Abstract
Hypoxia and hypoxia-related genes regulate tumor initiation and progression. However, the exact roles that hypoxia plays in hepatocellular carcinoma (HCC) remain unclear. In the present study, we calculated the hypoxia score of each sample in the GSE14520 training set by single-sample gene set enrichment analysis (ssGSEA). Then, weighted gene coexpression network analysis (WGCNA) was utilized to identify gene modules most correlated with hypoxia. Least absolute shrinkage and selection operator (LASSO) Cox regression analysis was utilized to further compress the candidate genes. We constructed the hypoxia-related prognostic risk score (HPRS) model based on the genes' corresponding Cox regression coefficients. Univariate and multivariate Cox analyses of the hypoxia score and clinicopathological characteristics showed that the hypoxia score and stage were the main risk factors affecting the overall survival of patients. Based on WGCNA, we identified 41 key hypoxia-related gene modules and screened out nine core genes to construct the HPRS model. Importantly, high-HPRS patients have a worse prognosis, while low-HPRS patients have a better prognosis. Further research showed that various immune cells, such as CD8 T cells, cytotoxic cells, and DCs, were significantly enriched in the low-HPRS group compared with the high-HPRS group. Notably, patients in the low-HPRS group were less likely to benefit from immunotherapy and chemotherapy than those in the high-HPRS group. In summary, we identified and validated a hypoxia-derived gene model that could serve as a potential biomarker to predict prognosis and therapeutic response in HCC.