[18F]CFT synthesis and binding to monoamine transporters in rats

Background We present the electrophilic synthesis of [18F]2β-carbomethoxy-3β-(4-fluoro)tropane [[18F]CFT] and the pharmacological specificity and selectivity of [18F]CFT for monoamine transporters in the brain and peripheral organs of rats. The human radiation dose is extrapolated from the animal data. Methods [18F]CFT was synthesized by electrophilic fluorination of a stannylated precursor by using post-target-produced [18F]F2 as a fluorinating agent. The ex vivo 18F-activity biodistribution of [18F]CFT in the brain of rats was studied by autoradiography. The binding of [18F]CFT to the monoamine transporters was studied using in vivo blocking experiments with dopamine transporter [DAT], norepinephrine transporter [NET], or serotonin transporter [SERT] inhibitors. In vivo animal positron emission tomography was used as a comparative method to determine tracer kinetics. Human radiation dose was assessed using OLINDA software. Results The radiochemical yield of [18F]CFT from the initial [18F]F-, decay corrected to the end of bombardment, was 3.2 ± 1.0%. The specific activity [SA] was 14.5 ± 3.4 GBq/μmol, decay corrected to the end of synthesis. Radiochemical purity exceeded 99%. DAT-specific binding was found in the striatum, locus coeruleus, and pancreas. NET-specific binding was found in the locus coeruleus. SERT-specific binding was not found in any of the studied organs. Effective dose equivalent [EDE] estimated for the standard human model was 12.8 μSv/MBq. Effective dose [ED] was 9.17 μSv/MBq. Conclusions Post-target-produced high-SA [18F]F2 was used to incorporate18F directly into the phenyl ring of [18F]CFT. The final product had high radiochemical and chemical purities and a high SA for DAT and NET studies in vivo. In periphery, [18F]CFT showed a specific uptake in the pancreas. EDE and ED corresponded well with other18F-radioligands.


Background
Dopamine transporters [DAT] are proteins located in the dopaminergic nerve terminals; they regulate the synaptic concentration of dopamine in the brain. Changes in the density and function of DAT in the brain are involved in many neurodegenerative and neuropsychiatric disorders, such as Parkinson's disease and schizophrenia. These changes can be imaged using positron emission tomography [PET].
Many radioligands, including [ 11 C]CFT [1] and several [ 18 F]F-labeled phenyl tropane analogs of cocaine [2], have been used to study dopamine reuptake in living subjects. However, none of these fulfill the requirements for an optimal radioligand for DAT imaging. [ 11 C]CFT suffers from slow kinetics compared with the short half-life of 11 C (T 1/2 = 20.4 min). The phenyl tropane analogs have a high or moderate affinity with other monoamine transporters (i.e., serotonin transporters [SERT] and norepinephrine transporters [NET]), or they undergo extensive metabolism. More recently, the new [ 18 F]F-labeled phenyl tropane analog [ 18 F]FE-PE2I has shown promise as a radioligand for DAT [3], despite its relatively fast metabolism [4].
Previously, electrophilic fluorination of a stannylated precursor, 2β-carbomethoxy-3β-(4-trimethylstannylphenyl)tropane (precursor) to achieve 2β-carbomethoxy-3β-(4-[ 18 F]-fluorophenyl)tropane [[ 18 F]CFT] (product) (see Figure 1) and preliminary evaluation of the radioligand in rats were reported by Haaparanta et al. [5] and by Bergman et al. [6]. A report on the ability of [ 18 F]CFT to reflect nigral dopaminergic cell loss in a rat model of Parkinson's disease [7] as well as a study comparing the brain accumulation, metabolism, and kinetics of [ 18 F] CFT and [ 18 F]CFT-FP [8] have shown that [ 18 F]CFT can be used to image DAT in rats. The suitability of [ 18 F] CFT as a radioligand for in vivo studies of DAT in humans has been evaluated [9], and [ 18 F]CFT has been used in human studies of Parkinson's disease [10][11][12][13][14][15], schizophrenia [16,17], and detached personality [18]. [ 18 F]CFT was proven to be a suitable radiotracer to image DAT by PET in humans due to its high target-tonontarget ratio and low metabolism [9] although [ 3 H] CFT has also been shown to have some affinity to SERT and NET [19,20]. The kinetics of [ 18 F]CFT are relatively slow, but the half-life of 18 F (T 1/2 = 109.8 min) allows equilibrium between specific and nonspecific binding during a human PET study.
A PET radioligand suitable for DAT must have a moderate to high specific activity [SA] to avoid saturation of transporter sites (with associated pharmacological effects) in patients. High SA can be easily achieved by nucleophilic fluorination. The 18 F label is usually incorporated into a molecule via an alkyl side chain, as in the case of [ 18 F] FE-PE2I [3] or [ 18 F]CFT-FP [21]. However, side chains are often prone to fast metabolism. A more metabolically stable configuration can be achieved by inserting 18 [23]. The latter method is more efficient than the former [24]. However, both production methods suffer from low SA. Post-target-produced [ 18 F]F 2 has 100-to 1,000-fold higher SA than the traditional methods [25]. Therefore, post-target-produced [ 18 F]F 2 offers the possibility of producing high-affinity radioligands through electrophilic labeling.
In the present study, we report the optimized electrophilic synthesis of [ 18 F]CFT with high SA and its quality assurance for clinical PET studies. The pharmacological specificity and selectivity of [ 18 F]CFT for monoamine transporters ex vivo are reported. The distribution of 18 F activity in the brain and peripheral organs of rats is reported ex vivo and in vivo in order to compare the methods in the determination of tracer kinetics. The human radiation dose is extrapolated from the animal data using organ level internal dose assessment [OLINDA]/EXM 1.0 software (OLINDA EXM, Vanderbilt University, Nashville, TN, USA) [26].

Chemicals and equipment
The stannylated precursor and the reference compound, 2β-carbomethoxy-3β-(4-fluorophenyl)tropane, were supplied by ABX (ABX GmbH, Radeberg, Germany). All other reagents that were purchased from commercial suppliers were either of synthesis grade or analytical grade and were used without further purification.

Synthesis of [ 18 F]CFT
The stannylated precursor (precursor, 250 to 500 μg, 0.6 to 1.2 μmol) was dissolved in a mixture of trichlorofluoromethane (Freon-11, 600 μl) and dry acetic acid (100 μl). [ 18 F]F 2 was bubbled through this mixture at room temperature. Freon-11 was evaporated using neon flow, and 600 μl of preparative HPLC eluent was added to the residue. With no further modifications, this solution was then loaded onto the preparative HPLC column.
[ 18 F]CFT was purified by semi-preparative HPLC using the system described in the 'Chemicals and equipment' section. The 3-ml fraction containing the [ 18 F] CFT was collected ( Figure 2). The fraction was evaporated to dryness with a vacuum evaporator, formulated into a 0.9% NaCl/0.1 M phosphate buffer (phosphate buffer pH 7, 3:2, v/v), and passed through the sterile filter into the end product vial.

Quality of [ 18 F]CFT
The 18 F activity, pH, and volume were measured from the end product. A sample from the end product was evaluated by the analytical HPLC system described in the 'Chemicals and equipment' section. Determinations of chemical purity, radiochemical purity [RCP], and SA were conducted by comparing HPLC retention times and peak intensities with a reference compound of known concentration and 18  . This corresponds to a 23 ± 10-nmol/kg (5 to 40 nmol/kg) administration of CFT, as calculated from the SA that was determined using the HPLC method at the time of injection. The animals were killed in a CO 2 chamber at 10 min (n = 2), 20 min (n = 3), 40 min (n = 6), or 120 min (n = 3) after injection of the tracer. The brains were rapidly removed, and a piece from the cerebellar cortex of each brain was dissected, measured for 18 F activity in a calibrated 3 × 3-in NaI(Tl) well counter (Bicron, Newbury, OH, USA), and weighed. After decay correction, these data, expressed as the percentage of injected dose per gram of tissue [%ID/g], were used to calibrate the absolute uptake of 18 F activity in autoradiographic brain images. The rest of the brain was frozen in isopentane, chilled with dry ice for sectioning with a cryomicrotome, and handled as described in the 'Digital autoradiography' section.
Other organs and tissue samples were rapidly dissected, weighed, and measured for 18 F activity. The decay-corrected uptake of 18 F activity in the organs and tissues was expressed as %ID/g.

Pharmacological studies
The specificity of [ 18 F]CFT binding to DAT in the brain in pretreated rats was assessed with a selective DAT antagonist, GBR12909 (5 mg/kg, Sigma-RBI, St. Louis, MO, USA). Selectivity was examined by injecting rats with a 5mg/kg dose of fluoxetine, a selective antagonist for SERT (Sigma-RBI, St. Louis, MO, USA), or with a 5-mg/kg dose of nisoxetine, a selective antagonist for NET (RBI, Natick, MA, USA). Binding profiles of the antagonists are presented in Table 1. GBR12909, fluoxetine, or nisoxetine dissolved in distilled H 2 O/0.9% NaCl (50:50, v/v, 2 mg/ml) were injected intravenously into rats 60 min prior to the injection of [ 18 F]CFT. The rats were killed by CO 2 inhalation 40 min after injection of [ 18 F]CFT. The brains and organs were handled as in the biodistribution studies. The regional distribution of 18 F activity in the brains of control rats (n = 6) and in the brains of rats that were pretreated with GBR12909 (n = 7), fluoxetine (n = 6), or nisoxetine (n = 6) were determined using digital autoradiography.
The imaging plates were scanned with the Fuji Analyzer BAS-5000.
The digital autoradiographic images were analyzed for count density (photo-stimulated luminescence per unit area [PSL/mm 2 ]) with a computerized image analysis program (Tina 2.1, Raytest Isotopenmessgeräte, GmbH, Straubenhardt, Germany). Regions of interest [ROIs] were drawn over the frontal cortex, striatum, locus coeruleus, and cerebellum, which were anatomically identified from the cryomicrotome sections using a rat brain atlas [27]. At least 10 sections were analyzed for each brain region, and the count densities for background areas were subtracted from the image data. PSL/mm 2 values were converted into %ID/g values as previously described in the 'Biodistribution studies' section.

Dosimetry
The animal %ID/g tissue data was extrapolated to humans using the percentage kilogram per gram method [28]. In this method, the animal %ID/g value is first multiplied with the animal's weight and then multiplied with the human organ weight/human weight ratio. Human radiation dose was estimated from these values using OLINDA/EXM 1.0 software [26].

Data analysis and statistical procedures
Statistical analyses were performed using the SPSS Statistics 17.0 software (SPSS Inc., Chicago, IL, USA). Means were considered significantly different when p < 0.05. Comparison of SAs was tested using Student's t test (paired, two samples for mean assuming unequal variances).
Effects of the pretreatments were tested using repeated measurement analysis of variance. Results are expressed as means ± SD for the indicated number of observations.

Synthesis and quality of [ 18 F]CFT
Electrophilic fluorination was applied to a stannylated precursor ( Figure 1) to synthesize [ 18 F]CFT product (n = 24). The initial [ 18 F]Factivity was 37 ± 3 GBq (range, 32 to 42 GBq), and the average synthesis time was 43 ± 3 min, including the synthesis of [ 18 F]F 2 , radiofluorination, and semi-preparative purification. In a semi-preparative HPLC purification, the [ 18 F]CFT fraction eluting at 10.5 min was collected ( Figure 2). Evaporation to dryness and formulation for injection took an additional 10 min.  Figure  3), the RCP exceeded 99% in all cases. The pH of the final product was 7. The final product was radiochemically stable for up to 6 h.
From selected batches (n = 19), the SA of the final product was determined by LC/MS. The SA of these batches measured by analytical HPLC was 14.9 ± 3.1 GBq/μmol. The SA measured by LC/MS was 18.2 ± 5.9 GBq/μmol. SAs calculated using the analytical HPLC method differed significantly from the SAs calculated using the LC/MS method (p = 0.04).

Biodistribution and pharmacological studies
The 18 F-activity accumulation in the striatum, locus coeruleus, frontal cortex, and cerebellum of control rats and monoamine inhibitor-pretreated rats 40 min after [ 18 F] CFT injection is presented in Table 2. Pretreatment with GBR12909 significantly reduced the [ 18 F]CFT uptake in the striatum (p = 0.006) and locus coeruleus (p = 0.02). In nisoxetine-pretreated rats, the [ 18 F]CFT uptake decreased significantly in the locus coeruleus (p < 0.005). Fluoxetine pretreatment had no effect on the accumulation of the 18 F activity in any region studied. Autoradiograms of representative brain sections from a control rat and from rats pretreated with GBR12909, fluoxetine, or nisoxetine are shown in Figure 4.
The region-to-cerebellum ratios at different time points from the ex vivo and in vivo studies are shown in Figure  5a,b, respectively. In ex vivo studies, the striatum-to-cerebellum ratio increased from 2.1 ± 0.2 at 10 min to 8.8 ± 2.2 at 120 min. The locus coeruleus-to-cerebellum ratio was 2.2 ± 0.3 at 10 min and 3.5 ± 1.6 at 120 min. The frontal cortex-to-cerebellum ratio was constant, ranging from 1.4 to 1.6 at all time points and with all pretreatments. All monoamine inhibitors used in this study significantly decreased the locus coeruleus-to-cerebellum ratio. Pretreatment with GBR12909 significantly reduced the striatum-to-cerebellum ratio.
The 18 F-activity accumulation in the peripheral organs and tissues after the injection of [ 18 F]CFT is presented in Table 3. The accumulation of [ 18 F]CFT-derived 18 F activity peaked at 20 min in most tissues and decreased thereafter (Figure 6). High levels of 18 F activity were recorded in the liver, kidneys, and spleen. In the liver, the highest uptake (8.3 ± 1.2% ID/g) was measured 120 min after injection of [ 18 F]CFT. 18 F-activity accumulation in the bone increased slowly with time, but it was still low (0.19 ± 0.15% ID/g) at 120 min.
Pretreatment of the rats with GBR12909 significantly decreased the uptake of 18 F activity in the pancreas (p = 0.03). No significant changes in the 18 F-activity uptake were recorded in the periphery of the rats pretreated with nisoxetine or fluoxetine.

PET imaging
The in vivo distribution and uptake of 18 F activity after [ 18 F]CFT injection in the brain of rats are presented in Figure 7. The uptake of 18 F activity in the striatum and cerebellum peaked during the first 5 to 10 min and decreased thereafter. The striatum-to-cerebellum ratio reached the maximum value of approximately 9 at 60 min. The time courses of the striatum-to-cortex and cortex-to-cerebellum ratios are shown in Figure 5b. The highest in vivo uptake of 18 F activity in the periphery was observed in the liver. The uptake increased in the liver during the first 60 min of PET imaging and was almost constant until the end of the scanning (i.e., until 120 min after injection).

Dosimetry
The effective dose equivalent [EDE] estimated for the standard human model was 12.8 μSv/MBq. Effective dose [ED] was 9.17 μSv/MBq.

Discussion
Post-target-produced [ 18 F]F 2 [25] offers a feasible method to produce PET tracers with high SA for neuroimaging through electrophilic fluorination. When using post-target-produced [ 18 F]F 2 , the SA depends on several factors, such as the initial amount of [ 18 F]Fand the amount of carrier fluorine used in the 19 F-18 F exchange reaction [25]. The SA of [ 18 F]CFT can potentially be increased by further optimizing this 19 F-18 F exchange reaction by decreasing the amount of carrier F 2 . However, in our hands, this resulted in a dramatic decrease in RCY. The amount of carrier F 2 used in this study (290 to 400 nmol) is a compromise, offering SA  and RCY that are high enough for several human PET studies from one production run. The radiochemical and chemical purities of the final product were verified by HPLC, and both were found to fulfill the requirements for human injection (i.e., RCP > 95.0% and the absence of unknown signals in the UV trace; Figure 3). Signals from compounds other than CFT observed in the UV trace were from the formulation solution.
For radioligands having very high SA, the sensitivity limitation of UV detection means that LC/MS can be the only method to determine the SA [29]. In the present case where [ 18 F]CFT is synthesized through electrophilic fluorination at high SA, we therefore compared the HPLC/UV absorption and the LC/MS technique for the determination of SA. In quantitative analyses, LC/MS is a faster and more sensitive method than HPLC combined with a UV detector. However, LC/MS is more easily affected by changes in the sample matrix. In this study, the SAs were significantly higher when determined by the LC/MS method than by the HPLC method (i.e., the concentration of CFT was lower when measured by LC/MS than by HPLC). This could be due to ion suppression in LC/MS [30]. The effect of ion suppression could be diminished by more extensive chromatographic separation or sample preparation prior to MS. However, both methods of analysis are suitable for analysis in the present case.
In the biodistribution study, the highest level of 18 F activity was found in the main excretory organs.
Additionally, the uptake in the bone was low even at 120 min, reflecting the good stability of the carbonfluorine bond. 18 F activity accumulated in the liver with the highest value at 120 min, indicating slow excretion and low metabolism. In all other organs studied, the 18 F activity peaked 20 min after injection (see Table 3). The results from the in vivo study were in accordance with  the ex vivo findings. [ 18 F]CFT has been reported to be relatively resistant to metabolism; in a microdialysis study of 120 min in rodents, the amount of unmetabolized [ 18 F]CFT was approximately 64% of the total 18 F activity [31].
In the periphery, non-neuronal DAT expression and DAT immunoreactivity have been found in the stomach, pancreas, and kidneys [32]. [ 18 F]CFT uptake in the pancreas, stomach, and kidneys was moderate in the present study. After GBR12909 pretreatment, the 18 F-activity uptake decreased significantly in the pancreas, indicating DAT-specific binding. In this study, no NET-or SERTspecific binding of 18 F activity was found in the periphery even though extraneuronal NET expression has been found in the lungs, adrenal medulla, and placenta [32].
In the brain, the accumulation of 18 F activity in the striatum was specific for DAT; it was significantly decreased with GBR12909 pretreatment (Figure 4). In addition to DAT, the striatum contains a low density of SERT, and NET is virtually absent [33]. Neither fluoxetine nor   nisoxetine pretreatment affected the 18 F-activity uptake in the striatum. High 18 F-activity accumulation was seen in the locus coeruleus, a brain region with high NET density as has been earlier shown by Burchett et al. [34]. A similar finding in monkeys using [ 3 H]CFT was observed by Kaufman and Madras [19]. The 18 F-activity accumulation in locus coeruleus was significantly decreased in rats pretreated with nisoxetine, which indicates that NET sites also bind with [ 18 F]CFT. Because NET is virtually absent in the striatum, [ 18 F]CFT is suitable for imaging striatal DAT sites. However, the accumulation of 18 F activity in locus coeruleus was also significantly decreased in rats pretreated with GBR12909. The affinity of CFT for DAT and NET is of the same order of magnitude (Table 1). With the relatively high dose of GBR12909 used in the pretreatment of the animals, it is evident that although the affinity of GBR12909 for NET is 50-fold less than for DAT, this dosage is high enough to displace [ 18 F]CFT from the NET sites in the locus coeruleus. It is noteworthy that the p value in the statistical analyses is higher for the locus coeruleus than for the striatum in the GBR12909 blocking study.
In the ex vivo study, the uptake ratios for the striatum, locus coeruleus, and cortex versus the cerebellum reached a maximum between 40 and 120 min. In the in vivo study, the maximum striatum-to-cerebellum ratio was reached at 60 min. In both studies, the absolute values for this ratio were similar in the range of 9 to 10. It is noteworthy that due to its small size, the locus coeruleus cannot be analyzed from the in vivo PET study. These parallel studies provide a good demonstration of the strengths and weaknesses of different methods in radiopharmacological studies. Overall, the distribution of 18 F activity in rats after [ 18 F]CFT injection was in good agreement with our earlier preliminary studies [5,8] and with studies using [ 3 H]CFT [19,35].  The human ED and EDE values for [ 18 F]CFT are 9.17 μSv/MBq and 12.8 μSv/MBq, respectively, and they are in line with those of another dopamine transporter ligand, 18 F-FPCIT [36]. Extrapolation of the animal data to humans to estimate the human radiation dose is inexact, but the order of magnitude of the EDE and ED values for [ 18 F]CFT correspond well with those of other 18 F-labeled radioligands.

Conclusions
Post-target-produced high-SA [ 18 F]F 2 was used to incorporate 18 F directly into the phenyl ring of [ 18 F]CFT. The final product had high radiochemical and chemical purities and a high SA for neurotransmitter studies in vivo. It is noteworthy that as [ 18 F]CFT shows a specific binding to NET in addition to DAT, [ 18 F]CFT can also be used for imaging NET. The finding that [ 18 F]CFT shows specific uptake in the pancreas also warrants future studies in humans with respect to potential utility in pancreatic imaging.