Pan-Fibrotic Gene Expression Signature in Major Chronic Diseases by Integrative Bulk and Single-Cell Transcriptomic...

Fibrosis is a shared feature of chronic liver, kidney and heart disease, yet no molecular signature captures the mechanisms common to these organs. We integrated human microarray, bulk and single-nucleus RNA sequencing (snRNA-seq) data across 2022 unique tissue samples to identify a conserved pan-fibrotic signature. The pan-fibrotic signature was defined in a microarray discovery cohort (n = 1051) and verified in a microarray liver (n = 325), two RNA-seq (n = 414) and four integrated snRNA-seq validation datasets (n = 69). Diagnostic models were built using five machine learning approaches in separate RNA-seq cohorts (training n = 209; liver validation n = 70; kidney validation n = 24). 492 fibrosis-associated genes were identified across organs, including 23 hallmark genes, a subset of which correlated with disease severity. A novel finding was consistent metallothionein (MT1E, MT1X) downregulation; snRNA-seq resolved cell-type-specific expression of 12 hallmark genes, including MT1E/MT1X predominantly expressed in hepatocytes and liver sinusoidal endothelial cells. Elastic Net performed best across all three gene sets (AUROC 0.765-0.830), with scores correlating significantly with interstitial fibrosis and tubular atrophy (%IFTA; rho 0.58-0.73). This study defines a cross-organ pan-fibrotic signature, providing potential candidate therapeutic targets and biomarkers applicable across tissues affected by fibrosis.