‘Disproportionate’ Suicide/Self-Harm Signal With Semaglutide: WHO Database

A deep dive into public reporting tracked by the World Health Organization (WHO) is raising the possibility that the GLP-1 receptor agonist semaglutide, but not liraglutide (both Novo Nordisk), is associated with suicide and self-injury.

The new analysis also showed that the number of reports for suicidality is increasing over recent years, “which may indicate a widening therapeutic scope in obesity and accumulating clinical experience,” Georgios Schoretsanitis, MD, PhD (Zucker Hillside Hospital/Northwell Health, Glen Oaks, NY), and colleagues write in their paper published online August 20, 2024, in JAMA Network Open.

The new numbers come on the heels of announcements by both European and US regulators that they had concluded their own investigations into suicidality with GLP-1 receptor agonists and determined that the link was not causal.

Schoretsanitis stressed that patients and clinicians should not interpret these latest data as proof of a causal relationship between use of semaglutide and suicide and self-harm, noting that pharmacovigilance studies such as theirs can only identify associations.

“This pharmacovigilance signal is like an early warning system and should not be immediately viewed as an alarm,” he told TCTMD. “A signal indicates that something unusual has been noticed and that more research is needed to figure out if the drug is indeed causing the issue.”

The study reaffirms the importance of “timely, proactive postmarketing surveillance” with the GLP-1 receptor agonists, particularly semaglutide, said Schoretsanitis.

“We encourage healthcare professionals to monitor and advise patients using GLP-1 receptor agonists to report new or worsening depression, suicidal thoughts, or any unusual changes in mood or behavior,” he said. “Future studies should provide a closer look into the risk of suicidal ideation associated with semaglutide in persons with previous history of mental disorders or in persons with comorbid mental disorders.”

This pharmacovigilance signal is like an early warning system and should not be immediately viewed as an alarm. Georgios Schoretsanitis

Roger McIntyre, MD (University of Toronto, Canada), who has studied the risk of suicidality with the GLP-1 receptor agonists, said that these new data, at first glance, suggest there is a “red flag” with semaglutide, marketed as Wegovy for type 2 diabetes and Ozempic for weight loss, but he stressed that pharmacovigilance databases provide incomplete information to make any conclusive causal link.

“In order for us to arrive at the conclusion of causation versus association, there’s a number of other additional pieces of information that are required, and unfortunately this type of pharmacovigilance database doesn't provide it,” he told TCTMD. “For example, we know very little about who these people are. We don't know their background history, we don't know their psychiatric history, and we don't know their prior treatment history. There could be aspects about people who are more likely to get semaglutide that makes them different than other people.”

To establish causation between a drug and serious adverse event or safety concern, several criteria need to be met. The Bradford-Hill criteria, for example, include nine different variables, such as the strength of the association, consistency of findings, dose-response relationship, and biological plausibility, among others, to help determine if observed association is causal. In an editorial published earlier this year, McIntyre said the some of the criteria may apply to GLP-1 receptor agonists, but there isn’t compelling evidence for cause and effect.  

Despite the inherent limitations, databases of individual case safety reports are important for picking up early signals of potential harm, he agreed.

“Now, a signal might mean it's something real, it's causative, and it's something we have to get on top of, or it could also just be confounding,” said McIntyre. “If I'm taking one of these treatments, I would like to know about all the reports around this treatment, but at the same time it needs to be interpreted in the broader context and interpreted carefully.”

Adds to Existing Surveillance Reports

Last year, the European Medicines Agency and US Food and Drug Administration both launched investigations into the potential harms associated with the GLP-1 receptor agonists, including liraglutide (Victoza/Saxenda; Novo Nordisk) and others, after concerns raised in postmarketing surveillance data. 

Based on data from the FDA Adverse Event Reporting System (FAERS), the agency concluded in January 2024 they could not find an association between use of GLP-1 receptor agonists and suicidal thoughts or actions. European regulators came to the same conclusion, stating in April 2024 there was “no evidence to support a causal association between GLP-1 receptor agonists and suicidal and self-injurious thoughts and actions.” 

Other analyses have supported those conclusions. One, in fact, showed that compared with non-GLP-1 receptor agonists, semaglutide was associated with a lower risk of incident and recurrent suicidal ideation in people who were overweight or obese prescribed the medication for weight loss. Another group, this was one led by McIntyre and colleagues, also looked into the disproportionate reporting of suicidal ideation and depression with semaglutide and liraglutide and concluded that no causal link existed once potential confounding factors were taken into account.

For the new study, investigators analyzed individual case safety reports from the WHO postmarketing surveillance database. They collected descriptive information on demographic and clinical characteristics, including different indications for GLP-1 receptor agonist use (eg, diabetes, weight loss, off-label indications) and performed a disproportionality analysis using frequentist and Bayesian methods to identify potential safety signals.

“In pharmacovigilance, researchers often compare ‘cases’ to ‘noncases’ to understand if a drug might be associated with certain adverse events,” explained Schoretsanitis. “Cases are reports where patients have experienced a specific adverse event after taking a drug. Noncases are reports of people who took the same drug but did not experience that specific adverse event. By comparing these groups, disproportionality analysis aims to detect unusual patterns in the data.”

As of August 2023, there were 36,172,078 total adverse reports in the database, including 107 cases of suicide and/or self-harm with semaglutide (median age 48 years, 55% female, median duration of treatment 28 days) and 162 cases with liraglutide (median age 47 years, 61% female, median duration of treatment 46 days). Roughly one-third of patients were prescribed the medication for off-label use, one-quarter for weight loss, and 20% to 24% for type 2 diabetes. Cases of semaglutide-associated suicidality were reported mainly by consumers/patients, but physicians reported the majority of cases with liraglutide.

A signal of disproportionality emerged for semaglutide-associated suicidal ideation (reported OR 1.45; 95% 1.18-1.77 and Bayesian information component 0.53; 95% CI 0.19-0.78) when compared with dapagliflozin, metformin, and orlistat, drugs chosen to offset the risk of confounding by indication (these drugs are also prescribed for obesity and type 2 diabetes). No signal emerged with liraglutide.

In the sensitivity analysis, the signal of disproportionality with semaglutide remained significant in patients who reported also taking antidepressants and benzodiazepines, and when compared with dapagliflozin, metformin, and orlistat.

For patients co-prescribed antidepressants and/or benzodiazepines, Schoretsanitis suggests physicians should inform them about the risks and assess their psychiatric history and current mental state before starting treatment with semaglutide.

“For patients with history of mental disorders or suicidal ideation/behaviors/attempts, physicians should be cautious and regularly monitor their mental state while taking semaglutide,” said Schoretsanitis. “If needed, such as in case of persisting suicidal ideation or in case of other relevant mental disorders, the treating physician should involve different specialists, including a psychiatrist and/or clinical psychologists, for a psychological and psychiatric evaluation.”  

GLP-1s Might Even Be Beneficial

As for why they observed a signal with semaglutide, but not liraglutide, Schoretsanitis stressed that despite belonging to the same drug class, the two medications differ in terms of effectiveness, with semaglutide being the more effective agent for glycemic control and weight loss. “Thus, it is not a surprise that they may also differ in terms of safety, although head-to-head comparisons are not available to my knowledge,” he said.

A signal might mean it's something real, it's causative, and it's something we have to get on top of, or it could also just be confounding. Roger McIntyre

To TCTMD, McIntyre pointed out there is evidence that GLP-1 receptor agonists might even have beneficial effects on psychopathology and may potentially be used as a treatment for depression and those at risk for suicide.

“There's a lot of interest in these drugs for the treatment of alcohol use disorder, nicotine or tobacco use, and cognitive impairment,” he said. “If anything, the evidence that we have seems to coalesce around a possibility—I’m choosing my words carefully, as these things are still very early—that these drugs might be helpful for mental illness rather than somehow engendering it or making people suicidal. But we leave open the door, there’s always a possibility [of harm]. You always have stay humble because anything is possible, and we have to go where the data take us.”

In an editorial, Francesco Salvo, MD, PhD (Université de Bordeaux, INSERM, France), and Jean-Luc Faillie, MD, PhD (Université de Montpellier, France), agree with position of the US FDA and say that continued caution with use of the drugs is reasonable, particularly in patients with a history of depression or suicide attempts.