Identification of oxidative stress-associated biomarkers for inflammatory bowel disease through integrated machine learning and weighted gene co-expression network analysis

Background Inflammatory bowel disease (IBD) is a complex chronic intestinal inflammatory disorder. Oxidative stress (OS) is crucial in the pathogenesis of IBD by promoting inflammation and disrupting the intestinal epithelial barrier. Methods Weighted gene co-expression network analysis and machine learning were applied to publicly available gene expression data to identify OS-related biomarkers for IBD. Receiver operating characteristic curve, immune infiltration analysis, and single-cell analysis were used to evaluate diagnostic performance and cellular localization. RT-qPCR and Western blotting were conducted to validate gene expression. Results Three hub genes (LCN2, APP, and TNFSF4) demonstrated strong diagnostic accuracy (AUC: 0.781, 0.805, and 0.754, respectively) in the discovery cohort (GSE3365). Furthermore, external validation in three independent cohorts also confirmed the robust diagnostic performance of the three hub genes. These genes were enriched in OS and inflammatory pathways, strongly correlated with infiltration of macrophages M0, monocytes, and neutrophils (p<0.01), and were highly expressed in goblet cells, stem cells, and T cells. DSS model validation confirmed significant upregulation at mRNA and protein levels. Conclusion These OS-associated genes represent novel molecular targets for precise diagnosis and personalized treatment of IBD. This finding highlights the interplay among OS, immune dysregulation, and inflammation in IBD pathogenesis.