Prediction of Poor Prognosis of HCC by Early Warning Model for Co-Expression of miRNA and mRNA Based on Bioinformatics Analysis

Objective: Hepatocellular Carcinoma (HCC) has the highest mortality rate worldwide with the intractability of its extremely complicated pathogenesis and unclear mechanism. The limited survival highlights the need for the further detection of prognosis for HCC. MicroRNAs (miRNAs) and messenger RNAs (mRNAs) have been identified as regulatory factors and target genes in human cancers, while some studies also found post-transcriptional modification plays a crucial role in the occurrence and development of HCC. The present study aimed to elucidate the prognostic significance of miRNA and mRNA models in HCC. Methods: Data were obtained from The Cancer Genome Atlas (TCGA), International Cancer Genome Consortium (ICGC), and Gene Expression Omnibus (GEO) databases. The miRNA and mRNA expressions were tested by the Wilcoxon and used funrich software to predict mRNA that might be related to miRNA. Then we determined the intersection with overlapped mRNA and miRNA Venn diagram, and screened out hub gene by using Degree algorithm in Cytoscape software. The COX models, with TCGA data as the training set and ICGC data as the test set, were constructed. All patients were divided into high-risk and low-risk groups. Data on overall survival of different groups were collected and analyzed by Kaplan-Meier method, and independent risk factors affecting prognosis were assessed by Cox analysis. Results: The miRNA and mRNA polygenic risk model showed a good true positive rate. Kaplan-Meier curve and Cox analysis suggested that the high-risk group was associated with poor prognosis, and the risk score could be used as an independent risk factor for HCC. Conclusion: Tumor risk models constructed in this study could effectively predict the prognosis of patients, which is expected to provide a reference for the prognostic stratification and treatment strategy development of HCC.