The value of dynamic contrast-enhanced MRI radiomics in predicting androgen receptor expression in breast cancer
10.3969/j.issn.1002-1671.2025.09.015
- VernacularTitle:动态增强MRI的影像组学在预测乳腺癌雄激素受体表达中的价值
- Author:
Na CHEN
1
;
Haiyue TIAN
;
Qinghong MENG
Author Information
1. 新泰市中医医院医学影像科,山东 泰安 271200
- Publication Type:Journal Article
- Keywords:
dynamic contrast-enhanced magnetic resonance imaging;
radiomics;
breast cancer;
androgen receptor
- From:
Journal of Practical Radiology
2025;41(9):1494-1497,1507
- CountryChina
- Language:Chinese
-
Abstract:
Objective To explore the value of dynamic contrast-enhanced MRI radiomics in predicting androgen receptor(AR)expression in breast cancer.Methods A total of 166 patients with pathological confirmed breast cancer and underwent dynamic con-trast-enhanced MRI were selected.All patients were randomly divided into training set(n=116)and test set(n=50)in a ratio of 7∶3.The 116 patients in the training set were divided into AR positive group(n=71)and AR negative group(n=45)according to the results of AR immunohistochemical.The clinical and radiomics prediction models were established by using the clinical data and radiomics characteristics of the patients,respectively.The combined prediction model was constructed by integrative both the Radiomics score(Radscore)and relevant clinical factors,and the prediction performance of the models was evaluated.Results The clinical model was 1.255+1.109×tumor diameter+0.801×lymph node metastasis+0.255×histological grade+0.215×TNM stage+0.226×estro-gen receptor(ER)+0.359×human epidermal growth factor receptor 2(HER-2)+0.311×Ki-67.The radiomics model was constructed based on the 5 potential predictors screened by least absolute shrinkage and selection operator(LASSO)model,Radscore=1.342+1.393×Orig_Shape_MAL+1.248×Wave_LHH_GLSZM_GLNU+0.888×Wave_HHH_GLDM_DE+0.714×Wave_HLL_FO_Minimum+1.138×Wave_HHH_FO_TE.The combined prediction model was 1.928+1.043×tumor diameter+1.012×lymph node metasta-sis+0.332×histological grade+0.309×TNM stage+1.059×ER+1.017×HER-2+0.893×Ki-67+1.259×Radscore.The receiver operating characteristic(ROC)curves showed that all of the three models had good prediction performance in both the training set and the test set.DeLong test showed that the area under the curve(AUC)of the combine model was significantly higher than that of the clinical and radiomics models in the training set and test set(P＜0.05).Calibration curve showed that the three models had good fitting effect in the training and test sets.The clinical deci-sion curve showed that the combine model had higher clinical practical value than the other two models both in the training set and test set.Conclusion The combined prediction model constructed by dynamic contrast-enhanced MRI radiomics and clinical characteristic can be used to predict AR expression and has high clinical application value.
