Suicidality among users of glucagon‐like peptide‐1 receptor agonists: An emerging signal?

Abstract

Glucagon-like peptide-1 receptor agonists (GLP-1RAs) have revolutionized the treatment of diabetes and obesity. Multiple agents are available which are injected twice daily (exenatide twice-daily), once daily (lixisenatide, liraglutide), or once weekly (exenatide extended-release, dulaglutide, semaglutide). Recently, the once-weekly injectable dual gastric inhibitory polypeptide/GLP-1RA tirzepatide has entered the global market. Compared with previous classes of diabetes and weight loss medications, these medications offer substantially better glycaemic control and weight loss,1-5 as well as cardiovascular benefits in high-risk adults with diabetes and obesity.6-9 The substantial degree of weight loss afforded by these medications, as well as aggressive marketing and adoption among influencer circles, have led to an explosion in popular awareness and use.10, 11 Given the expected future growth of GLP-1RA use,12 the recent identification of suicide as a potential adverse effect has raised significant concern.13-15 In this commentary, we will review the origin of the safety signal, potential explanations for the purported link, the limitations of the current evidence, and how clinicians and patients might navigate the potential risk of suicidality with GLP-1RAs. From 2020 to 2022, GLP-1RA prescriptions increased 300%.16 In the last 3 months of 2022 alone, there were 9 million prescriptions for liraglutide, semaglutide and tirzepatide in the United States. This explosion in GLP-1RA use appears to have been engendered in large part by their potent weight loss benefits, with slightly under half of prescriptions (46.2%) being made to adults without diabetes. Ozempic, the semaglutide product indicated for diabetes, has been diverted off-label for weight loss, possibly contributing to current global shortages of this medication.17 Faced with high drug prices and limited availability, many patients have sought GLP-1RAs from compounding pharmacies and online sources.18, 19 These practices are controversial due to variable manufacturing oversight at compounding pharmacies, lack of effectiveness and safety data for the specific compounded forms of these medications, and the risk of online scams.20, 21 The market for GLP-1RAs is exploding, increasing the potential for harm from emerging adverse events. Suicidality is defined as suicide or attempted suicide, as well as suicidal ideation or intent.22 In July 2023, the Icelandic Medicines Agency reported two patients experiencing suicidal thoughts and one patient reporting self-injurious behaviour while using liraglutide or semaglutide. The European Medicines Agency soon announced a formal review of approximately 150 reports of self-injury and suicidal thoughts.15, 23 In the United States, the Food and Drug Administration's Adverse Event Reporting System (FAERS) has received 662 reports of suicidality, including 66 deaths, in GLP-1RA users since exenatide was first approved in 2005 (Figure 1).24 In one published case report, an elderly woman with long-standing, stable major depression, developed severe worsening of her symptoms on exenatide, that resolved with exenatide discontinuation, and returned with rechallenge.25 The news agency Reuters summarized 113 narratives from FAERS, and found that many (>50%) reported symptom onset shortly after GLP-1RA initiation and dose escalation, with symptoms waning on GLP-1RA withdrawal (approximately 40%).13 Some reports (27%) noted a previous history of psychiatric comorbidity, while a few (4%) explicitly reported no previous psychiatric history. Cases have been reported among individuals using GLP-1RAs for both diabetes and obesity.13, 14 Consequently, multiple agencies have initiated reviews of GLP-1RAs for suicidality, including Health Canada26 and the UK Medicines and Healthcare Products Regulatory Agency.27 While the total number of suicidality-related events reported in FAERS is a small fraction of the total 199 177 GLP-1RA-related reports (0.3%), including 3955 deaths, since 2005, the presence of a signal, defined narrowly as ‘a new or known adverse event that may be caused by a medicine and requires further investigation’, is clear.28 Whether and how GLP-1RAs may cause suicidality remains unclear. Diabetes and obesity, the indications for GLP-1RAs, have themselves been associated with depression and suicidality. In one systematic review of 15 longitudinal studies, both overweight and obesity increased the risk of subsequent depression (odds ratios [OR] 1.27 and 1.55), and depression increased the risk of subsequent obesity (OR 1.58).29 Despite this reciprocal association between obesity and depression, increased BMI, in most studies, is associated with lower risk of completed suicide.30 Most of these studies excluded adults with depression. Whether obesity is actually associated with suicide attempts or ideation remains unclear, since studies examining these associations are far fewer, with marked methodological limitations, and heterogeneity in findings, both between studies, and in key risk groups (e.g., men and women).30 Notably, in one of the few studies in which obesity was associated with suicide completion, the most substantial risk factor was a sudden and unexplained loss of weight (hazard ratio [HR] 5.58).31 The relationship between diabetes and subsequent depression is well established in meta-analyses of longitudinal studies (OR ~1.24–1.33).32, 33 Diabetes appears to be associated with an increased risk of completed suicide and suicidality (Risk ratio ~1.5 for completed suicide),34, 35 though considerable heterogeneity is evident.36 The risk appears to be higher with type 1 diabetes than type 2 diabetes, for which fewer data are available. Possible mechanisms linking diabetes and obesity to suicidality include depression, accrual of diabetes complications, comorbid alcohol use and poor psychosocial coping strategies, and social stigma associated with obesity.30, 37 Diabetes may affect central nervous system (CNS) biochemistry and function, including through hypothalamic–pituitary–adrenal axis dysregulation, impaired neurogenesis and brain-derived neurotrophic factor synthesis, inflammation, and hypoglycaemia.37 Confounding by indication is one potential explanation for reports of suicidality among GLP-1RA users. A direct incretin effect on the CNS is another explanation for neuropsychiatric disturbances in GLP-1RA users. GLP-1-secreting neurons and GLP-1 receptors are found in multiple locations in the brain. GLP-1 receptor signalling influences neuronal activity in the hypothalamus and the mesolimbic system, inducing increased satiety, visceral malaise and taste aversion, and reducing food appeal and hedonic food intake. While beneficial for reducing caloric intake and addiction-related behaviours,38 reduced pleasure may be distressing to some.39 Disrupted reward learning is thought to influence depressive mood and behaviour.40 GLP-1 additionally modulates the release of several neurotransmitters, including serotonin, dopamine, gamma-aminobutyric acid and glutamate, believed to play a role in the pathophysiology of depression.41 The CNS effects of GLP-1 are complex, and many other depression-, anxiety-, addiction- and psychosis-related pathways have been theorized. However, the CNS actions of GLP-1RAs have hitherto been thought, on balance, to ameliorate depression,41-43 based on evidence of GLP-1-related neuroprotection, including some recent reassuring functional magnetic resonance imaging (MRI) evidence,43-46 and mouse models showing reductions in depression and psychosis-related behaviours.47, 48 While the evidence for a protective association between GLP-1RAs and incident diagnoses of depression is mixed,49 GLP-1RAs do appear to reduce depressive symptom scores.50 It is worth noting that antidepressants, effective as they are at reducing depressive symptomatology and pre-existing suicidal ideation,51, 52 appear to carry an increased risk of new-onset suicidal behaviour in young individuals via mechanisms that are hotly debated.53-55 Among treatments indicated for weight loss in adults with obesity, GLP-1RAs would not be alone in raising concerns of suicidality. Phenteramine, found in the weight loss combination phenteramine/topiramate, is a stimulant, a class of agents, which are over-represented in suicide completers relative to the general population.56 Anti-epileptic medications, such as topiramate, have controversially been associated with suicidal ideation, and anti-epileptic medications in the United States carry a black box warning for suicidality.57 Bupropion, found in the weight loss combination bupropion/naltrexone, is a new-generation serotonergic-noradrenergic antidepressant. This broad class of antidepressants has been associated with increased suicide in observational studies.58 There is a well-established association between bariatric surgery, and an increased postoperative risk of suicide.59, 60 Theorized mechanisms include unrealistic or thwarted postoperative expectations for weight loss, potentiation of the effects of alcohol from alterations in alcohol processing and absorption in the gut, and reduced anti-depressant pharmacokinetic activity due to relative malabsorption. Given the CNS activity of GLP-1 and other incretins, post-surgical alterations in GLP-1 and related peptide signalling may also link bariatric surgery and suicidality.60 The weight loss treatment with the best-established relationship with suicidality is rimonabant, an inverse agonist for the cannabinoid receptor CB1. Licensed in Europe and Brazil as Acomplia, rimonabant was withdrawn after reports of serious neuropsychiatric events, including suicide, which were later verified in a cardiovascular trial.61, 62 Successful obesity treatments appear to modulate the central nervous pathways also implicated in psychiatric conditions, and suicidality may be a rare, but important, adverse effect to be considered. The current evidence for an association between GLP-1RAs and suicidality is limited, based essentially on case reports and pharmacovigilance surveillance systems such as FAERS. Systematic reporting of clinical details from suicidality cases is virtually non-existent and heterogeneous, leaving us wondering whether cases arise primarily from those using GLP-1RAs for diabetes, or for obesity, at what doses they are used, whether cases arise from those with or without psychiatric comorbidity, or from those with or without substantial weight loss, or symptoms, and after what duration of GLP-1RA use the cases arise. Pharmacovigilance surveillance systems such as FAERS have well-characterized limitations. There is no certainty in surveillance reports that the reported event is causally related to the medication exposure. Reporting is heterogeneous, and often does not contain enough detail to evaluate an event.63 Many factors may influence whether an event is reported to pharmacovigilance systems. Many events are not reported, while others may be reported disproportionately due to notoriety and ripple biases, triggered by public perceptions or awareness of potential harms.64, 65 FAERS is known to contain duplicate reports. The lack of a source population denominator in surveillance systems generally precludes the estimation of incidence rates, which makes comparison between agents or over time difficult.63 Trials of GLP-1RAs have not shown any difference in GLP-1RA versus comparator arms on neuropsychiatric adverse outcomes. A recent meta-analysis of liraglutide weight loss trials (the SCALE trial programme) pooled data from 5325 participants. Suicidal ideation was reported in 0.3% (n = 9) of liraglutide-treated participants versus 0.1% (n = 2) of those receiving placebo (descriptive analysis only). The risk of suicidal ideation on the Columbia-Suicide Severity Rating Scale was similar (1.0%) in the two groups.66 The SELECT trial randomized 17 604 individuals without diabetes to high-dose semaglutide versus placebo.9 Serious psychiatric adverse events were reported in 0.7% of semaglutide patients (n = 59) compared to 0.6% of placebo patients (n = 49). Psychiatric events requiring treatment discontinuation occurred in 0.4% (n = 33) versus 0.3% (n = 25) of individuals (descriptive analysis only). These findings are reassuring that severe neuropsychiatric morbidity is at least rare and may be balanced between GLP-1RA and comparator groups. However, the rarity of psychiatric considerations means that these analyses are also underpowered. Many GLP-1RA trials excluded those at highest risk for suicidality. For example, the STEP trial programme that tested the efficacy of semaglutide for weight loss had six exclusion criteria related to mental health,67 perhaps recognizing the potential for harm from underlying psychiatric comorbidity in adults with obesity. Many GLP-1RA diabetes trials, as well as the SELECT trial, excluded patients with psychiatric disorders based on investigators' judgement.5, 9 These exclusions are common for trials of non-psychiatric medications to minimize losses to follow-up, but nonetheless leave open the possibility that those most likely to experience psychiatric harm from a GLP-1RA are not represented in the efficacy trial populations. Given these limitations, further evaluation of a potential GLP-1RA and suicidality association may require real-world post-market comparative effectiveness studies. These studies can leverage large healthcare databases to compare diverse GLP-1RA users to population-based controls appropriately matched on comorbidities and indication. Such studies may still struggle to ascertain impacts on patient-reported suicidal ideation and related symptoms, which are not commonly tracked in such data sources. Prospective registries tracking depressive symptoms and self-harm ideation may be an essential part of future pharmacovigilance efforts in diabetes and obesity.68 Such registries may allow more targeted case-centred investigations of individuals with potential GLP-1RA-associated neuropsychiatric symptoms. For those with symptoms that are not life threatening, such as depression or mood changes, further investigations could include placebo-controlled crossover rechallenge trials featuring both clinical and functional MRI outcomes.46 It may be that GLP-1RAs pose a specific risk in rare individuals, and the ability to identify such individuals may be a future target for precision medicine. For diabetes and obesity, GLP-1RAs offer substantially greater benefits than previously available medications. These benefits are well established by both placebo-controlled and active comparator-controlled randomized controlled trials. For adults with diabetes, they include weight loss, glycaemic control, reductions in cardiovascular and kidney outcomes in high-risk adults, diabetes-related treatment satisfaction, weight-loss-related quality of life, and, potentially, all-cause mortality.1, 8, 69-71 For adults with obesity, GLP-1RAs reduce cardiovascular outcomes in high-risk adults,9 and offer weight loss in magnitudes4, 5 that were not previously possible without bariatric surgery.72, 73 Compared to bariatric surgery, GLP-1RAs are non-invasive, reversible simply by stopping the medication, avoid the peri-operative risks of surgery,74 and require much less ongoing provider monitoring and intervention. For most patients, these benefits will more than outweigh the emerging and uncertain risk of suicidality. Although cases of suicidality are tragic, they are, at least, rare. The inability of large trials to detect this association indicates that, even presuming a causal relationship, the absolute attributable risk would be small. Although plausible mechanisms exist to explain how GLP-1RAs might increase suicidality, plausible mechanisms also exist to support anti-depressant effects, or alternative, non-causal explanations. Yet suicidality concerns should not be dismissed, given the compelling nature of some of the emerging narrative accounts, and how much remains to be elucidated about the CNS actions of these medications. Clinicians should continue to balance the risks and benefits of GLP-1RA use in adults with diabetes or obesity in a patient-centred manner. We recommend being mindful of guidance from regulatory authorities. Routine screening for diabetes-related depression, as recommended in some guidelines,75, 76 may be helpful for identifying higher-risk individuals, and particular caution may be warranted for patients with a history of suicidal ideation or psychiatric comorbidity. However, given reports of suicidality in individuals with no relevant psychiatric history, we recommend maintaining clinical vigilance for unusual mood changes or suicidality. The availability of GLP-1RAs from compounding pharmacies and online prescribers will complicate efforts to mitigate potential emerging risks, such as suicidality. Traditional and social media will continue to play an inevitable role in disseminating the potential risks of GLP-1RAs. Increasingly, clinicians will be called on to communicate the limitations and uncertainties of risk from diverse evidence sources. Clinicians will need to provide balanced counselling and monitoring in an increasingly fragmented social landscape. While the benefits of GLP-1RAs will outweigh the uncertain and rare risk of suicidality for most patients, the successful realization of these benefits for more and more adults with diabetes and obesity will require prompt and serious consideration of suicidality as a rare, but high-risk, safety signal. Darren Lau has received in-kind research support from Abbott Laboratories. He has spoken at Novo Nordisk-sponsored events but has not received speaking fees or honoraria from the company. The authors have no other conflicts, financial or otherwise, to disclose. The peer review history for this article is available at https://www.webofscience.com/api/gateway/wos/peer-review/10.1111/dom.15459. The data that support the figure in this study are openly available from https://fis.fda.gov.

Topics & Concepts