Identification of a cholesterol metabolism-related signature based on KIF20A and NDC80 for prognostic stratification and immunotherapy prediction in hepatocellular carcinoma

Abstract

Background The heterogeneity of hepatocellular carcinoma (HCC) complicates diagnosis and treatment, and previous prognostic models have failed to yield satisfactory results. Given that the metabolic reprogramming of cholesterol plays a crucial role in the pathogenic mechanism of hepatocellular carcinoma, this study was conducted as a breakthrough point. Methods Utilizing multi-cohort data from TCGA, ICGC, and GEO, we screened Cholesterol metabolism-related genes (CMRGs) to construct a CMRG-derived prognostic model based on two key genes: KIF20A and NDC80. Predictive performance was evaluated via Kaplan-Meier and ROC analyses. Additionally, the tumor immune microenvironment (TME) was characterized using CIBERSORT and ESTIMATE algorithms. Results The two-gene signature stratified patients into high- and low-risk groups, with high-risk patients showing poorer survival (HR = 2.71, P < 0.001). Elevated risk scores correlated with an immunosuppressive TME enriched in monocytes/macrophages. Exploratory anti-PD-1 analysis suggested higher risk scores were associated with non-response, warranting further validation. Conclusion The integrated nomogram achieved AUC values of 0.747 (95% CI: 0.665-0.830), 0.728 (95% CI: 0.652-0.804), and 0.778 (95% CI: 0.706-0.851) for 1-, 2-, and 3-year predictions. This KIF20A/NDC80-based model is a reliable prognostic tool for HCC; its immunotherapeutic utility remains exploratory and requires prospective validation.