Abstract
Introduction
Bariatric metabolic surgery (BMS) alone or glucagon-like peptide-1 receptor agonists (GLP-1RA) alone are known to reduce major adverse cardiovascular events (MACEs). However, the benefit for primary prevention of MACE with use of GLP-1RA treatment after BMS is unclear. This retrospective cohort study assessed the associations of post-BMS GLP-1RA treatment with first incidence of MACE/all-cause mortality, as well as with predicted cardiovascular (CVD) risk and body mass index (BMI) and hemoglobin-A1c (HbA1c) levels, among patients with obesity and diabetes.
Methods
Adults who initiated GLP-1RA post-BMS were compared with individuals who did not. Groups were matched based on age, sex, BMI, HbA1c, years since BMS, and the nearest-neighbor propensity score for the probability of receiving GLP-1RA. Follow-up began at GLP-1RA initiation and ended on December 31, 2023.
Results
The study population included 476 GLP-1RA initiators and 952 matched BMS-only patients (mean [SD] age: 48.7 [9.5] years; 1,046 [73.2%] women). GLP-1RA initiation after BMS was associated with reduction in BMI (−7.0%) and HbA1c (−13.0%) and higher diabetes remission (78.0% vs. 61.6%). Ten-year predicted CVD risk was similar between users and nonusers before treatment initiation (0.053%) but was significantly lower among users during follow-up (0.039% vs. 0.046%). During a mean (SD) follow-up of 1.4 (0.8) years (maximum ∼10.6 years), there were 12.7 per 1,000 person months (n = 17) and 10.6 per 1,000 person months (n = 7) new cases of MACE and all-cause mortality in the BMS-only and GLP-1RA post-BMS group, respectively. No overall statistically significant association was observed between initiation of GLP-1RA after BMS and MACE/all-cause mortality. However, significant interaction was observed with time since BMS (p = 0.04), with a lower risk of MACE/all-cause mortality as the time from BMS increased.
Conclusion
Initiation of GLP-1RAs after BMS was associated with lower BMI and HbA1c levels, higher diabetes remission, and reduced predicted CVD risk among patients with diabetes and obesity. Overall, GLP-1RA treatment after BMS did not further reduce the risk of MACE/all-cause mortality compared to BMS alone after a mean follow-up of 1.4 years. Given the observed time-dependent benefit of treatment after BMS, longer follow-up studies are needed to determine the optimal timing for initiation.
Keywords: Bariatric surgery, Glucagon-like peptide-1 receptor agonists, Obesity, Diabetes, Major adverse cardiovascular events
Introduction
Bariatric metabolic surgery (BMS) is effective for weight loss and reducing obesity-related complications [1, 2]. A pooled analysis of 16 high-quality cohort studies and one controlled trial demonstrated long-term survival advantage of BMS versus conventional care, with a greater benefit among patients with diabetes [1]. BMS, and the subsequent weight loss achieved, was also found to reduce the risk for incidence of cardiovascular disease (CVD) [2].
Challenges after BMS include maintaining both the weight-loss and remission of diabetes over time. Specifically, 3–13% of patients who underwent BMS experienced significant weight regain within 5 years after the surgery and this percentage further increases over time [3, 4]. In addition, 10–20% of patients experience a relapse of diabetes within 10 years [5].
To address these clinical relapses, treatment with glucagon-like peptide-1 receptor agonists (GLP-1RAs) post BMS has been suggested [6]. Unrelated to BMS, GLP-1RA treatment improves overall survival and lowers risk of CVD in patients with and without diabetes [7, 8]. However, the benefit of initiation of GLP-1RA treatment following BMS for the prevention of major adverse cardiovascular events (MACEs) and all-cause mortality is unclear. A recent position statement by the International Federation for the Surgery of Obesity highlighted the need for further research on the role of obesity management medications as adjunctive treatments to BMS [9]. Therefore, this study aimed to assess the associations of GLP-1RA treatment after BMS with first incidence of MACE and all-cause mortality, as well as with body mass index (BMI) and hemoglobin-A1c (HbA1c) levels, in patients with obesity and diabetes.
Methods
The study was approved by the Community Institutional Review Board and by data utilization committees of Clalit Health Services (Clalit). Study identification number: COM2-0186-22.
Study Design and Population
Data for this retrospective observational cohort study were obtained from electronic medical records of Clalit, the largest healthcare organization in Israel, which provides coverage for approximately 52% of the Israeli population. Prior studies have confirmed the high validity of certain disease diagnoses recorded in the Clalit database [10–12]. The public, universal and mandatory nature of healthcare insurance services in Israel enable patients living with diabetes and obesity to access GLP-1RA and BMS as part of their care package after clinical recommendation.
The study was comprised of post-BMS adults (first surgery year: 2008-2018) with diabetes and obesity who initiated GLP-1RA following the BMS. BMS types included laparoscopic banding, gastric bypass, and sleeve gastrectomy. Individuals who initiated GLP-1RA treatment after the BMS were matched to individuals who had not initiated the treatment. The index date was defined as the date of GLP-1RA treatment initiation. Follow-up began 6 months following treatment initiation at index date and ended on December 31, 2023.
Patients were included in the study if they met all of the following criteria: underwent BMS between 2008 and 2018; were aged ≥24 years at the time of surgery; had a diagnosis of diabetes at age ≥21; had a BMI ≥30 kg/m2 at the time of BMS; had no prior history of ischemic heart disease, ischemic stroke, or heart failure as of index date; and maintained continuous membership in Clalit Health Services during the 2 years preceding the surgery. Patients were excluded from the study if they had been treated with GLP-1RA prior to undergoing BMS, or if they had end-stage kidney disease, active cancer, or were pregnant before the index date.
The main exposure was based on an intention-to-treat approach, defined by 6 or more filled prescriptions within a period of 12 consecutive months. To create two comparable groups, individuals who initiated GLP-1RA treatment after the BMS were matched (1:2) to individuals who had not initiated the treatment by the index date based on age (5-year intervals), sex, number of years elapsed from BMS (2-year intervals), and the nearest-neighbor propensity score (PS) for the probability of receiving GLP-1RA. The following variables were included in the PS model: sex, age (5-years intervals), index year (+/− year), and levels and percent change in BMI (Kg/Height2) and HBA1c (%) following the BMS.
Outcomes
The primary outcome was 4-point MACE plus all-cause mortality. MACE was defined as the first incidence of myocardial infarction (MI), ischemic stroke, percutaneous transluminal coronary angioplasty, or coronary artery bypass grafting. Data on the cause of death were unavailable.
Secondary outcomes included levels of HbA1c and BMI at the end of the follow-up. For each secondary outcome, both the absolute levels and percentage changes were calculated.
In addition, 10-year predicted CVD risk and diabetes remission were calculated for each individual at the following time windows: (1) presurgery, defined as the most recent measurement within the 2 years before the surgery; (2) early post-surgery, defined as the most recent measurement within the first year after surgery; (3) pre-GLP-1RA initiation, defined as the most recent measurement within the year preceding treatment initiation; and (4) post-GLP-1RA initiation, defined as the most recent measurement within the 2 years after treatment initiation. CVD predicted risk was estimated using the SCORE2 Diabetes algorithm developed by the SCORE2 Diabetes Working Group and the European Society of Cardiology Cardiovascular Risk Collaboration [13]. The algorithm was designed for individuals with type 2 diabetes without established atherosclerosis CVD and was calculated based on the following variables: age (years), sex, current smoker (yes/no), systolic blood pressure (mm Hg), total cholesterol (mmol/L), HDL cholesterol (mmol/L), age of diabetes diagnosis (per-5 years), HbA1c (mmol/mol), and eGFR (mL/min/1.73 m2). In accordance with the ESC HeartScore regional calibration, Israel is classified as a low-risk region, and the corresponding calibration was applied. Diabetes remission was defined as HbA1c level of <6.5%.
Covariates as of Index Date
Demographic and behavioral variables included sex, age (years), socioeconomic status, and smoking status (never, past, or current smoker). Chronic disease diagnoses and microvascular complications were assessed based on International Classification of Diseases Ninth revision (ICD-9) codes and included hypertension, hyperlipidemia, neuropathy, retinopathy, and nephropathy. Laboratory measures included glucose, total cholesterol, high-density lipoprotein cholesterol (HDL-C), low-density lipoprotein cholesterol (LDL-C), and triglycerides. Duration of diabetes, HbA1c, and BMI were taken from the community healthcare medical records.
Statistical Analysis
The main characteristics of the study patients were described using means with standard deviations (SDs) or medians with ranges for continuous variables, and proportions for covariate balance between the groups after PS matching. Differences <0.1 indicated excellent covariate balance. Covariates imbalanced were adjusted for in the final models.
For the main analysis, unadjusted and adjusted Cox proportional-hazards regression models were used to evaluate the association between exposure to GLP-1RA post-BMS with the combined endpoints of MACE plus all-cause mortality incidence. To address potential residual confounding, the multivariable model was adjusted for the continuous values of variables used categorically in the matching and for variables with SMD >0.1 [14]. Specifically, the following covariates at the index date were included: age (years), sex, years elapsed since BMS (years), HbA1c level (%), BMI (kg/m2), change in HbA1c from BMS (%), change in BMI from BMS (%), level of HDL-C, triglycerides, and systolic and diastolic blood pressure. Termination of follow-up was defined as the earliest of incidence of primary outcome (MACE plus all-cause mortality), disruption in Clalit membership, or end of study. The proportional-hazards assumption was evaluated using Schoenfeld’s global test.
To assess whether the association between GLP-1RAs therapy and the main outcome varied by diabetes duration or by years elapsed from BMS, the model included post-BMS GLP-1RAs initiation x diabetes duration and post-BMS GLP-1RAs initiation x years since BMS. In secondary analysis, two linear models – one for HbA1c and one for BMI – were used to evaluate the association of post-BMS GLP-1RAs initiation with levels measured at the end follow-up. Both models were adjusted for the following covariates as of index date: age (years), sex, years elapsed since BMS (years), BMI and HbA1c as of index date, level of HDL-C, triglycerides, and systolic and diastolic blood pressure.
Linear and logistic mixed-effects models were used to assess longitudinal changes in the estimated 10 year predicted CVD risk based on SCORE2 and in diabetes remission, respectively, with a participant level random intercept to account for within individual correlation. Fixed effects included period, GLP-1RA use status, and their interaction. From these models, adjusted estimated marginal means from the linear model and predicted probabilities from the logistic model were derived for each treatment by period combination.
Period-specific between-group comparisons, defined as treatment users versus nonusers within each period, were conducted using estimated marginal means or predicted probabilities from the fitted mixed-effects model. Post hoc pairwise contrasts were performed with Tukey adjustment to account for multiple testing.
Data extraction was performed using the Structured Query Language (SQL) software. All statistical analysis conducted evaluated using R statistical software version 4.0.1 with the following freely available packages: survival (3.2-13), ggplot2 (3.3.5), ggpubr (0.4.0), survminer (0.4.9), matchit (4.3.3), and gtsummary (1.5.2).
Results
During the study period, 4,469 patients met the inclusion criteria as of BMS. After excluding 1,639 individuals who met one or more of the exclusion criteria, 2,830 were considered eligible for the study as of BMS date. Of them 595 (21.0%) initiated treatment with GLP-1RA after BMS (GLP-1RA post-BMS). The median (IQR) time for starting GLP-1RA was 8.0 (3.6) years. Following the exclusion of individuals who met >1 of the exclusion criteria between BMS and GLP-1RA initiation (n = 109) or had <6 months between GLP-1RA initiation and end of study (n = 7), 479 individuals were potentially eligible patients as of index date; of them 3 were excluded due to no matches. The final study population consisted of 476 GLP-1RA initiators and 952 matched BMS-only patients (shown in Fig. 1). Main characteristics of the two comparison groups as of index date and SMDs for all covariates between GLP-1RA post-BMS group and BMS-only group after matching are presented in Table 1.
Fig. 1.
Selection of study population.
Table 1.
Characteristics of patients as of index date
| Characteristic | BMS only, n = 952 | GLP-1RA post-BMS, n = 476 | SMD |
|---|---|---|---|
| Female, n (%) | 697 (73.2) | 349 (73.3) | |
| Age, mean (SD), years | 48.8 (9.5) | 48.5 (9.4) | 0.04 |
| BMI, mean (SD), kg/m2 | 33.4 (5.7) | 35.5 (5.7) | −0.37 |
| HbA1c, mean (SD), % | 6.2 (0.9) | 6.9 (1.3) | −0.66 |
| Duration of diabetes, mean (SD), m | 63.5 (62.1) | 68.6 (62.9) | 0.08 |
| Hypertension, n (%) | 471 (49.5) | 233 (48.9) | 0.01 |
| Hyperlipidemia, n (%) | 692 (72.7) | 348 (73.1) | −0.01 |
| Neuropathy, n (%) | 65 (6.8) | 30 (6.3) | 0.02 |
| Nephropathy, n (%) | 73 (7.7) | 41 (8.6) | −0.03 |
| Total cholesterol, mean (SD), mg/dL | 182.2 (36.8) | 183.6 (39.0) | −0.04 |
| LDL-C, mean (SD), mg/dL | 104.8 (31.5) | 105.3 (34.8) | −0.01 |
| HDL-C, mean (SD), mg/dL | 53.4 (12.8) | 51.4 (11.9) | 0.17 |
| Triglycerides, mean (SD), mg/dL | 120.2 (59.0) | 149.7 (151.1) | −0.26 |
| Smoking status, n (%) | |||
| Current | 31 (3.7) | 17 (4.1) | 0.17 |
| Never | 613 (72.6) | 287 (69.0) | |
| Past | 139 (16.5) | 92 (22.1) | |
| Unknown | 61 (7.2) | 20 (4.8) | |
| Type of BMS, n (%) | | | 0.04 |
| Laparoscopic banding | 125 (13.1) | 57 (12.0) | |
| Gastric bypass | 383 (40.2) | 190 (39.9) | |
| Sleeve gastrectomy | 444 (46.6) | 229 (48.1) | |
Association between GLP-1RA Initiation after BMS with BMI and HbA1c Levels
At the index date, the mean (SD) BMI was 35.5 (5.7) in the GLP-1RA post-BMS group and 33.4 (5.7) in the BMS-only group. From the index date to the end of follow-up, the BMS-only group maintained a similar mean (SD) BMI level at the end of follow-up (33.5 (5.9)), whereas the GLP-1RA post-BMS group showed a significant reduction to 33.0 (6.2), with a mean (SD) percentage decrease of −7.0% (9.0%). In the adjusted linear model, the estimated difference between groups was β = 6.74 (95% CI: 5.69–7.79, p < 0.001) (Table 2).
Table 2.
Association of GLP-1RA initiation following BMS with change in BMI and HbA1c levels
| | BMS only | GLP-1RA post-BMS |
|---|---|---|
| N | 952 | 476 |
| BMI, kg/m2 | ||
| As of index date, mean (SD) | 33.4 (5.7) | 35.5 (5.7) |
| Last level in follow-up period, mean (SD) | 33.5 (5.9) | 33.0 (6.2) |
| Percentage change between index date and end of follow up, mean (SD)a | 0.3% | −7.0% |
| β (95% CI)b | ref. | 6.74 (5.69–7.79) |
| HbA1c, kg/m2 | ||
| As of index date, mean (SD) | 6.2 (0.9) | 6.9 (1.3) |
| Last level in follow-up period, mean (SD) | 6.2 (1.0) | 6.0 (1.0) |
| Percentage change between index date and end of follow-up, mean (SD)a | 0 | −13.0% |
| β (95% CI)b | ref. | 10.05 (8.72–11.37) |
p < 0.05.
Linear regression model comparing GLP-1RA initiators with matched BMS-only patients; outcome: last level in follow-up period; adjusted for the following variables as of follow-up initiation: age (years), sex, years elapsed since BMS (years), BMI and HbA1c as of index date, level of HDL-C, triglycerides, and systolic and diastolic blood pressure.
With respect to HbA1c levels, at the index date, the HbA1c levels were 6.9 (1.3) in the GLP-1RA post-BMS group and 6.2 (0.9) in the BMS-only group. From the index date to the end of follow-up, no further decline in HbA1c was observed in the BMS-only group. In contrast, the GLP-1RA post-BMS group showed a significant reduction to 6.0 (1.0), with a mean percentage change of 13.0% from the index date to the end of follow-up. In the adjusted linear model, the estimated difference between groups was β = 10.05 (95% CI: 8.72–11.37, p < 0.001) (Table 2).
Diabetes Remission
Diabetes remission (HbA1c <6.5%) stratified by GLP-1RA use status across study periods is presented in Table 3. Based on the mixed-effects model with period-specific estimated marginal probabilities, the proportion of individuals with HbA1c <6.5 were nearly identical between GLP-1RA users and nonusers during the presurgery period (20.3% and 23.2%, respectively) and the early post-surgery period (79.6% and 81.4%, respectively). A significant difference was observed in the pretreatment initiation with fewer individuals who achieved diabetes remission among users (40.8%) compared with nonusers (59.6%). However, a higher proportion with diabetes remission was observed among those who initiated GLP-1RA after surgery compared with those who did not (78.0% and 61.6%, respectively).
Table 3.
Ten-year predicted CVD risk (SCORE2 Diabetes) and diabetes remission by GLP-1RA use status across study periods among individuals with diabetesa,b
| Period | HbA1c <6.5%, % | 10-year CVD risk, % | ||||
|---|---|---|---|---|---|---|
| BMS only | GLP-1RA post-BMS | p value | BMS only | GLP-1RA post-BMS | p value | |
| Presurgery | 23.2 | 20.3 | 0.217 | 0.0632 | 0.0612 | 0.5522 |
| Early post-surgery | 81.4 | 79.6 | 0.229 | 0.0407 | 0.0394 | 0.5609 |
| Pre-GLP-1RA initiation | 59.6 | 40.8 | <0.001 | 0.0526 | 0.0535 | 0.7489 |
| Post-GLP-1RA initiation | 61.6 | 78.0 | <0.001 | 0.0457 | 0.0387 | 0.0015 |
Probabilities are model based predicted values derived from mixed-effects models with a participant level random intercept and fixed effects for period, GLP-1RA use status, and their interaction. For CVD risk, adjusted estimated marginal means were obtained from the linear model. For diabetes remission, predicted probabilities were obtained from the mixed-effects logistic model.
p values correspond to the pairwise between-group comparison (GLP-1RA users vs. nonusers) within each period, based on estimated marginal means/probabilities.
Changes in Predicted CVD Risk
The 10-year predicted CVD risk, calculated using the SCORE2 Diabetes algorithm for individuals with type 2 diabetes, stratified by GLP-1RA use status across study periods, is presented in Table 3. Based on the mixed-effects model with period-specific estimated marginal means, predicted probabilities were nearly identical between GLP-1RA users and nonusers during the presurgery period, the first postoperative year, and the year preceding GLP-1RA initiation, with minimal nonsignificant absolute risk differences (−0.0020, −0.0013, and +0.0009, respectively). In contrast, during the 2 years after treatment initiation, a lower significant (p = 0.002) predicted probability was observed among GLP-1RA users (0.0387) compared with nonusers (0.0457), corresponding to an absolute risk difference of −0.0070 (−0.7 percentage points).
Association between GLP-1RA Initiation following BMS and MACE plus All-Cause Mortality
The mean (SD) follow-up time was 1.4 (0.8) years in both groups, with a maximum of ∼10.6 years. There were 12.67 per 1,000 person months (n = 17) and 10.60 per 1,000 person months (n = 7) new incidents of MACE plus all-cause mortality in the BMS-only and GLP-1RA post-BMS groups, respectively.
The unadjusted and adjusted Cox proportional-hazards models for MACE plus all-cause mortality, comparing treatment with GLP-1RA post-BMS to BMS alone, found HR = 0.86 (95% CI: 0.36–2.09) and HR = 0.70 (95% CI: 0.27–1.84), respectively (Table 4). No evidence of interaction was observed between GLP-1RA initiation after BMS and diabetes duration in relation to MACE plus all-cause mortality (p = 0.838), whereas a significant interaction was found between GLP-1RA initiation and time since BMS (p = 0.037). Specifically, as the time from BMS increased, a lower risk of MACE plus all-cause mortality was observed among those who initiated treatment with GLP-1RAs compared with those who did not. When stratified by the median time since BMS (≤8 years vs. >8 years), the HRs were 1.25 (95% CI: 0.36–4.38) and 0.32 (95% CI: 0.06–1.68), respectively.
Table 4.
Association of GLP-1RA initiation following BMS with long-term incidence of MACE
| | BMS only | GLP-1RA post-BMS |
|---|---|---|
| N | 952 | 476 |
| Follow-up, mean (SD), years | 1.4 (0.8) | 1.4 (0.8) |
| Total incidence of MACE, n (%) | 17 (1.79) | 7 (1.47) |
| MACE rate/1,000 person years | 12.67 | 10.60 |
| Hazard Ratio (95% CI) for MACE | ||
| Unadjusted | 1 (reference) | 0.86 (0.36–2.09) |
| Adjusted for potential confounders | 1 (reference) | 0.70 (0.27–1.84) |
p < 0.05.
Multivariate Cox proportional hazards model comparing GLP-1RA initiators with matched BMS-only patients; outcome: composite of MACE and all-cause mortality; adjusted for the following variables as of follow-up initiation: age (years) sex, years elapsed since BMS (years), HbA1c level (%), BMI (kg/m2), change in HbA1c from BMS (%), change in BMI from BMS (%), level of HDL-C, triglycerides, and systolic and diastolic blood pressure.
Discussion
In this retrospective cohort study among individuals living with diabetes and obesity who underwent BMS, a significant reduction in both BMI and HbA1c levels was observed among patients who initiated GLP-1RA after surgery, whereas no such reductions were observed among those who did not receive GLP-1RA. In addition, a higher proportion of diabetes remission was observed among patients who initiated GLP-1RA after surgery compared with those who did not. Ten-year predicted CVD risk was similar between GLP-1RA users and non-users before treatment initiation and became significantly lower among users during follow-up. While no overall statistically significant association was observed between initiation of GLP-1RA after BMS and MACE/all-cause mortality, after mean follow-up of 1.4 years, a statistically significant interaction was observed with time since BMS, indicating a potential time-dependent benefit of GLP-1RA, which increased as the time from surgery increased.
BMS has been found to be effective in reducing the risks of mortality and CVD events in individuals living with obesity compared to individuals treated with nonsurgical interventions [1, 2]. A meta-analysis of 39 prospective and retrospective cohort studies, with follow-up ranging from 2 to 24 years [2] showed a 41% risk reduction for CVD mortality, 42% for incidence of MI, and 36% for stroke. Specifically in patients living with diabetes and obesity, where weight and glycemic control are more difficult to achieve without surgical intervention, BMS reduced incidence of all-cause mortality and MACE compared to nonsurgical interventions [15] even more so than in patients without diabetes [1].
Several studies have demonstrated the association between treatment with GLP-1RA and improved CVD outcomes. A meta-analysis by Satar et al. [16] demonstrated 12% reduction in all-cause mortality and a 14% reduction in MACE across the studies evaluated, associated with GLP-1 treatment. In the seminal SELECT study, semaglutide 2.4 mg demonstrated reduction in the risk of MACE by 20% in people living with obesity [17]. The strong beneficial effect of GLP-1R in people living with diabetes was also demonstrated in the Flow trial [18] that was terminated earlier than planned due to the significant beneficial effect of semaglutide 1 mg on deterioration of chronic kidney disease, major cardiac events, and other causes of death.
The majority of the beneficial outcomes of BMS and GLP-1RA in patients living with diabetes and obesity have been attributed to significant weight loss [19, 20]. A recent study found that patients who underwent BMS had higher reduction in BMI compared to those treated with GLP-1RA [19], and this could have contributed to the increased advantage of BMS over GLP-1RA in prevention of all-cause mortality that was found. However, that study did not find an advantage for prevention of MACE, suggesting that GLP-1RA may contribute to the prevention of cardiac events through additional mechanisms. This notion could be further strengthened by the results of the Harmony study, which found that GLP-1RA’s benefit for reducing risk of MACE was independent of weight loss [21]. More specifically, beyond reductions in traditional CVD risk factors such as hyperglycemia, blood pressure, and postprandial lipemia, several cardioprotective mechanisms of GLP-1RA have been proposed [22, 23]. Anti-atherosclerotic effects of GLP-1RA have been demonstrated in two different mouse models of atherosclerosis, through reduction in inflammation and increase in plaque stability [24, 25]. Furthermore, a human study demonstrated reduction of C-reactive protein, a biomarker of inflammation, in individuals who are on GLP-1RA compared to placebo [26]. GLP-1RA may also be involved in the reduction of apoptosis and oxidative stress and the increase of glucose metabolism and coronary blood flow in the myocardium [22, 23]. In addition, reduction in the inflammation of the perivascular fat of the coronary artery and reduction in the volume and inflammation of the epicardial fat [27, 28] may be attributed to the non-weight loss beneficial effects of the GLP-1RA.
Despite the potential of GLP-1RA to reduce the risk of MACE/all-cause mortality through various mechanisms, no overall effectiveness was observed in the present study. Although the predicted SCORE2 Diabetes risk declined among GLP-1RA users and was lower than in those who did not receive GLP-1RA after BMS, this improvement did not translate into a detectable reduction in major clinical outcomes during the available follow-up. This finding may indicate that GLP-1RA provides no additional benefit beyond BMS, but it may also reflect the relatively short mean follow-up period, particularly in the context of primary prevention. Notably, in the peptide innovation for Early Diabetes Treatment 6 (PIONEER 6) trial [29], which had a follow up duration similar to the present study (1.3 years), the HR for the primary outcome of three-point MACE was not statistically significant (HR, 95% CI: 0.79, 0.57–1.11). A longer follow-up is necessary to reveal whether early improvements in predicted risk with GLP-1RA ultimately translate into reductions in clinical events. The statistically significant interaction with time since BMS that we observed in this study, indicating a potential time-dependent benefit of GLP-1RA, with greater benefits as the time from BMS increased. This finding may be explained by the initially strong effects of BMS, which reduce the need for additional therapy in the early post-procedure period. As the effects of the surgery diminish over time, the relative benefit of GLP-1RA treatment increases.
A recent review suggests that GLP-1RAs are safe and effective treatment for individuals who experienced suboptimal weight loss following BMS [3, 4]. Similar to previous studies [30–32], the present study also demonstrated the effectiveness of initiating GLP-1RA treatment after BMS for weight and diabetes control. Despite having poorer glycemic status at the index, individuals who initiated GLP-1RA treatment demonstrated substantial metabolic improvement, with diabetes remission rates approaching those observed early after surgery.
The results from this real-world study are consistent with previous publications demonstrating that during the first years after surgery most people remain in diabetes remission and the large majority of individuals are not prescribed GLP-1RA [3, 4]. In the current study, the median time to treatment initiation was approximately 8 years after BMS, indicating that in real-world practice, immediate initiation of treatment after BMS may not be required.
A strength of the current study is in the richness of the database, which allows continuous follow-up for the same cohort of individuals post-BMS surgery. This study has several limitations. First, the observational nature of the study limited the ability for causal inference regarding the effects of GLP-1RA over the sustained benefits of the BMS. Second, even after nearest-neighbor PS matching for the probability of receiving GLP-1RA, levels of HbA1c and BMI remained higher among individuals who initiated GLP-1RA compared to those who did not. While these differences likely reflect real-word clinical variability, residual confounding cannot be excluded. Although the remaining imbalance could have biased the results to a null, the benefit of GLP-1RA remained significant. In addition, covariates with residual imbalance were included in the multivariable model in their continuous form. Notably, predicted SCORE2 Diabetes CVD risk was similar between the groups at baseline, suggesting no substantial baseline risk difference between the groups. Third, the relatively short follow-up (1.4 years) and limited sample size likely reduced power to detect associations with MACE and all-cause mortality due to the small number of events. Nevertheless, improvements in BMI and HbA1c levels after GLP-1RA initiation and lower predicted CVD risk based on SCORE2 were observed. Future studies with longer follow-up and larger sample sizes are warranted. Fourth, the current database does not include information on lifestyle interventions (diet and exercise), which are important variables to consider in this population. Fifth, when considering mortality, there was no cause of death information, so only all-cause mortality was evaluated. Therefore, additional studies are needed to confirm these results.
Conclusions
In this retrospective cohort study, initiation of GLP-1RAs after BMS was associated with lower BMI and HbA1c levels, higher diabetes remission, and reduced predicted CVD risk among patients living with diabetes and obesity. Overall, adding GLP-1RA to BMS did not further reduce the risk for MACE and all-cause mortality in mean of 1.4-year follow-up. Given the observed significant interaction indicating a time-dependent benefit of treatment initiation after BMS, additional studies with longer follow-up are needed to determine the optimal timing for initiating this intervention.
Statement of Ethics
The study was approved by the Clalit Health Services (Clalit) Community Institutional Review Board (IRB) and by data Clalit utilization committee. Study identification number: COM2-0186-22. Due to the retrospective design, the study was exempt from obtaining informed consent from the patients, as approved by the Clalit Community IRB.
Conflict of Interest Statement
Drs. Reges, Dicker, Lavie, and Arbel reported receiving grants from the Israel Science Foundation outside the submitted work. Dr. Dicker reported receiving grants, personal fees, and nonfinancial support from NovoNordisk and Eli Lilly; and personal fees and nonfinancial support from Boehringer Ingelheim outside the submitted work. No other disclosures were reported.
Funding Sources
No financial or in-kind support was provided for the conduct of this study.
Author Contributions
All authors contributed to the study design and execution. E.B. extracted the data under the supervision of O.R.; W.A.A. and N.R. cleaned and analyzed the data with the guidance of Y.W. and O.R.; O.R., W.A.A., D.D., N.R., Y.W., and P.G drafted the manuscript; D.D. and G.L. approved all the clinical aspects of the study; O.R., D.D., G.L., and R.A. supervised the design and execution of the study. All authors revised the manuscript for critical content and clarity and approved it.
Funding Statement
No financial or in-kind support was provided for the conduct of this study.
Data Availability Statement
The data are not publicly available due to privacy restrictions. Further inquiries for aggregated data can be directed to the corresponding author and will be provided only upon approval from the relevant authorities at Clalit Health Services.
References
- 1. Syn NL, Cummings DE, Wang LZ, Lin DJ, Zhao JJ, Loh M, et al. Association of metabolic-bariatric surgery with long-term survival in adults with and without diabetes: a one-stage meta-analysis of matched cohort and prospective controlled studies with 174 772 participants. Lancet. 2021;397(10287):1830–41. [DOI] [PubMed] [Google Scholar]
- 2. van Veldhuisen SL, Gorter TM, van Woerden G, de Boer RA, Rienstra M, Hazebroek EJ, et al. Bariatric surgery and cardiovascular disease: a systematic review and meta-analysis. Eur Heart J. 2022;43(20):1955–69. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3. Arterburn DE, Telem DA, Kushner RF, Courcoulas AP. Benefits and risks of bariatric surgery in adults: a review. JAMA. 2020;324(9):879–87. [DOI] [PubMed] [Google Scholar]
- 4. Dreant A, Blanchard C, Jacobi D. Adjuvant glucose-like peptide 1 receptor agonist therapy for suboptimal weight loss after bariatric surgery: a systematic review. Obes Surg. 2024;34(5):1846–54. [DOI] [PubMed] [Google Scholar]
- 5. Conte C, Lapeyre-Mestre M, Hanaire H, Ritz P. Diabetes remission and relapse after bariatric surgery: a nationwide population-based Study. Obes Surg. 2020;30(12):4810–20. [DOI] [PubMed] [Google Scholar]
- 6. Dutta D, Nagendra L, Joshi A, Krishnasamy S, Sharma M, Parajuli N. Glucagon-Like Peptide-1 receptor agonists in post-bariatric surgery patients: a systematic review and meta-analysis. Obes Surg. 2024;34(5):1653–64. [DOI] [PubMed] [Google Scholar]
- 7. Nreu B, Dicembrini I, Tinti F, Sesti G, Mannucci E, Monami M. Major cardiovascular events, heart failure, and atrial fibrillation in patients treated with glucagon-like peptide-1 receptor agonists: an updated meta-analysis of randomized controlled trials. Nutr Metab Cardiovasc Dis. 2020;30(7):1106–14. [DOI] [PubMed] [Google Scholar]
- 8. Lincoff AM, Brown-Frandsen K, Colhoun HM, Deanfield J, Emerson SS, Esbjerg S, et al. Semaglutide and cardiovascular outcomes in obesity without diabetes. N Engl J Med. 2023;389(24):2221–32. [DOI] [PubMed] [Google Scholar]
- 9. Cohen RV, Busetto L, Levinson R, Le Roux CW, Salminen P, Prager G, et al. International consensus position statement on the role of obesity management medications in the context of metabolic bariatric surgery: expert guideline by the international Federation for the Surgery of Obesity and Metabolic Disorders (IFSO). Br J Surg. 2024;111(12):znae283. [DOI] [PubMed] [Google Scholar]
- 10. Rennert G, Peterburg Y. Prevalence of selected chronic diseases in Israel. Isr Med Assoc J. 2001;3(6):404–8. [PubMed] [Google Scholar]
- 11. Reges O, Weinberg H, Hoshen M, Greenland P, Rayyan-Assi H, Avgil Tsadok M, et al. Combining inpatient and outpatient data for diagnosis of non-valvular atrial fibrillation using electronic health records: a validation Study. Clin Epidemiol. 2020;12:477–83. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12. Saliba W, Mishchenko E, Cohen S, Rennert G, Preis M. Association between myelofibrosis and thromboembolism: a population-based retrospective cohort study. J Thromb Haemost. 2020;18(4):916–25. [DOI] [PubMed] [Google Scholar]
- 13. SCORE2-Diabetes Working Group and the ESC Cardiovascular Risk Collaboration . SCORE2-Diabetes: 10-year cardiovascular risk estimation in type 2 diabetes in Europe. Eur Heart J. 2023;44(28):2544–56. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14. Nguyen TL, Collins GS, Spence J, Daures JP, Devereaux PJ, Landais P, et al. Double-adjustment in propensity score matching analysis: choosing a threshold for considering residual imbalance. BMC Med Res Methodol. 2017;17(1):78. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15. Aminian A, Zajichek A, Arterburn DE, Wolski KE, Brethauer SA, Schauer PR, et al. Association of metabolic surgery with major adverse cardiovascular outcomes in patients with type 2 diabetes and obesity. JAMA. 2019;322(13):1271–82. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16. Sattar N, Lee MMY, Kristensen SL, Branch KRH, Del Prato S, Khurmi NS, et al. Cardiovascular, mortality, and kidney outcomes with GLP-1 receptor agonists in patients with type 2 diabetes: a systematic review and meta-analysis of randomised trials. Lancet Diabetes Endocrinol. 2021;9(10):653–62. [DOI] [PubMed] [Google Scholar]
- 17. Lincoff AM, Brown-Frandsen K, Colhoun HM, Deanfield J, Emerson SS, Esbjerg S, et al. Semaglutide and cardiovascular outcomes in obesity without diabetes. N Engl J Med. 2023;389(24):2221–32. [DOI] [PubMed] [Google Scholar]
- 18. Perkovic V, Tuttle KR, Rossing P, Mahaffey KW, Mann JFE, Bakris G, et al. Effects of semaglutide on chronic kidney disease in patients with type 2 diabetes. N Engl J Med. 2024;391(2):109–21. [DOI] [PubMed] [Google Scholar]
- 19. Dicker D, Sagy YW, Ramot N, Battat E, Greenland P, Arbel R, et al. Bariatric metabolic surgery vs glucagon-like Peptide-1 receptor agonists and mortality. JAMA Netw Open. 2024;7(6):e2415392. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20. Wolff Sagy Y, Lavie G, Ramot N, Battat E, Arbel R, Reges O, et al. Effectiveness of bariatric metabolic surgery versus glucagon-like peptide-1 receptor agonists for prevention of congestive heart failure. Nat Med. 2024;30(8):2337–42. [DOI] [PubMed] [Google Scholar]
- 21. Hernandez AF, Green JB, Janmohamed S, D’Agostino RB Sr, Granger CB, Jones NP, et al. Albiglutide and cardiovascular outcomes in patients with type 2 diabetes and cardiovascular disease (Harmony Outcomes): a double-blind, randomised placebo-controlled trial. Lancet. 2018;392(10157):1519–29. [DOI] [PubMed] [Google Scholar]
- 22. Wu Q, Li D, Huang C, Zhang G, Wang Z, Liu J, et al. Glucose control independent mechanisms involved in the cardiovascular benefits of glucagon-like peptide-1 receptor agonists. Biomed Pharmacother. 2022;153:113517. [DOI] [PubMed] [Google Scholar]
- 23. Ussher JR, Drucker DJ. Glucagon-like peptide 1 receptor agonists: cardiovascular benefits and mechanisms of action. Nat Rev Cardiol. 2023;20(7):463–74. [DOI] [PubMed] [Google Scholar]
- 24. Rakipovski G, Rolin B, Nohr J, Klewe I, Frederiksen KS, Augustin R, et al. The GLP-1 analogs liraglutide and semaglutide reduce atherosclerosis in ApoE(-/-) and LDLr(-/-) mice by a mechanism that includes inflammatory pathways. JACC Basic Transl Sci. 2018;3(6):844–57. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25. Jensen DM, Skovsted GF, Bonde MFB, Bentzon JF, Rolin B, Franck G, et al. Semaglutide treatment attenuates vessel remodelling in ApoE-/- mice following vascular injury and blood flow perturbation. Atheroscler Plus. 2022;49:32–41. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26. Verma S, Bhatta M, Davies M, Deanfield JE, Garvey WT, Jensen C, et al. Effects of once-weekly semaglutide 2.4 mg on C-reactive protein in adults with overweight or obesity (STEP 1, 2, and 3): exploratory analyses of three randomised, double-blind, placebo-controlled, phase 3 trials. EClinicalMedicine. 2023;55:101737. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27. Iacobellis G, Villasante Fricke AC. Effects of semaglutide versus dulaglutide on epicardial fat thickness in subjects with type 2 diabetes and obesity. J Endocr Soc. 2020;4(4):bvz042. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28. Malavazos AE, Meregalli C, Sorrentino F, Vignati A, Dubini C, Scravaglieri V, et al. Semaglutide therapy decreases epicardial fat inflammation and improves psoriasis severity in patients affected by abdominal obesity and type-2 diabetes. Endocrinol Diabetes Metab Case Rep. 2023;2023(3). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29. Husain M, Birkenfeld AL, Donsmark M, Dungan K, Eliaschewitz FG, Franco DR, et al. Oral semaglutide and cardiovascular outcomes in patients with type 2 diabetes. N Engl J Med. 2019;381(9):841–51. [DOI] [PubMed] [Google Scholar]
- 30. Miras AD, Perez-Pevida B, Aldhwayan M, Kamocka A, McGlone ER, Al-Najim W, et al. Adjunctive liraglutide treatment in patients with persistent or recurrent type 2 diabetes after metabolic surgery (GRAVITAS): a randomised, double-blind, placebo-controlled trial. Lancet Diabetes Endocrinol. 2019;7(7):549–59. [DOI] [PubMed] [Google Scholar]
- 31. Jensen AB, Renstrom F, Aczel S, Folie P, Biraima-Steinemann M, Beuschlein F, et al. Efficacy of the glucagon-like Peptide-1 receptor agonists liraglutide and semaglutide for the treatment of weight regain after bariatric surgery: a retrospective observational Study. Obes Surg. 2023;33(4):1017–25. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32. Mok J, Adeleke MO, Brown A, Magee CG, Firman C, Makahamadze C, et al. Safety and efficacy of liraglutide, 3.0 mg, once daily vs placebo in patients with poor weight loss following metabolic surgery: the BARI-OPTIMISE randomized clinical trial. JAMA Surg. 2023;158(10):1003–11. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Data Availability Statement
The data are not publicly available due to privacy restrictions. Further inquiries for aggregated data can be directed to the corresponding author and will be provided only upon approval from the relevant authorities at Clalit Health Services.