Multimodal PET-MR imaging with the radiotracer O-(2-¹⁸F-fluoroethyl)-L-tyrosine (FET) and dynamic susceptibility contrast (DSC) perfusion weighted imaging (PWI) with a contrast agent bolus is a useful tool to delineate and characterize human glioma pre-therapeutically. These methods are assumed to measure local tumor cell density and increased capillary volume, respectively^1,2. In FET-PET, the averaged static tracer uptake is a reliable method to distinguish normal brain tissue from glioma, but uptake kinetics have also been shown to offer valuable information about the tumor’s grade^3,4. The same abilities have been reported for DSC-based parameters, i.e. relative cerebral blood volume (rCBV), peak height (rPH) and percentage signal recovery (rPSR) of the first pass of contrast material bolus⁵. It is still unknown whether there is an association between static and dynamic parameters of FET-uptake and the above mentioned DSC-measures. Hence, we analyzed the dependence of these parameters in untreated high- and low-grade glioma voxel-wisely on a single subject-level and between patients.

45 patients (57.8±16.8, 26 men) with suspected glioma (30 °IV (glioblastoma multiforme), 5 °III, 7 °II, 3 other) underwent a simultaneous MRI (structural sequences and DSC-PWI) and dynamic ¹⁸F-FET-PET (0-40 min p.i) examination on a clinical 3 T mMR Biograph scanner (Siemens Medical Solutions). Mean activity maps for 10-20 min and 30-40 min were generated and normalized to mean background tracer uptake (rFET(10-20min) and rFET(30-40min), respectively, Fig. 1). Furthermore, time activity maps were analyzed for slope and time to peak (TTP). PWI comprised DSC imaging during a bolus injection of 15 ml Gd-DTPA using single-shot GE EPI (TR = 1500 ms, TE = 30 ms, α=90°, 60-80 dynamics) after a prebolus of 7.5 ml. Leakage corrected rCBV⁶, rPH and rPSR maps were calculated, each normalized to a contralateral ROI in normal appearing white matter (NAWM; Fig. 1). Tumor volume was defined by a FET-uptake with a tumor-to-brain ratio of 1.6 or above (VOI_{Tumor, FET}). Within VOI_{Tumor, FET}, FET-PET and DSC-parameters were correlated voxel-wisely within each tumor and averaged values of the whole tumor were correlated across all patients. Hotspots of FET’s and DSC-parameter’s peak values were defined and also correlated with each other.

In 30 patients with glioma (3 low-grade, 27 high-grade) complete dynamic FET and DSC data could be obtained. Voxelwise comparison of static rFET- and DSC-parameters within VOI_{Tumor, FET} revealed highly significant positive correlations of early and late rFET uptake with rCBV and rPH (Tab. 1A and Fig. 2). Voxelwise correlation of rFET values with rPSR showed weak negative correlations (Tab. 1A and Fig. 2). Peak FET- and rCBV-/rPH-values within VOI_{Tumor, FET} had highly significant positive associations (Tab. 1B; Fig. 3A,B), whereas averaged values of the whole tumor volume correlated weaker and only during the early tracer accumulation time period (Tab. 1C). rPSR failed to show significant associations in both cases (Tab. 1B,C; Fig. 3C). Considering dynamic parameters of FET tracer uptake, averaged slope of the time-activity-curves correlated negatively with rCBV and rPH, however again, no association was found for rPSR (Tab. 2). No significant correlation was observed for mean TTP of the time-activity-curve and DSC-parameters (Tab. 2).

Static FET-tracer uptake is strongly associated with rCBV and rPH both within the glioma on a single voxel level and also across patients, whereas only weak invers correlations for rPSR have been observed. rPH thereby shows even slightly stronger correlations than the more widely used rCBV. This is of special interest since rPH is a parameter that can be readily obtained and is not susceptible to contrast enhancement in the tumor due to a disturbed blood-brain-barrier and therefore has not to be corrected for leakage effects (unlike rCBV). We also observed a negative correlation of rCBV and rPH with the velocity of tracer accumulation but not with TTP in glioma. This finding might indicate that TTP represents perfusion independent tumor tissue properties.

In glioma, static and dynamic FET-tracer activity is highly associated with DSC-parameters, such as rCBV and rPH, within the tumor on a single voxel level as well as averaged across patients. However other valuable parameters of FET-PET (i.e. TTP) and DSC-MRI (i.e. rPSR) show no clear dependency. This points to a complementary role of these two modalities in pre-surgical delineation and characterization of glioma. If the simultaneous acquisition of PET-MR data yield additional diagnostic benefits remains to be elucidated in further studies.