The utility of multi-b-value diffusion and arterial spin labelling magnetic resonance imaging in gliomas grading and prediction of isocitrate dehydrogenase status.

Background: Although the possibilities for grading adult gliomas by conventional magnetic resonance imaging (MRI) are limited, several advanced techniques may be used in this respect. In this study, we evaluate the feasibility of multi-b-value diffusion MRI and arterial spin labelling (ASL) in grading gliomas and predicting their isocitrate dehydrogenase (IDH) molecular status. Methods: Preoperative brain MRI including multi-b-value diffusion sequences and ASL was performed prospectively in patients with gliomas. Three-dimensional (3D) masks of tumours were semi-automatically segmented and multiple diffusion parameters and cerebral blood flow (CBF) maps calculated. Histogram analysis of all parameters was performed and the parameters' diagnostic power to differentiate between high-grade glioma (HGG) and low-grade glioma (LGG) as well as between IDH-mutated and IDH wild-type subgroups was evaluated by receiver operating characteristic (ROC) analysis and least absolute shrinkage and selection operator (LASSO) regression method. Results: The study group included 107 patients (83 HGGs and 24 LGGs, 35 IDH-mutated and 72 IDH wild-type). Overall, 49 histogram diffusion and CBF parameters differed significantly between HGG and LGG subgroups and 42 parameters differed between IDH-mutated and wild-type subgroups. ROC analysis showed multi-b-diffusion parameters to be generally stronger predictors than were CBF parameters. Decision trees using 4 parameters selected by LASSO analysis achieved sensitivity of 0.988 and specificity 0.912 for differentiating HGG from LGG and sensitivity 0.857 and specificity 0.93 for predicting IDH status. Conclusions: Multi-b diffusion MRI and ASL may be valuable diagnostic tools for grading adult brain gliomas and predicting their IDH molecular status.