Background
Anxiety disorders (ADs) are among the most prevalent and disabling psychiatric disorders to occur in youth [
1]-[
4] and set children on a negative trajectory towards continued and additional comorbid psychological disorders during adulthood [
5]-[
7]. When left untreated, pediatric anxiety disorders can result in severe ongoing social impairment, decreased educational achievement, and interrupted employment [
7]-[
9]. The three most common anxiety disorders among youth include generalized anxiety disorder (GAD), social phobia (SP), and separation anxiety disorder (SAD) and are collectively referred to as the “pediatric anxiety triad” [
10],[
11]. High rates of comorbidity across these diagnoses suggest greater similarities than differences [
12], including sensitivity to perceived or actual negative life events [
13] and debilitating worry leading to avoidance patterns (DSM-5, 2013). Additionally, these three disorders respond to similar treatments [
12],[
14],[
15], further implicating diagnostic overlap and, perhaps, common neural mechanisms. Recent examinations of pediatric anxiety have moved towards a more dimensional approach by including children with comorbid ADs to evaluate neural correlates [
16], as well as the effectiveness of treatments, such as cognitive behavioral therapy (CBT), in reducing the severity of anxiety symptoms [
17].
Despite the prevalence and negative sequelae of the pediatric anxiety triad, research examining the underlying neural mechanisms is in its infancy. The amygdala is the most frequently studied region of interest in pediatric anxiety, given the robust human neuroimaging literature documenting amygdala activity and connectivity as it relates to emotional processing and regulation [
18],[
19]. Indeed, amygdala hyperactivation to perceived threat is a defining neuropathophysiological feature of anxiety disorders [
20]-[
22] and frontal regions are known to have dense bidirectional structural connections with the amygdala [
23],[
24]. The amygdala is one region contributing to what has been labeled the anterior limbic network (ALN; [
18]). This network encompasses connections between the amygdala, medial prefrontal cortex (mPFC), insula, anterior cingulate cortex (ACC), as well as the ventrolateral and dorsolateral prefrontal cortexes (vlPFC, dlPFC) [
18]. These regions modulate complex social and emotional behaviors and share architectural and functional features [
25]. Reciprocal connections within this network are hypothesized to contribute to monitoring of internal and external sensory information in order to maintain emotional equilibrium [
26].
The strongest evidence implicating aberrant ALN function in anxiety disorders derives from several task-based fMRI studies that measure connectivity of networks during emotional tasks. Altered functional connectivity patterns have been observed during emotional processing and fear responding in regions composing the ALN among adults with anxiety [
27],[
28] as well as among youth [
19],[
22],[
29],[
30]. Adults with ADs have demonstrated decreased connectivity between the amygdala and the rostral ACC and dlPFC while viewing fearful faces [
27]. An examination of functional connectivity during a face-emotion rating task found greater connectivity between the right amygdala and the insula in youth with GAD compared to healthy controls (HCs). Anxiety symptom severity (as measured by the Pediatric Anxiety Rating Scale (PARS)) was correlated with extent of with amygdala-insula connectivity [
29]. Functional connectivity studies of both adults and youth support the notion that ALN disruption contributes to symptoms of anxiety. Disruptions in this network may underlie core phenotypic features of the disorder across the lifespan [
31].
Functional connectivity can also be measured during the resting state (labeled rs-fMRI) and allows for the examination of the intrinsic functional connectivity (iFC)
in the absence of a specific emotional task. Rs-fMRI has proven useful in interrogating neural circuits implicated in anxiety-related processes, with several studies demonstrating the existence of disrupted connectivity
at baseline in amygdala-based networks among adults with anxiety disorders [
27],[
32]. Importantly, iFC methods have yielded reliable individual differences in neural connectivity that are correlated with self-report of behavior and symptoms [
33]-[
36]. This technique has been utilized in recent studies of healthy adults to demonstrate several iFC patterns that covaried with positive and negative affect [
35], and trait levels of anxiety modulated amygdala-mPFC connectivity [
37]. These results implicate the relevance of functional connectivity in the affective domain
even in the absence of an emotional challenge among adults with anxiety disorders. In addition, recent rs-fMRI studies have demonstrated altered resting state connectivity in regions considered part of the ALN, including the ACC, mPFC, and insula [
27],[
38],[
39].
A region outside of the ALN that has been implicated in social and general anxiety is the posterior cingulate cortex (PCC) and the adjacent precuneus. The PCC in particular may play a role in emotional evaluation [
40] and social behavior [
41]. The PCC may also contribute to modulation of the amygdala [
42]. Rs-fMRI data collected from adults with anxiety disorders found that reduced connectivity between the amygdala and the PCC/precuneus was associated with higher state anxiety [
43]. Among adolescents with GAD, one study that has examined connectivity of the amygdala during a task of emotional and neutral images found altered connectivity between the right amygdala and the posterior cingulate [
19]. In sum, task-based fMRI studies have identified abnormalities in the PCC among youth with anxiety disorders [
19],[
29], but limited work has examined this region at rest among youth.
Examinations of connectivity among youth with anxiety disorders is understudied to date, partly due to the difficulty in recruiting this population and acclimatizing them to the fMRI environment. However, the altered connectivity patterns observed among adults may not be applicable to pediatric populations due to the important structural and functional developmental changes known to occur in the brain during childhood and adolescence [
44]-[
46]. Examining the developmental trajectory of neural network abnormalities among youth with anxiety may elucidate predictive or modifiable biomarkers of anxiety in addition to illustrating the long-term effects of anxiety on neurodevelopment [
47]. To the best of our knowledge, only one study to date has used rs-fMRI to examine functional connectivity in youth with anxiety [
48]. This study documented perturbations in connectivity between the amygdala and the following regions: ACC, striatum, insula, superior temporal gyrus, as well as prefrontal regions including the ventromedial prefrontal cortex (vmPFC), dmPFC, vlPFC, and dlPFC among fifteen youth between 12 and 17 with a diagnosis of GAD. These differences support a more widespread disruption of network function than previously identified.
In the present study, we sought to contribute to the literature by examining rs-fMRI using bilateral amygdala seeds in a sample of 33 youth with primary ADs of GAD and/or SP with several comorbidity profiles and compared them to data for 23 healthy controls (ages 7 to 19). We chose to examine both the left and right amygdala seeds separately given the only pediatric anxiety rs-fMRI study to date detected laterality in amygdala connectivity [
48]. We sought to study a representative heterogeneous diagnostic group that would allow for greater generalizability of findings consistent with epidemiologic and intervention trials that demonstrate comorbidity across these disorders and commonality in treatment response [
14],[
49]. In line with emotion regulation models of ADs [
50], we hypothesized that relative to healthy peers, youth with ADs would demonstrate hyperconnectivity between the amygdala and insula. We also hypothesized that youth with AD would demonstrate hypoconnectivity between the amygdala and regions included in the ALN such as the ACC and mPFC. We also sought to explore amygdala-PCC connectivity but did not hypothesize a direction based on the paucity of findings to date.
Discussion
Consistent with our hypotheses, the youth with AD demonstrated aberrant amygdala connectivity with regions of the ALN including the vmPFC and insula when compared to HCs. Surprisingly, we did not find connectivity differences with the ACC but did observe amygdala-PCC hypoconnectivity among AD compared to HC youth. Our results replicate previous observations and extend upon the only study to date that has examined resting-state iFC in adolescents with AD [
48], suggesting these findings may be reliable and could even generalize across diagnostic categories - from GAD to SP and their comorbidities.
Specifically, we found hyperconnectivity between the right amygdala seed and the left insula among anxious youth compared to HC peers, consistent with the previous literature [
34],[
55]-[
57]. Insula and amygdala involvement in the detection of salience, emotion, and attention is well established [
34] and task-based fMRI findings have indicated the hyperactivity of these regions may be a key neural mechanism underlying anxiety-related processes [
34],[
58],[
59]. The amygdala has been found to be structurally connected to the insula [
60], and our results contribute to emerging evidence of a functional connection between the structures [
27],[
34],[
48]. Altered functional connectivity between the amygdala and insula has been previously observed in groups with anxiety disorders during task [
27],[
29] and more recently during rest [
27],[
38],[
39],[
48]. Given the insula’s role in interoceptive processing, increased connectivity with the amygdala at rest may reflect increased interactions between a region implicated in fear perception-expression (amygdala) and another implicated in anxious arousal-anticipation (insula).
The extant literature implicates dysfunction in amygdala connections to the prefrontal cortex [
19]. Our finding of decreased iFC between the amygdala and frontal regions such as the vmPFC among youth with ADs is consistent with prior findings in adult and pediatric resting-state studies. Specifically, previous research in healthy adults has demonstrated positive coupling between the amygdala and vmPFC at rest [
37],[
61] and has also found this relationship to be compromised in those with higher levels of self-reported anxiety [
37]. The latter study found those with high levels of anxiety displayed negative coupling between the amygdala and vmPFC. These findings have since been replicated within a sample of adolescents with GAD [
48]. This study documented perturbed amygdala-PFC circuitry, finding negative connectivity between amygdala and vmPFC and positive connectivity between amygdala and dmPFC, in the group of adolescents with GAD. The healthy control adolescents showed opposite patterns of coupling between the amygdala and these regions. Our findings of negative connectivity between the amygdala and vmPFC within the AD group contribute to the growing body of evidence implicating disruption of the dynamic interplay within amygdala-PFC circuitry among individuals with anxiety disorders. Further, our results suggest this aberrant connectivity pattern can be observed at rest. Taken together, these findings suggest inefficient crosstalk between the amygdala and mPFC may lead to increased anxiety levels. Additional research will be needed in order to determine if this compromised connectivity is a defining feature of the underlying neurocircuitry of anxiety disorders.
In the current study, we observed altered connectivity between the amygdala and PCC, which is consistent with the growing body of literature linking disruption of this functional connection to mood and anxiety disorders [
42],[
43],[
62]. Recent studies have implicated functional connections between the amygdala and posterior regions, such as the PCC and precuneus [
19],[
29], in the implementation of emotional processing [
62]. In addition, the PCC is a hub in the default mode network (DMN), a network that subserves processes such as mentalization and self-referential thinking [
19],[
29], which may contribute to hypervigilance to interoceptive cues of anxiety. Indeed, prior studies have observed altered amygdala-PCC connectivity in pediatric GAD cohorts during emotional processing tasks [
19],[
29] and at rest [
48]. Taken together, these convergent findings suggest a tonic (task-independent) versus phasic (task-dependent) disruption in amygdala-PCC connectivity and future research will be needed to elucidate whether this is a defining neural underpinning of pediatric anxiety disorders. Recent work in depression has shown that treatment normalizes posterior cingulate-amygdala connectivity [
52] and our findings suggest treatment targets for ADs and depression may overlap.
Amygdala-based connectivity correlated with PARS anxiety score across the entire sample, but this correlation was not significant within the AD group or HC group when considered independently, likely due to restriction of range. However, within the AD group, connectivity between the amygdala and PCC was positively correlated with age. Given this is the first documentation of this finding among youth and a cross-sectional study, we hesitate to over-interpret this finding. However, among HC youth, decreased connectivity between the amygdala and PCC has been observed across development [
63]. The PCC is a key node in the DMN and default mode regions are known to functionally connect in a more integrated fashion across development [
64], which may contribute to the current finding.
The current study is not without limitation. Although the sample size represents the largest to date, replication with a larger cohort of youth is necessary. However, the comorbidity profile of the current cohort may make our findings more generalizable, while noting that most patients (70%) had a GAD diagnosis. Participants in the current study met criteria for multiple ADs, similar to children presenting in clinical settings for treatment and to more recent clinical trials testing the efficacy of interventions in reducing overall anxiety. Moreover, although we observed one finding was significantly correlated with age, our sample size is underpowered for these analyses within the AD group. We captured a relatively wide age range in line with our desire to cast a wider net than previous studies. Data collection is ongoing, and a larger sample will allow for greater exploration of potential developmental effects. This larger sample may also allow for greater variability in anxiety levels within the AD group, making it more likely that variability in network functioning can be linked to severity of symptoms. We did not collect state anxiety symptoms at the time of the fMRI scan to relate to resting-state amygdala iFC. An additional limitation of the current data is that adolescent network functioning may differ from that of children and we look forward to future studies that can explore the nuances of healthy and disordered brain development. Clearly, the examination of the developmental trajectories of resting state networks among youth with and without ADs will be a groundbreaking work. Lastly, this is a cross-sectional observation and emotional face processing tasks (findings reported elsewhere) administered before the resting-state scan may have influenced connectivity in unexpected ways. Future work should address multiple resting state collection periods, acute influences of a preceding “emotional” task, as well as test order effects across tasks and rest, in order to determine the reliability of these networks.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made.
The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder.
Competing interests
The authors declare that they have no competing interests.
Authors’ contributions
LH participated in the acquisition, analysis, and interpretation of the data, along with drafting and revising the manuscript. RJ participated in the analysis and interpretation of the data, along with drafting and revising the manuscript. DF participated in the analysis and interpretation of the data, along with revising the manuscript. MJ participated in the acquisition and interpretation of the data, along with drafting the manuscript. KF participated in the conception and design of the study, along with revising the manuscript. SL participated in the analysis and interpretation of the data and in the revision of the manuscript. CM participated in the conception and design of the study, along with revising the manuscript. KLP participated in the conception and design of the study, along with interpretation of the data and drafting of the manuscript. All authors read and approved the final manuscript.