FDA Orders NAION Eye Risk Warning on All GLP-1 Drugs, From Ozempic to Zepbound

For most of the past decade, glucagon-like peptide-1 receptor agonists have been the fastest growing class of medicines in the United States, moving from a niche treatment for type 2 diabetes into household names prescribed for weight loss, cardiovascular risk reduction, kidney protection and sleep apnea. The FDA's own labeling table now lists twelve approved products in the class, beginning with Byetta (exenatide) in 2005 and running through Victoza, Trulicity, Saxenda, Xultophy, Soliqua, Ozempic, Rybelsus, Wegovy, Mounjaro, Zepbound and, most recently, Eli Lilly's oral pill Foundayo (orforglipron), approved in 2026.

Non-arteritic anterior ischemic optic neuropathy, or NAION, is a rare condition that arises when blood flow to the front of the optic nerve is reduced. Patients typically notice sudden, painless loss of vision in one eye, and a share of them never recover fully. Its underlying cause is incompletely understood, which is precisely why it has been so difficult to determine whether the cases reported among GLP-1 users are caused by the drugs, by the diabetes and metabolic disease that many users already have, or by chance.

The signal is not new. The FDA said its review was prompted by the July 2024 Hathaway et al. study published in JAMA Ophthalmology that linked semaglutide to NAION. European regulators moved earlier. The European Medicines Agency's safety committee said in June 2025 that NAION may affect up to 1 in 10,000 people taking semaglutide for at least a year, the first time a regulator confirmed the side effect, and the United Kingdom followed in February 2026. In Denmark, four patients who experienced vision loss after using Novo Nordisk's GLP-1 drugs were granted compensation.

On November 9, 2026, the FDA converted that accumulating signal into a class-wide labeling requirement, telling manufacturers of every approved GLP-1 receptor agonist to add information about a potential increased risk of NAION to the Warnings and Precautions section of their labels. The action does not withdraw any product or restrict any indication. It requires disclosure, and it attaches clinical advice about referral and about whether to continue treatment.

Key Facts

The FDA said in a Drug Safety Communication dated November 9, 2026 that it is requiring application holders for all GLP-1 receptor agonist medications to add the NAION information to labeling. The agency described NAION as a rare condition that can cause acute, painless, decreased vision in one eye, including permanent vision loss, and said cases have been reported for all currently marketed GLP-1 RA products.

The evidentiary core is a meta-analysis of 126 randomized, controlled clinical trials across multiple GLP-1 RAs, covering 134,253 patients: 78,217 GLP-1 RA users and 56,036 comparator patients. Together those trials represent 288,879 person-years of exposure, split into 156,947 person-years for GLP-1 users and 131,932 for comparators. The analysis identified 15 probable NAION events, 10 in GLP-1 RA patients, five in placebo patients and none among patients taking other non-GLP-1 glucose-lowering drugs. Incidence worked out to 6.4 events per 100,000 person-years on GLP-1 RAs versus 3.8 per 100,000 person-years for comparators.

A separate FDA Sentinel real-world study, a retrospective cohort of adults with type 2 diabetes who started treatment between 2019 and 2024, found 26 to 32 NAION events per 100,000 person-years, with GLP-1 RAs at 30 to 32, SGLT-2 inhibitors at 31 and DPP-4 inhibitors at 26. That study found no statistically significant increase versus SGLT-2 inhibitors (hazard ratio 1.19, 95 percent confidence interval 0.91 to 1.54) but did find an increased risk versus DPP-4 inhibitors (hazard ratio 1.43, 95 percent confidence interval 1.02 to 2.01). The SGLT-2 group includes widely used medicines such as Jardiance and Farxiga.

Reuters reported on November 9, 2026 that the action covers Eli Lilly's Mounjaro and Zepbound and Novo Nordisk's Ozempic and Wegovy, and that both companies disputed a causal link. A Lilly spokesperson said the communication applies to all medicines with GLP-1 RA activity and does not reference findings specific to Lilly medicines, adding that the company is carefully evaluating the update and will respond to the FDA promptly. Novo said it continues to believe the data do not establish a cause-and-effect relationship and that it remains confident in the benefit-risk profile of its GLP-1 medicines.

Analysis

The clinical guidance attached to the labeling change is modest and practical. Clinicians should refer suspected cases promptly to an ophthalmologist, and should consider discontinuing the GLP-1 RA in patients who develop NAION unless the benefits clearly outweigh the risk. That is a long way from a withdrawal, a restriction or a boxed warning, and it reflects an evidence base in which the absolute numbers remain small.

What this really means is that the FDA has decided the signal is strong enough to require disclosure but not strong enough to require limitation. Fifteen probable events across 288,879 person-years of exposure is a thin statistical base. The comparator rate of 3.8 per 100,000 person-years also sits inside the published general-population range of 2.3 to 10.2, which undercuts any claim of a dramatic class effect.

The real-world data are more awkward for the class than the trial data. The Sentinel study's failure to show a statistically significant increase versus SGLT-2 inhibitors, alongside a significant increase versus DPP-4 inhibitors, means the observational evidence points in different directions depending on which comparator is chosen. CNBC reported on November 10, 2026 that observational studies consistently reported increased risk versus both comparators even though the Sentinel analysis did not, a discrepancy that reflects differences in confounding and follow-up. Diabetes itself is a known risk factor for NAION, and separating drug effect from underlying disease is the central methodological problem.

The mechanistic gap matters too. The FDA said animal data did not identify a mechanism, which means there is no clean biological story to anchor the epidemiology. Without one, the case rests on a small number of events reported unevenly across products, with most of the attention so far concentrated on semaglutide. Citi analyst Geoff Meacham said the action may temporarily increase patient and investor concern, but the rarity of NAION and the substantial weight, glycemic, cardiovascular, renal and sleep apnea benefits should preserve a favorable benefit-risk profile and limit changes to prescribing.

Why It Matters

Scale is the reason this is not a niche label update. Semaglutide and tirzepatide are among the most widely prescribed medicines in the world. Even a very rare adverse event becomes a public health question when millions of patients are exposed, and a warning that reaches every product in the class reaches every prescriber in the country at once. Semaglutide and tirzepatide already carry a boxed warning over thyroid C-cell tumors seen in rodents and warnings about diabetic retinopathy.

Korea Biomedical Review reported on November 11, 2026 that Korea's Ministry of Food and Drug Safety issued a safety alert on November 10 covering 38 approved drug products from four pharmaceutical companies over the potential NAION risk. The 38 products include formulations of Novo's Wegovy, Ozempic, Rybelsus, Saxenda and Xultophy FlexTouch, Lilly Korea's Mounjaro and Trulicity, Hanmi Pharmaceutical's eppe (efpeglenatide) and Sanofi-Aventis Korea's Soliqua Pen. The ministry asked companies to submit safety information and plans for potential labeling changes and said it will review them before deciding whether to revise prescribing precautions. Korea had previously received one reported NAION case following semaglutide, in January 2024, and Food and Drug Safety Minister Oh Yu-kyoung told a parliamentary audit on October 13 that NAION had already been added as a very rare adverse reaction to Wegovy and Ozempic prescribing information.

The risk of the warning is behavioral. Kim Dae-jung, a professor of endocrinology and metabolism at Ajou University Hospital, cautioned against patients discontinuing GLP-1 treatment solely because of the warning, noting that most reported cases had been associated with semaglutide rather than other drugs in the class, that only 15 cases were identified in clinical trials and that no mechanism was established in animal studies. The cardiovascular and kidney benefits, he argued, generally outweigh the risk.

Next Up

Label revisions will roll out product by product, and the FDA will face pressure to reconcile its trial-based estimate with the Sentinel findings as more real-world data accumulate. The agency has not proposed restricting prescribing and has not identified a mechanism, so the practical question is whether the warning changes clinical behavior or simply adds a line to a label that already carries significant safety text.

Other regulators are moving in parallel. Korea's review of 38 products from four companies is the first clear sign that the American decision is being exported, and the European and British precedents from 2025 and early 2026 mean the class is now under active ophthalmologic scrutiny in three major markets. The next real test is not the label language but the prescribing data: whether clinicians add an eye question to the routine GLP-1 conversation, and whether patients who develop sudden painless vision loss are referred fast enough to matter.