Integrative Transcriptomics Across Etiologies Reveals Common and Disease-Specific Fibrogenic Signatures in Liver Fibrosis.

Abstract

Background: Chronic liver diseases caused by metabolic, viral, and mixed etiologies frequently converge on fibrosis and cirrhosis; however, the extent to which fibrogenic mechanisms are shared across etiologies versus disease-specific remains incompletely defined.

Methods: Four GEO datasets were analyzed: GSE135251 (NAFLD-related fibrosis), GSE84044 (HBV-related fibrosis), GSE197112 (mixed-etiology fibrosis), and GSE14323 (cirrhosis versus normal). Differential expression analysis was performed separately within each dataset using DESeq2 for RNA-seq and limma for microarray data. Shared genes were identified by cross-dataset intersection. For downstream network and ordination analyses, gene-level matrices were harmonized across platforms and batch-adjusted using ComBat, with PCA before and after correction provided in the supporting information. WGCNA and random forest were then applied to the integrated matrix, and the final seven hub genes were defined as genes supported by recurrent differential expression, co-expression prioritization, and random forest feature importance. The final seven-gene hub panel was further used for exploratory age-stratified visualization and regression analysis. Functional enrichment, miRNA-mRNA mapping, PPI analysis, PCoA, UMAP, RT-qPCR, and western blotting were performed.

Results: Across the four cohorts, 434-787 differentially expressed genes were identified per dataset, and 26 genes were consistently upregulated across all etiologies. Enrichment analyses converged on extracellular matrix organization, TGF-β, PI3K-Akt, MAPK, and Wnt-related signaling. The final seven hub genes were MAOA, LOC102724200, SLC16A3, GPM6B, CST7, MT3, and ZNF142. Exploratory age analyses in the age-annotated GSE84044 cohort suggested that a subset of the final seven hub genes varied with age; however, these findings should be interpreted cautiously because age metadata were not uniformly available across all public cohorts. RT-qPCR in 10 fibrotic and 10 nonfibrotic liver tissues confirmed upregulation of the seven hub genes, and western blotting supported increased protein abundance of CST7, MT3, SLC16A3, and MAOA.

Conclusions: This integrative analysis identifies both shared and etiology-associated transcriptional programs in liver fibrosis and defines a multistep strategy for prioritizing conserved hub genes. The seven validated hub genes represent candidate biomarkers for fibrotic liver injury, whereas the exploratory age-related findings based on this seven-gene panel remain hypothesis-generating. This workflow may support future cross-platform transcriptomic studies of hepatic fibrosis.