Identification of Shared Immune-Related Biomarkers and Construction of a Combined Diagnostic Model for Hepatic Fibrosis and Inflammatory Bowel Disease Using Bioinformatics and Machine Learning

Background Hepatic fibrosis (HF) and inflammatory bowel disease (IBD) frequently coexist; however, shared transcriptomic alterations and diagnostic biomarkers applicable to both conditions remain to be further elucidated.Methods GEO training datasets (HF: GSE84044; IBD: GSE59071) were analyzed to identify differentially expressed genes (DEGs), and the overlap was defined as common DEGs. Weighted gene co-expression network analysis (WGCNA) was performed separately to identify phenotype-related modules and hub genes; intersecting these with the ImmPort immune gene set yielded immune-related candidates. Eight machine-learning algorithms were applied for consensus feature selection to construct a four-gene combined diagnostic model, which was externally validated in independent cohorts (HF: GSE49541; IBD: GSE66407) by expression comparison and ROC analysis. Immune cell infiltration was estimated using CIBERSORT and correlated with hub genes; gene–miRNA and gene–transcription factor (TF) regulatory networks were constructed. Candidate compounds were explored via enrichment analysis using Enrichr-DSigDB.Results A total of 172 common DEGs were identified in the training datasets, mainly enriched in immune/inflammatory pathways (cytokine–cytokine receptor interaction, chemokine signaling, IL-17 signaling, and Toll-like receptor signaling) and remodeling-related pathways (ECM–receptor interaction, cell adhesion molecules, and leukocyte transendothelial migration). WGCNA intersection yielded 150 shared hub genes; integrating them with common DEGs resulted in 34 core targets, from which 13 immune-related candidate genes were obtained. Consensus selection across algorithms identified IFI16, CASP1, ANXA3, and THY1 as hub genes. All four genes were significantly upregulated in both training sets. In external validation, all four genes remained upregulated in the IBD cohort (GSE66407), whereas ANXA3, IFI16, and THY1 were significantly increased but CASP1 showed no significant difference in the HF cohort (GSE49541). The four-gene model achieved AUCs of 0.971 (GSE59071), 0.865 (GSE84044), 0.966 (GSE49541), and 0.784 (GSE66407). Immune infiltration analysis indicated enhanced myeloid-cell infiltration, particularly macrophage-lineage cells, in both diseases, and hub genes showed significant associations with multiple immune-cell subsets.Conclusions HF and IBD share transcriptomic features related to immune-inflammatory activation and extracellular matrix remodeling. IFI16, CASP1, ANXA3, and THY1 represent candidate diagnostic biomarkers for both conditions, and their combined model demonstrates discriminatory performance across cohorts, providing clues for subsequent mechanistic validation and translational studies.