Identification of CACNB1 protein as an actionable therapeutic target for hepatocellular carcinoma via metabolic dysfunction analysis in liver diseases: An integrated bioinformatics and machine learning approach for precise therapy.

In addition to histological evaluation for nonalcoholic fatty liver disease (NAFLD), a comprehensive analysis of the metabolic landscape is urgently needed to categorize patients into distinct subgroups for precise treatment. In this study, a total of 806 NAFLD and 267 normal liver samples were comprehensively analyzed. Alterations in 114 metabolic pathways were investigated and two distinct metabolic clusters were identified. Single-cell RNA sequencing (scRNA-seq) analysis was utilized to decipher the metabolic activities within the microenvironment of NAFLD-derived liver cirrhosis. A refined fibrosis prediction model was developed using a Gaussian Mixture Model (GMM), demonstrating superior performance in fibrosis discrimination across multiple independent cohorts. Additionally, using The Cancer Genome Atlas (TCGA), CACNB1 protein was identified as a promising therapeutic target for hepatocellular carcinoma (HCC) patients with elevated metabolic dysfunction scores (MBDS). Machine learning algorithms were applied to MBDS-related genes to select an optimal prognostic model for HCC. All the models were trained in an HCC cohort obtained from the Gene Expression Omnibus (GEO), and the best model was validated in two independent HCC datasets: the TCGA-HCC cohort and LIRI-JP cohort. Overall, we provide insights of metabolic molecular subtyping and its potential clinical applicability in risk stratification for NAFLD and HCC individuals.