ABSTRACT
Background: The benefit of Glucagon-like Peptide-1 (GLP-1) receptor agonists (RAs) in weight reduction against potential harms remains unclear. This study aimed at evaluating the benefit-harm balance of initiating GLP-1 RAs versus placebo for weight loss in people living with overweight and obesity but without diabetes. Methods: We performed benefit-harm balance modelling, which will be updated as new evidence emerges. We searched for randomised controlled trials (RCTs) in PubMed, controlled trials registry, drug approval and regulatory documents, and outcome preference weights as of April 10, 2024. We synthesize data using pairwise meta-analysis to estimate the effect of GLP-1 RAs to inform the benefit-harm balance modelling. We predicted the absolute effects of the positive and negative outcomes over 1 and 2 years of treatment using exponential models. We applied preference weights to the outcomes, ranging from 0 for least concerning to 1.0 for most concerning. We then calculated whether the benefit of achieving 5% and 10% weight loss outweighed the harms on a common scale. The analyses accounted for the statistical uncertainties of treatment effects, preference weights, and outcome risks. Findings: We included 8 RCTs involving 8847 participants. The pooled average age was 46.7 years, with the majority being women (74%) and people living with obesity (96%). Of 1000 persons treated with GLP-1 RAs for 2 years, 375 (95% confidence interval 352 to 399) achieved a 10% weight loss, and 318 (296 to 339) achieved a 5% weight loss compared to those treated with placebo. Several harm outcomes were more frequent in the GLP-1 RA group, including 41 abdominal pain events per 1000 persons over 2 years (19 to 69), cholelithiasis (8, 1 to 21), constipation (118, 78 to 164), diarrhoea (100, 42 to 173), alopecia (57, 10 to 176), hypoglycaemia (17, 1 to 68), injection site reactions (4, -3 to 19), and vomiting (110, 80 to 145) among others. Achieving a 10% weight loss with GLP-1 RA therapy outweighed the cumulative harms, with a net benefit probability of 0.97 at year 1 and 0.91 at year 2. The absolute net benefit was equivalent to 104 (100 to 112) per 1000 persons achieving a 10% weight loss over 2 years without experiencing any worrisome harm. A 5% weight loss did not show a net benefit, with probabilities of 0.13 and 0.01 at year 1 and year 2, respectively. However, these benefits were sensitive to preference weights, suggesting that even a 5% weight loss could be net beneficial for individuals with less concern about harm outcomes. The net benefit for a 10% weight loss was highest for semaglutide, followed by liraglutide and tirzepatide, with 2-year probabilities of 0.96, 0.72, and 0.60, respectively. Interpretation: The benefit of GLP-1 RAs exceeded the harms for weight loss in the first 2 years of treatment, yet the net benefit was dependent on individual' treatment goals (10% or 5% weight loss) and willingness to accept harms in pursuit of weight loss. This implies that treatment decisions have to be personalized to individuals to optimize benefits and reduce harms and overuse of treatments. Due to varying evidence, especially regarding harm outcomes across studies, it is necessary to continuously update and monitor the benefit-harm balance of GLP-1 RAs. Funding: SNSF and LOOP Zurich.
