Share this content in WeChat
Clinical Articles
Prediction of MGMT promoter methylation in gliomas with different radiomics models based on MRI
CHEN Sixuan  XU Yue  YE Meiping  LI Yang  YU Zhixuan  QING Zhao  WANG Zhengge  ZHANG Bing  ZHANG Xin 

Cite this article as: Chen SX, Xu Y, Ye MP, et al. Prediction of MGMT promoter methylation in gliomas with different radiomics models based on MRI[J]. Chin J Magn Reson Imaging, 2022, 13(3): 1-5, 36. DOI:10.12015/issn.1674-8034.2022.03.001.

[Abstract] Objective To investigate the efficacy of different radiomics models based on MRI for predicting the status of O6-methylguanine-DNA methyltransferase (MGMT) promoter methylation in gliomas before operation.Materials and Methods The MR data of 114 patients with gliomas confirmed by pathology were analyzed retrospectively, including T1WI, T2WI, ADC and Gd-enhanced T1WI. Among them, 58 cases were MGMT promoter methylation and 56 cases were MGMT promoter unmethylation. All patients were randomly divided into training set (91 cases) and validation set (23 cases) according to the 8∶2 ratio. Three dimensional manual segmentation was performed on tumor edema zone and tumor core zone respectively on T2WI and Gd-enhanced T1WI. A total of 688 radiomics features were extracted. Principal component analysis was used for feature dimension reduction and analysis of variance was used to select features. Support vector machine (SVM), Logistic regression (LR), Lasso's Logistic regression via Lasso (LR-Lasso) and Bayesian classifier (native Bayes, NB) were used to build a diagnostic model. The validation data set was used to evaluate the accuracy and diagnostic efficiency of the model prediction with 5-folder cross validation. The ROC curve was drawn to dynamically evaluate the sensitivity and specificity of the model prediction, and the area under curve (AUC) statistical index was used to quantify the prediction efficiency of the model.Results The AUC value and accuracy of LR model were 0.90 and 91%, the sensitivity and specificity were 92% and 91%, the AUC value and accuracy of LR-Lasso model were 0.80 and 74%, the sensitivity and specificity were 67% and 82%, the AUC value and accuracy of SVM model were 0.89 and 87%, the sensitivity and specificity were 83% and 91%, the AUC value and accuracy of NB model were 0.69 and 74%, the sensitivity and specificity were 75% and 72%. The performance based on LR model was the highest.Conclusions The diagnostic model of multimodal MRI radiomics parameters could be used to predict the status of MGMT promoter methylation in glioma before operation. Among the four models, LR model has the highest prediction performance.
[Keywords] glioma;O6-methylguanine-DNA methyltransferase promoter methylation;radiomics;magnetic resonance imaging

CHEN Sixuan1   XU Yue3   YE Meiping1   LI Yang1   YU Zhixuan1   QING Zhao1, 2   WANG Zhengge1   ZHANG Bing1, 2   ZHANG Xin1*  

1 Department of Radiology, the Affiliated Drum Tower Hospital of Nanjing University Medical School, Nanjing 210093, China

2 Institute of brain Science, Nanjing University, Nanjing 210093, China

3 National Institute of Healthcare Data Science at Nanjing University, Nanjing 210093, China

Zhang X, E-mail:

Conflicts of interest   None.

ACKNOWLEDGMENTS National Natural Science Foundation of China (No. 81971596);Project of Modern Hospital Management and Development Institute, Nanjing University, Aid project of Nanjing Drum Tower Hospital Health, Education & Research Foundation (NDYG2021005).
Received  2021-09-06
Accepted  2022-02-17
DOI: 10.12015/issn.1674-8034.2022.03.001
Cite this article as: Chen SX, Xu Y, Ye MP, et al. Prediction of MGMT promoter methylation in gliomas with different radiomics models based on MRI[J]. Chin J Magn Reson Imaging, 2022, 13(3): 1-5, 36. DOI:10.12015/issn.1674-8034.2022.03.001.

Lapointe S, Perry A, Butowski NA. Primary brain tumours in adults[J]. Lancet, 2018, 392(10145): 432-446. DOI: 10.1016/S0140-6736(18)30990-5.
Weller M, Tabatabai G, Kästner B, et al. MGMT Promoter Methylation Is a Strong Prognostic Biomarker for Benefit from Dose-Intensified Temozolomide Rechallenge in Progressive Glioblastoma: The DIRECTOR Trial[J]. Clin Cancer Res, 2015, 21(9): 2057-2064. DOI: 10.1158/1078-0432.CCR-14-2737.
Stupp R, Hegi ME, Mason WP, et al. Effects of radiotherapy with concomitant and adjuvant temozolomide versus radiotherapy alone on survival in glioblastoma in a randomised phase III study: 5-year analysis of the EORTC-NCIC trial[J]. Lancet Oncol, 2009, 10(5): 459-466. DOI: 10.1016/S1470-2045(09)70025-7.
Sanai N, Martino J, Berger MS. Morbidity profile following aggressive resection of parietal lobe gliomas[J]. J Neurosurg, 2012, 116(6): 1182-1186. DOI: 10.3171/2012.2.JNS111228.
Gillies RJ, Kinahan PE, Hricak H. Radiomics: Images Are More than Pictures, They Are Data[J]. Radiology, 2016, 278(2): 563-577. DOI: 10.1148/radiol.2015151169.
Lambin P, Rios-Velazquez E, Leijenaar R, et al. Radiomics: extracting more information from medical images using advanced feature analysis[J]. Eur J Cancer, 2012, 48(4): 441-446. DOI: 10.1016/j.ejca.2011.11.036.
Xue CQ, Du XH, Jin L, et al. Prediction of methylation status of MGMT promoter in WHO grade Ⅱ,Ⅲ glioma based on MRI deep learning model[J]. Chin J Radiol, 2021, 55(7): 734-738. DOI: 10.3760/cma.j.cn112149-20200825-01029.
Montazeri M, Montazeri M, Beigzadeh A. Machine learning models in breast cancer survival prediction[J]. Technol Health Care, 2016, 24(1): 31-42. DOI: 10.3233/THC-151071.
Song Y, Zhang J, Zhang YD, et al. FeAture Explorer (FAE): A tool for developing and comparing radiomics models[J]. PLoS One, 2020, 15(8): e0237587. DOI: 10.1371/journal.pone.0237587.
Kickingereder P, Götz M, Muschelli J, et al. Large-scale Radiomic Profiling of Recurrent Glioblastoma Identifies an Imaging Predictor for Stratifying Anti-Angiogenic Treatment Response[J]. Clin Cancer Res, 2016, 22(23): 5765-5771. DOI: 10.1158/1078-0432.CCR-16-0702.
Song SS, Lu J. Progresses in imaging study of MGMT promoter methylation in gliomas[J]. Chin J Med Imaging Technol, 2019, 35(5): 147-150. DOI: 10.13929/j.1003-3289.201811009.
Yu J, Shi Z, Lian Y, et al. Noninvasive IDH1 mutation estimation based on a quantitative radiomics approach for grade Ⅱ glioma[J]. Eur Radiol, 2017, 27(8): 3509-3522. DOI: 10.1007/s00330-016-4653-3.
Christodoulou E, Ma J, Collins GS, et al. A systematic review shows no performance benefit of machine learning over logistic regression for clinical prediction models[J]. J Clin Epidemiol, 2019, 110(202): 12-22. DOI: 10.1016/j.jclinepi.2019.02.004.
Wu HY, Gong CA, Lin SP, et al. Predicting postoperative vomiting among orthopedic patients receiving patient-controlled epidural analgesia using SVM and LR[J]. Sci Rep, 2016, 6(10): 2045-2322. DOI: 10.1038/srep27041.
Sasaki T, Kinoshita M, Fujita K, et al. Radiomics and MGMT promoter methylation for prognostication of newly diagnosed glioblastoma[J]. Sci Rep, 2019, 9(1): 14435. DOI: 10.1038/s41598-019-50849-y.
Xi YB, Guo F, Xu ZL, et al. Radiomics signature: A potential biomarker for the prediction of MGMT promoter methylation in glioblastoma[J]. J Magn Reson Imaging, 2018, 47(5): 1380-1387. DOI: 10.1002/jmri.25860.
Wei J, Yang G, Hao X, et al. A multi-sequence and habitat-based MRI radiomics signature for preoperative prediction of MGMT promoter methylation in astrocytomas with prognostic implication[J]. Eur Radiol. 2019, 29(2): 877-888. DOI: 10.1007/s00330-018-5575-z.
Jiang C, Kong Z, Liu S, et al. Fusion Radiomics Features from Conventional MRI Predict MGMT Promoter Methylation Status in Lower Grade Gliomas[J]. Eur J Radiol, 2019, 121(6): 108714. DOI: 10.1016/j.ejrad.2019.108714.
Han C, Ma S, Liu X, et al. Radiomics Models Based on Apparent Diffusion Coefficient Maps for the Prediction of High-Grade Prostate Cancer at Radical Prostatectomy: Comparison With Preoperative Biopsy[J]. J Magn Reson Imaging, 2021, 54(6): 1892-1901. DOI: 10.1002/jmri.27565.
Kuess P, Andrzejewski P, Nilsson D, et al. Association between pathology and texture features of multi parametric MRI of the prostate[J]. Phys Med Biol, 2017, 62(19): 7833-7854. DOI: 10.1088/1361-6560/aa884d.
Ringnér M. What is principal component analysis?[J]. Nat Biotechnol, 2008, 26(3): 303-304. DOI: 10.1038/nbt0308-303.
Modhukur V, Sharma S, Mondal M, et al. Machine Learning Approaches to Classify Primary and Metastatic Cancers Using Tissue of Origin-Based DNA Methylation Profiles[J]. Cancers (Basel), 2021, 13(15): 3768. DOI: 10.3390/cancers13153768.

NEXT The value of FLAIR vascular hyperintensity-diffusion weighted imaging mismatch in assessing collateral circulation in acute ischemic stroke

Tel & Fax: +8610-67113815    E-mail: