青岛大学学报（医学版）2023，Vol.59Issue(5)：651-657,7.DOI:10.11712/jms.2096-5532.2023.59.147
基于MRI影像组学预测乳癌NAC病理完全缓解价值
VALUE OF MRI RADIOMICS IN PREDICTING PATHOLOGICAL COMPLETE RESPONSE AFTER NEOADJUVANT CHEMO-THERAPY IN PATIENTS WITH BREAST CANCER
摘要
Abstract
Objective To investigate the value of the comprehensive model of multiparametric MRI radiomic features and clinical features in predicting pathological complete remission(pCR)after neoadjuvant chemotherapy(NAC)in patients with local-ly advanced breast cancer(LABC).Methods A retrospective analysis was performed for the clinical data of 387 patients with pathologically confirmed breast cancer,and all patients received NAC and underwent ultrasound-guided core needle biopsy and breast dynamic contrast-enhanced magnetic resonance imaging within one month before NAC.Pathological outcome was evaluated for the surgical specimens after chemotherapy,and according to the Miller-Payne(MP)grading of surgical specimens,the patients were divided into pCR group and non-pCR group.A total of 305 patients were included in Institution Ⅰ and were randomly divided into training cohort and testing cohort at a ratio of 7∶3,and 82 patients were included in Institution Ⅱ as an independent external validation cohort.Clinicopathological data and imaging features of primary breast cancer lesion were collected from all patients,and univariate and multivariate analyses were used to identify the independent predictive factors for pCR and establish a clinical model.The region of interest of primary breast lesion was delineated on the images of T1 WI+C,DWI,and ADC,and radiomic features were extracted.The LASSO regression model was used for feature selection,and a radiomics score(Radscore)was established.The clinical features obtained and radscore were included in a comprehensive predictive model to establish a nomogram.The area under the ROC curve(AUC)was used to evaluate the performance of each model,and decision curve analysis(DCA)and calibra-tion curve were used to evaluate the clinical application value of the nomogram.Results The analysis of clinical and radiological features showed that Ki-67 status,neoadjuvant chemotherapy,time-intensity curve pattern,and minimum ADC value were inde-pendent predictive factors for pCR after NAC in the patients with LABC.The comprehensive nomogram model based on the clinical features obtained and radscore had a better prediction efficiency than the radiomics model or the clinical model alone,with an AUC of 0.97 in the training cohort,0.90 in the testing cohort,and 0.86 in the external validation cohort.DCA showed that the comprehensive nomogram model had greater clinical benefits than the clinical model or the radiomics model alone.Conclusion The comprehen-sive nomogram model based on multiparametric MRI radiomic features and clinical-radiological factors can predict pCR after NAC in patients with LABC.关键词
乳腺肿瘤/肿瘤辅助疗法/磁共振成像Key words
breast neoplasms/neoadjuvant therapy/magnetic resonance imaging分类
医药卫生引用本文复制引用
王晓琳,韩珺琪,刘晶晶,钟鑫,陈静静..基于MRI影像组学预测乳癌NAC病理完全缓解价值[J].青岛大学学报（医学版）,2023,59(5):651-657,7.基金项目
国家自然科学基金资助项目(8207071895) （8207071895）
