MR imaging phenotypes and features associated with pathogenic mutation to predict recurrence or metastasis in breast cancer

Objectives Distant metastasis remains the main cause of death in breast cancer. Breast cancer risk is strongly influenced by pathogenic mutation.This study was designed to develop a multiple-feature model using clinicopathological and imaging characteristics adding pathogenic mutations associated signs to predict recurrence or metastasis in breast cancers in high familial risk women. Methods Genetic testing for breast-related gene mutations was performed in 54 patients with breast cancers. Breast MRI findings were retrospectively evaluated in 64 tumors of the 54 patients. The relationship between pathogenic mutation, clinicopathological and radiologic features was examined. The disease recurrence or metastasis were estimated. Multiple logistic regression analyses were performed to identify independent factors of pathogenic mutation and disease recurrence or metastasis. Based on significant factors from the regression models, a multivariate logistic regression was adopted to establish two models for predicting disease recurrence or metastasis in breast cancer using R software. Results Of the 64 tumors in 54 patients, 17 tumors had pathogenic mutations and 47 tumors had no pathogenic mutations. The clinicopathogenic and imaging features associated with pathogenic mutation included six signs: biologic features ( p = 0.000), nuclear grade ( p = 0.045), breast density ( p = 0.005), MRI lesion type ( p = 0.000), internal enhancement pattern ( p = 0.004), and spiculated margin ( p = 0.049). Necrosis within the tumors was the only feature associated with increased disease recurrence or metastasis ( p = 0.006). The developed modelIincluding clinico-pathologic and imaging factors showed good discrimination in predicting disease recurrence or metastasis. Comprehensive model II, which included parts of modelIand pathogenic mutations significantly associated signs, showed significantly more sensitivity and specificity for predicting disease recurrence or metastasis compared to Model I. Conclusions The incorporation of pathogenic mutations associated imaging and clinicopathological parameters significantly improved the sensitivity and specificity in predicting disease recurrence or metastasis. The constructed multi-feature fusion model may guide the implementation of prophylactic treatment for breast cancers at high familial risk women.