Identification of novel biomarkers and associated immunoregulatory network in pterygium through integrated transcriptome analysis and machine learning approaches.

Abstract

Pterygium is a common eye disease with unclear pathogenesis. This study aims to identify novel biomarkers for pterygium and explore their mechanisms by RNA sequencing (RNA-seq). Twelve conjunctival and sixteen pterygium samples were collected for RNA-seq. Differential expression analysis, weighted gene co-expression network analysis (WGCNA), and protein-protein interaction (PPI) network analysis were conducted. Multiple machine learning algorithms including XGBoost, Boruta, and LASSO were employed to identify biomarkers. The diagnostic value of biomarkers was evaluated through ROC curve analysis. To explore the functions of biomarkers, gene set enrichment analysis (GSEA), immune infiltration, lncRNA-miRNA-mRNA regulatory network, master regulator analysis (MRA), and transcription factor (TF) mRNA analysis were conducted. qRT-PCR was performed to validate the expression levels of biomarkers. Five novel biomarkers (HSPA8, HBB, ARRB1, IRS1, and FLT4) associated with pterygium were identified. The nomogram model incorporating these biomarkers showed excellent diagnostic performance (AUC = 1). GSEA revealed that these biomarkers were primarily involved in oxidative phosphorylation, cytokine receptor interaction, and chemokine signaling pathways. 168 miRNAs and 1039 lncRNAs related to these biomarkers were predicted in the ceRNA regulatory network. MRA identified Retinoic acid receptor alpha (RARα) as a key TF regulating these biomarkers. Immune infiltration analysis showed significant differences in 22 types of immune cells between pterygium and control groups. The differential expression of five biomarkers was validated using qRT-PCR (all p < 0.05). This study identified five novel pterygium biomarkers (HSPA8, HBB, ARRB1, IRS1, & FLT4) and revealed a complex immunoregulatory network involving immune cells, lncRNA, miRNAs, and TFs (RARα).