Semaglutide and Diabetic Retinopathy Risk: Evidence from FAERS Database Analysis and Protective Effects on High Glucose-Stimulated Retinal Microvascular Endothelial Cells

AIM Diabetic retinopathy (DR) is a leading cause of vision loss in type 2 diabetes, and concerns remain regarding semaglutide's potential association with DR progression. This study aimed to evaluate semaglutide-related DR risk using real-world pharmacovigilance data and explore its direct effects on retinal microvascular cells under high glucose conditions. METHODS Data from the FDA Adverse Event Reporting System (FAERS, 2004 Q1–2024 Q2) were analyzed, including 287,470 semaglutide users with 82,931 adverse events. Four disproportionality algorithms [reported odds ratio (ROR), proportional reporting ratio (PRR), Bayesian confidence propagation neural network (BCPNN), empirical Bayesian geometric mean (EBGM)] were used to detect DR-related signals. In vitro experiments were performed on human retinal microvascular endothelial cells (HRMECs) cultured in normal glucose (5 mM) or high glucose (30 mM). Semaglutide (5–1000 nM) cytotoxicity was assessed via MTT assay. Cell proliferation (EdU labeling), angiogenesis (Matrigel assay), migration (Transwell), and GLP-1 receptor (GLP-1R) expression (Western blotting) were evaluated. RESULTS FAERS analysis identified 2004 ocular adverse events in semaglutide users, with 88 cases of DR (ROR = 19.57, PRR = 19.55, IC025 = 3.67, EBGM05 = 15.37) that did not rank among top 30 adverse event terms. Gastrointestinal (n = 22,801) and metabolic disorders (n = 5,230) were the most frequently reported. High glucose significantly downregulated GLP-1R in HRMECs (P < 0.05). Semaglutide at non-cytotoxic concentrations (50, 100 nM) inhibited high glucose-induced hyperproliferation, excessive angiogenesis, and enhanced migration (all P < 0.01 vs. high glucose group). CONCLUSION Real-world FAERS data indicate a low clinical association between semaglutide and DR. Preclinically, semaglutide protects HRMECs from high glucose-induced dysfunction via GLP-1R modulation, supporting its safety profile in DR-prone individuals. These findings contribute to understanding GLP-1RA effects on diabetic microvascular complications.