Integrative single-cell atlas unveils heterogeneity and prognostic value of cancer-associated fibroblasts in gastric cancer

Gastric cancer (GC) exhibits molecular heterogeneity and diverse immune cell infiltration patterns closely associated with patient prognosis. However, a comprehensive understanding of the variations in immune cell phenotypes among different patient subgroups still needs to be improved. In this study, we performed a detailed analysis of the tumor microenvironment in GC by integrating 200,466 single cells from 72 patients across six datasets. We classified patients into immune-deserted, B, T, and myeloid cell subtypes. Using genomic and clinical data from TCGA samples, we identified cellular components associated with tumor histology and genotypes. GC patients were stratified into immune-deserted, B cell, T cell, and myeloid cell subtypes, and we described the pathway and transcription factor activity characteristics of different microenvironment subtypes. Integration of bulk RNA-seq data reveals that fibroblasts and endothelial cells were associated with adverse patient outcomes whereas NK and T cells were notably correlated with improved prognosis. Subsequently, we focused on characterizing cancer-associated fibroblasts (CAFs) and discovered that they acquire new functional properties within the tissue microenvironment, providing evidence of CAF plasticity. We constructed a novel four-gene CAF signature including SPARC, EFEMP1, RGS5 and SERPINE1 which may enhance patient stratification and prognostic prediction of GC patients. qPCR analysis revealed that the significant expressions of SPARC, EFEMP1, RGS5 and SERPINE1 were significantly upregulated in gastric cancer tissues compared to the normal tissues. Our study provides insights into the composition of the tumor microenvironment and construction of a four-gene CAF signature associated with clinical prognosis, offering new perspectives for the clinical management of gastric cancer.