A biological network approach for mining target genes in EBV-positive gastric cancer.

4056 Background: Epstein–Barr virus (EBV) discriminates a molecular group of gastric cancer (GC) that seems to have peculiar treatment responses and prognosis. Nevertheless, the gene expression pattern of such tumors needs additional investigations due to the heterogeneity of the tumor microbiome and the small number of studied samples. Biological networks analysis of gene expression patterns, in this complex scenario, might lead to the discovery of putative "hubs" genes, supposed to have important roles in cancer occurrence and consequences. This study aimed to explore the biological value of hub genes in EBV-positive gastric cancer patients. Methods: Fresh tumor samples collected during gastrectomies from individuals with gastric adenocarcinoma (n = 41, 59.0±11.0 years) were included. Gene expression of tumor samples was evaluated in the Illumina sequencing platform. A weighted correlation network analysis (WGCNA) was used to find highly correlated gene patterns. The gene's correlation to EBV status was also explored, including eight EBV positive and 33 EBV negative cases. Gene Ontology (GO) term enrichment was performed to predict the involved biological functions. The expression level of the hubs gene was measured by differential expression analysis. Results: A gene cluster including 636 genes able to discriminate EBV status (ρ = 0.65, p-value = 6e-05) was identified. GO analysis (padj <0.05) showed that this cluster of genes is related to T cell activation, regulation of immune effector process, and response to the biotic stimulus. In this cluster, 54 genes were differentially expressed (log2FoldChange (FC) >1 and padj< 0.05). The top hub genes were LAPTM5, PTPN22, C1QA, CD84, CD53, ADAMDEC1, DPYD, TYROBP, RARRES3, IFI16, CYBB, CMKLR1, PHF21A, GPNMB, and C1QC, according to cluster significance and gene expression level. In positive EBV-GC, all hub genes were overexpressed (FC 1.24 - 2.49) (Table). The average area under the ROC curve (AUC) for hub genes was 0.9 (0.84-0.99). Overexpression of LAPTM5 (p-value = 0.013) and DPYD (p-value =0.0065) and low expression of ADAMDEC1 (p-value = 0.05) were strongly related to poor survival outcome. Conclusions: EBV-positive status was correlated with overexpression of hub genes. Genes found in this investigation have high sensitivity and specificity to discriminate EBV status in GC. Some of them seem to be related to patients' survival, opening an avenue for future scientific explorations in this field. The biological network's approach may be a promising tool for mining target genes for potential clinical applications. [Table: see text]