A potential EBV-related classifier is associated with the efficacy of immunotherapy in gastric cancer

Abstract

Background: Gastric cancer (GC) is a global health problem. As a complicated heterogeneous disease, The Cancer Genome Atlas (TCGA) Research Network recognized four subtypes of GC, including Epstein-Barr virus (EBV)-positive GC (EBVaGC), which accounts for approximately 9% of all GC cases. The response to immunotherapy in GC is limited, and whether EBV status is a predictor of immunotherapy remains controversial.

Methods: The differential gene expression analysis was utilized to compare the gene expression between the EBV-positive group and the EBV-negative group in the TCGA-Stomach Adenocarcinoma (TCGA-STAD) cohort. Weighted gene co-expression network analysis (WGCNA), protein-protein interaction (PPI) network analysis, and gene functional enrichment analysis were used to investigate the most pivotal hub genes and their roles. The "Estimation of Stromal and Immune cells in Malignant Tumours using Expression data" (ESTIMATE) and CIBERSORT algorithms were performed to infer the immune compositions of tissue samples. Furthermore, quantitative real-time polymerase chain reaction (RT-qPCR) and survival analysis were used to validate the expression and prognosis of these hub genes in Sun Yat-sen University Cancer Center (SYSUCC) cohort. A GC cohort that received anti-programmed cell death 1 (PD-1) therapy was used to analyze the predicted efficacy based on the expression of hub genes.

Results: There is a total of 1,686 differentially expressed genes (DEGs) between the EBV-positive group and EBV-negative group, and WGCNA identified a yellow-colored module that was most related to EBV features. Functional enrichment analysis of 144 genes in this yellow module demonstrated that these genes primarily performed immune-related functions, and PPI network analysis through the CytoHubba plug-in identified 11 hub genes in the network. The RT-qPCR results in the SYSUCC cohort further validated that the hub genes were all increased in the EBV-positive group. Finally, we found that a potential classifier was associated with the efficacy of immunotherapy based on the expression of these 11 hub genes.

Conclusions: Our study identified several hub genes associated with EBV status that are closely related to the immune microenvironmental features of EBVaGC and may be used as molecular markers for predicting the immune response in GC.

Keywords: Epstein-Barr virus (EBV); Gastric cancer (GC); hub genes; immunotherapy.

Figure 1

The flowchart of this study. TCGA-STAD, The Cancer Genome Atlas-Stomach Adenocarcinoma; EBV, Epstein-Barr virus; WGCNA, weighted gene co-expression network analysis; PPI, protein-protein interaction; SYSUCC, Sun Yat-sen University Cancer Center; RT-qPCR, quantitative real-time polymerase chain reaction.

Figure 2

Identification of the DEGs and most related modules between the EBV-positive group and EBV-negative group. (A) Volcano plot for visualizing DEGs between the EBV-positive group and EBV-negative group. The red points represent the upregulated DEGs, and the blue points represent the downregulated DEGs. (B) Topological network analysis for choosing the optimal soft threshold. (C) Dynamic tree cut after module clustering. (D) Heatmap of the correlations between multiple modules and two groups of different EBV statuses. DEG, differentially expressed gene; EBV, Epstein-Barr virus.

Figure 3

Functional enrichment analysis for GO and KEGG. (A) BP; (B) MF; (C) CC; (D) KEGG. BP, biological process; MF, molecular function; CC, cellular component; GO, Gene Ontology; KEGG, Kyoto Encyclopedia of Genes and Genomes; MHC, major histocompatibility complex; CXCR, chemokine receptor; C-C, C-C motif.

Figure 4

Comparisons of the tumor tissue immune microenvironment between the EBV-positive group and EBV-negative group. (A) Immune score, (B) ESTIMATE score, (C) stromal score, and (D) tumor purity calculated by the ESTIMATE algorithm. (E) Comparison of the proportions of 22 different immune cells between the EBV-positive group and the EBV-negative group. *, P<0.05; **, P<0.01; ***, P<0.001; ****, P<0.0001; and ns means not significant. EBV, Epstein-Barr virus; NK, natural killer.

Figure 5

Identification and validation of hub genes in the PPI network. (A) Counts of the top 10 key genes in the protein-protein interaction network analyzed using the 12 algorithms in the CytoHubba plug-in. Red points are hub genes with counts greater than or equal to 5, and blue points are genes with a count less than 5. (B-L) The expression of these 11 hub genes in the SYSUCC cohort. *, P<0.05; ***, P<0.001; and ****, P<0.0001. EBV, Epstein-Barr virus; PPI, protein-protein interaction.

Figure 6

A cluster heatmap of the expression of 11 genes in a cohort of gastric cancer patients receiving immunotherapy. Each row represents a gene, and each column represents a sample. Expression values in the heatmap were standardized by scale parameter. C1, cluster 1; C2, cluster 2; C3, cluster 3; C4, cluster 4. EBV, Epstein-Barr virus; PD, progressive disease; SD, stable disease; PR, partial response; CR, complete response.