Abstract
184
Objectives Human gliomas exhibit considerable heterogeneity both inter-individually and within the same tumor. 18F-fluorethyltyrosine (FET) PET as well as MR dynamic susceptibility contrast (DSC) perfusion weighted imaging have been devised to characterize glioma biology and peak values have been shown to correlate well for different lesions. The introduction of integrated PET/MR imaging enhances spatial and temporal coregistration and enables the analysis of regional and voxel-wise correlation, which could be of use for identifying hot spots of higher malignancy or tumor recurrence. In this study, we present a thorough analysis of the association between perfusion MRI and FET PET in a cohort of untreated glioma patients receiving simultaneous FET-PET/MRI.
Methods Thirty patients with untreated gliomas (27 high-grade, 3 low-grade) underwent simultaneous MRI/PET examination on a 3 T mMR Biograph scanner (Siemens Medical Solutions). The MRI protocol comprised standard structural sequences and DSC imaging after a prebolus. Normalized FET SUV (10-20 min. and 30-40 min. p.i.) and relative cebral blood volume (rCBV), peak height (rPH) and percentage signal revovery (rPSR) maps were generated and coregistered. PET and DSC variables were compared within hotspots across patients and voxel-wisely within the tumor. Furthermore, tumor volumes were defined by a FET uptake tumor-to-brain-ratio of 1.6 and for each of the DSC-parameters at varying cut-off values and compared by applying Dice's similarity coefficient.
Results Inter-individual as well as voxelwise comparison of static FET-uptake and rCBV and rPH revealed strongly significant positive correlations, being highest for early FET uptake (10-20 min.) and rPH (inter-individual: r=0.561, p=0.001; voxel-wise: median r=0.398, p<0.001). Tumor volumes determined by FET and rCBV/rPH exhibited a median Dice coefficient of 0.37 for rCBV and 0.44 for rPH at optimal thresholds of 2.84 and 2.74, respectively. Only weak negative associations and low spatial congruence were observed for rPSR.
Conclusions The simultaneous acquisition of FET-PET and MR perfusion parameters was reliably feasible. Our results demonstrate a high inter- and intra-individual dependency and moderate spatial congruence of the acquired FET PET and DSC parameters. These findings indicate that the two modalities might provide both congruent and complimentary information on glioma biology, and lay the foundation for multiparametric imaging of brain tumors in the future.