In vitro and In vivo Assessment of Suitable Reference Region and Kinetic Modelling for the mGluR1 Radioligand [11C]ITDM in Mice

Purpose This study aimed at investigating binding specificity, suitability of reference region-based kinetic modelling, and pharmacokinetics of the metabotropic glutamate receptor 1 (mGluR1) radioligand [11C]ITDM in mice. Procedures We performed in vivo blocking as well as displacement of [11C]ITDM during positron emission tomography (PET) imaging using the specific mGluR1 antagonist YM-202074. Additionally, we assessed in vitro blocking of [3H]ITDM at two different doses of YM-202074. As an alternative to reference region models, we validated the use of a noninvasive image-derived input function (IDIF) compared to an arterial input function measured with an invasive arteriovenous (AV) shunt using a population-based curve for radiometabolite correction and characterized the pharmacokinetic modelling of [11C]ITDM in the mouse brain. Finally, we also assessed semi-quantitative approaches. Results In vivo blocking with YM-202074 resulted in a decreased [11C]ITDM binding, ranging from − 35.8 ± 8.0 % in pons to − 65.8 ± 3.0 % in thalamus. Displacement was also markedly observed in all tested regions. In addition, in vitro [3H]ITDM binding could be blocked in a dose-dependent manner. The volume of distribution (VT) based on the noninvasive IDIF (VT (IDIF)) showed excellent agreement with the VT values based on the metabolite-corrected plasma input function regardless of the metabolite correction (r2 > 0.943, p < 0.0001). Two-tissue compartmental model (2TCM) was found to be the preferred model and showed optimal agreement with Logan plot (r2 > 0.960, p < 0.0001). A minimum scan duration of 80 min was required for proper parameter estimation. SUV was not reliable (r2 = 0.379, p = 0.0011), unlike the SUV ratio to the SUV of the input function, which showed to be a valid approach. Conclusions No suitable reference region could be identified for [11C]ITDM as strongly supported by in vivo and in vitro evidence of specific binding in all brain regions. However, by applying appropriate kinetic models, [11C]ITDM PET imaging represents a promising tool to visualize mGluR1 in the mouse brain. Electronic supplementary material The online version of this article (10.1007/s11307-019-01435-1) contains supplementary material, which is available to authorized users.

(1 ml/min). Activity (5.7 ± 2.4 MBq) was injected in a trace dose (0.86±0.33 µg/kg) keeping the cold mass within 2.0 μg/kg as no mass effect could be observed within this limit (Suppl. Fig. 1). PET data were acquired in list-mode format. Following the PET scan, a 10 min 80 kV/500 μA CT scan was performed on the same gantry for attenuation correction and coregistration purposes.
To assess the possibility of a suitable reference region, two treatment studies (blocking and displacement) using the high affinity mGluR1 antagonist YM-202074 (Merck, Germany) were executed. YM-202074 was administered with an i.v. bolus injection of 50 µl with the highest soluble dose (20 mg/kg) in saline. During the first paradigm (blocking study), YM-202074 was injected 2 min before [ 11 C]ITDM injection and the start of the PET scan. In the second paradigm (displacement study), YM-202074 was injected 30 min after [ 11 C]ITDM injection. One animal died during the baseline scan therefore it was excluded from all analysis.
In order to compare the noninvasive IDIF with the gold standard arterial input function, an AV shunt was surgically inserted into the femoral vein and artery prior to the PET scan as we previously described [2]. Once the animal was positioned onto the scanner, the shunt was connected to a peristaltic pump with tubing from the artery led through the Twilite detector [4] and ran in the pump. The tubing coming from the vein was connected on the output line of the pump, together with a second line for tracer injection. The arteriovenous shunt surgery failed in 2 mice and 1 mouse died during the scan. These animals were omitted from the analysis.

Image reconstruction and processing
Acquired PET data were histogrammed and reconstructed into 39 frames of increasing length (12x10s, 3x20s, 3x30s, 3x60s, 3x150s and 15x300s) using a list-mode iterative reconstruction with proprietary spatially variant resolution modelling in 8 iterations and 16 subsets of the 3D ordered subset expectation maximization (OSEM 3D) algorithm [5]. Normalization, dead time, and CT-based attenuation corrections were applied. PET image frames were reconstructed on a 128x128x159 grid with 0.776x0.776x0.776 mm 3 voxels. PMOD 3.6 software (Pmod Technologies, Zurich, Switzerland) was used for analysis and processing of the PET data. In order to spatially normalize the PET images, a [ 11 C]ITDM PET template was generated using the 90 min static PET images of the baseline scans, modifying the previously described procedure [3] to [ 11 C]ITDM. Since the blocking and displacement experiments were lacking part of the specific signal, the PET template approach could not be applied. Thus, for these data, individual CT images were spatially normalized to the CT image of the PET template adapting the previously described procedure [3]. All images were visually checked for accuracy following spatial transformation.
During the PET scans with AV shunt, the arterial input function was obtained using a 1 s sampling interval from the whole blood activity derived from the Twilite count detection coupled with the AV shunt. The activity measured with the Twilite was decay and background corrected as well as cross-calibrated with the PET scanner each experimental day. A three-exponential function was fitted to the decaying part of the input function in order to reduce the noise in the Twilite data. The whole body of the animal was in the microPET scanner's field of view (FOV), hence it was possible to extract the IDIF as a volume-of-interest (VOI) (threshold set to 50% of max) in the lumen of the left ventricle of the heart. Half of the VOI's maximum was found to be the optimal image threshold in order to obtain a stable volume across subjects. The ventricular region was delineated on an early time frame exhibiting maximal activity as previously described [2] and shown in Suppl. Fig. 2.
The [ 11 C]ITDM PET template was generated in the same space of the Waxholm atlas [6], thus VOIs were adapted from this atlas. Regional time-activity curves (TACs) were extracted for striatum, thalamus, hippocampus, cerebellum, and pons.

Metabolite correction
In order to generate a population-based metabolite correction to account for peripheral radiometabolism, we measured parent fractions in a cohort of WT mice (n = 3 per time point) at 5, 15, and 30 min p.i. The procedure was done adapting the previously described methodology [7] to [ 11 C]ITDM. Briefly, mice were injected with the radioligand via the lateral tail vein and blood collected via cardiac puncture at the respective p.i times. Following centrifugation of blood at 2377×rcf for 5 min, both plasma and residual fractions were counted in a gamma counter (Wizard 2 , PerkinElmer) in order to determine the plasma-to-whole-blood ratio. Next, equal amounts of ice-cold acetonitrile and 10 µl of cold reference (1 mg/ml) were added to the plasma samples.
After another centrifugation at 2377×rcf for 5 min, supernatant was separated from the precipitate and both fractions were counted in the gamma counter to calculate the plasma extraction efficiency (95.8 ± 3.0%). Next, 100 μl of supernatant were loaded onto a pre-conditioned reverse-phase (RP)-HPLC system (Kinetex, 150×4.6 mm, 5 μm HPLC column + Phenomenex security guard pre-column) and eluted with 0.1% TFA in H20 and acetonitrile (67:33 v/v) buffer at a flow rate of 1.5 ml/min. RP-HPLC fractions were collected at 0.5 min intervals for 8 min and radioactivity was measured in the gamma counter. The radioactivity was expressed as a percentage of the total area of the peaks. Blood spiked in vitro with 37 kBq of radiotracer indicated no degradation occurred during the workup (parent = 98.0 ± 0.9%).
In PMOD 3.6 software, average parent fraction values were fitted using a sigmoid curve as this model provided the best fit. This fitted model was then used for populationbased metabolite correction and each individual input function (AV shunt or IDIF) was corrected with this parent fraction model as well as for the plasma-to-whole-blood ratio, calculated based on the radiometabolite data. Thus, the metabolite-corrected plasma input function (Corr) was derived and compared to the uncorrected one (Uncorr). The radiometabolite-corrected plasma activity curve derived from the AV shunt measurement was used as gold standard for analysis and assessment of the noninvasive IDIF.

Kinetic modelling
To determine the relative performance of each model to fit the regional PET data using both metabolite-corrected plasma AV shunt as well as IDIF uncorrected as input functions, the goodness-to-fit was calculated using two different approaches: the Akaike Information Criterion (AIC) [8] was applied to 1TCM, 2TCM, and Logan plot,  washed in 50 mM Tris-HCl buffer on ice, followed by 5 dips into distilled water, and dried for 2 hours at room temperature. Lastly, slides were exposed on imaging plates (BAS-TR2025, Fujifilm, Japan) for 90 h. A phosphor imager (Fuji FLA-700 image reader) was used to detect radioactivity, which was quantified based on intensity values calculated using tritium standards (American Radiolabeled Chemicals Inc., USA).
Regional specific binding was calculated by subtracting nonspecific binding from total binding in triplicate using ImageJ software (National Institute of Health, USA).
[ 3 H]ITDM binding was measured in regions of interest, namely striatum, hippocampus, thalamus, pons, and cerebellum, manually drawn on each section.

Supplementary Tables
Supplementary Table 1