Introduction
Attention-deficit/hyperactivity disorder (ADHD) is a neurodevelopmental disorder characterized by developmentally inappropriate and impairing inattention, hyperactive, and impulsive behaviors. Etiological models of ADHD postulate dysfunction in fronto-subcortical neural pathways involved in executive functions and motivation as contributing to deficient self-regulation of cognition, behavior, and emotion [
1,
2]. Executive function (EF) refers to the deliberate, top–down control of thoughts, actions, and emotions in the service of goal-directed behavior [
3] and is generally purported to rely on discrete cortico-striatal-thalamo-cortical loops [
4‐
9]. Cognition and motivation and the associated neural circuitry interact to produce adaptive and maladaptive behavior [
10]. The interaction of cognition and motivation guides reward-based decision-making in the form of delay discounting, a well-established phenomenon in which individuals discount the value of a reward as a function of delay to receiving the reward [
11,
12]. The ability to inhibit a response to an immediately available reward in pursuit of a larger or more valuable, albeit delayed, reward is a critical component of cognitive, emotional, and social development. Failure to inhibit such a response is thought to be a central feature of pathological behavior associated with impulsivity including ADHD, substance abuse, obesity, and gambling [
13‐
15].
Delay discounting is central to many theories of ADHD, which postulate altered reinforcement sensitivity [
16] either due to attenuation of dopamine signaling to delayed reward [
17], a failure of anticipatory dopamine cell firing [
18], or a breakdown in higher order control resulting in an inability to suppress the drive (i.e., resist temptation) to respond to the immediate option [
19]. Although delay discounting is typically described as reflecting reward sensitivity, there is growing evidence that delay aversion may also contribute to one’s preference for immediate over delayed rewards. Sonuga-Barke and colleagues proposed that delay is an aversive experience in and of itself, eliciting a negative affective state, which children with ADHD work to escape or avoid [
20,
21]. Delay aversion may also work in concert with an impulsive drive for immediate reward to exacerbate impulsive choice [
1,
22]. Neuroimaging research has implicated fronto-subcortical circuitry in delay discounting as part of a cognitive control network including the dorsolateral PFC (dlPFC) and anterior cingulate cortex (ACC) and a reward valuation network including the ventromedial (vmPFC)/orbitofrontal cortex (OFC) and ventral striatum (VS)/nucleus accumbens (NAcc) [
11]. In addition, task-based fMRI studies have shown involvement of the amygdala in delay discounting among individuals with ADHD [
23,
24], in support of the delay aversion theory of ADHD. Thus, variability in fronto-subcortical neural circuitry implicated in ADHD may be associated with individual differences in delay discounting.
Evidence of ADHD-associated disruptions in intrinsic fronto-subcortical functional connectivity (FC) using resting state functional MRI (rs-fMRI) has been inconsistent. In general, studies have shown aberrant FC of fronto-subcortical networks in children and adolescents with ADHD (see reviews by [
25,
26]). However, the specific regions involved and whether a group effect or an association with ADHD symptoms was observed and the direction of the observed group effect or symptom association have all varied [
27‐
29]. Studies examining striatum-vmPFC FC have reported greater FC [
30‐
32] and similar FC among children and adolescents with ADHD compared to controls [
33]. In contrast, studies of striatal-dlPFC FC have reported weaker FC with the VS [
34], dorsal caudate [
33], and putamen [
35] in children and adolescents with ADHD. Further, findings from the same researchers among a sample of partially overlapping participants reported both stronger [
31] and weaker NAcc-anterior PFC FC in ADHD [
32], possibly due to the different methods used to define the NAcc seed region or a more heterogeneous ADHD sample in the latter study. Finally, two studies examining a much wider age range (e.g., 8–30 years) did not find evidence of aberrant cortico-striatal networks in ADHD [
27,
29]. Only one study to date has examined associations between rs-fMRI FC and delay discounting in children with ADHD, reporting that increased NAcc-anterior PFC FC in ADHD positively correlated with delay discounting [
31]. While the majority of studies in the ADHD literature have used seed-based analyses, they have varied in their selection and definition of the seed regions. We chose to apply a combined data- and hypothesis-driven approach in which we use group-independent components analyses (ICA) to identify the intrinsic functional networks rather than defining seed-regions based on anatomical boundaries or as spheres centered around reported peaks of task activation with arbitrary radii. To focus our analyses on fronto-subcortical regions, we selected components with the greatest spatial overlap with anatomical regions of interest (ROIs) spanning ventral to dorsal regions of the PFC (OFC, ACC, dlPFC) and subcortical reward and limbic regions (striatum, amygdala).
Recent evidence suggests ADHD-related sex differences across behavioral and neural domains are another important inter-individual variable to consider. There is a surprising lack of research comparing girls and boys with ADHD to same-sex TD children despite reports that the proportion of males to females diagnosed with the disorder has fallen to approximately 2:1 [
36]. Evidence suggests boys with ADHD display greater motor deficits both in terms of behavior [
37‐
39] and the associated neural circuitry [
40‐
44]. In contrast, girls with ADHD tend to display equivalent or greater executive dysfunction both in terms of behavior [
39,
45] and the associated neural circuitry [
40,
41]. Moreover, girls with ADHD show greater delay discounting relative to TD girls and to boys with ADHD [
46], as well as distinct neuropsychological correlates of delay discounting [
47] and atypical behavioral response to reward [
48]. However, no study has explicitly examined whether FC of fronto-subcortical circuitry is differentially altered among girls and boys with ADHD compared to same-sex TD children.
The current study adds to the existing literature and builds off of our previous findings of greater delay discounting in girls, but not boys, with ADHD by examining ADHD-related sex differences in intrinsic FC of fronto-subcortical brain networks implicated in ADHD and delay discounting. We hypothesized that fronto-subcortical intrinsic FC would be disrupted in ADHD with the greatest differences involving ventromedial regions of the PFC. Given previous evidence of ADHD-related differences in delay discounting being greater among girls, we expected greater disruptions in fronto-subcortical FC among girls. We also examined correlations between delay discounting and intrinsic FC of fronto-subcortical networks.
Discussion
The current study adds to the existing ADHD neuroimaging and delay discounting literature by combining a data-driven approach to identify intrinsic functional networks with a theory-driven approach to examine ADHD-related sex differences in fronto-subcortical FC. Our findings suggest that children with ADHD show atypical FC between the vmPFC component and subcortical regions, including stronger positive FC with the striatum component and weaker negative FC with the amygdala component, with greater magnitude of effects among girls although the small effects among boys were in the same direction. In addition, girls with ADHD show atypical intrinsic FC between the striatum component and the relatively dorsal PFC components, including stronger positive FC with the ACC component and stronger negative FC with the dlPFC component. Further, girls but not boys, with ADHD, show heightened delay discounting on the real-time task compared to TD girls, as previously reported [
46], whereas no diagnostic effects were observed among boys. Examination of brain–behavior correlations showed that FC between the anterior dlPFC-striatal components correlated with money delay discounting across all participants, regardless of diagnosis. Further, FC of the amygdala component with both the ACC and dlPFC components was differentially related to real-time delay discounting among girls and boys with and without ADHD. These findings contribute to the growing literature examining functional connectivity of fronto-striatal networks implicated in ADHD using ICA methods and extend this literature through examination of ADHD-related sex differences and associations with multiple measures of delay discounting.
Consideration of these finding with the existing literature provides growing evidence for stronger vmPFC-striatum FC, thought to reflect greater integration [
80,
81], among children and adolescents with ADHD [
30‐
32]. Fewer studies have examined connectivity of the amygdala among children with ADHD, with evidence of greater PFC-amygdala FC in adolescents with ADHD during an emotional task [
82] and in relation to emotional lability [
83], whereas reduced negative FC of an amygdala subregion with the dlPFC has been reported among boys with ADHD [
84]. Our findings add to this literature, suggesting reduced negative FC, thought to reflect reduced segregation, [
80,
81] between the vmPFC-amygdala components in ADHD. Our findings of atypical intrinsic vmPFC-subcortical FC in children with ADHD may be related to the behavioral and emotional dysregulation observed in individuals with ADHD given the role of the vmPFC in top-down inhibitory control of bottom-up activity in subcortical areas. The vmPFC is a key component of the brain’s reward system and is highly interconnected with subcortical structures involved in reward and affective processing such as the striatum and amygdala [
85]. Research has shown that the vmPFC regulates behavior by inhibiting the influence of emotions, thoughts, and actions [
86]. Further, the vmPFC is involved in representing the actual and expected reward-value of stimuli, reward prediction errors, and reward-based decision-making [
87]. Although diagnostic groups did not differ in the spatial topography of the vmPFC component, FC between this component and subcortical components was atypical among children with ADHD, particularly girls, highlighting the importance of examining interactions between fronto-subcortical neural networks. Furthermore, these findings call attention to the influence of sex on ADHD-related differences in fronto-subcortical functional networks and emphasize the importance for replication of these results among larger samples of girls with ADHD using ICA- and seed-based methods.
Examination of fronto-subcortical FC within sex suggests girls with ADHD, but not boys, displayed stronger negative anterior dlPFC-striatum FC compared to same-sex TD children (
d = .74), and this correlated with money delay discounting. Thus, individuals showing stronger functional segregation between striatal regions involved in reward processing and prefrontal regions involved in cognitive control tend to show greater delay discounting (Fig.
2). In contrast, FC of the amygdala with relatively dorsal PFC components correlated with real-time discounting among TD girls and, to a lesser extent, among ADHD boys. The differential associations between dlPFC-striatum FC and money delay discounting and between dlPFC/ACC-amygdala FC and real-time delay discounting suggests the neural correlates of delay discounting depend on characteristics of the task. In particular, when delays and rewards are experienced in real-time, negative affect associated with waiting may contribute to the preference for immediate reward as suggested by delay aversion models of ADHD [
1,
22,
88]. This may be why functional connectivity of the amygdala is more strongly related to real-time delay discounting whereas decision-making on delay discounting tasks involving more abstract reasoning without a significant affective component relate to connectivity between brain regions governing cognitive control and reward.
One previous study using the identical money delay discounting task along with a seed-based analysis reported that increased positive NAcc-anterior PFC FC (a small region included in the anterior dlPFC component examined here) was positively correlated with delay discounting [
31]. Although both studies implicate atypical striatal-PFC FC in delay discounting, the direction of these effects differs. In the current study, we used ICA to functionally define a component that includes the caudate and putamen rather than focusing specifically on the NAcc, which may contribute to the discrepant findings. In addition, the dlPFC component is much larger than the anterior PFC component in the previous study, suggesting that distinct functional connectivity patterns may be observed across different regions of the PFC. However, the consistent involvement of striatal-PFC regions in relation to delay discounting suggests a possible neural mechanism of heightened delay discounting in ADHD. Importantly, children with ADHD did not significantly differ in their performance on the money delay discounting task involving choices about money (although they did differ in the task involving choices about gametime), consistent with some prior research [
23,
46,
89‐
91]. This might suggest a subgroup of children with ADHD who display atypical delay discounting and fronto-striatal FC, which may inform our understanding of heterogeneity in ADHD (e.g., [
32]).
The novel findings of ADHD-related sex differences in fronto-subcortical FC and associations with delay discounting must be considered within the limitations of this study. First, the majority of sample of children with ADHD included in this study were not naïve to stimulant medication and it is unclear what, if any, affect this might have on our findings. Second, in order to understand the pathophysiology of ADHD specifically, we excluded children with comorbid disorders other than ODD, which limits the generalizability of our results. Our results also may not generalize to children with more severe ADHD and behavioral problems due to the exclusion of participants with excessive motion during the resting-state scan. Future research must attempt to replicate these findings given the small sample of girls with ADHD as well as the inconsistent results in the ADHD neuroimaging literature and the lack of studies comparing girls and boys with ADHD, and to extend these findings using longitudinal methods to understand the developmental trajectory of anomalous fronto-subcortical FC in ADHD.
Conclusions
Our findings suggest functional fronto-subcortical networks are affected in children with ADHD, particularly girls, such that the striatum is intrinsically more strongly connected to frontal regions, being both more functionally segregated (e.g., negatively correlated) with the anterior dlPFC and more functionally integrated (e.g., positively correlated) with the vmPFC, while the amygdala/hippocampus is intrinsically less connected to the vmPFC. In addition, intrinsic FC of the striatum and amygdala is differentially related to money and real-time discounting, providing support for unique neural correlates of delay discounting tasks involving real versus hypothetical delays and rewards. These findings add to the extant literature implicating fronto-striatal circuitry in children with ADHD and expand upon these findings to reveal associations with a behavioral preference for immediate reward and atypical functional connectivity of the amygdala in ADHD. Moreover, this is the first study to show greater anomalies in fronto-subcortical functional networks among girls with ADHD. This study adds to our understanding of the neurobiological correlates of ADHD and suggests potential differences among school-age girls and boys with ADHD that relate to reward-based decision-making.