Radiomics based on multiparametric MRI for prediction of breast cancers sensitive to neoadjuvant chemotherapy
Objective: To predict the sensitivity of breast cancer to neoadjuvant therapy (NAT) based on multiparametric magnetic resonance imaging (mpMRI) combined with clinical variables.Materials and Methods: A total of 248 patients with pathologically confirmed breast cancer were enrolled in this study and randomly divided into a training group (173 cases) and a validation group (75 cases) in a 7:3 ratio. All patients underwent mpMRI examination before NAT. The Miller-Payne (MP) grading system was used to assess the effectiveness of NAT, with MP grades 1-2 considered as insensitive to NAT response, and MP grades 3-5 as sensitive. Based on dynamic contrast-enhanced MRI (DCE-MRI), T2WI, and diffusion weighted imaging (DWI) sequence images to delineate tumor regions, to extract and filter imaging radiomics features. A radiomics score (Rad-score) was derived using the least absolute shrinkage and selection operator algorithm. Univariate logistic regression was ultilized to analyze clinical and pathological variables, including age, menstrual status, molecular subtype, chemotherapy regimen, estrogen receptor (ER), progesterone receptor (PR), human epidermal growth factor receptor-2 (HER-2), and tumor proliferative index Ki-67. Significant clinical and pathological variables, along with the Rad-score, were included in the multivariate logistic regression analysis to establish an radiomics-clinical combined model and nomogram. The predictive performance of model was assessed using receiver operating characteristic (ROC) curves, calibration curves, and decision curve analysis (DCA). Results: Univariate logistic regression analysis showed that Rad-score (P<0.001), ER expression status (P=0.001), and chemotherapy regimen (P=0.031) were significantly associated with the sensitivity of NAT in breast cancer. The AUC of the radiomics-clinical combined model constructed by Rad-score with ER expression status and chemotherapy regimen was 0.845 (95％ CI:0.780-0.910) in the training cohort, and 0.820 (95％ CI: 0.718-0.923) in the validation cohort. The nomogram in prediction of breast cancer susceptibility to NAT had a higher degree of differentiation (C index: training queue is 0.842, validation queue is 0.822), the calibration curve shows good consistency. The clinical decision curve showed that the nomogram had a high overall net benefit. Conclusions: The integration of radiomics and clinical variables and nomogram show promise in predicting sensitivity of breast cancer to neoadjuvant therapy.
breast cancerradiomicsmulti-parametric magnetic resonance imagingmagnetic resonance imagingneoadjuvant therapysensitivity
国家科技期刊平台
NSTL
万方数据
