This study aimed to develop and validate a prognostic nomogram integrating baseline MRI and clinicopathological features to predict disease-free survival (DFS) in breast cancer patients undergoing neoadjuvant chemotherapy (NAC).

A retrospective cohort of 402 invasive breast cancer patients who underwent pre-treatment MRI, NAC, and surgery between January 2014 and December 2021 was analyzed. Patients were randomly assigned to a training group (n = 280) and a validation group (n = 122). Variables were selected via univariate Cox regression and Lasso-Cox analyses, with significant factors used to construct nomogram models. The clinicopathological, baseline MRI and combined models were constructed. Model performance was assessed using the area under the curve (AUC), concordance index (C-index), and calibration curves. A risk score derived from the combined model facilitated stratification into high- and low-risk groups, with log-rank test used for survival comparison.

Key predictors in the clinicopathological model included clinical T stage, pathological complete response (pCR) in primary tumor, pCR in axillary lymph nodes, and lymphovascular invasion. MRI-based predictors included multifocal or multicentric lesions, subcutaneous edema, and ipsilateral suspicious internal mammary lymph nodes. The combined model outperformed the clinicopathological (training C-index = 0.67, validation C-index = 0.754) and baseline MRI models (training C-index = 0.665, validation C-index = 0.605), achieving C-indices of 0.706 and 0.719 in the training and validation groups, respectively. A risk score cut-off of −0.35 effectively stratified patients into high- and low-risk groups.

This combined nomogram integrating clinicopathological and MRI features offers improved predictive accuracy for DFS in breast cancer patients after NAC, enabling enhanced risk stratification and individualized follow-up strategies.

Breast cancer

Neoadjuvant chemotherapy

Magnetic resonance imaging

Disease-free survival

Nomogram

© 2026 The Authors. Published by Elsevier Inc.