Dhivagaran · Neurology 2026 · systematic review and Bayesian meta-analysis · n=5 studies

Glucagon-like Peptide-1 Receptor Agonists and Risk of Nonarteritic Anterior Ischemic Optic Neuropathy: Systematic Review and Meta-Analysis.

Cited 3 times in the scientific literature.

Level 3 - non-randomized controlled study

Systematic review and meta-analysis of non-randomized observational studies

PubMed 42060879 · doi:10.1212/WNL.0000000000214864 · record verified 2026-08-26

What was done

A systematic review and Bayesian random-effects meta-analysis evaluated the risk of nonarteritic anterior ischemic optic neuropathy (NAION) associated with GLP-1 receptor agonist (GLP-1RA) exposure. Searches across Embase, Medline, and Cochrane CENTRAL through April 5, 2025, identified randomized trials and observational studies comparing GLP-1RA users with non-users. Risk of bias was assessed using the ROBINS-I tool, and certainty was evaluated with the GRADE framework.

What was found

Five studies encompassing 1,593,554 patients (682,456 semaglutide users and 911,098 non-users) were included in the primary analysis. Semaglutide use was associated with an increased risk of NAION compared with non-GLP-1RA use (RR 2.52, 95% CrI 1.56-4.72, I2 57.8%, P(RR > 1) = 99.9%), and specifically among patients with diabetes (RR 2.41, 95% CrI 1.57-4.10, I2 47.2%, P(RR > 1) = 99.9%). Across 5 studies and 7 comparisons (1,460,760 patients, 272 NAION events), the overall incidence was 85 cases per 100,000 GLP-1RA users (95% CrI 29-263, I2 98.8%). For semaglutide users (1,108,542 patients, 198 events), incidence was 118 cases per 100,000 (95% CrI 32-451, I2 98.8%).

Why it matters

This review demonstrates a more than twofold increased risk of NAION with semaglutide. These estimates provide quantitative data to guide clinical risk-benefit counseling.

Limits

The evidence base is graded low-to-moderate certainty, reflecting observational designs subject to residual confounding. The analysis relies on only 5 studies with few absolute events (272 cases total), and pooled incidence exhibited extreme between-study heterogeneity (I2 98.8%).

Cited by