PMID-41021211 — Cheng: Multi-Database Pharmacovigilance of GLP-1 Ophthalmic Risks

[DRAFT — authored 2026-04-19. Citation web-verified 2026-04-19 against PubMed.]

Citation

Cheng X, Jiang Z, Li G, Wang J, Han F. Multi-database pharmacovigilance assessment of GLP-1 receptor agonist-related ophthalmic risks using advanced signal detection in FAERS and VigiBase. Journal of Endocrinology and Investigation. 2026;49(2):425-433. doi: 10.1007/s40618-025-02712-3. PMID: 41021211.

External URL: DOI resolver

Study Design

Design: Multi-database disproportionality (signal-detection) pharmacovigilance analysis
Databases: FDA Adverse Event Reporting System (FAERS) + WHO VigiBase
Scope: Ophthalmic adverse events reported across the full GLP-1 receptor agonist class (semaglutide, liraglutide, dulaglutide, exenatide, tirzepatide)
Method: Advanced signal-detection statistics (reporting odds ratio and related disproportionality metrics)

Key Findings

Semaglutide: Strongest ocular signals across the class, particularly for NAION (non-arteritic anterior ischemic optic neuropathy):
- 2,878 NAION cases in VigiBase
- 2,047 NAION cases in FAERS
Dulaglutide: Elevated rates of visual impairment and diabetic retinopathy; numerous early-onset events (≤30 days post-initiation) flagged
Demographics: Age 45-64 years and female sex predominant in reporting
Labeling gap: Authors note that "ocular adverse events are incompletely characterized" in current prescribing information across the class
Triangulation context: Aligns with and extends the Hathaway cohort (2024) and Lakhani 180-country analysis (2025) — adds volumetric disproportionality data across two large pharmacovigilance databases

Clinical Relevance

Cheng 2026 is the most recent multi-database pharmacovigilance synthesis of GLP-1 RA ocular adverse events as of publication. Core points:

Semaglutide NAION signal is confirmed at scale — volumetric data (thousands of reports across two databases) corroborate the Hathaway HR and Lakhani ROR findings
Dulaglutide signal for visual impairment / diabetic retinopathy is a distinct-but-related class signal warranting monitoring
Labeling characterization is incomplete — a direct argument for updated prescriber education and informed consent
Teaching implication: The NAION triangulation framework (Hathaway → Danish/Norwegian → Lakhani → Cheng → EMA/WHO) is increasingly robust; US FDA labeling-inaction divergence is now at odds with multiple independent pharmacovigilance signals

Limitations (Author-acknowledged + inherent to pharmacovigilance)

Pharmacovigilance data is subject to reporting bias, notoriety bias (post-FDA alert and EMA action, reports can spike independent of incidence change), and confounding by indication (GLP-1 RA patients have higher metabolic/vascular baseline risk)
Disproportionality signals are hypothesis-generating, not causal
Heterogeneity in classification of "ophthalmic AE" across national reporting systems

Evidence Level

Level II (Oxford CEBM for pharmacovigilance disproportionality analyses) — hypothesis-generating, complementary to cohort and RCT evidence.

Linked Peptides

Related Studies

PMID-38958939 – Hathaway NAION Cohort — seminal single-center cohort
PMID-39696569 – Danish Norwegian NAION Cohort — nationwide cohort
PMID-40383360 – Lakhani 180 Countries GLP-1 Ocular Events — parallel 180-country pharmacovigilance
REG-EMA-PRAC-Semaglutide-NAION-2025 – EMA NAION Very Rare — EMA regulatory outcome
CONF-2025-EASD-Vilsboll-NAION — trial-pooled conference data

Orchestrator Notes

Citation web-verified 2026-04-19 via PubMed.
Adds pharmacovigilance-scale volumetric evidence to the NAION triangulation framework in Lesson 5.2 and Semaglutide profile NAION section.

PMID-41021211 – Cheng GLP1 Ophthalmic Pharmacovigilance FAERS VigiBase