Abstract
Identification of safe and valid PET radioligands for metabotropic glutamate receptor, type 5 (mGluR5), is essential to measure changes in brain mGluR5 in neuropsychiatric disorders, to confirm central mGluR5 occupancy of drug candidates, and to guide dose selection for obtaining an optimum therapeutic window. Here we present the results of a first-in-human study assessing the safety and effectiveness of a novel PET radiopharmaceutical, 18F-3-fluoro-5-[(pyridin-3-yl)ethynyl]benzonitrile (18F-FPEB), for quantifying regional brain concentrations of mGluR5. Methods: Quantification of whole-body biokinetics was conducted in 6 healthy adults (3 men and 3 women). The radiation safety profile was estimated with OLINDA/EXM software. Subsequently, pairs of dynamic brain scans were obtained for 11 healthy men to identify optimal methods for derivation of regional distribution volume and binding potential and to determine the repeatability of measurement. Results: The whole-body effective radiation dose was approximately 17 μSv/MBq (62 mrem/mCi), with the gallbladder receiving the highest dose of 190 μSv/MBq. In brain studies, time–activity curves showed high accumulation in the insula/caudate nucleus, moderate uptake in the thalamus, and the lowest concentration in the cerebellum/pons. The plasma reference graphical analysis method appeared optimal for 18F-FPEB; it showed acceptable test–retest variability of nondisplaceable binding potential (<10%) and identified the highest nondisplaceable binding potential values (from ∼0.5 in the globus pallidus to ∼3.5 in the insula) for target regions. Safety assessments revealed no clinically meaningful changes in vital signs, electrocardiogram, or laboratory values. Conclusion: 18F-FPEB is safe and well tolerated, and its regional cerebral distribution is consistent with previous reports in the literature for metabotropic glutamate receptors. The repeatability of measurement suggests that 18F-FPEB is suitable for quantifying mGluR5 in humans.
- positron emission tomography (PET)
- neuroreceptors
- metabotropic glutamate receptor subtype 5 (mGluR5)
- 3-fluoro-5-[(pyridin-3-yl)ethynyl]benzonitrile (FPEB)
- 18F-FPEB
- radioligand
- human radiation dosimetry
Metabotropic glutamate receptors offer a plausible alternative to ionotropic receptors for drug development, as they are involved in most aspects of normal brain function through the modulation of synaptic transmission and cell excitability (1,2). Specifically, metabotropic glutamate receptor, type 5 (mGluR5), seems to be an important regulator of the glutamatergic system because of its hypothesized involvement in anxiolytic, antidepressant, and antiaddictive effects; memory processes; and neurotransmitter release (3). Thus, several pharmaceutical companies have focused significant resources on the development of drugs to act directly at, or modulate, mGluR5; the clinical use of these drugs, however, could be limited by central side effects such as amnesia and psychotomimetic symptoms (4–7). A clinically established PET radiotracer would be valuable for demonstrating differences in brain mGluR5 expression in healthy individuals and patients, for confirming central occupancy of novel mGluR5 drug candidates, and for guiding dose selection.
The preclinical evaluation of several recently developed mGluR5 radiotracers (8–11) supports mGluR5 as a quantifiable target by PET in mice and nonhuman primates. One such radiotracer was 18F-3-fluoro-5-[(pyridin-3-yl)ethynyl]benzonitrile (18F-FPEB), a potent, selective mGluR5 inhibitor with excellent properties as a central nervous system ligand (12). 18F-FPEB is a low-molecular-weight radiotracer made at high specific activity that selectively labels mGluR5 in vitro and in vivo and provides high-quality mGluR5-specific images in rhesus monkey PET studies (12). PET imaging with 18F-FPEB results in a large, long-lived specific signal with the highest uptake in the caudate and lowest uptake in the cerebellum in nonhuman primates, consistent with autoradiographic results for mGluR5 receptors (12). These data strongly indicate that 18F-FPEB is a suitable candidate for further clinical development. Two other radiolabeled mGluR5 antagonists, 11C-ABP688 and 18F-SP203 (13,14), have been evaluated in humans. Figure 1 shows the chemical structure of these radioligands and 18F-FPEB.
Structures of radiolabeled mGluR5 antagonists.
This study included an initial whole-body (WB) study to define the tissue and organ radiation dose associated with a single intravenous administration of 18F-FPEB. The study also developed a method of dynamic brain imaging with test–retest reliability-of-measurement analyses. The paradigm of dynamic PET imaging of the human brain facilitated the validation of 18F-FPEB for quantifying mGluR5 by characterizing its regional kinetics and its ability to generate robust outcome measures of nondisplaceable binding potential (BPND) and distribution volumes (VT).
We report here the first human study designed with the following goals in view: demonstrate the clinical and radiation tolerability of 18F-FPEB, with the hypothesis that it is entirely safe for human use; demonstrate dynamic human brain and plasma metabolites, with the hypothesis that they are feasible for human studies; and initially quantify human brain mGluR5, with the hypothesis that binding is consistent with the known metabotropic glutamate receptor regional distribution and that test–retest reliability is good.
MATERIALS AND METHODS
Subjects, Eligibility, and Overall Study Design
This was an open-label, 2-part, first-in-human pilot study conducted under an exploratory U.S. investigational new drug application that included and followed animal toxicology and safety studies of the unlabeled imaging agent and initial nonhuman radiation dosimetry performed at Merck Research Laboratories (15). All procedures were approved by the Johns Hopkins Medicine Institutional Review Board, and all participants signed informed consent forms.
A contract research organization, Parexel, Inc., accrued and medically managed the research subjects. A total of 17 (14 men and 3 women) nonsmoking healthy volunteers participated after demonstrating good physical and mental health, as determined by medical history, physical examination, vital signs, clinical laboratory tests, and a 12-lead electrocardiogram. Of these subjects, 6 (3 men and 3 women; mean age ± SD, 35 ± 7 y) participated in the radiation dosimetry study, and the remaining 11 men (mean age, 42 ± 7 y) participated in the brain imaging study.
To be included, the participants had to be subjectively healthy, have a body mass index of 19–29 kg/m2, be shorter than 195 cm (6 ft 5 in), and, if a woman, have a negative pregnancy test. Subjects were excluded from participation if they had any physical or mental disease or condition; were taking a medication that could interfere with the safety, tolerability, or biokinetics of the tracer; or, in part 2, failed to complete the MR head scan.
Participants in part 1, the WB radiation dosimetry study, received a single administration of 18F-FPEB. They were inpatients at the Parexel clinical research center and then underwent a single PET/CT scan consisting of 8–9 bed positions at the Johns Hopkins University PET Center.
Participants in part 2 were also inpatients. They received 2 administrations of 18F-FPEB on different days within 16 d (range, 2–16 d), except for 1 subject. After each administration, dynamic brain scans were acquired for 90 min, with arterial and venous blood sampling for metabolite and input functions.
Chemical Structure and Radiosynthesis of 18F-FPEB
18F-FPEB was synthesized at high specific activity using a slight modification of a published method (12); the modification involved using isocratic preparative high-pressure liquid chromatography (HPLC) and cesium carbonate instead of potassium/Kryptofix (Merck). After production of 18F-fluoride, target water was passed through a Chromafix 30-PS-HCO3 QMA cartridge (Macherey-Nagel GmbH and Co.) to trap the 18F-fluoride. The cartridge was eluted with aqueous cesium carbonate and then acetonitrile into a glass reaction vessel. The 18F-fluoride and cesium carbonate solution was azeotropically dried with multiple additions of acetonitrile at 120°C under a nitrogen flow. A dimethyl sulfoxide solution of FPEB precursor was added to the reaction vessel and microwaved for 4 min at 50 W. The mixture was diluted with water and applied to a semipreparative HPLC column (XBridge Prep OBD, C18, 5 μm, 19 × 50 mm; Waters) and eluted with 30% acetonitrile:70% H2O triethylamine buffer at pH 7.2 at an initial flow rate of 5 mL/min for the first 2 min and then 15 mL/min until product was observed and collected.
The 18F-FPEB fraction was collected in water and loaded onto a C18 Sep-Pak (Waters) and washed with 10 mL of 0.9% NaCl. The 18F-FPEB was eluted with 1 mL of ethanol followed by 10 mL of 0.9% NaCl, via a sterilizing 0.22-μm filter, into a sterile vial containing 4 mL of 0.9% NaCl.
18F-FPEB was demonstrated to meet all quality control criteria for a PET radiotracer including appropriate visual appearance and pH, radionuclidic, chemical, and radiochemical purities. The final radiotracer product was shown to be sterile and nonpyrogenic and lacking the presence of residual organic solvents from its preparation.
Clinical Safety Assessments
The safety and tolerability of 18F-FPEB was assessed and monitored through clinical and laboratory safety evaluations, including physical examinations, frequent vital sign measurements, 12-lead electrocardiograms, and relevant laboratory tests (hematology, chemistry, and urinalysis). For women, a urine pregnancy test for exclusion criteria was also conducted before any PET or MR scan. The patient was followed up by telephone at about 72 h and 14 d after administration of 18F-FPEB. In the dosimetry study, WB PET scans were used to estimate radiopharmaceutical biokinetics and regional radiation exposure. The outcome measures included effective dose and absorbed radiation doses to individual organs.
Effective dose was defined in 1991 by the International Commission on Radiological Protection in its publication 60 (16). Effective dose is calculated by summing the “tissue-weighting factor”–weighted organ-absorbed doses, which are a function of their assumed relative radiosensitivity for expressing carcinogenesis or genetic defects. The tissue-weighting factor given in publication 60 was a revision of that given in publication 26.
WB Imaging, Dosimetry, and Subject Exposure Estimates for 18F-FPEB
Before the radiation dosimetry studies of 18F-FPEB, tracer biodistribution data from PET in 3 male and 2 female rhesus monkeys were used to forecast the initial human radioactivity dose administered for the WB human biokinetic studies (15). The single-study dosage limit for men and women was based on the calculated dose to the upper large intestine, which appeared to receive the highest dose per administration in nonhuman primates. The maximum single-study activity of 185 MBq (5 mCi) was estimated to provide an exposure of 33 mGy and 37 mGy (per injection) to the upper large intestine of men and women, respectively, and 2.5 mGy per WB study. The maximum amount of 18F-FPEB administered was set at 2 scans with approximately 185 MBq (5 mCi) to limit exposure to the large intestine, which was predicted to be the dose-limiting organ.
The subjects were positioned supine in a Discovery VCT PET/CT scanner (lutetium yttrium orthosilicate PET with a 15.7-cm axial field of view; GE Healthcare). To correct for photon absorption and scatter, as well as to provide anatomic information as a reference, a low-dose CT scan (64 slices; pitch, 0.984; tube voltage, 120 kVp; tube current, 20–200 mA; rotation time, 0.5 s) was performed before tracer administration. After intravenous injection of a mean of 188 MBq (range, 178–197 MBq) of 18F-FPEB through an indwelling catheter, a series of 4 sequential WB PET scans (bed position overlap, 11 slices (36 mm); coincidence window, 6.5 ns; random correction, singles; energy window, 425–650 keV) were obtained at 4 time points (0, 10, 30, and 60 min). The duration of the dosimetry scans lasted approximately 105 min when there were 9 bed positions.
Each scan covered the body from the head to the mid thigh and consisted of 8–9 bed positions scanned for 1 min (×8 = 8 min), 2 min (×8 = 16 min), 3 min (×8 = 24 min), and 5 min (×8 = 40 min) in 3-dimensional mode. The data were then reconstructed in a 128 × 128 matrix (slice thickness, 3.75 mm) and corrected for photon attenuation and scatter using information from the corresponding CT scan.
Curve fits extrapolated to a time of infinity were used to calculate the time–activity integrals. The OLINDA/EXM program (17) was used to estimate the absorbed radiation doses. The effective dose served as the primary measure of radiation safety.
Brain Imaging Study and Reconstruction
MR Imaging
Participants who underwent dynamic brain PET also underwent structural MR imaging of the brain. All imaging was done on a 3-T Magnetom Trio scanner (Siemens Medical Solutions) equipped with the 3T Head Matrix Coil. T1-MPRAGE (magnetization-prepared 180° radio-frequency pulses and rapid gradient-echo) and T2-weighted images were acquired, with all brain images read by an attending neuroradiologist to rule out clinical pathology.
PET Procedures
All brain-only scans were obtained on the second-generation High-Resolution Research Tomograph (CPS Innovations, Inc.), a lutetium oxyorthosilicate–based 2.5-mm-resolution dedicated brain PET scanner. Subjects had 1 venous catheter for the radioligand and 1 arterial catheter for input functions. The subjects were positioned in the scanner with their heads restrained by a custom-made thermoplastic mask. A 6-min transmission scan was acquired using a rotating 137Cs source for attenuation correction. Dynamic PET acquisition was performed in 3-dimensional list mode for 90 min after an intravenous bolus injection of 18F-FPEB. Arterial blood samples were collected, and selected samples were analyzed by HPLC for radioactive metabolites.
Reconstruction
High-Resolution Research Tomograph dynamic PET images were reconstructed using an iterative ordered-subset expectation maximization algorithm with 6 iterations, 16 subsets, data mashing (span) of 3, a maximum ring difference of 67, and corrections for attenuation, scatter, and dead time. This is the standard reconstruction algorithm for the High-Resolution Research Tomograph. The reconstructions were performed on a specialized 32-node (64 processors) cluster (IBM). The following 30-frame schedule was used from the list-mode collection: four 15-s, four 30-s, three 1-min, two 2-min, five 4-min, and twelve 5-min frames, for the 90-min scan.
Blood Sampling and Plasma Metabolite Collection
Total Radioactivity Plasma Curve
Arterial plasma samples (1.5 mL) were withdrawn manually at approximately 5, 10, 15, 20, 25, 30, 35, 40, 45, 50, 55, 60, 65, 70, 75, 80, 85, 90, 95, 100, 105, and 110 s and at 3, 4, 5, 8, 10, 15, 20, 30, 40, 60, 80, and 90 min; counted in a γ-counter; and cross-calibrated with the PET scanner.
18F-FPEB Metabolites
Additional 6-mL radial arterial samples taken at 0, 2, 5, 10, 20, 30, 60, and 90 min for assessing the fraction of unchanged 18F-FPEB in plasma were analyzed by HPLC using a general method developed for PET radioligands (18).
PET Data Analysis
WB Dosimetry Study
The injected dose of radioactivity was determined by dose calibrator after subtraction of the residual syringe activity. Because no voiding occurred during the WB scans, the entire injected activity was recovered in the PET/CT measures. Segmentations of each organ for volumes of interest (VOIs) were performed by a threshold method implemented with a Johns Hopkins University tool based on MATLAB (MathWorks) that allowed voxels with radioactivity above the current threshold to be displayed in a color map and then segmented into a VOI after elimination of unwanted connections to the other organs in all orthogonal slices (transverse, sagittal, and coronal). For the liver, cerebral cortex, bone marrow, spleen, thyroid, lung, heart, and kidneys, the optimum threshold was applied to the single time frame that most clearly showed the organ. The resulting VOIs were transferred to the remaining frames. For the urinary bladder, gallbladder, stomach, and small intestine, VOIs were determined on more than a single frame. The VOIs were applied to generate time–activity curves for a total of 12 organs. The standardized uptake values of each organ were calculated by dividing the measured activity concentration (Bq/mL) by the injected activity (Bq) per gram of body weight and then expressed as a percentage (i.e., definition of percentage standardized uptake value).
The percentage injected activity per organ was multiplied by reference standard organ masses (19), divided by the amount of activity administered, and then fit using SAAM II software (20). Time integrals of activity were entered into OLINDA/EXM software (version 1.1) (17) using the adult male model.
Accumulation of activity was observed in the bladder and integrated until 100 min, when accumulation was assumed to be negligible. The number of disintegrations in the remainder of the body was assumed to be equal to 100% of the activity administered, integrated to total decay of 18F minus the disintegrations in other body organs. The observed bladder doses were integrated since no voiding occurred, which probably yielded conservative dose estimates.
Brain Imaging Study
Cortical VOIs were automatically defined using Freesurfer software (21) and combined into standard regions, including frontal, temporal, parietal, and occipital cortices; fusiform gyrus; cingulate; and insula. Subcortical regions were defined with FIRST software (22) and manually adjusted on individual MR images. Subcortical regions included putamen, caudate nucleus, globus pallidus, thalamus, hippocampus, and amygdala. VOIs were transferred from MR images to PET spaces following the MR imaging–to–PET coregistration parameters given by SPM5 (23,24) to obtain time–activity curves.
Mathematic Kinetic Modeling
We prespecified that the study would identify the compartmental model that most precisely described the kinetics of 18F-FPEB in the brain using metabolite-corrected plasma time–activity curves. We expected that the VT of the nondisplaceable compartment (25) could be estimated accurately in cerebellum (8,26,27) or in white matter (12) for radioligands of mGluR5.
A set of standard plasma input methods was used to identify the optimal method for derivation of VT for 18F-FPEB. The following 4 methods were tested: a 1-tissue-compartment model with 3 parameters (K1 and k2′, defined by Koeppe et al. (28), and v0, the tissue vascular volume); 2-tissue-compartment models with 5 parameters (K1, k2, k3, k4, and v0, defined by Innis et al. (25)), with and without constraining the K1/k2 ratio (nondisplaceable VT (28)) to the cerebellum estimate (TTCM and TTCM-C, respectively); and plasma reference graphical analysis (PRGA) (30). In TTCM and TTCM-C, BPND was given as the k3/k4 ratio. In the data analyses, metabolite-corrected plasma time–activity curves were obtained by applying the metabolite-corrected input function, given by HPLC analysis, to total plasma time–activity curves after interpolating at plasma sample times using the piecewise cubic Hermite interpolation implemented in MATLAB.
BPND maps were generated using PRGA, spatially normalized to a SPM5 standard brain using SPM5 (23), and averaged across subjects to generate averaged BPND maps of 18F-FPEB.
Test–Retest Repeatability of Dynamic Brain Scans
Regional test–retest variability was calculated as follows and expressed as a percentage:
RESULTS
Radiochemical Yield at End of Synthesis
The mean (±SEM) radiochemical yield was 1.7 ± 0.292 GBq (45.6 ± 7.39 mCi), and the mean specific activity at the end of synthesis was 6,000 ± 823 GBq/μmol (16.2 ± 2.2 Ci/μmol).
Subjects
WB Radiation Dosimetry Study
Three men and 3 women (mean age ± SD, 35 ± 7 y; range, 24–45 y) were scanned on the Discovery VCT. The mean (±SD) 18F-FPEB dose was 188 ± 7 MBq (5.08 ± 0.18 mCi), with a mass of 0.10 ± 0.06 μg, specific activity of 5.3 × 105 ± 2.9 × 105 MBq/μmol, and a mass dose of 1.57 ± 0.96 pg/kg.
The free fraction was determined using methods developed by Hamill et al. (12). Plasma protein binding of 18F-FPEB was evaluated at tracer levels in human plasma. All calculations of protein binding were corrected for nonspecific binding of the tracer to the centrifugal filters. The average free fraction was determined to be 8.10% ± 0.67%, ensuring delivery of tracer to the brain.
Brain 18F-FPEB PET Study
For the first scan, the 11 subjects were given an average dose (±SD) of 181 ± 14 MBq (4.90 ± 0.38 mCi), with a mass of 0.15 ± 0.10 μg, specific activity of 3.7 × 105 ± 2.3 × 105 MBq/μmol, and a mass dose of 1.8 ± 1.3 pg/kg. For the second scan, the average dose was 179 ± 4 MBq (4.85 ± 0.11 mCi), with a mass of 0.23 ± 0.19 μg, specific activity of 2.9 × 105 ± 1.9 × 105 MBq/μmol, and a mass dose of 2.7 ± 2.5 pg/kg.
The dose, mass, and specific activity of test and retest scans were not statistically different (P > 0.1, paired t test). Means ± SDs for both test and retest scans were as follows: dose of 181 ± 10 MBq (4.88 ± 0.28 mCi), mass of 0.19 ± 0.15 μg, specific activity of 3.3 × 105 ± 2.1 × 105 MBq/μmol (8,963 ± 5,683 mCi/μmol), and mass dose of 2.2 ± 1.96 pg/kg.
Test–retest evaluation was performed for only 5 subjects because of technical problems with the HPLC (4 scans), uncorrectable head motion (1 scan), and withdrawal from the retest scan (1 subject). A total of 16 successful scans (i.e., those without the aforementioned problems) from 11 subjects (thus denoted as n = 11) were used for method evaluation and derivation of regional VT and BPND.
Safety
There were no significant changes in vital signs, laboratory values, or electrocardiograms during the PET scans that could be attributed to the radiotracer. Mean systolic blood pressure was 122 mm Hg (maximum, 147 mm Hg), mean diastolic blood pressure was 68 mm Hg (maximum, 93 mm Hg), and mean heart rate was 64 beats/min (maximum, 92 beats/min). Mean QTc was 397 ms. (QTc is the QT interval corrected for heart rate, used to assess the risk for tachyarrhythmia. Normal QTc is ≤400 ms.) No subjects experienced more than a 28-ms change from baseline in QTc, and no QTc values were greater than 441 ms.
Absorbed Doses for WB Radiation Dosimetry
Coronal WB PET images are shown in Figure 2A. The mean number of disintegrations (Bq-h/Bq administered) of 18F-FPEB (n = 6) calculated from WB PET images are summarized in Table 1. All 18F-FPEB data were well fit with 1 or 2 exponential functions. The mean radiation dose estimates (n = 6) in the various organs is reported in Table 2. Most organs received exposures of about 7.7–44 μSv/MBq (28–160 mrem/mCi). VOI analyses of an average time–activity curve of non–decay-corrected 18F-FPEB (n = 6) revealed that most of the 18F activity was seen in the hepatobiliary transit through the liver and gallbladder into the small intestine, with gradual retention of radioactivity in the urinary bladder (Fig. 2B). The gallbladder received the highest radiation dose, approximately 190 μSv/MBq (708 mrem/mCi). The urinary bladder received approximately 52 μSv/MBq (1,790 mrem/mCi). The effective dose was approximately 17 μSv/MBq (62 mrem/mCi).
(A) 18F-FPEB WB PET images 0–10, 10–30, 30–60, and 60–90 min after injection of 33-y-old man (73 kg) with 177.6 MBq (4.8 mCi) with specific activity of 555 GBq (15.0 Ci)/μmol and FPEB mass of 0.07 μg. Gallbladder was critical organ for ranges of percentage standardized uptake value (SUV), decay-corrected to individual scan start times but not to injection time. Images are displayed using same scale for direct comparison of percentage SUV between images. (B) Tissue time–activity curves of non–decay-corrected 18F activity using SUVs for various organs. Values are mean (±SEM) SUVs. GB = gallbladder; UB = urinary bladder; SI = small intestine; LV = liver; BR = brain; KD = kidney; BM = bone marrow; HT = heart; SP = spleen; TH = thyroid.
Output from OLINDA for Residence Times of 18F-FPEB from WB PET Images
Output from OLINDA for Radiation Doses of 18F-FPEB from PET Images and OLINDA/EXM
Dynamic Brain Imaging and mGluR5 Quantification
Radioactive Metabolites and Plasma Time–Activity Curves
Total radioactive metabolites were about 60% at 10 min after injection, about 80% at 30 min, and about 5% at the end of the scan (90 min) (Fig. 3). Plasma time–activity curves peaked within 1 min of injection and decreased monotonically (Fig. 3).
Line plots (means with SD bars, n = 11) of total radioactive metabolite fractions in plasma (A), and total and metabolite-corrected plasma time–activity curves expressed in standardized uptake value (SUV) (B). Inset shows first 7 min to visualize initial portions of time–activity curves.
Tissue Time–Activity Curves
Figure 4 shows decay-corrected tissue time–activity curves in percentage standardized uptake value for selected brain regions. Time–activity curves of good-retention regions (e.g., insular cortex and caudate nucleus) peaked at around 20 min and then gradually declined. Time–activity curves of medium-retention regions (e.g., thalamus) peaked within 10 min of the injection and declined monotonically. Time–activity curves of low-retention regions (e.g., cerebellum and pons) peaked within 10 min of injection and decreased monoexponentially.
Average percentage standardized uptake value time–activity curves for 18F-FPEB (n = 11) of representative regions. Data for thalamus include SD bars (SD in thalamus is representative of magnitude of error for all regions).
Selection of Methods for PET Outcome Variables
Akaike information criteria (38) supported TTCM-C over a 1-tissue-compartment model in all brain regions, including the pons, which showed the lowest accumulation of radioactivity. TTCM yielded relatively unstable estimates of VT at 60 min in time-consistency tests (y = 1.1x − 1.1; R2 = 0.964, where x and y stand for VT values given by PET frames 0 to 90 and 60 min, respectively), compared with TTCM-C and PRGA. Therefore, TTCM might not be appropriate for 90-min scans. TTCM-C and PRGA showed comparable results in time consistency tests for VT (TTCM-C: y = 1.0x − 0.51, R2 = 0.987; PRGA: y = 1.0x − 0.86, R2 = 0.988) and for BPND (TTCM-C: y = 1.0x − 0.08, R2 = 0.964; PRGA: y = 1.0x − 0.01, R2 = 0.986). Therefore, it appeared that 60 min could be sufficient for TTCM-C and PRGA to yield stable estimates of VT and BPND (and thus could be considered a time to reach steady state). The cerebellum was used for the VT of the nondisplaceable compartment for TTCM-C, whereas the pons was used to calculate BPND with PRGA because it showed the lowest VT among tested regions.
Test–Retest Variability
TTCM-C and PRGA showed similar test–retest variability estimates of VT across regions (Fig. 5A). Test–retest variability estimates were lower than 20% in most regions except for the caudate nucleus. Test–retest variability estimates of BPND were less than the 10% criterion in all regions (Fig. 5B), except for the globus pallidus. PRGA showed slightly better test–retest variability on BPND across regions than did TTCM-C (paired t test: t12 = −4.96.13, P < 0.001). Intraclass correlation coefficients for test–retest across regions were as follows: TTCM-C VT was 0.96 and BPND was 0.82; PRGA VT was 0.96 and BPND was 0.91.
Histograms (means with SE bars) of regional estimates of test–retest variability (Eq. 1) for VT (A) and BPND (B) given by TTCM-C and PRGA. The 10% level, which is the generally accepted criterion for acceptable reproducibility, is shown by horizontal dotted line. Fr = frontal lobe; Tp = temporal lobe; Pa = parietal lobe; Oc = occipital lobe; Fs = fusiform gyrus; Cg = cingulate gyrus; In = insula; Hp = hippocampus; Am = amygdala; Pu = putamen; CN = caudate nucleus; GP = globus pallidus; Th = thalamus; PO = pons.
Regional VT and BPND Values
Regional VT and BPND are shown in Figure 6. VT, which includes not only binding but also blood–brain transport properties, was a secondary variable for 18F-FPEB because PRGA yielded robust estimates of BPND. Insula, temporal, and cingulate cortices showed high BPND values that could be visually identified on averaged BPND maps (Fig. 7).
Histograms (means with SE bars, n = 11) of regional value of VT (A) and BPND (B) given by TTCM-C and PRGA. Fr = frontal lobe; Tp = temporal lobe; Pa = parietal lobe; Oc = occipital lobe; Fs = fusiform gyrus; Cg = cingulate gyrus; In = insula; Hp = hippocampus; Am = amygdala; Pu = putamen; CN = caudate nucleus; GP = globus pallidus; Th = thalamus; PO = pons.
Transaxial (left) and sagittal (right) BPND images of 18F-FPEB (top) and matching MR images (bottom) in SPM standard space. Regions with high BPND values, namely insular (In), temporal (Tp), and cingulate (Cg) cortices, are indicated on coregistered MR images.
DISCUSSION
The present study confirmed that PET imaging with 18F-FPEB, prepared in high radiochemical yield and purity (12), seems safe and feasible for use in humans. The organ and effective doses were lower than those of our preclinical study (15); nonhuman primates had higher doses to the intestinal walls and testes and a lower dose to the gallbladder wall than did these human participants.
18F-FPEB has low radiation doses (16.9 μSv/MBq), comparable to clinically used 18F-ligands such as 18F-FDG (19–24 μSv/MBq) (39).
The total radiation dose for an 18F-FPEB administered activity of about 185 MBq (5 mCi) per scan for 3 scans is about 9.4 mSv (0.93 rem). The addition of about 12 mSv (1.20 rem) from the CT scans leads to a total of about 21.4 mSv (2.13 rem), which is far less than the 50-mSv (5-rem) annual effective dose limit for WB imaging, as per title 21 of Code of Federal Regulations, part 361.1. WB dosimetry of 18F-SP203 showed a higher dose to red marrow (30.9 μSv/MBq) than that from other 18F-labeled tracers, whereas the effective doses were similar (17.8 μSv/MBq) (14).
The PET imaging method for dosimetry used in the current study, segmentation of each organ by threshold, has a strong advantage for the gallbladder, the urinary bladder, and other gastrointestinal organs over the CT-based traditional method, since the filling rate of pharmacokinetic uptake and the anatomic position will change over time with marked distension or movement (Wong lab, unpublished data).
The metabolism of 18F-FPEB was not atypical, with the percentage metabolism in plasma being 75% and 80% at 30 and 90 min, respectively, and no lipophilic metabolites, allowing a concrete characterization of the radioactivity plasma input function, although no preclinical studies of brain metabolites have been conducted. We confirmed that it is possible to routinely achieve specific activities at the end of synthesis of 2.2 × 105 to 2.96 × 105 MBq/μmol (6–8 GCi/μmol) and an average injected dose of about 185 MBq.
The evaluation of kinetic models indicated that PRGA might be a useful method for quantifying VT, but it is premature to choose between PRGA and TTCM-C. Because the lowest VT was in the pons, it may be an appropriate reference region for obtaining BPND with PRGA, whereas cerebellum is widely used for 11C-ABP 688 (8). Volumes of distribution kinetic modeling using PRGA demonstrated that the VT of the nondisplaceable compartment and BPNP is consistent with that for human mGluR5. Most important, our test–retest variability in 5 human subjects was quite favorable and was less than 10% for most regions for BPND, in contrast to another mGluR5 PET radiotracer (40). Slightly higher test–retest variability estimates that we observed on VT—including blood–brain barrier–related parameters—may not be a real issue in clinical applications of this radioligand when the preferred PET variable (BPND, not including these parameters) shows excellent reproducibility.
A comparison of FPEB with SP203 and ABP688 has been previously published (37). The high test–retest variability in humans with 11C-ABP688 may make it a less than ideal ligand for receptor quantification (41). A possible disadvantage of FPEB may be the low radiochemical yield, although the amounts of tracer that can be synthesized are sufficient for human studies and, at Johns Hopkins University, are more than adequate for routine human PET studies.
Although mGluR5 occupancy was not evaluated in this pilot study, studies in nonhuman primates demonstrated that a 10 mg/kg dose of 3-[(2-methyl-4-thiazolyl)ethynyl] pyridine, an mGluR5 antagonist, nearly completely blocked regional brain uptake of this radiotracer in rhesus monkeys (12). Measurement of mGluR5 occupancy with therapeutic candidates is important and should be feasible with 18F-FPEB. The use of mGluR5 human PET tracers for drug occupancy is increasingly relevant. For example, Novartis’ AFQ056 (mGluR5 antagonist) has shown improvements in several Fragile X syndrome behavioral symptoms; STX 107, an mGluR5 blocker, for early testing of Fragile X, is being developed by Seaside Therapeutics; and RO4917523, another mGluR5 antagonist, is being developed by Roche. Given the good test–retest variability, ease of synthesis, and longevity of the 18F label, there are considerable advantages in using 18F-FPEB to study neuropsychiatric conditions and facilitate future drug development studies that include receptor occupancy.
CONCLUSION
PET scanning with 18F-FPEB was shown in this first-in-human study to be safe and well tolerated, as all postscan clinical safety tests remained unremarkable. Regional brain uptake was consistent with known mGluR5 sites, and within-subject test–retest variability for most regions was favorable (<10%) for most brain regions. The initial quantification of BPND values and excellent test–retest estimates strongly support the use of 18F-FPEB in clinical imaging studies of mGluR5.
DISCLOSURE
The costs of publication of this article were defrayed in part by the payment of page charges. Therefore, and solely to indicate this fact, this article is hereby marked “advertisement” in accordance with 18 USC section 1734. This study was supported in part by NIH grants R33 MH66623P and K24DA000412 and by NIBIB, NIDA, NIAAA training grant 5T32EB006351-05 for clinician scientists in imaging research. Data from this study were presented in part at the Society of Nuclear Medicine annual meeting in San Antonio, Texas, on June 4–8, 2011. No potential conflict of interest relevant to this article was reported.
Acknowledgments
We thank the staff members of the Johns Hopkins PET Center for their radiochemical syntheses and PET data acquisitions; William Willis, Ayon Nandi, Emily Gean, Maria Guevara, Anil Kumar, Stephen Condouris, and Maria Thomas for technical support and coordination; Andrew Crabb and Arman Rahmin for High-Resolution Research Tomograph reconstruction and physics issues; and Drs. Igor Grachev and Richard Hargreaves from Merck for their collaboration and scientific advice. Changlae Lee and Hee-Joung Kim of Yonsei University, Wonju, Korea, contributed to the radiation dosimetry calculations. Parexel, Inc., accrued and cared for the subjects and performed most of the pharmacologic safety studies.
Footnotes
Published online Feb. 12, 2013.
- © 2013 by the Society of Nuclear Medicine and Molecular Imaging, Inc.
REFERENCES
- Received for publication May 8, 2012.
- Accepted for publication July 27, 2012.