Background and Purpose: The creation of new blood vessels, angiogenesis, plays a critical role in glioma development. The formation of irregular vasculature is not only varying between various glioma subtypes, but changes in the underlying vasculature can also herald malignant transformation (1). In particular in light of the recent 2021 WHO classification scheme, early and accurate diagnosis of glioma or malignant transformation is of utmost importance for prognosis and treatment decision making.
The most commonly assessed imaging biomarker of tumor vasculature is rCBV coming from dynamic susceptibility contrast (DSC) MRI. Although previous work illustrated that increased rCBV corresponds to increased malignancy in diffuse glioma, in practice this does not hold when oligodendroglioma - as defined in the WHO 2021 classification scheme - are included (2). This type of tumor is known for its “chicken wire” vasculature, which in turn can lead to high rCBV values (3). When extending DSC MRI to assess simultaneous acquisition of T₂­ and T₂^* weighted images to follow the passage of a gadolinium-based contrast agent (GBCA) through the cerebral vasculature, vessel size imaging (VSI) can be done, that has the potential to add information about microvascular structure. We explored using VSI in addition to rCBV to assess three diffuse glioma subtypes present within the WHO 2021 glioma classification.

Materials and Methods: A retrospective dataset consisting of 38 patients with confirmed non-enhancing glioma was used and classified in three groups: Oligo/IDH^MUT&1p/19q^-, Astro/IDH^MUT, and Glioblastoma/IDH^WT (4). All patients underwent 3T MRI scanning (GE, Milwaukee, WI, USA) prior to surgery. For DSC MRI, Hybrid EPI (HEPI) was used with following parameters: 122 repetitions, TR: 1500 ms, FOV: 15 slices, voxel size: 1.88x1.88x4.00 mm³, TE(GRE): 18.6 ms and TE(SE): 69 ms, with administration of 7.5 ml of GBCA (Gadovist, Bayer, Leverkusen, GE). A pre-load bolus of equal size was administrated 5 minutes before the DSC scan was acquired. Diffusion weighted imaging (2 b-values: 0, 1000 s/mm², voxel size: 1x1x3 mm³ and TE/TR of 63/5000 ms) was done to estimate the apparent diffusion coefficient (ADC) required for VSI. High resolution T1-weighed pre- and post-contrast (voxel size: 1x1x0.5 mm³; TE/TR: 2.1/6.1 ms), T2 (voxel size: 0.5x0.5x3.2 mm³; TE/TR: 107/10000 ms), and FLAIR (voxel size: 0.6x0.5x0.5 mm³; TE/TR: 106/6000 ms) images were acquired and used for tumor segmentation.
Estimates of mean vessel size and normalized rCBV maps were made according to previously described methods (5,6). HD-GLIO was used to generate tumor regions of interest (ROI) for all patient (7). Average tumor vessel size, normalized rCBV, and ADC were calculated for each group of patients. Additionally, correlation analysis was performed between rCBV and vessel size for each subgroup.

Results: Patient information is summarized in Table 1. Figure 1 shows an example slice of ADC, rCBV and vessel size maps for three patients, each selected from one subgroup. The average of the microvascular parameters (rCBV, vessel size) and ADC are presented in figure 2 for each patient and each group. In Oligo/IDH^MUT&1p/19q^-, rCBV was significantly higher compared to Astro/IDH^MUT (p < 0.05, unpaired t-test), but equal to Glioblastoma/IDH^WT. A trend of increased vessel size was found for Oligo/IDH^MUT&1p/19q^- and Astro/IDH^MUT compared to Glioblastoma/IDH^WT, but no significant differences were found. Vessel size and rCBV showed strong correlation in Glioblastoma/IDH^WT (r=0.82, p=0.02), moderate correlation in Oligo/IDH^MUT&1p/19q^- (r=0.56, p=0.01) and no correlation in Astro/IDH^MUT (r=0.17, p=0.57) (Fig. 3).

Discussion: The result of this study on nonenhancing glioma suggests that rCBV and vessel size alone cannot distinguish between three subgroups based on the 2021 WHO guidelines. However, combining these two parameters and measuring the correlation of these two parameters sheds light on the microvasculature characteristics of each subgroup.
Intuitively it might be expected that vessel size and rCBV are positively correlated, however this correlation is modulated by the vessel density. In other words, in an area with densely packed microvessels, slight change in vessel diameters can result in a considerable change in rCBV, whereas in an area with low vessel density, change in vessel diameters may lead to no or limited change in rCBV. Oligo/IDH^MUT&1p/19q^- with high rCBV and high VSI showing a moderate relationship between these two parameters, might be reflecting high vessel density. Low density might be present within the Astro/IDH^MUT, as we see no correlation - while in Glioblastoma/IDH^WT our findings indicate high vessel density. Note that this would be in line with the histopathological findings in (3). Future work validating these results should include histology measurements of vessel size and mean vessel density of targeted biopsies of tumor tissue based on rCBV and vessel size and investigation of the clinical applicability of VSI in glioma imaging diagnostics.

Acknowledgements

No acknowledgement found.