Abstract
BACKGROUND: Breast cancer (BRCA) is a prevalent and aggressive disease. Despite various treatments being applied, a significant number of patients continue to experience unfavorable prognoses. Accurate prognosis prediction in BRCA is crucial for tailoring individualized treatment plans and improving patient outcomes. Recent studies have highlighted the significance of immune cell infiltration in the tumor microenvironment (TME), but predicting survival remains challenging due to the heterogeneity of BRCA. The aim of this study was thus to produce an immune cell signature-based framework capable of predicting the prognosis of patients with BRCA. METHODS: The GSE169246 dataset was from the Gene Expression Omnibus (GEO) database, comprising single-cell RNA sequencing (scRNA-seq) data from 95 individuals with BRCA. Seurat, principal component analysis (PCA), the unified matrix polynomial approach (UMAP) algorithm, and linear dimensionality reduction were used to determine the heterogeneity of T cells. Overlapping analysis of differentially expressed genes (DEGs), genes associated with prognosis, and T-cell pharmacodynamics-related genes were used to obtain the T-cell core pharmacodynamics-related genes. The dimensionality of the T-cell core pharmacodynamics-related genes was reduced employing the least absolute shrinkage and selection operator (LASSO) Cox regression model and the LASSO model. The prognostic model was built via a Cox analysis of the overall survival (OS) information. The clinical sample included 95 patients with BRCA who underwent surgical treatment from October 2018 to October 2021 at the Second Affiliated Hospital of Qiqihar Medical University. Patients were divided into a good prognosis group and a poor prognosis group based on their prognostic outcomes. The predictive value of tumor characteristics and immune responses was validated through correlation analysis, logistic regression analysis, and receiver operating characteristic (ROC) analysis. RESULTS: A group of 95 genes was used to establish a prognostic model. In the GEO clinical sample, with a high-risk group demonstrating shorter median survival times (2,447 vs. 6,498 days, P=4.733e-12). Area under the curve (AUC) values of 0.75, 0.75, and 0.72 were obtained for 2-, 4-, and 6-year OS predictions, respectively. Clinical validation found that the 6-year OS of the favorable prognosis group was significantly higher than that of the unfavorable prognosis group (92.06% vs. 65.62%; P=0.005). Poor prognosis was positively correlated with age, tumor size, B-cell level, and CTLA4 level and negatively correlated with tumor stage (T1/T2), lymph node metastasis stage (N0), clinical stage I-II, CD3(+)T-cell, CD4(+)T-cell, CD8(+)T-cell, neutrophil, lymphocyte, natural kill cell, TIGIT expression and OS. The combined model of clinical parameters had an AUC value of 0.898. CONCLUSIONS: This study established a prognostic model that demonstrated excellent predictive value for OS of BRCA. The predictive model developed offers valuable insights into prognosis and treatment planning, emphasizing the importance of tumor characteristics and immune cell infiltration.