C-Q Su1, S-S Lu1, M-D Zhou1, H Shen1, H-B Shi1, X-N Hong2. 1. Department of Radiology, The First Affiliated Hospital of Nanjing Medical University, Nanjing, Jiangsu Province, 210029, China. 2. Department of Radiology, The First Affiliated Hospital of Nanjing Medical University, Nanjing, Jiangsu Province, 210029, China. Electronic address: hongxunning@sina.com.
Abstract
AIM: To examine whether texture analysis (TA) of diffusion-weighted imaging (DWI) combined with conventional magnetic resonance imaging (MRI) could non-invasively predict isocitrate dehydrogenase 1 (IDH1) mutational status in anaplastic gliomas. MATERIALS AND METHODS: Fifty-two patients with histologically confirmed anaplastic glioma was reviewed retrospectively. Conventional MRI was evaluated using the Visually Accessible Rembrandt Images (VASARI) scoring system. TA of DWI based on the entire tumour volume was compared between IDH1-mutant and wild-type tumours by using unpaired Student's t-test. Receiver operating characteristic curve (ROC) and logistic regression were used to assess their diagnostic performance. RESULTS: Significant statistical differences in VASARI features and TA of DWI were observed between IDH1-mutant and wild-type tumours (all p<0.05). Using multivariable logistic regression, the proportion of the tumour that was non-enhancing and the entropy of apparent diffusion coefficient (ADC) were found to possess higher prediction potential for IDH1 mutation with areas under the ROC curve (AUC) of 0.918 and 0.724, respectively. A combination of these for the identification of IDH1 mutations improved the AUC to 0.954, with a sensitivity and a specificity of 81% and 96%. CONCLUSIONS: The combined assessment of the conventional MRI and TA of DWI were useful for predicting IDH1 mutation in anaplastic gliomas.
AIM: To examine whether texture analysis (TA) of diffusion-weighted imaging (DWI) combined with conventional magnetic resonance imaging (MRI) could non-invasively predict isocitrate dehydrogenase 1 (IDH1) mutational status in anaplastic gliomas. MATERIALS AND METHODS: Fifty-two patients with histologically confirmed anaplastic glioma was reviewed retrospectively. Conventional MRI was evaluated using the Visually Accessible Rembrandt Images (VASARI) scoring system. TA of DWI based on the entire tumour volume was compared between IDH1-mutant and wild-type tumours by using unpaired Student's t-test. Receiver operating characteristic curve (ROC) and logistic regression were used to assess their diagnostic performance. RESULTS: Significant statistical differences in VASARI features and TA of DWI were observed between IDH1-mutant and wild-type tumours (all p<0.05). Using multivariable logistic regression, the proportion of the tumour that was non-enhancing and the entropy of apparent diffusion coefficient (ADC) were found to possess higher prediction potential for IDH1 mutation with areas under the ROC curve (AUC) of 0.918 and 0.724, respectively. A combination of these for the identification of IDH1 mutations improved the AUC to 0.954, with a sensitivity and a specificity of 81% and 96%. CONCLUSIONS: The combined assessment of the conventional MRI and TA of DWI were useful for predicting IDH1 mutation in anaplastic gliomas.
Authors: Michael E Berens; Anup Sood; Jill S Barnholtz-Sloan; John F Graf; Sanghee Cho; Seungchan Kim; Jeffrey Kiefer; Sara A Byron; Rebecca F Halperin; Sara Nasser; Jonathan Adkins; Lori Cuyugan; Karen Devine; Quinn Ostrom; Marta Couce; Leo Wolansky; Elizabeth McDonough; Shannon Schyberg; Sean Dinn; Andrew E Sloan; Michael Prados; Joanna J Phillips; Sarah J Nelson; Winnie S Liang; Yousef Al-Kofahi; Mirabela Rusu; Maria I Zavodszky; Fiona Ginty Journal: PLoS One Date: 2019-12-27 Impact factor: 3.240