Development and evaluation of clinical models based on ultrasound, serum CA153 and molecular subtype in predicting the efficacy of neoadjuvant chemotherapy in breast cancer

Breast cancer is a threat to health, and reliable predictors for neoadjuvant chemotherapy (NAC) efficacy remain limited. The aim of the present study was to develop a clinical model that combines ultrasonography with clinical and pathological characteristics to predict the effectiveness of NAC in patients with breast cancer prior to surgery. The present retrospective study included patients with pathologically confirmed primary breast cancer who underwent NAC followed by surgery based on predefined inclusion and exclusion criteria. Patients were randomly allocated to a test and a validation set in a 7:3 ratio. Clinical data, comprehensive pathological reports and ultrasound imaging features were collected from the Hospital Information System of Hubei Cancer Hospital (Wuhan, China). Univariate and multivariate logistic regression analyses were conducted to identify independent predictive factors, with significance set at a bilateral P<0.05. The receiver operating characteristic curve was used to assess the performance of the model. The present study included 323 female patients with breast cancer. Univariate analysis indicated significant differences in tumor diameter changes, mass margin and posterior echo patterns (classified as shadow, enhancement or no posterior features), change in serum carbohydrate antigen 153 (CA153) levels, clinical N stage (classified as cN0‑cN3), hormone receptor expression, Ki‑67 levels and molecular type between the pathological complete response (pCR) and pathological incomplete response (non‑pCR) group. Multivariate analysis identified margin, molecular subtype, changes in tumor diameter and serum CA153 levels as significant predictors included in the final model (P<0.05). Furthermore, the present study model was validated in the validation set. The area under the curve for the predictive model in test set was 0.823 (95% confidence interval, 0.765‑0.881), while in the validation set, it was 0.884 (95% confidence interval, 0.816‑0.953). The integrated model, which incorporated ultrasound, serum CA153 levels and molecular subtypes, demonstrated notable potential for preoperative assessment of NAC efficacy in the future.