Mapping the Development of the Basal Ganglia in Children With Attention-Deficit/Hyperactivity Disorder

Abstract

Objective:

The basal ganglia are implicated in the pathophysiology of attention-deficit/hyperactivity disorder (ADHD), but little is known of their development in the disorder. Here, we mapped basal ganglia development from childhood into late adolescence using methods that define surface morphology with an exquisite level of spatial resolution.

Method:

Surface morphology of the basal ganglia was defined from neuroanatomic magnetic resonance images acquired on 270 youth with DSM-IV-defined ADHD and 270 age- and sex-matched typically developing controls; 220 children were scanned at least twice. Using linear mixed model regression, we mapped developmental trajectories from age 4 through 19 years at approximately 7,500 surface vertices in the striatum and globus pallidus.

Results:

In the ventral striatal surfaces, there was a diagnostic difference in developmental trajectories (t=5.6, p<0.0001). Here, the typically developing group showed surface area expansion with age (increase of 0.54mm²/yr, SE 0.29mm²/yr ) whereas the ADHD group showed progressive contraction (decrease of 1.75mm²/yr, SE 0.28mm²/yr). The ADHD group also showed significant, fixed surface area reductions in dorsal striatal regions, which were detected in childhood at study entry and persisted into adolescence. There was no significant association between history of psychostimulant treatment and developmental trajectories.

Conclusions:

Progressive, atypical contraction of the ventral striatal surfaces characterizes ADHD, localizing to regions pivotal in reward processing. This contrasts with fixed, non-progressive contraction of dorsal striatal surfaces in regions that support executive function and motor planning.

Introduction

Pathology of the basal ganglia has been consistently implicated in the pathogenesis of attention-deficit/hyperactivity disorder (ADHD). The basal ganglia are core components of richly interconnected ‘loops’ that connect the cortex and thalamus, supporting many cognitive processes impaired in ADHD.¹ Dysfunction in the circuit spanning the ventral striatum (nucleus accumbens, ventral caudate and putamen) and limbic cortex is linked primarily with the abnormal processing of rewards found in ADHD² sag³. Problems with executive functions, such as cognitive control and working memory, have been tied to anomalies in the circuit linking the lateral prefrontal cortex with the head of caudate and anterior putamen.^4–7 Finally, problems in motor planning and control, another hallmark of ADHD, may be underpinned by disruptions in the links between the posterior/caudal regions of the basal ganglia and sensorimotor cortex.^8–12

Consonant with these functional deficits are findings of structural compromise. Reduced total striatal volumes have often but not always been reported, with meta-analyses of voxel-based morphometric studies localizing this loss to the putamen and head of caudate.^13–15 Two recent studies have mapped changes in the surface morphology of the striatum and found predominantly highly localized, multifocal surface contractions, undetectable by traditional volumetric techniques. In both studies, surface area contractions were found in the tail of caudate, the mid-body of the putamen, and medial, anterior globus pallidus, with less prominent expansion in the posterior putamen ¹⁶ and head of caudate.¹⁷

Despite these advances there are several major gaps in our knowledge. Foremost is the need to map the trajectory of basal ganglia development in ADHD, as all but one previous study¹⁸ have been cross-sectional and confined to a relatively narrow age range. This developmental mapping requires methods that afford a high level of spatial resolution given the recent evidence that ADHD is characterized by highly localized alterations to surface morphology.^{14, 16, 17}

A second question surrounds the possible effects of psychostimulant treatment on basal ganglia structure. A recent meta-analysis of cross-sectional studies suggested that treatment is associated with more normative dimensions, and one surface morphology study found that diagnostic differences were confined to medication-naïve ADHD participants.^{14, 17} However, no study has used longitudinal data to test for associations between the duration of psychostimulant treatment and basal ganglia development.

We thus map the development of the basal ganglia over childhood and adolescence, combining both longitudinal and cross-sectional neuroanatomic magnetic resonance data collected on a large cohort of participants.

Method

Two hundred and seventy children with ADHD participated. Details of this cohort can be found elsewhere,¹⁸ but in brief, diagnosis was based on the Parent Diagnostic Interview for Children and Adolescents,¹⁹ and Conners’ Teacher Rating Scales.²⁰ Exclusion criteria were a full-scale IQ of less than 80, evidence of medical or neurological disorders on examination or by clinical history, Tourette disorder, or any other axis I psychiatric disorder requiring treatment with medication at study entry. At study entry, 238 (88.2%) had combined-type ADHD, 17 (6.2%) had inattentive subtype, and 15 (5.6%) had hyperactive/impulsive subtype. Numbers and age of participants at each wave of scanning and their sex and IQ are given in Table 1. All ADHD participants retained the diagnosis at each assessment. The focus of the study was childhood and adolescence, and the age range was between age 4 and 18.9 years old. The proportion of time that participants were taking psychostimulant medication during the study was obtained from parental history. Conventional volumetric change in the caudate has previously been reported upon on 291 individuals in this study with a total of 544 scans.¹⁸ This study includes new data on 249 participants with a total of 325 new neuroanatomic images and reports on surface morphological change for the first time. Socioeconomic status was defined using the Hollingshead Four-Factor Index of Socioeconomic Status and IQ was estimated using an age-appropriate version of the Wechsler intelligence scales.

Table 1.

Demographic and clinical characteristics of the Attention-Deficit/Hyperactive Disorder (ADHD) and typically developing groups

	ADHD		Typically developing		Group difference
Age (yrs) Mean (SD) Time 1	N=270	9.8 (3.1)	N=270	10.3 (3.3)	t(538)=1.69, p=0.09
Time 2	N=110	11.9 (2.9)	N=110	12.6 (2.9)	t(218)=1.73, p=0.08
Time 3	N=43	13.9 (2.9)	N=45	14.4 (2.3)	t(86)=1.06, p=0.29
Time 4	N=9	16.3 (2.1)	N=12	16.0 (1.5)	t(19)=037, p=0.72
Sex	Male 184 (68%)		Male 184 (68%)		N/A
Male : Female	Female 86 (32%)		Female 86 (32%)
IQ Mean (SD)	108 (15)		108 (11)		t(500)=0.39, p=0.7
SES Mean (SD)	47 (24)		45 (20)		t(512)=1.15, p=0.25
On psychostimulant medication at study entry (yes/no)	168/99 (data missing on 3 participants)
Ethnicity					White, non-Hispanic vs. others
White, non-Hispanic	207 (77%)		219 (81%)
White, non-Hispanic	17 (6%)		8 (3%)		Χ(1)2=1.6, p=0.21
African American	36 (13%)		29 (11%)
Other	10 (4%)		14 (5%)
CBCL attention problems Mean (SD)	72 (9)		N/A
CBCL externalizing problems	65 (11)		N/A
DSM-IV diagnosed disorders, Number (%)
ODD	91 (32%)		N/A
Any mood disorder	15 (5%)		N/A
Any anxiety disorder	17 (6%)		N/A
Any tics	16 (6%)		N/A
Any learning disability	28 (10%)		N/A

Note: Missing data are reflected in the degrees of freedom. CBCL = Child Behavior Checklist; IQ= intelligence quotient; N/A = not applicable; ODD = oppositional defiant disorder; SES = socioeconomic status.

The typically developing participants were part of the National Institutes of Health (NIH) intramural project on typical brain development.²¹ The group was matched with the ADHD group on sex, IQ and number of scans. A parent of each child completed the Childhood Behavior Checklist (CBCL) as a screening tool and underwent a structured diagnostic interview to rule out other neuropsychiatric diagnoses. Both the participants with ADHD and those typically developing were recruited through advertisement and community contacts, and were predominately from the geographic area surrounding the study center. The institutional review board of NIH approved the research protocol for both the typically developing children and children with ADHD, and written informed consent and assent to participate in the study were obtained from parents and children, respectively.

Both the participants with ADHD and the typically developing participants had neuroanatomic magnetic resonance imaging (MRI) on the same 1.5-T General Electric (Milwaukee, Wisconsin) Signa scanner throughout the study. T1-weighted images with contiguous 1.5-mm slices in the axial plane were obtained using three-dimensional spoiled gradient-recalled echo in the steady state. Imaging parameters were echo time of 5 ms, repetition time of 24 ms, flip angle of 45°, acquisition matrix of 256 xh192, number of excitations = 1, and 24 cm field of view.

The basal ganglia were automatically identified using a recently developed segmentation method known as the MAGeT Brain algorithm.²² Here we define the basal ganglia as comprising the striatum (the caudate and putamen) and globus pallidus. To summarize the technique, a single atlas for the striatum and an atlas for the globus pallidus that were previously defined using a three-dimensional reconstruction of serial histological data²³ were warped to an MRI-based template. The atlases were then customized to a subset of the dataset (30 randomly selected participants between the ages of 5–18 years old) using a nonlinear transformation estimated in a region-of-interest defined around the subcortical structures.^{23, 24} This set of participants now acts as a set of templates and all other participants are now warped to these templates. This provides thirty candidate segmentations for each participant’s striata and globus pallidi. The final segmentation is decided upon using a voxel-wise majority vote, that is, the label occurring most frequently at a specific location is retained. These methods are reliable in comparisons against ‘gold standard’ manual definitions (Kappa = 0.86).²²

To determine shape, first surface-based representations of the basal ganglia defined on the input atlas were estimated using the marching cubes algorithm²⁵ and morphologically smoothed using the AMIRA software package (Visage Imaging; San Diego, CA, USA). The resulting surfaces have about 6,300 vertices per striatum and 1,300 per globus pallidus. The nonlinear portions of the 30 transformations that map each participant to the input template were concatenated and then averaged across the template library in order to limit the effects of noise and error and to increase precision and accuracy.²⁶ These surface-based representations were warped to fit each template, and as in the case of the segmentation, each of these surfaces was warped to match each participant. This yields thirty possible surface representations per participant that are then merged by creating a new surface representation of the striatum or pallidum by estimating the median coordinate representation at each location. At this point a third of the surface of each triangle is assigned to each vertex within the triangle. The surface area value stored at each vertex is the sum of all such assignments from all connected triangles.²⁷ Finally, surface area values were blurred with a surface-based diffusion smoothing kernel (5mm and 3mm for the striatum and pallidum, respectively).

Data analysis.

Morphometric group differences at study entry were determined using t tests. In the longitudinal analyses, we determined developmental trajectories for total volumes and the surface area at every vertex using mixed-model regression analysis ²⁸. This approach allows the inclusion of participants with a single observation and participants with multiple observations measured at different and irregular time periods. While most data were from individuals with repeated observations (549 observations, comprising 63% of all data), in the primary analysis we included data from individuals who had only a single scan, as these provide additional information about between-participant variation and overall curve shape. The inclusion of a random intercept per person accounts for the non-independence of the subcortical measures in participants with multiple observations. To determine the fit for each morphometric index of interest, a Likelihood Ratio test was used to determine if a model including a quadratic age terms accounted for significantly more variance at p<0.05 than a model including only a linear age term. For striatal volumes, a quadratic model was appropriate; for pallidal volumes and for surface area change at the vertex level, a linear model provided the best overall fit. In the linear model, the jth surface area of the ith individual in the kth group was modelled as

$${\text{Metric}}_{\text{ij}}=\text{intercept}+{\text{d}}_{\text{i}}+{\text{β}}_{\text{1age}}+{\text{β}}_{\text{2 group}}+{\text{β}}_{\text{3}\left(\text{age*group}\right)}+\text{e ijk}$$

where di is a random effect modelling within-person dependence, the intercept and β terms are fixed effects, and eijk represents the residual error (for further details see Supplemental Methods). Group differences in the slopes (or linear trajectory) at each vertex is given by the β3 term. T tests were used to test the significance of each parameter in the mixed-model regression, and the results were projected onto a brain template. To control for multiple comparisons, we used the false discovery rate procedure, which controls the expected proportion of incorrectly rejected null hypotheses ²⁹. We set this proportion at 5% (q=0.05). We also repeated all analyses including only participants with repeated observations.

To examine associations with psychostimulant treatment, we compared medication-naïve participants against those on psychostimulants at study entry. We also tested for an interaction between the proportion of time during the study on psychostimulants and trajectories using only those with longitudinal data.

Results

Baseline differences.

At study entry, basal ganglia volume and total surface areas were significantly reduced in ADHD (see Table S1, available online). Vertex-level analyses showed this reduction was most prominent in the lateral surface of the caudate bilaterally, extending from the head through the body of the caudate to the anterior regions of the tail (see Figure 1). Surface area reduction in the putamen was confined to a small region of the anterior-superior putamen, the mid-body, and posterior-inferior regions. In the globus pallidus, there was surface area reduction predominantly in posterior-inferior regions (see Figure S1, available online). There was relative sparing at baseline of the ventral regions of the globus pallidus and striatum, with no significant group difference.

Figure 1:

Regions where the group with attention-deficit/hyperactivity disorder (ADHD) showed a significant reduction in surface area at study entry, following adjustment for multiple comparisons. Note: A) R lateral view; B) L lateral view; c) Inferior (ventral) view.

Trajectory differences.

In both groups the striatal volumes increased with age, peaking around mid-adolescence and then beginning to decrease. Trajectories are given in Figure 2. The growth trajectories did not differ significantly by diagnosis, although there was a trend towards convergence between the ADHD and typical groups on the right (right F=2.9, p=0.053; left F=0.74, p=0.48). For the total striatal surface areas, there were also no diagnostic difference in trajectories (right F=1.9, p=0.15; left F=0.18, p=0.83). For the volumes and total surface areas of the globus pallidus, there were no group differences in trajectories (right volumes t=0.89, p=0.37; left volumes t=0.45, p=0.65; right total surface area t=1.1, p=0.29; left total surface area t=0.77, p=0.44). Adjustment for total brain volume did not alter the longitudinal results but did render non-significant the baseline differences in volume and surface area for the globus pallidus and the surface area of the right striatum (see Table S1, available online).

Figure 2:

Developmental trajectories (estimates with 95% CI) for the striatal and globus pallidus volumes and total surface areas. Note: There were no significant differences in the shapes of the curves. The group with attention-deficit/hyperactivity disorder (ADHD) is shown in red and typically developing in blue.

Vertex-level surface analyses showed a significant group difference in the trajectory of a region centered on the ventral striatum bilaterally (overall slope difference in this region, t=5.6, p<0.00001). Here, the typically developing group showed expansion of the surface area with age (increase of 0.54mm²/yr, SE 0.29mm²/yr). By contrast, the group with ADHD showed marked, progressive contraction of the ventral striatal surface (decrease of 1.75mm²/yr, SE0.28mm²/yr), shown in Figure 3. This diagnostic difference in ventral striatal development is illustrated in a time-lapse sequence (‘movie’) that shows the progression of significant group differences in this region with age (see Video S1, available online). A similar group difference was also seen in a small region of the right posterior putamen. Throughout the remainder of the putamen and caudate, there were no trajectory differences, as can be seen from Figure S2 (available online), which illustrates trajectories at points throughout the striatum. Thus, the dorsal striatal surface area reductions noted at study entry persisted into adolescence.

Figure 3:

Regions where there was a significant difference in trajectories following adjustment for multiple comparison (overall slope difference in this region, t=5.6, p<0.0001) and a graph showing the trajectories of surface area change for this region, with 95% CI). Note: ADHD = attention-deficit/hyperactivity disorder.

The pattern of results was almost identical when only individuals with two or more observations were included, when analyses were confined to the subgroup with ADHD without any comorbid disorders, and when analyses were confined to those with combined-type ADHD only (see Figures S3 and S4, available online). Males had greater striatal surface areas in both groups, although this sex effect was accentuated in the group with ADHD in a small region in the head of caudate, resulting in a significant sex by group interaction in this region (see Figure S5, available online). In the longitudinal analyses, there were no regions that showed a significant interaction between diagnosis and sex following adjustment for multiple comparisons.

In the globus pallidus, there was a group difference at a nominal level of significance in trajectories in ventral regions bilaterally (see Figure S6, available online). These group differences in trajectories did not survive adjustment for multiple comparisons.

There were no significant differences following adjustment for multiple comparisons between those on or off psychostimulants (treating medication status as a dichotomous variable) at study entry (see Table S2 and Figure S7, available online). Likewise there were no regions of significant interaction between the proportion of time on psychostimulants and the trajectories of surface area change, following adjustment for multiple comparisons (full results are given in Figure S8, available online).

Discussion

Three novel findings emerge from this study, which is the first to map basal ganglia development at a high level of spatial resolution in ADHD. First, we find a diagnostic difference in developmental trajectories of the ventral striatum throughout childhood and adolescence; this region shows progressive surface contraction in ADHD, but not in typical development. Second, multiple areas show surface area reductions in ADHD that are fixed and do not progress with age, including the head and tail of the caudate and inferior-posterior putamen. Third, we found no significant association between duration of treatment with psychostimulants during the study and the developmental morphology of basal ganglia surfaces.

The diagnostic difference in ventral striatal development is consonant with its dysfunction in patients with ADHD, particularly in abnormal reward processing. Many but not all studies find a preference for immediate over delayed rewards—even when this strategy is ultimately disadvantageous ^30–32—in patients with ADHD. Decreased activity within the ventral striatum during the anticipation and receipt of rewards is one of the more consistent functional imaging findings in patients with ADHD.^{33, 34} Atypical connectivity between the ventral striatum, the amygdala, and orbitofrontal cortex has been found in both the patterns of spontaneous resting brain activity and during tasks of reward processing.^34–36 This delineates dysfunction in a network that may contribute to motivational deficits in ADHD. Given the implication of ventral striatal dysfunction in risk for drug dependence, the hypothesis emerges that this progressive ventral striatal surface loss could contribute to the vulnerability to substance misuse found among young adults with ADHD.^{37, 38}

We find the dorsal striatum shows fixed surface area reduction in ADHD, detected in childhood and persisting into adolescence. This reduction was present in the so-called ‘associative’ regions of the striatum (head/body of caudate and anterior putamen) that receive input from and project via the thalamus to predominately lateral prefrontal cortical regions, supporting executive functions.^{39, 40} Dysfunction within this circuit in ADHD has been demonstrated by fMRI during tasks of attention and working memory.^{6, 7} The mid-body/tail of the caudate and posterior-inferior regions of the putamen also showed fixed surface area reduction in ADHD. These regions receive rich inputs from motor and premotor areas and parietal somatosensory areas.^{12, 39} Structural anomalies in this area could contribute to problems in the planning, control, and execution of motor behavior that are a hallmark feature of ADHD.^{8, 9} Hypoactivation in this circuit, specifically in the left putamen and supplementary motor area, was confirmed in a meta-analysis of fMRI studies of motor inhibitory control.⁶ The two other surface morphology studies found surface anomalies in similar regions, a notable convergence in results given the different approaches to mapping surface morphology.^{16, 17} It is important to note that these fronto-striato-thalamic circuits interact richly with one another, allowing flexible and adaptive behaviour. ⁴¹ Thus disruption in one striatal region is unlikely to be tied to a single neuropsychological deficit, but rather may have an impact on many cognitive functions.

Why do ventral striatal regions alone show diagnostically different trajectories? Nested within each cortico-striato-thalamic circuit is a direct pathway (monosynaptically linking striatal neurons with the internal globus pallidus and substantia nigra) and an indirect pathway (linking the striatum and internal globus pallidus through the external globus pallidus and subthalamic nucleus). Dopamine is thought to activate the direct pathway through D1 receptors and inhibit the indirect pathway through D2 receptors.^{42, 43} Different densities of these receptors throughout the striatum with high levels of D1 receptor in the medial caudate and limbic regions could be an important factor in contributing to the regional differences in growth we report.⁴⁴ Interestingly, the effects of action at these receptors on behaviors relevant to the symptoms of ADHD (such as behavioral inhibition) vary with striatal region. For example in rats, antagonists of D1 and D2 receptors in the dorsal but not ventral striatum have different effects on behavioral inhibition.⁴⁵ Other factors altering dopaminergic tone may also be important, such as the relatively low levels of dopamine transporters in the ventral striatum.⁴⁶ Animal studies link cytoarchitectural change with local perturbation in dopamine concentrations, and regional differences in dopamine transporter density might contribute to regional differences in growth. However, any trophic effects of psychostimulants are likely to be complex, varying by the mode of administration (chronic vs. intermittent vs. acute ^47–49) and by brain region.⁵⁰

Striatal projections to the globus pallidus are topographically organized, and thus we would expect and indeed found congruence between changes seen at the striatal and pallidal levels. At baseline, both the striatum and globus pallidus showed multifocal surface area reductions primarily in dorsal regions, with relative sparing of the ventral regions. In the longitudinal analyses, the diagnostic differences in the trajectories of ventral striatal surface morphology were mirrored by similar trajectory differences in ventral pallidal areas. In the ventral pallidum, while the typically developing group showed a clear increase in surface area with age, the ADHD group showed either progressive surface contraction or minimal change, as was found in the ventral striatum.

We found no association between psychostimulant medication and the structural development of the basal ganglia using our longitudinal data, nor any medication-related differences at study entry. While a meta-analysis of cross-sectional voxel-based morphometric studies reported more normative striatal volumes with psychostimulant treatment ¹⁴, studies using manual delineations and surface mapping techniques yield more inconsistent findings ^{16, 51, 52}. One limitation of our study is that we treated medication use as a dichotomous variable at study entry and did not take into account the amount of time on pscyhostimulants prior to study entry as a variable. We also did not consider the possible effects of non-pharmacological interventions.

How do these findings relate to other anomalies of developmental trajectories found in the prefrontal cortex? This study confirms regional heterogeneity for developmental trajectories in typical development: the prefrontal cortex and striatum appear to follow somewhat different growth patterns ⁵³. Given such regional heterogeneity in typical growth patterns, we might expect to find regional heterogeneity in the altered trajectories associated with ADHD, which complicates the integration of cortical and subcortical trajectories. Nonetheless, a unifying theme emerges in ADHD of disruption to developmental trajectories in interconnected prefrontal cortical and striatal regions. For example, we recently found that among children with ADHD that persists into adulthood, there is a fixed, non-progressive thinning during adolescence of the medial prefrontal cortex/cingulate regions that project heavily to the ventral striatum ⁵⁴. There is thus disruption at both the cortical and subcortical level in a network supporting functions frequently implicated in ADHD, including reward processing. It has also been hypothesized that the primary neuroanatomic anomaly in ADHD lies in deep structures such as the striatum, with the prefrontal cortical anomalies reflecting symptom severity ⁵⁵. This model might predict that the subcortex shows relatively fixed or even progressive anomalies, while trajectories of the prefrontal cortex may be more reflective of current symptom status, which is consonant with some of our findings. It is also possible that other methods for modelling growth, beyond the mixed-model regression we employed, could reveal a common, underlying anomaly in both cortical and subcortical growth in ADHD.

The ‘multi-atlas’ segmentation method we used has several possible advantages over methods using a single template, defined and labelled by an expert neuroanatomist. In the single template approach, the ‘labels’ of each structure are customized to a specific participant’s unique neuroanatomy through first estimating a participant-to-template nonlinear transformation and then applying the inverse transformation to the labels. These segmentations are prone to errors in the transformation estimation, irreconcilable neuroanatomical differences between the neuroanatomy of the template and the participant, and resampling errors after the application of the transformation to the labels. An approach incorporating multiple templates can mitigate these errors. For example, the use of thirty surface representations of the same structure may attenuate the impact on registration of any highly atypical features that might occur in a single template.

The inclusion criteria for the study resulted in a homogenous, severely affected phenotype at baseline; nearly all participants had combined-type ADHD and were relatively free of other major psychiatric comorbidities beyond oppositional defiant disorder. While this enhances the applicability of the results to the ‘pure’ syndrome, it limits the generalizability of the findings to children with either inattentive or hyperactive-impulsive subtype ADHD. Our participants were also drawn from a relatively socioeconomically affluent area, and a priority for future work is the inclusion of a more diverse sample. While there is evidence of sexual dimorphism in the development of subcortical structures,^{56, 57} we did not find an interaction between sex and diagnosis, although we may have been underpowered due to the preponderance of males. We did not measure reward sensitivity in our participants, although the interpretation of our ventral striatal finding benefits from the large behavioral and neuroimaging literature demonstrating its role in reward processing in individuals with ADHD. The study also used a mix of cross-sectional and longitudinal data, although the results held when analyses were confined to the longitudinal data only.

In this study, we demonstrate a highly atypical progressive surface area contraction in the ventral striatum in ADHD that contrasts with a fixed, non-progressive surface area loss in the dorsal striatum. The finding of both fixed and progressive surface alterations throughout the basal ganglia is in keeping with the concept of multiple neurocognitive deficits contributing to ADHD, and of multiple pathways to the disorder. ^{2, 58, 59}

Figure 4:

The emergence of significant group differences in surface area in the ventral striatal region from ages 8 to 18. Note: This is derived from the linear mixed model regression by re-centering age from age 8 to 18 and illustrates the progression of ventral striatal contraction in attention-deficit/hyperactivity disorder (ADHD).