GLP-1 receptor agonists and suicidal ideation/behavior: subject-level analysis of clinical trials

Background Glucagon-like peptide-1 receptor agonists (GLP-1 RA) are approved in the USA for indications including glycemic control in persons with type 2 diabetes (T2D) and weight reduction in persons with obesity/overweight. We conducted a subject-level analysis of clinical trials to evaluate a potential drug effect of GLP-1 RA on suicidal ideation and behavior (SI/B) and other psychiatric endpoints upon learning of postmarketing reports of SI/B in patients taking GLP-1RA. Methods We included prospective, randomized, placebo-controlled trials of GLP-1 RAs approved from 2005 to 2023 that evaluated glycemic control, weight reduction, or cardiovascular risk in persons with T2D or obesity/overweight. The primary outcome was the first occurrence of SI/B retrospectively identified by the Standard Medical Dictionary for Regulatory Activities Query and classified using the Columbia Classification Algorithm of Suicide Assessment from subject-level data. Other endpoints were identified using adverse event queries. Mantel–Haenszel methods stratified by trial were used to estimate the rate ratio (RR) and rate difference (RD) comparing GLP-1 RA to placebo. Results Of the 91 placebo-controlled trials ( N = 107,910; 271,000 person-years), 62 evaluated persons with T2D; 29 evaluated persons with obesity or overweight. The GLP-1 RA arm had 55 SI/B events during 14.3 × 10 4 PY, corresponding to an incidence rate of 3.6 SI/B events/10 4 PY. The placebo arm had 44 SI/B events during 12.8 × 10 4 PY, corresponding to an incidence rate of 3.6 SI/B events/10 4 PY. The RR for SI/B comparing GLP-1 RA to placebo was 1.0 (95% confidence interval [CI] 0.7, 1.5); the RD was 0.0 (95% CI − 1.4, 1.5). In subgroup analyses, the RR was 1.0 (95% CI 0.5, 1.7) in T2D trials and 1.0 (95% CI 0.6, 1.8) in obesity/overweight trials. Supplemental and sensitivity analysis results were consistent with the primary analysis of SI/B. The RR for anxiety, depression, irritability, and psychosis ranged from 0.9 to 1.3 with 95% CIs that included the null of 1.0. Conclusions In the largest known subject-level analysis of clinical trial participants across GLP-1 RA development programs, GLP-1 RA did not show an increased risk for suicidal ideation and behavior compared with placebo.