“…Unsupervised clustering analysis was adopted and found that the expression levels of some immune cells between clusters, such as "Mast cell resting (P < 0.001)", "T cell regulatory (P < 0.01)", "Macrophages M1 (P < 0.01)", "Eosinophils (P < 0.01)", and "T cell CD4 memory resting (P < 0.01)", were significantly different, demonstrating that differences in CAAS could lead to changes in the tumour immune microenvironment. Previous studies have shown that high infiltration of M1 [51] and CD4 T cells, E0 expression [52], mast cell expression [53], and low Treg expression [54] or other states are the best state of hot tumours, which can increase the efficacy of immunotherapy. What's more, the main difference between the three clusters was the infiltration level of innate and acquired immune cells, which was verified by differential analysis of the underlying immunophenotype and immune microenvironment.In addition, we also discovered the heterogeneity of the immune microenvironment in TNBC based on the consensus matrix heatmap, which could to some extent explain the clinical phenomenon that PD-1 / PD-L1 immunotherapy had different effects on TNBC patients.…”