Construction of a prognostic model for hepatocellular carcinoma based on necrosis by sodium overload-related genes and identification of ANKRD13B as a new prognostic marker

Hepatocellular carcinoma (HCC), a prevalent malignant tumor of the digestive tract worldwide, is characterized by poor prognosis and high mortality rates. Necrosis by sodium overload (NECSO) represents a novel form of cell death that has been implicated in various cancer types. However, its functional role in HCC pathogenesis remains poorly understood. We conducted a co-expression analysis of the NECSO-associated gene TRPM4, followed by clustering analysis and weighted gene co-expression network analysis (WGCNA) to identify NECSO-related genes. Through evaluation of 101 distinct machine learning algorithm combinations, we developed prognostic models for HCC, with the optimal model selected based on the highest mean concordance index (C-index) across training and validation cohorts. Patients were stratified into high-risk and low-risk groups according to computed risk scores. Subsequent analyses compared intergroup differences in biological functions, immune microenvironment characteristics, and therapeutic responses to immunotherapy and chemotherapy. To identify pivotal biomarkers, we employed three feature selection methodologies: LASSO, SVM-RFE, and random forest algorithms. The biological significance of the identified core gene ANKRD13B was experimentally validated through in vitro cellular experiments. Using a correlation coefficient (cor) > 0.6, we identified 78 co-expressed genes. Subsequent clustering analysis of HCC samples based on these genes revealed 1,402 NECSO-associated genes. Further WGCNA, differential expression, and prognostic analyses of these genes yielded 31 prognostically genes. Among 101 machine learning combinations, the StepCox[both] combined with GBM algorithm emerged as the optimal prognostic model, achieving the highest mean C-index across training and validation cohorts. Survival analysis confirmed significantly poorer prognosis in the high-risk group. Receiver operating characteristic (ROC) curve analysis demonstrated good predictive performance. Functional enrichment revealed distinct intergroup biological profiles, with the high-risk group and the low-risk group showing enrichment in immune-related pathways, metabolic regulation, and cell death mechanisms. Notably, the high-risk group exhibited enhanced immune activation status and superior response rates to immune checkpoint inhibitors therapy. Correlation analyses established significant associations between model genes/risk scores and cell death genes, including ferroptosis, pyroptosis, cuproptosis, and disulfidptosis. Drug sensitivity analysis identified eight chemotherapeutic agents with heightened sensitivity in high-risk patients: BI.2536, Bleomycin, Cisplatin, Doxorubicin, Epothilone B, Gemcitabine, Mitomycin C, and Paclitaxel. In vitro validation confirmed ANKRD13B promoted the proliferation, invasion and migration of HCC. We established a novel NECSO prognostic model demonstrating good predictive capacity for HCC prognosis and therapeutic responsiveness. This model helps with personalized clinical management.