Abstract
BACKGROUND: To explore the predictive value of conventional ultrasound combined with grayscale Ultrasound Radiomics for major pathological response (MPR) after neoadjuvant therapy (NAT) in triple-negative breast cancer (TNBC), with the aim of guiding the selection of neoadjuvant treatment strategies. METHODS: A retrospective analysis was performed on the clinical data of TNBC patients pathologically diagnosed at Fujian Cancer Hospital and treated with NAT from January 2023 to March 2025. All patients underwent conventional two-dimensional ultrasound examination before NAT, and modified radical mastectomy of the affected breast was performed after NAT to obtain gross pathological examination results. Pathological efficacy was evaluated according to the Miller Payne Grading (MPG) criteria, with MPG grades 4 and 5 defined as MPR. Univariate analysis was used to compare the conventional ultrasound and contrast-enhanced ultrasound features of the two groups before NAT. Multivariate logistic regression analysis was applied to identify independent predictors of NAT efficacy in TNBC, followed by the establishment of a predictive model. The receiver operating characteristic curve was plotted to evaluate the model’s diagnostic performance. RESULTS: A total of 71 TNBC patients were included, among whom 22 (30.99%) achieved MPR, while 49 (69.01%) were non-MPR. Binary logistic regression analysis revealed that tumor margin (Beta = 1.86, SE = 0.69; OR = 6.45, p = 0.007), tumor aspect ratio (Beta = -4.17, SE = 1.76; OR = 0.015, p = 0.017), and GLSZM-Small Area Low Gray Level Emphasis (Beta = -4.28, SE = 1.78; OR = 0.014, p = 0.016) were independent predictive factors for effective NAT in TNBC. The predictive model showed good discrimination ability, with an original AUC of 0.814 and a bootstrap-validated AUC of 0.786, sensitivity of 77.3%, specificity of 79.5%, and accuracy of 78.9%. Both the calibration plot and decision curve analysis (DCA) indicated favorable clinical utility for this model. CONCLUSIONS: Conventional ultrasound combined with grayscale ultrasound radiomics showed an association with MPR after NAT in TNBC. The developed model is promising and warrants further validation in larger, prospective cohorts.