INTRODUCTION

Several lines of evidence suggest that dopamine (DA) D3 receptors (DRD3) may be promising targets for drug addiction treatment (Heidbreder et al, 2005; Le Foll et al, 2005). One, the anatomy of the DRD3, associated with the mesolimbic DA system, suggests that it is well positioned to influence drug-seeking and relapse mechanisms (Murray et al, 1994). Two, preclinical and post-mortem human brain studies suggest elevated DRD3 levels after chronic exposure to drugs of abuse (Staley and Mash, 1996). Three, positron emission tomography (PET) data from our group and others have echoed preclinical findings in showing that in vivo DRD3 levels are also higher in individuals who abuse stimulants and correlates with addiction-relevant behavior and traits (Boileau et al, 2012; Matuskey et al, 2014; Payer et al, 2013). Finally, in animal models, DRD3-selective antagonists have been shown to decrease seeking of, and relapse to, a variety of drugs of abuse and it has been hypothesized that the DRD3 modulates the motivation to seek drugs and notably contributes to the relapse phenomenon (Heidbreder et al, 2005; Le Foll et al, 2005).

Unfortunately, there are no DRD3-selective antagonists currently available in the clinic. It has been recently reported that buspirone, a medication used for generalized anxiety disorder (Apter and Allen, 1999), has DRD3 antagonist properties, raising the possibility that it may be the only DRD3 antagonist currently available. The in vitro data indicate a twofold affinity and an 11-fold functional selectivity for DRD3 over 5HT1A, and 70-fold affinity over DRD2 (Kula et al, 1994), with metabolites of buspirone also binding with higher affinity to DRD3 relative to DRD2 (Bergman et al, 2013). In addition, buspirone is able to disrupt cocaine self-administration in non-human primates (Mello et al, 2013). For all these reasons, buspirone is a potential promising tool to study the impact of DRD3 blockade in humans. In support of this, a recent study in non-human primates has suggested that buspirone occupies the DRD3 at therapeutic doses (Kim et al, 2014).

Here we used the DRD3-preferring radiotracer [11C]-(+)-PHNO to investigate whether oral buspirone (0, 30, 60, and 120 mg) occupies DRD3 vs DRD2 in healthy subjects. Note that [11C]-(+)-PHNO binding can be interpreted in a region-dependent manner, with binding in dorsal striatum reflecting DRD2 receptor availability, and binding in hypothalamus and substantia nigra (SN), ventral pallidum (VP), globus pallidus (GP), and ventral limbic striatum (LST) reflecting 100%, 75%, 65%, and 26% D3 availability, respectively (Tziortzi et al, 2011). Our main working hypothesis was that orally administered buspirone would lead to a significant dose-dependent occupancy of DRD3; that is, selective decrease in [11C]-(+)-PHNO BPND in DRD3-rich areas vs DRD2 areas.

MATERIALS AND METHODS

Subject

All procedures were approved by the Centre for Addiction and Mental Health Research Ethics Board and complied with the 1975 Helsinki Declaration (5th revision, 2000). Subjects (adults, male or female (n=8), >19 years old) were recruited from the community and provided written informed consent and participated in a comprehensive screening interview. All met the following criteria: (1) no past/present significantmedical condition including neurological illnesses or head trauma; (2) normal physical exam (12-lead electrocardiogram, normal routine blood tests); (3) no past/present Axis I psychiatric diagnoses as per Mini-International Neuropsychiatric Interview version 5.0.0; (4) no condition that precludes use of buspirone; (5) no MR scanning contraindications; (6) no claustrophobia; (7) no current pregnancy/breastfeeding; (8) no current use or use during the previous month of medication that may affect the CNS, including monoamine oxidase inhibitor (MAOI) or positive during drug screening for drugs of abuse; and (9) no exposure to radiation in the past 12 months exceding limit for subjects participating in research with PET.

Procedure

We used a within-subjects, fixed-order drug schedule design to characterize the dose occupancy of buspirone at the DRD3 DA receptor. Subjects were blind to dosing regimen. Each completed four PET scans, on separate days, at least 2 days apart following the oral administration of placebo (0 mg), and 60, 30, and 120 mg of buspirone (in identical-looking capsules). Injection of the PET tracer [11C]-(+)-PHNO was timed 60 min after dosing, corresponding to the expected plasma peak of buspirone (Mahmood and Sahajwalla, 1999; Meltzer et al, 1983).

PET Image Acquisition

The radiosynthesis of [11C]-(+)-PHNO has been described in detail elsewhere (Wilson et al, 2005).

PET studies were performed using a high-resolution head-dedicated PET camera system, CPS-HRRT (Siemens Medical Imaging). The in-plane resolution of the scanner is ~2.8 mm full-width at half-maximum (FWHM). Transmission scans were acquired using a 137Cs (T½=30.2 years, E=662 keV) single-photon point source. The raw data were reconstructed by filtered-back projection. A custom-fitted thermoplastic mask (Tru-Scan Imaging) was made for each subject to reduce movement during the acquisition. A total of ~370±40 MBq (~10±1 mCi) of [11C]-(+)-PHNO was injected as a bolus into an antecubital vein. Scanning time was 90 min in list mode, and then 30 frames were defined: 1–15 of 1-min duration and 16–30 of 5-min duration.

Blood samples for plasma level of buspirone and metabolites were taken before and 60 and 180 min after dosing. Prolactin levels were measured before and 60, 90, and 180 min following dosing (to investigate DRD2 antagonist effect). Subjective assessments of drug effects were conducted at each PET visit with Visual Analogue Scale (VAS) (0, 60, and 180 min after dosing).

MRI Image Acquisition

Subjects underwent standard proton density-weighted brain magnetic resonance imaging (MRI) on a Discovery MR750 3T MRI scanner (General Electric, 3T MR750) (slice thickness 2 mm; interleaved; slice number, 84; repetition time, 6000 ms; echo time, 8 ms; number of excitations, 2; acquisition matrix, 256 × 192; FOV, 22 × 16.5 cm) to aid region-of-interest (ROI) delineation of the PET images.

Plasma Levels of Buspirone

Plasma levels of buspirone and two major metabolites, 5-hydroxybuspirone and 6-hydroxybuspirone, were measured in plasma by LC/MS/MS (see Supplementary Material) and were used in the one-site binding model described below.

PET Image Analysis

ROI-based analysis

ROI delineation and time activity curve analyses were performed using ROMI (details in Rusjan et al, 2006). Functional subcompartments of the striatum (Martinez et al, 2003) including the associative striatum (AST), limbic striatum (LST), and sensorimotor striatum (SMST) were chosen as ROIs. Delineation for the GP (whole), VP, and SN is described elsewhere (Boileau et al, 2012).

The [11C]-(+)-PHNO-specific binding (BPND) was estimated in each ROI using the simplified reference tissue method (SRTM; Lammertsma and Hume, 1996), with cerebellar cortex as reference region. The cerebellar cortex template excludes the vermis as well as lobules IX and X (uvula and nodulus) and a tissue-classification process (described in Rusjan, et al. 2006) removes all voxels with the cerebellar ROI that contain white matter (that is, voxels with a probability of gray matter <99% are excluded). Parameter estimation was performed using PMOD (version 2.8.5; PMOD Technologies, Zurich, Switzerland). Because of the low signal present in hypothalamus and the difficulty in delineating this structure, we did not attempt to quantify the [11C]-(+)-PHNO BPND in this brain area.

Receptor occupancy, defined as the percentage reduction in [11C]-(+)-PHNO BPND from 0 mg (placebo) scan to the buspirone-exposed state, was calculated for each subject using the following Equation (1):

Comparisons between [11C]-(+)-PHNO BPND in ROIs as well as regional receptor occupancies were conducted by using repeated-measures ANOVAs (SPSS 20.0, SPSS). Sphericity was assessed with the Mauchly test and, when indicated, correction was made with Greenhouse–Geisser adjustments. When appropriate, least significant difference t-tests, Bonferroni corrected, were applied to determine the significance of regional differences in BPND between conditions (doses).

As we are interested in the DRD3 vs DRD2 selectivity of buspirone, we estimated the DRD3 fraction (ƒD3) in areas of moderate to high DRD3 signal using previously established regional fractions (Tziortzi et al, 2011). This was done by multiplying [11C]-(+)-PHNO BPND in each ROI by their estimated regional fraction (as per Girgis et al, 2011). The average [11C]-(+)-PHNO BPND within the DRD3 ‘compartment’ (that is, ƒD3) as well as occupancy values within the ƒD3 were calculated (as described above) and compared with the D2 ‘compartment’ (that is, [11C]-(+)-PHNO BPND in dorsal striatum where 100% of the signal is associated with D2; ƒD2) by using a repeated-measures ANOVA (as described above). Relationship between [11C]-(+)-PHNO BPND, occupancies and continuous variables including plasma buspirone (and metabolites), prolactin, and self-reported drug effects were investigated using correlation analysis (Pearson’s product moment correlation).

Voxel-Wise Analysis

Voxel-wise parameter estimation of [11C]-(+)-PHNO BPND was implemented using receptor parametric mapping (RPM) (Gunn et al, 1997). The BPND map images were spatially normalized into the Montreal Neurological Institute (MNI) brain space by nearest neighbor interpolation and with a voxel size fixed in 2 × 2 × 2 mm using SPM8 software (Wellcome Trust Centre for Neuroimaging, London, UK). The normalized images were smoothed with a Gaussian filter in each coordinate direction with a kernel of 3 mm and an average [11C]-(+)-PHNO BPND was created at every dose for visualization purposes. Normalized [11C]-(+)-PHNO BPND images were also entered into a GLM and statistically investigated in SPM 8. To minimize multiple comparisons and given our a priori regional hypotheses, voxels investigated for significant effects were limited to voxels with a BPND value >0.1.

One-Site Modeling

Regional brain [11C]-(+)-PHNO occupancy and buspirone/metabolites dose–response relationship was analyzed by a one-site binding model. Plasma levels of buspirone and its major metabolites (presumably binding to DRD3/DRD2) 5-hydroxybuspirone and 6-hydroxybuspirone (average between 60 and 180 min after dose; corresponding to the start and end of the PET acquisition time and the peak in buspirone plasma levels) were entered into the analyses (Equation 2):

where ED50 represents the plasma level of buspirone or it metabolites that results in 50% receptor occupancy, A1 is the saturated occupancy, A2 is the baseline occupancy, and h is the Hill constant. The model constraints were ED50>0, h > 0, A1 ≤ 100, and │A2│ ≤ 10 allowing variance of binding at baseline given the test–retest variability, with A1 and A2 shared between buspirone and metabolites in each area being analyzed. Despite the presence of at least two brain binding sites for buspirone and metabolites in some brain areas, the simplified one-site model is useful to estimate the efficacy (maximal %Occupancy) and apparent efficiency (ED50) of blocking of tracer binding among the brain regions given variable dose adsorption among subjects (see Supplementary Figures 3 and 5). The one-site nonlinear least square fitting and comparison (Akaike information criterion (AIC)) with simpler one-site hyperbolic or two-sites functions (Graff-Guerrero et al, 2010) were performed in GraphPad Prism software (Version 4, GraphPad Software).

RESULTS

Eight subjects (4 male/4 female, all Caucasians) were recruited for this study. Two subjects were nauseous after [11C]-(+)-PHNO; this led to scan interruption and study termination. These two subjects provided partial data: one completed the baseline scan and the other completed the baseline scan as well as the 60 and 30 mg scans. The average age of the sample was 35.4±13.6 (23–56 years old), their body mass index was within normal range (23.4±3.7 kg/m2), all tested negative for drugs of abuse, and none were nicotine smokers. The dose per kg of buspirone corresponded to 0.44±0.11 mg/kg for the 30 mg scan, 0.87±0.22 mg/kg for the 60 mg scan, and 1.74±0.44 mg/kg for the 120 mg scan. There were no differences in scan parameters across doses (mass injected (μg) 2.0±0.4; amount injected (mCi) 9.1±1.0; specific activity (mCi/μmol) 1146.0±316.3).

ROI Analysis

The [11C]-(+)-PHNO binding was moderately blocked by buspirone in a dose-dependent manner (F(5, 25) = 6.420; P=0.005). Pairwise comparison revealed that [11C]-(+)-PHNO BPND was reduced from placebo after 120 and 60 mg of oral buspirone but not after 30 mg (−5%, P=0.50; Table 1) (effect size in D3-rich SN and dorsal striatum: Cohen’s d: 1.2 and 2.4, respectively). This effect corresponded to a mean decrease in [11C]-(+)-PHNO BPND (all ROIs included) of −22% (P=0.012) and −15% (P=0.007) after 120 and 60 mg of buspirone respectively relative to 0 mg. The [11C]-(+)-PHNO BPND at the 30 mg dose was also higher than at the 120 mg dose (−18%, P=0.027). The ROI × dose interaction was not significant (F(15, 75)=1.524; P=0.118), suggesting that buspirone administration did not differentially affect [11C]-(+)-PHNO BPND across ROIs (Table 1) (effect size for the interaction: Cohen’s d: 1.3). An ANOVA investigating [11C]-(+)-PHNO BPND associated with the ƒD3 and ƒD2 yielded the same results: that is, an effect of dose (F(3, 15) = 6.981; P=0.004), suggesting that 120 and 60 mg of orally administered buspirone reduced [11C]-(+)-PHNO BPND by 25% and 16%, respectively (P<0.05). Pairwise comparisons investigating differences in occupancy associated with the ƒD3 and ƒD2 suggested that at the lowest dose of buspirone, occupancy was greater in ƒD2 vs ƒD3 (P=0.015; Table 1 and Figure 1).

Table 1 Percentage Receptor Occupancy by Buspirone in ROIs
Full size table
Figure 1
figure 1

[11C]-(+)-PHNO BPND associated with ƒD3 and ƒD2.Shaded circles represent dose: from lightest to darkest, respectively 0, 30, 60, and 120 mg. *Significantly different from 0 mg. ƒD3 : DRD3 fraction calculated in areas of moderate to high DRD3 signal using previously established regional fractions (Tziortzi et al, 2011). This was done by multiplying [11C]-(+)-PHNO BPND in each ROI by their estimated regional fraction (as per Girgis et al, 2011). ƒD2:DRD2 fraction corresponding to [11C]-(+)-PHNO BPND in dorsal striatum.

PowerPoint slide

Full size image

We performed partial volume effect correction (as described in Rousset et al, 1998) and this did not change the results (Supplementary Figure 1). A main effect of dose was observed (F(3, 15)=4.791; P=0.016); however, the effect was not region dependant (F(12, 60)=0.564; P=0.862). Note that we did not correct for partial volume effect in the VP as our template for partial volume effect does not include this region.

One case had lower [11C]-(+)-PHNO BPND during placebo vs 60 and 120 mg doses that could not be explained by motion artifact, other drug on board (a clean toxicology screen was provided). Removing this case from the analysis (outlier test on raw data does not suggest that he is an outlier, but outlier test on occupancy value does) revealed a significant ROI × dose interaction (F(15, 60)=3.707; P<0.001). Follow-up pairwise comparison suggested that relative to placebo the 60 mg dose significantly decreased [11C]-(+)-PHNO BPND in all ROIs (SN: P=0.05, AST: P=0.002, LST: P=0.006, SMST: P=0.009, GP: P=0.007) except VP (P=0.08), and that the 120 mg dose reduced [11C]-(+)-PHNO BPND in all ROIs (AST: P=0.014, LST: P=0.015, SMST: P=0.04, GP: P=0.009, VP: P=0.002) except SN (P=0.09), suggesting overall a more variable occupancy in D3-rich SN and VP. The ANOVA investigating [11C]-(+)-PHNO BPND associated with the ƒD3 and ƒD2 without the ‘outlier’ yielded the same results: that is, no interaction (F(3, 12) = 1.981; P=0.225).

Effect on Prolactin

An ANOVA investigating whether buspirone (dose) affected prolactin levels yielded a significant dose × time interaction (F(9, 45)=4.479; P<0.001), suggesting that orally administered buspirone increased prolactin levels after 60 mg (P=0.02) and 120 mg (P=0.02) of buspirone. Peak effects occurred 60 min after buspirone. There were no significant differences in baseline, predrug prolactin level between doses (all P>0.05). See Supplementary Figure 2A. Plasma levels of prolactin correlated with occupancy after orally administered buspirone (120 mg in ƒD2: −0.87, P <0.05; see Figure 2b).

Figure 2
figure 2

(a) Average [11C]-(+)-PHNO BPND maps after 0, 30, 60, and 120 mg of orally administered buspirone, overlaid on top of PD MRI template in MNI space. (b) T-statistical map of difference between [11C]-(+)-PHNO BPND at baseline and at dose 120 mg. P (FEW-corrected)=0.008; KE=249, peak T=23.83 in (x: 18, y: 8, z: 10). SN: substantia nigra, VP: ventral pallidum.

PowerPoint slide

Full size image

Effect on Plasma Levels of Buspirone and Metabolites

Orally administered buspirone increased plasma levels of buspirone after 60 and 120 mg, at 60 and 180 min of dosing (F(6, 30)=8.906; P=0.01 and P<0.05), whereas the 30 mg dose only increased buspirone plasma levels 180 min after dosing (P<0.05). The ANOVA for 6-OH and 5-OH buspirone yielded the same finding (6-OH: dose × time interaction F(6, 30)=5.893; P<0.001; 5-OH dose × time interaction F(6, 30)=5.009 P=0.001), suggesting again that oral administration significantly increase metabolites after 60 and 120 mg, 60 and 180 min after dosing (P<0.05), whereas the 30 mg dose only increased metabolites 180 min after dosing (P<0.05). Supplementary Figure 3.

Effect on Subjective Measures

Buspirone significantly increased self-reported ‘Dizziness’ (F(3, 18)=14.798; P<0.001) and ‘Drowsiness’ (F(3, 21)=4.591; P=0.013). Pairwise comparisons revealed that the effect on ‘Dizziness’ was present at all doses, whereas ‘Drowsiness’ was only reported after the highest doses (60 and 120 mg; P<0.05; Supplementary Figure 4). Self-reported behaviors ‘Dizziness’ and ‘Drowsiness’ did not correlate with percentage of occupancy in any region.

Voxel-Wise Analysis

Results of our voxel-wise analysis is consistent with our ROI analyses in showing an effect of buspirone on [11C]-(+)-PHNO BPND at higher doses. Figure 2a illustrates average [11C]-(+)-PHNO BPND maps after orally administered buspirone showing a progressive decreases of binding at higher doses. The voxel-wise t-statistical comparison is concordant with this in showing a large (k=249) cluster of significant difference in dorsal and ventral striatum expanding to the ventral pallidum (Figure 2b). Peak (T=23.83) occurred in the dorsal striatum in (18, 8, 10) and survived correction for multiple comparisons (P(FEW-corrected)=0.008).

Model Fit

The plots of %Occupancy versus plasma levels of buspirone and metabolites (Figure 3 and Supplementary Figure 5) suggest that maximal blocking of [11C]-(+)-PHNO binding was reached at lower plasma levels of buspirone and metabolite in ƒD3 and in D3-rich brain areas (SN, VP, GP, and LST) as compared with that in ƒD2 and D2-rich areas (AST, SMST, or whole dorsal striatum). Curve fitting with the one-site model (Equation 2, see Table 2 and Supplementary Table 1, R2=0.19–0.71) confirmed that ƒD3 and the individual D3-rich areas have lower ED50 values of buspirone (3.1–5.2 ng/ml, P=0.003), 5-OH (2.1–2.7 ng/ml, P=0.003), and 6-OH (42–55 ng/ml, P=0.005) with larger Hill constants (2–44) than ƒD2 and D2 areas (buspirone, 6.1–13 ng/ml; 5-OH, 3.1–5.5 ng/ml; 6-OH, 62–129 ng/ml; Hill, 1.1–4.4). This analysis also showed a maximal occupancy of 30–36% in ƒD3 and D3 areas versus 42–51% in ƒD2 and D2 areas, with the latter having larger uncertainty (see 95% CI in Table 2 and Supplementary Table 1) as the fitted curves were not plateaued in the concentration range, suggesting that higher doses of buspirone could have bigger occupancy in the D2 areas but not for D3. Indeed, compared with a simple one-site hyperbolic function with Hill constant fixed to 1, Equation 2 was the preferred model for data in ƒD3 (ΔAIC=−12) and D3 areas (ΔAIC= −9.9, −8.1, and −1.9 for SN, VP, and GP, respectively), whereas the simpler model was preferred for data in ƒD2 (ΔAIC=+9.1) and D2 areas (ΔAIC=+5.8, +6.8, and +5.7 for SMST, AST, and LST, respectively). With the simpler model, the estimated maximal occupancy for ƒD2 and D2 areas increased to 55–90% and ED50 increased to 10–26, 5–12, and 102–265 ng/ml for buspirone, 5-OH, and 6-OH, respectively, reflecting the limited buspirone concentration range for D2 areas. A two-site model could not converge for data in ƒD3 and D3 areas and did not improve the fitting for data in ƒD2 and D2 areas, namely, the second site was not identifiable (data not shown).

Figure 3
figure 3

Plots of [11C]-(+)-PHNO occupancy against plasma concentration of Buspirone and 5OH and 6OH buspirone in ƒD2 and ƒD3. Solid lines represent model fits (via Equation 2 above) to the measured data. ƒD3 : DRD3 fraction calculated in areas of moderate to high DRD3 signal using previously established regional fractions (Tziortzi et al, 2011). This was done by multiplying [11C]-(+)-PHNO BPND in each ROI by their estimated regional fraction (as per Girgis et al, 2011). ƒD2:DRD2 fraction corresponding to [11C]-(+)-PHNO BPND in dorsal striatum.

PowerPoint slide

Full size image
Table 2 ED50 for Buspirone and Metabolites Calculated Using a One-Site Binding Model
Full size table

DISCUSSION

Our data indicate for the first time that acute doses of buspirone can occupy DRD2 and DRD3 in human subjects. However, this occupancy was in the same range for the DRD2 vs DRD3 and the maximum occupancy achieved during acute single-dose regimen was modest with the doses of buspirone tested.

One possible explanation for our finding of a modest occupancy may be that the plasma concentrations of the parent drug and/or metabolites reached by the dose of buspirone were not high enough. The average dose per kg body weight of the subjects was 0.87±0.22 mg/kg for the 60 mg scan (the highest dose approved for clinical use) and 1.74±0.44 mg/kg for the 120 mg scan, yielding plasma concentrations of buspirone (>8 ng/ml after 120 mg and >4 ng/ml after 60 mg) slightly higher than what is reported in the literature during daily 60 mg exposure of buspirone (2.7 ng/ml) (Dockens et al, 2006). A recent[11C]-(+)-PHNO study by Kim et al (2014) in female baboons indicated that 3 mg/kg of buspirone p.o. (but not 1 mg/kg) led to a significant occupancy in DRD3-rich areas (up to 74 % in midbrain). We cannot directly compare these data with ours, because plasma levels of buspirone were not reported and we cannot say whether higher dosage of buspirone (for example, 3 mg/kg) would have increased occupancy. However, our data would argue against the possibility that higher occupancy of DRD3 sites or preferential DRD3 over DRD2 occupancy would be achieved by increasing the dose of buspirone in humans (see Figure 3). It is possible that the differences when compared with the study of Kim et al (2014) may be because of interspecies differences. Although a repeated dosing regimen study is required to rule out that chronic exposure would have an accumulative effect increasing the DRD3 occupancy, our results showing a lack of selectivity for DRD3 over DRD2, similar to antipsychotic drugs, might induce a decreased occupancy of DRD3 after chronic exposure (Graff-Guerrero et al, 2009; McCormick et al, 2010).

Another issue is whether buspirone’s major metabolites had differential effects on [11C]-(+)-PHNO binding. In this regard, Bergman et al (2013) have shown that some of buspirone metabolites have high affinities for DRD3 vs DRD2. (ki were: 98, 261, and 795 nM for buspirone, 5-OH buspirone and 6-OH buspirone, respectively). In line with this, Kim et al (2014) have reported in baboons that there is significantly higher occupancy of DRD2 following intramuscular administration and that 6’-hydroxybuspirone affected [11C]-(+)-PHNO in midbrain, suggesting a possible important role of metabolites in the occupancy of DRD3. Curve fitting of our data suggests faster occupancy of DRD3 vs DRD2 sites that may be translated in a higher DRD3 affinity and is consistent with in vitro findings. Based on plasma levels of metabolites achieved in the current study (in literature range: 37 ng/ml; Dockens et al, 2006) our conditions allowed for (some) metabolism of buspirone and accumulation of high dosages of metabolites (see Supplementary Figure 3).

Another issue that could have affected our measure is the possible elevation of DA induced by DRD2 antagonism. It could also be speculated that because of the higher affinity of endogenous DA for the DRD3 (vs DRD2), increases in DA would have interfered with buspirone in ƒD3 more so than in ƒD2. However, several studies reported no significant effects of buspirone on DA release using microdialysis in striatal areas in rats (Kaariainen et al, 2008; Liu et al, 2004). In addition, the elevation of DA induced by acute antipsychotics is in the range of 100% (Ichikawa and Meltzer, 1991), and therefore it is unlikely that potential elevation of DA induced by buspirone would lead to significant occupancy of receptors as assessed by PET (Martinez and Narendran, 2010).

Our findings that higher doses of buspirone were associated with self-reported drowsiness and dizziness are in line with clinical observation in patients receiving this drug and were within tolerable limits. Subjects who reported negative side effects from buspirone reported maximal effects during the ascending limb of the plasma buspirone curve (that is, at 60 min) and were back to normal at the end of the session (~200 min after buspirone). We found that plasma prolactin levels were increased after buspirone and correlated with DRD2 occupancy: this is in line with the known effects of DRD2 antagonism on prolactin release from the anterior pituitary gland and suggests that even low brain occupancy of DRD2 stimulates prolactin increases. Note that because of the low signal, we did not measure [11C]-(+)-PHNO binding in hypothalamus (arcuate nucleus) where DA DRD2 antagonism would have, via hypothalamic hormones, stimulated prolactin release.

It may be assumed that one reason for a lack of difference in occupancy of the DRD2 and DRD3 by buspirone may be that it is not feasible to image both the DRD2 and the DRD3 using a single PET tracer. However, converging evidence suggests that it is possible to differentially measure these receptors that are differentially expressed in the brain. A number of occupancy studies (using the DRD3 selective/preferential drugs ABT-925 and GSK598809) have suggested that the DRD2 and DRD3 fraction of the [11C]-(+)-PHNO signal can be measured simultaneously (Graff-Guerrero et al, 2009, 2010; Searle et al, 2010). Preclinical studies performed with [3H]-(+)-PHNO and use of DRD3 knockout mice and of DRD3-selective antagonist also clearly demonstrated that [11C]-(+)-PHNO is suitable to visualize DRD3 sites (Rabiner et al, 2009). In contrast to the high occupancy that was reported with a highly selective DRD3 ligand such as GSK598809, here we have obtained modest occupancy of DRD3 sites using buspirone. It is unlikely that this is because of the inability of [11C]-(+)-PHNO to measure occupancy at the DRD3.

Nevertheless, our findings should be interpreted in light of some limitations including small sample size as well as caveats of the [11C]-(+)-PHNO PET approach, namely scanning at nontracer doses, and specific binding in the reference tissue (for a detailed discussion of these issues, refer to Shotbolt et al, 2012). Importantly, it is not likely that specific binding in the cerebellum contributed to the finding as AUC cerebellum was not different between conditions and mass was not different between scans. Furthermore, the difference in measurement sensitivity between D3-rich areas vs D2 areas (that is, D3-rich areas are more difficult to delineate, prone to greater partial volume effect, and in some cases may be affected by white matter content (globus pallidus)) may have limited our ability of finding a region-dependant effect. This study has some implications for the mechanism of action of buspirone. Although initially developed to be an antipsychotic drug, the initial clinical trial with buspirone showed a lack of antipsychotic activity. Our results indicating low occupancy of DRD2 is in line with this, as it has been shown that higher degree of DRD2 occupancy is required for antipsychotic activity. It has been previously reported that buspirone has a low occupancy (<26%) of the 5HT1A in clinical doses (Rabiner et al, 2000). We cannot exclude that the therapeutic efficacy is therefore related to a combined moderate occupancy of the HT1A, DRD2, and DRD3. In addition, it has some affinity toward the DRD4 (Bergman et al, 2013) that can also contribute to its effects on animal models of substance use disorders (Di Ciano et al, 2014). It is not currently feasible to determine the occupancy of DRD4 in humans because of the lack of a proper PET tracer.

There has been considerable interest in using buspirone for drug addiction treatment. In animal models of drug dependence, it was recently reported by Mello et al (2013) and Bergman et al (2013) that buspirone can significantly decrease cocaine self-administration in non-human primates. It is notable that, in the former study, buspirone was administered chronically, and that it has been previously demonstrated that i.v. or i.m. injections of higher, but not lower, doses of buspirone decreased cocaine seeking (Gold and Balster, 1992). It is possible that higher doses of buspirone are needed to disrupt the primary reinforcing properties of drug, because acute and moderate i.p. doses of buspirone decreased drug seeking as measured in the reinstatement model (Shelton et al, 2013). As mentioned above, it is difficult to compare human doses with animal doses, but it is possible that the dose of buspirone needed to treat addictions is too high to be considered free from adverse events.

In human clinical studies, promising effects of buspirone have been reported for treatment of substance dependence, including tobacco (Cinciripini et al, 1995; Hilleman et al, 1992, 1994), marijuana (McRae et al, 2006), and opiates (McRae et al, 2004), but not alcohol (Malec et al, 1996), although it was effective in reducing alcohol withdrawal (Dougherty and Gates, 1990). However, recently the clinical results of a trial evaluating buspirone for cocaine dependence have been reported and no significant effects were observed (Winhusen et al, 2014). If anything, there was higher rate of relapse to cocaine use in the female participants in the trial. This possible worsening effect has been reported in different trials performed with dopamine antagonists while being tested for substance use disorder treatment and may reflect the impact of DADRD2 blockers (as SUD has been already associated with lower DRD2 function, Volkow et al, 1993).

CONCLUSION

Our study provides the first evidence of DRD2/3 occupancy by buspirone in humans. However, contrary to our expectation, its occupancy at the DRD3 was modest despite plasma levels (of drugs and metabolites) higher than that obtained at the usual therapeutic regimen. Our data do not suggest that increasing the dose would lead to higher occupancy of DRD3 without concurrent occupancy of DRD2; chronic studies would be needed to conclude with certainty on this. Buspirone may not be a good candidate to test specifically the role of DRD3 in drug addiction. It is not excluded that the therapeutic effects of buspirone could be mediated by a combination of its effects at HT1A, DRD2, and DRD3.

FUNDING AND DISCLOSURE

Dr Le Foll has received grant and salary support for other studies from Pfizer and is a consultant for Richter Pharmaceuticals, Lundbeck, Mylan, Ethypharm, and Pfizer. Dr Graff-Guerrero has received grant support for other studies from Janssen (Johnson & Johnson) and consultant fee for Richter Pharmaceuticals. Dr Nakajima has received manuscript fees from Dainippon Sumitomo Pharma and Kyowa Hakko Kirin. Dr Meyer has received operating grant funds for other studies from Eli-Lilly, GlaxoSmithKline, Bristol Myers Squibb, Lundbeck, Janssen, and SK Life Sciences. Dr Meyer has consulted to several of these companies as well as Takeda, Sepracor, Trius, Mylan, and Teva. Drs Payer, Di Ciano, Guranda, Tong, Mansouri, Wilson, Houle, and Boileau declare no conflict of interest.