American Academy of Ophthalmology says tirzepatide's NAION risk is under-studied

The American Academy of Ophthalmology and the North American Neuro-Ophthalmology Society (NANOS) released a joint clinical statement in May 2026 on whether GLP-1 drugs raise the risk of non-arteritic anterior ischemic optic neuropathy (NAION), a rare and usually permanent form of sudden vision loss ^[1]. The groups concluded the evidence points to a possible association with semaglutide of roughly two-fold, that cause and effect has not been established, and that even a several-fold relative risk remains a small absolute risk for a rare condition ^[1]. They also said other GLP-1 drugs have been studied far less often and deserve more investigation, singling out tirzepatide because it is more potent than semaglutide ^[1].

What NAION is

NAION affects an estimated 2.3 to 10.2 people per 100,000 each year ^[1]. It is thought to involve reduced blood flow to the optic nerve, and most patients have a small, crowded optic disc — a "disc at risk" — that may make the nerve vulnerable ^[1]. People typically notice sudden, painless vision loss in one eye. About one-third have mild spontaneous improvement, but vision loss is often permanent, and up to one-quarter eventually develop it in the other eye ^[1]. There is no known effective treatment ^[1]. Diabetes, cardiovascular disease, high blood pressure, metabolic syndrome and obstructive sleep apnea are all established risk factors ^[1] — conditions common among people prescribed GLP-1 drugs.

The evidence is mixed

The concern began with a 2024 study of nearly 17,000 patients at a single neuro-ophthalmology referral clinic. Among 710 patients with type 2 diabetes, semaglutide use was tied to a hazard ratio of 4.28 (95% CI, 1.62–11.29) over 36 months, with cumulative NAION incidence of 8.9% versus 1.8% in controls; among 979 patients with obesity or overweight, the hazard ratio was 7.64 (95% CI, 2.21–26.36), with incidence of 6.7% versus 0.8% ^[1]. The statement flags major limits: possible confounding by diabetes severity, and selection bias at a center that treats most of the region's NAION cases, which likely overestimated risk ^[1].

Larger databases show smaller and inconsistent effects. An analysis of 37.1 million adults with type 2 diabetes in the OHDSI network found semaglutide carried a hazard ratio of 2.27 (95% CI, 1.16–4.46) versus empagliflozin under a specific case definition, but no increase versus sitagliptin or glipizide, and none at all under a broader definition ^[1]. A self-controlled analysis found incidence rate ratios of 1.32 and 1.50 depending on definition ^[1]. About half the included databases showed no increased risk ^[1].

TriNetX studies also split. One covering more than 170,000 patients found hazard ratios of 2.39, 2.44 and 2.05 at two, three and four years ^[1]. Another, of about 80,000 patients taking semaglutide or tirzepatide, found NAION in 0.04% versus 0.02% of controls over two years (HR 1.76; 95% CI, 1.01–3.07) ^[1]. But a study of roughly 130,000 patients with type 2 diabetes and 58,000 with overweight or obesity found no increased risk at any time point, with cumulative incidence in the semaglutide diabetes group of 0.039% at one year and 0.065% at five years ^[1]. A separate analysis of more than 185,000 GLP-1 users found a non-significant HR of 1.26 (95% CI, 0.94–1.70) ^[1].

Meta-analyses of randomized trials are similarly unsettled: one of 78 semaglutide trials found an odds ratio of 3.92 (95% CI, 1.02–15.02) based on just five trials reporting events, while one of 69 GLP-1 trials found no association (OR 1.53; 95% CI, 0.53–4.44) ^[1]. A third, of eight semaglutide trials, found no significant increase for diabetes (RR 1.76) or weight loss (RR 2.18) ^[1].

Why it matters for patients

The practical takeaway from the statement is that shared decision-making between a patient and their care team is warranted when deciding whether to start, continue or stop a GLP-1 drug ^[1]. Even at the higher end of reported estimates, NAION remains uncommon, and the groups say the overall magnitude of risk is low ^[1].

For tirzepatide (Mounjaro, Zepbound) specifically, the honest answer is that the data are thin. Only a handful of the studies reviewed included tirzepatide at all, and none isolated it ^[1]. Whether its greater potency translates into different eye risk is not yet known.

What happens next

The statement notes that prospective trials in this population are difficult because NAION is so rare — one analysis estimated a signal of 570 per 100,000 patient-years would be needed for adequate statistical power, versus the roughly 2.6 per 100,000 patient-years observed ^[1]. The groups called for further investigation, particularly of non-semaglutide GLP-1 drugs ^[1]. No timeline for new data was given.