An Angiogenic Gene Signature for Prediction of the Prognosis and Therapeutic Responses of Hepatocellular Carcinoma

Abstract

Among cancer-related deaths worldwide, hepatocellular carcinoma (HCC) ranks second. The hypervascular feature of most HCC underlines the importance of angiogenesis in therapy. This study aimed to identify the key genes which could characterize the angiogenic molecular features of HCC and further explore therapeutic targets to improve patients’ prognosis. Public RNAseq and clinical data are from TCGA, ICGC, and GEO. Angiogenesis-associated genes were downloaded from the GeneCards database. Then, we used multi-regression analysis to generate a risk score model. This model was trained on the TCGA cohort (n = 343) and validated on the GEO cohort (n = 242). The predicting therapy in the model was further evaluated by the DEPMAP database. We developed a fourteen-angiogenesis-related gene signature that was distinctly associated with overall survival (OS). Through the nomograms, our signature was proven to possess a better predictive role in HCC prognosis. The patients in higher-risk groups displayed a higher tumor mutation burden (TMB). Interestingly, our model could group subsets of patients with different sensitivities to immune checkpoint inhibitors (ICIs) and Sorafenib. We also predicted that Crizotinib, an anti-angiogenic drug, might be more sensitive to these patients with high-risk scores by the DEPMAP. The inhibitory effect of Crizotinib in human vascular cells was obvious in vitro and in vivo. This work established a novel HCC classification based on the gene expression values of angiogenesis genes. Moreover, we predicted that Crizotinib might be more effective in the high-risk patients in our model.

Keywords

• hepatocellular carcinoma
• bioinformatics
• prognostic prediction model
• angiogenesis effect