Weighted Gene Co-expression Network Analysis Identifies a Cancer-Associated Fibroblast Signature for Predicting Prognosis and Therapeutic Responses in Gastric Cancer

Abstract

Background: Cancer-associated fibroblasts (CAFs) are the most prominent cellular components in gastric cancer (GC) stroma that contribute to GC progression, treatment resistance, and immunosuppression. This study aimed at exploring stromal CAF-related factors and developing a CAF-related classifier for predicting prognosis and therapeutic effects in GC. Methods: We downloaded mRNA expression and clinical information of 431 GC samples from Gene Expression Omnibus (GEO) and 330 GC samples from The Cancer Genome Atlas (TCGA) databases. CAF infiltrations were quantified by the estimate the proportion of immune and cancer cells (EPIC) method, and stromal scores were calculated via the Estimation of STromal and Immune cells in MAlignant Tumors using Expression data (ESTIMATE) algorithm. Stromal CAF-related genes were identified by weighted gene co-expression network analysis (WGCNA). A CAF risk signature was then developed using the univariate and least absolute shrinkage and selection operator method (LASSO) Cox regression model. We applied the Spearman test to determine the correlation among CAF risk score, CAF markers, and CAF infiltrations (estimated via EPIC, xCell, microenvironment cell populations-counter (MCP-counter), and Tumor Immune Dysfunction and Exclusion (TIDE) algorithms). The TIDE algorithm was further used to assess immunotherapy response. Gene set enrichment analysis (GSEA) was applied to clarify the molecular mechanisms. Results: The 4-gene (COL8A1, SPOCK1, AEBP1, and TIMP2) prognostic CAF model was constructed. GC patients were classified into high– and low–CAF-risk groups in accordance with their median CAF risk score, and patients in the high–CAF-risk group had significant worse prognosis. Spearman correlation analyses revealed the CAF risk score was strongly and positively correlated with stromal and CAF infiltrations, and the four model genes also exhibited positive correlations with CAF markers. Furthermore, TIDE analysis revealed high–CAF-risk patients were less likely to respond to immunotherapy. GSEA revealed that epithelial–mesenchymal transition (EMT), TGF-β signaling, hypoxia, and angiogenesis gene sets were significantly enriched in high–CAF-risk group patients. Conclusion: The present four-gene prognostic CAF signature was not only reliable for predicting prognosis but also competent to estimate clinical immunotherapy response for GC patients, which might provide significant clinical implications for guiding tailored anti-CAF therapy in combination with immunotherapy for GC patients.