Neurobiological Processes in Adolescent Addictive Disorders

Abstract

The purpose of this review is to summarize the neurobiological factors involved in the etiology of adolescent addiction and present evidence implicating various mechanisms in its development. Adolescents are at heightened risk for experimentation with substances, and early experimentation is associated with higher rates of SUD in adulthood. Both normative (e.g., immature frontal-limbic connections, immature frontal lobe development) and non-normative (e.g., lowered serotonergic function, abnormal hypothalamic-pituitary-adrenal axis function) neurobiological developmental factors can predispose adolescents to a heightened risk for SUD. In addition, a normative imbalance in the adolescent neurobiological motivational system may be caused by the relative underdevelopment of suppressive mechanisms when compared to stimulatory systems. These neurobiological liabilities may correspond to neurobehavioral impairments in decision-making, affiliation with deviant peers and externalizing behavior; these and other cognitive and behavioral traits converge with neurobiological factors to increase SUD risk. The progression to SUD acts as an amplifying feedback loop, where the development of SUD results in reciprocal impairments in neurobehavioral and neurobiological processes. A clearer understanding of adolescent neurobiology is a necessary step in the development of prevention and treatment interventions for adolescent SUD.

INTRODUCTION

The consequences of substance use disorders (SUD) are well publicized and involve substantial costs to society.^1–3 Using data from the late 1990s, various government agencies have estimated that the annual cost of alcohol, drug, and nicotine use disorders was nearly five hundred billion dollars.^4–6 In large part, the initiation of addictive substance use appears to be an adolescent phenomenon: nearly 60% of individuals who initiate drug use do so at or before 18 years of age,⁷ and the rates of initiation rise to roughly 80% for alcohol⁷ and cigarettes.⁸ Furthermore, it appears that the early use of certain substances (e.g., cigarettes, methamphetamine, inhalants, or marijuana) is associated with accelerated use of other substances,^9,10 greater progression to SUD,^11–14 and psychiatric comorbidity.^13,15 The 2003 Youth Risk Behavior Survey stated that the use of alcohol, tobacco and illicit drugs by high school students markedly increased their likelihood of injury or death due to the four major causes of fatalities.¹⁶

Adolescence is a time of great neurobiological change.¹⁷ Evidence increasingly indicates that these changes impact the propensity of adolescents to experiment and experience persistent alterations from psychoactive substance use;¹⁸ substance use (and the consequent sequelae) in adolescence may correspond to accelerations in the development of SUDs in adulthood.^19–21 Thus, prevention or early treatment holds great promise for limiting the costs, morbidity, and mortality associated with addiction. In order to develop more effective treatment interventions, it is essential to understand the pathophysiology of addiction in youth.

This review will present an integrated etiology of the development and entrenchment of addiction in adolescents. A conceptual summary is provided in Figure 1. Research findings have led some to posit that all adolescents have neurobiological risks stemming from immature connectivity²² and/or imbalances in the expression of the motivational learning system.¹⁸ Thus, the changing neurobiology of adolescence (#1 in Figure 1) may underlie the increase in high-risk behaviors and disinhibition (#3) associated with the development of SUD.^23,24 Adolescents at high-risk (#5; e.g., children of probands with SUD) likely have neurobiological liabilities in serotonergic (5-HT), hypothalamic-pituitary-adrenal (HPA) axis and/or neurophysiological (e.g., P300) functioning above those of low-risk adolescents. These factors in high-risk youth may correspond to greater levels of conflict with parents and the formation of affiliative friendships with other high-risk youth (#2 and 4), both risk factors for SUD. Finally, following the initiation of psychoactive substance use (#6), adolescents appear to be more acutely and persistently affected than adults. One result appears to be a more rapid progression to SUD. The acute differences and persistent alterations may reflect neuroplastic changes that serve to entrench and accelerate use, resulting in greater neurobiological liability (#3) and SUD (#7).

FIGURE 1.

Etiology of SUD Development in Adolescents

This review will focus on factors associated with or leading to levels of substance use that would meet criteria for a diagnosis of abuse or dependence,²⁵ rather than factors leading only to experimental use. Many substance users remain experimenters,²⁶ and experimentation may be associated with outcomes that are no worse,²⁷ or are even better,²⁸ than outcomes in those who abstain. While the first step to addiction is experimentation,²⁹ infrequent use is significantly different than heavy use. Factors common to all adolescents (and thus, present in experimenters) will be examined only to create the foundation on which dysfunctional traits accelerate levels of substance use. In addition, this review will focus on adolescents with a familial history of SUD; such individuals have a significantly greater incidence of SUD than individuals without a family history^30,31 and are more likely to have dysfunctional neurobiological and neurobehavioral traits.³² Given the concentration of risk factors, high-risk adolescents are thought to be most likely to demonstrate pathways to SUD development.

THE DEVELOPMENTAL NEUROSCIENCE OF ADOLESCENCE: ANIMAL AND HUMAN STUDIES

Adolescence is perhaps the greatest time of neural growth, change, and maturation since infancy. The development of the executive functions (e.g., decision-making, self-monitoring, impulse control, delay of gratification) continues from childhood through adolescence,^33–38 with completion as late as early adulthood.^39–42 These neurocognitive traits correlate with prefrontal cortex (PFC) and anterior cingulate activity; the development of these traits appears dependent upon the maturation of PFC and limbic system interconnectivity. (The limbic system consists of diverse neural structures, including the cingulate, amygdala, and hippocampus, and serves to regulate emotional experience, memory, and motivational learning). Furthermore, the maturation of connections between the PFC, basal ganglia, and cerebellum also appear to be crucial for the development of higher cognitive functions.⁴³

PRUNING AND MYELINATION

In large part, these neurocognitive changes occur during adolescence and depend on large-scale myelination and synaptic pruning (#1 in Figure 1). The deposition of white matter, or myelin, increases the speed of neural transmission. White matter allows for quicker processing and more concentrated circuits to respond to the rapid demands of the environment. Using magnetic resonance imaging (MRI), longitudinal studies of human brain development have found that white matter increases linearly from ages 4 to 20.^44,45 In a study of men ranging in age from 19 to 76, Bartzokis et al.⁴⁶ reported that frontal lobe myelination peaked in adult males at 44 years of age.

Synaptic pruning is the process by which excess connections (synapses) between neurons are removed. Rakic and et al.⁴⁷ and Lictman et al.⁴⁸ theorize that pruning is determined by synapse use: connections that are employed to respond to the environment are retained and strengthened, while those that are used less often are eliminated. Synapse elimination is believed to reduce the childhood pattern of processing, which requires greater metabolic activity and the recruitment of a wider array of structures.^49–51 In addition, pruning appears to increase the efficiency of cognitive processing through the creation of dedicated neural networks.⁵² For instance, synaptic overproduction followed by selective pruning allows for maximum efficiency in associative memory functions.^53,54

Rakic and colleagues⁴⁷ have estimated that up to 30,000 synapses are pruned per second in the non-human primate adolescent brain. Thus, nearly one-half of the cortical synapses present before adolescence may be pruned during this period. Regional gray matter in humans tends to peak and decline, at least partially due to pruning, beginning from 12 to 20 years of age.^{44, 55–57} Using structural MRI, Sowell and colleagues^58,59 found large-scale cortical and subcortical brain changes in late adolescence and early adulthood. Numerous other studies document that concomitant synaptic pruning and increased myelination occur in the human frontal cortex during adolescence.^45,60–62 Referring back to Figure 1 (#1), it appears that normal adolescent neurobiology is characterized by lesser myelination, synaptic pruning, and integration than is found in adults; overall, these processes do not appear to culminate until early adulthood.⁶³

DEVELOPMENT OF NEUROTRANSMITTER SYSTEMS DURING ADOLESCENCE

In addition to connective and structural changes in the central nervous system (CNS), adolescents undergo dramatic alterations in virtually all neurotransmitter systems (also within #1 of Figure 1). Most relevant to the development of SUD are the changes experienced in dopamine-related systems. Dopamine (DA) plays a central role in the mesolimbic neural pathway. This circuit originates in the ventral tegmental area (VTA) and projects to the nucleus accumbens (NAc) and various limbic structures.^64,65 The mesostriatal release of DA occurs in response to a wide variety of environmental reinforcers, including water,⁶⁶ food,⁶⁷ and drugs of abuse.^68–70 Increased striatal concentrations of DA are essential in assigning value to these reinforcing stimuli.⁷¹

Both animal and human studies indicate that DA receptors reach a density peak early in development and undergo elimination during adolescence.^72–75 Synaptic pruning of DA receptors occurs in both the human and rat NAc in adolescence,^74,76,77 although some studies report otherwise.⁷⁸ In contrast, DA receptors in the PFC do not demonstrate significant pruning until late adolescence.^74,77,79 DA fiber density increases in the PFC of adolescent rats^80,81 and NAc of gerbils⁸² and DA inputs to the primate PFC peak in adolescence.^83,84 Teicher and colleagues⁸⁵ found age-related striatal differences in synaptic DA levels in rats; these were not seen in the NAc or medial PFC. Furthermore, increased DA synthesis has been observed in the striatum, NAc and PFC in adolescent rats.⁸⁶ Finally, striatal DA turnover is higher in adolescent than adult rats, with smaller (non-significant) differences in the NAc and medial PFC.⁸⁵ This study also found age-related differences in DA metabolism in rats.⁸⁵ Finally, DA systems in the adolescent rat display significant regenerative plasticity following neurotoxin administration.^87,88 In toto, these studies may indicate a functional increase in mesostriatal DA activity during adolescence,^18,89 though the region and time of these increases seem to differ from species to species.

Adolescent developmental maturation is also seen in the cannabinoid, glutaminergic, gamma-aminobutyric acid (or GA-BAergic) and serotonergic (5-HT) systems. Cannabinoid systems regulate mesocortical⁹⁰ and striatial DA systems,^91–93 and reach functional maturity in rats during adolescence.^94–96 Changes in the cannabinoid system may influence motivational learning, with preclinical studies highlighting the complex role of cannabinoid input on mesolimbic DA.^97–99 The behavioral effects of agonists for a specific glutamate receptor, the N-methyl-D-aspartate (NMDA) receptor, appear to peak late in the pre-adolescent period in rats.^100,101 This coincides with greater NMDA agonist sensitivity.¹⁰² NMDA receptor binding peaks differentially by subtype,¹⁰³ and glutaminergic inputs to the PFC decrease slightly during adolescence.^83,84

The nature of GABA and 5-HT alterations during adolescence are not as well established. GABA inhibits NAc activity and opposes the modulating excitatory effects of glutamate.¹⁰⁴ GABA receptors achieve maturity in adolescence,¹⁰⁵ with increased responsiveness of GABAergic systems linked to stress in animal models.^106,107 GABAergic input to the PFC appears to decrease strongly through adolescence in humans,^83,84 and, following a pre-adolescent peak, rat GABAergic neurons in the PFC decrease in size during adolescence.¹⁰⁸ 5-HT inhibits and opposes DA activity, particularly as DA relates to aggressive and impulsive behaviors.^109–117 Thus, DA is thought to promote motivated behaviors, whereas 5-HT is conceptualized as a break upon mesostriatal promotion of appetitive behavior. NAc 5-HT turnover is up to four times lower in adolescent rats than in younger or older rats,¹¹⁸ and 5-HT_1A receptor binding appears to decrease most dramatically in human males during adolescence.¹¹⁹ DA input to the PFC is up to three times greater than 5-HT input,¹²⁰ and PFC concentrations of a DA precursor are much greater than those of a 5-HT precursor in pubertal rhesus monkeys.¹²¹ Finally, there is evidence that early adolescent rats undergo significant 5-HT synaptic pruning in the basal forebrain.¹²²

In summary, many neurotransmitter systems demonstrate notable maturation during adolescence. This is captured in box 1 of Figure 1. DA systems display extensive pruning and plasticity with concurrent maturation of the cannabinoid, glutaminergic, and GABAergic systems during adolescence. The latter three systems all exert modulatory effects on mesolimbic DA, and it is probable that changes in these systems have consequences for the development of mesolimbic DA circuitry. Finally, 5-HT input may be underdeveloped when compared to DA NAc input during adolescence. The relevance of the adolescent DA to 5-HT ratio for behavior is explored in the next section.

IMPLICATIONS OF NEUROBIOLOGICAL CHANGES ON STIMULATORY AND SUPPRESSIVE PROCESSES

In examining the motivational learning system, the relevant circuits and neurotransmitters can be divided into stimulatory and suppressing aspects. The stimulatory substrates of this system encode for appetitive behaviors (e.g., drug-seeking), whereas the suppressive substrates encode for both regulatory and harm-avoidance behaviors (e.g., avoidance of drug use environment). This system is primarily composed of a neural network that loops from the PFC to the striatum and NAc (through the thalamus) and back to the PFC.^123,124 This system receives affect-related input from limbic structures as well as information concerning biologically motivated drives (e.g., thirst) from the hypothalamus.^71,125 In the motivational learning system, DA and glutamate are stimulatory neurotransmitters, 5-HT and GABA are suppressive, and the PFC functions as a regulatory and/or suppressive influence. As noted in box 1 of Figure 1, greater expression of the stimulatory over the suppressive aspects appears to be present during normal adolescent development.¹⁸

Chambers and colleagues¹⁸ posited that the release of NAc DA “is a principle neuromodulatory event implicated in the translation of encoded motivated drives into action, operating like a general ‘go’ signal” (p. 1045). The exact action of DA on motivational learning is unclear,⁷¹ although it appears to be related to novel stimuli and reward valence.¹²⁶ In an animal model, Hoglund et al.¹¹³ found that the administration of a DA precursor increased aggression, an aspect of stimulatory expression. DA release modulates NAc firing in response to glutaminergic input from the PFC, limbic regions, and the hippocampus.^127,128 DA also induces neuroplastic modifications in the NAc^129,130 and has been associated with learning.^131–133 Glutamatergic projections from the PFC and basolateral amygdala to the NAc core also appear to promote motivational learning.¹³⁴

Conversely, 5-HT appears to play a primary suppressive role in the motivational system. 5-HT projections involved originate in the raphe nucleus in the midbrain and project to the PFC, NAc, hippocampus, and limbic regions.⁶⁴ Taylor and Jentsch¹³⁵ found that following five days of 3,4-Methylenedioxy-methamphetamine (MDMA) administration, which is toxic for 5-HT axonal projections,^136,137 reward-related learning was impaired in rats. Unlike control rats, MDMA-administered rats performed conditioned behaviors for more than a week in the absence of cue for action.¹³⁵ Other studies have correlated impulsive or aggressive behaviors with lowered levels of 5-HT turnover,^109,112,115 and increases in 5-HT activity correlate with attenuated aggression and impulsivity in both humans and animals.^109,138,139

The PFC appears to be crucial for the regulation of motivationally driven behaviors^140–145 through a glutaminergic feedback loop with the NAc.^146–148 PFC damage has been associated with impulsiveness, dysfunctional affect,^149–152 and a higher risk for SUD.^153–155 Dysfunctional PFC input to the NAc could disrupt the motivational loop by reducing suppressive options; persistent dysfunction could impair the neuroplastic shaping of the motivational circuit, entrenching maladaptive stimulatory responses.¹⁵⁶ GABA also likely regulates motivationally driven responses through projections from the central amygdala to the VTA and through connections from the NAc core and ventral pallidum.¹³⁴

The activation of the motivational circuit appears to differ between adolescent and adult humans. In a task in which participants could either win or avoid losing money, Bjork et al.¹⁵⁷ found reduced activation in striatal and amygdalar structures during the anticipation of a gain in adolescents, relative to young adults. This implies differences in the ability of adolescents to use information to regulate motivational behavior. Ernst and collaborators¹⁵⁸ compared the neural responses of adolescents and adults on a gambling task. Adolescents had stronger NAc responses (likely stimulatory) to outcomes than adults. This study also found attenuated amygdalar responses (likely suppressive), implying that adolescents process stimulatory feedback better than inhibitory feedback.¹⁵⁸ Thus, both studies suggest that adolescents have impaired suppressive and enhanced stimulatory systems when compared to adults.

IMPLICATIONS FOR EXPERIMENTATION WITH ADDICTIVE SUBSTANCES AND DEVELOPMENT OF SUD

Given the evidence that DA, unlike 5-HT, may be close to a functional maximum,^89,118,159 adolescence appears to be marked by greater influence for activating substrates. Experimentation with addictive substances is certainly the product of many influences, one of which is likely to be this motivational imbalance. Furthermore, DA release in the NAc could interact synergistically with adolescent plasticity to promote the reinforcement-related learning of drug cues and hence continued drug use.¹⁶⁰ Glutamate-mediated learning also appears to influence behavioral adaptations to repeated drug use; for example, mice with mutated NMDA receptors did not develop conditioned place preference (CPP) or locomotor sensitization to cocaine administration.¹⁶¹ Greater NMDA agonist sensitivity in adolescence may indicate that motivated learning is altered, especially in relation to psychoactive substance use.

As mentioned previously, 5-HT systems appear to be functioning at a lesser level than DA systems during adolescence. Furthermore, GABA may attenuate drug-seeking,¹⁶² but the effects of GABA appear to decrease through adolescence. This may indicate attenuated GABA-related regulation of appetitive drives. Finally, Bjork et al.¹⁵⁷ and Ernst et al.¹⁵⁸ provide neuroimaging evidence that adolescents process stimulatory information more strongly than inhibitory or regulatory information. As the integration of diverse neural structures is crucial for the regulation of motivated drives,²² adolescents are likely not equipped to exert maximal control over their appetitive urges. Thus, it appears that a heightened ratio of the stimulatory relative to the suppressive aspects of motivation may be normal for the adolescent phase of development. This imbalance may have explanatory power for the impulsivity (#2) and experimentation with addictive substances (#6) seen across adolescents.

BEHAVIORAL FACTORS RELATING TO SUBSTANCE ABUSE IN ADOLESCENTS

Neurobiological changes during adolescence contribute to three behavioral factors that relate to the development of SUD: increases in peer affiliation, decreased parental monitoring, and risk-taking (#2–4 in Figure 1).¹⁶³ These occur across all adolescents and appear to be conserved across species.^164–166 The transition to adulthood necessitates a shift from dependence on parents and family to peer networks for support. The transition to greater dependence on peers co-occurs with increased levels of parent-adolescent conflict,^167,168 which also involves decreases in reported closeness and time that adolescents and parents spend together.¹⁶⁹ Changes in the parent-child relationship appear to be increasingly influenced by age in youngsters, potentially due to gene-environment interactions.¹⁷⁰ These changes may adversely impact parental monitoring. For even well-adjusted adolescents, this transition can be a time of heightened stress, as self-identity changes within the context of these shifting relationships. Finally, animal¹⁶⁶ and human adolescents^171,172 engage in a higher level of risk-taking behavior than during any other developmental period. Steinberg²⁴ posits that risk-taking in adolescents is a normal developmental consequence of the need for greater stimulation due to decreased reward sensitivity.

The processes mentioned in the preceding paragraph are seen across adolescents; other processes involved in SUD development are present in adolescents at high-risk (#2 in Figure 1). These include impulsivity, labile emotions, questionable decision-making, and other dysfunctional neurobehavioral processes thought to be regulated by the frontal lobes. While some combination of these is a hallmark of adolescence, those at high-risk appear to have amplified versions of these traits. Internalizing psychopathology,¹⁷³ externalizing behavior,¹⁷² and even alcohol use¹⁷⁴ are present in children and pre-adolescents; furthermore, such childhood phenomena can predict the development of SUD.^175,176 Thus, many of the risk factors for the development of SUD are present before the transition to adolescence occurs.^177,178

DECISION-MAKING, DIFFICULT TEMPERAMENT, AND INTERNALIZING PSYCHOPATHOLOGY

The evidence for decision-making as a risk factor for SUD has been mixed. Executive functioning (EF), which subsumes decision-making, is a minor risk for SUD and is mediated by aggressive behavior^179,180 and/or difficult temperament.^179,181 Overman and associates¹⁸² found that heavy alcohol and substance use decreased performance on a gambling task among adolescents; greater polysubstance use led to increasing decrements in performance. Given the design, it is unknown whether these decision-making liabilities predated or were sequelae of substance use. In a prominent comment, Steinberg²⁴ noted that many laboratory-based investigations of decision-making in adolescents may not be ecologically valid. Specifically, Gardner and Steinberg¹⁸³ have found that adolescents make riskier decisions than adults on a computer driving simulation task, but only when participants were with peers and emotionally aroused. These are common situational variables when adolescents use addictive substances.^24,183 Thus, the lack of compelling findings may be the result of measurement and design, and not the result of a weak relationship between EF and SUD.

Difficult temperament is defined as a set of traits that include negative affect, irritability, and problems with attention, persistence, and coping.¹⁸¹ While difficult temperament predicts SUD,¹⁸⁴ the relationship is mediated by aggression and deviant-peer-association.¹⁷⁹ Giancola and Parker¹⁸⁵ posited that difficult temperament was often a step on the path to SUD, but not a necessary step. Internalizing psychopathology (e.g., depressive or anxiety disorders) may also serve as a risk factor for the development of SUD, as studies in adolescents consistently find that depression and substance use co-occur.^186–189 While much of the evidence indicates that SUD precedes depressive diagnosis,^190,191 there does appear to be a subgroup for whom depressive symptoms come first.^192,193 Furthermore, Lopez and colleagues¹⁹⁴ found that post-traumatic stress disorder appeared to predict SUD. Despite some negative results,¹⁹⁵ it appears that a relationship between SUD and internalizing psychopathology exists. Aspects of this relationship, such as direction and strength of influence, however, cannot be stated given the current literature.¹⁹⁶

EXTERNALIZING BEHAVIOR

Externalizing behavioral syndromes (e.g., conduct disorder) appear to be an unmediated causal risk for the development of SUD in adolescents. Genetic research has found a pathway for SUD development beginning with parental antisocial personality disorder (ASPD) and progressing through aggressive behavior and conduct disorder (CD) in the child, resulting in SUD.^197,198 In an experiment that began with three-year-old children, a subset was classified as undercontrolled based on the presence of difficult temperament, impulsivity, hyperactivity, and aggressive, uncontrolled behavior. Under-controlled children were 2.9 times more likely to be diagnosed with ASPD, and undercontrolled males were 2.7 times more likely to have alcohol dependence at 21 years of age.¹⁹⁹ In all, Robins²⁰⁰ concluded that extensive but insufficient evidence exists to justify the idea that CD and SUD have a reciprocal causal relationship. That said, externalizing behavior may be a non-specific risk for SUD: initiation of problem behavior before the age of 15 was a risk factor for diagnosis of SUD and/or ASPD at the age of 20.²⁰¹

It is important to note that anxiety,^202,203 depressive,^145,202,204 and externalizing behavioral disorders²⁰⁵ share neurological circuits and neurobiological processes with SUD, which may speak to a common developmental mechanism. Fronto-limbic, HPA axis, DA, 5-HT, NE and/or GABA dysfunction are believed to play an important role in the development of internalizing, externalizing, and addictive disorders.^{18,202,203,205} While further examination of the comorbidity and common mechanisms of these disorders is beyond the scope of this review, it is crucial for the practicing clinician to be aware of issues in the combined assessment and treatment of other psychiatric disorders and SUD in adolescents).²⁰⁶ Treatment of the SUD and the comorbid psychiatric disorder(s) appears necessary to achieve the greatest reductions in SUD symptoms and to prevent relapse to substance use;²⁰⁷ furthermore, emerging evidence indicates that treatment of psychiatric disorders prior to the emergence of substance problems can prevent the development of SUD.^208,209

ASSOCIATION WITH DEVIANT PEERS

As mentioned previously, adolescence is a time of social transition: while affiliation with healthy peers is important for development, increased interaction with deviant peers is a risk factor for SUD development (#3 in Figure 1). Adolescents tend to befriend peers with similar identities and interests,²¹⁰ as demonstrated by a tendency for aggressive boys to affiliate with other aggressive boys.²¹¹ Friendships with deviant peers predict development of SUD among adolescents;²¹² these results have been replicated in adolescents without a familial SUD history²¹³ and across gender and ethnicity.^214–216 Ary and collaborators^217,218 proposed that increasing family conflict in adolescence discouraged parental monitoring, which led to SUD through increased affiliation with deviant peers. Indeed, affiliation with deviant peers was a more powerful predictor of SUD than conduct problems in the index subject, particularly in females.²¹⁵ The best evidence indicates that adolescent deviance and association with deviant peers amplify each other as risk factors.^219,220

Researchers at the University of Pittsburgh have investigated a construct termed neurobehavioral disinhibition, which combines measures of externalizing behavior, EF, and affect into a single factor.^221,222 It has been used to predict SUD at age 19^222,223 and marijuana use at age 16,²²³ and it is believed to be highly heritable.²²⁴ Although the concept does not account for peer affiliation and needs further investigation, neurobehavioral disinhibition is a good example of the push to aggregate neurobehavioral risk factors that can inform primary prevention. Overall, it appears that the combination of deviant peer associations and externalizing behavior problems poses the greatest risk for the development of SUD, with difficult temperament and decision-making serving as less important risk factors. This is demonstrated by McGue and Iacono,²⁰¹ though, it may be that these traits serve as non-specific risks for later diagnoses in the disinhibited spectrum rather than as specific risks for SUD.

TREATMENT IMPLICATIONS

As stated previously, best practice appears to be treatment of both the SUD and any co-occurring psychiatric disorders. It appears that selective serotonin reuptake inhibitors have efficacy in ameliorating both depressive and SUD symptoms.^225–227 For those with externalizing disorders and SUD, various medications, including divalproate,²²⁸ fluoxetine,²²⁷ and bupropion,²²⁹ have shown efficacy in reducing SUD symptoms. The effects were weaker on externalizing symptoms. All of these studies were of a pilot nature, so larger-scale randomized controlled trials are needed to fully evaluate the effects of these medications in adolescents with SUD.²⁰⁷

For behavioral treatment, there is evidence that cognitive-behavioral interventions (both in group or individual settings),²³⁰ motivational enhancement,²³¹ and family therapy²³² have efficacy in adolescents with SUD. The evidence for cognitive-behavioral intervention in a group setting is notable, given earlier evidence that group interventions with adolescents may increase high-risk behaviors, including substance use.²³³ It is also notable that improvements in familial interaction appear to co-occur with a reduction in adolescent substance use for those in family therapy.²³² Given the unique developmental transitions adolescents are undergoing, many of which specifically relate to familial relationships, family therapy could be an important component of any treatment regimen. Biglan and colleagues²³⁴ have written an informative book reviewing treatments for adolescents with multiple high-risk behaviors, including substance use; they note that many adolescents with SUD often have other externalizing problems and that such a multi-problem presentation requires multi-faceted treatment.²³⁴ Clinicians should assess patients with substance use problems for other high-risk behaviors (e.g., sexual risk behaviors) and act to treat the entire set of problem behaviors. Such treatment will likely necessitate both pharmacological and behavioral interventions.

THE NEUROBIOLOGY OF RISK FOR SUD IN HUMAN ADOLESCENTS

Neuroimaging: ERP, EEG, and fMRI

One of the best examined neurobiological risk markers (noted in #5 in Figure 1) for the development of SUDs is the P300 event-related potential (ERP). ERPs are EEG voltage changes that are temporally related to sensory, motor, or cognitive events. These electrical potentials represent the activity of large numbers of synchronous neuronal elements during information processing. There is considerable evidence that GABA and glutamate modulate, or perhaps directly cause, the expression of the P300 signal.²³⁵ The P300 ERP appears to correlate with the attention one devotes to a stimulus, and latency in attending to a stimulus.^235–237

Begleiter and collaborators²³⁸ initiated interest in P300 ERPs as a marker for SUD risk by demonstrating that the P300 response was blunted in male children of alcoholics. The P300 signal appeared to be a useful predictor of later alcohol-related disorders²³⁹ and was found to be moderately heritable.^240–242 A study of monozygotic twins also indicated that reduced P300 amplitudes predated SUD development, and persisted in twins both with and without SUD.²⁴³ The association of P300 abnormalities with SUD, however, may be a function of the relationship between P300 amplitude and non-specific disinhibition.²⁴⁴ Bauer and Hesselbrock²⁴⁵ found that P300 dysfunction might be better explained by comorbid CD, especially given the correlation between P300 signals and CD.²⁴⁶ Habeych and colleagues²⁴⁷ found that SUD was predicted both by P300 attenuations and trait disinhibition, with disinhibition mediating P300 signal and SUD. This is not to say that P300 decrements are without utility; instead, the P300 may serve as an endophenotype (or “downstream traits” between genes and the disorder²⁴⁸) to guide investigations into more specific risk factors for SUD. Furthermore, P300 signals have been used to predict treatment response in ADHD,²⁴⁹ so they could serve a similar function for SUD. Also, investigations that evaluated changes in P300 signals following treatment could lend insight into the mechanisms of interventions.

EEG studies have shown that right frontal activity during a gambling task, similar to the task mentioned above,^182,250 corresponds with risky decision-making.²⁵¹ Poor performance on these tasks may represent deficits in the participant’s ability to tie emotional state to the predicted outcome of a decision. Functional MRI (fMRI) studies of emotional regulation and processing, which are dysfunctional in individuals with negative affect, implicate bilateral activations of the PFC, the orbitofrontal cortex, and the right anterior cingulate in adolescents.²⁵² Adults activated a different pattern of structures.^253,254 Finally, on a go/no-go task (a measure of inhibition), substance-näive high-risk young adolescents had attenuated left frontal middle gyrus activity on fMRI when compared to low-risk adolescents. There was a trend toward less overall frontal activity in high-risk individuals.²⁵⁵

5-HT and DA Receptor Systems

Neurobiological measures of risk have focused on 5-HT functioning, prompted partially by the research mentioned earlier on the association of 5-HT function with appetitive behaviors.^115,117 Often, these studies have used peripheral markers, including platelet 5-HT, whole blood 5-HT concentrations, and/or platelet MAO activity, as proxies for central 5-HT function. In an examination of children with an alcoholic parent, the level of the child’s externalizing behavior was inversely correlated with that child’s level of whole blood 5-HT.^256,257 Askenazy et al.²⁵⁸ found that platelet 5-HT concentrations were higher in those with greater impulsivity, and Mezzich and colleagues²⁵⁹ reported that for adolescents with SUD, platelet MAO activity corresponded to difficult temperament. In addition, certain 5-HT transporter polymorphisms²⁶⁰ and 5-HT transporter gene combinations²⁶¹ appear to increase the risk for SUD, especially alcohol-related problems.

However, studies exploring the relationship of 5-HT to the development of SUD often contain methodological issues.²⁶² Results in animals indicate that 5-HT depletion is associated with motoric impulsivity, whereas the development of SUD in human adolescents may be associated instead with impulse choice or delayed discounting.^250,263,264 These terms denote the selection of smaller, immediate rewards over larger, delayed rewards. In all, 5-HT dysfunction may be a non-specific marker for disinhibition and negative affect rather than a specific risk factor for the development of addiction. Indeed, Oreland hypothesized that platelet MAO (as a peripheral marker of central 5-HT) may serve as a non-specific marker for CNS dysfunction corresponding to disinhibited behavior.^265,266 Nonetheless, disinhibition and negative affect are both risk factors for SUD, so 5-HT dysfunction remains relevant as a risk factor for SUD. The salience of the association between 5-HT and SUD is strengthened by data (mentioned previously) that treatment with fluoxetine can improve adolescent substance-related outcomes, with stronger effects on depressive symptoms.^225–227 It is likely that some of the improvements seen are due to the normalization of 5-HT function. That said, selective serotonin reuptake inhibitors have mixed results in treating SUD alone in adults;²⁶⁷ the efficacy of these medications is seen primarily in comorbid presentations, which may indicate that reductions in depressive symptoms aid abstinence efforts.

DA-related genotypes also appear to influence the development of SUD. In examining high-risk adolescents, the presence of A1 allele for the D2 receptor (DRD2) was related to greater use of alcohol, tobacco, and illicit drugs.²⁶⁸ The A1 allele has been linked to reduced binding to D2 receptors and lowered D2 receptor availability.^269–271 That said, measures of psychoticism and negative affect were also related to drug use outcomes in adolescents with the A1 allele, and no analyses were performed to assess mediation effects. Currently, it is unclear whether the DRD2 polymorphism serves as a specific risk factor for SUD development or whether, like 5-HT, it may be a non-specific risk factor of disinhibition and/or negative affect.

HPA Axis

Another major neurobiological measure corresponding to SUD is HPA axis function. The HPA axis is a neuroendocrine system that plays a significant role in the stress response, particularly through the downstream release of cortisol. Cortisol appears to have a permissive effect on the mesostriatal release of DA, much like drugs of abuse.^272–274 Moss et al.²⁷⁵ found that male offspring of individuals with SUD had a blunted cortisol response to a stressor. These high-risk males also had higher levels of impulsivity and externalizing behavior. Pajer et al.²⁷⁶ replicated these results. Dawes and collaborators²⁷⁷ found that the correlation of blunted cortisol response and disinhibition was greater in adolescent females than in males, with cortisol response accounting for 24% of the variance in disinhibition. Findings from our laboratory indicate that HPA activity and SUD are related, as increased HPA activity was associated with development of SUD during the follow-up period. This was particularly true when individuals had comorbid depressive and/or anxiety disorders.^193,278

These studies suggest that either HPA hyperactivity or hypoactivity can serve as a risk marker for SUD development. HPA dysfunction might mark either physiological under-reaction to a stressor, as in externalizing disorders, or overreaction to a stressor, often observed in adolescents with internalizing disorders. There is clear evidence that externalizing behavioral syndromes pose a risk for SUD development.^175,279 Internalizing symptoms pose a risk for SUD as well,²⁸⁰ possibly through HPA hyperactivity.²⁸¹ Thus, there may be two pathways to SUD involving HPA dysfunction: internalizing symptoms in tandem with HPA hyperactivity, or externalizing symptoms and HPA hypoactivity. While these may be etiological mechanisms, the limited results and speculative nature of this hypothesis highlight the need for further investigation. Concomitantly, investigations of medications with modulatory effects on the HPA system may be warranted in adolescents with SUD. Some antidepressants appear to regulate HPA expression,^282,283 and their use could be efficacious in the prevention or treatment of SUD and in the treatment of important co-occurring disorders, including major depression.²⁸⁴ Also, the opioid antagonists naltrexone and nalmefene appear to increase hypothalamic tone,^71,285 indicating potential utility in those with HPA hypoactivity and externalizing disorders.

THE CONSEQUENCES OF SUD ON THE ADOLESCENT BRAIN: ANIMAL FINDINGS

While an examination of the sequelae of substance use in adults is outside the scope of this review, the burgeoning literature concerning the neurobiology of substance use in adolescent animals is instructive. Alcohol, nicotine, and cocaine all appear to affect the adolescent brain in a distinct manner relative to their effect in adults (#6 in Figure 1). Less substantial literature supports a similar developmentally specific effect of cannabinoids,^286–288 morphine,²⁸⁹ and MDMA.²⁹⁰ Many of these unique effects predispose adolescents to greater substance use than adults, and a subset persist into adulthood, implying permanent alterations in structure and/or function.

Alcohol

After alcohol administration, adolescent rats experience less disruption of motor function^291,292 and less sedation²⁹³ than adults. Adolescent rats also experience attenuated acute alcohol withdrawal symptoms, including reductions in anxiety,^294,295 hyperthermia,²⁹⁶ and social isolation.²⁹⁴ These manifestations may be a result of less GABA-mediated inhibition in response to alcohol.^297,298 Linking human and animal findings, Schuckit and colleagues²⁹⁹ found that a lower self-reported response to alcohol in younger adolescents related to six-month drinking frequency and number of experienced alcohol problems. Compared to adult rats, adolescent rats display higher seizure thresholds and hypoactivity during withdrawal^300–302 and appear to develop tolerance to alcohol more quickly.²¹ In addition, rats administered high doses of alcohol both during adolescence and adulthood demonstrate more severe impairments in working memory³⁰³ and greater self-administration following uncontrollable stress³⁰⁴ during adulthood than rats given heavy doses only in adulthood.

The effects of alcohol on the hippocampus^305–311 in rats and humans, and on the cingulate^312,313 in rats, are greater in adolescents than adults. These effects may be mediated by the NMDA receptor. Pyapali et al.³⁰⁷ found that NMDA-mediated long-term potentiation, which is associated with synapse formation,³¹⁴ is more strongly inhibited in adolescent rats dosed with alcohol. Such functional suppression may result in impaired cortical restructuring.^315–317 Furthermore, adolescent rats administered alcohol in a binge fashion evidenced alterations in 5-HT binding not found in rats that only binge as adults.³¹⁸ Adolescent mice that are chronically administered alcohol have greater in vitro increases in VTA DA neuron firing in response to alcohol dosing and attenuated blunting of VTA DA firing following GABA administration.³¹⁹

Nicotine

Young adolescent rats exposed to nicotine are more likely to self-administer as adults than nicotine-näive adult rats and they exhibit CPP for nicotine^320,321 and self-administer nicotine more readily³²² than do older rats.¹⁹ In addition, late adolescent rats develop tolerance to the inhibitory effects of low-dose nicotine on locomotion more strongly and more quickly than adult rats.³²¹ Slawecki, Ehlers, and collaborators^323–325 found altered ERP and EEG outputs following five days of nicotine dosing in adolescent rats; these persisted into adulthood. Most strikingly, Trauth et al.^326,327 found increased expression of a set of genes (activated following cellular damage) in brain areas following nicotine administration in adolescent rats; the increased gene expression was most notable in the hippocampus.³²⁷ Nicotine dosing in adolescents produces immediate alterations in nicotinic, DA, and 5-HT transporter densities unlike those seen in adult rats.³²⁸ Finally, alterations from nicotine administration in adolescent rats that persist into adulthood have been seen in nicotinic,^19,329 DA,^330,331 and 5-HT systems.^332,333

Cocaine and Other Stimulants

Many,^334–336 but not all,³³⁷ studies have found that adolescent rats exhibit locomotor sensitization in response to cocaine dosing. Ehrlich and collaborators³³⁸ found that adolescents demonstrated greater increases in NAc levels of ΔFosB expression (implicated in neural plasticity) than adults following chronic cocaine dosing. Furthermore, Kosofsky and colleagues³³⁹ found that unique groups of genes were activated in the striatum following cocaine administration at different developmental periods. The genes involved have been implicated in regulation of the cell cycle and cell death,^340,341 neural plasticity,^342,343 and addiction.^344,345

Research on other psychostimulants found that young adolescent rats did not develop sensitization to methylphenidate, unlike adults.³³⁶ Dosing with d-amphetamine did not produce CPP in adolescent mice, unlike adult mice,³⁴⁶ but adolescent dosing does increase sensitization in adulthood.³⁴⁷ Bolanos et al.³⁴⁸ administered methylphenidate to young adolescent rats and found disruptions in preference for reinforcers and in stress sensitivity in adulthood. Adriani and Laviola³⁴⁶ found that adolescent mice responded more often during the non-reinforced portion of a reinforcement schedule than did adults following d-amphetamine administration. Adolescent rats given methylphenidate did not differentiate well between a novel and familiar object,³⁴⁹ and late adolescent rats displayed learning deficits following methamphetamine administration.³⁵⁰ Finally, chronic administration of d-amphetamine in adolescent rats increased ΔFosB expression in the NAc and striatum, a result not seen in adult or pre-adolescent rats.³³⁸

Thus, the literature on the neurobiology of substance abuse in adolescent animals indicates that adolescents differ from adults in response to nearly all drugs of abuse; perhaps more importantly, adult animals evidence alterations in function following the administration of alcohol, nicotine, or stimulants during adolescence (all #6 in Figure 1). Furthermore, there is evidence that the use of nicotine^351–353 or methylphenidate²⁰ in adolescent animals alters adult responses to cocaine and other psychostimulants. Taken together, especially in light of the evidence for greater neural plasticity of adolescents (#1 in Figure 1), these results raise questions about the ability of psychoactive substances to alter development and increase substance use.

SUMMARY

Adolescence is a time of several transitions, particularly involving maturational changes in the CNS; foremost among these are synaptic pruning, myelination, and neurotransmitter system modifications. Overall, the CNS structural and functional evidence indicates that adolescents have greater neurobiologically based tendencies for risk-taking with attenuated suppressive and regulatory controls on behavior (as depicted in #1 in Figure 1). Adolescent transitions often feature strong affect and stress, and adolescents may be more adversely affected than adults by stressful and/or affect-laden situations due to functional neurobiological immaturity. Decreases in parental monitoring during adolescence (#2 in the figure) can allow for experimentation with substances and affiliative relationships between disinhibited peers. These friendships appear to create an amplifying feedback cycle that increases both SUD risk and disinhibition levels (#3). Disinhibiton and other less robust risk factors (e.g., EF liabilities and difficult temperament) are likely to be the expression of both normative (#1) and non-normative neurobiological changes in those at high-risk for SUD (#5).

The abnormal neurobiological markers of those at risk for SUD development seem to correspond most clearly to disinhibition and/or negative affect. While the evidence for an unmediated pathway from these traits to SUD is limited, these neurobiological markers can serve as useful endophenotypes associated with SUD. Once adolescents begin substance use (#5 and #6 in the figure), the evidence indicates that they are more vulnerable to the effects of many substances of abuse. Most likely, this vulnerability is mediated by the heightened neuroplasticity of adolescents and the effects of stress. Substance-induced neurobiological alterations likely strengthen drug-use behaviors. Neuroplasticity-mediated increases in drug use accelerate neurobehavioral dysfunction, which accelerates substance use, eventually resulting in SUD and continued maladaptive neurobiological alteration.

FUTURE DIRECTIONS

While there are concerns about the methods used to estimate the heritability of various types of SUD,³⁵⁴ these do not obscure the fact that genetic factors play a prominent role in their development.^355,356 Stallings et al.³⁵⁷ found six genomic markers with high correspondence to SUD, which can be linked to various proteins and (eventually) functions for these products. Other research has highlighted DA^261,358 and 5-HT²⁶⁸ gene-related polymorphisms linked to externalizing behavior and SUD; it seems that the identification of specific genes or polymorphisms could be useful to narrow the potential neurobiological targets for intervention.

Further investigation of the contributions of 5-HT, the HPA axis, and the frontal-striatal-thalamic neural circuit is also needed. While the results on 5-HT have been somewhat problematic, it appears to be one of the primary suppressive neurotransmitters. Given its role in behavioral inhibition and other psychiatric disorders (e.g., depressive disorders), there is evidence that it plays a role in drug use. Further investigation into the developmental course and effects of 5-HT (along with the glutaminergic, GABAergic, and cannabinoid) systems are required as well. Such research could result in targets for pharmacological intervention in a fashion similar to 5-HT in depressive disorders. Research into the interaction of the HPA axis and psychopathology is also very promising, not least because of the ability of researchers to quantify HPA reactivity in stress paradigms. There is some thought that medications with regulatory effects on HPA function (e.g., mifepristone) may help stress-sensitive adults with SUD; if true, attempts should be made to extend the concept to adolescents. Finally, neuroimaging findings that implicate frontal-striatal-thalamic dysfunction are useful in linking neurobehavioral findings to specific brain areas. Further clarification is needed on the contribution of the hippocampus, striatum and limbic structures to SUD.

In addition to 5-HT and the HPA axis, other pharmacological targets could include the glutaminergic (especially NMDA-related aspects), GABAergic or cannabinoid system. Neuromodulators such as enkephalin or dynorphin could also serve as targets, pending clarification of their roles in addiction. Despite a lack of efficacy for DA agents in adults,⁷¹ the significant developmental differences between adolescents and adults may indicate that DA agents could have differential efficacy in adolescents. In addition to pharmacological intervention, behavioral interventions need to be further studied and utilized. While there are many SUD-specific interventions, interventions targeting disinhibition may be most effective in combating both SUD and risky behavior. Such an intervention could improve the neurobehavioral skills of individuals at high-risk for substance use. There is some evidence that interventions can reduce disinhibited behavior,^359,360 which calls for further investigation of programs aimed at risky behavior across domains.²³⁴

The concept of intervening on disinhibition is based on indications that there are few, if any, necessary or sufficient risk factors for the development of SUD. It appears that both multifinality and equifinality are at work in the development of SUD, clouding the causal picture: many different, separate paths seem to lead to SUD (equifinality) and the traits or paths involved lead to other disinhibited outcomes (multifinality) as well. It is essential that investigators continue to seek risk factors with greater specificity for the development of SUD. While current risk markers (e.g., P300, 5-HT, HPA dysfunction) are useful as warning signs of disinhibitory liabilities, the specific factors that lead to SUD are unknown. While such specific factors may not exist (or, more likely, may not be currently detectable), negative results are useful to rule out potential risk factors for the development of SUD.

The clearest area of investigative need is translational research, which bridges the findings concerning normal development, animal models of SUD, and disinhibition. While each area of research is compelling, the capability of investigators to link findings together remains problematic. Without translational research, the development of pharmacological and psychosocial interventions will be haphazard. The benefits of integrative research are compelling: targeted combinations of medication and psychosocial interventions that maximize the benefits to individual patients based on unique traits and risk patterns.