Participants
Seven healthy male volunteers (age, 27.4 ± 7.2 years; body weight, 68.3 ± 8.8 kg; values when applicable are expressed as mean ± SD hereafter) were included in this study. Subjects with current or past psychiatric disorders, substance abuse, current smoking, or organic brain disease were excluded based on their medical history and magnetic resonance imaging (MRI) findings. Subjects also underwent a physical examination and blood and urine analyses to exclude physical illnesses.
This study was approved by the Radiation Drug Safety Committee, and the Institutional Review Board of National Institutes for Quantum and Radiological Science and Technology, Chiba, Japan, and was carried out in accordance with the ethical standards laid down in the 1964 Declaration of Helsinki and its later amendments. After complete description of the study, written informed consent was obtained from all participants. The current study was registered with the Japan Registry of Clinical Trials (jRCTs031190054).
PET scan procedures
Radiosynthesis of 11C-MTP38 was carried out as described previously [14] (Supplementary Fig. 1). All PET scans were conducted with a Biograph mCT flow system (Siemens Healthcare, Erlangen, Germany), which provides 109 sections with an axial field of view (FOV) of 21.8 cm. The intrinsic spatial resolution of this device was 5.9 mm in-plane and 5.5 mm full-width at half-maximum (FWHM) axially. CT scan was performed prior to the emission scan for attenuation correction. Immediately after intravenous rapid bolus injection of 11C-MTP38 (injected dose, 349.2 ± 60.1 MBq; molar activities, 14.2 ± 6.8 GBq/μmol; mass, 7.9 ± 3.3 μg), three-dimensional list-mode emission data acquisition was started on a PET camera for 90 min. List-mode data were sorted and rebinned into sinograms with 33 frames of increasing duration from 10 s to 5 min (10 s × 6, 20 s × 3, 1 min × 6, 3 min × 4, and 5 min × 14). Signograms were reconstructed using a filtered back-projection algorithm with a Hanning filter (4.0 mm FWHM). All PET images were corrected for attenuation based on the CT images, for randoms using the delayed coincidence counting method, and for scatter using the single-scatter simulation method. A head fixation device was used to minimize the subject’s head movement during the PET measurements.
To evaluate the safety of 11C-MTP38, blood tests including a complete blood cell count and serum biochemistry were conducted before and 90 min after injection of 11C-MTP38.
Measurement of 11C-MTP38 in plasma
To obtain individual input functions to be used for the PET scan data analysis, arterial blood samples were taken manually 32 times after radioligand injection at 10-s intervals up to 120 s, 30-s intervals up to 3 min, 1-min intervals up to 10 min, and single 12-, 15-, 20-, 25-, 30-, and 10-min intervals up to 90 min after injection. Arterial blood sampling could not be technically performed in one subject. An aliquot of each blood sample was centrifuged to obtain plasma. Plasma and whole blood radioactivity concentrations were measured with an auto-gamma counter (WIZARD 1480, PerkinElmer, Waltham, MA). The plasma-free fraction was measured by ultrafiltration (Centrifree, Merck Millipore, Billerica, MA) in triplicate. The fractions of the parent and its radiometabolites in plasma were determined by HPLC from six samples in each subject (at 3, 10, 20, 30, 60, and 90 min). A plasma sample was deproteinized by adding an equivalent volume of acetonitrile, and an aliquot of the supernatant obtained by centrifugation was analyzed by radio-HPLC with a semi-preparative column (Capcell Pak C18 AQ, 5 μm, 10 × 250 mm connected with a guard column, Capcell Pak C18 AQ, 5 μm, 10 × 20 mm, Osaka Soda Co., Ltd., Osaka, Japan). An aqueous solution of acetonitrile (50%) was used as mobile phase at a flow rate of 4 mL/min.
MRI scan procedures
Structural T1-weighted images were acquired for all subjects with a 3-T MRI scanner (MAGNETOM Verio, Siemens, Germany). 3D volumetric acquisition of a T1-weighted gradient-echo sequence produced a gapless series of thin sagittal sections (TE/TR, 1.95/2300 ms; TI, 900 ms; flip angle, 9°; FOV, 250 mm; acquisition matrix, 256 × 256; slice thickness, 1 mm).
Brain image data processing
Head movement during the PET scan in each subject was corrected by registering all emission frames to the average image of the first 10 frames, with rigid transformation. Two authors (MK and CS) independently performed visual inspection of all PET images and confirmed that there was no apparent intra-frame or inter-frame motion, as well as no mis-registration between the CT and emission images in each subject. The motion-corrected PET images were co-registered to the corresponding individual T1-weighted MR images. For each T1-weighted image, surface-based cortical reconstruction and volumetric subcortical segmentation were performed with FreeSurfer tools (version 6.0.0; http://surfer.nmr.harvard.edu), and the following regions of interest (ROIs) were defined using its atlases [15,16,17]: frontal, temporal, parietal, occipital, anterior cingulate, posterior cingulate, and insular cortices, thalamus, caudate, putamen, globus pallidus, amygdala, hippocampus, cerebellar cortex, and pons. Motion correction, visual inspection of images, co-registration of PET images into MR images, and kinetic analyses were performed using PMOD® software ver. 3.8 (PMOD Technologies Ltd., Zurich, Switzerland).
Kinetic analyses of radioligand in the brain
Regional total distribution volume (VT), a sum of non-displaceable (VND) and specific binding (VS) distribution volumes, being equal to the tissue-to-plasma ratio of the radioligand concentration at equilibrium, was calculated with compartment models and graphical analyses using arterial input functions. Brain tissue and blood data from six subjects were used for the initial modeling evaluation. For compartment analyses, VT was determined with one- (1TCM) and two-tissue compartment (2TCM) models. For graphical analyses, VT was estimated by plasma input Logan plot [18] and Ichise multilinear analysis (MA1) [19]. For all kinetic analyses using input functions, the cerebral blood volume contribution to tissue radioactivity was fixed at 5%.
To investigate the minimal scan length required for reliable quantification of VT, we conducted the analysis by truncating PET data acquisition duration by every 10 min stepwise from 90 min down to 40 min.
Reference tissue model
To estimate the non-displaceable binding potential (BPND) of the radioligand in each brain region, we applied the original multilinear reference tissue model (MRTMO) [20] in ROI analysis and parametric imaging, with the latter allowing a voxel-based analysis. MRTMO is known to allow for BPND estimation with the smallest parameter estimation variability compared to other linear (MRTM and MRTM2) or non-linear models. Its limitation is a negative bias in the presence of noise in PET data, and its magnitude increases with noise. However, this negative bias is known to be minimal when the magnitude of BPND is small (< 1) [20].
Reference tissue models use time-activity data in a brain region devoid of specific binding components as input functions instead of arterial data, permitting quantitative measurements in all seven subjects. BPND was defined as
$$ {BP}_{\mathrm{ND}}=\left[{V}_{\mathrm{T}}\left(\mathrm{target}\right)/{V}_{\mathrm{T}}\left(\mathrm{reference}\right)\right]-1, $$
where VT (target) and VT (reference) are VT values of target and reference regions, respectively. We used the cerebellar cortex as a reference region, because only low levels of PDE7 mRNA are reported in the human cerebellar cortex [7]. BPND estimations were performed both with the indirect 2TCM method and with MRTMO.
SUVR method
Finally, we calculated (SUVR-1) to examine the possibility of quantification of 11C-MTP38 specific binding with shorter scan length without requirement of arterial blood data. We obtained (SUVR-1) ROI values from the summed PET images for 40–60, 50–70, 60–80, and 70–90 min normalized to the cerebellar cortex.
Statistical analysis
An optimal compartment model was chosen on the basis of the Akaike information criterion (AIC) [21], model selection criterion (MSC) [22], and goodness of fit assessed with F statistics [23]. In a model with better fitting, AIC showed lower values. A P value of less than 0.05 was considered significant for the F test. The standard error (SE) of kinetic parameters was given by the diagonal of the covariate matrix. Divided by the estimate of the parameter itself, SE was expressed as a percentage and used to assess parameter identifiability. A smaller percentage indicates better identifiability.
Pearson r and linear regression analyses were used to assess (1) correlations between VT values estimated with an optimal compartment model and those estimated with graphical analyses, (2) correlations between BPND values estimated by indirect kinetic method and those estimated with MRTMO on a ROI basis, (3) correlations between BPND values estimated with MRTMO on a ROI basis and those estimated with MRTMO on a voxel-by-voxel basis, and (4) correlations between BPND values estimated by indirect kinetic method and (SUVR-1) values using PET data from different scan intervals.
For one subject without arterial blood sampling, full kinetic analyses were not performed. The imaging data of this subject was used for investigation of the time-course of radioactivity and for reference tissue model analysis (BPND estimations with MRTMo on (1) a ROI basis and (2) a voxel-by-voxel basis, as well as correlations between BPND estimations by these two methods).