Regional and source-based patterns of [¹¹C]-(+)-PHNO binding potential reveal concurrent alterations in dopamine D₂ and D₃ receptor availability in cocaine-use disorder

Abstract

Dopamine type 2 and type 3 receptors (D₂R/D₃R) appear critical to addictive disorders. Cocaine-use disorder (CUD) is associated with lower D₂R availability and greater D₃R availability in regions primarily expressing D₂R or D₃R concentrations, respectively. However, these CUD-related alterations in D₂R and D₃R have not been concurrently detected using available dopaminergic radioligands. Furthermore, receptor availability in regions of mixed D₂R/D₃R concentration in CUD remains unclear. The current study aimed to extend investigations of CUD-related alterations in D₂R and D₃R availability using regional and source-based analyses of [¹¹C]-(+)-PHNO positron emission tomography (PET) of 26 individuals with CUD and 26 matched healthy comparison (HC) participants. Regional analysis detected greater binding potential (BP_ND) in CUD in the midbrain, consistent with prior [¹¹C]-(+)-PHNO research, and lower BP_ND in CUD in the dorsal striatum, consistent with research using non-selective D₂R/D₃R radiotracers. Exploratory independent component analysis (ICA) identified three sources of BP_ND (striatopallidal, pallidonigral, and mesoaccumbens sources) that represent systems of brain regions displaying coherent variation in receptor availability. The striatopallidal source was associated with estimates of regional D₂R-related proportions of BP_ND (calculated using independent reports of [¹¹C]-(+)-PHNO receptor binding fractions), was lower in intensity in CUD and negatively associated with years of cocaine use. By comparison, the pallidonigral source was associated with estimates of regional D₃R distribution, was greater in intensity in CUD and positively associated with years of cocaine use. The current study extends previous D₂R/D₃R research in CUD, demonstrating both lower BP_ND in the D₂R-rich dorsal striatum and greater BP_ND in the D₃R-rich midbrain using a single radiotracer. In addition, exploratory ICA identified sources of [¹¹C]-(+)-PHNO BP_ND that were correlated with regional estimates of D₂R-related and D₃R-related proportions of BP_ND, were consistent with regional differences in CUD, and suggest receptor alterations in CUD may also be present in regions of mixed D₂R/D₃R concentration.

1. Introduction

Individuals with cocaine-use disorder (CUD) exhibit alterations in subcortical dopamine including alterations in receptor availability, blunted dopamine release and reduced dopamine transport mechanisms (1–9). However, dopaminergic pharmacotherapies for stimulant-related addictions have displayed limited efficacy to date (see (10, 11)), thus a greater understanding of dopamine-related alterations in CUD may improve targeted interventions for addictions. Alterations in dopamine type-2 and type-3 receptors (D₂R/D₃R) are associated with CUD and other substance-use disorders (1–5, 12–17), suggesting related neurobiological mechanisms that may hold potential as targets in treating addictions. Specifically, D₂Rs and D₃Rs are implicated in the habitual and compulsive behaviors associated with the persistence of chronic drug use and relapse (18). Animal models demonstrate functional dissociation of D₂R and D₃R systems (19–21), suggesting these dopamine receptor subtypes may provide distinct contributions to addictive processes.

Research using dopamine-antagonist radioligands with equal affinities for D₂R and D₃Rs (e.g., [¹¹C]raclopride) indicate lower receptor availability in D₂R-rich striatal regions of individuals with CUD (1–5). By comparison, research using the D₃R-preferring dopamine-agonist radioligand [¹¹C]-(+)-PHNO in CUD and stimulant addiction indicate increased receptor availability in the D₃R-rich midbrain (8, 9, 22). These concurrently upregulated and downregulated dopamine systems in CUD may undermine dopaminergic therapies for cocaine addiction. Adding to this complexity, D₂R and D₃R are both expressed in structures central to addiction (e.g., midbrain, ventral striatum, pallidum) (23), and assessing potential alterations of specific receptor subtypes in these regions is challenging.

Regional binding of [¹¹C]-(+)-PHNO reflects local concentrations of both D₂Rs and D₃Rs (24, 25), allowing for simultaneous imaging of D₂R/D₃R not possible using other D₂-like receptor radioligands. To date however, [¹¹C]-(+)-PHNO research of CUD has not replicated findings of lower receptor availability in the D₂R-rich dorsal striatum, or detected alterations in mixed-D₂R/D₃R regions (8, 9). However, these results of non-altered D₂R-like availability in the striatum in CUD are consistent with research using alternative dopamine-agonist radioligands with equal affinities for D₂R/D₃R (e.g., [¹¹C]NPA)(26). The current study aimed to further examine D₂R and D₃R alterations in the midbrain, striatum and other subcortical structures (e.g., globus pallidus, ventral pallidum and hypothalamus) in CUD using [¹¹C]-(+)-PHNO positron emission tomography (PET) imaging. Investigation of regional [¹¹C]-(+)-PHNO BP_ND was conducted on individuals with CUD and matched healthy comparison (HC) participants.

In addition, exploratory spatial independent component analysis (ICA) was performed on [¹¹C]-(+)-PHNO BP_ND data to examine D₂R and D₃R using a source-based approach. ICA is a data-driven computational procedure that decomposes or ‘un-mixes’ a measured signal into its maximal spatially independent ‘sources’. While [¹¹C]-(+)-PHNO exhibits an estimated 20- to 50-fold selectivity in vivo for D₃Rs relative to D₂Rs (27–29), local BP_ND reflects a weighted sum of D₂R/D₃R concentrations. That is, [¹¹C]-(+)-PHNO BP_ND primarily reflects D₃R availability in D₃-rich regions (e.g., substantia nigra), D₂R availability in D₂-rich regions (e.g., dorsal striatum), and an aggregate of D₂R and D₃R availability in mixed D₂R/D₃R regions (e.g. ventral striatum) (24, 25, 30, 31). This unique mixed-signal nature of [¹¹C]-(+)-PHNO strongly motivates the utility of source-based analyses like ICA to un-mix the D₂R and D₃R components of BP_ND and examine distinct sources of receptor availability.

We hypothesized that CUD-related reductions in D₂R availability in the dorsal striatum would be detectable using [¹¹C]-(+)-PHNO in a large sample, and that CUD-related increases in D₃R availability in the midbrain would be confirmed. We also hypothesized that the exploratory ICA would identify sources of [¹¹C]-(+)-PHNO BP_ND consistent with estimates of regional proportions of D₂R-related and D₃R-related BP_ND, with mixed-D₂R/D₃R regions incorporated into multiple sources. Consistent with research demonstrating regional differences in D₂R and D₃R in CUD, we hypothesized that individuals with CUD would exhibit increased intensity of sources that include D₃-rich regions and reduced intensity of sources that include D₂-rich regions. Finally, we examined relationships between regional BP_ND and source intensities with duration of cocaine use.

2. Methods and Materials

2.1. Participants

Participants were 26 non-treatment seeking individuals with CUD and 26 age- and gender-matched HC participants, most of whom (22 CUD and 21 HC) were included in previous reports (8, 32) (Table 1). Physical exams with medical history, routine laboratory studies, pregnancy tests and electrocardiograms (ECGs) were performed to assess medical health eligibility criteria. Urine toxicology screening for a range of drugs (including cocaine, amphetamines, marijuana, opiates, benzodiazepines, barbiturates) were performed to confirm cocaine-use status in CUD participants and the absence of other recent drug use in both CUD and HC participants. Participants were assessed for DSM-IV diagnoses using clinical interviews (SCID (33); MINI (34)). Participants meeting criteria for cocaine dependence (i.e., presenting with 3 or more criteria) were included in the CUD sample. Exclusion criteria included the presence or history of a general medical illness or psychotic disorder, pregnancy or breast feeding, or any condition that would interfere with the ability to participate in PET or MRI protocols (e.g., claustrophobia, metallic implants). All study procedures were approved by the Yale Human Investigation, Yale University Radiation Safety, Yale-New Haven Hospital Radioactive Drug Research, and Yale MRI Safety Committees, and participants provided written informed consent.

Table 1.

Sample characteristics and injection data.

Variable	HC (N=26)	CUD (N=26)	p-value
Age, years (SD)	41.3 (6.9)	42.7 (6.4)	0.46
Gender, F (%)	5 (19.2)	5 (19.2)	1.00
Body-mass, BMI (SD)	28.7 (6.0)	29.9 (6.5)	0.47
Daily tobacco user, N (%)	1 (3.8)	22 (84.6)	<0.001
Daily tobacco use, cigarettes (SD)	0.4 (2.0)	9.2 (6.4)	<0.001
Weekly alcohol use, drinks (SD)	1.4 (2.2)	6.8 (14.6)	0.75
Amount of cocaine per use, $ (SD)	-	112.7 (148.5)	-
Frequency of use, days/month (SD)	-	19.0 (7.7)	-
Last cocaine use before PET, days (SD)	-	7.8 (4.9)	-
CUD chronicity, years (SD)	-	19.3 (6.3)	-
Radioactive dose, MBq (SD)	398 (155)	370 (146)	0.51
Injected mass, µg/kg (SD)	0.024 (0.007)	0.025 (0.006)	0.63
Specific activity, MBq/nmol (SD)	56.7 (30.3)	46.9 (19.5)	0.17

2.2. Radiochemistry and image acquisition

[¹¹C]-(+)-PHNO was prepared as previously described (35, 36). Radioactive dose, specific activity and injected mass for CUD and HC groups are listed in Table 1. PET imaging was performed on a Siemens high-resolution research tomograph (HRRT; Siemens/CTI, Knoxville, TN, USA). A transmission scan was obtained before the emission scan. PET scans (slices=207, slice separation=1.2mm, reconstructed image resolution ~3 mm) were acquired for 120 min at rest. Head motion was measured using an optical detector (Vicra, NDI Systems, Waterloo, Ontario, Canada). MR images were collected on a Siemens 3T Trio system (Siemens Medical Solutions, Malvern, PA) using a magnetization prepared rapid acquisition gradient echo (MPRAGE) sequence (repeat time (TR)/echo time (TE)=2530/3.34, flip angle=7°, in-plane resolution=0.98×0.98 mm, matrix=256×256, slice thickness=1mm, slices=176).

2.3. Image processing

Dynamic PET data (frame timing: 6×30s; 3×60s; 2×120s; 22×300s) were reconstructed with corrections (attenuation, normalization, scatter, randoms, deadtime and motion) using the MOLAR algorithm (37). Parametric images of [¹¹C]-(+)-PHNO BP_ND were computed using a simplified reference tissue model (SRTM2 (38)) with the cerebellum as reference (39). Registration of individual images into MNI152 space (40) was performed using SPM12 (Wellcome Trust Centre for Neuroimaging, London, UK). A summed [¹¹C]-(+)-PHNO uptake (0–10 min post-injection) image was created from the motion-corrected PET data and rigid-body registered to the subject's MR image. Nonlinear registration of the MR image to MNI152 space was determined using optimized unified segmentation (41). The combined rigid-body and nonlinear registrations were applied to the parametric BP_ND, and smoothed with a 4mm FWHM Gaussian kernel.

2.4. Analyses of regional [¹¹C]-(+)-PHNO BP_ND

Seven regions of interest (ROIs) were examined: dorsal caudate (DCA), dorsal putamen (DPU), globus pallidus (GP), hypothalamus (HY), substantia nigra (SN), ventral pallidum (VP), and ventral striatum (VS). ROIs were defined using previously described guidelines (24, 42). DCA, DPU and VS ROIs were obtained from FSL (Oxford Centre for Functional MRI of the Brain, Oxford, UK (24)). GP, HY, and VP ROIs were manually drawn on the ICBM-MNI52 template (Montreal Neurological Institute, Montreal, Canada). The SN ROI was manually defined on a group-average parametric [¹¹C]-(+)-PHNO BP_ND image from an independent sample (35). Regional [¹¹C]-(+)-PHNO BP_ND was determined from the individual smoothed parametric BP_ND images in standard space. ROI volumes were calculated by inverse transformation of template regions into individual native space. A confirmatory whole-brain, between-group analysis was performed on BP_ND images using a two-sample t-test in SPM12 with a relative threshold mask (0.8; displayed in Figure 1). Due to the relatively low BP_ND signal (less than 0.5) in the thalamic and amygdala regions included in previous reports, substantial portions of these structures did not survive threshold masking and thus were excluded from ROI analyses. Group differences were explored at an uncorrected significance level of p<0.005, and cluster extent (k) >20 contiguous voxels.

Figure 1.

Regional analyses of [¹¹C]-(+)-PHNO BP_ND. (a) Region of interest (ROI) analysis performed on smoothed parametric images. CUD relative to HC participants displayed reduced BP_ND in the dorsal putamen (DPU; *p<0.05) and greater BP_ND in the substantia nigra (SN, **p<0.01). Error bars indicate SD. (b) Whole-brain mean BP_ND across all participants with relative threshold mask borders used in GLM and ICA analyses indicated by the white outline (left) and GLM-identified regions of group differences in [¹¹C]-(+)-PHNO BP_ND (right). CUD relative to HC exhibited greater BP_ND in midbrain regions (red/yellow) and reduced BP_ND in the putamen (blue/green). Axial slices displayed at p<0.005, k>20 on MNI152. Abbreviations: DCA, dorsal caudate; DPU, dorsal putamen; GP, globus pallidus; HY, hypothalamus; SN, substantia nigra; VP, ventral pallidum; VS, ventral striatum.

2.5. Independent component analysis (ICA)

Exploratory ICA was performed on parametric images of [¹¹C]-(+)-PHNO BP_ND using the source-based morphometry (SBM (43)) module of the Group ICA of fMRI Toolbox (GIFT v2.0e; http://mialab.mrn.org/software/gift). Using higher order statistics, ICA identifies maximally independent vectors that comprise an unknown linear mixture (A) of mostly non-Gaussian sources (s) that generate a random variable (x). Given x = As, ICA solves for y =Wx, or the estimated un-mixing matrix (W) of approximated source maps (y). Parametric images of all 52 participants were entered into the ICA to allow direct comparison of source loadings. The analysis was constrained to data within the relative threshold mask determined in the GLM analysis (described above), and with no prior information as to group membership. A modified minimum-description-length (MDL) criterion (44) estimated the dimensionality of the dataset to consist of three sources which were extracted from the aggregate dataset using a neural-network algorithm designed to minimize mutual information of source outputs (InfoMax (45)).

Output from the ICA included the spatial source maps (y_M;M =1,2,3) and subject ICA-loading values (Ã; 1_M,i;i = 1, …, 52), where à denotes the estimate of the unknown mixing matrix A (i.e. the inverse of ICA-derived un-mixing matrix W). Note that applications of ICA to biomedical images typically employ principal component analysis (PCA) for data reduction prior to ICA; thus, subject ICA-loadings also incorporate the PCA-reduction factor (R) and are more accurately characterized as (R × Ã), but for convenience, we use à to represent the final subject ICA-loading parameters. The subject ICA-loading values represent the source ‘intensity’ or relative strength to which each subject’s BP_ND data contained a given source. The spatial source maps were ‘intensity-scaled’ such that (y_M × Ã) reflects the relative contribution of each source to the aggregate BP_ND signal, where Ã_M,iis the average subject intensity for source M, and:

$$\tilde{x}=(y_1\times {Ã}_{1,i})+(y_2\times {Ã}_{2,i})+(y_3\times {Ã}_{3,i})+\overline{x}$$

where x̃ is the approximate reconstruction of the original data x, and x̄ is the global mean BP_ND value (1.56) removed prior to ICA extraction. For analysis purposes, the source maps were scaled to ICA-estimated BP_ND (B̃P_M) through the product of (y_M × Ã_M,i). For display and anatomical assignment purposes, the scaled source maps were thresholded at B̃P_M≥0.5 (with no cluster-extent threshold) based on the relevant range of [¹¹C]-(+)-PHNO BP_ND of 1.5–4.

To examine source compositions relative to regional estimates of D₂R/D₃R availability, intensity-scaled source maps were generated for all subjects, and regional B̃P_M values were extracted from the seven ROIs and compared the D₂R-related and D₃R-related proportions of BP_ND calculated using reported regional fractions of [¹¹C]-(+)-PHNO BP_ND (24, 25). That is, regional BP_ND values obtained in the ROI-based analysis (see Section 2.4. and Figure 1) were multiplied by average D₂R-and D₃R-related fractions reported in (23, 24) (Supplementary Table 1) to obtain regional estimates of local D₂R/D₃R BP_ND (i.e., $B{P}_{ND}^{D2}=B{P}_{ND}\times f_{D2};B{P}_{ND}^{D3}=B{P}_{ND}\times f_{D3}$). Average regional B̃P_M for each source were then examined in relation to the calculated estimates of $B{P}_{ND}^{D2}$ and $B{P}_{ND}^{D3}$ using Pearson correlations to assess potential association with D₂R-related and D₃R-related BP_ND proportions.

2.6. Statistical procedures

Group differences in ROI BP_ND values and ICA source intensities were assessed with two separate multivariate analyses of variance (i.e., group × ROI and group × source) and post-hoc univariate analyses. Due to non-tracer conditions of [¹¹C]-(+)-PHNO for D₃Rs at the administered dose (25), multivariate analyses including the estimated free concentration of [¹¹C]-(+)-PHNO (based on the average mass concentration in the cerebellum from 90–120 mins, multiplied by the specific activity) as a covariate were performed and revealed no significant mass dose effect on results (p’s>0.4). Given reported associations between [¹¹C]-(+)-PHNO BP_ND and age and body-mass (46–48), all between-group multivariate and univariate analyses were repeated adjusting for age and body-mass. Additional analyses were performed to examine potential influences of alcohol use across all participants (12, 13) and cigarette smoking (14) within CUD participants. Exploratory correlational analyses were performed to examine relationships between regional and source-based binding measures with years of cocaine use.

3. Results

Participant characteristics and radioactivity information are provided in Table 1. There were no group differences in body-mass or radiotracer injection parameters. CUD participants were more likely to be daily tobacco smokers (χ²=32.4, p<0.001) and did not differ from HCs in weekly alcohol consumption. ROI volumes (in subject/native space) did not differ between groups (p’s>0.1).

3.1. Regional analyses of [¹¹C]-(+)-PHNO BP_ND

Results of ROI and whole-brain analyses are displayed in Figure 1. There was a main effect of group across ROIs (F_7,44=3.87, p=0.002), with CUD relative to HC displaying greater BP_ND in the SN (t₅₀=2.91, p=0.005) and lower BP_ND in the DPU (t₅₀=2.14, p=0.037), while a group difference in the DCA did not reach significance (t₅₀=1.79, p=0.080) (Figure 1a). Whole-brain analysis confirmed CUD relative to HC participants exhibited greater BP_ND in the midbrain and lower BP_ND in the dorsal striatum (Figure 1b). Consistent with previous reports (46, 48, 49), there was a significant association between age and BP_ND (F_7,42=3.25, p=0.008), and a marginal association between body-mass and BP_ND (F_7,42=2.23, p=0.051). Adjusting for influences of age and body-mass, group differences in the SN remained significant (F_1,48=11.29, p=0.002) while differences in the DPU did not reach significance (F_1,48=3.72, p=0.060). There was no main effect of alcohol use on BP_ND (F_7,44=0.87, p=0.54), and group differences survived adjusting for the number of cigarettes smoked daily (F_7,43=3.89, p=0.002). Within CUD participants, there was no association between BP_ND and daily cigarettes smoked (F_7,18=0.55, p=0.79) or years of cocaine use (F_7,18=1.07, p=0.42).

3.2. ICA of [¹¹C]-(+)-PHNO BP_ND

The exploratory ICA identified three sources of [¹¹C]-(+)-PHNO binding that represent systems of brain regions displaying coherent variation in BP_ND through the sample (Figure 2; Table 2). Source 1 represented a ‘striatopallidal source’, with largest contributions in the DPU, DCA, GP, VP and VS. Source 2 represented a ‘pallidonigral source’ comprised of the SN, GP, VP and HY. Source 3 represented a ventral ‘mesoaccumbens source’ encompassing HY, VP, VS and anterior regions of the midbrain extending into the SN. Although extracted solely from the data, the identified sources were spatially consistent with the distribution of D₂R-related and D₃R-related availability across regions of interest (Figure 3). Specifically, regional B̃P_M values of the striatopallidal source were associated with estimated D₂R-related BP_ND proportions (r=0.89, p=0.008), and regional B̃P_M of the pallidonigral source were associated with estimated D₃R-related BP_ND proportions (r=0.89, p=0.007). The mesoaccumbens source was not associated with either D₂R-related or D₃R-related binding proportions (p’s>0.1).

Figure 2.

ICA-identified sources of [¹¹C]-(+)-PHNO BP_ND displayed at B̃P_M ≥0.5 on different intensity scales as denoted in the color bars.

Table 2.

Regional composition of sources of [¹¹C]-(+)-PHNO BP_ND.

Component/region	k	x	y	z	peak B̃PMM
a. Striatopallidal
L DPU, DCA, GP, VP, VS	1467	−10	8	−8	3.20
R DPU, DCA, GP, VP, VS	1416	14	6	−4	3.06
b. Pallidonigral
R GP, HY, SN, VP	457	12	2	−10	1.60
L GP, HY, SN, VP	579	−14	−2	−8	1.97
c. Mesoaccumbens
R SN	5	10	−14	−16	0.59
L SN	6	−6	−12	−16	0.62
R HY, VP, VS	243	10	10	−12	1.23
L HY, VP, VS	259	−8	8	−14	1.39

Regional composition of ICA-identified sources at the threshold of ICA-estimated BP_ND contribution (B̃P_M) ≥0.5. Cluster information includes spatial extent (k, in voxels), peak location (x, y, z) in MNI152 space and peak B̃P_M.

Abbreviations: DCA, dorsal caudate; DPU, dorsal putamen; GP, globus pallidus; HY, hypothalamus; SN, substantia nigra; VP, ventral pallidum; VS, ventral striatum; R/L, right/left hemisphere.

Figure 3.

Source associations with estimated regional D₂R-related and D₃R-related BP_ND proportions, group intensity differences and correlation with years of cocaine use for the (a) striatopallidal and (b) pallidonigral sources. Regional estimates of D₂R-related and D₃R-related BP_ND were calculated using reported fractions of [¹¹C]-(+)-PHNO BP_ND (24, 25) (see Section 2.5). Subject ICA-loadings refer to the mixing parameter representing source intensity, or how strongly each source contributed to the aggregate BP_ND signal. Error bars indicate SD. *p<0.05.

ICA subject intensities for the three sources did not correlate with each other (p’s>0.2). There were no associations between source intensities and age (F_3,46=1.53, p=0.22) or body-mass F_3,46=1.80, p=0.16). There was no main effect of alcohol use on source intensities (F_3,46=1.48, p=0.23).

3.3. [¹¹C]-(+)-PHNO BP_ND sources and CUD

There was a main effect of group on ICA-intensity across the three identified sources (F_3,48=4.28, p=0.009). CUD relative to HC participants displayed lower intensities of the striatopallidal source (t₅₀=2.58, p=0.013; Figure 3a) and greater intensity of the pallidonigral source (t₅₀=2.03, p=0.047; Figure 3b). Group differences in the mesoaccumbens source did not reach significance (t₅₀=1.63, p=0.109). Different from the regional analysis above, the intensity of the striatopallidal source was negatively associated with years of cocaine use (r=0.39, p=0.048; Figure 3a) and pallidonigral source intensity was positively correlated with years of cocaine use (r=0.40, p=0.042; Figure 3b). Group differences survived adjusting for the number of cigarettes smoked daily (F_3,47=4.28, p=0.009), and within CUD participants there was no association between source intensities and the number of cigarettes smoked daily (F_3,22=0.75, p=0.53).

4. Discussion

The current study extends prior research of D₂R/D₃R availability in CUD using [¹¹C]-(+)-PHNO. Regional analyses revealed CUD-related dorsal striatal reductions in D₂R-related BP_ND, consistent with D₂R/D₃Rantagonist radioligand research (1–5) but not previously detected with the D₃R-preferring [¹¹C]-(+)-PHNO, and confirmed greater availability in the SN in CUD (8, 9). Exploratory ICA identified three sources of [¹¹C]-(+)-PHNO BP_ND that were correlated with regional proportions of D₂R-related and D₃R-related BP_ND and spatially consistent with known subcortical dopamine circuitry. Individuals with CUD relative to HC exhibited reduced intensity of a D₂R-related striatopallidal source and greater intensity of a D₃R-related pallidonigral source that were both associated with years of cocaine use, an association not found when analyzing regions individually. Notably, the ICA-identified sources suggest CUD-related alterations may be present in regions of mixed D₂R/D₃R concentration (e.g., ventral striatum, globus pallidus) not previously observed using traditional analyses of D₂R/D₃R radioligands. The current findings underscore the advantage of [¹¹C]-(+)-PHNO in simultaneous investigation of both D₂R and D₃R alterations in addictions, especially when combined with ICA in exploring sources of dopamine receptor availability. The patterns of concurrently increased and decreased receptor availability, particularly within structures (e.g., the ventral striatum), lend insight into the clinical challenges of dopaminergic pharmacotherapy for CUD.

4.1. Regional [¹¹C]-(+)-PHNO BP_ND in CUD

Alterations in D₂-like receptor systems in CUD have been reported using multiple radioligands, and the current ROI and whole-brain observations of greater midbrain and reduced dorsal striatal BP_ND are consistent with previous research (1–5, 8, 9). Reduced D₂R-related BP_ND in the striatum and greater D₃R-related BP_ND in the SN and have been linked to increased impulsivity in cocaine/stimulant addictions (9, 50). Lower striatal D₂R BP_ND has been associated with poor behavioral therapy outcomes in CUD (51). The relationship between reduced striatal BP_ND and years of cocaine use in the current study is consistent with preclinical evidence of the effects of cocaine exposure on D₂R availability (52). Similarly, increases in D₃R systems follow cocaine exposure in animal models (53, 54), though the previous finding of a relationship between D₃R-related SN BP_ND and years of cocaine use (8) was not replicated in the current regional analysis. Dissociating potential dopaminergic vulnerabilities to addictive disorders and dopaminergic consequences of substance use remains an important area of addictions research (55).

4.2. Sources of [¹¹C]-(+)-PHNO BP_ND

ICA identified three sources of [¹¹C]-(+)-PHNO BP_ND that were correlated with estimated regional proportions of D₂R-related and D₃R-related BP_ND and spatially consistent with dopamine circuitry. The striatopallidal source was identified in nodes of the ‘indirect’ dopamine pathway linking the striatum (DPU, DCA and VS) to the SN through connections with the pallidum (GP and VP). The indirect pathway is comprised of D₂R-expressing GABAergic medium spiny neurons interconnecting the striatum and pallidum. Consistent with this indirect pathway anatomy, the intensity of the identified striatopallidal source was strongly associated with estimated regional D₂R components of [¹¹C]-(+)-PHNO BP_ND (24, 25). Preclinical models indicate striatopallidal D₂R mechanisms may counterbalance reward-related responsivity, with impaired functioning associated with a greater sensitivity to drug-related rewards (56). Evidence also suggests a role of indirect-pathway striatal D₂R with behavioral inflexibility relative to aversive conditions (57, 58) and compulsive cocaine-seeking (59).

The pallidonigral source identified by ICA was composed of regions highly correlated with estimates of regional proportions of D₃R [¹¹C]-(+)-PHNO BP_ND (24, 25). The pallidonigral source is spatially consistent with dopamine circuitry between the pallidal structures and the midbrain. While less is understood regarding specific functionality of D₃R systems, integration of GP, VP and SN regions suggest a role of this system in drug-related reinforcement processing and motivational/appetitive states (60). Incentive salience models of addiction identify ‘wanting’ as a motivation state, dissociable from the hedonic state of ‘liking’, as a process that becomes sensitized over the course of an addiction (61), and is associated with greater D₃R availability in the SN in polysubstance and methamphetamine users (17, 22).

The mesoaccumbens source was identified in SN, VS and VP regions. While the SN and VP are D₃R-rich regions, there was no relationship between regional source intensities and estimated D₃R-related (or D3R-related) proportions of [¹¹C]-(+)-PHNO BP_ND across ROIs. Decreased D₂R/D₃R availability in the SN and VS is associated with increased sensitivity to reinforcing effects of stimulants mediated through increased striatal dopamine release (62). Neuroadaptive increases in SN and VS D₃R associated with cocaine exposure (54, 63) may act to downregulate the stimulant-induced release of dopamine in the striatum in individuals with CUD (4, 5). While a functional role of D₂R/D₃R in VS and midbrain circuitry appears critical in balancing dopamine transmission in CUD, mesoaccumbens source intensities were not altered in CUD (or associated with years of cocaine use), and additional research and/or validation of the reliability of this source is warranted.

4.3. Clinical implications of D₂-like receptor sources in addictive disorders

CUD-related differences in source intensities were consistent with regional analyses, broadly identifying lower D₂R-related and greater D₃R-related source intensities. ICA-identified sources extend individual regional observations, detecting CUD-related differences in regions characterized by mixed-D₂R/D₃R concentrations. Patterns of source-intensity difference suggest concurrent upregulation and downregulation of receptor availability in the VS and pallidal regions. In the current study, alterations in striatopallidal and pallidonigral sources, proposed to reflect complementary mechanisms of the indirect dopamine pathway, were associated with years of cocaine use. The intensities of these two sources were significantly related to CUD chronicity, suggesting receptor alterations may represent reciprocal compensatory adaptations through the course of addiction.

Despite extensive efforts to identify effective pharmacotherapies for CUD, there is currently no FDA-approved medication with an indication for CUD. Although it has been proposed that dopamine-agonist therapies may aid in the maintenance of cocaine abstinence (64), preclinical evidence of the influences of D₂R and D₃R agonists and antagonists on cocaine-related behaviors is mixed (65–68). The current findings illustrate the underlying complexity of targeting receptor subtypes that may integrate into multiple functional systems and influence drug-related behavior differently depending on the region of action (69). Adding to the challenge of dopaminergic pharmacotherapies are the neuroplastic adaptations associated with cocaine exposure and chronic use, suggesting dopaminergic interventions may be aimed at ‘moving targets’, resulting in heterogeneous response profiles across individuals at different stages of addiction.

4.4. Strengths and limitations

The current study investigated the largest sample of CUD participants completing [¹¹C]-(+)-PHNO PET to date (N=26 relative to previous samples that included 10 to 16 individuals (8, 9, 32)), and demonstrates [¹¹C]-(+)-PHNO sensitivity to reductions in dorsal striatal D₂R availability frequently reported in CUD using equal-affinity D₂R/D₃R antagonist radioligands. The current sample was largely comprised of previously reported subjects. ROI results of CUD-related BP_ND alterations in the DPU survived post-hoc analysis co-varying for cohort (p=0.03), suggesting greater statistical power (rather than potential differences in individual variability across independent cohorts) contributed to the significant detection of D₂R-related differences in individuals with CUD. This study also represents the first application of ICA to explore sources of covarying D₂R/D₃R availability and examine alterations associated with CUD.

The data-driven ICA extraction was blind to the spatial distributions of D₂R/D₃R concentrations and blind to subject group assignments. The current mixed-sample design, while allowing direct comparison of source intensities in CUD relative to HC, identified three sources of [¹¹C]-(+)-PHNO BP_ND and alternative strategies of component extraction (e.g., higher-dimension ICA) may identify alternative sources (70). Alternative computational strategies (e.g., principal component analysis; PCA) have also been used to investigate PET data using ROI-based and voxel-wise approaches (71, 72), and may identify different factor structures in [¹¹C]-(+)-PHNO data.

Proposed associations with D₂R-related and D₃R-related availability in identified sources were based on calculated estimates using fractions of D₃R-related [¹¹C]-(+)-PHNO BP_ND derived in healthy controls using antagonists that are not 100% selective for D₃R receptors (24, 25). Thus, imprecise estimates, and potentially altered regional D₂R/D₃R-related fractions in individuals with CUD, limit conclusive determination of source signals relative to receptor subtypes. Research using ICA in conjunction with specific D₂R and D₃R blocking agents would be needed to validate these exploratory source compositions. Similarly, interpretations of potential functional involvement of identified sources was based on spatial patterns of regional integration relative to the extant and evolving literature, and direct examination of source relationships with measures of impulsivity, compulsivity, and substance use-related domains (e.g., self-administration, craving and cue-reactivity) may reveal alternative associations. Potential influences of co-occurring nicotine dependence in CUD participants on D₂R/D₃R availability warrants consideration given group differences in regular tobacco smoking (14). Though daily smoking amounts were not associated with PET results within CUD participants, future studies including a nicotine-dependent comparison sample would allow examination of cocaine-specific alterations in D₂R/D₃R receptor systems. Additional research is needed to replicate current observations and explore these dopamine systems in other addictive disorders.

4.5. Conclusions

The current study extends previous research of regional D₂R/D₃R availability in CUD measured with [¹¹C]-(+)-PHNO, detecting lower BP_ND the dorsal striatum and replicating greater BP_ND in the midbrain. Source-based analysis using ICA identified three distinct sources of BP_ND, indicating lower intensity of a striatopallidal source and greater intensity of a pallidonigral source in CUD. These source-based patterns of BP_ND suggest cocaine-related alterations in D₂R and D₃R may not be limited to the dorsal striatum and midbrain respectively, but may extend into the pallidum and ventral striatum. Furthermore, the alterations in striatopallidal and pallidonigral sources were associated with duration of cocaine use, and may indicate reciprocal and compensatory mechanisms of dopaminergic function in addiction. Additional investigation into concurrent alterations in D₂-like dopamine receptor systems (e.g., prior to drug exposure, during initial use and withdrawal, and following extended abstinence) should provide important insight into treatment development efforts for CUD.

Highlights.

• [¹¹C]-(+)-PHNO binding potential reflects the relative local availability of both D₂ and D₃ receptors

• Regional analysis indicated alterations in D₂R-rich and D₃R-rich regions in cocaine-use disorder

• ICA identified source-based patterns of receptor availability consistent with DA circuitry

• Source intensities suggest CUD-related differences may also be present in D₂R/D₃R-mixed regions