Nine of these subjects continued to lose weight after stopping semaglutide. Notably, we observed highly heterogeneous responses of the individual subjects, both, while on treatment with semaglutide and in the post-treatment period. These findings indicate the potential of metformin therapy in stabilizing weight fluctuation. In the present study, all models found time-dependent associations of compensatory insulin secretion and of weight change from baseline with the risk of developing diabetes. In all 3 treatment groups, weight loss significantly improved compensatory insulin secretion (increased the distances away from the line) and with that the degree of glucose homeostasis which resulted in lower levels of glycemia. This suggested that improvements in insulin sensitivity and beta-cell function were more prominent from baseline to year 1; however, as for weight loss, the effects of both ILS and metformin were attenuated after year 1. As expected, the relationship between insulin secretion and sensitivity (ie, beta-cell function) changed over time for ILS and metformin participants, but not for those in placebo. Over time, the beneficial changes in plasma glucose, serum insulin, insulin sensitivity, and insulin secretion compensation followed a similar pattern to that of weight loss, suggesting that weight loss drives such changes. Comprehensive analysis of metformin effect on BMI. Comprehensive analysis of metformin effect… PRISMA flow diagram detailing process of study selection for the meta-analysis. Model 2 includes time-dependent distances away from the SMA line and time-dependent distances along the SMA line in addition to the covariates included in Model 1. Hazard ratios for baseline weight and weight gain from baseline are calculated per 5-kg decrease. Hazard ratios for distances away from the line and along the line are calculated per 1 baseline SD. Change in weight and its longitudinal effect on changes in position with respect to baseline curve. For year 1, predicted distances away from and along the baseline SMA line and their 95% CIs are presented elsewhere (Table S4 (16)). Thus, with weight loss, insulin sensitivity and insulin secretion improved at years 1, 2, and 3.These measures were calculated for sets of intravenous and oral indices of insulin secretion and sensitivity.The weight trajectories of individuals are graphed in Figure 2B.Baseline covariates included age, sex, race/ethnicity, and DPP treatment assignment.At year 1, the relationship between insulin sensitivity and secretion was examined within each treatment among 3 groups based on weight change from baseline.At week 16, semaglutide was discontinued. Time-dependent Associations of Distances Away From the Line and Along the Line With Type 2 Diabetes Further studies are needed to evaluate the long-term weight loss in patients with euglycemia. Mean and categorical weight losses were compared between patients with and without T2DM/PreDM. Patients who received metformin as a sole pharmacotherapy were identified, and data pertaining to demographics, medications, comorbidities, and weight changes during 1-year follow-up were obtained from their electronic medical records. Demographic and clinical details of study patients are included in Supplementary Table 3. This might reflect beneficial effects of these 2 interventions on compensatory insulin secretion and insulin secretory demand that are unrelated to weight change. ILS participants experienced the most beneficial changes over time, followed by metformin participants; such findings are consistent with the reduced incidence of diabetes in both groups. Model 3 includes baseline weight and time-dependent weight loss from baseline in addition to the covariates included in Model 1. Model 1 is the baseline model and includes sex, age at baseline, race/ethnicity, and treatment group and is the reference model used to determine the proportion of treatment effects mediated by the time-dependent covariates. According to the TVCMs, the changes in compensatory insulin secretion and the secretory demand combined mediated 44% and 63% of the ILS and metformin effects, respectively, while in the JMs they mediated 92% of the ILS effect and completely mediated the metformin effect. Insulin secretion levels by themselves (CIR30, not adjusted for insulin sensitivity) did not differ between treatments over time (Fig. 2G), while insulin sensitivity (ISI0) was highest at all time-points after randomization for ILS participants and lowest for placebo (Fig. 2H). ILS and metformin participants lost weight at year 1; over time, ILS participants had the highest weight loss followed by metformin participants, while placebo participants gained weight on average (Fig. 2A). Weight change from baseline was included as a time-dependent covariate in the multivariate linear mixed-effects model to investigate its longitudinal associations with the trajectories of these distances, when adjusted for its cross-sectional effect and several baseline covariates (sex, age, race/ethnicity, treatment assignment, weight). In participants with the same secretion compensation levels (same distances away from the baseline SMA line), the distances along the baseline SMA line allow the quantification of increase in insulin secretion per decrease in insulin sensitivity (9). We also studied whether weight change from baseline to years 1, 2, and 3 affected the relationship between insulin secretion and sensitivity within each treatment group. 3. Endocrine parameters For distance away from the baseline SMA line, Model 1 included sex, age at baseline, race/ethnicity, treatment, baseline distance away from the baseline SMA line, and baseline weight; Model 2 included weight change at year of measurement in addition to the variables in Model 1. Similar patterns for these curves were observed at years 2 and 3 (16), where Group 1 also included participants who lost the most weight, Group 2 those who lost the least weight, and Group 3 those who gained weight by the time of each of these examinations. Figure 4A includes data from all examinations in participants without diabetes, whether they subsequently developed diabetes or not. Withdrawal of semaglutide is frequently followed by weight regain due to compensatory biological changes that prevent the maintenance of long-term weight loss. Patients with severe insulin resistance lost significantly more weight as compared to insulin sensitive patients. A few small trials showed heterogenous effects on liver parameters in patients with NAFLD treated with MTF compared to placebo/control. The current results suggest that this success is achieved through improved insulin sensitivity (decreased insulin secretory demand) coupled with increased insulin secretion relative to demand. Our results extend previous DPP analyses that concluded that interventions to reduce the risk of diabetes should aim at weight loss as the primary determinant of success (12). 2 Cardiometabolic parameters After adjusting for plate ID, study centre ID and DNA extraction batch, the mtDNA-CN was significantly correlated with gender, age, white blood cell count, red blood cell count, and platelets count (Supplementary Table 2), in line with previous reports22,23. Since some patients were treated with other add-on oral hypoglycaemic agents after 24 weeks, the main end point was taken at 24-week. All the individuals involved in this study provided written informed consent for the main study and subsequent genetic investigations. Each study centre obtained institutional review board approval. Funnel plot of weighted mean difference in BMI. Funnel plot of weighted mean… BMI, body mass index; IFG, impaired… To investigate whether the relationship between insulin secretion and sensitivity changed over time, we compared the lines describing this relationship at baseline and at years 1, 2, and 3 separately for each treatment group. First, we expanded the work of Kitabchi et al (11) by describing the longitudinal relationship between insulin secretion and sensitivity in DPP participants by treatment group, not only at baseline and year 1, but also at years 2 and 3 after randomization. ILS participants had the biggest movement to the right of the baseline curve after 1 year of treatment (11), and among them weight loss was the primary factor contributing to reduced risk of diabetes (12). Six subjects started to regain weight after stopping semaglutide, but after 2 years their weight was still lower than at baseline. In 6 subjects, body weight remained stable after discontinuation of semaglutide. Until today, the longest observational period for semaglutide treatment for obesity is 2 years (STEP 5). Figure 1C and 1D illustrate how the distances away from the baseline SMA line and along the baseline SMA line were calculated for all data points of a participant. For each participant, we calculated perpendicular distances for all data points (ie, at baseline and at years 1, 2, and 3) from the baseline SMA line and from its orthogonal line and allowed these distances to be positive or negative to take on the same direction as the residuals (9). We estimated longitudinal associations of weight change with trajectories of insulin secretion compensation and secretory demand estimated as previously described (9). Weight loss caused by common diabetes drug tied to "anti-hunger" molecule in study This study investigated the progression of insulin secretion deficiency and decreased insulin sensitivity during the transition from IGR to diabetes and allowed us to understand how these factors interact with one another and what affects their interaction. By contrast, at the first examination with diabetes, insulin sensitivity and secretion had moved to the left and down, in other words, worsened, from their baseline relationship in all treatment groups (Fig. 4B). 1A, 1I, and 1J indicated that these 2 treatments decreased weight from baseline and insulin secretory demand and increased compensatory insulin secretion. A fasting blood sample was drawn for determination of glucose luteinizing hormone (LH), follicle-stimulating hormone (FSH), androstenedione, dehydroepiandrosterone sulphate (DHEAS), total and free testosterone (T), and lipids, followed by a 2-h OGTT. Patients were monitored for weight changes over 6 months. Additionally, we included 45 untreated patients as controls. There was no interaction by baseline BMI, MTF dose or duration, nor presence or absence of NAFLD. In all treatment groups, a 5-kg and a 10-kg decrease in weight would move the point to the right of the overall baseline curve fitted to all participants. Then, at time of diagnosis, they had moved further below and to the left, representing deterioration in both insulin sensitivity and compensatory insulin secretion. Generally, the points representing insulin secretion and sensitivity at baseline were below and to the left of the overall baseline curve. When adding these variables, both models indicated that an increase in compensatory insulin secretion significantly reduced diabetes risk at any point in time.Multicenter, placebo-controlled trial of lorcaserin for weight management.Increasing levels of insulin secretion compensation significantly reduced risk of diabetes while increasing levels of secretory demand increased it.Metformin-induced increases in GDF15 are important for suppressing appetite and promoting weight loss.Six subjects started to regain weight after stopping semaglutide, but after 2 years their weight was still lower than at baseline. They also used the ratio of insulin secretion divided by insulin sensitivity as a beta-cell demand index (BCDI), which estimated insulin secretion per degree of insulin sensitivity. They measured the homeostatic response using the disposition index (DI), the product of estimated insulin secretion and insulin sensitivity. The relationship between these 2 variables is usually described by a curve that allows an assessment of insulin secretion relative to insulin sensitivity and provides a measure of beta-cell function (6). Short-term low dose semaglutide treatment also resulted in normalization of fasting blood glucose in 80% of PCOS women with impaired fasting glucose (33). After 3 months, almost 80% of the obese PCOS patients in the study achieved a minimum 5% reduction in body weight. AOMs, including liraglutide and semaglutide, combined with lifestyle intervention could offer an upgraded treatment strategy for patients with PCOS and obesity (28). Insulin resistance (IR) occurs early in the course of PCOS and is frequently present even in the absence of obesity, accounting for 60-80% of women with PCOS (26). Despite these issues, all participants continued with metformin therapy. Two years after discontinuation of semaglutide (timepoint 3), 2 subjects reported occasional liquid stools, 2 experienced occasional nausea and dyspepsia and 1 had occasional bloating. None of the participants discontinued treatment due to gastrointestinal events or other reasons. After starting semaglutide treatment, 11 subjects reported experiencing nausea and dyspepsia, with 3 of them also suffering from constipation. In contrast, no significant association was detected between mtDNA-CN and weight loss in the acarbose arm. Multiple linear regression models showed that for both drugs individuals with higher baseline weight, older age and being female tended to lose more weight (Table 2). Similar as described in the original publication, there was no statistically significant difference between the metformin arm and acarbose arm after data quality control. Baseline characteristics of the participants remained in this study were summarized in Supplementary Table 3. Among the main reasons for discontinuation were price of the drug and fear of long-term side effects. The authors claimed that this is particularly important given the lack of data on long-term safety (28). Even though GLP-1 RAs seem to have a great potential to improve obesity-related metabolic and reproductive complications in PCOS, their use is currently not widely acknowledged and accepted in this population. In addition, basal insulin levels decreased, and insulin resistance (HOMA-IR) improved. Metformin was approved for the treatment of diabetes by the US Food and Drug Administration in 1994. Long noted that, while semaglutide drugs are injected into the bloodstream, metformin is an oral drug that is already prescribed to millions of people. Further research revealed that lac-phe is made by intestinal epithelial cells in the animals; blocking the ability of mice to make lac-phe erased the appetite suppression and weight loss previously observed. They saw significant increases in the levels of lac-phe in people after metformin compared with their levels before treatment. Baseline body mass index (BMI) was chosen as a self-control to compare the changes in BMI of different populations before and after treatment. We also assessed the rate of metformin discontinuation due to side effects, lack of efficacy or other reasons in the entire cohort. And K.X.Z. are the guarantor of this work and, as such, had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis. For the choice of secondary weight loss options, pharmacotherapy that is mechanistically less dependent on mitochondrial function would be preferred to achieve significant and sustained weight loss. Intervention with the latest AOMs even results in 20% weight loss and is already approaching the effectiveness of bariatric surgery.The diverse approaches regarding the use of semaglutide for weight reduction highlight the necessity to guide physicians and standardize the long-term treatment regimen in anti-obesity medicine.Safety clinical assessment was performed at the beginning and the end of the treatment with semaglutide period.Glucose levels were determined using a standard glucose oxidase method (Beckman Coulter Glucose Analyzer, Beckman Coulter, Inc., CA, USA).In men, research suggests that metformin treatment may lead to a decrease in serum testosterone levels independent of blood glucose control.These results are consistent with previous DPP analyses showing that effects of the ILS and metformin interventions on reducing diabetes incidence were largely, but not entirely, attributed to weight loss (12, 13).While weight loss is achievable, weight maintenance remains a lifetime challenge (1).This meta-analysis aimed to summarize the weight loss effect of metformin quantitatively. At each time-period, average insulin sensitivity and beta-cell function were best for ILS participants, followed by participants in the metformin group which were better than for placebo participants. The curves representing this relationship for ILS and metformin participants at years 1, 2, and 3 after randomization were above and to the right of the baseline curves, indicating that these 2 treatments improved insulin sensitivity and beta-cell function. In both the TVCMs and the JMs, baseline weight and weight loss from baseline mediated over 49% of the effects of both treatments (53% for ILS and 50% metformin in the TVCM; 82% for ILS and 72% for metformin in the JM). The purpose of this manuscript is to review the mechanisms that underlie metformin’s effects on obesity. Categorical weight losses (≥5% and ≥10% of baseline weight) were also similar. "These findings suggest there may be a way to optimize oral medications to affect these hunger and energy balance pathways to control body weight, cholesterol and blood pressure. I think what we're seeing now is just the beginning of new types of weight loss drugs." "The fact that metformin and sprint exercise affect your body weight through the same pathway is both weird and interesting," Long said. "Until now, the way metformin, which is prescribed to control blood sugar levels, also brings about weight loss has been unclear," said Jonathan Long, PhD, an assistant professor of pathology. Data availability statement The cardiometabolic improvements seen during semaglutide treatment reverted towards the baseline in our study, similar to the findings in the STEP 1 study (10). There are some data that metformin can attenuate weight regain after weight loss and prolong the maintenance of a stable weight after weight loss in PCOS. Contradictory to clinical research data confirming benefits of long-term treatment, semaglutide is frequently used for a significantly shorter period of time in clinical practice. The treatment led to substantial, sustained weight loss versus placebo and maintained improvements in cardiometabolic parameters. In contrast, participants in our cohort only regained about one-third of their prior weight loss during the two-year period after discontinuing semaglutide treatment. In the STEP 1 study, participants regained two-thirds of their previous weight loss one year after discontinuing semaglutide therapy. Furthermore, also in pregnant women with PCOS, administration of metformin during pregnancy resulted in less weight gain during pregnancy and a lesser degree of weight loss in the first year postpartum compared to those who received placebo (20). Based on its mode of action and some clinical data (17–23), we hypothesized that metformin may partially prevent weight regain after weight loss induced by semaglutide. Weight change, cardiometabolic, and endocrine parameters were assessed 2 years after semaglutide discontinuation. Treatment with metformin 2000 mg/day and promotion of lifestyle intervention were continued during the 2-year follow-up period. At week 16, semaglutide was discontinued. A two-way ANNOVA was used to assess the correlation between mtDNA-CN and repeated measurements of weight loss at multiple time points. In addition to its efficacy in lowering blood glucose, emerging evidence has shown a modest beneficial effect on weight loss2,3. There is growing body of evidence that GLP-1 RAs demonstrate numerous metabolic benefits in PCOS, including a high effect on prediabetes remission rate, a significant reduction in atherothrombotic markers, and an improvement in lipid profile (29–32). Comparison of clinical parameters at baseline (Timepoint 1) with treatment phase (Timepoint 2) and off treatment phase (Timepoint 3). The percentage of weight loss was independent of age, sex or BMI. After conducting all previous analyses, we determined whether the longitudinal measurements of distances away from the baseline SMA line, distances along the baseline SMA line, and weight loss mediated the effects of ILS and metformin on the risk of diabetes using the causal steps approach (28, 29). We analyzed data from 2931 participants with baseline FPG Surrogate indices of insulin secretion and sensitivity were calculated from plasma glucose and serum insulin concentrations measured at fasting (G0, I0) and 30 minutes (G30, I30) during an OGTT. Two years after semaglutide withdrawal, women with PCOS and obesity that continued with metformin treatment regained one-third of their semaglutide-induced weight loss. At baseline (timepoint 1), at the end of semaglutide intervention (timepoint 2), and at the end of the study (timepoint 3), all participants underwent body weight assessment. To date, the amount of weight regain after semaglutide withdrawal in patients with obesity continuing metformin treatment, including obese women with PCOS, has not yet been evaluated. Metformin is also being explored in the management of obesity’s sequelae such as hepatic steatosis, obstructive sleep apnea, and osteoarthritis. Metformin is a first-line therapy for type 2 diabetes. In her spare time, she enjoys textile arts, experimenting with new recipes and hiking in beautiful northwestern Montana, where she was raised and now lives. These were compared by first checking if lines for the same treatment group but at different yearly visits shared a common slope with a likelihood ratio test (17); if so, we used a Wald test for common intercepts (17). By contrast with ordinary least squares regression, which minimizes the vertical distance between each point and the line (only accounts for variability in y), SMA regression minimizes the area of the right triangle defined by vertical and horizontal distances from each point to the line. Baseline covariates included age, sex, race/ethnicity, and DPP treatment assignment. Written informed consent was obtained from all participants and the studies were approved by each clinical center's institutional review board. Definitions of drug response In general, weight loss contributed to rightward movements from the curve describing beta-cell function at baseline, while weight gain resulted in leftward movements. In general, beneficial changes in most longitudinal variables, including weight loss, were greatest from baseline to year 1, and after year 1 they tended to return toward baseline. In the TVCM, all these covariates together (model 4) mediated 81% and 98% of the ILS and metformin effects, respectively, and in the JM, they completely mediated the effects of both treatments; once we accounted for all the covariates together, the hazard ratios for ILS and metformin treatment effects exceeded 1 but not significantly. All JMs indicated that ILS did not significantly reduce the risk of diabetes compared with placebo when adjusted for distances away and along the SMA line (Tables 3 and 4). Both models indicated that 5-kg decreases in weight from baseline significantly decreased diabetes risk. However, the exact nature of the mitochondrial interaction with metformin is still poorly characterized. Metformin has also been reported to promote mitochondrial fission to improve mitochondrial respiration, restore the mitochondrial life cycle, and alleviate hyperglycaemia in obesity by activating AMP-activated protein kinase (AMPK)9. With the biological mechanisms of metformin action not being fully elucidated, both clinical and genomic factors have been suggested to affect how an individual responds to metformin5. Metformin is globally recommended as the first-line oral glycaemic agent for type 2 diabetes1. Figure 4A shows the baseline SMA curve fitted to all participants and the geometric means of insulin secretion and sensitivity for each treatment group at each of the yearly visits before they developed diabetes.Impaired mitochondrial function plays an important role in the development of insulin resistance and type 2 diabetes10.However, after stopping AOM, weight regains due to the compensatory metabolic adjustments described above (7–10).The TVCMs assumed that the insulin secretion compensation (distances away from the SMA line) stayed constant in between visits, which is biologically unlikely, and approximated the longitudinal trajectories using a step function that does not capture trends between visits.As weight management increasingly becomes an integrated goal in the treatment of type 2 diabetes37, the observed association between mtDNA-CN and metformin-induced weight loss is of particular interest.In healthy individuals, insulin secretion increases to compensate for decreased insulin sensitivity which maintains normal glucose tolerance (4–6).The trajectory of body weight change after discontinuation of the currently most widely used AOM, the glucagon like peptide-1 receptor agonist (GLP-1 RA) semaglutide, was observed in two studies from the STEP program (9, 10).Model 3 includes baseline weight and time-dependent weight loss from baseline in addition to the covariates included in Model 1.Based on its mode of action and some clinical data (17–23), we hypothesized that metformin may partially prevent weight regain after weight loss induced by semaglutide. For this purpose, we fitted 2 linear regression models separately for each year with the distances away from the baseline SMA line at that year as outcomes. This participant had lowest insulin secretory demand at year 1 (most negative distance along the baseline SMA line) and highest secretory demand at baseline (least negative distance along the baseline SMA line). Points located below and to the right of the orthogonal line have negative distances along the baseline SMA line and points located above and to the left of the orthogonal line have positive distances along the baseline SMA line. Written informed consent was obtained from all patients18. Acarbose was started from 50 mg once a day and titrated to 100 mg three times a day over the same period. Add-on therapy with insulin secretagogues was given to those with inadequate glycaemic control after week 24. The repository includes data only from consenting participants at clinics with institutional review board approval to distribute data to the repository and does not include data from the American Indian centers. The primary outcome was diabetes development which was diagnosed using an annual oral glucose tolerance test (OGTT) or a semiannual FPG test using the American Diabetes Association 1997 criteria (2-hour plasma glucose ≥11.1 mmol/L or FPG ≥7.0 mmol/L), with confirmation by a repeat test. Second, we determined whether the relationship changed over time separately at years 1, 2, and 3 within each treatment group. Later, measures of insulin secretion compensation and insulin secretory demand that are highly correlated with DI and BCDI, respectively, but which do not require the assumption of a hyperbolic relationship inherent in the DI/BCDI, were investigated in this same population (9). Insulin secretion and sensitivity regulate glycemia, with inadequately compensated deficiencies leading to diabetes. E.V.A. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. The JM possibly corrected this shortcoming by estimating the longitudinal trajectories of compensatory insulin secretion. For both treatments, the curves for years 2 and 3 moved to the left of the curve for year 1 ending between the baseline and year 1 curves. Models 2, 3, and 4 were fitted using time-varying Cox models (TVCM) and joint models (JM) for time-to-event and multivariate longitudinal data. After estimating the mtDNA-CN from deep whole genome sequencing (WGS) data, we evaluate the association between mtDNA-CN and metformin response, with the acarbose arm of the trial as a comparator group. A post hoc analysis of data from the Metformin and AcaRbose in Chinese as the initial Hypoglycaemic treatment (MARCH) trial is conducted. Impaired mitochondrial function plays an important role in the development of insulin resistance and type 2 diabetes10. Mechanistically metformin is thought to act by modulating mitochondrial respiratory chain Complex I and mitochondrial glycerophosphate dehydrogenase to exert its pleiotropic effects on metabolism6,7,8. After a run-in 20-week treatment period with semaglutide, enrolled participants were re-randomized to continue treatment with semaglutide for 48 weeks or switch to placebo. A new generation of anti-obesity medications (AOMs) has revolutionized the way obesity is treated, enabling most patients to lose 10-20% of their body weight. At the end of the follow up, 84% of women had a lower body weight than at baseline. In the control (placebo) group, side effects occurred in 18% of participants receiving metformin and 19.2% without metformin. This analysis demonstrated that GI side effects occurred in 37.6% of participants receiving semaglutide plus metformin, compared to 40.3% of those receiving semaglutide alone. When interpreting the endocrine and cardiometabolic outcomes, it is important to highlight that participants in our cohort were already treated with metformin before the semaglutide intervention, therefore prior the baseline assessment. Contrary to other beneficial changes, HDL cholesterol has been reduced during semaglutide treatment. Two years after semaglutide cessation, all observed metabolic parameters reverted toward baseline (Table 1). Table 1 Comparison of clinical parameters at baseline (Timepoint 1) with treatment phase (Timepoint 2) and off treatment phase (Timepoint 3). Safety clinical assessment was performed at the beginning and the end of the treatment with semaglutide period. STEP 1 study extension complements the observations from STEP 4 (10). However, after stopping AOM, weight regains due to the compensatory metabolic adjustments described above (7–10). Weight management and improvement in overall health are fundamental goals in treating obesity. Together, they harness the full potential of biomedicine through collaborative research, education and clinical care for patients. "The magnitude of effect of metformin on lac-phe production in mice was as great as or greater than what we previously observed with exercise. If you give a mouse metformin at levels comparable to what we prescribe for humans, their lac-phe levels go through the roof and stay high for many hours." "Now we know that it is acting through the same pathway as vigorous exercise to reduce hunger. Understanding how these pathways are controlled may lead to viable strategies to lower body mass and improve health in millions of people." Kitabchi et al (11) described the relationship between insulin secretion and sensitivity in DPP at baseline and 1 year after randomization, finding that there were rightward shifts from the baseline curve in points representing median insulin secretion and sensitivity for participants assigned to ILS or metformin. Overall, during semaglutide treatment phase, from timepoint 1 to timepoint 2, women lost a significant amount of body weight. We aimed to explore changes in body weight, cardiometabolic and endocrine parameters in obese women with PCOS who continued metformin treatment 2 years after semaglutide cessation. Continuation with semaglutide resulted in continued weight loss over the following 48 weeks, resulting in 17.4% of net weight loss from baseline, while those switched to placebo regained 6.9% of body weight after the switch to placebo (9). It is unknown whether weight loss outcomes differ with metformin monotherapy in patients with obesity with or without type 2 diabetes (T2DM)/prediabetes (PreDM). W.Y.Y., H.L., L.Z., Y.W., X.Y.Z. and J.P.W. collected the data. The authors are very grateful to all participants who took part in these studies. As the current study is a post hoc investigation of a published trial, it is largely of a hypothesis generation nature. These rightward shifts were consistent with increased insulin sensitivity and improved beta-cell function and with the reduced incidence of diabetes for participants in these 2 treatment groups.All subjects were informed of the study’s aims and provided written consent before they were enrolled in the study.Baseline characteristics of the participants remained in this study were summarized in Supplementary Table 3.At the end of the study, 21 out of 25 subjects had lower body weight compared to baseline.Thus, within each treatment, the relationship differed based on weight change group at year 1; similar results were obtained for weight change groups at years 2 and 3 (see (16)).Over time, the beneficial changes in plasma glucose, serum insulin, insulin sensitivity, and insulin secretion compensation followed a similar pattern to that of weight loss, suggesting that weight loss drives such changes.The relationship between these 2 variables is usually described by a curve that allows an assessment of insulin secretion relative to insulin sensitivity and provides a measure of beta-cell function (6).In 6 subjects, body weight remained stable after discontinuation of semaglutide. E.V.A. managed the data, conducted statistical analysis, and wrote the first draft of the manuscript. The data (and samples) from the DPP were supplied by the NIDDK Central Repository. Previous DPP studies only investigated the relationship of interest at baseline and year 1, while we also studied it at years 2 and 3. Thus, the results shown apply only to examinations prior to the diagnosis of diabetes. Nevertheless, this group experienced a high incidence rate of diabetes (11 cases/100 person-years) (10). Metformin, a biguanide anti-hyperglycemic agent, is the first-line treatment for type 2 diabetes (11). Intervention with the latest AOMs even results in 20% weight loss and is already approaching the effectiveness of bariatric surgery. There is some evidence of not only intensified appetite post-weight loss but also of an increased preference for high-calorie foods (4). Insulin secretion compensation combined with insulin secretory demand and baseline weight combined with weight loss mediated the ILS and metformin effects on the risk of diabetes. Figure 6A-6C shows how a point representing a participant with average (geometric mean) insulin secretion and sensitivity at baseline in the ILS, metformin, and placebo interventions, respectively, would be predicted to move at 1 year after randomization depending on weight changes. For the distances along the baseline line (insulin secretory demand), adding weight loss to the models attenuated the effects of ILS on them in all 3 years; this also happened for metformin but to a lesser extent. Figure 4A shows the baseline SMA curve fitted to all participants and the geometric means of insulin secretion and sensitivity for each treatment group at each of the yearly visits before they developed diabetes. Our results show that patients with diabetes with a higher mtDNA-CN lost more weight upon metformin treatment. It reduces liver glucose production, enhances insulin sensitivity in body tissues (11), decreases hyperinsulinemia by reducing insulin resistance and increases the secretion of growth/differentiation factor 15 (GDF15), which suppresses appetite and reduces caloric intake (13–16). Improvements in cardiometabolic parameters including decrease in total and LDL cholesterol, fasting glucose, and glucose after OGTT that had been seen during semaglutide-treatment phase reverted towards baseline two years after semaglutide cessation. 25 women with PCOS and obesity, aged 33.7 ± 5.3 years (mean ± SD), were treated with once-weekly subcutaneous semaglutide 1.0 mg as an adjunct to metformin 2000 mg/day and lifestyle intervention for 16 weeks. We restricted analyses to such persons to avoid effects of changes in insulin dynamics caused by diabetes or its treatments. First, we fitted an SMA line that estimated the relationship between log-transformed insulin secretion and sensitivity at baseline. We divided participants with data at year 1 into 3 groups depending on their weight change from baseline ranging from participants who lost the most weight to those who gained weight. Surrogate indices of insulin secretion and sensitivity were calculated from plasma glucose and serum insulin concentrations measured at fasting (G0, I0) and 30 minutes (G30, I30) during an OGTT. Table 3 presents associations of time-dependent compensatory insulin secretion and insulin secretory demand with diabetes risk adjusted for sex, baseline age, race/ethnicity, and treatment for the TVCM and the JM. At each year, the distances away from the baseline SMA line (changes in compensatory insulin secretion) were mostly explained by the weight loss from baseline at that year. Insulin sensitivity for ILS participants was higher than for metformin and placebo participants at all time points after the baseline visit. Points representing average insulin secretion and sensitivity moved furthest to the right for ILS and metformin participants from baseline to year 1. Thus, tests for common slopes and intercepts suggested that the relationship between insulin secretion and sensitivity changed over time only for ILS and metformin participants. Here we evaluated whether whole blood derived mitochondrial DNA copy number (mtDNA-CN) is a biomarker of metformin response as measured by glucose reduction or weight loss.The DPP was a multicenter, randomized clinical trial that investigated the effects of an ILS and metformin on the prevention or delay of type 2 diabetes onset in participants with IGR (10).Both models found a time-dependent association between compensatory insulin secretion with the risk of diabetes, and suggested that at any point in time an increase in compensatory insulin secretion (ie, the distance away from the line) significantly decreased diabetes risk.Table 2 presents the results of the cross-sectional linear regression models for distances away from the baseline SMA line and distances along the baseline SMA line fitted at years 1, 2, and 3 to determine whether weight change explained the effects of ILS and metformin on both distances.The ability of metformin to reduce androgen levels in women with hyperandrogenic PCOS is more pronounced in non-obese women (36, 37).The detailed study protocol of the trial had been described previously18.In both the TVCMs and the JMs, baseline weight and weight loss from baseline mediated over 49% of the effects of both treatments (53% for ILS and 50% metformin in the TVCM; 82% for ILS and 72% for metformin in the JM). Three subjects experienced intermittent diarrhea, bloating and nausea before the initiation of semaglutide treatment (timepoint 1), but the symptoms were not severe enough to warrant discontinuation of metformin therapy. Two years after semaglutide cessation, 18 women had normal glucose tolerance, 3 had impaired fasting glucose, 3 had impaired glucose tolerance, and 1 developed type 2 diabetes. At the end of semaglutide therapy (on timepoint 2), all women had normal glucose tolerance. According to the OGTT at the beginning of the study (timepoint 1), 21 women had normal glucose tolerance (NGT), 3 had impaired fasting glucose (IFG), and 1 had impaired glucose tolerance (IGT). Here we evaluated whether whole blood derived mitochondrial DNA copy number (mtDNA-CN) is a biomarker of metformin response as measured by glucose reduction or weight loss. The most recent international guidelines recommend that when discussing the use of GLP-1 RA with women with PCOS, the potential side effects and the need for long-term use in weight management should always be considered in the shared decision making with the patient. Metformin in addition to lifestyle intervention has been used in PCOS over the past few decades (27), and should be considered irrespective of glucose homeostasis in all women with PCOS with BMI ≥ 25 kg/m2, because of its expected beneficial effects on metabolic outcomes (28). Improvements of cardiometabolic variables achieved during semaglutide treatment phase reverted to baseline, whereas the reduced free testosterone levels observed during semaglutide treatment did not significantly deteriorate after semaglutide discontinuation. In the Diabetes Prevention Program Outcome Study, modest weight loss with metformin was maintained over a 10-year follow-up period, in contrast to the original lifestyle group from Diabetes Prevention Program, which partly regained weight during follow up period (17, 18). At the molecular level, it seems that the mechanisms of action vary depending on the dose of metformin used and duration of treatment. Early evidence highlighted the liver as the major organ involved in the effect of metformin on reducing blood levels of glucose. This marker could potentially be used to guide treatment with metformin. However, there is considerable variability in how patients respond to metformin and few markers or tests are available to guide prescribing. For comparison of continuous variables in different time points, Friedman (comparison of three time points) and Wilcoxon signed-rank test for related samples with post hoc Bonferroni corrections for multiple comparisons (comparison of two time points) were used. Similar situations may have occurred for the other time-dependent covariates. The percentage effects mediated were calculated using HRs from the 4 models shown in Table 5. Time-varying Cox models assume that covariates only change at time of examination while joint models used a mixed model to estimate the trajectory over time of the time-dependent covariates (ie, they estimate underlying true values of time-dependent covariates at any point in time). At year 1, in all treatment groups, the curves representing Group 1 were above and to the right of the curves representing Group 2, which were above and to the right of the curves representing Group 3. Thus, with weight loss, insulin sensitivity and insulin secretion improved at years 1, 2, and 3. At year 1, the relationship between insulin sensitivity and secretion was examined within each treatment among 3 groups based on weight change from baseline. It shows the geometric means of insulin secretion and sensitivity for subjects diagnosed with diabetes at the yearly visit of diagnosis and the geometric means of their baseline values. The same process was followed to estimate the geometric means of insulin sensitivity and insulin secretion at all other time-periods and for the other 2 treatment groups. Hjerpsted JB, Flint A, Brooks A, Axelsen MB, Kvist T, Blundell J. Semaglutide improves postprandial glucose and lipid metabolism, and delays first-hour gastric emptying in subjects with obesity. Dunaif A, Segal KR, Futterweit W, Dobrjansky A. Profound peripheral insulin resistance, independent of obesity, in polycystic ovary syndrome. Reduction in the incidence of type 2 diabetes with lifestyle intervention or metformin. Metformin-induced increases in GDF15 are important for suppressing appetite and promoting weight loss. The trajectory of body weight change after discontinuation of the currently most widely used AOM, the glucagon like peptide-1 receptor agonist (GLP-1 RA) semaglutide, was observed in two studies from the STEP program (9, 10). While weight loss is achievable, weight maintenance remains a lifetime challenge (1). In this review, we will discuss the use of metformin in weight loss, cardiovascular health, and longevity, highlighting the historic background, molecular mechanisms, and current evidence. Finally, we studied the time-dependent associations of insulin secretion compensation and secretory demand with the transition from IGR to type 2 diabetes development. We examined the longitudinal associations of weight loss with insulin secretion compensation and secretory demand. After 3 years of follow-up, participants assigned to an intensive lifestyle intervention (ILS) or metformin, compared with placebo, had a 58% and a 31% reduction in diabetes incidence, respectively (10). These indices are highly correlated with the gold standard measures of insulin secretion and sensitivity (15) and were used in the previous DPP analysis (11).Baseline covariates included age, sex, race/ethnicity, and DPP treatment assignment. We conducted a secondary analysis using data from the DPP repository (14), which contained information on 3081 of the 3234 original participants randomized to the ILS, metformin, or placebo group. The DPP was a multicenter, randomized clinical trial that investigated the effects of an ILS and metformin on the prevention or delay of type 2 diabetes onset in participants with IGR (10). GDF15 mediates the effects of metformin on body weight and energy balance. Randomised placebo-controlled trial of orlistat for weight loss and prevention of weight regain in obese patients. The diverse approaches regarding the use of semaglutide for weight reduction highlight the necessity to guide physicians and standardize the long-term treatment regimen in anti-obesity medicine. In conclusion, the role of metformin in attenuation of weight regain after semaglutide discontinuation needs to be explored in randomized controlled studies in different insulin resistant populations. The weight trajectory after discontinuation of short-term semaglutide treatment in obese women with PCOS who continued metformin treatment has not yet been evaluated. Comparable clinically significant weight loss was achieved with metformin monotherapy in patients with obesity with or without T2DM/PreDM. In this retrospective study, 6- or 12-month weight loss outcomes were compared in 222 patients with or without T2DM/preDM who completed metformin monotherapy. An "anti-hunger" molecule produced after vigorous exercise is responsible for the moderate weight loss caused by the diabetes medication metformin, according to a new study in mice and humans. However, it is unknown whether weight loss outcomes differ with metformin monotherapy in patients with excess weight and euglycemia compared to patients with T2DM/prediabetes (PreDM). Peer review We performed a standard 75 g oral glucose tolerance test (OGTT) (24). Blood samples for glucose were drawn at 0 and 120 min of OGTT. The starting dose of semaglutide was 0.25 mg/week for 2 weeks, followed by 0.5 mg/week for 2 weeks, escalating to 1.0 mg/week for 12 weeks. The use of both therapeutic groups, the established metformin and the newer AOMs, in combinational and intermittent regimens seems to be a reasonable option that needs more research attention in this population. An updated approach with more advanced therapeutic strategies is needed to control weight and cardiometabolic risk factors in the long term. The weight trajectory after discontinuation of short-term semaglutide treatment in obese women with PCOS who continued metformin treatment has not yet been evaluated, so we can only provide a comparison with the general population from the STEP 1 extension study. We explored changes in body weight, cardiometabolic and endocrine parameters in obese women with PCOS who continued treatment with metformin 2 years after discontinuation of short-term intervention with semaglutide. Before metformin treatment was started insulin sensitivity was determined in all patients by calculating HOMA index and Matsuda index after a 75 g oral glucose tolerance test. For the same levels of weight loss, metformin decreased insulin secretory demand more than ILS (predicted distances along the baseline SMA line were smaller for metformin, Table S4 (16)). BMI in the high dose group decreased by 1.01 units (WMD-1.01; 95% CI, -1.29 to -0.73) and BMI did not continue to decrease significantly after treatment of more than 6 months. Subgroup analysis was performed based on baseline BMI, age, daily dose, and duration of medication. Results were presented as weighted mean difference (WMD) with a 95% confidence interval (CI 95%). Women with PCOS face a 2.8-fold higher risk for obesity and a 1.7-fold higher risk for central obesity compared with individuals without PCOS (25). The improvement in endocrine parameters induced by semaglutide resulted in the reduction of free testosterone (Table 1). The weight trajectories of individuals are graphed in Figure 2B. We investigated effects of weight loss, an intensive lifestyle intervention (ILS), and metformin on the relationship between insulin secretion and sensitivity using repository data from 2931 participants in the Diabetes Prevention Program clinical trial in adults at high risk of developing type 2 diabetes. Beneficial weight loss effects in insulin secretion compensation were greater in the ILS intervention than in metformin (predicted distances away from the baseline SMA line at year 1 were greater for ILS, Table S4 (16)). In this study, we used repeated measurements of insulin secretion and sensitivity to investigate how abnormalities in both develop relative to each other and to determine whether treatment, time, weight change, or combinations of these variables affect their interaction during the progression from IGR to diabetes. These results are consistent with previous DPP analyses showing that effects of the ILS and metformin interventions on reducing diabetes incidence were largely, but not entirely, attributed to weight loss (12, 13). These curves and the average insulin secretion and sensitivity showed that for both treatments, the biggest movement to the right occurred from baseline to year 1. In DPP participants, weight loss was more pronounced from baseline to year 1 for those assigned to ILS or metformin (10). Model 4 includes time-dependent distances away from the SMA line, time-dependent distances along the SMA line, baseline weight, and time-dependent weight loss from baseline in addition to covariates included in Model 1. All TVCMs and JMs indicated that metformin did not reduce diabetes risk compared with placebo with adjustment for changes in compensatory insulin secretion and secretory demand (Tables 3 and 4). These abnormalities are summarized through the concepts of estimated insulin secretory demand and compensatory insulin secretion which we defined in terms of the baseline SMA line. There were only small changes, on average, in insulin sensitivity and secretion in the placebo group (Figs. 3 and 4A). The TVCMs assumed that the insulin secretion compensation (distances away from the SMA line) stayed constant in between visits, which is biologically unlikely, and approximated the longitudinal trajectories using a step function that does not capture trends between visits. According to all models, the changes in compensatory insulin secretion were stronger determinants of risk than insulin secretory demand and weight change. A 5-kg decrease in weight from baseline significantly decreased diabetes risk; this is consistent with movements up and to the right of the baseline curve per 5-kg decreases in weight in Fig. They ate less than their peers and lost about 2 grams of body weight during the nine-day experiment.Here, we hypothesize that type 2 diabetes patients with varying levels of mitochondrial function, as indicated by mtDNA-CN, would have differential response to metformin treatment.A 5-kg decrease in weight from baseline significantly decreased diabetes risk; this is consistent with movements up and to the right of the baseline curve per 5-kg decreases in weight in Fig.In men with type 2 diabetes, metformin therapy can reduce testosterone levels and counteract the testosterone elevation that may accompany improvements in blood glucose levels (38, 39).\"Now we know that it is acting through the same pathway as vigorous exercise to reduce hunger. Understanding how these pathways are controlled may lead to viable strategies to lower body mass and improve health in millions of people.\"We aimed to explore changes in body weight, cardiometabolic and endocrine parameters in obese women with PCOS who continued metformin treatment 2 years after semaglutide cessation. We calculated the percentage of effects mediated as described elsewhere (16). The absolute value of the numerators in these formulas were not taken to allow the distances to be positive or negative so they could take the same direction as the residuals. Figure 1A and 1B show the SMA line fitted to log-transformed baseline variables and the resulting curve in their original scale, respectively. Significantly different slopes and/or intercepts indicated that the relationship changed over time. The nonlinear relationship of secretion and sensitivity was evaluated by standardized major axis regression to account for variability in both variables. Metformin therapy in polycystic ovary syndrome reduces hyperinsulinemia, insulin resistance, hyperandrogenemia, and systolic blood pressure, while facilitating normal menses and pregnancy. Cosentino F, Grant PJ, Aboyans V, Bailey CJ, Ceriello A, Delgado V, et al. 2019 ESC Guidelines on diabetes, pre-diabetes, and cardiovascular diseases developed in collaboration with the EASD. Mechanisms, pathophysiology, and management of obesity. The study was supported by Slovenian Research Agency grants #P3-0298, #P3-0343. Withdrawal of semaglutide is frequently followed by weight regain due to compensatory biological changes that prevent the maintenance of long-term weight loss.Continuation with semaglutide resulted in continued weight loss over the following 48 weeks, resulting in 17.4% of net weight loss from baseline, while those switched to placebo regained 6.9% of body weight after the switch to placebo (9).In participants with the same secretion compensation levels (same distances away from the baseline SMA line), the distances along the baseline SMA line allow the quantification of increase in insulin secretion per decrease in insulin sensitivity (9).While distances along the baseline SMA line representing insulin secretory demand increase as points go from below and to the right of the orthogonal line to above and to the left of the orthogonal line (as they go from negative to positive).Metabolic improvements during semaglutide treatment phase from timepoint 1 to timepoint 2 included a statistically significant decrease in total and LDL cholesterol, fasting plasma glucose, and glucose on 120 min of OGTT.At the time of diagnosis, all women were advised to follow healthy lifestyle intervention and to start with metformin treatment up-titrated to 2000 mg/day.Only the TVCM that did not adjust for baseline weight and weight change from baseline indicated that ILS significantly reduced diabetes risk compared with placebo (Tables 3 and 4).However, the exact nature of the mitochondrial interaction with metformin is still poorly characterized. This perhaps surprising result may be due to the restriction of the insulin sensitivity and secretion analysis to participants who remained without diabetes. When adjusted for compensatory insulin secretion and insulin secretory demand, the TVCMs did not find an effect of metformin on diabetes risk and the JMs did not find an effect of the ILS or metformin. This suggests that the inability of beta-cells to adequately compensate for changes in insulin sensitivity was the strongest determinant of diabetes risk at any point in time. Metabolic improvements during semaglutide treatment phase from timepoint 1 to timepoint 2 included a statistically significant decrease in total and LDL cholesterol, fasting plasma glucose, and glucose on 120 min of OGTT. The participants were treated with semaglutide, 1.0 mg/week in addition to their ongoing treatment of metformin at a dosage of 2000 mg/day and lifestyle interventions for 16 weeks. At the time of diagnosis, all women were advised to follow healthy lifestyle intervention and to start with metformin treatment up-titrated to 2000 mg/day. Some studies also imply that metformin could stabilize natural course of progressive body weight regain in women with PCOS (20–23). One year after withdrawal of 68 week intervention with semaglutide, participants regained 2/3 of their prior weight loss with residual benefits in some changes in cardiometabolic variables (10).