Single-cell Sequencing and Multiple Machine Learning Algorithms to Identify Key T-cell Differentiation Gene for Progression of NAFLD Cirrhosis to Hepatocellular Carcinoma

Hepatocellular carcinoma (HCC), which is closely associated with chronicinflammation, is the most common liver cancer and primarily involves dysregulated immune responses in the precancerous microenvironment. Currently, most studies have been limited to HCC incidence. However, the immunopathogenic mechanisms underlying precancerous lesions remain unknown. We obtained single-cell sequencing data (GSE136103) from two nonalcoholic fatty liver disease (NAFLD) cirrhosis samples and five healthy samples. Using pseudo-time analysis, we systematically identified five different T-cell differentiation states. Ten machine-learning algorithms were used in 81 combinations to integrate the frameworks and establish the best T-cell differentiation-related prognostic signature in a multi-cohort bulk transcriptome analysis. LDHA was considered a core gene, and the results were validated using multiple external datasets. In addition, we validated LDHA expression using immunohistochemistry and flow cytometry. LDHA is a crucial marker gene in T cells for the progression of NAFLD cirrhosis to HCC.