¹¹C-GSK189254: A Selective Radioligand for In Vivo Central Nervous System Imaging of Histamine H₃ Receptors by PET

Chemistry

Solvents were of high-performance liquid chromatography (HPLC) grade and were purchased from Aldrich or Merck. Labeling reactions were performed using anhydrous Sure/Seal solvent from Aldrich. The ¹¹C-CO₂ was produced using the GE Healthcare PETtrace 200 cyclotron. The preparation of ¹¹C-methyl iodide was achieved using a PETtrace MeI system (GE Healthcare). The labeling procedure was performed using a fully automated system (Synthia (23)) including the methylation step, HPLC purification, rotary evaporation, and labeled product formulation.

GSK189254 and its precursor for radiolabeling, 6-[(3-cyclobutyl-2,3,4,5-tetrahydro-1H-3-benzazepin-7-yl)oxy]-pyridinecarboxamide (1), were synthesized according to a published procedure (Fig. 1) (24).

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 1. 

Radiosynthesis of ¹¹C-GSK189254.

Radiochemical Synthesis of ¹¹C-GSK189254 (25)

¹¹C-GSK189254 was prepared by N-alkylation of the carboxamide precursor 1 using cyclotron-produced ¹¹C-methyl iodide (Fig. 1). ¹¹C-CO₂ was produced by the ¹⁴N(p,α)¹¹C nuclear reaction using a nitrogen gas target (containing 1% oxygen) pressurized to 150 psi and bombarded with 16.5-MeV protons. Subsequently, ¹¹C-CO₂ was converted into ¹¹C-CH₃I by catalytic reduction (Ni), which gave the ¹¹C-CH₄ intermediate, followed by gas phase iodination with iodine. The precursor 1 (1.0 mg) dissolved in dimethylsulfoxide (300 μL) was placed in a 1-mL glass vial. Tetrabutylammonium fluoride (20 μL) was added. The ¹¹C-CH₃I was passed in a stream of helium gas through the solution containing the precursor at room temperature. After the ¹¹C-CH₃I was delivered, the sealed vessel was heated at 130°C for 5 min and injected onto the semipreparative HPLC column (250 × 10 mm) (Sphereclone ODS(2) C-18; Phenomenex). HPLC purification was performed at a 10 mL/min flow rate, with a mobile phase consisting of acetonitrile and a solution of ammonium formate (50 mM) (42:58). The product fraction collected after approximately 7.7 min was evaporated to dryness and reformulated in 10 mL of 0.9% NaCl. Quality control was performed on a Sphereclone ODS(2) C-18 (250 × 4.6 mm) or a Zorbax SB-C18 (150 × 4.6 mm; Agilent) using acetonitrile and a solution of ammonium formate (50 mM) (48:52) as mobile phase at a flow rate of 3 mL/min or 1.0 mL/min, respectively. To confirm the radiopharmaceutical identity, the dose was also spiked with a GSK189254 standard.

Lipophilicity Measurements

Measurement of the partition coefficient of the radiolabeled compound was performed according to the method described by Wilson et al. using phosphate-buffered saline (phosphate buffer, 0.01 M; NaCl, 0.138 M; and KCl, 0.0027 M) at pH 7.4 and 1-octanol (26). The 1-octanol phase was washed with the buffer solution before any lipophilicity determination to reduce the potential errors due to hydrophilic impurities. Two measurements were performed from 2 different ¹¹C-GSK189254 productions, each consisting of 4 samples.

In Vitro H₃ Receptor Autoradiography

To assess the in vitro imaging properties of GSK189254, the preparation of ³H-GSK189254 for autoradiography (specific activity, 3 GBq/μmol [81 Ci/mmol]) was outsourced to Amersham Biosciences. Autoradiographic studies were based on a previously reported method (27). Frozen sagittal sections (20 μm) of porcine brain (Yorkshire/Danish Landrace; ∼40 kg) were thaw-mounted onto gelatin-coated slides, and the slides were stored at −80°C until the time of assay. The sections were incubated in assay buffer (50 mM Tris-HCl, pH 7.7, and 5 mM ethylenediaminetetraacetic acid) containing 1 nM ³H-GSK189254 for 60 min at room temperature (22°C). On anatomically adjacent sections, nonspecific binding was determined in the presence of 10 μM imetit. After incubation, all sections were rinsed 5 times for 3 min at 4°C in Tris-HCl buffer with the addition of 5 mM MgCl₂. The sections were then quickly dipped in distilled water at 4°C to remove buffer salts, and they were dried in a stream of cool air. Once dried, the sections were apposed to Hyperfilm (Amersham) and exposed for 3 wk. After development, the film was analyzed using ANAlysis imaging software (Olympus Soft Imaging Solutions). Adjacent sections were also stained with cresyl fast violet to allow for anatomic orientation.

Porcine PET Studies

All animal studies were performed in accordance with the Danish Animal Experimentation Act on a license granted by the Danish Ministry of Justice. Five Yorkshire/Danish Landrace pigs (∼40 kg) were housed singly in thermostatically controlled (20°C) and naturally illuminated stalls and were scanned under terminal anesthesia. The animals were medicated in advance with an intramuscular injection of 50 mg of midazolam and 250 mg of (S)-ketamine. The anesthesia was induced by an intravenous injection of 50 mg of midazolam and 125 mg of (S)-ketamine and maintained by ventilation with isoflurane (2%) in an O₂/N₂O mixture (1/2). At the end of the scanning session, the pigs were euthanized by intravenous administration of an overdose of pentobarbitone.

After anesthesia was induced, the left femoral artery and vein of each animal were surgically cannulated (Avanti, size 4–7 French; Cordis Corporation). Blood samples were collected from the femoral artery, and the radiolabeled and nonlabeled agents were injected in the femoral vein. Animals were placed supine in a Siemens ECAT EXACT HR tomograph, with the head immobilized in a custom-made holding device. During the study, blood pH, pCO₂, and pO₂ levels were monitored and maintained within the normal physiologic range. In addition, blood pressure and heart rate were recorded throughout the study. In each PET experiment, 150–450 MBq of ¹¹C-GSK189254, which corresponded to an injected mass dose of GSK189254 of less than 2 μg, were administered intravenously in the femoral vein as a 1-min bolus injection. PET and arterial blood sampling commenced at the start of the radioligand administration. Dynamic emission data were collected in 3-dimensional mode for 90 min (26 frames; 8 × 0.25 min, 4 × 0.5 min, 2 × 1 min, 2 × 2 min, 4 × 5 min, and 6 × 10 min). For each study, the baseline PET summed image was coregistered with a 12-parameter affine registration to the Landrace porcine brain atlas using a mutual information metric. T1-weighted MRI scans from 22 female Yorkshire/Landrace pigs were used to generate an atlas according to methods described previously for the Gottingen minipig (28). Subsequently, the transformation parameters were applied to each frame in the dynamic image of all the PET scans acquired in that study. All registrations were assessed by visual inspection, and then the regional regions of interest (ROIs) were applied to the dynamic images to generate regional time–activity curves.

The present work was performed using 5 different pigs and consisted of a total of 14 scans. In each pig, a baseline scan was acquired before the administration of the challenge. The unlabeled drugs were all coadministered with the bolus injection of the radioligand. One animal received a 50-μg dose of GSK189254 per kilogram.

Two animals underwent scans involving the coadministration of ciproxifan: In the first animal, a PET measurement scan was acquired after the administration of each of 3 escalating doses of ciproxifan (0.006, 0.06, and 0.6 mg/kg), and in the second animal a PET scan was obtained after the administration of a 2.0-mg dose of ciproxifan per kilogram.

Two animals underwent scans involving the coadministration of doses of GSK189254 aiming to achieve partial occupancy: the first animal underwent 2 scans for which 0.08 μg/kg and 0.25 μg/kg doses of unlabeled GSK189254 were administered. The second animal underwent a similar protocol but with coadministered doses of 0.07 μg/kg and 0.5 μg/kg for GSK189254.

For the challenge experiments, GSK189254 hydrochloride and ciproxifan hydrochloride were dissolved in saline and were intravenously injected, immediately followed by the ¹¹C-GSK189254 dose and a saline flush.

Plasma Input Function

Additional arterial blood samples were taken from the femoral artery to determine the total radioactivity concentration in plasma. The samples (a total of 40 per scan) were collected with an automated sampling system for the first 5 min and manually thereafter. In addition, a subset of 18 samples was analyzed for whole-blood activity. The total radioactivity in arterial plasma and blood samples was measured using a γ-counter (Cobra II Auto Gamma; Packard).

Metabolite Analysis

Arterial plasma analysis of ¹¹C-GSK189254 metabolism was performed in all pigs and for each scan. Arterial blood samples (5 mL) were collected at 5, 10, 20, 30, 60, and 90 min after tracer injection. The blood was centrifuged. An equal volume of acetonitrile was added to the plasma sample, and the resulting suspension was then centrifuged at 10,000 rpm for 5 min. The supernatant was removed and filtered through a 0.45-μm nylon syringe filter before HPLC analysis. The deproteinized plasma sample (typically, 1.0–2.0 mL) was injected and analyzed by HPLC with an ultraviolet detector (254 nm) and a flow-through radioactivity detector. Chromatographic separation used a Phenomenex Sphereclone ODS column (4.6 × 250 mm), a mobile phase of acetonitrile/ammonium acetate (50 mM; 50/50), and a flow rate of 2.0 mL/min. The fraction of unchanged tracer was determined by the integration of the peak corresponding to the ¹¹C-GSK189254 (retention time, 6.6 min) and expressed as percentage of all the radioactive peaks observed. The input function of arterial blood plasma concentrations corrected for the presence of radiometabolites was generated for all studies. The metabolite correction was performed by fitting a 1-exponential model to the measured parent fraction in plasma.

Kinetic Analysis

Radiochemistry

GSK189254 was labeled in the N-methyl position by treatment of the carboxamide 1 with ¹¹C-methyl iodide in the presence of tetrabutylammonium fluoride. After a 40-min irradiation time at 40 μA, typical syntheses provided 1.0–2.6 GBq (mean ± SD, 1.67 ± 0.45 GBq; n = 14) of ready-to-inject ¹¹C-GSK189254 at 35 min after the end of bombardment. The incorporation of the ¹¹C-methyl iodide was good, with radiochemical yields of up to 80% based on the trapped ¹¹C-methyl iodide.

Analytic HPLC demonstrated that the radiolabeled GSK189254 was more than 99% radiochemically pure (99.6 ± 0.47, n = 14), with a specific activity of 151 ± 148 GBq/μmol (n = 14). The injected dose of ¹¹C-GSK189254 was 350 ± 136 MBq (n = 14), and the mass dose of unlabeled GSK189254 was 1.28 ± 0.56 μg (n = 14).

Lipophilicity

The lipophilicity of GSK189254 was determined to predict its ability to passively cross the blood–brain barrier and the extent of the nonspecific binding signal. The measured logD_7.4 of GSK189254 was 1.74 ± 0.03 (n = 8).

In Vitro Autoradiography

The incubation of porcine brain sections with 1 nM ³H-GSK189254 revealed a regional heterogeneous distribution pattern. As illustrated in Figure 2, there was a high level of ³H-GSK189254 binding in the striatal, cortical, thalamic, hippocampal, and hypothalamic areas and substantia nigra, with minimal binding in the white matter areas, consistent with the results of previous studies conducted using alternative in vitro H₃ radioligands (31,32). Coincubation with the selective H₃ agonist imetit (10 μM) reduced ³H-GSK189254–specific binding by 90% in the striatal region.

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 2. 

(A) In vitro autoradiography of ³H-GSK189254 (1 nM) binding to H₃ receptors in sagittal sections of porcine brain. Cortex, putamen, caudate, hypothalamus, thalamus, hippocampus, corpus callosum, and cerebellum are indicated. (B) Nonspecific binding was determined in presence of 10 μM imetit.

PET in Pig

A sagittal summed (0–90 min) PET image for ¹¹C-GSK189254 under baseline conditions and a structural MR image of a porcine brain are depicted in Figures 3A and 3B. ¹¹C-GSK189254 readily entered the porcine brain; its overall brain uptake was approximately 9.0 %ID/L at 20 min after injection. Typical time–activity curves describing the kinetics of ¹¹C-GSK189254 in the different ROIs and the whole-blood and parent plasma curves are shown in Figure 4A. The regional brain distribution of ¹¹C-GSK189254 reflected the known distribution of the histamine H₃ receptors (1,2,32), with highest accumulation in the striatum, moderate accumulation in the cortices, and low accumulation in the cerebellum.

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 3. 

Sagittal MR (A) and PET (B–F) images of ¹¹C-GSK189254 binding in porcine brain summed from 0 to 90 min after injection. (A) Aligned MR image. (B) Baseline. (C) +0.006 mg/kg ciproxifan. (D) +0.06 mg/kg ciproxifan. (E) +0.6 mg/kg ciproxifan. (F) +2.0 mg/kg ciproxifan. Scale bar expressed as standardized uptake value.

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 4. 

Tissue time–activity curves describing kinetics of ¹¹C-GSK189254 for selected ROIs in porcine brain. (A) Tracer alone. (B) Coadministration of 50 μg/kg dose of unlabeled GSK189254. × = striatum; □ = thalamus; • = cortex; ▴ = hippocampus; ○ = cerebellum; * and dotted line = activity corresponding to unmetabolized ¹¹C-GSK189254 in plasma; ⋄ and dotted line = total blood activity.

Metabolite Analysis in Pig

Subsequent to the intravenous administration of ¹¹C-GSK189254, arterial plasma samples were analyzed for labeled parent and metabolites using radio-HPLC. The initial plasma sample displayed a major radioactive peak corresponding to the unmetabolized parent. The fraction of unchanged ¹¹C-GSK189254 in plasma was determined with radiodetection by integration of the peak corresponding to the parent compound as a percentage of the total of all the radioanalytes observed. This percentage decreased from 88% at 5 min after injection to 58% at 30 min after injection and then remained stable until the end of the scanning session (55% at 90 min after injection). The major radioactive metabolite observed was more polar than the parent compound and eluted at the solvent front. A second metabolite was observed and was identified as the product of the cleavage of the N-cyclobutyl bond. Although the brain permeability of this compound was not investigated, its potential CNS penetration would contribute only to the level of nonspecific binding because this molecule was shown to have no significant affinity at the histamine H₃ receptor (fpKi [recombinant functional assay] < 6.0).

Kinetic Analysis

At baseline, the AIC values for the striatum, cortex, and cerebellum were 141, 143, and 180, respectively, for the 1TC model and 144, 147, and 113, respectively, for the 2TC model. When a 2 mg/kg dose of ciproxifan was administered, the corresponding AIC values were 166, 148, and 158 for the 1TC model and 97, 93, and 62 for the 2TC model. Regional V_T values obtained using the 2TC model were similar to those obtained using the 1TC model for all regions. For the purpose of brevity, we present only the results from the 2TC analyses. The regional ¹¹C-GSK189254 V_T values in the baseline scan ranged from 5.5 (mL of plasma/cm³ of tissue) for the cerebellum up to 19.8 (mL of plasma/cm³ of tissue) for the striatum (Table 1).

The results of the time-stability assessment of V_T (Supplemental Fig. 1; supplemental materials are available online only at http://jnm.snmjournals.org) show that stable estimates of V_T were achieved for all regions by 60 min.

PET with Pharmacologic Challenges

To further characterize the ¹¹C-GSK189254–specific and –nonspecific binding signals, a self-blocking study was performed. Subsequent to a baseline scan, ¹¹C-GSK189254 was coadministered with a dose of unlabeled GSK189254 (50 μg/kg), which blocked the uptake of radiotracer in all H₃-rich regions, leading to a homogeneous distribution of radioactivity throughout the brain. The corresponding time–activity curves are presented in Figure 4B. To demonstrate the ¹¹C-GSK189254 binding selectivity, a similar blocking experiment was conducted using escalating doses (0.006, 0.06, 0.6, and 2.0 mg/kg intravenously) of ciproxifan. Ciproxifan is a known selective and structurally unrelated histamine H₃ antagonist (14) whose affinity for the histamine H₃ receptor expressed in the porcine cerebral cortex was measured at 9.3 ± 3.6 nM (22). This study showed a dose-dependent decrease of ¹¹C-GSK189254 uptake in the H₃-rich regions of the brain (Fig. 3). The time–activity curves presented in Figure 5 showed that this dose-dependent decrease of activity uptake was observed in all regions including the cerebellum. The quantification of the ¹¹C-GSK189254 binding confirmed that the administration of ciproxifan induced a significant change in the radioligand V_T in all ROIs (Fig. 6A; Table 1). Moreover, after the coadministration of a 2.0 mg/kg dose of ciproxifan, the V_T values were reduced to similar values throughout the brain, with V_T values ranging from 3.48 to 2.33 mL of plasma/cm³ of tissue for the striatum and cerebellum, respectively, demonstrating the specificity of ¹¹C-GSK189254 binding. There was also a significant decrease in the cerebellar V_T values between the baseline scan and the ciproxifan scans, indicating that the cerebellum is not a suitable reference region. Because no region could be regarded as a reference region, the nondisplaceable volume of distribution (V_ND) was determined from the cerebellum in the self-blocking study with a 50 μg/kg dose, when all the H₃ sites were occupied, as V_ND = 2.47. The ¹¹C-GSK189254 binding potential values were then calculated for each ROI () (Table 1). The mean occupancy values for the H₃-rich regions (cortex, striatum, thalamus, and hippocampus) were 27%, 55%, 81%, and 96% when ciproxifan doses of 0.006, 0.06, 0.6, and 2 mg/kg, respectively, were administered.

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 5. 

Time–activity curves after bolus injection of ¹¹C-GSK189254 in striatum and in cerebellum after administration of escalating doses of ciproxifan. × = baseline; ▴ = 0.006 mg/kg; ○ = 0.06 mg/kg; • = 0.6 mg/kg; □ = 2.0 mg/kg. Points are measured values, and lines are 2TC model fits.

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 6. 

(A) V_T estimates from 2TC model before and after treatment with escalating doses of ciproxifan. (B) Graphical interpretation of PET data obtained with 2TC model to estimate receptor occupancy values after administration of escalating doses of ciproxifan.

Similar results were achieved with the 1TC (data not shown). However, to circumvent the errors that between-subject variability may generate—and, most important, to be able to use ¹¹C-GSK189254 in higher species for which full blocking experiments may not be achievable and therefore the V_ND may not be determinable—a graphical method that obviated a reference region was developed on the basis of work described previously (33–35). This method relied on the availability of several regions with different levels of specific binding and assumed that the occupancy and the V_ND were constant for all these regions. Under these circumstances, the following equation can be derived,Eq. 1where and are vectors of the regional distribution volumes at baseline and after a challenge, is the nondisplaceable distribution volume, and Occ is the occupancy at the target site.

Equation 1 yields a simple graphical representation and linear regression to derive the occupancy (the gradient) and the (the x-intercept). This analytic approach is shown in Figure 6B for the ciproxifan challenge. The receptor occupancy values measured using this method were 33%, 61%, 86%, and 92% at 0.006, 0.06, 0.6, and 2 mg/kg, respectively.

Determination of In Vivo Affinity from Homologous Coadministration Experiments

The simultaneous nonlinear fit model (Fig. 7) was able to obtain stable estimates of the parameter values (as determined by independent evaluation from a range of different initial parameter values). The parameter estimates for the tracer affinity and the H₃ receptor concentrations are given in Table 2, and these demonstrate that consistent results were obtained for both subjects. GSK189254 was estimated to have a subnanomolar in vivo affinity (0.09–0.10 nM). The B_max estimates for cerebellum (0.5–0.98 nM), cortex (1.4–2.7 nM), and striatum (2.1–3.2 nM) were consistent with the baseline volume of distribution estimates.

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 7. 

Simultaneous nonlinear model fit of time–activity curves of striatum, cortex, and cerebellum for 1 subject for 3 scans. ○ = baseline scan; ▴ = 0.08 μg/kg coadministration; □ = 0.25 μg/kg coadministration.