CDC20 May Serve as a Potential Biomarker-Based Risk Score System in Predicting the Prognosis of Patients with Hepatocellular Carcinoma.

Abstract

Background The specificity and sensitivity of hepatocellular carcinoma (HCC) diagnostic markers are limited, hindering the early diagnosis and treatment of HCC patients. Therefore, improving prognostic biomarkers for patients with HCC is urgently needed. Methods HCC-related datasets were downloaded from the public databases. Differentially expressed genes (DEGs) between HCC and adjacent nontumor liver tissues were then identified. Moreover, the intersection of DEGs in four datasets (GSE138178, GSE77509, GSE84006, and TCGA) was used in the functional enrichment, and module genes were obtained by a coexpression network. Cox and Kaplan-Meier analyses were used to identify overall survival- (OS-) related genes from module genes. Area under the curve (AUC) > 0.9 of OS-related genes was then carried out in order to perform the protein-protein interaction network. The feature genes were identified by least absolute shrinkage and selection operator (LASSO). Furthermore, the hub gene was identified through the univariate Cox model, after which the correlation analysis between the hub gene and pathways was explored. Finally, infiltration in immune cell types in HCC was analyzed. Results A total of 2,227 upregulated genes and 1,501 downregulated DEGs were obtained in all four datasets, which were mainly found to be involved in the cell cycle and retinol metabolism. Accordingly, 998 OS-related genes were screened to construct the LASSO model. Finally, 8 feature genes (BUB1, CCNB1, CCNB2, CCNA2, AURKB, CDC20, OIP5, and TTK) were obtained. CDC20 was shown to serve as a poor prognostic gene in HCC and was mainly involved in the cell cycle. Moreover, a positive correlation was noted between the high degree of infiltration with Th2 and CDC20. Conclusion High expression of CDC20 predicted poor survival, as potential target in the treatment for HCC.