Mining featured biomarkers associated with vascular invasion in HCC by bioinformatics analysis with TCGA RNA sequencing data.

This study aims to identify the feature genes associated with vascular invasion in hepatocellular carcinoma (HCC). Here, the RNA sequencing data related to vascular invasion in The Cancer Genome Atlas (TCGA) database, including 292 HCC patients with complete clinical data were included in our study as the training dataset for construction and E-TABM-36, including 41 HCC patients with complete clinical data was used as the validation dataset. Following data normalization, differentially expressed mRNA and copy number (CN) were selected between with and without vascular invasion samples. A support vector machine (SVM) classifier was constructed and validated in GSE9828 and GSE20017 datasets. Total 59 feature genes were found by the SVM classifier. Using Cox regression analysis, three clinical features, including Patholigic T, Stage and vascular invasion and 6 optimal prognostic genes, including ANO1, EPHX2, GFRA1, OLFM2, SERPINA10 and TKT were significantly correlated with prognosis. A risk score formula was developed to assess the prognostic value of 6 optimal prognostic genes, which were identified to possess the most remarkable correlation with overall survival in HCC patients. By performing in vitro experiments, we observed TKT was significantly increased, but OLFM2 was decreased in high metastatic potential HCC cell lines (SK-HEP-1 and MHCC-97 H) compared with low metastatic potential cell line Huh7 and normal human liver cell line LO2 using western blotting analysis. Knockdown of TKT in MHCC-97H or overexpression of OLFM2 in SK-HEP-1 significantly suppressed cell migration and invasion using transwell assays. Our results demonstrated that TKT and OLFM2 might be novel independent biomarkers for predicting survival based on the presence of vascular invasion in patients with HCC.