On Monday 9 November, the US Food and Drug Administration did something it has avoided for two years. It required every company holding an approval for a GLP-1 receptor agonist to add a warning about a rare eye condition to the Warnings and Precautions section of their label. Twelve marketed products, from Byetta-era exenatide through to the oral Foundayo, now carry a line about non-arteritic anterior ischemic optic neuropathy.
If you have been anywhere near peptide social media this week you have already seen the takes. One camp says the drugs are blinding people and the whole class is finished. The other camp says it is a nothingburger from a regulator with nothing better to do. Both are wrong, and the truth sits in a fairly narrow band in the middle.
What NAION actually is
Non-arteritic anterior ischemic optic neuropathy is a stroke of the optic nerve head. Blood flow to the front of the optic nerve drops, the nerve swells, and vision in that eye degrades. It is painless, it usually happens on waking, and there is no proven treatment that reverses it. Roughly speaking, it is the second most common cause of optic nerve blindness in older adults after glaucoma.
It is also rare. The usual figure quoted in ophthalmology is somewhere between two and ten cases per 100,000 people per year in the general population, and it clusters in people who already have the risk factors that GLP-1 users often have: type 2 diabetes, high blood pressure, sleep apnoea, smoking history, anaemia, kidney disease. That overlap is precisely why this has been so hard to untangle.
The FDA’s own summary lists the symptoms plainly: sudden painless vision loss in one eye, blind spots or tunnel vision, distorted colour perception, blurry or hazy central vision. If NAION happens in one eye, the risk of it happening in the other goes up. That is the part worth paying attention to.
The numbers behind the label change
This is where the coverage has been lazy, so here is the actual basis for the decision.
The FDA ran a meta-analysis across 126 clinical trials in GLP-1 development programs, covering 78,217 patients who received a GLP-1 receptor agonist. Within that dataset the agency identified 15 probable NAION events: 10 in patients on a GLP-1, five on placebo, and none in patients on other non-GLP-1 glucose-lowering medication.
That works out to an incidence of 6.4 events per 100,000 patient-years on a GLP-1, against 3.8 per 100,000 patient-years for comparators. Roughly a 1.7-fold relative increase on a very small absolute number.
Then the FDA did a second, separate piece of work, a retrospective study in its real-world evidence network looking at adults with type 2 diabetes. Compared with SGLT-2 inhibitors like empagliflozin, GLP-1 users showed no higher NAION risk. Compared with DPP-4 inhibitors like sitagliptin, they did.
Read those two analyses together and you get something more interesting than a headline. The signal is consistent enough across trial data and spontaneous reports for a regulator to act, but it is not strong enough to show up against every comparator in messy real-world data. That is what a rare adverse event looks like when you finally have enough exposure to see it.
The regulatory trigger, for what it is worth, was a 2024 study in JAMA Ophthalmology that found an association between semaglutide and NAION risk in a Harvard-affiliated cohort. That paper caused a brief media cycle, then went quiet. This week’s action is the FDA closing the loop on it.
What the regulator is and is not saying
The FDA did not restrict any indication. It did not pull anything off the shelf. It did not add a boxed warning. It required a Warnings and Precautions entry, and it gave clinicians two instructions: tell patients to report sudden vision changes immediately, and consider discontinuing the GLP-1 in anyone who develops NAION unless the benefits clearly outweigh the risk.
BMO analysts made the commercial read on the Monday, and it is bluntly sensible: the action formalises a signal that was already known, event rates are low, and a Warnings and Precautions update is not a restriction on approved indications. Nobody’s prescription volume changes because of this.
Where the clinical conversation actually needs to go is monitoring. Sudden painless vision loss is not something people reliably report, because it is painless and because it often happens in one eye while the other compensates. If you are on a GLP-1 and your vision changes in one eye, that is a same-day ophthalmology referral, not a wait-and-see.
What this means if you are not on a labelled product
Here is the uncomfortable part for this community.
Everything above applies to approved medicines, made by manufacturers with pharmacovigilance departments, dosed at known concentrations, taken under some form of clinical supervision, and covered by a reporting system that feeds events back into exactly the kind of analysis the FDA just published. None of that is true for a vial bought online.
There is no NAION reporting pathway for unapproved retatrutide or grey-market tirzepatide. There is no batch traceability, no dose certainty, and no one collecting outcomes. If the risk is genuinely around 6 per 100,000 patient-years at labelled doses, an unregulated population taking unknown doses, often stacked with other compounds, is not a population where anyone is counting. Absence of reports is not absence of events. It is absence of a system.
The TGA’s own testing of seized vials has repeatedly found products that did not contain what the label claimed, which makes any extrapolation from trial safety data to a grey-market product a guess dressed up as an inference. We have made this point before about liver toxicity and about the oesophagus injury case, and it holds here too.
What the Research Says
- The FDA issued a Drug Safety Communication on 9 November 2026 requiring all holders of approved GLP-1 receptor agonist applications to add information about a potential increased risk of NAION to the Warnings and Precautions section of labelling. Twelve marketed products are covered, including semaglutide (Ozempic, Rybelsus, Wegovy), tirzepatide (Mounjaro, Zepbound), liraglutide, dulaglutide, exenatide, lixisenatide and orforglipron. Source: US Food and Drug Administration, “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,” 9 November 2026.
- The agency’s meta-analysis covered 126 clinical trials and 78,217 GLP-1-treated patients, identifying 15 probable NAION events (10 on GLP-1, five on placebo, none on other glucose-lowering drugs), an incidence of 6.4 versus 3.8 events per 100,000 patient-years. Source: BioSpace, “FDA orders class-wide GLP-1 label update to include potential risk for rare eye disease,” 9 November 2026, reporting the FDA release.
- The regulatory action followed a 2024 cohort study in JAMA Ophthalmology that reported an association between semaglutide and NAION. Source: Hathaway JT et al., “Risk of Nonarteritic Anterior Ischemic Optic Neuropathy in Patients Prescribed Semaglutide,” JAMA Ophthalmology, 2024.
- FDA patient guidance lists sudden painless vision loss in one eye, visual field defects, colour distortion and reduced central sharpness as symptoms requiring immediate contact with a healthcare professional, and notes that NAION in one eye raises the risk in the other. Source: FDA MedWatch alert, content current as of 9 November 2026.
The short version
A rare event got slightly less rare in a very large dataset, a regulator responded proportionally, and 12 labels changed. That is the whole story. If you take a GLP-1, know the symptom and act on it same-day. If you take something that is not a labelled product, understand that the safety net described in that FDA release does not include you, and that the community’s job is to be honest about that rather than to pick whichever headline is more comfortable.
We are tracking how this develops, including whether the TGA mirrors the warning in Australian product information, in the Grey Highway Telegram group. Background on the compounds involved is on the semaglutide and tirzepatide pages.
Disclaimer: This article is for educational and informational purposes only. It does not constitute medical advice, therapeutic recommendations, or endorsements of any compound. Grey Highway is a research-education community. We do not sell, supply, or promote the use of research compounds. Always consult a qualified healthcare professional regarding health decisions. For Australian regulatory information, visit the TGA website.