Machine Learning-Based Identification of Core Regulatory Genes in Hepatocellular Carcinoma: Insights from Lactylation Modification and Liver Regeneration-Related Genes

Rongchun Xing^1,2Yu Yang^1,2Yiwei Hou^1,3Li Yi⁴Chongyuan Chen^1,2Xiang Li⁴Yashan Wang⁴Yinxi Fu⁴Mingzheng Hu^1,2*

• ¹The First College of Clinical Medical Science， China Three Gorges University, Yichang, China
• ²Department of Hepatobiliary Surgery, Yichang Central People’s Hospital, Yichang, China
• ³Department of Endocrinology, Yichang Central People’s Hospital, Yichang, China
• ⁴Medical Technology College of Qiqihar Medical College, Qiqihar, China

Hepatocellular carcinoma (HCC) progression shares metabolic-epigenetic features with physiological liver regeneration, yet the regulatory interplay remains poorly defined. Here, we integrate lactylation modification profiles with liver regeneration-associated transcriptomes and employ machine learning to prioritize core regulatory genes driving HCC pathogenesis. Multi-omics analysis of three murine liver regeneration datasets (GSE20426, GSE70593, GSE4528) revealed 793 differentially expressed genes (470 upregulated, 323 downregulated), with 18 overlapping lactylation-related candidates. Machine learning (LASSO regression and SVM-RFE) prioritised six core genes (Ccna2, Csrp2, Ilf2, Kif2c, Racgap1, Vars) enriched in cell cycle regulation and DNA repair pathways. Functional characterisation demonstrated their strong correlation with immune microenvironment remodelling, particularly CD8⁺ T cells and M1 macrophages. Prognostic validation in HCC cohorts revealed significant overexpression of these genes in tumours versus normal tissues (p < 0.05), with elevated Kif2c and Ccna2 predicting poor survival. Crucially, Csrp2 exhibited superior diagnostic efficacy (AUC > 0.8) compared to conventional biomarkers. Experimental validation via qPCR confirmed marked upregulation of all six genes in clinical HCC specimens (p < 0.0001). Furthermore, Western blot demonstrated significantly elevated protein levels of CCNA2, CSRP2, ILF2, KIF2C, RACGAP1, and VARS in HCC tissues compared to adjacent non-tumor controls, corroborating their tumor-specific overexpression. Previous studies have explored lactylation in HCC pathogenesis, while this work uniquely establishes lactylation as a metabolic-epigenetic nexus linking physiological regenerative pathways to oncogenesis—by leveraging liver regeneration models and machine learning-based gene prioritization—and proposes the identified gene panel as dual-purpose biomarkers for HCC diagnosis and therapeutic targeting.