Explore this article's sources with AI
Follow PeptideMethods on Google
FDA finds rare eye signal for GLP-1 drugs
FDA pooled 126 trials and found 15 probable NAION events, prompting a class-wide warning while causation remains uncertain.
Why we wrote this. The FDA warning needs its small event count, mixed comparator results, and causal limits stated together.
In this article (5 sections)
The FDA has required class-wide label warnings about a possible link between GLP-1 receptor agonists and non-arteritic anterior ischemic optic neuropathy, or NAION. Its pooled analysis covered 126 clinical trials and 134,253 patients. FDA reviewers identified 10 probable events among GLP-1 recipients and five among comparator recipients. That is a rare safety signal, not proof that these medicines cause the condition[1].
What the FDA pooled
The agency combined trials that compared a GLP-1 receptor agonist with placebo or another glucose-lowering medicine. The dataset included 78,217 GLP-1 recipients and 56,036 comparator recipients, with 288,879 patient-years of exposure. The 10 versus five event count came from this pooled trial evidence[1].
FDA reported exposure-adjusted rates of 6.4 probable events per 100,000 patient-years in the GLP-1 group and 3.8 per 100,000 patient-years among comparators. The communication did not provide a pooled hazard ratio, confidence interval, or inferential test for that rate comparison. The counts therefore support concern and further monitoring, but they do not quantify a settled causal effect[1].
Absolute numbers keep the signal in proportion. Across nearly 289,000 patient-years, FDA reviewers found 15 probable events. Even a real relative increase could still correspond to a small absolute risk, while the severity of sudden vision loss makes careful labeling reasonable.
The warning applies across marketed GLP-1 receptor agonists, including medicines discussed on our semaglutide evidence page and tirzepatide evidence page. A class label action does not mean every product has the same amount of supporting evidence or the same estimated risk.
The observational result depended on the comparator
FDA also ran a retrospective Sentinel cohort study in adults with type 2 diabetes who started a GLP-1 receptor agonist, an SGLT2 inhibitor, or a DPP-4 inhibitor between 2019 and 2024. Compared with SGLT2 inhibitors, the hazard ratio was 1.19 and the 95% confidence interval ran from 0.91 to 1.54. That interval included no difference[1].
Compared with DPP-4 inhibitors, the hazard ratio was 1.43, with a 95% confidence interval from 1.02 to 2.01. The contrast reached the conventional statistical threshold in that comparison. Different comparator results matter because patients receiving these drug classes may differ in health, prescribing history, access, or other factors that claims data cannot fully measure[1].
A cohort can detect associations in routine care, but it cannot reproduce random assignment. Statistical adjustment can balance recorded features. It cannot remove every unmeasured difference. The mixed Sentinel results are one reason the FDA used the phrase potential increased risk rather than declaring that causation had been established[1].
Why the earlier semaglutide study raised concern
A 2024 study used a neuro-ophthalmology registry from one academic institution. In a propensity-matched type 2 diabetes cohort, 17 NAION events occurred among 194 patients prescribed semaglutide and six among 516 patients prescribed non-GLP-1 medicines. The adjusted hazard ratio was 4.28, with a wide 95% confidence interval from 1.62 to 11.29[2].
In a separate overweight or obesity cohort, 20 events occurred among 361 patients prescribed semaglutide and three among 618 comparators. The adjusted hazard ratio was 7.64, with a 95% confidence interval from 2.21 to 26.36. The authors called this an association and said further work was needed to assess causality[2].
Those estimates should not be treated as general population incidence. The study came from a specialist neuro-ophthalmology setting, used prescription records rather than confirmed medicine exposure, and examined semaglutide rather than the whole GLP-1 class. Its unusual population and small event counts also help explain the broad confidence intervals[2].
What changed and what did not
FDA now requires application holders to add the potential NAION risk to the Warnings and Precautions section of prescribing information. The agency also says patients should seek immediate medical attention for sudden vision loss or rapidly worsening eyesight. That is safety advice attached to a rare but serious condition, not a reason to stop a prescribed medicine without clinical guidance[1].
The action did not establish a biological mechanism. FDA reported that its animal review did not identify one. Nor did the trial pool contain enough events to define how risk varies by medicine, dose, treatment duration, diabetes status, or baseline eye disease[1].
What we do not yet know
We do not yet know whether the association reflects a direct medicine effect, metabolic changes after treatment, patient selection, detection patterns, or a mixture of factors. A reliable answer will need more confirmed cases, clear exposure definitions, suitable active comparators, and analyses that separate individual medicines from the class.
The trial pool is large, but the event count is tiny. That combination can reveal a signal while leaving its size uncertain. Future updates should report product-specific exposure, absolute rates, adjudication methods, and confidence intervals. They should also preserve the difference between a regulatory precaution and proof of cause.
**Medical disclaimer:** This article is for educational and journalistic purposes only and does not constitute medical advice. Peptides discussed may be classified as prescription medicines or research chemicals depending on your jurisdiction. Always consult a qualified healthcare professional before using any peptide product. PeptideMethods.com does not sell, distribute, or facilitate the sale of any peptide product.
Frequently asked
Did the FDA prove that GLP-1 drugs cause NAION?
No. The FDA identified a potential increased risk from pooled trials, observational studies, and reports. It did not establish causation or a biological mechanism.
How common was NAION in the pooled trials?
FDA found 15 probable events among 134,253 patients: 10 in GLP-1 recipients and five among comparators. The reported rates were 6.4 and 3.8 per 100,000 patient-years.
Should patients stop a GLP-1 medicine because of the warning?
Patients should not change prescribed treatment based on this article. FDA advises immediate medical attention for sudden vision loss or rapidly worsening eyesight and discussion with a healthcare professional.
Sources
- [1]FDA safety labeling change for GLP-1 receptor agonists and NAIONTier 1 · primary↩
- [2]Semaglutide and NAION retrospective matched cohort (PMID 38958939)Tier 1 · primary↩
No revisions yet. First published .