Gastric adenocarcinoma (GA) a leading cause of cancer-related deaths with particularly high incidence in Asia, Eastern Europe, and Central America. It is a highly heterogeneous malignant disease that is featured with diverse subtypes and clinical behaviors. The cellular features of the tumor environment (TME) play an important role in enabling tumors to proliferate and metastasize. Diverse  and deregulated cellular states of components in the TME such as the immune cells and fibroblasts could influence tumor cell survival as well as response to treatments such as immune checkpoint blockade. In recent years, enormous effort has been made in the assessment of the prognostic value of multiple immune cell subsets. However, our understanding of mechanisms by which TME- resident cell types influence clinical outcomes of gastric cancer remains nascent. In this study, we proposed a data analysis framework to evaluate the prognostic values of cell subtypes. Firstly, we preprocessed the GSE183904 dataset measured by 10x Genomics including 15 cell types annotated by the canonical markers. Then, we collected eight independent bulk transcriptome datasets with survival time including 1,879 patients. Using univariate Cox regression, we obtained genes whose expressions were significantly correlated with the overall survival time for patients in each bulk dataset. For each cell type, these Cox associated genes were used to performed single sample gene set enrichment analysis (ssGSEA) to explore the degrees of different survival outcome. We found several cell types shown strong prognostic values. Favourable prognostic subpopulations included CD8⁺ T, CD4⁺ T and NK cells. Unfavourable prognostic subpopulations included cancer-associated fibroblasts, pericytes and  endothelial cells. Interestingly, both protective and risk related genes were enriched in the tumor-associated macrophages (TAMs) and it seems that TAMs have dual functions in the prognostic level. We further subtyped the TAMs into smaller clusters and annotated them based on the hallmark functional pathways. Four subtypes named pro-tumor TAM, anti-tumor TAM, proliferative TAM and TGF TAM were obtained. Moreover, we found the proliferative TAM has the largest enrichment score in the protective genes and pro-tumor TAM has the largest enrichment score in the risk related genes. These results suggest that the TAMs show heterogeneous prognostic values in GA and might be a potential good predictor of GA prognosis. Next, we used CellChat, which could systematically classify ligand-receptor pairs within cell-cell communication into functionally related signal pathways, to explore cell-cell interactions between cancer cells and tumor microenvironment.We highlighted several cell groups with strong cell-cell interaction signals including Endothelial,CAF and DC cell.It should be emphasized that TGF TAM has a stronger interaction with these cell types than other TAM subgroups.In addition,the  expressed  ligand-receptor pairs  is different across  different TAM subgroups.Results of this research would have theoretical and practical values for identification of prognostic cell groups based on single-cell transcriptome of GA.
 

Gastric adenocarcinoma Single-cell RNA-sequencing tumor-associated macrophages ovarall survival Cell-cell communications