ScRNA-seq based copy number variations (CNVs) detection identified TPSB2 and TPSAB1 as prognostic biomarkers in Gastric Cancer(GC).

Abstract

e16021 Background: Gastric Cancer (GC) is the third leading cause of cancer-related mortality around the world, with a relatively poor prognosis. Single-cell RNA-sequencing (scRNA-seq) is a powerful tool to analyze tumor microenvironment at a comprehensive resolution, revealing previously uncharacterized molecular complexities. CNVs detection based on scRNA-seq data can reveal genome heterogeneity, therefore identify prognostic biomarkers at the single cell level. Methods: ScRNA-seq data from publicly available dataset (Anuja S et.al, CCR 2020) with 3 gastric cancer tissues, and 3 paired adjacent normal tissues was used. We inferred CNVsusing non-malignant cells as reference to estimate the mutations of malignant cells based on InferCNV package. Survival analysis was made by Kaplan-Meier method on TCGA-STAD (Stomach adenocarcinoma) cohort to identify the CNV genes with prognostic ability. To further explore these genes, their correlations with CD8+T cells marker ( CD8A) and malignant marker ( EPCAM) were analyzed based on the Spearman Correlation Method. Results: From a total of 11,459 cells, comprising 3 gastric cancer tissues (7,634 cells), and 3 paired adjacent normal tissues (3,825 cells), we identified 23 clusters that belonged to different types of cells including Endothelial, Epithelial, Fibroblast, B cells, Myeloid cells, T cells and NK cells. InferCNV analysis predicted 5 CNV regions and 209 CNV genes according to the heterogeneity between malignant and non-malignant cells. Survival analysis based on RNA-seq data from TCGA-STAD cohort, enabled us to narrow down the list of candidate biomarkers to 22 genes with prognostic ability. Among them, the specific functions of TPSB2 and TPSAB1 in GC, especially in their prognosis value, remain unclear. Downregulations of these 2 genes were both associated with longer progression-free survival (PFS). More specifically, expression profiles of gene TPSB2 and TPSAB1 were positively correlated with Epithelial marker genes (p = 0.02 and 0.03 respectively) but negatively correlated with CD8+ T lymphocytes (p = 0.02 and 0.11 respectively). Conclusions: Our analysis suggested that the high expressions of gene TPSB2 and TPSAB1 are associated with poor prognosis in GC patients. And these two genes are correlated with low T lymphocyte infiltration and an immune-suppressive environment, which could be possible prognosis biomarkers to identify patients with a more aggressive tumor evolution.