Identification of Prognostic Factors in Cholangiocarcinoma Based on Integrated ceRNA Network Analysis.

Abstract

This study is aimed at screening prognostic biomarkers in cholangiocarcinoma (CHOL) based on competitive endogenous RNA (ceRNA) regulatory network analysis. Microarray data for lncRNAs, mRNA, and miRNAs were downloaded from the GEO and TCGA databases. Differentially expressed RNAs (DERs) were identified in CHOL and normal liver tissue samples. WGCNA was used to identify disease-related gene modules. By integrating the information from the starBase and DIANA-LncBasev2 databases, we constructed a ceRNA network. Survival analysis was performed, and a prognostic gene-based prognostic score (PS) model was generated. The correlation between gene expression and immune cell infiltration or immune-related feature genes was analyzed using TIMER. Finally, real-time quantitative PCR (RT-qPCR) was used to verify the expression of the 10 DERs with independent prognosis. A large cohort of DERs was identified in the CHOL and control samples. The ceRNA network consisted of 6 lncRNAs, 2 miRNAs, 90 mRNAs, and 98 nodes. Ten genes were identified as prognosis-related genes, and a ten-gene signature PS model was constructed, which exhibited a good prognosis predictive ability for risk assessment of CHOL patients (AUC value = 0.975). Four genes, ELF4, AGXT, ABCG2, and LDHD, were associated with immune cell infiltration and closely correlated with immune-related feature genes (CD14, CD163, CD33, etc.) in CHOL. Additionally, the consistency rate of the RT-qPCR results and bioinformatics analysis was 80%, implying a relatively high reliability of the bioinformatic analysis results. Our findings suggest that the ten-signature gene PS model has significant prognostic predictive value for patients with CHOL. These four immune-related DERs are involved in the progression of CHOL and may be useful prognostic biomarkers for CHOLs.