Identification of TXNIP, FTCD, and HAGH as Key Genes in a Cancer Stem Cell-driven Prognostic Model for Hepatocellular Carcinoma.

INTRODUCTION

This study developed a prognostic model for hepatocellular carcinoma (HCC) based on cancer stem cell (CSC)-related modules identified by high-dimensional WGCNA (hdWGCNA).

METHODS

The scRNA-seq data were filtered, dimensionally reduced, and clustered for downstream analysis. CSC-related modules were identified by hdWGCNA. Based on the TCGA liver hepatocellular carcinoma (LIHC) data, we developed a prognostic RiskScore model using univariate Cox, LASSO, and stepwise regression analyses and validated in the ICGC-LIRI-JP dataset. Correlations between the model and the tumor immune microenvironment (TME) and drug sensitivity were analyzed.

RESULTS

Nine cell subpopulations, including CSCs, were identified and were significantly enriched in tumors. hdWGCNA identified six CSC-related key modules. The prognostic RiskScore model, developed based on TXNIP, FTCD, and HAGH, exhibited an AUC > 0.6 in both cohorts. A high RiskScore was correlated with an immunosuppressive TME characterized by downregulated neutrophils, NK cells, and eosinophils, and higher sensitivity to chemotherapeutic agents such as Pyrimethamine and Vinorelbine.

DISCUSSION: This study identified CSC-related modules associated with HCC prognosis, TME, and drug response, providing a potential mechanistic basis for the clinical application of the model.

CONCLUSION

Potential prognostic and targeted therapy biomarkers were identified. The CSC-related module-based prognostic model may offer a new tool for personalized HCC treatment.