Signature gene expression model for quantitative evaluation of MASH-like liver injury in mice

Abstract

Metabolic dysfunction-associated steatotic liver disease (MASLD) is a spectrum of chronic pathologic conditions strongly associated with metabolic syndrome and affects approximately 38 % of the global population. Untreated MASLD may progress to metabolic dysfunction-associated steatohepatitis (MASH), fibrosis, and cirrhosis and is currently recognized as one of the main risk factors for hepatocellular carcinoma (HCC). The molecular determinants of MASLD stratification are not clearly defined and require additional investigation. In this study, we used a dietary preclinical model of MASH-like liver injury induced by feeding male and female Collaborative Cross CC042/GeniUnc mice a high-fat and high-sucrose diet (HF/HS) for up to 60 weeks and analyzed the global hepatic transcriptomic alterations. Chronic feeding the HF/HS diet induced profound alterations in liver gene expression associated with the key toxicity pathways, including cell death, cell proliferation, inflammation, fibrosis, and hyperplasia. We identified a panel of 74 differentially expressed genes, the expression of which significantly correlated with total MASH pathology scores in the livers of both male and female mice. Using these genes, we developed a machine-learning model that accurately predicted the severity of MASH-like liver injury in several different animal models of MASH and demonstrated high accuracy for a smaller model with 37 genes. We also used this signature to analyze human gene expression data and show its translational relevance. The results of this study demonstrate that a panel of MASH-related genes can assist in the assessment of MASH-like liver injury, its monitoring, and in development of mechanism-based drugs against MASH.

Keywords: Liver Pathology; MASH; Machine Leaning Pathologist Assistance; Mice; Transcriptomics.