The GLP-1 Eye-Safety Signal: NAION Cases Cluster Around Semaglutide, Not Tirzepatide

Non-arteritic anterior ischaemic optic neuropathy (NAION) is a rare eye condition in which the blood supply to the front of the optic nerve is disrupted, causing sudden painless vision loss in one eye. There is no treatment that reliably restores lost vision after NAION happens. Roughly 2 to 10 people per 100,000 develop NAION each year in the general population.

Starting in 2024, case reports and observational studies began to raise a question: are GLP-1 receptor agonist drugs, particularly semaglutide, associated with an increased risk of NAION in the people who take them? The signal was first flagged in a July 2024 JAMA Ophthalmology study out of Massachusetts Eye and Ear that documented elevated NAION rates in patients prescribed semaglutide. Regulatory agencies have been evaluating the accumulating data ever since.

On February 5, 2026, the UK's Medicines and Healthcare products Regulatory Agency (MHRA) issued a Drug Safety Update on semaglutide and NAION. On July 25, 2026, Australia's Therapeutics Goods Administration (TGA) went further: the agency issued a class-wide label update covering every GLP-1 receptor agonist sold in Australia (dulaglutide, semaglutide, tirzepatide, and liraglutide), with an important nuance in the accompanying analysis. The TGA Advisory Committee on Medicines concluded that the current evidence supports the NAION signal for semaglutide, but does not currently support it for dulaglutide or tirzepatide.

That drug-specific split is the piece of the story most worth understanding. This piece walks through what NAION is, what the TGA data actually shows (36 reported cases: 23 semaglutide, 10 tirzepatide, 3 liraglutide), why regulators concluded the signal is currently semaglutide-specific rather than class-wide, what the absolute risk numbers look like in real-world context, and what patients currently on GLP-1 therapy should ask their prescriber.

The optic nerve carries visual signals from the retina to the brain. The front portion of the optic nerve (the optic nerve head, visible during a routine dilated eye exam as the optic disc) receives its blood supply from a small network of vessels branching from the ophthalmic artery.

Non-arteritic anterior ischaemic optic neuropathy (NAION) occurs when the blood supply to the optic nerve head is acutely reduced or interrupted, damaging the nerve fibers that carry visual information out of the eye. The word 'non-arteritic' distinguishes NAION from a similar condition (arteritic AION) that involves inflamed arteries and behaves differently clinically.

The presentation is characteristic and abrupt. A person wakes up in the morning, or notices during the day, that vision in one eye has gone partially dark, blurry, or lost in a particular field. There is no pain. The affected eye typically shows a specific pattern on visual field testing, most often loss of the lower half of the field of vision.

Risk factors for NAION include age over 50, a small optic disc (the anatomical setup called 'disc at risk'), hypertension, diabetes, sleep apnea, and certain medications. The absolute risk in the general adult population is low: roughly 2 to 10 people per 100,000 per year, with higher rates in people over 50 and much higher rates in people with the anatomical risk factor.

Treatment options are limited. Once NAION has occurred, no intervention has been shown to reliably improve visual acuity outcomes. Some patients recover partial vision on their own over months. Most do not recover the lost vision. About 15-20% of people who have NAION in one eye will develop NAION in the other eye within five years.

That prognostic reality is why regulatory agencies pay close attention to any drug that might elevate NAION risk, even at low absolute rates. The condition is permanent, functionally significant, and untreatable after the fact.

The first widely-publicized concern about GLP-1 receptor agonists and NAION came from a July 2024 JAMA Ophthalmology study conducted at Massachusetts Eye and Ear (Harvard Medical School). The retrospective observational study analyzed 16,827 patients seen at the eye institute over six years. It documented elevated rates of NAION among patients prescribed semaglutide compared with matched controls: hazard ratios in the range of 4-7 depending on the analysis subgroup and the specific comparator.

The study drew scrutiny for its observational design and selection issues (the population was Mass Eye and Ear patients, not a general population), but it was consistent with a signal the FDA Adverse Event Reporting System (FAERS) had been accumulating: post-marketing reports of NAION in people taking semaglutide.

Subsequent research over the next 18 months tried to clarify the picture using larger, more representative data sources:

• A Danish national-registry study covered several hundred thousand patients on semaglutide and found a modest increase in NAION incidence.
• A US insurance-claims analysis using Optum data replicated the elevated hazard ratios for semaglutide.
• A Nordic cohort study covering both semaglutide and tirzepatide detected the signal for semaglutide but not, or much less clearly, for tirzepatide.

By early 2026, the accumulated evidence was strong enough for MHRA to issue a Drug Safety Update on February 5, 2026 specifically covering semaglutide. Then Australia's TGA took the further step of updating the entire GLP-1 receptor agonist class on July 25, 2026, on precautionary grounds, while the Advisory Committee on Medicines separately concluded that the specific signal supports semaglutide but not dulaglutide or tirzepatide.

The FDA in the US has been accumulating FAERS reports and has not yet issued a specific label update or Drug Safety Communication on NAION and GLP-1 receptor agonists. That may change as European and Australian regulatory positions harden. FDA class labeling changes typically follow international regulator moves by 6 to 18 months when the underlying evidence is being assembled from the same global observational data.

The TGA action on July 25, 2026 has two parts.

Part 1: Class-wide label update. All five GLP-1 receptor agonists currently marketed in Australia get updated Product Information (equivalent to the US FDA-approved label) that adds NAION under Adverse Reactions and includes the standard patient guidance about seeking urgent medical attention for sudden vision loss. The five products are:

• Trulicity (dulaglutide, Eli Lilly)
• Ozempic (semaglutide, Novo Nordisk)
• Wegovy (semaglutide, Novo Nordisk)
• Mounjaro (tirzepatide, Eli Lilly)
• Saxenda (liraglutide, Novo Nordisk)

Part 2: The Advisory Committee on Medicines analysis. The committee reviewed the underlying evidence base (published case-control studies, cohort studies, and pharmacovigilance data) and reached the more specific conclusion that the current evidence supports the NAION signal for semaglutide, but does not currently support it for dulaglutide or tirzepatide.

The underlying case data from the Australian Database of Adverse Event Notifications (DAEN) as of the review:

• Total cases of optic ischaemic neuropathy reported with GLP-1 receptor agonists: 36
• Semaglutide (Ozempic + Wegovy): 23 cases
• Tirzepatide (Mounjaro): 10 cases
• Liraglutide (Saxenda): 3 cases

The raw case counts alone do not resolve the drug-specific question because they reflect exposure volume as much as any real risk difference. Semaglutide has been on the Australian market longer and been prescribed to substantially more patients than tirzepatide (which launched later). What separates the semaglutide signal from the tirzepatide non-signal in the committee's analysis is the underlying observational and case-control evidence, which controls for exposure volume: hazard ratios for semaglutide are elevated in multiple independent datasets; hazard ratios for tirzepatide are not, or are much smaller with wider confidence intervals in the datasets analyzed to date.

That distinction is important for patients thinking about a switch or a comparison. Regulators updated the entire class label on precautionary grounds because the class shares a receptor mechanism, but the actual evidence signal at this point in time is not uniform across drugs in the class.

GLP-1 receptor agonists share a mechanism: they bind and activate the glucagon-like peptide-1 (GLP-1) receptor. But the individual drugs are structurally distinct molecules with different pharmacokinetic profiles, different receptor-binding kinetics, different secondary receptor activities, and different exposure duration.

The differences that could matter for NAION risk:

Half-life and steady-state exposure. Semaglutide has a half-life of about a week. Tirzepatide has a half-life of about five days. Dulaglutide has a half-life of about five days. Liraglutide has a half-life of about 13 hours (requires daily injection). Longer steady-state exposure to a drug that affects blood-vessel dynamics or glycemic swings might correlate with different vascular risk profiles.

Receptor specificity. Tirzepatide is a dual GIP/GLP-1 receptor agonist (activates both receptors). Semaglutide is a selective GLP-1 receptor agonist. GIP receptor activation has downstream effects on lipid handling and blood-vessel function that partially differentiate the classes.

Rate and magnitude of glycemic change. One hypothesis about the NAION signal is that it may be mediated by rapid changes in blood glucose during treatment initiation and dose escalation. Sudden reductions in chronic hyperglycemia can trigger transient changes in optic-nerve microvascular perfusion. Drugs with different titration schedules and different rates of glycemic reduction could produce different NAION risk patterns.

Weight loss velocity. Rapid weight loss has been documented as a risk factor for other ischemic events. Tirzepatide typically produces faster weight loss than semaglutide over the same time window, which if anything argues against the drug-specific split observed in the data.

The honest answer is that regulatory scientists do not yet know why semaglutide shows a clearer NAION signal in observational data than tirzepatide does. The next 12-24 months of prospective and retrospective analyses will probably clarify whether the pattern reflects a real drug-specific risk difference or an artifact of exposure duration, prescribing patterns, and study design.

What is clear right now: the observational evidence at hand supports the semaglutide signal. It does not currently support the same magnitude of signal for tirzepatide. Regulators updated the class label because it is the cautious move to make, not because the evidence supports the same risk for every drug in the class.

The elevated hazard ratios reported in observational studies (typically in the 2-7 range for semaglutide) sound dramatic, but the absolute risk is small.

The general-population NAION incidence is roughly 2-10 per 100,000 per year in adults. In older adults (over 50) with cardiovascular risk factors, the baseline rate is higher, in the range of 10-20 per 100,000 per year. If the true hazard ratio for semaglutide in a comparable population is 3-5, the absolute rate on semaglutide would be in the range of 30-100 per 100,000 patient-years.

That translates to something like 3-10 additional NAION cases per 10,000 patient-years of semaglutide exposure above the background rate. The excess is small in absolute terms and much smaller than the risk of the outcomes that GLP-1 receptor agonists are typically prescribed to reduce (cardiovascular death, myocardial infarction, stroke, all-cause mortality).

For context, the SELECT cardiovascular outcomes trial of semaglutide in patients with prior cardiovascular disease and obesity documented a 20% relative reduction in major adverse cardiovascular events (about 90 fewer events per 10,000 patient-years compared to placebo). The magnitude of the potential NAION excess is roughly 1/10 to 1/30 of the magnitude of the cardiovascular benefit for that indication.

That kind of risk-benefit calculation belongs with a prescribing clinician who can factor in the individual patient's baseline cardiovascular risk, the anatomical NAION risk factor (small optic disc, often documented on prior eye exam), diabetes status, and the specific indication being treated. But the class of magnitudes involved makes clear that NAION risk, while real and worth informing patients about, is not on its own a reason for a patient with a strong cardiovascular indication to discontinue therapy without discussion.

The TGA patient guidance and the standard NAION medical literature converge on a straightforward set of actions.

Seek urgent medical attention for sudden vision loss in one eye. The classic NAION presentation is painless loss of vision in one eye, often noticed on waking or during the day. Any acute vision change (partial loss, dark area, distorted vision, blurring that does not resolve within minutes) warrants urgent evaluation. Urgent means going to an emergency department or contacting an ophthalmologist that day, not scheduling an appointment for next week.

Do not stop GLP-1 therapy on your own. Sudden discontinuation of a GLP-1 receptor agonist can produce substantial glycemic swings in patients with diabetes and can lead to rapid weight regain in patients on the drug for obesity. Any decision to change therapy should be made with the prescribing clinician who knows the patient's full clinical picture. If you are worried about NAION risk, contact your prescriber to discuss rather than stop.

Ask about baseline eye exam status. Patients who have had an ophthalmology exam that documented a small optic disc (the anatomical 'disc at risk' setup) are at elevated NAION risk regardless of GLP-1 use. If you don't know your optic disc anatomy and you're starting a GLP-1 receptor agonist, a routine dilated eye exam can document the baseline. This is not required by any guideline but can be reasonable if you are in a higher-risk population.

Report symptoms to prescribers and pharmacies. In the US, the FDA MedWatch program accepts patient-reported adverse events. Reporting matters because pharmacovigilance systems depend on real-world event reporting to detect signals and update labels.

If you are currently on a GLP-1 receptor agonist or considering starting one, here are questions worth bringing to your next appointment.

What is my baseline NAION risk? Age, hypertension, sleep apnea, diabetes, and small optic disc are the main risk factors. Some risk factors are known from your medical history; others (like optic disc anatomy) require an eye exam. Knowing your baseline risk helps size the incremental risk from GLP-1 therapy relative to it.

Am I on semaglutide, tirzepatide, or another GLP-1? The regulatory evidence signal is currently specific to semaglutide. If you are on tirzepatide, the current evidence does not support the same NAION risk. If you are on semaglutide, the signal is real, though the absolute magnitude is small.

What is my cardiovascular indication for this drug? If you are on the drug for cardiovascular disease prevention or secondary prevention, the cardiovascular benefit is likely to substantially outweigh the NAION risk. If you are on the drug primarily for weight management without significant cardiovascular disease, the risk-benefit calculation is different.

Would tirzepatide be a reasonable alternative for me? For patients on semaglutide primarily for weight management who are concerned about NAION risk, switching to tirzepatide (Mounjaro or Zepbound) is a discussion worth having. Tirzepatide typically produces greater weight loss and does not currently show the same NAION signal. The switch has downsides (insurance coverage, injection frequency, side-effect profile) that a prescriber can walk through.

When should I get an eye exam? If it's been more than two years since your last dilated eye exam, this is a reasonable time to schedule one, particularly if you are over 50, have diabetes, or have not had a formal optic disc evaluation.

What do I do if I notice sudden vision change? The answer is emergency evaluation, not scheduled follow-up. Establish the plan before you need it.

Whether the signal will hold in prospective randomized data. Everything to date is observational (case reports, case-control studies, cohort studies from insurance claims and national registries). Randomized cardiovascular outcomes trials of semaglutide and tirzepatide were designed to detect cardiovascular endpoints, not NAION. Post-hoc analyses of NAION rates in the completed trials may or may not confirm the observational signal. Purpose-built prospective studies for NAION risk would take years to complete.

Whether the effect is dose-related. Semaglutide is prescribed at 0.25-2.4 mg per week depending on indication. The observational studies to date have not consistently disentangled dose effects from exposure duration. A dose-response relationship would strengthen the case for a real drug effect; a lack of one would argue for exposure-duration or other confounding as the driver.

Whether the mechanism is direct drug effect or mediated by glycemic change. If NAION risk is driven by rapid changes in blood glucose during treatment initiation, then slower dose titration might mitigate the risk. If the effect is a direct receptor-mediated action on the optic-nerve microvasculature, titration wouldn't help. Basic pharmacology work over the next 24 months should clarify this.

Whether the class label update from TGA and MHRA will trigger changes in US FDA labeling. The FDA has not yet updated GLP-1 receptor agonist labels for NAION. Historical patterns suggest that FDA updates typically follow European and Australian moves by 6-18 months when the underlying evidence is being assembled from the same global data. The FDA may act sooner given the volume of US GLP-1 prescriptions.

Whether tirzepatide will eventually show the same signal at longer exposure duration. Tirzepatide's post-marketing exposure is currently shorter than semaglutide's. The absence of a current signal for tirzepatide may reflect a true drug-specific difference, or it may reflect insufficient patient-years of tirzepatide exposure to detect a signal of the same magnitude. This will become clearer over the next 24-48 months.

The July 25, 2026 TGA class-wide GLP-1 label update is important. It reflects a real safety signal in observational data that regulators have been evaluating for two years. Patients on GLP-1 receptor agonists should know that NAION is a documented, if rare, adverse event associated with these drugs, and that sudden vision loss requires urgent medical attention.

The more specific piece of the story worth understanding is that the accumulated evidence signal currently points at semaglutide, not at tirzepatide or dulaglutide. Regulators updated the class label on precautionary grounds because the drugs share a receptor mechanism. But the underlying observational data supporting the signal is currently strongest for semaglutide alone.

For patients on semaglutide with a strong cardiovascular indication, the risk-benefit balance likely still favors continuing therapy because the magnitude of cardiovascular benefit is substantially larger than the magnitude of the NAION excess. For patients on semaglutide primarily for weight management who are concerned about the signal, tirzepatide represents a reasonable alternative that a prescriber can evaluate.

For patients on tirzepatide, the current evidence does not support the same NAION concern. Regulators may update the tirzepatide-specific evidence base as more data accumulates, but as of July 2026, the tirzepatide safety profile on NAION appears different from the semaglutide profile.

The US FDA has not yet issued a Drug Safety Communication or a specific label update on NAION and GLP-1 receptor agonists. That is likely to change over the next 12-18 months as the FDA reviews the same underlying observational evidence that MHRA and TGA have already acted on. Patients who want to stay ahead of the regulatory cycle can raise the topic proactively with their prescriber now rather than wait for a formal FDA action to prompt the conversation.