GLP-1 and Glaucoma: NAION Signal, IOP, and Dry Eye Evidence

In July 2024, a Mass Eye and Ear neuro-ophthalmology group published a case-control analysis (Hathaway 2024, JAMA Ophthalmology^[1]) reporting that patients prescribed semaglutide had a roughly four-fold higher hazard of non-arteritic anterior ischemic optic neuropathy (NAION) than matched non-GLP-1 comparators — sudden, painless, usually-irreversible monocular vision loss. The paper went viral, the methodology was sharply critiqued in an accompanying editorial (Mollan 2024^[4]), and two large population replications followed: a Danish-Norwegian cohort (Simonsen 2025^[2]) that found a much smaller association, and a US population-based study from the same Harvard pharmacoepidemiology group (Tesfaye 2026^[3]) that confirmed an elevated risk in the diabetic population but at substantially lower magnitude than the original signal. This article walks through what the eye literature actually says — NAION, glaucoma, intraocular pressure, dry eye — and the practical ophthalmology follow-up protocol for patients on Wegovy, Ozempic, Zepbound, and Mounjaro.

The honest summary

• NAION is rare; the relative-risk signal is real but smaller than the original headline. Hathaway 2024^[1] reported HR ~4 in diabetic patients prescribed semaglutide; Simonsen 2025^[2] and Tesfaye 2026^[3] replicated an elevated but smaller signal in population data. Absolute incidence remains a few cases per 10,000 person-years.
• Glaucoma data trend favorable. Sterling 2023 (Br J Ophthalmology^[5]) found GLP-1RA use was associated with a reduced incidence of new glaucoma diagnoses. Preclinical work (Sterling 2020^[6], Lawrence 2023^[7]) shows GLP-1R agonists rescue retinal ganglion cells in ocular-hypertension models.
• Intraocular pressure is not meaningfully changed.The published series do not show a consistent IOP signal in either direction; patients on glaucoma drops should continue them.
• Dry-eye symptoms are common during dose escalation and usually resolve. Tear-film improvement on stable dosing has been reported (Ottonelli 2025^[8]). Carboxymethylcellulose or hyaluronate artificial tears are first-line.
• Existing retinopathy is the main red flag.SUSTAIN-6 (Marso 2016^[9]) reported a HR of 1.76 for retinopathy complications in patients with pre-existing retinopathy — consistent with the DCCT-era pattern of early worsening following rapid glycemic improvement (DCCT 1998^[10]).

The NAION signal: what Hathaway 2024 actually reported

NAION is a stroke of the optic nerve head. Blood supply to the front of the optic nerve fails — usually overnight or on awakening — producing sudden, painless monocular vision loss with an altitudinal field defect. The classic patient is older, has a small “disc at risk” cup-to-disc ratio, hypertension, diabetes, or sleep apnea. There is no proven treatment; about a third of patients later lose vision in the fellow eye.

Hathaway 2024^[1] was a retrospective case-control analysis of 16,827 patients seen for any reason at the Mass Eye and Ear neuro-ophthalmology clinic between 2017 and 2023. Among 710 patients with type 2 diabetes, those prescribed semaglutide had a hazard ratio of roughly 4.28 for NAION compared with matched diabetics on other anti-hyperglycemic medications; in the obesity subgroup, the hazard ratio was approximately 7.6. Absolute incidence was 8.9% in the semaglutide-diabetes group versus 1.8% in the non-GLP-1 diabetes group at three-year follow-up — a striking signal in a referred neuro-ophthalmology cohort.

The accompanying editorial (Mollan 2024^[4]) raised the methodology concerns that have framed every subsequent discussion: single-center, referral-biased cohort (Mass Eye and Ear sees an enriched population of optic-nerve disease), small absolute case counts, no adjustment for the “disc at risk” cup-to-disc ratio, and possible confounding by indication (sicker diabetics get newer agents). Mollan's editorial explicitly cautioned against changing prescribing practice on a single observational study.

The replications: Simonsen 2025 and Tesfaye 2026

Two well-powered population studies have since attempted to replicate the signal in datasets that are less susceptible to single-center referral bias.

Simonsen 2025^[2] (Diabetes, Obesity and Metabolism) used linked Danish and Norwegian national registries to follow 424,152 patients with type 2 diabetes initiating either semaglutide or an SGLT2 inhibitor. Over two years of follow-up, NAION incidence was higher with semaglutide — but the hazard ratio was approximately 2.2, not 4 or 7, and absolute incidence remained low (roughly 2 cases per 10,000 person-years). The Danish-Norwegian analysis controlled for sleep apnea, hypertension, prior ophthalmic history, and HbA1c trajectory — confounders the case-control design could not handle.

Tesfaye 2026^[3] (Diabetes, Obesity and Metabolism, same Harvard pharmacoepidemiology group as the original Hathaway co-author) used US commercial-claims data across multiple GLP-1RAs and reported an elevated but attenuated signal in patients with type 2 diabetes, consistent with the Danish-Norwegian magnitude. The Tesfaye paper is particularly notable because Hathaway was a co-author — the original team confirming that the population-level magnitude is smaller than their single-center headline implied.

The honest synthesis is that NAION risk on semaglutide is probably modestly elevated in patients with type 2 diabetes, with absolute incidence still in the few-per-10,000-person-years range. That is rare in absolute terms but not negligible for a non-cosmetic indication, and it is irreversible if it happens. The mechanism is hypothesized to involve rapid glycemic improvement combined with an already-vulnerable small-vessel circulation at the optic nerve head — the same physiology behind DCCT-era early-worsening retinopathy (DCCT 1998^[10]).

Glaucoma: the data trend favorable

The glaucoma evidence is the inverse of the NAION pattern — consistently favorable in both observational and preclinical data.

Sterling 2023^[5] (British Journal of Ophthalmology) analyzed a US claims database and reported that GLP-1RA use was associated with a reduced risk of new-onset glaucoma diagnosis compared with matched users of non-GLP-1 anti-hyperglycemic agents. The effect was modest in magnitude but consistent across sensitivity analyses.

The preclinical story is more direct. Sterling 2020^[6] (Cell Reports) showed that NLY01, a long-acting GLP-1 receptor agonist, reduced retinal microglial activation and rescued retinal ganglion cells in a mouse model of ocular hypertension — the cell type whose loss defines glaucoma. Lawrence 2023^[7] (Frontiers in Cellular Neuroscience) extended the finding by showing that both topical (eye-drop) and systemic GLP-1R agonist administration rescued retinal ganglion cells in the same hypertensive-glaucoma model. GLP-1R is expressed on retinal ganglion cells and on retinal microglia, providing a plausible mechanism for the observational signal.

For a patient with known glaucoma starting a GLP-1, this means: the medication is not contraindicated, the data lean protective, and the practical advice is to continue all glaucoma drops and maintain routine ophthalmology follow-up.

Intraocular pressure: no consistent signal

Across the published series, GLP-1 receptor agonists do not appear to meaningfully change intraocular pressure in either direction. Patients on prostaglandin analogs (latanoprost, bimatoprost), beta-blockers (timolol), alpha agonists (brimonidine), or carbonic anhydrase inhibitors (dorzolamide) should continue these as prescribed. There is no published rationale for adjusting glaucoma medication around GLP-1 initiation or titration.

Dry eye during dose escalation

Dry eye is one of the more commonly reported but under-discussed eye complaints during the first two to three months of GLP-1 therapy. The mechanism is straightforward: nausea, reduced fluid intake, and the volume loss that accompanies rapid weight loss reduce tear-film volume and thicken tear-film osmolarity. Patients describe gritty eyes, burning on waking, fluctuating vision in the late afternoon, and contact-lens intolerance.

Two practical points stand out. First, the symptom usually resolves on stable dosing and adequate hydration. Ottonelli 2025^[8] (Clinical Ophthalmology) followed type 2 diabetics on GLP-1RA therapy and actually reported improvement in ocular-surface parameters — tear break-up time, Schirmer testing, and ocular-surface disease index — on stable medication. The dry-eye signal is a transient escalation effect, not a chronic ocular-surface disease.

Second, first-line management is over-the-counter: preservative-free carboxymethylcellulose drops (Refresh, TheraTears) four to six times daily, hyaluronate drops (Blink, Hylo) as an alternative, and a nightly lipid- based gel or ointment for patients with morning symptoms. For patients whose symptoms persist beyond three months on stable dosing, refer for ophthalmologic evaluation: meibomian-gland expression, punctal plugs, or short-course cyclosporine/lifitegrast prescription drops may be warranted.

Pre-existing retinopathy: the DCCT precedent

SUSTAIN-6 (Marso 2016 NEJM^[9]) reported a hazard ratio of 1.76 for retinopathy complications in patients with pre-existing retinopathy randomized to semaglutide. The finding is mechanistically consistent with the Diabetes Control and Complications Trial pattern (DCCT 1998^[10]) of early worsening of diabetic retinopathy in the first 6–12 months of any rapid glycemic improvement — not specific to GLP-1RAs but more visible in this class because the HbA1c drops are larger and faster. The long-term retinopathy outcomes still favor tight glycemic control; the short-term worsening is real and warrants a pre-treatment dilated eye exam in any diabetic with established retinopathy and ophthalmology follow-up at six months. The full retinopathy-monitoring protocol is covered in our companion article on diabetic retinopathy ophthalmology monitoring on GLP-1s.

Magnitude: the published eye signals on one chart

The practical ophthalmology follow-up protocol

1. Baseline eye history before starting a GLP-1.Ask about known glaucoma, diabetic retinopathy, prior NAION (absolute red flag for the fellow eye), sleep apnea, small cup-to-disc ratio noted on a prior exam, and current ophthalmology follow-up schedule.
2. Diabetic patients with established retinopathy: pre-treatment dilated eye exam and 6-month re-examination. Aligns with the SUSTAIN-6 and DCCT signal. Coordinate with the prescribing endocrinologist on titration pace if proliferative retinopathy is present.
3. Patients with known glaucoma: continue all glaucoma drops; maintain routine follow-up. The data lean protective. No drug interaction with prostaglandin analogs, beta-blockers, alpha agonists, or carbonic anhydrase inhibitors.
4. Dry-eye during dose escalation: preservative-free artificial tears first. Carboxymethylcellulose or hyaluronate drops, 4–6 times daily. Adequate hydration. Persistent symptoms past three months on stable dosing warrant ophthalmology referral for meibomian-gland assessment and possible punctal plugs.
5. Any new visual symptom on a GLP-1: stop dose escalation and refer urgently. Sudden monocular vision loss, an altitudinal field defect, or new optic-disc swelling on exam is a same-day neuro-ophthalmology referral regardless of whether the medication is the cause. NAION has no proven treatment; the value of urgent evaluation is diagnosis confirmation, fellow-eye risk assessment, and documentation.
6. Prior NAION: discuss risk-benefit explicitly before initiation. About one third of NAION patients subsequently lose vision in the fellow eye from independent causes; an additional GLP-1-associated hazard, even at the smaller population magnitude, deserves a documented conversation. For obesity indication without diabetes, the risk-benefit lean is less favorable; for a diabetic patient who needs cardiovascular risk reduction (semaglutide is the only GLP-1RA with a primary CV outcome benefit in SUSTAIN-6^[9]), the calculus may still favor treatment.
7. Provider routes. A general ophthalmologist handles routine dilated exams, dry-eye management, and IOP measurement. New visual symptoms route to neuro-ophthalmology. Diabetic retinopathy management routes to retina specialty.
8. Insurance and cost. An ophthalmology office visit is covered under any commercial or Medicare plan; self-pay typically runs $150–400 for an established-patient comprehensive exam, more for a dilated retinal exam with imaging. The cost is not a reason to skip baseline assessment if any risk factor is present.

Related research and tools

• Diabetic retinopathy and GLP-1 monitoring — the SUSTAIN-6 retinopathy signal and the ophthalmology schedule for diabetic patients
• GLP-1 first 30 days survival guide — the dose-escalation window when dry-eye symptoms peak alongside nausea and dehydration
• GLP-1 side effect questions answered — the broader hub covering FAERS signals, rare events, and how to think about reporting bias
• GLP-1 heart rate and palpitations — the adjacent vascular-symptom literature and when autonomic side effects warrant cardiology workup
• GLP-1 tinnitus and hearing evidence — another rare FAERS-flagged sensory signal with a similar “case-control vs population” replication arc

Important disclaimer. This article is educational and does not constitute medical advice. Any new visual symptom on GLP-1 therapy — sudden vision loss, an altitudinal field defect, persistent blurring beyond the dose-escalation window, or visible optic-disc swelling — warrants same-day evaluation by an ophthalmologist or emergency department, regardless of whether the medication is the cause. Decisions about starting, continuing, or stopping a GLP-1 in a patient with prior NAION, proliferative retinopathy, or advanced glaucoma should be made jointly with the prescribing clinician and an ophthalmologist. PMIDs were verified live against the PubMed E-utilities API on 2026-05-29.

Last verified: 2026-05-29. Next review: every 12 months, or sooner if a new prospective trial or meta-analysis on GLP-1 ophthalmic outcomes is published.