Long-term weight loss effects of semaglutide in obesity without diabetes in the SELECT trial

The study addresses the growing global burden of obesity and its cardiometabolic consequences, noting that BMI, while useful for population surveillance, imperfectly reflects individual fat amount and distribution. Excess visceral and ectopic fat drive cardiovascular disease and other chronic conditions. Achieving and sustaining substantial weight loss with lifestyle interventions alone is challenging; GLP-1 receptor agonists, including once-weekly subcutaneous semaglutide 2.4 mg, have demonstrated the ability to produce greater and more durable weight loss and reduce cardiovascular events in high-risk populations. The prespecified analysis in SELECT investigates, in adults with established cardiovascular disease and overweight/obesity without diabetes, the long-term effects of semaglutide versus placebo on weight and anthropometric measures (BMI, waist circumference, waist-to-height ratio), and evaluates safety and tolerability across baseline BMI categories.

Prior evidence indicates modest weight loss (≈5%) improves cardiometabolic risk factors, with greater losses yielding larger benefits. Pharmacotherapies that modulate appetite, particularly GLP-1 receptor agonists, enable clinically meaningful weight loss and sustainability. Semaglutide (up to 2.4 mg weekly) is approved for chronic weight management and, at lower doses, for type 2 diabetes, with prior trials showing cardiovascular benefit. The STEP 1 trial in non-diabetic individuals with overweight/obesity reported ~15% mean weight loss with semaglutide at 68 weeks versus ~2.4% with placebo. However, SELECT is a cardiovascular outcomes trial without targeted weight-loss counseling, allowing evaluation of long-term weight and anthropometric effects in a diverse, globally representative cohort. The study also references evidence that sex- and race-specific waist circumference and waist-to-height ratio cutoffs better predict cardiometabolic risk in some populations and that BMI may not fully capture fat distribution-related risk.

Design: Prespecified analysis of SELECT, a randomized, double-blind, placebo-controlled cardiovascular outcomes trial (NCT03574597). Participants were randomized 1:1 to once-weekly subcutaneous semaglutide 2.4 mg or placebo, in addition to standard-of-care management for secondary cardiovascular prevention. Participants: Adults ≥45 years with BMI ≥27 kg/m², established cardiovascular disease (prior MI, prior ischemic/hemorrhagic stroke, or symptomatic PAD), and without diabetes. A total of 17,604 participants (72.3% male) from 41 countries were enrolled between Oct 2018 and Mar 2021; mean age 61.6 years, mean BMI 33.3 kg/m². Intervention: Semaglutide initiated at 0.24 mg weekly with dose escalation every 4 weeks (to 0.5, 1.0, 1.7, then 2.4 mg by week 16). Dose adjustments, pauses, or maintenance below 2.4 mg were permitted for tolerability. Measurements: Weight measured on calibrated digital scales; height and waist circumference (WC) measured per standardized protocol; BMI and waist-to-height ratio (WHtR) calculated. Sex and race were self-reported. Endpoints (assessed at baseline and years 2, 3, and 4 unless noted): Primary for this analysis included change in body weight (%) over time; categorical weight loss proportions (≥5%, ≥10%, ≥15%, ≥20%, ≥25%) at week 104; change in WC (cm) and percentage change in WHtR; improvement in BMI category (healthy <25; overweight 25–<30; class I 30–<35; class II 35–<40; class III ≥40 kg/m²) at week 104; proportions below sex- and race-specific WC risk cutoffs at week 104 (Asian women <80 cm; non-Asian women <88 cm; Asian men <88 cm; non-Asian men <102 cm), stratified by baseline BMI <35 vs ≥35 kg/m². Analyses conducted for the in-trial (intention-to-treat) period and a first on-treatment period (until first time off treatment >35 days) to estimate on-drug effects. Statistical analysis: Continuous endpoints analyzed by ANCOVA with treatment as fixed factor and baseline endpoint value as covariate. Missing data at week 104 imputed via multiple imputation separately by treatment arm using models fit to those with observed week-104 data; 500 imputed datasets combined using Rubin’s rules. Binary endpoints analyzed by logistic regression with treatment and baseline value as covariates, with multiple imputation preceding categorization. Predefined subgroup analyses included sex, age, BMI, body weight, region, race, ethnicity, WHtR, glycemic status, HbA1c levels, eGFR categories, and cardiovascular disease history; interaction tests were two-sided without multiplicity adjustment. Safety analyses included adverse events (AEs) leading to discontinuation and serious adverse events (SAEs), including SAE rates per 100 person-years by BMI classes.

- Weight change over 4 years: In-trial mean weight change at week 208 was -10.2% with semaglutide vs -1.5% with placebo; estimated treatment difference (ETD) -8.7 percentage points (95% CI -9.42 to -7.88; P<0.0001). First on-treatment mean change at week 208 was -11.7% vs -1.5%; ETD -10.2 percentage points (95% CI -11.0 to -9.42; P<0.0001). - Categorical weight loss at week 104 (in-trial): With semaglutide vs placebo, proportions achieving ≥5%: 67.8% vs 21.3%; ≥10%: 44.2% vs 6.9%; ≥15%: 22.9% vs 1.7%; ≥20%: 11.0% vs 0.6%; ≥25%: 4.9% vs 0.1%. - Waist circumference (WC): At week 208, mean WC change was -7.7 cm with semaglutide vs -1.3 cm with placebo; ETD -6.4 cm (95% CI -7.18 to -5.61; P<0.0001). At week 104, more semaglutide-treated patients crossed below sex- and race-specific WC risk cutoffs, especially when baseline BMI <35 kg/m² (semaglutide: 15.0% baseline to 41.2% at week 104; placebo: 14.3% to 18.0%). - Waist-to-height ratio (WHtR): Baseline mean WHtR was 0.66. At week 208, relative WHtR reduction was -6.9% with semaglutide vs -1.0% with placebo; ETD -5.87 percentage points (95% CI -6.56 to -5.17; P<0.0001). - BMI category improvement at week 104 (in-trial): 52.4% of semaglutide vs 15.7% of placebo patients improved by ≥1 BMI class. Proportion with BMI <25 kg/m² at week 104: 12.0% semaglutide vs 1.2% placebo (0% at baseline by design). Proportion with obesity (BMI ≥30 kg/m²) fell from 71.0% to 43.3% with semaglutide vs 71.9% to 67.9% with placebo. - Subgroup effects on mean percentage weight loss to week 104: Women had greater ETD than men (-11.1% vs -7.5%); younger age associated with slightly greater weight loss; higher baseline BMI modestly associated with greater percentage loss; Asian race/Asia region showed slightly smaller ETDs (e.g., Asian race ETD -7.27% [95% CI -8.09 to -6.46]). No meaningful heterogeneity by ethnicity, glycemic status (prediabetes vs normoglycemia), or renal function (eGFR categories). - Safety: AEs leading to permanent discontinuation occurred in 16.6% of semaglutide vs 8.2% of placebo (P<0.001), with higher discontinuation rates in lower BMI classes for semaglutide but not placebo. SAEs were fewer with semaglutide: overall SAEs in 33.4% vs 36.4% (P<0.001). SAE rates per 100 person-years by BMI class were lower with semaglutide vs placebo: <30 kg/m²: 43.23 vs 50.48; 30–<35: 43.54 vs 49.66; 35–<40: 51.07 vs 52.73; ≥40: 47.06 vs 60.85, with reductions driven primarily by cardiovascular events and infections. No detectable differences in hepatobiliary or gastrointestinal SAEs across BMI classes.

Semaglutide 2.4 mg once weekly produced substantial, clinically meaningful, and durable weight loss through 65 weeks that was sustained to 208 weeks, with concordant improvements in waist circumference and waist-to-height ratio. Benefits were observed across sexes, races, BMI and WHtR strata, regions, age groups, glycemic categories, and renal function groups, with women and individuals with higher baseline BMI showing slightly larger percentage losses and Asian race/region showing slightly smaller mean differences. Many semaglutide-treated participants moved to lower BMI categories; notable proportions achieved a healthy BMI and fell below sex- and race-specific WC risk thresholds, indicating improved adiposity-related risk profiles. Safety signals were generally favorable, with fewer serious adverse events than placebo; however, tolerability issues, reflected in higher discontinuations particularly at lower BMI, suggest potential differences in exposure or other factors (e.g., motivation, cultural norms). Weight loss magnitudes in SELECT were smaller than in dedicated weight-loss trials like STEP 1, likely due to trial design (cardiovascular outcomes focus, standard-of-care counseling not targeted to weight loss, and different population characteristics). The sustained long-term weight effects and broad subgroup efficacy suggest potential for meaningful public health impact in reducing obesity-related morbidity alongside cardiovascular risk reduction.

This prespecified analysis of the SELECT trial demonstrates that once-weekly subcutaneous semaglutide 2.4 mg safely and effectively induces clinically significant, durable weight loss and favorable anthropometric changes over up to 4 years in adults with overweight or obesity without diabetes but with established cardiovascular disease. Benefits were consistent across key demographic and clinical subgroups, while serious adverse events were fewer than with placebo. These findings support the broad use of semaglutide 2.4 mg in this population, complementing its cardiovascular benefits. Future research should evaluate cardiovascular risk reduction and weight outcomes in Asian populations with high cardiovascular risk and BMI <27 kg/m², incorporate detailed body composition assessments (fat mass, lean and muscle mass), and further explore weight-loss physiology in individuals with BMI ≤30 kg/m².

- Trial focus and counseling: SELECT was a cardiovascular outcomes trial, not a dedicated weight-loss study; lifestyle counseling was standard-of-care and not targeted to weight reduction, potentially attenuating weight loss relative to weight-management trials. - Body composition: Lack of direct body composition data (fat mass, lean mass, muscle mass) limits interpretation of how weight changes relate to changes in specific fat depots and lean tissue. - Subgroup nuances: Lower BMI (≤30 kg/m²) groups exhibited smaller percentage weight loss, possibly reflecting proximity to a physiological weight ‘settling point’; implications require further study. - Generalizability to lower BMI thresholds: The trial enrolled individuals with BMI ≥27 kg/m²; effects in high cardiovascular risk populations with lower BMI (particularly some Asian groups) were not directly assessed and warrant investigation. - Multiplicity: Subgroup interaction P values and CIs were not adjusted for multiplicity and should be interpreted cautiously.