Development and validation of a stromal immune phenotype classifier for predicting immune activity and prognosis in triple-negative breast cancer

Abstract

Our study aims to construct a prognosis-related immune phenotype classifier for predicting clinical prognosis and immune activity in triple-negative breast cancer (TNBC). A total of 237 patients with TNBC from Sun Yat-sen University Cancer Center (SYSUCC) and 533 patients with TNBC from public datasets were included in our study. A stromal immune quantified index was generated with a LASSO Cox regression model based on five prognosis-related immune cells evaluated by CIBERSORT or IHC and was used to determine immune phenotypes. Immune features were evaluated in the samples before chemotherapy. A total of 119 patients in the SYSUCC training cohort were classified into immune Phenotypes A and B according to the density of stromal CD4+ T cells, γδ T cells, monocytes, M1 macrophages and M2 macrophages. Phenotype A predicted better survival than Phenotype B, and the classification was further validated in the testing cohort of 118 patients and the validation cohort of 533 patients. In the combined cohort, significant differences were found in Phenotype A compared to Phenotype B for the 5-year overall survival (83.5% vs 65.8%, respectively, P