Integrating clinical-pathological-MRI features to construct a prediction model for pathological complete remission of axillary lymph nodes after neoadjuvant therapy: a retrospective study.

BACKGROUND

Accurate assessment of axillary lymph node (ALN) metastasis is essential for developing an effective treatment strategy for breast cancer (BC). Despite advancements in imaging and surgical techniques, a critical need remains for reliable, non-invasive methods to predict axillary response to neoadjuvant therapy (NAT). This study aimed to identify key factors influencing axillary lymph node pathological complete response (pCR) following NAT and to develop a predictive model for axillary pCR (apCR) to support clinical decision-making regarding the necessity of axillary lymph node dissection (ALND).

MATERIALS AND METHODS

Clinical data from female patients diagnosed with breast cancer (BC) between January 2019 and December 2024 were retrospectively collected. All patients had biopsy-confirmed metastasis to ipsilateral axillary lymph nodes at initial presentation, received standardized neoadjuvant therapy (NAT), and subsequently underwent ALND. Patients were randomly divided into a training set (n = 354) and a test set (n = 151) in a 7:3 ratio. Based on ALND results, patients were classified into the apCR (axillary pathological complete response) and non-apCR groups, and their clinicopathological and magnetic resonance imaging (MRI) features were compared. Independent predictors of apCR were identified using multivariate logistic regression analysis, and feature selection was performed using the Least Absolute Shrinkage and Selection Operator (LASSO) method. Two predictive models were developed, a Clinical-Pathological-MRI model and a Clinical-Pathological-Delta-MRI model. The predictive performance of both models was evaluated and compared.

RESULTS

A total of 505 patients were enrolled, including 237 patients in the apCR group and 268 in the non-apCR group. The AUC values for the Clinical-Pathological-MRI model were 0.817 in the training set and 0.680 in the test set. For the Clinical-Pathological-Delta-MRI model, the AUC values were 0.844 in the training set and 0.793 in the test set, indicating superior predictive performance. Decision curve analysis (DCA) further demonstrated that the Clinical-Pathological-Delta-MRI model provided greater net clinical benefit compared to the Clinical-Pathological-MRI model in both the training and test sets.

CONCLUSION

This model may provide valuable support for individualized surgical decision-making and help guide the selective omission of axillary lymph node dissection in appropriate candidates.