Nomogram for Prediction of Hepatocellular Carcinoma Prognosis

Abstract Background A reliable nomogram for predicting prognosis of Hepatocellular Carcinoma (HCC) is lacking, and the cause of poor outcome still remains unclear. It is urgently to prolong the anticipation of overall survival (OS) time of HCC by exploring new therapeutic targets. Methods Different expression genes were calculated in R packages, hub genes defined as overlapped candidates across five datasets and estimated in Oncomine database. A prognostic nomogram constructed on multivariate Cox analysis and evaluated by receiver operating characteristic curve and concordance index analysis. The fraction of TME cell types were estimated by xCell, a gene signatures enriched method. Hypoxia scores were calculated by single sample gene set enrichment analysis upon the 15 hypoxia signatures. Statistically significance and correlation analyses were processed in R program, co-expression interaction plotted in Cytoscape. Results We screened 11 hub genes from five dataset across multi-platform datasets and covering main risk factors to HCC. Compared with two other prognostic models (AUC of 0.65, 0.65 respectively), we established a prognostic nomogram based on multivariate Cox analysis in the training set (AUC of 0.71). The nomogram performed robustly by evaluating in internal validation set (AUC of 0.72) and external cohorts (GSE144269, AUC of 0.70). With further analyze, we found the risk scores of the prognostic model were positively correlated with the fraction of T helper 2 cell (Th2), a kind of T helper cells, which had been approved to be a promising target for curing cancer. On the contrary, and the proportions of hematopoietic stem cell (HSC) and endothelial cell (EC) decreased in RiskHigh group and acted as good prognostic factors. However, signaling pathways analyses shown that formation and growth of angiogenesis were actively in RiskHigh group, accompanied with the increasing proliferation of EC. Taken together, EC played bidirectional roles in biological process of HCC. In addition, HIF1A related hypoxia status was identified as a feature of HCC providing a novel therapeutic target. Conclusions We constituted a robust prognostic nomogram for predicting OS time of HCC. Upon the model, Th2 and HIF1A related hypoxia signaling pathway may be the novel promising therapeutic targets for prolonging survival time in HCC.