Additional related information may be found at:
Neuropsychopharmacology: The Fifth Generation of Progress

Back to Psychopharmacology - The Fourth Generation of Progress

Pharmacotherapy of Alcoholism

Charles P. O’Brien, Margaret R. Rukstalis, and Michael F. Stromberg


Alcohol abuse and dependence (alcoholism) are serious health problems throughout the world.  In the United States alone, excessive alcohol ingestion costs society more than 150 billion dollars annually and results in over 40,000 deaths (128).  Household surveys show the lifetime prevalence of alcohol abuse/dependence to be between 11 and 16% amounting to over 30 million Americans afflicted with this disorder at some time in their lives (142).  Until recently, medical approaches to the treatment of alcoholism focused on detoxification while long-term rehabilitation was the province of drug-free, non-medical programs based on the 12 Steps/Alcoholics Anonymous philosophy.  Some patients have been treated successfully with this approach, but the overall rate of relapse is substantial even in the best drug-free programs.  During the past decade, even therapists who previously opposed all medications as a “crutch” have come to recognize the potential for an approach that combines psychotherapy with a medication that decreases the tendency to resume alcohol drinking.  Psychopharmacologists have approached alcoholism by developing animal models of excessive alcohol drinking, analyzing the neurochemical factors involved in the reinforcement produced by alcohol and then testing medications that block or diminish this reinforcement.  The implicit hypothesis has been that if a medication acceptable to patients could be found that reduces the urges to resume alcohol drinking or diminishes the rewarding effects of alcohol, the results of treatment could be improved.  In this chapter we will describe some of the progress that has been made in classifying subtypes of alcoholics, in understanding the pharmacology of alcohol and in testing new and possibly more effective treatments for alcohol dependence.

Classification of alcoholics

Various subtypes of alcoholics have been described by clinical researchers in recent years, the characteristics of which suggest specific pharmacotherapies may be beneficial for subgroups of alcoholics.  Theoretically, subtyping alcoholics could assist in understanding the risk factors, etiology, and natural history of alcoholism and could be useful for matching types of alcoholics with treatments.

Researchers have attempted to search for pure subtypes of alcoholism and this has led to the development of a confusing list of single domain typologies including familial alcoholics and antisocial alcoholics.  Other typologists have attempted to incorporate multiple defining characteristics into their typological models including Tarter (164) Cloninger (26), von Knorring (173) and Babor( 5).  Tarter et al. (164) described primary alcoholics as experiencing more severe alcohol-related problems who present at an earlier age compared to secondary alcoholics.  The primary alcoholics reported more energy and disinhibition with a history of more childhood signs and symptoms, suggestive of hyperactivity/minimal brain dysfunction.  Cloninger et al. (26) also described two subgroups of alcoholics:  Type II was characterized as male-limited, with an onset of alcohol-related problems before the age of 25, often accompanied by aggressive behavior.  Type I was characterized as milieu-limited, with an onset of alcohol-related problems after age 25.  This subtyping was replicated in Swedish adoption data (150).  von Knorring et al. (173) classified alcoholics into either Type 1 or Type 2 using more clinically based criteria.  Type 2 alcoholics had impaired social or occupational functioning due to alcohol use before 25 years of age.  Babor et al. (5) proposed a dichotomous typology derived by an empirical clustering technique based on data obtained at the initial evaluation and follow-up of a cohort of inpatient alcoholics.  Type A was characterized by later onset and less severe alcohol dependence, fewer alcohol related problems, less psychopathology, less drug use and fewer childhood risk factors.  Type B was characterized by early onset of alcohol problems, greater severity of dependence, greater psychopathology and more childhood and familial risk factors.

Most recently, treatment research has utilized the typologic approach to examine prognosis following treatment for alcoholism.  Babor reported that prognosis was better for the Type A group compared to the Type B group in response to traditional treatment for alcohol dependence (5) and the typology has demonstrated predictive validity (79, 100). Babor Type B (5) patients with alcohol dependence respond better to coping skills treatment than to interactional therapy (100).  The Cloninger Type I alcoholics are similar to those described as Babor Type A (5) who respond better to interactional therapy than to coping skills treatment (100).  Recently, the multifactorial typologic approach has been used to demonstrate a differential treatment response in the preliminary data from a large double blind placebo controlled trial of sertraline for depressed and non-depressed alcohol dependent subjects.  Sertraline significantly decreased drinking more than placebo in Babor Type A alcoholics, but significantly increased drinking more than placebo in Type B alcoholics (132).  Clearly, the typologic approach to examine prognosis and/or match types of alcoholics with treatments is an interesting approach that requires more empirical evaluation and testing (87).

Whether specific neurotransmitter systems underlie subtypes of alcoholism remains a research area that merges basic science with clinical treatment for alcohol dependence (26).  Cloninger (26) has suggested that certain personality characteristics such as novelty seeking, harm avoidance, and reward dependence, differentiate the subtypes in their interactions with the environment.  Buydens-Brachey et al (15,16) reported differences in serotonin function in subjects with early and late onset alcohol dependence.  Patients with early-onset alcoholism before 20 years of age were significantly more likely to have a history of incarceration for violent crimes compared to patients who developed alcohol dependence after the age of 20.  The investigators found an inverse relationship between plasma levels of tryptophan and measures of depression and aggressivity in the early onset alcoholics but not the late onset alcoholics, suggesting that anti-social behaviors and early onset alcohol dependence may involve serotonin deficiencies.  Whether certain types of alcoholics (i.e. those with serotonin deficiencies) might benefit from treatments that modify serotonin function remains to be tested. 

It is well documented that alcoholism is more prevalent in relatives of alcoholics than relatives of non alcoholics (29).  A significant heritability for an increased vulnerability to alcoholism is suggested by twin studies wherein the concordance of alcoholism in monozygotic twins is much higher than in dizygotic twins (134,80).  Adoption studies also suggest a significant genetic component to the trait of excessive consumption of alcoholic beverages (24,25,57).  Although some genetic marker studies have proposed that certain biological characteristics may be associated with risk of excessive alcohol consumption, to date, none has been widely accepted (54).  The complexity of gene-environment interactions is apparent in the observations that many people at risk for alcoholism do not develop the disorder and that several traits associated with alcoholism (length of sobriety, frequency of consumption, and quantity consumed) may be inherited separately (61,62).

Basic Pharmacology of Alcohol

Ethyl alcohol, called alcohol in this chapter, is pharmacologically nonspecific and of low potency.  Alcohol pharmacokinetics are characterized by rapid absorption, relatively low first pass metabolism and quick distribution throughout body water.  The time from the last drink to a maximal concentration in blood ranges from 30-90 minutes.  Alcohol's elimination approximates zero order kinetics and 90-98% of alcohol that enters the body is oxidized in the liver.  The pharmacodynamic effects of alcohol that are of particular interest for understanding the mechanisms of its excessive ingestion are the reinforcing effects.  Reinforcement refers to the effects produced by a drug that increase the likelihood that once having experienced these effects, the user will continue to take the drug.  In general, these effects are interpreted by the user as pleasant or rewarding.  It is likely that development of medications for the treatment of alcoholism would be aided by a better understanding of how alcohol affects behavior through its interaction with neurochemical systems.

Preclinical Research


Alcohol's reinforcing effects are currently thought to be primarily related to the release of dopamine and serotonin in the mesolimbic system (85).  In this context, particular attention has been focused on the effects of alcohol on extracellular dopamine concentrations in the nucleus accumbens.  Alcohol, administered either locally into the nucleus accumbens in anesthetized animals (182), or systemically in freely moving rats

(36,74,113,183,184) or orally self-administered (176) increases extracellular dopamine concentrations in the nucleus accumbens.  Similar effects have been shown in rats selectively bred for high ethanol preference (82,176,185).   One recent study found that systemically administered alcohol increased extracellular dopamine in the nucleus accumbens of awake, but not anesthetized rats (145).  These authors suggested that alcohol’s effects on extracellular dopamine in the nucleus accumbens may originate indirectly through its action in another area of the mesolimbic dopamine pathway, the ventral tegmental area.   The ventral tegmental area contains the cell bodies of those dopamine neurons whose terminal fields project to the nucleus accumbens.   In support of this, alcohol also has been shown to  increase the firing rate of ventral tegmental dopamine neurons which project to the limbic forebrain including the nucleus accumbens (12,30,49,100).  

Behaviorally,  both dopamine agonists  (121,139,152,175) and antagonists (138) have been shown to reduce consumption of alcohol.  However, a recent examination of several agonists and antagonists selective for the two major subtypes of dopamine receptor families, D1 and D2,  suggested that these two receptor subtypes interact in a complex fashion to regulate alcohol consumption (67). 

At the biological level, the alcohol-preferring P rats  have been found to have a lower density of dopamine receptors in the nucleus accumbens compared to the nonpreferring NP rats (187).  Similarly,  extracellular levels of both dopamine and its metabolites have been found to be lower in the nucleus accumbens of alcohol preferring strains  (108,116).  The effect of these variations in the mesolimbic dopamine system have on the differential preference for alcohol shown by the high and low consuming strains is unclear at this time.  However, most theories of alcohol reinforcement view these dopamine pathways as critical for the maintenance of alcohol consumption.  Although these effects are observed at alcohol concentrations which are relevant to human social alcohol consumption, the direct demonstration of alcohol increasing extracellular dopamine concentrations in the nucleus accumbens in humans is still lacking. 

During acute alcohol withdrawal, extracellular dopamine concentrations are reduced in the nucleus accumbens of rats.  This finding has been suggested to explain alcohol withdrawal-associated aversive effects in rodents and by inference dysphoric mood during alcohol withdrawal in humans (144).  Certain behavioral experiments, however, appear to yield evidence that conflicts with the notion that extracellular dopamine concentrations in the nucleus accumbens are important for alcohol reinforcement.  For example, destruction of the dopamine neurons projecting to the nucleus accumbens has been reported to be without effect on alcohol preference in rats (81).  Moreover, low dose dopamine2  receptor blockers do not alter alcohol-induced place preference (32).  Additional behavioral studies are clearly needed to settle this issue.

There is a long history of experiments showing that the effects of alcohol and other abused drugs can be classically conditioned (10).  Environmental cues may serve as conditioned stimuli  that may be important determinants of expected reinforcing effects of substances of abuse including alcohol.  Recently, environmentally conditioned expectancy of alcohol in rats has been reported to decrease dopamine2 receptor binding in the nucleus accumbens suggesting a conditioned increase of  endogenous dopamine that competes with the radioligand (168).  In another study using microdialysis, rats who had been trained to consume alcohol in a specific environment showed increases in extracellular dopamine in the nucleus accumbens when again tested in that environment prior to receiving alcohol (176).  If environmental cues such as the sights and smells of people drinking produce alcohol-like effects in those dopamine pathways in the brain that are hypothesized to underlie the euphoric effects of alcohol, this process could serve as a priming stimulus ultimately leading to relapse and loss of control in recovering alcoholics.

Endogenous opioids in alcohol reinforcement

Extensive behavioral evidence links the endogenous opioid system to alcohol reinforcement.  There are at least three discrete opioid receptor subtypes, m, d and k  (52) and their role in maintaining alcohol consumption has most often been investigated by examining the effects of  opioid agonists and antagonists  on alcohol consumption in animal models.  For example, low doses of the prototypical m agonist morphine have been demonstrated to increase the consumption of alcohol (71,156), while the opioid antagonists naloxone and naltrexone have been shown to reduce consumption of alcohol in both rodents and nonhuman primates (31,42,71,73,119,120,140,146,157,170).  A dose-response effect for naltrexone in decreasing oral self-administration of alcohol at doses ranging from 0.1 mg/kg to 10.0 mg/kg for rats has been demonstrated (157).  In addition, this effect can be maintained when naltrexone is administered to rats in repeated daily sessions across 60 days (158).   These results suggest that naltrexone can be administered to rats  for extended periods without the development of tolerance produced by  mechanisms such as receptor up-regulation or supersensitivity.

The role of m and d  receptors has been investigated in an attempt to determine the contribution of specific opioid receptor subtypes with somewhat equivocal results.   For example, the selective irreversible m antagonist, b-funaltrexamine,  has been shown to reduce alcohol consumption in both outbred (157) and an alcohol -preferring rat strain (91).  The results for d selective antagonists are somewhat less clear.  For example, the d selective antagonists ICI 174864,  naltrindole and naltriben have been shown to reduce ethanol consumption in ethanol preferring rats (43,89,90) and the alcohol preferring C57/B6 mouse (94).  However, other laboratories have failed to find an effect of naltrindole or ICI 174864 on alcohol consumption in outbred rats (157),  other strains of alcohol preferring rats (70,73) or monkeys (179).  Similarly, naltrindole failed to block the discriminative stimulus properties of alcohol (153).  Taken as a whole, these data suggest that the role of specific opioid receptor subtypes in modulating alcohol consumption may vary both across and within species and that the d receptor may be more important in strains that have been selected for high alcohol consumption.

Further support for this hypothesized difference between opioid subtypes and alcohol consumption comes from evidence showing that alcohol preferring strains of mice and rats have biologically based differences  in both the density  and activity in their endogenous opioid systems.  For example, the alcohol preferring P rats have been shown to have a higher density in d receptors in the nucleus accumbens when compared to the alcohol nonpreferring NP rats (108).  In the AA alcohol preferring strain, higher densities of m receptors  were found in the nucleus accumbens and ventral tegmental area when compared to the nonpreferring ANA strain (34).  Further, a higher density of d receptors was found in the nucleus accumbens and ventral tegmental area of the alcohol preferring C57BL/6  mice when compared to the DBA/2 nonpreferring mice (35).   These data are consistent with the behavioral effects found for m and d selective opioid antagonists discussed above and suggest that those mechanisms underlying a complex trait,  such as alcohol preference, are heterogeneous.

Despite this research, the precise role of  the opioid system and specific opioid receptor subtypes in determining the motivational state underlying alcohol consumption is not yet well characterized.  However, it is hypothesized that alcohol, acting either directly or indirectly with opioid receptors in the ventral tegmental area and nucleus accumbens, modulates the activity of the mesolimbic dopamine system (63).  In support of this view is evidence from microdialysis experiments showing that both naltrindole (1) and naltrexone (6) block the alcohol-induced increase in dopamine in the nucleus accumbens of anesthetized rats.  In addition,  a recent experiment using microdialysis in awake alcohol self-administering rats,  naltrexone concurrently reduced both alcohol consumption and extracellular dopamine levels in the nucleus  accumbens  (56).

Studies in human subjects support the concept that alcohol affects the endogenous opioid system acutely in the 30-90 minutes following ingestion.  Gianoulakis and colleagues (51), Figure 1, studied non-alcoholic volunteers with a strong family history of alcoholism (High Risk) and compared them to volunteers with no family history of alcoholism (Low Risk).  Baseline levels of plasma ß-endorphin were lower in the High Risk subjects but a 0.5 gm/kg test dose of alcohol produced a significantly greater rise in plasma ß-endorphin in the High Risk group. In a subsequent study, alcohol increased the plasma level of ß-endorphin-related peptides of the High Risk but not of the Low Risk subjects in a dose dependent manner (51).  In alcoholics, baseline plasma ß-endorphins were depressed after cessation of drinking for at least six months in one study (50), but returned to normal range after six weeks of abstinence in another study (169).  These results support an acute effect of alcohol on the endogenous opioid system.

Taken together, these data are consistent with the hypothesis that alcohol ingestion stimulates the release of endogenous opioids which potentially increases some of the rewarding effects of alcohol through opioid mechanisms.  A small amount of morphine may produce alcohol drinking in rodents by priming this effect, but larger doses of an opioid drug could replace alcohol and thus reduce alcohol preference.  Naltrexone, by blocking opiate receptors, could block this mechanism of reinforcement and thus reduce or eliminate alcohol preference.  This hypothesis has been tested in acute alcohol administration studies where alcohol was given to healthy normal volunteers  who received naltrexone or placebo before drinking.  Naltrexone reduced the subjective “high” from alcohol compared to placebo as measured on the Biphasic Alcohol Effect Scale (163,83).  In a study of male subjects, pretreatment with naltrexone attenuated the stimulation response to alcohol in High Familial Risk subjects (social drinkers with an alcoholic father) compared to subjects with no family history of alcohol dependence (83).  The findings that naltrexone reduced a stimulant response to alcohol in those social drinkers with a positive family history for alcohol dependence was particularly interesting in light of the study (Figure 1) that reported alcohol increased the plasma levels of  b-endorphin-related peptides in male and female social drinkers from families with a history of alcoholism but not from families without a history of alcoholism  (51).  These data support the hypothesis that naltrexone opioid antagonism blocks some of the acute subjective effects of alcohol.  Additional support of this hypothesis has been obtained from data from clinical trials of naltrexone in alcoholics reviewed below.


Serotonin (5-HT) is yet another neurotransmitter that has been intensively investigated as a possible mediator of alcohol's reinforcing effects.  A serotonergic involvement in alcohol's reinforcing properties is suggested by studies showing that the serotonin agonist,  8-OH-DPAT, (108,162), serotonin reuptake inhibitors  (53,59,117,118,143) and 5-HT3 antagonists reduce alcohol consumption in both outbred and alcohol-preferring rats (39,66,84,93).  Serotonin's role in mediating alcohol consumption is hypothesized to emanate from its modulation of mesolimbic dopamine activity.   Support for this comes from research showing that  5-HT3 antagonists reduce the rate of neuronal discharge in the ventral tegmental area (112,136) and decrease the alcohol-induced release of dopamine in the nucleus accumbens (19,181,185).  Conversely,  serotonin, itself, potentiates alcohol-induced excitation of putative dopamine neurons in the ventral tegmental area (13) while CPDG, a 5-HT3 agonist, increases  alcohol-induced release of dopamine in the nucleus accumbens (18).

Additionally,  preliminary studies conducted in our laboratory have shown that the selective 5-HT1A antagonist, WAY100635, when administered alone, reduces voluntary alcohol consumption in outbred rats and, when administered in combination with the serotonin reuptake inhibitor, sertraline, it reduced alcohol consumption more rapidly than sertraline alone.  These results have recently been demonstrated in alcohol-preferring rats as well (188).  In the raphe nuclei, 5-HT1A receptors act as somatodendritic autoreceptors that function to inhibit firing of the serotonin neuron, decrease serotonin release, and thereby reduce extracellular serotonin.  If, as hypothesized, serotonergic agents function to decrease alcohol consumption by increasing extracellular serotonin, it might be expected that antagonists of the 5-HT1A autoreceptor will increase serotonin in the terminal fields in the ventral tegmental area and nucleus accumbens.

These data demonstrate that the relationship between serotonin and alcohol consumption is complex.  Perhaps this is related to the potential for differential effects of alcohol at each of  the seven known serotonin receptors  and their various subtypes.  In addition, alcohol-preferring strains of rats have been shown to have lower levels of serotonin and its metabolites in the nucleus accumbens of the alcohol-preferring P (108) and HAD rats (55) when compared to the nonpreferring NP and LAD strains. Further, alcohol-preferring P rats have been shown to have higher densities of  5-HT1A receptors (108).   These data suggest that higher alcohol consumption in the alcohol preferring strains may be due to biologically based deficits in their serotonin systems.  Additional work is required to more thoroughly examine this system once ligands selective for the various serotonin receptor types and subtypes become available.

Excitatory and Inhibitory Amino Acids

In intoxicating doses to humans, alcohol has major effects on ion channels and ion channel-receptor complexes. This is particularly true for the N-methyl-D-aspartate (NMDA) and g-aminobutyric acid (GABA)  receptors.  Alcohol decreases the NMDA receptor agonist-induced cation currents (102,68) and enhances the GABA-A receptor agonist-induced chloride flux (161).  Alcohol's effects on these ion fluxes are thought to be particularly relevant to its intoxicating, amnestic and ataxic effects (115,160,161). One hypothesis suggests that alcohol has a biphasic effect with its positive reinforcing properties appearing on the ascending limb of the blood alcohol curve, while its sedative, ataxic and anxiolytic effects appear at higher blood alcohol levels (96). The full meaning of the alcohol’s effects on ion fluxes through these ion-gated receptor complexes is, however, not clear at this time.

Recent research has focused on acamprosate, a taurine derivative showing structural similarity to the GABA receptor ligand.  This drug has been found to be effective in reducing relapse in recovering alcoholics.  When chronically administered, alcohol produces a neuronal hypoexcitability by antagonizing glutamate and enhancing GABA, withdrawal from chronic alcohol produces just the opposite, a neuronal hyperexcitability.   Acamprosate is hypothesized to have its primary effects through antagonism of excitatory neurotransmission at the glutamate receptor on the NMDA receptor complex and by stimulating inhibitory neurotransmission at either the GABAA or GABAB receptor subtypes (101).  In support of this, research has shown that acamprosate reduces postsynaptic excitability in the rat neocortex (186). 

Alternatively, glutamate receptors in both the amygdala and hippocampus send afferent projections to the nucleus accumbens and could potentially serve to modulate DA release in these positive reinforcement pathways (174).  It has also been proposed that alcohol’s potentiation of dopamine activity at receptors in the ventral tegmental area may be due to the inhibition of interneurons by GABA that, in turn, disinhibit dopamine neurons (60).  These data have lead to the suggestion that alcohol’s reinforcing properties may be mediated, in part, by glutamate afferents to the nucleus accumbens as well as GABA neurotransmission that ultimately affects dopamine release in the nucleus accumbens (137). 

If acamprosate attenuates alcohol consumption because of its action at GABA or glutamate receptors, which, in turn, modulate accumbens dopamine release, these actions would appear to emerge from a complex interaction of acamprosate and alcohol.   While acamprosate has been shown to decrease baseline alcohol consumption in some animal models (9,69,137), in other models it decreased consumption in alcohol dependent rats, but not alcohol naive rats (95).  Further, while acamprosate has been shown to attenuate the deprivation-induced increase, above basal levels, of alcohol, it appears to be ineffective in reducing baseline alcohol consumption (65,154).   These findings suggest that acamprosate may be most effective in preventing relapse following a period of withdrawal from alcohol.

Summary of Preclinical Research

As  reviewed above , the dopamine-alcohol reinforcement hypothesis is central  to most theoretical accounts seeking to explain those mechanisms underlying alcohol consumption in animal models.  Findings from these studies suggest that dopaminergic, serotonergic, opioidergic and NMDA receptor effects may yield useful pharmacotherapies for alcoholism.  Because these systems appear to have reciprocal connections with one another in those brain areas seen as critical for alcohol’s reinforcing potential,  and the potential for heterogeneity in the biological differences determining the sensitivity of these individual systems to alcohol ,  it would not be surprising to discover that effective pharmacotherapy  would require multiple agents that alter the activity of more than on neurochemical system and may differ dependent upon the characteristics of a specific subpopulation.

Human experiments involving alcohol consumption

Following the lead provided by the animal data, Naranjo and Sellers at the Addiction Research Foundation of Ontario in Toronto (122,123,125),  conducted a series of innovative studies administering various serotonin uptake inhibitors and a serotonin 2 receptor antagonist to male heavy “social drinkers” most of whom fulfilled DSM III-R criteria for alcohol dependence.  The investigators recruited their subjects by newspaper advertisements, screened them medically and psychiatrically and asked them to continue their usual alcohol consumption and to document the level of consumption carefully.  Subjects were then administrated zimelidine, citalopram 20 or 40 mg, fluoxetine, viqualone or ritanserin 5 mg or 10 mg, each in controlled, double blind, cross over studies with placebo.  The general findings can be summarized as follows:

1)    All serotonin uptake inhibitors except citalopram 20 mg reduced significantly the amount of alcohol consumed.

2)    The quantitative effect was modest, but clinically meaningful, in the 50 to 70% of subjects who were classified as responders.

3)    The effect was immediate.

4)     There was a reduction of body weight which coincided with the administration of serotonin uptake inhibitors, but was independent of their effect on alcohol consumption.

5)     Cigarette smoking was not reduced by the serotonin uptake inhibitors.

6)    The 40 mg dose of citalopram seemed to reduce reinforcing effects of alcohol.

7)    The 10 mg dose of ritanserin increased the reinforcing effects of alcohol

8)    While ritanserin decreased the liking and craving and desire for alcohol, it did not reduce alcohol intake.

Even though theoretically interesting, clinically these studies present the reader with several problems. The first question involves whether a study of this nature is in accord with guidelines for studies of human subjects.  Is it reasonable for an investigator studying alcohol dependent subjects consuming in excess of six drinks per day to implicitly condone their continued drinking for experimental reasons on an outpatient basis for two months or more?  Clearly the study was conducted with appropriate committee review and oversight, but the issues presented by the protocol are difficult and debatable.  Second, the clinical applicability of these data is limited, because the current treatment of alcoholism involves initial cessation of consumption and treatment of withdrawal followed by treatments conducive to abstinence. An important insight gained from the Naranjo and Sellers studies is that the mechanism for reducing alcohol consumption and the antidepressant mechanisms of action of serotonin uptake inhibitors are probably different.  The alcohol consumption reducing effect is immediate, whereas the antidepressant effect is delayed and the alcohol consumption reducing effect may require higher doses than the antidepressant effect.  As such, the Naranjo and Sellers results parallel somewhat the results of studies on the efficacy of fluoxetine in bulimia (41).

Serotonin 3 receptor antagonists have been reported to reduce voluntary alcohol consumption in rats.  Therefore, Sellers studied ondansetron, a drug from this class that is approved for the treatment of vomiting.  Using the paradigm described above, Sellers found that in a very low but not in a medium dose, ondansetron reduced alcohol consumption in heavy social drinkers.  In a 6 week randomized, placebo-controlled study, of .25 mg or 2.0 mg oral  ondansetron for alcohol dependent patients,  the lower dose group reduced their drinking more than the higher dose group and both groups drank less than the placebo group.  Once the heaviest drinkers were excluded from the analysis, medication effects were more pronounced.  These data suggested that the heavier drinkers added variability and emphasized that the efficacy of a drug may be different based upon the patient populations and doses tested (147).  

Gorelick performed an inpatient study in heavy drinking alcohol dependent men in a Veterans' Administration Medical Center (58).   He administered placebo and up to 80 mg fluoxetine/24 hours for 4 weeks on a research ward where alcohol was available according to a fixed interval procedure.  Fluoxetine reduced alcohol consumption by 14% but only during the first week.  The author attributes the lack of efficacy during the last three weeks of the study to the development of tolerance to fluoxetine.  An equally plausible explanation is an alcohol-fluoxetine interaction.  The initial effects of fluoxetine alone could have been overcome by the adaptation produced by the combination of alcohol and fluoxetine.  In animal studies, co-administration of alcohol reverses receptor changes produced by noradrenergic antidepressants and neuroleptics.

Treatment of Alcoholism

The pre-clinical studies described above have stimulated clinical research that integrates medication into the existing treatment approaches for alcoholism.  There are two distinct phases in the current treatment of alcohol dependence: detoxification and rehabilitation.  Detoxification consists of stopping the ingestion of alcohol so that alcohol in the body is metabolized and the various organ systems affected by alcohol can re-adjust to the absence of alcohol.  During this re-adjustment after abrupt stopping of alcohol, withdrawal symptoms consisting of rebound effects may occur.  In most cases these symptoms are mild and do not require hospitalization.  Depending on the dose of alcohol and the general health of the patient, the withdrawal syndrome can be severe and life-threatening and medical evaluation is important.  Benzodiazepines are so effective in the management of the alcohol syndrome that the majority of alcoholics with uncomplicated withdrawal can be managed on an outpatient basis (61).  For a full description of the management of withdrawal, see  Jaffe et al (75).  While the withdrawal phase of treatment can be treated very successfully with modern medications, most patients will quickly relapse unless they are engaged in the rehabilitation phase.

Rehabilitation of alcoholics and other drug dependent patients should consist of a long-term (years) structured program aimed at helping the patient cope with the life problems likely to provoke relapse.  Currently there are many 28 day treatment programs that promote abstinence via group and individual therapy sessions in a hospital or residential sheltered living arrangement.  Follow-up care frequently consists only of referral to a local Alcoholics Anonymous (AA) group.  The AA program is extremely helpful to many patients, but without an individualized program directed by a therapist experienced in the management of substance abuse, relapse to uncontrolled drinking is common.  Even in good outpatient programs that combine AA and individualized treatment, relapse rate at three months is in the neighborhood of 50 per cent (38,40,44,45,127,129).  Clearly there is a need for medications that could be used in conjunction with psychosocial rehabilitation programs.

Clinical Trials

Methodological Issues

The course of alcoholism is quite variable so in any test of a medication or treatment approach, a randomly assigned control group is essential.  Evaluation prior to beginning treatment should be comprehensive.  Simply recording the daily intake of alcohol is not adequate.  Important variables that may influence outcome include presence of other psychiatric disorders, gender,  age, history of prior treatment, use of other drugs and the degree of problems in the family, social, occupational, legal or medical areas.  The sample size should be large enough so that there is general equality across all of these important areas between treatment and control groups.  Pre-, during and post-treatment evaluations should be conducted by investigators who are not aware of the patients group assignment.

Determining the optimal dose, the minimum effective dose, and the maximum tolerated dose of a medication are essential in pharmacotherapy trials in patients with alcohol dependence.  Clinical dosing strategies must consider possible alterations in compliance, drug metabolism or drug-alcohol interactions in alcoholics.  The therapeutic dose response curves that have been determined for a medication in non-drinking patients can not be extrapolated to patients with alcohol dependence since chronic and/or  acute alcohol ingestion may alter the absorption, distribution or metabolism of drugs (99).  Ciraulo et al (22) examined depressive symptoms and the steady state plasma levels of imipramine in 23 depressed patients with alcohol dependence compared to 12 non depressed alcoholics following 10 days of 150 mg oral imipramine.  Patients with alcohol dependence had significantly lower plasma levels of imipramine and it’s principle metabolite compared to the non alcoholic depressed patients.  Importantly, the lower plasma levels corresponded with a failure to respond to the antidepressant effects of imipramine, since the symptoms of depression did not improve in the alcohol dependent patients but did significantly improve in the non-alcohol dependent depressed patients.  Ciraulo et al hypothesized that the results are consistent with decreased oral bioavailability or enhanced clearance of imipramine in patients with alcohol dependence.  This hypothesis was supported in another study where Ciraulo et al (23) found increased clearance of imipramine and desipramine in 15 recently detoxified men with alcohol dependence, compared with 15 healthy volunteers.  In a third study, Mason (107) examined alcohol consumption, liver function, desipramine pharmacokinetics and treatment response to desipramine for depression in depressed patients with or without an alcohol use disorder.  The baseline hepatic microsomal functioning measured by the aminopyrine breath test (ABT) was higher in the 23 alcohol dependent subjects compared to the 13 non-alcoholic patients.  Patients were titrated up to clinically indicated oral doses of desipramine and maintained on the regimen for the treatment period.  At week 4, the mean level of desipramine in plasma was significantly (p< .02) lower among alcoholics (Mean, sd= 89.7, 52.1 ng/ml) than among nonalcoholic (mean, sd = 193.7, 11.6 ng/ml) depressed patients in spite of the fact that the mean dosage of oral desipramine was higher for the alcohol-dependent patients (204 mg/day) compared to the non-alcohol-dependent patients (172 mg/day).  The authors concluded that:  1) hepatic microsomal enzyme activity of alcohol-dependent patients was increased relative to nonalcoholic patients; 2)  the mean ratio of plasma desipramine/oral dose was lower in alcohol-dependent patients compared to non-alcohol dependent patients; 3) the ratio for alcoholic patients improved following 6 months of treatment, but remained lower compared to the non-alcohol dependent patients at baseline. 

Dose response relationships for the treatment for alcohol dependence have been reported in a clinical studies with other classes of medications including an oral opioid antagonist nalmefene and the serotonin reuptake inhibitors.  Results from 14 patients randomized to nalmefene 10 mg or 40 mg daily were compared to the results from 16 patients who were randomized to receive nalmefene 20 mg or 80 mg daily in addition to weekly cognitive-behavioral therapy (106,107).  In spite of the fact that the 80 mg group had indicators of more severe alcohol dependence at baseline, the patients who received the highest dose were more likely to complete the 12 week trial and showed a more stable response.  Those subjects who received the lowest dose of nalmefene (10 mg) had the poorest outcomes and patients on the intermediate doses (20 and 40 mg) had responses that fell between the low and high dose groups.  In a review of the use of serotonin selective pharmacotherapies in 13 clinical studies for alcohol dependence, Pettinati summarized that therapeutic effects were generally found in patients who received higher doses (131).  Thus, the results from studies with different classes of medications emphasize the importance of examining both the pharmacokinetic and pharmacodynamic dose response relationships in pharmacotherapy trials in patients with alcohol dependence.

Compliance with medication is an important variable in any clinical trial.  Studies conducted in populations of alcoholics or other drug abusers should be carefully evaluated for compliance because those who are sufficiently motivated to comply with the medication regimen may be more likely to remain abstinent irrespective of medication.  It is particularly misleading to measure compliance in the medication group without also measuring it in the placebo group.  Thus any compliance measure must apply equally to drug and placebo groups.  With most medications, one can measure compliance directly by monitoring plasma levels of the compound, but measures of placebo compliance are problematic.  In some studies riboflavin has been added to both drug and placebo and urine fluorescence is tested on each clinic visit.  This procedure presents several practical problems because of an inability to monitor riboflavin-containing dietary supplements and because the urine may be falsely negative if it is obtained within one hour or more than 12 hours after ingesting the study medication.  Also, the addition of riboflavin to a drug in the investigational phase is regarded as a new formulation and thus data from a study where riboflavin has been added could not be used for potential FDA approval of a new medication.  Residual pill counts are another method to assess compliance that does not interfere with the drug development process.  Of course this method too can be misleading if a patient is determined to deceive the investigator while not dropping out of the study.  Comparisons of the riboflavin method and pill counts indicate comparable accuracy (159).  Pill counts can be enhanced by giving research patients a container that has a different number of pills each week and determining how many were ingested by counting the remainder.  Yet another method involves special pill bottles containing a micro-chip in the lid that records the dates and times that the lid is removed.  Clearly compliance should be addressed in some fashion during all clinical trials of new medications.

Another variable that should be quantified in clinical trials is the level of psycho-social intervention.  Since some detoxified alcoholics do very well while receiving no medication or when assigned to placebo, the non-medical aspects of a rehabilitation program should be assessed.  Of course, in a clinical trial the amount of psychosocial treatment should be specified and equal for all research patients.  In order to insure this equality and to assess the psychosocial component in all patients, a treatment assessment instrument (109) has been developed that is patterned on the Addiction Severity Index (ASI).  Data are obtained directly from the patient as the number of minutes spent each week with a helping person or a self-help group working on problems in each of the seven areas of the ASI.

Any discussion of clinical trials methodology should include a consideration of the population that volunteers for a treatment requiring random assignment.  Studies (155) of the alcoholics who refuse to volunteer after the study is explained to them indicate that in general, the volunteer study subjects are more severely ill than the refusers.  Thus double-blind trials may under-estimate medication efficacy because of the nature of the patients being treated.

Historically, women have been underrepresented in pharmacotherapy trials and recent changes in the FDA and NIH guidelines mandate 1) the inclusion of women and minorities as subjects in clinical research and 2) safety and efficacy analyses according to sex (111,167).  Despite these recent mandates, limited data is available about women with alcohol dependence.  Numerous studies have reported an accelerated progression of alcoholism in women (133).  Although epidemiological studies have consistently reported female alcoholics drink less than male alcoholics (64,180), female alcoholics have a higher rate of medical (21,99,178) and psychiatric comorbidity (33,166) compared to male alcoholics.  Based upon smaller total body water and less gastric alcohol dehydrogenase activity in women compared to men, a given amount of alcohol may produce significantly higher blood ethanol concentrations in females as compared with males (78,99). These studies suggest that men and women with alcohol dependence may present with qualitative and quantitative differences and studies to examine gender effects on the efficacy and side effects in pharmacotherapy trials are needed.

The clinical trials to be discussed in the remainder of this chapter have strengths and weaknesses and none solves all of the methodological issues mentioned above.  While individual study results may be promising, replication of trials in different research centers lends credibility to the effectiveness of an intervention.

Aldehyde dehydrogenase inhibitors

Disulfiram (Antabuse) is used as a deterrent drug for the treatment of alcohol dependence.  Disulfiram inhibits the metabolism of alcohol and when combined with alcohol, relapsing patients may experience aversive effects including nausea, vomiting , sweating, flushing, hypotension, shortness of breath, tachycardia, blurred vision and possibly confusion described as the disulfiram-ethanol reaction (DER) (141).  A single daily dose of 125 mg or 150 mg disulfiram binds irreversibly and permanently inactivating aldehyde dehydrogenase (ALDH).  Although disulfiram has been used for the treatment of alcohol dependence for over 50 years, results from the few placebo controlled trials are equivocal (72).  A large multi-center study which randomized 600 male veterans with alcohol dependence to daily 1 mg or 250 mg disulfiram or a placebo control group found a direct correlation with compliance in all groups and abstinence (44).  The advantage of disulfiram was seen only in those subjects who resumed drinking; the therapeutic dose group had significantly fewer days drinking compared to those in the other two groups.  The most common side effects reported for disulfiram include hypertension, hepatotoxicity, lethargy, drowsiness and peripheral neuropathy.  Chick et al (20) conducted a randomized, partially blind six month follow-up study with 126 patients who received 200 mg disulfiram or 100 mg vitamin C under supervision.  The patients on disulfiram had more total days of abstinence, reduced the amount of alcohol consumed weekly, and had lower levels of serum GGT compared to the vitamin C group.  The authors report that in spite of ongoing drinking while on disulfiram, few patients reported minor side effects.  In addition, disulfiram has also been shown to reduce the clearance of other medications including phenytoin, warfarin, caffeine, chlordiazepoxide and diazepam (92).  Based on the risks involved in relapse, poor patient compliance, side effects and increased drug-drug interactions widespread use of disulfiram has not developed.        

However, disulfiram is an effective pharmacological agent for the treatment of alcohol dependence and a better understanding of which patients might benefit from disulfiram is warranted (92).  Azrin et al (4) developed programs to enhance compliance with disulfiram involving a written agreement with the therapist and a second person witnessing the ingestion of disulfiram.  The Community reinforcement Behavioral Therapy program (151) further developed the enhancement of compliance and involved mandatory therapy.    Kristenson suggests conditions for successful treatment with deterrent drugs including:  written contract and consent for treatment, acceptable supervision of ingestion of disulfiram, accessible outpatient treatment and possible in patient back up (92).

Serotonin 1A receptor partial agonists

Buspirone has been investigated in several studies for the treatment for alcohol dependence.  It is a serotonin lA receptor partial agonist with weak dopamine 2 receptor antagonist potency, and is quickly metabolized to 1-pyrimidinylpiperazine (1-PP) which is an a2 receptor antagonist.  Bruno in Italy investigated buspirone 20 mg/24 hours for 8 weeks in a double blind placebo controlled study (14).  There were significantly fewer dropouts and the patients reported greater anxiety reduction in the buspirone than in the placebo group.  In addition, patients reported a significant reduction in their craving for and intake of alcohol during the buspirone treatment.  Tollefson et al. (165) conducted a 24 week, double blind, placebo-controlled investigation of buspirone 40 mg per/24 hours.  All patients had an anxiety disorder in addition to alcohol dependence and they had maintained abstinence for 30 days prior to the start of the medication.  More patients on buspirone than placebo attained clinically meaningful anxiety reduction and were retained in the study, but no data were given on maintenance of abstinence.  Malcolm et al. (103) treated 67 anxious, alcohol dependent patients in a Veterans' Affairs Medical Center in a double blind, placebo-controlled, parallel groups study for 6 months with 45 to 60 mg buspirone/24 hours.  Several outcome measures were used in this study and no benefit over placebo could be documented for buspirone on either alcohol consumption or anxiety-related variables.  Kranzler et al (86) conducted a double-blind, randomized, placebo controlled trial of buspirone for subjects with alcohol dependence and anxiety disorder.  61 subjects participated in the twelve week pharmacotherapy trial which included weekly relapse prevention psychotherapy.  Compared to placebo, buspirone therapy reduced anxiety, delayed the return to excessive alcohol consumption and lowered the number of drinking days during follow-up.  Malec (104) et al conducted a double-blind placebo controlled study of buspirone on alcohol dependence and found significant improvements in several psychopathological measures in the medication group but no differences in measures of alcohol consumption between the two groups.  The authors of a meta analysis of the published controlled studies on the effects of buspirone in alcohol treatment concluded that buspirone is effective for the treatment of comorbid psychopathological symptoms, but does not appear to effect alcohol consumption per se (105).

The main criticisms of the buspirone studies to date are two fold: 1) Buspirone administered chronically is a relatively nonspecific compound to test the importance of the serotonergic system in the relapse process in alcohol dependent patients;  2) anxious alcoholics are more likely to represent late onset or Type I alcoholics, who have in biochemical and neuroendocrinological studies been found to demonstrate only subtle, if any, serotonergic abnormalities.  Thus, the very alcoholics who have shown the clearest signs of serotonergic dysfunction, who have early onset, antisocial traits, who are called Type II by Cloninger  (26), Type 2 by von Knorring (173), and Type B by Babor (5) and who rarely have clinically significant levels of anxiety, have been excluded from most of these studies.  Partial serotonin lA receptor agonists with a greater specificity than buspirone need to be tested for their efficacy as adjunctive pharmacotherapy in sobriety maintenance in early onset alcoholics with antisocial traits.

Serotonin 2 Antagonist

Ritanserin is a serotonin 2 receptor antagonist that was shown to decrease alcohol intake in some but not all animal studies.  In an open label clinical trial, newly abstinent alcohol dependent subjects treated with ritanserin reported less alcohol craving and lower rates of drinking relapses (114).  In a subsequent multi-center 12 week double blind, placebo-controlled study 423 subjects with alcohol dependence received either placebo, ritanserin 2.5 mg/day or 5 mg/day in combination with cognitive behavioral therapy (77).  No differences were found on any of the measures of alcohol drinking, craving or clinical outcomes.  All treatment groups improved as assessed by the Clinical Global Impression Scale and did not differ on compliance or reporting of side effects.  Ritanserin was associated with a dose-dependent prolongation of QTc interval on electrocardiogram (a risk factor for cardiac arrythmias), but did not lead to clinically significant difficulties.  The lack of efficacy for ritanserin could be due to:  1) that it might not have an effect on alcohol drinking in humans; 2) the low dose range tested; or 3) the low severity of alcohol dependence/high social functioning subjects in the study who benefited from the cognitive behavioral therapy that was provided.

Serotonin Reuptake Inhibitors

The results from studies using serotonin reuptake inhibitors in the treatment of patients with a confirmed diagnosis of alcohol dependence have been modest and variable (131).  In a study of alcohol dependent inpatients, open label fluoxetine briefly reduced alcohol consumption that did not persist into the second week of inpatient care (58).  In a randomized placebo controlled study, Kranzler et al (88) found that fluoxetine in combination with relapse prevention psychotherapy was not more effective than placebo in reducing alcohol consumption.  In another double-blind placebo controlled crossover design study comparing one week of citalopram to placebo, citalopram statistically significantly reduced the number of daily standard drinks and percent days drinking and significantly lowered the patient’s interest in, liking of and desire and craving for alcohol (124). 

Based upon clinical and experimental data, studies have examined whether certain types of alcoholics might benefit from SSRI treatment.  A small open-label study of severely depressed alcohol-dependent patients showed that fluoxetine reduced both drinking and depressive symptoms (27).  In another study, fluoxetine was found to be superior to placebo in reducing the alcohol consumption, depressive and obsessive-compulsive symptoms in alcohol-dependent patients with major depression during a twelve week trial (28). Gerra et al evaluated fluoxetine and acamprosate compared to placebo and found that subjects with a family history of alcoholism drank significantly less after fluoxetine treatment, but not after acamprosate (48).  In absence of a positive family history of alcoholism, patients’ did not significantly respond to fluoxetine.  However, acamprosate significantly reduced the number of alcoholic drinks consumed by those subjects without a family history of alcohol dependence.

In summary, there are several interesting leads in the animal and clinical literature that are suggestive of potential efficacy of serotonergic agents in reducing alcohol intake and perhaps preventing relapse in certain alcohol dependent patients.  Experiments in animals and in humans given serotonergic medications described above suggest that serotonergic systems play a role in alcohol reinforcement and may benefit certain subtypes of alcoholics, but not others.  Studies are currently underway to investigate the effect of administering different serotonergic agents to various alcoholic patient populations, with the goal to identify optimal treatments that can be provided based on patient characteristics.

GABA  receptor agonist/ Excitatory Amino Acid antagonist

The efficacy of oral acamprosate, calcium bisacetyl homotaurine, has been assessed in several well designed double-blind placebo-controlled trials.  The mechanisms of action for acamprosate appear to be quite complex with GABA agonist, excitatory amino acid antagonist and both inhibitory and excitatory NMDA properties depending upon the experimental conditions.  Oral acamprosate treatment for 3-12 months has been started 1-4 weeks following detoxification with various combinations of psychotherapy/behavioral therapies and treatments for comorbid conditions was started 1-4 weeks following detoxification with various combinations of psychotherapy/behavioral therapies and treatments for comorbid conditions.  The first study involved 85 alcoholics who were considered "hopeless" due to the severity of their illness and were, therefore, not considered to be good candidates for psychotherapy.  The dose of acamprosate was 25 mg/kg of body weight/24 hours and duration of treatment was 90 days.  Twenty of 33 patients on acamprosate and 12 of 37 on placebo (p<.02 by X2 test) did not relapse during the study (97).  The second multicenter study on sobriety maintenance with acamprosate enrolled 356 patients, 181 on acamprosate and 175 on placebo (98). The dose of acamprosate was significantly higher than in the prior study at 1.3 g/24 hours and the duration of treatment was similar, 90 days.  The main dependent variable was the liver enzyme, g-glutamyltransferase (gGT).  At the end of the study the gGT concentration was significantly higher in the placebo than in the gGT group suggesting that there was more alcohol use in the placebo group, but the number of subjects maintaining abstinence is not given in the article.  In subsequent trials, acamprosate has been shown to be more effective than placebo in a dose dependent fashion in preventing alcohol relapse assessed by cumulative, continuous and or absolute rate and duration for abstinence.  In a pooled analysis of data from 11 randomized placebo-controlled trials including 3338 patients with alcohol dependence, rates of abstinence with acamprosate were significantly superior to placebo at 6 and 12 month follow-up periods (6 month: 35 vs 25%; 12 month: 33 vs 21%) (177).  Concomitant administration of disulfiram enhanced the effectiveness of acamprosate in one study.  In a multicenter investigation of the efficacy of acamprosate for the treatment of alcohol dependence, 118 patients were randomly assigned to placebo or acamprosate and both groups were stratified for concomitant volunteer use of disulfiram (7).  Patients who received disulfiram were more severe on a number of measures for alcohol dependence with longer duration of alcoholism, significantly higher MAST scores, greater physiological dependence and greater baseline craving compared to those who did not take disulfiram.  The disulfiram users were more likely to complete the trial, and the patients who received both acamprosate and disulfiram had the best cumulative abstinence duration (CAD).  Trials comparing the efficacy of acamprosate with other active treatments such as naltrexone are warranted.

Dopamine precursors

L-DOPA + carbidopa combination in doses up to 100 mg of carbidopa and 800 mg of L-DOPA titrated according to patient's tolerance to side effects was compared to placebo in a one year follow up study on sobriety maintenance(47).  A total of 30 patients started the trial after an intensive six week inpatient treatment program.  All were encouraged to take advantage of AA meetings during the outpatient follow up period and they visited the outpatient clinic regularly. The study was discontinued after an interim analysis that showed no efficacy of L-DOPA and carbidopa over placebo.

Serotonin precursors

The above mentioned L-DOPA and carbidopa study also included a 5-hydroxytryptophan and carbidopa cell which consisted of 15 alcoholics.  According to the patients’ tolerance, the dose was titrated up to 100 mg of carbidopa and 400 mg of 5-hydroxytryptophan.  The combination was no more effective than placebo.

Dopamine receptor agonists

In the past, apomorphine has been used as an emetic to produce aversion to alcohol in alcoholics.  No controlled studies of apomorphine's efficacy used in aversion treatment are available.  In relatively low doses which do not provoke nausea or vomiting, apomorphine has in uncontrolled studies been reported to enhance sobriety maintenance.  In a two week controlled clinical study of 40 patients, started immediately after treatment for alcohol withdrawal, apomorphine 36 or 108 mg/24 hours was no better than placebo in maintaining sobriety (76).  Thus, there appears to be no evidence based on controlled studies that shows apomorphine to be efficacious in sobriety maintenance.

Bromocriptine, a relatively nonspecific dopamine2 receptor agonist, which undergoes  extensive first pass metabolism in humans, has been examined in studies to enhance sobriety maintenance. Two studies have yielded conflicting results.  Borg et al. (11) treated 50 alcoholics using a parallel group design for six months with 2.5 mg bromocriptine three times a day.  During months four through six, the dose was increased to 5 mg three times a day.  Borg reported significant efficacy for bromocriptine.  The bromocriptine group reported reduced craving, fewer depressive reactions, and only one of 19 patients reported drinking at the end of the trial.  The study had a remarkably low dropout rate in both the placebo and bromocriptine groups.  Dongier et al. (37) administered 2.5 mg bromocriptine three times a day for eight weeks to 84 alcoholics of whom 38 were available for analysis at the end of the treatment phase. Twenty patients could be investigated after a further eight week medication-free follow up.  In this study, which had a more typical dropout rate than the Borg et al. study, bromocriptine did not improve relapse rate compared to placebo.  However, various psychiatric symptoms showed significant improvement on bromocriptine compared to placebo.  These results were not confirmed in a double-blind placebo controlled clinical trial comparing bromocriptine and nortriptyline in male alcoholics with comorbid psychiatric disorders where bromocriptine did not attenuate drinking or psychiatric symptoms (135).  Based upon earlier results, a randomized, double-blind, international, multicenter study assessed the effects of a long-acting parenteral form of bromocriptine.  366 alcoholics were randomized to three groups and received six monthly injections of  placebo, 25 mg or 50 mg of bromocriptine.   No differences were seen between treatment groups in rates of abstinence or relapse (126).  Possible reasons for the lack of findings include the bromocriptine formulation (parenteral vs oral), dose or study design issues (treatment withdrawal, non-pharmacological interventions that were not standardized across the trial).

Dopamine receptor antagonists

Tiapride is a substituted benzamide which acts primarily as a dopamine2 receptor antagonist. Shaw et al. reported a double blind, six month, comparison of 100 mg. tiapride three times a day and placebo in 32 alcoholic dependent patients.  All patients had high anxiety and depression ratings at the end of their detoxification. Twenty patients completed the study. Tiapride halved the amount of ethanol consumed by the patients and roughly doubled the number of abstinent days compared to placebo.  Both findings were statistically significant.  Depression and anxiety ratings were also significantly reduced in the tiapride group (148).  In a second  six month double-blind randomized placebo controlled trial comparing 100 mg tiapride three times daily to placebo in 100 alcohol-dependent patients, tiapride significantly improved abstinence, self-esteem and satisfaction with life situations and reduced alcohol consumption, use of health service resources and levels of neuroticism (149).  Although the benefits for the treatment for alcohol dependence remain promising, the potential risk of tardive dyskinesia at the doses used and small risk for neuroleptic malignant syndrome merits careful evaluation and necessitates medical supervision (130).


Gamma-hydroxybutyrate is an endogenous substance with sedative properties whose exact mechanism of action is not known.  It has been demonstrated in low doses, to increase, and in high doses, to reduce dopamine turnover in the central nervous system.  Thus, it has been postulated to mimic alcohol's reinforcing effects.  In a double blind, placebo controlled, parallel group study with a 90 day treatment period, 36 patients received 50 mg/kg of body weight/24 hours of g-hydroxybutyrate and 35 patients received placebo (8).  During the third month, but not earlier, the number of patients abstaining or practicing "controlled drinking" was significantly (p<.001) higher in the group receiving g-hydroxybutyrate.  In an “open” multicenter study, 179 alcohol dependent subjects received oral 50 mg/kg three times daily for 24 weeks.  Although 70 subjects (39.1%) dropped out, g-hydroxybutyrate led to complete abstinence in 78% of the patients with a significant reduction of craving on the Alcohol Craving Scale (2).  At follow-up after drug discontinuation, 43 subjects were abstinent at 6 months and 30 were abstinent for 1 year.  However, 11 (10.1%) of the subjects craved g-hydroxybutyrate and escalated their doses to 6-7 times the recommended levels to obtain anxiolytic and hypnotic effects (2,3).  A subsequent case series documented the development of g-hydroxybutyrate abuse and physical dependence, and highlighted the therapeutic limitations of g-hydroxybutyrate (46). Although the results from the open label studies are promising, the potential for abuse and dependence remain important when considering a large double blind, placebo controlled trial needed to confirm efficacy of g-hydroxybutyrate for the treatment of alcohol dependence.

Opioid antagonists

As reviewed above, there are several animal models of alcohol drinking that suggest a role for endogenous opioids in the reinforcement produced by alcohol.  It was these reports from animal studies that caused Volpicelli and colleagues (171) to conduct a double-blind trial of naltrexone versus placebo in chronic alcoholics.  The patients in the study were applying for treatment at the Philadelphia Veterans Affairs Medical Center.  The 70 male subjects were on average 43 years old with a 20 year history of heavy drinking and all met at least five of the nine DSM-III-R criteria for alcohol dependence including physical signs of withdrawal sufficient to require medication.  After completion of detoxification, the patients began an outpatient rehabilitation program that consisted of one month of daily day hospital participation tapering to weekly counseling sessions during months two and three.  At the time of entering the rehabilitation program, the patients were randomly assigned to 50 mg naltrexone daily or to placebo.  During the three months of the study, the naltrexone-treated patients reported significantly less craving for alcohol and they reported significantly fewer days of alcohol use.  The overall survival curve during the three-month study is shown in Figure 2.  As is typical of chronic alcoholics in a rehabilitation program, 54% of the placebo patients met criteria for relapse while only 23% of those in the naltrexone group met these criteria.  Among the patients randomized to placebo, having a “slip” and drinking any alcohol almost invariably (95%) led to a relapse.  In contrast, those receiving naltrexone who used any alcohol had only a 23% relapse rate.  Two naltrexone patients who dropped out due to nausea were included in the outcome analysis.  There was no evidence that naltrexone impaired liver function.  To the contrary, there were more improvements in liver enzyme data in the naltrexone group probably reflecting their greater abstinence, but the differences did not reach statistical significance. 

The positive results in this single study of a new treatment for alcoholism have been confirmed in a second trial.  O’Malley and colleagues (129) heard a preliminary report of the Philadelphia group and conducted a similar study among 97 alcohol dependent subjects in New Haven.  Instead of a day hospital, the rehabilitation program consisted of coping skills/relapse prevention or supportive psychotherapy.   The results of this independent study involving a predominantly white male population were quite similar to those of the Philadelphia group.  Naltrexone was found to be clearly superior to placebo irrespective of the type of psychosocial intervention to which the patients were assigned.  Those randomly assigned to naltrexone drank on half as many days and consumed one third the number of standard drinks during the trial as did subjects who received placebo.  Craving for alcohol was significantly decreased only in those subjects who completed the study.  Relapse rates were significantly lower in the naltrexone-treated patients.  As in the Volpicelli et al study, relapse rate differences were particularly high among patients who drank some alcohol.  Among the patients who had at least one drink, those randomized to naltrexone and coping skills therapy had one fourth the risk of relapse compared with subjects taking placebo who received coping skills treatment.  Medication compliance was measured by the addition of riboflavin to the study medication and the testing of urines weekly with UV light.  This method indicated high compliance for both groups, but significantly higher for the naltrexone treated patients.  As in the previous study, there was no evidence of naltrexone induced-hepatotoxicity.  Aspartate aminotransferase levels were significantly lower in the naltrexone-treated patients at endpoint and a similar trend was noted for alanine aminotransferase.

Patients in the Volpicelli et al study who had a “lapse” and drank some alcohol were asked about the effects of the alcohol.  Those receiving placebo reported that the effects were no different from those experienced prior to the study.  Those receiving naltrexone reported significantly less euphoria from their drinking than they expected based on prior experience (172). This finding which suggests an effect of naltrexone on the immediate subjective effects of alcohol in a clinical setting supports the experimental findings that naltrexone attenuated the “high” in acute alcohol administration studies in healthy normal volunteers (83,163).  These data from clinical and laboratory studies give a clue as to a mechanism of action by which naltrexone might decrease relapse rates in alcoholics.

These two controlled studies comparing naltrexone to placebo in middle-aged alcoholics involved in outpatient rehabilitation programs had strikingly similar results.  More studies are required, but the available clinical evidence is consistent with evidence from rodent and non-human primate models.  It suggests that naltrexone by blocking µ and d opiate receptors is able to reduce the reinforcement produced by alcohol drinking.  Both studies had total abstinence as the stated goal of treatment and there was no intention of using naltrexone to teach controlled drinking.  Naltrexone clearly does not prevent alcoholics from relapsing.  Patients can stop the medication at any time with no withdrawal symptoms and resume drinking.  Naltrexone may, however, interact with the effects of the rehabilitation program, theoretically by reducing the reward produced by alcohol.  Naltrexone may help the patient to remain in treatment longer and may facilitate the benefits of long term behavior change.

published 2000