Considerable evidence has emerged suggesting that the dopamine system plays a central role in the biology of alcoholism. Mesolimbic dopamine A10 neurons are activated by alcohol, resulting in a release of the neurotransmitter in the nucleus accumbens and mediating positive reinforcement and reward.11 It is postulated that repeated alcohol use sensitizes the system, so that behavioral stimuli associated with alcohol also cause the release of dopamine and facilitate additional alcohol use.12 This sensitization may account for the craving and preoccupation with alcohol that are the hallmarks of alcohol dependence.

The endogenous opioid system seems to play a modulatory role on the dopaminergic system, whereby activation of opiate receptors stimulates the release of dopamine in the brain. Alcohol consumption increases the release of endorphins (which are endogenous opioid peptides) in the brain, thus indirectly activating the dopaminergic reinforcement/reward system.13 It has been postulated that individual differences in the sensitivity of endogenous opioid systems may underlie individual differences in the intensity of alcohol craving and the risk of becoming alcohol dependent.

The facilitation of inhibitory GABAergic and the inhibition of excitatory glutamatergic neurotransmission are important targets for the acute effects of alcohol. Potentiation of GABAergic inhibition is widely accepted as the underlying cause of the acute sedative effects of alcohol. Long-term adaptive changes to the sedative effects of alcohol in these two neurotransmitter systems are thought to underlie the development of alcohol dependence. After chronic exposure to alcohol, there is a compensatory up-regulation of the glutamatergic system (and down-regulation of the GABA system) in an attempt to balance alcohol's inhibitory action. The result is an increased tolerance for alcohol.14 When alcohol is abruptly withdrawn, however, a state of hyper-excitability emerges. This is perceived by the patient as a disagreeable state of arousal, anxiety and sleeplessness and is the core of the negative affective state which the alcoholic patient will drink to relieve. These plastic changes in the brain, brought about by change in protein synthesis, are only slowly reversible. This may explain the persistence of negative craving during alcohol withdrawal and why stable abstinence after acute detoxification is so difficult to achieve. Antiglutamamatergic agents, such as NMDA antagonists and anticonvulsant agents, have been proposed to reduce the motivation for drinking by suppressing symptoms of alcohol withdrawal. Recent data suggest that NMDA antagonists may have other beneficial effects in alcohol dependent patients, such as substituting for deficits in negative feedback signals or reducing the development of tolerance/sensitization to alcohol.15,16

The first agent to be approved for treatment of alcohol dependence was disulfiram. This substance was serendipitously discovered to be an agent causing alcohol aversion in Ohio rubber workers in 1939. The major metabolic pathway for alcohol metabolism is a two-step enzymatic process (ethanol → acetaldehyde → acetic acid). Disulfiram is an irreversible inhibitor that blocks the second stage of alcohol metabolism, causing an accumulation of toxic intermediate acetaldehyde, which results in hypotension, flushing, nausea, and vomiting. The objective of disulfiram treatment is thus to create an aversion to alcohol, rather than to modulate its neurochemical effects. Although many studies have been performed with disufiram, controlled clinical trials demonstrated inconsistent findings for alcohol drinking outcomes between disulfiram and placebo, and have failed to clearly establish the therapeutic benefit of this treatment in enhancing abstinence.17 However, it is difficult to conclude the efficacy of the treatment through classical double-blind, placebo-controlled trials, since it is the psychological deterrent effect of the drug rather than its biological effect that is useful.18 Fuller et al.19 observed no significant differences in abstinence rates or in the time to first relapse among groups taking placebo, 1 mg/day disulfiram (an inactive dose) or 250 mg/day disulfiram (the standard dose). The patients receiving 250 mg disulfiram, however, had fewer drinking days once they relapsed than did the other two groups.

The disulfiram dosage is usually 250 mg per day with a maximum of 500 mg per day. While normal results of alcohol consumption after taking disulfiram are palpitations, flushing, nausea, vomiting and headaches, more severe reactions could include myocardial infarction, congestive heart failure, respiratory depression, and death. The use of disulfiram appears to be most useful in adherent patients, in special high-risk patients, and when administration is supervised.

Naltrexone and acamprosate are at the forefront of currently available pharmacological options and they share many important features20 (Table 2). Most of the concerns in using medications for relapse prevention are probably related to the possible intrinsic dependence potential. However, there is no evidence that naltrexone or acamprosate have developed either tolerance or withdrawal symptoms, including rebound drinking when treatment is ceased, and nor do they have any overt psychoactive effects on the central nervous system. In addition, both drugs do not have pharmacological or pharmacokinetic interaction with alcohol and so the serious side effects associated with disulfiram are not troublesome when using naltrexone or acamprosate.

The important difference between naltrexone and acamprosate are mainly attributed to their mechanisms of action. Naltrexone is an antagonist at the opioid receptors, which are known to mediate the rewarding effects of alcohol, and is thus thought to reduce the desire or craving for rewarding. Although less well defined, acamprosate normalizes the dysregulation of NMDA-mediated glutamatergic excitation that occurs in alcohol withdrawal and early abstinence. This effect probably attenuates the desire or craving for reduce of tension. These different mechanisms of actions may endow the two drugs with efficacy for different components of drinking.

The USA FDA's approval of naltrexone for the treatment of alcohol dependence was based mainly on two small, double-blind, placebo-controlled trials demonstrating a reduced rate of relapse to heavy drinking, reduced craving, and less frequent drinking in naltrexone-treated patients.22,23 During the following years, several more trials have followed and three meta-analyses have concluded that naltrexone is efficacious in the treatment of alcohol dependence.24-26 The most consistent finding obtained with naltrexone is an increased time to first relapse (typically defined as more than 5 drinks/day in males, 4 drinks/day in females). The decrease in relapse rate, however, has not been observed in all studies. Such differences in results are seen even in the largest trial.27 Several factors may explain the discrepancies in results of the different clinical trials of naltrexone. The animal models of relapse indicate that naltrexone blocks cue-induced relapse but is less effective against stress-induced relapse.28 The effect of naltrexone on relapse may, to some extent, be dependent on associated psychotherapy, since naltrexone was found to be more effective in patients receiving training in coping skills than in those receiving supportive therapy alone.29 Compliance may be a limiting factor in naltrexone treatment.30-32 Monti et al.32 demonstrated a significant treatment outcome only when non-compliant subjects were excluded from the analysis. A large multi-site, double blind, placebo-controlled trial with an injectable, sustained-release formation, (a strategy to improve compliance) demonstrated that relapse to heavy drinking decreased in patients receiving depot preparation compared to placebo.33

A conceptual framework for integrating the clinical data on naltrexone has been proposed by Sinclair,34 who suggested that naltrexone is useful for preventing relapse rather than at maintaining absolute abstinence. Thus, the contingency of drinking alcohol and taking naltrexone is important in bringing to light treatment effects. In a recent systematic meta-analysis of 24 placebo-controlled trials, including a total of 2861 patients, short-term naltrexone therapy significantly decreased relapse rate (relative risk 0.64), but did not enhance absolute abstinence (relative risk 0.91).35

For the first 90 days of abstinence, when the risk of relapse is greatest, the recommended dosage of naltrexone is a single dose of 50 mg per day but doses of 25 mg to 100 mg daily are sometimes used. The most common side effects are nausea (10%), headache (7%), anxiety (2%) and sedation (2%).36 Naltrexone has been shown to have dose-related hepatotoxicity, although generally this occurs at doses of 300 mg per day, higher than those recommended for treatment of alcohol dependence. The drug is contraindicated in patients with hepatitis or liver failure, and a monthly check of hepatic transaminase levels is recommended, for the first three months and every three months thereafter.

Acamprosate was investigated in nearly 20 controlled, published trials with about 4000 patients and these studies have produced consistent results showing that acamprosate treatment is superior to placebos in maintaining abstinence.37 In all but three clinically controlled published studies, the proportion of acamprosate-treated patients abstaining at the end of the study was twice as high as patients receiving placebos. In addition, two studies38,39 evaluated long-term abstinence for 1-ear after the end of the treatment period, and both showed that treatment effects were maintained. A systematic metaanalysis37 in which clinical data from 17 trials were reanalyzed concluded that the treatment effect could increase with time.

Since most of the trials were undertaken in Europe, it was considered important to evaluate the efficacy and safety of acamprosate in different ethnic groups, given the emerging role of genetic and cultural issues in drug treatment. In a study by Namkoong et al.,40 however, acamprosate did not show any treatment benefits when compared to the placebo. This negative finding might be explained by the sample characteristics (i.e., a more severe form of alcohol dependence, a lower level of social support, a short interval between the last drink and the first medication), the dosage issues of acamprosate, the short study period (8 weeks), and the variable concomitant psychosocial treatment.

Acamprosate is available in 333-mg enteric coated tablets. Dosing is determined by weight (≥ 60 kg: 1998 mg, < 60 kg: 1332 mg). It is not metabolized but is eliminated by renal excretion, and should therefore be given cautiously with renal impairment. Acamprosate is well tolerated with limited side effects, most commonly transient diarrhea (10%) and headache (20%). Like naltrexone and disulfiram, acamprosate is FDA pregnancy category C, i.e., there have been adverse effects on the fetus in animal studies but no human trials have been performed.

Combining naltrexone and acamprosate in the treatment of alcohol dependence is an attractive concept for several reasons. Since naltrexone and acamprosate have different mechanisms of action and different target neurotransmitter systems, presumably, they affect different aspects of alcohol use behavior. (Naltrexone decreases alcohol consumption and acamprosate stabilizes abstinence.) Pharmacokinetic and behavioral assessments of combining naltrexone and acamprosate have found the combination to be safe.

Kiefer et al.41 performed a randomized, double-blind, placebo-controlled, clinical trial of 160 alcohol-dependent patients and assessed the efficacy of naltrexone and acamprosate, as monotherapy and in combination. It was demonstrated that the proportion of patients remaining absolutely abstinent at the end of the 12-week treatment period was around twice as high in the combination therapy group than in the monotherapy group (placebo 25%, naltrexone alone 65%, acamprosate alone 50%, combination therapy 73%). Even though further relapse occurred during the follow-up period, the relative treatment benefits between the three treatment groups and the placebo group was maintained at the end of the 3-month open label phase. There was, however, no significant difference between the three treatment groups (placebo 20%, naltrexone alone 47%, acamprosate alone 46%, combination therapy 66%).42 Although combination therapy was generally well tolerated, the incidence of diarrhea (13.8%) and nausea (5.6%) was significantly greater than in the monotherapy groups, perhaps due to a pharmacokinetic interaction.

There are several possible explanations for the superior efficacy of the combination treatment.43 First, there may be subgroups that respond selectively to naltrexone or acamprosate and thus the added benefit of combination therapy would be merely explained by the recruitment of additional responder patients. A hypothesis of such patient subgroups may be that 'reward' craving drinkers would respond better to naltrexone and 'relief' craving drinkers would respond better to acamprosate (Fig. 1). Attempts to find variables that might predict the response of medications will be discussed later. Second, the combination produces a synergic anticraving effect as the two drugs interfere with distinct biological aspects of the craving process. Third, pharmacokinetic interaction might underlie the observed treatment benefits, whereby bioavailability might be enhanced by co-administration of the other drug.

The ongoing COMBINE study44 plans to recruit 1,375 subjects at 11 sites to examine treatment interactions between naltrexone, acamprosate and two behavioral interventions (medical management and combined behavioral interventions). The data from COMBINE will answer important questions regarding the effectiveness of naltrexone and acamprosate both alone and in combination, as well as that of psychosocial treatment.

Since not all patients seem to benefit from the different anticraving drugs, the use of matching procedures might be important in improving the effectiveness of treatment with anticraving compounds.45 The type of craving could possibly be an important predictor or matching variable with anticraving compounds. In general, craving mainly refers to the strong desire or urge to experience the effect of a previously experienced psychoactive substance.46 Despite the simplicity of this definition, a wide variety of craving concepts are used in the research and clinical field. Verheul et al.47 have proposed a novel three pathway model of craving in alcoholics. This model suggests that craving is likely to result from distinct psychobiological mechanisms and that the efficacy of different anti-craving compounds is associated with individual differences in craving: (1) The reward pathway suggests that craving or desire for the rewarding, stimulating and/or enhancing effects of alcohol might result from either dopaminergic/opioidergic dysregulation or a personality style characterized by reward seeking and/or hedonism. (2) The relief pathway suggests that craving or desire for the reduction of tension, arousal or withdrawal might result from either GABAergic/glutamatergic dysregulation or a personality style characterized by stress reactivity, anxiety sensitivity, and/or hyperarousability. (3) The obsessive pathway can be defined as a lack of control over intrusive thoughts about drinking resulting in impaired functioning. This pathway of craving might result from serotonin deficiency or a personality style characterized by low constraint or disinhibition.

Some studies found that patients with high levels of alcohol craving are most likely to benefit from naltrexone treatment.48,49 No distinction, however, was made between reward, relief and obsessive craving. Palfai et al.50 conducted a study among hazardous drinkers and suggested that naltrexone may be particularly effective in reducing the cue-elicited positive reinforcement of alcohol for those with high positive alcohol outcome expectancies. Positive outcome expectancies also moderated the effects of naltrexone on subjective reports of stimulation following drinking.

Lesch et al.51 conducted a long-term, prospective study to assess the efficacy of acamprosate based on the four subtypes of alcohol dependent-patients. Consistent with their matching hypothesis, based on the three pathway model of Verheul et al.,48 acamprosate differentially reduced alcohol intake in those patients who use alcohol to counteract withdrawal symptoms (Type I), and in patients who use alcohol as a conflict-solving and anxiety reducing agent (Type II), but not in patients who ingest alcohol to self-medicate affective disorders (Type III) or patients with a history of cerebral impairment that precedes the development of alcohol dependence (Type IV). A recent study including the pooled data of 1,485 alcohol-dependent patients from seven randomized, controlled trials comparing acamprosate and a placebo, directly tested the matching hypothesis that acamprosate would be more effective in patients with high physical dependence at baseline, negative family history of alcoholism, late age of onset, serious anxiety symptomatology at baseline, severe craving at baseline, and female gender.52 In contrast to the expectations, the authors found that none of these theoretically relevant clinical matching variables predicted the treatment effectiveness of acamprosate. Similarly, another pooled analysis of 11 trials by Sass et al.53 did not find any demographical, psychopathological or biological predictors for acamprosate efficacy.

It should be noted that the three-pathway model by Verheul et al.48 and related matching hypotheses might not cover all the possible mechanisms of action of anti-craving medications. For example, a recent study showed that naltrexone not only includes an opioid-receptor blockade, but that it indirectly increased hypothalamic pituitary adrenocortical (HPA) activity, resulting in higher adrenocorticotropic hormone (ACTH) and cortisol levels, which may in turn be partially responsible for the effect of naltrexone on drinking and craving.54

Investigating the role of genetics in predicting treatment outcomes is one promising area of research since a genetic basis of for alcoholism (heritability rate, 50-60%) is well established.55 Monterossso et al.49 explored the predictors of naltrexone response in a double-blind, placebo-controlled trial and demonstrated significant interactions between treatment condition and family history of alcohol problems. Some studies have investigated individual differences with respect to sensitivity in the hypothalamic pituitary adrenal axis to naltrexone administration. King et al.56 examined the neuroendocrine and mood responses to oral naltrexone as a function of the biological family history of alcoholism. The results demonstrated the potential biological bases of altered opioidergic sensitivity in those persons with an assumed greater inherited vulnerability for future alcoholism. Meanwhile, studies of family history as a predictor of treatment response have led to speculation that naltrexone may function differently in genetically predisposed individuals.

Naltrexone has a high affinity for the µ-opioid receptor, which is hypothesized to be the principal site of action of the drug. It has been hypothesized that sequence variation in the gene encoding the µ-receptor may result in a receptor with altered expression, structure, or function, and as a consequence, may increase or decrease an individual's susceptibility to substance dependence.57 Oslin et al.58 examined the association between two specific polymorphisms of the gene encoding the µ-opioid receptor and treatment outcomes, measured over 12 weeks, in alcohol dependent patients who were prescribed naltrexone or a placebo. Patients with one or two copies of the Asp40 allele treated with naltrexone had significantly lower rates of relapse (26.1% versus 47.9%; p = 0.0044) and took a longer time to return to heavy drinking than those homozygous for the Asn40 allele. There were no differences in relapse rates or abstinence rates between the two genotype groups among those assigned to placebo. Meanwhile, Kim et al.59 reported that the allele frequency of the Asp40 allele was 39.7% in the Korean alcohol dependent group, which is consistent with previous studies demonstrating a higher Asp40 allele frequency in the Asian population. Within the alcohol dependent group, the Asp40 allele was associated with more drinking days. The finding of a genotype that predicts success with naltrexone, if replicated, will provide a pharmacogenetic tool to enhance the matching of patients to treatment and encourage the search for additional functional polymorphisms.

During the last two decades, a number of drugs acting on serotonergic neurotransmission have been studied in alcohol dependence, since serotonin is widely implicated in a variety of consummatory behaviors and impulsivity. These agents are either selective serotonin reuptake inhibitors or receptor agonist/antagonists. Most of this work, however, has used small samples with relatively short treatment periods. Selective serotonin reuptake inhibitors, despite their effectiveness in animals, have shown inconsistent or disappointing results in humans60-65 and so the usefulness is still controversial.17 Meanwhile, no evidence of clinical efficacy in alcohol-dependent has been obtained with ritanserin (5-HT₂ anatagonist).66,67 In addition, a meta-analysis of studies performed with bupropion (5-HT₁ partial agonist) concluded that any efficacy of bupropion was secondary to an anxiolytic effect, rather than on drinking perse.68

Of the numerous serotonergic drugs which have been suggested as pharmacotherapies for alcohol dependence treatment, ondansetron, a 5-HT₃ antagonist that is FDA-approved as an antiemetic, appears to be the most promising.69 The 5-HT₃ receptor is involved in the expression of alcohol's rewarding effects. Behavioral pharmacological studies show that many of the reward effects of alcohol are mediated by interactions between DA and the 5-HT₃ receptor in the mid-brain and cortex.70,71 5-HT₃ receptors are densely distributed in the terminals of mesocorticolimbic DA-containing neurons where they regulate DA release in these brain regions. Following a previous clinical trial,72 Johnson et al.73 evaluated ondansetron as a treatment for alcohol dependence in a 12-week, double-blind, placebo-controlled trial of 321 patients. The early-onset, alcohol dependent group treated with ondansetron (particularly 4 µg/kg b.i.d.) reported fewer drinks per day and fewer drinks per drinking days, while the late-onset group treated with ondansetron did not differ from those treated with placebo. It is interesting that while serotonine reuptake inhibitors have little effect on drinking among early-onset alcoholics, ondansetron, with functionally opposite effects in the serotonergic system, is effective for the early-onset subtype. Sufficient evidence exists that early-onset alcoholics are more prone to serotonergic dysfunction than late-onset alcoholics.74,75

Mood stabilizers and anticonvulsants decrease alcohol consumption in experimental animals. Clinical trials, however, have not provided clear evidence of the efficacy of treatment for alcohol dependence. The controlled trials of lithium did not demonstrate efficacy in either non-depressed or depressed alcohol-dependent patients and Garbutt et al.17 concluded that lithium lacks efficacy in the treatment of primary alcohol dependence. More promising are results with non-benzodiazepine anticonvulsants such as carbamazepinem, valproate, gabapentin, vigabatrin and topiramate.76

Topiramate, although only approved by the FDA for seizure disorders, was evaluated in the treatment of alcohol dependence because of its effects on GABAergic and glutamatergic systems. It has been shown to augment GABA function and inhibit specific glutamatergic pathways at the AMPA/kinate receptors. This may decrease the cortical expression of alcohol reward by decreasing the midbrain dopamine function.77 In a 12-week double-blind, placebo-controlled trial of 150 alcohol dependent patients, topiramate was more effective than placebo in initiating abstinence and in reducing self-reported drinks per day, drinks per drinking days, and heavy drinking days. Compared with a placebo, it also significantly reduced craving as assessed with Obsessive Compulsive Drinking Scale. Topiramate was equally effective in both early-onset and late-onset alcohol dependence.78 The study used an escalating dose of 25 to 300 mg per day. Hypersensitivity to the drug was the only known contraindication. No serious adverse events occurred. Paresthesia was the most common adverse effect, which includes dizziness, somnolence, diplopia and nausea.

Important advances have been made in the development of pharmacotherapy in alcohol dependence and it is experiencing a major shift in direction. There is clear evidence of the efficacy and safety of natrexone and acamprosate, and still more novel medications are under investigation. In addition, important questions remain regarding the optimal dose and duration of treatment, the role of combinations of medications, and the treatment subtypes of alcohol dependent patients.