The overall survival of hepatocellular carcinoma (HCC) remains poor, highlighting the need for better prognostic tools. Nucleotide metabolism fuels tumor progression, while the immune microenvironment dictates therapy response, but integrated models combining both features are lacking. Using TCGA-LIHC transcriptomic/clinical data, we identified nucleotide metabolism and immune-related differentially expressed genes (NMIRGs), which stratified HCC patients into two subtypes via non-negative matrix factorization. A nine-gene prognostic risk signature was constructed through LASSO/Cox regression and validated using independent GEO datasets, and the NMIRG signature was further validated experimentally via RT-qPCR in HCC cell lines and independently using the HPA database for protein-level evidence. As evaluated by our risk signature, high-risk patients exhibited altered immune profiles (T cells increasing, neutrophils decreasing), elevated tumor mutation burden and microsatellite instability, and worse predicted immunotherapy response. Gene set enrichment analysis linked high-risk genes to immune pathways and low-risk genes to metabolic processes. Our risk signature predicted HCC prognosis independent of demographic features and outperformed existing signatures with superior C-index accuracy, effectively predicting immune microenvironment status and therapy benefits. Together, this integrated NMIRG signature offers enhanced prognostication and identifies promising biomarkers for personalized HCC management.
Wang et al. (Mon,) studied this question.