Triple-negative breast cancer: predictive model of early recurrence based on MRI features

Abstract

To develop an integrated model based on preoperative magnetic resonance imaging (MRI) features for predicting early recurrence in patients with triple-negative breast cancer (TNBC). Women with TNBC who underwent breast MRI and surgery between 2009 and 2019 were evaluated retrospectively. Two breast radiologists reviewed MRI images independently based on the Breast Imaging Reporting and Data System Lexicon (BI-RADS), and classified the breast oedema scores on T2-weighted imaging (WI) as no oedema, peritumoural oedema, prepectoral oedema, or subcutaneous oedema. The relationship between disease-free survival (DFS) and MRI features was analysed by Cox regression, and a nomogram model was generated based on the results. 150 patients with TNBC were included and divided into a training cohort (n=78) and validation cohort (n=72). MRI features including subcutaneous oedema and rim enhancement showed a tendency to worsen DFS in univariate analysis. Multivariate analysis showed that subcutaneous oedema (p=0.049, HR [95% confidence interval {CI}=8.24 [1.01-67.52]) and rim enhancement (p=0.016, HR [95% CI]=4.38 [1.32-14.54]) were independent predictors for DFS. In the nomogram, the areas under the curves (AUCs) of the training cohort was 0.808, and that of the validation cohort was 0.875. The presence of subcutaneous oedema or rim enhancement on preoperative breast MRI was shown to be a good predictor of poor survival outcomes in patients with TNBC.