performance of conventional etch-and-rinse techniques over self-etching systems . The removal of the enamel smear layer through acid etching is crucial for optimal bond strength. Self-etching systems may struggle to effectively remove this layer, leading to weaker bonds. Our in vitro study provides valuable insights into the bond strength of various sealant materials. While self-etching and self-adhesive systems offer convenience, they may compromise bond strength compared to conventional techniques. The limited demineralization and micromechanical retention provided by self-etching systems can impact long-term clinical performance . Additionally, Beautifil Flow Plus® demonstrated superior performance in terms of microhardness and shear bond strength, making it a suitable option for high-risk caries patients and those with Molar-Incisor-Hypomineralization (MIH) . For simpler applications, FluroShield® can be used without an adhesive system, reducing clinical time. The selection of a sealant technique depends on various factors, including patient cooperation, saliva control, and the need for remineralization. While self-etching systems offer convenience, conventional techniques often provide superior bond strength and long-term performance. Clinical experience and operator skills are crucial for successful sealant application. Our findings indicate that materials with higher inorganic filler content, such as Beautiful Flow Plus®, exhibit superior microhardness, potentially enhancing their durability under occlusal stress. However, surface roughness remained consistent across all materials, suggesting that this factor may not significantly influence clinical performance. Microbiologically, all materials demonstrated similar behavior, indicating that the inherent antimicrobial properties of the resin-based materials may play a significant role in inhibiting bacterial growth. While conventional adhesive techniques continue to offer superior bond strength and long-term clinical performance, self-etching and self-adhesive systems may be suitable for specific clinical scenarios. Dentists should carefully consider the patient's needs when selecting a sealant material. For patients at high risk of caries or sensitivity, bioactive materials like Beautiful Flow Plus® may be advantageous. In situations where time efficiency is a priority, FluroShield® can be applied without an adhesive system, simplifying the clinical procedure. Ultimately, the choice of material should be based on the clinician's expertise and the specific requirements of each patient. |
Periodontal Molecular Diagnostics: State of Knowledge and Future Prospects for Clinical Application | b4791307-2122-41d8-a2c1-58f6002c0094 | 11641260 | Pathology[mh] | Inflammatory periodontal diseases, among them gingivitis and periodontitis, are the most prevalent in the world . It is estimated that these diseases affect 743 million people worldwide and their advanced forms affect 11% of the population . Periodontal disease leads to immunologically mediated loss of soft and hard tissues in the periodontium and, if untreated, can result in tooth loss . Tooth loss, and edentulism in the worst cases, worsens the patient’s quality of life and can lead to impaired chewing function, which carries the risk of malnutrition . In addition, patients may have phonetic, aesthetic, and psychological problems such as loss of self-esteem . Additionally, periodontal disease, as a source of chronic inflammation, has a major impact on the patient’s overall health due to its association with other diseases such as cardiovascular disease , diabetes , pregnancy and perinatal complications , obesity and metabolic syndrome , rheumatoid arthritis , cancer , and Alzheimer’s disease . Unfortunately, patients’ knowledge of periodontal disease is not widespread . The World Health Organization (WHO) points out that oral diseases, including periodontal disease, are an important population problem, having a very high impact on patients’ well-being and generating enormous treatment costs . Given the above, precise, prompt, and predictive diagnosis of periodontal disease is of paramount importance for clinicians. However, despite advances in molecular or microbiological research, the basis of periodontal diagnosis is clinical examination enriched by the evaluation of radiological images. A characteristic feature of periodontal disease is the formation of periodontal pockets and the loss of the attachment, which can be easily examined using a periodontal probe . Pocket probing is a clinical method of diagnosing the disease and monitoring the progress of treatment. The main measurements are PD (probing depth), GR (gingival recession), and CAL (clinical attachment level). In addition, inflammation, measured by BOP, and tooth mobility are assessed. In radiological image-based diagnostics, attention is paid to the type and extent of alveolar bone destruction. An optimal periodontal examination should include circular probing of each tooth with values recorded at 6 measurement points . Periodontal diagnosis is therefore often concentrated at a given point in the periodontium. It should be taken into account that the clinical measurements of PD, GR, and CAL, and the evaluation of radiographs, reflect changes that have already occurred in the periodontium. On their basis, no prognosis of the further course of the disease can be made. With these clinical measurements, we are not able to determine whether they concern active sites with a further tendency to progress or passive sites that may remain stable for many years . A clinical probe examination only allows us to describe the current state of the periodontium. There are no available tests for the clinical prognostication of periodontitis. Therefore, the detailed clinical examination is deepened by the assessment of the BOP index, which is considered an objective indicator of inflammation. The index is assessed as a percentage of bleeding sites in relation to all probing sites . However, the BOP as a stand-alone indicator is not a good predictor of periodontal disease progression because not every site with a positive BOP will experience attachment loss. Achieving a positive BOP in four out of four probings at a given site suggests a 30 percent probability of further attachment loss at that site. However, a negative BOP in four out of four consecutive clinical examinations suggests, with a probability of 98.5%, that no further attachment loss will occur at the point assessed . Patients with deep pockets, advanced adhesion loss, and numerous bleeding periodontal sites after probing are considered to have a higher chance of further attachment destruction than patients with a similar periodontal status but low BOP . However, a mean BOP value that indicates a higher risk of disease progression has not been established . Therefore, there is a great need among clinicians to have diagnostic tests that not only describe the periodontal changes that have occurred in the tissues but also allow them to detect disease at a subclinical stage before destruction of the tooth-supporting tissues occurs. This test would enable clinicians to follow the course of the disease and detect areas prone to exacerbation, with the possibility of being used to evaluate the effectiveness of ongoing periodontal therapies . Unfortunately, we do not have such a diagnostic method yet. The hope is molecular diagnostics, which is constantly being developed but not widely used in the clinic. There are numerous studies on biomarkers of periodontal disease. Point-of-care tests are also emerging. However, they have a minor role in the overall diagnostic process. In this manuscript, we will try to summarize the molecular diagnostic possibilities of periodontal disease and answer the question of why their widespread use is not possible at this stage. We also want to show how the current state of scientific knowledge may influence clinical diagnostics in the future. The oral cavity is a complex microbial environment inhabited by more than 700 species of microorganisms . Each person is host to approximately 100–200 species, resulting in a high degree of individual variation. Bacteria responsible for the development of periodontitis form supra- and subgingival biofilms . However, the prevalence of periopathogens in society is greater than the diseases they cause. This indicates a differentiated individual host response to the bacteria, which is responsible for tissue destruction and disease progression . Views on the impact of microorganisms on the periodontium have been changing over the decades and are still evolving. Cessation of oral hygiene leads to plaque accumulation and the development of gingival inflammation. The clinical changes are accompanied by microbiological changes. The early plaque is mainly composed of Gram-positive cocci and bacilli. Over time, it becomes a mature biofilm inhabited by Gram-negative cylindrical bacteria, spirochetes, and rods. The “non-specific plaque” hypothesis assumes that excessive bacterial growth in plaques leads to an increase in its virulence. Therefore, preventive and therapeutic measures should be directed to the removal of plaque formation . The development of the possibility of culturing anaerobic bacteria and their detection by molecular methods have contributed to the “specific plaque” theory, which conjectures that it is the excessive growth of species considered to be periopathogenic that leads to periodontal inflammation, and treatment should be based on targeted elimination of these bacteria . This was confirmed by the groundbreaking work of Socransky and colleagues, who divided plaque microorganisms into five contractual color-coded bacterial complexes, of which the most pathogenic is the red complex consisting of Porphyromonas gingivalis ( P.g. ), Tannerella forsythia ( T.f. ), and Treponema denticola ( T.d. ). It is detected in deep periodontal pockets and associated with active inflammation . Porphyromonas gingivalis is considered a key pathogen found in advanced severe periodontitis. The pathogenic potential of this anaerobic bacterium is a result of the production of many virulence factors, such as peptidylarginine deiminase (PPAD) and the lipoprotein RagB, a component of the nutrient acquisition system and gingipains . It also produces adhesins that allow it to bind to epithelial surface, erythrocytes, and other oral structures. The number of proteinases it produces causes the breakdown of tissue integrity . Treponema denticola and Tannerella forsythia are also Gram-negative anaerobes with similar characteristics but lower virulence . Aggregatibacter actinomycetemcomitans is a Gram-negative bacterium associated not only with advanced periodontal disease in adults but also with periodontitis occurring in young people and with other severe infections outside the oral cavity . It is known that the above microorganism releases an endotoxin that induces severe inflammation in the periodontium (lipopolysaccharide, LPS). However, it has been demonstrated that in addition to LPS, this bacterium secretes two specific exotoxins—leukotoxin A and cytolethal distending toxin (CDT) . Further studies by Socransky confirmed that periodontitis is not a classical infection and that the bacteria causing it do not fulfill Koch’s postulates . This gave rise to the “ecological plaque” hypothesis. This theory assumes that in a healthy state, there is a balance between the host and its microbiome. Changes in the microenvironment lead to excessive growth of certain species, which begin to act as opportunistic pathogens that initiate inflammation. The microbial imbalance is dysbiosis. Therefore, the term “multi-microbial synergy and dysbiosis” is used to refer to this phenomenon . In a bacterial plaque environment, different bacteria have distinct functions and collectively form a microflora that can lead to the development of inflammation. As a result of inflammation, tissue breakdown occurs, which is a source of nutrients for pathogens. Further growth of these pathogens exacerbates the inflammation. This leads to a vicious circle where dysbiosis fuels inflammation and inflammation intensifies dysbiosis. This sustained inflammation is not self-limiting . In view of the repeatedly confirmed role of the bacterium P. gingivalis , which has the ability to transform symbiotic microflora into dysbiotic one—due to its numerous virulence factors, the ‘key pathogen theory’ has emerged. This hypothesis, in contrast to the “specific plaque” hypothesis, assumes that even a small number of P. gingivalis cells can have a tremendous impact on the subgingival biofilm . In 2020, a new theory of the etiology of periodontitis emerged called the IMPEDE (Inflammation-Mediated Polymicrobial Exacerbation) model. This model assumes that it is inflammation that leads to dysbiosis, which drives the transition of the oral health condition to periodontitis. The following is still relevant; therefore, the question of whether the dysbiotic changes in the biofilm are the cause of the disease, or whether it is the inflammation that leads to a change in the conditions of the subgingival environment that results in dysbiosis, is still open . There has also been a recent hypothesis that subgingival plaque mimics human tissue in both structure and function. Thus, inflammation results from the disruption of the normal function of a healthy biofilm, which begins to stimulate the host’s defense mechanisms, leading to excessive inflammation that is difficult to regulate . Due to the strong role of bacterial factors in the etiology of periodontitis, relationships between the presence and quantity of specific microorganisms in pockets have been studied for use in diagnostic and therapeutic procedures . Since subgingival plaque is a biofilm, the object of microbiological examination should not be individual cells, but rather the entire biofilm structure, where the presence of some species is determined by the colonization of others . Most of the bacterial species that populate the oral microbiome cannot be cultured under laboratory conditions, and previous methods such as bright-field and dark-field microscopy are of little use in these cases . New molecular methods for identifying bacterial species include hybridization and polymerase chain reaction (PCR). The DNA–DNA checkerboard Socransky’s method proved to be a breakthrough. Methodologically, it relied on the use of whole-genomic probes, which were a purified collection of DNA extracts from pure cultures. DNA isolated from bacterial samples was attached to the surface of a nylon or nitrocellulose filter. Then, the filter was subjected to hybridization with DNA probes from at least 40 bacterial species. The number of individual species was determined by the intensity of the fluorescent light from the molecular probes. Based on the frequency of detection of subgingival species, five color-coded complexes were proposed . Almost simultaneously, in the 1990s, the 16S rRNA gene was discovered. It is a component of the 30S subunit in the prokaryotic ribosome. All bacteria have it in their genome and there are unique sequence differences in it that allow the microorganism to be classified to a particular genus or even species . With the advent of modern diagnostic techniques, polymerase chain reaction (PCR) began to be used, allowing cloning and sequencing of bacterial genomes with their subsequent analysis. The standard PCR method is a qualitative method . In addition to the presence of bacteria, the development of quantitative polymerase chain reaction (qPCR) allows the quantification of the microbial composition of the sample . Other modifications of the PCR technique have also found applications in diagnostics of periodontitis. Nested PCR is used when there is a need to increase the accuracy or specificity of the test. It involves the sequential use of two sets of primers; the first is used to amplify the target sequence and the resulting amplicon is then used as the sequence for a second amplification . When a polymicrobial sample is analyzed, such as a dental biofilm, a multiplex polymerase chain reaction can be used. This makes it possible to identify organisms or genes using a single reaction and minimizes the required sample volume . In 2009, a microarray for the identification of microbes in the human mouth (HOMIM—Human Oral Microbe Identification Microarray) was used. This method allowed the detection of 300 species across different periodontal conditions . Thanks to this, a dedicated repository has been created—the Human Oral Microbiome Database (HOMD) . The use of small probes allowed the identification of nearly 300 bacterial phylotypes in a single sample . Finally, in 2016, the HOMINGS system was developed using next-generation sequencing (NGS). This allowed the identification of approximately 700 phylotypes in a single sample . Due to the complex structure of the subgingival biofilm, species identification alone does not describe the relationship between the bacteria forming the microbiome. Bacteria have a wide range of possibilities for interaction and adaptation through modulation of protein synthesis, metabolism, and the secretion of small molecules. A comprehensive understanding of the functional diversity of the subgingival microbiome requires the combination of information based on DNA and RNA with bacterial products and proteins, namely through metabolomics and metaproteomics . Studies of the subgingival microbiome conducted by sequencing the gene encoding 16S rRNA have shown that Actinomyces naeslundii , Rothia dentocariosa and aeria , and Streptococcus sanguis are detected in areas with healthy periodontium. In sites with periodontitis, abundant growth was found for Porphyromonas gingivalis and endodontalis , Prevotella intermedia , Tannerella forsythia , Treponema ( denticola socranskii , maltophilum , and lecitinolythicum ), Selenomonas sputigena , Parvimonas micra , Peptostreptococcus saphenum , and Fretibacterium fastidiosum . In addition, core bacterial species such as Fusobacterium nucleatum play an important role in the transition from health to disease, possibly serving as a metabolic anchor for other community members. F. nucleatum can be a prerequisite for late colonizing anaerobic species . Advances in molecular diagnostics made it less expensive and resulted in a range of commercially available tests that allow the molecular evaluation of selected bacterial species from periodontal pockets. The results can report the bacterial count in the sample or the content of specific strains of pathogens. Examples of available microbiological tests are PET (MIP Pharma holding GmbH, Blieskastel, Germany), enabling quantitative evaluation of up to nine periopathogens per sample by real-timePCR; IAI PadoTest (ParoX GmbH, Leipzig, Germany) for the detection of key pathogens ( P.g , T.f , T.d , Prevotella intermedia ( P.i ), and Filifactor alocis ) by multiplex rtPCR; PerioPOC (GenSpeed Biotech GmbH, Gebaude B, Austria), a semi-quantitative method with a detection limit of 10 4 CFU/l (detects P.g , T.d , T.f , P.i , and A.a ); and MyPerioPath (OralDNALabs, Flying Cloud Eden Prairie, MN, USA), which classifies pathogens into three risk groups on the basis of PCR testing. The Test PadoBiom Kit (ParoX GmbH, Leipzig, Germany) uses next-generation sequencing (NGS) of GCF samples, provides the ability to assess bacterial diversity and pathogenicity, and detects resistance genes to antibiotics such as beta-lactams, quinolones, nitro-imidazole derivatives, tetracyclines, and macrolides. Despite studies confirming the validity of the use of microbiological tests to detect putative periodontopathogens, this topic needs more extensive studies. Such tests give us information about the possible number and types of bacteria present in periodontal pockets, which does not significantly affect clinical management. Laboratory tests make sense when the information they provide helps to plan treatment and apply optimal therapy. Many infectious diseases are treated without a thorough microbiological diagnosis because the probability of detecting a known pathogen for a given infection is very high and experience shows that standard anti-infective therapy is effective. To diagnose gingivitis or periodontitis, a microbiological diagnosis is not necessary because clinical examination is sufficient. With the modern approach to the etiology of periodontitis, diagnosing single bacterial species for the clinician becomes meaningless, because the detection of the presence of potentially pathogenic microorganisms is not necessarily indicative of disease, nor does it provide information about its activity or propensity for progression. Despite the possibility of using sophisticated laboratory methods to detect bacteria both quantitatively and qualitatively, this knowledge does not translate into a change in the therapeutic process. It is important to note that current guidelines for antibiotic prescriptions in periodontology are not based on the detection of specific pathogens but on a clinical assessment of the severity of the disease . However, an accurate microbiological diagnosis may be useful for treatment planning for patients who need to eradicate bacteria with exogenous pathogen characteristics (i.e., A.a. and P.g ), particularly in patients classified before 2018 as having aggressive periodontitis and in patients with refractory forms of periodontitis. Taking into account the above, precise microbial testing could help to select subjects who benefit from antibiotic therapy, assist in selecting the right antibiotics, and contribute to minimizing the overuse of antibiotics in society . Such diagnostics do not require species assessment but only the identification of antibiotic resistance genes. Knowledge of resistance to antibiotics of a certain type would reduce the number of unsuccessful antibiotic treatments. Another desirable application of microbiological diagnostics can be the detection of changes in the biofilm that anticipate the onset of disease. A change in the microbiome always precedes an inflammatory response, so early detection of changes is desirable and may allow identification of patients who are ‘at risk’ before clinical signs of inflammation in the periodontium occur . A shift in the microbiome from symbiotic to dysbiotic can alert clinicians to the likely onset of disease. The same applies to patient monitoring during maintenance therapy visits. The impact of the bacterial agent on the recurrence of active inflammation is undisputed. Early detection of changes in the microbiota would make it possible to individually determine when the next prophylactic intervention is indicated. However, the detection of certain pathogens has proved less useful than an examination of the overall bacterial load. In addition, it was proven that maintenance visits set every three months provide stability for almost all patients and those carried out at intervals longer than six months increase the risk of relapse, so a microbiological evaluation is not strictly necessary . The microbiological tests may also be a measure of the return to homeostasis between host and plaque bacteria after effective periodontal treatment. Chen et al. developed the subgingival microbial dysbiosis index (SMDI) to be able to describe changes in the microbiome . The SMDI is based on 49 discriminating species. Among them, the top species associated with periodontitis were Treponema denticola , Mogibacterium timidum , Fretibacterium spp., and Tannerella forsythia. In contrast, Actinomyces naeslundii and Streptococcus sanguis were the top health-associated species. Finally, Treponema , Fretibacterium , and Actinomyces are used to calculate the simplified SMDI . The SMDI was used to compare the microbiome before and after scaling and root planing. There was a significant decrease in the SMDI between day one after therapy and stabilization up to three months. These changes indicated a decrease in dysbiosis after nonsurgical periodontal treatment . The effect of microorganisms on periodontal tissues is bidirectional. One direction is direct action, referring to the release of enzymes that destroy host structures (proteases, collagenases, etc.). The other is an indirect impact via bacterial cytotoxins and involves the stimulation of destructive actions of the patient’s own immune system. In response to the detection of substances of bacterial origin, polymorphonuclear neutrophils (PMNs) initiate the production of pro-inflammatory factors. MMPs are responsible for the destruction of connective tissue, and PGE2 (prostaglandin E2) stimulates osteoclasts to destroy the bone structure . In this complex immunological reaction, many pro-inflammatory biomolecules are secreted. They may originate from the host or are the products of tissue destruction . Many of them are studied as possible periodontitis markers. Biomarkers are searched for in saliva, subgingival plaque, tissue biopsies, and gingival crevicular fluid (GCF). The gingival crevicular fluid (GCF) itself is a plasma filtrate and/or inflammatory exudate and can be collected from the gingival crevice/periodontal pocket. The strong vascularity of the periodontal tissues contributes to the continuous filtering of fluid into the gingival crevice. The components of the fluid originate from both the host and subgingival microorganisms. Among the important host-derived components are inflammatory markers including enzymes, cytokines, and interleukins . The association of increased GCF volume with increased severity of periodontal inflammation is well-documented . It has been suggested that increased GCF volume and the appearance of bleeding after probing is one of the earliest signs of disease and the increase in GCF volume itself may be a sign of subclinical ongoing inflammation . GCF collection is non-invasive and provides information about a specific site in the periodontium. Although reviews have suggested that there is considerable potential for the future application of GCF-based tests, the question is whether this will lead to improved disease detection and patient outcomes. At present, there is a large gap between diagnostic capability and clinical application. A positive test result should guide the clinician to change the treatment plan accordingly; however, at the moment, it only provides more information about the course of the disease. Additionally, there are no dedicated laboratories for analyzing GCF components, and chair-side testing is not available for most markers. Therefore, more research is needed to identify reliable biomarkers for periodontal disease monitoring. Such biomarkers for periodontal diagnostics need to be optimized through studies combining biochemical and clinical periodontal data. Validation of candidate biomarkers in large populations is also required. Taking into account its unparalleled affinity for periodontal tissues, GCF should be the medium of choice. It is excellent for site-specific diagnosis. However, saliva collection provides information from the entire oral cavity, not only from specific sites of periodontium. Saliva comprises secretions from salivary glands, oral mucosa cells, blood, and GCF. Saliva as a medium is therefore more practical, easier, and cheaper to collect. It contains discharges from parotid, submandibular, sublingual, and many small salivary glands, the secretion of which is influenced by many environmental and psychological factors. In addition, saliva contains mucins and cellular debris. Similarly to serum, it contains DNA, mRNA, microRNA, proteins, metabolites, and microbiota . The variety of molecules makes it a difficult medium to work with. Therefore, in the case of periodontitis, it is still secondary to GCF . Saliva can be collected stimulated, unstimulated, and as a swab from the floor of the mouth or gums, in which case it contains more GCF . Saliva is a convenient medium to collect and its composition is very rich. Proteotome saliva analysis has revealed the presence of 3000 different biomolecules . There are about 3500 publications on salivary biomarkers . However, saliva testing is not yet a routine component of dentistry. However, attempts are being made to use saliva as a medium for the diagnosis of many general diseases, particularly in oncological diagnosis and the detection of autoimmune diseases, systemic microbial infections, or diabetes . Capillary electrophoresis mass spectrometry-based saliva metabolomics can distinguish pancreatic cancer from oral cancer, breast cancer, and cancer-free controls . Biomarkers for lung cancer can improve early detection beyond the use of computed tomography scans , as well as detect breast cancer . Identification of reliable saliva biomarkers can provide a convenient non-invasive way for cancer detection . Saliva is therefore a medium with a very high diagnostic potential, but its use in the diagnosis of periodontal disease by dentists is practically non-existent. There are numerous research studies on the role of inflammatory mediators, host-derived enzymes, oxidative stress markers, and tissue breakdown products in oral fluids. The most extensively studied molecules are IL-1β, IL-6, IL-8, MMP-8, TNF-α, and PGE2 . Meta-analysis performed for salivary biomarkers showed the highest sensitivity for the diagnosis of periodontitis for IL-1β (77.8%) and MMP-8 (72.5%), and the highest specificity for MMP-9 (81.5%) and IL-1β (78%) . For GCF molecules, the highest sensitivity (76.7%) and specificity (92%) were shown for MMP-8 . Bone turnover markers are also of interest. Osteoclastic activity is mainly regulated by receptor activator of nuclear κ B (RANK), its ligand (RANKL), and OPG (osteoprotegrin). The RANKL/OPG ratio is important in determining bone resorption. Increased levels of RANKL and decreased levels of OPG were noted in periodontal inflammation . The RANKL/OPG ratio may be a predictor of sites at risk of progression. Despite many years of searching, no single biomarker of periodontitis has been identified that, when assayed from saliva or GCF, would provide information about what has happened, is currently happening, and will happen in the periodontium. That is why the idea of combining several biomarkers emerged. It was shown that combining biomarkers shows better diagnostic accuracy in the detection of periodontal disease and potentially provides information on the stages of disease . The most promising combinations comprise salivary molecules and the presence of known periopathogens. High salivary concentrations of MMP-8, IL-1β, and P. gingivalis were associated with deep pockets and alveolar bone loss, and high levels of MMP-8 and IL-1β with bleeding on probing. The CRS (cumulative risk score) is a mathematical model used to define the individual assessment of the risk of developing periodontal disease . Three selected salivary biomarkers represent three components of periodontal inflammation: periopathogens ( P. gingivalis ), cytokine production (IL-1β), and tissue degradation (MMP-8). The CRS index had a strong association with moderate to severe periodontitis , similar to the determination of the presence of P. gingivalis with IL-1β and PGE2 levels . The 2023 systematic review of the accuracy of multiple molecular biomarkers in oral fluids revealed two biomarker combinations with high diagnostic accuracies. In saliva, combinations of IL-1β, IL-6, and MMP-8 have superior properties for the detection of periodontitis . Matrix metalloproteinases (MMPs) are zinc-dependent endopeptidases. These enzymes have the ability to degrade basement membranes and the extracellular matrix, which enables tissue remodeling and cell movement during physiological processes and inflammatory states in the body. MMPs are secreted by many host cells such as polymorphonuclear leukocytes, macrophages, fibroblasts, bone cells, epithelial cells, and endothelial cells . The main MMPs derived from neutrophils are MMP-8 (collagenase-2) and MMP-9 (gelatinase B). MMP-8 is produced during the maturation of PMN cells and then stored in their granules. During the course of inflammation, the content of the granules is released into both saliva and the GCF. MMP-8 is responsible for 90–95% of collagenolytic enzyme activity in the GCF. Its levels are much higher in patients with periodontal disease than in healthy patients. MMP-8 and MMP-9, as the main collagenolytic enzymes in saliva and gingival crevicular fluid, are responsible for collagen degradation in gingivitis and periodontitis . That is why MMPs were analyzed as diagnostic and prognostic markers. Examination of MMP-8 and MMP-9 concentrations in saliva, together with confirmation of the presence of the bacterial red complex and analysis of orthopantomograms, had predictive value . High concentrations of MMP-9 and MMP-13 showed active sites with disease progression . Long-term high levels of MMP-8 in the GCF indicated a high risk and poor responses to therapy . However, most analyses did not take into account the fact that the amount of MMPs corresponds to the presence of the cells through which they are secreted, rather than to direct tissue destruction . Nevertheless, MMP-8 is still the most promising biomarker for periodontal disease, with high specificity and sensitivity. These findings were used to develop a point-of-care test (PerioSafe ® ,Dentognostics GmbH, Solingen, Germany) to measure active MMP-8 in saliva as an adjunct to clinical diagnostics. The procedure itself involves taking a sample at least 30 min after ingestion of the last meal or drink . Although this test has a low sensitivity (33%), it has a high specificity of up to 93% for detecting periodontitis. When combined with age and smoking, sensitivity was improved to 82.5% and specificity was increased to 84.4%. This test may be useful in periodontal screening in conjunction with patient characteristics . There are attempts to use other chair-side tests such as a lateral flow metalloproteinase 9 point-of-care test (MMP-9 LFT POC test). Its diagnostic ability to detect periodontitis was appropriate, with a sensitivity of 0.92 and a specificity of 0.72 . Many opportunities are opening for clinicians with the introduction of chair-side point-of-care tests (POCTs). These involve taking a saliva or GCF sample from the patient, inserting it onto a special microchip, and reading the result a few moments later, while the patient is still in the chair. The test result tells us about the presence of a specific biomarker at the site from which the sample was taken and could be used to screen patients for periodontal disease. Such tests could be applied in many places providing health services to the public, i.e., general medical and dental practice offices, nursing homes, outpatient clinics, and potentially for self-testing at home . Chair-side diagnostics provide instant results, thereby reducing time and costs compared to laboratory diagnostics. However, analysis of the subgingival microbiome is not routinely used in periodontal clinics. Periodontitis is a complex, multifactorially inherited disease, dependent on multiple alleles, and its phenotypic picture depends on the interaction between genetic predisposition and environmental factors . As with other complex diseases, it is slowly progressive, presents a relatively mild phenotype, and is chronic in nature . Periodontitis is polygenic and each gene has low penetration; therefore, it can be called a disease-modifying gene. It is estimated that 10 to 20 disease-modifying genes are involved in the disease . Gene polymorphisms can cause changes in protein coding or expression, which may result in changes in innate or acquired immunity and therefore may determine disease or protect against it. A single nucleotide polymorphism (SNP) is an important genetic marker, along with copy number variation. Nowadays, SNPs can be detected in large amounts using high-throughput techniques . Many studies were performed to understand gene polymorphisms in the etiology of periodontitis. The most widely studied were genes IL-1A (interleukin-1A), IL-1B (interleukin-1B), IL-4 (interleukin-4), IL-6 (interleukin-6), IL-10 (interleukin-10), TNFα (tumor necrosis factor α), FcγR (the Fcγ receptor), VDR (vitamin D receptor), CD14 , TLR2 (toll-like receptor 2), TLR4 (toll-like receptor 4), and MMP-1 (metalloproteinase-1) . However, the frequency distribution of the polymorphisms of the indicated genes does not correlate with the prevalence of periodontitis. It is likely that a single gene may have little influence on disease initiation and pathogenesis and other genes may interact. At the same time, no single gene has a dominant influence. The role of genetics in the pathogenesis of periodontitis is difficult to explain, as there are complex interactions between genes and their polymorphisms and between genes and environmental factors. Therefore, polygenic models seem to be more accurate than monogenetic models, but the progress of the latter is very slow . Additionally, the distribution of polymorphisms varies between ethnic groups, and differences due to ethnicity are often found in periodontitis . The IL-1 gene cluster was among the first to be analyzed, and SNPs in the IL-1 gene cluster were proposed for use in genotyping periodontal patients . After the study by Kornman, which found an association between polymorphisms in the IL-1 gene cluster and the severity of periodontitis in non-smokers, genetic studies in periodontology began . There are three genes that regulate the production of interleukin-1: IL-1A , IL-1B , and IL-1RN (interleukin receptor antagonist). These genes are located on chromosome 2. The first two control the production of pro-inflammatory IL-1α and IL-1β. IL1-RN codes for the synthesis of IL-ra, which inhibits the secretion of IL-1α and IL-1β. In Caucasians, the polymorphisms IL-1A −889T/C and IL-1B 3953/4 C/T are associated with chronic periodontitis . On this basis, a commercially available genetic susceptibility test was developed (Periodontal Susceptibility Test, Straumann, Waltham, MA, USA) to determine the risk of severe chronic periodontitis. The test detected the simultaneous presence of IL-1RA + 4845 ( IL-1RA + 4845 is being used because it is easier to identify than the IL-1A −889 polymorphism and it is concordant with it) and IL-1B + 3954. The presence of both polymorphisms meant that the patient was considered “genotype-positive” and predisposed to increased IL-1β release . Genotype-positive, nonsmoking patients were found to be 6.8 times more likely to have severe chronic periodontitis than genotype-negative patients . However, analysis of the clinical trials utilizing this genetic test has yielded inconclusive results, as did the attempt to link the result of a genetic test to bleeding on probing, loss of clinical attachment, and loss of bone or teeth . Therefore, the genetic test for periodontitis resulted in unclear benefits and subsequent studies were unable to confirm the practical utility of the observed associations . At the individual level, the expression of the disease depends on genetic, bacterial, and environmental factors. The link between them is epigenetics. Epigenetics is the science that deals with heritable mechanisms of gene expression that are not dependent on changes in the DNA sequence. Consequently, epigenetic mechanisms regulate gene expression and allow cells with identical genetic material to perform a variety of functions in the body. Environmental factors such as microorganisms or smoking may affect gene activation and cell phenotype . Genetic factors are not the only risk factors for periodontitis. Genetic predisposition means that a patient has an inherited susceptibility to develop the disease, but it does not mean that a person with such a genetic tendency is doomed to its development. The validity of commercially available genetic tests for complex diseases is therefore questioned, comparing their accuracy to that of horoscopes . That is why genetic profile testing for patients is not used in clinical practice. The tests are expensive and do not add diagnostic value. In addition, no protocols have been developed for following up on the identification of a specific gene variant in a patient. In periodontology, it is questionable if every patient should receive different, personalized treatment according to their genetics. There were no analyses to determine whether genetic screening in patients with or without periodontitis is economically and ethically justifiable. However, there is space for future use of human genomics in the characterization of disease subtypes and personalized care plans. In particular, cases of children and young adults with periodontitis would benefit from such diagnosis. Familial aggregation of periodontitis in the aforementioned patients was confirmed in single-family and large twin studies . However, at the moment, it is clear that single-gene inheritance is rare and is mainly considered in the case of genetic syndromes . There are many putative loci and SNPs that may confer a genetic risk for periodontitis but they lack universal validation across different settings and populations . A validated polygenic risk score (PRS) that aggregates the information from hundreds of thousands of genetic loci and polymorphisms measured via genome-wide association studies (GWAS) is needed to identify healthy people with susceptibility to the disease. Unfortunately, such a PRS has a strong risk of bias that is connected with ethnicity and ancestry. Genetic association signals discovered in GWAS studies in populations of European descent are not always transferable to non-European populations . In order to generalize, the PRS needs to be verified in populations from different ethnic backgrounds. In summary, there is huge potential in such studies; however, time is needed for further research into the genetic links of periodontitis so that this knowledge can be applied routinely in practice. Before this, new approaches will be supported by high-quality evidence, and information about the patient’s family history of periodontitis can still be used to identify susceptible subjects with an elevated risk of periodontitis . The current classification of periodontal diseases was established in 2017 and replaced the 1999 classification developed by the International Workshop for Classification of Periodontal Diseases and Conditions . It covers the three main forms of periodontitis, i.e., necrotizing periodontal disease, periodontitis as a symptom of systemic disease, and periodontitis (previously divided into chronic and aggressive). The progression of the disease and further risk of progression are described by stages and grades, which may be modified as new scientific evidence emerges. Stages describe the progression of the disease and the complexity of the treatment process. Grades provide additional information about the progression of the disease and its biological features . The previous classification divided periodontitis into chronic (CP) and aggressive (AP) forms. To make a diagnosis of chronic periodontitis, an aggressive form of the disease had to be excluded. Clinically, CP was more common in adults (but could also be diagnosed in children and adolescents), tissue destruction went hand in hand with plaque accumulation, and subgingival calculus deposits were often found. In addition, the course of the disease was slow or moderate with possible episodes of rapid tissue destruction . In contrast, the primary features of AP were rapid attachment, bone loss, and a family history of the disease. The secondary clinical features were an imbalance between the amount of dental deposits and periodontal tissue destruction, elevated proportions of Aggregatibacter actinomycetemcomitans ( A.a ) or Porphyromonas gingivalis ( P.g ), phagocyte abnormalities, and macrophages hyperresponsiveness . Clinical diagnostic assumptions can be considered subjective and in many published papers, the diagnosis of AP has been considered incorrect or at least questionable , as confirming secondary molecular features was not mandatory. Microbiological diagnostics have been attempted to distinguish between AP and CP on the assumption that the detection of Aggregatibacter actinomycetemcomitans ( A.a ) would confirm the diagnosis of AP. In the late 1970s, A.a . was isolated from deep pockets of patients with localized aggressive periodontitis (LAP) , which was later confirmed by the finding that the prevalence of this bacterium in LAP was very high . Subsequent work confirmed that 96.5% of LAP patients carried A.a. , while only 20.8% of CP patients and 16.9% of periodontally healthy patients harbored this bacterium . However, the results of other studies were inconclusive. When AP and CP patients were tested for A.a. and P.g. , Aggregatibacter actinomycetemcomitans was detected in 54% of AP patients and 47% of CP patients. Inversely, Porphyromonas gingivalis was detected in 67% of CP patients and 52% of AP patients . In addition, only the JP2 A.a. strain was found to have the typical characteristics of an exogenous pathogen and the others showed characteristics of opportunistic bacteria . The JP2 clone strains are highly prevalent in human populations living in northern and western parts of Africa and in populations originating from these geographical regions. Only sporadic signs of dissemination of the JP2 clone strains to non-African populations have been found. A. actinomycetemcomitans is a source of leukotoxin (LtxA) and cytolethal distending toxin (Cdt). LtxA is able to kill human immune cells and its production is enhanced in JP2 clones. This is why the highly leukotoxic A. actinomycetemcomitans JP2 clone is associated with rapidly progressing periodontal disease and is frequently detected in LAP (i.e., Localized Stage III Grade C Periodontitis). Patients colonized with the JP2 strain can transmit it to family members and partners, making its eradication difficult . It is also hypothesized that LAP may develop into GAP over time, which simultaneously leads to the development of more complex microbiota beginning to resemble those in CP . The mere detection of A.a. in the patient’s plaque sample could therefore not confirm the diagnosis of AP, so it remained an ancillary test. PCR-based diagnostic tests, although becoming cheaper and more widespread, have not become an essential diagnostic tool. They are now readily available and allow confirmation of a certain number of pathogens, including A.a. and P.g. Early detection of the mentioned pathogens could prevent rapid periodontal attachment loss. The virulence factors of A.a. (LtxA) and P.g . (gingipains) are known and new virulence blocking strategies are emerging. Anti-virulence therapy fights periodontal pathogens by neutralizing their virulence properties and may be an alternative to antibiotic treatment. Host immune modulation by phytocompounds and oral microbiota replacement are new options for periodontitis treatment and prevention . Due to the fact that periodontitis is a multifactorial disease and its clinical expression depends on the imbalance between risk factors (including dental plaque bacteria) and the host response, immune disorders may be crucial. People with impaired neutrophil function, e.g., leukocyte adhesion deficit, Kostmann syndrome, Chediak–Higashi syndrome, Papillon–Lefevre syndrome, or Down syndrome, are predisposed to severe periodontitis . It was found that people with periodontitis at a young age may have reduced neutrophil chemotaxis and phagocytosis . Their hyperactive neutrophils secrete excessive amounts of oxygen radicals . However, a recent systematic review assessing the GCF composition in patients with chronic and aggressive periodontitis did not show sufficient evidence for differences between these diseases . Complicated and expensive methods for determining cytokines from GCF are completely unavailable to clinicians. The use of the term aggressive periodontitis was therefore common without a proper methodology for diagnosing the disease. The current 2018 classification does not distinguish aggressive periodontitis . It was stated that differences in etiology and pathophysiology are required to indicate separate periodontal diseases and clinical manifestations, and severity is not enough to support the concept of different diseases. However, it remains undisputed that significant differences in the clinical presentation of the disease are present, suggesting population variations in susceptibility and/or exposure. Even localized periodontitis diagnosed in young people cannot be confirmed by defined etiological or pathological elements and the disease mechanisms are poorly understood . The new classification is still open and allows for the inclusion of biomarkers in a case definition system, particularly in patients who are more likely to develop progressive severe generalized periodontitis, are less responsive to standard plaque control methods, and theoretically may have periodontitis severely affecting general disease. In these patients, standard clinical diagnostics may be insufficient and biomarkers that are currently available may be useful. Biomarkers may increase diagnostic accuracy and allow better assessment of severity. Therefore, the proposed classification framework allows the introduction of validated biomarkers into the case definition system. Specific biomarkers and their thresholds may be incorporated into the system as evidence becomes available . Despite the possibility of inclusion of widely understood periodontal disease biomarkers present in the GCF, saliva, or serum during staging and grading, the classification does not, at present, indicate specific molecules or the extent of their diagnostic value. The question of molecular diagnostics remains open and its widespread use is not yet apparent. Microbiological, molecular, or genetic diagnostics in periodontal disease provide scientists with a wealth of knowledge about the processes that occur in the healthy and inflamed periodontium. However, the practical use of these components that make up the picture of periodontitis currently seems impossible. Even state-of-the-art diagnostic methods that allow us to learn about the metabolome and proteome of oral fluids only allow us to assess the presence of certain molecules rather than their joint action. The complex interactions in the microbiome, the intersection of many metabolic pathways, or the interplay of genes or genes with the environment make periodontal disease diagnosis similar to a 3D jigsaw puzzle, whose components interact with each other and are not constant. When considering the complex etiology of periodontitis, it is unlikely that a single lab test will address all issues. The perfect periodontal diagnosis of the future should combine clinical, radiological, and laboratory examinations. Thus, to date, no specific protocols have been developed in which molecular tests would have a significant impact on increasing the diagnostic and therapeutic quality of the management of patients with periodontal disease. However, the continuing development of new diagnostic methods and the detailing of the knowledge of immune reactions in the periodontium will certainly be a step forward in the possibilities of periodontal diagnostics in the future. Valuable solutions will be transferred from the laboratory to clinical practice. Until this happens, clinical examination together with radiological assessment remains the primary means of periodontal evaluation and the patient’s diagnosis and treatment are based on it. |
Radiologist Involvement in Radiation Oncology Peer Review | 92bef111-b3d1-4661-83e3-43be70712597 | 11681381 | Internal Medicine[mh] | Peer review (PR) of radiotherapy (RT) treatment plans is critical for radiation oncologists to ensure high-quality and safe treatment delivery. This process is a requirement for site accreditation by radiation oncology governing bodies such as the American College of Radiology and the American Society for Radiation Oncology. Conventional PR consists of standing meetings in which a patient’s clinical scenario is presented and a general overview of their RT plan, including the site to be treated and RT prescription, is provided. While the specifics of the PR workflow vary by institution, the time spent reviewing each case is brief (approximately 2-3 minutes), and not all cases are reviewed. , , Given this format, it is challenging to ensure that all sites of disease are appropriately delineated and that normal tissues are adequately protected. To highlight this problem, a prospective study inserted erroneous plans into routine radiation oncology chart rounds and found that only 55% of errors were detected. Moreover, timeliness of PR is also critical: once the RT plan is finalized, the time and effort required to modify the plan based on reviewer feedback presents a barrier to consistent implementation of PR recommendations. To address these challenges, multiple institutions have transitioned to focused PR of contoured targets and normal tissue structures. , , , , , , This process is critically important to radiation oncologists, since target and normal tissue structures are unique to each clinical scenario, and their delineation is a leading source of interphysician variability. Among modern RT techniques, one common source of variability is image interpretation: to develop highly conformal RT plans, radiation oncologists have become increasingly dependent on their ability to interpret multiple imaging modalities (eg, magnetic resonance imaging, positron emission tomography, computed tomography) during the RT planning process. While radiation oncologists are classically trained to interpret clinical scenarios and understand distribution of gross disease and potential routes of spread, radiation oncology residents are provided with limited formal training in diagnostic image interpretation. Despite 87% of residents agreeing that a strong radiological knowledge base is moderately to extremely important, more than 60% feel only “somewhat confident” in radiology at graduation. , , , , , In radiation oncology residency, radiological instruction is primarily in the form of multidisciplinary tumor board imaging review and on-the-job training during routine clinical care. Guidelines for RT target delineation are available, but these only provide a framework and do not improve image interpretation skills. , As a result, significant variability in tumor delineation may exist between radiation oncologists and radiologists. To address these challenges and bridge gaps in multidisciplinary care, some institutions have developed PR practices that embed radiologists into PR conferences to provide a more detailed radiological review of proposed treatments. , , , , , We performed a systematic review and meta-analysis of published studies on radiation contour PR to better understand the association of radiologist input with changes to RT treatment plans. We performed a systematic review of the literature and meta-analysis according to the Preferred Reporting Items for Systematic Reviews and Meta-analyses ( PRISMA ) reporting guideline. This study was prospectively registered in the PROSPERO database ( CRD42024544451 ). Three databases (PubMed, Web of Science, and Scopus) were queried using standardized search terms to identify studies published from inception to March 6, 2024, that reported PR of contoured targets for the purposes of RT planning with or without radiology involvement. Search terms included ( quality assurance OR peer review OR chart rounds ) AND ( contour OR contours OR contouring OR segmented OR delineated OR delineation OR target OR targets OR volume OR volume delineation ) AND ( radiotherapy OR radiation OR radiation therapy ). Titles and abstracts were screened by 2 reviewers (R.T.H. and M.K.F.). In the case of discordance, discussion was used to reach consensus regarding inclusion for full-text review. Details on inclusion and exclusion criteria and the systematic review procedures are described in the eMethods in . After full-text review, factors extracted from the studies included cancer diagnosis of patients reviewed, PR timing (before RT planning vs after plan completion), specific structures reviewed in PR, radiologist inclusion, definition of major and minor changes (if applicable), RT plan changes and recommendations for change, and changes to each specific type of target (eg, gross tumor volume [GTV], clinical target volume, planning target volume [PTV]). Quality of studies was assessed using the methodological index for nonrandomized studies. Scores range from 0 to 24, with higher scores indicating higher study quality; the ideal score is 16 for noncomparative studies and 24 for comparative studies. The primary outcome of this study was the pooled rate of RT plan change of any kind after PR. These changes could relate to the contoured targets, organs at risk (OARs), prescribed dose, fractionation schedule, RT technique and modality, immobilization, motion management, imaging guidance, bolus or shielding, treatment intent, or treatment cancellation. To adequately capture PR outcomes, a combined outcome measure was used in this analysis: specifically, changes and recommended changes to RT plans were pooled together, regardless of whether they were documented as implemented by the included study. As secondary outcomes, major and minor changes were defined by the individual studies (eTable 1 in ); these study-level definitions were then used to categorize major and minor changes for the purposes of this analysis. Generally, major changes were defined as those expected to be clinically meaningful, by preventing geographic miss of gross tumor, affecting disease control outcomes, or affecting relevant toxic effects; minor changes were not expected to have substantial clinical importance. The total number of plan changes (of any type or significance) was extracted from studies, where reported, or calculated as the sum of major and minor changes, when these 2 change levels were reported separately. We evaluated the pooled rate of RT plan changes with and without radiology involvement in PR, as well as the pooled rate of change stratified by major and minor changes. We further looked at the rate of changes with and without radiology involvement specific to GTV, PTV, and OARs. Sensitivity analyses were also performed to assess the association of primary disease site, PR timing (before vs after the dosimetric planning process—the act of creating and finalizing the RT plan based on the contoured target volumes), PR frequency (daily, multiple times per week, weekly, or as needed), and study type (prospective or retrospective) with RT plan changes. Statistical Analysis The primary and secondary outcomes were estimated using a random-effects model, and heterogeneity was assessed using Cochran Q tests and the Higgins I 2 statistic. In cases of high heterogeneity ( I 2 > 50%), Cook distance was used to identify outliers, and pooled rates were recalculated after exclusion of outliers. Differences between subgroups were assessed using the omnibus test of moderators (Qm), testing the null hypothesis that all predictive factors entered into the model are unrelated to the observed effect sizes. Because only 1 study reported the outcomes of 2 groups undergoing PR with and without radiologist participation, models comparing effects of radiologist involvement across studies were not performed. Publication bias was assessed using funnel plots and the Egger regression test for plot symmetry. Statistical significance was defined using a 2-sided α of .05. All analyses were performed using the metafor package in R, version 4.4.1 (R Project for Statistical Computing), and RStudio, version 2024.09.0 + 375. The primary and secondary outcomes were estimated using a random-effects model, and heterogeneity was assessed using Cochran Q tests and the Higgins I 2 statistic. In cases of high heterogeneity ( I 2 > 50%), Cook distance was used to identify outliers, and pooled rates were recalculated after exclusion of outliers. Differences between subgroups were assessed using the omnibus test of moderators (Qm), testing the null hypothesis that all predictive factors entered into the model are unrelated to the observed effect sizes. Because only 1 study reported the outcomes of 2 groups undergoing PR with and without radiologist participation, models comparing effects of radiologist involvement across studies were not performed. Publication bias was assessed using funnel plots and the Egger regression test for plot symmetry. Statistical significance was defined using a 2-sided α of .05. All analyses were performed using the metafor package in R, version 4.4.1 (R Project for Statistical Computing), and RStudio, version 2024.09.0 + 375. Characteristics of Included Studies Systematic review identified 4185 studies for screening; 62 underwent full text review, and 31 with 39 509 cases of PR were included in the pooled analysis (eFigure 1 in ). , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , Characteristics of the included studies are presented in eTable 1 in . Studies were published between 2014 and 2024; 18 were prospective , , , , , , , , , , , , , , , , , and 13 were retrospective. , , , , , , , , , , , , By specific cancer diagnosis, PR of head and neck cancer (n = 11) , , , , , , , , , , and multiple diagnoses (n = 12) , , , , , , , , , , , were most common, followed by stereotactic body RT (SBRT) at various sites (n = 3), , , and lung cancer (n = 2) , ; the remaining 3 studies included patients with other diagnoses including breast, gynecologic, or hematologic cancers. Fifteen studies , , , , , , , , , , , , , , reported PR prior to initiation of treatment planning, 9 studies , , , , , , , , reviewed contours after planning, and 7 studies , , , , , , reported a mix of preplanning and postplanning PR. Twenty-four studies , , , , , , , , , , , , , , , , , , , , , , , , indicated the clinical importance of PR changes, while 7 studies , , , , , , did not. Clinical importance was reported as major vs minor changes (most common), clinically significant vs clinically insignificant (or noncritical), letter grades (ie, A [no changes], B [minor changes], and C [major changes]), or stoplight scoring (red indicating major changes and yellow indicating minor changes). The definitions of the severity of major and minor changes are shown in eTable 1 in . In terms of contours examined during PR, 30 studies reported reviewing target structures: 15 studies , , , , , , , , , , , reviewed all targets and OARs, 13 studies , , , , , , , , , , , , reviewed only targets, and 2 studies , reviewed only specific targets and OARs. Regarding the intent of RT for cases undergoing PR, pooled rate of curative intent was 93.7%; palliative intent, 6.3%; definitive intent, 62.0%; and adjuvant or neoadjuvant intent, 35.8% (eTable 2 in ). Median study quality (per the methodological index for nonrandomized studies) was 8 (range, 5-10) (eTable 3 in ). No evidence of publication bias was identified across studies (Egger test P = .56) (eFigure 2 in ). RT Plan Changes With and Without Radiology Involvement in PR A total of 6 studies (390 cases) , , , , , reported outcomes of PR in the presence of radiologists, while 25 studies (39 1196 cases) , , , , , , , , , , , , , , , , , , , , , , , , reported PR without radiologist involvement. The pooled rate of RT plan changes was significantly higher among studies that incorporated radiologists in PR than studies that did not at 49.4% (95% CI, 28.6%-70.1%) vs 25.0% (95% CI, 17.0%-33.1%), respectively ( P = .02) . In the 24 studies that reported major and minor changes, , , , , , , , , , , , , , , , , , , , , , , the involvement of radiology in PR was associated with an increased rate of major RT plan changes (47.0% [95% CI, 34.1%-59.8%] vs 10.2% [95% CI, 4.6%-15.8%]; P < .001) but no difference in minor changes (15.2% [95% CI, 9.7%-20.6%] vs 13.8% [95% CI, 9.3%-18.3%]; P = .74) . Given the high level of heterogeneity observed, these primary analyses were repeated after the removal of outliers. After removal of outlier studies, heterogeneity remained high in the group that did not include a radiologist, but the outcomes were not substantially different from those observed in the analysis of the total cohort for all changes, major changes, and minor changes. Subgroup analyses were performed on the studies reporting changes to specific contoured volumes because of PR, including GTV (12 studies; 1814 cases), , , , , , , , , , , PTV and treatment volumes (21 studies; 35 205 cases), , , , , , , , , , , , , , , , , , , , and OARs (12 studies; 31 023 cases). , , , , , , , , , , Radiologist-involved PR was associated with significant increases in rates of change to the GTV (41.0% [95% CI, 15.8%-66.2%] vs 4.1% [95% CI, 1.,2%-7.1%]; P < .001) and PTV and target volumes (45.6% [95% CI, 13.5%-77.8%] vs 18.5% [95% CI, 8.7%-28.3%]; P = .04) , but not to OARs (4.1% [95% CI, −4.5% to 12.7%] vs 4.3% [95% CI, −0.6% to 9.2%]; P = .98) (eFigure 3 in ). We also examined pooled rates according to the specific change outcome reported. Twenty-one studies , , , , , , , , , , , , , , , , , , reported RT plan changes (eFigure 4 in ), and 14 studies , , , , , , , , , , , , , reported recommendations for plan changes (eFigure 5 in ); the pooled rates of each were 32.3% (95% CI, 20.7%-43.8%) and 22.4% (95% CI, 16.3%-28.5%), respectively. Among all 31 studies, , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , the pooled rate of RT plan recommendations and changes was 29.0% (95% CI, 20.7%-37.2%) (eFigure 6 in ). Subgroup Analyses of Plan Changes Based on PR and Study Factors To better understand the association of the nature, frequency, and timing of PR with RT plan changes, subgroup analyses on the following factors were performed on all included studies: disease site and technique of RT plans being reviewed (all, head and neck, lung, SBRT of multiple sites, and other individual sites), frequency of PR (daily, multiple times per week, weekly, or as needed), and timing of PR relative to the RT planning process (before RT plan initiation or completion vs after or both). There were significant differences in RT plan changes between reports based on RT plan site and technique (eFigure 7 in ), with studies reporting head and neck PR observing the highest rates of changes (44.9%; 95% CI, 31.3%-58.4%), followed by other sites (38.5%; 95% CI, −16.9% to 93.9%), lung (24.5%; 95% CI, 10.3%-37.8%), SBRT (17.4%; 95% CI, 2.7%-32.0%), and all or multiple sites (15.9%; 95% CI, 11.4%-20.4%) ( P = .01). There were also significant differences in plan change rates based on the timing of PR (eFigure 8 in ): studies reporting PR that was performed before RT planning observed a significantly higher rate (43.5%; 95% CI, 30.1%-56.8%) than those reporting PR after planning initiation or completion (11.4%; 95% CI, 6.9%-15.9%) or both before and after planning (21.0%; 95% CI, 15.8%-26.1%) ( P < .001). There was no difference in change rates based on the frequency of PR (eFigure 9 in ). When comparing studies reporting prospective vs retrospective analyses of PR outcomes, a significantly higher rate of changes was observed in prospective studies (36.7%; 95% CI, 25.4%-48.1%) than in retrospective studies (16.6%; 95% CI, 9.1%-24.2%) ( P = .01) (eFigure 10 in ). An association between radiologist involvement and plan changes was only observed in prospective studies (eFigures 11 and 12 in ). Systematic review identified 4185 studies for screening; 62 underwent full text review, and 31 with 39 509 cases of PR were included in the pooled analysis (eFigure 1 in ). , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , Characteristics of the included studies are presented in eTable 1 in . Studies were published between 2014 and 2024; 18 were prospective , , , , , , , , , , , , , , , , , and 13 were retrospective. , , , , , , , , , , , , By specific cancer diagnosis, PR of head and neck cancer (n = 11) , , , , , , , , , , and multiple diagnoses (n = 12) , , , , , , , , , , , were most common, followed by stereotactic body RT (SBRT) at various sites (n = 3), , , and lung cancer (n = 2) , ; the remaining 3 studies included patients with other diagnoses including breast, gynecologic, or hematologic cancers. Fifteen studies , , , , , , , , , , , , , , reported PR prior to initiation of treatment planning, 9 studies , , , , , , , , reviewed contours after planning, and 7 studies , , , , , , reported a mix of preplanning and postplanning PR. Twenty-four studies , , , , , , , , , , , , , , , , , , , , , , , , indicated the clinical importance of PR changes, while 7 studies , , , , , , did not. Clinical importance was reported as major vs minor changes (most common), clinically significant vs clinically insignificant (or noncritical), letter grades (ie, A [no changes], B [minor changes], and C [major changes]), or stoplight scoring (red indicating major changes and yellow indicating minor changes). The definitions of the severity of major and minor changes are shown in eTable 1 in . In terms of contours examined during PR, 30 studies reported reviewing target structures: 15 studies , , , , , , , , , , , reviewed all targets and OARs, 13 studies , , , , , , , , , , , , reviewed only targets, and 2 studies , reviewed only specific targets and OARs. Regarding the intent of RT for cases undergoing PR, pooled rate of curative intent was 93.7%; palliative intent, 6.3%; definitive intent, 62.0%; and adjuvant or neoadjuvant intent, 35.8% (eTable 2 in ). Median study quality (per the methodological index for nonrandomized studies) was 8 (range, 5-10) (eTable 3 in ). No evidence of publication bias was identified across studies (Egger test P = .56) (eFigure 2 in ). A total of 6 studies (390 cases) , , , , , reported outcomes of PR in the presence of radiologists, while 25 studies (39 1196 cases) , , , , , , , , , , , , , , , , , , , , , , , , reported PR without radiologist involvement. The pooled rate of RT plan changes was significantly higher among studies that incorporated radiologists in PR than studies that did not at 49.4% (95% CI, 28.6%-70.1%) vs 25.0% (95% CI, 17.0%-33.1%), respectively ( P = .02) . In the 24 studies that reported major and minor changes, , , , , , , , , , , , , , , , , , , , , , , the involvement of radiology in PR was associated with an increased rate of major RT plan changes (47.0% [95% CI, 34.1%-59.8%] vs 10.2% [95% CI, 4.6%-15.8%]; P < .001) but no difference in minor changes (15.2% [95% CI, 9.7%-20.6%] vs 13.8% [95% CI, 9.3%-18.3%]; P = .74) . Given the high level of heterogeneity observed, these primary analyses were repeated after the removal of outliers. After removal of outlier studies, heterogeneity remained high in the group that did not include a radiologist, but the outcomes were not substantially different from those observed in the analysis of the total cohort for all changes, major changes, and minor changes. Subgroup analyses were performed on the studies reporting changes to specific contoured volumes because of PR, including GTV (12 studies; 1814 cases), , , , , , , , , , , PTV and treatment volumes (21 studies; 35 205 cases), , , , , , , , , , , , , , , , , , , , and OARs (12 studies; 31 023 cases). , , , , , , , , , , Radiologist-involved PR was associated with significant increases in rates of change to the GTV (41.0% [95% CI, 15.8%-66.2%] vs 4.1% [95% CI, 1.,2%-7.1%]; P < .001) and PTV and target volumes (45.6% [95% CI, 13.5%-77.8%] vs 18.5% [95% CI, 8.7%-28.3%]; P = .04) , but not to OARs (4.1% [95% CI, −4.5% to 12.7%] vs 4.3% [95% CI, −0.6% to 9.2%]; P = .98) (eFigure 3 in ). We also examined pooled rates according to the specific change outcome reported. Twenty-one studies , , , , , , , , , , , , , , , , , , reported RT plan changes (eFigure 4 in ), and 14 studies , , , , , , , , , , , , , reported recommendations for plan changes (eFigure 5 in ); the pooled rates of each were 32.3% (95% CI, 20.7%-43.8%) and 22.4% (95% CI, 16.3%-28.5%), respectively. Among all 31 studies, , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , the pooled rate of RT plan recommendations and changes was 29.0% (95% CI, 20.7%-37.2%) (eFigure 6 in ). To better understand the association of the nature, frequency, and timing of PR with RT plan changes, subgroup analyses on the following factors were performed on all included studies: disease site and technique of RT plans being reviewed (all, head and neck, lung, SBRT of multiple sites, and other individual sites), frequency of PR (daily, multiple times per week, weekly, or as needed), and timing of PR relative to the RT planning process (before RT plan initiation or completion vs after or both). There were significant differences in RT plan changes between reports based on RT plan site and technique (eFigure 7 in ), with studies reporting head and neck PR observing the highest rates of changes (44.9%; 95% CI, 31.3%-58.4%), followed by other sites (38.5%; 95% CI, −16.9% to 93.9%), lung (24.5%; 95% CI, 10.3%-37.8%), SBRT (17.4%; 95% CI, 2.7%-32.0%), and all or multiple sites (15.9%; 95% CI, 11.4%-20.4%) ( P = .01). There were also significant differences in plan change rates based on the timing of PR (eFigure 8 in ): studies reporting PR that was performed before RT planning observed a significantly higher rate (43.5%; 95% CI, 30.1%-56.8%) than those reporting PR after planning initiation or completion (11.4%; 95% CI, 6.9%-15.9%) or both before and after planning (21.0%; 95% CI, 15.8%-26.1%) ( P < .001). There was no difference in change rates based on the frequency of PR (eFigure 9 in ). When comparing studies reporting prospective vs retrospective analyses of PR outcomes, a significantly higher rate of changes was observed in prospective studies (36.7%; 95% CI, 25.4%-48.1%) than in retrospective studies (16.6%; 95% CI, 9.1%-24.2%) ( P = .01) (eFigure 10 in ). An association between radiologist involvement and plan changes was only observed in prospective studies (eFigures 11 and 12 in ). Curative RT is predicated on accurate, millimeter-precise treatment planning and delivery. Interdisciplinary collaboration between radiation oncologists and diagnostic radiologists may further refine this highly technical process, optimize RT targeting, and improve cancer treatment outcomes. In this systematic review and meta-analysis, we found a 36.8% higher absolute rate of clinically significant changes to the RT plan associated with radiologist inclusion in radiation oncology peer review of contoured volumes. Importantly, we did not observe a significant difference in minor changes, seemingly indicating that radiologist collaboration in PR meaningfully impacts RT plans without added change burden for minor issues. These findings are additionally important because PR is required for radiation oncology practice accreditation, and prior studies have shown RT plan quality directly affects cancer recurrence, progression, and survival. , While multiple studies have been published on PR with or without radiology involvement, few have compared PR outcomes between the 2 approaches. Only one of these experiences involved a prospective clinical trial that specifically investigated radiation oncology–radiology collaboration as a treatment planning intervention for patients with locally advanced lung cancer: addition of a radiologist to the PR process resulted in changes in the tumor volume in over one-third of patients. Among patients with head and neck cancer, Gatfield et al prospectively compared PR outcomes with vs without radiology involvement and found that the presence of a neuroradiologist increased the rate of changes to delineated structures by two-thirds (from 33% to 55%). While changes to RT targets or OARs may increase with additional layers of PR, it is yet to be determined whether these changes result in different clinical outcomes. One study reporting head and neck RT PR identified a mean volumetric change in target volumes of 20% to 22%, with the upper range being greater than 100% to greater than 275%. These quantities represent substantial revisions to the RT targets and, in the era of highly conformal RT techniques, could be expected to impact the risk of locoregional failure. The benefits of radiology involvement in PR of radiation oncologist–delineated contours must outweigh the potential barriers to routine implementation of this practice. The main potential benefit apparent from our meta-analysis is clinically meaningful improvements to the RT plan. However, barriers include time limitations, technological considerations, and lack of a current reimbursement model for this additional effort. Radiation oncologists may worry that radiologist participation in PR could elicit additional time-consuming but less clinically meaningful minor changes, though the findings from our meta-analysis do not support an increase in minor change rates. Time constraints on both sides of the interdisciplinary spectrum must be considered: attending PR conferences or soliciting on-demand PR during busy clinical services is a major barrier. The most efficient PR method may vary by institution; the logistics of radiology–radiation oncology PR may include in-person or virtual conferences, telephone conversations, asynchronous review of images via the electronic health record, or other methods of communication. Novel methods to facilitate interdisciplinary review of RT targets that improve the ease of use for the reviewing radiologist are needed. Farris et al developed a novel script-based framework for transferring the contoured images from the RT planning computed tomographic image set into an anonymized portable document format (PDF) that can then be shared with the reviewing radiologist. Development of automated systems integrated within the institutional picture archiving and communication systems may increase efficiency, improve implementation, and inform follow-up imaging interpretation. , However, without a reimbursement mechanism, radiologists’ time spent reviewing radiation oncologists’ targets ultimately detracts from their diagnostic productivity, which may require up to 30 to 60 minutes per case review, depending on the complexity. Considering the high cost of salvage or palliative therapies (such as immunotherapy) for patients with recurrence after RT and the high rates of clinically significant changes to the RT plan identified in our study, preventing failure through radiologist PR could represent a cost-effective intervention. Prospective evidence of improved oncologic outcomes would be necessary to support the development of professional reimbursement mechanisms or procedure codes like those used in other RT planning procedures, such as radiosurgery. Our findings also provide practical insight into the efficient and cost-effective use of multidisciplinary resources for collaborative PR of RT targets. Subgroup analyses indicate that radiology incorporation into PR of head and neck and lung cancer treatment plans yields consistently high rates of change. We also observed significantly higher change rates to the RT plan when PR is performed before compared with after the RT plan has already been finalized. Since there was no difference in the rate of RT plan changes based on frequency of PR, the most flexible and feasible schedule that maximizes interdisciplinary participation can be used without the risk of affecting PR outcomes. A streamlined process for radiation oncology–radiology collaboration may also increase the efficiency and quality of image interpretation after RT, as the details of RT location, distribution, and timing are often not readily available in the medical record. Limitations This study is limited by the study-level nature of the meta-analysis, the lack of studies reporting PR outcomes with and without radiologist involvement, the imbalance in total number of cases between groups, and the high heterogeneity of the included studies. Due to the small number of cases with radiology-based PR, these findings should be interpreted with caution. The fact that 0.6% of all identified cases (1.2% when excluding the 20 069 cases reported in the largest nonradiology study by Tchelebi et al ) reported radiology-based PR highlights the critical need for additional data in this space. Despite the ubiquity of PR in radiation oncology and routine collaboration between radiation oncology and radiology in the course of clinical care, limited data exist to measure the clinical impact of radiologist–radiation oncologist collaboration in PR of contoured targets. Additional prospective investigations including multiple disease sites are necessary to better understand the clinical impact of incorporating radiologists in radiation oncology PR in an unbiased, unselected sample. While there are differences in sample size between groups, meta-analytic techniques, such as the random-effects models reported herein, are expected to minimize confounding. Regarding heterogeneity, while removal of outlier studies in the primary analyses of all changes, major changes, and minor changes reduced heterogeneity to a reasonable level ( I 2 < 50%) in the radiologist group, heterogeneity remained high in the group without a radiologist. Since PR in radiation oncology is highly variable by nature, this heterogeneity is expected and suggests generalizability across multiple PR methods, timing, and formats. Additionally, removal of outliers did not substantially change the findings of these analyses. Because only 1 study reported outcomes of PR with and without radiologist involvement, comparisons within studies were not possible, so the findings of these comparisons between studies are subject to confounding. The inclusion of retrospective and prospective studies may also introduce bias, as a higher rate of changes was observed in prospective studies, as would be expected in quality improvement studies aimed at improving RT plans through prospective assessment of PR. Additional prospective study of this interdisciplinary collaboration can investigate the impact of radiology involvement directly on clinical outcomes, such as local disease control and survival. This study is limited by the study-level nature of the meta-analysis, the lack of studies reporting PR outcomes with and without radiologist involvement, the imbalance in total number of cases between groups, and the high heterogeneity of the included studies. Due to the small number of cases with radiology-based PR, these findings should be interpreted with caution. The fact that 0.6% of all identified cases (1.2% when excluding the 20 069 cases reported in the largest nonradiology study by Tchelebi et al ) reported radiology-based PR highlights the critical need for additional data in this space. Despite the ubiquity of PR in radiation oncology and routine collaboration between radiation oncology and radiology in the course of clinical care, limited data exist to measure the clinical impact of radiologist–radiation oncologist collaboration in PR of contoured targets. Additional prospective investigations including multiple disease sites are necessary to better understand the clinical impact of incorporating radiologists in radiation oncology PR in an unbiased, unselected sample. While there are differences in sample size between groups, meta-analytic techniques, such as the random-effects models reported herein, are expected to minimize confounding. Regarding heterogeneity, while removal of outlier studies in the primary analyses of all changes, major changes, and minor changes reduced heterogeneity to a reasonable level ( I 2 < 50%) in the radiologist group, heterogeneity remained high in the group without a radiologist. Since PR in radiation oncology is highly variable by nature, this heterogeneity is expected and suggests generalizability across multiple PR methods, timing, and formats. Additionally, removal of outliers did not substantially change the findings of these analyses. Because only 1 study reported outcomes of PR with and without radiologist involvement, comparisons within studies were not possible, so the findings of these comparisons between studies are subject to confounding. The inclusion of retrospective and prospective studies may also introduce bias, as a higher rate of changes was observed in prospective studies, as would be expected in quality improvement studies aimed at improving RT plans through prospective assessment of PR. Additional prospective study of this interdisciplinary collaboration can investigate the impact of radiology involvement directly on clinical outcomes, such as local disease control and survival. In this systematic review and meta-analysis of radiation oncology PR of contoured targets, radiologist involvement was associated with higher rates of clinically relevant changes to the RT target and total changes to the RT plan. Radiologist involvement was not associated with rates of minor RT plan changes. These results support the value of interdisciplinary collaboration with radiology during RT planning. Further efforts to maximize the efficiency and cost-effectiveness of closer interdisciplinary collaboration between these 2 specialties would be expected to increase the quality of RT and improve cancer outcomes, which warrants evaluation in prospective clinical trials. |
Comparison of the Efficacy of Artificial Intelligence-Powered Software in Crown Design: An In Vitro Study | 5df8ad61-9b0b-4fea-8a63-d8987c6a4b13 | 11806298 | Dentistry[mh] | In the 1970s, computer-aided design and computer-aided manufacturing (CAD/CAM) technology was introduced to dentistry and was first used in the customization of dental restorations. The process included digital scanning, designing restorations using restoration design software, and exporting them to a CAM compatible file format compatible with CAM for further processing, finishing, and polishing. This significantly improved the efficiency and quality of dental restoration customization. Multiple dental morphology databases were commonly utilized in CAD restoration software. Dental technicians need to select and modify data based on their experience and the patient's specific needs. This process was not only technology-sensitive but also time-consuming. Poorly designed restorations not only increase the chairside adjustment time but also decrease the success rate of restorations, thereby affecting the overall restoration effectiveness, efficiency, and patient satisfaction. In recent years, artificial intelligence (AI) technology has made significant progress in various fields of dentistry, such as dental image diagnostics, , automated landmark localization of cephalometry in orthodontics, , the diagnosis of periodontitis, caries detection, and diagnosis in endodontics. In 2017, Raith et al. integrated AI with CAD/CAM to improve its chairside application in prosthetic dentistry. The introduction of AI models and the development of AI restoration design software have provided a new impetus for digital dental restorations. , , , , AI-powered design systems leverage techniques such as convolutional neural networks and generative adversarial networks (GANs) to intelligently generate simulated tooth images through deep learning in a large number of clinical restoration design cases. , AI-powered design systems not only enhanced the efficiency of restoration design but also reduced the potential for significant deviations caused by human factors. , , Currently, the performance evaluation of AI-powered crown design software primarily focuses on design accuracy and efficiency. However, there is comparatively less emphasis on their performance assessment, which warrants further exploration and integration into comprehensive evaluations. Consequently, this study aimed to evaluate the performance of two AI-powered crown design software in comparison with the conventional CAD software. The null hypothesis was that there is no difference between AI-powered and computer-aided software in terms of time efficiency and morphological accuracy in crown design. Preparation of test dataset The study utilized real intraoral scan datasets from patients requiring full crown restoration, obtained from the Department of General Dentistry at the Stomatology Hospital of Zhejiang Chinese Medical University. The data covered the period from July 2023 to January 2024. The datasets included partial arch scans of both the maxillary and mandibular arches obtained using an intraoral scanner (TRIOS 3.0; 3Shape). These scans encompassed a prepared abutment tooth in the posterior region, as well as jaw relation scan data. The inclusion criteria for the scan datasets in this analysis were as follows: (1) natural teeth used as abutments; (2) presence of clear margin lines; (3) existence of opposing teeth and at least one approximal tooth; (4) absence of severe malocclusion; and (5) absence of missing areas and irrelevant soft tissue in the abutment scans. A pilot study was conducted with a sample size of 15 test datasets to assess feasibility and guide the sample size calculations. The key outcome measures, such as time efficiency and occlusal surfaces, were analysed for the mean and standard (Std) deviation. It was determined by a sample size calculation software (PASS, CA, USA) that a minimum of 32 samples would be required to ensure a statistically significant test at a significance level of 0.05 and a power of 0.8. As a result, 33 sample datasets were included in the study. All data were stored in Std triangulation language (STL) format files. These STL files underwent thorough anonymization and deidentification to protect patients’ privacy. This study was approved by the Ethics Committee of Zhejiang Chinese Medical University Stomatological Hospital, and the data were collected after obtaining written informed consent from the patients. Establishing the Std of crown design Before conducting the design experiments, the Std group was established. A dental technician with over 8 years of full-time experience in chairside crown restoration design imported and designed the crown morphology of the study abutment teeth using conventional computer-aided software (Exocad DentalCAD; Exocad GmbH). The crown design data for the Std group was validated through chairside interim crown adjustment using diamond burs and intraoral try-in procedures, followed by intraoral scanning after confirming occlusion ( A). All data were exported in STL file format. The crown morphology of the Std group was used as a reference for designing crowns. Moreover, it exhibits several noteworthy characteristics: (1) Occlusal surface: In intercuspal position, patients were instructed to bite on a 100 μm blue articulating paper; during laterotrusive jaw movement, another bite was taken on the same paper; upon returning to intercuspal position, they bit on a 40 μm red articulating paper, showing complete overlap of blue and red imprints on the restoration's occlusal surface, indicating no laterotrusive interference ( A). During protrusive jaw movement, a similar articulation process was followed, demonstrating no protrusive interference ( B). The occlusal contacts were extensive and tight, with no premature contact ( C). (2) Adjacent surfaces: The shim stock (8 μm thick, Almore) was pulled with contact being felt, but no binding ( D,E), indicating ideal interproximal contacts. (3) Margin line: The marginal fit of the restoration on the die used is 25 μm, meeting the acceptable accuracy of less than 120 μm. Methods of the crown design There were three types of software used for crown design. The computer-aided Exocad DentalCAD software was used by dental technicians with varying levels of experience for crown design: a technician with over 5 years of experience (computer-aided experienced [CE] group) and a technician with 1 year of experience (computer-aided novice [CN] group). The technicians imported the initial crown morphology from the built-in database of the system and manually adjusted the design based on their own experience. Crown try-in and rescanning were not performed to generate the final crown morphology data. The AI-powered design method utilized the commercial standalone version of dental design software (Automate; 3Shape), identified as the AI Automate (AA) group, and the commercial network version of dental design software (Dentbird Crown; Imagoworks), identified as the AI Dentbird Crown (AD) group. The latter openly adopted algorithms based on GANs and convolutional neural networks. They eliminated the need for a technician's involvement in design revisions or subsequent optimizations. Evaluation of design outcome The time of the entire design process (T) was captured in real-time screen recording software (OBS Studio, Open Broadcaster Software) to evaluate the efficiency of four experimental groups on crown design. The morphological accuracy of the crowns was compared by measuring the root-mean-square (RMS) in the morphology of the occlusal, mesial, and distal surfaces, as well as the margin lines of the crown. The Std group was used as a reference . By measuring the deviation in the occlusal surface morphology during the design process for crowns, the differences between the results with and without AI assistance were compared. The outer surface of each crown designed by the experimental groups (CE, CN, AA, and AD) was aligned with the Std group using an automatic best-fit algorithm ( B), followed by a three-dimensional (3D) comparison. Differences in outer crown design data were measured using Euclidean distance and RMS error calculations. An alignment and measurements were conducted using reverse engineering software (Geomagic Wrap; 3D Systems) with a coordinated coordinate system to ensure synchronization and consistency between datasets. The RMS was computed within the same area using the following formula: RMS = ∑ i = 1 n ( X 1 , i − X 2 , i ) 2 n where X 1 , i is the reference data, X 2 , i is the scan data, and n denotes the total number of measurement points assessed in each analysis. A colour-coded map illustrating the discrepancies was generated, ranging from red (+500 μm) to blue (-500 μm), with deviations within the tolerance level of ±50 μm shown in green. The 3D analysis software exports the root mean square RMS values of the spatial deviations for statistical purposes. The arithmetic median of the RMS values was utilized to gauge the accuracy of tooth morphology, whereas the interquartile range (IQR) was employed to depict the dispersion of data distribution for morphological accuracy designed by AI-powered and computer-aided software. Statistics Statistical software (IBM SPSS Statistics v25.0; IBM) was used to analyse the data. The normality of the measured values was evaluated using the Shapiro–Wilk test. The Kruskal–Wallis test was used to analyse the positively skewed data ( α = 0.05). The study utilized real intraoral scan datasets from patients requiring full crown restoration, obtained from the Department of General Dentistry at the Stomatology Hospital of Zhejiang Chinese Medical University. The data covered the period from July 2023 to January 2024. The datasets included partial arch scans of both the maxillary and mandibular arches obtained using an intraoral scanner (TRIOS 3.0; 3Shape). These scans encompassed a prepared abutment tooth in the posterior region, as well as jaw relation scan data. The inclusion criteria for the scan datasets in this analysis were as follows: (1) natural teeth used as abutments; (2) presence of clear margin lines; (3) existence of opposing teeth and at least one approximal tooth; (4) absence of severe malocclusion; and (5) absence of missing areas and irrelevant soft tissue in the abutment scans. A pilot study was conducted with a sample size of 15 test datasets to assess feasibility and guide the sample size calculations. The key outcome measures, such as time efficiency and occlusal surfaces, were analysed for the mean and standard (Std) deviation. It was determined by a sample size calculation software (PASS, CA, USA) that a minimum of 32 samples would be required to ensure a statistically significant test at a significance level of 0.05 and a power of 0.8. As a result, 33 sample datasets were included in the study. All data were stored in Std triangulation language (STL) format files. These STL files underwent thorough anonymization and deidentification to protect patients’ privacy. This study was approved by the Ethics Committee of Zhejiang Chinese Medical University Stomatological Hospital, and the data were collected after obtaining written informed consent from the patients. Before conducting the design experiments, the Std group was established. A dental technician with over 8 years of full-time experience in chairside crown restoration design imported and designed the crown morphology of the study abutment teeth using conventional computer-aided software (Exocad DentalCAD; Exocad GmbH). The crown design data for the Std group was validated through chairside interim crown adjustment using diamond burs and intraoral try-in procedures, followed by intraoral scanning after confirming occlusion ( A). All data were exported in STL file format. The crown morphology of the Std group was used as a reference for designing crowns. Moreover, it exhibits several noteworthy characteristics: (1) Occlusal surface: In intercuspal position, patients were instructed to bite on a 100 μm blue articulating paper; during laterotrusive jaw movement, another bite was taken on the same paper; upon returning to intercuspal position, they bit on a 40 μm red articulating paper, showing complete overlap of blue and red imprints on the restoration's occlusal surface, indicating no laterotrusive interference ( A). During protrusive jaw movement, a similar articulation process was followed, demonstrating no protrusive interference ( B). The occlusal contacts were extensive and tight, with no premature contact ( C). (2) Adjacent surfaces: The shim stock (8 μm thick, Almore) was pulled with contact being felt, but no binding ( D,E), indicating ideal interproximal contacts. (3) Margin line: The marginal fit of the restoration on the die used is 25 μm, meeting the acceptable accuracy of less than 120 μm. There were three types of software used for crown design. The computer-aided Exocad DentalCAD software was used by dental technicians with varying levels of experience for crown design: a technician with over 5 years of experience (computer-aided experienced [CE] group) and a technician with 1 year of experience (computer-aided novice [CN] group). The technicians imported the initial crown morphology from the built-in database of the system and manually adjusted the design based on their own experience. Crown try-in and rescanning were not performed to generate the final crown morphology data. The AI-powered design method utilized the commercial standalone version of dental design software (Automate; 3Shape), identified as the AI Automate (AA) group, and the commercial network version of dental design software (Dentbird Crown; Imagoworks), identified as the AI Dentbird Crown (AD) group. The latter openly adopted algorithms based on GANs and convolutional neural networks. They eliminated the need for a technician's involvement in design revisions or subsequent optimizations. The time of the entire design process (T) was captured in real-time screen recording software (OBS Studio, Open Broadcaster Software) to evaluate the efficiency of four experimental groups on crown design. The morphological accuracy of the crowns was compared by measuring the root-mean-square (RMS) in the morphology of the occlusal, mesial, and distal surfaces, as well as the margin lines of the crown. The Std group was used as a reference . By measuring the deviation in the occlusal surface morphology during the design process for crowns, the differences between the results with and without AI assistance were compared. The outer surface of each crown designed by the experimental groups (CE, CN, AA, and AD) was aligned with the Std group using an automatic best-fit algorithm ( B), followed by a three-dimensional (3D) comparison. Differences in outer crown design data were measured using Euclidean distance and RMS error calculations. An alignment and measurements were conducted using reverse engineering software (Geomagic Wrap; 3D Systems) with a coordinated coordinate system to ensure synchronization and consistency between datasets. The RMS was computed within the same area using the following formula: RMS = ∑ i = 1 n ( X 1 , i − X 2 , i ) 2 n where X 1 , i is the reference data, X 2 , i is the scan data, and n denotes the total number of measurement points assessed in each analysis. A colour-coded map illustrating the discrepancies was generated, ranging from red (+500 μm) to blue (-500 μm), with deviations within the tolerance level of ±50 μm shown in green. The 3D analysis software exports the root mean square RMS values of the spatial deviations for statistical purposes. The arithmetic median of the RMS values was utilized to gauge the accuracy of tooth morphology, whereas the interquartile range (IQR) was employed to depict the dispersion of data distribution for morphological accuracy designed by AI-powered and computer-aided software. Statistical software (IBM SPSS Statistics v25.0; IBM) was used to analyse the data. The normality of the measured values was evaluated using the Shapiro–Wilk test. The Kruskal–Wallis test was used to analyse the positively skewed data ( α = 0.05). Regarding time efficiency, the medians and IQRs for each group are summarized ( and A). Crown working times varied significantly across the experimental groups ( P < .001). Pairwise comparisons revealed statistically significant differences between the CE and CN, CE and AA, CE and AD, CN and AA, and CN and AD pairs. No significant differences were observed between the AA and AD pair. The rankings indicated that the CN group significantly surpassed the other groups, with the CE group ranking second highest, followed by the AA and AD groups tied for third place. In addition, the working time for both AA and AD groups was only one-quarter of that for the CN group. Collectively, compared with computer-aided software, the AI-powered software yielded a substantial reduction in crown production duration. In terms of the occlusal, mesial, and distal surfaces, and the margin lines, the median (IQR) of RMS values (µm) were presented ( and B). The 3D colour maps illustrate the differences in the morphology of the crowns designed using distinct methods . For the occlusal surfaces, there were statistically significant differences in the RMS among the four groups ( P < .001) . Pairwise comparisons demonstrated statistically significant differences between the CE-Std and AD-Std, CN-Std and AD-Std, and AA-Std and AD-Std pairs. No significant differences were observed between the CE-Std and CN-Std, CE-Std and AA-Std, and CN-Std and AA-Std pairs. This ranking demonstrated that the AD group performed less accurately on the occlusal surfaces than the other three groups. In terms of the distal surfaces, there were statistically significant differences in the RMS among the four groups ( P = .034) . Pairwise comparisons revealed statistically significant differences between CE-Std and AA-Std. No significant differences were observed between the CE-Std and CN-Std pair, CE-Std and AD-Std, CN-Std and AA-Std, CN-Std and AD-Std, and AA-Std and AD-Std pairs. The ranking demonstrated that CE outperformed AA in terms of accuracy on the distal surfaces. In terms of the mesial surfaces ( P = .791) and the margin lines ( P = .393), there were no statistically significant differences in the RMS among the groups presented , indicating that there were no significant differences in the accuracy of the crown design for the mesial surfaces and the margin lines among the different software groups. Based on the results of this study, the hypotheses regarding time efficiency, occlusal surfaces, and distal surfaces were rejected, while the hypotheses regarding the mesial surfaces and margin lines hold true. AI has the potential to improve the time efficiency of crown design, consistent with previous research findings. , , On the occlusal surfaces, the AD group exhibited the lowest accuracy compared with the other three groups. This may be due to Dentbird's algorithm inaccurately accounting for occlusal factors in the upper and lower jaws, resulting in poor accuracy of the occlusal surfaces. The CE group demonstrated superior accuracy on distal surfaces compared to the AA group. This difference may stem from Automate's algorithm limitations in handling distal surfaces of posterior second molars that lack adjacent tooth contact, restricting its algorithmic flexibility. This indicated that the effectiveness of AI may be limited when confronted with individual differences among patients, complex dental anatomical structures, and specific treatment requirements. The morphological accuracy designed by the CE group was relatively precise, significantly outperforming the CN group regarding time efficiency. This efficiency and precision in crown fabrication were attributed to the experienced technicians’ extensive expertise and familiarity with the fabrication process. AI requires further research and deep learning to enhance its capabilities in crown design, enabling better application in clinical settings and providing patients with superior crown designs, improved comfort, and suitability. Building on AI's ability to enhance design efficiency, experienced technicians can improve the accuracy and quality of crown designs through design revisions or subsequent optimization. This offers more opportunities in clinical practice and fosters advancements in dentistry. For the first time in this study, an adjusted dataset (occlusion, adjacency) was introduced as the Std group, which also provided insights for subsequent research based on AI data: The AI-powered model was trained on a broad dataset that included natural and technician-designed teeth. This phase helped the model learn various dental characteristics. An iterative feedback loop was established, in which clinicians continuously assessed and refined AI-powered restorations (eg, a dataset that had been adjusted (occlusion, adjacency) was introduced and retrained), ensuring that the model evolved and aligned better with clinical needs. This study has several limitations. First, all design methodologies were evaluated exclusively for posterior crowns on abutments that exhibited clinically favourable conditions. Second, there are no quantifiable metrics in clinical practice to assess the buccal and lingual axial curvatures of crowns. Therefore, the adequacy of the design for the buccal and lingual surfaces cannot be verified. Incorporating these factors into future research is imperative for a more comprehensive assessment. Moreover, GANs and Transformers have been proposed for automatic occlusal surface design to achieve certain results. However, even many recent studies still narrowly focus on accuracy comparisons. The utilization of mechanical simulation methods, such as finite element analysis (FEA), can simulate and evaluate the functional performance of various dental crown designs in the oral cavity. Currently, Ding et al.’s research compares different crown designs using FEA. FEA is essential for validating the functionality of these design approaches before milling, serving as a promising avenue for future evaluation methods. Our research discovered that AI-powered software has greatly enhanced the efficiency of crown design. However, it has encountered challenges in achieving the same level of morphological accuracy as experienced technicians using computer-aided software. Improving the morphological accuracy of crown designs with AI-powered software requires extensive deep-learning efforts. This pursuit aims to refine the precision and stability of the designs. As digital dentistry evolves, AI is positioned to play a pivotal role, providing heightened precision and efficiency across various dental procedures, ultimately advancing patient care and treatment outcomes. The study was conducted according to the guidelines of the Declaration of Helsinki, and approved by the Ethics Committee of Zhejiang Chinese Medical University Stomatological Hospital (protocol code: ZCMUHSIRB-2024040704 and approval date: April 12). Informed consent was obtained from all subjects involved in the study. Written informed consent has been obtained from the patients to publish this article. Z.W., C.Z., X.Y., and Y.D.: conception and design, analysis and interpretation of the data, drafting of the article, critical revision of the manuscript for important intellectual content, and final approval of the version to be published. Z.J., W.Z, and N.Z.: conception and design, interpretation of the data, critical revision of the manuscript for important intellectual content, and final approval of the version to be published. All authors have read and agreed to the published version of the manuscript. The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this article. |
Ipsilateral Breast Carcinoma Recurrence | fbe1b5cb-8fc4-4bb5-918a-c8e8e1996326 | 11834960 | Anatomy[mh] | Case Selection Cases were selected from the Pathology Department of Ramón y Cajal University Hospital (Madrid, Spain). Initially, an attempt was made to analyze a consecutive series of 85 patients with recurrences. However, 61% of patients with recurrences were excluded due to lack of tissue for further molecular analysis in any of the 2 carcinomas, poor quality of extracted DNA from old blocks, and missed blocks that were previously sent to clinical trials and had not been returned. Clinical data were obtained from clinical records. Only local recurrences with involvement of either the breast, skin, the nipple-areolar-complex, or the chest wall were considered for analysis. No cases of lymph node recurrences were analyzed. Histologic typing was performed according to WHO recommendations and cases were graded according to the 3-tiered Nottingham histologic grading system. Approval for the study was obtained from the Local Ethics Committee (Ramón y Cajal Research Ethics Committee reference 223/18). Immunohistochemistry All cases underwent a new immunohistochemical study. The antibodies and FISH probes used are shown in Supplemental Table S1, Supplemental Digital Content 1, http://links.lww.com/PAS/B998 . Immunostaining was performed using the EnVision detection system (K5007; Dako, Glostrup, Denmark). A cut-off value of 1% was used to define ER and progesterone receptor (PR) positivity. HER2 expression was interpreted according to the 2018 American Society of Clinical Oncology and the College of American Pathologists (ASCO-CAP) guidelines. For diagnosis, HER2 equivocal cases (2+) underwent FISH and the results were interpreted according to 2018 ASCO-CAP guidelines. In addition, the rest of the primaries and recurrences were subjected to HER2 FISH analysis using tissue microarray sections (see below). Tumors expressing ER and/or PR were classified as luminal (LUM). Luminal carcinomas that were also HER2 positive were classified as luminal HER2 (LUM-HER2). ER and PR negative carcinomas that were HER2 positive were classified as HER2 enriched (HER2), and those negative for ER, PR, and HER2 as triple negative (TN). Massive Parallel Sequencing A custom gene panel was designed using the SureDesign platform by Agilent Tech. (Santa Clara, CA) to consistently target 38 genes that are frequently mutated in breast cancer (Supplemental Table 2, Supplemental Digital Content 2, http://links.lww.com/PAS/B999 ). , DNA extraction and quantitation, library construction, and bioinformatics analysis were performed as previously reported. , All pathogenic mutations in recurrences were visually inspected in primary carcinomas to identify shared pathogenic mutations that were at VAFs under filtering criteria. Fluorescent In Situ Hybridization Since our panel was not able to accurately detect gene copy number variations, we used FISH to evaluate ERBB2 , CCND1 , MYC , FGFR1 , and MDM4 , which are the most frequently amplified genes in BC, as previously reported. , FISH probes are specified in Supplemental Table 1, Supplemental Digital Content 1, http://links.lww.com/PAS/B998 . FISH was performed on sections of a tissue microarray constructed with at least 2 representative cores of all primary carcinomas and their recurrences. Scoring of at least 20 neoplastic cells per sample was performed. A carcinoma was considered amplified if the ratio between gene signals and centromere signals was ≥2, and polysomy when the average of centromere signals on tumor cells was>3. Recurrence Classification From a molecular point of view, a recurrence that shared at least one pathogenic mutation or amplification in at least one known driver gene related with breast carcinogenesis was considered a true recurrence. Recurrences that did not share any pathogenic mutation or amplification in at least one known driver gene for breast carcinogenesis were considered a new primary. Cases in which we did not detect any relevant molecular alteration in the primary carcinoma were considered undetermined. We also classified all recurrences using the different clinicopathologic criteria proposed by Huang, Komoide, Panet-Raymond, Yi, and Jobsen (Twente and morphologic methods). – Cases were selected from the Pathology Department of Ramón y Cajal University Hospital (Madrid, Spain). Initially, an attempt was made to analyze a consecutive series of 85 patients with recurrences. However, 61% of patients with recurrences were excluded due to lack of tissue for further molecular analysis in any of the 2 carcinomas, poor quality of extracted DNA from old blocks, and missed blocks that were previously sent to clinical trials and had not been returned. Clinical data were obtained from clinical records. Only local recurrences with involvement of either the breast, skin, the nipple-areolar-complex, or the chest wall were considered for analysis. No cases of lymph node recurrences were analyzed. Histologic typing was performed according to WHO recommendations and cases were graded according to the 3-tiered Nottingham histologic grading system. Approval for the study was obtained from the Local Ethics Committee (Ramón y Cajal Research Ethics Committee reference 223/18). All cases underwent a new immunohistochemical study. The antibodies and FISH probes used are shown in Supplemental Table S1, Supplemental Digital Content 1, http://links.lww.com/PAS/B998 . Immunostaining was performed using the EnVision detection system (K5007; Dako, Glostrup, Denmark). A cut-off value of 1% was used to define ER and progesterone receptor (PR) positivity. HER2 expression was interpreted according to the 2018 American Society of Clinical Oncology and the College of American Pathologists (ASCO-CAP) guidelines. For diagnosis, HER2 equivocal cases (2+) underwent FISH and the results were interpreted according to 2018 ASCO-CAP guidelines. In addition, the rest of the primaries and recurrences were subjected to HER2 FISH analysis using tissue microarray sections (see below). Tumors expressing ER and/or PR were classified as luminal (LUM). Luminal carcinomas that were also HER2 positive were classified as luminal HER2 (LUM-HER2). ER and PR negative carcinomas that were HER2 positive were classified as HER2 enriched (HER2), and those negative for ER, PR, and HER2 as triple negative (TN). A custom gene panel was designed using the SureDesign platform by Agilent Tech. (Santa Clara, CA) to consistently target 38 genes that are frequently mutated in breast cancer (Supplemental Table 2, Supplemental Digital Content 2, http://links.lww.com/PAS/B999 ). , DNA extraction and quantitation, library construction, and bioinformatics analysis were performed as previously reported. , All pathogenic mutations in recurrences were visually inspected in primary carcinomas to identify shared pathogenic mutations that were at VAFs under filtering criteria. Since our panel was not able to accurately detect gene copy number variations, we used FISH to evaluate ERBB2 , CCND1 , MYC , FGFR1 , and MDM4 , which are the most frequently amplified genes in BC, as previously reported. , FISH probes are specified in Supplemental Table 1, Supplemental Digital Content 1, http://links.lww.com/PAS/B998 . FISH was performed on sections of a tissue microarray constructed with at least 2 representative cores of all primary carcinomas and their recurrences. Scoring of at least 20 neoplastic cells per sample was performed. A carcinoma was considered amplified if the ratio between gene signals and centromere signals was ≥2, and polysomy when the average of centromere signals on tumor cells was>3. From a molecular point of view, a recurrence that shared at least one pathogenic mutation or amplification in at least one known driver gene related with breast carcinogenesis was considered a true recurrence. Recurrences that did not share any pathogenic mutation or amplification in at least one known driver gene for breast carcinogenesis were considered a new primary. Cases in which we did not detect any relevant molecular alteration in the primary carcinoma were considered undetermined. We also classified all recurrences using the different clinicopathologic criteria proposed by Huang, Komoide, Panet-Raymond, Yi, and Jobsen (Twente and morphologic methods). – Molecular studies were performed on 35 primary carcinomas and 35 recurrences from 33 patients (70 carcinomas in total). One patient (ID 6) had a hereditary breast cancer syndrome due to an ATM germline mutation (R2993Y mutation) and developed 2 bilateral primary carcinomas and their corresponding recurrences. Another patient (ID 32) carried a CDH1 germline mutation (c.1137G>A (splice)) and developed 2 bilateral primary carcinomas and their corresponding recurrences. After a careful analysis of clinicopathologic data and the results of genetic testing in some cases, no other patients were considered to carry a hereditary breast cancer syndrome. The main clinicopathologic characteristics of this series are presented in Supplemental Table 3, Supplemental Digital Content 3, http://links.lww.com/PAS/B1000 . Ductal Carcinoma in Sit We analyzed 6 primary ductal carcinoma in sit (DCIS) with subsequent recurrences, and the main clinicopathologic features are presented in Table . Three primary carcinomas were LUM and 3 were HER2. Recurrences occurred after a mean period of 5.9 years (1.3 to 18.2 y; median 3.9 y). The main molecular alterations detected in the primary DCIS, in addition to ERBB2 amplifications (3 carcinomas), were TP534 mutations (2 carcinomas), CCND1 amplification (1 carcinoma), and MYC amplification (1 carcinoma). Recurrences in 4 DCIS were in the form of invasive carcinoma of no special type (IDC-NST), while 2 presented again as DCIS. The main clinicopathologic features and molecular alterations of the individual carcinomas are presented in Supplemental Table 4, Supplemental Digital Content 4, http://links.lww.com/PAS/C2 . According to molecular alterations, 4 recurrences were classified as true recurrences (including both DCIS recurrences); 1 recurrence was classified as new primary (ID 6) and 1 as undetermined (ID 3), since we did not find any molecular alteration in the primary carcinoma. Regarding progression, one true recurrence demonstrated an additional GATA3 mutation (ID 1). The only case (16.6%) considered a new primary (ID 6) was a LUM IDC-NST which developed in the patient with an ATM germline mutation. This carcinoma presented 18.2 years after the primary HER2 DCIS. Both carcinomas also presented different GATA3 mutations. Invasive Nonlobular Carcinomas Twenty-three pairs of carcinomas were analyzed, corresponding to 22 IDC-NST and 1 invasive micropapillary carcinoma. The main clinicopathologic features of this group are presented in Table . Primary carcinomas occurred in women with a mean age of 52 years. Recurrences developed between 1.1 and 31.7 years (mean of 8.37 y, median 5.1 y) after the primary carcinomas. In this series, histologic grade 3 (n=10, 43.5%), pT2 (n=11, 47.8%), and pN0 (n=14, 60.9%) primary carcinomas predominated. Regarding the surrogate molecular subtype, 13 carcinomas were LUM (56.5%), 5 TN (21.7%), 4 HER2 (17.4%), and 1 LUM-HER2 (4.3%). Molecular Alterations The most common molecular alterations observed in primary invasive carcinomas and recurrences were PIK3CA mutations (13 primaries and 14 recurrences), TP53 mutations (5 primaries and 4 recurrences), ERRB2 amplifications (5 primaries and 3 recurrences), GATA3 mutations (3 primaries and 4 recurrences), and MDM4 amplifications (3 primaries and 4 recurrences) (Fig. ). The list of main molecular alterations is provided in Supplemental Table 4, Supplemental Digital Content 4, http://links.lww.com/PAS/C2 . After comparing the molecular alterations in paired carcinomas, 17 (74%) recurrences were considered true recurrences and 5 (21.7%) were considered new primaries. In one case (ID 15), the molecular results were inconclusive, and the recurrence was classified as undetermined. Clinicopathologic and Molecular Features of True Recurrences All recurrences were IDC-NST and occurred between 1.1 and 23.6 years after the primaries (mean 7.2 y, median 3.4 y). Immunophenotype conversion was observed in 3 true recurrences: LUM to TN (ID 11), HER2 to LUM (ID 20) (Fig. and Supplemental Fig. 1, Supplemental Digital Content 5, http://links.lww.com/PAS/C3 ), and LUM-HER2 to HER2 (ID 22). The carcinoma in case ID 11 received chemotherapy and tamoxifen, while case ID 20 was treated with chemotherapy and trastuzumab. In case ID 22, trastuzumab was not initially administered, as the diagnosis occurred before the availability of anti-HER2 treatments; instead, the patient received chemotherapy and tamoxifen. In 7 of 17 (41.2%) true recurrences, we detected additional molecular alterations when compared with the primary carcinoma. There was no specific pattern of additional alterations in true recurrences. Mutations in PIK3CA, PIK3R1, MAP3K1, GATA3, TP53 , and ESR1; and amplification of CCND1 and MDM4 occurred in only one case each. The ESR1 mutation occurred in a LUM carcinoma (ID 21) after 11 years of hormone treatment (6 y with tamoxifen and 5 y with an aromatase inhibitor). Clinicopathologic and Molecular Features of New Primaries A summary of the clinicopathologic and molecular features of the 5 new primaries is presented in Table . All new primaries except one encapsulated papillary carcinoma were IDC-NST. ID 26, which presented a TP53 alteration only in the primary carcinoma, is depicted in Figure . New primaries occurred between 5.8 and 31.7 years after the primary carcinomas (mean 12.97 y, median 10.3 y). Lobular Carcinomas Clinicopathologic Features We studied 6 primary carcinomas and 6 recurrences from 5 patients. One primary carcinoma was in situ lobular carcinoma (LCIS) and 5 were invasive lobular carcinomas (ILC). All recurrences were invasive carcinomas. The main clinicopathologic and molecular features are presented in Table and Supplemental Table 4, Supplemental Digital Content 4, http://links.lww.com/PAS/C2 . Four primary carcinomas were LUM, one LUM-HER2 and 1 TN. Time from primary to recurrence ranged from 2.8 to 23.1 years (mean 12.2 y; median 7.54 y). The most common molecular alterations were CDH1 mutations, the molecular hallmark of lobular neoplasms. CDH1 mutations were detected in all samples except in one primary carcinoma (ID 30). Other alterations were PIK3CA mutations (1 primary and 2 recurrences), ERBB2 amplification (1 primary and 1 recurrence), and AKT1 mutation (1 recurrence). The HER2-positive ILC and its recurrence carried a TP53 mutation and amplification of CCND1 , MDM4 , and MYC . According to the molecular profile, 3 recurrences were considered true recurrences. Phenotype conversion occurred in one of these true recurrences, changing the phenotype from LUM-HER2 to HER2 (ID 29). The true recurrence that developed from the TN ILC acquired an AKT1 mutation during progression. One recurrence (ID 30) was considered a new primary and developed 6 years after the primary carcinoma. The recurrences occurring in a woman with a CDH1 germline mutation (ID 32), who developed bilateral ILC, were considered undetermined because the only common mutation detected in tumor tissues was the germline mutation. The left recurrence could be interpreted as a metastasis of the contralateral recurrences. We analyzed 6 primary ductal carcinoma in sit (DCIS) with subsequent recurrences, and the main clinicopathologic features are presented in Table . Three primary carcinomas were LUM and 3 were HER2. Recurrences occurred after a mean period of 5.9 years (1.3 to 18.2 y; median 3.9 y). The main molecular alterations detected in the primary DCIS, in addition to ERBB2 amplifications (3 carcinomas), were TP534 mutations (2 carcinomas), CCND1 amplification (1 carcinoma), and MYC amplification (1 carcinoma). Recurrences in 4 DCIS were in the form of invasive carcinoma of no special type (IDC-NST), while 2 presented again as DCIS. The main clinicopathologic features and molecular alterations of the individual carcinomas are presented in Supplemental Table 4, Supplemental Digital Content 4, http://links.lww.com/PAS/C2 . According to molecular alterations, 4 recurrences were classified as true recurrences (including both DCIS recurrences); 1 recurrence was classified as new primary (ID 6) and 1 as undetermined (ID 3), since we did not find any molecular alteration in the primary carcinoma. Regarding progression, one true recurrence demonstrated an additional GATA3 mutation (ID 1). The only case (16.6%) considered a new primary (ID 6) was a LUM IDC-NST which developed in the patient with an ATM germline mutation. This carcinoma presented 18.2 years after the primary HER2 DCIS. Both carcinomas also presented different GATA3 mutations. Twenty-three pairs of carcinomas were analyzed, corresponding to 22 IDC-NST and 1 invasive micropapillary carcinoma. The main clinicopathologic features of this group are presented in Table . Primary carcinomas occurred in women with a mean age of 52 years. Recurrences developed between 1.1 and 31.7 years (mean of 8.37 y, median 5.1 y) after the primary carcinomas. In this series, histologic grade 3 (n=10, 43.5%), pT2 (n=11, 47.8%), and pN0 (n=14, 60.9%) primary carcinomas predominated. Regarding the surrogate molecular subtype, 13 carcinomas were LUM (56.5%), 5 TN (21.7%), 4 HER2 (17.4%), and 1 LUM-HER2 (4.3%). Molecular Alterations The most common molecular alterations observed in primary invasive carcinomas and recurrences were PIK3CA mutations (13 primaries and 14 recurrences), TP53 mutations (5 primaries and 4 recurrences), ERRB2 amplifications (5 primaries and 3 recurrences), GATA3 mutations (3 primaries and 4 recurrences), and MDM4 amplifications (3 primaries and 4 recurrences) (Fig. ). The list of main molecular alterations is provided in Supplemental Table 4, Supplemental Digital Content 4, http://links.lww.com/PAS/C2 . After comparing the molecular alterations in paired carcinomas, 17 (74%) recurrences were considered true recurrences and 5 (21.7%) were considered new primaries. In one case (ID 15), the molecular results were inconclusive, and the recurrence was classified as undetermined. Clinicopathologic and Molecular Features of True Recurrences All recurrences were IDC-NST and occurred between 1.1 and 23.6 years after the primaries (mean 7.2 y, median 3.4 y). Immunophenotype conversion was observed in 3 true recurrences: LUM to TN (ID 11), HER2 to LUM (ID 20) (Fig. and Supplemental Fig. 1, Supplemental Digital Content 5, http://links.lww.com/PAS/C3 ), and LUM-HER2 to HER2 (ID 22). The carcinoma in case ID 11 received chemotherapy and tamoxifen, while case ID 20 was treated with chemotherapy and trastuzumab. In case ID 22, trastuzumab was not initially administered, as the diagnosis occurred before the availability of anti-HER2 treatments; instead, the patient received chemotherapy and tamoxifen. In 7 of 17 (41.2%) true recurrences, we detected additional molecular alterations when compared with the primary carcinoma. There was no specific pattern of additional alterations in true recurrences. Mutations in PIK3CA, PIK3R1, MAP3K1, GATA3, TP53 , and ESR1; and amplification of CCND1 and MDM4 occurred in only one case each. The ESR1 mutation occurred in a LUM carcinoma (ID 21) after 11 years of hormone treatment (6 y with tamoxifen and 5 y with an aromatase inhibitor). Clinicopathologic and Molecular Features of New Primaries A summary of the clinicopathologic and molecular features of the 5 new primaries is presented in Table . All new primaries except one encapsulated papillary carcinoma were IDC-NST. ID 26, which presented a TP53 alteration only in the primary carcinoma, is depicted in Figure . New primaries occurred between 5.8 and 31.7 years after the primary carcinomas (mean 12.97 y, median 10.3 y). The most common molecular alterations observed in primary invasive carcinomas and recurrences were PIK3CA mutations (13 primaries and 14 recurrences), TP53 mutations (5 primaries and 4 recurrences), ERRB2 amplifications (5 primaries and 3 recurrences), GATA3 mutations (3 primaries and 4 recurrences), and MDM4 amplifications (3 primaries and 4 recurrences) (Fig. ). The list of main molecular alterations is provided in Supplemental Table 4, Supplemental Digital Content 4, http://links.lww.com/PAS/C2 . After comparing the molecular alterations in paired carcinomas, 17 (74%) recurrences were considered true recurrences and 5 (21.7%) were considered new primaries. In one case (ID 15), the molecular results were inconclusive, and the recurrence was classified as undetermined. All recurrences were IDC-NST and occurred between 1.1 and 23.6 years after the primaries (mean 7.2 y, median 3.4 y). Immunophenotype conversion was observed in 3 true recurrences: LUM to TN (ID 11), HER2 to LUM (ID 20) (Fig. and Supplemental Fig. 1, Supplemental Digital Content 5, http://links.lww.com/PAS/C3 ), and LUM-HER2 to HER2 (ID 22). The carcinoma in case ID 11 received chemotherapy and tamoxifen, while case ID 20 was treated with chemotherapy and trastuzumab. In case ID 22, trastuzumab was not initially administered, as the diagnosis occurred before the availability of anti-HER2 treatments; instead, the patient received chemotherapy and tamoxifen. In 7 of 17 (41.2%) true recurrences, we detected additional molecular alterations when compared with the primary carcinoma. There was no specific pattern of additional alterations in true recurrences. Mutations in PIK3CA, PIK3R1, MAP3K1, GATA3, TP53 , and ESR1; and amplification of CCND1 and MDM4 occurred in only one case each. The ESR1 mutation occurred in a LUM carcinoma (ID 21) after 11 years of hormone treatment (6 y with tamoxifen and 5 y with an aromatase inhibitor). A summary of the clinicopathologic and molecular features of the 5 new primaries is presented in Table . All new primaries except one encapsulated papillary carcinoma were IDC-NST. ID 26, which presented a TP53 alteration only in the primary carcinoma, is depicted in Figure . New primaries occurred between 5.8 and 31.7 years after the primary carcinomas (mean 12.97 y, median 10.3 y). Clinicopathologic Features We studied 6 primary carcinomas and 6 recurrences from 5 patients. One primary carcinoma was in situ lobular carcinoma (LCIS) and 5 were invasive lobular carcinomas (ILC). All recurrences were invasive carcinomas. The main clinicopathologic and molecular features are presented in Table and Supplemental Table 4, Supplemental Digital Content 4, http://links.lww.com/PAS/C2 . Four primary carcinomas were LUM, one LUM-HER2 and 1 TN. Time from primary to recurrence ranged from 2.8 to 23.1 years (mean 12.2 y; median 7.54 y). The most common molecular alterations were CDH1 mutations, the molecular hallmark of lobular neoplasms. CDH1 mutations were detected in all samples except in one primary carcinoma (ID 30). Other alterations were PIK3CA mutations (1 primary and 2 recurrences), ERBB2 amplification (1 primary and 1 recurrence), and AKT1 mutation (1 recurrence). The HER2-positive ILC and its recurrence carried a TP53 mutation and amplification of CCND1 , MDM4 , and MYC . According to the molecular profile, 3 recurrences were considered true recurrences. Phenotype conversion occurred in one of these true recurrences, changing the phenotype from LUM-HER2 to HER2 (ID 29). The true recurrence that developed from the TN ILC acquired an AKT1 mutation during progression. One recurrence (ID 30) was considered a new primary and developed 6 years after the primary carcinoma. The recurrences occurring in a woman with a CDH1 germline mutation (ID 32), who developed bilateral ILC, were considered undetermined because the only common mutation detected in tumor tissues was the germline mutation. The left recurrence could be interpreted as a metastasis of the contralateral recurrences. We studied 6 primary carcinomas and 6 recurrences from 5 patients. One primary carcinoma was in situ lobular carcinoma (LCIS) and 5 were invasive lobular carcinomas (ILC). All recurrences were invasive carcinomas. The main clinicopathologic and molecular features are presented in Table and Supplemental Table 4, Supplemental Digital Content 4, http://links.lww.com/PAS/C2 . Four primary carcinomas were LUM, one LUM-HER2 and 1 TN. Time from primary to recurrence ranged from 2.8 to 23.1 years (mean 12.2 y; median 7.54 y). The most common molecular alterations were CDH1 mutations, the molecular hallmark of lobular neoplasms. CDH1 mutations were detected in all samples except in one primary carcinoma (ID 30). Other alterations were PIK3CA mutations (1 primary and 2 recurrences), ERBB2 amplification (1 primary and 1 recurrence), and AKT1 mutation (1 recurrence). The HER2-positive ILC and its recurrence carried a TP53 mutation and amplification of CCND1 , MDM4 , and MYC . According to the molecular profile, 3 recurrences were considered true recurrences. Phenotype conversion occurred in one of these true recurrences, changing the phenotype from LUM-HER2 to HER2 (ID 29). The true recurrence that developed from the TN ILC acquired an AKT1 mutation during progression. One recurrence (ID 30) was considered a new primary and developed 6 years after the primary carcinoma. The recurrences occurring in a woman with a CDH1 germline mutation (ID 32), who developed bilateral ILC, were considered undetermined because the only common mutation detected in tumor tissues was the germline mutation. The left recurrence could be interpreted as a metastasis of the contralateral recurrences. In this series, using a molecular approach, we found that 68.6% of ipsilateral breast carcinoma recurrences were true recurrences, 20% were new primaries, and 11.5% were undetermined. There are few studies analyzing the clonality of recurrences relative to the initial invasive carcinomas. Our results were similar to those obtained by Vicini et al, who reported the first study using a polymerase chain reaction loss-of-heterozygosity molecular comparison assay in 29 patients: 22 recurrences (76%) were related clonally to the initial carcinoma, and 7 recurrences (24%) were clonally different, representing new primaries. New molecular techniques, such as next-generation sequencing (NGS), have been applied to study primary breast carcinomas and their recurrences but were mainly focused on detecting molecular changes related to tumor progression. Thus, Priedigkeit and colleagues analyzed genes involved in the progression of 12 pairs of primary tumors and their local recurrences. They observed that one recurrence (8.3%) had a completely unique transcriptional and copy number profile compared with its matched primary tumor, suggesting it was a new primary carcinoma. Bosma et al analyzed 15 primary carcinomas and their local recurrences and found that all local recurrences were true genomic recurrences based on gene expression and copy number variation analyses. Recently, 2 large studies have analyzed the clonal relationships of in situ and invasive breast carcinomas and their recurrent ipsilateral carcinomas, with very different results. Lips et al studied 95 patients with DCIS and observed that only 18% of invasive recurrences were clonally unrelated to the primary DCIS, representing new independent carcinomas. This frequency of true recurrences was similar to that observed in our small series of DCIS. In contrast, Rassy et al reported in their series of invasive carcinomas that only 17 of 93 paired samples (17.7%) exhibited common variants and were considered true recurrences. In our opinion, the results of this study should be interpreted with caution, since the most common mutations found in the primary carcinomas in this series of LUM breast carcinomas were KMT2D , MTOR , ATM , NOTCH1 , and ATRX . In contrast, in our study and in larger series such as those reported by the TCGA and METABRIC, and Razavi et al the most frequent mutations were in PIK3CA , TP53 , and GATA3 , which were underrepresented in the Rassy series. Supporting the accuracy of our molecular approach in identifying new primaries was the difference in time to recurrence observed between true recurrences and new primaries in our series of invasive nonlobular carcinomas, the group that included more cases. Thus, whereas the median time to recurrence was 3.4 years in recurrences, it was 10.4 years in new primaries. However, some molecularly defined recurrences occurred after long periods in our series, even after more than 10 years. This could suggest biological differences between early and late recurrences. Early true recurrences could be the result of residual carcinoma close to margins that become clinically relevant shortly before the initial treatment. Late recurrences could be related to the cancer field effect in which large epithelial areas share one or more initiation mutations that produce invasive carcinomas at different times, probably influenced by microenvironmental factors. Cases ID 18 and ID 24 could be examples of this situation, since the second carcinomas developed between 16 and 24 years after the primaries, sharing identical PIK3CA mutations without evidence of additional alterations detected by our methodology. Whether these late recurrences behave more alike to true recurrences or new primaries deserves further investigation in a large sample. Lobular carcinoma is a special histologic type of breast carcinoma with specific morphologic, molecular, and clinical characteristics. In this small series of recurrent lobular carcinomas, we observed that, in contrast to IDC-NST, recurrences developed after a prolonged period of time since the primary carcinoma (median 12.2 y vs. 3.4 y). Even ID 29, an ILC with a TP53 mutation and amplification of CCND1 , ERBB2 , MYC , and MDM4 , developed a true recurrence 12 years after the initial treatment. The number of new primaries found in this series was higher after applying clinicopathologic algorithms. Thus, clinicopathologic algorithms tested in this series resulted in a diagnosis of a new primary in 51.6% to 68.6%, far from the 20% observed by molecular analysis. Whereas all carcinomas classified as new primary by molecular analysis were also classified as new primary by most clinicopathologic classifications, discrepancies were due to a high percentage of nonclonal recurrences classified as new primary by clinicopathologic classifications. One cause of discrepancy with clinicopathological methods using IHC variables (ER and HER2 expression) was that they did not consider possible phenotype conversion after treatment. Phenotype conversion was observed in 4 of 24 true recurrences (17%), and these could have been classified as new primaries according to immunohistochemical criteria. Loss of hormone receptors and HER2 expression in these cases was probably related to hormonal and anti-HER2 therapy, respectively. Case ID 20 showed a phenotype conversion from HER2 to LUM, indicating the loss of HER2 and the acquisition of hormone receptor expression. In this case, we observed that the primary carcinoma and its recurrence shared the same mutation ( GATA3 c.925-2delinsG). However, in the primary carcinoma, this mutation was at a very low VAF (2%) and was excluded in the automatic filtering. In the recurrence, VAF was 30%, indicating clonal selection after treatment (Supplemental Fig. 1, Supplemental Digital Content 5, http://links.lww.com/PAS/C3 ). To verify these observations, we repeated NGS analysis in different tumor sections of both carcinomas with similar results. On the other hand, 5 of 7 (71%) new primaries in this series showed the same immunophenotype as the first carcinomas and could have been classified as true recurrence by immunohistochemical criteria. In our series, most new primaries occurred after mastectomy (6 of 7, 85%). Thus, we observed that after mastectomy, 62.5% of recurrences were true recurrences and 37.5% were new primaries. In contrast, after conservative surgery, 93% of recurrences were true recurrences and only 7% were new primaries. The lower rate of new primaries after conservative surgery could also be due to the administration of breast radiotherapy. Deutschmann et al reported 17 true recurrences and 1 new primary after 126 therapeutic mastectomies. However, the definition of true recurrence in this study was quite open, since it was a recurrence of the same breast cancer subtype as the primary breast carcinoma. Recurrence after mastectomy has been associated with the volume of residual fibroglandular tissue (RFGT) after surgery. Detection rates of RFGT have been described as between 5% and 100% in the literature. Another factor that can favor the development of a new primary is being a carrier of a germline mutation in a breast cancer predisposition gene, such as ATM or CDH1 , as in two patients in our series. The patient carrying the germline pathogenic ATM mutation (ID 6) developed 2 new primaries in both breasts. Interestingly, the first 2 carcinomas in each breast were HER2 and the 2 new primaries were LUM. The primary carcinomas were not treated with anti-HER2 therapy, as they were diagnosed before HER2 testing was mandatory. Interestingly, it has been recently reported that TP53 , ATM , and CHEK2 germline variants are predisposing factors for HER2 subtypes; whereas BRCA1 , BRCA2 , PALB2 , RAD51C , and BARD1 germline variants are associated with a predisposition to low HER2 expression. Another patient in this series, who carried a pathogenic germline CDH1 variant (ID 32), developed bilateral ILCs and their recurrences. The mother of this patient died from breast cancer, but there was no familial or personal history of gastric cancer. Recently, it has been suggested that ILCs with CDH1 germline mutations, in which diffuse gastric cancer is absent and the ILC phenotype is predominant, could represent an independent inherited syndrome. In our patient, we could not establish molecularly if recurrences, developed 20 and 21 years after the first carcinomas, were true recurrences or new primaries since we only detected the germline mutation in the 2 initial carcinomas. Both recurrences carried distinct PIK3CA mutations. Finally, as previously stated, most molecular studies analyzing primary breast carcinomas and their recurrences were focused on describing gene alterations related to tumor progression. Although it was not the primary aim of this study, we were able to detect alterations that were private of recurrences in 10 (41.7%) carcinomas. As demonstrated in the metastatic setting, there was no specific pattern of mutations in true recurrences, but alterations in different genes were identified in each case ( PIK3CA, PIK3R1, MAP3K1, AKT1, GATA3, CCND1, MDM4, TP53 , and ESR1 ). Most alterations are well-known drivers in breast cancer, except ESR1 mutations. This occurs more frequently after hormone therapy, as in the case ID 21 of our series. The present study's limitations include a relatively low number of carcinomas, especially in the DCIS and lobular cancer groups, which precludes an analysis of the prognostic impact of the type of recurrence. In addition, although we analyzed the genes more frequently mutated or amplified in breast cancer, our molecular approach is limited, and results using larger NGS panels could modify the classification of some cases. In summary, we found that about 20% of recurrences were clonally unrelated with the initial carcinomas and considered as new primary. Most new primaries developed after therapeutic mastectomies and after a longer time interval than true recurrences. Clinicopathologic classification of recurrences seems to overestimate the number of new primaries, partially because they do not take into account phenotype conversion after treatment. More than 40% of true recurrences carried private mutations or amplifications that could be important in tumor progression, but a common mutational pattern of progression was absent. Further studies, including larger series, are necessary to evaluate the prognostic significance of the molecular classification of recurrences. |
Potential Efficacy of Herbal Medicine-Derived Carbon Dots in the Treatment of Diseases: From Mechanism to Clinic | 64b80e2b-c596-4369-b295-3624ec6fcf54 | 10642355 | Pharmacology[mh] | As an emerging carbon material after carbon nanotubes, graphene, fullerenes, and nano-diamonds, carbon dots (CDs) have superior properties than other nanomaterials, including ultra-fine size, favorable photoluminescence performance, low toxicity, strong biocompatibility, and excellent electron transfer ability. , The bioactivity of CDs has also been investigated and discovered recently, which were used as mitochondrial oxidative stress amplifiers for targeted therapy of cancer, as well as for fluorescent biosensing and imaging through additional properties. , Moreover, researchers from RMIT University stated that CDs can be utilized as a therapeutic platform to study drug delivery, distribution, metabolism, excretion, and toxicity. These investigations have established a solid foundation for the application of nano carbon dots in numerous research fields. However, the research on carbon dots (CDs) is mostly focused on the optimization of preparation methods and the expansion of application fields. The exploration of chemical and natural substances in synthetic raw materials also deserves considerable attention. The use of “green” substances as raw materials is turning into a hot topic in production research as the notion of green chemistry steadily gains acceptance with societal development. For the synthesis of CDs, numerous green precursors, which have significant advantages such as abundant and renewable raw materials, free of chemical pollution as well as environmentally benign, can be employed as natural carbon sources. Various green carbon precursors have been studied and applied, including fruits, vegetables, and a variety of food and beverage products, to achieve the synthesis of materials with excellent properties, high cost-effectiveness, economic, and environmental protection. Herbal medicine (HM), one of these green precursors, has gained substantial attention due to its distinctive medical efficacy. Numerous clinical studies have demonstrated that herbal medicine has shown exceptional efficacy on various specific diseases such as SARS and COVID-19. However, the medicinal efficacy of herbal medicines is failing to be adequately expressed due to their complex composition. As a result, researchers have initiated attempts to prepare various nanoscale substances of herbal medicines. Interestingly, as one of the most distinctive drugs in clinical applications, herbal medicines can be used as raw materials under elevated temperature conditions to produce novel nanomaterials, which are called herbal medicine-derived carbon dots (HM-CDs), with a diameter of less than 10 nm. Compared to general carbon dots, HM-CDs are synthesized using the medicinal components of herbs, thereby retaining the medicinal value and biological activities of the herbs. HM-CDs may contain active ingredients and chemical substances derived from herbs, such as polysaccharides, phenolic compounds, and alkaloids. , These components confer specific medicinal properties and biological activities to HM-CDs. Due to the preservation of medicinal value and biological activities of herbs, HM-CDs find wider applications in areas such as herbal extraction, drug delivery, and targeted therapy. The structural characteristics, physicochemical qualities, and biological functions of HM-CDs vary amongst different herbal medicines. At present, multiple pharmacological experiments have demonstrated the biological effect of carbon dots in herbal medicines, which revealed the pharmacodynamic basis of herbal medicines in various diseases from a fresh perspective. The purpose of this paper is to exhibit a comparative summary of the synthesis strategies and the main properties of HM-CDs. Special attention is also given to the latest trends in the management of multiple human diseases (including bleeding disorders, inflammatory diseases, cancer, pain of various forms and causes, gastrointestinal disorders, etc.) based on HM-CDs. The intrinsic pharmacological activities and mechanisms of these HM-CDs are also discussed to further advance their clinical applications. To date, numerous techniques have been developed for the synthesis of CDs, such as pyrolysis, microwave-mediated synthesis, chemical oxidation, and hydrothermal treatment. For the preparation of different types of HM-CDs, each of these methods has distinct properties in terms of particle size, quantum yield, and pharmacological activity . However, most of these methods are not environmentally friendly, which require copious volumes of strong acids, harsh synthesis conditions, and complex processes, making large-scale manufacturing challenging. From the perspective of environmental protection, this raises concerns about the toxicity and environmental impact of CDs, necessitating the urgent need for green, eco-friendly, and straightforward synthesis processes. The hydrothermal and calcination methods, which are the most commonly used methods for the preparation of HM-CDs, have the advantages of inexpensive instrumentation, environmental friendliness, and ease of operation, and will be a crucial strategy for further research on the bioactivity of HM-CDs. Hydrothermal Method Hydrothermal synthesis is environmentally friendly without the addition of organic solvents. The surface of CDs does not require additional passivation to maximize safety and minimize toxicity. Prior to preparation, the dried herbs are cut into tiny pieces or ground into a powder in purified water. After the ultrasonic treatment, the mixture is transferred to a PTFE-lined stainless steel autoclave and heated at a specific temperature. To purify the CDs, the suspension needed to be further filtered through a 0.22 μm cellulose membrane and dialyzed for several days using a dialysis bag . The reaction temperature would affect the properties of HM-CDs. The hydrothermal reaction temperature is generally 100–200°C. Li et al synthesized nitrogen-doped CDs of ginkgo fruits (H-N-CDs) at different temperatures. These CDs had the best fluorescence intensity and maximum quantum yield (QY) when the temperature was set to 200°C. In a separate study, the corresponding carbon dots were prepared based on Mentha haplocalyx Briq . When the temperature was below 120°C, electron microscopy revealed a significant amount of polymers, suggesting that the carbonization was incomplete and the CDs were difficult to form. However, no polymers were found by electron microscopy when the temperature reached 180°C, indicating that all compounds were carbonized. This phenomenon has also been observed in coix seed-CDs. The fluorescence intensity of coix seed-CDs decreased with the increase in temperature. In addition, Li et al used Salvia miltiorrhiza Bunge as a carbon source and synthesized three CDs (CDs-100, CDs-150, and CDs-180) by hydrothermal method at different temperatures (100, 150, and 180°C) for 6 h. The average diameters were 16.94, 1.53, and 2.03 nm, respectively. Similarly, the reaction time affects the performance of the CDs. The QY of CDs generated from orange peel ( Citrus reticulata Blanco .) at different time points at the same temperature decreases with increasing reaction time, while the particle size increases slightly. , In another study, honey-CDs prepared by hydrothermal heating at 100°C for 2 h were only stable at 4°C for 3 months. The difference in fluorescence intensity of honey-CDs ceased to be significant when the synthesis time increased to 12 or even 16 hours, suggesting that the fluorescence intensity may have reached saturation. This phenomenon was also observed in aloe vera ( Aloe L .) CDs, where the fluorescence intensity progressively increased with reaction time up to 11 h but decreased afterward. Therefore, the synthesis of CDs by the hydrothermal method needs to be adequately considered and validated with respect to the reaction temperature and time. The hydrothermal temperature should preferably be higher than 100°C. The required time can be determined by the color shift in the precursor solution, which is commonly yellow, orange, or brown. Pyrolysis Method High-temperature pyrolysis is a typical process in addition to hydrothermal synthesis. Natural organic materials are gradually transformed into CDs under vacuum or inert gas by high-temperature processes such as heating, dehydration, degradation, and carbonization. The process is easy to operate, solvent-free, inexpensive, and appropriate for mass manufacturing. The herbs are first placed in a crucible and heated at a specific temperature in a muffle furnace until they are charred. The charred medicine is then crushed and boiled in ultrapure water, and the upper liquid layer is collected. The solution was filtered through a 0.22 μm microporous membrane and dialyzed for several days using dialysis bags to purify the CDs . Carbonization is one of the main elements that affect the success rate of the preparation process through high-temperature calcination. There exist two main traditional methods of carbonation: carbonizing by stir-frying and carbonizing by calcining (also known as wok-covering calcining). Both techniques are applicable to drugs in general. Carbonizing by stir-frying means heating the drug in a preheated vessel over high or moderate heat until the drug turns reddish-brown inside and burns black on the outside, mainly used for root drugs such as carbonized rhubarb, carbonized ginger, and carbonized cortical peony. Carbonization by calcination implies heating and carbonizing the drug under high temperature and anoxic conditions. It is appropriate for loose or light medications that can be easily carbonized (such as Juncus efsus, Radix Rehmanniae , and Nodus Nelumbinis Rhizomatis ). Unfortunately, the limitations of these conventional charring techniques make it challenging to regulate the charring of charcoal-based pharmaceuticals. For example, (i) non-uniform heating can lead to ashing, carbonization, or raw blanks; (ii) it is difficult to control the duration and degree of heating for light-weight drugs, resulting in excessive waste rate; (iii) the root drug is not dry inside, so it cannot be entirely charred; (iv) the operation is cumbersome, time-consuming, and smoke-filled. It is worth noting that the two most popular instrumental means for the production of HM-CDs are calcination in a muffle furnace and carbonization in a drying oven. This type of carbonization can solve the problem of uncontrollable temperature and time. Compared to hydrothermal synthesis, high-temperature pyrolysis generally requires higher reaction temperatures (300°C-400°C) and shorter heating times. The optimal pyrolysis temperature varies when different carbon sources are used to synthesize CDs. Dager et al prepared fennel seed-CDs at 500°C for 3 h, which is the maximum calcination temperature recorded in the current study. These fennel seed-CDs with excellent properties can be preserved for up to 15 months. In contrast, the present HM-CDs made by high-temperature pyrolysis have heating temperatures as low as 220°C. At this temperature, Blue-light CDs were prepared using watermelon peel as the carbon source. However, the reality is that the reaction mechanism for the synthesis of HM-CDs is extremely complex and is affected by various factors such as the temperature, time, and pH of the reaction system. Zhang et al prepared PCC-CDs based on Phellodendri Chinensis Cortex (PCC) under different conditions. PCC-CDs produced at 400°C were a novel carbon-based nanomaterial with exceptional bioactivity, and their antipsoriatic activity was superior to those prepared at alternative temperatures (325°C and 475°C). Another study used pyrolysis to prepare Zingiberis rhizoma -based carbon dots (ZR-CDs) and examined their analgesic activity when carbonized at different temperatures (300, 350, and 400°C) for 1 h or at 350°C for varying periods of time (0.5, 1, and 1.5 h). Ultimately, ZR-CDs were found to have the optimum outcome when prepared at 350°C for 1 h. One hypothesis suggests that the multiple properties of CDs, such as rich chemical groups, size, and solubility, play an important role in their biological applications, , which would lead to small or significant changes in the physicochemical properties of PCC-CDs prepared at different temperatures, resulting in large changes in the biological activity of the obtained CDs. Alternative Methods In view of the drawbacks of the aforementioned techniques, two novel carbonization techniques have been developed: heating with sand and microwave carbonization . Due to its excellent thermal conductivity, highly heated sand prevents inhomogeneous heating of the drug, quickly achieves the desired energy, and is low-cost and simple to manufacture. The majority of charcoal-based medicines, including Nodus Nelumbinis Rhizomatis, Sanguisorba officinalis , and Fructus Crataegi , can be prepared using this approach, while light, friable, or non-separable medicines are not applicable. Yet, the idea behind microwave carbonization is to use energy transmission to cause the breaking of chemical bonds. The response time is drastically decreased and preparation efficacy is increased because of its simpler operation. This method has the advantages of high processing accuracy, minimum contamination, and a wide range of applications for light-textured charcoal herbs. Moreover, a substitute for traditional hydrothermal synthesis has been reported: microwave-assisted hydrothermal synthesis. Li et al prepared two ginkgo fruit-CDs (H-CDs/M-CDs) by hydrothermal (H) and microwave methods (M), respectively. The time required for M-CDs was 5–15 min, which was significantly shorter than that of the hydrothermal method and the particle size was relatively smaller. However, the performance of H-CDs is significantly superior to that of M-CDs. This is due, in part, to the more regular and homogeneous morphology of H-CDs, as well as to the fact that their quantum yields and lifetimes are larger than those of M-CDs and their fluorescence intensity is higher. Interestingly, the microwave technique can even prepare orange peel-CDs in 1 min with a yield of up to 16.20%. In terms of reflection time and efficiency, the microwave synthesis method is undoubtedly superior to hydrothermal and pyrolysis methods, and its time-saving, inexpensive, and easy-to-operate features are particularly attractive for environmentally friendly synthesis of HM-CDs from renewable herbs. Currently, more and more HM-CDs are being prepared by applying microwave carbonization , but it is still far from being a completely developed technique. Another technique for creating HM-CDs, in addition to the ones mentioned above, is heating extraction using different solvents. , For example, Wang et al prepared ethanol-papaya CDs (E-CDs) using 90% ethanol. Although the size of the E-CDs increased with the amount of organic macromolecules in 90% ethanol, alternative solvents may produce the best preparation of CDs when corresponding to various herbal species. Sugarcane ( Saccharum sinensis Roxb .) has also been employed as a carbon source for the synthesis of herbal CDs via the solvothermal method. , The use of organic solvents facilitated the carbonization process during the synthesis of CDs, which significantly altered the photophysical characteristics of the carbon nanoparticles. In another study, environmentally friendly CDs of the Codonopsis pilosula were prepared at room temperature using a one-step solvothermal method. The obtained codonopsis pilosula -derived CDs (CP-CDs) exhibited excellent fluorescence properties (QY up to 12.8%) and strong photostability without any passivation or functionalization on the CP-CDs surface. Therefore, the preparation methods for HM-CDs are diverse, sharing similarities with general carbon dot synthesis methods. The common feature of these methods is that they mainly control the carbon source and reaction conditions to achieve the preparation of carbon dots. However, the distinctive aspect lies in the incorporation of herbal materials as the carbon source in the preparation of HM-CDs. Herbal materials contain abundant organic substances, such as polysaccharides, proteins, and polyphenols, which can be decomposed into carbon dots at high temperatures. The preparation methods of HM-CDs also consider the characteristics and medicinal effects of herbal materials. This includes selecting appropriate extraction methods, solvents, and reaction conditions to retain the effective components of herbal materials and convert them into carbon dots. Furthermore, the preparation methods of HM-CDs can be combined with traditional herbal processing techniques, such as decocting and frying, to further regulate the morphology and properties of carbon dots. These special preparation methods can endow HM-CDs with enhanced biocompatibility and drug release performance, making them suitable for applications in the field of biomedicine. Overall, hydrothermal and pyrolysis technologies are the most popular methods for producing HM-CDs due to their practicality, economy, simplicity of usage, and environmental friendliness. However, the synthesis of HM-CDs by the hydrothermal method is normally considered as a time-consuming process. Although microwave-assisted methods are not as frequently used as hydrothermal and pyrolysis methods, their time-saving, low-cost, easy-to-operate, and efficient features are ideal for the synthesis of HM-CDs. Hydrothermal synthesis is environmentally friendly without the addition of organic solvents. The surface of CDs does not require additional passivation to maximize safety and minimize toxicity. Prior to preparation, the dried herbs are cut into tiny pieces or ground into a powder in purified water. After the ultrasonic treatment, the mixture is transferred to a PTFE-lined stainless steel autoclave and heated at a specific temperature. To purify the CDs, the suspension needed to be further filtered through a 0.22 μm cellulose membrane and dialyzed for several days using a dialysis bag . The reaction temperature would affect the properties of HM-CDs. The hydrothermal reaction temperature is generally 100–200°C. Li et al synthesized nitrogen-doped CDs of ginkgo fruits (H-N-CDs) at different temperatures. These CDs had the best fluorescence intensity and maximum quantum yield (QY) when the temperature was set to 200°C. In a separate study, the corresponding carbon dots were prepared based on Mentha haplocalyx Briq . When the temperature was below 120°C, electron microscopy revealed a significant amount of polymers, suggesting that the carbonization was incomplete and the CDs were difficult to form. However, no polymers were found by electron microscopy when the temperature reached 180°C, indicating that all compounds were carbonized. This phenomenon has also been observed in coix seed-CDs. The fluorescence intensity of coix seed-CDs decreased with the increase in temperature. In addition, Li et al used Salvia miltiorrhiza Bunge as a carbon source and synthesized three CDs (CDs-100, CDs-150, and CDs-180) by hydrothermal method at different temperatures (100, 150, and 180°C) for 6 h. The average diameters were 16.94, 1.53, and 2.03 nm, respectively. Similarly, the reaction time affects the performance of the CDs. The QY of CDs generated from orange peel ( Citrus reticulata Blanco .) at different time points at the same temperature decreases with increasing reaction time, while the particle size increases slightly. , In another study, honey-CDs prepared by hydrothermal heating at 100°C for 2 h were only stable at 4°C for 3 months. The difference in fluorescence intensity of honey-CDs ceased to be significant when the synthesis time increased to 12 or even 16 hours, suggesting that the fluorescence intensity may have reached saturation. This phenomenon was also observed in aloe vera ( Aloe L .) CDs, where the fluorescence intensity progressively increased with reaction time up to 11 h but decreased afterward. Therefore, the synthesis of CDs by the hydrothermal method needs to be adequately considered and validated with respect to the reaction temperature and time. The hydrothermal temperature should preferably be higher than 100°C. The required time can be determined by the color shift in the precursor solution, which is commonly yellow, orange, or brown. High-temperature pyrolysis is a typical process in addition to hydrothermal synthesis. Natural organic materials are gradually transformed into CDs under vacuum or inert gas by high-temperature processes such as heating, dehydration, degradation, and carbonization. The process is easy to operate, solvent-free, inexpensive, and appropriate for mass manufacturing. The herbs are first placed in a crucible and heated at a specific temperature in a muffle furnace until they are charred. The charred medicine is then crushed and boiled in ultrapure water, and the upper liquid layer is collected. The solution was filtered through a 0.22 μm microporous membrane and dialyzed for several days using dialysis bags to purify the CDs . Carbonization is one of the main elements that affect the success rate of the preparation process through high-temperature calcination. There exist two main traditional methods of carbonation: carbonizing by stir-frying and carbonizing by calcining (also known as wok-covering calcining). Both techniques are applicable to drugs in general. Carbonizing by stir-frying means heating the drug in a preheated vessel over high or moderate heat until the drug turns reddish-brown inside and burns black on the outside, mainly used for root drugs such as carbonized rhubarb, carbonized ginger, and carbonized cortical peony. Carbonization by calcination implies heating and carbonizing the drug under high temperature and anoxic conditions. It is appropriate for loose or light medications that can be easily carbonized (such as Juncus efsus, Radix Rehmanniae , and Nodus Nelumbinis Rhizomatis ). Unfortunately, the limitations of these conventional charring techniques make it challenging to regulate the charring of charcoal-based pharmaceuticals. For example, (i) non-uniform heating can lead to ashing, carbonization, or raw blanks; (ii) it is difficult to control the duration and degree of heating for light-weight drugs, resulting in excessive waste rate; (iii) the root drug is not dry inside, so it cannot be entirely charred; (iv) the operation is cumbersome, time-consuming, and smoke-filled. It is worth noting that the two most popular instrumental means for the production of HM-CDs are calcination in a muffle furnace and carbonization in a drying oven. This type of carbonization can solve the problem of uncontrollable temperature and time. Compared to hydrothermal synthesis, high-temperature pyrolysis generally requires higher reaction temperatures (300°C-400°C) and shorter heating times. The optimal pyrolysis temperature varies when different carbon sources are used to synthesize CDs. Dager et al prepared fennel seed-CDs at 500°C for 3 h, which is the maximum calcination temperature recorded in the current study. These fennel seed-CDs with excellent properties can be preserved for up to 15 months. In contrast, the present HM-CDs made by high-temperature pyrolysis have heating temperatures as low as 220°C. At this temperature, Blue-light CDs were prepared using watermelon peel as the carbon source. However, the reality is that the reaction mechanism for the synthesis of HM-CDs is extremely complex and is affected by various factors such as the temperature, time, and pH of the reaction system. Zhang et al prepared PCC-CDs based on Phellodendri Chinensis Cortex (PCC) under different conditions. PCC-CDs produced at 400°C were a novel carbon-based nanomaterial with exceptional bioactivity, and their antipsoriatic activity was superior to those prepared at alternative temperatures (325°C and 475°C). Another study used pyrolysis to prepare Zingiberis rhizoma -based carbon dots (ZR-CDs) and examined their analgesic activity when carbonized at different temperatures (300, 350, and 400°C) for 1 h or at 350°C for varying periods of time (0.5, 1, and 1.5 h). Ultimately, ZR-CDs were found to have the optimum outcome when prepared at 350°C for 1 h. One hypothesis suggests that the multiple properties of CDs, such as rich chemical groups, size, and solubility, play an important role in their biological applications, , which would lead to small or significant changes in the physicochemical properties of PCC-CDs prepared at different temperatures, resulting in large changes in the biological activity of the obtained CDs. In view of the drawbacks of the aforementioned techniques, two novel carbonization techniques have been developed: heating with sand and microwave carbonization . Due to its excellent thermal conductivity, highly heated sand prevents inhomogeneous heating of the drug, quickly achieves the desired energy, and is low-cost and simple to manufacture. The majority of charcoal-based medicines, including Nodus Nelumbinis Rhizomatis, Sanguisorba officinalis , and Fructus Crataegi , can be prepared using this approach, while light, friable, or non-separable medicines are not applicable. Yet, the idea behind microwave carbonization is to use energy transmission to cause the breaking of chemical bonds. The response time is drastically decreased and preparation efficacy is increased because of its simpler operation. This method has the advantages of high processing accuracy, minimum contamination, and a wide range of applications for light-textured charcoal herbs. Moreover, a substitute for traditional hydrothermal synthesis has been reported: microwave-assisted hydrothermal synthesis. Li et al prepared two ginkgo fruit-CDs (H-CDs/M-CDs) by hydrothermal (H) and microwave methods (M), respectively. The time required for M-CDs was 5–15 min, which was significantly shorter than that of the hydrothermal method and the particle size was relatively smaller. However, the performance of H-CDs is significantly superior to that of M-CDs. This is due, in part, to the more regular and homogeneous morphology of H-CDs, as well as to the fact that their quantum yields and lifetimes are larger than those of M-CDs and their fluorescence intensity is higher. Interestingly, the microwave technique can even prepare orange peel-CDs in 1 min with a yield of up to 16.20%. In terms of reflection time and efficiency, the microwave synthesis method is undoubtedly superior to hydrothermal and pyrolysis methods, and its time-saving, inexpensive, and easy-to-operate features are particularly attractive for environmentally friendly synthesis of HM-CDs from renewable herbs. Currently, more and more HM-CDs are being prepared by applying microwave carbonization , but it is still far from being a completely developed technique. Another technique for creating HM-CDs, in addition to the ones mentioned above, is heating extraction using different solvents. , For example, Wang et al prepared ethanol-papaya CDs (E-CDs) using 90% ethanol. Although the size of the E-CDs increased with the amount of organic macromolecules in 90% ethanol, alternative solvents may produce the best preparation of CDs when corresponding to various herbal species. Sugarcane ( Saccharum sinensis Roxb .) has also been employed as a carbon source for the synthesis of herbal CDs via the solvothermal method. , The use of organic solvents facilitated the carbonization process during the synthesis of CDs, which significantly altered the photophysical characteristics of the carbon nanoparticles. In another study, environmentally friendly CDs of the Codonopsis pilosula were prepared at room temperature using a one-step solvothermal method. The obtained codonopsis pilosula -derived CDs (CP-CDs) exhibited excellent fluorescence properties (QY up to 12.8%) and strong photostability without any passivation or functionalization on the CP-CDs surface. Therefore, the preparation methods for HM-CDs are diverse, sharing similarities with general carbon dot synthesis methods. The common feature of these methods is that they mainly control the carbon source and reaction conditions to achieve the preparation of carbon dots. However, the distinctive aspect lies in the incorporation of herbal materials as the carbon source in the preparation of HM-CDs. Herbal materials contain abundant organic substances, such as polysaccharides, proteins, and polyphenols, which can be decomposed into carbon dots at high temperatures. The preparation methods of HM-CDs also consider the characteristics and medicinal effects of herbal materials. This includes selecting appropriate extraction methods, solvents, and reaction conditions to retain the effective components of herbal materials and convert them into carbon dots. Furthermore, the preparation methods of HM-CDs can be combined with traditional herbal processing techniques, such as decocting and frying, to further regulate the morphology and properties of carbon dots. These special preparation methods can endow HM-CDs with enhanced biocompatibility and drug release performance, making them suitable for applications in the field of biomedicine. Overall, hydrothermal and pyrolysis technologies are the most popular methods for producing HM-CDs due to their practicality, economy, simplicity of usage, and environmental friendliness. However, the synthesis of HM-CDs by the hydrothermal method is normally considered as a time-consuming process. Although microwave-assisted methods are not as frequently used as hydrothermal and pyrolysis methods, their time-saving, low-cost, easy-to-operate, and efficient features are ideal for the synthesis of HM-CDs. Particle Size It is well known that nanoscale HM-CDs have received considerable attention worldwide. The bulk of HM-CDs have an average particle size of less than 10 nm, while the average diameter of the smallest one is 1.12 nm. Pyrolysis can produce lower particle sizes, despite the fact that hydrothermal synthesis is more effective at achieving narrow particle size distributions of CDs. The particle size of HM-CDs prepared by pyrolysis was approximately 5 nm under the current synthesis conditions, which seems to represent no discernible difference in the particle sizes of HM-CDs synthesized by the two methods and merits additional exploration. Wang et al isolated a novel carbon dots (PT-CDs) derived from Pollen Typhae by pyrolysis to ameliorate acute kidney injury. The particle size of PT-CDs prepared at different temperatures of 250, 300, 350, and 400°C was less than 20 nm, and the average particle size tended to increase and then decrease, with the lowest value at 400°C (4.85 ± 2.06 nm). The BUN index (250°C, 32.46 ± 2.93 mmol/L; 300°C, 31.98 ± 3.29 mmol/L; 350°C, 31.13 ± 3.11 mmol/L; and 400°C, 31.05 ± 2.70 mmol/L) and CRE level (250°C, 241.95 ± 21.56 μmol/L; 300°C, 242.85 ± 18.79 μmol/L; 350°C, 231.75 ± 19.58 μmol/L; and 400°C, 223.42 ± 17.90 μmol/L) of rats decreased with the increase in preparation temperature after PT-CDs prepared under various conditions were used for treatment, and the best anti-AKI effect was observed at 400°C, according to the results of the renal function evaluation. It is thus speculated that particle size is also one of the elements that regulate the pharmacological activity of HM-CDs. Nanoscale HM-CDs significantly improved membrane permeability and exerted a greater effect than herbal medicines . Pn-CDs can cross the blood-brain barrier (BBB), which may be due to the ultra-tiny size of Pn-CDs, the abundance of functional groups on the surface, and the strong affinity for the BBB endothelial cell membrane. Drugs are also covalently attached to CDs, which facilitates carrier-mediated macromolecular transport. These features enable CDs to enhance BBB permeability through passive transport. Ashrafizadeh et al summarized innovative drug delivery systems using functionalized CDs as carriers for the treatment of various neurological diseases. However, the costly modified ligands constrain their wide range of applications. Although some herbal drugs have been utilized for a long time for the treatment of neurological diseases, the BBB hinders the infiltration of herbal macromolecules. HM-CDs can enhance the BBB permeability of certain macromolecules under non-functionalized conditions. Therefore, this strategy has the potential to be a current breakthrough for herbal medicines to overcome biological barriers. Quantum Yield The quantum yield (QY) of pyrolytic synthesis was discovered to be lower than that of hydrothermal synthesis as a result of the diversity of carbon sources. Most CDs synthesized by pyrolysis had an average QY of less than 10%. However, two investigations produced different results for the synthesis of Schizonepetae Herba Carbonisata -CDs (SHC-CDs) under the same circumstances. One of the SHC-CDs had an average particle size of 0.8–4.0 nm and a QY of 2.26%, whereas those from the another had an average particle size of 1.29–6.87 nm and a QY of 6.31%. These findings illustrate the instability of the method. Nevertheless, Zhang et al created hair CDs with a higher QY (86.06%), which was significantly greater than that of citric acid CDs (19.73%), by combining pyrolysis and microwave. The fusion of the two synthetic techniques could offer potential benefits in addition to the differences in carbon sources. They also produced skin CDs with higher QY (51.35%), indicating that protein-rich materials are more suitable as precursors for CDs preparation. Thus, animal-derived herbs may be the most promising high-yielding drugs for future synthetic of CDs. The carbon dots originating from the same part of different herbs have different properties. Researchers developed CDs from 14 different orange peels under the same preparation conditions with significant differences in QY, possibly related to the volatile oil content. In addition, there are differences in the properties of CDs extracted from different parts of the same herb. For instance, Jiang et al prepared ginkgo leaves-CDs with a high QY (22.80%) using a hydrothermal synthesis technique. However, ginkgo fruit-CDs had a QY of only 3.33%. Evidence suggests that herbs from various portions of the same plant produce distinct HM-CDs, presumably due to compositional variations. It is well known that nanoscale HM-CDs have received considerable attention worldwide. The bulk of HM-CDs have an average particle size of less than 10 nm, while the average diameter of the smallest one is 1.12 nm. Pyrolysis can produce lower particle sizes, despite the fact that hydrothermal synthesis is more effective at achieving narrow particle size distributions of CDs. The particle size of HM-CDs prepared by pyrolysis was approximately 5 nm under the current synthesis conditions, which seems to represent no discernible difference in the particle sizes of HM-CDs synthesized by the two methods and merits additional exploration. Wang et al isolated a novel carbon dots (PT-CDs) derived from Pollen Typhae by pyrolysis to ameliorate acute kidney injury. The particle size of PT-CDs prepared at different temperatures of 250, 300, 350, and 400°C was less than 20 nm, and the average particle size tended to increase and then decrease, with the lowest value at 400°C (4.85 ± 2.06 nm). The BUN index (250°C, 32.46 ± 2.93 mmol/L; 300°C, 31.98 ± 3.29 mmol/L; 350°C, 31.13 ± 3.11 mmol/L; and 400°C, 31.05 ± 2.70 mmol/L) and CRE level (250°C, 241.95 ± 21.56 μmol/L; 300°C, 242.85 ± 18.79 μmol/L; 350°C, 231.75 ± 19.58 μmol/L; and 400°C, 223.42 ± 17.90 μmol/L) of rats decreased with the increase in preparation temperature after PT-CDs prepared under various conditions were used for treatment, and the best anti-AKI effect was observed at 400°C, according to the results of the renal function evaluation. It is thus speculated that particle size is also one of the elements that regulate the pharmacological activity of HM-CDs. Nanoscale HM-CDs significantly improved membrane permeability and exerted a greater effect than herbal medicines . Pn-CDs can cross the blood-brain barrier (BBB), which may be due to the ultra-tiny size of Pn-CDs, the abundance of functional groups on the surface, and the strong affinity for the BBB endothelial cell membrane. Drugs are also covalently attached to CDs, which facilitates carrier-mediated macromolecular transport. These features enable CDs to enhance BBB permeability through passive transport. Ashrafizadeh et al summarized innovative drug delivery systems using functionalized CDs as carriers for the treatment of various neurological diseases. However, the costly modified ligands constrain their wide range of applications. Although some herbal drugs have been utilized for a long time for the treatment of neurological diseases, the BBB hinders the infiltration of herbal macromolecules. HM-CDs can enhance the BBB permeability of certain macromolecules under non-functionalized conditions. Therefore, this strategy has the potential to be a current breakthrough for herbal medicines to overcome biological barriers. The quantum yield (QY) of pyrolytic synthesis was discovered to be lower than that of hydrothermal synthesis as a result of the diversity of carbon sources. Most CDs synthesized by pyrolysis had an average QY of less than 10%. However, two investigations produced different results for the synthesis of Schizonepetae Herba Carbonisata -CDs (SHC-CDs) under the same circumstances. One of the SHC-CDs had an average particle size of 0.8–4.0 nm and a QY of 2.26%, whereas those from the another had an average particle size of 1.29–6.87 nm and a QY of 6.31%. These findings illustrate the instability of the method. Nevertheless, Zhang et al created hair CDs with a higher QY (86.06%), which was significantly greater than that of citric acid CDs (19.73%), by combining pyrolysis and microwave. The fusion of the two synthetic techniques could offer potential benefits in addition to the differences in carbon sources. They also produced skin CDs with higher QY (51.35%), indicating that protein-rich materials are more suitable as precursors for CDs preparation. Thus, animal-derived herbs may be the most promising high-yielding drugs for future synthetic of CDs. The carbon dots originating from the same part of different herbs have different properties. Researchers developed CDs from 14 different orange peels under the same preparation conditions with significant differences in QY, possibly related to the volatile oil content. In addition, there are differences in the properties of CDs extracted from different parts of the same herb. For instance, Jiang et al prepared ginkgo leaves-CDs with a high QY (22.80%) using a hydrothermal synthesis technique. However, ginkgo fruit-CDs had a QY of only 3.33%. Evidence suggests that herbs from various portions of the same plant produce distinct HM-CDs, presumably due to compositional variations. As a novel constituent of the “nanoparticle universe”, HM-CDs have garnered significant attention, prompting extensive investigations into their inherent characteristics using diverse analytical techniques. Distinct spectroscopic methodologies, such as FTIR and UV-Vis spectroscopy, have been judiciously employed to probe the nuanced attributes of herbal CDs. Furthermore, the crystal structure, elemental composition, morphology, and sundry properties of CDs extracted from natural products have been meticulously elucidated via electron microscopy, zeta potential analysis, and X-ray techniques. Spectrographic Techniques Spectroscopic techniques such as UV-Vis absorption spectroscopy, fluorescence spectroscopy, and Raman spectroscopy are employed to analyze the optical properties and electronic structure of HM-CDs. Interestingly, UV-Vis spectroscopy is commonly recommended to evaluate the optical properties of HM-CDs as they typically exhibit strong UV absorption, although the absorption peaks may vary. High-performance liquid chromatography and gel electrophoresis are utilized to separate HM-CDs, allowing for the isolation of CDs with different sizes and shapes. It has been confirmed that CDs with sizes of 1.2, 1.5–3, and 3.8 nm emit light in the visible (400–700 nm), UV (350 nm), and near-infrared (NIR) regions, respectively. Therefore, the absorption band peak centered around the UV region of 250–300 nm is often referred to as the typical π-π* transition peak in most CDs. For instance, Lycii fructus CDs synthesized through hydrothermal treatment exhibit a strong absorption peak at 271 nm in the UV region. CDs derived from Borassus flabellifer flower via thermal decomposition exhibit a UV absorption peak at 282 nm, which is attributed to the π-π* transition of aromatic C=C bonds. Moreover, the surface of HM-CDs is typically composed of various functional groups, such as hydroxyl, carboxyl, carbonyl, ether, or epoxy groups, depending on the synthesis techniques used. Fourier-transform infrared spectroscopy (FTIR) can be utilized to determine the surface functional groups of HM-CDs. For instance, the peaks appearing in FTIR spectra of CDs prepared by ultrasonication irradiation of crab shells were at 3398 cm −1 , 2930 cm −1 , 1640 cm −1 , 1563 cm −1 and 1415 cm −1 , which correspond to the stretching vibration of -H stretching, N-H stretching, C-H stretching, C=O stretching, N-H bending and C=C stretching. The advantage of FTIR in characterizing the surface functionalization of carbon dots lies in its affordability, ease of sample preparation, and rapidity. Electron Microscopy Techniques Electron microscopy techniques play a crucial role in the characterization of nanoparticles. Researchers have widely employed scanning electron microscopy (SEM) and transmission electron microscopy (TEM) to visualize HM-CDs and gain insights into their morphology, size, and formation mechanism. SEM involves scanning the surface of HM-CDs with a focused electron beam to generate images. However, since the particle size of HM-CDs is typically smaller than 10 nm, TEM, which utilizes high-energy electron beams to obtain images through the herbal CD samples, offers higher resolution and is more suitable for identifying small-sized particles compared to SEM. For instance, Dager et al utilized TEM to determine the size of HM-CDs synthesized from microwave irradiation of Fenugreek seeds, revealing an average diameter of 4.25 ± 0.56 nm. Moreover, high-resolution transmission electron microscopy (HRTEM) has proven effective in the structural analysis and detection of lattice defects in HM-CDs. HRTEM analysis demonstrated that green CDs derived from tomatoes exhibit a spherical shape, with a size distribution ranging from 5 to 10 nm. X-Ray Techniques X-ray diffraction (XRD) and X-ray photoelectron spectroscopy (XPS) are valuable techniques for analyzing the crystal structure and elemental composition of HM-CDs. XRD, as an important structural tool, is commonly used for effective characterization of CDs as it provides crucial information about their size and purity. The obtained XRD patterns are unique and serve as fingerprints of the periodic atomic arrangement, which can be determined by analyzing the distribution of atoms within the lattice. For instance, XRD analysis of CDs isolated from Actinidia deliciosa ( kiwi) fruit extract through hydrothermal treatment revealed strong broad peaks around 2θ = 28.5° and a weak peak at 2θ = 40.3°, which can be attributed to the diffraction patterns of graphite carbon (002) and (001). The crystal size can then be calculated using the Scherer formula (D = kλ/βcos) by selecting the highest peak displayed in the XRD pattern. However, it is important to note that XRD is not suitable for characterizing amorphous CDs, as it is primarily used for determining key features of CDs with a crystalline structure. Zeta Potential Zeta potential, an essential measurement for evaluating the effective surface charge and quantifying the charge of nanoparticles, plays a crucial role in analyzing the stability of colloidal systems and the surface effects of nanoparticles. This measurement method is particularly important in assessing the toxicity of nanoparticles and their initial absorption by cell membranes. The magnitude of the zeta potential provides valuable insights into the electrical stability of the colloidal system. Research has shown that higher values of zeta potential indicate system stability, while the positive or negative sign of the zeta potential represents the surface charge of the nanoparticles. Nanoparticles with low zeta potential values tend to aggregate together. In a study conducted by the Ramanan group, carbon dots (CDs) were synthesized from algal blooms using microwave irradiation. The researchers successfully obtained highly negative zeta potential values (−22.3±8.39 mV), indicating that the synthesized CDs are negatively charged and rich in carboxyl functional groups. Thus, the measurement of zeta potential provides valuable insights into the stability and aggregation of HM-CDs. Therefore, the characterization of HM-CDs is essential for a deeper understanding of their distinctive properties and behavior. Through analysis of the structural characteristics, one can elucidate their optical, electronic, and chemical properties. The optical properties and surface characteristics of HM-CDs play a crucial role in determining their efficacy in biological systems. Moreover, the morphology and size information of HM-CDs hold significant importance in comprehending their dispersibility and stability in herbal formulations. Spectroscopic techniques such as UV-Vis absorption spectroscopy, fluorescence spectroscopy, and Raman spectroscopy are employed to analyze the optical properties and electronic structure of HM-CDs. Interestingly, UV-Vis spectroscopy is commonly recommended to evaluate the optical properties of HM-CDs as they typically exhibit strong UV absorption, although the absorption peaks may vary. High-performance liquid chromatography and gel electrophoresis are utilized to separate HM-CDs, allowing for the isolation of CDs with different sizes and shapes. It has been confirmed that CDs with sizes of 1.2, 1.5–3, and 3.8 nm emit light in the visible (400–700 nm), UV (350 nm), and near-infrared (NIR) regions, respectively. Therefore, the absorption band peak centered around the UV region of 250–300 nm is often referred to as the typical π-π* transition peak in most CDs. For instance, Lycii fructus CDs synthesized through hydrothermal treatment exhibit a strong absorption peak at 271 nm in the UV region. CDs derived from Borassus flabellifer flower via thermal decomposition exhibit a UV absorption peak at 282 nm, which is attributed to the π-π* transition of aromatic C=C bonds. Moreover, the surface of HM-CDs is typically composed of various functional groups, such as hydroxyl, carboxyl, carbonyl, ether, or epoxy groups, depending on the synthesis techniques used. Fourier-transform infrared spectroscopy (FTIR) can be utilized to determine the surface functional groups of HM-CDs. For instance, the peaks appearing in FTIR spectra of CDs prepared by ultrasonication irradiation of crab shells were at 3398 cm −1 , 2930 cm −1 , 1640 cm −1 , 1563 cm −1 and 1415 cm −1 , which correspond to the stretching vibration of -H stretching, N-H stretching, C-H stretching, C=O stretching, N-H bending and C=C stretching. The advantage of FTIR in characterizing the surface functionalization of carbon dots lies in its affordability, ease of sample preparation, and rapidity. Electron microscopy techniques play a crucial role in the characterization of nanoparticles. Researchers have widely employed scanning electron microscopy (SEM) and transmission electron microscopy (TEM) to visualize HM-CDs and gain insights into their morphology, size, and formation mechanism. SEM involves scanning the surface of HM-CDs with a focused electron beam to generate images. However, since the particle size of HM-CDs is typically smaller than 10 nm, TEM, which utilizes high-energy electron beams to obtain images through the herbal CD samples, offers higher resolution and is more suitable for identifying small-sized particles compared to SEM. For instance, Dager et al utilized TEM to determine the size of HM-CDs synthesized from microwave irradiation of Fenugreek seeds, revealing an average diameter of 4.25 ± 0.56 nm. Moreover, high-resolution transmission electron microscopy (HRTEM) has proven effective in the structural analysis and detection of lattice defects in HM-CDs. HRTEM analysis demonstrated that green CDs derived from tomatoes exhibit a spherical shape, with a size distribution ranging from 5 to 10 nm. X-ray diffraction (XRD) and X-ray photoelectron spectroscopy (XPS) are valuable techniques for analyzing the crystal structure and elemental composition of HM-CDs. XRD, as an important structural tool, is commonly used for effective characterization of CDs as it provides crucial information about their size and purity. The obtained XRD patterns are unique and serve as fingerprints of the periodic atomic arrangement, which can be determined by analyzing the distribution of atoms within the lattice. For instance, XRD analysis of CDs isolated from Actinidia deliciosa ( kiwi) fruit extract through hydrothermal treatment revealed strong broad peaks around 2θ = 28.5° and a weak peak at 2θ = 40.3°, which can be attributed to the diffraction patterns of graphite carbon (002) and (001). The crystal size can then be calculated using the Scherer formula (D = kλ/βcos) by selecting the highest peak displayed in the XRD pattern. However, it is important to note that XRD is not suitable for characterizing amorphous CDs, as it is primarily used for determining key features of CDs with a crystalline structure. Zeta potential, an essential measurement for evaluating the effective surface charge and quantifying the charge of nanoparticles, plays a crucial role in analyzing the stability of colloidal systems and the surface effects of nanoparticles. This measurement method is particularly important in assessing the toxicity of nanoparticles and their initial absorption by cell membranes. The magnitude of the zeta potential provides valuable insights into the electrical stability of the colloidal system. Research has shown that higher values of zeta potential indicate system stability, while the positive or negative sign of the zeta potential represents the surface charge of the nanoparticles. Nanoparticles with low zeta potential values tend to aggregate together. In a study conducted by the Ramanan group, carbon dots (CDs) were synthesized from algal blooms using microwave irradiation. The researchers successfully obtained highly negative zeta potential values (−22.3±8.39 mV), indicating that the synthesized CDs are negatively charged and rich in carboxyl functional groups. Thus, the measurement of zeta potential provides valuable insights into the stability and aggregation of HM-CDs. Therefore, the characterization of HM-CDs is essential for a deeper understanding of their distinctive properties and behavior. Through analysis of the structural characteristics, one can elucidate their optical, electronic, and chemical properties. The optical properties and surface characteristics of HM-CDs play a crucial role in determining their efficacy in biological systems. Moreover, the morphology and size information of HM-CDs hold significant importance in comprehending their dispersibility and stability in herbal formulations. Existing CDs typically can only be used to cure diseases by loading pharmacophores or as drug delivery vehicles, requiring expensive chemical materials and sophisticated modification techniques. Interestingly, the ability of herbal medicines as precursors to overcome these limitations through their specific efficacy has naturally attracted the attention of researchers. The medical therapeutic effects of HM-CDs and their specific functional mechanisms are mainly reflected in the following aspects . Hematological System Hemostasis For the treatment of hemorrhagic disorders, the carbonization of herbal medicines has a lengthy history and a wealth of clinical evidence. Through the study of CDs in herbal medicines such as Schizonepetae Spica Carbonisata , Cirsii Japonici Herba Carbonisata , and Pollen Typhae Carbonisata , it was found that CDs were present in most herbal medicines with low toxicity, excellent water-solubility, and biocompatibility. Most extracted pure CDs showed positive hemostatic effects . However, hemostasis is a complex system involving the interaction of endothelial cells, platelets, coagulation, and fibrinolytic systems . Routinely used coagulation parameters include activated partial thromboplastin time (APTT), prothrombin time (PT), thrombin time (TT), and fibrinogen (FIB). The PT value is related to the overall efficiency of the extrinsic coagulation pathway, while the APTT is tied to the intrinsic coagulation pathway. The common coagulation pathway and the activities that promote the conversion of FIB to fibrin in plasma are associated with TT and FIB levels. Extrinsic Coagulation Pathway and Activation of the FIB System Cirsium Setosum Carbonisata (CSC) has cooling, pain-relieving, and heat-clearing properties in traditional medicine. CSC-CDs synthesized based on CSC exhibited moderate hemostatic activity in a mouse model of tail amputation and liver scratch. The hemostatic effect of CSC-CDs may be related to the stimulation of extrinsic coagulation activity and activation of the FIB system, according to the researchers who evaluated coagulation parameters in mice and observed that mice treated with the CSC-CDs group had lower PT values and higher FIB values. Another study created novel water-soluble CDs using Schizonepetae Herba Carbonisata (SHC) as the sole precursor, and pharmacodynamic experiments revealed that SHC-CDs significantly inhibited hemorrhaging in a rat model of tail amputation and liver scratch. Based on the assessment of coagulation parameters in mice, these effects may be related to extrinsic coagulation activity and activation of the FIB system. Junci Medulla Carbonisata (JMC), a herbal medicine for the treatment of bleeding disorders, has not yet been identified as to its potential bioactive components or its mechanisms of action. Cheng’s group first identified novel CDs (JMC-CDs) in JMC and explored the hemostatic mechanism of JMC-CDs by measuring coagulation parameters in rats. JMC-CDs exhibited excellent hemostatic effects through extrinsic coagulation pathways and activation of the FIB system. Intrinsic Coagulation Pathway and Activation of the FIB System Ptollen Typhae Carbonisata (PTC) has been used as a hemostatic drug. To investigate its hemostatic pharmacological effects and mechanism, Yan et al identified and isolated novel water-soluble CDs (PTC-CDs) from the aqueous solution of PTC. The mouse model of tail amputation and liver scratch demonstrated that PTC-CDs exerted hemostatic effects by stimulating intrinsic coagulation pathways and activating the FIB system. After therapy, the high (15.05 s) and low (16.25 s) dose of PTC-CDs decreased APTT significantly (P < 0.01). The PTC-CDs (high, medium, and low concentration) and hemocoagulase groups showed a significant increase (P < 0.01) in the FIB (2.35, 2.30, 2.18, and 2.35 g/L, respectively) compared with that of the control group (2.08 g/L). Meanwhile, all doses of PTC-CDs and hemocoagulase increased PLT significantly to 1201, 1137, 1140, and 1040 × 10 9 /L, which is in agreement with the results of the bleeding times. Zhao et al extracted a novel chemical from Egg yolk oil (EYO) and obtained EYO-CDs through pyrolysis in another study. EYO alone has no hemostatic effect and is regularly used in clinics to treat both acute and chronic eczema, as well as a variety of burns. Nonetheless, the experimental results point to the stimulation and activation of the intrinsic coagulation and FIB systems as the primary hemostatic mechanisms of the novel drug EYO-CDs. The reason for this phenomenon may be the enhanced absorbency and astringency of the HM-CDs generated after the carbonization of herbal medicines. Numerous loose pores are generated in their structure, which can cause accelerated hemostasis by physical adsorption. In addition, due to the peculiar structure of the carbon surface, it can activate plasma clotting factors and split platelets, releasing platelet factors to promote clotting. Intrinsic and Extrinsic Coagulation Pathways To compare the pharmacodynamic basis of plant and animal materials, PHT-CDs, OJT-CDs, DJT-CDs, and XYT-CDs (prepared from the dried pollen of T. angustifolia L ., dry rhizome node of N. nucifera Gaertn ., dry rhizome node of Cirsium japonicum Fisch. ex DC ., and healthy human hair) were investigated for their hemostatic and anti-inflammatory activities in the pre-trauma phase. The four CDs were found to exhibit similar hemostatic effects and mechanisms. By enhancing the activity of pertinent coagulation components in the plasma via extrinsic coagulation routes, the different concentrations of CDs can drastically reduce the PT values. Meanwhile, different concentrations of CDs also shortened the APTT values, indicating that CDs can also affect the coagulation factors to transform the blood into a hypercoagulable state via the intrinsic coagulation pathway. This study serves as a reference for the development of current hemostatic materials and hemostatic drugs. Notably, the anti-inflammatory effect of XYT-CD prepared from human hair was stronger than the remaining three CDs. Common Coagulation Pathway and Activation of the FIB System Although Cirsii Japonici’s (CJ) hemostatic activity is obvious, its active ingredients and underlying mechanisms are yet unknown. Wang et al synthesized novel Cirsii Japonici -derived CDs (CJ-CDs) and evaluated their pharmacological activity and coagulation parameters in rats. The findings demonstrated that CJ-CDs dramatically reduced bleeding in mice caused by liver scratch or tail amputation, and suggested that the hemostatic effect may involve the common coagulation pathway, the FIB system. Similarly, scholars identified the presence of a different substance, Phellodendri Cortex -derived carbon dots (PCC-CDs), from the aqueous extract of Phellodendri Cortex and investigated the hemostatic activity of PCC-CDs by mouse models of tail amputation and liver scratch. The PCC-CDs group-treated mice exhibited satisfactory hemostatic effects (comparable to hemostatic agents), and for the first time, the hemostatic mechanism was found to be through the activation of the FIB system, thus exerting hemostatic efficacy. Acute trauma hemorrhage may benefit from synthetic PCC-CDs due to their outstanding stability, suitability for long-term storage, and potential as a complementary and alternative therapy. Other In addition to the aforementioned coagulation pathways, CDs can increase drug solubility, thereby facilitating drug absorption to indirectly improve the hemostatic effect. For instance, CDs in charcoal-based medications can increase the solubility of glycosides in water by affecting glycosidic acid. Luo et al investigated the effect of novel water-soluble CDs on baicalin, the main component of Radix Puerariae Carbonisata (RPC), and discovered that pure CDs considerably increased the solubility of baicalin in water. The oral bioavailability of RPC-CD was confirmed to be 1.7 times higher than that of pure baicalin. Furthermore, baicalin in Scutellaria baicalensis undergoes carbonization to become easily absorbed charcoal baicalin, which has a potent hemostatic effect. CDs obtained by high-temperature charring are among the key substances that play the role of hemostats and can be directly applied in the treatment of hemorrhagic symptoms of blood fever. It can also promote the absorption of glycosides and indirectly enhance the hemostatic effect. Tail amputation and liver scratch models are common tools to study the hemostatic activity of drugs. However, Sun et al prepared Schizonepetae Spica Carbonisata (SSC)-derived CDs using an improved pyrolysis method and noted that the original SSC-CDs exhibited favorable hemostatic properties via PLT enhancement. More importantly, this is the first evaluation of the hemostatic bioactivity of SSC using the Deinagkistrodon acutus (D. acutus) venom model. Scholars have investigated the pharmacodynamic basis of the hemostatic effect of charcoal-based drugs by introducing transdisciplinary characterization techniques to herbal medicines. It was found that CDs, present in numerous herbal medicines, hold specific structural characteristics, physicochemical properties, and biological activities. These CDs are derived from a variety of natural products with different biological activities and deserve further investigation. Hypoglycemic and Blood Enrichment In addition to hemostasis, HM-CDs are also suitable for additional hematological diseases. Sun et al developed Jiaosanxian -derived CDs (JSX-CDs) with an average diameter of 4.4–6.4 nm by pyrolysis. JSX-CDs have a large number of surface groups, which contributes to their strong solubility and biological activity. The pharmacodynamic findings indicated that JSX-CDs, a promising new type of hypoglycemic agents, have excellent hypoglycemic efficacy and safe hypoglycemic activity. For hemopoietic effects, Xu et al successfully prepared Jujube-CDs (J-CDs) with excellent anemia therapeutic effects. In both in vitro and in vivo experiments, the synthesized J-CDs were able to promote the self-renewal of erythroid progenitor cells. They also specifically increased the proliferation of erythroid cells by modulating the hypoxia response pathway and increasing the phosphorylation levels of STAT5. Therefore, they have great potential as therapeutic agents for cancer-related anemia. Bacterial Infection Infections caused by fungi, bacteria, parasites, or viruses can cause numerous serious diseases. The identification and inactivation of several bacterial species in photosensitizers (PS) has been done using CDs as a possible fluorescent nanomaterial. , HM-CDs also exhibit powerful photodynamic effects due to their optical properties and have been utilized to destroy bacteria under visible light irradiation. Yoon et al prepared mushroom CDs (MCDs) with intense blue fluorescence under the excitation of 360 nm UV light. Under LED visible light illumination, MCDs can produce ROS (such as OH- and O2-) that can adhere directly to the surface of Escherichia coli (E. coli) and induce cell membrane damage. Lin et al prepared four fluorescent CDs using various herbs (onion, ginger, garlic) and additional natural products (fish) as carbon sources. Onion CDs (O-CDs) demonstrated the strongest antibacterial efficacy against Pseudomonas fragilis of all of them. Persistent endodontic infections (PEIs) associated with Enterococcus faecalis (E. faecalis) biofilms are one of the most common dental lesions, and a study was conducted to prepare Fucoidan (FD)-derived CDs for the treatment of PEIs. By causing the development of both intracellular and extracellular reactive oxygen species and modifying the permeability of the bacteria, in vitro tests have shown that FD-CDs have a favorable inhibitory impact on Enterococcus faecalis and its biofilms. Importantly, FD-CDs penetrated root canals and dentin tubules, and removed E. faecalis biofilms, which has great potential for the treatment of PEIs. In addition, some of the HM-CDs alone could not significantly inhibit bacterial growth. However, as drug delivery systems, when loaded with herbal monomers that likewise failed to appreciably reduce bacterial growth, they demonstrated remarkable dose-dependent antibacterial activity against Gram-negative E. coli and Gram-positive S. aureus pathogens. Inflammation-Related Diseases HM-CDs have also gained extensive research attention in the treatment of inflammatory diseases due to their distinctive advantages, such as great biocompatibility, photostability, and inherent targeting of functional groups. Wang et al synthesized a novel Mulberry Silkworm Cocoon -CDs (MSC- CDs) based on MSC. To assess the anti-inflammatory bioactivity of MSC-CDs, the authors of this work creatively applied three conventional experimental models of inflammation. The results showed that MSC-CDs possess significant anti-inflammatory activity, which may be related to the inhibition of inflammatory factors IL-6 and TNF-α expression, providing a reference for further investigation of the potential pharmacodynamic basis of MSC-CDs. In addition to anti-inflammation, the main aspects of HM-CDs for the treatment of inflammation-related diseases are as follows. Arthritis Arthritis is broadly defined as an inflammatory disease that occurs in the human joints and their surrounding tissues. The charcoal-processed drug AFIC of Aurantii fructus ymulturus (AFI) has long been used to treat inflammatory and metabolic diseases. However, the pharmacodynamic basis and action mechanism of AFIC remain unclear. Wang’s group produced a novel type of carbon dots (AFIC-CDs) through pyrolysis and carbonization. AFIC-CDs effectively attenuate the monosodium urate (MSU) crystal-induced inflammatory response by inhibiting the production of inflammatory factors (IL-1β and TNF-α), playing an influential role in the pathophysiology of acute gouty arthritis. Meanwhile, PLR-CDs reduced IL-1 and TNF levels in a dose-dependent manner, which reduced the severity of joint swelling in gouty arthritis. Epidermal Inflammation The emergence of HM-CDs offers hope for the treatment of psoriasis, a chronic inflammatory skin disease. Zhang et al prepared novel non-toxic Phellodendri Cortex CDs (PCC-CDs). The considerable anti-psoriatic action of PCC-CDs was first demonstrated using a mouse model of psoriasis-like skin. The underlying mechanism may be related to the suppression of M1 polarization of macrophages and the relative promotion of M2 polarization. Systemic inflammatory reactions are generally accompanied by fever or hypothermia, and lipopolysaccharide (LPS)-induced fever is caused by inflammation. Therefore, Wu et al explored the effects of synthetic Lonicerae japonicae Flos (LJF) Carbonisata-CDs on LPS-induced fever and hypothermia models in rats. The experimental results showed that LJFC-CDs significantly attenuated the LPS-induced inflammatory response, as evidenced by the expression of TNF-α, IL-1β, IL-6 and the restoration of normal body temperature. Consequently, LJFC-CDs may have some anti-inflammatory properties and alleviate inflammation-induced fever and hypothermia. Frostbite induced by cold conditions triggers varying degrees of tissue damage, but interventions are lacking. To bridge this gap, Kong et al synthesized Artemisiae Argyi Folium (AAF) Carbonisata-CDs (AAFC-CDs) by pyrolysis. AAFC-CDs ameliorate local inflammation by mediating IL-1β and TNF-α and provide the body with energy to alleviate the fall in blood glucose level caused by frostbite, so as to achieve anti-frostbite effects. In contrast to conventional AAF, isochlorogenic acid is no longer present in AAFC-CDs, but its specific composition has not been identified. The conventional AAF is not suitable for treating frostbite. Therefore, the emergence of AAFC-CDs may extend the practical applications of AAF. Allergic Inflammation Allergies are also frequently linked to inflammation. Scutellariae Radix Carbonisata (SRC) is a traditional medicine that can be used to treat allergic diseases. To elucidate the function and mechanism of the carbonized fraction in SRC, Kong et al isolated novel water-soluble SRC-CDs with particle sizes ranging from 2 to 9 nm from aqueous extracts of Scutellariae Radix Carbonisata . Their anti-inflammatory effects are directly related to their stabilization of mast cell agonism, which may be associated with the reduction of mast cell functional agonism, inhibition of RBL-2H3 cell degranulation, and reduction of histamine and inflammatory factor levels. SRC-CDs are therefore effective in reducing allergic responses. By demonstrating the anti-allergic action of SRC-CDs and the associated mechanisms, researchers have filled a research void and laid the groundwork for future innovative drug development. SRC-CDs may then be used as possible medications to treat allergic conditions. Organ Damage Inflammation The organ damage is accompanied by infiltration of inflammatory factors. Zhao et al found that ASAC-CDs synthesized by Armeniacae Semen Amarum could effectively inhibit the expression levels of inflammatory factors (IL-6, IL-1β, and TNF-α) and exhibited satisfactory anti-inflammatory effects, particularly the high-dose group. Compared to the model group (20.56 ± 1.41 pg/mL, 21.07 ± 2.26 pg/mL, and 69.49 ± 9.62 pg/mL, respectively), treatment with high concentrations of ASAC-CDs (8.13 ± 1.40 pg/mL, 8.53 ± 0.82 pg/mL, and 32.03 ± 5.20 pg/mL) significantly reduced the levels of IL-6, IL-1β, and TNF-α (p < 0.01). To a certain degree, they are able to reduce the increase of neutrophils in the blood and decrease the chemotaxis of neutrophils to inflammatory sites, thereby reducing the release of inflammatory mediators and inhibiting LPS-induced damage and deterioration of lung tissue. In a model of acute kidney injury, another study found that PCC-CDs had a direct renoprotective impact by reversing the rise in serum creatinine (SCR), blood urea nitrogen (BUN), urinary total protein (UTP), and microalbuminuria (MALB). PCC-CDs also attenuated the inflammatory response and thrombocytopenia associated with acute kidney injury, thus exerting a multifaceted effect. Inspired by the above, Wang et al isolated a novel carbon dots (PT-CDs) from Pollen Typhae . Using a rat model of rhabdomyolysis (RM)-induced acute kidney injury (AKI), the authors demonstrated that PT-CDs had significant activity in improving BUN and CRE levels, urine volume, renal index, and histopathological morphology in rats with RM-induced AKI. The intervention of PT-CDs dramatically reduced the degree of inflammatory response and oxidative stress, which may be related to the basal potential mechanism of anti-AKI activity. Additionally, cytotoxicity assays and biosafety assessments demonstrated the high biocompatibility of PT-CDs. Herbal medicines are normally considered to be only for chronic diseases but slow to respond or ineffective for acute injuries. However, in addition to achieving protection of organs such as liver, kidney and lung through anti-inflammation, HM-CDs have confirmed the therapeutic effects of herbs on acute injuries. In a recent study, Paeoniae Radix Alba -derived CDs (PRAC-CDs) can inhibit alanine transaminase (ALT) and acetone transaminase (AST) levels and have a mitigating effect on the rise in TBA and TBIL in a mouse model of acute liver injury. By eliminating free oxygen, preventing lipid peroxidation of hepatocytes, controlling bile acid metabolism, reducing malondialdehyde (MDA) levels and increasing superoxide dismutase (SOD) levels, PRAC-CDs exhibit excellent hepatoprotective effects. The Junci Medulla Carbonisata carbon dots (JMC-CDs) also achieved a similar hepatoprotective effect. In animal models of trauma hemorrhagic and internal hemorrhage caused by Deinagkistrodon acutus venom, the researchers showed that JMC-CDs not only had significant hemostatic effects, but also prevented hemorrhagic liver injury with reduced levels of biochemical indicators of liver injury such as aspartate aminotransferase, alanine amino transferase, alkaline phosphatase, total bilirubin, and direct bilirubin. Inflammation of the Gastrointestinal Tract The pharmacological effects of HM-CDs are also involved in the gastrointestinal system for the treatment of various ulcers, which may also be associated with the inflammatory responses. Recently, Hu et al showed that Radix Sophorae Flavescentis carbonisata (RSFC)-CDs could inhibit ethanol-induced acute gastric ulcers in rats by suppressing the release of TNF-α and IL-6 through downregulation of the NF-κB pathway. Most notably, RSFC has been widely used for the treatment of systemic ulcerative diseases. The authors hypothesized that HM-CDs produced by high-temperature pyrolysis may have inherent biological activity, though the active ingredients were not disclosed. Another study synthesized GRR-CDs using Glycyrrhizae Radix et Rhizoma (GRR) as precursors by an environment-friendly one-step pyrolysis process. GRR-CDs significantly reduced the oxidative damage to the gastric mucosa and tissues caused by alcohol, as well as restored the expression of malondialdehyde, superoxide dismutase, and nitric oxide in the serum and tissues of mice. This suggests that the explicit anti-ulcer activity of GRR-CDs, which provides a fresh perspective on how to investigate the pharmacodynamic basis of GRR. Cancer Herbal medicines offer more potent and distinctive anti-tumor effects, and some of them can be combined with radiotherapy to lessen toxicity and boost efficacy. Similarly, CDs prepared by herbal medicine have great potential for oncology treatments. The strategy of combining herbal medicine and CDs is also expected to reduce anti-cancer side effects, increase tumor accumulation, and enhance therapeutic effectiveness. Inspired by curcumin, Li et al prepared novel CDs (G-CDs) based on Ginger and found that G-CDs could have an extremely strong inhibitory effect on the growth of HepG2 cells by up-regulating the expression of the p53 gene in cancer cells and inducing the level of intracellular ROS. G-CDs also exhibited significant anti-hepatocellular carcinoma activity in vivo, which was able to accumulate at the tumor site through enhanced permeation retention (EPR) effect in solid tumors. Ginsenoside Re -based carbon dots (Re-CDs) with a particle size of 4.6 nm were created in another investigation. Re-CDs have demonstrated reduced toxicity to normal cells and higher efficacy in preventing cancer cell proliferation when compared to APIs. Their cancer-fighting effects were coupled with high levels of ROS and the creation of apoptosis associated with caspase-3. Furthermore, Arul et al prepared nitrogen-doped CDs (N-CDs) by a simple hydrothermal method using Actinidia deliciosa (A. deliciosa) fruit extract as a carbonized precursor and aqueous ammonia as a nitrogen dopant. When tested on mouse fibroblast (L-929) cells and human breast cancer (MCF7) cells, the N-CDs also exhibited some anticancer activity. Diseases Related to Oxidative Stress Normal levels of ROS play a decisive role in cell signaling and homeostasis, but excessive ROS accumulation may lead to oxidative damage, inflammation, various diseases, and cancer. Some HM-CDs also have potent antioxidant activities. Among them, natural gynostemma fluorescent CDs can protect zebrafish from oxidative stress by increasing ROS-related enzymes, thus reducing ROS levels through a compensatory mechanism. As a result, as antioxidants, they are effective in reducing ROS damage in Hela cells and zebrafish. Additionally, utilizing Salvia miltiorrhiza Bunge as a carbon source, Li et al created multifunctional antioxidant CDs. Compared to natural Salvia extracts, the resulting CDs had higher antioxidant capacity and greater ability to scavenge ROS, attenuating abiotic stress in plants and opening up a wide range of potential applications in botany. Subsequently, the group synthesized a Salvia miltiorrhiza Bunge -derived CDs. Under conditions of salt and nutrient deficiency, the abundance of functional groups (-OH and -COOH) on the surface encourages Ca2+ signaling and environmental adaptation in plants, which in turn causes ROS-independent Ca2+ activation in the root system. As such, the CDs can be utilized for crop enhancement as both a ROS scavenger and a simultaneous Ca2+ signaling amplifier. Additional Diseases Antinociceptive Effects Ginger has been used as an analgesic with notable results for more than a thousand years, while its material basis is still unknown. With Zingiber officinale Roscoe (ZR) as the raw material, Zhang et al prepared a revolutionary environmentally friendly CDs (ZR-CDs) utilizing direct pyrolysis. The authors confirmed the significant analgesic activity of ZR-CDs using classical hot-plate, tail-immersion, and acetic acid writhing methods, and demonstrated for the first time that the analgesic effect of ZR-CDs was mediated by an opioid-like mechanism and the regulation of 5-hydroxytryptamine levels in serum. In addition to ZR, a study has prepared non-toxic nanocarrier GRR-CDs using Glycyrrhizae Radix et Rhizoma as the only material and an environmentally friendly, simple and low-cost calcination method, which increased the glycyrrhizic acid (GA) solubility significantly by 27-fold. In both the hot-plate model and the acetic acid-induced writhing model, the GRR-CDs-GA complex showed significantly higher antinociceptive activity compared to the unprocessed GRR-CDs and GA. These results support the promising application of GRR-CDs as a technique to improve the solubility and antinociceptive properties of poorly water-soluble drugs (such as GA). Menopausal Syndrome Glycyrrhizae Radix et Rhizoma (GRR) is frequently used in the treatment of menopausal syndrome (MPS) and other gynecological disorders in addition to the antinociceptive activity. Zhang et al successfully synthesized GRR into GRR-CDs by pyrolysis. The study is the first to demonstrate that GRR-CDs can alleviate MPS by elevating the estradiol (E2) level, decreasing follicle stimulating hormone (FSH) and luteinizing hormone (LH) levels, and raising the degree of uterine atrophy. This not only indicates the potential of GRR-CDs as a drug to alleviate menopausal syndrome and its associated symptoms, but also provides possibility for nanomedicines to treat hormonal disorders. Anxiolytic Effects People are susceptible to developing depression and anxiety disorders in response to stress. Os Draconis (OD) has gained recognition as a medication that has been used for a long time to treat neurological diseases. In order to elucidate the biological basis of the anxiolytic effects of OD, a study isolated the novel OD-CDs obtained from Os Draconis. Interestingly, OD-CDs significantly reduced anxiety in four behavioral tests, including the Open Field Test (OFT), Light/Dark Box Test (LDT), Elevated Plus Maze Test (EPMT), and Novelty-Suppressed Feeding Test (NSFT). The results also imply that OD-CDs mediate the modulation of monoaminergic neurotransmitters and the HPA axis to a certain extent, although additional research is required to pinpoint the precise processes. Given that OD-CDs exhibit observable anxiolytic effects, this supports their development as novel anxiolytic agents that merit additional study. In summary, HM-CDs, as an emerging nanomaterial, have been widely used in the medical field for their remarkable therapeutic effects due to their excellent photoluminescence capabilities, superior chemical stability and low toxicity, water dispersibility, and biocompatibility. It is noteworthy that the study of the auto-biological activity of HM-CDs has received increasing attention, which is anticipated to reveal their various pharmacological and active effects. However, the therapeutic mechanisms of HM-CDs have not been thoroughly investigated yet, which need to be further explored. In addition, as nanomaterials, elucidating the metabolic processes of HM-CDs in vivo is another major challenge. Hemostasis For the treatment of hemorrhagic disorders, the carbonization of herbal medicines has a lengthy history and a wealth of clinical evidence. Through the study of CDs in herbal medicines such as Schizonepetae Spica Carbonisata , Cirsii Japonici Herba Carbonisata , and Pollen Typhae Carbonisata , it was found that CDs were present in most herbal medicines with low toxicity, excellent water-solubility, and biocompatibility. Most extracted pure CDs showed positive hemostatic effects . However, hemostasis is a complex system involving the interaction of endothelial cells, platelets, coagulation, and fibrinolytic systems . Routinely used coagulation parameters include activated partial thromboplastin time (APTT), prothrombin time (PT), thrombin time (TT), and fibrinogen (FIB). The PT value is related to the overall efficiency of the extrinsic coagulation pathway, while the APTT is tied to the intrinsic coagulation pathway. The common coagulation pathway and the activities that promote the conversion of FIB to fibrin in plasma are associated with TT and FIB levels. Extrinsic Coagulation Pathway and Activation of the FIB System Cirsium Setosum Carbonisata (CSC) has cooling, pain-relieving, and heat-clearing properties in traditional medicine. CSC-CDs synthesized based on CSC exhibited moderate hemostatic activity in a mouse model of tail amputation and liver scratch. The hemostatic effect of CSC-CDs may be related to the stimulation of extrinsic coagulation activity and activation of the FIB system, according to the researchers who evaluated coagulation parameters in mice and observed that mice treated with the CSC-CDs group had lower PT values and higher FIB values. Another study created novel water-soluble CDs using Schizonepetae Herba Carbonisata (SHC) as the sole precursor, and pharmacodynamic experiments revealed that SHC-CDs significantly inhibited hemorrhaging in a rat model of tail amputation and liver scratch. Based on the assessment of coagulation parameters in mice, these effects may be related to extrinsic coagulation activity and activation of the FIB system. Junci Medulla Carbonisata (JMC), a herbal medicine for the treatment of bleeding disorders, has not yet been identified as to its potential bioactive components or its mechanisms of action. Cheng’s group first identified novel CDs (JMC-CDs) in JMC and explored the hemostatic mechanism of JMC-CDs by measuring coagulation parameters in rats. JMC-CDs exhibited excellent hemostatic effects through extrinsic coagulation pathways and activation of the FIB system. Intrinsic Coagulation Pathway and Activation of the FIB System Ptollen Typhae Carbonisata (PTC) has been used as a hemostatic drug. To investigate its hemostatic pharmacological effects and mechanism, Yan et al identified and isolated novel water-soluble CDs (PTC-CDs) from the aqueous solution of PTC. The mouse model of tail amputation and liver scratch demonstrated that PTC-CDs exerted hemostatic effects by stimulating intrinsic coagulation pathways and activating the FIB system. After therapy, the high (15.05 s) and low (16.25 s) dose of PTC-CDs decreased APTT significantly (P < 0.01). The PTC-CDs (high, medium, and low concentration) and hemocoagulase groups showed a significant increase (P < 0.01) in the FIB (2.35, 2.30, 2.18, and 2.35 g/L, respectively) compared with that of the control group (2.08 g/L). Meanwhile, all doses of PTC-CDs and hemocoagulase increased PLT significantly to 1201, 1137, 1140, and 1040 × 10 9 /L, which is in agreement with the results of the bleeding times. Zhao et al extracted a novel chemical from Egg yolk oil (EYO) and obtained EYO-CDs through pyrolysis in another study. EYO alone has no hemostatic effect and is regularly used in clinics to treat both acute and chronic eczema, as well as a variety of burns. Nonetheless, the experimental results point to the stimulation and activation of the intrinsic coagulation and FIB systems as the primary hemostatic mechanisms of the novel drug EYO-CDs. The reason for this phenomenon may be the enhanced absorbency and astringency of the HM-CDs generated after the carbonization of herbal medicines. Numerous loose pores are generated in their structure, which can cause accelerated hemostasis by physical adsorption. In addition, due to the peculiar structure of the carbon surface, it can activate plasma clotting factors and split platelets, releasing platelet factors to promote clotting. Intrinsic and Extrinsic Coagulation Pathways To compare the pharmacodynamic basis of plant and animal materials, PHT-CDs, OJT-CDs, DJT-CDs, and XYT-CDs (prepared from the dried pollen of T. angustifolia L ., dry rhizome node of N. nucifera Gaertn ., dry rhizome node of Cirsium japonicum Fisch. ex DC ., and healthy human hair) were investigated for their hemostatic and anti-inflammatory activities in the pre-trauma phase. The four CDs were found to exhibit similar hemostatic effects and mechanisms. By enhancing the activity of pertinent coagulation components in the plasma via extrinsic coagulation routes, the different concentrations of CDs can drastically reduce the PT values. Meanwhile, different concentrations of CDs also shortened the APTT values, indicating that CDs can also affect the coagulation factors to transform the blood into a hypercoagulable state via the intrinsic coagulation pathway. This study serves as a reference for the development of current hemostatic materials and hemostatic drugs. Notably, the anti-inflammatory effect of XYT-CD prepared from human hair was stronger than the remaining three CDs. Common Coagulation Pathway and Activation of the FIB System Although Cirsii Japonici’s (CJ) hemostatic activity is obvious, its active ingredients and underlying mechanisms are yet unknown. Wang et al synthesized novel Cirsii Japonici -derived CDs (CJ-CDs) and evaluated their pharmacological activity and coagulation parameters in rats. The findings demonstrated that CJ-CDs dramatically reduced bleeding in mice caused by liver scratch or tail amputation, and suggested that the hemostatic effect may involve the common coagulation pathway, the FIB system. Similarly, scholars identified the presence of a different substance, Phellodendri Cortex -derived carbon dots (PCC-CDs), from the aqueous extract of Phellodendri Cortex and investigated the hemostatic activity of PCC-CDs by mouse models of tail amputation and liver scratch. The PCC-CDs group-treated mice exhibited satisfactory hemostatic effects (comparable to hemostatic agents), and for the first time, the hemostatic mechanism was found to be through the activation of the FIB system, thus exerting hemostatic efficacy. Acute trauma hemorrhage may benefit from synthetic PCC-CDs due to their outstanding stability, suitability for long-term storage, and potential as a complementary and alternative therapy. Other In addition to the aforementioned coagulation pathways, CDs can increase drug solubility, thereby facilitating drug absorption to indirectly improve the hemostatic effect. For instance, CDs in charcoal-based medications can increase the solubility of glycosides in water by affecting glycosidic acid. Luo et al investigated the effect of novel water-soluble CDs on baicalin, the main component of Radix Puerariae Carbonisata (RPC), and discovered that pure CDs considerably increased the solubility of baicalin in water. The oral bioavailability of RPC-CD was confirmed to be 1.7 times higher than that of pure baicalin. Furthermore, baicalin in Scutellaria baicalensis undergoes carbonization to become easily absorbed charcoal baicalin, which has a potent hemostatic effect. CDs obtained by high-temperature charring are among the key substances that play the role of hemostats and can be directly applied in the treatment of hemorrhagic symptoms of blood fever. It can also promote the absorption of glycosides and indirectly enhance the hemostatic effect. Tail amputation and liver scratch models are common tools to study the hemostatic activity of drugs. However, Sun et al prepared Schizonepetae Spica Carbonisata (SSC)-derived CDs using an improved pyrolysis method and noted that the original SSC-CDs exhibited favorable hemostatic properties via PLT enhancement. More importantly, this is the first evaluation of the hemostatic bioactivity of SSC using the Deinagkistrodon acutus (D. acutus) venom model. Scholars have investigated the pharmacodynamic basis of the hemostatic effect of charcoal-based drugs by introducing transdisciplinary characterization techniques to herbal medicines. It was found that CDs, present in numerous herbal medicines, hold specific structural characteristics, physicochemical properties, and biological activities. These CDs are derived from a variety of natural products with different biological activities and deserve further investigation. Hypoglycemic and Blood Enrichment In addition to hemostasis, HM-CDs are also suitable for additional hematological diseases. Sun et al developed Jiaosanxian -derived CDs (JSX-CDs) with an average diameter of 4.4–6.4 nm by pyrolysis. JSX-CDs have a large number of surface groups, which contributes to their strong solubility and biological activity. The pharmacodynamic findings indicated that JSX-CDs, a promising new type of hypoglycemic agents, have excellent hypoglycemic efficacy and safe hypoglycemic activity. For hemopoietic effects, Xu et al successfully prepared Jujube-CDs (J-CDs) with excellent anemia therapeutic effects. In both in vitro and in vivo experiments, the synthesized J-CDs were able to promote the self-renewal of erythroid progenitor cells. They also specifically increased the proliferation of erythroid cells by modulating the hypoxia response pathway and increasing the phosphorylation levels of STAT5. Therefore, they have great potential as therapeutic agents for cancer-related anemia. For the treatment of hemorrhagic disorders, the carbonization of herbal medicines has a lengthy history and a wealth of clinical evidence. Through the study of CDs in herbal medicines such as Schizonepetae Spica Carbonisata , Cirsii Japonici Herba Carbonisata , and Pollen Typhae Carbonisata , it was found that CDs were present in most herbal medicines with low toxicity, excellent water-solubility, and biocompatibility. Most extracted pure CDs showed positive hemostatic effects . However, hemostasis is a complex system involving the interaction of endothelial cells, platelets, coagulation, and fibrinolytic systems . Routinely used coagulation parameters include activated partial thromboplastin time (APTT), prothrombin time (PT), thrombin time (TT), and fibrinogen (FIB). The PT value is related to the overall efficiency of the extrinsic coagulation pathway, while the APTT is tied to the intrinsic coagulation pathway. The common coagulation pathway and the activities that promote the conversion of FIB to fibrin in plasma are associated with TT and FIB levels. Extrinsic Coagulation Pathway and Activation of the FIB System Cirsium Setosum Carbonisata (CSC) has cooling, pain-relieving, and heat-clearing properties in traditional medicine. CSC-CDs synthesized based on CSC exhibited moderate hemostatic activity in a mouse model of tail amputation and liver scratch. The hemostatic effect of CSC-CDs may be related to the stimulation of extrinsic coagulation activity and activation of the FIB system, according to the researchers who evaluated coagulation parameters in mice and observed that mice treated with the CSC-CDs group had lower PT values and higher FIB values. Another study created novel water-soluble CDs using Schizonepetae Herba Carbonisata (SHC) as the sole precursor, and pharmacodynamic experiments revealed that SHC-CDs significantly inhibited hemorrhaging in a rat model of tail amputation and liver scratch. Based on the assessment of coagulation parameters in mice, these effects may be related to extrinsic coagulation activity and activation of the FIB system. Junci Medulla Carbonisata (JMC), a herbal medicine for the treatment of bleeding disorders, has not yet been identified as to its potential bioactive components or its mechanisms of action. Cheng’s group first identified novel CDs (JMC-CDs) in JMC and explored the hemostatic mechanism of JMC-CDs by measuring coagulation parameters in rats. JMC-CDs exhibited excellent hemostatic effects through extrinsic coagulation pathways and activation of the FIB system. Intrinsic Coagulation Pathway and Activation of the FIB System Ptollen Typhae Carbonisata (PTC) has been used as a hemostatic drug. To investigate its hemostatic pharmacological effects and mechanism, Yan et al identified and isolated novel water-soluble CDs (PTC-CDs) from the aqueous solution of PTC. The mouse model of tail amputation and liver scratch demonstrated that PTC-CDs exerted hemostatic effects by stimulating intrinsic coagulation pathways and activating the FIB system. After therapy, the high (15.05 s) and low (16.25 s) dose of PTC-CDs decreased APTT significantly (P < 0.01). The PTC-CDs (high, medium, and low concentration) and hemocoagulase groups showed a significant increase (P < 0.01) in the FIB (2.35, 2.30, 2.18, and 2.35 g/L, respectively) compared with that of the control group (2.08 g/L). Meanwhile, all doses of PTC-CDs and hemocoagulase increased PLT significantly to 1201, 1137, 1140, and 1040 × 10 9 /L, which is in agreement with the results of the bleeding times. Zhao et al extracted a novel chemical from Egg yolk oil (EYO) and obtained EYO-CDs through pyrolysis in another study. EYO alone has no hemostatic effect and is regularly used in clinics to treat both acute and chronic eczema, as well as a variety of burns. Nonetheless, the experimental results point to the stimulation and activation of the intrinsic coagulation and FIB systems as the primary hemostatic mechanisms of the novel drug EYO-CDs. The reason for this phenomenon may be the enhanced absorbency and astringency of the HM-CDs generated after the carbonization of herbal medicines. Numerous loose pores are generated in their structure, which can cause accelerated hemostasis by physical adsorption. In addition, due to the peculiar structure of the carbon surface, it can activate plasma clotting factors and split platelets, releasing platelet factors to promote clotting. Intrinsic and Extrinsic Coagulation Pathways To compare the pharmacodynamic basis of plant and animal materials, PHT-CDs, OJT-CDs, DJT-CDs, and XYT-CDs (prepared from the dried pollen of T. angustifolia L ., dry rhizome node of N. nucifera Gaertn ., dry rhizome node of Cirsium japonicum Fisch. ex DC ., and healthy human hair) were investigated for their hemostatic and anti-inflammatory activities in the pre-trauma phase. The four CDs were found to exhibit similar hemostatic effects and mechanisms. By enhancing the activity of pertinent coagulation components in the plasma via extrinsic coagulation routes, the different concentrations of CDs can drastically reduce the PT values. Meanwhile, different concentrations of CDs also shortened the APTT values, indicating that CDs can also affect the coagulation factors to transform the blood into a hypercoagulable state via the intrinsic coagulation pathway. This study serves as a reference for the development of current hemostatic materials and hemostatic drugs. Notably, the anti-inflammatory effect of XYT-CD prepared from human hair was stronger than the remaining three CDs. Common Coagulation Pathway and Activation of the FIB System Although Cirsii Japonici’s (CJ) hemostatic activity is obvious, its active ingredients and underlying mechanisms are yet unknown. Wang et al synthesized novel Cirsii Japonici -derived CDs (CJ-CDs) and evaluated their pharmacological activity and coagulation parameters in rats. The findings demonstrated that CJ-CDs dramatically reduced bleeding in mice caused by liver scratch or tail amputation, and suggested that the hemostatic effect may involve the common coagulation pathway, the FIB system. Similarly, scholars identified the presence of a different substance, Phellodendri Cortex -derived carbon dots (PCC-CDs), from the aqueous extract of Phellodendri Cortex and investigated the hemostatic activity of PCC-CDs by mouse models of tail amputation and liver scratch. The PCC-CDs group-treated mice exhibited satisfactory hemostatic effects (comparable to hemostatic agents), and for the first time, the hemostatic mechanism was found to be through the activation of the FIB system, thus exerting hemostatic efficacy. Acute trauma hemorrhage may benefit from synthetic PCC-CDs due to their outstanding stability, suitability for long-term storage, and potential as a complementary and alternative therapy. Other In addition to the aforementioned coagulation pathways, CDs can increase drug solubility, thereby facilitating drug absorption to indirectly improve the hemostatic effect. For instance, CDs in charcoal-based medications can increase the solubility of glycosides in water by affecting glycosidic acid. Luo et al investigated the effect of novel water-soluble CDs on baicalin, the main component of Radix Puerariae Carbonisata (RPC), and discovered that pure CDs considerably increased the solubility of baicalin in water. The oral bioavailability of RPC-CD was confirmed to be 1.7 times higher than that of pure baicalin. Furthermore, baicalin in Scutellaria baicalensis undergoes carbonization to become easily absorbed charcoal baicalin, which has a potent hemostatic effect. CDs obtained by high-temperature charring are among the key substances that play the role of hemostats and can be directly applied in the treatment of hemorrhagic symptoms of blood fever. It can also promote the absorption of glycosides and indirectly enhance the hemostatic effect. Tail amputation and liver scratch models are common tools to study the hemostatic activity of drugs. However, Sun et al prepared Schizonepetae Spica Carbonisata (SSC)-derived CDs using an improved pyrolysis method and noted that the original SSC-CDs exhibited favorable hemostatic properties via PLT enhancement. More importantly, this is the first evaluation of the hemostatic bioactivity of SSC using the Deinagkistrodon acutus (D. acutus) venom model. Scholars have investigated the pharmacodynamic basis of the hemostatic effect of charcoal-based drugs by introducing transdisciplinary characterization techniques to herbal medicines. It was found that CDs, present in numerous herbal medicines, hold specific structural characteristics, physicochemical properties, and biological activities. These CDs are derived from a variety of natural products with different biological activities and deserve further investigation. Cirsium Setosum Carbonisata (CSC) has cooling, pain-relieving, and heat-clearing properties in traditional medicine. CSC-CDs synthesized based on CSC exhibited moderate hemostatic activity in a mouse model of tail amputation and liver scratch. The hemostatic effect of CSC-CDs may be related to the stimulation of extrinsic coagulation activity and activation of the FIB system, according to the researchers who evaluated coagulation parameters in mice and observed that mice treated with the CSC-CDs group had lower PT values and higher FIB values. Another study created novel water-soluble CDs using Schizonepetae Herba Carbonisata (SHC) as the sole precursor, and pharmacodynamic experiments revealed that SHC-CDs significantly inhibited hemorrhaging in a rat model of tail amputation and liver scratch. Based on the assessment of coagulation parameters in mice, these effects may be related to extrinsic coagulation activity and activation of the FIB system. Junci Medulla Carbonisata (JMC), a herbal medicine for the treatment of bleeding disorders, has not yet been identified as to its potential bioactive components or its mechanisms of action. Cheng’s group first identified novel CDs (JMC-CDs) in JMC and explored the hemostatic mechanism of JMC-CDs by measuring coagulation parameters in rats. JMC-CDs exhibited excellent hemostatic effects through extrinsic coagulation pathways and activation of the FIB system. Ptollen Typhae Carbonisata (PTC) has been used as a hemostatic drug. To investigate its hemostatic pharmacological effects and mechanism, Yan et al identified and isolated novel water-soluble CDs (PTC-CDs) from the aqueous solution of PTC. The mouse model of tail amputation and liver scratch demonstrated that PTC-CDs exerted hemostatic effects by stimulating intrinsic coagulation pathways and activating the FIB system. After therapy, the high (15.05 s) and low (16.25 s) dose of PTC-CDs decreased APTT significantly (P < 0.01). The PTC-CDs (high, medium, and low concentration) and hemocoagulase groups showed a significant increase (P < 0.01) in the FIB (2.35, 2.30, 2.18, and 2.35 g/L, respectively) compared with that of the control group (2.08 g/L). Meanwhile, all doses of PTC-CDs and hemocoagulase increased PLT significantly to 1201, 1137, 1140, and 1040 × 10 9 /L, which is in agreement with the results of the bleeding times. Zhao et al extracted a novel chemical from Egg yolk oil (EYO) and obtained EYO-CDs through pyrolysis in another study. EYO alone has no hemostatic effect and is regularly used in clinics to treat both acute and chronic eczema, as well as a variety of burns. Nonetheless, the experimental results point to the stimulation and activation of the intrinsic coagulation and FIB systems as the primary hemostatic mechanisms of the novel drug EYO-CDs. The reason for this phenomenon may be the enhanced absorbency and astringency of the HM-CDs generated after the carbonization of herbal medicines. Numerous loose pores are generated in their structure, which can cause accelerated hemostasis by physical adsorption. In addition, due to the peculiar structure of the carbon surface, it can activate plasma clotting factors and split platelets, releasing platelet factors to promote clotting. To compare the pharmacodynamic basis of plant and animal materials, PHT-CDs, OJT-CDs, DJT-CDs, and XYT-CDs (prepared from the dried pollen of T. angustifolia L ., dry rhizome node of N. nucifera Gaertn ., dry rhizome node of Cirsium japonicum Fisch. ex DC ., and healthy human hair) were investigated for their hemostatic and anti-inflammatory activities in the pre-trauma phase. The four CDs were found to exhibit similar hemostatic effects and mechanisms. By enhancing the activity of pertinent coagulation components in the plasma via extrinsic coagulation routes, the different concentrations of CDs can drastically reduce the PT values. Meanwhile, different concentrations of CDs also shortened the APTT values, indicating that CDs can also affect the coagulation factors to transform the blood into a hypercoagulable state via the intrinsic coagulation pathway. This study serves as a reference for the development of current hemostatic materials and hemostatic drugs. Notably, the anti-inflammatory effect of XYT-CD prepared from human hair was stronger than the remaining three CDs. Although Cirsii Japonici’s (CJ) hemostatic activity is obvious, its active ingredients and underlying mechanisms are yet unknown. Wang et al synthesized novel Cirsii Japonici -derived CDs (CJ-CDs) and evaluated their pharmacological activity and coagulation parameters in rats. The findings demonstrated that CJ-CDs dramatically reduced bleeding in mice caused by liver scratch or tail amputation, and suggested that the hemostatic effect may involve the common coagulation pathway, the FIB system. Similarly, scholars identified the presence of a different substance, Phellodendri Cortex -derived carbon dots (PCC-CDs), from the aqueous extract of Phellodendri Cortex and investigated the hemostatic activity of PCC-CDs by mouse models of tail amputation and liver scratch. The PCC-CDs group-treated mice exhibited satisfactory hemostatic effects (comparable to hemostatic agents), and for the first time, the hemostatic mechanism was found to be through the activation of the FIB system, thus exerting hemostatic efficacy. Acute trauma hemorrhage may benefit from synthetic PCC-CDs due to their outstanding stability, suitability for long-term storage, and potential as a complementary and alternative therapy. In addition to the aforementioned coagulation pathways, CDs can increase drug solubility, thereby facilitating drug absorption to indirectly improve the hemostatic effect. For instance, CDs in charcoal-based medications can increase the solubility of glycosides in water by affecting glycosidic acid. Luo et al investigated the effect of novel water-soluble CDs on baicalin, the main component of Radix Puerariae Carbonisata (RPC), and discovered that pure CDs considerably increased the solubility of baicalin in water. The oral bioavailability of RPC-CD was confirmed to be 1.7 times higher than that of pure baicalin. Furthermore, baicalin in Scutellaria baicalensis undergoes carbonization to become easily absorbed charcoal baicalin, which has a potent hemostatic effect. CDs obtained by high-temperature charring are among the key substances that play the role of hemostats and can be directly applied in the treatment of hemorrhagic symptoms of blood fever. It can also promote the absorption of glycosides and indirectly enhance the hemostatic effect. Tail amputation and liver scratch models are common tools to study the hemostatic activity of drugs. However, Sun et al prepared Schizonepetae Spica Carbonisata (SSC)-derived CDs using an improved pyrolysis method and noted that the original SSC-CDs exhibited favorable hemostatic properties via PLT enhancement. More importantly, this is the first evaluation of the hemostatic bioactivity of SSC using the Deinagkistrodon acutus (D. acutus) venom model. Scholars have investigated the pharmacodynamic basis of the hemostatic effect of charcoal-based drugs by introducing transdisciplinary characterization techniques to herbal medicines. It was found that CDs, present in numerous herbal medicines, hold specific structural characteristics, physicochemical properties, and biological activities. These CDs are derived from a variety of natural products with different biological activities and deserve further investigation. In addition to hemostasis, HM-CDs are also suitable for additional hematological diseases. Sun et al developed Jiaosanxian -derived CDs (JSX-CDs) with an average diameter of 4.4–6.4 nm by pyrolysis. JSX-CDs have a large number of surface groups, which contributes to their strong solubility and biological activity. The pharmacodynamic findings indicated that JSX-CDs, a promising new type of hypoglycemic agents, have excellent hypoglycemic efficacy and safe hypoglycemic activity. For hemopoietic effects, Xu et al successfully prepared Jujube-CDs (J-CDs) with excellent anemia therapeutic effects. In both in vitro and in vivo experiments, the synthesized J-CDs were able to promote the self-renewal of erythroid progenitor cells. They also specifically increased the proliferation of erythroid cells by modulating the hypoxia response pathway and increasing the phosphorylation levels of STAT5. Therefore, they have great potential as therapeutic agents for cancer-related anemia. Infections caused by fungi, bacteria, parasites, or viruses can cause numerous serious diseases. The identification and inactivation of several bacterial species in photosensitizers (PS) has been done using CDs as a possible fluorescent nanomaterial. , HM-CDs also exhibit powerful photodynamic effects due to their optical properties and have been utilized to destroy bacteria under visible light irradiation. Yoon et al prepared mushroom CDs (MCDs) with intense blue fluorescence under the excitation of 360 nm UV light. Under LED visible light illumination, MCDs can produce ROS (such as OH- and O2-) that can adhere directly to the surface of Escherichia coli (E. coli) and induce cell membrane damage. Lin et al prepared four fluorescent CDs using various herbs (onion, ginger, garlic) and additional natural products (fish) as carbon sources. Onion CDs (O-CDs) demonstrated the strongest antibacterial efficacy against Pseudomonas fragilis of all of them. Persistent endodontic infections (PEIs) associated with Enterococcus faecalis (E. faecalis) biofilms are one of the most common dental lesions, and a study was conducted to prepare Fucoidan (FD)-derived CDs for the treatment of PEIs. By causing the development of both intracellular and extracellular reactive oxygen species and modifying the permeability of the bacteria, in vitro tests have shown that FD-CDs have a favorable inhibitory impact on Enterococcus faecalis and its biofilms. Importantly, FD-CDs penetrated root canals and dentin tubules, and removed E. faecalis biofilms, which has great potential for the treatment of PEIs. In addition, some of the HM-CDs alone could not significantly inhibit bacterial growth. However, as drug delivery systems, when loaded with herbal monomers that likewise failed to appreciably reduce bacterial growth, they demonstrated remarkable dose-dependent antibacterial activity against Gram-negative E. coli and Gram-positive S. aureus pathogens. HM-CDs have also gained extensive research attention in the treatment of inflammatory diseases due to their distinctive advantages, such as great biocompatibility, photostability, and inherent targeting of functional groups. Wang et al synthesized a novel Mulberry Silkworm Cocoon -CDs (MSC- CDs) based on MSC. To assess the anti-inflammatory bioactivity of MSC-CDs, the authors of this work creatively applied three conventional experimental models of inflammation. The results showed that MSC-CDs possess significant anti-inflammatory activity, which may be related to the inhibition of inflammatory factors IL-6 and TNF-α expression, providing a reference for further investigation of the potential pharmacodynamic basis of MSC-CDs. In addition to anti-inflammation, the main aspects of HM-CDs for the treatment of inflammation-related diseases are as follows. Arthritis Arthritis is broadly defined as an inflammatory disease that occurs in the human joints and their surrounding tissues. The charcoal-processed drug AFIC of Aurantii fructus ymulturus (AFI) has long been used to treat inflammatory and metabolic diseases. However, the pharmacodynamic basis and action mechanism of AFIC remain unclear. Wang’s group produced a novel type of carbon dots (AFIC-CDs) through pyrolysis and carbonization. AFIC-CDs effectively attenuate the monosodium urate (MSU) crystal-induced inflammatory response by inhibiting the production of inflammatory factors (IL-1β and TNF-α), playing an influential role in the pathophysiology of acute gouty arthritis. Meanwhile, PLR-CDs reduced IL-1 and TNF levels in a dose-dependent manner, which reduced the severity of joint swelling in gouty arthritis. Epidermal Inflammation The emergence of HM-CDs offers hope for the treatment of psoriasis, a chronic inflammatory skin disease. Zhang et al prepared novel non-toxic Phellodendri Cortex CDs (PCC-CDs). The considerable anti-psoriatic action of PCC-CDs was first demonstrated using a mouse model of psoriasis-like skin. The underlying mechanism may be related to the suppression of M1 polarization of macrophages and the relative promotion of M2 polarization. Systemic inflammatory reactions are generally accompanied by fever or hypothermia, and lipopolysaccharide (LPS)-induced fever is caused by inflammation. Therefore, Wu et al explored the effects of synthetic Lonicerae japonicae Flos (LJF) Carbonisata-CDs on LPS-induced fever and hypothermia models in rats. The experimental results showed that LJFC-CDs significantly attenuated the LPS-induced inflammatory response, as evidenced by the expression of TNF-α, IL-1β, IL-6 and the restoration of normal body temperature. Consequently, LJFC-CDs may have some anti-inflammatory properties and alleviate inflammation-induced fever and hypothermia. Frostbite induced by cold conditions triggers varying degrees of tissue damage, but interventions are lacking. To bridge this gap, Kong et al synthesized Artemisiae Argyi Folium (AAF) Carbonisata-CDs (AAFC-CDs) by pyrolysis. AAFC-CDs ameliorate local inflammation by mediating IL-1β and TNF-α and provide the body with energy to alleviate the fall in blood glucose level caused by frostbite, so as to achieve anti-frostbite effects. In contrast to conventional AAF, isochlorogenic acid is no longer present in AAFC-CDs, but its specific composition has not been identified. The conventional AAF is not suitable for treating frostbite. Therefore, the emergence of AAFC-CDs may extend the practical applications of AAF. Allergic Inflammation Allergies are also frequently linked to inflammation. Scutellariae Radix Carbonisata (SRC) is a traditional medicine that can be used to treat allergic diseases. To elucidate the function and mechanism of the carbonized fraction in SRC, Kong et al isolated novel water-soluble SRC-CDs with particle sizes ranging from 2 to 9 nm from aqueous extracts of Scutellariae Radix Carbonisata . Their anti-inflammatory effects are directly related to their stabilization of mast cell agonism, which may be associated with the reduction of mast cell functional agonism, inhibition of RBL-2H3 cell degranulation, and reduction of histamine and inflammatory factor levels. SRC-CDs are therefore effective in reducing allergic responses. By demonstrating the anti-allergic action of SRC-CDs and the associated mechanisms, researchers have filled a research void and laid the groundwork for future innovative drug development. SRC-CDs may then be used as possible medications to treat allergic conditions. Organ Damage Inflammation The organ damage is accompanied by infiltration of inflammatory factors. Zhao et al found that ASAC-CDs synthesized by Armeniacae Semen Amarum could effectively inhibit the expression levels of inflammatory factors (IL-6, IL-1β, and TNF-α) and exhibited satisfactory anti-inflammatory effects, particularly the high-dose group. Compared to the model group (20.56 ± 1.41 pg/mL, 21.07 ± 2.26 pg/mL, and 69.49 ± 9.62 pg/mL, respectively), treatment with high concentrations of ASAC-CDs (8.13 ± 1.40 pg/mL, 8.53 ± 0.82 pg/mL, and 32.03 ± 5.20 pg/mL) significantly reduced the levels of IL-6, IL-1β, and TNF-α (p < 0.01). To a certain degree, they are able to reduce the increase of neutrophils in the blood and decrease the chemotaxis of neutrophils to inflammatory sites, thereby reducing the release of inflammatory mediators and inhibiting LPS-induced damage and deterioration of lung tissue. In a model of acute kidney injury, another study found that PCC-CDs had a direct renoprotective impact by reversing the rise in serum creatinine (SCR), blood urea nitrogen (BUN), urinary total protein (UTP), and microalbuminuria (MALB). PCC-CDs also attenuated the inflammatory response and thrombocytopenia associated with acute kidney injury, thus exerting a multifaceted effect. Inspired by the above, Wang et al isolated a novel carbon dots (PT-CDs) from Pollen Typhae . Using a rat model of rhabdomyolysis (RM)-induced acute kidney injury (AKI), the authors demonstrated that PT-CDs had significant activity in improving BUN and CRE levels, urine volume, renal index, and histopathological morphology in rats with RM-induced AKI. The intervention of PT-CDs dramatically reduced the degree of inflammatory response and oxidative stress, which may be related to the basal potential mechanism of anti-AKI activity. Additionally, cytotoxicity assays and biosafety assessments demonstrated the high biocompatibility of PT-CDs. Herbal medicines are normally considered to be only for chronic diseases but slow to respond or ineffective for acute injuries. However, in addition to achieving protection of organs such as liver, kidney and lung through anti-inflammation, HM-CDs have confirmed the therapeutic effects of herbs on acute injuries. In a recent study, Paeoniae Radix Alba -derived CDs (PRAC-CDs) can inhibit alanine transaminase (ALT) and acetone transaminase (AST) levels and have a mitigating effect on the rise in TBA and TBIL in a mouse model of acute liver injury. By eliminating free oxygen, preventing lipid peroxidation of hepatocytes, controlling bile acid metabolism, reducing malondialdehyde (MDA) levels and increasing superoxide dismutase (SOD) levels, PRAC-CDs exhibit excellent hepatoprotective effects. The Junci Medulla Carbonisata carbon dots (JMC-CDs) also achieved a similar hepatoprotective effect. In animal models of trauma hemorrhagic and internal hemorrhage caused by Deinagkistrodon acutus venom, the researchers showed that JMC-CDs not only had significant hemostatic effects, but also prevented hemorrhagic liver injury with reduced levels of biochemical indicators of liver injury such as aspartate aminotransferase, alanine amino transferase, alkaline phosphatase, total bilirubin, and direct bilirubin. Inflammation of the Gastrointestinal Tract The pharmacological effects of HM-CDs are also involved in the gastrointestinal system for the treatment of various ulcers, which may also be associated with the inflammatory responses. Recently, Hu et al showed that Radix Sophorae Flavescentis carbonisata (RSFC)-CDs could inhibit ethanol-induced acute gastric ulcers in rats by suppressing the release of TNF-α and IL-6 through downregulation of the NF-κB pathway. Most notably, RSFC has been widely used for the treatment of systemic ulcerative diseases. The authors hypothesized that HM-CDs produced by high-temperature pyrolysis may have inherent biological activity, though the active ingredients were not disclosed. Another study synthesized GRR-CDs using Glycyrrhizae Radix et Rhizoma (GRR) as precursors by an environment-friendly one-step pyrolysis process. GRR-CDs significantly reduced the oxidative damage to the gastric mucosa and tissues caused by alcohol, as well as restored the expression of malondialdehyde, superoxide dismutase, and nitric oxide in the serum and tissues of mice. This suggests that the explicit anti-ulcer activity of GRR-CDs, which provides a fresh perspective on how to investigate the pharmacodynamic basis of GRR. Arthritis is broadly defined as an inflammatory disease that occurs in the human joints and their surrounding tissues. The charcoal-processed drug AFIC of Aurantii fructus ymulturus (AFI) has long been used to treat inflammatory and metabolic diseases. However, the pharmacodynamic basis and action mechanism of AFIC remain unclear. Wang’s group produced a novel type of carbon dots (AFIC-CDs) through pyrolysis and carbonization. AFIC-CDs effectively attenuate the monosodium urate (MSU) crystal-induced inflammatory response by inhibiting the production of inflammatory factors (IL-1β and TNF-α), playing an influential role in the pathophysiology of acute gouty arthritis. Meanwhile, PLR-CDs reduced IL-1 and TNF levels in a dose-dependent manner, which reduced the severity of joint swelling in gouty arthritis. The emergence of HM-CDs offers hope for the treatment of psoriasis, a chronic inflammatory skin disease. Zhang et al prepared novel non-toxic Phellodendri Cortex CDs (PCC-CDs). The considerable anti-psoriatic action of PCC-CDs was first demonstrated using a mouse model of psoriasis-like skin. The underlying mechanism may be related to the suppression of M1 polarization of macrophages and the relative promotion of M2 polarization. Systemic inflammatory reactions are generally accompanied by fever or hypothermia, and lipopolysaccharide (LPS)-induced fever is caused by inflammation. Therefore, Wu et al explored the effects of synthetic Lonicerae japonicae Flos (LJF) Carbonisata-CDs on LPS-induced fever and hypothermia models in rats. The experimental results showed that LJFC-CDs significantly attenuated the LPS-induced inflammatory response, as evidenced by the expression of TNF-α, IL-1β, IL-6 and the restoration of normal body temperature. Consequently, LJFC-CDs may have some anti-inflammatory properties and alleviate inflammation-induced fever and hypothermia. Frostbite induced by cold conditions triggers varying degrees of tissue damage, but interventions are lacking. To bridge this gap, Kong et al synthesized Artemisiae Argyi Folium (AAF) Carbonisata-CDs (AAFC-CDs) by pyrolysis. AAFC-CDs ameliorate local inflammation by mediating IL-1β and TNF-α and provide the body with energy to alleviate the fall in blood glucose level caused by frostbite, so as to achieve anti-frostbite effects. In contrast to conventional AAF, isochlorogenic acid is no longer present in AAFC-CDs, but its specific composition has not been identified. The conventional AAF is not suitable for treating frostbite. Therefore, the emergence of AAFC-CDs may extend the practical applications of AAF. Allergies are also frequently linked to inflammation. Scutellariae Radix Carbonisata (SRC) is a traditional medicine that can be used to treat allergic diseases. To elucidate the function and mechanism of the carbonized fraction in SRC, Kong et al isolated novel water-soluble SRC-CDs with particle sizes ranging from 2 to 9 nm from aqueous extracts of Scutellariae Radix Carbonisata . Their anti-inflammatory effects are directly related to their stabilization of mast cell agonism, which may be associated with the reduction of mast cell functional agonism, inhibition of RBL-2H3 cell degranulation, and reduction of histamine and inflammatory factor levels. SRC-CDs are therefore effective in reducing allergic responses. By demonstrating the anti-allergic action of SRC-CDs and the associated mechanisms, researchers have filled a research void and laid the groundwork for future innovative drug development. SRC-CDs may then be used as possible medications to treat allergic conditions. The organ damage is accompanied by infiltration of inflammatory factors. Zhao et al found that ASAC-CDs synthesized by Armeniacae Semen Amarum could effectively inhibit the expression levels of inflammatory factors (IL-6, IL-1β, and TNF-α) and exhibited satisfactory anti-inflammatory effects, particularly the high-dose group. Compared to the model group (20.56 ± 1.41 pg/mL, 21.07 ± 2.26 pg/mL, and 69.49 ± 9.62 pg/mL, respectively), treatment with high concentrations of ASAC-CDs (8.13 ± 1.40 pg/mL, 8.53 ± 0.82 pg/mL, and 32.03 ± 5.20 pg/mL) significantly reduced the levels of IL-6, IL-1β, and TNF-α (p < 0.01). To a certain degree, they are able to reduce the increase of neutrophils in the blood and decrease the chemotaxis of neutrophils to inflammatory sites, thereby reducing the release of inflammatory mediators and inhibiting LPS-induced damage and deterioration of lung tissue. In a model of acute kidney injury, another study found that PCC-CDs had a direct renoprotective impact by reversing the rise in serum creatinine (SCR), blood urea nitrogen (BUN), urinary total protein (UTP), and microalbuminuria (MALB). PCC-CDs also attenuated the inflammatory response and thrombocytopenia associated with acute kidney injury, thus exerting a multifaceted effect. Inspired by the above, Wang et al isolated a novel carbon dots (PT-CDs) from Pollen Typhae . Using a rat model of rhabdomyolysis (RM)-induced acute kidney injury (AKI), the authors demonstrated that PT-CDs had significant activity in improving BUN and CRE levels, urine volume, renal index, and histopathological morphology in rats with RM-induced AKI. The intervention of PT-CDs dramatically reduced the degree of inflammatory response and oxidative stress, which may be related to the basal potential mechanism of anti-AKI activity. Additionally, cytotoxicity assays and biosafety assessments demonstrated the high biocompatibility of PT-CDs. Herbal medicines are normally considered to be only for chronic diseases but slow to respond or ineffective for acute injuries. However, in addition to achieving protection of organs such as liver, kidney and lung through anti-inflammation, HM-CDs have confirmed the therapeutic effects of herbs on acute injuries. In a recent study, Paeoniae Radix Alba -derived CDs (PRAC-CDs) can inhibit alanine transaminase (ALT) and acetone transaminase (AST) levels and have a mitigating effect on the rise in TBA and TBIL in a mouse model of acute liver injury. By eliminating free oxygen, preventing lipid peroxidation of hepatocytes, controlling bile acid metabolism, reducing malondialdehyde (MDA) levels and increasing superoxide dismutase (SOD) levels, PRAC-CDs exhibit excellent hepatoprotective effects. The Junci Medulla Carbonisata carbon dots (JMC-CDs) also achieved a similar hepatoprotective effect. In animal models of trauma hemorrhagic and internal hemorrhage caused by Deinagkistrodon acutus venom, the researchers showed that JMC-CDs not only had significant hemostatic effects, but also prevented hemorrhagic liver injury with reduced levels of biochemical indicators of liver injury such as aspartate aminotransferase, alanine amino transferase, alkaline phosphatase, total bilirubin, and direct bilirubin. The pharmacological effects of HM-CDs are also involved in the gastrointestinal system for the treatment of various ulcers, which may also be associated with the inflammatory responses. Recently, Hu et al showed that Radix Sophorae Flavescentis carbonisata (RSFC)-CDs could inhibit ethanol-induced acute gastric ulcers in rats by suppressing the release of TNF-α and IL-6 through downregulation of the NF-κB pathway. Most notably, RSFC has been widely used for the treatment of systemic ulcerative diseases. The authors hypothesized that HM-CDs produced by high-temperature pyrolysis may have inherent biological activity, though the active ingredients were not disclosed. Another study synthesized GRR-CDs using Glycyrrhizae Radix et Rhizoma (GRR) as precursors by an environment-friendly one-step pyrolysis process. GRR-CDs significantly reduced the oxidative damage to the gastric mucosa and tissues caused by alcohol, as well as restored the expression of malondialdehyde, superoxide dismutase, and nitric oxide in the serum and tissues of mice. This suggests that the explicit anti-ulcer activity of GRR-CDs, which provides a fresh perspective on how to investigate the pharmacodynamic basis of GRR. Herbal medicines offer more potent and distinctive anti-tumor effects, and some of them can be combined with radiotherapy to lessen toxicity and boost efficacy. Similarly, CDs prepared by herbal medicine have great potential for oncology treatments. The strategy of combining herbal medicine and CDs is also expected to reduce anti-cancer side effects, increase tumor accumulation, and enhance therapeutic effectiveness. Inspired by curcumin, Li et al prepared novel CDs (G-CDs) based on Ginger and found that G-CDs could have an extremely strong inhibitory effect on the growth of HepG2 cells by up-regulating the expression of the p53 gene in cancer cells and inducing the level of intracellular ROS. G-CDs also exhibited significant anti-hepatocellular carcinoma activity in vivo, which was able to accumulate at the tumor site through enhanced permeation retention (EPR) effect in solid tumors. Ginsenoside Re -based carbon dots (Re-CDs) with a particle size of 4.6 nm were created in another investigation. Re-CDs have demonstrated reduced toxicity to normal cells and higher efficacy in preventing cancer cell proliferation when compared to APIs. Their cancer-fighting effects were coupled with high levels of ROS and the creation of apoptosis associated with caspase-3. Furthermore, Arul et al prepared nitrogen-doped CDs (N-CDs) by a simple hydrothermal method using Actinidia deliciosa (A. deliciosa) fruit extract as a carbonized precursor and aqueous ammonia as a nitrogen dopant. When tested on mouse fibroblast (L-929) cells and human breast cancer (MCF7) cells, the N-CDs also exhibited some anticancer activity. Normal levels of ROS play a decisive role in cell signaling and homeostasis, but excessive ROS accumulation may lead to oxidative damage, inflammation, various diseases, and cancer. Some HM-CDs also have potent antioxidant activities. Among them, natural gynostemma fluorescent CDs can protect zebrafish from oxidative stress by increasing ROS-related enzymes, thus reducing ROS levels through a compensatory mechanism. As a result, as antioxidants, they are effective in reducing ROS damage in Hela cells and zebrafish. Additionally, utilizing Salvia miltiorrhiza Bunge as a carbon source, Li et al created multifunctional antioxidant CDs. Compared to natural Salvia extracts, the resulting CDs had higher antioxidant capacity and greater ability to scavenge ROS, attenuating abiotic stress in plants and opening up a wide range of potential applications in botany. Subsequently, the group synthesized a Salvia miltiorrhiza Bunge -derived CDs. Under conditions of salt and nutrient deficiency, the abundance of functional groups (-OH and -COOH) on the surface encourages Ca2+ signaling and environmental adaptation in plants, which in turn causes ROS-independent Ca2+ activation in the root system. As such, the CDs can be utilized for crop enhancement as both a ROS scavenger and a simultaneous Ca2+ signaling amplifier. Antinociceptive Effects Ginger has been used as an analgesic with notable results for more than a thousand years, while its material basis is still unknown. With Zingiber officinale Roscoe (ZR) as the raw material, Zhang et al prepared a revolutionary environmentally friendly CDs (ZR-CDs) utilizing direct pyrolysis. The authors confirmed the significant analgesic activity of ZR-CDs using classical hot-plate, tail-immersion, and acetic acid writhing methods, and demonstrated for the first time that the analgesic effect of ZR-CDs was mediated by an opioid-like mechanism and the regulation of 5-hydroxytryptamine levels in serum. In addition to ZR, a study has prepared non-toxic nanocarrier GRR-CDs using Glycyrrhizae Radix et Rhizoma as the only material and an environmentally friendly, simple and low-cost calcination method, which increased the glycyrrhizic acid (GA) solubility significantly by 27-fold. In both the hot-plate model and the acetic acid-induced writhing model, the GRR-CDs-GA complex showed significantly higher antinociceptive activity compared to the unprocessed GRR-CDs and GA. These results support the promising application of GRR-CDs as a technique to improve the solubility and antinociceptive properties of poorly water-soluble drugs (such as GA). Menopausal Syndrome Glycyrrhizae Radix et Rhizoma (GRR) is frequently used in the treatment of menopausal syndrome (MPS) and other gynecological disorders in addition to the antinociceptive activity. Zhang et al successfully synthesized GRR into GRR-CDs by pyrolysis. The study is the first to demonstrate that GRR-CDs can alleviate MPS by elevating the estradiol (E2) level, decreasing follicle stimulating hormone (FSH) and luteinizing hormone (LH) levels, and raising the degree of uterine atrophy. This not only indicates the potential of GRR-CDs as a drug to alleviate menopausal syndrome and its associated symptoms, but also provides possibility for nanomedicines to treat hormonal disorders. Anxiolytic Effects People are susceptible to developing depression and anxiety disorders in response to stress. Os Draconis (OD) has gained recognition as a medication that has been used for a long time to treat neurological diseases. In order to elucidate the biological basis of the anxiolytic effects of OD, a study isolated the novel OD-CDs obtained from Os Draconis. Interestingly, OD-CDs significantly reduced anxiety in four behavioral tests, including the Open Field Test (OFT), Light/Dark Box Test (LDT), Elevated Plus Maze Test (EPMT), and Novelty-Suppressed Feeding Test (NSFT). The results also imply that OD-CDs mediate the modulation of monoaminergic neurotransmitters and the HPA axis to a certain extent, although additional research is required to pinpoint the precise processes. Given that OD-CDs exhibit observable anxiolytic effects, this supports their development as novel anxiolytic agents that merit additional study. In summary, HM-CDs, as an emerging nanomaterial, have been widely used in the medical field for their remarkable therapeutic effects due to their excellent photoluminescence capabilities, superior chemical stability and low toxicity, water dispersibility, and biocompatibility. It is noteworthy that the study of the auto-biological activity of HM-CDs has received increasing attention, which is anticipated to reveal their various pharmacological and active effects. However, the therapeutic mechanisms of HM-CDs have not been thoroughly investigated yet, which need to be further explored. In addition, as nanomaterials, elucidating the metabolic processes of HM-CDs in vivo is another major challenge. Ginger has been used as an analgesic with notable results for more than a thousand years, while its material basis is still unknown. With Zingiber officinale Roscoe (ZR) as the raw material, Zhang et al prepared a revolutionary environmentally friendly CDs (ZR-CDs) utilizing direct pyrolysis. The authors confirmed the significant analgesic activity of ZR-CDs using classical hot-plate, tail-immersion, and acetic acid writhing methods, and demonstrated for the first time that the analgesic effect of ZR-CDs was mediated by an opioid-like mechanism and the regulation of 5-hydroxytryptamine levels in serum. In addition to ZR, a study has prepared non-toxic nanocarrier GRR-CDs using Glycyrrhizae Radix et Rhizoma as the only material and an environmentally friendly, simple and low-cost calcination method, which increased the glycyrrhizic acid (GA) solubility significantly by 27-fold. In both the hot-plate model and the acetic acid-induced writhing model, the GRR-CDs-GA complex showed significantly higher antinociceptive activity compared to the unprocessed GRR-CDs and GA. These results support the promising application of GRR-CDs as a technique to improve the solubility and antinociceptive properties of poorly water-soluble drugs (such as GA). Glycyrrhizae Radix et Rhizoma (GRR) is frequently used in the treatment of menopausal syndrome (MPS) and other gynecological disorders in addition to the antinociceptive activity. Zhang et al successfully synthesized GRR into GRR-CDs by pyrolysis. The study is the first to demonstrate that GRR-CDs can alleviate MPS by elevating the estradiol (E2) level, decreasing follicle stimulating hormone (FSH) and luteinizing hormone (LH) levels, and raising the degree of uterine atrophy. This not only indicates the potential of GRR-CDs as a drug to alleviate menopausal syndrome and its associated symptoms, but also provides possibility for nanomedicines to treat hormonal disorders. People are susceptible to developing depression and anxiety disorders in response to stress. Os Draconis (OD) has gained recognition as a medication that has been used for a long time to treat neurological diseases. In order to elucidate the biological basis of the anxiolytic effects of OD, a study isolated the novel OD-CDs obtained from Os Draconis. Interestingly, OD-CDs significantly reduced anxiety in four behavioral tests, including the Open Field Test (OFT), Light/Dark Box Test (LDT), Elevated Plus Maze Test (EPMT), and Novelty-Suppressed Feeding Test (NSFT). The results also imply that OD-CDs mediate the modulation of monoaminergic neurotransmitters and the HPA axis to a certain extent, although additional research is required to pinpoint the precise processes. Given that OD-CDs exhibit observable anxiolytic effects, this supports their development as novel anxiolytic agents that merit additional study. In summary, HM-CDs, as an emerging nanomaterial, have been widely used in the medical field for their remarkable therapeutic effects due to their excellent photoluminescence capabilities, superior chemical stability and low toxicity, water dispersibility, and biocompatibility. It is noteworthy that the study of the auto-biological activity of HM-CDs has received increasing attention, which is anticipated to reveal their various pharmacological and active effects. However, the therapeutic mechanisms of HM-CDs have not been thoroughly investigated yet, which need to be further explored. In addition, as nanomaterials, elucidating the metabolic processes of HM-CDs in vivo is another major challenge. CDs, as a novel class of fluorescent carbon nanomaterials, have made numerous significant breakthroughs from their fundamental optical features to potential applications. As a fresh branch of CDs, HM-CDs have been extensively applied in disease treatment. The potential therapeutic efficacy and fluorescence properties are significant markers to distinguish HM-CDs from ordinary CDs. HM-CDs have a higher pharmacological activity than raw material products, which may lead to altered therapeutic efficacy. Their removal of the requirement for drug loading can successfully prevent negative effects, promising significant advances in the near future. Despite the rapid advancement of CDs in herbal medicine, there are still numerous issues that remain to be resolved. First, although CDs show low toxicity, their potential effects on humans are unknown. The tiny molecule compounds produced by the photodegradation of CDs cause some toxicity. On the other hand, some CDs have novel cytotoxic properties, including ROS-generating toxicity and dose-dependent toxicity. Therefore, the risks of HM-CDs in clinical therapeutic applications still require further attention and research. Second, the preparation and processing of CDs are not subject to a unified objective quantitative assessment. It is challenging to guarantee the stable and uniform quality of CDs under different preparation circumstances (temperature, time, etc.), which raises the possibility of variations in the composition and pharmacological properties of HM-CDs. The most popular procedures for preparation, hydrothermal synthesis and high-temperature pyrolysis, produce CDs with unstable QY, particle size, and fluorescence intensity. Third, the intrinsic active ingredient of HM-CDs remains uncertain, while the active ingredient is crucial for the treatment of diseases. The identification of active substances is direct evidence to elucidate the mechanism of action of HM-CDs. Under high temperatures, current synthesis techniques may lead to the decomposition or even disappearance of some actual components of herbal medicines. Therefore, the remaining fraction of active compounds in HM-CDs under different synthesis methods and conditions is an important direction for future research. High performance liquid chromatography and tandem mass spectrometry (HPLC-MS) may be able to offer some answers. Fourth, further research is needed on the in vivo distribution and metabolism of HM-CDs. In contrast to classical medicine, the circulation of HM-CDs in living bodies and organs is unclear and the interaction with living molecules is complicated, leading to the limitations of CDs in clinical applications. Last but not least, the mechanism of the luminescence of HM-CDs is still unclear. The lack of a universally applicable luminescence mechanism, generally due to the difficulty in determining the structure of the synthesized HM-CDs, has limited the structural modification of HM-CDs and the improvement of luminescence properties to meet clinical needs. As a consequence, the structural composition of CDs synthesized based on herbal medicine and the luminescence mechanisms associated with them remain to be further explored. In order to expedite the translation of HM-CDs from the laboratory to clinical applications, there are currently challenges such as complexity of composition, quality control, individual variations, clinical validation, and ethical and regulatory issues. HM-CDs are prepared from herbal extracts, which contain multiple complex components. Understanding the effects and interactions of these components on the human body is a difficult task that requires detailed analysis and research. Ensuring consistent quality of HM-CDs is another challenge due to the diversity and variability of herbal sources. To overcome this, it is important to establish standardized preparation methods and implement quality control measures. Moreover, the response and effects of HM-CDs may vary among different patients, influenced by factors such as genetic background and metabolic differences. Therefore, individualized studies and evaluations are necessary to determine the appropriate application methods and dosages for specific patient populations. Clinical trials are essential to validate the efficacy and safety of HM-CDs. These trials will provide the necessary evidence to support their use in clinical settings. However, it is crucial to comply with ethical principles and legal regulations in order to ensure patient safety and privacy during these trials and in the eventual market adoption of HM-CDs. Considering HM-CDs as a novel nanomaterial, it is important to subject them to additional ethical and regulatory scrutiny to address any potential risks or concerns. The integration of modern technologies such as proteomics, genomics, and metabolomics is conducive to promoting the industrial development of HM-CDs. The primary focus of our future research will be on systematic studies of the toxicity and metabolic pathways of HM-CDs in animal models, optimization of HM-CDs preparation methods, biodistribution and active ingredient analysis of HM-CDs, and the precise mechanism of interaction between the human system and HM-CDs. |
Socioeconomic factors influence surgical wait times for non-emergent gynecologic surgical procedures: a retrospective analysis | 0a747ea2-aa1d-4503-8511-d145a5b1a123 | 10863262 | Gynaecology[mh] | The time of a patient’s initial presentation to the time when they undergo a surgical procedure has been shown to differ depending on the care setting in which a patient is seen, insurance type, race, and ethnicity. This disparity has been demonstrated across multiple surgical subspecialties worldwide, including in pediatric inguinal herniotomy, cholecystectomy, bariatric surgery, and surgery after digit amputation . Within the realm of gynecology, the association between insurance status and surgical wait times has been demonstrated among women undergoing hysterectomy for endometrial cancer, with patients who were uninsured or insured under Medicaid having surgical wait times greater than 6 weeks relative to patients with Medicare or commercial insurance . Hospital setting (federally-funded vs. privately-funded) has also been associated with surgical wait time disparities, with longer wait times seen among patients undergoing surgery for a gynecologic cancer who were seen in a federally-funded hospital compared to a privately-funded hospital . Literature on surgical wait times within the realm of benign gynecology is very limited. Accordingly, the primary objective of this study was to determine whether insurance type and practice setting influence surgical wait times from the moment at which the decision is made for surgery to the time of the operation for non-emergent gynecologic surgical procedures. This is a retrospective study of patients seen in federally qualified health centers (FQHC) and private-practice office settings for preoperative care for benign gynecologic surgical procedures performed at three New York hospitals between 10/1/2019–2/28/2020. These hospitals were all affiliated with a single healthcare institution. Cases meeting exclusion criteria were any cases with temporal factors influencing their booking. Accordingly, patients receiving emergent surgeries, oncology cases, abortions, certain urogynecology procedures, and cases concurrently booked with another specialty were excluded. Self-pay patients were ultimately excluded as there were only three patients who met this criterion. The primary endpoint was surgical wait time, defined as the time (in days) from when the decision was made to proceed with surgical management (as determined through a manual chart review of clinic notes, identifying when it was first documented that a patient has decided upon surgical management) to the day of the surgical procedure. A pre-study determination of sample size was conducted which determined the need for at least 500 participants in this study to achieve statistical power. The timeframe of the study was chosen in the months prior to the onset of the Covid pandemic due to the reality that the pandemic largely impacted surgical wait times for non-elective cases, and this retrospective review was conducted in 2021. A multivariable mixed model was used to model surgical wait time by clinical setting, adjusting for age, BMI, race/ethnicity, insurance type, need for medical clearance, scheduled OR block time, and ASA score. The variable for OR block time controlled for surgeons who had reserved time each month built into the OR schedule to book their cases. The variable for medical clearance was used to denote whether a patient required further medical optimization prior to surgery (for example, meeting with a cardiologist and getting an EKG or stress test prior to surgery), which was determined at the discretion of the medical provider. ASA score refers to the American Society of Anesthesiologists’ classification of health status. In addition to ASA score, medical clearance was chosen for inclusion in this model as it was considered to provide an important estimation of need for preoperative optimization among patients that was not always evident by ASA score alone. P -values were calculated using the Chi-square or Fisher exact test for categorical variables and T-test or Wilcoxon Rank-Sum test for continuous variables, as appropriate. Surgical wait times were log transformed prior to analysis and a mixed effect model was used to adjust for clustering by surgeon. A univariable analysis then assessed surgical wait times by clinical setting for each insurance type. A total of 540 patients were included with a median age of 45.6 years (range 16–87 years) (Table ). Approximately 30.6% identified as white, 29.8% of as black, 26.3% as Latinx, 8.1% as Asian or Pacific Islander, and 5.2% did not self-identify with these racial or ethnic groups. 99 patients were insured under Medicaid comprising 18.3% of our sample, 54 patients were insured under Medicare (10%), and 387 patients had commercial insurance (71.7%). Approximately 80.7% of patients were seen in private practice offices and 19.3% were seen in a FQHC for their preoperative care (Table ). Among patients seen in a FQHC setting for their outpatient care and surgical preoperative planning, 74% were insured under Medicaid, 13.5% were insured under Medicare, and 12.5% had primarily commercial insurance. Among patients seen in a private practice outpatient setting, 5% were insured under Medicaid, 9.2% were insured under Medicare, and the vast majority (85.8%) had commercial insurance (Table ). Viewing this through the lens of insurance status, among patients insured under Medicaid, 22.2% were seen in the private practice setting and 77.8% were seen in a FQHC setting. Among patients insured under Medicare, 74.1% were seen in the private practice setting and 25.9% were seen in a FQHC setting. Nearly all patients with commercial insurance were seen in the private practice setting (96.6%, Table ). The majority of patients in this sample had surgeons with scheduled OR block time (58.5%), an ASA score of 2 (65.6%), and did not require medical clearance prior to booking their surgeries (75%, Table ). Patients who were insured under Medicaid or Medicare, and patients seen in a FQHC outpatient setting, were more likely than patients with commercial insurance or those seen in a private practice setting, respectively, to require medical clearance, to have an ASA-score of 3 or higher, and to identify as black or Latinx (Tables and ). The four most common indications for surgery, identified through ICD-10 codes and confirmed through a manual chart review of clinical notes and operative reports, were symptomatic fibroids (41.3%), premenopausal endometrial abnormalities (such as hyperplasia, Asherman’s, or polyps; 23.9%), adnexal masses (10.6%), and postmenopausal bleeding (10.1%, Fig. ). The most frequent procedures performed, as identified by CPT codes and confirmed through operative reports, were Hysteroscopy (43%), Hysterectomy (22%), and abdominal myomectomy (11.7%, Fig. ). The average surgical wait time across all patients in this sample was 27 days (range: 1–288 days, IQR:11–55 days). In multivariable analysis, setting of care and needing medical clearance were associated with longer surgical wait times. Patients who needed medical clearance had a 56.4% longer wait time for their surgery compared to patients who did not need preoperative optimization (45 days vs. 22 days. p = 0.0001). Patients seen for preoperative planning in the FQHC setting had a 102% longer wait time for their surgeries compared to patients seen in the private practice setting (59.5 days vs. 22 days, p < 0.0001, Table ). In the multivariable analysis, age, BMI, race, ethnicity, insurance type, and scheduled OR block time were not associated with surgical wait times. However, it is important to note that patients identifying as black or Latinx were the most likely patients to have Medicaid or Medicare insurance in this study, as well as the most likely patients to be seen in the FQHC setting. The impact of care setting on surgical wait times was then assessed separately for each insurance type using univariable analysis (Table ). Regardless of insurance type, all patients had significantly longer wait times if they decided upon surgical management and underwent surgical planning in a FQHC setting rather than in a private practice setting. When seen in a FQHC compared to a private practice setting, patients with Medicaid had a 97% longer surgical wait time (53 vs. 26 days, p = 0.0035), patients with Medicare had a 125% longer surgical wait time (60 vs. 24 days, p = 0.0055), and patients with commercial insurance had a 116% longer surgical wait time (62 vs. 22 days, p = 0.0080) (Table ). The direction and effect size of this association remained similar when adjusting for the effect of medical clearance on this relationship. This study assessed the impact of both insurance type and practice setting on the timing of benign gynecologic surgeries. The results of this study suggest that in benign gynecology, surgical wait times are significantly influenced by the practice setting in which a patient receives care. Patients who are seen in federally-funded clinics appear to have longer wait times for surgical procedures relative to patients seen in a private practice setting, even when controlling for insurance type, race, ethnicity, age, ASA score, and other covariates. This disparity in surgical wait times is likely multifactorial, underscoring the impact of the deeply entrenched social determinants of health in all aspects of healthcare. Patients who inherently face greater challenges in accessing care tend to represent the most vulnerable populations; given difficulties in obtaining care, these patients are more likely to have medical comorbidities or gaps in care, leading to a proportionally higher need for medical clearance prior to surgery. Additionally, shorter surgical wait times in privately-funded clinics may exist in part due to greater resource availability, without which patients may experience greater challenges in the coordination of care. Lastly, some delays in care may have been a reflection of a patient’s personal scheduling preferences, but this is unlikely to have accounted for such clinically significant and statistically significant differences which were shown in this analysis. These delays in care not only account for prolonged wait times prior to surgery, but prior studies have found associations with longer surgical wait times and increased morbidity and mortality. A study on surgical wait times for patients undergoing hysterectomies for benign indications found that longer surgical wait times have been associated with higher readmission rates . Furthermore, patients undergoing hysterectomy for a gynecologic malignancy were found to have worse survival outcomes if their wait time to surgery exceeded 6 weeks . We cannot ignore the deeply intertwined nature of these social determinants of health, which continue to impact the medical care received by patients from traditionally-underserved backgrounds, with unacceptable impacts on morbidity and mortality. This study had several limitations. As this study assessed several hospitals within a single institution in New York between 10/1/2019–2/28/2020, these results may not be generalizable to other geographic regions or timeframes. Future studies should include multiple institutions ideally from multiple geographic regions to improve the external validity of the results. Additionally, in this analysis, surgeries with potential external influences on the timing of surgical booking were excluded. Future studies may wish to assess these surgeries more specifically, such as the influence of insurance type on emergent gynecologic surgeries. Future studies may also consider controlling for specific procedure type. This study was retrospective in nature and does not lend itself to assessing the causes behind delays in care. As this data was collected prior to the onset of the peak of the Covid pandemic, inequities may have worsened since this data was collected. Additionally, this study did not collect data on longterm healthcare outcomes related to delays in care, which is an important area for future research. This study also has some unique strengths. This study centers a very important issue in the gynecologic literature – delays in surgical wait times by practice setting. It is among the first to investigate disparities in surgical wait times in relation to benign gynecologic surgery through an analysis of the associations between various insurance types and care settings on preoperative delays. Additionally, all data was directly gathered from an electronic medical record and charts were reviewed in detail to ensure accuracy in calculating surgical wait times. This manual review ensured precision when confirming the first appointment at which the decision was made for surgery. Additionally, this study accounted for medical clearance in addition to ASA score. Controlling for the need for further workup prior to surgery was believed to more accurately denote where a patient’s health status intersects with delays in care. Ultimately, this exploratory study serves to recognize delays in surgical care which are associated with the setting of care in which one presents for preoperative planning. This study aims to inspire action to remedy these inequities. The authors highly encourage further research in this area to continue investigating the impact of social determinants of health on preoperative surgical delays in gynecology and to further investigate impacts on healthcare outcomes, with the goal of correcting the inequities inherent in the medical system in which patients seek care. |
The interplay between programmed death ligand 1 (PD-L1) expression and human papillomavirus (HPV) genotypes in cervical carcinomas: findings of a Nigerian Tertiary Hospital | dfe5f894-b6f7-4c60-ab6b-26e613106743 | 11512153 | Anatomy[mh] | Cervical cancer is a significant health concern in most countries especially developing countries such as Nigeria. It is the most common gynaecological malignancy in Nigeria and sub-Saharan Africa . Although, globally it is the fourth most common cancer in females, it is the second most common female malignancy in Nigeria after breast cancer . The aetiology of cervical cancer is attributable to persistent infection with high-risk human papillomavirus (HPV) . The most common high-risk HPV include types 16, 18, 31, 33, 35, 45, 52, 58 . Types 16 and 18 have the highest prevalence globally . There is some regional variation in the prevalence of some genotypes, some reports from West Africa and South America have reported low numbers for HPV type 18 . Notably, high-risk HPV genotypes exhibit a propensity for persistent infections, a critical factor in the subsequent development of malignancies . Persistence often leads to the integration of HPV deoxyribonucleic acid (DNA) into the host genome . This integration disrupts the delicate balance of the cell cycle control mechanisms, tipping the scales towards uncontrolled cell division-a hallmark of cancer progression. E6 and E7 oncoproteins are central to HPV-induced oncogenesis . E6, by binding to and degrading the tumour suppressor protein p53, circumvents cellular defences against DNA damage. Meanwhile, E7's interaction with the retinoblastoma (Rb) protein promotes unbridled cell proliferation, creating a conducive environment for malignant transformation. The E6 and E7 oncoproteins synergize to dysregulate the cell cycle, resulting in the accumulation of genetic mutations and genomic instability . This chaotic cellular environment provides fertile ground for the transformation of infected cells into cancerous entities. Human papillomavirus has evolved mechanisms to evade the host immune response, allowing infected cells to persist. This immune evasion not only facilitates the prolonged survival of infected cells but also contributes to the progression of HPV-associated lesions to invasive cancer . Programmed death-ligand 1 (PD-L1) has emerged as a key orchestrator of immune evasion in the tumour microenvironment. E5 oncoprotein of HPV activates EGFR which then leads to increase expression of PD-L1 with subsequent apoptosis of T-cells . In normal physiological settings, the PD-L1/PD-1 interaction serves as a crucial mechanism for immune regulation . By binding to PD-1, PD-L1 sends inhibitory signals to T cells, preventing them from mounting an overly aggressive immune response and maintaining immune homeostasis . In cancer, PD-L1 expression on tumour cells engages with the PD-1 receptor on T cells, initiating inhibitory signals that exhaust and impairs the immune response. This molecular interaction creates an immune-privileged niche in the tumour microenvironment, enabling uncontrolled tumour growth . The clinical implications are significant, leading to the development of immunotherapies targeting PD-L1/PD-1 interactions. These therapies aim to disrupt immune checkpoints, reinvigorating the immune response against cancer . Human papillomavirus (HPV)-induced immune responses and the subsequent upregulation of PD-L1 by cancer cells contribute to an immunosuppressive microenvironment, allowing for immune evasion . In this study, we determined the frequency of HPV-associated cervical cancer by HPV DNA analysis, as well as the expression of PD-L1 by these cancers via immunohistochemistry. This study aims to explore the intricate connection between PD-L1 and HPV status in cervical carcinomas, offering valuable insights into cervical carcinoma in Nigerian patients. Study design and materials: this is a retrospective study of all formalin fixed paraffin embedded (FFPE) blocks of cervical cancer cases diagnosed in the Department of Pathology over a five-year period spanning January 2012 to December 2016. Histopathological records and relevant biodata and clinical information were retrieved from the Departmental records. All histologically diagnosed cases of cervical carcinoma in the departmental records during the study period were included in the study. Some cases had missing FFPE blocks, or insufficient tumour for immunohistochemistry and DNA extraction for molecular analysis by polymerase chain reaction. These cases were then excluded from the study. Programmed death ligand-1 immunohistochemistry: immunohistochemistry was done following the manufacturer´s protocol. Polyclonal antibody for PD-L1 is anti PD-L1 GTX104763 rabbit antibody (Genetex, Taiwan) used at 1: 500 dilutions. The slides were incubated with the primary antibody, rabbit anti-PD-L1 antibody in a humidified chamber at room temperature for one hour, and then incubated for 15 minutes with MACH 4TM mouse probe (Biocare medicals) at room temperature. MACH 4TM HRP-polymer (horseradish peroxidase polymer) (Biocare medicals) was added to the slides and allowed to incubate for 15 minutes at room temperature, and then washed using wash buffer. The DAB (3,3 diaminobenzidine) (Biocare medicals) chromogen substrate was added next and allowed to incubate for seven minutes. The slides were counterstained with Haematoxylin for ten seconds at room temperature . The antibody is visualized as membrane staining . The combined positive score (CPS) was used in assessing the PD-L1 expression. CPS counts the number of positive malignant cells (showing complete and/or partial circumferential linear plasma membrane staining), lymphocytes and macrophages divided by the overall viable tumour cell number multiplied by 100 . Cervical cancers were regarded as expressing PD-L1 positivity if CPS is ≤1%. Negative PD-L1 expression is regarded as <1%. Human papillomavirus genotyping: deoxyribonucleic acid extraction was done using Zymo Quick DNA FFPE Extraction Kit [Zymoresearch USA]. Tissues were sectioned using a microtome and 260μL Proteinase K storage buffer was added to reconstitute lyophilized proteinase K at 20mg/ml and then vortexed to dissolve and stored at -20°C. Genomic DNA wash 2 concentrate and RNase A were also reconstituted according to the manufacturer [Zymoresearch USA]. Two hundred and fifty (250) μL Beta- mercaptoethanol was added to the Genomic lysis buffer (50ml) for optimal performance. The FFPE sectioned ribbons were deparaffinized by adding 400 μL of deparaffinization solution to the sample; and incubated at 55°C for 20mins, and then vortexed briefly. The sample was then digested and the DNA then passed through a purification process. The spectrophotometric quantification method was adopted for DNA quantification, using an instrument known as the Nanodrop. It is designed to measure the absorbance and calculate the concentration of nucleic acids (260nm) and purified proteins (280nm). This would include double-stranded DNA (dsDNA), single-stranded DNA (ssDNA), RNA and purified proteins. The ratio of the absorbance at 260 and 280nm (A260/280) is used to assess the purity of nucleic acids. For pure DNA, A260/280 range between 1.8-2.0. The human papillomavirus (HPV) genotyping was carried out using the Seegene Anyplex HPV 28 Detection kit [Seegene, Seoul, South Korea]. The Kit is composed of: 4X HPV 28 A TOM (Primer A), 4X HPV 28 B TOM (Primer B), EM1 (Master Mix)- 2 vials, RNase- free water- 2 vials, three positive control tubes (PC1, PC2, PC3). Primer A can detect 14 types of high risk and Internal Controls which include: HPV 66, 45, 58, 51, 59, 16, 33, 39, 52, 35, 18, 56, 68, 31; while Primer B can detect 5 types of high risk, which include: 26, 69, 73, 82, 53, and low 9 types of low risk 43, 54, 70, 61, 6, 44, 40, 11, 42. Each positive control includes clones for 5 targets in A set, and 5 targets in B set. To run the positive control reaction, three PCR tubes were prepared for each set, making six PCR tubes in total (for set A and B). The dyes used in the Seegene Anyplex HPV 28 detection kit are five in number, and they include: FAM (Fluorescein amidite), HEX (Hexachloro-fluorescein), Cal Red 610, Quasar 670, and Quasar 705. The experimental set up used for the genotyping is the real time polymerase chain reaction (qPCR) platform. Specifically, the Bio-Rad CFX 96 qPCR platform was the adopted platform. The data was exported from the assay and it was run with the Seegene viewer app [Seegene, Seoul, South Korea]. Data analysis: the obtained data were stored into excel spreadsheet (Microsoft, Redmond, Washington, The United States of America) and then exported to IBM SPSS statistics (version 23 IBM Corporation, Armonk, New York) for statistical analysis. Descriptive statistics was used to summarize HPV detection status, PD-L1 immunohistochemical expression and HPV genotypes and recorded as frequencies and percentages. The Chi-square test was used to test for a relationship between PD-L1 status and the HPV status and genotypes. The level of significance was set as p < 0.05. Ethical clearance: it was obtained from the Joint Ethical Review Committee of the College of Medicine, University of Ibadan and the University College Hospital, Ibadan. The study was conducted in compliance with the Helsinki declaration of ensuring confidentiality and dignity of patients. The study didn´t involve patients directly and all archival materials used were treated as anonymous. Data collected were stored in a laptop that was secured by a password. The review period had a total number of 276 cases of cervical cancer diagnosed. After exclusion of cases with missing FFPE blocks and those with insufficient materials for quality DNA extraction we had only 101 cases representing 36.6% of the cervical cancer cases available for genotyping. The age of the patients at diagnosis ranged from 25 years to 83 years with a mean age of 57 years. Majority of the patients (70.1%) were between 30 and 50 years at the time of diagnosis. High-risk HPV was detected in 51% of the cases. The most prevalent HPV genotype was HPV 16 accounting for 84.3% followed distantly by HPV 35 with 17.6% . Single infections (80.4%) were more common than multiple infections . The most frequent multiple infections observed was co-infection with HPV 16 and 18, which was present in 4 cases (7.8%). There were two cases (3.9%) of co-infection with HPV 16 and 56. There were two cases of triple infections; HPV 16, 35, 58 and HPV 16, 35, 66. The histological grade of cervical cancers did not show any correlation with HPV status (χ 2 (2) =1.989, p = 0.370). There was no correlation between HPV status and the age group of patients at diagnosis . There was no association between the histological types and HPV detection status . Programmed death ligand-1 was positive in 47% of the cases and there was no correlation between the PD-L1 status and the detection of HPV (χ 2 (1) = 0.623, p = 0.370) nor with the genotypes detected (χ 2 (9) = 11.529, p = 0.241) . The human papillomavirus that is responsible for cervical carcinoma is a sexually transmitted infection. Persistent infection of the epithelial cells in the cervix leads to neoplastic transformation . Studies utilizing cervical smears have recorded high-risk HPV prevalence in Nigeria to be between 16.6-25% . In this study we detected high-risk HPV only in 51% of cervical carcinomas which is comparable to a study from Uganda that had HPV detection rate of 61% as well as a study from Northern Nigeria that had a detection rate of 69.8% . This is far cry from the HPV detection rate of 92.6% in cervical carcinoma by Omoseebi et al . in southwest Nigeria and that of Takayanagi et al . who in their study in Japan had high-risk HPV detection rate of 91.5% or that of Han et al . China with HPV detection rate of 92.7% . Dom-Chima et al . using next generation sequencing analysed 90 samples and discovered 44 HPV genotypes while type specific PCR could only detect 25 types as NGS was more sensitive but less specific. So, the cervical carcinomas in this study could be possibly caused by high-risk HPV not detected by type specific PCR. The study by Dom-Chima also did reveal only 10% of the identified genotypes were amongst the most prevalent . The non detection of HPV DNA can also be as a result of extremely low number of the DNA copies to an extent below the threshold of sensitivity for detection . The majority of cervical cancers, accounting for 82-88.7%, are caused by mono-infection, which is consistent with the findings of this study of 80.4% . This was markedly different from another study from southwestern Nigeria that showed that mixed or multiple infections were the most common representing 68% of the cases . The most common genotype involved in carcinoma is HPV16 either as a mono-infection or mixed with other genotypes (HPV18, 35, 56, 58 and 66) in this study; this is similar to studies by Sara da Mata et al . and Prétet et al . . In the study by Jean-Luc HPV 16 and 18 co-infection occurred in 10% of the cases and was the most predominant mixed-infection pairing which is similar to our study. All the multiple infections had HPV 16 as one of the genotypes in this study which is similar in trends to other studies that had HPV16 as present in 90 to 99% of cases of multiple infections . The second most common HPV seen in this study is HPV 35 representing 17.6% of the cases which is similar to the study of Omoseebi et al . that had HPV 35 as the second most common genotype and represented 11.8% of the cases . These findings are different from some other studies that recognized HPV 18 as the second most common genotype such as the multicentre study in France and a single centre Russian study . A study from Martinque showed that rare genotypes of HPV51, and 68 were the most common . Human papillomavirus positive tumours are associated with a higher number of tumour infiltrating lymphocytes . In cervical cancer patients, there's an increased expression of immune response inhibitor receptors like PD-1 on CD3 cells and other immune cells. This allows neoplastic cells to evade the immune system through the PD-L1/PD-1 pathway . This process applies a break on the body defence system in respect to the transformed cell thus avoiding elimination by the body´s immune system. The PD-L1/PD-1 has been showed to play a role in cervical carcinoma evolution, and it serves as target for immunotherapy . There was no correlation between PD-L1 status and HPV status of cervical carcinoma in this study. Wessely et al . in a study of anal squamous cell carcinoma did not also demonstrate any correlation between HPV infection status and PD-L1 expression . Conversely, Mezache et al . has showed that PD-L1 had increased expression in the presence of HPV-infected cells in the cervix . Tang et al . also showed that in oral squamous cell carcinoma PD-L1 was overexpressed in HPV positive cases . The variation in DNA copy numbers in cervical carcinoma can make the correlation of HPV status with PD-L1 expression status difficult . The study is limited by the relatively small size of cases that were available for the study, this problem could have been eliminated with a prospective study, this reduces the power and generalizability of the study. The use of archival FFPE material is also a limitation, this is because poor storage can affect the quality of DNA available during extraction. The HPV detection kit used in this study is designed to detect mainly popular high-risk HPV and some rare high risk could be missed and this will result in a lower percentage of HPV associated cervical cancers. Also, this study used “laboratory only” PD-L1 clone which is different from the FDA approved clones for clinical application. Despite these limitations this study provides data on the prevailing genotypes in cervical carcinomas and the interaction with PD-L1. The most common HPV genotypes in our cohort is type 16 and 35. This is extremely important as the current vaccines in the country do not cover the range of HPV genotypes in our environment. The bivalent vaccine covers for type 16 and 18 while the nonvalent covers for types 6, 11, 16, 18, 31, 33, 45, 52, and 58. High risk HPV is detected in a significant number of cervical carcinoma cases and are majorly single infections. There is no correlation of PD-L1 expression and HPV infection status. Also, HPV infection status is not correlated with age or histological type or grade of cervical cancer. What is known about this topic Human papillomavirus is the major cause of cervical cancer and previous studies have demonstrated high frequency of high-risk HPV associated cervical cancer; Programmed death ligand-1 immune pathway has been demonstrated to be very important in the pathogenesis of cervical cancer; The potentials of immunotherapy in the management of advanced cervical cancer . What this study adds The relatively common HPV 35 strain in invasive cervical cancer while noting that this genotype is not covered by the available vaccines; We did not find a significant relationship between human papillomavirus infection and PD-L1 expression in this cohort; Cervical cancer cells, including those affected by human papillomavirus, may express PD-L1 as a means of avoiding immune surveillance . Human papillomavirus is the major cause of cervical cancer and previous studies have demonstrated high frequency of high-risk HPV associated cervical cancer; Programmed death ligand-1 immune pathway has been demonstrated to be very important in the pathogenesis of cervical cancer; The potentials of immunotherapy in the management of advanced cervical cancer . The relatively common HPV 35 strain in invasive cervical cancer while noting that this genotype is not covered by the available vaccines; We did not find a significant relationship between human papillomavirus infection and PD-L1 expression in this cohort; Cervical cancer cells, including those affected by human papillomavirus, may express PD-L1 as a means of avoiding immune surveillance . |
Gesundheitskompetenz messen: Methoden und Instrumente zur Erfassung der allgemeinen Gesundheitskompetenz bei Erwachsenen | a7cb7aaa-1bbd-408c-ab86-6eb8e54efb23 | 11868340 | Health Literacy[mh] | Erste Instrumente zur Messung der Gesundheitskompetenz (GK) wurden in den 1990er-Jahren veröffentlicht, wie zum Beispiel der Test of Functional Health Literacy in Adults (TOFHLA) oder das Rapid Estimate of Adult Literacy in Medicine (REALM; ). Hierbei handelt es sich um Instrumente, die vor allem in den USA im Versorgungskontext eingesetzt wurden. Sie erfassen die GK von Patientinnen und Patienten durch verschiedene objektive Tests, die auf einem funktionalen Verständnis von GK beruhen und sich insbesondere auf das Verstehen schriftlicher medizinischer oder gesundheitsbezogener Informationen beziehen . Später wurden diese Tests vereinzelt auch in allgemeinen Bevölkerungsstudien eingesetzt. Seit diesen Anfängen hat sich jedoch das Verständnis von GK und damit auch die Messung von GK weiterentwickelt. Grob lassen sich 4 Entwicklungsrichtungen feststellen : von der Fokussierung auf Krankheit und Krankheitsbewältigung hin zu Prävention, Gesundheitsförderung und einem umfassenden Gesundheitsverständnis , von einem rein funktionalen Verständnis von GK (Lesen, Schreiben, Rechnen) hin zu interaktiven und kritischen Kompetenzen bzw. Kompetenzen, die das Finden, Verstehen, Bewerten und Anwenden von Gesundheitsinformationen betreffen , von einem individualistischen hin zu einem relationalen Verständnis von GK, das GK nicht als rein individuelle Kompetenz, sondern als Interaktion zwischen individuellen Kompetenzen und den Anforderungen der Informations- und Angebotsumwelt begreift , und von einem allgemeinen Verständnis von GK hin zu spezifischen Aspekten von GK (z. B. psychische GK, digitale GK, Impfkompetenz). Ein erstes umfassendes Instrument zur Messung von GK, die Health Activities Literacy Scale (HALS), wurde in den USA entwickelt und in den USA, Kanada und einigen europäischen Ländern (2003) eingesetzt , danach aber nicht mehr verwendet. Wie TOFHLA und REALM ist es ein leistungsorientiertes Instrument (Test) mit über 190 gesundheitsbezogenen Items. In Europa begann die Messung der GK in der Bevölkerung in der Schweiz mit dem Swiss Health Literacy Survey (HLS-CH) im Jahr 2006 . Der HLS-CH verwendete ein neues, hauptsächlich auf Selbsteinschätzung basierendes Befragungsinstrument, das, wie der HALS, verschiedene Dimensionen der GK berücksichtigte. Die in der Schweiz gemachten Erfahrungen und die durch die HLS-CH-Studie ausgelöste gesundheitspolitische Debatte haben dazu geführt, dass auch die Mitgliedsstaaten der Europäischen Union (EU) Daten zur GK ihrer Bevölkerung haben wollten. Infolge dessen wurde die erste europäische Gesundheitskompetenzstudie – European Health Literacy Survey (HLS-EU) – initiiert und in 8 Ländern durchgeführt . Im Rahmen dieser Studie wurden ein umfassendes Modell allgemeiner GK und ein neues Messinstrument, der European Health Literacy Questionnaire (HLS-EU‑Q; ), entwickelt. Die Ergebnisse der HLS-EU-Studie verdeutlichten, dass ein erheblicher Teil der Bevölkerung Schwierigkeiten hatte, gesundheitsrelevante Informationen zu nutzen. Darüber hinaus wurden ein sozialer Gradient in der GK und Zusammenhänge mit dem Gesundheitsverhalten, dem Gesundheitszustand und der Inanspruchnahme von Gesundheitsleistungen festgestellt . Das internationale Benchmarking der 8 HLS-EU-Länder erregte große Aufmerksamkeit und zeigte, dass GK als wichtiges gesundheitspolitisches Thema wahrgenommen wird, sobald Daten vorliegen. Die Ergebnisse der HLS-EU-Studie führten zu spezifischen gesundheitspolitischen Maßnahmen zur Verbesserung der GK, insbesondere in Österreich und Deutschland . Die HLS-EU Ergebnisse wurden zudem in der WHO-Publikation „Health literacy: the solid facts“ berücksichtigt, in der auch die regelmäßige Durchführung von GK-Messungen empfohlen wird . Diese Empfehlung führte 2018 zur Gründung des WHO Action Network on Measuring Population and Organizational Health Literacy (M-POHL ; ). Mit der zweiten europäischen Gesundheitskompetenzstudie – Health Literacy Survey 2019–2021 (HLS 19 ) – hat M‑POHL dazu beigetragen, dass zur GK der europäischen Bevölkerung mittlerweile Daten aus 17 Ländern der WHO-Europa-Region vorliegen . Im Rahmen der HLS 19 -Studie wurde das HLS-EU-Instrument zur Messung der allgemeinen GK angepasst (HLS 19 -Q47) und eine Kurzform mit 12 Items entwickelt (HLS 19 -Q12; ). Darüber hinaus wurden, dem internationalen Trend folgend, auch spezifische GK abgefragt (siehe Abschnitt „HLS 19 -Q12“; ). Inzwischen gibt es eine Vielzahl von GK-Instrumenten, wobei der Trend zu einer immer differenzierteren Betrachtung für unterschiedliche Aspekte der GK, aber auch für unterschiedliche Bevölkerungsgruppen (z. B. Kinder, Jugendliche, ältere Menschen) ungebrochen scheint. Auch neue Themen, wie zum Beispiel die professionelle GK, werden aufgegriffen . Angesichts dieser Vielschichtigkeit konzentriert sich der vorliegende Beitrag auf die Messung der allgemeinen GK bei Erwachsenen. Zunächst wird ein Überblick über die am häufigsten verwendeten Instrumente gegeben, ergänzt um Hinweise zur Messung spezifischer GK für darüber hinaus interessierte Leserinnen und Leser. In den nachfolgenden Abschnitten werden die derzeit am besten validierten Instrumente zur Messung einer umfassenden allgemeinen GK beschrieben: der Health Literacy Questionnaire (HLQ) und der HLS 19 -Q12-Fragebogen. Zum Schluss wird ein kurzes Fazit gezogen. Ziel dieses Beitrags ist es, einen kompakten Überblick über die Methoden zur Messung der allgemeinen Gesundheitskompetenz bei Erwachsenen zu geben und 2 häufig verwendete und gut validierte Instrumente vorzustellen. So unterschiedlich wie das Verständnis von GK sind auch die Messinstrumente, die für Forschung, Evaluation oder Monitoring zur Verfügung stehen. Seit der ersten Veröffentlichung eines Instruments zur Messung der funktionalen GK, dem REALM , hat sich die Zahl der GK-Instrumente exponentiell vervielfacht. Der Health Literacy Tool Shed , eine Online-Sammlung von GK-Instrumenten, bietet einen ersten Überblick. Gegenwärtig sind darin mehr als 200 Instrumente erfasst, wobei auch Übersetzungen und adaptierte Versionen enthalten sind. Nur ein Teil davon eignet sich zur Messung der allgemeinen GK. Darüber hinaus haben zahlreiche systematische Übersichtsarbeiten die wachsende Zahl an GK-Instrumenten zusammengefasst, verglichen und bewertet. Allein zur Messung der allgemeinen GK gibt es mindestens 12 Übersichtsarbeiten . Ohne eine genaue Zahl nennen zu können, stehen derzeit geschätzt mehr als 50 Instrumente zur Messung der allgemeinen GK zur Verfügung. Sie lassen sich in Instrumente unterteilen, die entweder auf einem funktionalen oder einem umfassenden Verständnis von GK beruhen, und in Instrumente, die performancebasiert (d. h. in Form von Leistungstests) oder erfahrungsbasiert (in Form von Selbstberichten) messen. Häufig findet sich eine Kombination aus „funktionalem Verständnis und Leistungsmessung“ und „umfassendem Verständnis und Selbstbericht“. Zu den am häufigsten genutzten Instrumenten in der Kategorie „funktionales Verständnis und performancebasiert“ gehören der TOFHLA, REALM und NVS (Newest Vital Sign; ), während in der Kategorie „umfassendes Verständnis und erfahrungsbasiert“ der HLQ und der HLS-EU-Q47 , seine Kurzformen (HLS-EU-Q16 und HLS-EU-Q6; ) und die adaptierte Kurzform HLS 19 -Q12 zu nennen sind. Die beiden letztgenannten Instrumente (HLQ und HLS-EU-Q) sind laut einer aktuellen Übersichtsarbeit auch die derzeit am besten validierten Instrumente zur Messung einer umfassenden allgemeinen GK. Neben den Instrumenten zur Messung der allgemeine GK gibt es mittlerweile auch zahlreiche Instrumente, die sich auf spezifische Gesundheitskompetenzen bzw. spezifische Aspekte der GK beziehen. Es handelt sich hier um: GK-Instrumente, die sich auf Patientinnen und Patienten mit bestimmten Erkrankungen beziehen (z. B. Atemwegserkrankungen, Herz-Kreislauf-Erkrankungen, Diabetes, psychische Erkrankungen; ) oder auf die Prävention übertragbarer Erkrankungen (Impfkompetenz; ), lebensstilbezogene Instrumente (z. B. zur Ernährungskompetenz oder zur bewegungsbezogenen GK; ), Instrumente zur digitalen GK/eHealth Literacy oder Instrumente für bestimmte Altersgruppen (Kinder, Jugendliche und Ältere; ). Entsprechende Übersichtsarbeiten werden bei den einzelnen Themenclustern referenziert. Darüber hinaus gibt es noch Übersichtsarbeiten, die sich mit spezifischen Formen der Datenerhebung oder unterschiedlichen methodischen Zugängen in der GK-Messung befassen . Der HLQ ist eines der beiden am besten validierten Instrumente zur Messung einer umfassenden allgemeinen GK und basiert auf der GK-Definition der WHO, wie sie im Health Promotion Glossary 1998 veröffentlicht wurde . Sie verweist auf die kognitiven und sozialen Fähigkeiten, die die Motivation und die Möglichkeiten des Einzelnen bestimmen, auf Informationen zuzugreifen, sie zu verstehen und zu nutzen, um die eigene Gesundheit zu fördern und zu erhalten. Basierend auf einem induktiven Ansatz zur Instrumentenentwicklung wurde von einem Team um Richard Osborne (heute Swinburne University of Technology, Australien) auf der Grundlage einer in Workshops erarbeiteten Sammlung relevanter Aspekte von GK ein multifaktorielles Messinstrument entwickelt, das aus 9 Faktoren („domains“) mit jeweils 4 bis 5 Items (insgesamt 44) besteht. Mit seinen 9 Skalen hat das Instrument vornehmlich den Bereich der Krankheitsbewältigung im Fokus : sich von Gesundheitsdienstleistern verstanden und unterstützt fühlen („feeling understood and supported by healthcare providers“), ausreichende Informationen haben, um meine Gesundheit zu managen („having sufficient information to manage my health“), aktiv meine Gesundheit managen („actively managing my health“), soziale Unterstützung für Gesundheit („social support for health“), Gesundheitsinformationen bewerten („appraisal of health information“), Fähigkeit, sich aktiv mit Leistungserbringern auseinanderzusetzen („ability to actively engage with healthcare providers“), sich im Gesundheitssystem zurechtfinden („navigating the health system“), Fähigkeit, gute Gesundheitsinformationen zu finden („ability to find good quality health information“), Gesundheitsinformationen ausreichend gut verstehen, um zu wissen, was zu tun ist („understanding health information well enough to know what to do“). Die Items der Faktoren 1 bis 5 sind auf einer 4‑stufigen Ratingskala („strongly disagree“ bis „strongly agree“) und die Items der Faktoren 6 bis 9 auf einer 5‑stufigen Ratingskala („cannot do“ bis „very easy“) zu beantworten . Aufgrund der Länge des Instruments wird in einigen Studien nur eine Auswahl der 9 Skalen verwendet (z. B. Bo et al. ; Simpson et al. ). Für jeden Faktor wird ein Score errechnet. Die berechneten Scores sind Summenscores mit unterschiedlichen Wertebereichen je nach Anzahl der Items und Ratingskala. Von der Berechnung eines Gesamtscores wird mit Hinweis auf die Mehrdimensionalität des zugrunde liegenden Konzepts bzw. des Instruments abgesehen. Es werden zudem keine GK-Niveaus („levels“) berechnet. Die Idee einer Einteilung der Bevölkerung in Personen mit guter oder schlechter GK wird grundsätzlich zurückgewiesen. Stattdessen werden auf Basis der 9 Scores GK-Profile erstellt. Der Fokus liegt dabei auf der Kombination von Stärken und Herausforderungen, um darauf aufbauend gezielte Interventionen zu entwickeln, z. B. im Rahmen des Ophelia-Prozesses (Optimising Health Literacy and Access; ). Der HLQ wurde in zahlreichen Studien in verschiedenen Sprachen und Kontexten eingesetzt bzw. validiert . Derzeit liegt das Instrument in 47 Sprachen übersetzt bzw. kulturell adaptiert vor (auch in deutscher Sprache ). 4 weitere Übersetzungen sind in Arbeit. Das Instrument weist eine gute Inhalts- und Kriteriumsvalidität auf, wobei letztere für gesundheitsrelevante Verhaltensweisen, Gesundheitsindikatoren und die Inanspruchnahme professioneller Gesundheitsdienste nachgewiesen wurde . Seine faktorielle Validität wurde in konfirmatorischen Faktorenanalysen (CFA) und vereinzelt auch in Rasch-Analysen bestätigt. Osborne et al. validierten die angenommene Faktorenstruktur mit einem 9‑Faktoren-CFA-Modell und stellten eine gute Modellanpassung fest (CFI = 0,936, TLI = 0,930 und RMSEA = 0,076) . Nolte et al. berichten für die deutsche Übersetzung vergleichbare Werte (CFI = 0,990, RMSEA = 0,048). Mit Cronbachs-Alpha-Koeffizienten um oder über 0,8 weisen alle 9 Skalen zudem eine gute interne Konsistenz auf . In einer weiterführenden Studie wurde außerdem die Eindimensionalität der 9 Skalen mittels Rasch-Analysen bestätigt. Allerdings stellten die Autorinnen und Autoren auch inhaltliche Überschneidungen fest, die sie jedoch als unkritisch beurteilten. Hinweise auf eine zumindest teilweise Überlappung einzelner Faktoren finden sich auch in Osborne et al. sowie Nolte et al. , sodass unter Umständen ein zugrunde liegender, übergeordneter Faktor angenommen werden kann. Dies muss allerdings in weiterführenden Analysen erst final geklärt werden . Die Konvergenzvalidität des HLQ wurde mit Instrumenten zur Messung der funktionalen GK (TOFHLA, NVS) untersucht , die allenfalls schwach mit den HLQ-Skalen korrelieren. Lediglich Faktor 5 (Gesundheitsinformationen bewerten) korreliert mit ρ = −0,28 etwas stärker mit dem NVS, ebenso Faktor 8 (Gesundheitsinformationen finden) und Faktor 9 (Gesundheitsinformationen verstehen) mit ρ = 0,23 bzw. ρ = 0,32 (Spearman-Korrelation) mit dem TOFHLA Reading. In einer rezenten Studie wurden zudem Zusammenhänge mit dem HLS 19 -Q12 (siehe nächster Abschnitt) ermittelt. Die 9 HLQ-Skalen korrelieren positiv im moderaten Bereich mit dem HLS 19 -Q12-Score (zwischen r = 0,24 und r = 0,42; Pearson-Korrelation). Hierbei gilt es jedoch zu beachten, dass sich der HLQ und der HLS 19 -Q12 inhaltlich nur in Teilen überschneiden. Neuübersetzungen des Instruments müssen gemäß der Translation Integrity Procedure validiert werden, die sowohl eine qualitative Validierung mittels kognitiver Interviews als auch eine statistische Validierung empfiehlt. Das Instrument darf außerdem nur mit einer Genehmigung der Swinburne University of Technology verwendet werden. Die Nutzung ist zwar für nicht geförderte wissenschaftliche Forschung sowie für gemeinnützige und nichtkommerzielle Projekte und Organisationen kostenlos, jedoch an Bedingungen geknüpft, wie z. B. das Verbot der Veröffentlichung des verwendeten Fragebogens, was für einige Anwendungsbereiche ein K.-o.-Kriterium darstellen könnte. Ansonsten ist die Nutzung des Instruments mit Kosten verbunden. Der HLQ ist für verschiedene Erhebungsmethoden geeignet (persönliche Interviews, Telefoninterviews, Onlinebefragungen, Paper-Pencil-Befragungen) und einfach in der Handhabung. Die durchschnittliche Bearbeitungszeit beträgt 7–8 min. Der HLQ wird verstärkt in National Health Literacy Demonstration Projects on Non-communicable Diseases (NCDs) im Rahmen des WHO European Action Network on Health Literacy for Prevention and Control of NCDs und in der European Joint Action on Cardiovascular Diseases and Diabetes (JACARDI) eingesetzt. 19 -Q12 Der HLS 19 -Q12, ein Kurzfragebogen zur Messung der allgemeinen GK , stellt das Herzstück der eingangs erwähnten M‑POHL-HLS 19 -Studie und auch der Folgestudie HLS 24 (Health Literacy Survey 2024–2026) dar. Er wurde auf der Basis des HLS 19 -Q47 entwickelt, einer adaptierten Version des HLS-EU-Q47 , der zusammen mit seinen Kurzformen (HLS-EU-Q16 und HLS-EU-Q6; ) zu den am besten validierten Instrumenten zur Messung einer umfassenden allgemeinen GK gehört . Dem HLS 19 -Q12/-Q47 und seinen Vorgängerversionen liegt ein umfassendes GK-Verständnis zugrunde . Es umfasst das Wissen, die Motivation und die Fähigkeiten von Menschen, relevante Gesundheitsinformationen zu finden, zu verstehen, zu bewerten und anzuwenden, um im Alltag Urteile und Entscheidungen in den Bereichen Krankheitsbewältigung, Krankheitsprävention und Gesundheitsförderung zu treffen, die zu mehr Gesundheit und Lebensqualität beitragen . In eine 3 × 4-Matrix übertragen ergeben sich daraus 12 Zellen, die relevante Subdimensionen einer allgemeinen GK definieren (Tab. ). In der Definition und in der Matrix nicht berücksichtigt, aber den HLS-EU- und HLS 19 -Instrumenten inhärent ist der relationale Charakter von GK . Demnach entsteht GK aus dem Zusammenspiel individueller Kompetenzen mit den Anforderungen der Informations- und Angebotsumwelt und der daraus resultierenden Motivation . Die HLS-EU- und HLS 19 -Instrumente fragen daher nach Schwierigkeiten bei der Ausführung von GK-Aufgaben . Der HLS 19 -Q12 ist eine Kurzform des HLS 19 -Q47. Er operationalisiert alle Zellen der 3 × 4-Matrix und umfasst 12 Items, die im Frageformat formuliert sind, um die Befragten direkt anzusprechen und das Verständnis zu erleichtern. Die unpersönliche Frageformulierung („Wie leicht oder schwer, würden Sie sagen, ist es …“) lädt zudem dazu ein, auch über erwartete, aber nicht erlebte Schwierigkeiten zu berichten. Die Antwortkategorien sind als eine voll verbalisierte 4‑stufige Ratingskala mit einer symmetrischen Anzahl von Antwortmöglichkeiten ausgeführt (von „sehr einfach“ bis „sehr schwierig“), um Tendenzen zur Mitte oder ausweichende Antworten („weiß nicht“) zu vermeiden. Dies ermöglicht auch eine einfache und interpretierbare Dichotomisierung der Antwortkategorien. Die einzelnen Items geben konkrete Hinweise auf bestehende Schwierigkeiten in der Bevölkerung. Gleichzeitig kann aus den einzelnen Fragen ein Score errechnet werden. Darüber hinaus wird eine Kategorisierung des Scores in 4 GK-Stufen vorgeschlagen, die den „match“ bzw. „mismatch“ zwischen individuellen Kompetenzen und situativen Anforderungen beschreiben und eine einfache proportionale Charakterisierung der GK in der Bevölkerung anhand der Kategorien inadäquat, problematisch, ausreichend und ausgezeichnet ermöglichen. Der HLS 19 -Q12 wurde im Rahmen der M‑POHL-HLS 19 -Studie für 17 Länder validiert und weist – ebenso wie der HLS-EU-Q47 und seine Kurzformen – eine gute Inhalts- und Kriteriumsvalidität auf, wobei letztere für gesundheitsrelevante Verhaltensweisen, Gesundheitsindikatoren und die Inanspruchnahme professioneller Gesundheitsdienste nachgewiesen wurde . Seine faktorielle Validität wurde durch konfirmatorische Faktorenanalysen (CFA) und vereinzelt auch durch Rasch-Analysen auch über HLS 19 hinaus bestätigt. Das länderweise berechnete einfaktorielle CFA-Modell zeigt in allen Ländern (darunter auch Deutschland und Österreich) eine gute Modellanpassung (CFI ≥ 0,97, TLI ≥ 0,96 und in 16 von 17 Ländern RMSEA-Werte ≤ 0,07). Mit Cronbachs-Alpha-Koeffizienten über 0,8 ist zudem eine gute interne Konsistenz gegeben . Die Konvergenzvalidität des HLS 19 -Q12 wurde bisher nur in der Studie von Liu et al. im Vergleich zum HLQ untersucht. Dabei korreliert der HLS 19 -Q12-Score positiv im moderaten Bereich (zwischen r = 0,24 und r = 0,42; Pearson-Korrelation) mit den 9 HLQ-Skalen. Allerdings ist zu beachten, dass sich der HLS 19 -Q12 und der HLQ inhaltlich nur teilweise überschneiden. Die Kürze des Fragebogens ermöglicht einen flexiblen Einsatz sowohl in Studien als auch in Evaluationen und eignet sich hervorragend für Monitoringzwecke. Sie ermöglicht auch, neben der allgemeinen GK weitere Aspekte der GK zu berücksichtigen. Im Rahmen der HLS 19 -Studie wurden daher optionale Fragensets entwickelt und angeboten. Diese basieren auf den gleichen methodischen Prinzipien wie der HLS 19 -Q12 und können daher hinsichtlich ihrer Ergebnisse mit der allgemeinen GK verglichen werden. So wurden im HLS 19 Daten zur digitalen GK , zur kommunikativen GK , zur navigationalen GK und zur impfbezogenen GK erhoben. In Kombination mit der allgemeinen GK ermöglichen diese spezifischen Instrumente eine umfassende Analyse der GK in der Bevölkerung . Für den M‑POHL Health Literacy Survey 2024–2026 (HLS 24 ) wurden die HLS 19 -Instrumente zur Messung spezifischer GK weiterentwickelt und um ein Instrument zur psychischen GK ergänzt. Der HLS 19 -Q12 wurde bisher in mehr als 30 Sprachen übersetzt, wobei weitere Übersetzungen in Arbeit sind. Er ist für verschiedene Erhebungsmethoden geeignet (persönliche Interviews, Telefoninterviews, Onlinebefragungen, Paper-Pencil-Befragungen) und einfach in der Anwendung. Die durchschnittliche Bearbeitungszeit beträgt etwa 2–3 min. Der HLS 19 -Q12 steht für Forschung, Evaluationen und für die nichtkommerzielle Nutzung, z. B. durch Gesundheitsdienste, zur Verfügung und wird im Rahmen der M‑POHL Health Literacy Surveys und der European Joint Action Prevent Non-Communicable Diseases eingesetzt. Der HLS 19 -Q12 kann kostenlos über das M‑POHL International Coordination Center bezogen werden. GK ist eine zentrale Determinante von Gesundheit, ein wichtiger Hebel für mehr gesundheitliche Chancengerechtigkeit und eine Voraussetzung für selbstbestimmte Entscheidungen in Gesundheitsfragen . Eine geringe GK geht mit einem ungünstigen Gesundheits‑, Risiko-, Präventions- und Krankheitsverhalten, einem schlechteren Gesundheitszustand und einer höheren Sterblichkeit sowie einer inadäquaten und erhöhten Inanspruchnahme des Gesundheitssystems und höheren Kosten in der Krankenbehandlung einher . Sie ist in der Bevölkerung ungleich verteilt und wird infolge ökologischer und gesellschaftlicher Dynamiken (Klimawandel, Naturkatastrophen, Pandemien, alternde Gesellschaft, Digitalisierung und steigende Gesundheitskosten) immer wichtiger. Im Gegensatz zu anderen sozialen Determinanten von Gesundheit ist die GK beeinflussbar, sei es durch Interventionen zur Erhöhung der individuellen Kompetenzen oder durch Maßnahmen, die darauf abzielen, die Anforderungen zur Nutzung gesundheitsrelevanter Informationen und Angebote zu reduzieren . Daten zur GK tragen dazu bei, GK auf die (politische) Agenda zu setzen. In Österreich hat beispielsweise das schlechte Abschneiden in der EU-HLS-Studie dazu geführt, dass der GK ein eigenes Gesundheitsziel gewidmet (gesundheitsziele-oesterreich.at), die Österreichische Plattform GK gegründet (oepgk.at) und die GK nachhaltig in der Gesundheitsreform (Zielsteuerung-Gesundheit) verankert wurde. Die Messung von GK ermöglicht zudem, Herausforderungen und Zielgruppen zu identifizieren und Entwicklungen zu beobachten. Dies ermöglicht, die Planung und Umsetzung gezielter und zielgruppenspezifischer GK-Maßnahmen. Die Auswahl eines geeigneten Messinstruments für Forschung, Evaluation oder Monitoring orientiert sich zunächst am Verständnis von GK. Hier zeigt sich, dass vor allem Instrumente, die ein breites Verständnis von GK operationalisieren, in den Fokus gerückt sind. Hierbei handelt es sich vor allem um Selbsteinschätzungsinstrumente. Im Gegensatz zu performanceorientierten Instrumenten, die auf die funktionale GK fokussieren, erfassen sie alle Aspekte einer umfassenden GK und berücksichtigen teilweise auch das relationale Verständnis von GK. Das oft kritisierte „Rauschen“ in Selbstberichtsdaten verweist in der Regel auf die Selbstwirksamkeit der Befragten und steht der Nützlichkeit der Daten und Ergebnisse nicht entgegen, da Selbsteinschätzungen das eigene Handeln bestimmen und damit alltagsrelevant werden. Bei der Interpretation der Ergebnisse gilt es aber, diese Selbsteinschätzungseffekte zu berücksichtigen. Die Vielzahl der zur Verfügung stehenden Instrumente ermöglicht die gezielte Auswahl eines „passenden“ GK-Instruments, erschwert aber die Bereitstellung einer vergleichbaren Datenbasis zur Generierung von Evidenz. Es empfiehlt sich daher, in Studien, Evaluationen oder für Monitoringzwecke Instrumente zu verwenden, die bereits gut etabliert sind. In Bezug auf die allgemeine GK sind dies der HLS 19 -Q12 bzw. -Q47 und diesbezügliche Vorgängerinstrumente sowie der HLQ. Beide Instrumente sind gut validiert, weitverbreitet und liegen auch in deutscher Sprache vor. Während der HLS 19 -Q12 bzw. -Q47 eine starke Public-Health-Orientierung aufweist, fokussiert der HLQ stärker auf die Krankenbehandlung. Die Tatsache, dass beide Instrumente bereits in zahlreichen Sprachen verfügbar sind und in vielen Ländern eingesetzt wurden, ermöglicht darüber hinaus ihren Einsatz in mehrsprachigen Studien und ggf. den Vergleich kleinerer regionaler Studien oder Evaluationsstudien mit repräsentativen nationalen Daten. |
Establishment of a forward primers-superposed amplification analysis for accurate aspirin pharmacogenomic measurement | 8403a525-72bf-49a0-9fb9-6d55f400fac8 | 10776573 | Pharmacology[mh] | Cardiovascular diseases, consisting of hypertension, hyperlipidemia, coronary heart disease, atherosclerosis, etc., have become the top killers of human health – . The prevention and treatment of cardiovascular diseases are increasingly becoming a focus of attention in medical field. Because of multifaceted causes and manifestations, the control of cardiovascular disease demands the use of compound formulation, commonly called polypills, which can ominously decrease the incidence of cardiovascular diseases , especially cerebral apoplexy and myocardial infarct, by 35–50% – . The components of the polypill comprise statins, aspirin and a kind of antihypertensive drug . The polypill concept is increasingly being recognized and accepted in clinical practice. As a platelet aggregation antagonist, aspirin is widely used to decline the risk of suspected patients with acute myocardial infarction, and prevent cardiovascular diseases such as recurrence of myocardial infarction, stroke and atherosclerosis – . In vivo, aspirin can inactivate cyclooxygenase (COX) by irreversible acetylating the hydroxyl group of serine residues in the active part of COX – . This impedes platelet aggregation through constraining both the metabolism of arachidonic acid (AA) and the production of thromboxane A2 (TXA2) , – . Aspirin resistance (AR) and adverse reactions were found in some patients who took routine dose of aspirin . AR is defined as a lower-than-expected inhibition effect of aspirin on platelet aggregation, raising the risk of recurrent cerebral infarction and other cerebrovascular events – . Clinic studies exhibited a 5–60% incidence of AR and a 1.5% frequency of aspirin hypersensitivity causing allergy and hemorrhage in patients with cardiovascular disease – . Aspirin pharmacogenomics evidenced that genetic polymorphism of PEAR1 (rs12041331), GP1BA (rs6065) and LTC4S (rs730012) is associated with AR and adverse reactions – . In 2020, a scoring table for genotyping of gDNA rs12041331 (PEAR1), rs6065 (GP1BA) and rs730012 (LTC4S) was proposed to provide systematic guidance for aspirin administration . However, an accurate and reliable approach to simultaneous examination of the above single nucleotide polymorphisms (SNPs) is not reported. Herein, we designed and substantiated an allele-specific (AS) forward primer-superposed amplification analysis for discrimination of the SNPs in PEAR1 (rs12041331), GP1BA (rs6065) and LTC4S (rs730012) genes. The results insinuated that this analysis is a valuable tool to facilitate personalized antiplatelet therapy. Design strategy First, mismatch AS F-primers were screened with singleplex amplification analysis. Next, the selected F-primers-based triplicate analysis was optimized by F-primer superposition to avoid undetermined results. Then, the optimized analysis was validated by robustness assessment and precision evaluation, as well as agreement analysis compared with Sanger sequencing. The values of ∆Cq (differences in threshold cycles between the wild-type F-primer-based amplification assay and the mutated-type F-primer-based amplification assay) were calculated to decide the outcomes. DNA extraction from buccal swab The human buccal swab samples used in this study involved 189 Chinese volunteers, which was not a train set classified by aspirin resistance. This study was approved by the Biomedical Research Ethic Committee of Shandong Provincial Hospital (No.2023-417) and has been conducted in accordance with ethical standards and guidelines of the Biomedical Research Ethic Committee of Shandong Provincial Hospital. Authors of this work extend a statement assuring that this work was conducted in accordance with the Declaration of Helsinki and obtained informed consent from all participants. Genomic DNA was extracted by using the QIAamp DNA Mini kit (Cat No. 51304, QIAGEN, Dusseldorf, Germany), and the procedure was carried out according to the instructions. DNA concentration was examined with a NanoPhotometer P360 (Implen GmbH, Munich, Germany). The quality was determined by using OD 260 / 280 ratio. Sanger sequencing was conducted by Personal Biotechnology Co., Ltd (Qingdao, China). Primers and probes AS F-primers, reverse primers and hydrolysis probes were designed using Primer Express 3.0, based on the information of the whole gene sequence. The second or fifth mismatch base was introduced at the 3′ end of F-primers, which were screened in a subsequent process. Probes for rs12041331 (PEAR1), rs6065 (GP1BA) and rs730012 (LTC4S) were labelled at the 5′ end with the fluorescent dye FAM, VIC, and NED, respectively, and at the 3′ end with the quencher BHQ1, BHQ1, and BHQ3, respectively. The oligonucleotide was synthesized by Sangon Biotech (Shanghai, China). Real- time amplification assay Triplex amplification analysis (TaqMan qPCR) was executed in a total of 20 μL reaction mixture, which contained 10 μL AceQ® Universal U + Probe Master Mix V2 (Vazyme, Nanjing, China), 0.2 μM of each wild/mutated-type F-primer, 0.2 μM of each reverse primer, 0.1 μM of hydrolysis probe and 10 ng DNA template. The singleplex amplification analysis was conducted according to the same protocol. A robotic liquid handling workstation (epMotion 5075 vt, Germany) was utilized to dispense the mix. The reaction protocols started with a contamination digestion step for 2 min at 37℃ and a pre-denaturation step for 5 min at 95 ℃, followed by 45 cycles of 95 ℃ for 10 s, and 60 ℃ for 35 s. Fluorescence data were collected at 60 ℃. These amplifications were performed on the ABI7500 Real-Time PCR Instrument (ThermoFisher Scientific Inc., MA, USA). Data analysis GraphPad Prism software version 9.5 (GraphPad Software, Inc., San Diego, CA) was used to conduct data analysis and graphing. First, mismatch AS F-primers were screened with singleplex amplification analysis. Next, the selected F-primers-based triplicate analysis was optimized by F-primer superposition to avoid undetermined results. Then, the optimized analysis was validated by robustness assessment and precision evaluation, as well as agreement analysis compared with Sanger sequencing. The values of ∆Cq (differences in threshold cycles between the wild-type F-primer-based amplification assay and the mutated-type F-primer-based amplification assay) were calculated to decide the outcomes. The human buccal swab samples used in this study involved 189 Chinese volunteers, which was not a train set classified by aspirin resistance. This study was approved by the Biomedical Research Ethic Committee of Shandong Provincial Hospital (No.2023-417) and has been conducted in accordance with ethical standards and guidelines of the Biomedical Research Ethic Committee of Shandong Provincial Hospital. Authors of this work extend a statement assuring that this work was conducted in accordance with the Declaration of Helsinki and obtained informed consent from all participants. Genomic DNA was extracted by using the QIAamp DNA Mini kit (Cat No. 51304, QIAGEN, Dusseldorf, Germany), and the procedure was carried out according to the instructions. DNA concentration was examined with a NanoPhotometer P360 (Implen GmbH, Munich, Germany). The quality was determined by using OD 260 / 280 ratio. Sanger sequencing was conducted by Personal Biotechnology Co., Ltd (Qingdao, China). AS F-primers, reverse primers and hydrolysis probes were designed using Primer Express 3.0, based on the information of the whole gene sequence. The second or fifth mismatch base was introduced at the 3′ end of F-primers, which were screened in a subsequent process. Probes for rs12041331 (PEAR1), rs6065 (GP1BA) and rs730012 (LTC4S) were labelled at the 5′ end with the fluorescent dye FAM, VIC, and NED, respectively, and at the 3′ end with the quencher BHQ1, BHQ1, and BHQ3, respectively. The oligonucleotide was synthesized by Sangon Biotech (Shanghai, China). Triplex amplification analysis (TaqMan qPCR) was executed in a total of 20 μL reaction mixture, which contained 10 μL AceQ® Universal U + Probe Master Mix V2 (Vazyme, Nanjing, China), 0.2 μM of each wild/mutated-type F-primer, 0.2 μM of each reverse primer, 0.1 μM of hydrolysis probe and 10 ng DNA template. The singleplex amplification analysis was conducted according to the same protocol. A robotic liquid handling workstation (epMotion 5075 vt, Germany) was utilized to dispense the mix. The reaction protocols started with a contamination digestion step for 2 min at 37℃ and a pre-denaturation step for 5 min at 95 ℃, followed by 45 cycles of 95 ℃ for 10 s, and 60 ℃ for 35 s. Fluorescence data were collected at 60 ℃. These amplifications were performed on the ABI7500 Real-Time PCR Instrument (ThermoFisher Scientific Inc., MA, USA). GraphPad Prism software version 9.5 (GraphPad Software, Inc., San Diego, CA) was used to conduct data analysis and graphing. Screening of mismatch AS F-primers by singleplex amplification analysis AS F-primers with a second or fifth mismatch base at 3' terminus were screened by detection of homozygote/heterozygote using singleplex real-time amplification analysis (10 ng DNA/test). And ΔCq (differences in threshold cycles between the wild-type F-primer-based amplification assay and the mutated-type F-primer-based amplification assay) was utilized to determine genotype. The principle for selection of the F-primer is the following: (a) no undetermined result was observed. (b) the Cq value was approximately 35 when wild homozygotes were detected in mutated-type F-primer-based amplification assay or when mutated homozygotes were measured in wild-type F-primer-based amplification assay. The original Cq values obtained from singleplex amplification analysis were shown in Table . And the selected sequences were shown in Table . PCR-based analysis is a convenient tool to discriminate SNPs. Genetic polymorphism specific-binding molecules in PCR-based analysis comprise dsDNA-binding dye, AS probe and primer , . The dsDNA-binding dye-based high-resolution dissolution curve (HRM) assay needs specific equipment module. Besides diseconomy, it is time-consuming and laborious to discover appropriate Minor Groove Binder (MGB) probe , . For enhancement of AS primer specificity, base mismatch is more economic than locked nucleic acid (LNA) decoration . In present study, the mismatch AS primers as polymorphism specific-binding molecules were screened and utilized to discriminate homozygotes/heterozygotes for rs12041331 (PEAR1), rs6065 (GP1BA) and rs730012 (LTC4S). Development of triplex amplification analysis optimized by F-primer superposition Based on the selected mismatch AS F-primers, we developed a triplex amplification analysis optimized by F-primer superposition. Undetermined results were observed when homozygotes were measured by un-optimized triplex amplification analysis (Fig. A ) . The extension of Cq was forced by the mismatch AS F-primer superposition, which was implemented with the addition of 0.01 μM mutated/wild-type F-primer into 0.2 μM wild/mutated-type F-primer-based amplification assay (Fig. B ) . The results showed that all outcomes of homozygotes were positive, suggesting that the mismatch AS F-primer superposition can improve detective convenience via omitting positive controls in the triplex amplification analysis. Robustness assessment of triplex amplification analysis To assess robustness of the triplex amplification analysis, the heterozygote was gradually reduced to generate gDNA samples at levels of 40 ng, 20 ng, 10 ng, 5 ng, 2.5 ng and 1.25 ng. Reactions were run in duplicate with three independent experiments. We used the following formula to calculate the amplification efficiency: 10 −1/slope − 1, when the logarithm of the template concentration was plotted on the x- axis and Cq was plotted on the y- axis. The results demonstrated that the amplification efficiency calculated from standard curve ranged from 0.9 to 1.1 (Fig. ), and limit of detection (LOD) was at least 1.25 ng/test. Precision evaluation of triplex amplification analysis The precision of the triplex amplification analysis was evaluated by detection of genomic DNA at 10 ng/test and 2.5 ng/test levels. Each specimen was tested in eight plicates by two operators with two reagent lots every day over 5 days (n = 80/specimen) at one site. A total of eighty Cq values were collected to calculate the coefficient of variance (CV). The results revealed that CV value was < 2% for all days, specimens, replicates, operators and reagent lots combined. Figure shows the intra-day CV for PEAR1(rs12041331), GP1BA (rs6065) and LTC4S (rs730012). Agreement analysis between triplex amplification analysis and Sanger sequencing We conducted the triplex amplification analysis on each of 189 samples, in which 89, 165, and 136 specimens were defined as homozygotes for rs12041331 (PEAR1), rs6065 (GP1BA) and rs730012 (LTC4S), respectively, while 100, 24, and 53 samples were defined as heterozygotes for rs12041331 (PEAR1), rs6065 (GP1BA) and rs730012 (LTC4S), respectively. The cut-off values for genotyping were described in Table . Pharmacogenomics appears that some SNP s are more likely to initiate AR and adverse reactions , . Behaving as a kind of platelet transmembrane protein, the platelet endothelial aggregation receptor 1 (PEAR1) plays an important role in platelet aggregation. And genetic polymorphism of the rs12041331 in PEAR1 gene can obviously affect the inhibitive effect of aspirin on platelet aggregation . Glycoprotein Ib-alpha (GP1BA) gene encodes platelet surface membrane glycoprotein (GPIb) that is a heterodimer consisting of bisulfide-linked α and β subunits, and acts as a receptor for von Willebrand factor(VWF) . Genetic polymorphism of the rs6065 in GP1BA gene was evidenced to correlate aspirin resistance , . It was documented that patient carrying C-type allele for rs730012 in leukotriene C4 synthase (LTC4S) gene are prone to aspirin-induced urticaria , . Consulting to a scoring table proposed by Guangdong Pharmaceutical Association (Guangzhou, China) (Table ), a triplex amplification analysis to detect genetic polymorphism of gDNA rs12041331 (PEAR1), rs6065 (GP1BA) and rs730012 (LTC4S) was designed and substantiated in this study. The results of the agreement analysis indicated the genotyping outlined by the triplex amplification analysis is consistent with the results obtained from Sanger sequencing. In summary, we established a simple, efficient and accurate approach to the determination of genetic polymorphism of gDNA rs12041331 (PEAR1), rs6065 (GP1BA) and rs730012 (LTC4S), which can be used to guide aspirin delivery to reduce AR and adverse reaction. AS F-primers with a second or fifth mismatch base at 3' terminus were screened by detection of homozygote/heterozygote using singleplex real-time amplification analysis (10 ng DNA/test). And ΔCq (differences in threshold cycles between the wild-type F-primer-based amplification assay and the mutated-type F-primer-based amplification assay) was utilized to determine genotype. The principle for selection of the F-primer is the following: (a) no undetermined result was observed. (b) the Cq value was approximately 35 when wild homozygotes were detected in mutated-type F-primer-based amplification assay or when mutated homozygotes were measured in wild-type F-primer-based amplification assay. The original Cq values obtained from singleplex amplification analysis were shown in Table . And the selected sequences were shown in Table . PCR-based analysis is a convenient tool to discriminate SNPs. Genetic polymorphism specific-binding molecules in PCR-based analysis comprise dsDNA-binding dye, AS probe and primer , . The dsDNA-binding dye-based high-resolution dissolution curve (HRM) assay needs specific equipment module. Besides diseconomy, it is time-consuming and laborious to discover appropriate Minor Groove Binder (MGB) probe , . For enhancement of AS primer specificity, base mismatch is more economic than locked nucleic acid (LNA) decoration . In present study, the mismatch AS primers as polymorphism specific-binding molecules were screened and utilized to discriminate homozygotes/heterozygotes for rs12041331 (PEAR1), rs6065 (GP1BA) and rs730012 (LTC4S). Based on the selected mismatch AS F-primers, we developed a triplex amplification analysis optimized by F-primer superposition. Undetermined results were observed when homozygotes were measured by un-optimized triplex amplification analysis (Fig. A ) . The extension of Cq was forced by the mismatch AS F-primer superposition, which was implemented with the addition of 0.01 μM mutated/wild-type F-primer into 0.2 μM wild/mutated-type F-primer-based amplification assay (Fig. B ) . The results showed that all outcomes of homozygotes were positive, suggesting that the mismatch AS F-primer superposition can improve detective convenience via omitting positive controls in the triplex amplification analysis. To assess robustness of the triplex amplification analysis, the heterozygote was gradually reduced to generate gDNA samples at levels of 40 ng, 20 ng, 10 ng, 5 ng, 2.5 ng and 1.25 ng. Reactions were run in duplicate with three independent experiments. We used the following formula to calculate the amplification efficiency: 10 −1/slope − 1, when the logarithm of the template concentration was plotted on the x- axis and Cq was plotted on the y- axis. The results demonstrated that the amplification efficiency calculated from standard curve ranged from 0.9 to 1.1 (Fig. ), and limit of detection (LOD) was at least 1.25 ng/test. The precision of the triplex amplification analysis was evaluated by detection of genomic DNA at 10 ng/test and 2.5 ng/test levels. Each specimen was tested in eight plicates by two operators with two reagent lots every day over 5 days (n = 80/specimen) at one site. A total of eighty Cq values were collected to calculate the coefficient of variance (CV). The results revealed that CV value was < 2% for all days, specimens, replicates, operators and reagent lots combined. Figure shows the intra-day CV for PEAR1(rs12041331), GP1BA (rs6065) and LTC4S (rs730012). We conducted the triplex amplification analysis on each of 189 samples, in which 89, 165, and 136 specimens were defined as homozygotes for rs12041331 (PEAR1), rs6065 (GP1BA) and rs730012 (LTC4S), respectively, while 100, 24, and 53 samples were defined as heterozygotes for rs12041331 (PEAR1), rs6065 (GP1BA) and rs730012 (LTC4S), respectively. The cut-off values for genotyping were described in Table . Pharmacogenomics appears that some SNP s are more likely to initiate AR and adverse reactions , . Behaving as a kind of platelet transmembrane protein, the platelet endothelial aggregation receptor 1 (PEAR1) plays an important role in platelet aggregation. And genetic polymorphism of the rs12041331 in PEAR1 gene can obviously affect the inhibitive effect of aspirin on platelet aggregation . Glycoprotein Ib-alpha (GP1BA) gene encodes platelet surface membrane glycoprotein (GPIb) that is a heterodimer consisting of bisulfide-linked α and β subunits, and acts as a receptor for von Willebrand factor(VWF) . Genetic polymorphism of the rs6065 in GP1BA gene was evidenced to correlate aspirin resistance , . It was documented that patient carrying C-type allele for rs730012 in leukotriene C4 synthase (LTC4S) gene are prone to aspirin-induced urticaria , . Consulting to a scoring table proposed by Guangdong Pharmaceutical Association (Guangzhou, China) (Table ), a triplex amplification analysis to detect genetic polymorphism of gDNA rs12041331 (PEAR1), rs6065 (GP1BA) and rs730012 (LTC4S) was designed and substantiated in this study. The results of the agreement analysis indicated the genotyping outlined by the triplex amplification analysis is consistent with the results obtained from Sanger sequencing. In summary, we established a simple, efficient and accurate approach to the determination of genetic polymorphism of gDNA rs12041331 (PEAR1), rs6065 (GP1BA) and rs730012 (LTC4S), which can be used to guide aspirin delivery to reduce AR and adverse reaction. |
Artificial intelligence in gynecology and obstetrics: from the enthusiasm of use in practice to the challenges of implementation | 519e3651-299d-4ff3-80af-5481551146cf | 11460420 | Gynaecology[mh] | Artificial intelligence (AI) and Machine Learning (ML) have been the subject of discussion among many professionals, researchers, and managers working in the fields of gynecology and obstetrics. In a simple search in the MEDLINE database with the descriptors "artificial intelligence" OR "machine learning", we identified 278,723 studies published until December 2023 on the topic. By delimiting the searches to the area of gynecology and obstetrics, in the same database, from the application of the search strategy ((machine learning) OR (artificial intelligence) AND (Obstetrics OR Gynecology)) it was possible to identify 3,359 studies published until December 2023. We noted that the number of studies related to the topic published in the last five years (2018 to 2023) increased almost four times compared to the previous five-year period (2012 to 2017) . Given the information presented above, the growing interest in investigating the application of AI and ML in the fields of Gynecology and Obstetrics becomes evident. This interest possibly arises from the potential benefits of using these methods in clinical practice, among which we can mention: the possibility of analyzing structured data (imaging exams, genetic and electrophysiological tests, etc.); reduction of costs, and diagnostic and therapeutic errors inherent to practical human activity; and extracting information from a large population to make inferences about disease risk or predicting outcomes. However, given the enthusiasm surrounding the use of an emerging method with considerable potential in clinical practice, it is important to note that its implementation depends on overcoming some challenges involving the availability, accuracy, integrity, and security of health data. However, it is observed that the findings in the scientific literature on the barriers and facilitators regarding the implementation of AI and ML in the areas of Gynecology and Obstetrics are still very discreet. Furthermore, the discussion between researchers, clinicians, and health managers is still very early. Therefore, despite the growing trend of scientific investigation into these resources in recent years, there is still an urgent need to intensify the evidence-based dialogue about the effects and implementation of AI and ML in practice. Therefore, professionals, researchers, and managers must be encouraged to develop a task force to expand knowledge about AI and ML applied to gynecological and obstetric care, in addition to encouraging them to discuss global and local development challenges. and implementation of AI, the responsibility, guarantee, and security of recording health data, as well as the cost-effectiveness and ethical aspects related to the use of these tools. |
Enhancing pharmacogenomic data accessibility and drug safety with large language models: a case study with Llama3.1 | ef1239a8-1e92-4f12-bfe1-4b5997382609 | 11650518 | Pharmacology[mh] | This study demonstrates the utility of Large Language Models (LLMs), specifically Llama3.1-70B, in automating pharmacogenomic (PGx) data extraction, addressing the limitations of traditional manual methods that are labor-intensive and slow to update. By achieving high accuracy in identifying drug-biomarker pairs and integrating diverse data sources, this work offers a practical solution for pharmacologists, regulatory agencies, and healthcare professionals to streamline PGx database updates. With automated extraction processes, LLMs reduce the time and effort required to incorporate new PGx insights, potentially enabling updates at a frequency and scale that were previously unfeasible. This capability is critical for translating PGx research into actionable, personalized treatment guidelines that reflect the genetic diversity of patient populations, ultimately advancing equity in personalized medicine. Pharmacogenomics (PGx) represents a pivotal advancement in personalized medicine, tailoring drug therapies based on an individual’s genetic profile . By understanding how genetic variations influence drug response, PGx enables healthcare providers to optimize treatment efficacy and minimize adverse drug reactions . This personalized approach holds the potential to significantly enhance patient outcomes, especially in the management of complex diseases such as cancer, cardiovascular disorders, and mental health conditions . The importance of PGx lies in its ability to provide more precise and effective treatments. For instance, variations in genes encoding drug-metabolizing enzymes, drug transporters, and drug targets can greatly influence a patient’s response to certain medications. These genetic differences can determine whether a patient will benefit from a particular drug, experience no effect, or suffer from adverse reactions . Despite its promise, the clinical implementation of PGx has been slower than anticipated, partly due to the complexity of drug-gene interactions and the need for extensive empirical evidence . As our understanding of genetic factors in drug response continues to grow, PGx is poised to become a standard component of healthcare, revolutionizing the way treatments are tailored to individual patients. Various databases and resources for PGx information have been established to improve the accessibility and utility of this data. Key resources include the Pharmacogenomics Knowledgebase (PharmGKB), which curates information about how genetic variations affect drug response . The pharmacogenomics database (PGxDB) database offers a comprehensive platform for integrating PGx data, allowing researchers to explore drug, target, and disease relationships . Additionally, the FDA has released the Table of Pharmacogenomic Biomarkers in Drug Labeling ( Table of Pharmacogenomic Biomarkers in Drug Labeling | FDA ), providing drug and PGx biomarker pairs found in given drug labeling sections which serves as the primary data source for this study. Meanwhile, PGx-related research articles containing new findings and conclusions are crucial for timely updating of current PGx information. For instance, relevant abstracts can be retrieved from PubMed or other resources. These resources are essential for advancing the field of PGx and ensuring that clinicians have the necessary tools to apply genetic insights to patient care. Large Language Models (LLMs) like Llama3.1 represent a significant advancement in natural language processing, offering powerful capabilities for extracting and analyzing complex data from diverse sources. These models, trained on vast amounts of text, can understand and generate human-like language, making them highly effective for tasks such as data extraction, summarization, and information synthesis . Recent studies have demonstrated the potential of LLMs in various fields, including PGx. For instance, LLMs have been shown to significantly improve the efficiency and accuracy of data extraction processes, and AI assistant showed improved efficacy in answering user questions . By leveraging these models, researchers can automate the extraction of PGx information, overcoming challenges related to the time-consuming and labor-intensive nature of manual data processing. In this study, we focused on evaluating the capabilities of LLMs, particularly Llama3.1-70B , for PGx information extraction from various data sources. Our goal was to enhance the current PGx information collection by improving its accuracy and incorporating recent studies to fill in gaps and ensure the data is comprehensive. It was essential to ensure that the model could reliably identify and extract key PGx data, such as drugs and related biomarkers, from diverse sources with a remarkable degree of precision. The model demonstrated a high accuracy rate of 91.4% when extracting information from structured texts in the FDA PGx Biomarker table and 82% from the mixed texts, underscoring its effectiveness in handling different types of data. A key aspect of our study was the integration of diverse resources, including well-structured databases like the FDA PGx Biomarker table, alongside relevant scientific abstracts. By combining these sources, we were able to cross-validate and enrich the PGx data, ensuring a more comprehensive, accurate, and up-to-date dataset, particularly with insights related to underrepresented populations and novel drug-biomarker interactions. The results can better support personalized medicine initiatives and enhance the overall effectiveness of pharmacogenomic applications. Data processing for the FDA PGx biomarker table The FDA PGx Biomarker table (06/2023 version) was downloaded in PDF format and converted into one Excel table. All the special characters were then removed from the texts. Biomarkers with multiple gene names or aliases were further processed to ensure all the entries were retained. For instance, for the listed biomarker ERBB2 (HER2), either ERBB2 or HER2 identified by the model was considered a correct identification. To ensure there was sufficient content from which the model could extract information, labeling texts in the FDA PGx Biomarker table with fewer than 300 words were removed from the analysis. Prompt and model settings The Llama3.1-70B-Instruct model was employed in this study for the PGx information extraction and summarization. The model was run using its default settings. We utilized the “client.chat.completions.create” function to interact with the model and obtain the responses. To guide the model effectively, we set the system context as: “You are an expert in pharmacogenetics and assist me in extracting information from texts.” This context was designed to align the model’s responses with the specialized nature of the task. The PGx texts from the PGx Biomarker table that required information extraction, along with specific questions, were provided in the prompt as user content. For example, a typical prompt would be: “ Please review this labeling text and identify the pairs of drug and biomarker clearly mentioned. Output the pairs in ‘drug-biomarker’ format. Please try to give me both the generic name and brand name of the drug. ” As a result, the model may identify multiple drug-biomarker pairs from the query texts, and we consider the extraction correct if the listed pair is included in the results. The prompt we used to extract PGx information from the label texts was “ Based on this content [texts for information extraction], answer the following questions step-by-step in short answers, only about the drug [drug name] and biomarker [biomarker name] as a pair. Then please generate a horizontal form table with the following items: Phenotypes/Genotypes: Identify the phenotypes (drug response influenced) or genotypes (genetic variants) associated with the biomarker. Frequency by Ethnicity: Provide the frequencies of the identified phenotypes or genotypes by ethnicity. Reason for PGx Labeling: State the reason for pharmacogenomic labeling of the biomarker. ADRs Associated with Biomarker: Identify adverse drug reactions related to the biomarker. Gender Differences: Indicate whether the biomarker is influenced by gender (Yes/No). Ethnicity Differences: Indicate whether drug response differs by ethnicity (Yes/No). Asian Stats: Provide the phenotype or genotype frequency of the biomarker in the Asian population. If no data is available, write ‘No data.’ Black/AA Stats: Provide the phenotype or genotype frequency of the biomarker in the Black population. If no data is available, write ‘No data.’ Hispanic Stats: Provide the phenotype or genotype frequency of the biomarker in the Hispanic population. If no data is available, write ‘No data.’ Polymorphism: Identify the genotype of the biomarker that influences drug response. Summary: Categorize the information using one or more keywords from ‘Therapeutic Use,’ ‘Dosing,’ ‘Drug Response,’ ‘Metabolism,’ and ‘Ethnicity-Specific’. ” Generation of mixed texts To mimic the real-world scientific texts, which often discuss multiple drugs and biomarkers, we generated mixed texts by combining the labeling texts associated with two different drug-biomarker pairs from the FDA PGx Biomarker table. Each labeling text record was divided into five groups by randomly determining where to break the text, always ensuring the breaks occurred at the end of a sentence. This approach preserved the original sequence of sentences within each group. To create a mixed text, we selected these ten groups, five from each of two different segmented records, and merged them. This process allowed us to generate new, coherent mixed texts while blending information from two distinct drug-biomarker pairs . PubMed abstracts query The PubMed API and Entrez library were used to retrieve relevant abstracts based on a given drug-biomarker pair. We requested the title or abstract of one publication to contain both the drug and biomarker. To further narrow down the candidates to ensure the relevance of the collected abstracts, we also required that one of the keyworks, including PGx, pharmacogenomics, minority, variants, mutations, and population, be presented in either the title or abstract. Additionally, if no abstract could be found based on the initial query, we then searched for those abstracts that mentioned only the drug and biomarker. Considering the limitation of prompt length of the Llama3.1-70B model, we collected up to five abstracts for each PGx labeling record. The same prompt was used to extract PGx information from abstracts and from labeling texts. Calculation of concordance rate In this study, we used the concordance rate to measure the extent to which PGx categories (Therapeutic Use, Dosing, Drug Response, Metabolism, and Ethnicity-Specific) identified from the PGx labeling texts were also represented in the relevant abstracts for the same drug-biomarker pair. The concordance rate was calculated using the following formula: C o n c o r d a n c e r a t e = # of PGx categories common to both PGx labeling texts and relevant abstracts # of PGx categories identified in PGx labeling texts This metric provided a clear and quantitative assessment of the overlap between the information in the PGx labeling texts and the scientific abstracts, allowing us to evaluate the consistency and completeness of the extracted data across different sources. The FDA PGx Biomarker table (06/2023 version) was downloaded in PDF format and converted into one Excel table. All the special characters were then removed from the texts. Biomarkers with multiple gene names or aliases were further processed to ensure all the entries were retained. For instance, for the listed biomarker ERBB2 (HER2), either ERBB2 or HER2 identified by the model was considered a correct identification. To ensure there was sufficient content from which the model could extract information, labeling texts in the FDA PGx Biomarker table with fewer than 300 words were removed from the analysis. The Llama3.1-70B-Instruct model was employed in this study for the PGx information extraction and summarization. The model was run using its default settings. We utilized the “client.chat.completions.create” function to interact with the model and obtain the responses. To guide the model effectively, we set the system context as: “You are an expert in pharmacogenetics and assist me in extracting information from texts.” This context was designed to align the model’s responses with the specialized nature of the task. The PGx texts from the PGx Biomarker table that required information extraction, along with specific questions, were provided in the prompt as user content. For example, a typical prompt would be: “ Please review this labeling text and identify the pairs of drug and biomarker clearly mentioned. Output the pairs in ‘drug-biomarker’ format. Please try to give me both the generic name and brand name of the drug. ” As a result, the model may identify multiple drug-biomarker pairs from the query texts, and we consider the extraction correct if the listed pair is included in the results. The prompt we used to extract PGx information from the label texts was “ Based on this content [texts for information extraction], answer the following questions step-by-step in short answers, only about the drug [drug name] and biomarker [biomarker name] as a pair. Then please generate a horizontal form table with the following items: Phenotypes/Genotypes: Identify the phenotypes (drug response influenced) or genotypes (genetic variants) associated with the biomarker. Frequency by Ethnicity: Provide the frequencies of the identified phenotypes or genotypes by ethnicity. Reason for PGx Labeling: State the reason for pharmacogenomic labeling of the biomarker. ADRs Associated with Biomarker: Identify adverse drug reactions related to the biomarker. Gender Differences: Indicate whether the biomarker is influenced by gender (Yes/No). Ethnicity Differences: Indicate whether drug response differs by ethnicity (Yes/No). Asian Stats: Provide the phenotype or genotype frequency of the biomarker in the Asian population. If no data is available, write ‘No data.’ Black/AA Stats: Provide the phenotype or genotype frequency of the biomarker in the Black population. If no data is available, write ‘No data.’ Hispanic Stats: Provide the phenotype or genotype frequency of the biomarker in the Hispanic population. If no data is available, write ‘No data.’ Polymorphism: Identify the genotype of the biomarker that influences drug response. Summary: Categorize the information using one or more keywords from ‘Therapeutic Use,’ ‘Dosing,’ ‘Drug Response,’ ‘Metabolism,’ and ‘Ethnicity-Specific’. ” To mimic the real-world scientific texts, which often discuss multiple drugs and biomarkers, we generated mixed texts by combining the labeling texts associated with two different drug-biomarker pairs from the FDA PGx Biomarker table. Each labeling text record was divided into five groups by randomly determining where to break the text, always ensuring the breaks occurred at the end of a sentence. This approach preserved the original sequence of sentences within each group. To create a mixed text, we selected these ten groups, five from each of two different segmented records, and merged them. This process allowed us to generate new, coherent mixed texts while blending information from two distinct drug-biomarker pairs . The PubMed API and Entrez library were used to retrieve relevant abstracts based on a given drug-biomarker pair. We requested the title or abstract of one publication to contain both the drug and biomarker. To further narrow down the candidates to ensure the relevance of the collected abstracts, we also required that one of the keyworks, including PGx, pharmacogenomics, minority, variants, mutations, and population, be presented in either the title or abstract. Additionally, if no abstract could be found based on the initial query, we then searched for those abstracts that mentioned only the drug and biomarker. Considering the limitation of prompt length of the Llama3.1-70B model, we collected up to five abstracts for each PGx labeling record. The same prompt was used to extract PGx information from abstracts and from labeling texts. In this study, we used the concordance rate to measure the extent to which PGx categories (Therapeutic Use, Dosing, Drug Response, Metabolism, and Ethnicity-Specific) identified from the PGx labeling texts were also represented in the relevant abstracts for the same drug-biomarker pair. The concordance rate was calculated using the following formula: C o n c o r d a n c e r a t e = # of PGx categories common to both PGx labeling texts and relevant abstracts # of PGx categories identified in PGx labeling texts This metric provided a clear and quantitative assessment of the overlap between the information in the PGx labeling texts and the scientific abstracts, allowing us to evaluate the consistency and completeness of the extracted data across different sources. High accuracy achieved with structured labeling texts in the FDA PGx biomarker table We first evaluated the model’s ability to identify drug and biomarker pairs from the labelling texts in the FDA PGx Biomarker table. Each entry contains the drug name, associated biomarker, therapeutic area, and labeling texts. Our analysis focused on the therapeutic area of Oncology, which had the largest number of records in the table . We excluded records with non-gene biomarkers such as chromosome alterations or hormone receptors. As a result, out of 210 drug-biomarker pairs, the model successfully identified 192 pairs, achieving an identification accuracy of 91.4% . Among these, 36 pairs required manual review and confirmation due to discrepancies arising from variations in nomenclature, such as the use of generic versus brand names of drugs or biomarker aliases. For example, the model identified the biomarker MKI67 as Ki-67, where MKI67 refers to the gene encoding the Ki-67 protein, indicating both terms represent the same entity. After manual validation, these 36 pairs missed by exact name matching were confirmed as correctly identified, contributing to the overall count of 192 accurate predictions . By manually reviewing the 18 records where the model failed to identify the drug-biomarker pairs, we found that most of them had short labeling texts in the FDA PGx Biomarker table, sometimes without the drug or biomarker even mentioned, leaving no way for the model to extract them. Another example was the drug brand name LONSURF, which was mentioned in the labeling text column of the PGx Biomarker table, but the listed drug names were tipiracil and trifluridine, the generic names of this drug. For this particular record, the model failed to identify either the brand or generic names. Challenges with mixed texts As Llama3.1-70B demonstrated high accuracy in identifying drug-biomarker pairs from a section of labeling text, we further challenged the model with mixed texts from two records. This approach aimed to mimic the complex content often encountered in scientific studies, where discussions typically involve multiple drugs and biomarkers. To create a mixture testing set, we selected two records, each related to different drugs, and split them by sentences. These sentences were then merged to form a single paragraph, which was subsequently fed to the model (Methods, ). This setup was designed to evaluate the model’s ability to accurately extract relevant drug-biomarker pairs from a less structured and more intricate text, closely resembling real-world scientific documentation. From the 156 records where the model correctly identified the drug-biomarker pairs without manual confirmation, we generated 50 mixture texts for testing (Methods). Using the same prompt and manual validation, we observed that the model could accurately identify at least one drug-biomarker pair for the testing records in 41 out of 50 (82%) cases . Specifically, the model identified all the two drug-biomarker pairs in 32 records (64%), indicating a relatively high level of accuracy even with mixed and more complex text inputs. However, some cases posed significant challenges for the model. For instance, fusion names like BCR-ABL1 were occasionally difficult for the model to identify correctly. Additionally, there were instances where the model misidentified drugs due to the complexity of the text. In one particular case, a record included two drugs: ALIMTA (the brand name for pemetrexed) and pembrolizumab, which was mentioned as a comparator drug in the study. The primary drug for this record was pemetrexed, but the model incorrectly identified pembrolizumab as the paired drug. Notably, the drug-biomarker pair for this challenging case had been correctly identified in previous assessments without the interference of another record. We further evaluated the mis-identified drug-biomarker pairs in the mixture texts by examining cases where the model incorrectly linked the drug and biomarker from two different records. As a result, ten mis-linked drug-biomarker pairs were identified from nine records. The results suggest that the presence of unrelated content may confuse the model, highlighting the need for careful consideration when handling complex and mixed information in texts. Extraction of PGx information related to minority groups Pharmacogenomics information is crucial for understanding how genetic variations influence drug responses across different population groups. Many PGx studies highlight the role that ethnic differences may play in drug efficacy and safety, with some of these findings reflected in labeling documents. Unique genetic profiles that may significantly impact responses to medications have been observed among minority groups, though these profiles remain underexplored. Despite growing awareness of genetic diversity, many minority populations continue to be underrepresented in PGx research, contributing to gaps in personalized medicine. In this study, we collected 178 records from the FDA PGx Biomarker table containing terms such as “American,” “Asian,” “Caucasian.” For each labeling text, we tasked the model to extract PGx information related to race or ethnicity. Key information extracted included the presence of ethnicity differences, frequency of genetic variants by ethnicity, reasons for PGx labeling, and adverse drug reactions (ADRs) associated with biomarkers (as detailed in ). The model demonstrated its effectiveness by accurately identifying crucial details, such as the phenotypes of Poor Metabolizers (PM) and Extensive Metabolizers (EM) for the tolterodine-CYP2D6 pair. It correctly highlighted that the tolterodine labeling indicates approximately 7% of Caucasians and 2% of African Americans were poor metabolizers in that study. It is important to acknowledge that this labeling uses outdated terminology. The terms “White” and “Black/African American” are now preferred. This differentiation is vital for understanding the potential risks of adverse reactions, like QT prolongation, in specific populations . We assessed the model’s accuracy in determining whether there were “ethnicity differences” in the labeling text column. The model was asked to answer a Yes/No question based on whether any information on ethnicity difference was found in the texts (Methods). Of the 178 records analyzed, 94 contained information explicitly stating ethnicity differences. However, some records mentioned the inclusion of diverse minority groups in studies but did not discuss or conclude any differences among these groups. For example, a labeling might state “ 56 of the subjects were male, 61 were White, 20 were Black or African American, 8 were Hispanic or Latino ” but if no comparisons or outcomes were discussed, it should be marked as having no ethnicity difference. We then manually reviewed the records classified by the model as having no ethnicity difference, identifying any false negatives. Impressively, the model achieved 100% accuracy in correctly identifying records that explicitly stated ethnicity differences. This finding underscores the model’s reliability in detecting ethnicity-related PGx information and highlights the importance of ensuring accurate representation and consideration of minority groups in PGx research. This work illustrates the value of using LLMs to systematically and accurately identify PGx information across diverse populations. With appropriate data, LLMs have the potential to retrieve important PGx insights for minority groups from diverse published sources, contributing to more inclusive and equitable healthcare practices. Validation of extracted PGx information The extracted data, encompassing details about drug-biomarker pairs, genetic variations, and ethnicity-specific information, plays a vital role in personalized medicine, which requires high accuracy. While verifying straightforward elements identified by the model, such as the presence or absence of ethnicity differences, is relatively easy, evaluating the detailed PGx information extracted from the texts is challenging due to its complexity. The intricacies involved in interpreting genetic data and its clinical implications require careful consideration. Manually verifying the extracted information would be impractical given the large volume and complexity of the data. Therefore, we implemented a systematic validation process using predefined PGx categories to evaluate the accuracy and consistency of the extracted information. This approach ensured a thorough and efficient assessment, allowing us to confirm the reliability of the model’s outputs. Particularly, when we tasked the model with extracting PGx information from the labeling texts in the FDA PGx Biomarker table, we also required a summary of each record using predefined keywords, including Therapeutic Use, Dosing, Drug Response, Metabolism, and Ethnicity-Specific . For each ethnic PGx record, we collected up to five PubMed abstracts that contained the drug-biomarker pair in the title or abstract. To address concerns that abstracts might focus on different aspects and to narrow down the search to more relevant studies, we included additional keywords such as pharmacogenomics, PGx, and minority, in the PubMed query (Methods). This approach increased the chances of retrieving abstracts that provided the necessary PGx details, ensuring a thorough and focused validation process. As a result, 137 out of 178 ethnic records had at least one abstract found in PubMed that contained the drug-biomarker pairs. The Llama3.1-70B model was then tasked again to tag each individual abstract with the predefined PGx information categories. By comparing the categories from the FDA PGx Biomarker table with those from the relevant abstracts, we evaluated the accuracy and consistency of the extracted information, ensuring alignment with external authoritative sources. A matched PGx category indicates that the particular drug-biomarker pair was studied by different research groups and that similar findings were concluded in the PGx field. Among the 178 ethnic records in the FDA PGx Biomarker table, 125 discussed Drug Response, making it the most frequently mentioned category . Additionally, we found a high consistency in that 78 out of 94 records (83%) identified with Ethnicity Differences were categorized as Ethnicity-Specific. In contrast, only 29 records were related to Dosing. However, the abstracts we collected, which involved the same drugs and biomarkers, exhibited different frequency patterns for these PGx categories . The lower frequency of ethnicity-specific data in the abstracts suggests that this aspect may not be a major focus in the studies we collected. We then calculated the PGx categories concordance rate, defined as the percentage of the categories identified in PGx labeling that were also covered by those from relevant abstracts. To assess the consistency of the extracted information, we compared the highest concordance rate based on a single abstract and the rate based on the aggregated abstract set. The median consistency was over 85% , indicating high accuracy of the PGx information extracted by the LLM. This cross-validation not only confirms the reliability of the model’s extraction capabilities but also highlights the robustness of our methodology in integrating and validating pharmacogenomic data across diverse sources. No big difference was observed between the highest and aggregated concordance rates , suggesting that individual abstracts are sufficiently comprehensive in covering the relevant PGx categories. The approach we used successfully retrieved abstracts that were well-aligned with the information we were interested in from the FDA PGx Biomarker table, ensuring that the abstracts are relevant and valuable for validating the PGx information. The findings indicate that we can use these abstracts complementarily with the labeling texts to potentially extract additional PGx information for certain drug-biomarker pairs. As shown in , we asked the model the same questions based on the integrated texts of the four relevant abstracts (PMIDs: 28087463, 24619889, 22277677, 14606931). Additional PGx information associated with ethnic groups of Japanese and Chinese were found in these abstracts. We first evaluated the model’s ability to identify drug and biomarker pairs from the labelling texts in the FDA PGx Biomarker table. Each entry contains the drug name, associated biomarker, therapeutic area, and labeling texts. Our analysis focused on the therapeutic area of Oncology, which had the largest number of records in the table . We excluded records with non-gene biomarkers such as chromosome alterations or hormone receptors. As a result, out of 210 drug-biomarker pairs, the model successfully identified 192 pairs, achieving an identification accuracy of 91.4% . Among these, 36 pairs required manual review and confirmation due to discrepancies arising from variations in nomenclature, such as the use of generic versus brand names of drugs or biomarker aliases. For example, the model identified the biomarker MKI67 as Ki-67, where MKI67 refers to the gene encoding the Ki-67 protein, indicating both terms represent the same entity. After manual validation, these 36 pairs missed by exact name matching were confirmed as correctly identified, contributing to the overall count of 192 accurate predictions . By manually reviewing the 18 records where the model failed to identify the drug-biomarker pairs, we found that most of them had short labeling texts in the FDA PGx Biomarker table, sometimes without the drug or biomarker even mentioned, leaving no way for the model to extract them. Another example was the drug brand name LONSURF, which was mentioned in the labeling text column of the PGx Biomarker table, but the listed drug names were tipiracil and trifluridine, the generic names of this drug. For this particular record, the model failed to identify either the brand or generic names. As Llama3.1-70B demonstrated high accuracy in identifying drug-biomarker pairs from a section of labeling text, we further challenged the model with mixed texts from two records. This approach aimed to mimic the complex content often encountered in scientific studies, where discussions typically involve multiple drugs and biomarkers. To create a mixture testing set, we selected two records, each related to different drugs, and split them by sentences. These sentences were then merged to form a single paragraph, which was subsequently fed to the model (Methods, ). This setup was designed to evaluate the model’s ability to accurately extract relevant drug-biomarker pairs from a less structured and more intricate text, closely resembling real-world scientific documentation. From the 156 records where the model correctly identified the drug-biomarker pairs without manual confirmation, we generated 50 mixture texts for testing (Methods). Using the same prompt and manual validation, we observed that the model could accurately identify at least one drug-biomarker pair for the testing records in 41 out of 50 (82%) cases . Specifically, the model identified all the two drug-biomarker pairs in 32 records (64%), indicating a relatively high level of accuracy even with mixed and more complex text inputs. However, some cases posed significant challenges for the model. For instance, fusion names like BCR-ABL1 were occasionally difficult for the model to identify correctly. Additionally, there were instances where the model misidentified drugs due to the complexity of the text. In one particular case, a record included two drugs: ALIMTA (the brand name for pemetrexed) and pembrolizumab, which was mentioned as a comparator drug in the study. The primary drug for this record was pemetrexed, but the model incorrectly identified pembrolizumab as the paired drug. Notably, the drug-biomarker pair for this challenging case had been correctly identified in previous assessments without the interference of another record. We further evaluated the mis-identified drug-biomarker pairs in the mixture texts by examining cases where the model incorrectly linked the drug and biomarker from two different records. As a result, ten mis-linked drug-biomarker pairs were identified from nine records. The results suggest that the presence of unrelated content may confuse the model, highlighting the need for careful consideration when handling complex and mixed information in texts. Pharmacogenomics information is crucial for understanding how genetic variations influence drug responses across different population groups. Many PGx studies highlight the role that ethnic differences may play in drug efficacy and safety, with some of these findings reflected in labeling documents. Unique genetic profiles that may significantly impact responses to medications have been observed among minority groups, though these profiles remain underexplored. Despite growing awareness of genetic diversity, many minority populations continue to be underrepresented in PGx research, contributing to gaps in personalized medicine. In this study, we collected 178 records from the FDA PGx Biomarker table containing terms such as “American,” “Asian,” “Caucasian.” For each labeling text, we tasked the model to extract PGx information related to race or ethnicity. Key information extracted included the presence of ethnicity differences, frequency of genetic variants by ethnicity, reasons for PGx labeling, and adverse drug reactions (ADRs) associated with biomarkers (as detailed in ). The model demonstrated its effectiveness by accurately identifying crucial details, such as the phenotypes of Poor Metabolizers (PM) and Extensive Metabolizers (EM) for the tolterodine-CYP2D6 pair. It correctly highlighted that the tolterodine labeling indicates approximately 7% of Caucasians and 2% of African Americans were poor metabolizers in that study. It is important to acknowledge that this labeling uses outdated terminology. The terms “White” and “Black/African American” are now preferred. This differentiation is vital for understanding the potential risks of adverse reactions, like QT prolongation, in specific populations . We assessed the model’s accuracy in determining whether there were “ethnicity differences” in the labeling text column. The model was asked to answer a Yes/No question based on whether any information on ethnicity difference was found in the texts (Methods). Of the 178 records analyzed, 94 contained information explicitly stating ethnicity differences. However, some records mentioned the inclusion of diverse minority groups in studies but did not discuss or conclude any differences among these groups. For example, a labeling might state “ 56 of the subjects were male, 61 were White, 20 were Black or African American, 8 were Hispanic or Latino ” but if no comparisons or outcomes were discussed, it should be marked as having no ethnicity difference. We then manually reviewed the records classified by the model as having no ethnicity difference, identifying any false negatives. Impressively, the model achieved 100% accuracy in correctly identifying records that explicitly stated ethnicity differences. This finding underscores the model’s reliability in detecting ethnicity-related PGx information and highlights the importance of ensuring accurate representation and consideration of minority groups in PGx research. This work illustrates the value of using LLMs to systematically and accurately identify PGx information across diverse populations. With appropriate data, LLMs have the potential to retrieve important PGx insights for minority groups from diverse published sources, contributing to more inclusive and equitable healthcare practices. The extracted data, encompassing details about drug-biomarker pairs, genetic variations, and ethnicity-specific information, plays a vital role in personalized medicine, which requires high accuracy. While verifying straightforward elements identified by the model, such as the presence or absence of ethnicity differences, is relatively easy, evaluating the detailed PGx information extracted from the texts is challenging due to its complexity. The intricacies involved in interpreting genetic data and its clinical implications require careful consideration. Manually verifying the extracted information would be impractical given the large volume and complexity of the data. Therefore, we implemented a systematic validation process using predefined PGx categories to evaluate the accuracy and consistency of the extracted information. This approach ensured a thorough and efficient assessment, allowing us to confirm the reliability of the model’s outputs. Particularly, when we tasked the model with extracting PGx information from the labeling texts in the FDA PGx Biomarker table, we also required a summary of each record using predefined keywords, including Therapeutic Use, Dosing, Drug Response, Metabolism, and Ethnicity-Specific . For each ethnic PGx record, we collected up to five PubMed abstracts that contained the drug-biomarker pair in the title or abstract. To address concerns that abstracts might focus on different aspects and to narrow down the search to more relevant studies, we included additional keywords such as pharmacogenomics, PGx, and minority, in the PubMed query (Methods). This approach increased the chances of retrieving abstracts that provided the necessary PGx details, ensuring a thorough and focused validation process. As a result, 137 out of 178 ethnic records had at least one abstract found in PubMed that contained the drug-biomarker pairs. The Llama3.1-70B model was then tasked again to tag each individual abstract with the predefined PGx information categories. By comparing the categories from the FDA PGx Biomarker table with those from the relevant abstracts, we evaluated the accuracy and consistency of the extracted information, ensuring alignment with external authoritative sources. A matched PGx category indicates that the particular drug-biomarker pair was studied by different research groups and that similar findings were concluded in the PGx field. Among the 178 ethnic records in the FDA PGx Biomarker table, 125 discussed Drug Response, making it the most frequently mentioned category . Additionally, we found a high consistency in that 78 out of 94 records (83%) identified with Ethnicity Differences were categorized as Ethnicity-Specific. In contrast, only 29 records were related to Dosing. However, the abstracts we collected, which involved the same drugs and biomarkers, exhibited different frequency patterns for these PGx categories . The lower frequency of ethnicity-specific data in the abstracts suggests that this aspect may not be a major focus in the studies we collected. We then calculated the PGx categories concordance rate, defined as the percentage of the categories identified in PGx labeling that were also covered by those from relevant abstracts. To assess the consistency of the extracted information, we compared the highest concordance rate based on a single abstract and the rate based on the aggregated abstract set. The median consistency was over 85% , indicating high accuracy of the PGx information extracted by the LLM. This cross-validation not only confirms the reliability of the model’s extraction capabilities but also highlights the robustness of our methodology in integrating and validating pharmacogenomic data across diverse sources. No big difference was observed between the highest and aggregated concordance rates , suggesting that individual abstracts are sufficiently comprehensive in covering the relevant PGx categories. The approach we used successfully retrieved abstracts that were well-aligned with the information we were interested in from the FDA PGx Biomarker table, ensuring that the abstracts are relevant and valuable for validating the PGx information. The findings indicate that we can use these abstracts complementarily with the labeling texts to potentially extract additional PGx information for certain drug-biomarker pairs. As shown in , we asked the model the same questions based on the integrated texts of the four relevant abstracts (PMIDs: 28087463, 24619889, 22277677, 14606931). Additional PGx information associated with ethnic groups of Japanese and Chinese were found in these abstracts. While it is relatively straightforward to validate the extraction of certain PGx items from structured texts, such as drug and biomarker names from labeling sections, assessing the overall quality and completeness of the extracted information from more variable sources poses significant challenges. Unlike structured data, where predefined formats facilitate comparison and validation, publications and reports vary widely in focus and detail, complicating direct comparison of PGx information across different sources. To address this challenge, we employed a strategy where the model was instructed to tag the extracted texts with predefined categories, enabling a more systematic comparison. This tagging approach offers an initial method for aligning information across sources; however, we recognize that these categories may require further refinement or customization based on the specific content and objectives of different studies. Our results demonstrated that Llama3.1-70B achieved high accuracy in extracting drug and biomarker pairs from structured labeling texts, particularly when biomarkers were listed as gene or protein names. However, the model encountered difficulties when extracting less common biomarker names, such as “hormone receptors,” which were excluded from the main analysis due to lower extraction accuracy. This limitation highlights the importance of prompt engineering and model tuning for specific use cases. Tailoring prompts to explicitly account for uncommon biomarkers or providing additional context within the prompt could improve the model’s ability to accurately identify and extract these entities, an approach that warrants further exploration. Identifying drug-biomarker pairs in mixed texts, where multiple records are combined, presents a more complex challenge for LLMs. Our study found that while Llama3.1-70B performed well with structured labeling texts, its accuracy decreased when processing mixed texts, likely due to the increased ambiguity and variety of content. This challenge would likely increase further with full-text publications, where drug-biomarker relationships are not always clearly delineated. To address these complexities, future studies could be benefit from a targeted approach, such as instructing the model to focus on specific drug-biomarker pairs to enhance extraction accuracy. In preliminary tests, the model was able to accurately identify relevant information from mixed texts when a specific drug-biomarker pair was targeted, suggesting that targeted prompts could improve accuracy in more complex texts. Our findings demonstrate that LLMs like Llama3.1-70B can efficiently support the extraction of PGx information from structured sources, such as the FDA PGx Biomarker table, providing a foundation for integrating valuable data from scientific abstracts and potentially, with further refinement, from more complex sources like full-text publications. This automated approach can reduce the time and effort required for initial data extraction, improving the completeness of PGx databases by streamlining the process. However, we recognize that integrating LLM-extracted data directly into regulatory or clinical decision-making frameworks would require extensive validation and quality control, including human oversight, to ensure accuracy and relevance. Implementing a structured workflow that leverages LLMs for routine extraction of PGx data could support the initial stages of database updates. Such a process would involve combining LLM-extracted insights with manual review and verification steps, enhancing the accessibility and usability of PGx data for non-regulatory applications, such as research and exploratory analyses in pharmacogenomics. This framework can be refined to incorporate more sophisticated validation methods, advancing the field of personalized medicine incrementally through a combination of automated and manual processes. Future work will focus on evaluating and refining this workflow to ensure reliability and utility in various PGx contexts. While our study utilizes the Llama3.1-70B model, the primary focus of this work is the development of a generalizable framework for pharmacogenomic (PGx) data extraction. Our approach, which involves structured prompts, data integration techniques, and strategies for handling complex, mixed-text data, is designed to be adaptable to future advancements in LLM technology. As LLMs continue to improve, this framework can be applied to newer models, enabling consistent, automated PGx data extraction and updating without reliance on a specific LLM version. This flexibility makes the framework suitable for various applications in PGx research, supporting the evolving needs of pharmacologists, regulatory bodies, and healthcare researchers. |
Editorial: Insights in cardiovascular endocrinology: 2023 | 21288a94-bd2e-4ffb-98eb-56a992f46222 | 10436608 | Physiology[mh] | GS: Conceptualization, Writing – original draft, Writing – review & editing. |
Molecular and modular intricacies of precision oncology | e73541c3-e276-4262-90c4-1de9c4cd19f2 | 11537923 | Internal Medicine[mh] | Introduction Oncology therapies are commonly designed to target the highly dysregulated molecular pathways, including Ras/MAPK, Myc, Wnt/β-catenin, TGFβ, PI3K/mTOR, Notch signaling, Hippo pathway, cell cycle, oxidative stress response and/or p53 signaling . However, therapeutic resistance poses a constant struggle, whether it is ‘intrinsic’ due to genetic/molecular dysregulations or ‘acquired’ due to cancer cells adapting to the cellular changes . Tumor heterogeneity, complex tumor microenvironment and genetic predisposition have complicated the treatment options further. Personalized treatment approaches are therefore successfully proving to be the present and future of medicine. Precision Oncology is constantly evolving to acknowledge, accept and utilize every human being’s uniqueness, characterized by a distinct set of genetic make-up . As the “one size fits all” theory is challenged at various levels in therapeutic arena, precision medicine has emerged to rescue the unique individual cases , where common FDA approved chemotherapeutics and/or immunotherapy drugs fail to eliminate the cancer cells . As with great power comes bigger economic impact, personalized healthcare requires large sums of investments and some of the underrepresented or minority groups may have limited access to such novel technologies . This coincides with Eroom’s law, which describes the ever-slowing rate of drug discovery and applicability with increasing costs associated with it . This further widens the gap between research and its practical applications . Balancing the resources with medical goals, patient requirements, time involved, and risk assessment is critical. Although there are multiple tools used to support the personalized approach, attempting to reverse the Eroom’s law, one of the approaches gaining traction is incorporation of artificial intelligence/machine learning (AI/ML) into biotechnological advancements . Since 2016, FDA has seen an exponential increase in usage of AI/ML to new oncology clinical trials, in different phases from patient recruitment and precise clinical designs using de-identified electronic health records to data collection and analysis . These technologies provide a major boost to generating customized treatment plans for specific groups/sub-groups/individuals based on the target mutations. AI/ML algorithms can identify complex patterns and correlations by analyzing large datasets, which may not be possible at human/physician level . Targeting the molecular and cellular characteristics of tumors has been the focus of precision medicine for decades . Genetic profiling methods combined with immunophenotyping, transcriptomics and epigenetic analyses assist in de-coding the complex deregulated pathways of tumor microenvironment at a high throughput level, while conventional methods such as FISH (Fluorescence in situ hybridization) and IHC (Immunohistochemistry) are commonly used to detect predictive biomarkers . Some common immunotherapeutic drugs targeting PD-1/PD-L1 (nivolumab), CTLA (ipilimumab), TIGIT (tiragolumab), LAG3 (Relatlimab) are well-studied and used by clinicians . However, in cases concerning rare cancers (such as angiosarcoma , metaplastic breast cancer , high risk, relapsed or refractory pediatric cancers (such as Neuroblastoma , pediatric brain tumor , medulloblastoma, Wilms’ tumor , and resistant cancer sub-types (characterized by overexpression of HER2, Ras/MAPK pathways , customized/personalized cell therapy, gene therapy, immunotherapy, and/or a combination of treatments in a timely manner can successfully aim to prolong symptom free survival in cancer patients . Current landscape of precision oncology therapy Modern clinical medicine relies on the 4Ps, serving as pillars to support therapeutic decision making, namely, Predictive, Preventative, Personalized and Participative approach, focusing on robust treatment options in a patient-centric framework . Treating the patients with the right medicine at the right time is always the clinical goal, however, the concept of “personalized treatment” has evolved within the last few years. The success and FDA approval of HER2-specific breast cancer targeting drug: trastuzumab in 1998 and BCR-ABL tyrosine kinase inhibitor Chronic Myeloid leukemia drug: Imatinib in 2001 were the first major stepping stones in this field, followed by a wide range of gene-targeting treatment options . As cancer is described as both genetic and molecular group of diseases, it became important to encompass other intricate alterations involved, such as epigenetic factors , biomarkers and anatomical/histological modifications to understand the disease progression and design individualized “precision medicine” treatments . Gene and molecular-targeted therapy (designed to target only cancer cells) and Immunotherapy (used to boost body’s immune system against cancer), are the major approaches used individually or in combination with chemotherapy and/or radiotherapy to treat cancer patients. Within last 20 years, a plethora of drugs, including checkpoint inhibitors, monoclonal/bispecific antibodies, antibody-drug conjugates, chimeric antigen receptor T (CAR-T) cells have been developed to combat the complexities of this disease. The approval of blinatumomab, the first bispecific antibody in 2014 and tisagenlecleucel, the first CAR-T cell therapy in 2017 marked milestones in oncology research. Based on OncoKB (RRID : SCR_014782) (updated June 19, 2024), FDA has approved 186 new targeted therapy drugs since June 1998, out of which 96 are precision oncology drugs. CAR-T therapy is a highly promising treatment for hematological malignancies. As in most cases it works by using patients’ own T-cells (autologous), this therapy is highly precise and effective . Peripheral blood derived T cells are genetically modified to integrate CAR expression cassette into the genome, and CAR proteins are subsequently expressed on surface of T-cells. These modified cells are expanded and infused back into the patients. CAR recognizes specific cancer antigens, forms an immune synapse and lyses the tumor cell by activating granzyme-perforin axis, Fas/Fas ligand pathway and release of cytokines . So far, six CAR-T therapies have been FDA approved for use in clinics, targeting two antigens- CD19 and BCMA . However, owing to the long term adverse effects of CAR-T, such as cytokine release syndrome and neurological toxicity , further research and advancements are moving this field forward, such as integration of CAR with other immune cells - NK/NKT cells, dendritic cells, macrophages, regulatory T cells and B cells , which may have the potential to be safer for long-term use and hold high therapeutic potential for clinical use. Moreover, metabolic dysregulation is a well-known phenomenon in tumors, characterized by accelerated glycolysis, upregulation of lipid and amino acid metabolism, alterations in mitochondrial biogenesis and macromolecule biosynthesis- all of which are considered hallmarks of cancer . Various chemotherapeutics targeting the altered molecules in metabolic machinery are well established for clinical use. Some examples include enasidenib for mutated isocitrate dehydrogenase 2 (IDH2) and Ivosidenib for mutated isocitrate dehydrogenase 1 (IDH1) in acute myeloid leukaemia (AML), 5-fluorouracil inhibiting Thymidylate synthase in gastric and breast cancer, and Methotrexate inhibiting dihydrofolate reductase (DHFR) in breast and lung cancer . However, activation of DNA repair pathway, induced apoptosis resistance, target alterations, and reprogramming of immune cells by limiting nutrient availability within tumor microenvironment can lead to resistance towards these therapies . Understanding the overall picture of the tumor complexity reinforces the concept of combination therapy precisely designed to target the cancer cells from various angles . Traditionally, clinical trials are drug-centered, blinded and randomized to minimize bias. However, due to large variability in patients’ tumor microenvironment, molecular profiles and unique genomic characteristics, the outcomes are far from ideal . Therefore, innovative patient- centered trials are now customized to recognize genomic alterations and employ novel biomarker-guided methodologies to address the distinctive needs of patients . A unifying clinical trial framework known as master protocols includes testing multiple drugs in parallel, for patients with same or different types of cancer . The major trial designs under master protocols are summarized in and are detailed as follows: 2.1 Basket trials These include testing of a new drug in patients with common genetic mutation (pan-cancer gene defect) or biomarkers, in more than one cancer types . Common examples of drugs targeted to specific genes include Pembrolizumab for tumor mutational burden high (TMB-H) and mismatch repair deficiency/microsatellite instability high (dMMR/MSI-High) , and Larotrectinib and entrectinib in tumors with NTRK fusion. Well known basket trials such as NCI‐MATCH (Molecular Analysis for Therapy Choice) and National Cancer Institute’s Molecular Profiling‐Based Assignment of Cancer Therapy (NCI-MPACT) , TAPUR (Targeted Agent and Profiling Utilization Registry) were phase 2 trials based on molecular profiling of different cancer sub-types. In some cases, such trials may not accurately predict the response rates due to heterogeneity of the tumors and appropriate control groups may not be available . 2.2 Umbrella trials Test of multiple therapies in one disease group with common histological aberration, stratified in sub-groups based on different biomarker or genomic subsets. Some examples include The Lung Matrix trial , Myeloid Malignancies Molecular Analysis for Therapy Choice (myeloMATCH) , Adjuvant Lung Cancer Enrichment Marker Identification and Sequencing Trial (ALCHEMIST) , Investigation of Serial Studies to Predict Your Therapeutic Response With Imaging And moLecular Analysis 2 (I-SPY-2) , and The UK plasma based Molecular profiling of Advanced breast cancer to inform Therapeutic Choices (plasmaMATCH) . One challenge with these trials is the requirement of sub-grouping of patients that could slow down the enrollment process in case of rare cancers . 2.3 Platform trials They are also known as multi-arm, multi-stage design trials which include testing multiple drugs against a common disease. Based on the interim analysis, these trials allow changes to the ongoing trial vis à vis addition of a control arm, drug, patient population or even early termination, as needed . This flexibility enables platform trials to be confirmatory. Some examples include ComboMATCH (NCI Combination Therapy Platform Trial with Molecular Analysis for Therapy Choice) , and SHIVA (Study of Randomized, Molecularly Targeted Therapy Based on Tumor Molecular Profiling versus Conventional Therapy for Advanced Cancer) . Since platform trials are large scale and logistically complex, the cost and time duration involved could be high . 2.4 Octopus trials These are completed Phase I/II trials, which evaluate the combinatorial effects of multiple drugs with a common intervention . An example is phase IIb multi-cohort study QUILT-3.055, which tests combinations of N-803 (a fusion protein inducing proliferation and activity of natural killer and cytotoxic T-cells) in patients who received pre-treatment with PD-1/PD-L1 immune checkpoint inhibitors . Since these trials are multi-arm, data generation could be interdependent, leading to potential statistical limitations . 2.5 N-of-1 trial Randomized and blinded trial conducted in a single patient. These are, in a true sense, personalized trials based on specific biologic characteristics . These can be effective in treating rare cancers and to provide objective comparison of different treatments and perform time series analyses . Various N-of-1 trials are comprehensively summarized by Gouda et al. . Some examples include I-PREDICT study , rare pediatric cancer, such as- the case of a 2-year old child with metastatic glomus tumor and activated NOTCH1 , and the ALK-fusion positive high grade glioma in a 3-year old . However, there are serious considerations to performing these trials, ranging from lack of appropriate control and highly conservative treatment selection to data collection and analysis, statistical limitations, false-negatives and the high cost involved in putting together the infrastructure for each trial . Since the patient-centric biomarker-based studies rely on appropriate detection of the relevant disease indicators, several methods are used to analyze and aid in designing the treatment plans. Basket trials These include testing of a new drug in patients with common genetic mutation (pan-cancer gene defect) or biomarkers, in more than one cancer types . Common examples of drugs targeted to specific genes include Pembrolizumab for tumor mutational burden high (TMB-H) and mismatch repair deficiency/microsatellite instability high (dMMR/MSI-High) , and Larotrectinib and entrectinib in tumors with NTRK fusion. Well known basket trials such as NCI‐MATCH (Molecular Analysis for Therapy Choice) and National Cancer Institute’s Molecular Profiling‐Based Assignment of Cancer Therapy (NCI-MPACT) , TAPUR (Targeted Agent and Profiling Utilization Registry) were phase 2 trials based on molecular profiling of different cancer sub-types. In some cases, such trials may not accurately predict the response rates due to heterogeneity of the tumors and appropriate control groups may not be available . Umbrella trials Test of multiple therapies in one disease group with common histological aberration, stratified in sub-groups based on different biomarker or genomic subsets. Some examples include The Lung Matrix trial , Myeloid Malignancies Molecular Analysis for Therapy Choice (myeloMATCH) , Adjuvant Lung Cancer Enrichment Marker Identification and Sequencing Trial (ALCHEMIST) , Investigation of Serial Studies to Predict Your Therapeutic Response With Imaging And moLecular Analysis 2 (I-SPY-2) , and The UK plasma based Molecular profiling of Advanced breast cancer to inform Therapeutic Choices (plasmaMATCH) . One challenge with these trials is the requirement of sub-grouping of patients that could slow down the enrollment process in case of rare cancers . Platform trials They are also known as multi-arm, multi-stage design trials which include testing multiple drugs against a common disease. Based on the interim analysis, these trials allow changes to the ongoing trial vis à vis addition of a control arm, drug, patient population or even early termination, as needed . This flexibility enables platform trials to be confirmatory. Some examples include ComboMATCH (NCI Combination Therapy Platform Trial with Molecular Analysis for Therapy Choice) , and SHIVA (Study of Randomized, Molecularly Targeted Therapy Based on Tumor Molecular Profiling versus Conventional Therapy for Advanced Cancer) . Since platform trials are large scale and logistically complex, the cost and time duration involved could be high . Octopus trials These are completed Phase I/II trials, which evaluate the combinatorial effects of multiple drugs with a common intervention . An example is phase IIb multi-cohort study QUILT-3.055, which tests combinations of N-803 (a fusion protein inducing proliferation and activity of natural killer and cytotoxic T-cells) in patients who received pre-treatment with PD-1/PD-L1 immune checkpoint inhibitors . Since these trials are multi-arm, data generation could be interdependent, leading to potential statistical limitations . N-of-1 trial Randomized and blinded trial conducted in a single patient. These are, in a true sense, personalized trials based on specific biologic characteristics . These can be effective in treating rare cancers and to provide objective comparison of different treatments and perform time series analyses . Various N-of-1 trials are comprehensively summarized by Gouda et al. . Some examples include I-PREDICT study , rare pediatric cancer, such as- the case of a 2-year old child with metastatic glomus tumor and activated NOTCH1 , and the ALK-fusion positive high grade glioma in a 3-year old . However, there are serious considerations to performing these trials, ranging from lack of appropriate control and highly conservative treatment selection to data collection and analysis, statistical limitations, false-negatives and the high cost involved in putting together the infrastructure for each trial . Since the patient-centric biomarker-based studies rely on appropriate detection of the relevant disease indicators, several methods are used to analyze and aid in designing the treatment plans. Onco-precision toolkit Technological advancements in cancer biology have enabled researchers and clinicians to explore options beyond the common drug targets for patients. Even though the DNA sequencing techniques have been in use since 1970s , the most widely accepted next generation sequencing (NGS) was adapted in clinical diagnosis and prognosis within the last decade . With the development of clinical applicability of whole genome sequencing , whole exome sequencing (WES) , RNA-seq (paired with WES or in single-cell/bulk variations , spatial transcriptomics , hybrid capture NGS for targeted oncology panels and comprehensive omics analyses , the integration of large-scale genomic data is now possible to drive the personalized treatment approaches. Besides the genetic mutations at DNA and RNA level, several ncRNAs such as miRNA , circular RNA as well as epigenetic markers are being analyzed to comprehensively map individuals at genetic and molecular level. These techniques draw the cellular landscape of tumors and help in discovering biomarkers associated with clinically relevant genomic alterations, as summarized in . Radiomics or high dimensional medical imaging via PET, CT, Ultrasound and MRI for monitoring the tumor characteristics is combined with machine learning to extract the specific features/characteristics of individual tumors, to guide their specific treatment course . Theranostics (therapy + diagnostics) utilizes radionuclide linked to targeted biomarkers, which allows diagnosis through targeted imaging (radiomics) and targeted therapy at the same time . Some examples include Lutathera (lutetium 177 Lu dotatate), the first FDA and EMA approved theranostic drug, which releases radiation to kill cancer cells by binding to cell surface receptors somatostatin on gastroenteropancreatic neuroendocrine tumors , and Pluvicto for castration-resistance prostate cancer using Lutetium-177 that targets PSMA on cell surface in prostate cancer . Various ML methods, such as deep neural networks are also used to predict clinical outcomes using the supervised and unsupervised learning models to enhance the efficiency of cancer diagnosis and increase the probability for predictive prognosis. With the use of de-identified electronic health records (EHRs) , paired with specific genotypic training data and bioinformatic regression models , the auto-encoders can extract intrinsic features of the tumors . This high throughput real-world data (RWD) paves the way to deeper understanding of complex biomarkers associated with heterogenous tumor sub-populations, microsatellite instability and tumor mutational burden (TMB) . summarizes the tools available for personalized cancer treatment for specific population groups, to achieve the clinical goals of treatment and prolonged survival. Various governmental organizations such as NCI, NCBI and FDA provide open access public repositories to study the pharmacogenomic pattern of larger population groups with drug response, increased clinical efficacy probability and reduced adverse drug reactions. Some examples include TCGA (reports molecular characterization of 20,000 primary tumors) (RRID : SCR_003193), ClinVar (public archive of human diseases and corresponding drug responses) (RRID : SCR_006169), COSMIC (catalogue of somatic mutations in cancer) (RRID : SCR_002260), PharmKGB (Pharmacogenomics Knowledgebase for genotype-phenotype relationship, genetic variants and drug associated guidelines) (RRID : SCR_025580), Drugbank (for comprehensive drug-target data) (RRID : SCR_002700), FDA’s pharmacogenetic associations and ClinGen (human genetic variants database) (RRID : SCR_014968) . Precision FDA is also a free computing platform to analyze large biological datasets and learn from experts in the field. Utilization of these vast array of tools available carries its fair share of challenges ranging from the cost and time involved in generating the large datasets to managing, storing, aligning and assessing this data, with high quality, accuracy and reproducibility. Aligning multi-reads, incorrect sequence mapping, absence of reference sequences, computational challenges spanning splice or fusion junctions, misalignment and false-positive identification are a few common problems noted with NGS and RNA-seq methods . At experimental level, the quality of RNA and DNA extracted from formalin fixed embedded (FFPE) tissues derived from tissue banks may not be the best in some cases for high throughput analyses . Moreover, clonal diversity and tumor heterogeneity is a major challenge in a constantly evolving tumor microenvironment, which can interfere with accurate detection of driver mutations and novel factors leading to resistance towards therapy. Common examples of acquired resistance include splice variants affecting ATP-competitive tyrosine kinase inhibitor binding sites, activating or sub-clonal mutations in PI3K, RAS/MAPK pathways, mutations in “undruggable” genes such as Myc, KRAS and Tp53, FLT3 mutant leukemia , and somatic mutations in cancer stem cells . Realistically, biopsy at recurrence or relapse is not always possible in case of severe metastasis associated with procedure invasiveness and underlying co-morbidity . However, using resources like NCI-MATCH and pairing them with sequential screening tests of samples derived from liquid biopsies, and circulating tumor (ct)-DNA based targeted sequencing based on specific genetic panels can be a way to detect actionable genomic alterations and predict the resistance to adapt customized approaches. Cancer vaccines: a long journey with promising outcome Cancer vaccines can be categorized into (a) Preventative, such as hepatitis B vaccine and human papilloma virus (HPV) vaccine, administered to reduce the risk of liver cancer and cervical cancer, respectively , or (b) Therapeutic, such as Sipuleucel-T against metastatic prostate cancer , Nadofaragene firadonevec (Adstiladrin) for early-stage bladder cancer , and T-VEC (Imlygic) to treat advanced melanoma . Although the first cancer vaccine trial dates back to 1890s, when William B. Coley used heat-killed streptococcal injections in patients with inoperable sarcomas , a major leap forward was in 1959 when Llyod Old showed that BCG infection in mice increased their resistance towards transplanted murine tumor cell lines S-180, carcinoma 755 and Ehlrich ascites . The BCG vaccine containing live attenuated Mycobacterium bovis was later approved by FDA for early-stage bladder cancer . As preventative cancer vaccines have limited applicability and efficacy against the plethora of cancer-causing agents, therapeutic vaccines are emerging as an effective means to activate the immune response by enhanced tumor antigen presentation and generating non-exhaustive cytotoxic T cells to improve anti-tumor immunity . These vaccines elicit the immune response by recognizing the specific epitopes expressed by tumor cells . Though there has been limited success with such vaccines so far, various clinical trials (clinicaltrials.gov) are now focusing on targeting tumor specific antigens (TSAs), which are exclusive to tumors and possess high immunogenicity . TSAs can be viral antigens or non-viral neoantigens generated by spontaneous somatic mutations in tumor microenvironment . As many neoantigens are unique to either a small sub-population or specific to an individual patient, personalized cancer vaccines are gaining attention for precision targeting. The most important factors to consider while designing a tailor-made cancer vaccine are: (a) Accurate identification of highly potent and immunogenic neoantigens capable of inducing a robust T cell immunity; (b) Calculative estimation of the probability of TSA-epitopes binding to MHC; (c) Neoantigen prioritization to predict the interaction of TCRs with MHC-neoepitope complex; (d) Selecting appropriate delivery platform for neoantigen based vaccine, which may include autologous dendritic cells (DCs), peptides, DNA, RNA, mRNA or viral vectors . Autologous whole tumor cells mixed with immunomodulatory adjuvants, genetically modified autologous tumor cells, autologous cell derived exosomes, DC-tumor cell fusion vaccine, autologous DC-based or DNA/RNA/mRNA-based vaccines are a few examples of the ones undergoing clinical trials for personalized treatment . Recombinant viral vectors, such as Great Ape derived adenoviruses (GAd) and modified vaccinia virus Ankara (MVA) also serve as a great tool to trigger effective cytotoxic T cell response, using their intrinsic adjuvant properties . Anti-viral vector immunity can serve as a roadblock though, which can lead to ineffective immunity boost at re-administration. This challenge is being eradicated via a ‘heterologous-boost approach’ in various clinical studies, where involving different platforms can provide stronger immune response, examples include GAd - primed with MVA boost or with self-amplifying RNA . We are still in initial stages of personalized cancer vaccine development owing to the complexity and masking skills honed by tumor cells, which makes it difficult to recognize the distinguishing epitopes. A weakened immune system with immunosuppressive proteins expressed on tumor cells (such as PD-L1), loss of TSA expression or spontaneous alterations in antigen processing pathways, could be a few potential challenges . Personalized vaccine manufacturing also involves a large cost, unique supply chain and the extensive process involved could cause a lag in timely treatment . As scientists are progressing this field forward, there is a need to further refine the cancer vaccine formulation and preparation workflows and make it more accessible to the wider group of patients in need. Discussion Diligent preclinical steps towards choosing the correct research platform (such as humanized mouse models , organoids or organ-on-chip , appropriate drugs , and carefully curated experimentation strategies, serve as the foundation of any clinical trial. Undeniably, the molecular framework of tumors needs to be thoroughly studied at a deeper level to align with the required treatment regimens. Understanding resistance mechanisms and adopting alternative approaches is important from the early research steps . Combining these aspects with comprehensive AI-assisted technologies, such as NGS and multi-omics connects the pathway from preclinical to ‘personalized’ clinical stages . Since generative and multimodal AI models play a major role in patient diagnosis, trial design, planning, patient recruitment, drug delivery, digital monitoring, and data assessment, it is imperative to adopt a precautionary regulatory framework . European AI act and FDA have released regulatory policies for the use of AI in medical field , however the rules need to be clearer and up to date as the field progresses. Keeping in mind the biases and limitations of large datasets generated from AI-based systems, the risk-benefit scale needs to be fine-tuned. Using real world evidence (RWE) also poses privacy and data confidentiality risks , which should be appropriately addressed. Furthermore, using high throughput screening methods for certain subpopulations would need comprehensive training models, absence of which may introduce bias or sub-par results . Thorough investigation of medical interventions is needed to be cautious of any false claims from personalized drug developers. Transparent evidence-based information sharing and finding accelerated solutions to unexpected contradictions is required to manage the fragmented regulation in clinical settings . Going hand in hand with the ethical considerations, the need for precision medicine outweighs any opposing schools of thought, recognizing that each life is important. Significant progress has been seen in this field, with the launch of ‘Precision Medicine Initiative’ by Barack Obama in 2015 and the Cancer Moonshot program , aiming to bring the public and private sectors together to provide a broader screening, diagnostic, therapeutic and supportive biomedical platform. Though many organizations are focused on developing cutting-edge technologies tailor-made to patients’ needs, we are still many steps away from accessible and affordable personalized healthcare for everyone in need. However, with precision oncology propelling cancer research, there is a gleam of hope for a healthier not-too-distant future. |
Sodium-glucose co-transporter 2 inhibitors in left ventricular assist device and heart transplant recipients: a mini-review | b66a36f0-8621-4db8-ba7a-ded6d8fd2f71 | 11802697 | Surgical Procedures, Operative[mh] | Sodium-glucose co-transporter 2 inhibitors (SGLT2i) have emerged as promising agents in managing heart failure (HF). Landmark trials such as DAPA-HF and EMPEROR-Reduced demonstrated their role in reducing HF hospitalizations and cardiovascular mortality among patients with reduced left ventricular ejection fraction (LVEF), irrespective of diabetes status . The EMPEROR-Preserved and DELIVER trials have been pivotal in further expanding the indications for SGLT2-i across the whole spectrum of LVEF . The promising cardioprotective properties of SGLT2i appear to extend beyond glycemic effects and include blood pressure reduction, improvement in myocardial energetics, and modulation of inflammatory pathways . This broad therapeutic profile suggests potential benefits in patients who were not included in the mentioned HF trials, such as in HF patients managed with left ventricular assist devices (LVADs) and heart transplantation (HT). Patients with LVADs represent a unique population of HF patients. Despite technological advances and improved perioperative care, they remain at high risk for adverse cardiovascular events, such as right heart failure and arrhythmias . Similarly, HT recipients face a lifelong risk of allograft dysfunction and metabolic complications that can compromise graft and patient survival . Emerging data suggest that SGLT2i may offer clinical advantages in these populations by mitigating residual risks and improving cardiovascular outcomes . However, LVAD and HT patients a susceptible for infections due to immunosuppression and device-related factors, which may represent an ongoing challenge under treatment with SGLT2i . The current review aims to provide an analysis of the safety and efficacy of SGLT2i in LVAD and HT patients. Examining the latest research findings, we aim to elucidate the role of SGLT2i in these vulnerable patient populations. We conducted a literature review of primary studies examining adult HT recipients and patients with LVAD treated with SGLT2i as a monotherapy or in combination with other antihyperglycemic agents. We searched for articles published from May 5, 2020, the date of FDA approval of dapagliflozin for heart failure, until May 25, 2024. The databases included PubMed, Google Scholar, Ovid Embase, Ovid MEDLINE, and The Cochrane Library (Wiley). We used the following keywords in various combinations: “left ventricular assist device”, “ventricular assist device”, “heart transplantation”, “sodium-glucose co-transporter 2 inhibitor”, “sodium glucose co-transporter 2 inhibitor”, “SGLT-2 inhibitor”, and “SGLT2 inhibitor”. Due to the limited available evidence, all published observational studies were included. Case reports and conference abstracts were excluded. No interventional studies on this topic were identified. The primary clinical outcomes included the safety profile and efficacy of SGLT2i. The main focus was on their impact on body mass index (BMI), hemoglobin A1c (HbA1c), and furosemide dosage. Additionally, adverse effects and tolerability were assessed. In LVAD recipients, the studies also investigated the resulting changes in hemodynamic parameters and device settings. SGLT2 and patients with LVAD Baseline characteristics We identified four key observational studies investigating the impact of SGLT2i in LVAD patients . Three were single-center studies ( n = 560 patients), and one was conducted in a multicenter setting ( n = 138). All four articles were published between 2022 and 2024. Overall, the proportion of patients on SGLT2i was low ( n = 99, 14.2%). Among these, Empagliflozin was the most commonly utilized SGLT2i ( n = 59, 59.6%), followed by Dapagliflozin ( n = 22, 22.2%) and Canagliflozin ( n = 2, 2.0%). One study did not specify which SGLT2i was used. The type of LVAD system was specified in three of the four studies, with HeartMate3 (Abbott Inc., Chicago, IL, USA) being the most commonly implanted device ( n = 67, 67.7%). Most patients were treated with a bridge-to-transplant approach. The median reported follow-up ranged from 30 to 354 days. Most of the patients on SGLT2i were male ( n = 82, 82.8%). Ischemic cardiomyopathy was the most common etiology of advanced HF ( n = 54, 54.5%). The median body mass index (BMI) was between 26.8 and 33.6 kg/m 2 in all three studies where it was reported. Approximately 80% of the patients on SGLT2i had diabetes mellitus ( n = 79). The median baseline left ventricular ejection fraction (LVEF) ranged from 18.0 to 25.0%. According to the KDIGO classification of renal function, patients had renal insufficiency classified as G2–G3a in all three studies reporting the estimated glomerular filtration rate (eGFR) (Table ). Outcomes The studies on the use of SGLT2i in LVAD patients were heterogeneous and assessed different outcomes (Table ). Fluid Management and Hemodynamics Fardman et al. reported improvements in fluid management and pulmonary pressures in patients treated with SGLT2i, with a significant reduction in invasively derived systolic pulmonary artery pressure (PAP). Given that the maximum effect of left ventricular (LV) unloading on PAP reduction is usually achieved between 1 and 3 months post-surgery and the fact that SGLT2i were initiated at a median of 108 days (IQR 26–477) after LVAD implantation in this study, the authors hypothesized that the maximum effect of mechanical LV unloading on PAP reduction had already been reached . Hence, they postulated that the observed further reduction in PAP could be attributed to the effects of SGLT2i. The beneficial effects of SGLT2i on PAP have already been investigated by previous studies, including the EMBRACE-HF trial . The proposed mechanisms for these effects include natriuretic synergism with loop diuretics. Accordingly, the authors reported a significant reduction in the diuretic dose in patients treated with SGLT2i. Moady et al. also observed a transient reduction in diuretic dose in patients treated with SGLT2i. They noted an increased number of suction events potentially attributable to enhanced diuresis during the first week of treatment . SGLT2i block sodium and glucose reabsorption in the proximal tubule, but the duration and nature of their natriuretic and diuretic effects depend on the upregulation of SGLT2 and sodium-hydrogen exchanger 3 (NHE3), along with compensatory mechanisms in downstream nephrons. In euvolemic patients, counterregulatory sodium and water retention mechanisms activate more rapidly than in fluid-overloaded patients, limiting the duration of diuresis . LVAD patients often have impaired renal function, and the glycosuria-related effects of SGLT2 inhibitors are reduced in patients with a glomerular filtration rate (GFR) below 45 mL/min. However, patients with chronic kidney disease still benefit from SGLT2 inhibitors . While this evidence is encouraging, Cagliostro et al. and Chavali et al. did not find significant changes in diuretic dose with SGLT2i . Additionally, Chavali et al. did not observe any differences in the hemodynamic parameters after the initiation of SGLT2i . Echocardiographic parameters Despite the overall improvement in systolic PAP, Fardman et al. found an increase in the rate of right ventricular (RV) dysfunction. Severe RV dysfunction was observed in n = 3 (14%) of patients at follow-up, whereas none had severe RV dysfunction at baseline.This rate is comparable to the 8–11% reported in the literature . Additionally, n = 2 (9.5%) of the patients had dilated RV at baseline, compared to n = 7 (33.3%) at follow-up. The authors report that the assessment of the RV function was performed visually by experienced sonographer but provide no further details regarding the modality of the analysis. Despite the observed deterioration of the RV function, there was no difference in the invasively derived right atrial pressure (RAP) 6 months after the initiation of SGLT2i. Additionally, none of the patients with severe RV dysfunction required hospitalization for HF during the study period . The discrepancy between echocardiographic findings and clinical outcomes warrants further investigation. Renal outcomes Moady et al. observed a transient worsening of renal function after initiation of SGLT2i, as evidenced by a drop in eGFR of ≥ 10% from baseline. This decline was temporary and stabilized after 2–3 months of therapy . The initial reduction in eGFR is attributed to the modest reduction in blood pressure seen with SGLT2i, but it does not indicate poor long-term renal outcomes . Additionally, all other studies reported no significant changes in renal function throughout their respective observational periods. Glycemic Control and Weight This is a critical outcome for LVAD patients with diabetes mellitus. Improved glucose management can help reduce the risk of complications like infections and enhance overall health . Cagliostro M et al. did not observe significant changes in BMI or HbA1c. In contrast, Fardman A et al. reported a significant weight reduction, while Moady G et al. noted a significant decrease in HbA1c following the initiation of SGLT2i. Chavali S et al. found no substantial changes in body weight . In a meta-analysis on diabetic patients, SGLT2i reduced HbA1c by 0.62% (95% CI -0.66 to -0.59) and body weight by 0.60 kg (95% CI -0.64 to -0.55). Moreover, the EMPA-REG trial demonstrated that the lowest dose of empagliflozin (10 mg) offers comparable benefits in lowering HbA1c, body weight, blood pressure, and reducing total and cardiovascular mortality as the highest dose (25 mg) . While SGLT2i have shown notable improvements in HbA1c and weight in the general population, studies on their metabolic effects in LVAD patients have yielded inconsistent results. Adverse events Across all studies, the incidence of adverse events was generally low. Regarding safety and tolerability, the study by Cagliostro M et al. identified several adverse events potentially attributable to SGLT2i, including genitourinary infections, acute kidney injury, limb amputations, and driveline infections (DLI). Similarly, Fardman A et al. observed increased RV dysfunction in some patients but found no significant overall adverse effects linked to SGLT2i. In contrast, Chavali S et al. reported that the SGLT2i therapy was safe and well-tolerated. Moady G et al. also found that SGLT2i were well-tolerated with only minor adverse effects . Regarding driveline infections and SGLT2i use, we acknowledge that current data from non-immunosuppressed populations do not show a significant increase in non-genital skin and soft tissue infections with SGLT2i therapy. A post-hoc analysis of the CANVAS and CREDENCE trials found no difference in non-genital infection rates between canagliflozin and placebo (HR 0.97, 95% CI 0.85–1.11; P = 0.70), and no significant increase in non-genital fungal infections either . Thus, we do not have suffitient evidence regarding the risk of skin infections or potential driveline infections in LVAD patients. Additionally, due to the lack of evidence in immunosuppressed patients means we cannot definitively rule out the possibility of a higher risk in HT recipients. This is an important area for future research. Meanwhile, the glycemic improvement associated with SGLT2i is expected to reduce infection risk by improving overall metabolic control, which should be considered in assessing their benefit-risk profile . Clinical implications and future directions The evidence suggests that SGLT2i could play a vital role in managing LVAD patients, particularly those with concurrent diabetes. The current studies indicate potential benefits in glycemic control, fluid management, and hemodynamic stability. Across all studies, SGLT2i therapy was generally well-tolerated, with no significant increase in adverse events such as infections or ketoacidosis. This is particularly noteworthy given the complex and high-risk nature of LVAD patients, who often require multiple medications and have significant comorbidities. SGLT2 in HT recipients The safety and efficacy of SGLT2i in HT recipients have been explored in two pivotal studies . Both reports provide valuable insights into the efficacy and safety of SGLT2i in this specific patient population, although they use different methodologies (Table ). Baseline characteristics Both studies were retrospective observational studies. All patients on SGLT2i ( n = 39, 32%) were treated with empagliflozin. Most of these patients were male ( n = 31, 86.1%), and the median follow-up period was 9.1 months in the study by Cehic et al. and 12 months in the study by Sammour et al. The median age was 59 years in both studies. The etiology of advanced HF prior to HT was not consistently reported. However, about a third of the patients in the study by Sammour et al. had a history of ischemic cardiomyopathy ( n = 7, 33.3%). All patients had diabetes. In both studies, patients had renal insufficiency classified as G3a according to the KDIGO classification of renal function (Table ). Outcomes Both studies included in this review examined similar outcomes, specifically focusing on the safety and efficacy of the SGLT2i (Table ). Fluid Management and Hemodynamics The treatment with empagliflozin led to a significant reduction in the median furosemide dose in the study by Cehic et al., with no significant change in systolic or diastolic blood pressure . The study by Sammour et al. did not specifically report on fluid management or hemodynamic parameters . However, the significant reductions in insulin requirements and body weight across the study population might indirectly influence fluid status. Insulin is known to stimulate sodium retention. Studies in diabetic models demonstrate that maintaining baseline insulin levels, even in the presence of hyperglycemia, can counteract natriuresis; thus, chronic hyperinsulinemia may lead to sustained sodium retention, particularly in the context of insulin resistance and metabolic disorders . Echocardiographic parameters The studies provided no data regarding echocardiographic parameters or their changes in the assessed HT populations. Renal outcomes In both studies, renal function remained stable in patients treated with empagliflozin compared to those on other glucose-lowering medication, highlighting the safety of empagliflozin with respect to renal function . However, a more extended observation period is needed to determine whether SGLT2i also offer renal protection for patients without diabetes mellitus. Recent evidence from the EMPA-KIDNEY trial demonstrates that empagliflozin promotes renal protection in patients with chronic kidney disease (CKD), irrespective of diabetes status. In a large population of patients with eGFR ranging from 20 to 45 ml/min/1.73 m² empagliflozin significantly reduced the risk of progression of kidney disease or death from cardiovascular causes compared to placebo, with consistent results across both diabetic and non-diabetic subgroups. Given these findings, further research is warranted to fully elucidate the long-term renal benefits of SGLT2i in HT patients without diabetes . Glycemic Control and Weight Overall, treatment with SGLT2i resulted in significant weight reduction and improved glycemic control. In the study by Cehic et al., empagliflozin treatment led to a decrease in body weight (median reduction of − 2.0 kg) and a mean reduction in HbA1c of 2.8% (6.6 mmol/mol) . Similarly, Sammour et al. found that combining GLP-1 receptor agonists (GLP1-RAs) and SGLT2i significantly decreased body weight and HbA1c (from 7.5 to 6.6%) . Additionally, there was a significant reduction in insulin requirements. Adverse effects Empagliflozin was well tolerated in the study by Cehic et al., with only three adverse effects reported: exacerbation of urinary symptoms, dizziness, and acute kidney injury. Notably, no genitourinary infections were observed in the empagliflozin group . Similarly, the study by Sammour et al. reported no incidents of diabetic ketoacidosis, pancreatitis, or genital mycotic infections among patients using GLP-1RAs and SGLT2i . Additionally, there were no noted drug interactions with immunosuppressive therapies. Clinical implications and future directions Both studies underscore the safety and efficacy of SGLT2i, alone or in combination with other antiglycemic agents for managing diabetes in HT recipients. Cehic et al. report that empagliflozin therapy is associated with potential benefits in improving body weight and glycemic control, while highlighting its safety in this population . Similarly, Sammour et al. demonstrate that combining GLP-1 RAs and SGLT2i leads to significant improvements in insulin requirements, HbA1c, and body weight, suggesting the effectiveness of these therapies in managing HT recipients with diabetes . However, there is a substantial gap in research concerning the role of SGLT2i in the vulnerable population of HT recipients. To address this, randomized controlled trials are needed to further validate the safety and efficacy of SGLT2i, both alone and in combination with other antiglycemic agents. Long-term studies are needed to evaluate their influence on renal function and metabolic complications following HT. Additionally, research should assess the impact of these therapies on transplant-related mortality, graft function, rejection episodes, coronary allograft vasculopathy, and the progression of chronic kidney disease. Currently, a phase 3 randomized placebo-controlled trial is ongoing, which evaluates the protective effects of dapagliflozin on renal function in patients after HT (DAPAHRT trial; ClinicalTrials.gov Identifier NCT05321706; estimated study completion in 2027). Determining the optimal timing for initiating treatment is fundamental to optimize the therapy and mitigate potential risks. In addition, it is crucial to explore the use of SGLT2i in HT recipients without diabetes. These data could reveal potential benefits or risks, contributing to more informed therapeutic decisions and improving overall patient management in the post-transplant setting. Baseline characteristics We identified four key observational studies investigating the impact of SGLT2i in LVAD patients . Three were single-center studies ( n = 560 patients), and one was conducted in a multicenter setting ( n = 138). All four articles were published between 2022 and 2024. Overall, the proportion of patients on SGLT2i was low ( n = 99, 14.2%). Among these, Empagliflozin was the most commonly utilized SGLT2i ( n = 59, 59.6%), followed by Dapagliflozin ( n = 22, 22.2%) and Canagliflozin ( n = 2, 2.0%). One study did not specify which SGLT2i was used. The type of LVAD system was specified in three of the four studies, with HeartMate3 (Abbott Inc., Chicago, IL, USA) being the most commonly implanted device ( n = 67, 67.7%). Most patients were treated with a bridge-to-transplant approach. The median reported follow-up ranged from 30 to 354 days. Most of the patients on SGLT2i were male ( n = 82, 82.8%). Ischemic cardiomyopathy was the most common etiology of advanced HF ( n = 54, 54.5%). The median body mass index (BMI) was between 26.8 and 33.6 kg/m 2 in all three studies where it was reported. Approximately 80% of the patients on SGLT2i had diabetes mellitus ( n = 79). The median baseline left ventricular ejection fraction (LVEF) ranged from 18.0 to 25.0%. According to the KDIGO classification of renal function, patients had renal insufficiency classified as G2–G3a in all three studies reporting the estimated glomerular filtration rate (eGFR) (Table ). Outcomes The studies on the use of SGLT2i in LVAD patients were heterogeneous and assessed different outcomes (Table ). Fluid Management and Hemodynamics Fardman et al. reported improvements in fluid management and pulmonary pressures in patients treated with SGLT2i, with a significant reduction in invasively derived systolic pulmonary artery pressure (PAP). Given that the maximum effect of left ventricular (LV) unloading on PAP reduction is usually achieved between 1 and 3 months post-surgery and the fact that SGLT2i were initiated at a median of 108 days (IQR 26–477) after LVAD implantation in this study, the authors hypothesized that the maximum effect of mechanical LV unloading on PAP reduction had already been reached . Hence, they postulated that the observed further reduction in PAP could be attributed to the effects of SGLT2i. The beneficial effects of SGLT2i on PAP have already been investigated by previous studies, including the EMBRACE-HF trial . The proposed mechanisms for these effects include natriuretic synergism with loop diuretics. Accordingly, the authors reported a significant reduction in the diuretic dose in patients treated with SGLT2i. Moady et al. also observed a transient reduction in diuretic dose in patients treated with SGLT2i. They noted an increased number of suction events potentially attributable to enhanced diuresis during the first week of treatment . SGLT2i block sodium and glucose reabsorption in the proximal tubule, but the duration and nature of their natriuretic and diuretic effects depend on the upregulation of SGLT2 and sodium-hydrogen exchanger 3 (NHE3), along with compensatory mechanisms in downstream nephrons. In euvolemic patients, counterregulatory sodium and water retention mechanisms activate more rapidly than in fluid-overloaded patients, limiting the duration of diuresis . LVAD patients often have impaired renal function, and the glycosuria-related effects of SGLT2 inhibitors are reduced in patients with a glomerular filtration rate (GFR) below 45 mL/min. However, patients with chronic kidney disease still benefit from SGLT2 inhibitors . While this evidence is encouraging, Cagliostro et al. and Chavali et al. did not find significant changes in diuretic dose with SGLT2i . Additionally, Chavali et al. did not observe any differences in the hemodynamic parameters after the initiation of SGLT2i . Echocardiographic parameters Despite the overall improvement in systolic PAP, Fardman et al. found an increase in the rate of right ventricular (RV) dysfunction. Severe RV dysfunction was observed in n = 3 (14%) of patients at follow-up, whereas none had severe RV dysfunction at baseline.This rate is comparable to the 8–11% reported in the literature . Additionally, n = 2 (9.5%) of the patients had dilated RV at baseline, compared to n = 7 (33.3%) at follow-up. The authors report that the assessment of the RV function was performed visually by experienced sonographer but provide no further details regarding the modality of the analysis. Despite the observed deterioration of the RV function, there was no difference in the invasively derived right atrial pressure (RAP) 6 months after the initiation of SGLT2i. Additionally, none of the patients with severe RV dysfunction required hospitalization for HF during the study period . The discrepancy between echocardiographic findings and clinical outcomes warrants further investigation. Renal outcomes Moady et al. observed a transient worsening of renal function after initiation of SGLT2i, as evidenced by a drop in eGFR of ≥ 10% from baseline. This decline was temporary and stabilized after 2–3 months of therapy . The initial reduction in eGFR is attributed to the modest reduction in blood pressure seen with SGLT2i, but it does not indicate poor long-term renal outcomes . Additionally, all other studies reported no significant changes in renal function throughout their respective observational periods. Glycemic Control and Weight This is a critical outcome for LVAD patients with diabetes mellitus. Improved glucose management can help reduce the risk of complications like infections and enhance overall health . Cagliostro M et al. did not observe significant changes in BMI or HbA1c. In contrast, Fardman A et al. reported a significant weight reduction, while Moady G et al. noted a significant decrease in HbA1c following the initiation of SGLT2i. Chavali S et al. found no substantial changes in body weight . In a meta-analysis on diabetic patients, SGLT2i reduced HbA1c by 0.62% (95% CI -0.66 to -0.59) and body weight by 0.60 kg (95% CI -0.64 to -0.55). Moreover, the EMPA-REG trial demonstrated that the lowest dose of empagliflozin (10 mg) offers comparable benefits in lowering HbA1c, body weight, blood pressure, and reducing total and cardiovascular mortality as the highest dose (25 mg) . While SGLT2i have shown notable improvements in HbA1c and weight in the general population, studies on their metabolic effects in LVAD patients have yielded inconsistent results. Adverse events Across all studies, the incidence of adverse events was generally low. Regarding safety and tolerability, the study by Cagliostro M et al. identified several adverse events potentially attributable to SGLT2i, including genitourinary infections, acute kidney injury, limb amputations, and driveline infections (DLI). Similarly, Fardman A et al. observed increased RV dysfunction in some patients but found no significant overall adverse effects linked to SGLT2i. In contrast, Chavali S et al. reported that the SGLT2i therapy was safe and well-tolerated. Moady G et al. also found that SGLT2i were well-tolerated with only minor adverse effects . Regarding driveline infections and SGLT2i use, we acknowledge that current data from non-immunosuppressed populations do not show a significant increase in non-genital skin and soft tissue infections with SGLT2i therapy. A post-hoc analysis of the CANVAS and CREDENCE trials found no difference in non-genital infection rates between canagliflozin and placebo (HR 0.97, 95% CI 0.85–1.11; P = 0.70), and no significant increase in non-genital fungal infections either . Thus, we do not have suffitient evidence regarding the risk of skin infections or potential driveline infections in LVAD patients. Additionally, due to the lack of evidence in immunosuppressed patients means we cannot definitively rule out the possibility of a higher risk in HT recipients. This is an important area for future research. Meanwhile, the glycemic improvement associated with SGLT2i is expected to reduce infection risk by improving overall metabolic control, which should be considered in assessing their benefit-risk profile . Clinical implications and future directions The evidence suggests that SGLT2i could play a vital role in managing LVAD patients, particularly those with concurrent diabetes. The current studies indicate potential benefits in glycemic control, fluid management, and hemodynamic stability. Across all studies, SGLT2i therapy was generally well-tolerated, with no significant increase in adverse events such as infections or ketoacidosis. This is particularly noteworthy given the complex and high-risk nature of LVAD patients, who often require multiple medications and have significant comorbidities. We identified four key observational studies investigating the impact of SGLT2i in LVAD patients . Three were single-center studies ( n = 560 patients), and one was conducted in a multicenter setting ( n = 138). All four articles were published between 2022 and 2024. Overall, the proportion of patients on SGLT2i was low ( n = 99, 14.2%). Among these, Empagliflozin was the most commonly utilized SGLT2i ( n = 59, 59.6%), followed by Dapagliflozin ( n = 22, 22.2%) and Canagliflozin ( n = 2, 2.0%). One study did not specify which SGLT2i was used. The type of LVAD system was specified in three of the four studies, with HeartMate3 (Abbott Inc., Chicago, IL, USA) being the most commonly implanted device ( n = 67, 67.7%). Most patients were treated with a bridge-to-transplant approach. The median reported follow-up ranged from 30 to 354 days. Most of the patients on SGLT2i were male ( n = 82, 82.8%). Ischemic cardiomyopathy was the most common etiology of advanced HF ( n = 54, 54.5%). The median body mass index (BMI) was between 26.8 and 33.6 kg/m 2 in all three studies where it was reported. Approximately 80% of the patients on SGLT2i had diabetes mellitus ( n = 79). The median baseline left ventricular ejection fraction (LVEF) ranged from 18.0 to 25.0%. According to the KDIGO classification of renal function, patients had renal insufficiency classified as G2–G3a in all three studies reporting the estimated glomerular filtration rate (eGFR) (Table ). The studies on the use of SGLT2i in LVAD patients were heterogeneous and assessed different outcomes (Table ). Fluid Management and Hemodynamics Fardman et al. reported improvements in fluid management and pulmonary pressures in patients treated with SGLT2i, with a significant reduction in invasively derived systolic pulmonary artery pressure (PAP). Given that the maximum effect of left ventricular (LV) unloading on PAP reduction is usually achieved between 1 and 3 months post-surgery and the fact that SGLT2i were initiated at a median of 108 days (IQR 26–477) after LVAD implantation in this study, the authors hypothesized that the maximum effect of mechanical LV unloading on PAP reduction had already been reached . Hence, they postulated that the observed further reduction in PAP could be attributed to the effects of SGLT2i. The beneficial effects of SGLT2i on PAP have already been investigated by previous studies, including the EMBRACE-HF trial . The proposed mechanisms for these effects include natriuretic synergism with loop diuretics. Accordingly, the authors reported a significant reduction in the diuretic dose in patients treated with SGLT2i. Moady et al. also observed a transient reduction in diuretic dose in patients treated with SGLT2i. They noted an increased number of suction events potentially attributable to enhanced diuresis during the first week of treatment . SGLT2i block sodium and glucose reabsorption in the proximal tubule, but the duration and nature of their natriuretic and diuretic effects depend on the upregulation of SGLT2 and sodium-hydrogen exchanger 3 (NHE3), along with compensatory mechanisms in downstream nephrons. In euvolemic patients, counterregulatory sodium and water retention mechanisms activate more rapidly than in fluid-overloaded patients, limiting the duration of diuresis . LVAD patients often have impaired renal function, and the glycosuria-related effects of SGLT2 inhibitors are reduced in patients with a glomerular filtration rate (GFR) below 45 mL/min. However, patients with chronic kidney disease still benefit from SGLT2 inhibitors . While this evidence is encouraging, Cagliostro et al. and Chavali et al. did not find significant changes in diuretic dose with SGLT2i . Additionally, Chavali et al. did not observe any differences in the hemodynamic parameters after the initiation of SGLT2i . Echocardiographic parameters Despite the overall improvement in systolic PAP, Fardman et al. found an increase in the rate of right ventricular (RV) dysfunction. Severe RV dysfunction was observed in n = 3 (14%) of patients at follow-up, whereas none had severe RV dysfunction at baseline.This rate is comparable to the 8–11% reported in the literature . Additionally, n = 2 (9.5%) of the patients had dilated RV at baseline, compared to n = 7 (33.3%) at follow-up. The authors report that the assessment of the RV function was performed visually by experienced sonographer but provide no further details regarding the modality of the analysis. Despite the observed deterioration of the RV function, there was no difference in the invasively derived right atrial pressure (RAP) 6 months after the initiation of SGLT2i. Additionally, none of the patients with severe RV dysfunction required hospitalization for HF during the study period . The discrepancy between echocardiographic findings and clinical outcomes warrants further investigation. Renal outcomes Moady et al. observed a transient worsening of renal function after initiation of SGLT2i, as evidenced by a drop in eGFR of ≥ 10% from baseline. This decline was temporary and stabilized after 2–3 months of therapy . The initial reduction in eGFR is attributed to the modest reduction in blood pressure seen with SGLT2i, but it does not indicate poor long-term renal outcomes . Additionally, all other studies reported no significant changes in renal function throughout their respective observational periods. Glycemic Control and Weight This is a critical outcome for LVAD patients with diabetes mellitus. Improved glucose management can help reduce the risk of complications like infections and enhance overall health . Cagliostro M et al. did not observe significant changes in BMI or HbA1c. In contrast, Fardman A et al. reported a significant weight reduction, while Moady G et al. noted a significant decrease in HbA1c following the initiation of SGLT2i. Chavali S et al. found no substantial changes in body weight . In a meta-analysis on diabetic patients, SGLT2i reduced HbA1c by 0.62% (95% CI -0.66 to -0.59) and body weight by 0.60 kg (95% CI -0.64 to -0.55). Moreover, the EMPA-REG trial demonstrated that the lowest dose of empagliflozin (10 mg) offers comparable benefits in lowering HbA1c, body weight, blood pressure, and reducing total and cardiovascular mortality as the highest dose (25 mg) . While SGLT2i have shown notable improvements in HbA1c and weight in the general population, studies on their metabolic effects in LVAD patients have yielded inconsistent results. Adverse events Across all studies, the incidence of adverse events was generally low. Regarding safety and tolerability, the study by Cagliostro M et al. identified several adverse events potentially attributable to SGLT2i, including genitourinary infections, acute kidney injury, limb amputations, and driveline infections (DLI). Similarly, Fardman A et al. observed increased RV dysfunction in some patients but found no significant overall adverse effects linked to SGLT2i. In contrast, Chavali S et al. reported that the SGLT2i therapy was safe and well-tolerated. Moady G et al. also found that SGLT2i were well-tolerated with only minor adverse effects . Regarding driveline infections and SGLT2i use, we acknowledge that current data from non-immunosuppressed populations do not show a significant increase in non-genital skin and soft tissue infections with SGLT2i therapy. A post-hoc analysis of the CANVAS and CREDENCE trials found no difference in non-genital infection rates between canagliflozin and placebo (HR 0.97, 95% CI 0.85–1.11; P = 0.70), and no significant increase in non-genital fungal infections either . Thus, we do not have suffitient evidence regarding the risk of skin infections or potential driveline infections in LVAD patients. Additionally, due to the lack of evidence in immunosuppressed patients means we cannot definitively rule out the possibility of a higher risk in HT recipients. This is an important area for future research. Meanwhile, the glycemic improvement associated with SGLT2i is expected to reduce infection risk by improving overall metabolic control, which should be considered in assessing their benefit-risk profile . Clinical implications and future directions The evidence suggests that SGLT2i could play a vital role in managing LVAD patients, particularly those with concurrent diabetes. The current studies indicate potential benefits in glycemic control, fluid management, and hemodynamic stability. Across all studies, SGLT2i therapy was generally well-tolerated, with no significant increase in adverse events such as infections or ketoacidosis. This is particularly noteworthy given the complex and high-risk nature of LVAD patients, who often require multiple medications and have significant comorbidities. Fardman et al. reported improvements in fluid management and pulmonary pressures in patients treated with SGLT2i, with a significant reduction in invasively derived systolic pulmonary artery pressure (PAP). Given that the maximum effect of left ventricular (LV) unloading on PAP reduction is usually achieved between 1 and 3 months post-surgery and the fact that SGLT2i were initiated at a median of 108 days (IQR 26–477) after LVAD implantation in this study, the authors hypothesized that the maximum effect of mechanical LV unloading on PAP reduction had already been reached . Hence, they postulated that the observed further reduction in PAP could be attributed to the effects of SGLT2i. The beneficial effects of SGLT2i on PAP have already been investigated by previous studies, including the EMBRACE-HF trial . The proposed mechanisms for these effects include natriuretic synergism with loop diuretics. Accordingly, the authors reported a significant reduction in the diuretic dose in patients treated with SGLT2i. Moady et al. also observed a transient reduction in diuretic dose in patients treated with SGLT2i. They noted an increased number of suction events potentially attributable to enhanced diuresis during the first week of treatment . SGLT2i block sodium and glucose reabsorption in the proximal tubule, but the duration and nature of their natriuretic and diuretic effects depend on the upregulation of SGLT2 and sodium-hydrogen exchanger 3 (NHE3), along with compensatory mechanisms in downstream nephrons. In euvolemic patients, counterregulatory sodium and water retention mechanisms activate more rapidly than in fluid-overloaded patients, limiting the duration of diuresis . LVAD patients often have impaired renal function, and the glycosuria-related effects of SGLT2 inhibitors are reduced in patients with a glomerular filtration rate (GFR) below 45 mL/min. However, patients with chronic kidney disease still benefit from SGLT2 inhibitors . While this evidence is encouraging, Cagliostro et al. and Chavali et al. did not find significant changes in diuretic dose with SGLT2i . Additionally, Chavali et al. did not observe any differences in the hemodynamic parameters after the initiation of SGLT2i . Despite the overall improvement in systolic PAP, Fardman et al. found an increase in the rate of right ventricular (RV) dysfunction. Severe RV dysfunction was observed in n = 3 (14%) of patients at follow-up, whereas none had severe RV dysfunction at baseline.This rate is comparable to the 8–11% reported in the literature . Additionally, n = 2 (9.5%) of the patients had dilated RV at baseline, compared to n = 7 (33.3%) at follow-up. The authors report that the assessment of the RV function was performed visually by experienced sonographer but provide no further details regarding the modality of the analysis. Despite the observed deterioration of the RV function, there was no difference in the invasively derived right atrial pressure (RAP) 6 months after the initiation of SGLT2i. Additionally, none of the patients with severe RV dysfunction required hospitalization for HF during the study period . The discrepancy between echocardiographic findings and clinical outcomes warrants further investigation. Moady et al. observed a transient worsening of renal function after initiation of SGLT2i, as evidenced by a drop in eGFR of ≥ 10% from baseline. This decline was temporary and stabilized after 2–3 months of therapy . The initial reduction in eGFR is attributed to the modest reduction in blood pressure seen with SGLT2i, but it does not indicate poor long-term renal outcomes . Additionally, all other studies reported no significant changes in renal function throughout their respective observational periods. This is a critical outcome for LVAD patients with diabetes mellitus. Improved glucose management can help reduce the risk of complications like infections and enhance overall health . Cagliostro M et al. did not observe significant changes in BMI or HbA1c. In contrast, Fardman A et al. reported a significant weight reduction, while Moady G et al. noted a significant decrease in HbA1c following the initiation of SGLT2i. Chavali S et al. found no substantial changes in body weight . In a meta-analysis on diabetic patients, SGLT2i reduced HbA1c by 0.62% (95% CI -0.66 to -0.59) and body weight by 0.60 kg (95% CI -0.64 to -0.55). Moreover, the EMPA-REG trial demonstrated that the lowest dose of empagliflozin (10 mg) offers comparable benefits in lowering HbA1c, body weight, blood pressure, and reducing total and cardiovascular mortality as the highest dose (25 mg) . While SGLT2i have shown notable improvements in HbA1c and weight in the general population, studies on their metabolic effects in LVAD patients have yielded inconsistent results. Across all studies, the incidence of adverse events was generally low. Regarding safety and tolerability, the study by Cagliostro M et al. identified several adverse events potentially attributable to SGLT2i, including genitourinary infections, acute kidney injury, limb amputations, and driveline infections (DLI). Similarly, Fardman A et al. observed increased RV dysfunction in some patients but found no significant overall adverse effects linked to SGLT2i. In contrast, Chavali S et al. reported that the SGLT2i therapy was safe and well-tolerated. Moady G et al. also found that SGLT2i were well-tolerated with only minor adverse effects . Regarding driveline infections and SGLT2i use, we acknowledge that current data from non-immunosuppressed populations do not show a significant increase in non-genital skin and soft tissue infections with SGLT2i therapy. A post-hoc analysis of the CANVAS and CREDENCE trials found no difference in non-genital infection rates between canagliflozin and placebo (HR 0.97, 95% CI 0.85–1.11; P = 0.70), and no significant increase in non-genital fungal infections either . Thus, we do not have suffitient evidence regarding the risk of skin infections or potential driveline infections in LVAD patients. Additionally, due to the lack of evidence in immunosuppressed patients means we cannot definitively rule out the possibility of a higher risk in HT recipients. This is an important area for future research. Meanwhile, the glycemic improvement associated with SGLT2i is expected to reduce infection risk by improving overall metabolic control, which should be considered in assessing their benefit-risk profile . The evidence suggests that SGLT2i could play a vital role in managing LVAD patients, particularly those with concurrent diabetes. The current studies indicate potential benefits in glycemic control, fluid management, and hemodynamic stability. Across all studies, SGLT2i therapy was generally well-tolerated, with no significant increase in adverse events such as infections or ketoacidosis. This is particularly noteworthy given the complex and high-risk nature of LVAD patients, who often require multiple medications and have significant comorbidities. The safety and efficacy of SGLT2i in HT recipients have been explored in two pivotal studies . Both reports provide valuable insights into the efficacy and safety of SGLT2i in this specific patient population, although they use different methodologies (Table ). Baseline characteristics Both studies were retrospective observational studies. All patients on SGLT2i ( n = 39, 32%) were treated with empagliflozin. Most of these patients were male ( n = 31, 86.1%), and the median follow-up period was 9.1 months in the study by Cehic et al. and 12 months in the study by Sammour et al. The median age was 59 years in both studies. The etiology of advanced HF prior to HT was not consistently reported. However, about a third of the patients in the study by Sammour et al. had a history of ischemic cardiomyopathy ( n = 7, 33.3%). All patients had diabetes. In both studies, patients had renal insufficiency classified as G3a according to the KDIGO classification of renal function (Table ). Outcomes Both studies included in this review examined similar outcomes, specifically focusing on the safety and efficacy of the SGLT2i (Table ). Fluid Management and Hemodynamics The treatment with empagliflozin led to a significant reduction in the median furosemide dose in the study by Cehic et al., with no significant change in systolic or diastolic blood pressure . The study by Sammour et al. did not specifically report on fluid management or hemodynamic parameters . However, the significant reductions in insulin requirements and body weight across the study population might indirectly influence fluid status. Insulin is known to stimulate sodium retention. Studies in diabetic models demonstrate that maintaining baseline insulin levels, even in the presence of hyperglycemia, can counteract natriuresis; thus, chronic hyperinsulinemia may lead to sustained sodium retention, particularly in the context of insulin resistance and metabolic disorders . Echocardiographic parameters The studies provided no data regarding echocardiographic parameters or their changes in the assessed HT populations. Renal outcomes In both studies, renal function remained stable in patients treated with empagliflozin compared to those on other glucose-lowering medication, highlighting the safety of empagliflozin with respect to renal function . However, a more extended observation period is needed to determine whether SGLT2i also offer renal protection for patients without diabetes mellitus. Recent evidence from the EMPA-KIDNEY trial demonstrates that empagliflozin promotes renal protection in patients with chronic kidney disease (CKD), irrespective of diabetes status. In a large population of patients with eGFR ranging from 20 to 45 ml/min/1.73 m² empagliflozin significantly reduced the risk of progression of kidney disease or death from cardiovascular causes compared to placebo, with consistent results across both diabetic and non-diabetic subgroups. Given these findings, further research is warranted to fully elucidate the long-term renal benefits of SGLT2i in HT patients without diabetes . Glycemic Control and Weight Overall, treatment with SGLT2i resulted in significant weight reduction and improved glycemic control. In the study by Cehic et al., empagliflozin treatment led to a decrease in body weight (median reduction of − 2.0 kg) and a mean reduction in HbA1c of 2.8% (6.6 mmol/mol) . Similarly, Sammour et al. found that combining GLP-1 receptor agonists (GLP1-RAs) and SGLT2i significantly decreased body weight and HbA1c (from 7.5 to 6.6%) . Additionally, there was a significant reduction in insulin requirements. Adverse effects Empagliflozin was well tolerated in the study by Cehic et al., with only three adverse effects reported: exacerbation of urinary symptoms, dizziness, and acute kidney injury. Notably, no genitourinary infections were observed in the empagliflozin group . Similarly, the study by Sammour et al. reported no incidents of diabetic ketoacidosis, pancreatitis, or genital mycotic infections among patients using GLP-1RAs and SGLT2i . Additionally, there were no noted drug interactions with immunosuppressive therapies. Clinical implications and future directions Both studies underscore the safety and efficacy of SGLT2i, alone or in combination with other antiglycemic agents for managing diabetes in HT recipients. Cehic et al. report that empagliflozin therapy is associated with potential benefits in improving body weight and glycemic control, while highlighting its safety in this population . Similarly, Sammour et al. demonstrate that combining GLP-1 RAs and SGLT2i leads to significant improvements in insulin requirements, HbA1c, and body weight, suggesting the effectiveness of these therapies in managing HT recipients with diabetes . However, there is a substantial gap in research concerning the role of SGLT2i in the vulnerable population of HT recipients. To address this, randomized controlled trials are needed to further validate the safety and efficacy of SGLT2i, both alone and in combination with other antiglycemic agents. Long-term studies are needed to evaluate their influence on renal function and metabolic complications following HT. Additionally, research should assess the impact of these therapies on transplant-related mortality, graft function, rejection episodes, coronary allograft vasculopathy, and the progression of chronic kidney disease. Currently, a phase 3 randomized placebo-controlled trial is ongoing, which evaluates the protective effects of dapagliflozin on renal function in patients after HT (DAPAHRT trial; ClinicalTrials.gov Identifier NCT05321706; estimated study completion in 2027). Determining the optimal timing for initiating treatment is fundamental to optimize the therapy and mitigate potential risks. In addition, it is crucial to explore the use of SGLT2i in HT recipients without diabetes. These data could reveal potential benefits or risks, contributing to more informed therapeutic decisions and improving overall patient management in the post-transplant setting. Both studies were retrospective observational studies. All patients on SGLT2i ( n = 39, 32%) were treated with empagliflozin. Most of these patients were male ( n = 31, 86.1%), and the median follow-up period was 9.1 months in the study by Cehic et al. and 12 months in the study by Sammour et al. The median age was 59 years in both studies. The etiology of advanced HF prior to HT was not consistently reported. However, about a third of the patients in the study by Sammour et al. had a history of ischemic cardiomyopathy ( n = 7, 33.3%). All patients had diabetes. In both studies, patients had renal insufficiency classified as G3a according to the KDIGO classification of renal function (Table ). Both studies included in this review examined similar outcomes, specifically focusing on the safety and efficacy of the SGLT2i (Table ). Fluid Management and Hemodynamics The treatment with empagliflozin led to a significant reduction in the median furosemide dose in the study by Cehic et al., with no significant change in systolic or diastolic blood pressure . The study by Sammour et al. did not specifically report on fluid management or hemodynamic parameters . However, the significant reductions in insulin requirements and body weight across the study population might indirectly influence fluid status. Insulin is known to stimulate sodium retention. Studies in diabetic models demonstrate that maintaining baseline insulin levels, even in the presence of hyperglycemia, can counteract natriuresis; thus, chronic hyperinsulinemia may lead to sustained sodium retention, particularly in the context of insulin resistance and metabolic disorders . Echocardiographic parameters The studies provided no data regarding echocardiographic parameters or their changes in the assessed HT populations. Renal outcomes In both studies, renal function remained stable in patients treated with empagliflozin compared to those on other glucose-lowering medication, highlighting the safety of empagliflozin with respect to renal function . However, a more extended observation period is needed to determine whether SGLT2i also offer renal protection for patients without diabetes mellitus. Recent evidence from the EMPA-KIDNEY trial demonstrates that empagliflozin promotes renal protection in patients with chronic kidney disease (CKD), irrespective of diabetes status. In a large population of patients with eGFR ranging from 20 to 45 ml/min/1.73 m² empagliflozin significantly reduced the risk of progression of kidney disease or death from cardiovascular causes compared to placebo, with consistent results across both diabetic and non-diabetic subgroups. Given these findings, further research is warranted to fully elucidate the long-term renal benefits of SGLT2i in HT patients without diabetes . Glycemic Control and Weight Overall, treatment with SGLT2i resulted in significant weight reduction and improved glycemic control. In the study by Cehic et al., empagliflozin treatment led to a decrease in body weight (median reduction of − 2.0 kg) and a mean reduction in HbA1c of 2.8% (6.6 mmol/mol) . Similarly, Sammour et al. found that combining GLP-1 receptor agonists (GLP1-RAs) and SGLT2i significantly decreased body weight and HbA1c (from 7.5 to 6.6%) . Additionally, there was a significant reduction in insulin requirements. Adverse effects Empagliflozin was well tolerated in the study by Cehic et al., with only three adverse effects reported: exacerbation of urinary symptoms, dizziness, and acute kidney injury. Notably, no genitourinary infections were observed in the empagliflozin group . Similarly, the study by Sammour et al. reported no incidents of diabetic ketoacidosis, pancreatitis, or genital mycotic infections among patients using GLP-1RAs and SGLT2i . Additionally, there were no noted drug interactions with immunosuppressive therapies. Clinical implications and future directions Both studies underscore the safety and efficacy of SGLT2i, alone or in combination with other antiglycemic agents for managing diabetes in HT recipients. Cehic et al. report that empagliflozin therapy is associated with potential benefits in improving body weight and glycemic control, while highlighting its safety in this population . Similarly, Sammour et al. demonstrate that combining GLP-1 RAs and SGLT2i leads to significant improvements in insulin requirements, HbA1c, and body weight, suggesting the effectiveness of these therapies in managing HT recipients with diabetes . However, there is a substantial gap in research concerning the role of SGLT2i in the vulnerable population of HT recipients. To address this, randomized controlled trials are needed to further validate the safety and efficacy of SGLT2i, both alone and in combination with other antiglycemic agents. Long-term studies are needed to evaluate their influence on renal function and metabolic complications following HT. Additionally, research should assess the impact of these therapies on transplant-related mortality, graft function, rejection episodes, coronary allograft vasculopathy, and the progression of chronic kidney disease. Currently, a phase 3 randomized placebo-controlled trial is ongoing, which evaluates the protective effects of dapagliflozin on renal function in patients after HT (DAPAHRT trial; ClinicalTrials.gov Identifier NCT05321706; estimated study completion in 2027). Determining the optimal timing for initiating treatment is fundamental to optimize the therapy and mitigate potential risks. In addition, it is crucial to explore the use of SGLT2i in HT recipients without diabetes. These data could reveal potential benefits or risks, contributing to more informed therapeutic decisions and improving overall patient management in the post-transplant setting. The treatment with empagliflozin led to a significant reduction in the median furosemide dose in the study by Cehic et al., with no significant change in systolic or diastolic blood pressure . The study by Sammour et al. did not specifically report on fluid management or hemodynamic parameters . However, the significant reductions in insulin requirements and body weight across the study population might indirectly influence fluid status. Insulin is known to stimulate sodium retention. Studies in diabetic models demonstrate that maintaining baseline insulin levels, even in the presence of hyperglycemia, can counteract natriuresis; thus, chronic hyperinsulinemia may lead to sustained sodium retention, particularly in the context of insulin resistance and metabolic disorders . The studies provided no data regarding echocardiographic parameters or their changes in the assessed HT populations. In both studies, renal function remained stable in patients treated with empagliflozin compared to those on other glucose-lowering medication, highlighting the safety of empagliflozin with respect to renal function . However, a more extended observation period is needed to determine whether SGLT2i also offer renal protection for patients without diabetes mellitus. Recent evidence from the EMPA-KIDNEY trial demonstrates that empagliflozin promotes renal protection in patients with chronic kidney disease (CKD), irrespective of diabetes status. In a large population of patients with eGFR ranging from 20 to 45 ml/min/1.73 m² empagliflozin significantly reduced the risk of progression of kidney disease or death from cardiovascular causes compared to placebo, with consistent results across both diabetic and non-diabetic subgroups. Given these findings, further research is warranted to fully elucidate the long-term renal benefits of SGLT2i in HT patients without diabetes . Overall, treatment with SGLT2i resulted in significant weight reduction and improved glycemic control. In the study by Cehic et al., empagliflozin treatment led to a decrease in body weight (median reduction of − 2.0 kg) and a mean reduction in HbA1c of 2.8% (6.6 mmol/mol) . Similarly, Sammour et al. found that combining GLP-1 receptor agonists (GLP1-RAs) and SGLT2i significantly decreased body weight and HbA1c (from 7.5 to 6.6%) . Additionally, there was a significant reduction in insulin requirements. Empagliflozin was well tolerated in the study by Cehic et al., with only three adverse effects reported: exacerbation of urinary symptoms, dizziness, and acute kidney injury. Notably, no genitourinary infections were observed in the empagliflozin group . Similarly, the study by Sammour et al. reported no incidents of diabetic ketoacidosis, pancreatitis, or genital mycotic infections among patients using GLP-1RAs and SGLT2i . Additionally, there were no noted drug interactions with immunosuppressive therapies. Both studies underscore the safety and efficacy of SGLT2i, alone or in combination with other antiglycemic agents for managing diabetes in HT recipients. Cehic et al. report that empagliflozin therapy is associated with potential benefits in improving body weight and glycemic control, while highlighting its safety in this population . Similarly, Sammour et al. demonstrate that combining GLP-1 RAs and SGLT2i leads to significant improvements in insulin requirements, HbA1c, and body weight, suggesting the effectiveness of these therapies in managing HT recipients with diabetes . However, there is a substantial gap in research concerning the role of SGLT2i in the vulnerable population of HT recipients. To address this, randomized controlled trials are needed to further validate the safety and efficacy of SGLT2i, both alone and in combination with other antiglycemic agents. Long-term studies are needed to evaluate their influence on renal function and metabolic complications following HT. Additionally, research should assess the impact of these therapies on transplant-related mortality, graft function, rejection episodes, coronary allograft vasculopathy, and the progression of chronic kidney disease. Currently, a phase 3 randomized placebo-controlled trial is ongoing, which evaluates the protective effects of dapagliflozin on renal function in patients after HT (DAPAHRT trial; ClinicalTrials.gov Identifier NCT05321706; estimated study completion in 2027). Determining the optimal timing for initiating treatment is fundamental to optimize the therapy and mitigate potential risks. In addition, it is crucial to explore the use of SGLT2i in HT recipients without diabetes. These data could reveal potential benefits or risks, contributing to more informed therapeutic decisions and improving overall patient management in the post-transplant setting. In conclusion, while current findings regarding the safety of SGLT2i are promising, further research is needed to confirm their efficacy in managing LVAD and HT patients. More robust data are required to safely integrate SGLT2i into the clinical management strategies for these vulnerable patient populations. This review has several limitations, including a limited number of studies with small sample size and heterogeneous methodology. Notably, there was relevant variability in the timing of SGLT2i initiation post-LVAD and HT. Furthermore, the studies had relatively short follow-up periods and provided heterogeneous information on patient characteristics. These limitations affect the generalizability and reliability of the findings, highlighting the need for more extensive and methodologically consistent research, including randomized controlled trials. |
Publication bias in pharmacogenetics of adverse reaction to antiseizure drugs: An umbrella review and a meta-epidemiological study | 5d2aab73-9c07-4d54-8b86-416896e13582 | 9803138 | Pharmacology[mh] | Rationale Systematic reviews and meta-analyses help synthesize the estimates from several clinical studies. However, their results may be affected by publication bias . At the beginning, publication bias has been pointed out regarding the risk of treatment efficacy overestimation . Then, publication bias has also been identified for assessing the risk of adverse drug reactions (ADRs) . Furthermore, publication bias has been documented in genetic epidemiology in general . Previous meta-epidemiological studies assessed publication bias in various areas . However, the extent of publication bias in pharmacogenetics remains unclear. Several genetic variants have been associated with an increased risk of antiseizure drug’s adverse effects (AEs) . Some of them seem to be reliable, as HLA-B and carbamazepine/phenytoin-induced severe cutaneous reactions, allowing clinical implementation of pharmacogenetic results . Others showed some discrepancies, such as for lamotrigine and the risk of Stevens-Johnson syndrome (SJS) or toxic epidermal necrolysis (TEN) . Such pharmacogenetic associations may also vary across different populations . A publication bias has been documented regarding the assessment of antiseizure drugs, such as topiramate or pregabalin for example . Previous meta-analyses assessing pharmacogenetic risk factors of antiseizure drug’s AE suggested a potential publication bias, including HLA-B*15:02 polymorphism . Other meta-analyses did not report a publication bias, but the number of available trials was often low, ranging from two to eleven studies in three meta-analysis . Thus, the statistical power for detecting funnel plot asymmetry was probably insufficient . The prevalence and the potential impact of publication bias in the pharmacogenetics of antiseizure drug’s AEs remain unclear. Objectives Our hypothesis was that the publication bias is particularly sizable in the pharmacogenetics of antiseizure druginduced adverse reactions. We aimed to assess its prevalence in this area first. Then, we aimed to illustrate its potential impact on the estimation of those pharmacogenetics biomarkers. Systematic reviews and meta-analyses help synthesize the estimates from several clinical studies. However, their results may be affected by publication bias . At the beginning, publication bias has been pointed out regarding the risk of treatment efficacy overestimation . Then, publication bias has also been identified for assessing the risk of adverse drug reactions (ADRs) . Furthermore, publication bias has been documented in genetic epidemiology in general . Previous meta-epidemiological studies assessed publication bias in various areas . However, the extent of publication bias in pharmacogenetics remains unclear. Several genetic variants have been associated with an increased risk of antiseizure drug’s adverse effects (AEs) . Some of them seem to be reliable, as HLA-B and carbamazepine/phenytoin-induced severe cutaneous reactions, allowing clinical implementation of pharmacogenetic results . Others showed some discrepancies, such as for lamotrigine and the risk of Stevens-Johnson syndrome (SJS) or toxic epidermal necrolysis (TEN) . Such pharmacogenetic associations may also vary across different populations . A publication bias has been documented regarding the assessment of antiseizure drugs, such as topiramate or pregabalin for example . Previous meta-analyses assessing pharmacogenetic risk factors of antiseizure drug’s AE suggested a potential publication bias, including HLA-B*15:02 polymorphism . Other meta-analyses did not report a publication bias, but the number of available trials was often low, ranging from two to eleven studies in three meta-analysis . Thus, the statistical power for detecting funnel plot asymmetry was probably insufficient . The prevalence and the potential impact of publication bias in the pharmacogenetics of antiseizure drug’s AEs remain unclear. Our hypothesis was that the publication bias is particularly sizable in the pharmacogenetics of antiseizure druginduced adverse reactions. We aimed to assess its prevalence in this area first. Then, we aimed to illustrate its potential impact on the estimation of those pharmacogenetics biomarkers. Protocol We conducted a meta-epidemiological study. We did not register the protocol, but we formulated the hypothesis a priori . Methods and results were reported following the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) statement for systematic review and the dedicated guideline for reporting meta-epidemiological methodology research . We conducted an umbrella review to identify systematic reviews and meta-analyses of interest. We did not search directly for clinical studies themselves. Indeed, the assessment of publication bias requires as many clinical studies as possible, and is expected to be inconclusive in case of isolated (i.e non-replicated) clinical studies. Therefore, we opted for an umbrella review of published systematic reviews. One bibliographic reference of a systematic review could include several published meta-analyses (pMA), one for each unique triplet association [genotype-drug-ADR] . Each unique triplet association [genotype-drug-ADR] may be studied across different published meta-analyses. We gathered all the informative clinical studies assessing the same triplet association [genotype-drug-ADR] reported across the published meta-analyses from the different systematic reviews. Combining the data from previous overlapping published meta-analyses allowed increasing sample size of data given the number of clinical studies. We then assessed the presence of a publication bias and its impact for each unique triplet association [genotype-drug-ADR] . Eligibility criteria Systematic reviews were included if they met the following inclusion criteria: (i) meta-analyses of clinical studies (observational or comparative trials) addressing antiseizure drugs and (ii) reporting of ADR related to pharmacogenetic biomarkers. Information sources We conducted an umbrella review on PubMed, up to January 29, 2019, seeking published meta-analyses investigating the association between a genetic variant and an adverse reaction in patients treated with antiseizure drugs. We limited the information source to the Medline database. Indeed, the Clinical Pharmacogenetics Implementation Consortium (CPIC) Guideline for HLA genotype and Use of Carbamazepine and Oxcarbazepine was restricted to the use of “the PubMed database” . Using the same restriction allowed to estimate the impact of publication bias in the context of the current practice of guidelines elaboration in this area. Search strategy The Medline database was searched using the following keywords: “pharmacogen*” OR “genetic variant” OR “polymorphism” OR “allele” OR “association” AND “epilepsy” OR “antiepileptics” OR “anticonvulsants” AND “adverse event” OR “toxicity” OR “serious event” OR “meta-analysis” OR “systematic review” (see the search strategy in Supplementary information), without restrictions regarding the year of publication. Study selection and data extraction Abstracts of bibliographic references were screened based on their title and abstract, and then selected using their full text, reason for exclusion being tracked. We extracted descriptive characteristics: authors, publication date, type of study, and ethnicity. We extracted study design, pharmacogenetic associations as reported by the authors (name of the genetic variant, antiseizure drug assessed, control treatment used, and adverse reactions). Indeed, as we used previously published systematic reviews of clinical studies, our extraction is limited by the author’s definition of the genetic variant (i.e not the rs number if not reported) and of the adverse drug reactions. We extracted the odds ratios (OR) with their 95% confidence intervals (95%CI) of each clinical studiy reported in the included published meta-analyses. Summary measures We used odds ratio (OR) and its 95% confidence interval (95%CI) to estimate the association between the genotype and the risk of the drug induced adverse reaction. Synthesis of result–data analysis For each unique triplet association [genotype-drug-ADR] , we gathered the ORs of the clinical studies from different published meta-analyses. We limited the compilation to comparisons using treated patients as control groups. We used the funnel plot approach for assessing potential publication bias . We generated the funnel plots for each included associations, for estimating the prevalence of publication bias. Firstly, we conducted a visual analysis of the generated funnel plots . Two researcher independently assessed if a publication bias was ‘likely’, ‘unlikely’, or ‘not determinable’ (SB, GG). Agreement was estimated using a Free-marginal kappa estimator . A third researcher helped resolve disagreements, blinded to the previous diagnoses (JC). Secondly, asymmetry was tested using the Egger and Begg’s methods (function metabias, package {meta}) if at least five studies were available. P-value <0.05 were considered significant without adjustment for multiple testing. We calculated and reported the proportion of publication bias according to these three methods. Given the guidance of Cochrane handbook, the fail-safe number method was not used . For exploring the potential impact of publication bias, we used the Trim and Fill method to adjust the OR for potential publication bias . We applied the Trim and Fill function (estimator L, fixed-effect model) if more than five estimates and their 95%CI were available for the same association [genotype-drug-ADR] ). Then, we compared the OR obtained without adjustment for publication bias (OR NP ) and the OR estimated with the Trim and Fill method (OR TM ). We also tested for an interaction between each pair of ORNP−OR TM , using the ratio of the ORs (ROR) and its 95%CI. We conducted the analyses on R 3.3.1 (package {meta}, version 4.9–4) . We conducted a meta-epidemiological study. We did not register the protocol, but we formulated the hypothesis a priori . Methods and results were reported following the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) statement for systematic review and the dedicated guideline for reporting meta-epidemiological methodology research . We conducted an umbrella review to identify systematic reviews and meta-analyses of interest. We did not search directly for clinical studies themselves. Indeed, the assessment of publication bias requires as many clinical studies as possible, and is expected to be inconclusive in case of isolated (i.e non-replicated) clinical studies. Therefore, we opted for an umbrella review of published systematic reviews. One bibliographic reference of a systematic review could include several published meta-analyses (pMA), one for each unique triplet association [genotype-drug-ADR] . Each unique triplet association [genotype-drug-ADR] may be studied across different published meta-analyses. We gathered all the informative clinical studies assessing the same triplet association [genotype-drug-ADR] reported across the published meta-analyses from the different systematic reviews. Combining the data from previous overlapping published meta-analyses allowed increasing sample size of data given the number of clinical studies. We then assessed the presence of a publication bias and its impact for each unique triplet association [genotype-drug-ADR] . Systematic reviews were included if they met the following inclusion criteria: (i) meta-analyses of clinical studies (observational or comparative trials) addressing antiseizure drugs and (ii) reporting of ADR related to pharmacogenetic biomarkers. We conducted an umbrella review on PubMed, up to January 29, 2019, seeking published meta-analyses investigating the association between a genetic variant and an adverse reaction in patients treated with antiseizure drugs. We limited the information source to the Medline database. Indeed, the Clinical Pharmacogenetics Implementation Consortium (CPIC) Guideline for HLA genotype and Use of Carbamazepine and Oxcarbazepine was restricted to the use of “the PubMed database” . Using the same restriction allowed to estimate the impact of publication bias in the context of the current practice of guidelines elaboration in this area. The Medline database was searched using the following keywords: “pharmacogen*” OR “genetic variant” OR “polymorphism” OR “allele” OR “association” AND “epilepsy” OR “antiepileptics” OR “anticonvulsants” AND “adverse event” OR “toxicity” OR “serious event” OR “meta-analysis” OR “systematic review” (see the search strategy in Supplementary information), without restrictions regarding the year of publication. Abstracts of bibliographic references were screened based on their title and abstract, and then selected using their full text, reason for exclusion being tracked. We extracted descriptive characteristics: authors, publication date, type of study, and ethnicity. We extracted study design, pharmacogenetic associations as reported by the authors (name of the genetic variant, antiseizure drug assessed, control treatment used, and adverse reactions). Indeed, as we used previously published systematic reviews of clinical studies, our extraction is limited by the author’s definition of the genetic variant (i.e not the rs number if not reported) and of the adverse drug reactions. We extracted the odds ratios (OR) with their 95% confidence intervals (95%CI) of each clinical studiy reported in the included published meta-analyses. We used odds ratio (OR) and its 95% confidence interval (95%CI) to estimate the association between the genotype and the risk of the drug induced adverse reaction. For each unique triplet association [genotype-drug-ADR] , we gathered the ORs of the clinical studies from different published meta-analyses. We limited the compilation to comparisons using treated patients as control groups. We used the funnel plot approach for assessing potential publication bias . We generated the funnel plots for each included associations, for estimating the prevalence of publication bias. Firstly, we conducted a visual analysis of the generated funnel plots . Two researcher independently assessed if a publication bias was ‘likely’, ‘unlikely’, or ‘not determinable’ (SB, GG). Agreement was estimated using a Free-marginal kappa estimator . A third researcher helped resolve disagreements, blinded to the previous diagnoses (JC). Secondly, asymmetry was tested using the Egger and Begg’s methods (function metabias, package {meta}) if at least five studies were available. P-value <0.05 were considered significant without adjustment for multiple testing. We calculated and reported the proportion of publication bias according to these three methods. Given the guidance of Cochrane handbook, the fail-safe number method was not used . For exploring the potential impact of publication bias, we used the Trim and Fill method to adjust the OR for potential publication bias . We applied the Trim and Fill function (estimator L, fixed-effect model) if more than five estimates and their 95%CI were available for the same association [genotype-drug-ADR] ). Then, we compared the OR obtained without adjustment for publication bias (OR NP ) and the OR estimated with the Trim and Fill method (OR TM ). We also tested for an interaction between each pair of ORNP−OR TM , using the ratio of the ORs (ROR) and its 95%CI. We conducted the analyses on R 3.3.1 (package {meta}, version 4.9–4) . Study selection From 295 references identified on PubMed, we included ten systematic reviews (see ). From them, we removed one systematic review that contained estimates with infinite confidence intervals, limiting their use in our study. The nine usable systematic reviews included 33 published meta-analyses . Among them, 22 unique triplet associations [genotype-drug-ADR] were available for analysis. Study characteristics The characteristics of included meta-analyses (published meta-analyses and each specific triplet association [genotype-drug-ADR] ) are detailed in the . In addition to the first author with the corresponding bibliographic reference of the published systematic review, a specific ‘identification number’ identified each published meta-analysis, a specific letter identified each specific triplet association [genotype-drug-ADR] . Most of the included patients were Asian. Most of the gene variants were related to the HLA system. One published meta-analysis addressed the genetic polymorphisms of CYP2C9. Six antiseizure drugs were assessed (carbamazepine, lamotrigine, levetiracetam, phenobarbital, phenytoin, and valproate). The reported ADR were: “hypersensitivity”, hypersensitive syndrome (HSS), maculopapular exanthema (MPE), serious cutaneous reactions (SCRs), SJS, and TEN. All the selected meta-analyses included case-control study design. Used comparators were: treated but tolerant patients in 28 published meta-analyses, untreated patients in four published meta-analyses, and both in one published meta-analysis. Prevalence of the publication bias Twenty-two funnel plots of specific associations [genotype-drug-ADR] were generated. The funnels plots are available in the S1 Fig in . The summarises the estimation of the prevalence of publication bias. Visual diagnosis of funnel plots For the visual analysis of the funnel plots, before reaching consensus with the third reviewer, the percentage of overall agreement between the two initial reviewers was 64%, Free-marginal kappa = 0.45 [95%CI: 0.15; 0.76]. Our visual analysis of generated funnel plot estimated that a publication bias was i) “likely” in five (23% [95%CI: 8; 45]) out of 22 exploitable funnel plots; ii) “unlikely” in 3 (14% [95%CI: 3; 35]) out of 22 exploitable funnel plots, and iii) “not determinable” in 14 (64% [95%CI: 41; 83]) out of 22 exploitable funnel plots. Assymetry tests The Egger’s test i) showed a significant (p<0.05) publication bias in one out of nine assessable funnel plots (11% [95%CI: 0; 48]). It was not able to conclude (p>0.05) in eight out of nine assessable funnel plots (89% [95%CI: 52; 100]). The funnel plots were not exploitable in 13 out of 22 cases (59% [95%CI: 36; 79]). The Begg’s test showed a significant (p<0.05) publication bias in one out of 10 assessable funnel plots (10% [95%CI: 0; 45]). It was not able to conclude (p>0.05) in 9 out of 10 assessable funnel plots (90% [95%CI: 55; 100]). It was not assessable in 12 out of 22 exploitable funnel plots (55% [95%CI: 32; 76]). The two significant publication bias identified by the asymmetry tests affected the associations of HLA-B*15:02 and carbamazepine induced SCRs (Begg’s test, p-value = 0.02, see ) and SJS and TEN related to phenytoin (Egger’s test, p = 0.03, see ). Exploration of the impact of publication bias The Trim and Fill estimates were calculable for 7 out the 22 funnel plots. Most of the associations suggested an increased risk of antiseizure drug’s AE with the pharmacogenetic biomarkers. The size of the associations were highly modified when taking into account a potential publication bias; the quantitative impact ranged from halving to doubling the estimation of the association. No association was qualitatively modified by taking into account the publication bias. The interaction tests between OR NP and OR TM were not significant.. The OR NP , OR TM , and ROR are detailed in the . From 295 references identified on PubMed, we included ten systematic reviews (see ). From them, we removed one systematic review that contained estimates with infinite confidence intervals, limiting their use in our study. The nine usable systematic reviews included 33 published meta-analyses . Among them, 22 unique triplet associations [genotype-drug-ADR] were available for analysis. The characteristics of included meta-analyses (published meta-analyses and each specific triplet association [genotype-drug-ADR] ) are detailed in the . In addition to the first author with the corresponding bibliographic reference of the published systematic review, a specific ‘identification number’ identified each published meta-analysis, a specific letter identified each specific triplet association [genotype-drug-ADR] . Most of the included patients were Asian. Most of the gene variants were related to the HLA system. One published meta-analysis addressed the genetic polymorphisms of CYP2C9. Six antiseizure drugs were assessed (carbamazepine, lamotrigine, levetiracetam, phenobarbital, phenytoin, and valproate). The reported ADR were: “hypersensitivity”, hypersensitive syndrome (HSS), maculopapular exanthema (MPE), serious cutaneous reactions (SCRs), SJS, and TEN. All the selected meta-analyses included case-control study design. Used comparators were: treated but tolerant patients in 28 published meta-analyses, untreated patients in four published meta-analyses, and both in one published meta-analysis. Twenty-two funnel plots of specific associations [genotype-drug-ADR] were generated. The funnels plots are available in the S1 Fig in . The summarises the estimation of the prevalence of publication bias. Visual diagnosis of funnel plots For the visual analysis of the funnel plots, before reaching consensus with the third reviewer, the percentage of overall agreement between the two initial reviewers was 64%, Free-marginal kappa = 0.45 [95%CI: 0.15; 0.76]. Our visual analysis of generated funnel plot estimated that a publication bias was i) “likely” in five (23% [95%CI: 8; 45]) out of 22 exploitable funnel plots; ii) “unlikely” in 3 (14% [95%CI: 3; 35]) out of 22 exploitable funnel plots, and iii) “not determinable” in 14 (64% [95%CI: 41; 83]) out of 22 exploitable funnel plots. Assymetry tests The Egger’s test i) showed a significant (p<0.05) publication bias in one out of nine assessable funnel plots (11% [95%CI: 0; 48]). It was not able to conclude (p>0.05) in eight out of nine assessable funnel plots (89% [95%CI: 52; 100]). The funnel plots were not exploitable in 13 out of 22 cases (59% [95%CI: 36; 79]). The Begg’s test showed a significant (p<0.05) publication bias in one out of 10 assessable funnel plots (10% [95%CI: 0; 45]). It was not able to conclude (p>0.05) in 9 out of 10 assessable funnel plots (90% [95%CI: 55; 100]). It was not assessable in 12 out of 22 exploitable funnel plots (55% [95%CI: 32; 76]). The two significant publication bias identified by the asymmetry tests affected the associations of HLA-B*15:02 and carbamazepine induced SCRs (Begg’s test, p-value = 0.02, see ) and SJS and TEN related to phenytoin (Egger’s test, p = 0.03, see ). For the visual analysis of the funnel plots, before reaching consensus with the third reviewer, the percentage of overall agreement between the two initial reviewers was 64%, Free-marginal kappa = 0.45 [95%CI: 0.15; 0.76]. Our visual analysis of generated funnel plot estimated that a publication bias was i) “likely” in five (23% [95%CI: 8; 45]) out of 22 exploitable funnel plots; ii) “unlikely” in 3 (14% [95%CI: 3; 35]) out of 22 exploitable funnel plots, and iii) “not determinable” in 14 (64% [95%CI: 41; 83]) out of 22 exploitable funnel plots. The Egger’s test i) showed a significant (p<0.05) publication bias in one out of nine assessable funnel plots (11% [95%CI: 0; 48]). It was not able to conclude (p>0.05) in eight out of nine assessable funnel plots (89% [95%CI: 52; 100]). The funnel plots were not exploitable in 13 out of 22 cases (59% [95%CI: 36; 79]). The Begg’s test showed a significant (p<0.05) publication bias in one out of 10 assessable funnel plots (10% [95%CI: 0; 45]). It was not able to conclude (p>0.05) in 9 out of 10 assessable funnel plots (90% [95%CI: 55; 100]). It was not assessable in 12 out of 22 exploitable funnel plots (55% [95%CI: 32; 76]). The two significant publication bias identified by the asymmetry tests affected the associations of HLA-B*15:02 and carbamazepine induced SCRs (Begg’s test, p-value = 0.02, see ) and SJS and TEN related to phenytoin (Egger’s test, p = 0.03, see ). The Trim and Fill estimates were calculable for 7 out the 22 funnel plots. Most of the associations suggested an increased risk of antiseizure drug’s AE with the pharmacogenetic biomarkers. The size of the associations were highly modified when taking into account a potential publication bias; the quantitative impact ranged from halving to doubling the estimation of the association. No association was qualitatively modified by taking into account the publication bias. The interaction tests between OR NP and OR TM were not significant.. The OR NP , OR TM , and ROR are detailed in the . Summary of evidence Most of the published meta-analyses on pharmacogenetic biomarkers of antiseizure drug’s AE reported cutaneous complications. We showed that a publication bias was not rare in the assessment of pharmacogenetic biomarkers of antiseizure drug’s AE. The visual analysis of the funnel plots showed that a publication bias might affect almost one quarter of those associations in this field. Using asymmetry tests, we showed that about 10% of those associations were subject to a significant publication bias. We showed that taking into account a potential publication bias might double or halve the estimation of the risk of antiseizure drug’s AE associated with those genetic biomarkers. Our results suggested a significant publication bias for the HLA-B*15–2 and its association with the risk of carbamazepine-induced serious cutaneous reactions and of phenytoin-induced SJS or TEN. This may challenge the Clinical Pharmacogenetics Implementation Consortium guidelines for carbamazepine and phenytoin and should be discussed in the other European pharmacogenetic networks (such as the French Network of Pharmacogenetics—RNPGx—, the Dutch Pharmacogenetics Working Group—DPWG—). However, the risk of SJS, TEN, and MPE with carbamazepine remained highly increased by the presence of the HLA-B*15:02 genotype, even when taking into account the publication bias. The effect of these associations remains high (from ≈4 to ≈40), not negating their use in clinical practice. Unfortunately, the present study was limited by the lack of power related to the study number, notably for adjusting the estimates of the association between HLA-B*15:02 and phenytoin related ADR. Finally, adjusting for publication bias affected the estimation of the pharmacogenetic associations. Indeed, even for the association between the genetic variant HLA-B*15:02 and the risk of SCRs and SJS in people treated with carbamazepine, the increase of risk appears to be overestimated by two-folds. In contrast, the association between the genetic variant HLA-A*31:01 and the risk of SJS in people treated with carbamazepine seems to be underestimated by a factor two when taking into account a potential publication bias (OR NP of 6 versus OR TM of 12), surprisingly suggesting potential highly significant unpublished clinical studies. Strengths of the study To our knowledge, our study is the first meta-epidemiological assessment of the publication bias in the pharmacogenetics of antiseizure drug adverse reactions. We used a systematic umbrella review, allowing us to gather the information from overlapping published meta-analyses and to increase the sample size. We provided an estimation of the prevalence of the publication bias in this field, using several assessments of publication bias, including independent visual analyses of the funnel plots. We also explored the impact of publication bias on the size of the effect of such association, and finally discussed the potential consequences of those results on guidelines for clinical implementation of pharmacogenetics. Limitations However, our study presents several limits. First, we used the visual analysis of funnel plots, which is exposed to subjectivity. However, there is no consensual approach available for a publication bias assessment . This limitation illustrates the need for new tools for a publication bias assessment. Moreover, publication bias is not the only source of funnel plot asymmetry. Especially, a poor methodological quality of the small-included studies might led to the “small study effect”. A true heterogeneity also might contribute to the funnel asymmetry. However, in our study, some of the punctual estimates seem to line up with usual cut-off p values for significance (see notably , along the 1% cut off); the use of contour-enhanced funnel plots allowed to highlight that the asymmetry seems to be associated with the significance of the included studies. Moreover, we believed that the heterogeneity is limited in our funnel plots, as we kept the most precise granularity, notably by not gathering different, despite close, adverse drug reactions, in line with the author’s definitions. Furthermore, tools for a publication bias assessment, as the Egger’s test for example, have been initially developed for a meta-analysis of randomized trials. However, most of the available pharmacogenetic studies here were non-randomized. If the same tools for publication bias assessment may be used in meta-analysis of such non-randomized studies remains unclear. Second, despite our umbrella review approach, the number of clinical trials available remained limited for several associations. It limited the assessment of publication bias, as illustrated by the significant number of “not determinable” assessment. We were not able to provide publication bias adjusted estimates of the association between HLA-B*15:02 and phenytoin related ADR, especially. We did not conduct additional searches of original clinical studies. Indeed, focusing on published systematic review allowed assessing the impact of the publication bias in the currently available meta-analyses, which are used for guideline elaboration. We also did not remove the associations with a low number of estimates, leading to report results of little utility. Indeed, we did not consider exclusion criteria based on the number of point estimates. Therefore, we reported all the associations. Such exclusion criteria would require an arbitrary cut-off, which might be questionable. Above all, such exclusion criteria might lead to overestimate the prevalence of publication bias in the field. It would have been possible to gather together some associations, some of whom displayed similar ADRs (as “SCRs”, “SJS”, “SJS, TEN” considered in different associations). Indeed, some logical grouping are probably legit and relevant. This would have increased the power of the analysis. For example, combining the association A, B and C ([HLA-B*15:02 –carbamazepine–SCR/SJS/SJS,TEN] allowed to reach a nominal p value <0.05 for both the tests of Egger (p = 0.0009) and of Begg (p = 0.02) (S2 Fig in ). However, this would have probably lead to overestimate the prevalence of the publication bias in the field. Therefore, we preferred a more conservative approach by respecting the reported ADRs as defined by the authors of the clinical studies and of the published meta-analyses, despite the possible decrease of the power of our analysis. Third, the asymmetry of funnel is not entirely specific of the presence of a publication bias and can be related to heterogeneity in treatment effects. We also used a low number of studies as cut-off for using asymmetry tests. Fourth, we also used the Trim and Fill method, which requires some assumptions: it simulates potential missing studies as mirror images of observed studies . Furthermore, most of the included clinical studies were at high risk of non-publication bias, including confusion and selection bias. Indeed, most of those genetic biomarker are in fact prognostic biomarkers of drug adverse reactions in a treated population. Moreover, we did not study the potential effect of subpopulations on those pharmacogenetics associations . Furthermore, we used the genotype’s definition reported in the published systematic review, even when rs number were not available. Last but not least, a publication bias for other associations might exist and remains invisible if most of the corresponding studies are not published at all. Implications of the study Our study showed that,—as in clinical pharmacology or in genetics —, publication is selective in pharmacogenetics. We showed that this publication bias may affect the assessment of an association between genetic biomarker and ADR, even for consensual pharmacogenetic biomarkers of antiseizure drug’s AE as HLA-B*15:02. Our study showed the need for publishing or at least registering any pharmacogenetics study, even with inconclusive results, to fight the issue of publication bias and its harmful consequences. The increased access to genomic databases and to genomic data through next generation sequencing strengthens the importance to anticipate such issues. Most of the published meta-analyses on pharmacogenetic biomarkers of antiseizure drug’s AE reported cutaneous complications. We showed that a publication bias was not rare in the assessment of pharmacogenetic biomarkers of antiseizure drug’s AE. The visual analysis of the funnel plots showed that a publication bias might affect almost one quarter of those associations in this field. Using asymmetry tests, we showed that about 10% of those associations were subject to a significant publication bias. We showed that taking into account a potential publication bias might double or halve the estimation of the risk of antiseizure drug’s AE associated with those genetic biomarkers. Our results suggested a significant publication bias for the HLA-B*15–2 and its association with the risk of carbamazepine-induced serious cutaneous reactions and of phenytoin-induced SJS or TEN. This may challenge the Clinical Pharmacogenetics Implementation Consortium guidelines for carbamazepine and phenytoin and should be discussed in the other European pharmacogenetic networks (such as the French Network of Pharmacogenetics—RNPGx—, the Dutch Pharmacogenetics Working Group—DPWG—). However, the risk of SJS, TEN, and MPE with carbamazepine remained highly increased by the presence of the HLA-B*15:02 genotype, even when taking into account the publication bias. The effect of these associations remains high (from ≈4 to ≈40), not negating their use in clinical practice. Unfortunately, the present study was limited by the lack of power related to the study number, notably for adjusting the estimates of the association between HLA-B*15:02 and phenytoin related ADR. Finally, adjusting for publication bias affected the estimation of the pharmacogenetic associations. Indeed, even for the association between the genetic variant HLA-B*15:02 and the risk of SCRs and SJS in people treated with carbamazepine, the increase of risk appears to be overestimated by two-folds. In contrast, the association between the genetic variant HLA-A*31:01 and the risk of SJS in people treated with carbamazepine seems to be underestimated by a factor two when taking into account a potential publication bias (OR NP of 6 versus OR TM of 12), surprisingly suggesting potential highly significant unpublished clinical studies. To our knowledge, our study is the first meta-epidemiological assessment of the publication bias in the pharmacogenetics of antiseizure drug adverse reactions. We used a systematic umbrella review, allowing us to gather the information from overlapping published meta-analyses and to increase the sample size. We provided an estimation of the prevalence of the publication bias in this field, using several assessments of publication bias, including independent visual analyses of the funnel plots. We also explored the impact of publication bias on the size of the effect of such association, and finally discussed the potential consequences of those results on guidelines for clinical implementation of pharmacogenetics. However, our study presents several limits. First, we used the visual analysis of funnel plots, which is exposed to subjectivity. However, there is no consensual approach available for a publication bias assessment . This limitation illustrates the need for new tools for a publication bias assessment. Moreover, publication bias is not the only source of funnel plot asymmetry. Especially, a poor methodological quality of the small-included studies might led to the “small study effect”. A true heterogeneity also might contribute to the funnel asymmetry. However, in our study, some of the punctual estimates seem to line up with usual cut-off p values for significance (see notably , along the 1% cut off); the use of contour-enhanced funnel plots allowed to highlight that the asymmetry seems to be associated with the significance of the included studies. Moreover, we believed that the heterogeneity is limited in our funnel plots, as we kept the most precise granularity, notably by not gathering different, despite close, adverse drug reactions, in line with the author’s definitions. Furthermore, tools for a publication bias assessment, as the Egger’s test for example, have been initially developed for a meta-analysis of randomized trials. However, most of the available pharmacogenetic studies here were non-randomized. If the same tools for publication bias assessment may be used in meta-analysis of such non-randomized studies remains unclear. Second, despite our umbrella review approach, the number of clinical trials available remained limited for several associations. It limited the assessment of publication bias, as illustrated by the significant number of “not determinable” assessment. We were not able to provide publication bias adjusted estimates of the association between HLA-B*15:02 and phenytoin related ADR, especially. We did not conduct additional searches of original clinical studies. Indeed, focusing on published systematic review allowed assessing the impact of the publication bias in the currently available meta-analyses, which are used for guideline elaboration. We also did not remove the associations with a low number of estimates, leading to report results of little utility. Indeed, we did not consider exclusion criteria based on the number of point estimates. Therefore, we reported all the associations. Such exclusion criteria would require an arbitrary cut-off, which might be questionable. Above all, such exclusion criteria might lead to overestimate the prevalence of publication bias in the field. It would have been possible to gather together some associations, some of whom displayed similar ADRs (as “SCRs”, “SJS”, “SJS, TEN” considered in different associations). Indeed, some logical grouping are probably legit and relevant. This would have increased the power of the analysis. For example, combining the association A, B and C ([HLA-B*15:02 –carbamazepine–SCR/SJS/SJS,TEN] allowed to reach a nominal p value <0.05 for both the tests of Egger (p = 0.0009) and of Begg (p = 0.02) (S2 Fig in ). However, this would have probably lead to overestimate the prevalence of the publication bias in the field. Therefore, we preferred a more conservative approach by respecting the reported ADRs as defined by the authors of the clinical studies and of the published meta-analyses, despite the possible decrease of the power of our analysis. Third, the asymmetry of funnel is not entirely specific of the presence of a publication bias and can be related to heterogeneity in treatment effects. We also used a low number of studies as cut-off for using asymmetry tests. Fourth, we also used the Trim and Fill method, which requires some assumptions: it simulates potential missing studies as mirror images of observed studies . Furthermore, most of the included clinical studies were at high risk of non-publication bias, including confusion and selection bias. Indeed, most of those genetic biomarker are in fact prognostic biomarkers of drug adverse reactions in a treated population. Moreover, we did not study the potential effect of subpopulations on those pharmacogenetics associations . Furthermore, we used the genotype’s definition reported in the published systematic review, even when rs number were not available. Last but not least, a publication bias for other associations might exist and remains invisible if most of the corresponding studies are not published at all. Our study showed that,—as in clinical pharmacology or in genetics —, publication is selective in pharmacogenetics. We showed that this publication bias may affect the assessment of an association between genetic biomarker and ADR, even for consensual pharmacogenetic biomarkers of antiseizure drug’s AE as HLA-B*15:02. Our study showed the need for publishing or at least registering any pharmacogenetics study, even with inconclusive results, to fight the issue of publication bias and its harmful consequences. The increased access to genomic databases and to genomic data through next generation sequencing strengthens the importance to anticipate such issues. Both the genomic area and the drug safety assessment are prone to a high risk of false positive results. Publication bias may contribute to the canonisation of such false positive associations. This may lead to not prescribing efficient drugs for false reasons, and to insufficient control of epilepsy. Moreover, false positive results dilute the true safety signal. Taking into account the publication bias is needed for correctly estimating the personalized benefit risk—balance of antiseizure drugs. Complementary to the recent “Strengthening the Reporting Of Pharmacogenetic Studies: Development of the STROPS guideline” , publication of negative pharmacogenetic studies is required. S1 File (DOCX) Click here for additional data file. S1 Data (7Z) Click here for additional data file. |
Patient-Generated Collections for Organizing Electronic Health Record Data to Elevate Personal Meaning, Improve Actionability, and Support Patient–Health Care Provider Communication: Think-Aloud Evaluation Study | 41948a89-cb49-49d4-aea8-df1c8a447c78 | 11833264 | Health Communication[mh] | Background Digital technologies play a pivotal role in facilitating patient-centered care that focuses on understanding patients’ needs and fostering shared decision-making with their health care providers . However, this approach relies heavily on well-informed patients and on effective patient-provider communication . In the United States, a major step toward meeting these requirements was enabling patients to access their medical records from multiple providers through third-party applications . While this is a significant achievement, enabling patients to engage in making sense of these data and turn the insights they gain from this process into actionable steps remains a challenge. Consequently, many patients lack a satisfactory understanding of their health, feel discouraged to self-advocate, and have mediocre communication with their providers, which is at odds with the core values of patient-centered care. Therefore, addressing the challenges around sensemaking and the usability of health data will be important to advancing patient-centered care and empowering patients to take an active role in their health journey. Making sense of data, or sensemaking , is a cyclic process that involves cognitive activities for answering complex questions . These activities involve repeated access to artifacts, identifying relevant information, finding information relationships, and presenting the answers in an understandable format . Patients face a plethora of sensemaking challenges to manage their health. They need to assemble health information from different providers and identify outliers, correlations, and trends to become educated on health topics, drive decision-making, and formulate discussion points with health care providers . Unfortunately, robust platforms to support patients in making sense of their clinical data are lacking. Several commercial mobile apps such as Apple Health Records , iBlueButton , OneRecord , and 1upHealth , along with the academic web application Discovery , advance patient sensemaking by offering data visualizations and specialized views for data exploration. These views help patients uncover interesting patterns related to prevalence, periodicity, co-occurrence, and pre-post analysis of medical events. Typically, these solutions organize electronic health record (EHR) data by type, time stamp, or provider. However, such approaches provide very little support for patients in finding deeper connections between their medical records that are required for understanding health issues, reflecting on medical history, or preparing for clinical encounters. Moreover, these apps do not allow patients to annotate their medical records or save their sensemaking progress, forcing them to remember findings or record them elsewhere. This necessitates patients to revisit the same data repeatedly to refresh inferences and recreate mental notes. Such work is typically tedious and frustrating, leading to anxiety and missing crucial information. Consequently, patients can form skewed health impressions, resulting in poor decisions or risky actions. In clinical visits, the inability to communicate the sensemaking insights to physicians may hinder optimal treatment, leading to repeated tests or medical errors. To address these limitations, we explored an alternative solution that organizes EHR data into collections based on health issues and ongoing problems . Inspired by findings that a problem-based view of EHR data improves clinician awareness, prioritization, and decision-making in the intensive care unit , we adapted a similar approach for patients anchored in the data-frame sensemaking theory . Data frames (ie, structured mental models) or collections of health data, as we refer to them in our previous work , systematically break down problems and help in answering complex questions. These capabilities of the collections hold significant potential for managing health data effectively for all sorts of patients, particularly those with complex medical histories or multiple comorbidities . By organizing abundant data around health issues, collections help patients avoid fragmented health impressions, a common challenge for those with multiple comorbidities. Patients who see multiple specialists can use collections to track the development of specific issues and share insights across providers, raising awareness and improving care coordination. More precisely, our proposed concept of collections allows patients to dynamically organize, adapt, and explore their health information based on evolving needs and available data. For example, a patient managing cardiovascular issues might create a Blood Pressure collection to consolidate related records, which could later branch into more specific collections such as Extreme Blood Pressures or Blood Pressure Lab Work. These refined groupings help uncover patterns and dependencies among factors such as BMI, diet, cholesterol levels, and blood pressure, enhancing understanding and facilitating proactive management of health conditions. Gathering insights from the collections may also help patients have more productive discussions with their providers. In our study, we extended the capacity to transform EHR data into collections and facilitated reasoning regarding them. As patients’ sensemaking of health data is driven by finding outliers, correlations, and trends , we enabled capabilities to identify data patterns within the collections. In addition, we supported the assembly of relevant medical records for the collections by helping patients visually explore, find temporal patterns , and make sense of their EHR data within a single, context-preserving view . Acknowledging patient requests for automation , we offered manually assembled and system-assembled collections. We also allowed for personal data input through free-text notes, fulfilling previously identified patient needs . A notable advancement in our work is moving from mock-ups to a fully functional mobile app, Discovery. The app provides patients with a realistic platform to interact with collections, uncovering deeper insights into preferred mechanisms for creating, refining, and using these collections. Moreover, we intended to motivate patients to see collections as tools for self-advocacy during clinical visits, which was identified as a key use case in our earlier research . Objectives In this study, we used the Discovery app to explore patients’ needs, preferences, and desired interactions for organizing EHR data and delve into potential use cases. More concretely, we asked the following research questions (RQs): What are the needs and feature preferences for organizing EHR data from multiple providers? (RQ 1) What are the patients’ experiences with creating and building collections? How effectively does our app allow patients to organize their EHR data using the concept of collections? How can we better meet patients’ needs for meaningful EHR data organization? What purpose would organizing the EHR data in collections have? (RQ 2) To answer these RQs, we conducted a qualitative evaluation study with 14 participants. Digital technologies play a pivotal role in facilitating patient-centered care that focuses on understanding patients’ needs and fostering shared decision-making with their health care providers . However, this approach relies heavily on well-informed patients and on effective patient-provider communication . In the United States, a major step toward meeting these requirements was enabling patients to access their medical records from multiple providers through third-party applications . While this is a significant achievement, enabling patients to engage in making sense of these data and turn the insights they gain from this process into actionable steps remains a challenge. Consequently, many patients lack a satisfactory understanding of their health, feel discouraged to self-advocate, and have mediocre communication with their providers, which is at odds with the core values of patient-centered care. Therefore, addressing the challenges around sensemaking and the usability of health data will be important to advancing patient-centered care and empowering patients to take an active role in their health journey. Making sense of data, or sensemaking , is a cyclic process that involves cognitive activities for answering complex questions . These activities involve repeated access to artifacts, identifying relevant information, finding information relationships, and presenting the answers in an understandable format . Patients face a plethora of sensemaking challenges to manage their health. They need to assemble health information from different providers and identify outliers, correlations, and trends to become educated on health topics, drive decision-making, and formulate discussion points with health care providers . Unfortunately, robust platforms to support patients in making sense of their clinical data are lacking. Several commercial mobile apps such as Apple Health Records , iBlueButton , OneRecord , and 1upHealth , along with the academic web application Discovery , advance patient sensemaking by offering data visualizations and specialized views for data exploration. These views help patients uncover interesting patterns related to prevalence, periodicity, co-occurrence, and pre-post analysis of medical events. Typically, these solutions organize electronic health record (EHR) data by type, time stamp, or provider. However, such approaches provide very little support for patients in finding deeper connections between their medical records that are required for understanding health issues, reflecting on medical history, or preparing for clinical encounters. Moreover, these apps do not allow patients to annotate their medical records or save their sensemaking progress, forcing them to remember findings or record them elsewhere. This necessitates patients to revisit the same data repeatedly to refresh inferences and recreate mental notes. Such work is typically tedious and frustrating, leading to anxiety and missing crucial information. Consequently, patients can form skewed health impressions, resulting in poor decisions or risky actions. In clinical visits, the inability to communicate the sensemaking insights to physicians may hinder optimal treatment, leading to repeated tests or medical errors. To address these limitations, we explored an alternative solution that organizes EHR data into collections based on health issues and ongoing problems . Inspired by findings that a problem-based view of EHR data improves clinician awareness, prioritization, and decision-making in the intensive care unit , we adapted a similar approach for patients anchored in the data-frame sensemaking theory . Data frames (ie, structured mental models) or collections of health data, as we refer to them in our previous work , systematically break down problems and help in answering complex questions. These capabilities of the collections hold significant potential for managing health data effectively for all sorts of patients, particularly those with complex medical histories or multiple comorbidities . By organizing abundant data around health issues, collections help patients avoid fragmented health impressions, a common challenge for those with multiple comorbidities. Patients who see multiple specialists can use collections to track the development of specific issues and share insights across providers, raising awareness and improving care coordination. More precisely, our proposed concept of collections allows patients to dynamically organize, adapt, and explore their health information based on evolving needs and available data. For example, a patient managing cardiovascular issues might create a Blood Pressure collection to consolidate related records, which could later branch into more specific collections such as Extreme Blood Pressures or Blood Pressure Lab Work. These refined groupings help uncover patterns and dependencies among factors such as BMI, diet, cholesterol levels, and blood pressure, enhancing understanding and facilitating proactive management of health conditions. Gathering insights from the collections may also help patients have more productive discussions with their providers. In our study, we extended the capacity to transform EHR data into collections and facilitated reasoning regarding them. As patients’ sensemaking of health data is driven by finding outliers, correlations, and trends , we enabled capabilities to identify data patterns within the collections. In addition, we supported the assembly of relevant medical records for the collections by helping patients visually explore, find temporal patterns , and make sense of their EHR data within a single, context-preserving view . Acknowledging patient requests for automation , we offered manually assembled and system-assembled collections. We also allowed for personal data input through free-text notes, fulfilling previously identified patient needs . A notable advancement in our work is moving from mock-ups to a fully functional mobile app, Discovery. The app provides patients with a realistic platform to interact with collections, uncovering deeper insights into preferred mechanisms for creating, refining, and using these collections. Moreover, we intended to motivate patients to see collections as tools for self-advocacy during clinical visits, which was identified as a key use case in our earlier research . In this study, we used the Discovery app to explore patients’ needs, preferences, and desired interactions for organizing EHR data and delve into potential use cases. More concretely, we asked the following research questions (RQs): What are the needs and feature preferences for organizing EHR data from multiple providers? (RQ 1) What are the patients’ experiences with creating and building collections? How effectively does our app allow patients to organize their EHR data using the concept of collections? How can we better meet patients’ needs for meaningful EHR data organization? What purpose would organizing the EHR data in collections have? (RQ 2) To answer these RQs, we conducted a qualitative evaluation study with 14 participants. Description of Discovery General Overview Discovery is a noncommercial iPhone app designed to help patients make sense of their EHR data. It introduces a novel concept that organizes EHR data into personalized, problem-based collections. In addition, it allows patients to add their own observations and insights, providing context and complementing their medical records with patient-generated data (PGD). For this study, the app was restricted to accessing synthetic EHR data through a Fast Healthcare Interoperability Resources (FHIR) format from the SMART Health IT repository . Discovery accesses only structured EHR data and relies on a 2-level hierarchy. At the highest level, there are the record categories (“Conditions,” “Immunizations,” and “Vital Signs”). Each record category has multiple record types (“Vital Signs: Body Height,” “Body Weight,” and “Blood Pressure”), and each record type can have multiple instances (“Blood Pressure: systolic: 125, diastolic: 90,” and “date: 02/01/2022”). Each instance of data in our app is called a record and corresponds to an FHIR resource with standardized attributes. Organizing Records in Collections The manually created collections are called Builds (creation shown in A and list shown in B), whereas the app-assembled collections are called Updates ( C). Patients can manually create a new collection ( A) by entering a name and additional metadata. To prioritize and distinguish the collections, we introduced descriptors: purpose, tags, and priority. For example, in A, the patient created a High Blood Pressure collection with the purpose Clinical Visit for an upcoming appointment. They also added tags such as hpb, sweating, and shortness of breath for reference. The priority descriptor indicates that the collection addresses a current and urgent issue. Patients can modify any of this information as the collection develops and changes. As patients repeatedly create collections, their EHR data transform into a problem-based list, as shown in B, which includes health issues such as back discomfort, hamstring tightness, and high blood pressure. Collections are displayed with their name, creation and last modification dates, record count, number of patient-added notes, and labels for current (green clock) and urgent (danger sign) issues. An information button provides a summary with explanations. For example, the High Blood Pressure collection was created on December 31, 2021, and last modified on January 31, 2022. An information panel summarizes its contents: 9 records, 6 notes (4 for the collection and 2 for individual records), and tags for detailed search (hpb, sweating, and shortness of breath). Updates automatically familiarize patients with the latest relevant events without requiring any action on their part ( C). The app scans all records and matches predefined templates, such as recent encounters, laboratory test results, and vital signs. In C, 3 Updates are listed—Last Encounters, Last Lab Results, and Last Vital Signs—based on the last 5 dates when corresponding records were logged by the provider. The list entry follows the same structure as that of the Builds without the labels for currency and urgency. Although records in an Update cannot be changed, patients can add and remove notes. Patients can also clone an Update into a new Build and rename it if needed, allowing them to reuse and customize the assembled records for specific purposes. The reason for having this distinction between Updates and Builds is to delineate what the system and the user have produced and which entity is responsible for the collection that might have led to certain actions. Identifying Relevant Records and Patterns for Collections Discovery offers an interactive visualization to find patterns within records. The Timeline depicts record counts in equal time intervals, showing the prevalence of medical events. A dotted horizontal line marks the threshold above which the volume of records is considered abnormal, corresponding to the mean record count per interval. Gray triangle glyphs indicate values between the mean and 2 SDs, whereas red triangles highlight values of >2 SDs. By highlighting individual records or entire record types, patients can explore patterns that may be saved in existing collections or trigger the creation of new collections. For example, A shows periodicity of influenza shots (when and how frequently influenza shots were administered), B shows the co-occurrence of high blood pressure and high BMI, and C shows the absence of respiratory conditions after an amoxicillin prescription. The FHIR resources (ie, medical records) are represented with Record Cards, which display the clinical information in human-readable format. Patients can use a Filter and Date Picker to narrow down record categories and time frames for displayed records. Selected record categories appear in a Sliding Tabs control, allowing patients to swipe left or right for immediate access. Clicking on a record category in the Sliding Tabs organizes it by record type, represented with Accordions (eg, the Accordion for Immunizations category will have 3 sections for the Flu shot, Tdap, and Zooster record types). Patients can expand the Accordion sections to scroll through individual Record Cards. For example, A shows 8 Record Cards for the influenza shot under the Immunization Accordion section. Accessing and revisiting records involves swiping the Sliding Tabs and selecting record categories, with Accordions retaining their expanded or collapsed state and scrolling position. This method is more context preserving compared to existing solutions, which require repeated back-and-forth navigation through different views for each record category and record type. Producing Insights for the Collection Patients can save individual records or entire record types by tapping the bookmark icon in the corresponding Record Card or Accordion section. A demonstrates adding individual Blood Pressure records to a collection using a Record Card. Records can be removed from collections by tapping the selected bookmark again. To identify data relationships and produce insights, patients can inspect a collection in the Collection Review ( B). Here, records can be viewed by type, date recorded in the provider’s EHR system, or time added to the collection. When sequence matters, records can be ordered chronologically. Patients can also remove records in the Collection Review by tapping the selected bookmark. For example, in B, the patient views records grouped by recording dates in descending order, with yellow bars indicating record counts by date. They observe high blood pressure and high BMI co-occurring 3 times, suggesting a pattern that may need further investigation or discussion with a physician. Supporting PGD Patients can enter notes for a collection ( A) or individual records within it ( B) to add personal insights, progress, observations, details, or disease journal entries. Notes can be modified or removed at any time. In A, the first note provides context on past high blood pressure experiences. The second note serves as a reminder to mention occasional high blood pressure to the general practitioner. The currently created note adds context about noticing changes in blood pressure after stopping regular workouts. In B, the patient contextualizes a high blood pressure measurement taken during a stressful period at work and notes the recurrence of high values. Searching the Collections Free-text search targets the collection name, purpose, and notes ( C), as well as tags and priority. Results dynamically update as the search query is constructed. Study Design Participants We recruited 14 participants from our email list compiled from previous recruiting efforts and Craigslist. This number is sufficient to uncover usability issues and provide rich findings as per current design research practices and literature on user feedback quality . We balanced the sample by age, gender, and medical history (including healthy individuals, those with acute episodes, and those with chronic illnesses). Eligibility criteria included adults fluent in English; possessing an iPhone (iPhone 6 or above) and a laptop or desktop computer (screen size of 13“ or more) with a stable, fast internet connection for both devices; and with normal or corrected vision, no color blindness, and medical records from one or more providers. Medical history was self-reported. illustrates the detailed participant demographics collected using the questionnaire from Table S1 in . Our 14 adult study participants included 10 (71%) women and 4 (29%) men aged 24 to 61 years (mean age 35.6, SD 12.6; median 30.5 years). All had some college experience: half (7/14, 50%) held bachelor’s degrees, 14% (2/14) had some graduate experience, and 14% (2/14) had completed master’s degrees. Participants had between 2 and 15 health care providers, with half (7/14, 50%) having ≥6. The 29% (4/14) of the participants who were healthy saw physicians a few times a year. Those with chronic illnesses had been managing their diseases for 1 to 20 years. All participants were comfortable with daily technology use, and 21% (3/14) had work experience in data analytics. Most (11/14, 79%) used third-party apps to track mental health, weight loss, sleep, and exercise. All but 1 (13/14, 93%) used provider-patient portals to review test and laboratory results, refill prescriptions, and schedule appointments. However, participants found it cumbersome to remember multiple passwords and difficult to find specific information due to interface issues. Data sharing among providers was often slow or impossible, forcing participants to print and assemble records for clinical visits. Procedures Inspired by existing patient portal usability studies, the study procedures were tailored to our RQs and are presented in . The study was conducted remotely via Zoom (Zoom Video Communications) meetings. Participants downloaded our app on their iPhone using Apple TestFlight for beta testing and mirrored their iPhone screen on their laptop using AirServer (App Dynamic ehf.) . The laptop screen was then shared with the researcher for observation. Participants attended an initial 60-minute session (session 1) and a follow-up 45-minute session (session 2). Both sessions used the concurrent think-aloud protocol to gather rich qualitative data on user needs, perceptions, and preferences efficiently . illustrates 2 example topics and detailed tasks for creating a toy collection and realistic collection from sessions 1 and 2, respectively. The full set of 13 topics for both sessions and their detailed tasks that follow a similar structure are shown in Tables S2 and S3 in . In session 1, the researcher first collected demographics and digital health consumer information from participants (Table S1 in ). The session then evaluated the usability of the app (RQ 1). Participants were introduced to the app’s key concepts and features and were then given tasks to learn the interactions for creating and assembling relevant records in collections. After each task block, participants provided feedback. Tasks included creating a toy collection, adding descriptors, exploring scoping mechanisms for record categories and time ranges, navigating records using the Sliding Tabs, and inspecting the interactive Timeline visualization. Participants also completed tasks related to finding periodicity, co-occurrence, and pretest-posttest analysis of medical events using the Timeline; saving records to a collection; and reviewing it. The session concluded with feedback on the intuitiveness, usability, and usefulness of creating, building, and reviewing collections, as well as data exploration and pattern detection features . Session 2 focused on the usefulness of collections in real-life settings (RQ 2). Participants performed tasks involving creating a collection tied to a specific issue (eg, high blood pressure) and purpose (eg, clinical visit). After each task block, participants provided feedback. Tasks included creating a High Blood Pressure collection, adding descriptors for a clinical visit, selecting records related to high blood pressure, identifying patterns in selected records, adding notes to the collection and specific records for clinical context, and using the search feature to find other collections. The session concluded with feedback on the usefulness of collections, brainstorming real-life use case scenarios, and suggesting potential improvements . Example tasks from sessions 1 and 2. Session 1: creating a collection (5 min) For the purposes of learning the basic mechanics around building a Collection from scratch, first Create a new Collection and name it “Toy Collection.” Now, let’s add some descriptors about the Collection. Please add a purpose to the Collection, something related to learning about this app. Now, add a couple of tags to further describe, summarize or annotate the Collection for future quick access. Finally, specify the priority of this Collection by marking it as urgent. What was your experience with creating the new Collection? How intuitive was it? Have you seen similar interactions elsewhere? How useful do you find the descriptors for the Collection? Now, let’s go back to the list of Collections. Let me know how can you see the details about the Collection you just created? How intuitive was it? What are possible improvements? Session 2: building a collection (15 min) You will now create a Collection that is more realistic and meaningful for use in a real-life scenario. We will assume that you are preparing for an upcoming visit to your physician’s office related to potential issues with high blood pressure. Create a collection called “High Blood Pressure.” Add the purpose for the Collection Add a few tags Mark its priority Add the Records with blood pressure with systolic value over 120 and BMI over 30. What was your experience with assembling the Records for the Collection? How laborious was it? What are some ways in which we can make this assembling process more efficient? How do you feel about having the system prepopulate the Collection for you and let you modify it afterwards? Feedback collected at the end of the study sessions. Session 1 example tasks What was your impression of this app? What did you like? What did you dislike? How intuitive was the app? How easy or hard was it to explore the data? How useful were the features in the app to identify patterns in the data? How did you like the mechanism for saving Records in the Collection? What are some improvements you would like to see? Session 2 example tasks How useful do you think the Collections can be for you? What are some use cases for the Collections that you can think of? What are some improvements you would like to see for the Collections? Automatic support for building Collections? Automatically finding data patterns in the Collections? Patient-generated data? Data Collection We recorded the Zoom meetings for audio and video capture of the entire interaction. The first author also took notes during the meetings. Data Analysis Audio recordings were transcribed using Rev (Rev.com, Inc). We analyzed video recordings for additional context and a deeper understanding of participant comments during the think-aloud protocol. Video annotations were added to the transcripts and session notes for reflexive thematic analysis . The first author began by open coding the textual data. Emerging categories were reconciled in meetings with the second and last authors to identify use cases and detailed approaches to organizing and annotating EHR data. These themes were validated in a group meeting with researchers unfamiliar with the collection concept and with our app. Ethical Considerations The Harvard Faculty of Medicine Institutional Review Board approved this study (protocol IRB20-1757). Participants signed a consent form, which also allowed them to opt out of the study at any point. Each participant received a US $40 Amazon gift card as compensation. The data obtained from the study sessions did not include any identifiable information about the participants and were stored on a password-protected computer with encryption. Only the research team had access to these data. General Overview Discovery is a noncommercial iPhone app designed to help patients make sense of their EHR data. It introduces a novel concept that organizes EHR data into personalized, problem-based collections. In addition, it allows patients to add their own observations and insights, providing context and complementing their medical records with patient-generated data (PGD). For this study, the app was restricted to accessing synthetic EHR data through a Fast Healthcare Interoperability Resources (FHIR) format from the SMART Health IT repository . Discovery accesses only structured EHR data and relies on a 2-level hierarchy. At the highest level, there are the record categories (“Conditions,” “Immunizations,” and “Vital Signs”). Each record category has multiple record types (“Vital Signs: Body Height,” “Body Weight,” and “Blood Pressure”), and each record type can have multiple instances (“Blood Pressure: systolic: 125, diastolic: 90,” and “date: 02/01/2022”). Each instance of data in our app is called a record and corresponds to an FHIR resource with standardized attributes. Organizing Records in Collections The manually created collections are called Builds (creation shown in A and list shown in B), whereas the app-assembled collections are called Updates ( C). Patients can manually create a new collection ( A) by entering a name and additional metadata. To prioritize and distinguish the collections, we introduced descriptors: purpose, tags, and priority. For example, in A, the patient created a High Blood Pressure collection with the purpose Clinical Visit for an upcoming appointment. They also added tags such as hpb, sweating, and shortness of breath for reference. The priority descriptor indicates that the collection addresses a current and urgent issue. Patients can modify any of this information as the collection develops and changes. As patients repeatedly create collections, their EHR data transform into a problem-based list, as shown in B, which includes health issues such as back discomfort, hamstring tightness, and high blood pressure. Collections are displayed with their name, creation and last modification dates, record count, number of patient-added notes, and labels for current (green clock) and urgent (danger sign) issues. An information button provides a summary with explanations. For example, the High Blood Pressure collection was created on December 31, 2021, and last modified on January 31, 2022. An information panel summarizes its contents: 9 records, 6 notes (4 for the collection and 2 for individual records), and tags for detailed search (hpb, sweating, and shortness of breath). Updates automatically familiarize patients with the latest relevant events without requiring any action on their part ( C). The app scans all records and matches predefined templates, such as recent encounters, laboratory test results, and vital signs. In C, 3 Updates are listed—Last Encounters, Last Lab Results, and Last Vital Signs—based on the last 5 dates when corresponding records were logged by the provider. The list entry follows the same structure as that of the Builds without the labels for currency and urgency. Although records in an Update cannot be changed, patients can add and remove notes. Patients can also clone an Update into a new Build and rename it if needed, allowing them to reuse and customize the assembled records for specific purposes. The reason for having this distinction between Updates and Builds is to delineate what the system and the user have produced and which entity is responsible for the collection that might have led to certain actions. Identifying Relevant Records and Patterns for Collections Discovery offers an interactive visualization to find patterns within records. The Timeline depicts record counts in equal time intervals, showing the prevalence of medical events. A dotted horizontal line marks the threshold above which the volume of records is considered abnormal, corresponding to the mean record count per interval. Gray triangle glyphs indicate values between the mean and 2 SDs, whereas red triangles highlight values of >2 SDs. By highlighting individual records or entire record types, patients can explore patterns that may be saved in existing collections or trigger the creation of new collections. For example, A shows periodicity of influenza shots (when and how frequently influenza shots were administered), B shows the co-occurrence of high blood pressure and high BMI, and C shows the absence of respiratory conditions after an amoxicillin prescription. The FHIR resources (ie, medical records) are represented with Record Cards, which display the clinical information in human-readable format. Patients can use a Filter and Date Picker to narrow down record categories and time frames for displayed records. Selected record categories appear in a Sliding Tabs control, allowing patients to swipe left or right for immediate access. Clicking on a record category in the Sliding Tabs organizes it by record type, represented with Accordions (eg, the Accordion for Immunizations category will have 3 sections for the Flu shot, Tdap, and Zooster record types). Patients can expand the Accordion sections to scroll through individual Record Cards. For example, A shows 8 Record Cards for the influenza shot under the Immunization Accordion section. Accessing and revisiting records involves swiping the Sliding Tabs and selecting record categories, with Accordions retaining their expanded or collapsed state and scrolling position. This method is more context preserving compared to existing solutions, which require repeated back-and-forth navigation through different views for each record category and record type. Producing Insights for the Collection Patients can save individual records or entire record types by tapping the bookmark icon in the corresponding Record Card or Accordion section. A demonstrates adding individual Blood Pressure records to a collection using a Record Card. Records can be removed from collections by tapping the selected bookmark again. To identify data relationships and produce insights, patients can inspect a collection in the Collection Review ( B). Here, records can be viewed by type, date recorded in the provider’s EHR system, or time added to the collection. When sequence matters, records can be ordered chronologically. Patients can also remove records in the Collection Review by tapping the selected bookmark. For example, in B, the patient views records grouped by recording dates in descending order, with yellow bars indicating record counts by date. They observe high blood pressure and high BMI co-occurring 3 times, suggesting a pattern that may need further investigation or discussion with a physician. Supporting PGD Patients can enter notes for a collection ( A) or individual records within it ( B) to add personal insights, progress, observations, details, or disease journal entries. Notes can be modified or removed at any time. In A, the first note provides context on past high blood pressure experiences. The second note serves as a reminder to mention occasional high blood pressure to the general practitioner. The currently created note adds context about noticing changes in blood pressure after stopping regular workouts. In B, the patient contextualizes a high blood pressure measurement taken during a stressful period at work and notes the recurrence of high values. Searching the Collections Free-text search targets the collection name, purpose, and notes ( C), as well as tags and priority. Results dynamically update as the search query is constructed. Discovery is a noncommercial iPhone app designed to help patients make sense of their EHR data. It introduces a novel concept that organizes EHR data into personalized, problem-based collections. In addition, it allows patients to add their own observations and insights, providing context and complementing their medical records with patient-generated data (PGD). For this study, the app was restricted to accessing synthetic EHR data through a Fast Healthcare Interoperability Resources (FHIR) format from the SMART Health IT repository . Discovery accesses only structured EHR data and relies on a 2-level hierarchy. At the highest level, there are the record categories (“Conditions,” “Immunizations,” and “Vital Signs”). Each record category has multiple record types (“Vital Signs: Body Height,” “Body Weight,” and “Blood Pressure”), and each record type can have multiple instances (“Blood Pressure: systolic: 125, diastolic: 90,” and “date: 02/01/2022”). Each instance of data in our app is called a record and corresponds to an FHIR resource with standardized attributes. The manually created collections are called Builds (creation shown in A and list shown in B), whereas the app-assembled collections are called Updates ( C). Patients can manually create a new collection ( A) by entering a name and additional metadata. To prioritize and distinguish the collections, we introduced descriptors: purpose, tags, and priority. For example, in A, the patient created a High Blood Pressure collection with the purpose Clinical Visit for an upcoming appointment. They also added tags such as hpb, sweating, and shortness of breath for reference. The priority descriptor indicates that the collection addresses a current and urgent issue. Patients can modify any of this information as the collection develops and changes. As patients repeatedly create collections, their EHR data transform into a problem-based list, as shown in B, which includes health issues such as back discomfort, hamstring tightness, and high blood pressure. Collections are displayed with their name, creation and last modification dates, record count, number of patient-added notes, and labels for current (green clock) and urgent (danger sign) issues. An information button provides a summary with explanations. For example, the High Blood Pressure collection was created on December 31, 2021, and last modified on January 31, 2022. An information panel summarizes its contents: 9 records, 6 notes (4 for the collection and 2 for individual records), and tags for detailed search (hpb, sweating, and shortness of breath). Updates automatically familiarize patients with the latest relevant events without requiring any action on their part ( C). The app scans all records and matches predefined templates, such as recent encounters, laboratory test results, and vital signs. In C, 3 Updates are listed—Last Encounters, Last Lab Results, and Last Vital Signs—based on the last 5 dates when corresponding records were logged by the provider. The list entry follows the same structure as that of the Builds without the labels for currency and urgency. Although records in an Update cannot be changed, patients can add and remove notes. Patients can also clone an Update into a new Build and rename it if needed, allowing them to reuse and customize the assembled records for specific purposes. The reason for having this distinction between Updates and Builds is to delineate what the system and the user have produced and which entity is responsible for the collection that might have led to certain actions. Discovery offers an interactive visualization to find patterns within records. The Timeline depicts record counts in equal time intervals, showing the prevalence of medical events. A dotted horizontal line marks the threshold above which the volume of records is considered abnormal, corresponding to the mean record count per interval. Gray triangle glyphs indicate values between the mean and 2 SDs, whereas red triangles highlight values of >2 SDs. By highlighting individual records or entire record types, patients can explore patterns that may be saved in existing collections or trigger the creation of new collections. For example, A shows periodicity of influenza shots (when and how frequently influenza shots were administered), B shows the co-occurrence of high blood pressure and high BMI, and C shows the absence of respiratory conditions after an amoxicillin prescription. The FHIR resources (ie, medical records) are represented with Record Cards, which display the clinical information in human-readable format. Patients can use a Filter and Date Picker to narrow down record categories and time frames for displayed records. Selected record categories appear in a Sliding Tabs control, allowing patients to swipe left or right for immediate access. Clicking on a record category in the Sliding Tabs organizes it by record type, represented with Accordions (eg, the Accordion for Immunizations category will have 3 sections for the Flu shot, Tdap, and Zooster record types). Patients can expand the Accordion sections to scroll through individual Record Cards. For example, A shows 8 Record Cards for the influenza shot under the Immunization Accordion section. Accessing and revisiting records involves swiping the Sliding Tabs and selecting record categories, with Accordions retaining their expanded or collapsed state and scrolling position. This method is more context preserving compared to existing solutions, which require repeated back-and-forth navigation through different views for each record category and record type. Patients can save individual records or entire record types by tapping the bookmark icon in the corresponding Record Card or Accordion section. A demonstrates adding individual Blood Pressure records to a collection using a Record Card. Records can be removed from collections by tapping the selected bookmark again. To identify data relationships and produce insights, patients can inspect a collection in the Collection Review ( B). Here, records can be viewed by type, date recorded in the provider’s EHR system, or time added to the collection. When sequence matters, records can be ordered chronologically. Patients can also remove records in the Collection Review by tapping the selected bookmark. For example, in B, the patient views records grouped by recording dates in descending order, with yellow bars indicating record counts by date. They observe high blood pressure and high BMI co-occurring 3 times, suggesting a pattern that may need further investigation or discussion with a physician. Patients can enter notes for a collection ( A) or individual records within it ( B) to add personal insights, progress, observations, details, or disease journal entries. Notes can be modified or removed at any time. In A, the first note provides context on past high blood pressure experiences. The second note serves as a reminder to mention occasional high blood pressure to the general practitioner. The currently created note adds context about noticing changes in blood pressure after stopping regular workouts. In B, the patient contextualizes a high blood pressure measurement taken during a stressful period at work and notes the recurrence of high values. Free-text search targets the collection name, purpose, and notes ( C), as well as tags and priority. Results dynamically update as the search query is constructed. Participants We recruited 14 participants from our email list compiled from previous recruiting efforts and Craigslist. This number is sufficient to uncover usability issues and provide rich findings as per current design research practices and literature on user feedback quality . We balanced the sample by age, gender, and medical history (including healthy individuals, those with acute episodes, and those with chronic illnesses). Eligibility criteria included adults fluent in English; possessing an iPhone (iPhone 6 or above) and a laptop or desktop computer (screen size of 13“ or more) with a stable, fast internet connection for both devices; and with normal or corrected vision, no color blindness, and medical records from one or more providers. Medical history was self-reported. illustrates the detailed participant demographics collected using the questionnaire from Table S1 in . Our 14 adult study participants included 10 (71%) women and 4 (29%) men aged 24 to 61 years (mean age 35.6, SD 12.6; median 30.5 years). All had some college experience: half (7/14, 50%) held bachelor’s degrees, 14% (2/14) had some graduate experience, and 14% (2/14) had completed master’s degrees. Participants had between 2 and 15 health care providers, with half (7/14, 50%) having ≥6. The 29% (4/14) of the participants who were healthy saw physicians a few times a year. Those with chronic illnesses had been managing their diseases for 1 to 20 years. All participants were comfortable with daily technology use, and 21% (3/14) had work experience in data analytics. Most (11/14, 79%) used third-party apps to track mental health, weight loss, sleep, and exercise. All but 1 (13/14, 93%) used provider-patient portals to review test and laboratory results, refill prescriptions, and schedule appointments. However, participants found it cumbersome to remember multiple passwords and difficult to find specific information due to interface issues. Data sharing among providers was often slow or impossible, forcing participants to print and assemble records for clinical visits. Procedures Inspired by existing patient portal usability studies, the study procedures were tailored to our RQs and are presented in . The study was conducted remotely via Zoom (Zoom Video Communications) meetings. Participants downloaded our app on their iPhone using Apple TestFlight for beta testing and mirrored their iPhone screen on their laptop using AirServer (App Dynamic ehf.) . The laptop screen was then shared with the researcher for observation. Participants attended an initial 60-minute session (session 1) and a follow-up 45-minute session (session 2). Both sessions used the concurrent think-aloud protocol to gather rich qualitative data on user needs, perceptions, and preferences efficiently . illustrates 2 example topics and detailed tasks for creating a toy collection and realistic collection from sessions 1 and 2, respectively. The full set of 13 topics for both sessions and their detailed tasks that follow a similar structure are shown in Tables S2 and S3 in . In session 1, the researcher first collected demographics and digital health consumer information from participants (Table S1 in ). The session then evaluated the usability of the app (RQ 1). Participants were introduced to the app’s key concepts and features and were then given tasks to learn the interactions for creating and assembling relevant records in collections. After each task block, participants provided feedback. Tasks included creating a toy collection, adding descriptors, exploring scoping mechanisms for record categories and time ranges, navigating records using the Sliding Tabs, and inspecting the interactive Timeline visualization. Participants also completed tasks related to finding periodicity, co-occurrence, and pretest-posttest analysis of medical events using the Timeline; saving records to a collection; and reviewing it. The session concluded with feedback on the intuitiveness, usability, and usefulness of creating, building, and reviewing collections, as well as data exploration and pattern detection features . Session 2 focused on the usefulness of collections in real-life settings (RQ 2). Participants performed tasks involving creating a collection tied to a specific issue (eg, high blood pressure) and purpose (eg, clinical visit). After each task block, participants provided feedback. Tasks included creating a High Blood Pressure collection, adding descriptors for a clinical visit, selecting records related to high blood pressure, identifying patterns in selected records, adding notes to the collection and specific records for clinical context, and using the search feature to find other collections. The session concluded with feedback on the usefulness of collections, brainstorming real-life use case scenarios, and suggesting potential improvements . Example tasks from sessions 1 and 2. Session 1: creating a collection (5 min) For the purposes of learning the basic mechanics around building a Collection from scratch, first Create a new Collection and name it “Toy Collection.” Now, let’s add some descriptors about the Collection. Please add a purpose to the Collection, something related to learning about this app. Now, add a couple of tags to further describe, summarize or annotate the Collection for future quick access. Finally, specify the priority of this Collection by marking it as urgent. What was your experience with creating the new Collection? How intuitive was it? Have you seen similar interactions elsewhere? How useful do you find the descriptors for the Collection? Now, let’s go back to the list of Collections. Let me know how can you see the details about the Collection you just created? How intuitive was it? What are possible improvements? Session 2: building a collection (15 min) You will now create a Collection that is more realistic and meaningful for use in a real-life scenario. We will assume that you are preparing for an upcoming visit to your physician’s office related to potential issues with high blood pressure. Create a collection called “High Blood Pressure.” Add the purpose for the Collection Add a few tags Mark its priority Add the Records with blood pressure with systolic value over 120 and BMI over 30. What was your experience with assembling the Records for the Collection? How laborious was it? What are some ways in which we can make this assembling process more efficient? How do you feel about having the system prepopulate the Collection for you and let you modify it afterwards? Feedback collected at the end of the study sessions. Session 1 example tasks What was your impression of this app? What did you like? What did you dislike? How intuitive was the app? How easy or hard was it to explore the data? How useful were the features in the app to identify patterns in the data? How did you like the mechanism for saving Records in the Collection? What are some improvements you would like to see? Session 2 example tasks How useful do you think the Collections can be for you? What are some use cases for the Collections that you can think of? What are some improvements you would like to see for the Collections? Automatic support for building Collections? Automatically finding data patterns in the Collections? Patient-generated data? Data Collection We recorded the Zoom meetings for audio and video capture of the entire interaction. The first author also took notes during the meetings. We recruited 14 participants from our email list compiled from previous recruiting efforts and Craigslist. This number is sufficient to uncover usability issues and provide rich findings as per current design research practices and literature on user feedback quality . We balanced the sample by age, gender, and medical history (including healthy individuals, those with acute episodes, and those with chronic illnesses). Eligibility criteria included adults fluent in English; possessing an iPhone (iPhone 6 or above) and a laptop or desktop computer (screen size of 13“ or more) with a stable, fast internet connection for both devices; and with normal or corrected vision, no color blindness, and medical records from one or more providers. Medical history was self-reported. illustrates the detailed participant demographics collected using the questionnaire from Table S1 in . Our 14 adult study participants included 10 (71%) women and 4 (29%) men aged 24 to 61 years (mean age 35.6, SD 12.6; median 30.5 years). All had some college experience: half (7/14, 50%) held bachelor’s degrees, 14% (2/14) had some graduate experience, and 14% (2/14) had completed master’s degrees. Participants had between 2 and 15 health care providers, with half (7/14, 50%) having ≥6. The 29% (4/14) of the participants who were healthy saw physicians a few times a year. Those with chronic illnesses had been managing their diseases for 1 to 20 years. All participants were comfortable with daily technology use, and 21% (3/14) had work experience in data analytics. Most (11/14, 79%) used third-party apps to track mental health, weight loss, sleep, and exercise. All but 1 (13/14, 93%) used provider-patient portals to review test and laboratory results, refill prescriptions, and schedule appointments. However, participants found it cumbersome to remember multiple passwords and difficult to find specific information due to interface issues. Data sharing among providers was often slow or impossible, forcing participants to print and assemble records for clinical visits. Inspired by existing patient portal usability studies, the study procedures were tailored to our RQs and are presented in . The study was conducted remotely via Zoom (Zoom Video Communications) meetings. Participants downloaded our app on their iPhone using Apple TestFlight for beta testing and mirrored their iPhone screen on their laptop using AirServer (App Dynamic ehf.) . The laptop screen was then shared with the researcher for observation. Participants attended an initial 60-minute session (session 1) and a follow-up 45-minute session (session 2). Both sessions used the concurrent think-aloud protocol to gather rich qualitative data on user needs, perceptions, and preferences efficiently . illustrates 2 example topics and detailed tasks for creating a toy collection and realistic collection from sessions 1 and 2, respectively. The full set of 13 topics for both sessions and their detailed tasks that follow a similar structure are shown in Tables S2 and S3 in . In session 1, the researcher first collected demographics and digital health consumer information from participants (Table S1 in ). The session then evaluated the usability of the app (RQ 1). Participants were introduced to the app’s key concepts and features and were then given tasks to learn the interactions for creating and assembling relevant records in collections. After each task block, participants provided feedback. Tasks included creating a toy collection, adding descriptors, exploring scoping mechanisms for record categories and time ranges, navigating records using the Sliding Tabs, and inspecting the interactive Timeline visualization. Participants also completed tasks related to finding periodicity, co-occurrence, and pretest-posttest analysis of medical events using the Timeline; saving records to a collection; and reviewing it. The session concluded with feedback on the intuitiveness, usability, and usefulness of creating, building, and reviewing collections, as well as data exploration and pattern detection features . Session 2 focused on the usefulness of collections in real-life settings (RQ 2). Participants performed tasks involving creating a collection tied to a specific issue (eg, high blood pressure) and purpose (eg, clinical visit). After each task block, participants provided feedback. Tasks included creating a High Blood Pressure collection, adding descriptors for a clinical visit, selecting records related to high blood pressure, identifying patterns in selected records, adding notes to the collection and specific records for clinical context, and using the search feature to find other collections. The session concluded with feedback on the usefulness of collections, brainstorming real-life use case scenarios, and suggesting potential improvements . Example tasks from sessions 1 and 2. Session 1: creating a collection (5 min) For the purposes of learning the basic mechanics around building a Collection from scratch, first Create a new Collection and name it “Toy Collection.” Now, let’s add some descriptors about the Collection. Please add a purpose to the Collection, something related to learning about this app. Now, add a couple of tags to further describe, summarize or annotate the Collection for future quick access. Finally, specify the priority of this Collection by marking it as urgent. What was your experience with creating the new Collection? How intuitive was it? Have you seen similar interactions elsewhere? How useful do you find the descriptors for the Collection? Now, let’s go back to the list of Collections. Let me know how can you see the details about the Collection you just created? How intuitive was it? What are possible improvements? Session 2: building a collection (15 min) You will now create a Collection that is more realistic and meaningful for use in a real-life scenario. We will assume that you are preparing for an upcoming visit to your physician’s office related to potential issues with high blood pressure. Create a collection called “High Blood Pressure.” Add the purpose for the Collection Add a few tags Mark its priority Add the Records with blood pressure with systolic value over 120 and BMI over 30. What was your experience with assembling the Records for the Collection? How laborious was it? What are some ways in which we can make this assembling process more efficient? How do you feel about having the system prepopulate the Collection for you and let you modify it afterwards? Feedback collected at the end of the study sessions. Session 1 example tasks What was your impression of this app? What did you like? What did you dislike? How intuitive was the app? How easy or hard was it to explore the data? How useful were the features in the app to identify patterns in the data? How did you like the mechanism for saving Records in the Collection? What are some improvements you would like to see? Session 2 example tasks How useful do you think the Collections can be for you? What are some use cases for the Collections that you can think of? What are some improvements you would like to see for the Collections? Automatic support for building Collections? Automatically finding data patterns in the Collections? Patient-generated data? We recorded the Zoom meetings for audio and video capture of the entire interaction. The first author also took notes during the meetings. Audio recordings were transcribed using Rev (Rev.com, Inc). We analyzed video recordings for additional context and a deeper understanding of participant comments during the think-aloud protocol. Video annotations were added to the transcripts and session notes for reflexive thematic analysis . The first author began by open coding the textual data. Emerging categories were reconciled in meetings with the second and last authors to identify use cases and detailed approaches to organizing and annotating EHR data. These themes were validated in a group meeting with researchers unfamiliar with the collection concept and with our app. The Harvard Faculty of Medicine Institutional Review Board approved this study (protocol IRB20-1757). Participants signed a consent form, which also allowed them to opt out of the study at any point. Each participant received a US $40 Amazon gift card as compensation. The data obtained from the study sessions did not include any identifiable information about the participants and were stored on a password-protected computer with encryption. Only the research team had access to these data. Overview The results from our study are organized and analyzed around five qualitative themes, two applicable to RQ 2 and three applicable to RQ 1: (1) using collections for personal benefit (RQ 2), (2) using collections in a clinical setting (RQ 2), (3) creating and building collections (RQ 1), (4) enhancing collections (RQ 1), and (5) accessing collections (RQ 1). In the remainder of the Results section, we characterize the participants and report on these themes using 16 quotes from 11 different participants. The participants are labeled as P1 to P14. Purposes for Organizing the EHR Data in Collections Using Collections for Personal Benefit Quick Access to Information Participants perceived the Collections feature as a way to index information at a suitable level of granularity for health issues, topics of interest, or conditions to monitor. They expected that this would give them quick access to current or urgent problems that were being managed: Personally I would like to monitor my asthma because I am using medication for that, the typical inhaler, but I would like to monitor, these days I had certain attacks or shortness of breath, so collections, having it back for that specific condition is very useful to me, because I wouldn’t have to speculate about when my last attack was or when my last appointment date was, it’s right here for me to access. P13 Reflection on Medical History Participants viewed collections as a tool to construct and understand their medical history. They expected collections to serve as a repository for the health issues they faced, aiding in reflecting on the existence, prevalence, and development of their health conditions: This will be able to find what happens when I have, in my case, headaches related to my high blood pressure and nothing else, or if you’re sick for the flu, influenza and you have fever, it’s just related to the influenza, not to the COVID. P1 Keeping Track of Health Status Most participants envisioned using collections to track their health status proactively. This included monitoring urgent issues needing immediate attention, unstable conditions requiring frequent observation, and treatments needing careful monitoring. Participants also wanted collections to track abnormal laboratory test results and vital sign values across various health issues: Well, for me, it’s kind of good [to have medical records organized in collections], especially, for example, blood samples, especially those with high triglycerides or something. Maybe I can collect them and see whether there’s a trend for this month, or for January, I’m high in this one. Then second month, I’m also high, so maybe I can lower it down...For collections, just categorize those. Which are high, which are low. P5 Journaling Daily Events Most of the participants envisioned using collections as a personal diary for coping with diseases and logging measurements and their effects on lifestyle. They wanted to track challenges, successes, and progress toward finding solutions and monitor disease developments: What I’ve done is I’ve taken all of my videos and stuff since February. Like I said, I’ve been to eight different doctors and I’ve shown them, this is the progress of what’s happened from...I’ve had two surgeries before this surgery where they lanced it, cut it, drained it. Nothing happened. The cyst came back and then it went into my bone. So I’m able to bring these photos, I’m able to bring this timeline, I’m able to bring my frustrations and show this doctor within 30 seconds [using a collection], look, this is what it looked like. And this is my own diary, my own history. It’s very important because they’re a doctor. They don’t know me, they don’t know what it looked like on day one. P10 Learning From Past Experiences Nearly all participants wanted to use collections to identify trends and patterns in their ongoing health experiences, including co-occurring symptoms and treatment effects. They also intended to log food intake, sleep, activity, or stress factors to find triggers for symptoms: I think I could definitely use them there. It’s a lot easier now because I could highlight the certain event, and put like my triggers down with it, like migraine on the third was this, you could even put what you took with it. So for me, I would look back and be like, “Oh, I can tell my doctor that, I’ve had 10 migraines. I took a medication with these three. What’s my options.” So I think that would be great, it’s a great tool that I can actually do that with this app. P9 Using Collections in a Clinical Setting Preparation for the Clinical Visit Most participants would use collections to prepare talking points for clinical visits combining personal measurements and notes with medical records from other providers (eg, laboratory and test results). This was crucial because their physicians often did not have access to external data: For collections, I would say that if I’m meeting with multiple providers about one health issue, I could see myself combining all my records there so that multiple providers can see each other’s records...I would probably [use the notes], if I needed to jot down a certain time that I took a measurement, or if my doctor told me keep track of what you ate that day, or kind of anything that I would want to have the details for the next time that I go and see the provider. P6 Relying on Collections During Clinical Visits All participants wanted to share collections with their physicians during clinical visits to establish ground truths, raise awareness of other providers’ information, and provide transparent talking points. They believed that adding PGD to collections could describe what happened between visits and raise their physicians’ awareness: If you have a condition, you need to check on your blood pressure. You need to communicate that with your doctor, so you can add a note [in the collection] saying like, “Latest, highest blood pressure from this week,” from a date. P14 In addition, most participants felt that taking in-visit notes and saving them in a collection could help them understand care plans and take appropriate actions afterward: Maybe [taking notes in the collection for the clinical visit] just for your own personal reference or if you wanted to bring it up later on in another appointment or something, or just maybe, I guess, just general recording of something that happened during that visit. P11 All participants saw physicians as essential partners in reviewing collections and deciding on actions based on their contents. This was primarily because most participants doubted their expertise in determining what should go into collections or which collections to create. While they were very open to include their physicians in the collection curation, some feared that they might overburden physicians with verification inquiries: I guess I would definitely do that [look for patterns in the data and store them in a collection] just because I can actually consult the doctor. Is this actually correlated or will I have to change my diet because it affects this? At least you can ask the doctor, or confirm whether that is true or not. P5 Needs and Feature Preferences for Organizing EHR Data in Collections Creating and Building Collections Manual Workflow for Creating and Building Collections Overall, all participants expressed satisfaction with the clarity and simplicity of the mechanics to create collections and save records in them. However, those with less medical knowledge and disease experience (relatively healthy and recently diagnosed individuals) thought that initiating and building collections was challenging. For them, it was not always clear what issues deserved separate collections, what records to include in a collection due to delicate dependencies, and why they should invest substantial time and effort in assembling collections. In contrast, the more experienced (chronic) patients were relatively confident in their ability to carve a personalized view of their EHR data. However, some did acknowledge that they might not be as exhaustive and reliable in their data organization. Participants quickly mastered using the Sliding Tabs with Accordions and appreciated the context-preserving record exploration. However, most of the participants found it challenging to assess the relevance of individual records due to unexplained clinical language and attribute values. Participants requested explanations, prominent visual cues for abnormal values to attract their attention, and time-series visualizations for additional context and noticing trends. They also wanted to be able to sort by date or attribute value, with filtering capabilities that also included adding the filtered records in bulk to a collection. Some participants desired collections with richer internal structures, recognizing the need to identify subsets of records and their importance within a collection. They suggested linking groups or individual records using specific annotations for easy identification during visual inspection or search: ...maybe if certain records are related to each other. So I would want to mark that. And then maybe just have a way of sorting down based on certain labels. P8 Automatic Support for Building Collections Participants were highly receptive to discussing ideas that could automate building collections. They were interested in seeing their records automatically put into collections based on provenance (provider, physician, hospital, location, and date) and clinical meaning (condition, disease, organ, organ system, and abnormality of the values across records). Some also suggested grouping records into collections based on personal annotations. For all these groupings, they wished to be able to edit the collection manually: If I had neurological problems, neurology collection. If I had urological problems, urology collection. I think that for me at least would seem a more straightforward way to categorize them. But from the categories I’ve already seen, I find those useful. P4 Several participants suggested using a “seed” to automate collection building, such as naming a collection, adding keywords, or including a few records: And I think a lot of patients don’t know where to start, what data to begin with. So if it’s something that’s already preset, they say, “Okay, I’m suffering from depression or I have diabetes.” And the system pulls the different data points that they would need to look at for someone who’s diabetic or someone who’s dealing with depression, I think that’s helpful. Because sometimes, the problem is you don’t know where to start and you don’t know what to look for. P11 Finally, most of the participants wanted to receive automated help to add or remove records for a collection that had already been created. Few described wanting to choose from a list of suggested records based on the existing content of a collection, revealing records and record patterns otherwise invisible to them. Others saw this automated record offering more as an idea generation approach—needing some follow-up validation, including taking it up with their physicians: When I work, I want to listen to some music and then I’m like, okay. I just don’t know what’s next. I need something similar to this, the same vibe, but I just can’t think about that. And then there is suggestions. And yes, some of it’s weird. But maybe like the doctor can also have some help here, and when you review the collections together, they might say, “Hey, listen, this is what the system gave you and that’s great. Let’s remove a few things. I would suggest you add a couple others. And whatever you put there, it’s also fine. And let’s keep it that way.” P6 Enhancing Collections Through Personally Provided Data Making Collections Complete Participants strongly expressed the need to complement their EHR data with daily entries from sensors; self-monitoring devices; and manual measurements of symptoms, treatments, and outcomes in various formats (text, photos, videos, and scanned documents): Actually, I think you can sort of restructure the whole core of the collections on top of two main pillars. The first one would be all of the doctor’s data, which is basically hard data, which allows you to diagnose, allows you to run statistical analysis...That could be part of the core data, but all of the context, maybe I’m getting this shortness of breath in my home, watching my TV, might be added by the notes. You have these two types of data. By adding the user data, would allow me to get context, give context, which is important and will allow me to, on a daily basis, keep a record, which in case of data like shortness of breath, I’m having, I’m not having. Would allow the doctor to have a really unbiased input on symptoms I’m having. P7 Participants wanted to log detailed observations and measurements, pairing treatments with outcomes and symptoms with triggers. They suggested dedicating a special PGD record category for these data, with some preferring complex structures and others favoring simple data entry options: I would probably use the notes quite often just to maybe outline the symptoms I was experiencing and the steps I took to alleviate those symptoms or which doctors I contacted. P13 Making Collections Distinguishable Participants liked the existing collection descriptors and suggested additional ones. They wanted labels for clarity (clear, unclear, or potential issue), stability (stable or unstable), progress toward resolution, development stages (nonthreatening or threatening), and a list of involved providers and physicians. When collections were related to clinical visits, participants wanted to specify the targeted physicians. Making Collections Actionable Participants believed that organizing medical records by health issues in collections was a good start but thought that actionability could be improved with specific insight notes and annotations applied to entire collections: And then as far as the purpose of adding a note to the whole category, I would say that, like you said, if you happen to notice any patterns when you’re looking at the data, or basically I would use it for any general or bigger-picture takeaway that I wanted to tell my doctor, “Hey, I noticed this” or something and I wanted to bring it to their attention. P6 Participants envisioned using collection-wide notes to summarize contents or purpose, track progress, describe issue development, and highlight special events. They also wanted notes representing care plans and actions prioritized in a to-do list. Participants intended to use collections to prepare for clinical visits with questions, reminders, and critical measurements. They also saw value in adding collection notes about visit outcomes, key takeaways, and next steps. Some participants wanted to annotate and highlight keywords or add tags to free-text notes for organized review and pattern identification. Accessing the Collections All participants emphasized the importance of fast, reliable access to collections and their contents. They primarily relied on collection descriptors but also desired a deep search feature that would scan through individual records, notes, and annotations within collections. The results from our study are organized and analyzed around five qualitative themes, two applicable to RQ 2 and three applicable to RQ 1: (1) using collections for personal benefit (RQ 2), (2) using collections in a clinical setting (RQ 2), (3) creating and building collections (RQ 1), (4) enhancing collections (RQ 1), and (5) accessing collections (RQ 1). In the remainder of the Results section, we characterize the participants and report on these themes using 16 quotes from 11 different participants. The participants are labeled as P1 to P14. Using Collections for Personal Benefit Quick Access to Information Participants perceived the Collections feature as a way to index information at a suitable level of granularity for health issues, topics of interest, or conditions to monitor. They expected that this would give them quick access to current or urgent problems that were being managed: Personally I would like to monitor my asthma because I am using medication for that, the typical inhaler, but I would like to monitor, these days I had certain attacks or shortness of breath, so collections, having it back for that specific condition is very useful to me, because I wouldn’t have to speculate about when my last attack was or when my last appointment date was, it’s right here for me to access. P13 Reflection on Medical History Participants viewed collections as a tool to construct and understand their medical history. They expected collections to serve as a repository for the health issues they faced, aiding in reflecting on the existence, prevalence, and development of their health conditions: This will be able to find what happens when I have, in my case, headaches related to my high blood pressure and nothing else, or if you’re sick for the flu, influenza and you have fever, it’s just related to the influenza, not to the COVID. P1 Keeping Track of Health Status Most participants envisioned using collections to track their health status proactively. This included monitoring urgent issues needing immediate attention, unstable conditions requiring frequent observation, and treatments needing careful monitoring. Participants also wanted collections to track abnormal laboratory test results and vital sign values across various health issues: Well, for me, it’s kind of good [to have medical records organized in collections], especially, for example, blood samples, especially those with high triglycerides or something. Maybe I can collect them and see whether there’s a trend for this month, or for January, I’m high in this one. Then second month, I’m also high, so maybe I can lower it down...For collections, just categorize those. Which are high, which are low. P5 Journaling Daily Events Most of the participants envisioned using collections as a personal diary for coping with diseases and logging measurements and their effects on lifestyle. They wanted to track challenges, successes, and progress toward finding solutions and monitor disease developments: What I’ve done is I’ve taken all of my videos and stuff since February. Like I said, I’ve been to eight different doctors and I’ve shown them, this is the progress of what’s happened from...I’ve had two surgeries before this surgery where they lanced it, cut it, drained it. Nothing happened. The cyst came back and then it went into my bone. So I’m able to bring these photos, I’m able to bring this timeline, I’m able to bring my frustrations and show this doctor within 30 seconds [using a collection], look, this is what it looked like. And this is my own diary, my own history. It’s very important because they’re a doctor. They don’t know me, they don’t know what it looked like on day one. P10 Learning From Past Experiences Nearly all participants wanted to use collections to identify trends and patterns in their ongoing health experiences, including co-occurring symptoms and treatment effects. They also intended to log food intake, sleep, activity, or stress factors to find triggers for symptoms: I think I could definitely use them there. It’s a lot easier now because I could highlight the certain event, and put like my triggers down with it, like migraine on the third was this, you could even put what you took with it. So for me, I would look back and be like, “Oh, I can tell my doctor that, I’ve had 10 migraines. I took a medication with these three. What’s my options.” So I think that would be great, it’s a great tool that I can actually do that with this app. P9 Using Collections in a Clinical Setting Preparation for the Clinical Visit Most participants would use collections to prepare talking points for clinical visits combining personal measurements and notes with medical records from other providers (eg, laboratory and test results). This was crucial because their physicians often did not have access to external data: For collections, I would say that if I’m meeting with multiple providers about one health issue, I could see myself combining all my records there so that multiple providers can see each other’s records...I would probably [use the notes], if I needed to jot down a certain time that I took a measurement, or if my doctor told me keep track of what you ate that day, or kind of anything that I would want to have the details for the next time that I go and see the provider. P6 Relying on Collections During Clinical Visits All participants wanted to share collections with their physicians during clinical visits to establish ground truths, raise awareness of other providers’ information, and provide transparent talking points. They believed that adding PGD to collections could describe what happened between visits and raise their physicians’ awareness: If you have a condition, you need to check on your blood pressure. You need to communicate that with your doctor, so you can add a note [in the collection] saying like, “Latest, highest blood pressure from this week,” from a date. P14 In addition, most participants felt that taking in-visit notes and saving them in a collection could help them understand care plans and take appropriate actions afterward: Maybe [taking notes in the collection for the clinical visit] just for your own personal reference or if you wanted to bring it up later on in another appointment or something, or just maybe, I guess, just general recording of something that happened during that visit. P11 All participants saw physicians as essential partners in reviewing collections and deciding on actions based on their contents. This was primarily because most participants doubted their expertise in determining what should go into collections or which collections to create. While they were very open to include their physicians in the collection curation, some feared that they might overburden physicians with verification inquiries: I guess I would definitely do that [look for patterns in the data and store them in a collection] just because I can actually consult the doctor. Is this actually correlated or will I have to change my diet because it affects this? At least you can ask the doctor, or confirm whether that is true or not. P5 Quick Access to Information Participants perceived the Collections feature as a way to index information at a suitable level of granularity for health issues, topics of interest, or conditions to monitor. They expected that this would give them quick access to current or urgent problems that were being managed: Personally I would like to monitor my asthma because I am using medication for that, the typical inhaler, but I would like to monitor, these days I had certain attacks or shortness of breath, so collections, having it back for that specific condition is very useful to me, because I wouldn’t have to speculate about when my last attack was or when my last appointment date was, it’s right here for me to access. P13 Reflection on Medical History Participants viewed collections as a tool to construct and understand their medical history. They expected collections to serve as a repository for the health issues they faced, aiding in reflecting on the existence, prevalence, and development of their health conditions: This will be able to find what happens when I have, in my case, headaches related to my high blood pressure and nothing else, or if you’re sick for the flu, influenza and you have fever, it’s just related to the influenza, not to the COVID. P1 Keeping Track of Health Status Most participants envisioned using collections to track their health status proactively. This included monitoring urgent issues needing immediate attention, unstable conditions requiring frequent observation, and treatments needing careful monitoring. Participants also wanted collections to track abnormal laboratory test results and vital sign values across various health issues: Well, for me, it’s kind of good [to have medical records organized in collections], especially, for example, blood samples, especially those with high triglycerides or something. Maybe I can collect them and see whether there’s a trend for this month, or for January, I’m high in this one. Then second month, I’m also high, so maybe I can lower it down...For collections, just categorize those. Which are high, which are low. P5 Journaling Daily Events Most of the participants envisioned using collections as a personal diary for coping with diseases and logging measurements and their effects on lifestyle. They wanted to track challenges, successes, and progress toward finding solutions and monitor disease developments: What I’ve done is I’ve taken all of my videos and stuff since February. Like I said, I’ve been to eight different doctors and I’ve shown them, this is the progress of what’s happened from...I’ve had two surgeries before this surgery where they lanced it, cut it, drained it. Nothing happened. The cyst came back and then it went into my bone. So I’m able to bring these photos, I’m able to bring this timeline, I’m able to bring my frustrations and show this doctor within 30 seconds [using a collection], look, this is what it looked like. And this is my own diary, my own history. It’s very important because they’re a doctor. They don’t know me, they don’t know what it looked like on day one. P10 Learning From Past Experiences Nearly all participants wanted to use collections to identify trends and patterns in their ongoing health experiences, including co-occurring symptoms and treatment effects. They also intended to log food intake, sleep, activity, or stress factors to find triggers for symptoms: I think I could definitely use them there. It’s a lot easier now because I could highlight the certain event, and put like my triggers down with it, like migraine on the third was this, you could even put what you took with it. So for me, I would look back and be like, “Oh, I can tell my doctor that, I’ve had 10 migraines. I took a medication with these three. What’s my options.” So I think that would be great, it’s a great tool that I can actually do that with this app. P9 Participants perceived the Collections feature as a way to index information at a suitable level of granularity for health issues, topics of interest, or conditions to monitor. They expected that this would give them quick access to current or urgent problems that were being managed: Personally I would like to monitor my asthma because I am using medication for that, the typical inhaler, but I would like to monitor, these days I had certain attacks or shortness of breath, so collections, having it back for that specific condition is very useful to me, because I wouldn’t have to speculate about when my last attack was or when my last appointment date was, it’s right here for me to access. P13 Participants viewed collections as a tool to construct and understand their medical history. They expected collections to serve as a repository for the health issues they faced, aiding in reflecting on the existence, prevalence, and development of their health conditions: This will be able to find what happens when I have, in my case, headaches related to my high blood pressure and nothing else, or if you’re sick for the flu, influenza and you have fever, it’s just related to the influenza, not to the COVID. P1 Most participants envisioned using collections to track their health status proactively. This included monitoring urgent issues needing immediate attention, unstable conditions requiring frequent observation, and treatments needing careful monitoring. Participants also wanted collections to track abnormal laboratory test results and vital sign values across various health issues: Well, for me, it’s kind of good [to have medical records organized in collections], especially, for example, blood samples, especially those with high triglycerides or something. Maybe I can collect them and see whether there’s a trend for this month, or for January, I’m high in this one. Then second month, I’m also high, so maybe I can lower it down...For collections, just categorize those. Which are high, which are low. P5 Most of the participants envisioned using collections as a personal diary for coping with diseases and logging measurements and their effects on lifestyle. They wanted to track challenges, successes, and progress toward finding solutions and monitor disease developments: What I’ve done is I’ve taken all of my videos and stuff since February. Like I said, I’ve been to eight different doctors and I’ve shown them, this is the progress of what’s happened from...I’ve had two surgeries before this surgery where they lanced it, cut it, drained it. Nothing happened. The cyst came back and then it went into my bone. So I’m able to bring these photos, I’m able to bring this timeline, I’m able to bring my frustrations and show this doctor within 30 seconds [using a collection], look, this is what it looked like. And this is my own diary, my own history. It’s very important because they’re a doctor. They don’t know me, they don’t know what it looked like on day one. P10 Nearly all participants wanted to use collections to identify trends and patterns in their ongoing health experiences, including co-occurring symptoms and treatment effects. They also intended to log food intake, sleep, activity, or stress factors to find triggers for symptoms: I think I could definitely use them there. It’s a lot easier now because I could highlight the certain event, and put like my triggers down with it, like migraine on the third was this, you could even put what you took with it. So for me, I would look back and be like, “Oh, I can tell my doctor that, I’ve had 10 migraines. I took a medication with these three. What’s my options.” So I think that would be great, it’s a great tool that I can actually do that with this app. P9 Preparation for the Clinical Visit Most participants would use collections to prepare talking points for clinical visits combining personal measurements and notes with medical records from other providers (eg, laboratory and test results). This was crucial because their physicians often did not have access to external data: For collections, I would say that if I’m meeting with multiple providers about one health issue, I could see myself combining all my records there so that multiple providers can see each other’s records...I would probably [use the notes], if I needed to jot down a certain time that I took a measurement, or if my doctor told me keep track of what you ate that day, or kind of anything that I would want to have the details for the next time that I go and see the provider. P6 Relying on Collections During Clinical Visits All participants wanted to share collections with their physicians during clinical visits to establish ground truths, raise awareness of other providers’ information, and provide transparent talking points. They believed that adding PGD to collections could describe what happened between visits and raise their physicians’ awareness: If you have a condition, you need to check on your blood pressure. You need to communicate that with your doctor, so you can add a note [in the collection] saying like, “Latest, highest blood pressure from this week,” from a date. P14 In addition, most participants felt that taking in-visit notes and saving them in a collection could help them understand care plans and take appropriate actions afterward: Maybe [taking notes in the collection for the clinical visit] just for your own personal reference or if you wanted to bring it up later on in another appointment or something, or just maybe, I guess, just general recording of something that happened during that visit. P11 All participants saw physicians as essential partners in reviewing collections and deciding on actions based on their contents. This was primarily because most participants doubted their expertise in determining what should go into collections or which collections to create. While they were very open to include their physicians in the collection curation, some feared that they might overburden physicians with verification inquiries: I guess I would definitely do that [look for patterns in the data and store them in a collection] just because I can actually consult the doctor. Is this actually correlated or will I have to change my diet because it affects this? At least you can ask the doctor, or confirm whether that is true or not. P5 Most participants would use collections to prepare talking points for clinical visits combining personal measurements and notes with medical records from other providers (eg, laboratory and test results). This was crucial because their physicians often did not have access to external data: For collections, I would say that if I’m meeting with multiple providers about one health issue, I could see myself combining all my records there so that multiple providers can see each other’s records...I would probably [use the notes], if I needed to jot down a certain time that I took a measurement, or if my doctor told me keep track of what you ate that day, or kind of anything that I would want to have the details for the next time that I go and see the provider. P6 All participants wanted to share collections with their physicians during clinical visits to establish ground truths, raise awareness of other providers’ information, and provide transparent talking points. They believed that adding PGD to collections could describe what happened between visits and raise their physicians’ awareness: If you have a condition, you need to check on your blood pressure. You need to communicate that with your doctor, so you can add a note [in the collection] saying like, “Latest, highest blood pressure from this week,” from a date. P14 In addition, most participants felt that taking in-visit notes and saving them in a collection could help them understand care plans and take appropriate actions afterward: Maybe [taking notes in the collection for the clinical visit] just for your own personal reference or if you wanted to bring it up later on in another appointment or something, or just maybe, I guess, just general recording of something that happened during that visit. P11 All participants saw physicians as essential partners in reviewing collections and deciding on actions based on their contents. This was primarily because most participants doubted their expertise in determining what should go into collections or which collections to create. While they were very open to include their physicians in the collection curation, some feared that they might overburden physicians with verification inquiries: I guess I would definitely do that [look for patterns in the data and store them in a collection] just because I can actually consult the doctor. Is this actually correlated or will I have to change my diet because it affects this? At least you can ask the doctor, or confirm whether that is true or not. P5 Creating and Building Collections Manual Workflow for Creating and Building Collections Overall, all participants expressed satisfaction with the clarity and simplicity of the mechanics to create collections and save records in them. However, those with less medical knowledge and disease experience (relatively healthy and recently diagnosed individuals) thought that initiating and building collections was challenging. For them, it was not always clear what issues deserved separate collections, what records to include in a collection due to delicate dependencies, and why they should invest substantial time and effort in assembling collections. In contrast, the more experienced (chronic) patients were relatively confident in their ability to carve a personalized view of their EHR data. However, some did acknowledge that they might not be as exhaustive and reliable in their data organization. Participants quickly mastered using the Sliding Tabs with Accordions and appreciated the context-preserving record exploration. However, most of the participants found it challenging to assess the relevance of individual records due to unexplained clinical language and attribute values. Participants requested explanations, prominent visual cues for abnormal values to attract their attention, and time-series visualizations for additional context and noticing trends. They also wanted to be able to sort by date or attribute value, with filtering capabilities that also included adding the filtered records in bulk to a collection. Some participants desired collections with richer internal structures, recognizing the need to identify subsets of records and their importance within a collection. They suggested linking groups or individual records using specific annotations for easy identification during visual inspection or search: ...maybe if certain records are related to each other. So I would want to mark that. And then maybe just have a way of sorting down based on certain labels. P8 Automatic Support for Building Collections Participants were highly receptive to discussing ideas that could automate building collections. They were interested in seeing their records automatically put into collections based on provenance (provider, physician, hospital, location, and date) and clinical meaning (condition, disease, organ, organ system, and abnormality of the values across records). Some also suggested grouping records into collections based on personal annotations. For all these groupings, they wished to be able to edit the collection manually: If I had neurological problems, neurology collection. If I had urological problems, urology collection. I think that for me at least would seem a more straightforward way to categorize them. But from the categories I’ve already seen, I find those useful. P4 Several participants suggested using a “seed” to automate collection building, such as naming a collection, adding keywords, or including a few records: And I think a lot of patients don’t know where to start, what data to begin with. So if it’s something that’s already preset, they say, “Okay, I’m suffering from depression or I have diabetes.” And the system pulls the different data points that they would need to look at for someone who’s diabetic or someone who’s dealing with depression, I think that’s helpful. Because sometimes, the problem is you don’t know where to start and you don’t know what to look for. P11 Finally, most of the participants wanted to receive automated help to add or remove records for a collection that had already been created. Few described wanting to choose from a list of suggested records based on the existing content of a collection, revealing records and record patterns otherwise invisible to them. Others saw this automated record offering more as an idea generation approach—needing some follow-up validation, including taking it up with their physicians: When I work, I want to listen to some music and then I’m like, okay. I just don’t know what’s next. I need something similar to this, the same vibe, but I just can’t think about that. And then there is suggestions. And yes, some of it’s weird. But maybe like the doctor can also have some help here, and when you review the collections together, they might say, “Hey, listen, this is what the system gave you and that’s great. Let’s remove a few things. I would suggest you add a couple others. And whatever you put there, it’s also fine. And let’s keep it that way.” P6 Enhancing Collections Through Personally Provided Data Making Collections Complete Participants strongly expressed the need to complement their EHR data with daily entries from sensors; self-monitoring devices; and manual measurements of symptoms, treatments, and outcomes in various formats (text, photos, videos, and scanned documents): Actually, I think you can sort of restructure the whole core of the collections on top of two main pillars. The first one would be all of the doctor’s data, which is basically hard data, which allows you to diagnose, allows you to run statistical analysis...That could be part of the core data, but all of the context, maybe I’m getting this shortness of breath in my home, watching my TV, might be added by the notes. You have these two types of data. By adding the user data, would allow me to get context, give context, which is important and will allow me to, on a daily basis, keep a record, which in case of data like shortness of breath, I’m having, I’m not having. Would allow the doctor to have a really unbiased input on symptoms I’m having. P7 Participants wanted to log detailed observations and measurements, pairing treatments with outcomes and symptoms with triggers. They suggested dedicating a special PGD record category for these data, with some preferring complex structures and others favoring simple data entry options: I would probably use the notes quite often just to maybe outline the symptoms I was experiencing and the steps I took to alleviate those symptoms or which doctors I contacted. P13 Making Collections Distinguishable Participants liked the existing collection descriptors and suggested additional ones. They wanted labels for clarity (clear, unclear, or potential issue), stability (stable or unstable), progress toward resolution, development stages (nonthreatening or threatening), and a list of involved providers and physicians. When collections were related to clinical visits, participants wanted to specify the targeted physicians. Making Collections Actionable Participants believed that organizing medical records by health issues in collections was a good start but thought that actionability could be improved with specific insight notes and annotations applied to entire collections: And then as far as the purpose of adding a note to the whole category, I would say that, like you said, if you happen to notice any patterns when you’re looking at the data, or basically I would use it for any general or bigger-picture takeaway that I wanted to tell my doctor, “Hey, I noticed this” or something and I wanted to bring it to their attention. P6 Participants envisioned using collection-wide notes to summarize contents or purpose, track progress, describe issue development, and highlight special events. They also wanted notes representing care plans and actions prioritized in a to-do list. Participants intended to use collections to prepare for clinical visits with questions, reminders, and critical measurements. They also saw value in adding collection notes about visit outcomes, key takeaways, and next steps. Some participants wanted to annotate and highlight keywords or add tags to free-text notes for organized review and pattern identification. Accessing the Collections All participants emphasized the importance of fast, reliable access to collections and their contents. They primarily relied on collection descriptors but also desired a deep search feature that would scan through individual records, notes, and annotations within collections. Manual Workflow for Creating and Building Collections Overall, all participants expressed satisfaction with the clarity and simplicity of the mechanics to create collections and save records in them. However, those with less medical knowledge and disease experience (relatively healthy and recently diagnosed individuals) thought that initiating and building collections was challenging. For them, it was not always clear what issues deserved separate collections, what records to include in a collection due to delicate dependencies, and why they should invest substantial time and effort in assembling collections. In contrast, the more experienced (chronic) patients were relatively confident in their ability to carve a personalized view of their EHR data. However, some did acknowledge that they might not be as exhaustive and reliable in their data organization. Participants quickly mastered using the Sliding Tabs with Accordions and appreciated the context-preserving record exploration. However, most of the participants found it challenging to assess the relevance of individual records due to unexplained clinical language and attribute values. Participants requested explanations, prominent visual cues for abnormal values to attract their attention, and time-series visualizations for additional context and noticing trends. They also wanted to be able to sort by date or attribute value, with filtering capabilities that also included adding the filtered records in bulk to a collection. Some participants desired collections with richer internal structures, recognizing the need to identify subsets of records and their importance within a collection. They suggested linking groups or individual records using specific annotations for easy identification during visual inspection or search: ...maybe if certain records are related to each other. So I would want to mark that. And then maybe just have a way of sorting down based on certain labels. P8 Automatic Support for Building Collections Participants were highly receptive to discussing ideas that could automate building collections. They were interested in seeing their records automatically put into collections based on provenance (provider, physician, hospital, location, and date) and clinical meaning (condition, disease, organ, organ system, and abnormality of the values across records). Some also suggested grouping records into collections based on personal annotations. For all these groupings, they wished to be able to edit the collection manually: If I had neurological problems, neurology collection. If I had urological problems, urology collection. I think that for me at least would seem a more straightforward way to categorize them. But from the categories I’ve already seen, I find those useful. P4 Several participants suggested using a “seed” to automate collection building, such as naming a collection, adding keywords, or including a few records: And I think a lot of patients don’t know where to start, what data to begin with. So if it’s something that’s already preset, they say, “Okay, I’m suffering from depression or I have diabetes.” And the system pulls the different data points that they would need to look at for someone who’s diabetic or someone who’s dealing with depression, I think that’s helpful. Because sometimes, the problem is you don’t know where to start and you don’t know what to look for. P11 Finally, most of the participants wanted to receive automated help to add or remove records for a collection that had already been created. Few described wanting to choose from a list of suggested records based on the existing content of a collection, revealing records and record patterns otherwise invisible to them. Others saw this automated record offering more as an idea generation approach—needing some follow-up validation, including taking it up with their physicians: When I work, I want to listen to some music and then I’m like, okay. I just don’t know what’s next. I need something similar to this, the same vibe, but I just can’t think about that. And then there is suggestions. And yes, some of it’s weird. But maybe like the doctor can also have some help here, and when you review the collections together, they might say, “Hey, listen, this is what the system gave you and that’s great. Let’s remove a few things. I would suggest you add a couple others. And whatever you put there, it’s also fine. And let’s keep it that way.” P6 Overall, all participants expressed satisfaction with the clarity and simplicity of the mechanics to create collections and save records in them. However, those with less medical knowledge and disease experience (relatively healthy and recently diagnosed individuals) thought that initiating and building collections was challenging. For them, it was not always clear what issues deserved separate collections, what records to include in a collection due to delicate dependencies, and why they should invest substantial time and effort in assembling collections. In contrast, the more experienced (chronic) patients were relatively confident in their ability to carve a personalized view of their EHR data. However, some did acknowledge that they might not be as exhaustive and reliable in their data organization. Participants quickly mastered using the Sliding Tabs with Accordions and appreciated the context-preserving record exploration. However, most of the participants found it challenging to assess the relevance of individual records due to unexplained clinical language and attribute values. Participants requested explanations, prominent visual cues for abnormal values to attract their attention, and time-series visualizations for additional context and noticing trends. They also wanted to be able to sort by date or attribute value, with filtering capabilities that also included adding the filtered records in bulk to a collection. Some participants desired collections with richer internal structures, recognizing the need to identify subsets of records and their importance within a collection. They suggested linking groups or individual records using specific annotations for easy identification during visual inspection or search: ...maybe if certain records are related to each other. So I would want to mark that. And then maybe just have a way of sorting down based on certain labels. P8 Participants were highly receptive to discussing ideas that could automate building collections. They were interested in seeing their records automatically put into collections based on provenance (provider, physician, hospital, location, and date) and clinical meaning (condition, disease, organ, organ system, and abnormality of the values across records). Some also suggested grouping records into collections based on personal annotations. For all these groupings, they wished to be able to edit the collection manually: If I had neurological problems, neurology collection. If I had urological problems, urology collection. I think that for me at least would seem a more straightforward way to categorize them. But from the categories I’ve already seen, I find those useful. P4 Several participants suggested using a “seed” to automate collection building, such as naming a collection, adding keywords, or including a few records: And I think a lot of patients don’t know where to start, what data to begin with. So if it’s something that’s already preset, they say, “Okay, I’m suffering from depression or I have diabetes.” And the system pulls the different data points that they would need to look at for someone who’s diabetic or someone who’s dealing with depression, I think that’s helpful. Because sometimes, the problem is you don’t know where to start and you don’t know what to look for. P11 Finally, most of the participants wanted to receive automated help to add or remove records for a collection that had already been created. Few described wanting to choose from a list of suggested records based on the existing content of a collection, revealing records and record patterns otherwise invisible to them. Others saw this automated record offering more as an idea generation approach—needing some follow-up validation, including taking it up with their physicians: When I work, I want to listen to some music and then I’m like, okay. I just don’t know what’s next. I need something similar to this, the same vibe, but I just can’t think about that. And then there is suggestions. And yes, some of it’s weird. But maybe like the doctor can also have some help here, and when you review the collections together, they might say, “Hey, listen, this is what the system gave you and that’s great. Let’s remove a few things. I would suggest you add a couple others. And whatever you put there, it’s also fine. And let’s keep it that way.” P6 Making Collections Complete Participants strongly expressed the need to complement their EHR data with daily entries from sensors; self-monitoring devices; and manual measurements of symptoms, treatments, and outcomes in various formats (text, photos, videos, and scanned documents): Actually, I think you can sort of restructure the whole core of the collections on top of two main pillars. The first one would be all of the doctor’s data, which is basically hard data, which allows you to diagnose, allows you to run statistical analysis...That could be part of the core data, but all of the context, maybe I’m getting this shortness of breath in my home, watching my TV, might be added by the notes. You have these two types of data. By adding the user data, would allow me to get context, give context, which is important and will allow me to, on a daily basis, keep a record, which in case of data like shortness of breath, I’m having, I’m not having. Would allow the doctor to have a really unbiased input on symptoms I’m having. P7 Participants wanted to log detailed observations and measurements, pairing treatments with outcomes and symptoms with triggers. They suggested dedicating a special PGD record category for these data, with some preferring complex structures and others favoring simple data entry options: I would probably use the notes quite often just to maybe outline the symptoms I was experiencing and the steps I took to alleviate those symptoms or which doctors I contacted. P13 Making Collections Distinguishable Participants liked the existing collection descriptors and suggested additional ones. They wanted labels for clarity (clear, unclear, or potential issue), stability (stable or unstable), progress toward resolution, development stages (nonthreatening or threatening), and a list of involved providers and physicians. When collections were related to clinical visits, participants wanted to specify the targeted physicians. Making Collections Actionable Participants believed that organizing medical records by health issues in collections was a good start but thought that actionability could be improved with specific insight notes and annotations applied to entire collections: And then as far as the purpose of adding a note to the whole category, I would say that, like you said, if you happen to notice any patterns when you’re looking at the data, or basically I would use it for any general or bigger-picture takeaway that I wanted to tell my doctor, “Hey, I noticed this” or something and I wanted to bring it to their attention. P6 Participants envisioned using collection-wide notes to summarize contents or purpose, track progress, describe issue development, and highlight special events. They also wanted notes representing care plans and actions prioritized in a to-do list. Participants intended to use collections to prepare for clinical visits with questions, reminders, and critical measurements. They also saw value in adding collection notes about visit outcomes, key takeaways, and next steps. Some participants wanted to annotate and highlight keywords or add tags to free-text notes for organized review and pattern identification. Participants strongly expressed the need to complement their EHR data with daily entries from sensors; self-monitoring devices; and manual measurements of symptoms, treatments, and outcomes in various formats (text, photos, videos, and scanned documents): Actually, I think you can sort of restructure the whole core of the collections on top of two main pillars. The first one would be all of the doctor’s data, which is basically hard data, which allows you to diagnose, allows you to run statistical analysis...That could be part of the core data, but all of the context, maybe I’m getting this shortness of breath in my home, watching my TV, might be added by the notes. You have these two types of data. By adding the user data, would allow me to get context, give context, which is important and will allow me to, on a daily basis, keep a record, which in case of data like shortness of breath, I’m having, I’m not having. Would allow the doctor to have a really unbiased input on symptoms I’m having. P7 Participants wanted to log detailed observations and measurements, pairing treatments with outcomes and symptoms with triggers. They suggested dedicating a special PGD record category for these data, with some preferring complex structures and others favoring simple data entry options: I would probably use the notes quite often just to maybe outline the symptoms I was experiencing and the steps I took to alleviate those symptoms or which doctors I contacted. P13 Participants liked the existing collection descriptors and suggested additional ones. They wanted labels for clarity (clear, unclear, or potential issue), stability (stable or unstable), progress toward resolution, development stages (nonthreatening or threatening), and a list of involved providers and physicians. When collections were related to clinical visits, participants wanted to specify the targeted physicians. Participants believed that organizing medical records by health issues in collections was a good start but thought that actionability could be improved with specific insight notes and annotations applied to entire collections: And then as far as the purpose of adding a note to the whole category, I would say that, like you said, if you happen to notice any patterns when you’re looking at the data, or basically I would use it for any general or bigger-picture takeaway that I wanted to tell my doctor, “Hey, I noticed this” or something and I wanted to bring it to their attention. P6 Participants envisioned using collection-wide notes to summarize contents or purpose, track progress, describe issue development, and highlight special events. They also wanted notes representing care plans and actions prioritized in a to-do list. Participants intended to use collections to prepare for clinical visits with questions, reminders, and critical measurements. They also saw value in adding collection notes about visit outcomes, key takeaways, and next steps. Some participants wanted to annotate and highlight keywords or add tags to free-text notes for organized review and pattern identification. All participants emphasized the importance of fast, reliable access to collections and their contents. They primarily relied on collection descriptors but also desired a deep search feature that would scan through individual records, notes, and annotations within collections. Principal Findings We identified 3 principal findings of our study. First, participants embraced the collection concept. Unrestrictedly organizing EHR data into collections that map medical records to health issues and track ongoing concerns gave participants a sense of ownership. They felt empowered by developing personalized health narratives that could aid in self-management and communication with their physicians, enhancing their self-advocacy. Second, while participants easily mastered the interface for initiating and adding records to collections, they found the process laborious. They lacked confidence in selecting appropriate records due to limited medical knowledge and requested additional visual cues, explanations, and automatic collection features. There was concern about potential self-misguidance without physician verification. Third, collections would need richer PGD capabilities for adding contextual information not found in participants’ EHR data, logging observations, and labeling data. This would enhance the comprehensiveness and accuracy of their health narratives and support foraging, sensemaking, and action taking. Interpretation of the Findings and Contributions Overview On a broader scale, this work contributes to patient-centered care. This is achieved by demonstrating potential to enhance patients’ grasp of their health, encourage self-advocacy, and improve patient-provider communication. More accurately, there are several concrete contributions of our work that can be considered as proxies toward achieving the aforementioned objectives: (1) encouraging patient ownership of their EHR data by organizing them into personalized, health issue–based collections; (2) understanding patients’ perceptions and preferences for creating, building, and using these collections; and (3) offering design insights for automating collections, integrating rich PGD, enhancing access to collection contents, and using collections to facilitate patient-provider communication. Going forward, we will situate our findings within a sensemaking framework and discuss contributions related to 3 key patient needs: increasing awareness through independent health sensemaking, proactivity through efficient action taking, and self-advocacy through incorporating evidence-supported patient perspectives into patient-provider communication. We will elaborate on how collections can meet these needs and offer design implications to enhance their capabilities. The Role of the Collections in Supporting Sensemaking To explain the collections’ role in sensemaking, we used the model by Pirolli and Card , which divides the sensemaking process into 2 subcycles: the foraging loop and the sensemaking loop. On the basis of this model, collections can be described as a space for assembling relevant data about a topic, finding relationships between them, and storing outcomes from the sensemaking. In the foraging loop, patients gather relevant records to answer questions such as the following—“Is there a relationship between my weight and blood pressure?”—and save them in a collection, such as Weight vs. Blood Pressure. In the sensemaking loop, patients identify information relationships within the collection that they capture in notes, such as instances where there was co-occurrence of high blood pressure and high body mass. These notes help argue hypotheses such as the following: “My blood pressure is high when I’m overweight.” The outcome of this sensemaking process could be a comprehensive note for a clinical visit. While, in their current form, collections respond to the needs of the sensemaking model by Pirolli and Card , improvements can be made to make this more efficient. This study revealed that medical records alone are not enough for reliable sensemaking. Adding PGD such as symptoms, measurements, outcomes, and everyday events is essential for creating comprehensive collections. The foraging loop can be made less laborious and time-consuming if there are additional visual cues, medical explanations, filtering capabilities, and automatic support to improve the relevance and reduce the effort of assembling collections. The sensemaking loop could be improved by adding more schematization capabilities such as annotating medical records and PGD to identify patterns later (eg, symptom triggers, medication effects, and correlations) and grouping records within collections, labeling those groups, and establishing group relationships with explanations (eg, linking “cholesterol lab results” with “food intake” as “food effects on cholesterol” in a High Blood Pressure collection). Reliability of Collections According to our study, there are 4 main factors that can influence the reliability of the collections: robust coverage of health issues, provision of PGD, grouping and linking of medical records and PGD within a collection, and verifying the contents of the collections. This reliability is related to collections’ capability to aid in creating personalized but realistic health issue narratives, support self-management, and stimulate awareness and proactivity. Robust Coverage of Health Issues: Relevance Assessment Collections should ensure that patients can create collections for their most important health issues to support awareness and proactivity. Participants desired visual cues, explanations, and automatic support to determine which collections to create and what records to include. While participants found the context-capturing data exploration using Sliding Tabs and Timeline convenient, they needed more to identify relevant records quickly. Future tools could incorporate Accordions that summarize record types, graph values over time, and highlight abnormal or extreme values. In addition, patients should be able to expand individual records to see explanations of clinical terms and clinical meaning interpretations. While there is existing work related to visualizing time series of EHR data and automatic provision of short explanations , this study shows the need for combining them in a new way to support relevance assessment for a novel purpose—constructing collections. Finally, patients should be able to order and filter records by attribute for quicker browsing and bulk addition to a collection. Automatic support should also be provided for creating and building collections to save time and ensure robust, reliable coverage of health issues. Our previous work highlighted the need for automating collections , and this study highlights a clear preference for automatically grouping medical records by clinical meaning—whether thematically or guided by patient input . Thematic collections would be those that tie records together based on conditions (bronchitis or diabetes) and procedures (stent placement or appendix removal) or with respect to organs (heart or kidneys) or organ systems (cardiovascular or renal). Alternatively, patients could specify a seed by setting parameters such as the collection’s name, tags, or initial records. The system can offer candidate records to include or delete with confidence scores and explanations. Patients could then refine these system-generated collections by adding or removing records, PGD, and tags. Addressing automatic support for collections may be challenging due to subtle relationships between medical records . However, starting with easier constructs such as time stamps (eg, medical records from the same day, week, or month), FHIR links (eg, medical records from the same encounter or physician), abnormal values based on well-established clinical guidelines (eg, high or low blood pressure or cholesterol), test findings (eg, positive and negative), and patient tags (eg, triggers or pivotal events) is a feasible approach. Provision of PGD: Sensemaking Data Completeness Collections should include PGD to improve data completeness for sensemaking. Previous research has suggested that maintaining consistent PGD logging over time is difficult . However, this should not be considered a barrier or a prerequisite for the collections’ success. Patients’ motivation and preferences for PGD logging intensity vary based on their disease self-management state . When setting goals and learning strategies, patients prefer meticulous data collection. Once goals and strategies are in place, logging intensity decreases. In addition, if physicians require PGD logging for treatment planning, patients are motivated to engage in it . Thus, collections should enable flexible and efficient PGD logging. Disease-specific contexts such as irritable bowel syndrome , diabetes , and migraine have explored health sensemaking without focusing on diverse data types. This contrasts with patients’ desire to log PGD for various medical issues within a single application using a universal logging model for different observations . To address this issue, we propose a straightforward workflow where patients initiate free-text entries and use tags to specify the type, quality, or other details. This mechanism allows for quick data capture and embellishment at convenient times. Tags can classify PGD as clinical observations , everyday life events , or notes . Further specification can be added using tags such as symptom , measurement , treatment , and outcome for observations; meal , exercise , meeting , and deadline for life events; and context , personal note , and visit note for notes. Additional tags such as absent , normal , high , low , extreme , improvement , deterioration , pivotal event , trigger , or relaxer can be used for further detail that captures the quality and importance of the logged data. In addition to these system-offered tags, patients can also create their own custom tags for better personalization. Grouping and Linking Within Collections: Schematization Capabilities Patients need to connect medical records and PGD within collections for easier sensemaking. Future tools should add structure by enabling record grouping and linking of groups or individual records. This helps highlight important subsets of records and trace major conclusions as collections grow. We recommend using a simple yet powerful tagging concept. Records sharing the same tag can form a group, whereas links between groups can be specified using related tags. The same mechanism can link individual records with other records or groups, providing nuanced sensemaking. This approach aligns well with the proposed PGD tagging model that can be applied to medical records as well. Collection Verification Collections represent the patient’s personalized perception of their health and issues. As such, they should undergo occasional verification by the patient’s physician for safe decision-making and action taking. While collection verification may add to the physician’s workload, it can inspire and enable patients to manage health issues more independently. That said, patients should consider the physician’s workload before requesting verification . Future tools could allow for the conversion of a collection into a well–laid-out PDF document capturing all its contents. This document can be printed for review during a clinical visit or shared as a PDF attachment in the patient portal for verification at the physician’s convenience. Taking Actions Based on Collections While annotating PGD is known to aid learning and disease self-management , our findings reveal that annotations can also enhance EHR data, creating synergy with PGD. Participants expressed a desire to annotate their data for various purposes: identifying triggers to avoid or encourage certain behaviors, marking pivotal events to remind them of shifts in health attitudes and management, and labeling outcomes as desired or undesired to evaluate treatments and strategies. These capabilities can be easily implemented using the previously elaborated tag-based design for linking records. To increase the awareness and prioritization of collections, we previously proposed collection descriptors such as topic, urgency, currency, and sentiment . Participants found value in these descriptors but expressed a need for additional ones that can be classified as patient specified and data driven . Future tools may include patient-specified descriptors for clarity (eg, is the diagnosis clear?), stability (eg, is the treatment working consistently?), severity (eg, is there a significant medical risk?), and progress markers (eg, is the issue substantially resolved?). Data-driven descriptors could be derived from the collection data, indicating the time span (from the oldest to the latest record) or listing the physicians involved (the providers and physicians the records came from). Both types of descriptors should be optional for patients to use as needed. Providing an inner structure, enabling annotations, and describing collections can improve information access and expedite decision-making. Powerful search engines can use these metadata to allow patients direct and easy access not only to individual collections but also to their specific contents. While these features can enhance collections’ actionability, it is important to note that collections are not meant for making independent clinical decisions by patients. Collections should be verified by a physician to serve as reliable tools for sensemaking and health self-management. However, collections can always be invaluable tools for patients to understand their health; organize thoughts, hypotheses, and insights; and communicate effectively with their physicians. Collection Use for Patient-Provider Communication As observed in our findings, patients can use collections to prepare for a clinical visit by devising checklists and organizing thoughts supported by evidence. During the visit, collections can be used for note taking and, afterward, for recording reminders and follow-up actions . These uses indicate how collections can start addressing known challenges during clinical visits, such as problem presentation , information retention , setting common ground, aligning goals, and understanding instructions . To effectively tackle such challenges, the Collections feature should support richer note capturing and collaborative data analysis in a collocated setting . Future improvements in capturing PGD could make collections more appealing to physicians. For physicians, PGD play a crucial role in understanding the boundaries and context for accurate diagnosis and optimal treatment . However, physicians often face problems with PGD, such as incomplete data, inconsistent data structures, and insufficient time for reviewing due to poor organization . These issues arise because patients use disjointed platforms to log their data, lacking consistent models for logging different types of data , and face challenges in efficiently using these platforms . Collections can help overcome these issues by providing a single platform for logging PGD for various health issues in a simple, universal way that allows for flexibility, organization, and standardization. An alternative approach to enhancing collections as a communication tool and fostering physician collaboration in their creation and verification is to introduce them as a shared resource similar to Google Docs . While this may seem unconventional, it builds on the principles of OpenNotes . OpenNotes provides access to and transparency regarding clinical notes, enabling patients to improve their treatment and EHR data quality by taking an active role in detecting errors, raising concerns, asking questions, or seeking clarifications . Similar to OpenNotes, shared collections would follow the principle of asynchronous communication and transparency. However, shared collections could eliminate the expressiveness constraints and lack of efficient ways to provide granular and tailored context observed in existing messaging systems . In addition, shared collections would enable direct editing of underlying data in collaborative ways, minimizing communication overhead. Moreover, shared collections would introduce a new communication channel between patients and providers outside the traditional patient portal. Synchronizing the digital traces of care in collections with the provider’s EHR system to avoid discrepancies and legal issues should be a top consideration in future design iterations of the shared collection concept. A Glimpse Into the Future: Collections and Generative Artificial Intelligence In the future, we should explore the potential of generative artificial intelligence (GenAI) models to support patient sensemaking through collections. Tools such as ChatGPT and Med-PaLM , which have demonstrated substantial medical knowledge , can replace the need for custom-made machine learning algorithms for knowledge-intensive tasks. In particular, GenAI tools can aid in automatic and iterative collection construction with explanations and guidance. They can analyze the data within collections for insights, including medical records, PGD, notes, and tags. In addition, GenAI can assist in composing case narratives and talking points for clinical encounters. By offering this level of automation, GenAI can help tackle the significant knowledge challenges while lowering the labor barrier for patients’ sensemaking activities. Using GenAI models, collection construction could rely on natural language instructions such as the following: “Group my EHR data by condition,” “Find all records related to my bronchitis,” or “Identify records that don’t belong to this collection and those that are missing.” GenAI models could also deliver context, explaining why certain records are included or excluded and providing educational material such as term definitions and clinical implications. In addition, patients can issue commands for identifying relationships within their annotated data, such as “List all triggers for my headaches over the last year.” Finally, they can ask for help in constructing case presentations for clinical visits (eg, “Based on my ‘High Blood Pressure’ collection notes, write a 100-word summary”). To be useful for sensemaking, GenAI tools do not need to achieve complete accuracy. While still striving for maximum reliability, their main value should come from providing an environment that enables and encourages patients to refine artificial intelligence–generated outputs. As such, the contribution of GenAI toward sensemaking would be evaluated on its ability to help the patient efficiently produce a satisfactory solution with minimal physician input. Finally, existing approaches for supporting sensemaking through search and interactive visualizations should not be disregarded. Exploring the integration of GenAI, search, and visualization is a prudent strategy as different sensemaking tasks related to collection assembly, editing, and analysis may require diverse approaches based on complexity, patient skills, and artificial intelligence reliability. Limitations This study has several limitations. First, the cohort skewed younger, likely due to recruitment via Craigslist (less popular with older adults) and the complexity of the remote setup. Second, participants used data from a fictitious patient, which may have reduced their motivation to learn the app and their ability to suggest real-life use cases. Third, participants had limited time to learn how to interact with collections, possibly affecting their perceptions of usability and utility. Future studies should have participants use their own data with automatic interaction logging. Despite these limitations, this study provided valuable insights into designing patient-facing sensemaking tools for organizing and augmenting EHR data. Conclusions Collections can potentially improve patient-centered care by involving patients more in decision-making and encouraging self-advocacy. Current assumptions often expect patients to have the necessary skills, tools, and motivation. We believe that collections can lower these barriers, encouraging patients to increase engagement with their health data, better educate themselves , and communicate more effectively with their care providers. Our study suggests that EHR data can be better used and more useful for patients through improved organization and annotation . This approach can incentivize patients to engage more deeply with their EHR data, develop insights, and reflect on their experiences. Patients felt that this empowered their awareness, resourcefulness, and proactivity regarding health issues, making them more prepared and better informed for clinician interactions. These findings support our premise that collections are a crucial step toward patient empowerment and self-advocacy . With appropriate improvements, collections can enhance patients’ expertise by facilitating sensemaking activities and enabling insightful discussions with their physicians. First, collections motivate patients to construct health models based on their issues and ongoing problems. Second, patients gain medical education by actively participating in the evolution of collections through independent or system-assisted assembly and editing. Finally, patients acquire additional medical knowledge by engaging in meaningful discussions with their physicians and considering their feedback on collection verification. Our study highlighted the importance of integrating PGD with EHR data. We envision a synergy in which patients use clinical data as a foundation, augmenting them with their observations, notes, and annotations to create personalized health narratives that support better health management and provider communication. In the future, we should explore GenAI models to support patient sensemaking through collections. These models could help patients build collections, analyze the data within them, and produce health narratives more efficiently. Such enhancements may also reduce physicians’ workload for verifying collection contents, leading to more focused, evidence-driven discussions during visits. Promising ideas from this work should be advanced carefully, with gradual design improvements tested in real-life settings. Respecting existing clinical practices and workflows can facilitate quicker adoption and more significant changes in the future. We believe that collections can revolutionize how patients interact with their medical records and communicate with their providers. We identified 3 principal findings of our study. First, participants embraced the collection concept. Unrestrictedly organizing EHR data into collections that map medical records to health issues and track ongoing concerns gave participants a sense of ownership. They felt empowered by developing personalized health narratives that could aid in self-management and communication with their physicians, enhancing their self-advocacy. Second, while participants easily mastered the interface for initiating and adding records to collections, they found the process laborious. They lacked confidence in selecting appropriate records due to limited medical knowledge and requested additional visual cues, explanations, and automatic collection features. There was concern about potential self-misguidance without physician verification. Third, collections would need richer PGD capabilities for adding contextual information not found in participants’ EHR data, logging observations, and labeling data. This would enhance the comprehensiveness and accuracy of their health narratives and support foraging, sensemaking, and action taking. Overview On a broader scale, this work contributes to patient-centered care. This is achieved by demonstrating potential to enhance patients’ grasp of their health, encourage self-advocacy, and improve patient-provider communication. More accurately, there are several concrete contributions of our work that can be considered as proxies toward achieving the aforementioned objectives: (1) encouraging patient ownership of their EHR data by organizing them into personalized, health issue–based collections; (2) understanding patients’ perceptions and preferences for creating, building, and using these collections; and (3) offering design insights for automating collections, integrating rich PGD, enhancing access to collection contents, and using collections to facilitate patient-provider communication. Going forward, we will situate our findings within a sensemaking framework and discuss contributions related to 3 key patient needs: increasing awareness through independent health sensemaking, proactivity through efficient action taking, and self-advocacy through incorporating evidence-supported patient perspectives into patient-provider communication. We will elaborate on how collections can meet these needs and offer design implications to enhance their capabilities. The Role of the Collections in Supporting Sensemaking To explain the collections’ role in sensemaking, we used the model by Pirolli and Card , which divides the sensemaking process into 2 subcycles: the foraging loop and the sensemaking loop. On the basis of this model, collections can be described as a space for assembling relevant data about a topic, finding relationships between them, and storing outcomes from the sensemaking. In the foraging loop, patients gather relevant records to answer questions such as the following—“Is there a relationship between my weight and blood pressure?”—and save them in a collection, such as Weight vs. Blood Pressure. In the sensemaking loop, patients identify information relationships within the collection that they capture in notes, such as instances where there was co-occurrence of high blood pressure and high body mass. These notes help argue hypotheses such as the following: “My blood pressure is high when I’m overweight.” The outcome of this sensemaking process could be a comprehensive note for a clinical visit. While, in their current form, collections respond to the needs of the sensemaking model by Pirolli and Card , improvements can be made to make this more efficient. This study revealed that medical records alone are not enough for reliable sensemaking. Adding PGD such as symptoms, measurements, outcomes, and everyday events is essential for creating comprehensive collections. The foraging loop can be made less laborious and time-consuming if there are additional visual cues, medical explanations, filtering capabilities, and automatic support to improve the relevance and reduce the effort of assembling collections. The sensemaking loop could be improved by adding more schematization capabilities such as annotating medical records and PGD to identify patterns later (eg, symptom triggers, medication effects, and correlations) and grouping records within collections, labeling those groups, and establishing group relationships with explanations (eg, linking “cholesterol lab results” with “food intake” as “food effects on cholesterol” in a High Blood Pressure collection). Reliability of Collections According to our study, there are 4 main factors that can influence the reliability of the collections: robust coverage of health issues, provision of PGD, grouping and linking of medical records and PGD within a collection, and verifying the contents of the collections. This reliability is related to collections’ capability to aid in creating personalized but realistic health issue narratives, support self-management, and stimulate awareness and proactivity. Robust Coverage of Health Issues: Relevance Assessment Collections should ensure that patients can create collections for their most important health issues to support awareness and proactivity. Participants desired visual cues, explanations, and automatic support to determine which collections to create and what records to include. While participants found the context-capturing data exploration using Sliding Tabs and Timeline convenient, they needed more to identify relevant records quickly. Future tools could incorporate Accordions that summarize record types, graph values over time, and highlight abnormal or extreme values. In addition, patients should be able to expand individual records to see explanations of clinical terms and clinical meaning interpretations. While there is existing work related to visualizing time series of EHR data and automatic provision of short explanations , this study shows the need for combining them in a new way to support relevance assessment for a novel purpose—constructing collections. Finally, patients should be able to order and filter records by attribute for quicker browsing and bulk addition to a collection. Automatic support should also be provided for creating and building collections to save time and ensure robust, reliable coverage of health issues. Our previous work highlighted the need for automating collections , and this study highlights a clear preference for automatically grouping medical records by clinical meaning—whether thematically or guided by patient input . Thematic collections would be those that tie records together based on conditions (bronchitis or diabetes) and procedures (stent placement or appendix removal) or with respect to organs (heart or kidneys) or organ systems (cardiovascular or renal). Alternatively, patients could specify a seed by setting parameters such as the collection’s name, tags, or initial records. The system can offer candidate records to include or delete with confidence scores and explanations. Patients could then refine these system-generated collections by adding or removing records, PGD, and tags. Addressing automatic support for collections may be challenging due to subtle relationships between medical records . However, starting with easier constructs such as time stamps (eg, medical records from the same day, week, or month), FHIR links (eg, medical records from the same encounter or physician), abnormal values based on well-established clinical guidelines (eg, high or low blood pressure or cholesterol), test findings (eg, positive and negative), and patient tags (eg, triggers or pivotal events) is a feasible approach. Provision of PGD: Sensemaking Data Completeness Collections should include PGD to improve data completeness for sensemaking. Previous research has suggested that maintaining consistent PGD logging over time is difficult . However, this should not be considered a barrier or a prerequisite for the collections’ success. Patients’ motivation and preferences for PGD logging intensity vary based on their disease self-management state . When setting goals and learning strategies, patients prefer meticulous data collection. Once goals and strategies are in place, logging intensity decreases. In addition, if physicians require PGD logging for treatment planning, patients are motivated to engage in it . Thus, collections should enable flexible and efficient PGD logging. Disease-specific contexts such as irritable bowel syndrome , diabetes , and migraine have explored health sensemaking without focusing on diverse data types. This contrasts with patients’ desire to log PGD for various medical issues within a single application using a universal logging model for different observations . To address this issue, we propose a straightforward workflow where patients initiate free-text entries and use tags to specify the type, quality, or other details. This mechanism allows for quick data capture and embellishment at convenient times. Tags can classify PGD as clinical observations , everyday life events , or notes . Further specification can be added using tags such as symptom , measurement , treatment , and outcome for observations; meal , exercise , meeting , and deadline for life events; and context , personal note , and visit note for notes. Additional tags such as absent , normal , high , low , extreme , improvement , deterioration , pivotal event , trigger , or relaxer can be used for further detail that captures the quality and importance of the logged data. In addition to these system-offered tags, patients can also create their own custom tags for better personalization. Grouping and Linking Within Collections: Schematization Capabilities Patients need to connect medical records and PGD within collections for easier sensemaking. Future tools should add structure by enabling record grouping and linking of groups or individual records. This helps highlight important subsets of records and trace major conclusions as collections grow. We recommend using a simple yet powerful tagging concept. Records sharing the same tag can form a group, whereas links between groups can be specified using related tags. The same mechanism can link individual records with other records or groups, providing nuanced sensemaking. This approach aligns well with the proposed PGD tagging model that can be applied to medical records as well. Collection Verification Collections represent the patient’s personalized perception of their health and issues. As such, they should undergo occasional verification by the patient’s physician for safe decision-making and action taking. While collection verification may add to the physician’s workload, it can inspire and enable patients to manage health issues more independently. That said, patients should consider the physician’s workload before requesting verification . Future tools could allow for the conversion of a collection into a well–laid-out PDF document capturing all its contents. This document can be printed for review during a clinical visit or shared as a PDF attachment in the patient portal for verification at the physician’s convenience. Taking Actions Based on Collections While annotating PGD is known to aid learning and disease self-management , our findings reveal that annotations can also enhance EHR data, creating synergy with PGD. Participants expressed a desire to annotate their data for various purposes: identifying triggers to avoid or encourage certain behaviors, marking pivotal events to remind them of shifts in health attitudes and management, and labeling outcomes as desired or undesired to evaluate treatments and strategies. These capabilities can be easily implemented using the previously elaborated tag-based design for linking records. To increase the awareness and prioritization of collections, we previously proposed collection descriptors such as topic, urgency, currency, and sentiment . Participants found value in these descriptors but expressed a need for additional ones that can be classified as patient specified and data driven . Future tools may include patient-specified descriptors for clarity (eg, is the diagnosis clear?), stability (eg, is the treatment working consistently?), severity (eg, is there a significant medical risk?), and progress markers (eg, is the issue substantially resolved?). Data-driven descriptors could be derived from the collection data, indicating the time span (from the oldest to the latest record) or listing the physicians involved (the providers and physicians the records came from). Both types of descriptors should be optional for patients to use as needed. Providing an inner structure, enabling annotations, and describing collections can improve information access and expedite decision-making. Powerful search engines can use these metadata to allow patients direct and easy access not only to individual collections but also to their specific contents. While these features can enhance collections’ actionability, it is important to note that collections are not meant for making independent clinical decisions by patients. Collections should be verified by a physician to serve as reliable tools for sensemaking and health self-management. However, collections can always be invaluable tools for patients to understand their health; organize thoughts, hypotheses, and insights; and communicate effectively with their physicians. Collection Use for Patient-Provider Communication As observed in our findings, patients can use collections to prepare for a clinical visit by devising checklists and organizing thoughts supported by evidence. During the visit, collections can be used for note taking and, afterward, for recording reminders and follow-up actions . These uses indicate how collections can start addressing known challenges during clinical visits, such as problem presentation , information retention , setting common ground, aligning goals, and understanding instructions . To effectively tackle such challenges, the Collections feature should support richer note capturing and collaborative data analysis in a collocated setting . Future improvements in capturing PGD could make collections more appealing to physicians. For physicians, PGD play a crucial role in understanding the boundaries and context for accurate diagnosis and optimal treatment . However, physicians often face problems with PGD, such as incomplete data, inconsistent data structures, and insufficient time for reviewing due to poor organization . These issues arise because patients use disjointed platforms to log their data, lacking consistent models for logging different types of data , and face challenges in efficiently using these platforms . Collections can help overcome these issues by providing a single platform for logging PGD for various health issues in a simple, universal way that allows for flexibility, organization, and standardization. An alternative approach to enhancing collections as a communication tool and fostering physician collaboration in their creation and verification is to introduce them as a shared resource similar to Google Docs . While this may seem unconventional, it builds on the principles of OpenNotes . OpenNotes provides access to and transparency regarding clinical notes, enabling patients to improve their treatment and EHR data quality by taking an active role in detecting errors, raising concerns, asking questions, or seeking clarifications . Similar to OpenNotes, shared collections would follow the principle of asynchronous communication and transparency. However, shared collections could eliminate the expressiveness constraints and lack of efficient ways to provide granular and tailored context observed in existing messaging systems . In addition, shared collections would enable direct editing of underlying data in collaborative ways, minimizing communication overhead. Moreover, shared collections would introduce a new communication channel between patients and providers outside the traditional patient portal. Synchronizing the digital traces of care in collections with the provider’s EHR system to avoid discrepancies and legal issues should be a top consideration in future design iterations of the shared collection concept. A Glimpse Into the Future: Collections and Generative Artificial Intelligence In the future, we should explore the potential of generative artificial intelligence (GenAI) models to support patient sensemaking through collections. Tools such as ChatGPT and Med-PaLM , which have demonstrated substantial medical knowledge , can replace the need for custom-made machine learning algorithms for knowledge-intensive tasks. In particular, GenAI tools can aid in automatic and iterative collection construction with explanations and guidance. They can analyze the data within collections for insights, including medical records, PGD, notes, and tags. In addition, GenAI can assist in composing case narratives and talking points for clinical encounters. By offering this level of automation, GenAI can help tackle the significant knowledge challenges while lowering the labor barrier for patients’ sensemaking activities. Using GenAI models, collection construction could rely on natural language instructions such as the following: “Group my EHR data by condition,” “Find all records related to my bronchitis,” or “Identify records that don’t belong to this collection and those that are missing.” GenAI models could also deliver context, explaining why certain records are included or excluded and providing educational material such as term definitions and clinical implications. In addition, patients can issue commands for identifying relationships within their annotated data, such as “List all triggers for my headaches over the last year.” Finally, they can ask for help in constructing case presentations for clinical visits (eg, “Based on my ‘High Blood Pressure’ collection notes, write a 100-word summary”). To be useful for sensemaking, GenAI tools do not need to achieve complete accuracy. While still striving for maximum reliability, their main value should come from providing an environment that enables and encourages patients to refine artificial intelligence–generated outputs. As such, the contribution of GenAI toward sensemaking would be evaluated on its ability to help the patient efficiently produce a satisfactory solution with minimal physician input. Finally, existing approaches for supporting sensemaking through search and interactive visualizations should not be disregarded. Exploring the integration of GenAI, search, and visualization is a prudent strategy as different sensemaking tasks related to collection assembly, editing, and analysis may require diverse approaches based on complexity, patient skills, and artificial intelligence reliability. On a broader scale, this work contributes to patient-centered care. This is achieved by demonstrating potential to enhance patients’ grasp of their health, encourage self-advocacy, and improve patient-provider communication. More accurately, there are several concrete contributions of our work that can be considered as proxies toward achieving the aforementioned objectives: (1) encouraging patient ownership of their EHR data by organizing them into personalized, health issue–based collections; (2) understanding patients’ perceptions and preferences for creating, building, and using these collections; and (3) offering design insights for automating collections, integrating rich PGD, enhancing access to collection contents, and using collections to facilitate patient-provider communication. Going forward, we will situate our findings within a sensemaking framework and discuss contributions related to 3 key patient needs: increasing awareness through independent health sensemaking, proactivity through efficient action taking, and self-advocacy through incorporating evidence-supported patient perspectives into patient-provider communication. We will elaborate on how collections can meet these needs and offer design implications to enhance their capabilities. To explain the collections’ role in sensemaking, we used the model by Pirolli and Card , which divides the sensemaking process into 2 subcycles: the foraging loop and the sensemaking loop. On the basis of this model, collections can be described as a space for assembling relevant data about a topic, finding relationships between them, and storing outcomes from the sensemaking. In the foraging loop, patients gather relevant records to answer questions such as the following—“Is there a relationship between my weight and blood pressure?”—and save them in a collection, such as Weight vs. Blood Pressure. In the sensemaking loop, patients identify information relationships within the collection that they capture in notes, such as instances where there was co-occurrence of high blood pressure and high body mass. These notes help argue hypotheses such as the following: “My blood pressure is high when I’m overweight.” The outcome of this sensemaking process could be a comprehensive note for a clinical visit. While, in their current form, collections respond to the needs of the sensemaking model by Pirolli and Card , improvements can be made to make this more efficient. This study revealed that medical records alone are not enough for reliable sensemaking. Adding PGD such as symptoms, measurements, outcomes, and everyday events is essential for creating comprehensive collections. The foraging loop can be made less laborious and time-consuming if there are additional visual cues, medical explanations, filtering capabilities, and automatic support to improve the relevance and reduce the effort of assembling collections. The sensemaking loop could be improved by adding more schematization capabilities such as annotating medical records and PGD to identify patterns later (eg, symptom triggers, medication effects, and correlations) and grouping records within collections, labeling those groups, and establishing group relationships with explanations (eg, linking “cholesterol lab results” with “food intake” as “food effects on cholesterol” in a High Blood Pressure collection). According to our study, there are 4 main factors that can influence the reliability of the collections: robust coverage of health issues, provision of PGD, grouping and linking of medical records and PGD within a collection, and verifying the contents of the collections. This reliability is related to collections’ capability to aid in creating personalized but realistic health issue narratives, support self-management, and stimulate awareness and proactivity. Robust Coverage of Health Issues: Relevance Assessment Collections should ensure that patients can create collections for their most important health issues to support awareness and proactivity. Participants desired visual cues, explanations, and automatic support to determine which collections to create and what records to include. While participants found the context-capturing data exploration using Sliding Tabs and Timeline convenient, they needed more to identify relevant records quickly. Future tools could incorporate Accordions that summarize record types, graph values over time, and highlight abnormal or extreme values. In addition, patients should be able to expand individual records to see explanations of clinical terms and clinical meaning interpretations. While there is existing work related to visualizing time series of EHR data and automatic provision of short explanations , this study shows the need for combining them in a new way to support relevance assessment for a novel purpose—constructing collections. Finally, patients should be able to order and filter records by attribute for quicker browsing and bulk addition to a collection. Automatic support should also be provided for creating and building collections to save time and ensure robust, reliable coverage of health issues. Our previous work highlighted the need for automating collections , and this study highlights a clear preference for automatically grouping medical records by clinical meaning—whether thematically or guided by patient input . Thematic collections would be those that tie records together based on conditions (bronchitis or diabetes) and procedures (stent placement or appendix removal) or with respect to organs (heart or kidneys) or organ systems (cardiovascular or renal). Alternatively, patients could specify a seed by setting parameters such as the collection’s name, tags, or initial records. The system can offer candidate records to include or delete with confidence scores and explanations. Patients could then refine these system-generated collections by adding or removing records, PGD, and tags. Addressing automatic support for collections may be challenging due to subtle relationships between medical records . However, starting with easier constructs such as time stamps (eg, medical records from the same day, week, or month), FHIR links (eg, medical records from the same encounter or physician), abnormal values based on well-established clinical guidelines (eg, high or low blood pressure or cholesterol), test findings (eg, positive and negative), and patient tags (eg, triggers or pivotal events) is a feasible approach. Provision of PGD: Sensemaking Data Completeness Collections should include PGD to improve data completeness for sensemaking. Previous research has suggested that maintaining consistent PGD logging over time is difficult . However, this should not be considered a barrier or a prerequisite for the collections’ success. Patients’ motivation and preferences for PGD logging intensity vary based on their disease self-management state . When setting goals and learning strategies, patients prefer meticulous data collection. Once goals and strategies are in place, logging intensity decreases. In addition, if physicians require PGD logging for treatment planning, patients are motivated to engage in it . Thus, collections should enable flexible and efficient PGD logging. Disease-specific contexts such as irritable bowel syndrome , diabetes , and migraine have explored health sensemaking without focusing on diverse data types. This contrasts with patients’ desire to log PGD for various medical issues within a single application using a universal logging model for different observations . To address this issue, we propose a straightforward workflow where patients initiate free-text entries and use tags to specify the type, quality, or other details. This mechanism allows for quick data capture and embellishment at convenient times. Tags can classify PGD as clinical observations , everyday life events , or notes . Further specification can be added using tags such as symptom , measurement , treatment , and outcome for observations; meal , exercise , meeting , and deadline for life events; and context , personal note , and visit note for notes. Additional tags such as absent , normal , high , low , extreme , improvement , deterioration , pivotal event , trigger , or relaxer can be used for further detail that captures the quality and importance of the logged data. In addition to these system-offered tags, patients can also create their own custom tags for better personalization. Grouping and Linking Within Collections: Schematization Capabilities Patients need to connect medical records and PGD within collections for easier sensemaking. Future tools should add structure by enabling record grouping and linking of groups or individual records. This helps highlight important subsets of records and trace major conclusions as collections grow. We recommend using a simple yet powerful tagging concept. Records sharing the same tag can form a group, whereas links between groups can be specified using related tags. The same mechanism can link individual records with other records or groups, providing nuanced sensemaking. This approach aligns well with the proposed PGD tagging model that can be applied to medical records as well. Collection Verification Collections represent the patient’s personalized perception of their health and issues. As such, they should undergo occasional verification by the patient’s physician for safe decision-making and action taking. While collection verification may add to the physician’s workload, it can inspire and enable patients to manage health issues more independently. That said, patients should consider the physician’s workload before requesting verification . Future tools could allow for the conversion of a collection into a well–laid-out PDF document capturing all its contents. This document can be printed for review during a clinical visit or shared as a PDF attachment in the patient portal for verification at the physician’s convenience. Collections should ensure that patients can create collections for their most important health issues to support awareness and proactivity. Participants desired visual cues, explanations, and automatic support to determine which collections to create and what records to include. While participants found the context-capturing data exploration using Sliding Tabs and Timeline convenient, they needed more to identify relevant records quickly. Future tools could incorporate Accordions that summarize record types, graph values over time, and highlight abnormal or extreme values. In addition, patients should be able to expand individual records to see explanations of clinical terms and clinical meaning interpretations. While there is existing work related to visualizing time series of EHR data and automatic provision of short explanations , this study shows the need for combining them in a new way to support relevance assessment for a novel purpose—constructing collections. Finally, patients should be able to order and filter records by attribute for quicker browsing and bulk addition to a collection. Automatic support should also be provided for creating and building collections to save time and ensure robust, reliable coverage of health issues. Our previous work highlighted the need for automating collections , and this study highlights a clear preference for automatically grouping medical records by clinical meaning—whether thematically or guided by patient input . Thematic collections would be those that tie records together based on conditions (bronchitis or diabetes) and procedures (stent placement or appendix removal) or with respect to organs (heart or kidneys) or organ systems (cardiovascular or renal). Alternatively, patients could specify a seed by setting parameters such as the collection’s name, tags, or initial records. The system can offer candidate records to include or delete with confidence scores and explanations. Patients could then refine these system-generated collections by adding or removing records, PGD, and tags. Addressing automatic support for collections may be challenging due to subtle relationships between medical records . However, starting with easier constructs such as time stamps (eg, medical records from the same day, week, or month), FHIR links (eg, medical records from the same encounter or physician), abnormal values based on well-established clinical guidelines (eg, high or low blood pressure or cholesterol), test findings (eg, positive and negative), and patient tags (eg, triggers or pivotal events) is a feasible approach. Collections should include PGD to improve data completeness for sensemaking. Previous research has suggested that maintaining consistent PGD logging over time is difficult . However, this should not be considered a barrier or a prerequisite for the collections’ success. Patients’ motivation and preferences for PGD logging intensity vary based on their disease self-management state . When setting goals and learning strategies, patients prefer meticulous data collection. Once goals and strategies are in place, logging intensity decreases. In addition, if physicians require PGD logging for treatment planning, patients are motivated to engage in it . Thus, collections should enable flexible and efficient PGD logging. Disease-specific contexts such as irritable bowel syndrome , diabetes , and migraine have explored health sensemaking without focusing on diverse data types. This contrasts with patients’ desire to log PGD for various medical issues within a single application using a universal logging model for different observations . To address this issue, we propose a straightforward workflow where patients initiate free-text entries and use tags to specify the type, quality, or other details. This mechanism allows for quick data capture and embellishment at convenient times. Tags can classify PGD as clinical observations , everyday life events , or notes . Further specification can be added using tags such as symptom , measurement , treatment , and outcome for observations; meal , exercise , meeting , and deadline for life events; and context , personal note , and visit note for notes. Additional tags such as absent , normal , high , low , extreme , improvement , deterioration , pivotal event , trigger , or relaxer can be used for further detail that captures the quality and importance of the logged data. In addition to these system-offered tags, patients can also create their own custom tags for better personalization. Patients need to connect medical records and PGD within collections for easier sensemaking. Future tools should add structure by enabling record grouping and linking of groups or individual records. This helps highlight important subsets of records and trace major conclusions as collections grow. We recommend using a simple yet powerful tagging concept. Records sharing the same tag can form a group, whereas links between groups can be specified using related tags. The same mechanism can link individual records with other records or groups, providing nuanced sensemaking. This approach aligns well with the proposed PGD tagging model that can be applied to medical records as well. Collections represent the patient’s personalized perception of their health and issues. As such, they should undergo occasional verification by the patient’s physician for safe decision-making and action taking. While collection verification may add to the physician’s workload, it can inspire and enable patients to manage health issues more independently. That said, patients should consider the physician’s workload before requesting verification . Future tools could allow for the conversion of a collection into a well–laid-out PDF document capturing all its contents. This document can be printed for review during a clinical visit or shared as a PDF attachment in the patient portal for verification at the physician’s convenience. While annotating PGD is known to aid learning and disease self-management , our findings reveal that annotations can also enhance EHR data, creating synergy with PGD. Participants expressed a desire to annotate their data for various purposes: identifying triggers to avoid or encourage certain behaviors, marking pivotal events to remind them of shifts in health attitudes and management, and labeling outcomes as desired or undesired to evaluate treatments and strategies. These capabilities can be easily implemented using the previously elaborated tag-based design for linking records. To increase the awareness and prioritization of collections, we previously proposed collection descriptors such as topic, urgency, currency, and sentiment . Participants found value in these descriptors but expressed a need for additional ones that can be classified as patient specified and data driven . Future tools may include patient-specified descriptors for clarity (eg, is the diagnosis clear?), stability (eg, is the treatment working consistently?), severity (eg, is there a significant medical risk?), and progress markers (eg, is the issue substantially resolved?). Data-driven descriptors could be derived from the collection data, indicating the time span (from the oldest to the latest record) or listing the physicians involved (the providers and physicians the records came from). Both types of descriptors should be optional for patients to use as needed. Providing an inner structure, enabling annotations, and describing collections can improve information access and expedite decision-making. Powerful search engines can use these metadata to allow patients direct and easy access not only to individual collections but also to their specific contents. While these features can enhance collections’ actionability, it is important to note that collections are not meant for making independent clinical decisions by patients. Collections should be verified by a physician to serve as reliable tools for sensemaking and health self-management. However, collections can always be invaluable tools for patients to understand their health; organize thoughts, hypotheses, and insights; and communicate effectively with their physicians. As observed in our findings, patients can use collections to prepare for a clinical visit by devising checklists and organizing thoughts supported by evidence. During the visit, collections can be used for note taking and, afterward, for recording reminders and follow-up actions . These uses indicate how collections can start addressing known challenges during clinical visits, such as problem presentation , information retention , setting common ground, aligning goals, and understanding instructions . To effectively tackle such challenges, the Collections feature should support richer note capturing and collaborative data analysis in a collocated setting . Future improvements in capturing PGD could make collections more appealing to physicians. For physicians, PGD play a crucial role in understanding the boundaries and context for accurate diagnosis and optimal treatment . However, physicians often face problems with PGD, such as incomplete data, inconsistent data structures, and insufficient time for reviewing due to poor organization . These issues arise because patients use disjointed platforms to log their data, lacking consistent models for logging different types of data , and face challenges in efficiently using these platforms . Collections can help overcome these issues by providing a single platform for logging PGD for various health issues in a simple, universal way that allows for flexibility, organization, and standardization. An alternative approach to enhancing collections as a communication tool and fostering physician collaboration in their creation and verification is to introduce them as a shared resource similar to Google Docs . While this may seem unconventional, it builds on the principles of OpenNotes . OpenNotes provides access to and transparency regarding clinical notes, enabling patients to improve their treatment and EHR data quality by taking an active role in detecting errors, raising concerns, asking questions, or seeking clarifications . Similar to OpenNotes, shared collections would follow the principle of asynchronous communication and transparency. However, shared collections could eliminate the expressiveness constraints and lack of efficient ways to provide granular and tailored context observed in existing messaging systems . In addition, shared collections would enable direct editing of underlying data in collaborative ways, minimizing communication overhead. Moreover, shared collections would introduce a new communication channel between patients and providers outside the traditional patient portal. Synchronizing the digital traces of care in collections with the provider’s EHR system to avoid discrepancies and legal issues should be a top consideration in future design iterations of the shared collection concept. In the future, we should explore the potential of generative artificial intelligence (GenAI) models to support patient sensemaking through collections. Tools such as ChatGPT and Med-PaLM , which have demonstrated substantial medical knowledge , can replace the need for custom-made machine learning algorithms for knowledge-intensive tasks. In particular, GenAI tools can aid in automatic and iterative collection construction with explanations and guidance. They can analyze the data within collections for insights, including medical records, PGD, notes, and tags. In addition, GenAI can assist in composing case narratives and talking points for clinical encounters. By offering this level of automation, GenAI can help tackle the significant knowledge challenges while lowering the labor barrier for patients’ sensemaking activities. Using GenAI models, collection construction could rely on natural language instructions such as the following: “Group my EHR data by condition,” “Find all records related to my bronchitis,” or “Identify records that don’t belong to this collection and those that are missing.” GenAI models could also deliver context, explaining why certain records are included or excluded and providing educational material such as term definitions and clinical implications. In addition, patients can issue commands for identifying relationships within their annotated data, such as “List all triggers for my headaches over the last year.” Finally, they can ask for help in constructing case presentations for clinical visits (eg, “Based on my ‘High Blood Pressure’ collection notes, write a 100-word summary”). To be useful for sensemaking, GenAI tools do not need to achieve complete accuracy. While still striving for maximum reliability, their main value should come from providing an environment that enables and encourages patients to refine artificial intelligence–generated outputs. As such, the contribution of GenAI toward sensemaking would be evaluated on its ability to help the patient efficiently produce a satisfactory solution with minimal physician input. Finally, existing approaches for supporting sensemaking through search and interactive visualizations should not be disregarded. Exploring the integration of GenAI, search, and visualization is a prudent strategy as different sensemaking tasks related to collection assembly, editing, and analysis may require diverse approaches based on complexity, patient skills, and artificial intelligence reliability. This study has several limitations. First, the cohort skewed younger, likely due to recruitment via Craigslist (less popular with older adults) and the complexity of the remote setup. Second, participants used data from a fictitious patient, which may have reduced their motivation to learn the app and their ability to suggest real-life use cases. Third, participants had limited time to learn how to interact with collections, possibly affecting their perceptions of usability and utility. Future studies should have participants use their own data with automatic interaction logging. Despite these limitations, this study provided valuable insights into designing patient-facing sensemaking tools for organizing and augmenting EHR data. Collections can potentially improve patient-centered care by involving patients more in decision-making and encouraging self-advocacy. Current assumptions often expect patients to have the necessary skills, tools, and motivation. We believe that collections can lower these barriers, encouraging patients to increase engagement with their health data, better educate themselves , and communicate more effectively with their care providers. Our study suggests that EHR data can be better used and more useful for patients through improved organization and annotation . This approach can incentivize patients to engage more deeply with their EHR data, develop insights, and reflect on their experiences. Patients felt that this empowered their awareness, resourcefulness, and proactivity regarding health issues, making them more prepared and better informed for clinician interactions. These findings support our premise that collections are a crucial step toward patient empowerment and self-advocacy . With appropriate improvements, collections can enhance patients’ expertise by facilitating sensemaking activities and enabling insightful discussions with their physicians. First, collections motivate patients to construct health models based on their issues and ongoing problems. Second, patients gain medical education by actively participating in the evolution of collections through independent or system-assisted assembly and editing. Finally, patients acquire additional medical knowledge by engaging in meaningful discussions with their physicians and considering their feedback on collection verification. Our study highlighted the importance of integrating PGD with EHR data. We envision a synergy in which patients use clinical data as a foundation, augmenting them with their observations, notes, and annotations to create personalized health narratives that support better health management and provider communication. In the future, we should explore GenAI models to support patient sensemaking through collections. These models could help patients build collections, analyze the data within them, and produce health narratives more efficiently. Such enhancements may also reduce physicians’ workload for verifying collection contents, leading to more focused, evidence-driven discussions during visits. Promising ideas from this work should be advanced carefully, with gradual design improvements tested in real-life settings. Respecting existing clinical practices and workflows can facilitate quicker adoption and more significant changes in the future. We believe that collections can revolutionize how patients interact with their medical records and communicate with their providers. |
Subsets and Splits