Reduced Central Serotonin Transporters in Alcoholism

Several lines of research suggest that dysfunction of central dopaminergic and serotonergic neurotransmission predisposes to alcohol dependence and contributes to the maintenance of excessive alcohol consumption (1–3). In rodents, ethanol-induced dopamine release in the mesolimbic reward system reinforces alcohol intake (4, 5). Reduced homovanillic acid concentrations in cerebrospinal fluid (CSF) and a relatively increased density of dopamine uptake sites in the basal ganglia have been observed in alcohol-naive vervet monkeys that later developed alcohol preference (6). It has been hypothesized that alcohol releases dopamine in the ventral striatum by reversing dopamine uptake, so that dopamine transporters begin to release intra­cellular dopamine (7). Chronic alcohol consumption, on the other hand, has been associated with an increased dopamine turnover rate and decreased dopamine uptake sites and sensitivity of postsynaptic dopamine receptors (6, 8–12).

Another series of studies suggest that ethanol-induced reward may be independent of the activation of dopamine D₂ receptors and may be mediated through serotonin 5-HT_1B and 5-HT₂ receptors (13, 14). Increased sensitivity of postsynaptic serotonin receptors may be due to reduced serotonin turnover, which in rodents and nonhuman primates has been associated with impaired impulse control and increased alcohol consumption (3, 6, 15). In humans, low 5-hydroxyindoleacetic acid in CSF has been associated with aggressive behavior and early-onset alcoholism (16, 17). A facilitatory role for reduced functional capacity of serotonin transporters in rodents has been suggested as a pathogenic factor in alcohol dependence (18). Therefore, we examined dopamine and serotonin uptake sites in recently abstinent alcohol-dependent patients and healthy volunteers with the use of single photon emission computed tomography (SPECT) and the radioligand [¹²³I]β-CIT, which shows specific displacement by pharmaceuticals that bind to dopamine transporters in the basal ganglia and to serotonin uptake sites in the brainstem (19, 20). In the brainstem, the highest density of serotonin transporters is found in the raphe nuclei (21).

We assessed 1) the association between dopamine and serotonin uptake sites and chronic alcohol consumption and 2) the impact of these factors on behavioral measures that have previously been associated with them (e.g., impulsivity, anhedonia, depression, and anxiety). Moreover, we controlled for the effects of smoking on monoamine uptake sites, because over 80% of chronic alcoholics smoke, and nicotine may alter central dopamine release and metabolism (22, 23).

METHOD

All subjects provided written informed consent for this study under protocols approved by the institutional review boards of the Intramural Research Program of the National Institute on Alcohol Abuse and Alcoholism (NIAAA) or the National Institute of Mental Health. Twenty-two male patients who met the DSM-III-R criteria for alcohol dependence participated in the study. The exclusion criteria were current drug abuse (a positive urine drug screen) and a past history of dependence on a drug other than alcohol, serious head trauma, Korsakoff’s syndrome, or the presence of psychiatric and neurological diseases (axis I diagnoses) unrelated to alcoholism. The subjects (mean age=38 years, SD=7, range=27–51) had undergone alcohol withdrawal as inpatients in the Intramural Research Program of the NIAAA. SPECT scans were acquired after 3–5 weeks of supervised abstinence (verified by random breath testing) to avoid the confounding effects of alcohol withdrawal on monoamine neurotransmission (24, 25). Thirteen age-matched, healthy male volunteers (mean age=36 years, SD=10, range=22–55) served as comparison subjects. They did not have any DSM-III-R axis I diagnoses and had no history of drug or alcohol abuse. Fourteen alcoholics and two healthy comparison subjects were cigarette smokers.

Impulsivity was assessed with the Impulsive Nonconformity Scale (26), and anhedonia and apathy were rated with the Scale for the Assessment of Negative Symptoms (SANS) (27), by a psychiatrist (A.H.) on the day of the SPECT scan for each subject. Further ratings, acquired weekly by a group of trained research nurses (including C.G.), were available for 19 of the 22 patients after their admission to the NIAAA treatment and research program. These ratings included anxiety as assessed with the State-Trait Anxiety Inventory (28) and depression as appraised with the Beck Depression Inventory (29) and the Hamilton Depression Rating Scale (30).

[¹²³I]β-CIT SPECT Procedure

On the day before SPECT scanning and on 3 subsequent days, subjects received five drops of Lugol’s solution orally to reduce uptake of radioactive iodine into the thyroid. [¹²³I]β-CIT has been shown to bind with high affinity to dopamine and serotonin transporters (31); in the brainstem, the radioactivity is specifically displaced by ligands binding to serotonin uptake sites, while in the striatum, the binding is displaced by ligands binding to dopamine transporters (19, 20). Preparation of [¹²³I]β-CIT has been described previously (32). Each subject received a dose of 222–259 MBq (6–7 mCi) of [¹²³I]β-CIT. Free concentrations of [¹²³I]β-CIT in blood plasma were assayed by thin-layer chromatography of plasma concentrated ultrafiltrates (30-kDa cutoff) (33). A 60-minute SPECT scan was acquired at a mean of 22 hours (SD=1, range=21–26) after injection, when equilibrium at the dopamine and serotonin transporters is established (19, 20). Specific displacement in the brainstem of the radioligand by citalopram has been demonstrated in humans over this time range (20). During this time, smokers continued ad-lib use of nicotine to avoid effects of nicotine withdrawal; however, they were asked to abstain from smoking in the 60 minutes before SPECT scanning.

SPECT data were acquired with a CERASPECT gamma camera (Digital Scintigraphics, Waltham, Mass.) with a high-resolution (7.5 mm full width at half maximum) collimator in 120-projection step-and-shoot mode. The photopeak (143–175 keV) and two windows used for scatter correction (127–143 keV and 175–191 keV) were acquired. Reconstruction by back-projection with a 10th-order Butterworth filter (1-cm cutoff) generated an isotropic volume (1.67-mm voxels) of 64 128×128 transverse slices. The SPECT camera was calibrated before each scan session by imaging a 1-liter uniform flood phantom of known radioactivity similar to that observed in the brain (156 nCi/ml at 22 hours after injection).

Individual regions of interest were drawn for each subject on the basis of magnetic resonance images, which were coregistered with the SPECT scans. Raters drawing regions of interest and analyzing SPECT data were blind to diagnosis and subject ratings. Regions of interest at the level of the midbrain demarcated the ventral (substantia nigra) and dorsal (raphe nuclei) brainstem; cerebellar regions of interest were drawn at the level of the pons. Regions of interest drawn at a midthalamic level demarcated the caudate nucleus, putamen, thalamus, and various cortical regions of interest (frontal, temporal, anterior cingulate, and occipital) (figure 1). Of these regions, the highest density of serotonin transporters is found in the dorsal brainstem area, which contains the raphe nuclei (21), while dopamine transporters are predominantly found in the basal ganglia (caudate and putamen) (34). Regions of interest were measured in five consecutive slices, forming a volume of interest, and the average count per minute per milliliter in each volume of interest was tabulated and corrected for decay. Subtraction of the cerebellum measurement corrected for nonspecific binding. The effective binding potential (BP′=B_avail/K_d) was determined as the ratio of the specific binding to the free [¹²³I]β-CIT concentration in plasma, which may be assumed to be equal to the free synaptic concentration of ligand (35). This effective binding potential differs from the true binding potential (BP=B_max/K_d) by virtue of endogenous neurotransmitter binding to transporter sites (B_endog), which reduces transporter availability (B_avail=B_max–B_endog) (19). BP′ was chosen as the outcome measure instead of the ratio of the specific binding to the cerebellar uptake (V₃″) (35) for three reasons. First, assessment of BP′ corrects for possible differences in radioligand metabolism between alcoholics and comparison subjects, which could affect the availability of the unmetabolized radioligand in the brain. Second, cerebellar atrophy in alcoholism can affect the assessment of nonspecific binding and thus systematically bias V₃″. To test this hypothesis, the distribution volume, V₂(35), for the cerebellum (ratio of cerebellar β-CIT concentration to free concentration in plasma) was analyzed. Third, β-CIT retention in the cerebellum is exceedingly low at 22 hours after injection (19, 20, 35); this leads to small and relatively variable SPECT measurements for the cerebellum. As a subtractive correction for nonspecific binding, the increased variability of the cerebellum data is of little consequence, since the correction itself is a small one. However, as a normalization divisor, the full percentage variability of the cerebellum data propagates directly into V₃″.

Statistical Analyses

Statistical analyses were performed with the use of Statistica for Windows, Version 5.0 (Stat Soft, Tulsa, Okla.). Differences between groups in β-CIT binding in the measured volumes of interest and in the clinical data across groups were compared by means of Student’s t tests. Correlations between β-CIT binding and behavioral data were assessed with Pearson’s linear correlation coefficients (r).

RESULTS

[¹²³I]β-CIT Binding in Alcoholics and Normal Comparison Subjects

The effective binding potential (BP′) of [¹²³I]β-CIT was significantly reduced in the dorsal brainstem (raphe nuclei) area (t=2.53, df=33, p<0.02) (figure 2). No significant differences between alcoholics and comparison subjects were found in any other volumes of interest (t>1.60, df=33, p>0.12), and free parent β-CIT concentrations in blood plasma at 22 hours after injection showed no significant differences (comparison subjects: mean=125 pCi/ml, SD=40; patients: mean=113 pCi/ml, SD=42) (t=0.83, df=33, p=0.41). Cortical and cerebellar uptake was more than 30-fold lower than basal ganglia uptake and more than fivefold lower than brainstem uptake, making these volumes of interest difficult to measure accurately.

Post hoc division of alcoholics into type I and type II (onset before 25 years of age [17]; N=15) showed a significant reduction in dorsal brainstem (raphe nuclei) BP′ in the type I alcoholics (t=2.71, df=18, p=0.01) but not in the type II alcoholics (t=1.73, df=26, p<0.10). When only nonsmokers in each group were compared, the reduction in dorsal brainstem (raphe nuclei) BP′ remained significant (t=2.27, df=16, p<0.04), with no significant differences arising in any other volumes of interest.

Psychopathological and Behavioral Measurements

Compared with the healthy subjects, the alcoholics at admission exhibited significantly greater state anxiety (t=4.49, df=20, p=0.0002) and trait anxiety (t=3.37, df=20, p=0.003) on the State-Trait Anxiety Inventory and significantly greater depression on both the Beck Depression Inventory (t=2.88, df=20, p=0.009) and the Hamilton depression scale (t=2.91, df=20, p=0.009). During the first week of inpatient treatment, depression decreased significantly in the alcoholics according to both the Beck inventory (t=2.62, df=17, p=0.01) and the Hamilton depression scale (t=5.48, df=18, p<0.0001).

At the time of scanning, there was significantly greater anhedonia (t=4.47, df=33, p<0.001), apathy (t=3.27, df=33, p<0.003), and affective flattening (t=3.11, df=33, p<0.004) among the alcoholics than among the comparison subjects. No significant increase in impulsivity-nonconformity was observed in the type II alcoholics or in all of the alcoholics as compared with the healthy subjects. The type II alcoholics had significantly higher levels of lifetime alcohol consumption than the type I alcoholics (t=2.94, df=15, p=0.01, two-sided separate variance estimates); the type II alcoholics also exhibited significantly greater anhedonia than the type I alcoholics (t=2.53, df=20, p=0.02). Other clinical variables did not significantly differ between the type I and type II alcoholics.

Correlations Between Neurobiological and Psychopathological Data

The BP′ of β-CIT in the dorsal brainstem (raphe nuclei) was negatively correlated with lifetime alcohol consumption (r=–0.37, N=34, p=0.03) but not with age (r=–0.06, N=35, p=0.75). Moreover, raphe nuclei BP′ was negatively correlated with initial anxiety assessments in all subjects on both the state scale (r=–0.55, N=22, p=0.008) and the trait scale (r=–0.46, N=22, p=0.03) of the State-Trait Anxiety Inventory. Considering only alcoholics on the first day after admission to the ward, the score on the state scale was significantly negatively correlated with the dorsal brainstem BP′ (r=–0.64, N=19, p=0.003). Furthermore, the assessment of depression in the alcoholics by the Hamilton depression scale closest to the time of the SPECT scan was also negatively correlated with the dorsal brainstem BP′ (r=–0.49, N=18, p=0.04).

DISCUSSION

We observed a significant reduction in the effective binding potential of β-CIT in the dorsal brainstem of alcoholics, the site of the raphe nuclei, which is an area rich in serotonin cell bodies and serotonin uptake sites (21, 36). Further, the effective binding potential of β-CIT was negatively correlated with lifetime alcohol consumption. Reduced β-CIT binding to serotonin transporters may be due to increased serotonin concentrations in the synapse or to a decrease in transporter density or affinity (19, 37). Previous studies have addressed the impact of endogenous neurotransmitters on β-CIT binding in vivo. Laruelle et al. (19) observed a decrease in β-CIT binding in the basal ganglia of nonhuman primates after in vivo amphetamine administration to stimulate release of endogenous dopamine. The decrease they observed was greater than would be expected from direct competition between β-CIT and amphetamine for transporter sites, implying displacement of β-CIT by endogenous dopamine (19). Similarly, positron emission tomography studies with [¹¹C]cocaine in baboons have demonstrated that this radioligand is sensitive to pretreatment with drugs that inhibit dopamine release (38). However, other studies have failed to show an effect of endogenous dopamine concentrations on the binding of [¹¹C]d-threo-methylphenidate to monoamine transporter sites in the baboon striatum (39). Because serotonin turnover is known to decrease to normal or subnormal levels during abstinence in alcoholics (24, 40), it seems unlikely that the reduction in β-CIT BP′ that we observed in the raphe nuclei area is due to increased serotonin concentrations in the synapse. We suggest that the decrease in the availability of raphe serotonin transporters most likely reflects an actual reduction in transporter density due to cumulative toxic effects of ethanol consumption.

In accordance with the study of Tiihonen et al. (8), we did not observe a reduction in dopamine transporter availability in this group of mainly type II alcoholics. No significant reduction in cortical β-CIT BP′ was found, which may be because of the relatively low density of serotonin transporters and consequently low β-CIT signal in the cortex. The highest density of cortical serotonin transporters is found in the visual cortex; yet it is still about 10-fold lower than that in the raphe nuclei area of the brainstem (21). Similarly, the β-CIT signal in the cerebellum was very low, making determination of differences between groups impractical.

A dysfunction in the raphe system is likely to interfere with cortical and subcortical serotonergic projections (36) and to have a secondary impact on dopamine neurotransmission (41, 42). Overall reduced functional capacity of serotonin transporters in rodents and a reduction in serotonin turnover rate in humans and nonhuman primates have been implicated in the pathogenesis of alcoholism and in impulsivity (15–18). In the present study, however, we did not find a significant correlation between impulsivity and serotonin transporter availability in the raphe area, nor was impulsivity associated with the amount of lifetime alcohol consumption. The reduction in raphe β-CIT BP′ was significantly correlated with increased levels of anxiety, but curiously, the raphe β-CIT BP′ was associated with anxiety levels only during early abstinence. On the other hand, one measure of depression (Hamilton depression scale) obtained close to the time of the SPECT imaging was negatively correlated with raphe β-CIT BP′. As in previous studies (12), we observed a rapid decrease in depression in alcoholics after withdrawal, which may be due to therapeutic effects of the clinical setting or to short-term changes in serotonin, dopamine, and norepinephrine turnover rates after detoxification (3, 24, 43, 44. Serotonin transporter density in the raphe nuclei, on the other hand, is known to respond with considerable delay to changes in synaptic serotonin concentrations (45, 46) and may be a stable measure of the functional capacity of the serotonergic system. Reduced availability of raphe serotonin transporters may thus predispose an individual to depression and anxiety under certain conditions, such as chronic ethanol intoxication or withdrawal-associated detoxification stress. To clarify these issues, further studies are required that simultaneously measure depression and both the serotonin turnover rate and the availability of serotonin transporters during withdrawal and in abstinent alcoholics. Anxiety-related traits have recently been associated with a variation in the promoter region of the serotonin transporter gene (47). Interactions between serotonergic neurotransmission and anxiety and depression have been postulated before (48, 49), and the effect of selective serotonin uptake inhibitors on anxiety and depression is well documented (50). Our findings support the hypothesis that the availability of serotonin uptake sites is a factor related to anxiety and depression in alcoholism.

In conclusion, we observed a significant reduction in the availability of serotonin transporters in the raphe nuclei area of the brainstem of recently abstinent alcoholics. Reduced availability of serotonin transporters was correlated with lifetime alcohol consumption and was significantly associated with increased levels of anxiety and depression during early abstinence, factors that are known to influence short- and long-term treatment outcome in alcoholics (51). It is tempting to conclude that chronic alcohol intoxication reduces serotonin transporter density, which in turn affects anxiety and depression, increasing the risk of relapse in alcoholics and generating a vicious cycle of alcohol dependence. Further studies are necessary to assess whether the observed changes in indexes of monoamine transmission might be successfully targeted by pharmacotherapeutic approaches to alcoholism.

Received July 23, 1997; revision received March 30, 1998; accepted May 8, 1998. From the Clinical Brain Disorders Branch, NIMH, Neuroscience Center at St. Elizabeths Hospital, Washington, D.C.; and the Laboratory of Clinical Studies, Division of Intramural Clinical Biological Research, National Institute on Alcohol Abuse and Alcoholism, Bethesda, Md. Address reprint requests to Dr. Heinz, Department of Neurology of the Ruhr-Universität Bochum, St. Josefs Hospital, Gudrunstr. 56, 44791 Bochum, Germany; [email protected] (e-mail). Dr. Linnoila died in February 1998. Supported in part by grant Az: He 2597/1-1 from the Deutsche Forschungsgemeinschaft.