Exploring the oxidative stress-related molecular clusters in the cornea of patients with keratoconus

Keratoconus (KC), a progressive corneal disorder, severely compromises vision and quality of life, underscoring the urgency for biomarker identification and therapeutic advancements. This study investigates oxidative stress-related molecular clusters in KC and develops a robust predictive framework. Comprehensive analysis of gene expression datasets revealed oxidative stress-specific hub genes through WGCNA and delineated immune microenvironment features. The most effective prediction model, selected via a comparative assessment of four machine learning algorithms, was validated using ROC curves and external datasets. Furthermore, drug-gene interaction analysis identified candidate compounds targeting these hub genes. Seven key oxidative stress-related genes and two pathological clusters with distinct immune signatures were identified. Functional enrichment implicated immune-related pathways predominantly in Cluster 1. The XGB model demonstrated superior predictive performance (AUC = 0.957), and a nomogram incorporating five hub genes (TYMP, VGLL3, DISC1, TNF, and PRDM1) showed significant diagnostic potential. Lastly, three therapeutic candidates were proposed based on drug-gene interaction analysis. This study provides novel insights into the interplay between oxidative stress and KC pathogenesis while presenting a high-precision predictive model, paving the way for future diagnostic and therapeutic strategies.