The COMT Val158 allele is associated with impaired delayed-match-to-sample performance in ADHD

Attention-deficit/hyperactivity disorder (ADHD) is a common neuropsychiatric disorder, characterized by age-inappropriate symptoms of inattention, motor over-activity and impulsiveness, observed before the age of seven [1]. The disorder has an estimated prevalence of 3-8% in school-aged children [2] and causes significant lifetime academic, social and occupational impairment [3]. Family, twin and adoption studies suggest a significant genetic contribution to ADHD, with heritability estimates between 70-90% [4]. Despite this strong genetic loading mapping specific genes has proven difficult, in part due to the heterogeneous clinical presentation of ADHD. It has recently been proposed that cognitive endophenotypes, such as working memory ability, may increase the ability to detect subtle genetic effects by providing an index of neurobiological processes that are more closely related to the products of gene expression than diagnostic categories [5].

Working memory enables the temporary maintenance, updating and manipulation of relevant information in a limited capacity cognitive system and has a significant genetic component, with heritability estimates ranging from 33-49% [6]. Working memory processes have long been implicated in theoretical models of ADHD and a growing literature confirms the robust nature of working memory deficits in the disorder [7].

Given that working memory ability permits internal representations of information to guide decision-making and behaviour, impairments in working memory are likely to have important functional consequences for many higher-order processes including language, planning and goal-directed behaviour [8]. Consequently, measures of impaired executive function, including working memory, have been linked to increased risk for grade retention and decreased academic achievement in children with ADHD [9]. Understanding the genetic determinants of individual differences in working memory ability in children with ADHD may therefore have important prognostic value.

Working memory is sub-served by a broad neural network involving the ventrolateral and dorsolateral prefrontal cortices, the striatum and the inferior parietal lobes [10, 11]. However, within this network there is mounting evidence for a critical role for the prefrontal cortex, particularly for visuospatial working memory [12]. Within the prefrontal cortex spatial working memory ability is strongly mediated by the catecholamines dopamine and noradrenaline with both displaying an inverted U-shaped function whereby either too much or too little of either will result in sub-optimal working memory performance [13, 14].

Given the sensitivity of working memory ability to catecholamine levels, genes affecting these systems in prefrontal cortex may have utility for explaining individual differences in working memory ability. One of the most widely studied genes of relevance to prefrontal cognition is the catechol-O-methyltransferase (COMT) gene. COMT produces an enzyme that breaks down catecholamines, thus clearing them from the synaptic cleft. COMT is the primary mechanism of dopamine clearance in prefrontal cortex [15], in part because other regulators of synaptic dopamine, such as the dopamine transporter (DAT) are sparse in this region. A single nucleotide polymorphism (SNP: known as Val158Met or rs4680) comprising a guanine (G) to adenine (A) mutation results in an amino acid substitution of methionine (met) for valine (val) in enzyme synthesis. The more thermostable val allele is associated with greater dopamine degradation and hence less synaptic dopamine than the less stable met allele [16]. This change in thermostability may have consequences for working memory ability. The val rather than the met allele has been associated with reduced prefrontal cortical activation in functional neuroimaging studies of verbal working memory in healthy adults [17], and with poorer working memory performance in both adult populations [18, 19] and in healthy children [20]. It is reasonable therefore to hypothesize that working memory deficits in ADHD could be underpinned, at least in part, by the increased turnover of prefrontal dopamine that is associated with the val variant. Nevertheless, further examination of other genes of the catecholamine system that could also influence working memory is now also warranted.

To date, studies investigating the COMT variant as a risk factor for ADHD have been inconclusive. The majority of family-based studies that have examined the COMT polymorphism in ADHD have found no significant association, while associations with both the met and val alleles have also been reported [21]. Only a few studies have specifically investigated the relationship between COMT genotype and cognitive variables in ADHD populations and the results are inconclusive [22–24].

One explanation for the lack of consistency reported in the ADHD literature might be the varied characteristics of the tasks used to assay prefrontal function. There is evidence that prefrontal cortical dopamine is particularly important in the updating and stabilization of representations in working memory [25]. We therefore hypothesized that any associations with COMT genotype would be most pronounced in tasks requiring the active maintenance of a stable representation rather than the dynamic updating and manipulation of items stored in working memory.

The present study explored working memory in children with ADHD across a number of tasks that differed in the demands they placed on the necessity for maintenance of a stable representation versus dynamically updating working memory representations. Moreover, here we focus on non-verbal working memory tasks as previous research has suggested larger effect sizes for spatial, as compared to verbal, working memory tasks in ADHD [7]. We also explored for the first time the relationship between working memory ability in ADHD and allelic variation across a range of catecholamine genes, including those for the noradrenaline transporter (NET1), the dopamine D4 and D2 receptor genes (DRD4; DRD2), the gene encoding dopamine beta hydroxylase (DBH) and COMT. Each of these genes has been implicated as potentially increasing genetic susceptibility to ADHD or has been related to executive function ability [26] but, there has not been a systematic investigation of their association with spatial working memory phenotypes in a large ADHD cohort.

A robust literature demonstrates that working memory ability is impaired in individuals with ADHD [7]. Working memory is known to be reliant upon prefrontal catecholamine levels [14], with COMT being an important regulator of this system [15]. Here, for the first time in children with ADHD, we explored the relationship between allelic variation in a broad set of catecholamine genes, including COMT, and measures of working memory.

In children with ADHD there was a significant relationship between COMT genotype and performance on the DMTS task. Val/val homozygotes had lower percent correct scores compared to those who carried at least one met allele. There was no relationship between COMT genotype and any of the other working memory measures investigated. Performance on a working memory task requires the co-ordination of a number of abilities: information must be accurately encoded into working memory, maintained and/or manipulated depending on task demands, and finally retrieved. COMT genotype may be related to one or more of these sub-processes in children with ADHD.

Bruder et al. [43] investigated the relationship between COMT genotype and performance on a number of working memory tasks in a healthy adult population. They found that COMT genotype was related to performance on a Letter-Number Sequencing Task, but not to performance on a Spatial Delayed Response task, n-back task, or Wisconsin Card Sorting Task. The authors argued that the Letter-Number Sequencing task was more taxing of executive processes and thus, that COMT genotype is more closely related to the manipulation of items in working memory than to maintenance.

The present results suggest an alternative interpretation; here there was an association between COMT and performance on the DMTS task, which relies on successful working memory maintenance, but not with the SP or SWM tasks, which place greater executive demands on the updating and manipulation of items. Within the DMTS task there was no interaction between COMT genotype and increasing working memory delay, suggesting that COMT genotype was not related to task difficulty. Similarly, there was no relationship between COMT genotype and performance on the simultaneous matching condition of the DMTS task, which provides a measure of perceptual accuracy. Therefore, it may be concluded that the relationship between COMT genotype and DMTS performance may be most sensitive to either impairments in the encoding of a stable representation into working memory or in retrieval stages.

One potential explanation for our pattern of results may be found in the tonic-phasic hypothesis of dopamine regulation [44]. According to this hypothesis dopamine action is orchestrated through the co-ordination of tonic and phasic states. The COMT met allele increases tonic dopamine transmission, which regulates the stability of cortical activation states. Tonic dopamine stimulation has thus been hypothesized to be important for maintaining stable representations in working memory. In contrast, phasic dopamine stimulation regulates the plasticity of activation states and is believed to be important for updating and manipulating working memory representations. Our findings are consistent with the role of the met allele in enhancing the stability and maintenance of representations required for performance on the DMTS task. These findings suggest that the determining factor in observing associations between neurocognitive measures and COMT genotype may depend more on the nature of the representation required to perform the task, rather than, as suggested by Bruder et al. [43], the executive load of the task per se.

Interestingly, both chronic and acute administration of methylphenidate, a stimulant drug that alters extracellular catecholamine levels [45] has been shown to improve performance on the CANTAB DMTS task, but not on the spatial working memory task in children with ADHD [33, 46, 47]. It has been proposed that the therapeutic effect of methylphenidate is mediated through the enhancement of tonic rather than phasic dopamine release [45]. Therefore the DMTS task in ADHD may show sensitivity both to treatments that alter tonic prefrontal dopamine levels and to variation in COMT genotype. Further work is needed to fully characterize the specific aspects of DMTS task performance that are related to COMT genotype. Many of the commonly used measures of working memory, such as the n-back task and the Wisconsin Card Sorting Task, confound in time both the requirement for stable and flexible representations, and the working memory sub-process (be it encoding, maintenance, or retrieval) being engaged. Selection of tasks that allow for the compartmentalization of these processes will help in disentangling the relationship between COMT genotype and neurocognitive functioning in ADHD.

The finding that impaired performance on the DMTS was associated with the val rather than the met allele in children with ADHD supports previous research showing an association between poorer performance on executive tasks and the val allele in both healthy adults [18, 19] and children [20]. Only a few previous studies have specifically explored the relationship between COMT genotype and cognitive abilities in ADHD populations, with results conflicting. Two studies found no relationship between COMT genotype and executive function in children with ADHD [22, 23], while Bellgrove et al. [24] found that the met rather than the val variant was associated with impaired sustained attention performance in children with ADHD. There are a number of participant-factors that may influence the finding of a relationship between COMT genotype and executive function in children with ADHD, including history of medication and participant age. Perhaps more importantly, given the nature of the findings of the present study, it may be that cognitive task selection is also critically important.

The current study imposed a stringent correction for multiple comparisons that accounted both for the number of tasks studied and the number of SNPs and genetic models tested. The observation that the recessive model of the val allele survived this correction is interesting in light of other studies that have also reported recessive effects. For example, among children with ADHD val/val homozygotes have impaired task oriented performance [48] and increased antisocial behaviour [49] relative to carriers of at least one met allele. Nevertheless, we note that a nominally significant effect of the additive model (p = 0.039) was also found that might survive multiple comparison testing in larger samples. We repeated our analysis in just the sub-sample of participants who were medication free on the day of test (co-varying for gender, age, and IQ). Although this analysis did not survive correction for multiple comparisons it was nonetheless nominally significant (p = 0.02).

Since we failed to observe a relationship between COMT genotype and performance on either the SP or SWM tasks, one may conclude that performance on these tasks is perhaps influenced by other catecholamine gene variants. Nevertheless, we also failed to find evidence of association between performance on the working memory tasks and allelic variation in any of the other catecholamine genes (NET1, DRD2, DBH, DRD4) under study. There are however, a number of methodological considerations that should be considered. First, it should be noted that the low minor allele frequency associated with a number of the catecholamine SNPs may have reduced our power to detect significant associations in the relatively small sample size under study. Second, the three working memory tests used in this study were not matched for psychometric characteristics or task difficulty. In addition, although all our analyses co-varied for the medication status of the participants, it is possible that medication-related factors may have obscured associations between the catecholamine gene variants and working memory measures. The sample of children with ADHD presented here was recruited from a hospital service that specialized in assessment of children with severe behavioural disturbances and as such they presented with a high rate of conduct disorder co-morbidity. Previous research has demonstrated that the association between COMT val/val genotype and ADHD is modified by coexisting extreme anti-social behaviour and conduct disturbances [49]. However, Langley et al. [50] investigated whether impaired social functioning or executive control mediated the relationship between COMT and anti-social behaviour in the context of children with ADHD. Their analysis revealed a mediating effect of social but not executive functioning, suggesting that the association between COMT genotype and working memory ability reported herein is unlikely to be mediated by co-morbid conduct disturbance.

In summary, the results of the present study indicate that in children with ADHD the val variant of the COMT gene polymorphism is associated with impaired performance on a DMTS task that requires the stable maintenance of representations in working memory, but not with performance on working memory tasks that additionally require the dynamic updating of information. Although it remains uncertain whether the val allele of the COMT genotype confers susceptibility to ADHD, our results suggest that this polymorphism is associated with working memory impairments in ADHD children. The association between COMT genotype and working memory impairment may have important functional significance for ADHD, given that the latter has been reliably linked to poor educational and clinical outcomes.