Identification of circulating biomarkers and immune subtypes for diabetic retinopathy through multi-cohort integrative analysis.

PurposeTo elucidate the dynamic molecular evolution and immune heterogeneity in diabetic retinopathy (DR) and identify diagnostic biomarkers.MethodsMulti-omics integrative analysis was performed on peripheral blood transcriptomic data from diabetic patients with and without DR. Techniques included differential expression analysis, immune infiltration profiling, unsupervised clustering, and LASSO regression for biomarker selection. Findings were validated in an independent high-glucose-treated endothelial cell model.ResultsDR progression exhibits staged molecular evolution from diabetes (immune suppression) to early DR (cell cycle inhibition, immune burst) and advanced DR (metabolic stress, persistent immune dysregulation). Six co-downregulated immune modules indicated systemic immune suppression. Patients were classified into three immune subtypes: vascular-inflammatory, metabolic disorder, and oxidative stress-tissue remodeling. A diagnostic model based on IK, PLBD2, FLNA, H6PD, and ZMAT2 achieved an AUC of 0.876 for distinguishing DR from diabetes alone.ConclusionThis study delineates the dynamic molecular landscape and immune heterogeneity of DR, providing a potential multi-gene diagnostic signature and a framework for patient stratification, which may facilitate early detection and personalized management.