Abstract
For datasets of gastric cancer collected by TCGA (The Cancer Genome Atlas) and GEO (Gene Expression Omnibus) repositories, we applied a bioinformatics approach to obtain expression data for the ISLR (immunoglobulin superfamily containing leucine-rich repeat) gene, which is highly expressed in gastric cancer tissues and closely associated with clinical prognosis. Although we did not observe an overall association of ISLR mutation, high expression or copy number variation with survival, hypomethylation of four methylated sites (assessed by the probes cg05195566, cg17258195, cg09664357, and cg07297039) of ISLR was negatively correlated with high expression levels of ISLR and was associated with poor clinical prognosis. In addition, we detected a correlation between ISLR expression and the infiltration levels of several immune cells, especially CD8(+) T cells, macrophages and dendritic cells. We also identified a series of genes that were positively and negatively correlated with ISLR expression based on the TCGA-STAD, GSE13861, and GSE29272 datasets. Principal component analysis and random forest analysis were employed to further screen for six hub genes, including ISLR, COL1A2, CDH11, SPARC, COL3A1, and COL1A1, which exhibited a good ability to differentiate between tumor and normal samples. GO (Gene Ontology) and KEGG (Kyoto Encyclopedia of Genes and Genomes) pathway and gene set enrichment analysis data also suggested a potential relationship between ISLR gene expression and epithelial-mesenchymal transition (EMT). ISLR expression was negatively correlated with sensitivity to PX-12 and NSC632839. Taken together, these results show that the ISLR gene is involved in gastric carcinogenesis, and the underlying molecular mechanisms may include DNA methylation, EMT, and immune cell infiltration.