• FDA
• Safety and quality
• Science & evidence
• Clinical trials

GLP-1s just got a new label warning. It's about your eyes, not your waistline.

By Peptide Policy Watch staff · Published November 10, 2026 · 10 min read · 2 sources, 2 checked at the source

Most of the conversation around GLP-1 drugs still starts with the scale. On November 9, 2026, FDA pointed somewhere else. The agency said every approved GLP-1 receptor agonist must carry a new warning about a rare eye condition, non-arteritic anterior ischemic optic neuropathy, or NAION, that can cause loss of vision in one eye, and that loss can be permanent. [V]

Facts are as of November 10, 2026. Sources, with retrieval dates, are at the bottom.

What actually changed

This is a labeling order, not a recall and not a new approval. FDA is telling every company that holds an application in the class to put the NAION information in the Warnings and Precautions section. [V] That is the part of a drug's official label reserved for serious risks. The products stay on the market.

FDA's wording is careful: "potential increased risk." The agency is not saying NAION is common, and it is not saying every person who takes one of these medicines will have an eye problem. It is saying the evidence it reviewed is enough to warn about a possible rise in a rare event. [V]

The order applies to approved GLP-1 receptor agonists, the medicines whose applications FDA already holds. [V] It is a tool for labeled, approved products. This page does not extend that order to compounded copies; FDA's November 9 notice does not discuss compounding. [U] For how approved, compounded, and research-labeled GLP-1 products differ, see the GLP-1 decoder.

What NAION is, as FDA describes it

NAION is damage at the front of the optic nerve, the cable that carries signals from the eye to the brain, after blood flow there drops. FDA says the deeper "why" is not fully worked out. [V] The agency's list of possible risk factors includes diabetes, sleep apnea, smoking, anemia, kidney disease, and high blood pressure. If it has happened in one eye, FDA says the chance it happens in the other eye goes up. [V]

The symptoms FDA flags are sudden and painless: loss of sight in one eye, sometimes first noticed on waking; blurry or missing patches of the visual field, including blind spots or tunnel vision; colors that look wrong; and a soft or hazy center. [V]

Published background rates in people over 50, FDA notes, fall between 2.3 and 10.2 events a year for every 100,000 adults. [V] That range is the backdrop for the trial numbers below. Rare, in this case, is not a soft word. It is a rate measured in single digits per 100,000 person-years.

The trial numbers, in absolute terms

Because single trials saw few NAION events, or none, FDA combined 126 randomized, controlled trials from GLP-1 development programs. The pool included 134,253 people: 78,217 on a GLP-1 medicine and 56,036 on a comparator, either placebo or a glucose-lowering drug that is not a GLP-1. Exposure totaled 288,879 person-years, 156,947 of them on a GLP-1 medicine and 131,932 on a comparator. [V]

FDA identified 15 events it judged probable NAION. Ten were in people taking a GLP-1 medicine, five were in people on placebo, and none were in people on another non-GLP-1 glucose-lowering drug. That produced an incidence of 6.4 events per 100,000 person-years on a GLP-1 medicine and 3.8 on a comparator. [V]

• 126randomized trials in FDA's pooled analysis [V]
• 134,253people in that analysis [V]
• 6.4 vs 3.8probable NAION events per 100,000 person-years, GLP-1 vs comparator [V]

Those two rates sit inside, or next to, the general-population range FDA cited for adults over 50. [V] The difference is real in the pooled trials: about three extra events per 100,000 person-years, on top of a very small base. Relative change can look larger than that because the starting number is tiny. That is what "rare, and possibly more frequent" looks like in these data. The reading of absolute versus relative risk here is a way of stating FDA's figures, not a separate finding. [U]

What FDA's real-world study added

Trials are not the whole record. FDA also ran a retrospective cohort study in its Sentinel Real-World Evidence Data Enterprise Commercial Network. The study looked at adults with type 2 diabetes who started a GLP-1 medicine, an SGLT-2 inhibitor, or a DPP-4 inhibitor between 2019 and 2024. [V]

In that dataset, NAION incidence sat in a band of 26–32 events for every 100,000 person-years: 30 to 32 on a GLP-1 medicine, 31 on an SGLT-2 inhibitor, and 26 on a DPP-4 inhibitor. [V] Those absolute rates are higher than the trial pool. The groups, the way cases were found, and the length of follow-up are not the same, so the two sets of rates should not be treated as one number. [U]

The comparisons are the part FDA leaned on. Against SGLT-2 inhibitors, the hazard ratio came in at 1.19 (interval 0.91–1.54), which was not statistically significant. Against DPP-4 inhibitors it came in at 1.43 (interval 1.02–2.01). [V]

FDA also reviewed published observational studies. It said the more rigorous ones consistently found a higher NAION risk in people with type 2 diabetes who started a GLP-1 medicine, compared with those who started a DPP-4 or SGLT-2 inhibitor, and that those papers generally lined up with the trial analysis and the Sentinel study. [V]

If you want a fuller account of what a Sentinel-style records study can and cannot show, and why FDA still treats randomized trials as the stronger footing for proving a drug works, the real-world evidence explainer walks through the agency's own framework.

How FDA got here

The closer look started with a July 2024 paper by Hathaway and colleagues in JAMA Ophthalmology, on NAION risk in people prescribed semaglutide. FDA said that study had limitations, but that it was enough to justify a broader assessment. [V] The agency then reviewed trial data, animal data, and postmarketing material, including observational studies and case reports, plus NAION reports in its adverse-event system (now called AEMS, formerly FAERS). [V]

Those spontaneous reports were hard to interpret, FDA said, because the usual NAION risk factors overlap with conditions that are common in people who take GLP-1 medicines. [V] That is why the agency put more weight on the pooled trials and the Sentinel cohort.

On mechanism, the animal data were a blank. FDA wrote that its review of the available animal studies did not identify a way GLP-1 use could raise the chance of NAION. [V] The warning is going on the label anyway, on the strength of the human data, with that gap left open.

FDA also said NAION cases have been reported with every GLP-1 receptor agonist that is currently marketed. [V]

Which medicines are covered

FDA published a table of approved GLP-1 receptor agonists. It includes, among others, exenatide (Byetta, whose brand application was later withdrawn and is now sold only as a generic, FDA says, not for safety or effectiveness reasons), liraglutide (Victoza, Saxenda, and the liraglutide-insulin combination Xultophy), dulaglutide (Trulicity), the lixisenatide-insulin combination Soliqua, semaglutide (Ozempic, Rybelsus, Wegovy), tirzepatide (Mounjaro and Zepbound), and orforglipron (Foundayo). [V] FDA notes that tirzepatide also acts at the GIP receptor, and that Foundayo is the 2026 oral entry in the class. Foundayo is a small molecule, not a peptide; the approval explainer covers that distinction.

Several of these medicines have approved uses beyond type 2 diabetes and long-term weight reduction, including, for certain products, lowering the chance of major cardiovascular events, slowing kidney decline, treating obstructive sleep apnea in adults with obesity, and treating MASH — metabolic dysfunction-associated steatohepatitis — once fibrosis has reached a moderate or advanced stage. [V] Wegovy's MASH decision is covered in the semaglutide liver explainer. The new warning is class-wide. It is not limited to the weight-loss brands.

What FDA told patients and clinicians

FDA's advice to patients and caregivers is specific, and it is the agency's, not a recommendation from this site. The notice tells patients and caregivers that these medicines may carry a higher chance of NAION, and it lists the symptoms above.

If someone has already had a NAION event, FDA says the next step is a conversation with their clinician about staying on the GLP-1 medicine or stopping it. [V]

To clinicians, FDA says to know about the potential increased risk, to tell patients to report sudden painless vision changes in one eye right away, and, if NAION is suspected, to refer the person promptly to an ophthalmologist. FDA also told clinicians to consider stopping the GLP-1 medicine if NAION develops, except when they judge the benefits of staying on treatment to be clearly greater than that risk. [V]

What this does not settle

A few limits are baked into the notice itself.

• The event is rare. Even in the pooled trials, FDA is talking about 15 probable cases across more than 134,000 people and nearly 289,000 person-years. [V]
• The real-world comparison is mixed. One Sentinel contrast reached statistical significance; the other did not. [V]
• No mechanism showed up in animal studies. FDA said so plainly. [V]
• Risk factors overlap. People who take these medicines often have diabetes, high blood pressure, or sleep apnea, which are also NAION risk factors, and FDA flagged that overlap as a reason spontaneous reports were hard to read. [V]
• This is a label change for approved products. It is not a verdict on compounding, and it is not a new use. [V]

The useful next step is the updated label itself, once companies add the new language, and any later FDA follow-up if more data arrive. Until then, the November 9 Drug Safety Communication is the primary source for what changed and why.

Sources (retrieved November 10, 2026 unless noted)

1. FDA, "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," Drug Safety Communication, November 9, 2026 [V] — https://www.fda.gov/drugs/drug-safety-communications/fda-requires-safety-labeling-changes-glucagon-peptide-1-receptor-agonists-glp-1-ras-due-potential
2. Hathaway JT, Shah MP, Hathaway DB, et al., "Risk of Nonarteritic Anterior Ischemic Optic Neuropathy in Patients Prescribed Semaglutide," JAMA Ophthalmology, August 1, 2024; 142(8):732-739 [V] (cited by FDA as the paper that prompted the review) — https://doi.org/10.1001/jamaophthalmol.2024.2296

Spot a mistake? See how we source and our corrections policy.

Image: AI-generated. Image credits