An exosome mRNA-related gene risk model to evaluate the tumor microenvironment and predict prognosis in hepatocellular carcinoma

Abstract

Background: The interplay between exosomes and the tumor microenvironment (TME) remains unclear. We investigated the influence of exosomes on the TME in hepatocellular carcinoma (HCC), focusing on their mRNA expression profile.

Methods: mRNA expression profiles of exosomes were obtained from exoRBase. RNA sequencing data from HCC patients' tumors were acquired from The Cancer Genome Atlas (TCGA) and the International Cancer Genome Consortium (ICGC). An exosome mRNA-related risk score model of prognostic value was established. The patients in the two databases were divided into high- and low-risk groups based on the median risk score value, and used to validate one another. Functional enrichment analysis was performed based on a differential gene prognosis model (DGPM). CIBERSORT was used to assess the abundance of immune cells in the TME. The correlation between the expression levels of immune checkpoint-related genes and DGPM was analyzed alongside the prediction value to drug sensitivity.

Results: A prognostic exosome mRNA-related 4-gene signature (DYNC1H1, PRKDC, CCDC88A, and ADAMTS5) was constructed and validated. A prognostic nomogram had prognostic ability for HCC. The genes for this model are involved in extracellular matrix, extracellular matrix (ECM)-receptor interaction, and the PI3K-Akt signaling pathway. Expression of genes here had a positive correlation with immune cell infiltration in the TME.

Conclusions: Our study results demonstrate that an exosome mRNA-related risk model can be established in HCC, highlighting the functional significance of the molecules in prognosis and risk stratification.

Keywords: Exosome; Extracellular vesicle; Hepatocellular carcinoma; PRKDC; Prognostic signature; Risk score; Tumor immune microenvironment.

Fig. 1

Candidate mRNAs in HCC. A Volcano plot of DEGs between 130 benign and 118 healthy blood samples of exosome. B Volcano plot of DEGs between 112 HCC and 118 healthy blood samples of exosome. C Interaction analysis of downregulated (C) and upregulated (D) DEGs in both compared groups

Fig. 2

Frequently mutated genes in HCC. A The frequently mutated genes in HCC from TCGA cohort were depicted with Oncoplot. B The frequently mutated genes in HCC from ICGC cohort were displayed by waterfall plot. C Venn diagram of genes covered by both TCGA and ICGC cohorts. D Interaction analysis for DEMGs in both compared groups

Fig. 3

Candidate prognostic related DEMGs in the TCGA cohort. A Heatmap of the prognostic related DEMGs between normal and tumour tissues. B Forest plots showing the results of the univariate Cox regression analysis between gene expression and OS. C Landscape of prognostic related DEMGs alteration in HCC. D The correlation network of candidate genes

Fig. 4

Prognostic value of the risk score in the TCGA cohort. A Distribution and median value of the risk scores. B PCA plot and C t-SNE analysis. D Distributions of OS status, OS and risk score. E Kaplan-Meier curves for the OS of patients in the high- and low-risk group. F AUC of time-dependent ROC curves confirmed the prognostic performance of the risk score. G Comparison of TMB between low- and high-risk groups. H Survival analysis based on the TMB. I Survival analysis for 4 groups by combining TMB and DEMG-based risk signature

Fig. 5

Validation of the risk score in the ICGC cohort. A Distribution and median value of the risk scores. B PCA plot and C t-SNE analysis. D Distributions of OS status, OS and risk score. E Kaplan-Meier curves for the OS of patients in the high-risk group and low-risk group. F AUC of time-dependent ROC curves confirmed the prognostic performance of the risk score. G Comparison of TMB between low- and high-risk groups. H Survival analysis based on the TMB. I Survival analysis for 4 groups by combining TMB and DEMG-based risk signature

Fig. 6

Univariate and multivariable Cox regression analyses of risk score. A Univariate and B multiple Cox regression analyses were performed in the TCGA cohort. C Univariate and D multiple Cox regression analyses were performed in the ICGC validation cohort. E Heatmap of clinical parameters for the TCGA cohort

Fig. 7

Functional analysis of the DGPM in TCGA and ICGC cohort. A Bubble graph for GO enrichment and B barplot graph for KEGG pathways in the TCGA cohort. C Bubble graph for GO enrichment and D barplot graph for KEGG pathways in the ICGC cohort

Fig. 8

Immune cell infiltrations of TCGA and ICGC cohorts. Relative proportion of immune cell infiltration in (A) TCGA and (B) ICGC (B). Correlation analysis of immune cells in (C) TCGA and (D) ICGC

Fig. 9.

SsGSEA scores between different risk groups in TCGA and ICGC cohort. A The scores of 16 immune cells and B 13 immune-related functions in TCGA cohort are displayed in boxplots. C The scores of 16 immune cells and D 13 immune-related functions in ICGC cohort are displayed in boxplots. Adjusted P values were showed as: *, p < 0.05; **, p < 0.01; ***, p < 0.001

Fig 10

Correlation of DGPM with clinical features and construction of clinicopathological nomogram. A Correlation of risk score with A tumor grade, B clinical stage, and C T status. D Nomogram was constructed by grade, stage and risk signature for predicting survival. E AUCs for predicting 1-, 3-, and 5-year survival. F The 1-, 3-, and 5-year nomogram calibration curves

Fig. 11

The clinical significance of PRKDC in HCC. A PRKDC are overexpressed in HCC tumor tissue. B Four analyses from ONCOMINE platform showing high expression of PRKDC. C-D Protein expression level of PRKDC was shown by The Human Protein Altas by immunohistochemistry. E Correlation of risk score with tumor grade. F The TIMER shows that the clinical outcome increased risk with the increase of PRKDC gene expression. G Lower PRKDC level predicts longer OS

Fig. 12

Association between PRKDC and immune checkpoint genes. A Correlation analysis between immune checkpoints CD274, PDCD1, PDCD1LG2, CTLA4, HAVCR2, and IDO1 and risk score. B Comparison of the expression levels of ICB-related genes between low- and high-PRKDC groups. Association between PRKDC and C CD274, D CTLA4, E HAVCR2, F IDO1, G PDCD1LG2, and H PDCD1

Fig. 13

The role of PRKDC in TME features. A Comparison of (A1) immune score, (A2) ESTIMATE score and (A3) tumor purity between low- and high-PRKDC groups. B TIMER showed that low PRKDC was associated with good OS. Copy number of C CD4+ T-cells and D CD8+ T-cells in HCC. E Relationship between PRKDC with (E1) CD4+ T-cells, (E2) neutrophils, (E3) macrophages and (E4) myeloid dendritic cells. F Comparison of ssGSEA enrichment between low- and high-PRKDC groups.