Pretreatment MR‐Based Radiomics Signature as Potential Imaging Biomarker for Assessing the Expression of Topoisomerase II alpha (TOPO‐IIα) in Rectal Cancer

• Published in: Journal of magnetic resonance imaging Vol. 51; no. 6; pp. 1881 - 1889
• Main Authors: Chen, Jiayou, Chen, Ying, Zheng, Dechun, Pang, Peipei, Lu, Jianping, Zheng, Xiang
• Format: Journal Article
• Language: English
• Published: Hoboken, USA John Wiley & Sons, Inc 01.06.2020; Wiley Subscription Services, Inc
• Subjects: Biomarkers; Calibration; Cancer; Chemotherapy; Colorectal cancer; DNA topoisomerase (ATP-hydrolysing); DNA topoisomerases; DNA Topoisomerases, Type II; Feature extraction; Field strength; Humans; Magnetic Resonance Imaging; Mathematical models; Medical imaging; Nomograms; Population studies; Radiomics; rectal cancer; Rectal Neoplasms - diagnostic imaging; Rectum; Regression analysis; Regression models; Retrospective Studies; Statistical analysis; Statistical tests; Training; radiomics; magnetic resonance imaging; DNA topoisomerases; rectal cancer
• ISSN: 1053-1807; 1522-2586
• DOI: 10.1002/jmri.26972

Background Rectal cancer (RC) is one of the most common cancers throughout the world. Chemotherapy or neoadjuvant chemotherapy play an important role in the treatment of advanced RC. Whether to add topoisomerase inhibitor to individualized chemotherapy is a puzzling question for clinicians. Purpose To investigate whether pretreatment MR‐based radiomics signature can assess the expression of topoisomerase II alpha (TOPO‐IIα) in RC. Study Type Retrospective. Population In all, 122 patients with RC. Field Strength/Sequence: Pretreatment 3.0T; T2WI turbo spin echo (TSE) sequence. Assessment A training group (n = 85) and a test group (n = 37) with pathologically confirmed RC. Patients underwent TOPO‐IIα expression. A total of 180 radiomics features were extracted from oblique axial T2WI TSE images of the entire primary tumor. The least absolute shrinkage and selection operator (LASSO) regression model was used to reduce the dimension of the data and select the features. Statistical Tests The assessment models were established by multivariable logistic regression analysis. The performance of the model was assessed by the receiver operating characteristic (ROC) curve, nomogram, and calibration. Results The radiomics signature, which consisted of 10 selected optimal features, was significantly associated with TOPO‐IIα expression (P < 0.01 for both training and test groups). The area under the curve (AUC), the sensitivity, and the specificity for assessing TOPO‐IIα expression, were 0.859, 0.872, and 0.739, respectively, in the training group, while they were 0.762, 0.941, and 0.600 in the test group. The nomogram model of the radiomics signature (Rad‐score) had good calibration. Calibration curves were plotted to assess the calibration of the radiomics nomogram that was accompanied with the Hosmer–Lemeshow test (P = 0.52). Data Conclusion The proposed pretreatment MR‐based radiomics signature was associated with TOPO‐IIα expression. A radiomics nomogram might be helpful in the individualized assessment of TOPO‐IIα expression in patients with RC. Level of Evidence: 2 Technical Efficacy: Stage 2 J. Magn. Reson. Imaging 2020;51:1881–1889.