While attempts to identify these traits have largely been spurred by their potential for gene discovery, the identification of endophenotypes has enabled our understanding of the increasingly apparent co-morbidity between disorders, as well. As will be discussed at depth in the sections that follow, certain quantitative traits may be stable indicators of neural systems-level dysfunction that represents risk for multiple, partially-overlapping disorders at once, providing a mechanistic explanation for the apparent co-morbidity. This review argues that this is true for two clearly associated psychiatric disorders: substance abuse/dependence and attention deficit/hyperactivity disorder (ADHD).

Taken individually, these two psychiatric conditions can have a profound impact on the long-term functional capacities of an individual. As both of these syndromes often appear first prior to full adulthood, their consequences can be pervasive and life-long, by altering the trajectory of an individual during the period of most significant biological, behavioral and social development. What is particularly concerning is that these two syndromes co-occur in individuals at a rate much higher than that predicted by chance alone; in other words, individuals are often “co-morbid”, in that they suffer from symptoms of both disorders at once (Gordon et al. 2004; Wilens 2004).

ADHD is an early-onset disorder that is behaviorally identified by impulsive actions (trouble waiting turns and disruptive behavior), hyperactivity (fidgety) and inattention (difficulty focusing, distractibility, poor organizational skills and forgetfulness) (American-Psychiatric-Association 1994). Other cognitive impairments (poor working memory, executive function impairments) are common in ADHD patients and may represent key indicators of genetic liability to ADHD (Aron and Poldrack 2005; Barkley 1997; Castellanos and Tannock 2002; Doyle et al. 2005; Nigg et al. 2004), although they are not diagnostic features. The consequences of the constellation of symptoms and deficits are, oftentimes, academic and/or occupational problems.

On the other hand, substance abuse involves the maladaptive, non-medical use of illicit substances that leads to functional impairments and/or to undue hazards or risks (“driving under the influence”); additional physiological criteria (namely, the presence of tolerance or withdrawal) differentiate substance dependence from abuse (American-Psychiatric-Association 1994). Conceptually, the critical factors that identify “drug addiction” include pre-occupation with drug-seeking and —taking, despite knowledge of the associated risks and despite repeated attempts to stop and the functional consequences (socially, occupationally and otherwise) of drug use.

Though the phenotypic manifestations of ADHD and substance dependence seem considerably divergent (e.g., impulsive speech and scholastic impairment versus drug tolerance and compulsive drug-seeking), substance abuse/dependence disorders and ADHD are syndromes with long-recognized relationships (Figure 1). First of all, the addiction phenotype is a pervasive form of impulsive drug-seeking and —taking behavior (Jentsch and Taylor 1999; Robinson and Berridge 2003), indicating that it shares phenotypic aspects with ADHD. Second, ADHD, when left untreated, is a significant risk factor for the later development of substance abuse or dependence (ADHD→substance abuse/dependence). Third, individuals diagnosed with substance abuse/dependence often exhibit symptomatic features of ADHD, and studies from animal models (see Section 4.2) indicate that these effects could plausibly be caused by drug intake (substance abuse→ADHD-like traits). Finally, ADHD is often effectively treated with the very drugs that support addictive behavior (e.g., stimulants), albeit at doses and via routes of administration that do not typically produce the required brain levels of the drug that support reward and abuse (Volkow and Swanson 2003).

The many aspects of the relationship between ADHD and substance abuse have led to various neural and psychological explanations that attempt to explain their apparent co-morbidity; many proposals have focused on a common pattern of dysfunction within circuitry associated with motivational and cognitive processes. For example, several investigators have proposed that ADHD can be characterized as a state of aberrant motivational and reinforcement processes, enhanced sensitivity to delay, inability to properly allocate and sustain attention, poor motor planning and impaired executive functions, each resulting from hypofunctional dopamine system (Sagvolden et al. 2005). Correspondingly, some descriptive theories describe substance abuse/dependence disorders as resulting from drug-induced dysregulation of systems involved in reward and motivation (Koob and Le Moal 1997; Berridge and Robinson 2000) and executive control over reward-related behavior (Jentsch and Taylor 1999). Therefore, while there are clearly components of ADHD not shared with substance abuse problems (and vice versa), there is considerable empirical support for a partially overlapping set of problems with reward sensitivity, motivation and cognitive control.

A lack of cognitive control over behavior is likely to directly underpin the impulsive behavior that is a cardinal feature of ADHD and substance abuse/dependence. In naturalistic settings, children with ADHD exhibit difficulty suppressing situationally-inappropriate behavior, and it is the consequences of these control failures that lead to the disruptive behaviors that characterize the disorder and contribute to the scholastic deficits. Furthermore, addictive disorders critically include a failure of effective, voluntary control over reward-directed behavior (Jentsch and Taylor 1999).

Because multiple dimensions of psychological dysfunction are found in these disorders, it does not immediately follow that the impulsive behavioral patterns in ADHD and substance abuse stem from an empirically-measurable lack of cognitive control; direct investigation of this possibility, using laboratory measures, is necessary. A variety of tests have been used to evaluate the ability to stop or change responses in these clinical populations, and together, these different experimental tasks have provided convergent evidence for a substantial deficit in response inhibition in ADHD and substance abuse. One dimension of cognitive control that has been the focus of considerable study is response inhibition. Response inhibition encompasses the ability to adaptively suppress behavior when environmental contingencies demand it. Laboratory measures of response inhibition normally involve the establishment of a response that becomes the default (“pre-potent”) response. Each of the empirically-validated tasks (see Table 1) incorporates situational requirements for inhibiting, stopping, delaying or modifying the pre-potent response. The nature of control exerted over the pre-potent response can vary widely (e.g. stopping an ongoing behavior, inhibiting responding when reward is no longer delivered, eliminating inappropriate or excessive responding, or inhibiting responses to a previously rewarded stimulus); additionally, the cognitive processes response control may vary, as well (e.g. motor inhibition, ability to bridge a delay, ability to shift responding to new stimuli, ability to weigh magnitude of reward effectively). Importantly, the measures are not suggested to index a singular, invariant construct, but they do all appear to be procedures that allow one to quantify aspects of neural systems dysfunction that occurs in ADHD and substance abuse/dependence.

Using laboratory tests of response inhibition and cognitive control, patients with ADHD have been shown to exhibit difficulties with withholding, stopping or changing an established response (Aron et al. 2003; Casey et al. 1997; Chamberlain et al. 2007; Clark et al. 2007; Itami and Uno 2002; Kuntsi et al. 2005; Schachar et al. 1995), and these effects are generally associated with physiological dysfunction within a network that includes inferior frontal cortical regions (Aron and Poldrack 2005; Casey et al. 1997; Clark et al. 2007; Itami and Uno 2002). With respect to addiction, deficits in response inhibition have been found in patients diagnosed with drug abuse and dependence, particularly those that abuse stimulants (Ersche et al. 2008; Fillmore and Rush 2002; 2006; Monterosso et al. 2005). As in ADHD, anatomical and physiological dysfunction within ventrolateral frontal cortex is associated with these deficits (London et al. 2004; Thompson et al. 2004), suggesting that a common neural adaptation within this part of the brain likely mediates the deficiencies in inhibitory control function in both disorders.

Neuroimaging studies have revealed a common pattern of brain dysfunction that extends beyond anatomical and functional abnormalities in ventrolateral prefrontal cortex. Molecular imaging studies have demonstrated that both ADHD and substance-dependent individuals have altered dopaminergic function and production, particularly in striatal regions (Ernst et al. 1998; Heinz et al. 2005; Ludolph et al. 2008; Martinez et al. 2007). Beyond these dopaminergic alterations, both disorders show a consistent pattern of lower gray matter density in prefrontal regions (Matochik et al. 2003; Semrud-Clikeman et al. 2006) and striatal areas (Castellanos et al. 1994; Jacobsen et al. 2001). Functional imaging has also demonstrated hypoactivation of the anterior cingulate when performing a response inhibition task in both ADHD and substance dependent individuals (Hester and Garavan 2004; Leland et al. 2008). Together, these data indicate a shared neural dysfunction and dysregulation that may contribute to the shared behavioral deficits, indicative of a parallel neuronal pathway. Although the relationship between these functional, anatomical and biochemical alterations is not well understood, the fact that similarities exist beyond the behavioral output substantiates the claim that a shared neural pathway exists between ADHD and substance abuse.

Amongst neuropsychiatric disorders, ADHD is somewhat unique in that the available pharmacological treatments, while not without side effects, are remarkably effective at controlling symptomatology (Arnsten 2006b; Biederman et al. 2006). Additionally, methylphenidate and atomoxetine lessen deficits of inhibitory control in ADHD when given at therapeutically effective doses (Aron et al. 2003; Chamberlain et al. 2007; Scheres et al. 2003; Tannock et al. 1989). Pre-clinical studies have provided clues as to the neurotransmitter systems that mediate its effects on response inhibition measures. Methylphenidate and amphetamine (both of which non-selectively increase monoamine output in brain) have mixed effects on response inhibition tasks in rats that vary depending upon dose, route of administration and procedure (Cardinal et al. 2000; Cole and Robbins 1987; Eagle et al. 2007; Richards et al. 1999). On the other hand, atomoxetine, a selective norepinephrine reuptake inhibitor, appears to consistently improve response inhibition in a variety of pre-clinical measures (Robinson et al. 2007; Seu et al. 2008). Selective norepinephrine transporter inhibitors differ from traditional stimulant treatments in that stimulants increase extracellular dopamine levels in the striatum, while atomoxetine does not (Bymaster et al. 2002). Notably, however, both stimulants and atomoxetine both increase dopamine and norepinephrine in prefrontal regions (Berridge et al. 2006; Bymaster et al. 2002). These similar prefrontal monoaminergic effects of different drug classes suggest a critical role of prefrontal dopamine and norepinephrine in regulating and recruiting the neural systems that are believed to be critical for inhibiting behavior.

What is less clear, however, is the relationship between early, effective treatment of symptoms of ADHD and later risk for substance abuse disorders. Pre-clinical studies have suggested that developmentally-early treatment with stimulant medications used to treat ADHD reduce sensitivity to addictive drugs in adulthood (Andersen et al. 2002; Mague et al. 2005), but see also (Brandon et al. 2003). These results are seemingly congruent with recent prospective studies indicating that early methylphenidate treatment in ADHD does not increase, and may actually decrease, risk for substance use disorders (Biederman et al. 2008; Mannuzza et al. 2008). Because of the potential reductions in substance abuse risk associated with effective treatment of ADHD, it is of empirical interest to further determine whether clinical improvement in treated patients tracks along with effective modulation of the deficits in response inhibition (Nigg et al. 2006); if that were the case, it would strengthen support for the idea that the relationship between ADHD and substance abuse depends upon an aberrant response inhibition mechanism.

Virtually nothing is known about the pharmacological regulation of response inhibition deficits in substance abuse. Theoretically, effective treatments for ADHD may be expected to accomplish this effect; however, the abuse liability of methylphenidate and amphetamine make them practically problematic in the treatment of substance abuse. On the other hand, atomoxetine lacks abuse liability (Michelson et al. 2003), but its effects on response inhibition have not yet been evaluated and/or reported in substance abuse. If effective at modulating these deficits in substance-dependent persons, atomoxetine would represent an important tool in determining whether agents that lessen response inhibition deficits could be expected to enable voluntary cessation of drug intake, as is hypothesized by earlier models (Jentsch and Taylor 1999).

An important, albeit limited effect-size, gene that has received attention for its relationship to ADHD and substance abuse is DAT1 (Faraone et al. 2005), the gene that encodes the dopamine transporter. A variable number tandem repeat polymorphism in the 3′-untranslated region of the gene has been repeatedly associated with ADHD (Cornish et al. 2005; Gill et al. 1997; Lee et al. 2007b; Roman et al. 2001; Swanson et al. 2000) and, more recently, with cocaine abuse and inflexible smoking behavior (Guindalini et al. 2006; Stapleton et al. 2007). Notably, because a significant association between the DAT1 risk genotype and response inhibition remains after controlling for degree of ADHD symptoms (Cornish et al. 2005), the proximal effect of the risk genotype may lie at the level of response inhibition, not disorder severity. It therefore appears to be the case that certain genes influencing risk for both ADHD and substance abuse behavior may do so by mediating a common endophenotype of poor response inhibition, consistent with the assertion that this behavioral mechanism lies at the heart of the comorbidity.

Neuroimaging studies are needed in order to specify the molecular mechanisms that co-vary with individual variation in response inhibition and to determine whether these biological phenotypes will ultimately be useful as quantitative estimates of genetic liability for ADHD and/or substance abuse or dependence. For example, it is known that drug-dependent individuals show lower D2 availability (Volkow et al. 2001), and cocaine self-administration in non-human primates decreases D2 receptor availability (Nader et al. 2006). For example, a recent study in rats has suggested that low ventral striatal D2/D3 receptor availability is a predictor of disinhibited responding and susceptibility for drug taking (Dalley et al. 2007); remarkably, low nucleus accumbens D2/D3 receptor availability predicts poor ability to cease an on-going response, as measured by the stop signal reaction time task, in humans (London et al. 2007). Notably, recent pharmacological studies further directly tie low D2/D3 receptor function to response inhibition, as measured by reversal learning, in monkeys (Lee et al. 2007a). Therefore, it appears that alterations in D2 receptor levels may be linked to response inhibition in a manner that confers risk to substance abuse/dependence, while chronic exposure to drugs of abuse simultaneously encourages disinhibited responding via similar D2 mechanisms. Further investigation of modifications in monoamine signaling in brain in human and animal subjects may expose additional mechanisms of relevance to the association between response inhibition and disorder phenotypes.

Finally, investigating the extent to which naturally-occurring variation in impulsivity is a risk factor for ADHD- and addiction-like traits, more work should focus on exploring the consequences of this natural variation in animal models, rather than focusing on altering the physiology of subjects to make them useful for research purposes. Recent studies in both rats, monkeys and humans have shown that natural variation in impulsive responding is predictive of a range of cognitive control-related mechanisms, including working memory maintenance and updating (Cools et al. 2007; Dellu-Hagedorn 2006; James et al. 2007). Animal models may be particularly useful in understanding the genetic and neurochemical determinants of poor response inhibition and its consequences; for instance, variation in the DRD4 gene associates with impulsive behavior and associated cognitive deficits (James et al. 2007), precisely as it may do in humans (Lynn et al. 2005). The identification of animal models with both genetic and phenotypic variation nearly identical to that in humans suggests that there are new tools for understanding the genomic determination of complex behavioral phenotypes in humans.