The present study gives neurobiological support to the dimensional view of ADHD that typically developing children with no psychiatric diagnoses who have hyperactive/impulsive symptoms demonstrate neurodevelopmental changes resembling those found in youth with the syndrome of ADHD. Specifically, we found that the rate of cortical thinning during late childhood and adolescence was linked to the severity of hyperactive and impulsive symptoms. More symptoms accompanied a slower rate of cortical thinning in predominantly prefrontal cortical regions, mirroring the slower rate of adolescent cortical thinning seen in ADHD. This study also lends support to the concept of a perturbation of cortical trajectory as a fundamental deficit in the pathogenesis of ADHD, rather than reflecting features such as its treatment or comorbidities or the effects of nondisorder-specific functional impairments.
The localization of this effect may have dual significance. First, the regions overlap with cortical areas frequently reported to be structurally compromised in ADHD. Studies of gray matter density or thickness report deficits in the lateral prefrontal cortex, particularly in the right inferior and superior gyri (25
), as well as prefrontal medial wall deficits extending to the anterior cingulate gyrus (26
). Similarly, for some of the posterior regions showing a sensitivity to hyperactive/impulsive symptoms, structural alteration is reported, especially in the right middle/inferior temporal lobe (29
) and posterior cingulate/precuneus (25
). Second, many of the cortical regions highlighted in this study are components of the neural substrates supporting certain key cognitive processes, and impairments in these domains are in turn linked to hyperactive and impulsive symptoms. These cognitive functions include the ability to inhibit responses in order to achieve later, internally represented goals (34
). This cognitive control skill recruits a distributed cortico-striatal network with key involvement of the lateral prefrontal cortex, particularly the right inferior frontal gyrus (36
). Aberrant processing of reward and punishment, with a tendency to prefer immediate over delayed but larger rewards, has also been reported in ADHD (38
). Studies in healthy subjects implicate a neural network incorporating interactions between the ventral striatum and the limbic system, including orbitofrontal regions (40
), where we find cortical development to be linked to hyperactive/impulsive symptoms. Other investigators argue that ADHD partly results from anomalies of the spontaneous intrinsic brain activity that typifies nontask-related cognition (default-mode network), specifically its tendency to intrude into periods of active task-specific processing (41
). The default mode network in healthy subjects is thought to comprise medial (medial prefrontal cortex, posterior cingulate/precuneus) and lateral (posterior parietal) brain regions. Interestingly, anomalous resting state activity of the posterior cingulate has been reported in ADHD (42
), a region where, again, we link severity of hyperactive/impulsive symptoms to cortical change. It is noteworthy that there was no effect of hyperactive/impulsive symptoms in the superior portions of the motor strip bilaterally. We reported in a previous study that this superior region of the motor strip had the distinction of attaining a peak cortical thickness earlier in ADHD (7
) and it also did not show a diagnostic difference in the rate of adolescent cortical change. Thus, we would predict and indeed found that there was no effect of the severity of hyperactivity/impulsive symptoms on cortical change in this region. The association with symptoms was confined to the middle and inferior portions of the motor strip.
In the present study, the stepwise decrease in the rate of prefrontal cortical thinning in typically developing youth, moving from those with no symptoms to those with mild symptoms and then to those with moderate symptoms, was found to extend to youth with a diagnosis of ADHD. This is in keeping with the more severe symptoms of hyperactivity and impulsivity in the clinical group, which had slower rates of thinning than the typically developing youth, affording more support for dimensionality. In the posterior cortical region, this stepwise effect was not found because the ADHD group had rates of cortical thinning comparable to typically developing youth with mild and moderate symptoms. This is not unexpected, since we found less evidence in previous work for diagnostic differences in the properties of cortical trajectories in the posterior cortex (7
Rates of cortical thinning also showed some links with the presence of conduct problems in typically developing youth, albeit in less extensive cortical regions. Both conduct problems and hyperactive/impulsive symptoms were linked to cortical thinning rates in the right motor/supplementary area, and some degree of overlap is unsurprising given the correlation between these symptom domains. Perhaps more striking is the degree of specificity to symptom domain: the regions where cortical thinning was linked to hyperactive/impulsive symptoms were mostly spatially distinct from those linked to conduct problems. This is congruent with reports of distinct anomalies of brain activation in children with conduct disorder and ADHD (44
) and with the lack of similarity between the structural brain anomalies reported in children with conduct disorder and those most frequently reported in ADHD (45
The 48-item Conners’ Parent Rating Scale we utilized has many strengths, including its psychometric robustness, especially regarding the factor analyses used to derive the subscales for hyperactive/impulsive and conduct problem symptoms (10
). During the course of this longitudinal study, the scale was updated (47
), but in the interest of maximizing data we continued to use the 48-item version. Both the revised and 48-item versions are similar in their composition, particularly regarding the items loading onto the factor reflecting hyperactivity/impulsivity. As mentioned earlier, the version we used did not return a factor that reflects purely inattentive symptoms. Additionally, while there is evidence from latent class analyses of the 18 DSM-IV symptoms of ADHD that the dimensions of hyperactivity and impulsivity may be in part separable (48
), the Conners’ Rating Scales do not allow for independent assessment of these domains nor the possibility of partly distinct neural correlates.
We expected that any differences in cortical dynamics within the typically developing youth would be subtle, and thus we focused on the age period with the greatest data density, namely late childhood and adolescence. Previous investigations have established that the dominant effect of age is linear during this age period (51
). Further, when our typically developing cohort was divided on the basis of symptom score, there were insufficient data to examine higher-order effects of age because of a relative lack of data for this very young age group. Future inclusion of more data from a younger age group would allow consideration of more complex growth trajectory differences. Our cohort had a relatively high socioeconomic status and IQ, reflecting the self-selected nature, the relative affluence of the geographical area surrounding the study center, and the absence of any mental illness in typically developing participants, which are factors that may limit the generalizability of the findings. We also did not systematically collect measures of potentially important environmental and lifestyle factors such as diet and tobacco, alcohol, and illicit drug use.
Alterations in cortical growth rates will be only one of the many neural changes shown to reflect or drive ADHD. Many perturbations of brain structure and function have been found in the disorder, and it is unclear whether these will also show the dimensionality we report for cortical change.