|Home | About | Journals | Submit | Contact Us | Français|
[18F]Nifene is an α4β2* nicotinic acetylcholine receptor (nAChR) agonist PET radioligand developed to provide accelerated in vivo equilibrium compared to existing α4β2* radioligands. The goal of this work is to analyze the in vivo kinetic properties of [18F]nifene with both kinetic modeling and graphical analysis techniques.
Dynamic PET experiments were performed on four rhesus monkeys (4F, 9–13 yr) using a MicroPET P4 Scanner. Studies began with a high specific activity [18F]nifene injection followed by a coinjection of [18F]nifene and unlabeled nifene at t=60 minutes. Sampling of arterial blood with metabolite analysis was performed throughout the experiment to provide a parent radioligand input function. In vivo kinetics were characterized with one-tissue (1TCM) and two-tissue (2TCM) compartment models, Logan graphical methods (both with and without blood sampling), and the multilinear reference tissue model (MRTM). Total distribution volumes (VT) and nondisplaceable binding potentials (BPND) were used to compare regional binding of [18F]nifene. Regions examined include the antereoventral thalamus (AVT), lateral geniculate body (LG), frontal cortex (FC), subiculum (SB), and cerebellum (CB).
There was rapid uptake and binding of [18F]nifene in nAChR-rich regions of the brain, which was appropriately modeled using the 1TCM. No evidence for specific binding of [18F]nifene in the CB was detected based on the co-injection studies, suggesting suitability as a reference region. VT values in the CB were found to be 6.91±0.61 ml/cm3. BPND values calculated with the 1TCM were 1.60±0.17, 1.35±0.16, 0.26±0.08 and 0.30±0.07 in the AVT, LG, FC, and SB respectively. For all brain regions, there was a less than 0.04 absolute difference in the average BPND values calculated with each of the 1TCM, MRTM, and Logan methods.
The fast kinetic properties and specific regional binding of [18F]nifene promote extension of the radioligand into preclinical animal models and human subjects.
Much work in the last 15 years has been devoted towards developing suitable PET radioligands to image α4β2* nicotinic acetylcholine receptors (nAChRs). These ligands bind with varying affinities to nAChRs containing either the β2 or α4 subunit, but most commonly to the α4β2 subtype, resulting in the notation α4β2*. The impetus for developing these probes stems from the interest of this receptor system's involvement in neurodevelopment, tobacco and alcohol addiction, and neuropathology. The disruption of cholinergic neurotransmission has been implicated in Alzheimer's disease, which has led to the application of acetylcholinesterase inhibitors as a treatment option for the symptoms of Alzheimer's disease(1,2). Evidence of declining nAChR densities has been associated with Alzheimer's disease, Parkinson's disease, and healthy aging(3,4). Alterations from normal nAChR functioning have been implicated in other neurodegenerative diseases, including epilepsy(5), tobacco abuse(6), schizophrenia(7), and deficiencies in neurodevelopment, such as effects due to fetal alcohol exposure(8).
[11C]Nicotine was the first radioligand developed for targeting α4β2* nAChR binding, however, it was found to have high non-specific binding and rapid dissociation rendering it unsuitable for PET imaging studies(9). The radioligand 2-[18F]-FA-85380 (2-[18F]FA) has been used the most extensively for PET studies of α4β2* nAChRs(10,11). Due to its slow in vivo behavior, PET scanning experiments using 2-[18F]FA require more than 5 hours of imaging for accurate measurement of α4β2* nAChR binding throughout all regions of the brain(12). Applications of this radioligand for PET studies have included studies of healthy aging(13), Alzheimer's disease(14), Parkinson's disease(15), and epilepsy(16).
The prolonged scanning procedure requirement of 2-[18F]FA has spurred the development of various new α4β2* nAChR radioligands with faster kinetic properties, including both agonists and antagonists(17). Agonist radioligands in particular are of great interest since other receptor systems, including the dopamine and serotonin systems, have found agonist radioligands to exhibit increased sensitivity to competing endogenous neurotransmitter levels(18,19). This feature will prove vital nAChR radioligands in evaluating acetylchoinestease inhibitors for applications with Alzheimer's disease. The α4β2* agonist radioligand 2-fluoro-3-[2-((S)-3-pyrrolinyl)methoxy]pyridine ([18F]nifene) was developed with the aim of improving upon the success of 2-[18F]FA by creating an analog with faster kinetic properties and a similar binding profile(20). Previous studies have shown that [18F]nifene exhibits fast transient equilibrium times of ~30 minutes, resulting in required scanning procedures of less than an hour. Elevated [18F]nifene binding in α4β2* nAChR-rich regions of the brain was also observed, with target to background binding levels suitable for research applications in preclinical research(21). These studies have demonstrated the viability of [18F]nifene PET experiments for translation into human subjects.
The goal of the work presented herein is to extend previous work with [18F]nifene by examining its behavior in arterial blood to obtain quantitative measures of α4β2* nAChR binding with both model-based and graphical analysis techniques. Analysis was conducted in the thalamic regions of the antereoventral thalamus and lateral geniculate body due to the high binding levels and their role in a variety of neurodegenerative deficits, while the subiculum and frontal cortex were also examined as targets due to alterations in binding associated with Alzheimer's disease and tobacco addiction(2,22). Blocking studies with unlabeled nifene were performed to evaluate the viability of the cerebellum to serve as a region of negligible detectable specific binding for reference region graphical analysis techniques. These studies provide a necessary step towards validating the use of [18F]nifene for studying the nAChR system during neurodevelopment and in disease specific models.
A secondary goal of this work was to improve the chemical purity of [18F]nifene from previously reported methods(20,21). Specifically, HPLC methods were modified to enhance the separation of intermediate N-Boc-[18F]nifene from precursor after substitution. A 16 MeV GE PETtrace cyclotron bombarded [18O]-enriched water with protons, creating [18F]fluoride, which was separated from the enriched water with a QMA cartridge (Waters). After elution of 18F with 1.2 ml kryptofix-K2CO3 solution, the 18F was azeotropically distilled in a customized chemistry processing control. Once dry, 0.5 mg of nitro precursor (2-nitro-3-[2-((S)-N-tertbutoxycarbonyl-3-pyrroline)methoxy]pyridine, ABX) in 250 μl anhydrous acetonitrile and 150 μl anhydrous dimethylsulfoxide was added for the reaction, which was heated to 120° C for 10 min. The mixture was then extracted with 4 ml methylene chloride and passed through a neutral alumina sep-pak. The methylene chloride was dried and the product purified with HPLC. The use of a new separation method, consisting of a C-18 Prodigy 10 μm 250×10 mm column (Phenomenex) with a mobile phase of 55% 0.05 M sodium acetate, 27% methanol, and 18% tetrahydrofuran at a flow rate of 8.0 ml/min, was incorporated into the [18F]nifene purification. This HPLC method was previously developed for the synthesis of [18F]MPPF, a serotonin 5-HT1A radioligand whose precursor contains a nitro- leaving group(23). Retention time of the N-Boc-[18F]nifene was approximately 14 minutes. The collected eluate (~5 cc) was then diluted in 50 ml water, trapped on a C-18 sep-pak, and eluted with 1 ml acetonitrile. Deprotection was performed with the addition of 200 μl 6 N HCl followed by heating at 80° C for 10 min. The mixture was then dried and pH adjusted to 7.0 with sodium bicarbonate. Ethanol (0.5 ml) and sterile saline were added for a total volume of 10 ml. This final product was purified with a preconditioned C-18 sep-pak (Waters) to remove any residual intermediate species (i.e. remaining Boc protected product), followed by sterile filtration with a 0.22 μm millipore filter for final formulation.
PET scans with [18F]nifene were acquired with 4 Macaca mulatta (rhesus monkey) subjects (4F, 6.6–11.9 kg). All housing and experimental procedures obeyed institutional guidelines and were approved by the institutional animal care and use committee (IACUC). Subjects were anesthetized prior to PET procedures with 10 mg/kg ketamine (i.m.) and maintained on 1–1.5 % isoflurane for the duration of the experiment. Atropine sulfate (0.27 mg i.m.) was administered to minimize secretions. Body temperature, breathing rate, heart rate, and SpO2 levels were recorded for the duration of the experiment. Radiotracer was administered via bolus injection in the saphenous vein, and arterial blood samples were withdrawn from the tibial artery in the opposing limb. After completion of the experiment, the subject was returned to its cage and monitored until fully alert.
A Concorde microPET P4 scanner was used for acquisition of PET data. This scanner has an axial field of view of 7.8 cm, a transaxial field of view of 19 cm, and a reported in-plane spatial resolution of 1.75 mm(24), which is slightly degraded to a 2.80 mm spatial resolution in the reconstructed image using our experimental conditions and processing methods. The subject's head was held in a stereotaxic headholder to obtain consistency in subject placement. A 518 s transmission scan was then acquired with a 57Co rotating point source. Data acquisition began simultaneously with a bolus injection of 89–126 MBq (2.4–3.4 mCi) [18F]nifene and continued for 60 minutes. A second injection of 76–124 MBq (2.1–3.4 mCi) [18F]nifene mixed with unlabeled nifene was administered at 60 minutes to examine specific binding in the cerebellum. Details of each study are presented in Table 1.
For blood analysis, a 2” NaI(Tl) well counter cross-calibrated with the PET scanner was used for radioactivity assay. Arterial blood was obtained in 500 μl volumes, with rapid sampling immediately following each [18F]nifene injection and slowing to 10 min sampling by the end of the experiment. Once withdrawn, whole blood samples were mixed with 50 μl heparinized saline, assayed for radioactivity, and centrifuged for 5 min. Plasma samples in 250 μl volumes were extracted, mixed with 50 μl sodium bicarbonate, and assayed for radioactivity. To denature the proteins in the plasma, 1 ml acetonitrile was added to the samples followed by 30 s centrifugation. The supernatant was extracted in volumes of 850 μl for radioactivity assay. Select samples were concentrated and spotted on aluminum backed silica gel TLC plates (Whatman). The plates were developed in a mobile phase of 50% methanol: 50% 0.1M ammonium acetate and exposed to a phosphor plate for at least 3 hours. Plates were read with a Cyclone storage phosphor system (PerkinElmer) to determine the relative concentrations of metabolites in the plasma for each sample.
To account for radiometabolites, TLC data were analyzed with ImageJ (NIH) and the fraction of radioactivity in the [18F]nifene peak relative to the total spotted radioactivity was measured for each sample. The measured time course of parent [18F]nifene in the plasma was parameterized by fitting the data to a bi-exponential function, producing a unique parent curve for each subject. This function describing the fraction of non-metabolized parent was then applied as a correction to the radioactivity measured in the acetonitrile extract to produce values conveying the time course of [18F]nifene parent present in the arterial blood. To examine the sensitivity of compartment model parameter estimates to variations in the metabolite correction, analysis of the cerebellum region was also performed with a group-based (averaged) metabolite correction.
Plasma protein binding was examined with Centrifree ultrafiltration units (Millipore) to determine the free fraction (fP) of radioligand in the plasma. Prior to the administration of radiotracer, blood was drawn from subjects and centrifuged to yield a 250 μl plasma sample. [18F]Nifene was added to each sample in 25 μl volumes and samples were incubated for 15 min at 37° C prior to separation, which occurred via ultrafiltration for 15 min at 2,000 g. A similar procedure was performed with 250 μl saline to correct for nonspecific binding of radioligand to the filtration unit. The stability of the free fraction has not yet been fully characterized for [18F]nifene and was therefore not incorporated into the subsequent analysis of the data; it is reported solely to provide additional information concerning the in vivo behavior of [18F]nifene.
PET data were histogrammed from list mode into time bins of 8×30 s, 6×1 min, 24×2 min, 12×30 s, 6×1 min, 18×2 min. The sinograms were reconstructed with two-dimensional filtered backprojection employing a 0.5 cm−1 ramp filter. Corrections for arc, scatter, attenuation, and scanner normalization were applied during reconstruction. The reconstructed images were processed with a denoising algorithm(25) using a 3×3×4 voxel filtering kernel. The final images had a matrix size of 128×128×63 corresponding to voxel dimensions of 1.90×1.90×1.21 mm3.
Circular regions of interest (ROIs) were drawn over brain regions defined on the PET images. Cerebellum ROIs were hand drawn on early summed images (0–8 min) with 4-voxel diameter circles over 3 consecutive transverse slices to include mainly grey matter while avoiding the vermis region, resulting in a region volume of 663 mm3 (152 voxels). Regions of high and moderate binding were identified in late summed images, from 20 to 40 minutes. Two general regions were selected from the thalamus, the antereoventral thalamus and lateral geniculate body, both drawn with 3-voxel diameter circles over 3 consecutive transverse slices, yielding corresponding volumes of 231 mm3 (53 voxels) and 161 mm3 (37 voxels). The frontal cortex was identified with 4-voxel diameter circles over 3 consecutive sagittal slices resulting in a volume of 763 mm3 (175 voxels), while the subiculum was delineated with 3-voxel diameter circles over 3 consecutive transverse planes giving a volume of 314 mm3 (72 voxels). Time-activity curves were extracted for all regions of interest.
Model-based and graphical analysis methods were both applied for analysis of the [18F]nifene PET data. One- and two-tissue compartment models (1TCM, 2TCM) were used for the model-based analysis. The compartment model analyses are described by the following equations:
where CA is the concentration of parent radioligand in the arterial plasma. K1 (ml/min/ml) and k2 (min−1) describe plasma-to-tissue transport and tissue clearance, respectively. CND represents the concentration of radiotracer present in the nondisplaceable compartment, which contains both free and nonspecifically bound radioligand, and CS represents the concentration of radiotracer in the specifically bound compartment. Transfer between these compartments conveying the reversible process of specific binding is described by the association rate k3 and the dissociation rate k4, which both have units of min−1. The 1TCM contains only the concentration of total radioactivity in the tissue, CT, and includes only the parameters K1 and . In regions of negligible specific binding, is equal to k2, whereas in regions with rapidly equilibrating specific binding, is equivalent to k2(1+k3/k4)−1. Volumes of distribution (VT) were calculated for each region using data from the high specific activity injection (t<60min) with the equation VT = K1/ for the 1TCM and VT=K1/k2(1+k3/k4) for the 2TCM(26). VT were also calculated with the Logan graphical method with blood sampling using a linearization time of t*=20 min for all regions to provide for comparison with a graphical analysis parameterization technique(27).
For modeling calculations, the decay corrected PET signal in a given region of tissue represents CT = CND + CS + fVCWB , where fV is the fractional blood volume, assumed to be 0.04, and CWB is the concentration of radioactivity in the whole blood. Parameter estimations were performed with COMKAT software(28) with configurations for the 1TCM and 2TCM. Model comparison was evaluated with the corrected Aikaike information criteria where k is the number of parameters in the model, n is the number of observations used for the fit and RSS is the residual sum of squares for the fit(29).
Analysis of data from the cerebellum included PET data for the second low specific activity [18F]nifene injection (t>60 min). The plasma input function from 15 to 60 minutes was fit to a bi-exponential function, and then extrapolated out past the time of the second injection. These extrapolated values were then subtracted from the observed data to strip away the first injection's residual radioactivity from the second injection. A similar method was used to correct the cerebellum TACs by fitting the cerebellum data from 20 to 60 minutes to a bi-exponential function, then subtracting the extrapolated data from the second injection. Cerebellum VT values were calculated for the second injection with the compartment modeling techniques described above, with distinct parameters determined for each separate injection.
Receptor specific binding of [18F]nifene was characterized by estimation of the binding potential (BPND). BPND values were calculated with two graphical analysis techniques, the Logan graphical method(30) and the multilinear reference tissue model(MRTM, 31). The three methods previously discussed for VT calculation (1TCM, 2TCM, and Logan with blood sampling) were also used to result in five total different approaches to calculate BPND. For approaches with VT estimation, the relation BPND=VT/VND-1 was used, where VND is the volume of distribution for nondisplaceable uptake, and is equivalent to the estimation BPND=DVR-1 employed in the reference region approaches. The cerebellum was assumed to be a reference region devoid of specific binding where VT=VND, consistent with findings presented here (see results) and reported previously(32,21). Differences in parameter estimates between analysis methods were investigated by plotting BPND values calculated from a give method against each of the other methods, and fitting the data to a line. The slope was then subjected to a linear regression t-test to determine the deviation of the slope from unity.
Because Logan reference region analysis is widely used for binding quantification, the proper linearization time of this approach was closely examined. Values reported herein employed a linearization time of t*=20min for all regions and omitted the mean efflux term . The necessary time for stability of BPND values calculated with this method was examined by varying the length of data analyzed from 30 to 60 minutes. Additionally, the inclusion of min−1 and a t*=5 min was also examined to increase the number of points on the Logan plot for improved quantification.
The new HPLC purification procedure improved separation of the remaining nifene precursor from the intermediate N-Boc-[18F]nifene after substitution, as demonstrated by Figure 1. This modified first purification also eliminated the need for a second HPLC separation after deprotection, since this step could instead be performed with a simple C-18 sep-pak extraction. As a result, the overall synthesis time was reduced from 2.5 h to 1.5 h, with overall batch yields ranging from 500–1500 MBq, corresponding to 10–20% decay corrected radiochemical yield. Chemical purities were improved by 380% with the new HPLC purification method, primarily from the elimination of the N-Boc species, while end of synthesis specific activities were consistently in excess of 200 GBq/μmol.
[18F]Nifene is rapidly metabolized in the plasma, with all radiolabeled metabolite species being more polar than the parent compound (Figure 2). The radioactivity in the plasma due to metabolites was predominately due to two overlapping peaks, along with trace amounts of a third species. Parent radioligand concentrations in plasma quickly fell to 40% after 15 minutes and 25% in 50 minutes following [18F]nifene administration (Figure 3A). After the initial peak of [18F]nifene, radioactive parent monotonically decreased, consistent with a two-exponent function. The slower of these rates was 0.024±0.003 min−1 corresponding to a half-life in the plasma of 29 min (Figure 3B). Plasma protein binding was also assessed to measure fP for nifene. An average of 46±4% of the total radioactivity in the plasma was due to free [18F] nifene.
The [18F]nifene analysis with 1TCM and 2TCM using the measured arterial input parent functions revealed that the 2TCM was not statistically justified based upon the cAIC in all regions with the exception of one subject (M2). For this subject, however, VT estimates were within 2% for both the 1TCM and 2TCM methods. An illustrative example of a subject fit to the 1TCM is shown in Figure 4. VT values were also estimated using the Logan graphical method with blood sampling, and are shown with the 1TCM results in Table 2 (2TCM results are not shown due to the model preference of the 1TCM). The thalamic regions of the antereoventral thalamus and lateral geniculate body yielded the highest [18F]nifene VT values of 17.95±1.66 ml/cm3 and 16.17±1.64 ml/cm3 with the 1TCM, while the frontal cortex and subiculum were found to have intermediate values of 8.69±0.43 ml/cm3 and 8.96±0.48 ml/cm3. The cerebellum had the lowest measure of VT, 6.91±0.61 ml/cm3. The Logan method and 1TCM were in close agreement with an average difference across all regions of 3.4% and a greatest discrepancy of 8%. The VT values in the cerebellum calculated with a global metabolite corrected input function differed from VT calculated with individual corrections by less than 4% in all cases.
Blocking studies with unlabeled nifene (i.e. low specific activity [18F]nifene) were conducted to closely examine if α4β2* specific radioligand binding in the cerebellum could be detected. For three subjects, compartment modeling of the high mass injection in the cerebellum yielded an average VT of 6.8±0.1 ml/cm3, which closely matches the average high specific activity VT for these subjects of 6.7±0.6 ml/cm3. A visual comparison of the high and low specific activity time courses in the cerebellum was also made by subtracting the residual signal from the first injection from the second and normalizing the curves to the injected dose. An example is shown in Figure 5 for the subject where blood sampling was unavailable for the second injection. All curves follow highly similar time courses, suggesting negligible perturbation due to the presence of blocking doses of nifene.
BPND values for the various methods are shown in Table 2. In the high binding thalamic regions, values of 1.60±0.17 and 1.35±0.16 were measured with the 1TCM in the antereoventral thalamus and lateral geniculate body. The lower binding regions of the subiculum and frontal cortex yielded respective values of 0.30±0.07 and 0.26±0.08. The largest difference in BPND between all methods for all regions was 15% (in the frontal cortex), while the average difference was 1%, suggesting good agreement between the various methods. No statistically significant (α<0.1) difference was detected between any of the methods presented here, indicating that all of the methods yielded consistent results.
Using a t*=20 min with omission of for the Logan reference region method, a scan time of 45 minutes or less yielded BPND values within 5% of those generated with a scan time of 60 minutes in all subjects under the aforementioned analysis conditions, indicating good stability of the metric. Additionally, the inclusion of min−1 and a t*=5 min for this analysis method revealed BPND values within 2% of values calculated without , indicating agreement between the two methods. A voxel-wise image of BPND values calculated with the Logan graphical method was generated for visualization of [18F]nifene uptake (Figure 6).
[18F]Nifene was developed to fulfill the need for a rapidly equilibrating α4β2* PET radioligand to advance research of this system by the neuroimaging community. The fast equilibration times of [18F]nifene provide advantages in both greatly reducing the time of scan procedures and potentially detecting changes in endogenous acetylcholine levels. The 45 minute imaging requirement for [18F]nifene quantification is approximately 7-fold shorter compared to the current α4β2* standard radiotracer, 2-[18F]FA, providing reductions in both experimental complexity and cost. More recently developed α4β2* radioligands, including [18F]AZAN(33), [18F]ZW-104(34) and (−)-[18F]NCFHEB(35), show improvements over 2-[18F]FA both in increased binding levels and reduced scan times, although each requires at least 90 min acquisitions for quantification, at least double the time required for [18F]nifene. To build upon our earlier studies, we have made significant improvements in radiochemical production, and included the measurement of an arterial input function for use in the assay of specific binding and blocking studies with unlabeled nifene.
Arterial blood samples were acquired to examine the time course of [18F]nifene available to the tissue and to quantify the presence of radiolabeled metabolites. Radio-TLC provided a well separated profile of [18F]nifene and radiolabeled metabolites, which allowed for characterization of radiolabeled species in the plasma while avoiding the use of HPLC analysis due to poor data quality resulting from low count rates and injectate purification. Metabolism of [18F]nifene occurred rapidly at first, then slowed to a metabolism rate with a half-life of 63 min. The increase of radiolabeled metabolites in arterial blood samples was consistent between all subjects, ranging between 68%–74% at 30 minutes post-injection. Differences in VT in the cerebellum between the global and individual metabolite corrected input functions were less than 4%, indicating that a moderate level of uncertainty could be tolerated in the measurement of radiolabeled metabolites. This finding suggests the potential use of a global parent metabolite correction, although additional validation would be required to examine age and gender dependent variations in nifene metabolism. All detected radiolabeled metabolites were less lipophilic than the [18F]nifene parent, suggesting that these metabolites cross the blood-brain barrier at a substantially lower rate than nifene. These observations are in agreement with previous studies of [18F]nifene and its metabolites in the rat brain(36). Additionally, the ratio of radioactivity in the cerebellum to radioactivity of the parent [18F]nifene in the plasma was found to be constant or slowly decreasing after 40 minutes, suggesting that there was no buildup of radiolabeled metabolites in the brain.
The possible presence of α4β2* binding in the cerebellum was examined by introducing a second injection of [18F]nifene co-injected with unlabeled nifene. Previously, it was found that 0.03 mg/kg (−)nicotine qualitatively had no effect on cerebellum time-activity curves, suggesting suitability as a reference region(21). The work presented herein confirms this result with quantitative analysis performed with blood sampling. For the three subjects where blood sampling was available to the end of the study, the introduction of high-mass nifene yielded VT values (6.8±0.1 ml/cm3) which were consistent with the values calculated from the first injection (6.7±0.6 ml/cm3) with the same 1TCM analysis, found here to be appropriate in evaluating regions of elevated [18F]nifene binding. The similarity in VT despite high levels of unlabeled nifene suggests that the specific binding component (VS) in the cerebellum is lower than the sensitivity limits of the PET scanner and analysis techniques used here. This finding of negligible VS is in agreement with other studies examining 2-[18F]FA in the cerebellum of the rhesus monkey(32). Other PET studies have found small but significant cerebellar α4β2* nAChR expression in baboons(10) and humans(12), rendering it problematic as a reference region in these species. The moderate BPND values of [18F]nifene, however, may allow for the use of a valid reference region in humans in white matter regions such as the corpus callosum or the pons, as previously demonstrated with 2-[18F]FA(14). The corpus callosum was briefly examined here as a reference region, however, the lack of MRI data resulted in high noise from the extracted TACs and consequently higher variability in thalamic BPND values compared to the values presented herein.
In brain regions with elevated [18F]nifene binding, compartment modeling yielded similar quality of fitting results (ie. sum of squares) with both the 1TCM and 2TCM, suggesting the additional parameters of the 2TCM were not statistically warranted as specified by the cAIC. The selection of the 1TCM in regions of elevated binding suggests a lack of binding parameter identification due to fast equilibration between the nondisplaceable and specifically bound compartments. This result was also found in previous PET studies with [11C]nicotine(9), however, no other α4β2* nAChR radioligands to our knowledge exhibit this behavior. The fast kinetics of [18F]nifene are also reflected in the short 45 min scanning procedure requirement, which is advantageous in minimizing discomfort to the subjects when extending these imaging methods to diseased populations. The rapid binding and dissociation of [18F]nifene may also provide increased sensitivity of [18F]nifene binding to changes in endogenous levels of acetylcholine in vivo, as previously observed by in vitro work with acetylcholinesterase inhibitors(37).
BPND values were highest in thalamic regions of the brain, with intermediate levels of binding in the frontal cortex and subiculum. The level of [18F]nifene binding in the frontal cortex was consistently lower than that of the subiculum. Similarly, the reduction in specific binding after the second (high mass) [18F]nifene injection followed the same rank order decrease across these regions. Furthermore, In the study with the lowest receptor occupancy by unlabeled ligand, the measured occupancy in the frontal cortex (24%) and subiculum (31%) compared well with the value measured in the thalamus (34%). Although the coefficient of variation across the 4 subjects was slightly higher in the intermediate binding regions (~30%) compared to the antereoventral thalamus (11%), the consistent rank order and agreement in receptor occupancy levels in the frontal cortex and subiculum suggest that [18F]nifene provides adequate sensitivity to α4β2* binding in regions of intermediate uptake. We therefore do not rule out potential applications of [18F]nifene detecting small changes in α4β2* nAChR density in cortical regions, such as findings examining 2-[18F]FA uptake in patients with Alzheimer's Disease(14). Partial volume effects should be considered in these cortical and hippocampal regions, particularly in patients with brain atrophy. Such a correction was not included here because no MRI data were acquired potentially lowering the measured levels of binding.
Comparison of VND between [18F]nifene and other α4β2* radioligands provides insight into the speculated differences in their imaging properties. Multiple-injection studies of 2-[18F]FA in baboons found VND values of 4.90±0.46 g/ml in the thalamus and 4.25±0.48 g/ml in the cerebellum(38). Studies of 2-[18F]FA in the rhesus monkey yielded a VND value 4.32±0.17 ml/cm3 (when removing the correction for plasma free fraction)(32). Our present work indicates VND values of 6.9±0.6 ml/cm3 for [18F]nifene in the rhesus monkey. Similarly, the clearance of radioligand from the blood is much faster for [18F]nifene (0.024±0.003 min−1) than 2-[18F]FA (0.0056±0.0017 min−1). This large VND value for [18F]nifene indicates that it is more readily taken up from the blood into the nondisplaceable compartment (eg. free and nonspecifically bound radiotracer) and retained.
To gauge the level of α4β2* receptor occupancy in target regions, we have also examined data from the second injection in regions with specific binding. The rapid in vivo kinetics of [18F]nifene allows for approximations of the change in specific binding following high mass injections of nifene. These data can be used to estimate the in vivo equilibrium dissociation constant (KDapp) of [18F]nifene for α4β2* nAChRs through the use of a scatchard-type analysis(39). For this analysis, the 1TCM was used to calculate BPND values for both the first and second injection in the thalamus. The ratio of these two values was used as a measure of in vivo receptor fractional occupancy (Occ=1– BPND(Block)/BPND(Baseline)). The free radioligand (F) was estimated by averaging the radioligand signal in the reference region 20 minutes post-coinjection (t=80 min) to the end of the study, and dividing by the specific activity. Plotting receptor occupancy against F yielded a nonlinear Scatchard plot, which was fit to the equation Occ=F/(KD+F) to estimate a value of KDapp, as shown in Figure 7. The analysis yielded a preliminary KDapp value of 3±1 pmol/ml. This value is 2–3 times greater than the thalamic KDapp value for 2-[18F]FA reported by Gallezot and colleagues(38). When compared to 2-[18F]FA, the larger KDapp of nifene is consistent with its smaller BPND, assuming they compete for the same pool of receptors. This analysis provides only an approximation of KDapp due to the small number of subjects and the uncertainty in the precise measurement of the free nifene concentration, thus additional studies will be required to establish the precision and variability of this estimate. The present work provides a basis for guiding future experimental design of improved identification of Bmax and KD.
The present work characterized the behavior of [18F]nifene in the blood and found the 1TCM to most appropriately describe the data, further demonstrating the rapid equilibration times of [18F]nifene and suggesting potential applications in measuring changes in endogenous acetylcholine levels. The cerebellum was quantitatively confirmed as a suitable reference region in the rhesus monkey, and sensitivity of [18F]nifene to small changes in binding in areas of low uptake was found. These characteristics, combined with the requirement of 45 minute scan times for accurate quantification, give [18F]nifene unique advantages over other available α4β2* nAChR radioligands and promote the extension of [18F]nifene to disease specific animal models with the potential for studies in human subjects.
The authors thank the following for their contributions to this research: Professor R. Jerry Nickles and Drs. Jonathan Engle and Greg Severin for technical discussions and gracious assistance in isotope production. Julie Larson, Leslie Resch, and the staff at Harlow Center for Biological Psychology (RR000167) for assistance in animal handling and data acquisition. This work was supported by NIH grants AA017706, CA142188.
Financial Support: NIH Grants AA017706, CA142188