Predicting Breast Cancer Recurrence Following Breast-Conserving Therapy: A Single-Institution Analysis Consisting of 764 Chinese Breast Cancer Cases

Abstract

The purpose of this study was to identify prognostic factors related to recurrence in women treated with breast-conserving surgery (BCS) and to predict the recurrence following breast-conserving therapy (BCT) by constructing a prediction model. The retrospective analysis included 764 consecutive invasive breast cancer patients treated with BCT in Shanghai Cancer Center between 1995 and 2008. Univariate and multivariate analysis were performed to identify independent risk factors for locoregional recurrence (LRR) and all the recurrence events. Logistic regression was used to construct a recurrence prediction model, which was further evaluated by receiver operating characteristics (ROC) curves. The 5-year locoregional recurrence-free survival (LRRFS) and recurrence-free survival (RFS) rates were 90.8 and 88.4%, respectively. Multivariate analysis revealed 1 independent predictive factor for LRRFS (lymph node, P=.0049) and three independent predictive factors for RFS (lymph node, P=.0036; molecular subtype, P=.0021; histological grade, P=.041). These three variables entered into logistic regression to establish a recurrent prediction model. ROC curve showed that the area under the curve (AUC) of the established model was 0.70 (95% confidence interval: 0.61-0.78). This model could classify patients into "high-risk recurrence" and "low-risk recurrence" groups and could successfully predict their prognosis (P<.00001). The information of lymph node status, molecular subtype, and grade may help doctors to evaluate recurrence risk of a woman treated with BCT. Our new model might be helpful in clinical practice for recurrence prediction after BCT in Chinese patients, though further validation studies are needed.