August 20, 2024

expert reaction to study looking at semaglutide, liraglutide, and suicidality

A study published in JAMA Network Open looks at semaglutide use and suicidality. 

Dr Riccardo De Giorgi, Clinical Lecturer in the Department of Psychiatry, University of Oxford, said:

“Disproportionality analysis is used for “hypothesis generation” (rather than “testing”), meaning that potentially harmful associations need to be verified and quantified in other studies with a different design.  This is an important, robust study of disproportionality analysis of a rare but serious adverse event, namely suicidality, in people using semaglutide.  There is a degree of mismatch with previous similar pharmacovigilance studies using the same methodology (one reporting a potentially harmful association between semaglutide and sucidality, one reporting no association) and with cohort studies using routinely collected electronic healthcare records (again one reporting no association between semaglutide and sucidality, one in fact reporting a beneficial association).  As noted by the authors, one advantage of their analysis is that it can be generalised to patients taking GLP-1 RAs off-label (i.e., without a diagnosis of diabetes or obesity).  Currently, we do not have any plausible biological explanation behind these potential associations – whether positive, negative, or null.  A call for caution and for further investigations in these groups of people, including those with pre-existing psychiatric disorders such as depression, is probably justified.”

Dr Nerys Astbury, Senior Researcher – Diet & Obesity, Nuffield Department of Primary Health Care Sciences, University of Oxford, said:

“The results of this study show that a person taking semaglutide in this database of reported adverse drug reactions is 1.45 is times as likely to mention suicidal ideation than any other adverse drug reaction in all other drugs.  In people who also take antidepressants in this database suicidal ideation is reported 4.45 times as likely for semagulatide than for any other adverse drug reaction in all other drugs.

“Because of this disproportionality, these results suggest further close investigation is needed to explore possible effects of GLP-1 agonists, now popular weight loss medications, on adverse mental health outcomes.

“Because of the inherent limitations of the data used in the study the authors only have data on suicidality in people who take semaglutide and they don’t know how many semaglutide users don’t have this reaction.  For this reason they can’t say how likely this reaction is.

“It should be noted that the number of suicidal ideation ADRs are very low (94 in total).

“We typically expect ADR (adverse drug reactions) related to drugs to go away after discontinuation – however here the authors report that they do go away in 63% of cases, but that also means they persist in 37%; this warrants further investigation as it could be that at least some of the effects seen are not causally related to the drugs, or that the effects related to the drugs persist even after discontinuation?

“This is not the first time that anti-obesity medications which target the reward responses to food (and other substances) have been associated with adverse side effects.  Rimonabant, was originally approved medication which helped supress appetite through selectively blocking the Canabanoid 1 receptor, and was withdrawn in 2008 due to the risk of serious psychiatric problems, including suicide because the European Medicines Agency determined that these risks outweighed the benefits.

“Whilst the findings reported here are observational, and cannot infer causality, they send a clear message that there is a lot still to be learnt about the newer classes of obesity mediations (GLP-1 agonists).

“All medications are associated with side effects, with varying frequency and severity.  People thinking about taking these classes of drug for weight loss, should do so under supervision.”

Dr Giuseppe Fanelli, Chair of the ECNP Network on suicide research and prevention; psychiatrist, and Junior Assistant Professor in the Department of Biomedical and Neuromotor Sciences, University of Bologna, said:

“The recent discussions surrounding GLP-1 receptor agonists (GLP-1Ras) and their potential neuropsychiatric effects have brought to light both promising findings, such as those related to cognitive function improvement and possible positive effects on addictive behaviours, and significant concerns related to a possible increase of suicidality.  The studies by Schoretsanitis et al. (2024) and the accompanying commentary by Salvo et al. (2024) contribute valuable insights into this debate, particularly given the rising use of GLP-1Ras like semaglutide and liraglutide for diabetes management and weight loss.

“Schoretsanitis et al. conducted a disproportionality analysis using the World Health Organization’s (WHO) global database of suspected adverse drug reactions (ADRs).  Their findings suggest a significant signal for semaglutide-associated suicidal ideation, which persisted after taking into account potential confounding effects related to the concomitant use of antidepressants or benzodiazepines.  This association was robust across various sensitivity analyses, which compared semaglutide to other medications like dapagliflozin, metformin, and orlistat.  However, liraglutide, despite earlier concerns, did not show a similar signal in this study, adding complexity to the risk profile of GLP-1Ras.

“The methodology employed by Schoretsanitis et al. is commendable, particularly their use of a large-scale pharmacovigilance database, which allows for the detection of rare but serious ADRs.  However, it is important to interpret these findings with caution.  While disproportionality analyses are useful for identifying potential safety signals, they do not establish causality.  The study’s findings should be regarded as preliminary indicators that necessitate further investigation through more controlled and longitudinal studies.

“The commentary by Salvo et al. provides a broader context, emphasizing the need for caution when prescribing GLP-1Ras, especially to individuals with pre-existing psychiatric conditions.  The authors highlight the limitations of premarketing clinical trials, which often exclude patients with a history of depression or suicidality.  This exclusion likely underestimates the true risk of psychiatric side effects in the general population.  Salvo et al. effectively argue for closer monitoring of patients on these medications, particularly as their off-label use for weight management increases.

“Adding to this discussion, the findings from Possidente et al. (2023) contribute to the existing body of knowledge regarding the neuropsychiatric effects of GLP-1Ras, offering a valuable perspective for continued research in this field.  While most research on this topic has been conducted in animal models, there is some evidence from human studies suggesting potential antidepressant and cognitive-enhancing effects of GLP-1Ras.  For example, a meta-analysis by Pozzi et al. (2019) found that GLP-1Ras may be superior to control treatments in reducing depression scores in patients with diabetes.  However, this meta-analysis also had significant limitations, including a small number of studies, possible severe bias, and high heterogeneity among the data.

“Additionally, Mansur et al. (2017) conducted a small open-label trial that suggested liraglutide might improve depressive symptoms and executive functions in non-diabetic patients with major depressive disorder (MDD) or bipolar disorder (BD) who had below-average cognitive performance.  While these results are promising, they are preliminary and based on a very small sample size, indicating the need for larger, more rigorous studies.

“When integrating these findings with those of Schoretsanitis et al. and Salvo et al., a complex picture emerges.  On one hand, there is evidence suggesting that GLP-1Ras may have beneficial effects on mood and cognitive function.  On the other hand, the potential for serious adverse effects, such as suicidality, cannot be overlooked, especially in vulnerable populations.  This duality underscores the importance of individualized treatment decisions, where the potential benefits of GLP-1Ras must be carefully weighed against the risks, particularly in patients with a history of psychiatric disorders.

“In conclusion, while GLP-1Ras hold significant therapeutic promise, especially in the management of diabetes and obesity, their expanding use raises important safety concerns that must be addressed.  The findings from these recent studies highlight the need for continued vigilance in monitoring the psychiatric side effects of these drugs.  Future research should aim to clarify the conditions under which these medications may be safely used, ensuring that the benefits outweigh the risks for all patient populations.  As we move forward, it is important that clinicians and patients remain informed and cautious, balancing the exciting potential of GLP-1Ras with the emerging evidence of their multifaceted effects on mental health.”

Prof Kevin McConway, Emeritus Professor of Applied Statistics, Open University, said:

“This study uses data from a large WHO database, called VigiBase, of reports of adverse drug reactions (ADRs).  The data come from 140 countries, but in very broad terms the data from each country works in a similar way to the UK’s Yellow Card scheme for reporting ADRs (and data from the UK scheme is included in the WHO database).

“The use of this system is an excellent, and relatively rapid, way to be vigilant about possible adverse reactions that arise after drugs are marketed.  But, really, data from such a system can’t by any means establish everything we would want to know about adverse reactions to a drug.  That needs further investigation.

“This new study is subject to those overall limitations, which is why its authors can’t go beyond asking for ‘urgent clarification’ of what their findings might mean in more detail.

“The researchers’ main finding could be interpreted as saying that a bigger proportion of the reports on the database about the drug semaglutide mentioned suicidal ideation (i.e. suicidal thoughts) than is the case for the overall pattern of reports for all the very many drugs in the database.

“But, appropriately, the researchers list, in the ‘Limitations’ section of their paper, several inevitable limitations of what they found.

“They point out that there’s no way of telling, from the data they used, that this disproportionally high rate of mentions of suicidal ideation for semaglutide is actually caused by the patients having used semaglutide.  That’s for a number of reasons – partly because there will be many differences between patients who took semaglutide and patients who took other drugs, apart from whether they took semaglutide, and the high rate of mentions could be caused by one or more of these other differences rather than by the drug.

“Those other reasons may not all be on the database at all – and there’s no information there about people who did not have any kind of reported adverse reaction.

“So we just can’t know, from this research, whether or not taking semaglutide causes an increase in the chance that a patient will have suicidal thoughts.  It might, or it might not.  We can’t even tell whether it’s the semaglutide that’s causing the higher-than-average proportion of reactions of suicidal ideation for reports on the database, the measure that the researchers actually looked at.  It’s important to know more about what the cause might be, and this study can’t tell us, so that’s one reason that the researchers ask for ‘urgent clarification’.

“(The researchers also looked at the drug liraglutide, and at some other possible reactions that also relate to suicide or self-harm, but did not find any associations that went beyond what you might expect by chance.)

“Whatever might be causing suicidal thoughts in patients who use semaglutide, information in the paper indicates that this is very rare, at least in terms of the suspected adverse reactions that got onto the WHO database.  They found only 94 reports of suicidal ideation in patients who took semaglutide, and indeed only 107 patients who took semaglutide and had any report of a potential reaction that related to suicide or self-harm (in thought or action), out of the large number of people who have used the drug since it was introduced.

“Again, though, this study can’t tell us what the chance is that a patient who takes semaglutide will have an adverse reaction related to suicide or self-harm, or indeed will have any kind of ADR.  That’s because the study does not use data on patients where no report of an adverse reaction was submitted.  If we don’t know how many patients there are in all, we can’t know what the chance of an ADR is.

“Maybe you’re wondering why research based only on reports of (possible) adverse reactions is done at all, given all these limitations.  An analysis like this one is still very useful, because it gives a warning that some adverse effect may be more common in users of the drug in question, and it can be produced quickly and relatively cheaply.

“However, it’s reassuring to some extent that, of the ADRs on the database for patients who have taken semaglutide, only about 8 in every thousand (in 2023) related to suicide or self-harm (in thought or action).  So the great majority of reported ADRs were about some other possible reaction.  And to repeat myself, there’s no way of knowing how many patients had no adverse reaction at all, because they just aren’t on this database of ADRs.”

Prof Ian Douglas, Professor of Pharmacoepdidemiology, London School of Hygiene & Tropical Medicine, said:

“This paper presents, at best, weak evidence of an association between semaglutide and suicidality.  The level of association would not trigger a signal alert at MHRA as far as I am aware, since they would require a larger association.  But this is by the by, since the regulators already considered there to be a signal of this potential problem with GLP-1 agonists and have been investigating it.  Signal detection studies in pharmacovigilance databases are good for generating hypotheses, but are not suitable for assessing whether there is a causal association between a drug and an outcome.  This means that whatever this study had found, it would not improve our understanding of GLP-1 agonists and the risk of suicidal behaviour.”

Prof Stephen Evans, Emeritus Professor of Pharmacoepidemiology, London School of Hygiene & Tropical Medicine, said:

“In my view this is not a significant or noteworthy paper, and it has major limitations.  This paper is based just on spontaneous reports which are sent to regulatory authorities in the country of the person reporting a suspected adverse reaction.  These are sent by health professionals and patients to authorities, but are very subject to bias, including effects of media reporting.  The evidence is extremely weak for a genuine effect in this instance.

“A useful remark in the accompanying editorial is “Khouri et al showed a wide variability of results coming from disproportionality analyses, depending both on the method and model specifications, thus opening the door for a selective reporting of results”.

“Spontaneous report databases are especially prone to biases in relation to suicidal effects.  Reference 5 in the editorial analysing only US data did not find an adverse effect.

“Reference 3 in the editorial, Wang et al, is a much more reliable type of study, and not only does not show an effect but suggested protection: “semaglutide compared with non-GLP1R agonist anti-obesity medications was associated with lower risk for incident (HR = 0.27, 95% CI = 0.200.32–0.600.36) and recurrent (HR = 0.44, 95% CI = 0.32–0.60) suicidal ideation, consistent across sex, age and ethnicity stratification”.

“We can’t conclude from the study by Schoretsanitis et al that semaglutide itself is responsible for suicidality.

“There are other grounds, based on previous evidence and with other drugs, for being cautious in the use of semaglutide, and being aware of patients’ mental health when prescribing it is sensible, even though semagutide itself seems not to increase mental health problems.”

Dr Stephen Burgess, Group Leader at the MRC Biostatistics Unit, University of Cambridge, said:

“I am concerned that these analyses combine and conflate on-target effects of GLP-1R agonists with the broader consequences of weight loss.  There are biological effects of the mechanism of action of these drugs.  And there is the wider impact of losing as much as 20% of one’s bodyweight, as has been seen in trials of these drugs.  This is a life-changing intervention that will have largely positive consequences for many.  But for others, the impact of losing a large amount of weight could have negative emotional consequences.  If an individual thinks that losing weight is the key to happiness, but losing weight does not bring happiness, could this lead to depression and suicidal ideation?  And if it did, would we say that GLP-1R agonists caused suicidal thoughts?  While any mechanism that increases suicide risk demands action, these two potential causal pathways require different management.  We need to understand whether these results represent a specific side-effect of these drugs, or an uncommon but tragic consequence of some individuals’ weight loss journey.”

* ‘Disproportionality Analysis From World Health Organization Data on Semaglutide, Liraglutide, and Suicidality’ by Georgios Schoretsanitis et al. was published in JAMA Network Open at 16:00 UK time on Tuesday 20 August 2024.

DOI: 10.1001/jamanetworkopen.2024.23385

Declared interests

Dr Riccardo De Giorgi: “No COI.”

Dr Nerys Astbury: “No conflicts.”

Prof Kevin McConway: “I am a Trustee of the SMC and a member of its Advisory Committee.  My quote above is in my capacity as an independent professional statistician.”

Prof Ian Douglas: “I have research grants from GSK and AstraZeneca and shares in GSK.”

Prof Stephen Evans: “SE has no conflicts of interest in relation to this issue.”

Dr Stephen Burgess: “No relevant conflict of interest to disclose.”

For all other experts, no reply to our request for DOIs was received.