• Jo Sung Jung, Yoon Seong Choi, Sung Soo Ahn, Seong Yi, Se Hoon Kim, and Seung-Koo Lee.
• Department of Radiology and Research Institute of Radiological Science, College of Medicine, Yonsei University College of Medicine, 50 Yonsei-ro, Seodaemun-gu, Seoul, 120-752, South Korea.
• Neuroradiology. 2019 Mar 1; 61 (3): 313-322.

PurposeDiffuse midline glioma with histone H3 K27M mutation is a new entity described in the 2016 update of the World Health Organization Classification of Tumors of the Central Nervous System. The purpose of this study was to evaluate the clinical and imaging characteristics to predict the presence of H3 K27M mutation in spinal cord glioma using a machine learning-based classification model.MethodsA total of 41 spinal cord glioma patients consisting of 24 H3 K27M mutants and 17 wild types were enrolled in this retrospective study. A total of 17 clinical and radiological features were evaluated. The random forest (RF) model was trained with the clinical and radiological features to predict the presence of H3 K27M mutation. The diagnostic ability of the RF model was evaluated using receiver operating characteristic (ROC) analysis. Area under the ROC curves (AUC) was calculated.ResultsMR imaging features of spinal cord diffuse midline gliomas were heterogeneous. Hemorrhage was the only variable that was able to differentiate H3 K27M mutated tumors from wild-type tumors in univariate analysis (p = 0.033). RF classifier yielded 0.632 classification AUC (95% CI, 0.456-0.808), 63.4% accuracy, 45.8% sensitivity, and 88.2% specificity.ConclusionOur findings indicate that clinical and radiological features are associated with H3 K27M mutation status in spinal cord glioma.

36 keywords...

hide…