Abstract
BACKGROUND: M2-like tumor-associated macrophages (TAMs) promote an immunosuppressive microenvironment and contribute to tumor progression and metastasis. However, their molecular characterization and prognostic value have not been fully explored in the field of breast cancer. METHODS: Weighted gene co-expression network analysis (WGCNA) was used to identify modules significantly associated with M2-like TAMs. Consensus clustering analysis identified three molecular subtypes with distinct clinical features, and we explored potential differences in genomic mutations, pathway enrichment, and immune infiltration in patients between subtypes. Machine learning algorithms were used to screen key genes and construct M2-like macrophage-associated prognostic models. Comprehensive transcriptomic analysis and in vitro phenotyping and polarization experiments were performed on the key gene DLG3. RESULTS: M2-like TAMs infiltration was strongly associated with the prognosis of BC patients, and the associated gene characterization revealed three molecular subtypes, of which C2 has the worst prognosis with high M2 macrophages, immune desert phenotype, and immunotherapeutic resistance; C1 had the best prognosis, rich in stromal and immune cell infiltration, and metabolic pathway activation; and C3 had a high level of TILs and genomic mutations, with a high degree of immunogenicity and immunotherapeutic Potential. Risk scores can effectively predict the prognosis and immunotherapy response of BC patients, in which DLG3 is a key gene that may be involved in shaping the immunosuppressive microenvironment of breast cancer, and down-regulation of DLG3 can inhibit M2 polarization of macrophages. DISCUSSION: We constructed and comprehensively solved a model of M2-like TAM-related molecular subtypes and prognosis, which helps stratify and customize treatment regimens for BC patients. We also explored the role of DLG3 in BC progression and macrophage polarization.