Medications as adjuncts to psychosocial treatments for adolescent alcohol use disorders hold the promise of improved efficacy over psychosocial treatments alone. Specific components should be included in the design and implementation of these medication studies. Included should be assessment of the developmental risks of the chosen medication, consideration of short-term effects on the clinical disorder, factors affecting compliance and retention, age-specific pharmacokinetics, and systematic safety monitoring. A risk–benefit analysis should be conducted on the potential benefits of the medication to decrease alcohol use versus the potential long-term effects of medication use on brain development. To select clinically meaningful subtypes of adolescents with alcohol use disorders for medication trials, classification systems should be derived from multi-factorial models of complex neurodevelopmental disorders. Multi-factorial models will be required to select samples wherein specific gene–gene and gene–environment interactions predict medication treatment response. In samples of adolescents with alcohol use disorders, clinically meaningful subtypes are likely to have differential medication treatment response as a function of age of onset, family history of disorder, and comorbid psychopathology. Findings from preclinical and treatment studies in adults, along with pilot treatment findings in adolescents, suggest that particular serotonergic agents, opioid antagonists, and agents that modulate excitatory amino acids and GABAergic transmission might be effective. Future medication trials for adolescents with alcohol use disorders should use specific combinations of medications, based on specific hypotheses involving key neurotransmitter systems that putatively modulate treatment response. Combinations of medications may have additive effects on particular neurotransmitter systems or synergistic effects across two or more neurotransmitter systems, to further decrease alcohol consumption when compared with single-agent treatment. Is alcoholism a disorder.
(Received 15 November 2003, in revised form 13 January 2004, accepted 14 January 2004)
The prevalence of adolescent alcohol use and related problems is a major public health concern worldwide. The European School Survey Project on Alcohol and other Drugs surveyed more than 90 000 adolescents 15–16 years old from a total of 30 countries ( Hibbell et al., 2001 ) and found that the proportion of youth who had drunk to the point of intoxication at least 10 times in the past 12 months ranged from greater than 25% in Ireland, the UK, Finland and Denmark to less than 5% in the Former Yugoslav Republic of Macedonia, Greece, Malta, Portugal, France, Romania, Italy and Cyprus. Between 1995 and 1999, alcohol consumption had remained stable or risen in most countries, except in Cyprus and Italy, where drinking levels decreased ( Plant and Miller, 2001 ).
While alcohol and illicit drug use by teenagers in the US has declined slightly from the early 1990s, alcohol and other drug use in this population remains common. ‘Monitoring the Future’ findings from 2002 showed statistically significant decreases in any alcohol use in the past year and in the past 30 days for grades 8 and 10 ( NIDA News Release, 2002 ). For grades 8, 10, and 12, those adolescents who stated they were drunk at least once in the past 30 days were 7, 18 and 30%, respectively, in 2002. Findings from the ‘Youth Risk Behavior Surveillance System’ for 2001 have shown that for students in grades 9 through 12, almost 30% have consumed at least five drinks on at least one occasion in the 30 days prior to the survey ( Grunbaum et al., 2002 ). Taken together, these epidemiologic findings suggest that a minority of students initiate drinking in early adolescence, and these rates of initiation of drinking rise dramatically by middle adolescence ( Clark et al., 2002 ).
In contrast to slight declines in rates of drug use in the US, drug-related problems remain high. The ‘2001 National Household Survey on Drug Abuse’ tracked substance misuse problems in 10.1 million (28.5%) adolescents aged 12–20 years ( Substance Abuse and Mental Health Services Administration (SAMHSA), 2003 ). Included in the analysis was current alcohol misuse with associated problems. Of note, nearly 6.8 million (19%) were binge drinkers, and 2.1 million were heavy drinkers. The rate of driving under the influence of alcohol increased from 10.0 to 11.1% between 2000 and 2001. High binge drinking rates among adolescents may, at least partially, account for the growing rates of driving while intoxicated.
Despite the recent small decreases in prevalence, adolescent alcohol use remains a major public health concern due to the illegality of drinking in the adolescent years and the socioeconomic, psychological, and physical consequences that occur. Binge drinking to the point of intoxication is particularly associated with road accidents, suicides, homicides and violence ( Grunbaum et al., 2002 ). Heavy alcohol use also increases the probability of contracting sexually transmitted diseases (including HIV infection), pregnancy and performing poorly at school ( Grunbaum et al., 2002 ). Regional studies in the US indicate that up to 10% of adolescents are in need of treatment for alcohol use disorders ( Kaminer, 2001 ).
Prospective data show that when left untreated, adolescent alcohol use disorders often lead to adult alcohol dependence, increased frequency of other drug use and increased risk of adult drug dependence. Thus, the effective treatment of adolescent alcohol use disorders may prevent progression to adult alcohol dependence and other drug misuse. Left untreated, resistance to treatment may increase due to the worsening pathogenesis of alcohol use disorders. The pathogenesis of adolescent alcohol use disorders is likely to involve multiple neurotransmitter systems, including serotonergic, dopaminergic and opioid systems in the brain. These findings underscore the need for more effective treatments for adolescents who misuse alcohol.
Most treatments for adolescent alcohol use disorders have been limited to psychosocial interventions. Psychosocial treatments that have shown promise include family therapies, motivational interviewing, community reinforcement, 12-step approaches, cognitive-behavioural therapy and contingency management reinforcement ( Deas and Thomas, 2001 ). While results are especially promising for cognitive behavioural therapy and family-based therapies, most of the relevant studies have not used similar validated outcome measures ( Deas and Thomas, 2001 ). Therefore, the direct comparison of efficacy between studies using different psychosocial treatments has not been possible.
Not all adolescents with alcohol use disorders benefit from psychosocial treatment. Joanning et al. (1992) showed, in a study of 132 substance-misusing adolescents who received family system therapy, adolescent group therapy or family drug education, that the percentage of abstainers was 54% for family system therapy, 28% for adolescent group therapy and 16% for family drug education. Even for family system therapy, about half of subjects continued to use significant levels of illicit drugs. Liddle et al. (2001), in a study of 182 clinically referred adolescents with marijuana and alcohol misuse, showed that multi-dimensional family therapy was superior to adolescent group therapy, and multi-family educational intervention, at study end and at 1-year follow-up. However, even within the multi-dimensional family therapy group, subjects continued to use marijuana regularly at study end and at 1-year follow-up. At study termination, only 42% of adolescents who received multi-dimensional family therapy reported clinically significant reductions in drug use, compared with 25% of those adolescents who had group therapy and 32% of recipients who had multi-dimensional educational intervention. At 1-year follow-up, 45% of adolescents who received multi-dimensional family therapy, 32% of youth who received adolescent group therapy and 26% of teens assigned to multi-family educational intervention reported academic failure, moderate to heavy drug use and related behaviour problems.
Even among those adolescents who benefit from psychosocial treatments, relapse rates remain high. Brown et al. (2000) reported that for 157 adolescents recruited from inpatient alcohol and drug treatment followed 6 and 12 months after discharge, relapse rates were 79% for one or more substances. Further, alcohol triggered 46% of the initial episodes of relapse. Spear et al. (1999) reported on 113 adolescents evaluated 12 months after initial treatment. Of those adolescents with alcohol dependence, 45.9% relapsed to pre-treatment consumption levels within 12 months. Little is known about the impact of aftercare interventions on relapse rates.
RATIONALES FOR MEDICATIONS AS ADJUNCTS TO PSYCHOSOCIAL TREATMENTS
In contrast to research on treatment for adult alcohol dependence, few medication trials have been conducted to determine the efficacy of putative therapeutic agents for the treatment of adolescent alcohol use disorders. Bias and ethical concerns exist against the use of medications to treat adolescent alcohol use disorders. Nevertheless, the potential benefits of medication trials to treat adolescent alcohol use disorders should be weighed against the specific controversies and challenges. The following sections address biases against use of medications in vulnerable populations, ethical concerns regarding the use of placebos, challenges in diagnostic screening, and neurodevelopmental and other biological factors that should be considered in designing and conducting medication trials for treating adolescent alcohol use disorders. Later sections will provide details on promising medications.
Bias against the use of medications
Some groups have objected to the use of medications during recovery from alcohol misuse or dependence. For example, some 12-step support groups deny the importance of accurate diagnosis and treatment with medications. Other support groups, parents and treatment-seeking adolescents may be concerned that medications will be misused or act as a poor substitute for individual growth–which is considered to be what is needed. That is, adolescents with alcohol problems are in a developmental phase that will be outgrown. While this latter argument may be true for some adolescents who use alcohol, research shows that disease progression in alcohol-dependent adolescents is more typical ( Brown et al., 2001 ). Some adolescents with alcohol dependence will not live to adulthood if left untreated. Therefore, medications that are shown to be useful adjuncts to psychosocial treatments for adolescent alcohol use disorders may help to reduce disease prevalence, morbidity and mortality.
Ethical concerns regarding the use of placebos
In the US, the ethics of using placebos in clinical trials for vulnerable populations (see Miller, 2000 , Roberts et al., 2001 , and Charney et al., 2002 for reviews) has been debated. Opposition to the use of placebos in psychiatric clinical trials centres on two core issues–the ethical perspective to help and not harm treatment-seeking subjects and concern that psychiatric patients may be unable to make informed decisions about their care ( Roberts et al., 2001 ). These concerns are especially relevant in vulnerable populations, including children and adolescents ( Roberts et al., 2001 , Charney et al., 2002 ). Notably, however, no medication has been well validated for the treatment of alcohol use disorders in children or adolescents. Therefore, the use of placebos in research on treatment for alcohol use disorders is necessary to establish the efficacy and side-effect profile of experimental agents, reduce biases based upon the subject's perception of whether or not he or she is taking active medication, and separate out the ‘pill-taking’ expectancy effects from the pharmacological actions of the medication.
METHODOLOGY OF TRIALS USING MEDICATIONS AS ADJUNCTS TO PSYCHOSOCIAL TREATMENTS
Challenges in diagnostic screening
While the rationale for conducting controlled medication trials for adult alcohol-dependent individuals is based on a disease model, considerable controversy exists regarding how to best define the adolescent syndromal counterpart. That is, ‘Does a disease model apply to adolescent alcohol use disorders?’ Even for those who accept that adolescent alcohol use disorders are a disease, the current diagnostic nosology is inadequate. Further, assessment can be complicated by other psychiatric comorbidities as well as family and peer influences, and diagnostic and environmental factors need to be evaluated thoroughly to fully characterize the clinical presentation.
Diagnosis of adolescent alcohol use disorders
Validated measures should be used to diagnose psychiatric disorders and substance use outcome ( Clark and Winters, 2002 ). Instruments should also assess whether a respondent is ‘faking good’ or lying ( Deas and Thomas, 2001 ).
The evaluating clinician should be aware of the limitations in the current nosology for alcohol use disorders in adolescents, such as the fourth edition of the Diagnostic and Statistical Manual, revised text (DSM-IV-TR) ( American Psychiatric Association, 2000 ). Limitations include insufficient specificity of the developmental progression of symptoms, such that misuse symptoms may not precede dependence symptoms ( Martin et al., 1995 ). Regular users of alcohol may have one or two dependence symptoms without misuse symptoms ( Bukstein et al., 1989 , Martin et al., 1995 , Winters, 1999 ). Furthermore, impairment due to alcohol use may be masked by comorbid psychiatric disorders as well as by family dysfunction ( Brown and D'Amico, 2003 ). For adolescents with alcohol dependence, tolerance is often less of a specific symptom than in adults ( Martin et al., 1995 ). Adolescents may differ from adults in their pattern of reported withdrawal symptoms, with more mood lability, in addition to physiological and cognitive symptoms ( Stewart and Brown, 1995 ). Adolescents withdrawing from alcohol typically have more complicated outcomes than adults with the same condition, due to their increased propensity for comorbid polysubstance misuse ( Stewart and Brown, 1995 ) and the greater impact of alcohol on neurocognitive functioning in youths compared with adult counterparts ( Tapert and Brown, 1999 ). Adolescents with clinically significant alcohol symptoms may not meet criteria for a DSM-IV-TR Alcohol Use Disorder and are called ‘diagnostic orphans’ ( Pollock and Martin, 1999 ). Over 1 year, these diagnostic orphans have similar temporal profiles of problems to those of alcohol misusers, with tolerance often occurring before misuse symptoms ( Pollock and Martin, 1999 ), thereby questioning the external validity of the diagnostic scheme.
Diagnosis of psychiatric comorbidity
Even with the use of validated measures, diagnosis of co-occurring psychiatric disorders adds to the complexity of assessment of treatment-seeking adolescents with alcohol use disorders. The high prevalence of psychiatric comorbidity, and overlapping ages of onset and developmental patterning of comorbid psychiatric disorders, can be vexing in the initial assessment of adolescents presenting for comprehensive evaluation and treatment.
Most studies have shown that the earlier the onset of first alcohol and other substance use during adolescence, the greater is the risk for later substance use disorder. Anthony and Petronis (1995), using retrospective data from the ‘Epidemiologic Catchment Area’ study, showed that the risk of adult drug problems was directly related to age at onset of first drug use, with risk being two-fold higher in adults who reported first use before 13 years of age, compared with those who reported first use after the age of 17 years. The typical time from first use to problem use was 4 years, regardless of the age at first use. In the ‘National Longitudinal Alcohol Epidemiologic Survey’, a national probability sample from the US, age at first alcohol use was a powerful predictor of lifetime alcohol misuse and dependence ( Grant and Dawson, 1997 ). Prospective data suggested that by adulthood, those adolescents who began drug use prior to the age of 15 years had greater impairment compared with those who began drug use at the age of 16–17 years ( Fleming et al., 1982 ). Also, the ‘New York Longitudinal Study’ showed that adolescents with early-onset drug use (13–15 years) had a worse prognosis in adulthood than those adolescents with later-onset drug use ( Tubman et al., 1990 ). Similarly, Costello et al. (1999) reported that in a rural sample, boys who had early first use were more likely to become substance misusers in later life.
The onset of non-substance-related psychiatric disorders generally precedes the onset of substance use disorders. Longitudinal data suggest that disruptive behavioural disorders generally precede adolescent alcohol use disorders ( Biederman et al., 1997 ). For example, conduct disorder, which is often associated with attention-deficit hyperactivity disorder, often precedes adolescent alcohol use disorders and other substance use disorders ( Biederman et al., 1997 , White et al., 2001 ). Furthermore, the persistence ( Molina and Pelham, 2003 ) and severity ( White et al., 2001 ) of conduct disorder symptoms and attention deficit hyperactivity symptoms predict the severity of adolescent substance use disorders. Robins and McEvoy (1990) have shown that the greater the number of conduct disorder symptoms, the earlier the age of initial use and the greater the likelihood of developing substance use disorders. Anxiety and mood disorders can complicate the development of adolescent alcohol use disorders and other substance misuse ( Clark and Neighbors, 1996 ). Among adolescents with alcohol use disorders, increased prevalence of anxiety and mood disorders has been reported, especially among female adolescents ( Deykin et al., 1987 ). Furthermore, major depression in children, unlike in adults, is commonly found with substance use disorder ( Clark et al., 1998 ). Prospective research is needed to establish the pathophysiological and developmental sequence of childhood anxiety and mood disorders and adolescent alcohol use disorders.
Treatment-seeking adolescents who misuse alcohol and other drugs have high rates of externalizing behavioural disorders as well as anxiety and mood disorders ( Bukstein et al., 1989 , Crowley and Riggs, 1995 , Whitmore et al., 1997 , Cornelius et al., 2003 ). For example, in a sample of 279 adolescents in the US aged 15–19 years, who met DSM-III-R criteria for dependence on alcohol or other drugs, the lifetime prevalence of post-traumatic stress disorder was 29.6% ( Deykin and Buka, 1997 ). In another clinical sample in the US of 98 adolescents, prevalence of lifetime diagnoses included 78.6% for conduct disorder, 16.3% for attention-deficit hyperactivity disorder, 12.2% for major depression and 7.1% for generalized anxiety disorder. The total prevalence of greater than 100% was due to adolescents having more than one diagnosis. Furthermore, in this same sample, 16.3% had a current oppositional disorder and 6.1% had current post-traumatic stress disorder symptoms ( Mikulich et al., 2001 ). In a case series of 103 Icelandic adolescent outpatients aged 12–18 years presenting with substance misuse, about 75% of patients presented with psychiatric comorbidity: conduct disorder (44%), depression (28%) or post-traumatic stress disorder (11%) ( Hannesdottir et al., 2001 ).
Is alcoholism a mental disorder
In conclusion, the very high rates of comorbid psychiatric disorder with adolescent substance use disorder in clinical samples are likely to negatively affect retention and engagement in treatment, and may require specific medications as adjuncts to psychosocial treatments to address specific patterns of psychiatric comorbidity.
Collateral informants, serial serum and urine drug screens, and other appropriate biochemical tests should be used at initial assessment and throughout the study to monitor outcomes ( Winters et al., 2001 ).
Primary drinking outcome variables should include self-reported alcohol consumption: frequency (drinks per day), severity (drinks per drinking day and percentage of heavy drinking days), and abstinence (percentage of days abstinent). The usefulness of objective measures of alcohol consumption such as serum carbohydrate-deficient transferrin levels and serum gamma-glutamyl transferase in adolescent clinical trials has not been investigated and should be an area for future research.
Recruitment and retention
Enhanced methods are needed for recruitment and retention. Specific strategies are needed during the pre-treatment, treatment, and post-treatment phases ( Meyers et al., 2003 ). Prior to beginning recruitment, the research team should update basic information, including street guides, maps and community resource directories. Representatives from the research treatment team should meet with various agencies to provide an overview of the study, ask for cooperation in referrals and follow-up, identify specific contact personnel, explain issues related to confidentiality, and review the consent forms and additional release of information forms to ensure confidentiality. During treatment, the research team should obtain locator information for tracking adolescent participants. Contacts should include personal, community, social and peer information. Personal contact information should include the participant's name (including nickname and street name), identifying physical findings, telephone and pager numbers and e-mail addresses. Community contacts should include teachers, guidance counsellors or coaches, sports teams and schedule and location, places of work, systems or service involvement, outside treatment providers, and locations where the participant ‘hangs out’. Household and family information should include current address and phone numbers, pager and cell phone numbers of siblings and family members, full names, addresses and phone numbers of several relatives, and at least one neighbour. Peer contact information should include full names, addresses and telephone/pager information for girlfriend(s)/boyfriend(s), best friend and one other friend. Research staff should send initial notes to the adolescents and parent(s)/guardian(s) thanking them for agreeing to participate in the study. The research teams should send to all patients a personal letter congratulating them on their specific progress during the study. Financial incentives for participation should be used whenever possible.
Neurodevelopmental and biological factors that may influence outcomes and retention
Careful attention should be paid to the psychosocial and cognitive development of adolescent subjects in the design of clinical research. The adolescent patient often presents with increasing judgement and critical thinking and must be involved in the consent process. Patients 18 years of age and older should sign the consent form as the participant, those younger than 18 years of age should give their assent, and the parent or guardian can then provide written consent for them to participate in a pharmacotherapy trial.
Treatment-seeking adolescents with alcohol use disorders are more likely to have significant problems with attention, risk appraisal and coping skills ( Crowley and Riggs, 1995 , Meyers et al., 2003 ). For adolescents with alcohol use disorders, role transitions, including joining the work force, independent living, and choices for continued education, may be delayed or derailed. Furthermore, legal problems and mounting stressors may further impede their development. The cumulative effects of these developmental factors may be distorted perceptions of problems, unwillingness to report them, low motivation to change drug use behaviour and lower receptivity to engage in treatment compared with alcohol-dependent adults ( Meyers et al., 2003 ). Therefore, motivational interviewing along with enhanced tracking methods may be necessary to recruit, engage and retain adolescents with alcohol use disorders in all forms of treatment, including medication studies.
Factors that are unique to medication trials for adolescents with alcohol use disorders include the need to address concerns about short-term effects and age-specific pharmacokinetics, special considerations on safety monitoring, and concerns about developmental risks and long-term outcomes on brain development.
Specific issues in medication trials for adolescents
Short-term effects, compliance and age-specific pharmacokinetics
Medications with tolerable side-effects should be selected to enhance compliance. Compliance should be assessed with pill counts, automated dispensing devices (e.g. MEMS caps) and, when feasible, blood levels. Age-specific pharmacokinetics should be used to determine the most appropriate dosing strategy. Blood levels of the medication and measures of safety and toxicity should be collected from all participants ( Rogers, 1994 ).
Drug safety ascertainment
While there are increased concerns over the short-term effects of medications in controlled clinical trials in paediatric populations, most medication studies in children have used passive surveillance and general inquiry methods ( Greenhill et al., 2003 ). New methods of active surveillance using standardized, drug-specific questionnaires are, however, beginning to be used ( Ioannidis and Lau, 2001 , Greenhill et al., 2003 ). In addition, common definitions should be adopted for studies in children, using attribution-free terminology, standard definitions of severity of adverse events and manual-based approaches for the assessment of adverse events, reporting and attribution ( Vitiello et al., 2003 ). For adolescent samples, particularly younger adolescents, additional research is needed to determine which elicitation methods work best for which type of informant (i.e. adolescent or parent, or both).
Developmental risks, protection from risk and long-term outcomes on brain development
The biological and behavioural changes that occur during adolescence are profound. The effects of body growth, sexual maturation, and synaptic reorganization with consequent cognitive changes may alter pharmacokinetics and pharmacodynamics. The concept of developmental age should be considered, in addition to chronological age ( Finkelstein, 1994 ). Measures of developmental age include body height, weight, body mass index and pubertal development, as well as measures of self-image such as the ‘Offer Self-Image Questionnaire’ ( Finkelstein, 1994 , Patton and Noller, 1994 ).
In selecting medications to treat adolescents with alcohol use disorders, the clinician should identify whether there are theoretical or biological bases for the agent to affect brain or somatic growth and development. The clinician should also specify what is likely to be an adequate dose, the duration of treatment and whether there is a theoretical or biological basis for treatment response that justifies the risk for exposure to the agent ( Rogers, 1994 ).
For adolescents with alcohol use disorders, additional neurodevelopmental factors should be considered in the choice of medications, when to use these medications and for how long. The treating clinician should weigh the potential negative effects of chronic alcohol use on brain development against the potential risks and benefits of the specific medication.
Normal brain development during adolescence includes structural changes in grey and white matter, reorganization of cortical synapses and changes in various neurotransmitter systems. In vulnerable individuals, elimination and reorganization of cortical synapses and changes in neurotransmitter systems are hypothesized to increase the risk for developing alcohol use disorders. Huge numbers of synapses are lost from neocortical brain regions or reorganized during adolescence ( Rakic et al., 1994 ). Almost half of cortical synapses are lost by the end of puberty ( Huttenlocher, 1984 , Rakic et al., 1994 ). In human adolescents there is a profound increase in the amount of information that can be processed independently by each hemisphere ( Merola and Liederman, 1985 ). There are also different growth trajectories for specific brain regions, with the adolescent cortex not approximating the adult counterpart until the age of 20 years.
Receptors from a variety of neurotransmitter systems have progressive then regressive changes from infancy through adolescence ( Lidow et al., 1991 , Lidow and Rakic, 1992 ). Included are changes in the density and distribution of dopamine (DA), serotonin (5-HT), acetylcholine, glutamate and gamma-amino-butyric acid (GABA) systems. Co-occurring with synaptic reorganization, the number and density of specific receptors are thought to decline throughout the brain at different rates throughout adolescence, producing a relative decline in excitatory stimulation to the neocortex ( Rakic et al., 1994 ). During adolescence, decreases in excitatory input to the cortex can be seen with reductions in glucose metabolism ( Chugani et al., 1987 ). Commensurate with decreases in excitatory input to the cortex, DA input to the prefrontal cortex increases to levels higher than those seen earlier or later in development ( Spear, 2000 ).
DA systems undergo substantial reorganization during adolescence. At least one-third to half of DA D1-like and D2-like receptors in the striatum present in juveniles are lost by adulthood ( Seeman et al., 1987 ). DA transporter activity increases by about 30% at the onset of adolescence ( Gelbard et al., 1989 , Tarazi et al., 1998, 1999 ). Also, the expression in early life of DA autoreceptor-like modulation of DA synthesis that occurs in the prefrontal cortex disappears during adolescence ( Teicher et al., 1991 , Anderson et al., 1997 ). The changes within the DA system are hypothesized to produce a shift in the balance of DA activity to the prefrontal cortex and the anterior cingulate gyrus (‘top-down control’) from the striatal or mesolimbic terminal regions (‘bottom-up control’). During normal adolescent brain development, the result is hypothesized to be a predominance of DA activity in the prefrontal cortex early in adolescence, as opposed to the mesolimbic brain region. The mesolimbic region, thought to be critical for the modulation of the salience of various incentive stimuli, has connections with the orbitofrontal and anterior cingulate gyrus ( Spear, 2000 , Goldstein and Volkow, 2002 ). The functional DA deficits in the accumbens and related mesolimbic brain regions, compared with activation of frontal cortical structures, could produce a ‘reward deficiency syndrome’ characteristic of abstinent adult drug users and other at-risk adults ( Spear, 2000 ). With experimental alcohol use, these functional DA deficits might lead to increased alcohol consumption in vulnerable individuals.
Repeated alcohol and other drug use in adolescence is postulated to produce allostatic changes in brain circuits that subserve motivation, behavioural control and drug reinforcement ( Crews et al., 2000 , Dawes et al., 2000 , Spear, 2002 , Chambers et al., 2003 , Koob, 2003 ). Mounting evidence suggests that in the brain of adolescents with alcohol use disorders, glutaminergic cortico-accumbens circuitry is recruited ( Dawes et al., 2000 , Spear, 2002 ), along with augmented DA transmission, from the ventral tegmental area to the prefrontal cortex, nucleus accumbens (NAc), ventral pallidum and amygdala ( Kalivas and Nakamura, 1999 , Dawes et al., 2000 , Goldstein and Volkow, 2002 ). With repeated drug use, primary and secondary motivational circuits, modulated in part by the serotonergic and opioid systems, are hypothesized to produce greater drug reinforcement and drug use, particularly in those adolescents who have a positive family history for alcohol use disorders. Immature inhibitory circuits that are exposed to repeated binge drinking and other drug use may worsen behavioural control and impulsivity ( Dawes et al., 2000 , Chambers et al., 2003 ). Hence, it is tempting to speculate that repeated binging produces neuroplastic changes in specific brain regions, such that behavioural under-control and alcohol consumption worsen, thereby increasing the risk of adolescent alcohol use disorders. Allostatic changes in key neurotransmitter systems, including serotonergic and opioid systems, as well as excitatory amino acids and GABAergic systems, might also modulate the pathogenesis of the disorder, and possibly the medication treatment response.
Implications for clinical subtyping
Innovative uses of clinical subtyping for adolescents with alcohol use disorders will require careful selection of clusters of individuals, classified within a theoretical framework, that identifies and organizes predicted treatment response based on relatively enduring, continuously distributed factors that are predominantly heritable ( Lesch et al., 2001 ). Individual differences in medication treatment response are likely to result from gene–gene and gene–environment interactions in the regulation of brain neuroplasticity ( Lesch et al., 2001 ). A multi-factorial model of complex disorders ( Falconer, 1965 , Lander and Schork, 1994 ) will be necessary to provide a sufficiently broad theoretical framework to encompass specific gene–gene and gene–environment interactions that predict medication treatment response in adolescents with alcohol use disorders. In a multi-factorial neurodevelopmental model, several factors should be particularly useful in identifying clusters of adolescents with alcohol use disorders who have underlying neurochemical abnormalities that are differentially responsive to particular medications, these factors include age of onset, family history of disorder and comorbid psychopathology ( Lander and Schork, 1994 ).
RATIONALES FOR SPECIFIC MEDICATIONS
Assuming that a multi-factorial neurodevelopmental model of medication treatment response is valid, which medications are likely to be effective in decreasing craving and alcohol use in adolescents with alcohol use disorders? The following section synthesizes findings from preclinical and treatment studies in adults, where there are also at least some pilot treatment data in adolescents. Serotonergic agents and opioid antagonists currently meet these criteria. Also included are discussions of agents that antagonize glutaminergic neurotransmission, facilitate GABAergic function, or both. When appropriate, functional relationships and hypotheses are provided.
Preclinical human studies have shown that serotonergic neurotransmitter systems are involved in alcohol consumption, mood regulation and impulse control ( LeMarquand et al., 1994b , Young et al., 1996 ). Acute alcohol administration generally produces serotonin release, whereas chronic administration in most cases has been shown to decrease serotonin levels in the midbrain ( LeMarquand et al., 1994a, b , Fils-Aime et al., 1996 ). Animal studies have generally demonstrated decreases in alcohol consumption with a host of serotonergic agents ( LeMarquand et al., 1994a ).
If there exist subtypes of adolescent alcohol use disorders with underlying serotonergic abnormalities, then these subtypes should have differential treatment responses to specific serotonergic agents that ameliorate underlying serotonergic abnormalities. The typologies that are most likely to show differential serotonergic medication treatment response include those classification systems that use age of onset. Examples include the type 1 and type 2 ( Cloninger, 1987 ), and type A and type B ( Babor et al., 1992 ), typologies, and age of onset of problem drinking ( Johnson et al., 2000a ).
receptor has been implicated in the molecular mechanisms regulating alcohol consumption and mediating the actions of alcohol, alcohol reinforcement and ondansetron treatment response. Behavioural pharmacological studies have shown that the rewarding effects of alcohol are modulated by ethanol-induced activation of mesocorticolimbic DA receptors interacting with 5-HT
receptor, a receptor-gated ion channel, is densely distributed at the neuron terminals regulating DA release within the mesocorticolimbic system at the level of the cell body and in DA projection regions. Mesocorticolimbic DA pathways, particularly those that include the posterior ventral tegmental area and the NAc, are activated by administered or ingested alcohol, and are hypothesized to mediate its rewarding effects ( Di Chiara and Imperato, 1988 , Lovinger, 1999 , McBride et al., 2004 ). Results from three different types of animal studies suggest that 5-HT
receptor antagonism diminishes DA activity, which may decrease the rewarding effects of misused drugs including alcohol. For example, 5-HT
antagonism: decreases hyperlocomotion in rats, after DA or ethanol injection into the NAc ( Bradbury et al., 1985 ), attenuates neurokinin-induced hyperlocomotion ( Hagan et al., 1990 ), and reduces ethanol consumption in a variety of paradigms–see McBride et al. (2004) for a review. Blockade of 5-HT
receptors is hypothesized to decrease DA release, thereby reducing alcohol consumption and craving ( Johnson and Ait-Daoud, 2000 ). 5-HT
receptor blockade may also decrease serotonin turnover and reduce mood disturbance ( Johnson et al., 2003b ).
Human laboratory studies have shown that ondansetron reduces alcohol reward and preference. Ondansetron pre-treatment also attenuates low-dose ethanol-induced positive subjective effects ( Johnson and Cowen, 1993 ). High doses of ondansetron pre-treatment decrease alcohol preference ( Swift et al., 1996 ).
Ondansetron has been shown in studies from two different research groups to decrease alcohol consumption ( Johnson et al., 2000b , Kranzler et al., 2003 ) in adults with early-onset alcohol dependence. Johnson et al. (2000b) showed in a sample of 321 alcohol-dependent individuals that ondansetron, at 1, 4 and 16 μg/kg twice per day, was superior to placebo in decreasing drinks per day and drinks per drinking day. The 4 μg/kg b.i.d. dose, compared with placebo, also significantly increased percentage of days abstinent and total days abstinent per study week ( Johnson et al., 2000b ). In a prospective, open-label study of ondansetron in early- versus late-onset alcohol-dependent adults (n = 40, 20 early-onset alcohol-dependent adults, EOA, 20 late-onset alcohol-dependent adults, LOA), together with weekly relapse prevention therapy, EOA reported significantly greater decreases in drinks per day and drinks per drinking day, and reductions in alcohol-related problems, compared with LOA ( Kranzler et al., 2003 ). Results from a recent open-label study of adolescents with alcohol use disorders (n = 12) who received 4 μg/kg b.i.d. also showed significant within-group decreases in self-reported alcohol consumption (Dawes et al., unpublished observations). Side-effects were mild. Because alcohol releases 5-HT, and alcohol reward appears to be mediated in part through the 5-HT
receptor, we hypothesize that the mechanism underlying ondansetron treatment response in adult and adolescent EOA is blockade of up-regulated post-synaptic 5-HT
receptors. Double-blind, placebo-controlled clinical trials, human laboratory studies, and imaging studies will be needed to establish this hypothesis in adolescents with alcohol use disorders.
Selective serotonin reuptake inhibitors
Several double-blind, placebo-controlled trials with different selective serotonin reuptake inhibitors (SSRI) (zimelidine, citalopram, viqualine and fluoxetine) have shown reductions in alcohol consumption in adult heavy social drinkers ( Naranjo and Knoke, 2001 ). Several subsequent controlled trials of fluoxetine in alcohol-dependent adults have not demonstrated significant reductions in drinking, compared with placebo ( Gorelick and Paredes, 1992 , Kranzler et al., 1995 , Kabel and Petty, 1996 ). These latter studies led to the conclusion that some patients who drink excessively or are alcohol dependent may not have underlying serotonergic abnormalities or, at least, serotonergic abnormalities that are responsive to SSRI.
A more specific test of SSRI response would be to select different clinical subtypes of alcohol-dependent individuals with central 5-HT abnormalities, to detect differential treatment response. To test this hypothesis, two randomized controlled trials used different SSRI to treat alcohol-dependent adults classified by two different typologies of alcohol dependence severity ( Kranzler et al., 1996 , Pettinati et al., 2000 ). The hypothesis guiding these two studies was that 5-HT abnormalities would exist in certain subtypes of alcohol dependence, and would be successfully treated by SSRI, resulting in reduced drinking and improved abstinence. The outcomes were the opposite of the prediction: type B or early-onset-like alcohol-dependent adults were not responsive to fluoxetine and sertraline ( Kranzler et al., 1996 ). Indeed, Kranzler et al. (1996) reported that type B alcohol-dependent adults drank more frequently during the trial when treated with fluoxetine, compared with placebo treatment. Pettinati et al. (2000), in a more recent double-blind, placebo-controlled study with alcohol-dependent adults in a 14-week trial of 200 mg/day of sertraline, classified subjects into lower risk/severity (type A: n = 55) and higher risk/severity (type B: n = 45). This latter study showed high rates of total abstinence (64%) compared with placebo (<,10%) for the type A subtype. Taken together, these two studies provide further evidence that distinct subtypes of alcohol-dependent individuals may respond differently to particular serotonergic agents, in this case to SSRI.
Preliminary findings suggest that SSRI are beneficial in adolescents with major depression and an alcohol use disorder (AUD). Cornelius et al. (2001) have shown, in an open-label study of fluoxetine (20 mg/day) in 13 adolescents with an AUD and major depression, significant within-group decreases in comorbid major depression and in drinking days per week and drinks per drinking day. Fluoxetine was well-tolerated. Cornelius is conducting a prospective, double-blind, placebo-controlled study of fluoxetine versus placebo to treat adolescents with an alcohol use disorder and major depression. In this latter study, adolescents are being assigned to either of two treatment arms: fluoxetine or placebo, all subjects are receiving treatment as usual as the psychosocial intervention (Cornelius, personal communication, 2003).
Pre-synaptic serotonin reuptake enhancers: tianeptine
In contrast to SSRI, tianeptine enhances pre-synaptic 5-HT reuptake, decreasing 5-HT availability at the post-synaptic 5-HT receptor ( Curzon et al., 1992 ). Acutely, tianeptine increases 5-HT uptake in brain synaptosomes (cortex and hippocampus) ( Mennini et al., 1987 , Fattaccini et al., 1990 ), as well as platelets ( Chamba et al., 1991 ), presumably with the long-term effect of decreasing serotonin turnover. In addition, chronic tianeptine treatment decreases the stress response of the hypothalamic– pituitary–adrenal axis, and modulates neuroendocrine response to cytokines ( Castanon et al., 2003 , Nickel et al., 2003 ). Tianeptine's mechanisms of action would, therefore, make it a promising medication for treating anxiety-related and affective disorders.
Thus it is of interest that tianeptine has been shown to be effective in adults for the treatment of depression with comorbid alcohol misuse or dependence. In a double-blind, placebo-controlled study, tianeptine treatment significantly improved depression after withdrawal compared with placebo ( Lôo et al., 1988 ). Tianeptine does not appear promising for the treatment of non-depressed alcohol-dependent individuals. For instance, in a multi-centre, double-blind, controlled trial, tianeptine (12.5 mg t.i.d.) showed no difference from placebo after 9 months of treatment (n = 327) ( Favre et al., 1997 ).
Notably, the use of tianeptine has been piloted as a treatment for adolescent alcohol use disorders. Niederhofer et al. (2003) reported on a study of 26 adolescent inpatients with alcohol use disorders, aged 16–19 years, assigned randomly to 37.5 mg of tianeptine or placebo for 90 days. All subjects received psychotherapy three times per week, physiotherapy and ergotherapy. Continuous abstinence and cumulative abstinence duration were significantly greater for the tianeptine group compared with placebo. This study did not, however, report the comparative length of inpatient versus outpatient care for the two treatment groups. It is therefore tempting to speculate that tianeptine may be effective in adolescents with alcohol use disorders, compared with their adult counterparts, because it treats their relatively higher level of subsyndromal anxiety-related and affective disorders.
In summary, larger empirical studies are needed to establish tianeptine's effectiveness in treating adolescent alcohol use disorders.
Is alcoholism a genetic disorder
Mu- and delta-opioid receptors mediate the rewarding effects of ethanol ( Hemby et al., 1997 ). The mechanisms for opiate antagonist-induced decreases in ethanol consumption are hypothesized to involve direct and indirect effects of beta-endorphin. Beta-endorphin pathways appear to disinhibit tonic inhibition of GABAergic transmission to DA cells in the NAc ( Matthews and German, 1984 , Johnson and North, 1992 ). Second, beta-endorphin pathways directly stimulate DA cells in the NAc ( Gianoulakis, 1998 ). Third, beta-endorphins may modulate the hypopituitary-adrenal axis response to a variety of stressors ( Gianoulakis, 1998 ).
Opioid antagonists decrease overall alcohol consumption (preference), reinforcement, and craving. These agents produce modest decreases in ethanol consumption ( Swift et al., 1994 ), alcohol stimulant effects ( King et al., 1997 ), and cue-induced craving ( Monti et al., 1999 ), and increase latency to drink alcohol ( Palfai et al., 1999 ) and time to initiating drinking in a natural setting ( Davidson et al., 1996 ). Naltrexone, a mu-opioid antagonist, when given prior to alcohol reinstatement, diminishes the amount of alcohol consumed and the alcohol deprivation effect ( Froehlich et al., 2003 ). (The alcohol deprivation effect is defined as the transient increase in alcohol consumption that takes place in laboratory animals after a period of deprivation from alcohol.)
Three meta-analyses have supported the use of naltrexone to attenuate alcohol consumption or prevent relapse in alcohol-dependent patients ( Kranzler and Van Kirk, 2001 , Streeton and Whelan, 2001 , Srisurapanont and Jarusuraisin, 2002 ). Srisurapanont and Jarusuraisin's (2002) Cochrane review is the most comprehensive to date, with 19 randomized clinical trials or controlled clinical trials examined, and with a total number of 2102 alcohol-dependent adults included in the meta-analysis. The main findings showed that the primary outcome variables of number of patients who returned to drinking, and percentage or number of drinking days, significantly favoured naltrexone compared with placebo treatment. However, discontinuation rates in short-term studies (≤12 weeks of treatment) were high, without statistically significant differences between naltrexone and placebo. The US multi-site study by Krystal et al. (2001) was negative in the ‘2003 Cochrane Systematic Review’. Of note, the Krystal et al. (2001) study enrolled a more severe and chronic alcohol-dependent group than previous studies and the special population of military veterans. Srisurapanont and Jarusuraisin (2002) concluded that for adults, the optimal duration is likely to be longer than 3 months, with the greatest effectiveness of naltrexone being seen for alcohol-dependent subjects who used moderate amounts of alcohol, combined with psychosocial treatment. Particular psychosocial treatments, such as coping skills therapy, may be particularly effective when combined with naltrexone ( O'Malley et al., 1992 , Heinala et al., 2001 ).
Naltrexone is currently being investigated for the treatment of adolescent alcohol use disorders. In an open-label study of naltrexone for adolescents with alcohol use disorders, where five subjects were enrolled for 6 weeks, drinks per drinking day was decreased significantly by the third week compared with baseline (Deas et al., unpublished observations). Of note, adolescents in this open-label study tolerated 50 mg of naltrexone poorly due to extreme nausea, and the dose was reduced to 25 mg/day. No elevations in liver functions were reported. Deas and colleagues are now conducting a double-blind, placebo-controlled, 12-week naltrexone trial with adjunctive, weekly, individual cognitive behavioural therapy for 100 male and female adolescents with alcohol use disorders (Deas, personal communication, 2003).
Excitatory amino acids and gaba
Medications that antagonize glutaminergic neurotransmission, that facilitate GABAergic function, or both, have been shown to be effective in the treatment of adult alcohol dependence. Because alterations in the glutaminergic and GABAergic systems, as well as their interactions, are involved in the pathogenesis of adolescent alcohol use disorders ( Spear, 2002 ), medications that affect these neurotransmitter(s) might be effective treatment agents. Specifically, preliminary data on acamprosate suggest its ability to diminish adolescent alcohol consumption. On a theoretical basis, both topiramate and gamma-hydroxybutyrate (GHB) might be effective in this population.
Acamprosate (calcium acetylhomotaurinate) is a derivative of the amino acid taurine. In animals, acamprosate tends to block alcohol consumption ( Le Magnen et al., 1987 ). Mechanistically, acamprosate's effects might be due to its ability to quell N-methyl-D-aspartate (NMDA) excitation that occurs in alcohol withdrawal and early abstinence ( De Witte et al., 1996 , Spanagel and Holter, 2000 ).
In a series of large double-blind, placebo-controlled European studies, but not in the US multicentre study, acamprosate has demonstrated efficacy as a treatment for adult alcohol dependence (see Johnson and Ait-Daoud (2000) for a review). Preliminary work suggests that acamprosate might be effective in treating adolescents with alcohol dependence. In a small-sample, double-blind, placebo-controlled study of adolescents with alcohol dependence, acamprosate was shown to be effective and well tolerated as a pharmacological adjunct to psychosocial treatment ( Niederhofer and Staffen, 2003 ). Large-scale definitive studies testing acamprosate's efficacy in adolescent alcohol use disorders are now warranted.
Topiramate is a sulfamate-substituted fructopyranose derivative. Its dual mechanisms of action that combine to decrease extracellular midbrain dopamine levels are the potentiation of GABA
function at a non-benzodiazepine site on the GABA
receptor ( Moghaddam and Bolinao, 1994 , White et al., 2000 ) and the blockade of glutaminergic activity at alpha-amino-3-hydroxy-5-methylisoxazole-4-propionic acid and kainate receptors ( Skradski and White, 2000 ).
In a recent proof-of-concept trial of this hypothesis, topiramate was used to treat adults with alcohol dependence ( Johnson et al., 2003a ). In that study, alcohol-dependent adults were enrolled in a double-blind, randomized, 12-week trial comparing topiramate (up to 300 mg/day) with placebo (n = 150, 75 per group). Topiramate, compared with placebo, was effective in decreasing craving and heavy drinking and in promoting abstinence ( Johnson et al., 2003a ).
Taken together, these findings suggest that topiramate would be a promising candidate medication to study for the treatment of adolescent alcohol use disorders. We are currently pursuing this possibility.
Gamma-hydroxybutyrate (GHB) is a naturally occurring analogue of GABA in the mammalian brain that is thought to mediate sleep cycles, body temperature, cerebral glucose metabolism, and memory ( Mamelak, 1989 , Li et al., 1998 ). Also, GHB may modulate endogenous dopamine levels via interactions with GABA receptors ( Tunnicliff, 1992 ). Therapeutic uses of GHB have included use as a general anaesthetic, as well as for narcolepsy and alcohol dependence ( Myrick et al., 2001 , Smith et al., 2002 ). Several double-blind trials have shown that GHB treatment rapidly decreases alcohol withdrawal symptoms, especially for symptoms of anxiety, agitation and depression ( Gallimberti et al., 1989 , Addolorato et al., 1999 , Nimmerrichter et al., 2002 ). These properties would make GHB an interesting candidate medication for treating adolescents with alcohol use disorders who have a propensity for subsyndromal anxiety-related and affective disorders. Maremmani et al. (2001) have shown that GHB may be efficacious in the chronic treatment of treatment-resistant, alcohol-dependent adults. GHB has been reported to have significant misuse potential ( Smith et al., 2002 )–an effect that would limit its usefulness in the treatment of both alcohol-dependent adults and adolescents with alcohol use disorders.
RATIONALES FOR COMBINATIONS OF MEDICATIONS
While it is premature to discuss combination strategies in detail for the treatment of adolescent alcohol use disorders because no single agent has proved effective, some general themes are worthy of some consideration. For example, combinations of medications may have additive effects on particular neurotransmitter systems or synergistic effects across two or more neurotransmitter systems, to decrease alcohol consumption compared with single-agent treatment. A second rationale for combination pharmacotherapy is to treat patients with both alcohol use disorders and other comorbid psychopathologies. Patients who present with both a comorbid psychiatric disorder and a substance use disorder may require a second medication to be added, to treat one or more disorders. As adolescents with alcohol use disorders are especially likely to have comorbid psychopathology, combined pharmacotherapeutic interventions that target both the comorbid substance use disorder and other psychopathology might be a reasonable treatment approach. Thirdly, combinative medications might be prescribed to enable one compound to offset the adverse effects of the other(s), thereby facilitating compliance and the potential for finding effectiveness.
CONCLUSIONS AND FUTURE DIRECTIONS
To date, findings from open-label studies, case reports, and small randomized studies suggest that various medications may be efficacious in treating adolescent alcohol use disorders. Although well-designed studies are under way, no adequately powered, double-blind, placebo-controlled medication studies for adolescents with alcohol use disorders have been published. Controlled studies of medications as adjunctive treatments to psychosocial interventions are clearly warranted. Medication treatment studies for adolescents with alcohol use disorders should target the commonly associated comorbid disorders.
Future controlled studies for adolescents with alcohol use disorders should make clear the optimal dose, duration and sequence of medication treatments, test optimal combinations of pharmacotherapy and psychotherapy, and use combinations of medications that produce changes within the same or different neurotransmitter systems to decrease alcohol consumption and craving. Findings from efficacy studies should be replicated in ‘real-world’ settings, such as juvenile detention centres and state hospitals.
Supported in part by grants to B.A.J. from the National Institute on Alcohol Abuse and Alcoholism (N01 AA01016, R01 AA010522, R01 AA012964–03 and 2U10 AA011776).