Construction and validation of a macrophage polarization-related prognostic index to predict the overall survival in patients with early-stage triple-negative breast cancer

BACKGROUND: Triple-negative breast cancer (TNBC) is a highly heterogeneous disease and the current prognostic system cannot meet the clinical need. Interactions between immune responsiveness and tumor cells plays a key role in the progression of TNBC and macrophages are vital component of immune cells. A prognostic model based on macrophages may have great accuracy and clinical utility. METHODS: For model development, we screened early stage (without metastasis) TNBC patients from The Cancer Genome Atlas (TCGA) database. We extracted messenger RNA (mRNA) expression data and clinical data including age, race, tumor size, lymph node status and tumor stage. The follow up time and vital status were also retrieved for overall survival calculation. Cell-type Identification by Estimating Relative Subsets of RNA Transcripts (CIBERSORT) was used to calculate the immune cell composition of each sample. Weighted gene co-expression network analysis (WGCNA) was used to identify M1-like macrophage-related genes. Combining least absolute shrinkage and selection operator (LASSO) with multivariate Cox regression, the M1-like macrophage polarization-related prognostic index (MRPI) was established. We obtained TNBC patients in Gene Expression Omnibus (GEO) database through PAM50 method and retrieved the mRNA expression data and survival data. The Harrell’s concordance index (CI), the area under the receiver operating characteristic (ROC) curves (AUCs) and the calibration curve were used to evaluate the developed model. RESULTS: We obtained 166 early TNBC cases and 113 normal tissue cases for model building, along with 76 samples from GSE58812 cohort for model validation. CIBERSORT analysis suggested obvious infiltration of macrophages, especially M1-like macrophages in early TNBC. Four genes were eventually identified for the construction of MPRI in the training set. The AUCs at 2 years, 3 years, and 5 years in the training cohort were 0.855, 0.881 and 0.893, respectively; and the AUCs at 2 years, 3 years, and 5 years in the validation cohort were 0.887, 0.792 and 0.722, respectively. Calibration curves indicated good predictive ability and high consistency of our model. CONCLUSIONS: MRPI is a promising biomarker for predicting the prognosis of early-stage TNBC, which may indicate personalized treatment and follow-up strategies and thus may improve the prognosis.