Fluorine-18 Radiolabeled PET Tracers for Imaging Monoamine Transporters: Dopamine, Serotonin, and Norepinephrine

Synopsis

This review focuses on the development of fluorine-18 radiolabeled PET tracers for imaging the dopamine transporter (DAT), serotonin transporter (SERT), and norepinephrine transporter (NET). All successful DAT PET tracers reported to date are members of the 3β-phenyl tropane class and are synthesized from cocaine. Currently available carbon-11 SERT PET tracers come from both the diphenylsulfide and 3β-phenyl nortropane class, but so far only the nortropanes have found success with fluorine-18 derivatives. NET imaging has so far employed carbon-11 and fluorine-18 derivatives of reboxetine but due to defluorination of the fluorine-18 derivatives further research is still necessary.

Introduction

The dopamine transporter (DAT), serotonin transporter (SERT), and norepinephrine transporter (NET) are plasma membrane biogenic monoamine transporters which belong to the family of Na⁺/Cl⁻ dependent transporters.[1–8] In the central nervous system (CNS) the DAT, SERT, and NET are located on presynaptic neurons and function to remove their respective neurotransmitter from the synapse thereby terminating the action of that neurotransmitter. These three transporters have each been implicated in numerous psychiatric disorders such as depression, suicide, schizophrenia, Parkinson’s Disease (PD), and attention-deficit hyperactivity disorder (ADHD) and are also the target of drugs of abuse such as cocaine, amphetamines, and MDMA (Ecstasy). As such, these transporters have become therapeutic targets to treat psychiatric disorders and drug addiction.[9–11] The ability to image the DAT, SERT, and NET with positron emission tomography (PET) or single-photon emission computed tomography (SPECT) may aid in the diagnosis and management of psychiatric disease by providing a means to measure the density of these transporters in specific brain regions.[12–17] Additionally, the availability of radiolabeled tracers for these transporters may aid in the development of new therapeutics by enabling the occupancy of the therapeutic to be measured.[18–24]

Numerous PET tracers for the DAT, SERT, and NET have been developed that are radiolabeled with carbon-11 but these are limited to use in the location where they are prepared and only allow for imaging times of up to about 2 hours due to the short half-life of ¹¹C (t_1/2 = 20.4 min).[25–27] The longer half-life of ¹⁸F (t_1/2 = 109.8 min) allows for longer synthesis times and imaging sessions as well as for the transport of the ¹⁸F-labeled tracer to locations away from the cyclotron facility which allows for PET imaging centers without onsite cyclotrons to employ these tracers. In addition to the longer half-life of ¹⁸F, the positrons emitted from ¹⁸F-nuclides have a lower maximum energy (0.64 MeV)[26] than the positrons emitted from ¹¹C-nuclides (0.97 MeV) which therefore deposits less energy into tissue and also results in a shorter linear range that allows for higher spatial resolution.[28,29] Radiolabeling tracers with ¹¹C is convenient due to the ubiquitous nature of carbon in organic compounds whereas fluorine is far less common. Fluorine, though, has been shown to impart unique properties to organic molecules and is now being exploited extensively in medicinal chemistry.[30–36] Thus, numerous methods have been developed to introduce ¹⁸F or ¹⁹F into molecules.[37–41] This review will focus on fluorine-18 radiolabeled PET tracers for imaging the DAT, SERT, and NET. Several carbon-11 PET tracers are also included to allow for comparisons in instances where a tracer can be radiolabeled with either isotope or where fluorinated analogs of existing ¹¹C-labeled tracers have been developed.

There are several performance criteria that should be met in order for a candidate brain PET tracer to become a useful tracer. High binding affinity to the target, especially if the target is of low density, will enable highly specific and selective binding to the target. The goal is to obtain the highest possible target-to-nontarget uptake ratios which will result in PET images with high signal-to-noise ratios. If the tracer does not bind strong enough to the target then the tracer will not be retained in the tissue of interest and will just pass through. But, if the tracer binds too strongly then it will not dissociate from the target during the course of the PET study and will accumulate in the tissue of interest and only blood flow will be measured. A balance must, therefore, be obtained which will allow for the achievement of peak uptake in the target tissue in a short time frame (¹⁸F allows for longer time frames compared to ¹¹C) followed by a steady washout to allow for kinetic modeling of the behavior of the tracer.[42–47] Moderate lipophilicity in the range of logP = ~1–3 is necessary to allow for rapid entry into the brain and to limit nonspecific binding. [48–50] Low binding to plasma proteins is necessary to make as many tracer molecules as possible available for brain entry. Metabolism of the tracer is unavoidable but the resulting metabolites that are generated in the periphery, if they are radiolabeled, should be hydrophilic so that they cannot enter the brain. The generation of radiolabeled metabolites in the brain is also undesirable and any radiolabeled metabolites that are produced should have little or no affinity for the target of interest. As will become apparent below, meeting all of these criteria simultaneously is a difficult task.

Dopamine Transporter (DAT)

The human DAT is a 620-amino acid transmembrane protein which is 98.9% homologous to the monkey DAT and 92% homologous to the rat DAT.[4,51–53] The DAT is found in high densities in the caudate, putamen, nucleus accumbens, and olfactory tubercle with lower densities in the substantia nigra, amygdala, and hypothalamus.[54–57] The DAT has been associated with numerous neuropsychiatric diseases including PD,[58–60] supranuclear palsy,[59] ADHD,[61] and Tourette’s Syndrome.[62] The ability to image the DAT with PET may therefore aid in the diagnosis, monitoring, and treatment of these diseases.[13,15,17,63]

Early work towards developing an ¹⁸F-labeled DAT imaging agent focused on the GBR-compounds[64,65] [¹⁸F]1 and [¹⁸F]2 (Figure 1).[66–71] PET imaging with [¹⁸F]2 in monkeys[72] showed rapid uptake into the striatum and cerebellum with maximum levels achieved within 2–3 min followed by a slow but steady washout. Washout was faster from the cerebellum than the striatum which resulted in a striatum-to-cerebellum ratio of 1.76 at the end of the study. Radiotracer uptake was highest in the liver and kidneys due to metabolism of the tracer but low bone uptake indicated that the aryl-fluorine bond was stable to metabolic defluorination. The imaging results with [¹⁸F]2 were promising but the tracer still suffered from high lipophilicity and non-specific binding. The synthesis and radiolabeling of the GBR-derivatives [¹⁸F]3 and [¹⁸F]4 has also been reported but imaging data was not included.[73]

Figure 1.

Structures of GBR-based DAT PET tracers.

Cocaine (5) (Fig 2, Table 1) binds to the DAT, SERT, and NET[74,75] but the DAT was initially implicated as the site related to substance abuse.[76–78] As a tool to study cocaine addiction [¹¹C]cocaine PET imaging has been employed[79–83] but imaging times are limited due to the short half-life of ¹¹C. 4′-[¹⁸F]Fluorococaine ([¹⁸F]6) allows for extended imaging studies due to the longer half-life of ¹⁸F and [¹⁸F]6 was shown to have identical kinetic behavior to [¹¹C]cocaine in PET imaging studies.[82]

Fig 2.

Cocaine, 4′-Fluorococaine, and 3β-Phenyl Tropane-Based Dopamine Transporter Ligands.

Table 1.

Compound		R1 =	R2 =	X =
7	WIN 35,065-2	Me	Me	H
8	WIN 35,428; β-CFT	Me	Me	F
9	RTI-31	Me	Me	Cl
10	RTI-51; CBT	Me	Me	Br
11	RTI-55; β-CIT	Me	Me	I
12	RTI-32	Me	Me	Me
13	β-CFT-FE	CH2CH2F	Me	F
14	β-CFT-FP	CH2CH2CH2F	Me	F
15	FE-β-CIT; β-CIT-FE	CH2CH2F	Me	I
16	FP-β-CIT; β-CIT-FP	CH2CH2CH2F	Me	I
17		CH2CH2CH2F	i-Pr	I
18	FECNT	CH2CH2F	Me	Cl
19	FPCT	CH2CH2CH2F	Me	Cl
20	FPCMT	CH2CH2CH2F	Me	Me
21	FPCBT	CH2CH2CH2F	Me	Br
22	FETT	Me	CH2CH2F	Me
23	FECT	Me	CH2CH2F	Cl
24	MCL322	Me	CH2CH2F	Br
25	MCL301; FE@CIT	Me	CH2CH2F	I

Cocaine metabolism can take several possible routes: (a) ester hydrolysis, (b) N-demethylation, (c) N-oxidation, (d) aryl hydroxylation and/or epoxidation, and (e) dehydrobenzoylation.[82,84–89] The benzoic ester hydrolysis route was rendered obsolete when Clarke and co-workers demonstrated that replacement of the benzoyloxy group with a phenyl group attached directly to the 3β-position of the tropane skeleton produced compounds with higher binding affinities at the DAT than cocaine itself.[90] This new class of 3β-phenyl tropanes has been exploited extensively in the search for therapeutics to treat cocaine addiction[10,11,91–95] and hundreds of compounds have already been reported in the past[96] and more continue to be reported to this day.[97,98] The structures of some of the first 3β-phenyl tropane compounds prepared, 7–12, are shown in Fig 2 and Table 1.[90–92,99] The high affinity of these compounds for the DAT naturally made them candidates for use as PET tracers when labeled with ¹¹C as they can all be labeled on either the N-methyl group or the methyl ester.[88,100–104] Unfortunately, these ¹¹C-labeled compounds do not reach binding equilibrium during the course of the imaging study which limits their usefulness in kinetic modeling. Furthermore, compound 11 has a similar affinity for the DAT and SERT and compounds 9 and 10 are only slightly selective for the DAT over the SERT[94,96,105] thus limiting the utility of these compounds as DAT-selective imaging agents.

Out of compounds 7–12, only compound 8 contains a ¹⁹F atom that could be replaced with ¹⁸F. Additionally, 8 is more selective for the DAT over the SERT and NET than compounds 9–11,[94,96] and previous work with [¹¹C]8 had demonstrated that it was capable of imaging the DAT.[100,101] Thus, [¹⁸F]8 was synthesized and evaluated by biodistribution studies in rats[106] and then by PET imaging in humans.[107] In rat brain the striatum-to-cerebellum ratio reached a maximum of ~9.6 after 2 h and this uptake could be blocked by pretreatment with the DAT ligand GBR 12909 (1).[64,65] In rat biodistribution studies the highest uptake at all times was observed in the liver and urine whereas uptake in the kidney, spleen, and lungs was initially high but then decreased. Very little bone uptake was observed indicating that the aryl-fluorine bond of [¹⁸F]8 is stable to metabolic defluorination, similar to what was observed for [¹⁸F]2. In human subjects the total activity of [¹⁸F]8 in the caudate and putamen peaked at 225 min and then slowly declined whereas the specific binding remained at a plateau from 3–5 h post-injection. Application of [¹⁸F]8 to studying human subjects with early PD demonstrated that [¹⁸F]8 was capable of detecting presynaptic dopaminergic hypofunction.[108]

In an effort to reduce the time required to reach peak uptake of [¹⁸F]8, the N-[¹⁸F]fluoroethyl[105] and N-[¹⁸F]fluoropropyl[109,110] derivatives, [¹⁸F]13 and [¹⁸F]14, respectively, were prepared. In vitro binding assays demonstrated that both 13 and 14 had a reduced affinity for the DAT, SERT, and NET relative to 8, but 13 and 14 both had a slightly improved selectivity for the DAT over the SERT relative to 8.[105] PET imaging with [¹⁸F]13 in conscious rhesus monkeys demonstrated that [¹⁸F]13 reaches peak uptake after about 20 min followed by a washout phase, thus demonstrating reversible binding. A comparison study with [¹¹C]8 showed only irreversible binding during the 91 min PET scan as the radioactivity continuously increased throughout the study. Therefore, replacing the N-methyl group of 8 with an N-fluoroethyl group as in 13 resulted in improved tracer performance. A blocking study with 1 demonstrated that [¹⁸F]13 binding was DAT-specific. Metabolite analysis of [¹⁸F]13 in plasma showed only polar metabolites and no lipophilic metabolites that could cross the blood-brain barrier. Compound [¹⁸F]14 was evaluated ex vivo in rats and demonstrated rapid entry into the brain with a striatum-to-cerebellum ratio of ~3.1 after 5 min followed by a continuous decrease. This uptake could be blocked by pretreatment with 1. A continuous uptake was observed in the skull indicating a slow defluorination of the [¹⁸F]fluoropropyl group.

Compound 11 has a nearly equal affinity for the DAT and SERT[94,96,105] and PET imaging with [¹¹C]11 showed uptake in the striatum as well as the thalamus and neocortex.[88,103] This uptake in the thalamus and neocortex could be displaced by the SERT ligand citalopram[111,112] thus demonstrating binding to both the DAT and SERT in vivo, which is in agreement with the in vitro binding data.[96,105,113] In an effort to obtain ¹⁸F-labeled derivatives of 11 the N-fluoroalkyl compounds 15 and 16 were prepared.[113,114] Replacement of the N-methyl group of 11 with an N-fluoroethyl group (15) or an N-fluoropropyl group (16) resulted in a reduced binding affinity at the DAT and NET but an increased binding affinity at the SERT relative to 11.[113] Replacement of the methyl ester group of 16 with an isopropyl ester to give 17 increased the binding affinity at the DAT relative to 11 and significantly reduced the binding affinity at both the SERT and NET.[113,115] A SPECT imaging comparison[116] of [¹²³I]15, [¹²³I]16, and [¹²³I]17 in baboons demonstrated that [¹²³I]15 and [¹²³I]16 had higher peak uptake in the striatum than [¹²³I]17 thus ruling out [¹²³I]17 for adaptation to PET imaging as [¹⁸F]17. Additionally, [¹²³I]15 had more rapid kinetics than [¹²³I]16. Comparison between [¹²³I]11 and [¹²³I]16 in humans with SPECT imaging showed that [¹²³I]16 had higher non-specific binding than [¹²³I]11 but a lower radiation burden to the basal ganglia.[117]

PET studies in anesthetized cynomolgus monkeys with [¹¹C]15 showed high uptake in the putamen with lesser uptake in the thalamus and neocortex.[118] A blocking study with 1 reduced the uptake of [¹¹C]15 in the putamen by 75 % but did not affect the uptake in the thalamus or neocortex thus indicating that SERT binding was still an issue. Compound [¹⁸F]16 has been radiolabeled and evaluated in anesthetized cynomolgus monkeys[119] and conscious humans.[120] In monkeys high uptake of [¹⁸F]16 was observed in the putamen (this was blockable with 1) with peak uptake achieved in 30–40 min followed by only a very slight washout. After 70 min the striatum-to-cerebellum ratio was 4.5–5. Uptake in the thalamus was at levels similar to the cerebellum thus demonstrating an improvement over [¹¹C]11, [¹¹C]15, and [¹¹C]16. Metabolite analysis did not detect any lipophilic metabolites but [¹⁸F]fluoride was detected in plasma indicating that defluorination was occurring. Initial studies in humans with [¹⁸F]16 showed high uptake in the putamen of a healthy normal volunteer whereas uptake was reduced 65% in a PD patient.[120] Additional studies[121] in healthy normal humans with [¹⁸F]16 showed that uptake in the striatum increased rapidly after injection but then leveled off after 40 min and did not wash out. Uptake in the thalamus peaked at ~20 min and then began to wash out. In PD patients uptake in the putamen peaked at 30 min and then slowly washed out. Additional studies have been performed with [¹⁸F]16 to acquire data for parametric mapping,[122] dosimetry,[123] and modeling.[124] A one-step high-yield radiosynthesis of [¹⁸F]16 has been developed.[125]

Compounds 18 and 19 are the N-fluoroalkyl derivatives of 9 which has been previously radiolabeled as [¹¹C]9 and evaluated in rats.[102] PET studies with [¹⁸F]19 in anesthetized rhesus monkeys showed high uptake in the striatum with putamen-to-cerebellum and caudate-to-cerebellum ratios of 3.35 and 2.28, respectively, after 115 min.[126] The uptake in the cerebellum washed out after an initial peak at 13.5 min but the uptake in the caudate and putamen remained nearly constant after peaking thus indicating irreversible binding to the DAT. Biodistribution studies in rats showed continuously increasing bone uptake indicating that the fluoropropyl group was not resistant to defluorination which is in agreement with the detection of [¹⁸F]fluoride in plasma reported for the metabolism of [¹⁸F]16 and the observed skull uptake in studies with [¹⁸F]14. PET imaging with [¹⁸F]18 in an anesthetized rhesus monkey showed high uptake in the caudate and putamen (displaceable with 11) with peak uptake achieved in 60–75 min followed by washout at a rate of 8%/h.[127] At 115 min the putamen-to-cerebellum ratio was 12.7 and the caudate-to-cerebellum ratio was 12.3. When an imaging study with [¹⁸F]19 was performed in the same rhesus monkey the uptake in the caudate and putamen continuously increased throughout the course of the study,[127] similar to what was previously reported.[126] Thus, [¹⁸F]18 displays reversible binding and can achieve binding equilibrium whereas [¹⁸F]19 binds irreversibly. Compound [¹⁸F]18 has since been used to measure DAT occupancy of cocaine[128] and radiation dosimetry studies have been performed in rats and monkeys.[129,130] Two different automated radiosynthesis methods of [¹⁸F]18 have been reported.[131,132] Initial studies[133] in 6 healthy humans with [¹⁸F]18 showed that the time for peak uptake in the caudate and putamen was in the range of 70–130 min with caudate-to-cerebellum ratios of 7.6–10.5 and putamen-to-cerebellum ratios of 7.1–9.3. Comparison of the data obtained in healthy humans to that obtained in PD patients demonstrated that [¹⁸F]18 has the potential to measure DAT density in the brain and to detect the reduction in DAT density that is associated with PD. Metabolism studies in humans using arterial samples after injection of [¹⁸F]18 identified a polar non-ether-extractable component.[133] Subsequent studies[134] in rats, monkeys, and humans identified this polar metabolite as either [¹⁸F]fluoroethanol, [¹⁸F]fluoroacetaldehyde, or [¹⁸F]fluoroacetic acid (or a combination of all three as they are metabolically interchangeable). This work demonstrated that the polar metabolite is generated in the periphery but then passes into the brain and distributes evenly throughout the brain. The authors therefore suggest that the tissue reference method should not be used for analyzing PET data obtained with [¹⁸F]18 but rather an arterial input function should be used.

This N-dealkylation of [¹⁸F]18 is not unique to [¹⁸F]18 but has also been observed with cocaine,[84] [¹¹C]cocaine (to give [¹¹C]CO₂ in baboons),[79,82] and [¹¹C]11,[88] as well as other tracers.[135–139] Thus, metabolic cleavage of radiolabeled N-methyl and N-fluoroethyl groups is unavoidable and it should be expected that any tracer radiolabeled with an N-[¹⁸F]fluoroethyl group will be metabolized in the periphery to [¹⁸F]fluoroethanol, [¹⁸F]fluoroacetaldehyde, or [¹⁸F]fluoroacetic acid, and such an effect has recently been reported with the amyloid imaging agent [¹⁸F]FDDNP.[140] Switching to an N-[¹⁸F]fluoropropyl group is not necessarily a better alternative because the N-[¹⁸F]fluoropropyl group is not stable to defluorination as was demonstrated with [¹⁸F]14, [¹⁸F]16, and [¹⁸F]19.

Compounds 20 and 21 are the N-fluoropropyl derivatives of 12 and 10, respectively. A procedure for preparing [¹⁸F]20 has been reported but imaging data was not included.[141] PET imaging with [¹⁸F]21 in baboons showed the highest uptake in the putamen with peak uptake achieved in 7–10 min. The washout half-life was 54 min for the striatum whereas it was 19 min for the midbrain and 16 min for the cerebellum. A striatum-to-cerebellum ratio of 3.3 was achieved at 45–60 min post injection. Metabolite analysis detected a “less-lipophilic” metabolite which was presumed to be the 2β-carboxylic acid resulting from hydrolysis of the methyl ester.

As an alternative to radiolabeling with N-fluoroalkyl groups several compounds containing fluoroethyl esters have been prepared (22–25). Biodistribution studies in rats with [¹⁸F]22 and [¹⁸F]23 showed high uptake in the striatum and olfactory tubercles.[102] Compound [¹⁸F]23 showed higher uptake in the striatum than [¹⁸F]22 but also significantly higher liver uptake. Bone uptake was minimal for both tracers indicating that the [¹⁸F]fluoroethyl ester was stable to defluorination. An improved synthesis and metabolic stability analysis in rats of [¹⁸F]23 has recently been reported.[142] This work found that after 3 h plasma radioactivity consisted of ~93% intact [¹⁸F]23 and the radioactivity in homogenized cerebrum and cerebellum extracts each consisted of ~96% intact [¹⁸F]23. Furthermore, there was no trace of [¹⁸F]fluoroacetaldehyde or [¹⁸F]fluoroacetic acid which would result from hydrolysis of the [¹⁸F]fluoroethyl ester.

Compounds 24 and 25 are the fluoroethyl ester derivatives of 10 and 11, respectively.[143,144] Biodistribution studies of [¹⁸F]24 in rats showed high uptake in the striatum (blockable with 1) with peak uptake around 60 min followed by a slow washout.[145] Uptake in the adrenals and kidneys was initially high but then washed out whereas uptake in the liver continuously increased throughout the study. The tracer was rapidly metabolized in rat plasma to a polar metabolite which accounted for 85% of radioactivity in plasma after 60 min. In rat striatal homogenates the radioactivity was >90% intact parent compound after 60 min. Low bone uptake in the biodistribution studies indicated that the [¹⁸F]fluoroethyl ester was stable to defluorination. The striatal uptake observed in the biodistribution studies was further confirmed by autoradiography. MicroPET studies with [¹⁸F]24 in rats showed a steady uptake in the striatum for the first 20 min which then peaked and remained stable for 30–100 min post-injection. At the end of the study (115 min p.i.) the striatum-to-cerebellum ratio was 2.8. Evaluation of [¹⁸F]25 in rats showed peak uptake in the striatum at 15 min followed by a steady washout.[146] At 60 min the striatum-to-cerebellum ratio was 3.73 and the striatum-to-thalamus ratio was 2.99, while at 120 min the thalamus-to-cerebellum ratio was 1.65. High uptake in the kidneys suggested a renal excretion route whereas low bone uptake indicated stability to defluorination. A metabolic stability study with porcine carboxyl esterase found that [¹⁸F]25 and [¹²³I]11 have similar stabilities in regards to resistance to enzymatic hydrolysis of the ester but this stability was less than that found for [¹²³I]16.[147] The authors suggest that the N-fluoropropyl group of [¹²³I]16 is bulky enough to impede access to the ester bond by the enzyme thus inhibiting this route of metabolism. This bulkiness of the N-[¹⁸F]fluoropropyl group and the subsequent prevention of ester hydrolysis may, therefore, account for why defluorination of [¹⁸F]14, [¹⁸F]16, and [¹⁸F]19 is observed as the major metabolic route.

It had been previously shown that replacing the methyl ester of 9 with an isopropyl ester significantly reduced binding affinity at the SERT and NET while only slightly reducing DAT binding affinity.[115] Thus, the fluoroisopropyl derivatives (R)-26 and (S)-26 (Figure 3) were prepared and evaluated.[148] Competitive binding assays in murine kidney cells with transfected DAT or SERT showed that (S)-26 had a nearly 5-fold higher affinity (K_i = 0.67 nM) than (R)-26 (K_i = 3.2 nM) at the DAT and a lower affinity at the SERT (K_i’s = 85 nM (S) and 65 nM (R)). For comparison, 11 was included and had binding affinities of K_i = 0.48 nM (DAT) and K_i = 0.67 nM (SERT) which is in agreement with the previously reported nearly equal affinities at each transporter.[96,105,113] Thus, (S)-26 has only a slight reduction in DAT affinity compared to 11 but has a significant improvement in DAT vs. SERT selectivity. Biodistribution studies in rats with [¹⁸F](R/S)-26 showed highest uptake in the liver but very little bone uptake indicating that the fluoroisopropyl ester was stable to defluorination. In rat brain biodistribution studies [¹⁸F](R)-26 reached peak uptake in the striatum at 30 min and had a striatum-to-cerebellum ratio of 6.0 at 60 min, whereas [¹⁸F](S)-26 also reached peak uptake in the striatum at 30 min but at 60 min the striatum-to-cerebellum ratio was 11.8 and at 120 min it was 12.7. This higher uptake ratio for [¹⁸F](S)-26 compared to [¹⁸F](R)-26 is in agreement with the higher in vitro DAT binding affinity observed for [¹⁸F](S)-26. PET imaging in rhesus monkeys with [¹⁸F](R)-26 and [¹⁸F](S)-26 showed that for [¹⁸F](R)-26 uptake peaked in the putamen and caudate around 45 min and remained nearly stable until the end of the study (115 min) while uptake of [¹⁸F](S)-26 was still increasing at 115 min when the study ended. At 115 min [¹⁸F](R)-26 had a putamen-to-cerebellum ratio of 3.5 and a caudate-to-cerebellum ratio of 2.5 whereas [¹⁸F](S)-26 had a putamen-to-cerebellum ratio of 2.5 and a caudate-to-cerebellum ratio of 2.5. Additionally, the cerebellum uptake washed out significantly faster for [¹⁸F](R)-26 than for [¹⁸F](S)-26 which allowed [¹⁸F](R)-26 to achieve a transient equilibrium at 75 min. Metabolite analysis of rhesus monkey arterial plasma indicated that both [¹⁸F](R)-26 and [¹⁸F](S)-26 are metabolized to a non-ether-extractable polar compound but that [¹⁸F](R)-26 is metabolized more rapidly with 25% unmetabolized [¹⁸F](R)-26 remaining after 14 min but 30% unmetabolized [¹⁸F](S)-26 remaining after 30 min.

Figure 3.

3β-Phenyl Tropane-Based Dopamine Transporter Ligands.

Other variations on the tropane structure have included the N-benzyl 2β-ethyl ketones 27 and 28.[149] PET studies in rhesus monkeys with [¹⁸F]27 and [¹⁸F]28 demonstrated that both tracers reached peak uptake in the basal ganglia in ~20 min followed by a steady washout, thus indicating reversible binding. Compound [¹⁸F]28 was the superior of the two tracers with a basal ganglia-to-cerebellum uptake ratio of 2.6 vs. 1.5 for [¹⁸F]27, presumably due to the higher affinity [¹⁸F]28 has for the DAT.[149] Both tracers were rapidly metabolized in a rhesus monkey with <30% [¹⁸F]27 remaining and <20% [¹⁸F]28 remaining after 20 min. Metabolite analysis of arterial plasma samples from both a rhesus monkey and rats showed that [¹⁸F]28 was metabolized to a lipophilic metabolite but this metabolite was not observed in rat whole brain samples indicating that the metabolite cannot cross the blood-brain barrier.

The N-(E)-fluorobutenyl tropanes 29–32 (Fig 4, Table 2) have been previously reported.[150,151] Subsequently, the tolyl-derivative 33 was also reported.[152–156] Compounds 29–33 are thus the N-(E)-fluorobutenyl derivatives of 8–12, respectively. The binding affinities of 29–32 are shown in Fig 4 and Table 2.[151] Compounds 31 and 32 have a high affinity at both the DAT and SERT with little or no selectivity, similar to what has been reported for 10 and 11.[11,96,105,113,148] Compound 30 has a reduced affinity at the DAT and SERT relative to 31 and 32 without a significant improvement in selectivity. Compound 29 has a 10-fold lower affinity at the DAT than 32 along with a greatly reduced affinity at the SERT which results in a significantly improved selectivity for the DAT over the SERT compared to 30–32.

Fig 4.

Binding Affinities of 29–32 At Transfected Human Monoamine Transporters.

Table 2.

	Ki (nM)			Selectivity
Compound	DAT	SERT	NET	SERT/DAT	NET/DAT
29	1.70	85.5	>10,000	50.3	>5882
30	2.54	24.2	>10,000	9.5	>3937
31	0.24	0.85	91	3.5	379
32	0.17	0.21	57	1.2	335

Our initial imaging work with [¹⁸F]30-[¹⁸F]32 was performed on a Siemens 951 PET scanner. We have since reevaluated [¹⁸F]30 and [¹⁸F]31 along with [¹⁸F]29 in anesthetized cynomolgus monkeys with a Concorde microPET P4 which allows for higher resolution images.[157] The highest uptake of [¹⁸F]29 is seen in the putamen with peak uptake achieved after 25 min followed by a steady washout. Uptake in the caudate is also high but less than that seen in the putamen. This uptake in the caudate peaks after ~30 min but then remains stable until ~100 min post-injection followed by only a slight washout. Uptake in the substantia nigra peaks after 10 min and then slowly washes out, whereas uptake in the cerebellum peaks after 10 min and rapidly washes out. Injection of the DAT ligand RTI-113[158–161] at 90 min post-injection completely displaces [¹⁸F]29 from the caudate and the putamen. Compound [¹⁸F]30 shows a rapid uptake in the putamen with peak uptake achieved after 20 min followed by only a slight washout. Uptake in the caudate increases rapidly for the first 20 min, then increases slowly for 20–90 min, and then remains level for 90–235 min. The uptake in the caudate is less than that observed in the putamen and remains less than in the putamen even after the slight washout of activity from the putamen. The uptake in the substantia nigra peaks after 15 min and then washes out at a slower rate than that observed for [¹⁸F]29. Uptake of [¹⁸F]31 in the putamen peaks after 30 min and only slightly washes out whereas uptake in the caudate peaks after 30 min and then remains constant for 30–235 min. The initial uptake in the putamen was 1.5x that observed in the caudate. Uptake in the substantia nigra peaked after 20 min and then washed out steadily but slower than that observed for [¹⁸F]29. This rapid and high uptake of [¹⁸F]30 and [¹⁸F]31 in the putamen and the caudate followed by a lack of washout is similar to that reported for [¹¹C]33.[152,153,156]

Compounds [¹⁸F]29-[¹⁸F]31 and [¹¹C]33 all show higher uptake in the putamen than in the caudate. It has been previously shown that anesthesia can influence the behavior of PET tracers and can cause trafficking of the DAT to the plasma membrane.[162–168] Therefore, in an effort to evaluate whether the uptake observed for [¹⁸F]29-[¹⁸F]31 was influenced by the anesthesia used in the microPET studies a PET study with [¹⁸F]29 was performed in an awake rhesus monkey on a Siemens/CTI high resolution research tomograph (HRRT).[157] The uptake of [¹⁸F]29 in the caudate and putamen was equal in each hemisphere of the brain in the awake rhesus monkey thus indicating that the higher uptake in the putamen relative to the caudate observed in the microPET studies with [¹⁸F]29-[¹⁸F]31 was most likely an anesthesia effect. This is presumably also the cause of the higher uptake observed in the putamen than in the caudate with [¹¹C]33 in an anesthetized baboon as well as the reported putamen-to-cerebellum and caudate-to-cerebellum uptake ratios of 30 and 24.6, respectively, in one study[152] and 28.9 and 23.6, respectively, in another study.[153] The uptake of [¹⁸F]29 in the putamen and the caudate of the awake rhesus monkey peaked at 30 min and then slowly washed out until the end of the study (85 min), whereas the uptake ratios peaked around 60 min and then began to decline. Thus, [¹⁸F]29 is able to achieve peak uptake, reversible binding, and attain a transient equilibrium in an acceptable time frame while also achieving excellent uptake ratios vs. cerebellum uptake in an awake state.

The microPET images obtained with [¹⁸F]29-[¹⁸F]31 and the PET images obtained with [¹⁸F]29 all show skull uptake thus indicating defluorination of the [¹⁸F]-(E)-fluorobutenyl group. In an effort to by-pass this [¹⁸F]-defluorination and eliminate bone uptake of [¹⁸F]fluoride we are currently working towards radiolabeling [¹⁸F]29 on the aromatic ring similar to what has been reported for [¹⁸F]1-[¹⁸F]4, [¹⁸F]6, and [¹⁸F]8. Skull uptake of [¹⁸F]fluoride resulting from defluorination of [¹⁸F]33 can be avoided by using [¹¹C]33 but both [¹⁸F]33 and [¹¹C]33 presumably will still suffer from benzylic hydroxylation and the generation of radiolabeled metabolites similar to what has recently been reported for [¹¹C]PE2I.[139,147] It is expected that the size of the N-(E)-fluorobutenyl group will prevent enzymatic hydrolysis of the methyl ester similar to what has been previously reported.[147]

Extensive research over the past 20 years directed at developing fluorine-18 radiolabeled PET tracers for imaging the DAT has identified numerous potential candidates from the 3β-phenyltropane class. The most promising are those that can achieve reversible binding and a transient equilibrium in a short time frame along with high specific binding and selectivity while also minimizing the interference from radiolabeled metabolites. So far both [¹⁸F]16 and [¹⁸F]18 have found use in human imaging and both have demonstrated their utility in detecting a reduction of DAT density in human PD subjects. But both compounds show a slower washout than may be desirable and both suffer from complications due to metabolism. Compound [¹⁸F]29 offers improved kinetics but it will need to be radiolabeled on the aromatic ring in order to avoid the skull uptake that results from defluorination of the [¹⁸F]fluorobutenyl group.

Serotonin Transporter (SERT)

The human SERT is a 630-amino acid protein[6] that is 98.6% homologous to rhesus monkey SERT, 93% homologous to mouse SERT, and 90% homologous to rat SERT.[4,169] The SERT is found in high densities in the dorsal and median raphe nuclei, putamen, caudate, thalamus, hypothalamus, and amygdala, with lower levels in the cortex and still lower levels in the cerebellar cortex.[170–177] The serotonin system has been associated with numerous psychiatric conditions including PD,[178,179] obsessive-compulsive and panic disorders,[180] stress,[181] and depression and suicide.[182–184] A reduced density of SERT has been observed postmortem in depressed individuals and victims of suicide.[185–187] The SERT is, therefore, the target of the selective serotonin reuptake inhibitor (SSRI) class of antidepressants.[111,112] The availability of specific and selective SERT PET tracers would allow for the measurement of SERT density in the brain and may allow for the detection and diagnosis of SERT-related psychiatric illness as well as a means of measuring SSRI occupancy and the monitoring of SSRI therapy.[16,188–192]

The SSRI Fluoxetine (Prozac) 34[193] (Figure 5) has been radiolabeled with both ¹¹C[194] and ¹⁸F.[195] In both cases the uptake of the tracer was characterized by non-specific binding and high lipophilicity. A PET study with [¹⁸F]34 in rhesus monkeys showed uptake in all brain regions and an autoradiographic study in rats demonstrated irreversible binding due to subcellular uptake of the tracer.

Figure 5.

SERT Ligands.

The antidepressant (+)-McN5652[196] 35 has been radiolabeled with ¹¹C and evaluated in rats and mice[197] and in humans.[198] In the human brain the uptake of [¹¹C]35 in the midbrain reached a plateau after ~90 min but did not wash out before the end of the study (115 min) thus necessitating the need for the longer-lived ¹⁸F. The fluorinated derivatives 36 and 37 were therefore prepared and evaluated. Ex vivo autoradiography with [¹⁸F]37 in mice showed accumulation in the hypothalamus, substantia nigra, and raphe nuclei but whole-brain uptake, specific binding, and tissue-to-cerebellum ratios were significantly less than that observed for [¹¹C]35. Replacement of the methyl group of 35 with a fluoromethyl group to give 36 resulted in about a 3-fold loss in binding affinity at the SERT.[199] Biodistribution studies in rats with [¹⁸F]36 showed uptake in the raphe nuclei, substantia nigra, locus coeruleus, hypothalamus, thalamus, and amygdala, and this uptake could be blocked with 34.[200] The uptake in these regions washed out slower than cerebellum washout and this resulted in uptake ratios of 7–9 after 3 h. High uptake was also observed in the adrenal gland, lung, intestine, spleen, kidney, liver, and bone marrow. A continuous increase of radioactivity in the skull indicated that [¹⁸F]36 was not stable to defluorination in rats. Additional studies with [¹⁸F]36 were performed in pigs.[201–203]

The fluorinated derivatives of 6-nitroquipazine,[204] 38 and 39, have been prepared and evaluated as possible SERT PET tracers.[205,206] In rats [¹⁸F]38 rapidly entered the brain and showed the highest accumulation in the frontal- and posterial cortex followed by the striatum and this uptake could be reduced 10–20% with the SSRI citalopram.[111,112] Lesser uptake was observed in the thalamus but this uptake could be reduced 20–30% with citalopram. PET imaging in a monkey showed rapid uptake into the brain with a frontal cortex-to-cerebellum ratio of 1.53 and a striatum-to-cerebellum ratio of 1.25 at 80 min post-injection. The uptake in the thalamus was less than that observed in the cerebellum indicating that [¹⁸F]38 was not a good candidate for imaging the SERT. Biodistribution studies in mice with [¹⁸F]39 showed the highest uptake after 60 min in the frontal cortex and the lowest uptake in the cerebellum. High uptake was also observed in the olfactory tubercle, hypothalamus, thalamus, hippocampus, and striatum. Continuous bone uptake was also observed indicating a slow defluorination of the fluoropropyl group.

The diphenylsulfide 40 (Figure 6, Table 3) has been previously reported to be an inhibitor of the SERT and NET.[207,208] Numerous compounds have now been reported which exploit the diphenylsulfide motif in order to take advantage of its high binding affinity at the SERT and compounds 41–51 have been radiolabeled with ¹¹C.[209–218] Comparison of [¹¹C]35, [¹¹C]41, [¹¹C]42, [¹¹C]44, and [¹¹C]49 indicated that [¹¹C]42 had the fastest kinetics whereas [¹¹C]49 had the highest signal-to-noise ratio.[213] Comparison between [¹¹C]42 and [¹¹C]45 in the same anesthetized rhesus monkey indicated that [¹¹C]45 achieved higher uptake ratios than [¹¹C]42 and also reached peak uptake faster.[218] Both [¹¹C]42 and [¹¹C]45 have been employed in humans[219–221] and both have demonstrated that they are valuable tracers for imaging the SERT with PET, but both are ¹¹C-labeled and are still, therefore, limited to use where they are prepared and to imaging sessions of less than 2 h.

Fig 6.

Diaryl Sulfide-Based Serotonin Transporter Ligands.

Table 3.

					Ki (nM)
Compound		R1	R2	R3	SERT	DAT	NET	Ref.
41	ADAM	I	H	H	0.013	840	699	[209]
42	DASB	CN	H	H	1.10	1423	1350	[211]
43	DAPP	OMe	H	H	1.89	2651	1992	[211]
44	DAPA	Br	H	H	0.38	N/A	N/A	[213]
45	HOMADAM	CH2OH	H	H	0.57	>1000	144	[218]
46	EADAM	Et	H	H	0.17	>1000	367	[217]
47	AFA,	F	H	H	1.46	>10,000	141.7	[215]
	4-F-ADAM	F	H	H	0.08	2267	117	[222]
48		CF3	H	H	0.33	2038	1205	[211]
49	AFM	CH2F	H	H	1.04	>10,000	664	[214]
50	AFE	CH2CH2F	H	H	1.80	>10,000	946	[216]
51	AFP	CH2CH2CH2F	H	H	3.58	>10,000	505	[223]
52	5-F-ADAM	H	F	H	0.47	N/A	N/A	[224]
53	ACF	Cl	F	H	0.05	3020	650	[225]
54		H	H	OCH2CH2F	0.25	340	7.5	[226]
55		F	H	OCH2CH2F	0.10	>1000	37	[226]
56		Cl	H	OCH2CH2F	0.03	847	97	[226]
57		Br	H	OCH2CH2F	0.05	>1000	114	[226]
58		H	H	OCH2CH2CH2F	1.4	299	12	[226]
59		F	H	OCH2CH2CH2F	0.95	>1000	95	[226]
60		Br	H	OCH2CH2CH2F	0.15	>1000	>1000	[226]
61		CH2OH	H	F	1.26	>2000	618	[227]

Compounds 47, 49, and 50 all have a high affinity for the SERT and are selective for the SERT over the DAT and NET. Compound 51 has a slightly reduced SERT affinity but still shows selectivity over the DAT and NET.[223] Compounds [¹¹C]47, [¹¹C]49, and [¹¹C]50, have been evaluated in baboons.[214–216] Biodistribution studies in rats were performed to compare [¹⁸F]47, [¹⁸F]49, [¹⁸F]50, and [¹⁸F]51.[223] All four compounds rapidly entered the brain and accumulated in the thalamus, hypothalamus, frontal cortex, striatum, and hippocampus. This uptake could be blocked with citalopram thus demonstrating SERT-specific binding. Compound [¹⁸F]49 reached peak uptake after 30 min followed by washout and this uptake was the highest among the four tracers. Compounds [¹⁸F]47, [¹⁸F]50, and [¹⁸F]51 all reached peak uptake after 10 min followed by washout while the highest uptake observed was with [¹⁸F]47. PET imaging comparisons in the same baboon with [¹⁸F]47, [¹⁸F]49, and [¹⁸F]50 showed that [¹⁸F]47 and [¹⁸F]50 reached peak uptake in the thalamus in 15–35 min whereas [¹⁸F]49 reached peak uptake in 40–60 min but showed significantly higher specific binding.[228] Another reported biodistribution study in rats with [¹⁸F]47 had shown peak uptake at 30 min followed by washout.[222] This study also showed that [¹⁸F]47 accumulated rapidly in the muscle, lung, liver, and kidney but then cleared whereas bone uptake continuously increased indicating that, at least in rats, defluorination was a problem. In a PET study with [¹⁸F]47 in a baboon the radioactivity in the skull did not increase with time suggesting that defluorination was isolated to rats.[222] In this PET study peak uptake in the midbrain and striatum was reached in ~30–40 min with midbrain-to-cerebellum ratios of 3.2 at 2 h and 4.2 at 3 h and striatum-to-cerebellum ratios of 2.4 at 2 h and 2.7 at 3 h.

Compound 52 is an isomer of 47 where the fluorine atom has been moved from the 4-position (para to the sulfur atom) to the 5-position (meta to the sulfur atom). This change resulted in a significant loss in affinity for the SERT[224] compared to 47[222] as well as a slight decrease in lipophilicity (logP = 2.47 vs. 2.73). Biodistribution studies in rats with [¹⁸F]52 showed an initial high uptake in the lungs, kidneys, and heart but this cleared quickly whereas bone uptake continued to increase indicating defluorination.[229] Brain uptake was rapid with peak uptake at 2 min followed by a fast washout. The hypothalamus-to-cerebellum ratio was 2.97 at 60 min which is less than that observed with [¹⁸F]47. Thus, changing the position of the fluorine-atom did not produce an improvement in imaging properties compared to [¹⁸F]47.

Compound 53 retains the fluorine atom in the 5-position and places a chlorine atom in the 4-position which produced a compound with a high SERT affinity and selectivity.[225] Biodistribution studies with [¹⁸F]53 in rats showed initial high accumulation in lung, muscle, liver, kidney, and skin which then declined.[225] Bone uptake was slightly less than that observed with [¹⁸F]52[229] and significantly less than that observed with [¹⁸F]47.[222] Brain uptake of [¹⁸F]53 was rapid with the highest uptake observed at 2 min followed by washout. At 60 min the hypothalamus-to-cerebellum ratio was 3.5 (compared to 3 for [¹⁸F]52) and the striatum-to-cerebellum ratio was 2.9. Thus, [¹⁸F]52 and [¹⁸F]53 behave similarly in rats with regards to uptake ratios and defluorination but both have too rapid of kinetics when compared to [¹⁸F]47.

The fluoroalkyl ethers 54–60 have a high affinity for the SERT but 54 and 58 also have an appreciable affinity for the NET [226] which may cause interference in PET imaging if the cerebellum was to be used as the reference region. In rat biodistribution studies [¹⁸F]55-[¹⁸F]57, [¹⁸F]59, and [¹⁸F]60 showed high brain uptake but slow washout whereas [¹⁸F]54 and [¹⁸F]58 also showed high brain uptake but a faster washout. Of compounds [¹⁸F]54-[¹⁸F]60, the highest hypothalamus-to-cerebellum ratios were observed with [¹⁸F]54 and [¹⁸F]58 (~7.8 and ~7.7, respectively, at 120 min) due to the faster washout of these compounds.[226] Additional rat biodistribution studies with [¹⁸F]58 showed initial high uptake in the lung, skin, muscle, liver, and kidney which then slowly declined.[230] Bone uptake was high and remained high indicating defluorination. Brain uptake was rapid followed by a steady washout. Autoradiography showed uptake in the olfactory tubercles, thalamic nuclei, hypothalamic nuclei, substantia nigra, superior colliculus, dorsal raphe, medial raphe, and locus coeruleus, all of which could be blocked with the SSRI escitalopram[231,232] except for the locus coeruleus which could be blocked with the NET ligand nisoxetine.[233,234] Thus, some NET binding in the locus coeruleus is observed in conjunction with the desired binding in the SERT-rich brain regions when performing autoradiography but uptake in the cerebellum was not observed in the rat when PET imaging was performed with [¹⁸F]58 which will, therefore, allow for use of the cerebellum as the reference region. In the rat PET images obtained with [¹⁸F]58, clear localization was observed in the thalamus, midbrain, and striatum with peak uptake achieved in 10–20 min followed by a steady washout.[230] The region-to-cerebellum ratios peaked at 100–110 min with a value of ~4 followed by a slow decline. The uptake in the midbrain, thalamus, and striatum could be displaced by escitalopram (2 mg/kg) but also, to a lesser extent, by nisoxetine (10 mg/kg). A similar displacement of the SERT PET tracer [¹¹C]mZIENT during a chase study with the NET ligand (±)-reboxetine•mesylate (1 mg/kg) has also been observed.[235] Both nisoxetine and reboxetine[236] have a weak affinity for the SERT (K_i = 30 and 60 nM, respectively) but are selective for the NET over the SERT (21.4 and 6.7, respectively).[237] This weak binding to the SERT may be the cause of the observed displacement of these two tracers but alternatively, the displacement may be an indirect effect mediated through interactions between the noradrenergic and serotonergic systems.[238–240]

Compound 61 is a fluorinated derivative of 45.[227] Introduction of the fluorine atom resulted in a slight reduction in SERT affinity but also a reduction in DAT and NET affinity which produced similar selectivities for the SERT over the DAT and NET. The presence of the fluorine atom also increased the lipophilicity somewhat (logP_7.4 = 2.06 and 1.60, respectively). MicroPET imaging with [¹¹C]61 in an anesthetized cynomolgus monkey showed high uptake in the midbrain, pons, thalamus, and medulla with little uptake in the frontal cortex and cerebellum. No uptake was observed in the caudate or putamen which is surprising since this brain region does contain SERT.[177] Peak uptake of [¹¹C]61 in the midbrain occurred between 12.5 and 27.5 min post-injection followed by a steady washout. At 85 min the midbrain-to-cerebellum ratio was 3.4 and the thalamus-to-cerebellum ratio was 2.3. The observed uptake was displaceable with (R/S)-citalopram•HBr thus demonstrating SERT-selective binding. These promising results suggest that [¹⁸F]61 may be a viable SERT PET tracer.

Replacement of one of the N-methyl groups of 41 with a fluoroethyl group resulted in a significant loss of SERT affinity (K_i = ~19 nM vs. 0.4 nM for 41)[241] indicating that functionalizing this position was detrimental to binding. Nevertheless, the para-fluorobenzyl derivatives 62 (SERT K_i = ~4 nM) and 63 (SERT K_i = ~510 nM) were prepared and evaluated.[242] As seen from the SERT binding affinities the large p-fluorobenzyl group can be tolerated to a certain extent if the nitrogen atom is substituted secondarily (62) but replacement of the hydrogen atom with a methyl group (63) completely abolished any affinity for the SERT. Biodistribution studies in rats with [¹⁸F]63 showed low brain uptake and low selectivity in SERT-rich brain regions. Placement of the p-fluorobenzyl group on the anilino-nitrogen (64) produced a compound with moderate SERT affinity (K_i = ~10 nM) and replacement of the p-fluorobenzyl group with a p-fluorobenzoyl group (65) further reduced the SERT affinity (K_i = ~14 nM).[243]

During the search for tropane-based cocaine addiction therapeutics as mentioned in the DAT section above, it was discovered that removal of the N-methyl group of a tropane to give a nortropane produced compounds with enhanced SERT and NET affinity.[244] Numerous compounds have been prepared in an attempt to obtain SERT-selective nortropane derivatives[245–251] and several have been radiolabeled with ¹¹C.[235,252–255] The fluoroethyl ester nortropanes 66 and 67 (Figure 8) have been reported but, similar to 11, the affinities for the SERT and DAT were nearly equal.[144]

Figure 8.

Fluoroethyl Ester Nortropane-based SERT Ligands.

The results of microPET and HRRT PET imaging in monkeys with the meta- and para-vinyl iodides [¹⁸F]68 and [¹⁸F]69, respectively, have recently been reported.[256,257] Both 68 and 69 have a high affinity for the SERT (K_i = 0.43 and 0.08 nM, respectively) but substitution in the para-position provides about a 5-fold higher affinity. These differences in binding affinity to the SERT translate into significant differences in the imaging behavior of the compounds. MicroPET imaging with [¹⁸F]68 and [¹⁸F]69 was performed in anesthetized cynomolgus monkeys (Figure 9) and PET imaging on an HRRT with [¹⁸F]68 and [¹⁸F]69 was performed in an awake rhesus monkey (Figure 10). The time required to reach peak uptake is significantly different for the two isomers with the stronger binding para-isomer [¹⁸F]69 taking longer to reach peak uptake. Also, the times required for each tracer to reach peak uptake are about the same in both the anesthetized and awake states indicating that anesthesia does not significantly affect the performance of these tracers. After reaching peak uptake the activity of [¹⁸F]68 in the midbrain, putamen, thalamus, medulla, and caudate remains nearly steady for about 20 min followed by a constant washout during which time (65–175 min post-injection) a pseudo-equilibrium is achieved. Compound [¹⁸F]69 takes significantly longer to reach peak uptake in the SERT-rich brain regions and then only slightly washes out. The uptake of [¹⁸F]69 in the cerebellum reaches peak uptake after 45 min followed by a steady washout which provides ever increasing tissue-to-cerebellum ratios as the study progresses. Because the washout of [¹⁸F]68 from the SERT-rich brain regions is nearly parallel to the washout from the cerebellum the tissue-to-cerebellum ratios for [¹⁸F]68 slowly increase throughout the course of the study. Thus, [¹⁸F]68 provides superior imaging kinetics and achieves a pseudo-equilibrium whereas [¹⁸F]69 provides higher uptake ratios. Both of the tracers can be displaced with citalopram but the observed displacement is significantly greater for [¹⁸F]69 because of the minor washout that occurs (under baseline conditions) after peak uptake is achieved which results in only slightly declining time-activity curves. When a citalopram chase is then performed the slope of the time-activity curves of [¹⁸F]69 changes drastically. Thus, each of these tracers is a promising candidate for human use because each has its own unique imaging properties. Compound [¹⁸F]68 can be used for kinetic analysis because of its ability to achieve a pseudo-equilibrium whereas compound [¹⁸F]69 can be used for SSRI occupancy studies due to the large change in slope of the time-activity curves that occurs when an SSRI is administered.

Figure 9.

MicroPET time-activity curves obtained by injection of [¹⁸F]68 (left) and [¹⁸F]69 (right) into anesthetized cynomolgus monkeys.

Figure 10.

HRRT PET images obtained by injection of [¹⁸F]68 (left) and [¹⁸F]69 (right) into an awake rhesus monkey.

The goal of developing an ¹⁸F-radiolabeled PET tracer for imaging the SERT has benefited from the availability of high affinity and selective compounds from both the diphenylsulfide and nortropane classes. Compounds [¹¹C]42 and [¹¹C]45 have both demonstrated their utility in imaging normal human volunteers but the adaptation of these compounds to ¹⁸F-labeled tracers has so far proven difficult. The ¹⁸F-labeled nortropanes [¹⁸F]68 and [¹⁸F]69 have been shown to have improved kinetics relative to their ¹¹C-labeled counterparts, [¹¹C]mZIENT[235] and [¹¹C]pZIENT,[255] respectively, and we look forward to evaluating [¹⁸F]68 and [¹⁸F]69 in healthy human volunteers.

Norepinephrine Transporter (NET)

The human NET is a 617-amino acid transmembrane protein and is 98.9% homologous to the monkey NET.[4,174,258] The NET is found in high densities in the locus coeruleus, cerebellum, dorsal raphe nuclei, thalamus, and hypothalamus.[233,234,239,259,260] The norepinephrine system has been associated with a variety of neuropsychiatric diseases including PD,[261] ADHD,[262,263] posttraumatic stress disorder (PTSD),[264] depression and anxiety,[240,265–267] and seizure,[268] and a reduction in NET density has been demonstrated postmortem in depressed subjects.[269] The availability of specific and selective NET PET imaging agents will allow for density measurements of the NET and may help to clarify the role of the NET in certain psychiatric diseases as well as help to identify and evaluate NET therapeutics.[270–272]

The antidepressant reboxetine 70 (Figure 11, Table 4) is a selective norepinephrine reuptake inhibitor (SNRI) which is marketed as the racemic mixture.[236,273,274] The ethyl group of 70 has been replaced with a [¹¹C]methyl group to give [¹¹C]71 and PET imaging comparisons in baboon between racemic-[¹¹C]71, (S,S)-[¹¹C]71, and (R,R)-[¹¹C]71 has shown that (S,S)-[¹¹C]71 is the active isomer based on a significantly higher distribution volume in the thalamus and the ability to block the uptake of (S,S)-[¹¹C]71, but not (R,R)-[¹¹C]71, in the thalamus and cerebellum with nisoxetine.[275] The uptake of (S,S)-[¹¹C]71 in the striatum could not be blocked with nisoxetine thus demonstrating either low affinity binding or non-specific binding in this brain region. In biodistribution studies in rats (S,S)-[¹¹C]71 showed a hypothalamus-to-striatum ratio of 2.5 at 60 min post-injection whereas (R,R)-[¹¹C]71 showed only a homogenous distribution.[276] Plasma analysis indicated that (S,S)-[¹¹C]71 was metabolized rapidly in the periphery with 50% unmetabolized radiotracer remaining after 30 min and 20% remaining after 60 min but whole rat brain extracts showed that 95% of radioactivity in the brain was unmetabolized (S,S)-[¹¹C]71. PET imaging with (S,S)-[¹¹C]71 in cynomolgus monkeys showed the highest uptake in the lower brainstem, mesencephalon, and thalamus while the lowest uptake was in the striatum.[277] Thus, three groups independently verified that (S,S)-[¹¹C]71 could be used to image the NET in vivo with PET although specific peak binding equilibrium is not achieved during the course of these studies. Based on these results fluoroalkyl ether derivatives of reboxetine have been developed in pursuit of an ¹⁸F-labeled NET PET tracer which would enable longer imaging times and possibly the achievement of peak specific binding equilibrium.

Fig 11.

(S,S)-Reboxetine Derivative-Based Norepinephrine Transporter Ligands.

Table 4.

Compound		X =	R =
70	Reboxetine	O	OCH2CH3
71	MeNER	O	OCH3
72	FMeNER	O	OCH2F
73	FMeNER-D2	O	OCD2F
74	FERB	O	OCH2CH2F
75	FERB-D2	O	OCD2CD2F
76	MENET	O	CH3
77	MESNET	S	CH3
78	FENET	O	CH2CH2F
79	FPNET	O	CH2CH2CH2F
80		S	OCH2CH2F
81		S	OCH2CH2CH2F

Compound (S,S)-[¹⁸F]72 is a fluorinated derivative of (S,S)-[¹¹C]71 but PET imaging in a cynomolgus monkey with (S,S)-[¹⁸F]72 showed rapid defluorination and skull uptake. Thus, the deuterated derivative (S,S)-[¹⁸F]73 was developed and PET imaging showed that defluorination was reduced but not totally inhibited.[278] During PET imaging the total brain uptake of (S,S)-[¹⁸F]72 peaked at 8 min and then washed out whereas total brain uptake of (S,S)-[¹⁸F]73 peaked at 12 min and then washed out. The uptake of both (S,S)-[¹⁸F]72 and (S,S)-[¹⁸F]73 could be blocked with desipramine but blocking with 1 or citalopram did not have any effect on the uptake of either tracer. At 110 min after injection of (S,S)-[¹⁸F]72 the tissue-to-striatum ratios were 1.2, 1.2, and 1.3 for the lower brainstem, mesencephalon, and thalamus, respectively. At 160 min after injection of (S,S)-[¹⁸F]73 the tissue-to-striatum ratios were 1.5, 1.6, 1.3, and 1.5 for the lower brainstem, mesencephalon, thalamus, and temporal cortex, respectively. Peak specific binding was achieved at 90–120 min for (S,S)-[¹⁸F]72 and at 120–160 min for (S,S)-[¹⁸F]73. Autoradiography with (S,S)-[¹⁸F]73 using post-mortem human brain slices showed the highest accumulation of radioactivity in the locus coeruleus with lower accumulation in the cerebellum, cortex, thalamus, and hypothalamus along with non-specific binding in the caudate nucleus and putamen.[279] A low signal-to-noise ratio was seen outside the locus coeruleus and the authors suggest that the binding affinity of (S,S)-[¹⁸F]73 (K_i = 3.1 nM) may not be sufficient enough for visualization and quantification of the NET in low density regions. Biodistribution and radiation dosimetry studies have been performed with (S,S)-[¹⁸F]73 in cynomolgus monkeys[280] and humans[281] and (S,S)-[¹⁸F]73 has been used to measure the occupancy of the SNRI atomoxetine.[282] Initial human imaging studies with (S,S)-[¹⁸F]73 have recently been reported.[283] Using 4 healthy male subjects the mean peak uptake in the whole brain was 2.6 ± 0.5% at 12.5 min after injection. Uptake in the thalamus achieved peak equilibrium with a thalamus-to-caudate ratio of 1.5. Cortical uptake was high due to high skull uptake resulting from defluorination. Thus, (S,S)-[¹⁸F]73 is a promising PET tracer for imaging the NET in humans but the inability to totally stop defluorination suggests that further structural improvements are still needed.

Compound 74 is a fluorinated derivative of 70 and compound 75 is the deuterated analog of 74, the deuteration strategy again being employed to inhibit defluorination.[284] Biodistribution studies in mice with (S,S)-[¹⁸F]74 and (S,S)-[¹⁸F]75 showed high accumulation of both tracers in the kidneys, liver, and intestines. Brain uptake was moderate and washout was slow for both (S,S)-[¹⁸F]74 and (S,S)-[¹⁸F]75. At 2 h post-injection the average bone uptake of (S,S)-[¹⁸F]74 was 0.83% ID/g and that of (S,S)-[¹⁸F]75 was 0.25% ID/g thus demonstrating a reduction of defluorination via the deuterium isotope effect. A PET imaging comparison in baboons with (S,S)-[¹¹C]71 and (S,S)-[¹⁸F]75 (along with several other ¹¹C-labeled NET ligands) determined that (S,S)-[¹¹C]71 still has the best PET imaging characteristics of the prospective NET PET tracers available at that time due to the higher signal-to-noise ratio obtained with (S,S)-[¹¹C]71 and its faster washout from the striatum.[271,285]

Compounds 76–81 have recently been developed as potential NET PET tracers.[286,287] Compounds 76–79 have directly connected the alkyl group to the aryl ring thereby eliminating the phenyl alkyl ether functionality and thus possibly enhancing the metabolic stability of the radiolabel. Compounds 77, 80, and 81 have incorporated a sulfur atom in replacement of the benzyl phenyl ether oxygen atom. Compounds 76 and 77 each have a high affinity for the NET (K_i = 1.02 ± 0.11 and 0.30 ± 0.03 nM, respectively) which is very similar to the NET affinity of 70 (K_i = 1.04 ± 0.16 nM) and 71 (0.95 ± 0.03 nM).[287] Thus, the phenyl methyl ether oxygen atom is not an important determinant of binding affinity at the NET and a slight enhancement in affinity results from replacing the benzyl phenyl ether oxygen atom with a sulfur atom. Compound 77 has an enhanced SERT affinity (K_i = 14.8 ± 2.83 nM) relative to 76 (K_i = 93 ± 20 nM) indicating that the benzyl phenyl ether oxygen atom is necessary for low SERT affinity. Compounds 78 and 79 each had a reduced NET affinity (K_i = 3.14 ± 0.17 and 3.68 ± 0.92 nM, respectively) compared to 76 indicating that larger alkyl groups, or fluorine substitution, or both, is detrimental to the NET affinity of this series of compounds. MicroPET imaging with [¹¹C]76 in an anesthetized rhesus monkey showed peak uptake around 20 min followed by a steady washout with tissue-to-caudate ratios of 1.30, 1.45, 1.40, and 1.30 at 45 min post-injection and 1.30, 1.43, 1.44, and 1.25 at 85 min post-injection for the thalamus, midbrain, pons, and cerebellum, respectively, indicating that a quasi-equilibrium had been achieved. MicroPET imaging with [¹¹C]77 in an anesthetized rhesus monkey showed peak uptake around 30–40 min followed by a steady washout (but slower than that of [¹¹C]76) with a thalamus-to-caudate ratio of 1.34 and a midbrain-to-caudate ratio of 1.33 at 85 min post-injection. Because of the moderate affinity 77 has for the SERT, a chase study was performed with (R,S)-citalopram•HBr (1.5 mg/kg) at 40 min post-injection. The uptake of radioactivity in the caudate did not change indicating that this uptake was most likely the result of nonspecific binding rather than SERT binding. PET imaging on an HRRT was performed in awake rhesus monkeys with [¹¹C]76 and [¹¹C]77 to assess the effects of anesthesia on the performance of these tracers. With [¹¹C]76 high uptake was observed in the thalamus, locus coeruleus, midbrain, and pons and the kinetics were very similar to that observed in the microPET study with an anesthetized rhesus monkey but the tissue-to-caudate ratios were slightly reduced in the awake state. Compound [¹¹C]77, on the other hand, showed significantly different kinetics than that observed in the microPET study with an anesthetized rhesus monkey with prolonged retention observed over the course of the study and very little washout. The tissue-to-caudate ratios were also slightly reduced in the awake state. It is known that the anesthetic ketamine (which is used to initially anesthetize the animal before administration of isoflurane) binds to the NET[288] and this may be the cause of the differences observed between the anesthetized and awake states. In microPET studies with [¹⁸F]78 and [¹⁸F]79 in anesthetized rhesus monkeys peak uptake was achieved in all regions of interest after 15 min followed by a fast washout. High uptake was also observed in the caudate with tissue-to-caudate ratios for the thalamus, midbrain, and cerebellum of 1.13, 1.13, and 1.05, respectively, for [¹⁸F]78 and 1.18, 1.18, and 0.95, respectively, for [¹⁸F]79. Thus, the poor performance of [¹⁸F]78 and [¹⁸F]79 eliminates them as potential ¹⁸F-labeled NET PET tracers.

As mentioned in the DAT section above, cocaine binds to the NET and numerous tropane derivatives have been reported as ligands for the NET.[289–294] Preliminary microPET evaluation of a ¹¹C-labeled NET tropane derivative showed both SERT and NET binding and, therefore, a lack of NET selectivity.[295] The NET ligands talopram[296] and talsupram have been radiolabeled with ¹¹C and evaluated with PET but neither compound entered the brain in sufficient amounts to become a viable PET tracer.[297]

The goal of developing an ¹⁸F-labeled PET tracer for the NET is still a work-in-progress but progress is indeed being made. The results above with reboxetine derivatives demonstrate that some minor structural variation can be tolerated without a significant loss in NET affinity but it appears that placing an ¹⁸F-radiolabel on the methoxy or ethoxy group does reduce the NET affinity. A similar loss in affinity was observed when going from 35 to 36. Furthermore, defluorination also is a problem that can only be reduced, but not stopped, by incorporation of deuterium into these fluoroalkoxy groups. Defluorination was also observed with [¹⁸F]36 which, in combination with the ¹⁸F-NET ligand data above, suggests that radiolabeling with [¹⁸F]methyl and [¹⁸F]ethyl ethers and thioethers should be avoided if at all possible. Thus, the development of a reboxetine-based ¹⁸F-labeled PET tracer for the NET may require that the ¹⁸F-radiolabel be placed on one of the aromatic rings.

Summary

Progress is being made towards the development of fluorine-18 radiolabeled PET tracers for the DAT, SERT, and NET. The greatest number of potential tracers developed so far has been for the DAT due to the greater number of years that have been devoted to this target but also due in part to the concurrent research effort focused on developing a cocaine addiction therapeutic based on the tropane skeleton. The development of a library of potential therapeutic tropane compounds with high DAT affinity that could be radiolabeled with ¹¹C aided in the discovery of PET tracers for the DAT as well as the SERT, and this knowledge was then successfully translated to ¹⁸F-labeled tropane derivatives for both the DAT and SERT. The diphenylsulfide class of SERT ligands has already found success with ¹¹C-labeled PET tracers and a viable ¹⁸F-labeled PET tracer may soon be realized. Fewer tracers for the NET have been reported because PET imaging of the NET has received considerable less attention than the DAT or SERT but promising, though not optimal, compounds have already been reported. Further work towards this goal should also eventually provide a viable ¹⁸F-labeled NET PET tracer.

It is very difficult to predict if a proposed compound will have the desired properties of a PET tracer and so numerous compounds generally have to be synthesized and evaluated. As demonstrated with [¹⁸F]1-[¹⁸F]4, [¹⁸F]34, and [¹¹C]35, radiolabeling an antidepressant will not necessarily produce a good PET tracer. On the other hand, derivatives of antidepressants may produce candidate PET tracers as evidenced by the derivatives of 40 shown in Figure 6 and Table 3 and the derivatives of 70 shown in Figure 11 and Table 4. A high binding affinity to the target and a selectivity for that target are the first requirements that need to be met and any compound that cannot meet these requirements will most likely not succeed. Compounds with a high binding affinity and selectivity must also have a lipophilicity in the range of logP = ~1–3 or they will also fail. Furthermore, compounds which appear to have the appropriate affinity, selectivity, and lipophilicity can still fail due to poor kinetics or problems resulting from metabolism. Thus, the development of suitable PET tracers for the DAT, SERT, and NET is not an easy task, but with continued hard work by many research groups the goal will eventually be realized. It should, in fact, be possible to develop more than one tracer for each target and, as shown above, this has already been done. Which of the available tracers is the “best” would then have to be determined from side-by-side comparisons in the same subject under the same conditions (preferably in a conscious state) while taking into consideration the influence of metabolites that are generated from each individual tracer. This has already been done with many of the tracers reported above and this should continue to be done as part of tracer development. Because the goals of different studies of the same target may vary, the availability of several different tracers with slightly different properties may be of greater benefit to the PET imaging community than having only one “best” tracer for a given target.

Figure 7.

Diaryl Sulfide-Based Serotonin Transporter Ligands.