A Nomogram Combined Radiomics and Kinetic Curve Pattern as Imaging Biomarker for Detecting Metastatic Axillary Lymph Node in Invasive Breast Cancer

• Published in: Frontiers in oncology Vol. 10; p. 1463
• Main Authors: Shan, Yan-na, Xu, Wen, Wang, Rong, Wang, Wei, Pang, Pei-pei, Shen, Qi-jun
• Format: Journal Article
• Language: English
• Published: Frontiers Media S.A 28.08.2020
• Subjects: algorithm; axillary lymph node; breast cancer; magnetic resonance imaging; Oncology; radiomics
• ISSN: 2234-943X; 2234-943X
• DOI: 10.3389/fonc.2020.01463

Objective: To construct and validate a nomogram model integrating the magnetic resonance imaging (MRI) radiomic features and the kinetic curve pattern for detecting metastatic axillary lymph node (ALN) in invasive breast cancer preoperatively. Materials and Methods: A total of 145 ALNs from two institutions were classified into negative and positive groups according to the pathologic or surgical results. One hundred one ALNs from institution I were taken as the training cohort, and the other 44 ALNs from institution II were taken as the external validation cohort. The kinetic curve was computed using dynamic contrast-enhanced MRI software. The preprocessed images were used for radiomic feature extraction. The LASSO regression was applied to identify optimal radiomic features and construct the Radscore. A nomogram model was constructed combining the Radscore and the kinetic curve pattern. The discriminative performance was evaluated by receiver operating characteristic analysis and calibration curve. Results: Five optimal features were ultimately selected and contributed to the Radscore construction. The kinetic curve pattern was significantly different between negative and positive lymph nodes. The nomogram model showed a better performance in both training cohort [area under the curve (AUC) = 0.91, 95% CI = 0.83–0.96] and external validation cohort (AUC = 0.86, 95% CI = 0.72–0.94); the calibration curve indicated a better accuracy of the nomogram model for detecting metastatic ALN than either Radscore or kinetic curve pattern alone. Conclusion: A nomogram model integrated the Radscore and the kinetic curve pattern could serve as a biomarker for detecting metastatic ALN in patients with invasive breast cancer.