Most people who follow metabolic health research know that incretin-based therapies have attracted enormous scientific interest. These are peptide-based compounds that mimic or amplify the body's own gut-derived hormones to influence blood sugar and appetite signaling. In recent years, reports began surfacing that one specific incretin peptide, semaglutide, might be associated with a rare condition called non-arteritic anterior ischemic optic neuropathy, or NAION for short. That condition involves a reduction in blood flow to part of the optic nerve, the cable that carries visual information from the eye to the brain, and it can result in sudden, painless vision loss.
The concern was serious enough that European regulators launched a formal review. But a key question remained unanswered: is this a problem unique to semaglutide, or does it extend to other drugs in the same broad class? A comparative analysis published in the journal Diabetes, Obesity and Metabolism set out to probe exactly that, using one of the largest repositories of real-world adverse event data available.
The database behind the study
Researchers turned to the FDA Adverse Event Reporting System, commonly called FAERS. This is a publicly accessible database where healthcare providers, patients, and manufacturers can submit reports of suspected adverse drug reactions after a product reaches the market. It is not a clinical trial. Reports are voluntary, incomplete, and subject to biases, but the database is large enough that statistical patterns can emerge when a particular drug-outcome combination is reported far more often than would be expected by chance.
The team used a tool called OpenVigil 2.1 to extract reports for six incretin-based therapies, covering all submissions from each drug's approval date through the third quarter of 2025. They then applied a standard pharmacovigilance technique called disproportionality analysis. The idea is straightforward: if a specific adverse event is reported alongside a specific drug at a rate much higher than it is reported for all other drugs in the database combined, that excess is called a signal. Two measures were used, the reporting odds ratio and the proportional reporting ratio, both reported with 95 percent confidence intervals.
What the numbers showed
The results were striking for semaglutide. Researchers found 355 reports linking the peptide to optic ischemic neuropathy, producing a reporting odds ratio of 94.45. In plain terms, the optic neuropathy outcome was reported alongside semaglutide roughly 94 times more often than the baseline rate across the rest of the database. That is an exceptionally large disproportionality signal in pharmacovigilance terms.
Tirzepatide, a dual-acting incretin peptide that targets both GLP-1 and GIP receptors, also produced a significant signal, though considerably smaller, with 49 reports and a reporting odds ratio of about 2.94. Liraglutide, another GLP-1 receptor agonist peptide, likewise showed a signal based on 13 reports and an odds ratio of roughly 4.58.
By contrast, three other incretin peptides, dulaglutide, exenatide, and lixisenatide, produced no detectable signal at all. The authors note that this variation within the same drug class is itself informative, suggesting whatever mechanism might be at work is not simply a consequence of GLP-1 receptor activation in general.
Patient characteristics in semaglutide reports
Among the semaglutide-linked reports, the average age of the individuals was 59 years. Disability was documented in 18 percent of cases, and hospitalization was recorded in 11 percent. These figures hint that the outcomes, when they did occur, were not trivial, though it is important to stress that adverse event reports describe suspected associations, not confirmed drug-caused harm. Many patients in these reports may have had pre-existing risk factors for optic nerve problems, including hypertension, diabetes itself, sleep apnea, or structural features of the optic disc.
Why the class is not uniform
The finding that some incretin peptides carry a signal while others do not is one of the study's most scientifically interesting observations. All six drugs in the analysis work through overlapping hormonal pathways, yet they differ in molecular structure, receptor selectivity, half-life, and dosing patterns. Semaglutide, for instance, is formulated for once-weekly dosing and achieves relatively high and sustained plasma concentrations compared with some shorter-acting agents.
The researchers argue that the absence of a signal with dulaglutide, exenatide, and lixisenatide argues against what pharmacologists call a uniform class effect. In other words, the data do not support the idea that any drug touching the GLP-1 receptor automatically raises this risk. Whether the differences reflect pharmacokinetics, peak exposure levels, off-target receptor activity, or something else entirely is not resolved by this analysis. That question will require prospective studies with standardized diagnostic criteria.
Limits of pharmacovigilance data
Pharmacovigilance signals are hypothesis generators, not proof of causation. Several well-known limitations apply here. Reporting is voluntary, so only a fraction of actual adverse events ever reach the database. More widely prescribed drugs naturally accumulate more reports simply because more patients are using them, a phenomenon called notoriety bias. Once a safety concern gains media coverage, reporting rates often spike further, inflating apparent signals.
Semaglutide is by far the most widely used of the six peptides examined, and it has already received extensive media coverage related to optic neuropathy specifically. That context could contribute to its outsized numbers. The authors acknowledge these constraints directly and call for prospective studies in which diagnoses are confirmed by neuro-ophthalmologists using standardized criteria. Only that kind of rigorous follow-up can establish whether a true causal relationship exists and, if so, what the absolute risk actually is.
What researchers say should come next
The study concludes that significant post-marketing safety signals for optic ischemic neuropathy exist for three incretin peptides, with semaglutide showing by far the strongest association. The authors call for formal prospective investigations rather than relying on voluntary reporting data. They also note that European regulators have already moved toward formal review, which suggests the scientific community is treating the signal seriously even while recognizing its preliminary nature.
For those tracking peptide research, the broader takeaway is methodological as much as anything else. Pharmacovigilance databases, when analyzed carefully, can surface patterns that randomized trials are not designed or powered to detect. The field of incretin biology is now large enough that these tools are revealing meaningful variation within what was once assumed to be a uniform drug class. Early data points at molecular-level differences mattering considerably when it comes to safety profiles, a reminder that peptide structure and pharmacokinetics carry real-world consequences worth studying carefully.
