Analysis of fractional anisotropy facilitates differentiation of glioblastoma and brain metastases in a clinical setting

Abstract

PurposeDifferentiating glioblastoma from brain metastases is important for therapy planning. Diffusion tensor imaging (DTI) was described as a promising tool, however with conflicting results. Aimof this study was to analyze the clinical utility of DTI for the differentiation of brain metastases and glioblastoma. Methods294 patients (165 glioblastoma, 129 brain metastases) with preoperative DTI were included in this retrospective study. Fractional anisotropy (FA) was measured via regions of interest (ROIs) in the contrast-enhancing tumor, the necrosis and the FLAIR-hyperintense non-enhancing peritumoral region (NEPTR). Two neuroradiologists classified patient cases as glioblastoma or brain metastases without and with knowledge of FA values. ResultsGlioblastoma showed significantly higher FAcontrast (median glioblastoma=0.33, metastases=0.23; P<0.001) whereas no significant difference was observed for FANEPTR (0.21 vs. 0.22; P=0.28) and for FAnecrosis (0.17 vs. 0.18, P=0.37). FA improved diagnostic accuracy of the neuroradiologists significantly from an AUC of 0.84/0.85 (Reader1/Reader2) to 0.89/0.92. ConclusionsGlioblastoma show significantly higher FA values in the contrast enhancing tumor part than brain metastases. Implementation of a ROI-based measurement of FA values and FA color maps in clinical routine helps to differentiate between glioblastoma and brain metastases.