Introduction: Magnetic resonance (MR) tractography can be used to study the spatial relations between gliomas and white matter (WM) tracts. Various spatial patterns of WM tract alterations have been described in the literature. We reviewed classification systems of these patterns, and investigated whether low-grade gliomas (LGGs) and high-grade gliomas (HGGs) demonstrate distinct spatial WM tract alteration patterns. Methods: We conducted a systematic review and meta-analysis to summarize the evidence regarding MR tractography studies that investigated spatial WM tract alteration patterns in glioma patients. Results: Eleven studies were included. Overall, four spatial WM tract alteration patterns were reported in the current literature: displacement, infiltration, disruption/destruction and edematous. There was a considerable heterogeneity in the operational definitions of these terms. In a subset of studies, sufficient homogeneity in the classification systems was found to analyze pooled results for the displacement and infiltration patterns. Our meta-analyses suggested that LGGs displaced WM tracts significantly more often than HGGs (n = 259 patients, RR: 1.79, 95% CI [1.14, 2.79], I2 = 51%). No significant differences between LGGs and HGGs were found for WM tract infiltration (n = 196 patients, RR: 1.19, 95% CI [0.95, 1.50], I2 = 4%). Conclusions: The low number of included studies and their considerable methodological heterogeneity emphasize the need for a more uniform classification system to study spatial WM tract alteration patterns using MR tractography. This review provides a first step towards such a classification system, by showing that the current literature is inconclusive and that the ability of fractional anisotropy (FA) to define spatial WM tract alteration patterns should be critically evaluated. We found variations in spatial WM tract alteration patterns between LGGs and HGGs, when specifically examining displacement and infiltration in a subset of the included studies.
Purpose: We investigated whether quantitative tractography-based spatial metrics can objectively characterize visual tumor-induced white matter (WM) tract displacement in low-grade glioma. Specifically, the frontal aslant tract (FAT) was evaluated using the weighted Dice similarity coefficient (wDSC) and Hausdorff distance (HD) as spatial metrics. Methods: We retrospectively analyzed 45 patients with IDH-mutated frontal low-grade glioma (LGG), in whom preoperative diffusion-weighted MRI and probabilistic tractography of the FAT were performed. Patients were classified into one of three spatial WM tract alteration pattern groups (displacement, infiltration, combination) based on expert visual assessment. The wDSC and HD were investigated in both MNI-152 normalized and native anatomical space. In normalized space, inter-hemispheric comparisons (within patients) and intra-hemispheric comparisons (across patients) of spatial metrics were performed. In native space, comparisons were performed within patients, between mirrored ipsilesional and contralesional tracts. Results: In normalized MNI space, FAT displacement led to a decreased wDSC and increased HD during inter-hemispheric comparisons within patients, while infiltrated tracts showed no such differences. Conversely, intra-hemispheric spatial metrics, assessed across patients and independently of contralateral tract anatomy, could not differentiate FAT displacement from infiltration in MNI space. In native space, displacement of the FAT could be characterized objectively, using tract mirroring. Conclusion: Visual tumor-induced displacement of the FAT can be objectively confirmed with measurable geometric tract alterations using spatial metrics. While displacement could be characterized through inter-hemispheric comparisons in MNI space, native space might be more robust for detecting intra-hemispheric geometric differences. This study provides the foundation for an objective, quantitative framework to evaluate spatial WM tract alterations in glioma. These results might advance non-invasive radiomics-based tumor subtype predictions.