Transcriptional and microbial profile of gastric cancer patients infected with Epstein-Barr virus

Abstract

Introduction: Gastric cancer (GC), which has low survival rates and high mortality, is a major concern, particularly in Asia and South America, with over one million annual cases. Epstein-Barr virus (EBV) is recognized as a carcinogen that may trigger gastric carcinogenesis by infecting the stomach epithelium via reactivated B cells, with growing evidence linking it to GC. This study investigates the transcriptional and microbial profiles of EBV-infected versus EBV-non-infected GC patients.

Methods: Using Illumina NextSeq, cDNA libraries were sequenced, and reads were aligned to the human genome and analyzed with DESeq2. Kegg and differential analyses revealed key genes and pathways. Gene sensitivity and specificity were assessed using ROC curves (p < 0.05, AUC > 0.8). Non-aligned reads were used for microbiome analysis with Kraken2 for bacterial identification. Microbial analysis included LDA score, Alpha and Beta diversity metrics, with significance set at p ≤ 0.05. Spearman's correlation between differentially expressed genes (DEGs) and bacteria were also examined.

Results: The data revealed a gene expression pattern in EBV-positive gastric cancer, highlighting immune response, inflammation, and cell proliferation genes (e.g., GBP4, ICAM1, IL32, TNFSF10). ROC analysis identified genes with high specificity and sensitivity for discriminating EBV+ gastric cancer, including GBP5, CMKLR1, GM2A and CXCL11 that play pivotal roles in immune response, inflammation, and cancer. Functional enrichment pointed to cytokine-cytokine receptor interactions, antigen processing, and Th17 immune response, emphasizing the role of the tumor microenvironment, shaped by inflammation and immunomodulation, in EBV-associated GC. Microbial analysis revealed changes in the gastric microbiota in EBV+ samples, with a significant reduction in bacterial taxa. The genera Choristoneura and Bartonella were more abundant in EBV+ GC, while more abundant bacteria in EBV- GC included Citrobacter, Acidithiobacillus and Biochmannia. Spearman's correlation showed a strong link between DE bacterial genera and DEGs involved in processes like cell differentiation, cytokine production, digestion, and cell death.

Conclusion: These findings suggest a complex interaction between the host (EBV+ GC) and the microbiota, possibly influencing cancer progression, and offering potential therapeutic targets such as microbiota modulation or gene regulation. Comparing with EBV- samples further highlights the specific impact of EBV and the microbiota on gastric cancer pathogenesis.

Keywords: EBV; carcinogenesis; gastric cancer; gastric microbiome; metatranscriptomics.

Figure 1

Volcano plot of differentially expressed genes in CG EBV+ versus CG EBV- samples.

Figure 2

Heatmap of differentially expressed genes, selected by ROC curve analysis, showing EBV-positive and EBV-negative statuses.

Figure 3

Genes with the highest AUC values, representing the optimal trade-off between sensitivity and specificity in distinguishing EBV+ from EBV- samples.

Figure 4

Kegg analysis of differentially expressed genes in EBV+ GC.

Figure 5

Differentially abundant genera in EBV+ and EBV- GC samples.

Figure 6

Heatmap of differentially abundant bacterial genera between EBV-positive and EBV-negative statuses.

Figure 7

Boxplot comparing Linear Discriminant Analysis (LDA) scores between EBV-negative and EBV-positive samples. EBV-positive samples have higher LDA scores compared to the EBV-negative samples (p. value < 0.001), suggesting a clear separation between the two conditions. Alpha diversity analysis shows significant differences in the diversity and uniformity of bacterial communities between the groups, based on the Shannon (p-value = 0.038) and Simpson (p-value = 0.024) indexes. The Shannon index indicates that EBV+ patients have a more balanced distribution between species, reflecting less dominance of some species, while the Simpson analysis suggests lower bacterial diversity in this group. Beta diversity analysis demonstrate that the overall composition of the bacterial community (PCoA plot; PERMANOVA, p-value = 0.141) did not exhibit statistically significant separation between the groups, suggesting that both share a substantial overlap in taxonomic composition.

Figure 8

Correlation plot between differentially abundant bacteria and differentially expressed genes. (ρ > |0.3|) with at least 10% of DE bacteria.

Figure 9

Gene Ontology of DE genes correlated to DE bacteria genera.