Establishment and Analysis of an Individualized EMT-Related Gene Signature for the Prognosis of Breast Cancer in Female Patients

Abstract

Background. The current high mortality rate of female breast cancer (BC) patients emphasizes the necessity of identifying powerful and reliable prognostic signatures in BC patients. Epithelial-mesenchymal transition (EMT) was reported to be associated with the development of BC. The purpose of this study was to identify prognostic biomarkers that predict overall survival (OS) in female BC patients by integrating data from TCGA database. Method. We first downloaded the dataset in TCGA and identified gene signatures by overlapping candidate genes. Differential analysis was performed to find differential EMT-related genes. Univariate regression analysis was then performed to identify candidate prognostic variables. We then developed a prognostic model by multivariate analysis to predict OS. Calibration curves, receiver operating characteristics (ROC) curves, $C$ -index, and decision curve analysis (DCA) were used to test the veracity of the prognostic model. Result. In this study, we identified and validated a prognostic model integrating age and six genes (CD44, P3H1, SDC1, COL4A1, TGFβ1, and SERPINE1). $C$ -index values for BC patients were 0.672 (95% CI 0.611–0.732) and 0.692 (95% CI 0.586–0.798) in the training cohort and test set, respectively. The calibration curve and the DCA curve show the good predictive performance of the model. Conclusion. This study offered a robust predictive model for OS prediction in female BC patients and may provide a more accurate treatment strategy and personalized therapy in the future.