A Single-Cell Transcriptomic Evaluation of Livers from MALSD-HCC Mouse Models and Comparison to Humans

Abstract

Hepatocellular carcinoma arising from metabolic dysfunction-associated steatotic liver disease (MASLD-HCC) is an increasingly prevalent cancer subtype globally, but its heterogeneous progression poses challenges for preclinical modeling. Two widely used murine models of MASLD-HCC—the carbon tetrachloride (CCl₄) fibrosis-based model and the diethylnitrosamine (DEN) genotoxic model—have not been systematically compared using cell type–resolved analysis. This study integrates histological, phenotypic, and single-cell transcriptomic analyses to evaluate how each model recapitulates human MASLD-HCC. Both models developed steatosis, fibrosis, and tumors, but histological profiling revealed key distinctions. DEN mice exhibited greater liver-to-body weight ratios, lesion counts, and steatosis and ballooning scores, consistent with lipid-rich tumor burden. In contrast, CCl₄ mice showed more pronounced fibrosis and inflammation, reflecting a fibrosis-dominant phenotype. Using single-cell RNA-seq data from murine livers and publicly available human MASLD-HCC datasets, we compared hepatocytes, hepatic stellate cells (HSCs), macrophages, and endothelial cells (ECs) across conditions. Cross-species correlation and pathway enrichment analyses highlighted distinct strengths and limitations of each model. In human MASLD-HCC, hepatocytes exhibited inflammatory activation, mesenchymal remodeling, and metabolic suppression. DEN hepatocytes mirrored inflammatory and lipid-processing signatures, while CCl₄ hepatocytes showed broad metabolic collapse. Human HSCs adopted an immunomodulatory phenotype captured by DEN, whereas CCl₄ retained a fibrotic, myofibroblastic signature. Human macrophages exhibited mesenchymal and lipid-handling traits; DEN macrophages aligned with immunosuppressive signaling, while CCl₄ macrophages resembled metabolically exhausted inflammatory cells. Human ECs showed vascular activation and metabolic engagement. CCl₄ ECs shared inflammatory features but not metabolic ones, while DEN ECs were transcriptionally repressed. These findings highlight that each model captures distinct facets of MASLD-HCC pathophysiology. DEN more closely reflects immunosuppressive stromal activation, while CCl₄ recapitulates fibrosis-associated dysfunction. By integrating cell type–resolved, multi-level analysis across species, this study provides a roadmap for strategic preclinical model selection based on disease stage and biological focus.