US Peptide Science Research Team
October 29, 2026
Since 2023, reports of nonarteritic anterior ischemic optic neuropathy (NAION)—a sudden, often permanent vision loss condition—have raised questions about glucagon-like peptide-1 (GLP-1) receptor agonists, particularly semaglutide. Researchers and clinicians distinguish between observational signals and randomized controlled trial (RCT) evidence to accurately characterize this risk. The distinction matters because Phase 3 RCT data, which employ rigorous blinding and randomization protocols, carry greater weight in establishing causal relationships than retrospective observational studies.
A comprehensive meta-analysis published in 2025 that stratified findings by study design and clinical indication found markedly different results between RCTs and observational studies. For diabetes populations, randomized controlled trials reported 5 ischemic optic neuropathy (ION) events in semaglutide-treated participants versus 1 event in controls, yielding a risk ratio of 1.76 (95% CI 0.43–7.25)—a finding that did not reach statistical significance. This means the confidence interval is wide enough to include the null hypothesis (no increased risk).
In contrast, observational studies examining diabetes patients on semaglutide versus non-users reported a pooled hazard ratio of 1.85 (95% CI 1.20–2.85), suggesting a potential 1.85-fold increased hazard. Critically, this observational signal was not replicated in the RCT data, which represents the higher-quality evidence standard. The 2025 meta-analysis, which included 8 RCTs (31,174 patients) and 8 observational studies (1,611,278 patients), concluded that the "stratified analysis provides a more nuanced interpretation than previous pooled reviews, isolating a potential increased risk signal to observational data in the diabetes population only, a finding not corroborated by RCTs."
Observational studies, which analyze electronic health records and administrative claims databases, are subject to confounding by indication—patients prescribed more potent GLP-1 agents often have worse cardiometabolic profiles or more severe diabetes, which themselves increase NAION risk independent of the medication. Additionally, observational studies may be affected by detection bias, where heightened clinical vigilance for ocular complications in semaglutide users leads to increased case identification. RCTs, by contrast, randomly assign participants and typically employ standardized outcome monitoring across treatment groups, reducing these biases.
Source research
Referenced literature: PubMed Central / NIH. No direct link is attached to this article.
The small absolute number of NAION events in RCTs (5 semaglutide vs. 1 control for diabetes indications) also means that statistical power is limited. A 2025 meta-analysis noted that trial sequential analysis indicated "the number of trials included was not sufficient to definitively determine a link between semaglutide and NAION," highlighting the challenge of detecting rare adverse events in controlled settings.
Absolute incidence matters more clinically than relative risk. Observational data reported a pooled NAION incidence of 26.7 per 100,000 person-years in semaglutide users versus 18.9 per 100,000 person-years in non-users—a difference of approximately 7.8 additional cases per 100,000 person-years. To contextualize: NAION is classified as a rare condition with a baseline incidence in the general population of roughly 2–10 per 100,000 per year, depending on age and risk factors. Even a several-fold elevation in relative risk translates to a small absolute risk increase, which is essential for informed clinical decision-making.
For weight-loss indications, both observational studies (HR 1.57, 95% CI 0.69–3.59) and RCTs (RR 2.18, 95% CI 0.33–14.34) showed no statistically significant evidence of increased NAION risk.
No. The proposed mechanisms remain speculative. The aao.org consensus statement from the North American Neuro-Ophthalmology Society and American Academy of Ophthalmology notes that GLP-1 receptors are present in the retina and central nervous system, making direct toxic effects theoretically possible. However, animal model studies of retinal diseases suggest that GLP-1 receptor activation is neuroprotective, arguing against direct toxicity.
The leading hypothesis involves rapid glycemic correction. When hemoglobin A1c drops quickly after GLP-1 initiation, optic disc edema may develop—particularly in individuals with a small optic cup-to-disc ratio (a "disc at risk"). In a confined anatomical space, tissue edema could trigger a compartment syndrome-like pathology leading to ischemia. This mechanism would be secondary to rapid metabolic correction rather than a direct drug effect. Supporting this concept, previous studies have documented that rapid blood glucose reduction can cause asymptomatic optic disc edema, a potential precursor to NAION.
A systematic review and meta-analysis of 78 trials examining semaglutide and ocular adverse events found that semaglutide did not increase or reduce the overall risk of eye disorders (odds ratio 1.01, 95% CI 0.91–1.12) or diabetic retinopathy (OR 1.04, 95% CI 0.92–1.17). This broader ocular safety signal is reassuring and suggests that if a NAION association exists, it represents a specific rather than class-wide phenomenon.
In June 2025, the European Medicines Agency (EMA) recommended that product information for semaglutide include NAION as a "very rare" side effect and advised immediate discontinuation if NAION is confirmed. However, the aao.org North American Neuro-Ophthalmology Society and American Academy of Ophthalmology do not support automatic discontinuation, citing the low absolute risk and potential harm from withdrawing effective therapy for patients with difficult-to-manage diabetes or obesity.
The consensus recommendation is shared decision-making rather than blanket avoidance. Patients with a history of vision loss in one eye or prior NAION should be informed of the possible risk. However, the clinical significance of the potential association remains unclear, and causation has not been demonstrated. For patients with severe obesity, poorly controlled diabetes despite other therapies, or significant cardiovascular comorbidities, the benefits of GLP-1 receptor agonist therapy may outweigh the small, unconfirmed ocular risk. This determination requires individualized assessment by the patient and their care team, including ophthalmologists, endocrinologists, and primary care physicians.
Most evidence focuses on semaglutide. The association between other GLP-1 receptor agonists (liraglutide, dulaglutide, tirzepatide) and NAION has been less frequently studied and warrants further investigation. Until additional data emerge, the consensus is not to assume that any potential risk is specific to semaglutide alone, but evidence is currently insufficient to characterize risk for other agents in the class.
The existing literature has not evaluated patients with a prior NAION diagnosis who are subsequently exposed to GLP-1 receptor agonists, so safety in this specific population remains unknown. Additionally, most RCTs were not designed with ocular adverse events as a primary outcome, meaning systematic ophthalmological assessment was not standardized across trials. Future postmarketing surveillance studies and large-scale observational studies employing rigorous methodology—such as propensity score matching to control for confounding by indication—are needed to clarify the conflicting findings.
Phase 3 randomized controlled trial evidence does not demonstrate a statistically significant increased risk of NAION with semaglutide use, though absolute event numbers remain small. Observational studies suggest a possible signal in diabetes populations, but this finding is not corroborated by RCT data. The absolute incidence of NAION in semaglutide users is low, and causality has not been established. Individuals experiencing sudden vision loss should undergo immediate ophthalmological evaluation. Clinical decisions regarding initiation, continuation, or discontinuation of GLP-1 receptor agonist therapy should involve shared decision-making informed by individual risk factors and clinical context rather than categorical recommendations.