GLP-1 drugs: eye and mood safety signals

A pharmacovigilance study published online on 22 August 2026 in Naunyn-Schmiedeberg's Archives of Pharmacology reports that eye disorders and suicidal ideation were reported disproportionately often in adverse-event databases for people using semaglutide and liraglutide, compared with dulaglutide^[1]. That is a real signal worth understanding. It is not proof that these drugs cause either problem. The distinction matters, and the study's own authors say so directly.

The short version: this is a signal-detection study built from spontaneous adverse-event reports, the kind of data that is good at raising questions and bad at answering them. Two separate regulators, the European Medicines Agency (EMA) and the UK's MHRA, have already reviewed both questions using stronger evidence. On eye disorders, they found a real but very small risk and updated the semaglutide label accordingly. On suicidal thoughts, the EMA went further and found no such link at all. Both conclusions predate this new paper.

What kind of study this actually is

The authors, led by Ilaria Ammendolia at the University of Messina, pulled spontaneous reports of suspected adverse reactions to liraglutide, semaglutide and dulaglutide from EudraVigilance, the EU's adverse-event database, covering 2020 through 2025^[1]. They ran a disproportionality analysis using the reporting odds ratio (ROR), a statistical technique that asks whether a given side effect shows up more often in reports about one drug than in reports about a comparison drug.

That method has a real limitation worth spelling out. Spontaneous reports are filed whenever a clinician or patient suspects a drug caused a problem, not collected systematically from every user. There is no denominator of total users, so a report count cannot become a rate the way a trial or cohort study can. Reporting also spikes after media coverage, and semaglutide has drawn far more attention than liraglutide or dulaglutide, which alone could inflate its count regardless of any biological effect. Dulaglutide, the comparison drug here, is not a placebo or an untreated group. It is simply the least-discussed drug in the same class.

The eye-disorder finding, and what regulators already confirmed

Eye disorders were reported more often with semaglutide than with liraglutide (ROR 2.67, 95% CI 1.54 to 4.63) or dulaglutide (ROR 1.89, 95% CI 1.30 to 2.75)^[1]. Unlike the psychiatric finding below, this direction of signal lines up with an independent regulatory review that used a different kind of evidence entirely.

In June 2025, the EMA's safety committee concluded a formal review of non-arteritic anterior ischemic optic neuropathy (NAION), a condition that causes sudden vision loss in one eye, and semaglutide. Drawing on non-clinical studies, clinical trials, post-marketing data and published epidemiological studies rather than spontaneous reports alone, the PRAC found NAION is a very rare side effect of semaglutide, meaning it may affect up to 1 in 10,000 people^[2]. The epidemiological studies behind that conclusion suggested roughly a two-fold increase in NAION risk in adults with type 2 diabetes compared with people not taking the medicine, working out to about one additional case per 10,000 person-years of treatment. The EMA recommended updating the product information for Ozempic, Rybelsus and Wegovy to list NAION as a very rare side effect, and advised that treatment stop if NAION is confirmed^[2].

The UK's MHRA followed with its own Drug Safety Update on 5 February 2026, warning of the same association and telling patients on semaglutide who notice sudden vision loss or rapidly worsening eyesight in one or both eyes to attend eye casualty or A&E without delay^[3]. Two things are true here at once: the eye-disorder signal is real enough that regulators acted on it, and the absolute risk is small, on the order of one extra case per 10,000 patient-years. The new pharmacovigilance paper's finding is consistent with that established picture rather than a new discovery.

The suicidal-ideation finding, and what regulators actually concluded

Suicidal ideation was reported disproportionately often with semaglutide (ROR 6.49, 95% CI 1.57 to 26.81) and with liraglutide (ROR 8.61, 95% CI 1.87 to 39.45), compared with dulaglutide^[1]. Those confidence intervals are wide, which reflects how few suicidal-ideation reports exist in the database at all. A wide interval like 1.87 to 39.45 means the true effect size is poorly pinned down, not that the upper number is a reliable estimate.

This exact question, whether GLP-1 receptor agonists raise the risk of suicidal or self-injurious thoughts, was already the subject of a formal EMA safety review that concluded in April 2024. After examining post-marketing data, clinical-trial data and two large electronic health record studies, the EMA's Pharmacovigilance Risk Assessment Committee stated plainly that the available evidence does not support a causal association between GLP-1 receptor agonists, including dulaglutide, exenatide, liraglutide, lixisenatide and semaglutide, and suicidal or self-injurious thoughts and actions^[2]. That review used broader, more systematic evidence than a spontaneous-report disproportionality analysis alone can provide.

The new paper's own authors do not overstate their finding. They frame it as a reason not to underestimate the possibility of a link and to keep watching, not as proof of one, and they explicitly note that drug agencies have found no such relationship. That is the right way to read a disproportionality signal sitting next to a formal regulatory review reaching a different, better-supported conclusion.

What we still don't know

Neither the new paper nor the regulatory reviews settle everything. The EMA's 2024 conclusion is a statement that current evidence does not support causation, not a guarantee that no risk exists in any subgroup, and the agency said it would keep monitoring. Whether people with a personal history of depression, anxiety or prior suicidal ideation face a different risk profile on semaglutide is not resolved by any source here. The mechanism connecting GLP-1 receptor activity to eye or mood-related adverse events, if one exists beyond chance and reporting bias, is not established. And because EudraVigilance reporting shifts with media attention, the true magnitude of both signals could look different once more years of stable, less novelty-driven data are available.

The authors of the new paper recommend caution when prescribing GLP-1 receptor agonists to patients with existing eye conditions or psychiatric vulnerability, and further research into both signals^[1]. That is a reasonable clinical posture regardless of how the causation question eventually resolves. If you are taking semaglutide and notice sudden vision changes, or if you experience new or worsening thoughts of self-harm, contact your prescriber or seek urgent care. This is not a reason to stop a prescribed medicine on your own.

Medical disclaimer

This article is for educational and journalistic purposes only and does not constitute medical advice. Semaglutide and other GLP-1 receptor agonist medicines discussed here may be classified as prescription-only in your jurisdiction. Always consult a qualified healthcare professional before starting, stopping or adjusting any medicine. PeptideMethods.com does not sell, distribute or facilitate the sale of any medicine or peptide product.