Identification and mechanistic investigation of novel biomarkers for diabetic retinopathy through hypoxia- and cuproptosis-associated gene analysis

Background Emerging evidence indicates that hypoxia- and cuproptosis-related molecular alterations may contribute to diabetic retinopathy (DR). This study aimed to identify biomarkers associated with these processes and investigate their biological relevance in DR. Methods The DR-related datasets GSE189005 and GSE221521 were sourced from public databases, while hypoxia-related and cuproptosis-related genes were extracted from published studies. Genes consistently differentially expressed across both datasets were intersected with those associated with hypoxia- and cuproptosis-related expression scores to derive candidate genes. Selection of candidate biomarkers was performed using machine-learning algorithms and expression consistency analysis, with diagnostic performance assessed via receiver operating characteristic (ROC) curves. Functional enrichment, immune cell infiltration, regulatory network analysis, candidate compound prediction, and reverse transcription quantitative PCR (RT-qPCR) were also conducted. Results Intersection screening yielded 13 candidate genes. Machine-learning analysis identified three candidate biomarkers. Notably, MED19 and CA11 were consistently downregulated in DR, exhibiting area under the curve (AUC) values exceeding 0.70 in both datasets, thus qualifying as biomarkers. Functional enrichment analysis suggested MED19's involvement in “proteasome” and “spliceosome” pathways, while CA11 was linked to “regulation of autophagy” and “basal cell carcinoma” pathways. Differential infiltration of three immune cell populations (eosinophils, M2 macrophages, activated natural killer cells) was noted between DR and control groups in GSE189005, with M2 macrophages demonstrating a significant negative correlation with both biomarkers. Regulatory network analysis highlighted several candidate transcriptional regulators for MED19 and CA11, including ELF1–MED19 and KDM5B–CA11. Thirty candidate compounds/interventions were identified, including schizandrin B and pirinixic acid. RT-qPCR confirmed the significant downregulation of MED19 and CA11 in DR samples, providing preliminary support. Conclusion MED19 and CA11 were identified as biomarkers, presenting a potential framework for therapeutic intervention in DR.