LAUSR

Aim: Neoadjuvant chemotherapy (NAC) plays an important role in the treatment and prognosis of breast cancer. The early identification of patients who can truly benefit from preoperative NAC is crucial in clinical practice. The purpose of this study was to explore whether the ultrasound features and clinical characteristics combined with tumor-infiltrating lymphocyte(TIL) levels can improve the performance of predicting NAC efficacy in breast cancer patients.Material and methods: In this retrospective study, 202 invasive breast cancer patients who underwent NAC followed by surgery were included. The baseline ultrasound features were reviewed by two radiologists. Miller-Payne Grading (MPG) was used to assess pathological response, and MPG 4-5 was defined as major histologic responders (MHR). Multivariable logistic regression analysis was used to evaluate independent predictors for MHR and build the prediction models. The receiver operating characteristic (ROC) curve was used to evaluate the performance of the models.Results: Of the 202 patients, 104 patients achieved MHR and 98 patients achieved non-MHR. Multivariate logistic regression analysis showed the US size (p=0.042), molecular subtypes (p=0.001), TIL levels (p<0.001), shape (p=0.030), and posterior features (p=0.018) were independent predictors for MHR. The model combined the US features, clinical characteristics, and TIL levels had a better performance with an area under the curve (AUC) of 0.811, a sensitivity of 0.663, and a specificity of 0.847.Conclusion: The model combined US features, clinical characteristics, and TIL levels had a better performance in predicting pathological response to NAC in breast cancer