Li H-Y, et al. GLP-1 receptor agonists and risk of optic nerve or vision-threatening events in patients with type 2 diabetes or cardiometabolic diseases: A meta-analysis of randomized controlled trials. Diabetes Care 2026 Mar 1; 49:526. DOI: 10.2337/dc25-1929.

Heberer K, et al. New-onset nonarteritic anterior ischemic optic neuropathy and initiators of semaglutide in US veterans with type 2 diabetes. JAMA Ophthalmol 2026 Feb 12; [e-pub]. DOI: 10.1001/jamaophthalmol.2025.6262.

Noh Y, et al. Glucagon-like peptide 1 receptor agonists and risk of nonarteritic anterior ischemic optic neuropathy in patients with type 2 diabetes. Diabetes Care 2026 Feb 17; [e-pub]. DOI: 10.2337/dc25-2577.

Nonarteritic anterior ischemic optic neuropathy is the kind of diagnosis that makes every clinician's stomach drop: sudden, often permanent vision loss, and not much we can do about it. It has always been rare, but a growing body of work is now pointing to a possible link with one of the most widely discussed drug classes in medicine: GLP-1 receptor agonists.

Three new studies add fuel to that conversation. First, a large meta-analysis pooled 20 randomized trials with about 80,000 participants-mostly people with type 2 diabetes followed for roughly three years. In that dataset, GLP-1 agonists did not increase a composite of serious ocular events and did not show a signal for ischemic optic neuropathy specifically. On the surface, that sounds reassuring.

But the observational data tell a more worrying story. In a U.S. veterans cohort of around 100,000 patients with type 2 diabetes already on metformin, investigators compared add-on semaglutide to add-on empagliflozin over a median of two years. The rate of NAION was higher with semaglutide-about 123 versus 67 events per 100,000 person-years. A separate analysis using a U.K. primary care database of roughly 500,000 people with type 2 diabetes found a similar pattern: those starting a GLP-1 agonist had a higher 1-year risk of NAION than those starting a DPP-4 inhibitor (18.5 vs. 7.2 events per 100,000 person-years).

These new results line up with prior observational work suggesting roughly a doubling of NAION incidence among GLP-1 users. So why the disconnect with the meta-analysis of randomized trials? It's almost certainly about design rather than biology. None of the trials were built to capture rare, unexpected eye events: vision outcomes weren't prespecified, routine eye exams weren't mandated, and the definitions of ocular safety events were inconsistent. In that setting, a signal as uncommon as NAION can easily be undercounted or missed entirely.

What should clinicians do with this? For most patients, the cardiometabolic benefits of GLP-1 agonists will still far outweigh a very small absolute risk of a rare optic neuropathy. But when we start or continue these drugs, especially in patients who already have vascular risk factors for eye disease, it's reasonable to add one more line to the counseling script: there is a rare association with NAION, and any sudden change in vision warrants urgent evaluation. This isn't a reason to abandon GLP-1s-but it is a reminder that even our most promising therapies can carry risks we only discover once they're widely used.