A Nomogram To Predict Breast Cancer Probability in Patients with Nipple Discharge.

Abstract

Abstract Background: Nipple discharge is a symptoms of breast disease that lead to the diagnosis of breast cancer in 5% of cases. Our objective was to develop a scoring system based on clinical and radiological factors to predict the risk of breast cancer (BC).Methods: Predictive factors of breast cancer were identified in a population of 90 women operated for unilateral nipple discharge, at Institut Beauregard, in France. A scoring system that gives individual probability of BC was developed based on univariate and multivariate logistic regression models. This model was validated in an external independent data set of 263 patients treated at Institut Gustave Roussy, in France.Results: The frequency of BC was 50% and 19.1%, respectively in the training set and in the validation. set. Factors associated with BC were age (p=0.04), presence of a clinical tumor (p=0,03), presence of a nodular lesion (p=0,03) or microcalcifications (p= 0.006) at mammography, in univariate analysis. In multivariate analysis. Only age (p= 0.03), microcalcifications (p= 0,007), or nodular lesion (p=0.06) at mammography were significance or tend to reach significance in multivariate analysis. Based on AIC, factors entered in the model were: age, clinical mass, abnormal images at mammography. The model has a good auc (76.6%). In the validation set, the auc was lower (66.6%).Conclusion: A clinico-radiological model can be built to predict BC in women with nipple discharge to help the decision-making and plan the suitable surgery. A further validation and testing of biological parameters might improve our model. Citation Information: Cancer Res 2009;69(24 Suppl):Abstract nr 5031.