Abstract
BACKGROUND: Gastric cancer (GC) is known for its aggressive nature and considerable genomic diversity. M2-like tumor-associated macrophage (TAM) infiltration is closely associated with cancer progression and prognosis. Therefore, in this study, a prognostic model for patients with GC derived from M2-like TAM infiltration-related genes (M2RGs) was developed, and the relationship between the model and prognosis was investigated. METHODS: The RNA sequencing (RNA-seq) data were sourced from two publicly available databases: The Cancer Genome Atlas (TCGA) and Gene Expression Omnibus (GEO). M2RGs were obtained by CIBERSORT, weighted gene coexpression network analysis (WGCNA) was used to analyze the TCGA-stomach adenocarcinoma (STAD) messenger RNA (mRNA) sequencing data, and the most relevant module genes of M2-like TAMs were selected. The construction of the M2-like TAM infiltration-related prognostic model involved univariate Cox, least absolute shrinkage and selection operator (LASSO) Cox, and multivariate Cox regression analyses. A comprehensive nomogram that integrated the model with clinical variables was subsequently constructed in R using the "rms" package. To further investigate the biological relevance of the model, the immune infiltration landscape of the tumor microenvironment was analyzed, providing additional insight into the prognostic value of M2RGs. Immunohistochemistry analyses verified the expression of M2RGs in GC. RESULTS: Kaplan-Meier survival analysis revealed that patients with GC with low M2-like TAM content had longer survival. Analysis revealed eight M2RGs that were associated with GC prognosis. Four M2RGs (NPC2, AKR1B1, ENTPD2, and SNCG) were constructed and validated as prognostic models for the overall survival (OS) of patients with GC. A higher risk score was significantly correlated with poorer OS in both the training and validation cohorts. Correlation analysis revealed that different risk groups had different clinicopathological features, immune cell infiltration levels, and immune checkpoint inhibitor targets. Immunohistochemical staining analysis validated the reliability of AKR1B1, SNCG, NPC2, and ENTPD2 as prognostic biomarkers for GC. CONCLUSIONS: The model developed in this study can guide GC prognosis and provide a theoretical basis for GC survival research. Additionally, risk stratification based on this model facilitates more personalized management of GC.