To construct a clinical model, an imaging model, and a combined model for predicting pathological complete response (pCR) based on the clinicopathological characteristics and ultrasound imaging characteristics of breast cancer patients before neoadjuvant chemotherapy (NAC), and to compare the clinical value of the three models in predicting the efficacy of NAC.

A total of 202 female patients who underwent NAC for breast cancer at Huangshi Central Hospital from January 2019 to January 2024 and had the results of pathologic evaluation were retrospectively collected. The cases from two different hospital districts were classified into a training set (107 cases) and a validation set (95 cases). In the training set, univariate and multivariate logistic regression analyses were performed to identify clinically significant features and construct a clinical model. The minimum absolute contraction and selection operator regression were used to screen the radiomics features and establish a radiomics model. The Radscore of each patient was calculated according to the characteristic parameters and their regression coefficients. Multivariate logistic regression was then used to construct a combined prediction model based on statistically significant clinicopathological features and Radscore. Receiver operating characteristic (ROC) curve and decision curve were used to evaluate and verify the predictive efficacy and clinical value of the three models.

In the training set, hormone receptor (HR) status and human epidermal growth factor receptor 2 (HER2) status were included in the multivariate logistic regression to establish a clinical model. Lasso regression was used to select five best radiomics features, namely, contrast, correlation, entropy, gray unevenness, and percentage of run, to establish a radiomics model. Univariate logistic regression showed that Radscore was significantly different between patients with and without pCR. HR status (odds ratio [OR]: 0.31, 95% confidence interval [CI]: 0.15-0.64, P < 0.01), HER2 status (OR: 2.96, 95%CI: 1.43-6.12, P < 0.01), and Radscore (OR: 1.19, 95%CI: 1.07-1.33, P < 0.01) were included to build a combined model. In the training set, the area under the ROC curve (AUC) of the clinical model, radiomics model, and combined model in predicting pCR was 0.68, 0.75, and 0.82, respectively. In the validation set, the AUC of the clinical model, radiomics model, and combined model in predicting pCR was 0.68, 0.72, and 0.79, respectively. The combined model had the highest net benefit value in the decision curve analysis, followed by the radiomics model and the clinical model.

The clinical model, radiomics model, and combined model are all clinically valuable for predicting NAC efficacy in breast cancer, with the combined model having the best efficacy, predictive performance, and clinical applicability.