Glucagon-like Peptide-1 Receptor Agonists and Suicidal Ideation: Analysis of Real-Word Data Collected in the European Pharmacovigilance Database

On 11 July 2023, the European Medicine Agency (EMA) started an ongoing safety revision of glucagon-like peptide-1 receptor agonists (GLP-1 RAs), which is being carried out by the EMA Pharmacovigilance Risk Assessment Committee (PRAC) [1]. In particular, the data review focuses on the risk of suicidal thoughts and thoughts of self-harm in patients treated with these drugs.

GLP-1 RAs are incretin-mimetic agents, being mainly recommended for type 2 diabetes mellitus (T2DM) [2]. By stimulating GLP-1 receptors, this class of drugs can lower blood glucose, increase insulin secretion induced by hyperglycemia, and suppress glucagon release in hyper- or euglycemia. To date, in Europe, six GLP-1 RAs have been authorized. Exenatide was the first one, being authorized in 2006 for T2DM. Subsequently, the European Agency approved liraglutide (in 2009), lixisenatide (in 2013), combination liraglutide/insulin degludec, and dulaglutide (both in 2014), and most recently, semaglutide (in 2018). Another GLP-1 RA, albiglutide, was withdrawn from the European market upon request of the marketing authorization holder in 2018 for commercial reasons. All GLP-1 RAs are administered subcutaneously. The new oral semaglutide formulation, introduced in 2020, represents a pharmaceutical innovation aimed at increasing compliance in diabetic patients. Moreover, GLP-1 RAs have demonstrated cardiovascular benefits in high-risk patients, such as those with coronary syndromes, heart failure, or chronic kidney diseases, representing pharmacological progress in reducing the burden not only of diabetes mellitus but also of its complications [3]. The most common adverse events of GLP-1 RAs are gastrointestinal symptoms (nausea, vomiting, diarrhoea), which occur at start of the treatment and during dose escalation. Therefore, to reduce the onset of gastrointestinal events, gradual up-titration should be performed [3].

Beyond hypoglycaemic, endocrine, and cardiovascular effects, several decades of research have also shown the extended actions of GLP-1 RAs, which include decelerated gastric emptying and a reduction in food intake and body weight [4]. These are due to the expression of GLP-1 receptors in the central nervous system, including the area of the brain that regulates appetite [5]. For this reason, GLP-1 RAs increase feelings of fullness and decrease feelings of hunger. These additional actions have promoted their use for body weight control. Indeed, two GLP-1 RAs (liraglutide and semaglutide) were authorized in 2015 and 2022, respectively, for obesity or overweight treatment. In particular, Saxenda^® (liraglutide) and Wegovy^® (semaglutide) were approved as adjuncts to a reduced-calorie diet and increased physical activity for weight management in adults with a body mass index (BMI) of ≥30 kg/m² (obesity) or ≥27 kg/m² to <30 kg/m² (overweight) in the presence of at least one weight-related comorbidity (e.g., prediabetes, type 2 diabetes mellitus, hypertension, dyslipidaemia, obstructive sleep apnoea, or cardiovascular disease) and as an adjunct to healthy nutrition and increased physical activity for weight management in obese adolescents (≥12 years) with a BMI corresponding to 30 kg/m² for adults or with body weight above 60 kg.

Recently, the increased demand for semaglutide and the possible misuse of these antidiabetics as a weight loss treatment in non-obese people have been described in the literature and by the media [6,7,8]. Almost simultaneously, reports of suicidal thoughts in people using liraglutide and semaglutide were highlighted [1]. For this reason, the PRAC started a safety review initially involving semaglutide and liraglutide and subsequently expanded it to the entire class of GLP-1 RAs [1]. The evidence evaluating this risk is meagre. Few recent studies have questioned this association [9,10,11], and no study has compared this risk between different GLP-1 RAs. Clinical trials have several limitations related to their duration and patient inclusion criteria that can impede investigations on psychiatric adverse events. Consequently, psychiatric safety issues may emerge from clinical practice through the analysis of real-world safety data [12]. Considering the clinical relevance of this potential suicidal risk associated with GLP-1 RAs and the availability of pharmacovigilance databases as a source of real-world information, the present study aimed to evaluate the characteristics of suicidal events reported with GLP-1 RAs through the analysis of the pharmacovigilance database EudraVigilance (EV).

The present study is the first pharmacovigilance study to evaluate the reporting of suicidal events as suspected adverse drug reactions associated with the GLP1 RAs, by using the European pharmacovigilance database. GLP1 RA-induced suicidal events were mentioned in the European risk management plans for such medicines. They were brought to the attention of the Icelandic Medicines Agency after the reporting of cases of suicidal thoughts and self-harm in people using liraglutide and semaglutide [1]. This study provides reporting frequencies of suicide-related events with GLP-1 RAs. However, the causation requires further investigation, and a conclusion will be provided by the PRAC in the future. In this study, we found a 2-fold increase in the reporting of suicidal events with semaglutide compared with exenatide and dulaglutide. This estimate was even higher in the comparison between liraglutide and exenatide or dulaglutide, with 4- and 3.5-fold increases in the reporting of suicidal events, respectively. Moreover, when the comparison was between semaglutide and liraglutide, semaglutide showed a lower reporting of suicidal events. The more recent marketing authorization of semaglutide may explain this finding and the lower reporting of suicidal events. Both medicines also have the central effect of reducing appetite and are used not only for the treatment of diabetes mellitus but also for the treatment of obesity and overweight [13,14]. Generally, all centrally acting anti-obesity drugs draw attention for their neuropsychiatric safety [15,16]. A pooled post hoc analysis of neuropsychiatric safety data from the liraglutide weight management clinical trials program on liraglutide 3.0 mg found that the incidences of depression, anxiety, and insomnia were low (≤4%) in both the liraglutide and placebo groups, but with a slight increase in insomnia and suicidal ideation for liraglutide [17]. On the contrary, clinical trials with lower doses of liraglutide (up to 1.8 mg) for T2DM did not find any safety signal for neuropsychiatric events [18,19]. Therefore, it seems that the suicidal risk of GLP1 RAs could be related to the use of higher doses for weight management. Indeed, based on this consideration, the US prescribing information already states that patients should be monitored during treatment with liraglutide 3.0 mg for depression or suicidal thoughts and should discontinue the drug if these symptoms appear [20]. The same warning is also reported in the US prescribing information of semaglutide for obesity (Wegovy^®), which is also used at higher dosages for weight control than T2DM [21]. However, these warnings are not reported in any European summary of the product characteristics (SmPCs) for any GLP-1 RA.

A possible mechanism explaining the risk of suicidal events with GLP-1 RAs is related to their action in the hypothalamus [5] since the hyperactivity of the hypothalamic–pituitary–adrenal axis has been associated with suicidal behaviours [22]. However, the evidence for the role of GLP1 in the nervous system is conflicting. There is also evidence showing the ability of GLP1 to attenuate neuroinflammation, protect neurons and glia from oxidative stress, and improve neurotransmitter balance [23]. Moreover, the potential interaction between GLP-1 and serotonin pathways has been the objective of recent studies for energy balance regulation [24,25]. We cannot exclude the possibility that this interaction may also play a role in the pathogenesis of depression and suicidal ideation. Therefore, any causal conclusion is even more complicated. Indeed, many risk factors may contribute to or precipitate the onset of suicidal behaviours in patients treated with GLP1 RAs, including psychosocial and biological factors.

Considering that suicide is a public health problem, the identification of risk factors is fundamental, representing the fourth leading cause among the 15–29-years-old group. According to the latest World Health Organization (WHO) data, more than 700,000 people die each year due to suicide [22]. Suicides require timely, evidence-based, and adequate interventions based on the identified risk factors [26].

Multiple biological factors can also contribute to the pathophysiology of adverse events [27,28], including suicidal events [29]. Among these, sex can contribute to suicidal behaviours. In the literature, suicide attempts are more frequent in females, while completed suicides are three times more common in males [22]. In line with this, in our analysis, females more frequently described suicidal ideation and suicide attempts in ICSRs, while the rate of completed suicide was three times higher in males than females. Moreover, underlying or concomitant diseases and concomitant drug treatments are possible confounding factors for suicide. Several diseases and pharmacological treatments can increase suicide risk. First of all, diabetic and obese patients represent a fragile population, being more exposed to the risk of suicide than the general population because of their underlying pathologies [30]. Both diabetes [31,32] and obesity [33,34] have been identified as suicidal risk factors. Among obese patients, the association between obesity and suicide seems to be influenced by sex, being greater for women than men [35]. Moreover, suicidal ideation could also be a consequence of the therapeutic inefficacy of GLP1 RAs used for weight control. A cross-sectional study suggested an association between weight control failure and suicidal ideation in overweight and obese adults, especially in women [36]. However, in our dataset, no ICSRs reported therapeutic ineffectiveness. Regarding diabetes, both types of diabetes were associated with an increased occurrence of psychiatric disorders, including suicidal ideation. Psychiatric disorders, in particular depression, are well-known suicide risk factors [22]. In our dataset, antidepressants were the most reported suspected as well as concomitant drugs. Moreover, most suspected drugs were also antiepileptics. Even if a recent meta-analysis seems to question it [37], the link between suicide and these medications is widely described in the literature [38,39,40] and reported in all SmPCs of these drugs. Finally, other suspected and concomitant medicines reported in our ICSRs may have influenced the development of symptoms of suicidal behaviour. Statins, for example, have been associated with higher reporting of psychiatric adverse events in a previous pharmacovigilance study [41]. Proton pump inhibitors have also been investigated for their psychiatric effects, including suicidal ideation and depression [42]. Renin–angiotensin–aldosterone system (RAAS) blockers have also been found to be related to the risk of suicide due to the central role played by RAAS in mood disorders [43]. Considering the many contributing risk factors for suicide, understanding its causal relationship with medicine may be very hard, thus requiring further research on this topic.

The main strength of our study is the use of a broad data source, the EV database, which allows the evaluation of safety cases all over Europe. Moreover, a pharmacovigilance database has a low cost and is helpful for characterizing drug safety profiles in the real world. However, this study also has several limitations. First of all, some cases of suicide with GLP1 RAs may not have been reported to the national drug authorities and thus not submitted to EV (the underreporting phenomenon). Underreporting is a notable limitation of pharmacovigilance systems, as only 6–10% of all adverse events are reported to regulatory authorities [44,45]. This underreporting can prevent the quantification of the incidence of adverse events and the risk estimates [45], and it can delay the identification of safety signals, with repercussions for public health [45,46,47]. Another limitation is the quality of information reported. Indeed, ICSRs may be incomplete and lacking useful clinical information, such as clinical history and concomitant comorbidities and medications. This lack of information can impede the evaluation of confounding factors. From our data source, we cannot evaluate data on lifestyle factors, psychiatric comorbidities, or other medical conditions that could be associated with the risk of suicide. We also cannot retrieve information on the exact dates of administration and event onset. Both underreporting of ICSRs and possibly missing information can introduce information biases in the analysis. Moreover, the exact number of patients exposed to GLP1 RAs is unavailable in EV (real users), where we can only use the total number of events of each drug as a denominator for disproportionality analyses. Finally, our analyses were restricted to comparisons within the drug class of GLP1 RAs. Considering these limitations, our study only aimed to analyse ICSRs related to suicide with GLP1 RAs and to show the reporting probability of these adverse events but refrained from asserting any direct causal association between GLP-1 RAs and the risk of suicidal events.

Our study found 230 ICSRs describing suicidal events with GLP1 RAs. The most reported drugs were semaglutide and liraglutide, for which the disproportionality analyses found higher reporting probabilities than other GLP1 RAs. Among semaglutide and liraglutide ICSRs, 13.1% and 68.2% were related to the formulation authorized for weight management, respectively. The most reported events were suicidal ideation and suicide attempts in females and completed suicide in men. However, considering all the possible contributing factors for suicide, further investigations that take into account these aspects are needed, because the clinical trials were not powered enough to evaluate specifically neuropsychiatric events. This research underscores the need for strictly adhering to evidence-based practices when prescribing GLP1 RAs. Moreover, it highlights the importance of strengthening pharmacovigilance activities to facilitate the availability of more data concerning GLP-1 RAs in the future.

The following supporting information can be downloaded at: https://www.mdpi.com/article/10.3390/ph17020147/s1↗. Supplementary Table S1. Other suspected drugs reported in ICSRs of suicidal events with dulaglutide, exenatide, liraglutide, liraglutide/insulin degludec, or semaglutide classified by the Anatomical Therapeutic Chemical (ATC) Classification. Supplementary Table S2. Active ingredients of other suspected drugs reported in ICSRs of suicidal events with dulaglutide, exenatide, liraglutide, liraglutide/insulin degludec, or semaglutide. Supplementary Table S3. Concomitant drugs reported in ICSRs of suicidal events with dulaglutide, exenatide, liraglutide, liraglutide/insulin degludec, or semaglutide classified by the Anatomical Therapeutic Chemical (ATC) Classification. Supplementary Table S4. Active ingredients of concomitant drugs reported in ICSRs of suicidal events with dulaglutide, exenatide, liraglutide, liraglutide/insulin degludec, or semaglutide.

R.R.: Conceptualization, Methodology, Data curation, Writing—original draft. A.M.: Conceptualization, Methodology, Data curation, Writing—original draft. A.S.: Methodology, Data curation. C.C.: Methodology, Data curation. D.T.: Investigation, Writing—review and editing, Supervision. L.S.: Investigation, Writing—review and editing, Supervision. F.R.: Conceptualization, Validation, Project administration. G.P.: Writing—review and editing, Validation. A.C.: Validation, Supervision, Funding acquisition. Final approval of manuscript: all authors. Accountable for all aspects of the work: all authors. All authors have read and agreed to the published version of the manuscript.

Not applicable.

Not applicable.

EV data are publicly available at https://www.adrreports.eu/↗, accessed on 13 July 2023.

The authors declare no conflicts of interest.

This research was funded under the Italian fund “Manutenzione e continuità delle attività del Centro Regionale di Farmacovigilanza e Farmacoepidemiologia Regione Campania 2012–2014” of the Italian Medicine Agency (AIFA), funding number B61G17000170005.