Introduction
Attention-deficit/hyperactivity disorder (ADHD) is a common neurodevelopmental disorder, affecting 5–7% of children/adolescents worldwide [
1,
2]. It is characterized by an age-inappropriate, persistent pattern of inattention, and/or hyperactivity and impulsive behaviors that interferes with functioning and development [
3]. ADHD is a multifactorial disorder, with both genetic and environmental factors, as well as their interaction, contributing to its etiology [
4‐
6].
One of the most prominent neurocognitive biomarkers of ADHD is impaired response inhibition, which refers to the ability to voluntarily stop or suppress behaviors that are inappropriate for the context and/or individual goals [
7,
8]. Functional magnetic resonance imaging (fMRI) studies on response inhibition have shown that individuals with ADHD exhibit decreased activation during action cancellation and restraint compared with controls in specific brain areas: the fronto-parietal network, consisting of the prefrontal and superior parietal regions, and the fronto-striatal network, involving the prefrontal cortex and basal ganglia [
9‐
15]. By focusing on action cancellation assessed by the stop-signal task [
16], previous research has also reported that impairments in behavioral performance [
14‐
17] and aberrant neural activity associated with response inhibition [
14] are not only present in children and adolescents with ADHD but also in their unaffected first-degree relatives. This led researchers to propose impaired inhibitory control as a possible endophenotype or candidate neurocognitive biomarker of ADHD that shares familial loading with the phenotype [
18].
Family-based studies have indeed shown that both behavioral performance of response inhibition and response inhibition-related brain activity are heritable, up to 60%. Two twin studies reported significant contribution of additive genetic variance to action cancellation during the stop-signal task [
19,
20]. There has only been one genome-wide association study (GWAS) regarding behavioral response inhibition performance as assessed by the stop-signal task, but no significant loci were detected, probably due to the small sample size of 4,611 participants from a general population cohort [
21]. As for the neural correlates, there is no functional magnetic resonance imaging (fMRI) study directly investigating the heritability of response inhibition-related brain activation, but two twin studies addressing action restraint using a Go/NoGo task also reported that 50 to 60% of the variance in amplitudes of response inhibition-related-event-related potential components in adolescents and adults are attributable to genetic factors [
22,
23].
ADHD is a heritable disorder with a highly polygenic nature involving the combined effect of many genetic variants with small individual effects on the overall disease risk. The largest published GWAS of ADHD to date (comprising 20,183 cases and 35,191 controls) reported a SNP heritability of 22% and identified twelve genome-wide significant loci across the genome [
24]. This GWAS of ADHD not only confirms the polygenic architecture of ADHD [
24], but also enables the construction of polygenic risk scores (PRS) for ADHD to investigate a potentially shared genetic etiology between ADHD and cognitive and neural measures [
25]. A systematic review of the existing literature on PRS-ADHD revealed that PRS-ADHD has also been linked with ADHD traits, other externalizing behaviors, impaired working memory, and reduced brain volume [
26]. The relation between PRS-ADHD and hyperactivity-impulsivity symptoms may be partially mediated by neuroanatomical variation [
27]. Moreover, inhibitory control (as assessed by the Stroop task) has been linked with PRS-ADHD and was found to partially mediate the link between PRS-ADHD and symptoms of ADHD [
28]. A recent study investigating whether PRS-ADHD influenced attention regulation and response inhibition in ADHD reported significant associations of PRS-ADHD with reaction time variability but not with the number of commission errors during the Go/No-Go task [
29]. The aforementioned studies point to genetic sharing between ADHD and different cognitive traits and neuroimaging-derived variables and also suggest that certain (response inhibition-related) cognitive and neural processes mediate the link between genetic liability to ADHD as reflected in PRS-ADHD and ADHD symptomatology. However, to date, a possible shared genetic background between ADHD and brain activation during response inhibition has not been investigated yet.
Therefore, in the present study, we investigated whether genetic liability to ADHD (PRS-ADHD) was associated with neural activity related to response inhibition (i.e., as measured during a stop-signal task), and whether such response inhibition-related neural activity would mediate the link between PRS-ADHD and ADHD symptoms. Moreover, we aimed to expand previous evidence pointing to genetic sharing between ADHD and inhibitory control [
29] by investigating whether PRS-ADHD would be related to behavioral performance measures during a stop-signal task, as well as investigating a possible mediating role of these behavioral correlates in the relation between PRS-ADHD and ADHD symptoms, in a relatively modest sample of individuals with ADHD, their unaffected siblings, and controls (NeuroIMAGE).
Discussion
This study investigated the relationships between genetic liability to ADHD (PRS-ADHD), its core symptoms, and behavioral and functional neural correlates of response inhibition in a sample of children, adolescents, and young adults with ADHD, their unaffected siblings, and healthy controls. A higher genetic liability to ADHD was associated with higher levels of symptom severity in both symptom domains of inattention and hyperactivity-impulsivity, as well as with total ADHD symptom severity. Further, PRS-ADHD were found to be associated with slower and more variable responses to go-stimuli in the stop-signal task and with altered neural activity in several regions of the bilateral fronto-striatal network during response inhibition. We identified behavioral performance in the stop-signal task (MRT and IRT) as partial mediators of the association between PRS-ADHD and ADHD symptoms in both symptom domains; activity in the left temporal pole and anterior parahippocampal gyrus (PHG) during failed inhibition was observed to be a mediator in the relationship of PRS-ADHD with hyperactivity-impulsivity symptoms.
Our finding that PRS-ADHD were positively associated with not only total ADHD symptom scores, but also with inattention and hyperactivity-impulsivity symptom scores, is consistent with our hypothesis and previous studies [
26,
50,
51]. The explained variance by PRS-ADHD for both ADHD symptom domains was similar (3.9% for inattention and 4% for hyperactivity-impulsivity).
The significant positive associations that we found between the PRS-ADHD and latency of go responses, as indexed by MRT, and intra-individual reaction time variability, as indexed by IRT, point to an overlap between genetic effects on ADHD and MRT and IRT; it also further supports the hypothesis of increased intra-individual response variability as an endophenotype of ADHD [
52]. Elevated reaction time and greater reaction time variability in a cognitive task are among the most consistent findings in the literature of childhood ADHD [
53‐
55] and have been repeatedly observed in stop-signal task studies [
14,
55‐
58]. The unaffected siblings of individuals with ADHD have been shown to have levels of IRT intermediate between probands with ADHD and controls [
14]. Further to that, multivariate genetic analyses of ADHD cases and unaffected sibling pairs showed that cognitive impairment in ADHD related to response time (variability) during a Go/No-Go task [
59], and a link between PRS-ADHD and reaction time variability in response inhibition tasks have also been recently reported [
29,
60].
There was no significant association of PRS-ADHD with SSRT, the core measure of inhibitory control during the stop-signal task. Impaired response inhibition, as indexed by greater SSRT values, is thought to be one of the primary deficits associated with ADHD [
58,
61,
62]. Nevertheless, previous stop-signal task studies (sample sizes ranging from 45 to 170) indeed reported shorter, but also
similar SSRT values in children with ADHD compared to healthy controls [
63‐
66]. Moreover, our results are in line with a recent study that reported no association between PRS-ADHD and inhibitory control, as indexed by commission errors during a Go/No-Go task [
29] and SSRT during stop-signal tasks [
60,
67]. However, a link between PRS-ADHD and cognitive interference, measured in the Stroop task, has also been reported [
28]. These somewhat inconsistent results in the current literature might result from different samples (the previous studies were limited to individuals with ADHD [
28,
29], whereas we also included unaffected siblings and healthy controls) and/or different experimental tasks (since each inhibitory paradigm has its own measure of inhibitory control). As suggested earlier [
29,
60,
67], the genetic variants captured by the PRS-ADHD might not be directly related to the core behavioral measures of response inhibition, but rather to other response inhibition-related components (MRT, IRT) and neural activity. It is also possible that shared genetic effects between ADHD and certain behavioral correlates of response inhibition (i.e., SSRT) are more subtle than what we can detect with our current ‘base’ and ‘target’ samples. Therefore, larger future studies are needed to more robustly confirm genetic sharing of ADHD with different behavioral performance measures of response inhibition across different experimental paradigms. All in all, our findings point to the influence of genetic liability to ADHD on attentional processing during response inhibition rather than inhibitory control as such.
To our knowledge, this is the first study to investigate a possible shared genetic background between ADHD and brain activation in the response inhibition network by using individual-level PRS and fMRI data. PRS-ADHD was significantly related to activity of several regions in the bilateral fronto-striatal-thalamo-cortical network associated with response inhibition. We identified a cluster within the left fronto-insular regions and putamen, for which activity during successful inhibition was negatively associated with the PRS-ADHD. To further investigate the brain activation related to failed response inhibition, we used two separate contrasts, “
failed inhibition—go” and “
failed—successful inhibition”, which provide complementary information about the neural mechanisms underlying response inhibition (the first contrast is thought to reflect such as the engagement of the inhibitory control network and the detection of a stop signal, while the second contrast compares error processing and adjustment) [
68,
69]. The failed inhibition—go contrast revealed two clusters positively associated with PRS-ADHD, located in the left temporal pole and anterior PHG, and in the right putamen. In failed—successful inhibition contrasts, positive associations of PRS-ADHD with neural activation were found within two clusters localized in the left fronto-insular, putamen, anterior temporal, and parahippocampal regions, and in the right thalamus and basal ganglia.
Our findings regarding neural activity converge with a previous meta-analysis (607 participants; 287 ADHD cases and 320 healthy controls), which reported aberrant activation in individuals with ADHD during response inhibition for a large neural network encompassing these same areas [
12]. Moreover, decreased activation in bilateral fronto-parietal and fronto-striatal regions during the stop-signal task has also been reported in unaffected siblings of individuals with ADHD when compared to healthy controls in a previous study that also used the NeuroIMAGE sample (420 participants; 185 ADHD cases, 111 of their unaffected siblings, and 124 healthy controls) [
14]. During failed inhibition, we also identified a cluster in the left anterior temporal pole and PHG, in addition to areas of inhibition in the basal ganglia whose activities were positively associated with PRS-ADHD. The PHG with its surrounding areas, such as the entorhinal cortex and hippocampus, has been associated with post-error processing and error-driven learning strategy [
70]. The positive association between PRS-ADHD and activity in the left PHG may therefore reflect different strategies adopted by the individuals with higher PRS-ADHD for performance-monitoring and error-processing during the stop-signal task. These results, combined with our findings regarding the significant associations of PRS-ADHD with the activity of key nodes in the response inhibition network such as the prefrontal areas, anterior cingulate cortex, basal ganglia, and thalamus, suggest that common risk variants for ADHD play a role in altered neural substrates of inhibitory control in ADHD.
Our finding that MRT and IRT mediated the association between PRS-ADHD and the total ADHD, inattention, and hyperactivity-impulsivity symptom scores confirms disrupted attentional processing during response inhibition as a key cognitive variable in the context of ADHD. Individuals who had a greater polygenic risk for ADHD showed slower go responses with greater variability in response time, which in turn partially mediated the link between PRS-ADHD and ADHD symptom severity. MRT and IRT varied with regard to how much of the total effect they mediated. MRT mediated 7.7% of the total association of PRS-ADHD with total ADHD symptom severity, whereas the corresponding percentage for IRT was 15.7%. Similar patterns emerged for the inattention and hyperactivity-impulsivity symptom scores. A recent study also demonstrated that reaction time variability is associated with PRS-ADHD and also partially mediated the relationship between PRS-ADHD and ADHD traits [
60]. Furthermore, increased reaction time variability has repeatedly been reported in children and adults with ADHD [
54,
71‐
73]. Moreover, reaction time variability in particular has been suggested to be a robust and reliable feature of ADHD across stop-signal and other cognitive tasks [
71]. Thus, higher genetic liability to ADHD, as indexed by higher PRS-ADHD, might lead to the development of more ADHD symptoms somewhat more clearly through altered IRT than MRT as a possible intermediate phenotype.
As for the neural correlates of response inhibition, the association of PRS-ADHD with the severity of hyperactivity-impulsivity symptoms was partially mediated by lower activity in the left temporal pole and anterior PHG during failed inhibition. As the indirect effect of the neural activation was negative while the total effect had a positive sign, the effect of neural activation as a mediator indirectly reduced the effect of PRS-ADHD on ADHD symptoms, which may suggest a ‘suppressive’ or inhibiting (neural mediation) effect. More specifically, a subset of the SNPs that drive the direct effect of PRS-ADHD on the severity of hyperactivity-impulsivity symptoms might also be involved in the above-mentioned indirect pathway involving activation in the left temporal pole and anterior PHG during failed inhibition. Given the association between the PHG with its surrounding areas and post-error processing [
70], the increased activation of the anterior temporal pole and PHG in participants with a high genetic liability to ADHD may represent a compensatory response to error processing. It can be speculated that these individuals leverage their errors during failed inhibition to help optimize future behavior in upcoming trials in the stop-signal task and might develop a strategy to develop better general behavioral control skills to self-regulate their non-optimal impulsive behavior. However, because the association between the neural activity in the left temporal pole and anterior PHG and the hyperactivity-impulsivity symptoms was only nominally significant, further research is warranted.
Our results should be interpreted in light of the study’s potential strengths and limitations. As a unique feature, we consider the inclusion of individuals with ADHD, their unaffected siblings, and controls in our ‘target’ sample, which together may represent a more comprehensive and representative range of ADHD symptomatology and response inhibition correlates than a case-only study. Our PRS-ADHD—based on a well-powered GWAS of ADHD—showed a robust association with (both inattention and hyperactivity-impulsivity) ADHD symptoms in our ‘target’ sample. Nevertheless, a possible limitation of our current study could be the somewhat modest sample size of NeuroIMAGE. Future studies would benefit from larger ‘target’ sample sizes and probably even more powerful GWAS to allow for more definite conclusions about the shared genetic architecture between ADHD and behavioral and neural correlates of response inhibition. It is also worth keeping in mind that fMRI detects the hemodynamic changes in cerebral blood flow rather than direct neural activity, and our findings of fMRI analysis reflect altered brain activation related to cognitive processing during a stop-signal task. The clinical interpretation of altered brain activity (i.e., whether findings may reflect pathological or just altered physiological brain functioning) remains complicated. Although we used cross-sectional rather than longitudinal data to investigate mediation effects, the use of genetic risk scores together with the nature of brain functioning and (neurodevelopmental) behavioral symptomatology means that in this case inference of causality with regard to temporal precedence is not necessarily limited by the study design.
To conclude, our findings provide evidence for and better understanding of a shared genetic etiology between ADHD and behavioral measures and neural activity related to response inhibition in youth with a diagnosis of ADHD, unaffected siblings, and controls, corroborating response inhibition as a potential endophenotype. Partial mediation effect of brain activation in the left temporal pole and anterior PHG during failed inhibition on the association of PRS-ADHD with severity of hyperactivity-impulsivity symptoms may point to a possible pathway from genetic liability for ADHD to the expression of hyperactivity-impulsivity symptoms through altered brain activation during response inhibition. Moreover, MRT and IRT partially mediated the relationships of PRS-ADHD and ADHD symptom severity, suggesting that genetic liability to ADHD influences attention regulation, which in turn may affect the severity of both inattention and hyperactivity-impulsivity symptoms. Overall, our findings lend support for the conceptualization of response inhibition as a neurobiological mechanism underlying the etiology of ADHD. Our findings also provide novel insights regarding the genetic sharing of ADHD symptomatology with cognitive and underlying neural processing related to response inhibition.