Tumor immune microenvironment-based clusters in predicting prognosis and guiding immunotherapy in breast cancer

The development and progression of breast cancer (BC) depend heavily on the tumor microenvironment (TME), especially tumor infiltration leukocytes (TILs). TME-based classifications in BC remain largely unknown and need to be clarified. Using the bioinformatic analysis, we attempted to construct a prognostic nomogram based on clinical features and TME-related differentially expressed genes (DEGs). We also tried to investigate the association between the prognostic nomogram and clinical characteristics, TILs, possible signaling pathways, and response to immunotherapy in BC patients. DEGs for BC patients were identified from The Cancer Genome Atlas Breast Invasive Carcinoma database. TME-related genes were downloaded from the Immunology Database and Analysis Portal. After intersecting DEGs and TME-related genes, 3985 overlapping TME-related DEGs were selected for non-negative matrix factorization clustering, microenvironment cell populations-counter (MCP-counter), LASSO Cox regression, tumor immune dysfunction, and exclusion (TIDE) algorithm analyses. BC patients were divided into three clusters based on the TME-related DEGs and survival data, in which cluster 3 had the best overall survival (OS). Of note, cluster 3 exhibited the highest infiltration or lowest infiltration of CD3+ T-cells, CD8+ T-cells, cytotoxic lymphocytes, B-lymphocytes, monocytic lineage, and myeloid dendritic cells (MDCs). A total of 33 TME-related DEGs were identified as a prognostic gene signature by the LASSO regression analysis. The prognostic gene signature separated BC patients into low- and high-risk groups with significant differences in OS (p<0.01) and demonstrated powerful effectiveness (TCGA all group: 1-year area under the curve [AUC] = 0.773, 3-year AUC = 0.770, 5-year AUC = 0.792). By integrating demographic features, tumor-node metastasis (TNM) stages, and prognostic gene signature, we constructed a nomogram with better predictive value than other clinical features alone. TME-related DEGs in the low-risk BC patients (with better OS) were enriched in chemokine, cytokine–cytokine receptor interaction, and JAK-STAT and Toll-like receptor signaling pathways. BC patients in the low-risk group exhibited higher TIDE scores associated with worse immune checkpoint blockade response. A prognostic nomogram based on TME-related DEGs and clinical characteristics could predict prognosis and guide immunotherapy in BC patients.