Longevity Lab Report

News

FDA Orders a NAION Vision-Loss Warning on Every GLP-1 Drug Label, Citing 15 Probable Cases in 126 Trials

What the agency required on Monday, the numbers underneath it, the comparison that came back null, and what the warning does not say.

Saved articles

Ask ChatGPT about this article ↗

Copy the prompt, then paste it into ChatGPT to explore the findings and sources.

Open ChatGPT ↗

What happened

The FDA is requiring every GLP-1 receptor agonist label to carry a warning about a potential increased risk of NAION, a rare eye condition that can cause sudden, permanent vision loss in one eye. Its own pooling of 126 trials found 15 probable cases in 288,879 person-years, at 6.4 per 100,000 person-years on the drugs against 3.8 on comparators.

The Food and Drug Administration published a drug safety communication on Monday, November 9, requiring the makers of every GLP-1 receptor agonist to add information about a potential increased risk of non-arteritic anterior ischemic optic neuropathy, or NAION, to the Warnings and Precautions section of their labeling. NAION is a rare condition that can produce acute, painless vision loss in one eye, and the loss can be permanent, as the agency's MedWatch alert of the same day puts it. The requirement is class-wide: it covers semaglutide, tirzepatide, and every other marketed drug in the class, because NAION cases have been reported for all of them. It is a labeling requirement. It is not a recall, not a contraindication, and not an instruction to stop treatment.

The word the agency uses throughout is potential, and this piece keeps it in every sentence where the risk appears, because the FDA did not conclude that the drugs cause NAION. Its own conclusion, in its words, is that "there is a potential increased risk of rare NAION events in patients taking GLP-1 RAs."

What the FDA found in the trials

The agency pooled the clinical trial programs behind the class: 126 trials, 134,253 patients, and 288,879 person-years of follow-up. Across all of that exposure it found 15 probable NAION events: 10 in patients taking a GLP-1 drug and 5 on placebo, with none on other glucose-lowering medicines. The incidence works out to 6.4 probable events per 100,000 person-years on the drugs against 3.8 per 100,000 on the comparators.

Probable NAION events per 100,000 person-years, FDA trial pooling
GLP-1 drugs6.4 per 100,000 person-yearsComparators3.8 per 100,000 person-years

FDA drug safety communication, November 9, 2026: 15 probable events in 288,879 person-years across 126 pooled trials.

Both halves of that sentence matter. The relative difference is real in the agency's arithmetic, and the absolute numbers are very small: 15 probable events in nearly 290,000 person-years, in a condition whose risk factors, diabetes and vascular disease among them, overlap heavily with the conditions these drugs treat. The agency said as much about the spontaneous reports it reviewed, calling them challenging to interpret for exactly that reason. A reader who takes away "these drugs damage eyes" has read more into the data than the FDA did; a reader who takes away "nothing there" has read less.

The second dataset, and the comparison that came back null

The trial pooling is not the agency's only evidence. Its Sentinel analysis, a retrospective cohort of adults with type 2 diabetes starting therapy between 2019 and 2024, measured NAION incidence at 30 to 32 per 100,000 person-years on GLP-1 drugs, 31 on SGLT-2 inhibitors, and 26 on DPP-4 inhibitors. Against SGLT-2 inhibitors, the hazard ratio was 1.19 with a confidence interval of 0.91 to 1.54, which is not a statistically significant increase. Against DPP-4 inhibitors, the hazard ratio was 1.43 with a confidence interval of 1.02 to 2.01, which is an increase. The two comparisons point in different strengths, and any account of this warning that quotes only one of them is giving half the result.

The agency also noted that the published observational literature points the same way its own analyses do: methodologically rigorous studies have consistently reported increased NAION risk in patients with type 2 diabetes starting GLP-1 drugs compared with the other two classes. The signal that started the inquiry was a 2024 single-centre study in JAMA Ophthalmology linking semaglutide to NAION; the FDA states it found several limitations in that study and ran its own evaluation rather than adopting the finding.

One more finding belongs in any honest account, because the agency printed it plainly: its review of the animal data identified no mechanism by which GLP-1 drugs could increase NAION risk. The human data show an association the agency judges worth labeling; the animal data offer no pathway that explains it. Both facts are part of the record.

The European context

Europe's regulator moved first, on a narrower scope. In June 2025 the European Medicines Agency's safety committee concluded that NAION is a very rare side effect of semaglutide specifically, citing an observational association of roughly twofold in patients with type 2 diabetes. Monday's action is broader (the whole class, not one drug) and rests on the FDA's own trial pooling and Sentinel cohort rather than on the European read. The two agencies agree on the direction and differ on the scope.

What this means for the library's grades

Our library grades semaglutide and tirzepatide A- on evidence and D on worth it, at $1,000 or more a month without insurance in the United States. The worth-it D is structural: both are prescription medicines, not supplement purchases, and the grade says so. Those grades were set in October, before this warning existed. The grades do not change: what follows is a safety amendment, not a regrade. On November 17, 2026, both records absorbed the labeling action as a safety-note amendment, carrying the FDA's warning with its absolute rates in their safety notes, in the open and recorded in each record's changelog, rather than edited in quietly under cover of a news piece.

What the warning changes for a reader today is narrower than the headline cycle will make it sound. The labels will carry the risk. A patient who develops sudden vision symptoms in one eye should contact their clinician right away, and a patient who has already had a NAION event should discuss whether to continue. That is the agency's own guidance, and it is the whole of the practical instruction in Monday's communication.

Frequently asked questions

Did the FDA ban these drugs or tell patients to stop? No. The action is a labeling requirement covering every GLP-1 receptor agonist. It is not a recall, not a contraindication, and not an instruction to stop treatment.

How large is the potential risk in absolute terms? In the FDA's pooling of 126 trials, 15 probable NAION events occurred in 288,879 person-years: an incidence of 6.4 per 100,000 person-years on the drugs against 3.8 on comparators. Rare by any reading, and serious when it happens, because the vision loss can be permanent.

Is this only about Ozempic? No. The requirement is class-wide, and the agency states that NAION cases have been reported for all currently marketed GLP-1 products.

Does anyone know how the drugs could cause this? On the current record, no. The FDA's review of the animal data identified no mechanism by which GLP-1 drugs could increase NAION risk, and the agency's conclusion is a potential increased risk, not a demonstrated cause.

How we grade Full records: Semaglutide and Tirzepatide

Limitations: what we do not know

The 15 probable events are probable, adjudicated from trial records not designed to detect an event this rare, and the Sentinel cohort's absolute rates are roughly five times the trial pooling's, a gap between datasets that neither the agency nor this piece resolves. The comparison against SGLT-2 inhibitors is null and the comparison against DPP-4 inhibitors is not, and no dataset here explains why the two comparators differ. The overlap between NAION's risk factors and the diseases these drugs treat is the standing confound of the whole literature. What would change the picture: prospective eye monitoring inside a large trial, or a mechanism, neither of which exists today.

Sources

  • FDA Drug Safety Communication, "FDA Requires Safety Labeling Changes for Glucagon-Like Peptide-1 Receptor Agonists (GLP-1 RAs) Due to Potential Increased Risk of Rare But Serious Eye Condition," November 9, 2026, read in full at fda.gov; FDA MedWatch alert of the same date. Verification record: research/fda-glp1-naion-verification-2026-11-11.md.
  • European Medicines Agency, June 2025 conclusion on semaglutide and NAION, as characterized in the verification record.
  • Hathaway et al., JAMA Ophthalmology 2024;142(8):732-739, the single-centre cohort the FDA names as its trigger, with the agency's stated limitations.
  • Longevity Lab Report records: semaglutide and tirzepatide (grades and costs as graded October 17, 2026).

Changelog

  • 2026-11-12: Published as the news desk's Nov 12 piece, assigned by the LLR Newsroom on its first slate. Written against the FDA communication itself under Research's six print conditions (research/fda-glp1-naion-verification-2026-11-11.md): "potential" in every risk sentence, absolute rates beside relative figures, the Sentinel split printed whole, the animal review's no-mechanism finding printed with the item, class scope throughout, and the EMA context sourced to the EMA. The library records predate the warning and carry no NAION content; the piece says so rather than implying otherwise.

Medications mentioned

Grades are research summaries, not medical advice. Decisions about your own health belong with your clinician.

Keep exploring

View all

The weekly newsletter

Keep up with what’s changing.

New research, useful tools, and the people shaping health and longevity. Selected and explained in one email.

See what you’ll receive ↗

Free. Confirm your email to subscribe. Unsubscribe anytime.