Identifying Dendritic Cell-Related Genes Through a Co-Expression Network to Construct a 12-Gene Risk-Scoring Model for Predicting Hepatocellular Carcinoma Prognosis

Chaoyuan Huang; Xiaotao Jiang; Yuancheng Huang; Lina Zhao; Peiwu Li; Fengbin Liu

doi:10.3389/fmolb.2021.636991

Identifying Dendritic Cell-Related Genes Through a Co-Expression Network to Construct a 12-Gene Risk-Scoring Model for Predicting Hepatocellular Carcinoma Prognosis

Front Mol Biosci. 2021 May 24:8:636991. doi: 10.3389/fmolb.2021.636991. eCollection 2021.

Authors

Chaoyuan Huang¹, Xiaotao Jiang¹, Yuancheng Huang¹, Lina Zhao², Peiwu Li², Fengbin Liu²

Affiliations

¹ The First Clinical Medical School, Guangzhou University of Chinese Medicine, Guangzhou, China.
² Department of Gastroenterology, The First Affiliated Hospital of Guangzhou University of Chinese Medicine, Guangzhou, China.

Abstract

The prognostic prediction of hepatocellular carcinoma (HCC) is still challenging. Immune cells play a crucial role in tumor initiation, progression, and drug resistance. However, prognostic value of immune-related genes in HCC remains to be further clarified. In this study, the mRNA expression profiles and corresponding clinical information of HCC patients were downloaded from public databases. Then, we estimated the abundance of immune cells and identified the differentially infiltrated and prognostic immune cells. The weighted gene co-expression network analysis (WGCNA) was performed to identify immune-related genes in TCGA cohort and GEO cohort. The least absolute shrinkage and selection operator (LASSO) Cox regression model was applied to establish a risk-scoring model in the TCGA cohort. HCC patients from the GSE14520 datasets were utilized for risk model validation. Our results found that high level of dendritic cell (DC) infiltration was associated with poor prognosis. Over half of the DC-related genes (58.2%) were robustly differentially expressed between HCC and normal specimens in the TCGA cohort. 17 differentially expressed genes (DEGs) were found to be significantly associated with overall survival (OS) by univariate Cox regression analysis. A 12-gene risk-scoring model was established to evaluate the prognosis of HCC. The high-risk group exhibits significantly lower OS rate of HCC patients than the low-risk group. The risk-scoring model shows benign predictive capacity in both GEO dataset and TCGA dataset. The 12-gene risk-scoring model may independently perform prognostic value for HCC patients. Receiver operating characteristic (ROC) curve analysis of the risk-scoring model in GEO cohort and TCGA cohort performed well in predicting OS. Taken together, the 12-gene risk-scoring model could provide prognostic and potentially predictive information for HCC. SDC3, NCF2, BTN3A3, and WARS were noticed as a novel prognostic factor for HCC.

Keywords: co-expression network construction; hepatocellular carcinoma; immune-related gene; overall survival; risk-scoring model.