If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
2 DSA and LA contributed equally to this work as co-first authors.
Affiliations
Medical Investigation of Neurodevelopmental Disorders (MIND) Institute and Department of Psychiatry and Behavioral Sciences, University of California Davis, Davis, California
2 DSA and LA contributed equally to this work as co-first authors.
Leon Aksman
Footnotes
2 DSA and LA contributed equally to this work as co-first authors.
Affiliations
Stevens Neuroimaging and Informatics Institute, Keck School of Medicine of the University of Southern California, Los Angeles, CaliforniaCentre for Medical Image Computing, Department of Medical Physics and Biomedical Engineering, University College London, London, United Kingdom
Medical Investigation of Neurodevelopmental Disorders (MIND) Institute and Department of Psychiatry and Behavioral Sciences, University of California Davis, Davis, California
Medical Investigation of Neurodevelopmental Disorders (MIND) Institute and Department of Psychiatry and Behavioral Sciences, University of California Davis, Davis, California
Medical Investigation of Neurodevelopmental Disorders (MIND) Institute and Department of Psychiatry and Behavioral Sciences, University of California Davis, Davis, California
Medical Investigation of Neurodevelopmental Disorders (MIND) Institute and Department of Psychiatry and Behavioral Sciences, University of California Davis, Davis, California
1 CWN and DGA contributed equally to this work as senior primary investigators.
Christine Wu Nordahl
Footnotes
1 CWN and DGA contributed equally to this work as senior primary investigators.
Affiliations
Medical Investigation of Neurodevelopmental Disorders (MIND) Institute and Department of Psychiatry and Behavioral Sciences, University of California Davis, Davis, California
1 CWN and DGA contributed equally to this work as senior primary investigators.
David G. Amaral
Footnotes
1 CWN and DGA contributed equally to this work as senior primary investigators.
Affiliations
Medical Investigation of Neurodevelopmental Disorders (MIND) Institute and Department of Psychiatry and Behavioral Sciences, University of California Davis, Davis, California
The amygdala is widely implicated in both anxiety and autism spectrum disorder. However, no studies have investigated the relationship between co-occurring anxiety and longitudinal amygdala development in autism. Here, the authors characterize amygdala development across childhood in autistic children with and without traditional DSM forms of anxiety and anxieties distinctly related to autism.
Methods
Longitudinal magnetic resonance imaging scans were acquired at up to four time points for 71 autistic and 55 typically developing (TD) children (∼2.5–12 years, 411 time points). Traditional DSM anxiety and anxieties distinctly related to autism were assessed at study time 4 (∼8–12 years) using a diagnostic interview tailored to autism: the Anxiety Disorders Interview Schedule-IV with the Autism Spectrum Addendum. Mixed-effects models were used to test group differences at study time 1 (3.18 years) and time 4 (11.36 years) and developmental differences (age-by-group interactions) in right and left amygdala volume between autistic children with and without DSM or autism-distinct anxieties and TD children.
Results
Autistic children with DSM anxiety had significantly larger right amygdala volumes than TD children at both study time 1 (5.10% increase) and time 4 (6.11% increase). Autistic children with autism-distinct anxieties had significantly slower right amygdala growth than TD, autism–no anxiety, and autism–DSM anxiety groups and smaller right amygdala volumes at time 4 than the autism–no anxiety (−8.13% decrease) and autism–DSM anxiety (−12.05% decrease) groups.
Conclusions
Disparate amygdala volumes and developmental trajectories between DSM and autism-distinct forms of anxiety suggest different biological underpinnings for these common, co-occurring conditions in autism.
Clinically significant anxiety in children with autism spectrum disorder and varied intellectual functioning [published correction appears online ahead of print, J Clin Child Adolesc Psychol, Mar 27, 2020].
). No studies of autism have investigated the development of the amygdala longitudinally in relation to anxiety or the associations between different forms of anxiety and the amygdala in autism.
Clinical anxiety can manifest in several forms, including generalized anxiety disorder, separation anxiety, specific phobia, and social phobia (henceforth referred to as DSM anxiety) (
). Recently developed tools recognize classically defined symptoms of anxiety (e.g., anticipatory anxiety, fearful avoidance) that manifest within contexts that are somewhat unique to autism. These symptoms would not be captured by traditional assessments (
Anxiety Disorders Interview Schedule–Autism Addendum: Reliability and validity in children with autism spectrum disorder [published correction appears in J Clin Child Adolesc Psychol 2020 49:iv].
). Such autism-distinct anxieties (henceforth referred to as distinct anxiety) include fears related to social confusion (as opposed to fear of negative evaluation, which is required for a DSM diagnosis of social phobia), uncommon phobias (e.g., specific sounds, facial features), excessive worry related to losing access to materials related to circumscribed interests, and fears of change (
Clinically significant anxiety in children with autism spectrum disorder and varied intellectual functioning [published correction appears online ahead of print, J Clin Child Adolesc Psychol, Mar 27, 2020].
Gray matter volume in adolescent anxiety: An impact of the brain-derived neurotrophic factor Val(66)Met polymorphism? [published correction appears in J Am Acad Child Adolesc Psychiatry 2013; 52:329].
J Am Acad Child Adolesc Psychiatry.2013; 52: 184-195
). These studies used a range of anxiety assessments including dimensional measures that are nonspecific to the domains of anxiety described in the DSM (e.g., the Child Behavior Checklist [CBCL] and Screen for Child Anxiety Related Emotional Disorders) (
The Child Behavior Checklist and related instruments.
in: Maruish M.E. The Use of Psychological Testing for Treatment Planning and Outcomes Assessment. 2nd ed. Lawrence Erlbaum Associates Publishers. 429–466,
Mahwah, NJ1999
) and grouping individuals across DSM anxiety domains according to diagnostic interviews (e.g., Schedule for Affective Disorders and Schizophrenia for School-Age Children-Present and Lifetime version) (
Schedule for Affective Disorders and Schizophrenia for School-age Children-Present and Lifetime Version (K-SADS-PL): Initial reliability and validity data.
J Am Acad Child Adolesc Psychiatry.1997; 36: 980-988
A longitudinal MRI study of amygdala and hippocampal subfields for infants with risk of autism.
in: Zhang D. Zhou L. Jie B. Liu M. Graph Learning in Medical Imaging: First International Workshop, GLMI 2019, Held in Conjunction with MICCAI 2019, Shenzen, China, October 17, 2019, Proceedings. Springer International Publishing,
Cham, Switzerland2019: 164-171
Cortical and subcortical brain morphometry differences between patients with autism spectrum disorder and healthy individuals across the lifespan: Results from the ENIGMA ASD working group.
) methodologies suggest that the amygdala in autism exhibits an initial volumetric overgrowth during early childhood that is then followed by a slowed trajectory of growth into adulthood.
Despite high rates of anxiety in ASD, only three studies have investigated associations of amygdala structure with anxiety in autism. The first found enlarged right amygdala volumes to be associated with increased scores on the CBCL anxious/depressed subscale in children with autism (∼4–15 years) (
). Thus, two studies indicate a relationship between anxiety and right amygdala volume but, taken together, provide contradictory evidence as to whether amygdala volume is larger, smaller, or unrelated to anxiety in autism. Determining the relationship of amygdala volume with autistic development, with and without co-occurring anxiety, may identify biological correlates specific to these conditions and thus provide a valuable prodromal biological marker of anxiety in autism.
In this study, we characterize and test for anxiety-associated differences in the trajectories of volumetric development of the amygdala across childhood (∼2.5–12 years), as well as amygdala volumes during early (∼3 years) and late (∼11 years) childhood, in autistic children with and without anxiety disorders. We used clinical interviews for both DSM and distinct anxiety to test if these forms of anxiety had different associations with amygdala volume and development in autism. We hypothesize that DSM anxiety will be associated with larger right amygdala volumes and faster development. We further expect that anxieties distinctly related to autism will be associated with effects on the amygdala, but because this has not been previously investigated, we have no a priori prediction as to the nature of this relationship.
Methods and Materials
Participants
Participants were enrolled in the UC Davis Medical Investigation of Neurodevelopmental Disorders Institute Autism Phenome Project, a longitudinal study consisting of MRI scanning at four time points, enrollment/baseline at 24 to 42 months of age (time 1), follow-up at annual intervals for two time points (time 2 and 3), and ∼9 to 12 years of age (time 4). This study included data from all participants who completed MRI and anxiety assessments at time 4 (Table 1 and Figure S1). At time 1, ASD diagnosis was confirmed using the Autism Diagnostic Observation Schedule (ADOS)–Generic (
Autism Diagnostic Interview-Revised: A revised version of a diagnostic interview for caregivers of individuals with possible pervasive developmental disorders.
). Informed consent was obtained from the parent or guardian of each participant. All aspects of the study protocol were approved by the University of California Davis Institutional Review Board. See the Supplement for details.
Table 1Time 4 Participant Demographics
Measure
ASD, n = 71
TD, n = 55
Sex, Female/Male, n
14/57
21/34
Age, Years, Mean (SD)
11.18 (1.39)
11.34 (0.67)
IQ, Mean (SD)
80.54 (28.77)
111.58 (12.82)
ADOS CSS, Mean (SD)
7.5 (1.96)
–
ADOS Social CSS, Mean (SD)
7.3 (1.73)
–
ADOS Repetitive Restricted Behavior CSS, Mean (SD)
Anxiety Disorders Interview Schedule–Autism Addendum: Reliability and validity in children with autism spectrum disorder [published correction appears in J Clin Child Adolesc Psychol 2020 49:iv].
). ADIS includes modules relating to separation, social, specific phobia, and generalized anxiety disorders (DSM anxiety). In addition, the ADIS-ASA assesses forms of anxiety distinctly related to autism, including idiosyncratic fears, fear relating to social confusion, special interest fears, and fears of change (distinct anxiety). For each module, a clinical severity rating (CSR) ranging from 0 (no interference) to 8 (severe interference) is prescribed, with 4 being the cutoff for clinical determination of significant interference. Rates of anxiety and assessment within this cohort have been described in detail previously (
Clinically significant anxiety in children with autism spectrum disorder and varied intellectual functioning [published correction appears online ahead of print, J Clin Child Adolesc Psychol, Mar 27, 2020].
The Child Behavior Checklist and related instruments.
in: Maruish M.E. The Use of Psychological Testing for Treatment Planning and Outcomes Assessment. 2nd ed. Lawrence Erlbaum Associates Publishers. 429–466,
Mahwah, NJ1999
All T1-weighted structural MRI scans were acquired at the UC Davis Imaging Research Center on a 3T Siemens Trio (Siemens Corp.). Time 1 to time 3 scans were acquired during natural nocturnal sleep (
). Volumes from the left and right amygdala and hemispheres were exported for statistical modeling. See the Supplement for details.
Statistical Modeling
Linear mixed-effects modeling was performed using R (version 3.6; R Core Team, 2019). We first compared differences in amygdala volumes and development associated with anxiety between autistic children with (ADIS-DSM CSR ≥ 4) and without DSM anxiety and with (ADIS-ASA distinct anxiety CSR ≥ 4) and without distinct anxiety. Models included all autistic individuals with categorical factors for DSM and distinct anxiety; sex, age in months, and hemispheric volume as covariates; and individual as a random effect with age as a random slope. Here, overlapping anxieties are accounted for by separate DSM and distinct categorical variables.
A secondary analysis was conducted to compare amygdala volumes and development between five groups of interest: ASD with 1) only DSM anxiety, 2) only distinct anxiety, 3) both DSM and distinct anxiety, and 4) no clinical anxiety and 5) typically developing (TD) children without anxiety. Effects of interest for all analyses were mean group differences at the average age between study time 1 (38.11 mo/3.18 years) and time 4 (136.30 mo/11.36 years) and differences in developmental trajectories between groups (age-by-group interactions). Age was modeled by selecting from a range of polynomial terms (−3, −2, −1, −0.5, 0.5, 1, 2, 3) that returned the lowest log likelihood (left = 0.5, right = −0.5) (
Children with autism and TD children did not significantly differ (p > .05, two-tailed t test) for age of scan at time 4 and age of ADIS-ASA assessment. Compared with the TD sample, children with autism had significantly lower overall IQ scores (p < .001) and fewer longitudinal time points per individual (p = .008). Compared with the TD sample, the ASD sample also contained a significantly higher proportion of males (χ21 = 4.39, p = .03).
Within the autism sample, 61% (43/71) of participants met diagnostic criteria for at least one form of clinical anxiety, with 45% (32/71) meeting criteria for one or more DSM anxieties, 39% (28/71) meeting criteria for one or more distinct anxieties, and 24% (16/71) meeting criteria for both a DSM and distinct anxiety. In total, 39% (28/71) of children with autism did not reach clinical thresholds for any anxiety type (Table 1). No significant differences (analysis of variance [ANOVA] p > .05) were found between ASD groups with 1) only DSM anxiety, 2) only distinct anxiety, 3) both DSM and distinct anxiety, and 4) no clinical anxiety in terms of time 4 scan age, number of MRI time points, IQ, or ADOS calibrated severity score. No significant differences were found for the male-to-female ratio between the four ASD groups (χ21 = 5.10, p = .16).
Differences in Amygdala Volumes Associated With Clinical DSM Anxieties and Autism-Distinct Anxieties
Compared with autistic children without DSM anxieties, those with DSM anxieties (CSR ≥ 4) had larger right amygdala volumes at time 4 (4.94% mean increase, p = .017, FDR p = .038); no statistically significant differences were observed in the left amygdala. Compared with autistic children without distinct anxiety, those with distinct anxiety (CSR ≥ 4) showed smaller right and left amygdala volumes at both time 1 (right: −4.02% mean decrease, p = .019, FDR p = .38; left: −5.54%, p = .006, FDR p = .018), and time 4 (right: −4.91% mean decrease, p = .001, FDR p = .038; left: −7.14%, p = .002, FDR p = .012). Autistic children with autism-distinct anxiety also showed a statistical trend (p = .049, FDR p = .074) of slower right amygdala development than autistic children without distinct anxiety (Figure 1 and Table 2).
Figure 1Associations between amygdala volume and DSM and autism-distinct anxieties. Longitudinal development of (A) left and (B) right amygdala volumes are plotted for autistic children with and without DSM anxiety (ASD-DSM ANX and ASD-no DSM ANX, respectively). Compared with autistic children without DSM anxiety, those with DSM anxiety showed larger right amygdala volumes at both time 1 (5.87% larger, p = .005, false discovery rate–adjusted p = .048) and time 4 (6.15% larger, p = .005, false discovery rate–adjusted p = .048). Compared with autistic children without autism-distinct anxieties (ASD-no DIST ANX), those with distinct anxieties (ASD-DIST ANX) showed statistical trends (p < .05, false discovery rate–adjusted p > .05) of (C) smaller left amygdala volumes at both time 1 (−5.44% smaller, p = .024) and time 4 (−6.75% smaller, p = .011) and (D) right amygdala volumes at time 4 (−5.75% mean decrease, p = .020) compared with ASD-no DIST ANX. Vertical dotted lines indicate mean ages at study time 1 (38.11 mo/3.18 years) and time 4 (136.30 mo/11.36 years). Note that here, some individuals within the DSM/DIST ANX groups have duel DSM and distinct anxiety diagnoses; in addition, individuals within the no DSM group may have distinct clinical severity rating scores ≥4 (and vice versa for the no DIST group). These overlapping anxieties are accounted for by modeling separate DSM and distinct categorical variables.
Regression table for models of left and right amygdala volume between children with ASD with and without DSM anxiety (ASD-DSM, ASD-no DSM) and with and without distinct anxiety (ASD-DIST, ASD-no DIST). Effects of interest are shown for mean age of T1 (38.11 mo/3.18 years) and T4 (136.30 mo/11.36 years) as well as developmental differences (age-by-group interactions). Note that some individuals within the DSM/DIST anxiety groups have dual DSM and distinct anxiety diagnoses; in addition, individuals within the no DSM group may have distinct CSR scores ≥4 (and vice versa for the no DIST group). These overlapping anxieties are accounted for by modeling separate DSM and distinct categorical variables. Reference group is always indicated first.
% Diff, percent difference; ASD, autism spectrum disorder; ASD-DIST, autism spectrum disorder with distinct anxiety; ASD-DSM, autism spectrum disorder with DSM anxiety; ASD-no DIST, autism spectrum disorder without distinct anxiety; ASD-no DSM, autism spectrum disorder without DSM anxiety; Coef., coefficient; CSR, clinical severity rating; FDR p, false discovery rate–adjusted p value; T1, time 1; T4, time 4.
Differences Between Autistic Children With Only DSM Anxiety, Only Distinct Anxiety, Both DSM and Distinct Anxieties, and No Clinical Anxiety and TD Children
The above analyses indicate differential associations of amygdala volume between autistic children with DSM (increased volume) and distinct (decreased volume) anxieties. Given the overlap within participants of DSM and distinct anxiety diagnoses and to investigate the relationship of amygdala volume with TD children, we conducted a secondary analysis of amygdala volume and development between five groups of interest: 1) autistic children with only DSM anxieties (ASD-DSM) (n = 16), 2) autistic children with only distinct anxieties (ASD-distinct) (n = 11), 3) autistic children with both DSM and distinct anxieties (ASD-both anxieties) (n = 16), 4) autistic children with no clinical anxiety (ASD-no anxiety) (n = 28), and 5) TD children without clinical anxiety (n = 55) (Figure 2, Table 3, and Table S3).
Figure 2Associations between amygdala volume and different forms of anxiety in autism spectrum disorder (ASD). Longitudinal development of (A) left and (B) right amygdala volumes are plotted for autistic children with only DSM anxiety (ASD-DSM ANX), with only autism-distinct anxiety (ASD-DIST ANX), with both DSM and distinct anxieties (ASD-BOTH ANX) and without anxiety (ASD-no ANX) and typically developing (TD) children without anxiety. Vertical dotted lines indicate mean ages at study time 1 (38.11 mo/3.18 years) and time 4 (136.30 mo/11.36 years). ASD-DSM ANX was found to have significantly larger right amygdala volumes at both study time 1 (4.59% larger, p = .012, false discovery rate [FDR]–adjusted p = .044) and time 4 (5.19% larger, p = .01, FDR-adjusted p = .044) compared with TD children. ASD-DIST ANX had a significantly altered developmental trajectory of the right amygdala marked by slower growth compared with the TD (p = .009, FDR-adjusted p = .044), ASD-no ANX (p = .009, FDR-adjusted p = .044), and ASD-DSM ANX (p = .007, FDR-adjusted p = .044) groups, which resulted in significantly smaller right amygdala volumes at time 4 compared with the ASD-no ANX (−8.85% smaller, p = .004, FDR-adjusted p = .044), ASD-DSM (−11.04% smaller, p < .001, FDR-adjusted p = .013), and ASD-both anxieties (−8.26% smaller, p = .011, FDR-adjusted p = .044) groups.
Age-by-group: ASD-DSM ANX vs. ASD-DSM and DIST ANX
254.07
583.68
278
0.44
.664
.796
–
Regression results for models of left and right amygdala volume between children with autism spectrum disorder with only DSM anxiety (ASD-DSM ANX), only autism distinct anxiety (ASD-DIST ANX), both anxieties (ASD-DSM and DIST ANX), without anxiety (ASD-no ANX), and TD. Effects of interest are shown for mean age of T1 (38.11 mo/3.18 years) and T4 (136.30 mo/11.36 years) as well as developmental differences (age-by-group interactions). Reference group is always indicated first. Regressions were repeated switching reference group to provide all group comparisons.
% Diff., percent difference; ANX, anxiety; ASD, autism spectrum disorder; ASD-DIST ANX, autism spectrum disorder with distinct anxiety; ASD-DSM ANX, autism spectrum disorder with DSM anxiety; Coef., coefficient; FDR p, false discovery rate–adjusted p value; T1, time 1, T4, time 4; TD, typically developing.
The ASD-DSM group had significantly larger right amygdala volumes at both study time 1 (5.10% mean increase, p = .008, FDR, p = .038) and time 4 (6.11% mean increase, p = .009, FDR, p = .038) compared with the TD group. No differences in amygdala developmental trajectories between the ASD-DSM group and ASD-no anxiety or TD groups were observed. The ASD-distinct group was found to have a significantly altered developmental trajectory of the right amygdala marked by slower growth compared with the TD (p = .009, FDR p = .038), ASD-no anxiety (p = .009, FDR p = .038), and ASD-DSM (p = .006, FDR p = .038) groups. Slower right amygdala development in the ASD-distinct group resulted in significantly smaller right amygdala volume at time 4 (11.36 years) compared with the ASD-no anxiety (−8.13% mean decrease, p = .004, FDR p = .038) and ASD-DSM (−12.05% mean decrease, p < .001, FDR p = .010) groups. No results for the left amygdala reached statistical significance after FDR correction.
Effects of IQ and Autism Severity on Anxiety Group Differences
No significant differences in IQ or ADOS calibrated severity score measures were found between the four autism groups (ANOVA p > .05); inclusion of these variables within models did not change statistically significant differences between the groups, and they were not significantly associated with amygdala volumes. Compared with TD children, children with autism showed nonsignificant increases in left (t125 = 1.85, p = .066) and right (t125 = 1.76, p = .079) hemispheric volumes. Using the same five-group mixed-effect model structure, no significant differences between groups in either left or right hemisphere volumes were observed at time 1 or time 4 or developmentally (i.e., age-by-group interactions), suggesting that our findings are not due to global hemispheric effects (Table S1). No significant age-by-group-by-baseline amygdala volume effects between the distinct anxiety and other groups were found (Table S2).
Post Hoc Behavioral Analyses
To determine if differences in behaviors associated with autism could explain amygdala differences between the DSM and distinct anxiety groups, we performed exploratory post hoc analyses to investigate differences between the ASD-DSM, ASD-distinct, ASD-both anxieties, and ASD-no anxiety groups in measures including the CBCL subscales (
The Child Behavior Checklist and related instruments.
in: Maruish M.E. The Use of Psychological Testing for Treatment Planning and Outcomes Assessment. 2nd ed. Lawrence Erlbaum Associates Publishers. 429–466,
Mahwah, NJ1999
). At time 4, the ASD-DSM and ASD-both anxieties groups had significantly higher CBCL anxious depressed and DSM anxiety problems t scores compared with the ASD-distinct and ASD-no anxiety groups (ANOVA, p < .05; Tukey honestly significant difference, p < .05). These results indicate concordance between CBCL and ADIS assessments of DSM anxiety. The ASD-DSM group also showed significantly higher CBCL internalizing behavior scores compared with the ASD-distinct and ASD-no anxiety groups (ANOVA, p < .05; Tukey honestly significant difference, p < .05). In addition, at time 4, autism groups with anxiety (of any kind) had higher measures of CBCL thought problems, total SSP 2, and total RBS scores than the ASD-no anxiety group; however, Tukey honestly significant difference tests showed these elevated scores to be significant for only the ASD-both anxieties group (CBCL thought problems and RBS) or the ASD-both and ASD-DSM groups (SSP 2) compared with the ASD-no anxiety group (ANOVA, p < .05; Tukey honestly significant difference, p < .01). These findings indicate that autistic children with DSM and/or distinct anxiety experience elevated levels of sensory sensitivities, unusual thought processes, and repetitive and restricted behaviors—all potential indications of elevated distress in anxious children with autism (
The relationship between anxiety, external structure, behavioral history and becoming locked into restricted and repetitive behaviors in autism spectrum disorder.
). The ASD-both anxieties group also had significantly higher time 4 SRS scores than the ASD-no anxiety group (ANOVA, p = .02; Tukey honestly significant difference, p < .01). No significant differences between groups in CBCL subscales, total SRS, RBS, or SSP were observed at time 1 (Tables S3–S6).
Discussion
The primary aim of this study was to characterize amygdala volume and development across childhood in relation to different types of anxiety in children with autism. We examined two categories of problematic anxiety in ASD: traditional DSM anxieties and anxieties distinct to autism contexts (
Clinically significant anxiety in children with autism spectrum disorder and varied intellectual functioning [published correction appears online ahead of print, J Clin Child Adolesc Psychol, Mar 27, 2020].
). Initial analyses comparing autistic children with and without these anxieties revealed DSM anxiety to be associated with enlarged amygdala volumes and distinct anxieties to be associated with smaller amygdala volumes. A second analysis comparing TD children and four subgroups of children with autism (only DSM anxiety, only distinct anxiety, both anxieties, and no anxiety) found DSM anxiety to be associated with larger right amygdala volumes compared with TD children at both ∼3 (time 1) and ∼11 (time 4) years of age but no differences in developmental trajectory. Autistic children with distinct anxiety had slower development of the right amygdala from the ages of ∼3 to 11 compared with TD and other autistic children and had smaller right amygdala volumes at ∼11 years of age compared with other autistic children. These results support an association, albeit a complex one, between amygdala volume, ASD, and co-occurring anxiety. Our results also identify a novel association between the development of amygdala volume and anxieties distinctly related to autism.
The only three previous studies of amygdala volumes and anxiety in ASD report conflicting results, finding both larger (
). These results support traditional DSM anxieties being associated with larger amygdala volumes. Autism-distinct anxieties are estimated to occur at relatively high frequencies in children with autism (
Clinically significant anxiety in children with autism spectrum disorder and varied intellectual functioning [published correction appears online ahead of print, J Clin Child Adolesc Psychol, Mar 27, 2020].
) and have not been previously accounted for in imaging research. Accordingly, associations of smaller amygdala volumes in autistic children with these distinct anxieties may partially explain inconsistent findings in ASD.
We found that autistic children with anxiety have elevated sensory sensitivities (SSP), RBS, and thought problems (CBCL), which may be indicative of elevated levels of distress. We also noted that autistic children with DSM anxiety had higher CBCL internalizing behavior scores than those with distinct anxiety or no anxieties. However, no differences in CBCL internalizing behaviors were observed between the distinct anxiety and no anxiety groups. Despite finding no differences in measures of core autism features (e.g., ADOS calibrated severity score) between children with and without distinct anxieties, we hypothesize that a latent variable related to the autism phenotype contributes to both autism-distinct anxieties and smaller amygdala volumes. Others have reported smaller amygdala volumes in ASD to be associated with decreased joint attention, eye fixation, and emotional face processing speeds (
). Emergence of smaller right amygdala volumes due to slower development is indicative of a brain-behavior relationship between amygdala development and onset of distinct anxieties. Replication and future studies are needed to further examine the phenomenology of distinct anxieties and the ways in which their behavioral and neurobiological profiles relate to autism and vary from those of DSM anxiety.
We found that autistic children with DSM anxiety had the largest amygdala volumes compared with other groups, with significantly larger right amygdala volumes than TD children at both ∼3 and ∼11 years of age. It is important to emphasize that amygdala enlargement in autistic children with DSM anxiety was already present at 3 years of age, before clinical anxiety is typically diagnosed. This and the finding that there were no differences in the trajectory of amygdala development between ∼2.5 and 12 years would suggest that the process responsible for enlarged amygdala related to DSM anxiety occurred either prenatally or during an early postnatal period. This indicates that enlarged amygdala may be a potential prodromal marker of anxiety in autism. Amygdala enlargement also predated elevated CBCL measures of anxiety in the ASD-DSM group; however, this may be attributable to low sensitivity of the CBCL within a sample of children with autism who were likely to be developmentally below the cognitive start age of the CBCL (1.5 years) at study enrollment (
Clinically significant anxiety in children with autism spectrum disorder and varied intellectual functioning [published correction appears online ahead of print, J Clin Child Adolesc Psychol, Mar 27, 2020].
Autistic children without anxiety also showed statistical trends toward having larger amygdala volumes than TD children, while the group of autistic children with both DSM and distinct anxieties had marginally smaller amygdala volumes than autistic children without anxiety. Opposing associations of amygdala volume with DSM (larger) and distinct (smaller) anxieties is also supported by autistic children with both forms of anxiety having amygdala volumes between those of the DSM and distinct anxiety groups. Further studies with larger samples will be needed to confirm these trends. While the underlying mechanisms contributing to smaller and larger amygdala volumes are unclear, larger volumes could be a product of atypical amygdala neurogenesis, which has been noted in ASD (
Relationships between amygdala volume and activity during emotional information processing tasks in depressed and never-depressed individuals: An fMRI investigation.
) could contribute to stunted amygdala development. The process of untangling these various factors may be difficult because we suspect that neurophenotypic differences between autistic individuals with and without anxiety arise from complex interactions between biological and environmental variables that likely differ on an individual basis.
Significant differences in amygdala volume between autistic individuals with and without autism-distinct anxiety were found bilaterally. However, effects between TD and different autism anxiety groups (DSM only, distinct only, both DSM and distinct, and no anxiety) were only significant within the right hemisphere. This is consistent with the two previous studies that have reported relationships between anxiety and amygdala volume in autism (
). Meta-analyses have reported predominately leftward lateralization of amygdala activation in response to various emotional processing tasks, which has been postulated to result from increased right amygdala habituation (
). Indeed, emotional processing functional imaging studies suggest the right amygdala to be more engaged in rapid processing of potentially threatening stimuli and the left in prolonged stimulus evaluation (
). Thus, while the amygdala is likely to be affected bilaterally in autistic children with anxiety, a failure of the right amygdala to habituate to anxiolytic stimuli may contribute to these lateralized findings.
Consistent with the Research Domain Criteria framework (
), previous findings suggest that effects of DSM forms of anxiety on the amygdala may cut across diagnostic boundaries. For example, functional MRI studies report increased amygdala activation in response to facial processing tasks and resting-state amygdala-prefrontal decoupling to be related to anxiety both in ASD (
) populations. Furthermore, a recent longitudinal study found positive associations of amygdala volume with anxiety levels from 4 to 18 years of age in TD children (
). This study was limited by a lack of a control group comprising nonautistic individuals with clinically significant DSM anxiety. However, we would hypothesize such amygdala enlargement in nonautistic anxious individuals based on our findings of larger amygdala volumes being associated with DSM anxieties in ASD. In this view, divergent findings of smaller amygdala volumes associated with distinct anxieties may indicate that these manifestations of anxiety are not only distinctly related to autism but also distinct from traditional DSM classification of anxieties. However, given that distinct anxieties are intrinsically linked to autism and less commonly reach clinical levels in individuals without autism, direct comparison of distinct anxieties between autistic and nonautistic groups will be challenging.
This study benefited from a longitudinal design that allowed for the characterization of amygdala growth across early to midchildhood as well as clinician-based assessments of both traditional DSM and autism-distinct anxieties. However, it is important to note limitations. The individual anxiety groups had relatively small sample sizes. Both DSM and distinct anxieties incorporate multiple anxiety subtypes that may, in themselves, have different developmental effects on amygdala volume (
); our sample sizes precluded investigating differences between individual forms of anxiety. Complicating this further are multiple co-occurring anxiety diagnoses. For example, of the 32 children with autism in our sample with a DSM anxiety diagnosis, 37.5% (n = 12) met ADIS criteria for two or more DSM anxiety diagnoses. We assessed autism-distinct anxiety using the ADIS-ASA, a diagnostic clinical interview. However, future studies could use autism-specific parental and patient anxiety reports, which may provide more dimensional assessments of anxiety and be easier to widely implement across large samples (
). We focused solely on the amygdala due to the structure’s broad implications in both anxiety and ASD. However, the amygdala is highly interconnected with other brain regions and thus investigating a broader neural network critical for social and emotional processing in a multimodal manner would undoubtedly be informative. Higher-resolution imaging would afford the ability to investigate individual contributions of the 13 subnuclei, which are exceedingly challenging to segment given the standard ∼1-mm3 resolution of current structural MRI scans. Finally, consistent with previous findings in TD (
A longitudinal MRI study of amygdala and hippocampal subfields for infants with risk of autism.
in: Zhang D. Zhou L. Jie B. Liu M. Graph Learning in Medical Imaging: First International Workshop, GLMI 2019, Held in Conjunction with MICCAI 2019, Shenzen, China, October 17, 2019, Proceedings. Springer International Publishing,
Cham, Switzerland2019: 164-171
), we found a significant effect of sex, with males having larger amygdala volumes than females. This study included an insufficient number of females with autism to investigate sex-by-anxiety interactions in amygdala volumes, which is critically important given findings by our group indicating sex-specific relationships of amygdala volumes with psychopathology (
In conclusion, this study aimed to investigate the effect of different forms of anxiety on amygdala volume and development in children with autism. Traditional DSM forms of anxiety were found to be associated with larger right amygdala volumes while anxieties distinctly related to autism were associated with smaller right amygdala volumes and slower right amygdala development. While additional studies are needed to clarify amygdala-anxiety relationships, considering previous findings, these results support DSM anxiety having common effects on amygdala volume across diagnostic classifications. Opposing amygdala volumetric relationships between autism-distinct anxieties compared with DSM anxieties suggest that these autism-related anxiety presentations may be associated with a distinct syndrome of anxiety closely related to the autistic phenotype. Collectively, these results indicate that the amygdala is an important brain region for future efforts to identify and stratify those individuals with autism who endure debilitating co-occurring anxiety.
Acknowledgments and Disclosures
This research was supported by an Autism Center of Excellence grant awarded by the National Institute of Child Health and Development (Grant No. P50 HD093079 [to DGA, CWN, MS, SJR] ), the National Institute of Mental Health (Grant Nos. R01MH104438 [to CWN], R01MH103284 [to MS], and R01MH103371 [to DGA]), the Medical Investigation of Neurodevelopmental Disorders (MIND) Institute Intellectual and Developmental Disabilities Research Center (Grant No. P50 HD103526 ), the MIND Institute Autism Research Training Program (Grant No. T32MH073124 [to DSA]), and a Medical Research Council eMedLab Medical Bioinformatics Career Development Fellowship (Grant No. MR/L016311/1 [to AA]).
We thank the families and children who participated in the Autism Phenome Project study and all members of the research study staff.
DGA is on the scientific advisory board for Stemina Biomarker Discovery and Axial Biotherapeutics. CMK receives royalties for an edited book on anxiety and autism published by Academic Press. In addition, she has received honoraria and consulting fees for training other researchers on the Autism Spectrum Addendum. All other authors report no biomedical financial interests or potential conflicts of interest.
Clinically significant anxiety in children with autism spectrum disorder and varied intellectual functioning [published correction appears online ahead of print, J Clin Child Adolesc Psychol, Mar 27, 2020].
Anxiety Disorders Interview Schedule–Autism Addendum: Reliability and validity in children with autism spectrum disorder [published correction appears in J Clin Child Adolesc Psychol 2020 49:iv].
Gray matter volume in adolescent anxiety: An impact of the brain-derived neurotrophic factor Val(66)Met polymorphism? [published correction appears in J Am Acad Child Adolesc Psychiatry 2013; 52:329].
J Am Acad Child Adolesc Psychiatry.2013; 52: 184-195
The Child Behavior Checklist and related instruments.
in: Maruish M.E. The Use of Psychological Testing for Treatment Planning and Outcomes Assessment. 2nd ed. Lawrence Erlbaum Associates Publishers. 429–466,
Mahwah, NJ1999
Schedule for Affective Disorders and Schizophrenia for School-age Children-Present and Lifetime Version (K-SADS-PL): Initial reliability and validity data.
J Am Acad Child Adolesc Psychiatry.1997; 36: 980-988
A longitudinal MRI study of amygdala and hippocampal subfields for infants with risk of autism.
in: Zhang D. Zhou L. Jie B. Liu M. Graph Learning in Medical Imaging: First International Workshop, GLMI 2019, Held in Conjunction with MICCAI 2019, Shenzen, China, October 17, 2019, Proceedings. Springer International Publishing,
Cham, Switzerland2019: 164-171
Cortical and subcortical brain morphometry differences between patients with autism spectrum disorder and healthy individuals across the lifespan: Results from the ENIGMA ASD working group.
Autism Diagnostic Interview-Revised: A revised version of a diagnostic interview for caregivers of individuals with possible pervasive developmental disorders.
The relationship between anxiety, external structure, behavioral history and becoming locked into restricted and repetitive behaviors in autism spectrum disorder.
Relationships between amygdala volume and activity during emotional information processing tasks in depressed and never-depressed individuals: An fMRI investigation.