Reynolds · Annals of internal medicine 2026 · retrospective cohort study (target trial emulation) · n=?

Glucagon-Like Peptide-1 Receptor Agonists and Risk for Ischemic Optic Neuropathy : A Target Trial Emulation.

Cited 0 times in the scientific literature.

Level 3 - non-randomized controlled study

Non-randomized controlled cohort study using target trial emulation with inverse probability weighting.

PubMed 42441967 · doi:10.7326/ANNALS-25-00860 · record verified 2026-08-26

What was done

Researchers conducted an observational target trial emulation using a large US commercial claims database (January 2017 to December 2022) in patients aged 18 to 65 years with type 2 diabetes. They compared the risk of incident ischemic optic neuropathy (ION, evaluated as a proxy for nonarteritic anterior ischemic optic neuropathy) over 18 months among new users of GLP-1 receptor agonists (GLP-1RAs) versus sodium-glucose cotransporter-2 inhibitors (SGLT2is) and dipeptidyl peptidase-4 inhibitors (DPP4is). Models adjusted for more than 80 covariates using inverse probability of treatment weighting to estimate 18-month cumulative incidence and risk differences (RDs) per 10,000 patients.

What was found

The 18-month risk for ION was 8.5 versus 5.5 per 10,000 among GLP-1RA users versus SGLT2i users (RD, 3.0 [95% CI, 0.4 to 5.7]; number needed to harm [NNH], 3,333). Compared with DPP4i users, the risk was 7.8 versus 4.2 per 10,000 (RD, 3.6 [95% CI, 1.1 to 6.1]; NNH, 2,778). Among GLP-1RA users, 69 of 81 ION events (85.2%) occurred in patients older than 50 years and 57 (70.3%) occurred in men. Risk differences were attenuated among metformin monotherapy users (2.0 vs. SGLT2is; 4.1 vs. DPP4is) compared with users of 2 or more diabetes drugs (5.7 vs. SGLT2is; 4.0 vs. DPP4is). Risk differences were higher in men, patients aged 50 years or older, and those with pre-existing cardiovascular or ophthalmic conditions, with minimal differences seen in women and individuals younger than 50 years.

Why it matters

This study provides large-scale comparative safety data on the controversial link between GLP-1RAs and optic neuropathy, showing that while a small relative increase in risk exists, the absolute excess risk is minimal (around 3 to 4 cases per 10,000 patients over 18 months).

Limits

The total sample size was not reported in the abstract. Diagnostic codes specifically identifying NAION were unavailable, necessitating the use of broader ION diagnostic codes as a proxy. The administrative database lacked data on critical clinical confounders, including body mass index and duration of type 2 diabetes, leaving potential for residual confounding that precludes causal conclusions. The study population was restricted to commercially insured patients aged 18 to 65 years.

Cited by