Complementary role and spatial relationship of 18F-FAZA PET/CT, perfusion and diffusion brain MRI for hypoxia characterization in high-grade glioma

Purpose: To define the complementary value of 18F-FAZA PET/CT, perfusion and diffusion weighted MRI in assessing the functional status of high-grade glioma, by assessing the mutual spatial relationship of co-registered separated PET and MR acquisitions. Patients and Methods: Prospective single Centre clinical study. Twenty patients with brain MRI suggestive for HGG have been enrolled. Eighteen out of 20 pts underwent pre-treatment 18F-FAZA PET/CT. For each 18-FAZA PET scan, standardized uptake values (SUV) have been derived. All patients underwent MRI (FLAIR, T2w, 3D-T1w pre- and post-contrast, diffusion and perfusion images), but only 11 of them have been considered for analysis with 18F-FAZA in this preliminary work. FAZA-PET, MR Perfusion and Diffusion images were firstly co-registered to 3D-T1 MR. Ktrans, Vp and CBV maps were calculated from Dynamic contrast enhanced (DCE) and Dynamic Susceptibility contrast (DSC) MR images. Mean Diffusivity (MD) has been calculated from Diffusion Tensor MR (DTI). 18F-FAZA SUV maps were divided by the muscle value uptake to calculate defined maps. A threshold of tumour-to-muscle (T/M) >1.2 was used to define FAZA-based VOIs (VOIsFAZA). By using these VOIsFAZA, Spearman correlations between FAZA SUV and perfusion parameters have been calculated for each subject. Furthermore, a second VOI has been defined in pw-maps and MD based on the Gd-enhanced regions on 3D T1 post contrast images (VOIsGd). Finally, the spatial coordinates of center-of-mass (CoM) of VOI_Gds in each single parametric map (Ktrans, Vp, CBV,MD, FAZA) were found and relative distances were calculated for each patient.

Results: By using a T/M threshold of 1.2 for FAZA SUV, and the 3D T1 Gd-enhancement, the mean values ± SD of the volumes were: 12.4 ± 12.3cc for VOIsFAZA and 26.6±22.1cc for VOIsGd respectively. The tumor volumes were significantly different between the two modalities (p<0.004). Using the FAZA and VOIsGd, mean values of Dice and Jaccard coefficients were 0,588±0,166 and 0,434±0,165, while volume overlap was 0,612±0,254. The CoMs of pw-maps and MD in the VOIsGd, were in different positions depending on the map. The relative distances from FAZA CoMs of some MR CoMs were significantly correlated, specifically: Ktrans and Vp (p=0.000), Vp and CBV (p=0.002), Vp and MD (p=0.007) and CBV and MD (p=0.002). No significant difference in the distances were found between the different CoMs, neither normalizing with the volume extension.

Conclusions: Low similarity indexes indicate that FAZA PET and Gd-MR enhancement provide different information regarding tumour hypoxia status, while the similar relative distances of MR CoMs from FAZA CoM suggest that all the MR parameters are more correlated among them compared to FAZA. This may imply that FAZA PET and MRl give complementary information in HGG patients.