FDA Orders NAION Warning on All GLP-1 Drugs: What the Data Show

Complete guide: GLP-1 drugs: a complete guide to semaglutide and beyond

What the FDA ordered on November 9, 2026

On November 9, 2026, the US Food and Drug Administration (FDA) published a Drug Safety Communication requiring all GLP-1 receptor agonist application holders to add information on non-arteritic anterior ischemic optic neuropathy (NAION) to the Warnings and Precautions section¹. The communication specifies no deadline for the labeling change¹. Its conclusion, verbatim: "FDA concluded that there is a potential increased risk of rare NAION events in patients taking GLP-1 RAs"¹.

The list covers generic exenatide (the original Byetta application was voluntarily withdrawn, not for safety reasons), liraglutide (Victoza, Saxenda), dulaglutide (Trulicity), Xultophy, Soliqua, semaglutide (Ozempic, Rybelsus, Wegovy), tirzepatide (Mounjaro, Zepbound) and orforglipron (Foundayo, approved in 2026)¹. Reuters framed the decision around Eli Lilly and Novo Nordisk, which did not immediately respond to its requests for comment².

The trial meta-analysis: a small effect on few events

The analysis of randomized trials totaled 288,879 person-years (156,947 on GLP-1 drugs and 131,932 on comparators)¹. The 15 probable cases give a rate of 6.4 per 100,000 person-years on GLP-1 drugs versus 3.8 on comparators¹. The absolute difference is about 2.6 cases per 100,000 person-years (our calculation from FDA's figures). With only 15 events, shifting one or two cases would change the rate ratio considerably, so the number is best read as a signal, not a precise measurement.

Sentinel and observational studies: the comparator matters

In the Sentinel retrospective cohort (adults with type 2 diabetes, 2019-2024), incidence was 26 to 32 per 100,000 person-years. The HR versus SGLT-2 inhibitors was 1.19 (95% CI 0.91-1.54), not significant, while versus DPP-4 inhibitors it was 1.43 (1.02-2.01)¹. Media summarized the second figure as a 43% higher risk³. FDA adds that methodologically rigorous observational studies were generally consistent with its findings¹.

The published literature points the same way but with different magnitudes:

• OHDSI (37.1 million adults with type 2 diabetes): IRR 1.32 (1.14-1.54) in the self-controlled case series, but in the cohort design the HR did not differ from glipizide or sitagliptin and was higher only versus empagliflozin under the specific definition (HR 2.27; 1.16-4.46). The authors describe a "modest" increase⁷.
• TriNetX (174,584 patients per arm): no clear risk at 1 year (HR 1.94; 0.93-4.02), but HR 2.39 (1.37-4.18) at 2 years and 2.44 (1.44-4.12) at 3; the authors stress that a retrospective design can only infer association⁸.
• US veterans (102,361): 123 vs 67 per 100,000 person-years for semaglutide versus SGLT-2 inhibitors; HR 2.33 (1.54-3.54), with absolute risk of 0.29% vs 0.13%⁹.
• Other groups analyzed older adults with diabetes¹¹ and a comparison of GLP-1 RAs with SGLT-2 inhibitors¹²; the first report in 2024 came from a single center¹³.

Most of these data concern semaglutide. FDA states that NAION cases have been reported with all currently marketed GLP-1 drugs, which is not the same as measuring the risk of each molecule¹.

What to do in clinic: refer, inform and decide with the patient

FDA asks professionals to be aware of the potential increased risk, to instruct patients to report immediately any acute, painless decrease in vision or visual field deficit in one eye, and to refer to an ophthalmologist if NAION is suspected¹. If it occurs, the agency says to "consider" discontinuing the GLP-1 drug unless the benefits clearly outweigh the risk¹. Patients with prior NAION have a higher recurrence risk, and if one eye is affected the other eye is at increased risk¹.

FDA lists smoking, diabetes, high blood pressure, sleep apnea, anemia and kidney disease as possible risk factors¹. That profile overlaps with many patients treated with GLP-1 drugs, which is why FDA found spontaneous reports in its adverse event system difficult to interpret¹.

Europe moved first: EMA and WHO in 2025

The European Medicines Agency's safety committee (PRAC) concluded on June 6, 2025 that NAION is a "very rare" side effect of semaglutide, possibly affecting up to 1 in 10,000 people, and recommended stopping treatment if it is confirmed⁵. Its review cited roughly a two-fold increase in adults with type 2 diabetes, equal to about one additional case per 10,000 person-years⁵. The World Health Organization issued a medical product alert on June 27, 2025, noting that the vision loss is generally irreversible with no effective treatment available⁶. Reuters also notes that Denmark granted compensation to four patients who lost vision after using Novo Nordisk GLP-1 drugs².

In the US, as of September 2026 the label carried no such warning and about 100 NAION-related lawsuits were reported, with Novo Nordisk calling them "without merit"¹⁶¹⁷. Investing.com reports dozens of plaintiffs and says manufacturers maintain that no direct causal link has been proven³.

The other side: causation unproven, absolute risk low