Fully hybrid 11C-Methionine PET/MRI in brain oncological lesions: spatial relationship between brain tumor regions assessed by 11C-Methionine and perfusion weighted imaging (PWI)

Purpose: 11C-Methionine PET uptake (MET-PET) and perfusion MRI (PW-MRI) are tightly coupled, but it is not clear whether tumor regions (VOIs) defined by PWI and MET are overlapped and how the values of the corresponding parametric maps are distributed within each VOI. This preliminary work aims at studying the spatial relationship between PW-MRI maps and MET-PET in brain tumors by using a fully integrated PET/MRI system. Patients and Methods: Ten patients (13 lesions) with primary or secondary cerebral neoplastic diseases, characterized by Gd enhancement, underwent a simultaneous 11C-MET PET/MRI (SIGNA PET/MR system; General Electric Healthcare, Waukesha, WI). MR protocol included T1w, T2w, FLAIR, DWI, DCE and DSC PWI sequences with double Gd bolus injections and, finally, a 3D T1w post-contrast scan. Simultaneous MET-PET acquisition has been collected for 20 minutes on the brain level. ZTE-based attenuation correction was used to account for the head bone. After the assessment of the optimal co-registration of PW-MRI and MET-PET over 3D post-contrast T1w images, a neuro-radiologist and a nuclear medicine physician manually defined the segmentation of the lesions using PMOD software. MRI lesions (Gd-VOIs) and PET lesions (PET-VOIs) have been defined on T1w enhancement and on the PET images, respectively. PET-VOIs were used to measure SUVmax, MTV and SUVmean for each lesion seen on MET-PET. Ktrans, Vp and rCBV maps have been calculated using OLEA (Olea Medical, La Ciotat, France) software and the Gd-VOIs have been used to mask Ktrans, Vp and rCBV maps. DICE coefficient for each lesion of each patient was calculated to measure the overlap between the corresponding VOIs on PET and Gd-enhanced MRI. The Center of Masses (CoM) of each masked map (Ktrans, Vp, CBV and MET-PET) have been localized and the distances between MET-PET CoM and PWI CoMs have been calculated.

Results: Eleven/13 lesions showed both Gd-enhancement and MET-PET uptake (mean SUVmax: 4,5; range: 2,5-10,8; mean SUVmean: 2,7; range: 1,5-6,5; MTV: 21,7, range: 0,5-94,6). PET-VOIs were significantly larger than Gd-VOIs (median PET-VOI: 47 cc; range 9.9-2499; median Gd-VOI: 33cc; range 4,4-1097). Mean DICE index between PET and Gd-VOIs of the lesions was moderate (0.58 ± 015). The distances between PWI-CoMs and PET CoM were: mean Ktrans-PET distance = 2,79 mm ± 2,60; mean Vp-PET distance = 2,82 mm ± 2,32, mean rCBV-PET distance = 3,81 mm ± 3,52, without statistical difference between distances of PWI-CoMs from PET.

Conclusions: From this preliminary spatial relationship analysis, areas defined by MET-PET and Gd-MRI seem to characterize different aspects of the tumor (metabolic active lesions within not-enhancement and vice-versa) with moderate overlapping. No statistical difference between distances of PWI-CoMs from PET-CoM have been observed, although the heterogeneous histology of the lesions included in this study may have hampered the analysis. Our findings indicate that MET-PET and PW-MRI provide complementary information on brain tumor biology.