Identification of Breast Cancer Subtypes by Integrating Genomic Analysis with the Immune Microenvironment

[Image: see text] Objectives: We aim to identify the breast cancer (BC) subtype clusters and the crucial gene classifier prognostic signatures by integrating genomic analysis with the tumor immune microenvironment (TME). Methods: Data sets of BC were derived from the Cancer Genome Atlas (TCGA), METABRIC, and Gene Expression Omnibus (GEO) databases. Unsupervised consensus clustering was carried out to obtain the subtype clusters of BC patients. Weighted gene coexpression network analysis (WGCNA), least absolute shrinkage and selection operator (LASSO), and univariate and multivariate regression analysis were employed to obtain the gene classifier signatures and their biological functions, which were validated by the BC dataset from the METABRIC database. Additionally, to evaluate the overall survival rates of BC patients, Kaplan–Meier survival analysis was carried out. Moreover, to assess how BC subtype clusters are related to the TME, single-cell analysis was performed. Finally, the drug sensitivity and the immune cell infiltration for different phenotypes of BC patients were also calculated by the CIBERSORT and ESTIMATE algorithms. Results: TCGA–BC samples were divided into three subtype clusters, S1, S2, and S3, among which the prognosis of S2 was poor and that of S1 and S3 were better. Three key pathways and 10 crucial prognostic-related gene signatures are screened. Finally, single-cell analysis suggests that S1 samples have the most types of immune cells, S2 with more sensitivity to tumor treatment drugs are enriched with more neutrophils, and more multilymphoid progenitor cells are involved in subtype cluster S3. Conclusions: Our novelty was to identify the BC subtype clusters and the gene classifier signatures employing a large-amount dataset combined with multiple bioinformatics methods. All of the results provide a basis for clinical precision treatment of BC.