GLP-1 Medications and Eye Health: NAION, Retinopathy, and Vision Risks

Introduction

For most of their lives in clinical practice, GLP-1 receptor agonists were not considered "eye drugs." That began to change in mid-2024, when neuro-ophthalmologists at Mass Eye and Ear in Boston published a retrospective cohort study suggesting that patients prescribed semaglutide had a four- to seven-fold higher rate of nonarteritic anterior ischemic optic neuropathy (NAION) than matched non-users. NAION is rare, but it causes sudden, painless, and largely irreversible loss of vision in one eye — and the signal moved quickly from a single-center paper to a global regulatory review.

By June 2025 the World Health Organization's Advisory Committee on Safety of Medicinal Products had recommended adding NAION to the semaglutide Risk Management Plan, and the European Medicines Agency had ordered Novo Nordisk to list it as a "very rare" adverse reaction in the Ozempic, Wegovy, and Rybelsus product information. The U.S. label was updated in 2025 with parallel language. None of those actions established that semaglutide causes NAION — the data are still observational and the absolute risk remains tiny — but they did formalize what had been an emerging signal into a recognized safety concern that prescribers and patients now need to discuss.

Evidence: "Semaglutide-prescribed patients with type 2 diabetes had a significantly higher risk of NAION over 36 months than matched non-GLP-1 users (cumulative incidence 8.9% vs 1.8%; HR 4.28, 95% CI 1.62–11.29)." — Hathaway JT, et al. JAMA Ophthalmology. 2024;142(8):732–739. DOI: 10.1001/jamaophthalmol.2024.2296

NAION is only half the eye story. The older — and in many ways more clinically relevant — concern is diabetic retinopathy, which the SUSTAIN-6 cardiovascular outcomes trial flagged a decade ago and which the dedicated FOCUS trial is still working to settle. This article walks through both: where each signal came from, what the strongest 2024–2026 evidence actually shows, and how the modern monitoring picture looks for someone starting a GLP-1 today.

NAION: What the 2024–2026 Evidence Says

NAION is the most common cause of sudden optic-nerve vision loss in adults over 50. It happens when blood flow to the front of the optic nerve is acutely interrupted, leading to permanent partial or complete loss of vision in the affected eye. Background incidence is roughly 2–10 cases per 100,000 people per year in the general population — uncommon enough that even modest absolute increases can produce dramatic relative-risk numbers in observational studies.

The Mass Eye and Ear Study

The 2024 JAMA Ophthalmology paper that started the conversation was a retrospective matched-cohort analysis of patients seen at a single neuro-ophthalmology center between December 2017 and November 2023. After propensity matching:

Population	NAION cumulative incidence (36 mo) — semaglutide	NAION cumulative incidence (36 mo) — non-GLP-1	Hazard ratio
Type 2 diabetes (n=710 matched pairs)	8.9%	1.8%	4.28 (95% CI 1.62–11.29)
Overweight or obesity (n=979 matched pairs)	6.7%	0.8%	7.64 (95% CI 2.21–26.36)

The authors were careful to flag the limitations: a single tertiary referral center, neuro-ophthalmology patients (not a general population), no causal claim. But the magnitude of the effect was unusual enough that the paper drew immediate regulatory attention.

Replication Attempts

Subsequent studies produced a more nuanced picture. A 2025 follow-up cohort of 174,584 patients with type 2 diabetes from a multicenter electronic-health-record network found a smaller but still significant increase, with a hazard ratio of approximately 2.2 for NAION in semaglutide users versus DPP-4 inhibitor users.

A separate population-based observational study using the WHO VigiBase pharmacovigilance database covering 180 countries reported a disproportionate signal for optic-nerve and retinal events with semaglutide compared with other GLP-1 receptor agonists, dulaglutide and liraglutide included. The signal was strongest for NAION specifically and was not seen at the same magnitude with the other drugs in the class.

Evidence: "Semaglutide showed significantly higher reporting of vision impairment compared with other GLP-1 receptor agonists, DPP-4 inhibitors, SGLT2 inhibitors, and metformin." — Bell SJ, et al. American Journal of Ophthalmology. 2025. DOI: 10.1016/j.ajo.2025.01.030

A 2025 systematic review and meta-analysis of the emerging evidence concluded that, while the absolute risk remains small, the consistency of the signal across data sources warranted regulatory action and patient-level disclosure.

Regulatory Response

The regulatory cascade in 2025 moved in three steps:

EMA (April 2025): Pharmacovigilance Risk Assessment Committee classified NAION as a "very rare" adverse reaction (≤1 in 10,000) and ordered semaglutide product information updates across the EU.
WHO ACSoMP (June 2025): Recommended that the global Risk Management Plan for semaglutide formally include NAION as a potential risk.
FDA (2025 label updates): Wegovy (Reference ID 5676363) and Ozempic prescribing information were updated to mention NAION reports under "Postmarketing Experience," without elevating it to a boxed warning.

Importantly, none of these actions characterized NAION as a confirmed causal effect of semaglutide. They formalized the signal as one that prescribers must mention and that pharmacovigilance systems must continue to track.

Diabetic Retinopathy: The Older, Bigger Question

The retinopathy concern is older than the NAION one and applies to a much larger share of patients on GLP-1 therapy.

What SUSTAIN-6 Found

The 2016 SUSTAIN-6 cardiovascular outcomes trial randomized 3,297 patients with type 2 diabetes and high cardiovascular risk to subcutaneous semaglutide or placebo. Semaglutide reduced major adverse cardiovascular events by 26%. But it also raised the rate of a composite retinopathy outcome — vitreous hemorrhage, blindness, or need for intravitreal anti-VEGF or photocoagulation — by 76% (3.0% vs 1.8%; HR 1.76, p=0.02).

That signal was concentrated in patients who entered the trial with preexisting retinopathy and who had the largest, fastest reductions in HbA1c — a pattern consistent with the long-known phenomenon of "early worsening" of retinopathy after rapid glycemic improvement, first described in the DCCT trial of intensive insulin therapy.

Evidence: "Among patients with type 2 diabetes and high cardiovascular risk, rates of complications of diabetic retinopathy were significantly higher with semaglutide than with placebo (HR 1.76, 95% CI 1.11–2.78, p=0.02)." — Marso SP, et al. New England Journal of Medicine. 2016;375:1834–1844. DOI: 10.1056/NEJMoa1607141

What Later Meta-Analyses Show

Pooled analyses of the broader GLP-1 trial program have not consistently reproduced the SUSTAIN-6 signal. A 2023 meta-analysis of randomized trials reported no statistically significant increase in diabetic retinopathy progression across the class as a whole, and a 2025 Ophthalmology Retina comparative-effectiveness study of routine clinical practice — covering semaglutide, dulaglutide, liraglutide, and exenatide — found no increased risk of sight-threatening diabetic retinopathy compared with non-GLP-1 diabetes therapy.

A 2024 Ophthalmology Science tertiary-care study and a 2025 OHDSI network analysis of multinational electronic-health-record data both reached similar conclusions: in real-world cohorts, semaglutide does not appear to worsen diabetic retinopathy outcomes, although the early-worsening pattern can still be observed in patients with rapid HbA1c drops.

The phenomenon best fits a glycemic-control mechanism rather than a drug-specific toxic effect: any therapy that lowers HbA1c quickly in someone with longstanding poor control can trigger transient retinopathy progression, and the long-term trajectory is favorable.

The FOCUS Trial

The dedicated answer to this question is FOCUS (NCT03811561), a randomized double-masked trial of weekly semaglutide 1 mg versus placebo in 1,500 patients with type 2 diabetes and HbA1c 7–10%, with serial standardized ophthalmic assessments over five years. Primary readout is progression of diabetic retinopathy on the Early Treatment Diabetic Retinopathy Study scale, with secondary endpoints capturing anti-VEGF use, laser photocoagulation, and vitrectomy.

FOCUS is expected to report in early 2027 and will be the most rigorous evidence base for resolving whether the SUSTAIN-6 signal reflects a true class effect of semaglutide on retinal outcomes or a transient consequence of glycemic improvement.

How the Two Signals Differ

NAION and diabetic retinopathy are sometimes lumped together in patient-facing coverage as "eye risks of Ozempic," but mechanistically and clinically they are very different:

Feature	NAION	Diabetic retinopathy progression
Anatomic site	Anterior optic nerve head	Retinal microvasculature
Onset	Acute, hours to days	Gradual, weeks to months
Reversibility	Largely irreversible	Often reversible with treatment
Affected population	Both diabetes and obesity indications	Primarily preexisting diabetes
Strength of evidence	Observational, signal across multiple databases	Randomized signal in one trial; not consistently replicated
Plausible mechanism	Hemodynamic shift at the optic disk; rapid metabolic change	Early-worsening from rapid HbA1c reduction
Absolute risk	~1 in 10,000 (EMA classification)	Trial-level: ~3% over 2 years in high-risk diabetics on semaglutide
Regulatory status (2026)	Listed in EMA, WHO, and FDA postmarketing sections for semaglutide	Mentioned in semaglutide labels for diabetes indications

This separation matters because the screening response to each is also different. There is no validated way to predict NAION risk in advance for a given patient. Diabetic retinopathy, by contrast, is screenable — and pre-existing retinopathy is the single strongest modifier of how aggressively a clinician should pace HbA1c reduction.

Tirzepatide and Other GLP-1 Drugs

Most of the eye-safety literature concentrates on semaglutide because it has both the largest patient exposure base and the strongest published signals. Data on other agents are thinner but instructive.

Tirzepatide (Mounjaro, Zepbound): Cardiovascular and metabolic outcome trials so far have not shown an increased retinopathy or NAION signal, although exposure time is shorter. The SURPASS program and the SURMOUNT obesity trials have not flagged optic-nerve adverse events at rates exceeding placebo.
Liraglutide (Victoza, Saxenda): Cardiovascular outcomes data from LEADER did not show a retinopathy or NAION signal at the population level.
Dulaglutide (Trulicity): REWIND trial data and subsequent pharmacovigilance studies have not flagged increased optic-nerve events.
Exenatide, oral semaglutide, retatrutide, and survodutide: Insufficient long-term post-marketing data to draw firm conclusions, but no strong signals to date.

The 2025 American Journal of Ophthalmology pharmacovigilance analysis covering 180 countries was specifically designed to compare across the class, and it found that the optic-nerve signal was disproportionately concentrated in semaglutide reports rather than evenly distributed.

This pattern — combined with the absence of strong signals for tirzepatide, liraglutide, or dulaglutide — has shaped how regulators are responding: the labels are being updated for semaglutide specifically, not class-wide.

Practical Guidance for Patients and Prescribers

Both safety signals lead to a similar set of practical steps, even though their mechanisms and probabilities are very different.

Before starting a GLP-1

Baseline ophthalmologic exam for diabetic patients with poor glycemic control. Especially anyone with HbA1c above 9–10% who is about to start a drug expected to lower it rapidly. The American Academy of Ophthalmology and several diabetes societies now recommend this for patients beginning intensive therapy.
Risk stratification for NAION. No validated predictive model exists, but a "disk at risk" — a small, crowded optic nerve head with little or no physiologic cup — is the strongest known anatomic predictor of NAION generally. Patients with a prior episode in the contralateral eye, sleep apnea, or significant cardiovascular risk factors warrant explicit discussion before starting.
Patient education on warning signs. The single most useful intervention for NAION is rapid recognition: a sudden, painless decrease in vision in one eye, often noticed on waking, is the classic presentation and warrants emergency ophthalmology evaluation.

While on therapy

Pace glycemic improvement when retinopathy is present. If pre-existing diabetic retinopathy is documented, slower initial titration and closer ophthalmologic follow-up — typically every 3 to 6 months for the first 12 to 18 months — is advised by retinal specialists.
Symptom-directed evaluation rather than routine retinal screening. Beyond the standard annual diabetic eye exam, no specialty society recommends extra imaging solely because a patient is on a GLP-1.
Stop the drug for confirmed NAION. WHO and EMA guidance both recommend discontinuation if NAION is confirmed in a patient on semaglutide, even though no causal link has been established. The reasoning is precautionary: NAION in the contralateral eye is a known longitudinal risk after a first event, and removing any contributing exposure is reasonable while the science is still maturing.

What is not recommended

Routine pre-prescription neuro-ophthalmology referral. Not warranted for the general patient.
Blanket avoidance of semaglutide for high cardiovascular risk patients. The cardiovascular and metabolic benefits demonstrated in SUSTAIN-6, SELECT, and STEP-HFpEF substantially outweigh the absolute eye risk for most candidates.
Switching from semaglutide to tirzepatide solely on the basis of the NAION signal. Long-term tirzepatide eye data are not yet mature enough to call it definitively safer. Switching may be reasonable for a patient with strong baseline NAION risk factors, but it should be a clinical conversation, not a default policy.

For more on how rapid weight loss interacts with other organ systems, see our companion review on GLP-1 medications and kidney health and the broader GLP-1 side effects overview.

Key Takeaways

The 2024 Mass Eye and Ear study suggested a 4- to 7-fold higher rate of NAION in semaglutide users versus matched controls; replication studies have shown a smaller but still elevated signal.
The European Medicines Agency classifies NAION as a very rare (≤1 in 10,000) adverse reaction of semaglutide as of April 2025; the WHO ACSoMP and FDA have followed with formal Risk Management Plan and labeling updates.
The diabetic retinopathy signal traces back to the SUSTAIN-6 trial (HR 1.76 for the composite outcome) and is best explained by rapid HbA1c reduction in patients with preexisting retinopathy.
The ongoing FOCUS trial (NCT03811561), reporting in early 2027, will be the definitive randomized evidence base on long-term retinopathy outcomes with semaglutide.
Tirzepatide, liraglutide, and dulaglutide have not shown comparable optic-nerve or retinal signals to date, although follow-up time is shorter.
Practical response: baseline retinal exam for patients with poorly controlled diabetes, paced HbA1c reduction in those with pre-existing retinopathy, and prompt evaluation for any sudden monocular vision change.
The cardiovascular and weight-related benefits of semaglutide remain large enough that the rare eye risks should be discussed and monitored, not used as a default reason to avoid therapy.

The eye-safety profile of GLP-1 receptor agonists has matured significantly since 2024. The NAION signal is real and now formally on the label, but it remains a low-probability event. The retinopathy signal, while older, is increasingly understood as a glycemic-improvement phenomenon rather than a direct toxic effect of the drug. Both deserve a place in the informed-consent conversation; neither, by current evidence, is a reason to broadly discourage GLP-1 therapy in patients who would otherwise benefit.

References

Hathaway JT, Shah MP, Hathaway DB, et al. Risk of Nonarteritic Anterior Ischemic Optic Neuropathy in Patients Prescribed Semaglutide. JAMA Ophthalmology. 2024;142(8):732–739. DOI: 10.1001/jamaophthalmol.2024.2296
Chou T-H, et al. Semaglutide and Nonarteritic Anterior Ischemic Optic Neuropathy Risk Among Patients With Diabetes. JAMA Ophthalmology. 2025. PubMed: 40146102
Marso SP, Bain SC, Consoli A, et al. Semaglutide and Cardiovascular Outcomes in Patients with Type 2 Diabetes (SUSTAIN-6). New England Journal of Medicine. 2016;375(19):1834–1844. DOI: 10.1056/NEJMoa1607141
Bell SJ, et al. Association of Glucagon-Like Peptide-1 Receptor Agonists With Optic Nerve and Retinal Adverse Events: A Population-Based Observational Study Across 180 Countries. American Journal of Ophthalmology. 2025. DOI: 10.1016/j.ajo.2025.01.030
World Health Organization. The use of semaglutide medicines and risk of non-arteritic anterior ischemic optic neuropathy (NAION). WHO News. June 27, 2025. Statement
U.S. Food and Drug Administration. WEGOVY (semaglutide) Prescribing Information. Reference ID: 5676363. 2025. Label
Novo Nordisk. FOCUS — A Research Study to Look at How Semaglutide Compared to Placebo Affects Diabetic Eye Disease in People With Type 2 Diabetes. ClinicalTrials.gov NCT03811561. Registration
American Academy of Ophthalmology. Update on Semaglutide Risks. EyeNet. 2025. Article

Last updated: 2026-05-08 Medical review: Dr. James Chen, MD, PhD, FACE