INTRODUCTION
Glucagon-like peptide-1 receptor agonists (GLP-1 RAs) are analogs of the incretin hormone GLP-1. GLP-1 stimulates insulin secretion, inhibits glucagon secretion in the pancreas, and reduces gastrointestinal motility, thereby regulating blood glucose levels [1,2]. GLP-1 RAs are classified by their duration of action, with short-acting agents including exenatide and lixisenatide and long-acting agents including liraglutide, albiglutide, dulaglutide, extended-release exenatide, semaglutide, and tirzepatide, a dual GLP-1 and gastric inhibitory polypeptide (GIP) receptor agonist [1,3]. GLP-1 RAs were first approved for treating type 2 diabetes mellitus (T2DM) because of their glycemic control effects, and multiple studies have demonstrated they also exert cardiovascular protective effects [2,4]. Recently, owing to their weight loss effects by suppressing appetite and reducing gastrointestinal motility [2,5], GLP-1 RAs were additionally approved for weight loss inpatients with obesity/overweight. This led to a sharp rise in their use, including off-label use for cosmetic weight loss [6-8].
Clinical trials of liraglutide for obesity/overweight revealed slight imbalances in suicidal behavior and ideation rates compared with placebo arm (in adults: 0.3% vs. 0.1%; in pediatrics: 0.8% vs. 0%) [9,10], leading the U.S. Food and Drug Administration (FDA) to include suicide risk warnings on the label of GLP-1 RA agents for weight loss. According to the European Medicines Agency (EMA), 150 reports of suicide/self-injury associated with GLP-1 RAs prompted a review of this risk in July 2023 and concluded in April 2024 that the available evidence did not indicate a potential relationship between GLP-1 RAs and suicide/self-injury [11,12]. Previous studies have provided limited evidence. One cohort study suggested a lower suicide risk with semaglutide then with other anti-obesity or anti-T2DM drugs, but it did not consider the suicide risk of comparator drugs or analyze other GLP-1 RAs [13]. Another study found no significant suicide risk for GLP-1 RAs versus sodium-glucose cotransporter 2 inhibitors but did not examine non-T2DM patients use or individual components of GLP-1 RAs [14]. Both studies excluded off-label use for cosmetic weight loss. Disproportionality analyses using FDA Adverse Event Reporting System (FAERS) found no potential relationship between GLP-1 RAs and suicidality [15,16], but global pharmacovigilance analyses assessing the risk for each indication have not been conducted.
This highlights the need for a comprehensive study to assess the suicide/self-injury risk of GLP-1 RA, particularly for each indication. To address this, we performed a disproportionality analysis using the largest global pharmacovigilance database, World Health Organization (WHO) VigiBase, to further investigate potential relationships and identify the potential risk of each indication.
METHODS
Data source
We used the WHO-VigiBase, one of the largest international pharmacovigilance databases. As of April 11, 2024, it contained over 40 million of individual case safety reports from over 170 member countries, including those recorded in the FAERS, EudraVigilance, and other national databases from Asia, Africa, Latin America, and Oceania. VigiBase contains information related to case/demographic (variables: age group, gender, region, report type, date of reporting, notifier type, etc.), reaction (variables: Medical Dictionary for Regulatory Activities [MedDRA] term code, outcome, seriousness, etc.), and drug (variables: route of administration, dosage regimen, amount, frequency, time to onset, indication, dechallenge, and rechallenge, etc.) [17,18]. Since this study used publicly available, pre-existing, and anonymized data, ethical approval was not necessary.
Study design
Disproportionality analyses following case—non-case design were used to assess potential relationship between suicide/self-injury and GLP-1 RAs. VigiBase data were extracted from April 28, 2005, when GLP-1 RAs were first approved, to August 31, 2023. To analyze adverse events at the preferred terms (PTs) level and conduct analyses by drug ingredient and indication, duplicate drug-adverse event pairs (hereinafter referred to as reports) were removed based on the relevant variables (ReportId, drug record number [Drecno], indication, PT code). Only reports where the drug was considered ‘suspected’ were included. Duplicate reports, reports without adverse events, and reports of adverse events occurring before drug administration were excluded.
Analyses were conducted on GLP-1 RA components, together and separately, with integrated and stratified analyses by indication (T2DM treatment and weight loss). In the main analyses, all other drugs were selected as comparative drugs. In the stratified analyses by indication, to mitigate confounding by indications, drugs with similar indications but different mechanisms were selected as the comparative drug: the glucose-lowering agent excluding insulin for the T2DM treatment group and orlistat, the only peripherally acting anti-obesity agent, for the weight loss group. Central nervous system-acting appetite suppressants (e.g., bupropion/naltrexone and topiramate/phentermine) were not included in comparative drug because the risk of suicide/self-injury has already been identified (Supplementary Table 1).
Drugs and cases of interest
Among the GLP-1 RAs, exenatide, liraglutide, semaglutide, dulaglutide, albiglutide, and lixisenatide, which are approved by the FDA and EMA, and tirzepatide, a GLP-1, GIP dual receptor agonist, were selected as the drugs of interest (Supplementary Table 2). Cases retrieved using the PT included in the suicide/self-injury standardized MedDRA query were selected as cases of interest (Supplementary Table 3). To retrieve reports for each indication, 90 keywords were used (59 for T2DM treatment and 31 for weight loss) (Supplementary Table 4).
Statistical analysis
Descriptive analysis
To identify the characteristics of GLP-1 RA users related to suicide/self-injury reports, the frequencies and proportions of reports were examined across four age groups (17 years and younger, 18–44 years, 45–64 years, 65 years and older), sex, and indications (T2DM treatment and weight loss).
Disproportionality analysis
Disproportionality analyses were conducted to assess whether the reports of suicide/self-injury among GLP-1 RA users were higher than expected compared with the comparative drugs. Following a two-by-two contingency table (Supplementary Table 5), the main analyses were conducted using all adverse events reported for all other drugs as comparators. Analyses stratified by indication were conducted using all adverse events reported for the comparative drugs (T2DM treatment: glucose-lowering agents, excluding insulin; weight loss: orlistat) as active comparators for each indication. To detect signals, reporting odds ratios (ROR) and Bayesian confidence propagation neural network information component (IC) were utilized as signal detection methods, and thresholds for each measurement were established as follows: lower bound of 95% confidence interval (CI) of ROR ≥ 1; lower bound of 95% CI of IC (IC025) > 0 [19,20]. Additionally, disproportionality analyses for each dose of semaglutide and liraglutide were conducted to examine potential dose-response associations, and analyses by four age groups were conducted given the lack of post-marketing studies in pediatric patients.
RESULTS
Report characteristics
After applying the exclusion criteria, 111,752,363 reports (47.25%) out of 236,498,321 in the VigiBase were identified. Of these, 788,714 reports (0.71%) were associated with GLP-1 RAs, and 485 reports (0.06%) were linked to suicide/self-injury (Fig. 1). Among the GLP-1 RA reports, 475,069 (60.23%) were included in the stratified analysis by indication (T2DM treatment: 426,468 [54.07%]; weight loss: 48,601 [6.16%]). The T2DM group had 180 reports (0.04%) related to suicide/self-injury, and the weight loss group had 68 reports (0.14%), indicating a higher proportion in the weight loss group (Fig. 1). Among the 485 suicide/self-injury reports, liraglutide had the highest proportion (184 reports [37.94%]), followed by exenatide, semaglutide, dulaglutide, tirzepatide, and albiglutide, with no reports of lixisenatide. The 45–64 years age group had the most suicide/self-injury reports, and reports in females constituted over 50% of all GLP-1 RAs except dulaglutide (Table 1).
Disproportionality analysis
The main analyses were conducted by comparing GLP-1 RAs with all other drugs. In these analyses, no disproportionality signals for suicide/self-injury reports were detected for any GLP-1 RAs; total (ROR, 0.15; 95% CI, 0.14 to 0.16; IC025, –2.81), exenatide (ROR, 0.09; 95% CI, 0.07 to 0.11; IC025, –3.73), liraglutide (ROR, 0.32; 95% CI, 0.28 to 0.37; IC025, –1.83), semaglutide (ROR, 0.27; 95% CI, 0.22 to 0.33; IC025, –2.16), dulaglutide (ROR, 0.09; 95% CI, 0.07 to 0.12; IC025, –3.74), tirzepatide (ROR, 0.08; 95% CI, 0.04 to 0.15; IC025, –4.44), and albiglutide (ROR, 0.03; 95% CI, 0.01 to 0.09; IC025, –6.08) (Table 2).
Disproportionality analyses stratified by therapeutic indications were conducted comparing GLP-1 RAs with active comparative drugs (T2DM treatment: glucose-lowering agents other than insulin; weight loss: orlistat). InT2DM treatment group analyses, no disproportionality signals for suicide/self-injury reports were detected for any GLP-1 RAs; total (ROR, 0.09; 95% CI, 0.08 to 0.10; IC025, –3.26), exenatide (ROR, 0.07; 95% CI, 0.06 to 0.09; IC025, –3.89), liraglutide (ROR, 0.16; 95% CI, 0.12 to 0.22; IC025, –2.98), semaglutide (ROR, 0.12; 95% CI, 0.08 to 0.19; IC025, –3.52], dulaglutide (ROR, 0.10; 95% CI, 0.07 to 0.13; IC025, –3.73), tirzepatide (ROR, 0.18; 95% CI, 0.08 to 0.37; IC025, –3.29), and albiglutide (ROR, 0.03; 95% CI, 0.01 to 0.10; IC025, –6.28) (Table 3). However, in the weight loss group analyses, disproportionality signals for suicide/self-injury reports were detected in the overall class of GLP-1 RAs (ROR, 2.01; 95% CI, 1.34 to 3.01; IC025, 0.01), as well as in the liraglutide (ROR, 2.26; 95% CI, 1.47 to 3.49; IC025, 0.14) and semaglutide (ROR, 2.33; 95% CI 1.31 to 4.14; IC025, 0.22) (Table 3).
In the additional analyses, no potential dose-response associations were found for semaglutide and liraglutide, and findings from the analyses by age group were consistent with those of the main analyses (Supplementary Tables 6-8).
DISCUSSION
This is the first study, to our knowledge, using the largest global database, VigiBase, to assess the potential relationship between GLP-1 RAs and suicide/self-injury and to assess the potential risk for each indication compared with drugs with similar indications but different mechanisms of action. In the main analyses, we found no potential relationship between GLP-1 RAs and suicide/self-injury. However, in the stratified analyses by indication, we detected disproportionality signals, with ROR greater than 2 and IC025 greater than 0 for total GLP-1 RAs, liraglutide, and semaglutide in the weight loss group but not in the T2DM treatment group. This highlights the need for targeted safety monitoring in non-T2DM populations using GLP-1 RAs for weight loss.
From April 28, 2005, to August 31, 2023, the number of reports related to suicide/self-injury associated with GLP-1 RAs in the VigiBase was 485, accounting for 0.06% of all the GLP-1 RAs reports. Upon examining the clinical characteristics of GLP-1 RA users who reported suicide/self-injury, the proportion of female reports was significantly higher than that of male reports (female: 55.48%; male: 38.35%; unknown: 6.19%). Global epidemiology of suicide from the WHO Global Health Estimates for 2000–2019 found that males had a suicide rate 2.3 times higher than females [21]. This contrasts with the findings in GLP-1 RA users, where reports of suicide/self-injury were more prevalent among females. This discrepancy can be attributed to two factors. First, given the predominance of GLP-1 RA use in females than in males among patients with T2DM and obesity/overweight [22,23], the higher number of suicide/self-injury reports in females could be explained by the higher usage of GLP-1 RAs. Second, epidemiological studies found that the prevalence of suicide/self-injury is higher in females among patients with T2DM and obesity [24,25]. Therefore, the higher number of suicide/self-injury reports in females could be explained by the inherent suicide/self-injury risk in these underlying conditions.
Disproportionality analyses were conducted to investigate the potential relationship between GLP-1 RAs and suicide/self-injury. The results of the indication-integrated disproportionality analysis of GLP-1 RAs showed no potential relationship with suicide/self-injury for any GLP-1 RAs. These findings are consistent with those of previous disproportionality analysis studies using FAERS data [15,16].
Disproportionality analyses stratified by therapeutic indications were performed using an active comparator to analyze potential suicide/self-injury risks. The analysis results showed that disproportional signals were not detected in patients using GLP-1 RAs for T2DM treatment but were detected among patients using GLP-1 RAs for weight loss in the overall class of GLP-1 RAs, as well as in liraglutide and semaglutide.
The disproportionality analysis result of patients using GLP-1 RAs for T2DM treatment were consistent with those of previous cohort studies that found no risk of suicide/self-injury associated with the use of GLP-1 RAs in patients with T2DM [13,14]. However, disproportionality analysis result of populations using GLP-1 RAs for weight loss differed from a previous cohort study that found no risk of suicide/self-injury associated with semaglutide use in patients with obesity or overweight [13]. This discrepancy may be because of the inherent limitations of disproportionality analysis using pharmacovigilance databases. Considering that the cohort study used anti-obesity medications as comparative drugs, which included ingredients associated with suicide risk (bupropion/naltrexone, topiramate/phentermine), whereas this study used only orlistat, the discrepancy in results could be attributed to the differences in the comparators. On the other hand, the results in the weight loss group are consistent with findings from a community-based cohort study that shows a significant association between liraglutide and semaglutide treatment and a 98% increase in the risk of psychiatric disorders [26]. This cohort study focused solely on obese patients, while this study was conducted on a weight loss group that includes obese patients. Considering the significant increase in off-label use of GLP-1 RAs for cosmetic weight loss, the consistency of these results underscores the need for further investigation of suicide/self-injury risk associated with GLP-1 RAs in the weight loss group.
Given the differences in the results by indication, this study suggests that the characteristics of the non-T2DM populations using GLP-1 RAs for simple weight loss may be associated with the risk of suicide/self-injury. For example, considering that bupropion/naltrexone and topiramate/phentermine, which act on the hypothalamus to suppress appetite, similar to GLP-1 RAs, have a known risk of suicide/self-injury [27], the central nervous system-mediated appetite-suppressing mechanism of GLP-1 RAs may have a greater toll on individuals using GLP-1 RAs for simple weight loss than for T2DM treatment. Moreover, because GLP-1 RAs can be administered at higher doses for simple weight loss than for T2DM treatment, high-dose GLP-1 RA administration may be associated with suicidality. Additional dose-specific disproportionality analyses of semaglutide and liraglutide performed in this study did not identify a potential dose-response association; however, because of the low number of suicide/self-injury reports for each dose, the reliability of this analysis was expected to be low (Supplementary Tables 6, 7).
The data source for this study, VigiBase, contains reports based on subjective judgments, not medical diagnoses, and most reports lack detailed information such as dosage, indication, complications, and concomitant medication; therefore, data reliability is low. To mitigate this limitation, we only included reports where the drug was reported as ‘suspected,’ ensuring that only reports with a high probability of causality between the drug and the adverse event were analyzed. Furthermore, we did not consider factors that induce suicide/self-injury (such as a history of depression or suicide/self-injury and the concomitant use of drugs associated with suicide/self-injury risk). It is possible to identify and remove reports involving the concomitant use of drugs that are associated with suicide/self-injury risk using the information contained in the VigiBase, but the sequence of when these medications and GLP-1 RAs were administered cannot be determined; thus, to avoid excessive control, such reports were not excluded from the dataset [28]. Considering these limitations, caution should be exercised when interpreting the results.
In conclusion, using the global pharmacovigilance database VigiBase, we found no potential relationship between GLP-1 RAs use and suicide/self-injury when the indication was not considered. However, in the analysis by indication compared with active comparators, we found a potential relationship in patients using GLP-1 RAs for weight loss but not for T2DM treatment. Given this difference in results by indication, the characteristics of non-T2DM populations using GLP-1 RAs for simple weight loss may be associated with suicide/self-injury risk. Therefore, this study highlights the need for focused post-marketing safety monitoring of suicidality in non-T2DM populations using GLP-1 RAs for weight loss, along with more in-depth analysis of psychiatric safety in these populations, and that regulatory decisions should reflect these findings to mitigate potential risks and ensure public health safety.
PDF Links
PubReader
ePub Link
XML Download
Full text via DOI
Download Citation
Supplement1
Print
