Abstract OT3-02-02: Radiomics of multi-parametric MRI associated with axillary lymph node metastasis and prognostic in patients with breast cancer: A multicenter RBC-001 study

Abstract

Abstract Background: Axillary lymph node status has a guiding significance in the operative schemes. However, sensitivity and noninvasive for prediction of lymph node metastasis and survival of currently evidences is limited. This study aims to establish a deep learning algorithms of multi-parametric MRI radiomics for identifying lymph node metastasis and prognostic prediction, and investigate the relationship between the radiomics and tumor microenvironment in early stage breast cancer. Methods: RBC-01 is a multicentre, ambispective cohort study aims to assess multi-parametric MRI radiomics-based prediction model for identifying metastasis lymph nodes and prognostic prediction in early stage breast cancer. Patients are recruited from Sun Yat-sen Memorial Hospital of Sun Yat-sen University, Sun Yat-sen University Cancer Center, Tungwah hospital of Sun Yat-sen University, and Shunde Hospital of Southern Medical University in China. Patients from Sun Yat-sen Memorial Hospital of Sun Yat-sen University are training cohort, and the independent validation cohort were built by patients from other centers. Eligible women could be aged from 18 to 75 years old with pathological diagnosis of early stage invasive breast cancer, without distant organ metastasis, and completed the breast MRI examination before therapy. PyRadiomics is used to extract patients’ breast cancer region, paratumor region, and axillary lymph nodes in multi-parametric MRI. To avoid MRI data heterogeneity bias, all MRI data are subjected to imaging normalization and resampled to the same resolution before feature extraction. The least absolute shrinkage and selection operator, random forest and supporting vector machine models are used to select the most useful radiomics characteristics from the training cohort, and then test by validation cohorts. The correlation between radiomics features and tumor microenvironment is also planned to investigate. Combined with independent risk factors, a radiomics nomogram was built with a multivariate logistic regression model. Decision curve analysis was performed with the combined training and validation set to estimate the clinical usefulness of the nomogram. The primary endpoint is disease-free survival. Secondary endpoints are lymph node metastasis prediction and overall survival. The trial is registered with ClinicalTrials.gov, number NCT04003558, and Chinese Clinical Trail Registry, number ChiCTR1900024020. Citation Format: Herui Yao, Yujie Tan, Yunfang Yu, Chuanmiao Xie, Qiugen Hu, Jie Ouyang, Chenchen Li, Kai Chen, Nian Lu, Xiaohong Li, Rong Zhang, Jiafan Ma, Ying Wang, Jianli Zhao. Radiomics of multi-parametric MRI associated with axillary lymph node metastasis and prognostic in patients with breast cancer: A multicenter RBC-001 study [abstract]. In: Proceedings of the 2019 San Antonio Breast Cancer Symposium; 2019 Dec 10-14; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2020;80(4 Suppl):Abstract nr OT3-02-02.