Dopamine D2/3 receptor availability in alcohol use disorder and individuals at high risk: Towards a dimensional approach

Alcohol use disorder (AUD) is the most common substance use disorder worldwide. Although dopamine‐related findings were often observed in AUD, associated neurobiological mechanisms are still poorly understood. Therefore, in the present study, we investigate D2/3 receptor availability in healthy participants, participants at high risk (HR) to develop addiction (not diagnosed with AUD), and AUD patients in a detoxified stage, applying 18F‐fallypride positron emission tomography (18F‐PET). Specifically, D2/3 receptor availability was investigated in (1) 19 low‐risk (LR) controls, (2) 19 HR participants, and (3) 20 AUD patients after alcohol detoxification. Quality and severity of addiction were assessed with clinical questionnaires and (neuro)psychological tests. PET data were corrected for age of participants and smoking status. In the dorsal striatum, we observed significant reductions of D2/3 receptor availability in AUD patients compared with LR participants. Further, receptor availability in HR participants was observed to be intermediate between LR and AUD groups (linearly decreasing). Still, in direct comparison, no group difference was observed between LR and HR groups or between HR and AUD groups. Further, the score of the Alcohol Dependence Scale (ADS) was inversely correlated with D2/3 receptor availability in the combined sample. Thus, in line with a dimensional approach, striatal D2/3 receptor availability showed a linear decrease from LR participants to HR participants to AUD patients, which was paralleled by clinical measures. Our study shows that a core neurobiological feature in AUD seems to be detectable in an early, subclinical state, allowing more individualized alcohol prevention programs in the future.


| INTRODUCTION
Traditionally, mental diseases are diagnosed and treated based on discrete categories of symptom collections or patterns, observed to be present over a certain amount of time. 1 However, a categorical view of mental disease has recently been brought into question, in favor of a more dimensional and continuous approach. [2][3][4][5] This approach goes in line with an etiological model of mental disease, the diathesisstress model. 6 Further support for a dimensional view comes from psychological and neurobiological studies, showing that severity of symptoms as well as temporal aspects of mental diseases are associated with rather continuous neurobiological changes in the brain. 5 In addition, during remission of symptoms following psychotherapeutic or somatic treatment, neurobiological alterations occur, which predict future remission. [7][8][9][10] Even though a large body of research on neurobiological changes in association with mental disease (e.g., depression and addiction) has been conducted, it is in general still poorly understood how neurobiological changes interact with psychological changes in the course of the development of mental diseases.
One of the major health related issues in modern society is addiction, especially related to alcohol. According to the World Health Organization (WHO), 3.3 million people worldwide are dying because of excessive alcohol consumption and its consequences every year. 11 With respect to the dimensional approach, boundaries of symptoms and consequences of moderate alcohol consumption, risky alcohol use, and alcohol dependence are fluent rather than categorical. 12 Although alcohol consumption broadly affects structural, functional, and neurochemical characteristics in the brain, specifically the dopamine (DA) system plays a key role in the development and maintenance of addiction. 13, 14 For instance, acute alcohol consumption causes increased DA release in the striatum. 15 This increased DA release may further contribute to the rewarding effects of alcohol and seems to play an important role especially in early stages of alcohol use disorder (AUD). 13, [15][16][17] Additionally, in later stages of addiction, increased DA levels in the striatum have been found to be linked to cue-induced craving and further may predict relapse to a certain extent. [18][19][20] Chronic alcohol consumption may lead to further longlasting neurobiological changes, such as reduced dopamine receptor availability in the striatum, which seems to contribute to the maintenance of addictive behavior. 19,21 However, because radioligand binding competes with endogenous dopamine, reduced receptor availability could also be due to sensitized dopamine release. 22,23 Recently, the important role of decreased dopamine D2 receptor availability in addiction was described in a meta-analysis including mainly in vivo neuroimaging studies, showing that reduced dopamine D2/3 receptors in the striatum (particularly in the caudate and putamen) play a crucial role in the emergence and maintenance of AUD. 14 This finding is further supported by a recent meta-analysis including mainly in vivo neuroimaging studies, showing reduced dopamine D2/3 receptors in the striatum, particularly in the caudate and putamen. 14 Further, reduced D2/3 receptor availability in the putamen is associated with increased craving and frequency of alcohol intake. 19,20 Eventually, successful treatment of AUD also may lead to beneficial changes in brain structure and function, including changes associated to dopaminergic neurotransmission. 24,25 The dimensional approach towards mental disease predicts that dopamine receptors undergo progressive changes during the development of the disease. However, to our knowledge, dopamine D2/3 receptor availability and related clinical phenotypes have yet not been investigated in individuals at high risk (HR) to develop AUD. To gain more insight into the development of subclinical addiction, it is therefore important to investigate neurobiological trajectories of dopamine D2/3 receptor availability starting in preclinical states. Therefore, here we investigated D2/3 receptor availability in individuals with (1) low risk (LR) to develop AUD (healthy controls), (2) individuals at HR to develop AUD (HR participants), and (3) detoxified patients with a diagnosis of AUD. Further, we studied possible associations between drinking behavior (measured by clinical questionnaires) and D2/3 receptor availability.
In association with findings from earlier studies, we expect lower striatal D2/3 receptor availability in the HR and AUD groups compared with the LR group. We further hypothesize that D2/3 receptor availability in striatal subdivisions is different in HR participants compared with AUD patients. Additionally, we expect interactions between D2/3 receptor availability, region of interest (caudate nucleus and putamen nucleus accumbens), and group (LR/HR/AUD).
Further, we expect clinical measures to be associated with changes in D2/3 availability.

| Participants
The present study is part of a multicenter project investigating neurobiological, reward-related deficiencies in AUD (see www.leadstudie.de; clinical trial number: NCT01679145; present sample contains data from subprojects "P2" and "P5"). Altogether, 58  LR controls and HR participants were recruited via local online platforms and advertisements in supermarkets and newspapers. Participants were assigned to the LR/HR group based on the individual score in the Alcohol Use Disorders Identification Test (AUDIT). 28,29 Subjects with an AUDIT score equal or below 8 were defined as LR control, a score above 8 was classified as HR control as described in the literature earlier. 28 More detailed information on recruitment and general procedure strategy is included in Figures S1, S2, and S4.

| Clinical assessment and psychological testing
At least 1 day before the MRI measurement and minimum 2 days before PET scanning, participants were informed and instructed on the background and procedures of the experiment. After signing written informed consent, participants underwent a standardized clinical assessment (neuropsychological testing and questionnaires). In AUD patients, additional clinical information was collected from patient history files (see Tables 1 and S2 for  (F(2,55) (factor group) = 0.98; p = 0.38). 33,34 PET data were acquired for 4 h after FP administration in three successive blocks with a break between each block 35 and in three successive blocks each of 30-min duration ( Figure S3). The first block (3 × 20 s, 3 × 1 min, 3 × 2 min, 2 × 5 min, 1 × 10 min) started with the intravenous tracer injection.
The second and third blocks (both 3 × 10 min) started 60 and 210 min after tracer injection, respectively. A low-dose CT for attenuation correction was performed before each block. Transaxial PET images were reconstructed using the iterative LOR-RAMLA algorithm of the scanner software with default parameter settings for brain (three iterations, 33 subsets, "normal" relaxation).
Spatial resolution in the reconstructed PET images was about

| Processing of PET data
Voxel-by-voxel parametric maps of the nondisplaceable FP binding potential (BP ND ) were obtained by the two-step simplified reference tissue method (SRTM) for noise reduction by use of a global rate constant of tracer clearance from the reference region. 36   Additionally, post hoc group comparisons were conducted as two sample t tests. We further calculated Pearson correlation coefficients (partial correlations; ρ) to investigate correlations between BP ND with the ADS scale. Additionally, mainly for visual purposes, exploratory whole brain analyses were conducted (voxel-wise, oneway, between-group ANOVA; factor: "group" [LR, HR, AUD]; covariates: age in years, smoking status), followed by post hoc tests (two-sample t tests).

| Voxel-wise whole brain analysis of dopamine receptor availability
Voxel-wise, whole brain analyses throughout the sample (LR, HR, and AUD; baseline contrast and for group comparisons LR-AUD; HR-AUD) revealed the highest BP ND in striatal areas as expected, alongside with broad but weaker BP ND in cortical areas. Mainly striatal differences were observed between the LR and AUD group.
Comparing HR with AUD subjects, again striatal differences were visible, but also differences in BP ND in neocortical areas were observed ( Figure 1). No differences were present between LR and HR groups (not shown).

| Multivariate results
The MANOVA results regarding BP ND in the CA, PU, and NA revealed global differences in mean BP ND between the groups (W = 0.77, F(6,104) = 2.42, p = 0.03, partial eta squared = 0.13).  Table 2; individual data are presented in Figure 2.

| BP ND and ADS scale-related findings
The groups differed significantly from each other in terms of ADS score (see Table 1 for descriptives and ANOVA). The LR group had a lower mean ADS score compared to the HR group (t = −2.16, df = 30, p = 0.04), whereas the AUD group had a higher ADS mean score than had the HR group (t = −5.80, df = 30, p = 0.01). In the whole sample, we observed a significant negative correlation between the ADS score and BP ND in the CA (df = 47, r = −0.32, p = 0.03, Figure 3) and a marginally significant correlation with the PU (df = 47, r = −0.27, p = 0.06, Figure 3). No significant correlations were observed within subgroups.

| DISCUSSION
In the present study, we investigated striatal D2/3 receptor availability in recently abstinent alcohol addicted patients, subjects with increased risk for AUD and low alcohol consuming healthy controls.
Differences in striatal D2/3 receptor availability between the three reference groups were observed in the PU and CA subregions.
Although, striatal D2/3 receptor availability in the PU and CA did not differ between the LR and HR group in direct comparison, a significant linear relationship was present between disease state and D2/3 availability in the PU throughout the whole sample. Those findings support the view of a continuous decrease of striatal D2/3 receptor availability during the development of AUD but may also reflect a predisposition to a differing extent in the subgroups. HR participants showed increased D2/3 receptor availability compared to LR and AUD groups in the CA (marginally significant). Further, D2/3 receptor availability was related to the sum score of the ADS scale in the whole sample.

| Group differences
Along with our hypotheses, we observed differences in the striatal postsynaptic dopaminergic neurotransmitter system between groups of subjects with different alcohol use behavior. Specifically, alterations in BP ND measured by 18 F-fallypride PET were shown in the PU and the CA. The present findings are in line with several recent PET studies, which found decreased D2-receptors in the striatum in AUD patients compared with controls. 19,20,[46][47][48][49][50] However, in earlier studies using [ 11 C]raclopride-PET, decreased D2/3 receptor availability was observed in the NA of AUD patients, 46,51,52 in contrary to the present study. Still, in line with the present study, two studies that applied 18 Ffallypride-PET reported no significant baseline differences in striatal D2/3 receptor availability, 53 or no baseline difference, but increased striatal D2/3 receptor availability during detoxification. 43 Subsequently, it seems that in studies applying 18 Ffallypride-PET in AUD, ventral-striatal differences in D2/3 receptor availability are consistently absent, whereas in studies applying [ 11 C]raclopride-PET, they are consistently present.
In general, dopamine neurotransmission (including receptor availability/density) in AUD may further depend on progress of addiction, as well as environmental, psychological, and biological factors (e.g., genetics). 9 It is important to note that HR participants had much broader and more variable D2/3 receptor availability in many different brain regions, including prefrontal, insular, and hippocampal areas (see Figure 1). This is in line with earlier studies, where it was proposed that transition towards addiction may involve dopaminergic factors in the orbitofrontal cortex, prefrontal cortex, cingulate gyrus, and the extended amygdala. 13 We believe that extra-striatal changes in D2/3 receptor availability might be modulated via changes in cortico-striatal connectivity, involving additional neurotransmitter systems (e.g., glutamate and GABA) as proposed in addiction and other mental diseases in earlier studies. 59 However, the discussion of those results is beyond the aim of the present study (manuscript in preparation).

| Dopamine receptor availability and alcohol dependence scale
Severity of alcohol dependence, measured with the ADS-Score, was related to DR2/3 availability in the CA in the whole sample. In general, the negative relationship between the severity of alcohol dependence and striatal D2/3 receptor availability underlines the theory of a compensatory downregulation of D2/3 receptors due to chronic alcohol intake but may also reflect lowered D2/3 receptor availability as a potential predisposition to developing an AUD. However, in the framework of the proposed dimensional view, we believe that the relationship between the severity of alcohol dependence and lowered dorso-striatal D2/3 receptor availability represents a gradual loss of control over drinking behavior and its possible neurobiological correlates. 58

| LIMITATIONS
We did not find differences in D2/3 receptor availability in the striatum between the LR and HR groups. Still, as described above, we found a negative correlation between addiction severity and D2/3 receptor availability in the PU.
Further, during voxel-based group comparisons of D2/3 receptor availability, we observed broad extrastriatal differences between the groups. The discussion of those results is beyond the scope of the present study.
It is important to note that in past research on D2/3 receptors in human individuals, highly variable methods of investigation and analysis were utilized. For instance, instead of rather than 18