Abstract
Tumor immune microenvironment is associated with tumor progression. However, previous studies have not fully explored the breast cancer (BC) immune microenvironment. All the data analyzed in this study were obtained from the open-access database, including The Cancer Genome Atlas, Gene Expression Omnibus (TCGA), and cBioPortal databases. R software v4.0 and SPSS 13.0 were used to perform all the statistical analysis. Firstly, the clinical and expression profile information of TCGA, GSE20685, GSE20711, GSE48390, GSE58812, and METABRIC cohorts was collected. Then, 53 immune terms were quantified using the single-sample Gene Set Enrichment Analysis (ssGSEA) algorithm. A prognosis model based on HER2_Immune_PCA, IL12_score, IL13_score, IL4_score, and IR7_score was established, which showed great prognosis prediction efficiency in both training group and validation group. A nomogram was then established for a better clinical application. Clinical correlation showed that elderly BC patients might have a higher riskscore. Pathway enrichment analysis showed that the pathway of oxidative phosphorylation, E2F targets, hedgehog signaling, adipogenesis, DNA repair, glycolysis, heme metabolism, and mTORC1 signaling was activated in the high-risk group. Moreover, Tumor Immune Dysfunction and Exclusion and Genomics of Drug Sensitivity in Cancer analysis showed that low-risk patients might be more sensitive to PD-1 therapy, cisplatin, gemcitabine, paclitaxel, and sunitinib. Finally, four genes, XCL1, XCL2, TNFRSF17, and IRF4, were identified for risk group classification. In summary, our signature is a useful tool for the prognosis and prediction of the drug sensitivity of BC.