Xiang Zhang1,2, Zehong Yang1,2, Wenju Cui3,4, Chushan Zheng1,2, Haojiang Li5, Yudong Li2,6, Liejing Lu1,2, Jiaji Mao1,2, Weike Zeng1,2, Xiaodong Yang3, Jian Zheng7, Jun Shen8,9. 1. Department of Radiology, Sun Yat-Sen Memorial Hospital, Sun Yat-Sen University, No. 107 Yanjiang Road West, Guangzhou, 510120, People's Republic of China. 2. Guangdong Provincial Key Laboratory of Malignant Tumor Epigenetics and Gene Regulation, Medical Research Center, Sun Yat-Sen Memorial Hospital, Sun Yat-Sen University, No. 107 Yanjiang Road West, Guangzhou, 510120, People's Republic of China. 3. Medical Imaging Department, Suzhou Institute of Biomedical Engineering and Technology, Chinese Academy of Sciences, No. 88 Keling Road, Suzhou, 215163, People's Republic of China. 4. Institute of Biomedical Engineering, School of Communication and Information Engineering, Shanghai University, No. 99 Shangda Road, Shanghai, 200444, People's Republic of China. 5. Department of Radiology, Sun Yat-Sen University Cancer Center, Sun Yat-Sen University, No. 651 Dongfeng Road East, Guangzhou, 510060, People's Republic of China. 6. Department of Breast Surgery, Sun Yat-Sen Memorial Hospital, Sun Yat-Sen University, No. 107 Yanjiang Road West, Guangzhou, 510120, People's Republic of China. 7. Medical Imaging Department, Suzhou Institute of Biomedical Engineering and Technology, Chinese Academy of Sciences, No. 88 Keling Road, Suzhou, 215163, People's Republic of China. zhengj@sibet.ac.cn. 8. Department of Radiology, Sun Yat-Sen Memorial Hospital, Sun Yat-Sen University, No. 107 Yanjiang Road West, Guangzhou, 510120, People's Republic of China. shenjun@mail.sysu.edu.cn. 9. Guangdong Provincial Key Laboratory of Malignant Tumor Epigenetics and Gene Regulation, Medical Research Center, Sun Yat-Sen Memorial Hospital, Sun Yat-Sen University, No. 107 Yanjiang Road West, Guangzhou, 510120, People's Republic of China. shenjun@mail.sysu.edu.cn.
Abstract
OBJECTIVES: To develop and validate a multiparametric MRI-based radiomics nomogram for pretreatment predicting the axillary sentinel lymph node (SLN) burden in early-stage breast cancer. METHODS: A total of 230 women with early-stage invasive breast cancer were retrospectively analyzed. A radiomics signature was constructed based on preoperative multiparametric MRI from the training dataset (n = 126) of center 1, then tested in the validation cohort (n = 42) from center 1 and an external test cohort (n = 62) from center 2. Multivariable logistic regression was applied to develop a radiomics nomogram incorporating radiomics signature and predictive clinical and radiological features. The radiomics nomogram's performance was evaluated by its discrimination, calibration, and clinical use and was compared with MRI-based descriptors of primary breast tumor. RESULTS: The constructed radiomics nomogram incorporating radiomics signature and MRI-determined axillary lymph node (ALN) burden showed a good calibration and outperformed the MRI-determined ALN burden alone for predicting SLN burden (area under the curve [AUC]: 0.82 vs. 0.68 [p < 0.001] in training cohort; 0.81 vs. 0.68 in validation cohort [p = 0.04]; and 0.81 vs. 0.58 [p = 0.001] in test cohort). Compared with the MRI-based breast tumor combined descriptors, the radiomics nomogram achieved a higher AUC in test cohort (0.81 vs. 0.58, p = 0.005) and a comparable AUC in training (0.82 vs. 0.73, p = 0.15) and validation (0.81 vs. 0.65, p = 0.31) cohorts. CONCLUSION: A multiparametric MRI-based radiomics nomogram can be used for preoperative prediction of the SLN burden in early-stage breast cancer. KEY POINTS: • Radiomics nomogram incorporating radiomics signature and MRI-determined ALN burden outperforms the MRI-determined ALN burden alone for predicting SLN burden in early-stage breast cancer. • Radiomics nomogram might have a better predictive ability than the MRI-based breast tumor combined descriptors. • Multiparametric MRI-based radiomics nomogram can be used as a non-invasive tool for preoperative predicting of SLN burden in patients with early-stage breast cancer.
OBJECTIVES: To develop and validate a multiparametric MRI-based radiomics nomogram for pretreatment predicting the axillary sentinel lymph node (SLN) burden in early-stage breast cancer. METHODS: A total of 230 women with early-stage invasive breast cancer were retrospectively analyzed. A radiomics signature was constructed based on preoperative multiparametric MRI from the training dataset (n = 126) of center 1, then tested in the validation cohort (n = 42) from center 1 and an external test cohort (n = 62) from center 2. Multivariable logistic regression was applied to develop a radiomics nomogram incorporating radiomics signature and predictive clinical and radiological features. The radiomics nomogram's performance was evaluated by its discrimination, calibration, and clinical use and was compared with MRI-based descriptors of primary breast tumor. RESULTS: The constructed radiomics nomogram incorporating radiomics signature and MRI-determined axillary lymph node (ALN) burden showed a good calibration and outperformed the MRI-determined ALN burden alone for predicting SLN burden (area under the curve [AUC]: 0.82 vs. 0.68 [p < 0.001] in training cohort; 0.81 vs. 0.68 in validation cohort [p = 0.04]; and 0.81 vs. 0.58 [p = 0.001] in test cohort). Compared with the MRI-based breast tumor combined descriptors, the radiomics nomogram achieved a higher AUC in test cohort (0.81 vs. 0.58, p = 0.005) and a comparable AUC in training (0.82 vs. 0.73, p = 0.15) and validation (0.81 vs. 0.65, p = 0.31) cohorts. CONCLUSION: A multiparametric MRI-based radiomics nomogram can be used for preoperative prediction of the SLN burden in early-stage breast cancer. KEY POINTS: • Radiomics nomogram incorporating radiomics signature and MRI-determined ALN burden outperforms the MRI-determined ALN burden alone for predicting SLN burden in early-stage breast cancer. • Radiomics nomogram might have a better predictive ability than the MRI-based breast tumor combined descriptors. • Multiparametric MRI-based radiomics nomogram can be used as a non-invasive tool for preoperative predicting of SLN burden in patients with early-stage breast cancer.