Integrating Machine Learning and Multiomics Analyses to Identify Immune-Related Biomarkers and Mechanisms in Primary Biliary Cholangitis.

Abstract

Introduction: Primary biliary cholangitis (PBC) is a chronic autoimmune liver disease that gradually progresses, making early diagnosis and treatment challenging. Reliable biomarkers could enhance diagnostic accuracy and therapeutic development.

Methods: This study analyzed 3 publicly available gene expression data sets from the Gene Expression Omnibus database: GSE119600 (90 patients with PBC and 47 healthy controls), GSE159676 (12 PBC patients and 6 controls), and GSE61260 (11 patients with PBC and 38 controls). To identify genes closely linked to PBC, we applied machine learning techniques, including Least Absolute Shrinkage and Selection Operator, Support Vector Machine-Recursive Feature Elimination, and random forest. We subsequently conducted gene set enrichment and immune cell infiltration analyses to investigate their biological significance. IN addition, potential drug interactions were explored through the Drug Gene Interaction Database, and a competing endogenous RNA regulatory network was developed to examine gene regulation. Finally, the expression of selected genes was validated through multiplex immunofluorescence staining of liver tissue samples from patients with PBC.

Results: We identified proteasome subunit beta 7, TRAF family member associated nuclear factor kappa-light-chain-enhancer of activated B cells activator Albumin (TANK)-binding kinase 1, solute carrier family 29 member 1, and natural killer cell receptor 2B4 as key genes associated with PBC; these genes were significantly enriched in immune-related pathways and strongly correlated with immune regulation. Drug target prediction indicated that some genes could interact with existing immunomodulators or anticancer drugs. Competing endogenous RNA network analysis revealed that TANK-binding kinase 1, solute carrier family 29 member 1, and natural killer cell receptor 2B4 interact with multiple miRNAs and long noncoding RNAs, potentially regulating the immune microenvironment of PBC through noncoding RNA mechanisms. Immunofluorescence staining confirmed that these genes were highly expressed in liver tissues from patients with PBC.

Discussion: By integrating machine learning and functional analyses, this study identified 4 genes that may serve as potential biomarkers for PBC. Their involvement in immune regulation suggests possible applications in both diagnosis and therapy. Further studies are necessary to explore their clinical relevance and therapeutic potential.