Construction and validation of an immune-related LncRNA prognostic model for hepatocellular carcinoma.

Abstract

Herein, based on mRNA data from TCGA database, hepatocellular carcinoma (HCC) samples were subjected to a single-sample Gene Set Enrichment Analysis (ssGSEA). Then, HCC samples were finally classified into high-, middle-, and low-immunity groups using K-means consensus clustering (K=3) according to ssGSEA scores. After the tumor microenvironment of HCC patients was further analyzed using ESTIMATE algorithm, the results indicated high immune score, stromal score, ESTIMATE score and low tumor purity in high-immunity group. HLA family genes and PD-L1(CD274) were remarkably highly expressed in high-immunity group. Immune-related lncRNAs were required by analyzing differentially expressed genes in high- and low-immunity groups. Differential expression analysis was undertaken on HCC samples, with normal samples as the control. After immune-related lncRNAs and differentially expressed lncRNAs were intersected, 321 differentially expressed immune-related lncRNAs were acquired. Later, the prognostic model based on immune-related lncRNAs was obtained following the Cox regression analysis of previous samples. According to the riskScore, the samples in TCGA-LIHC were divided into high- and low-risk groups. Kaplan-Meier survival analysis, ROC curve, and independence analysis confirmed that the immune-related lncRNAs prognostic model was an important factor independent from clinical characteristics. We further analyzed the difference in immune microenvironment and mutational landscapes in both risk groups. Prominent differences were shown in multiple immunity-related gene sets and immune cells in both groups. The mutation rate of TP53 in high-risk group was much higher than the low-risk one. All these conclusions offered references to prognostic evaluations and personalized treatments for patients with HCC.