Multivariable Models Based on Baseline Imaging Features and Clinicopathological Characteristics to Predict Breast Pathologic Response after Neoadjuvant Chemotherapy in Patients with Breast Cancer.

Introduction: Currently, the accurate evaluation and prediction of response to neoadjuvant chemotherapy (NAC) remains a great challenge. We developed several multivariate models based on baseline imaging features and clinicopathological characteristics to predict the breast pathologic complete response (pCR).

Methods: We retrospectively collected clinicopathological and imaging data of patients who received NAC and subsequent surgery for breast cancer at our hospital from June 2014 till September 2020. We used mammography, ultrasound, and magnetic resonance imaging (MRI) to investigate the breast tumors at baseline.

Results: A total of 308 patients were included and 111 patients achieved pCR. The HER-2 status and Ki-67 index were significant factors for pCR on univariate analysis and in all multivariate models. Among the prediction models in this study, the ultrasound plus MRI model performed best, producing an area under curve of 0.801 (95% CI 0.749-0.852), a sensitivity of 0.797, and a specificity of 0.676.

Conclusion: Among the multivariable models constructed in this study, the ultrasound plus MRI model performed best in predicting the probability of pCR after NAC. Further validation is required before it is generalized.