Ahypoxia-related signature enhances the prediction of the prognosis in hepatocellular carcinoma patients and correlates with sorafenib treatment response

Hepatocellular carcinoma (HCC) is one of the leading cancer death and is the primary malignancy of the liver. Tumor hypoxia is the stressor that is involved in tumorigenesis and significantly increased the aggressiveness of HCC. Here, we systematically analyzed the expression profiles and prognostic values of 84 hypoxia associated genes in HCC. mRNA expression of 84 hypoxia associated genes and clinical parameters of HCC patients were downloaded from TCGA, GSE14520, GSE109211 and ICGC. Consensus clustering analysis was performed for unsupervised classes on the basis of 84 hypoxia associated genes. Univariate and LASSO analysis were used to develop the risk signature. A risk signature was developed, including the expression of APEX1, ATR, CTSA, DNAJC5, ENO1, EPO, HMOX1, LDHA, NDRG1, and PER', and found to be significantly related with OS and DFS of HCC patients. We stratified HCC patients into the high-risk group and low-risk group by means of the risk signature. Patients of high-risk group had shorter OS and DFS, while that of the low-risk group had longer OS and DFS. The risk signature showed better predictive efficiency than the TNM staging in predicting OS and DFS. Also, macrophage MO cells, regulatory T cells, and neutrophils were found to be significantly enriched in patients of high-risk group. Next, we validated the discrimination and prognostic value of the risk signature in GSE14520 and the ICGC HCC cohort. Finally, significantly lower risk scores were found in sorafenib treatment responders of GSE109211 cohort, and the AUC for predicting sorafenib treatment response was 0.881. In conclusion, a risk signature developed with the expression of 10 hypoxia associated genes improved the prognosis prediction of HCC and correlated with sorafenib treatment response.