Mri-Based Radiomics Signature for Predicting Response to Neoadjuvant Chemotherapy in Patients with Locally Advanced Rectal Cancer: A Multicentre Study
TingDan Hu,Jing Gong,Yanfen Cui,XiaoYan Zhang,Menglei Li,ChongPeng Cai,Ying-Shi Sun,YiQun Sun,Tong
DOI: https://doi.org/10.2139/ssrn.4270327
2022-01-01
SSRN Electronic Journal
Abstract:Background: We aimed to develop and validate a radiomics signature based on pretreatment MRI images to predict a good response (GR) to neoadjuvant chemotherapy (nCRT) in patients with locally advanced rectal cancer (LARC) and to research its prognostic value in predicting distant metastasis.Methods: A total of 922 patients with LARC were retrospectively recruited from 3 Chinese hospitals between Dec 1, 2009, and Jan 30, 2018. The training dataset was comprised of 509 patients from centre I; external validation dataset 1 (EVD1) was comprised of 242 patients from centre II; and EVD2 was comprised of 171 patients from centre III. The radiomics features were extracted from the T2WI and ADC images. A three-step procedure, including the least absolute shrinkage and selection operator (LASSO) algorithm, minority oversampling technique (SMOTE) and support vector machine (SVM) classifier, was applied for feature selection and signature building. The prediction performance was evaluated using the area under the curve (AUC), accuracy, sensitivity and specificity. A nomogram integrating the radiomics signature and clinical-MRI features to predict distant metastasis (DM) was constructed.Findings: Nineteen, 3, and 22 radiomic features were selected in the T2WI-based, ADC-based, and fusion models for GR prediction, respectively. Among the three models, the fusion model showed satisfactory discrimination of GR to nCRT and yielded AUCs of 0.89±0.02 (95% CI, 0.87−0.92), 0.69±0.05 (95% CI, 0.61−0.78), and 0.72±0.04 (95% CI, 0.65−0.79) in the training and two external validation datasets, respectively. In addition, the combined nomogram was successfully built with AUCs of 0.69, 0.78 and 0.66 in the three datasets. Kaplan‒Meier curves of disease-free survival(DFS) between the high-risk and low-risk DM patients who were grouped by the nomogram showed statisticalsignificance with p<0.05 in the three datasets.Interpretation:The MRI-based radiomics signature exhibited promising performance for response prediction to nCRT in LARC patients and can also provide prognostic value for DM prediction, which may assist in individualized clinical treatment.Funding Information: This study was supported by the National Natural Science Foundation of China (No. 82001776, 81971687, 82271946), Shanghai AntiCancer Association Young Eagle Program (SACA-CY20C130),Shanghai Natural Science Foundation (No. 20ZR1412700). Declaration of Interests: All authors declare no competing interests. Ethics Approval Statement: This retrospective multicentre study was conducted in accordance with the Declaration of Helsinki. Ethics approval was obtained from the Institutional Review Board of all participating centres, and the requirement for informed consent was waived due to the retrospective nature. A total of 509 consecutive patients with LARC were enrolled in Fudan University Shanghai Cancer Center (FUSCC) from 2016/05 to 2018/01; 171 consecutive patients from 2013/03 to 2016/11 were enrolled in Shanxi Cancer Hospital (SXPCH); and 242 consecutive patients from 2009/12 to 2015/05 were enrolled in Beijing Cancer Hospital (BJCH).