Deep sequencing and comprehensive expression analysis identifies several molecules potentially related to human poorly differentiated hepatocellular carcinoma.

Hepatocellular carcinoma (HCC) that is graded histologically as poorly differentiated has a high recurrence, metastasis and poor prognosis. We sought to determine the regulatory mechanisms of HCC tumorigenesis and to identify molecules closely related to poorly differentiated HCC. High-throughput sequencing was used to construct microRNA (miRNA) and mRNA expression profiles for poorly differentiated HCC tissues and adjacent tissues. Network analysis was carried out to study miRNA-target interactions. Integrating the miRNA and mRNA data of HCC with four tumor grades from The Cancer Genome Atlas (TCGA) portal enabled the identification of potential closely related molecules for early diagnosis of poorly differentiated HCC. Electronic validation of RNA-seq data and survival analysis was also performed. In total, 1051 differentially expressed genes and 165 differentially expressed miRNAs were identified between HCC tumor and paired non-tumorous tissue. Based on 3718 miRNA-target interactions, we established an miRNA-target interaction network; the target genes were mainly involved in bile acid biosynthesis and bile secretion. Integrating expression data of HCC from TCGA indicated that two proteins, TM4SF1 and ANXA2, are convincing indicators for initial diagnosis of poorly differentiated HCC. According to the survival analysis, three proteins, ANXA2, C8orf33 and IGF2BP3, were identified as being associated with the survival time of HCC patients. Moreover, we suggest that hsa-miR-1180 may be an effective biomarker for poorly differentiated HCC. Three molecules, TM4SF1, ANXA2 and C8orf33, are potential biomarkers for distinguishing poorly differentiated from well-differentiated HCC.