Abstract
OBJECTIVE: Because triple-negative breast cancer has a poor prognosis, adjuvant intensive therapy can effectively improve its prognosis. How to make accurate decisions is lacking of research.This study aimed to develop and validate a model to predict disease-free progression in triple-negative breast cancer (TNBC) using breast color Doppler ultrasound and magnetic resonance imaging (MRI), to facilitate precision in clinical intervention. METHODS: A retrospective analysis was conducted on data from 380 individuals with TNBC between June 2018 and June 2022. Collected variables included patient demographics, pathological characteristics, and imaging parameters. Predictive models were developed using variable selection through Cox regression analysis, random forest, and eXtreme gradient boosting (XGBoost). Model performance was evaluated using receiver operating characteristic (ROC) curves, area under the ROC curve (AUC) values, calibration curves, and measures such as net reclassification improvement and integrated discrimination improvement (IDI). The optimal model was visualized and subjected to clinical testing. RESULTS: Comparative analysis revealed that the Cox model outperformed the Rf.cox and XGBoost.cox models. Specifically, at the 48-month time point in the validation set, the XGBoost.cox model demonstrated inferior performance compared to the Cox model. The Cox model was chosen as the optimal model, incorporating seven variables: Age, T-Stage, N-Stage, Ki-67, SE-Score, time-signal intensity curve, and early-phase enhancement. The AUC was 0.937 (0.904-0.971) in the training set and 0.906 (0.855-0.957) in the validation set. Decision curve analysis and clinical impact curve supported the potential utility of the model in guiding clinical interventions. CONCLUSION: The predictive model for disease-free progression in TNBC, based on imaging parameters from breast color Doppler ultrasound and MRI, demonstrates feasibility. Further studies are recommended to confirm its clinical applicability.