Psychosocial Stress and Brain Function in Adolescent Psychopathology
Abstract
Objective:
The authors sought to explore how conduct, hyperactivity/inattention, and emotional symptoms are associated with neural reactivity to social-emotional stimuli, and the extent to which psychosocial stress modulates these relationships.
Method:
Participants were community adolescents recruited as part of the European IMAGEN study. Bilateral amygdala regions of interest were used to assess the relationship between the three symptom domains and functional MRI neural reactivity during passive viewing of dynamic angry and neutral facial expressions. Exploratory functional connectivity and whole brain multiple regression approaches were used to analyze how the symptoms and psychosocial stress relate to other brain regions.
Results:
In response to the social-emotional stimuli, adolescents with high levels of conduct or hyperactivity/inattention symptoms who had also experienced a greater number of stressful life events showed hyperactivity of the amygdala and several regions across the brain. This effect was not observed with emotional symptoms. A cluster in the midcingulate was found to be common to both conduct problems and hyperactivity symptoms. Exploratory functional connectivity analyses suggested that amygdala-precuneus connectivity is associated with hyperactivity/inattention symptoms.
Conclusions:
The results link hyperactive amygdala responses and regions critical for top-down emotional processing with high levels of psychosocial stress in individuals with greater conduct and hyperactivity/inattention symptoms. This work highlights the importance of studying how psychosocial stress affects functional brain responses to social-emotional stimuli, particularly in adolescents with externalizing symptoms.
Common mental health problems that emerge during adolescence, such as symptoms of depression, anxiety, attention deficit hyperactivity disorder (ADHD), and conduct problems, are frequent and debilitating (1). These symptom domains are associated with negative adult outcomes, including substance dependence (2–4), familial discord (2, 4), poor educational attainment (2, 5), and poor vocational attainment (2, 4, 5). Psychosocial stress is an important contributor to the emergence of child and adolescent psychopathology. Family, personal, and interpersonal stressors as well as trauma have been associated with externalizing symptoms, such as conduct problems and ADHD symptoms, as well as internalizing symptoms of depression and anxiety (6, 7). Therefore, identifying the interplay between symptoms of psychopathology, the environment, and biology is important to help prevent or ameliorate mental illness.
Adolescents with externalizing symptoms have difficulties with emotion recognition, particularly anger and disgust (8, 9), while people with internalizing symptoms are more accurate at recognizing sad and angry faces and tend to misinterpret neutral faces as angry or sad (10). Functional neuroimaging research suggests that the amygdala, an area crucial for emotional processing and emotional response, exhibits different activation patterns in individuals with externalizing and internalizing symptoms compared with control subjects (11–15). While these findings provide insight into neural differences associated with both internalizing and externalizing psychopathology, they have not taken into account important moderators, such as psychosocial stress, which itself is known to increase face processing–related amygdala activation (16–18). Despite evidence that psychopathology and psychosocial stress may affect emotion perception at behavioral and neural levels, to our knowledge there have been no studies exploring how psychopathology-related symptoms and psychosocial stress interact to modulate amygdala activation related to emotional processing. We therefore sought to better understand the effects of psychosocial stress and psychopathology on brain responses to angry and ambiguous faces, as well as the interaction of stress and psychopathology on brain function. We conducted amygdala-based region-of-interest analyses and functional connectivity analyses using psychophysiological interactions, as well as exploratory whole brain functional MRI (fMRI) analyses, in 1,288 adolescents from the IMAGEN study, a large community-recruited European cohort. We hypothesized that the relationship between adolescent externalizing symptoms (conduct symptoms, hyperactivity/inattention symptoms) and internalizing symptoms (emotional symptoms) and neural reactivity to emotional stimuli is influenced by the experience of psychosocial stress.
Method
Participants
After quality control checks for neuroimaging and behavioral tests, data for 1,288 community-recruited adolescents were eligible; 583 of the participants were male, and the sample’s mean age was 14.4 years (SD=0.40; range=13.2–15.4). Participants were assessed at eight study sites in England, France, Germany, and Ireland. Each site sought and received approval from the relevant local research ethics committee. Written consent was obtained from each participant and a parent or guardian. A detailed description of recruitment and assessment methods has been published elsewhere (19).
Clinical Symptoms
Because the adolescents were recruited from the community, we used continuous severity scores of internalizing and externalizing symptoms using the Strengths and Difficulties Questionnaire (20). For externalizing symptoms, we used the conduct symptom and hyperactivity/inattention symptom subscales. For internalizing symptoms, we used the emotional symptom subscale. Combined symptom scores from both the adolescents and their parents were calculated for these three symptom domains for the 6 months prior to the assessment (for more details, see the data supplement that accompanies the online edition of this article). The majority of participants’ scores were within the average range (see Table S1 in the data supplement).
Psychosocial Stress Score
A self-report measure (the Life Events Questionnaire) was used to record the occurrence of stressful events, both lifetime and during the previous 12 months (21). A score was calculated for number of stressful life events that had occurred during the previous 12 months only to avoid inaccurate recall (see the data supplement).
Additional Covariates
Covariates of no interest included pubertal status (22); socioeconomic status, indexed using the family stresses subsection of the Development and Well-Being Assessment (23); and verbal IQ, based on the WISC-IV (24). Substance use was measured using the European School Survey Project on Alcohol and Drugs (25) and was defined in a binary fashion (ever/never smoked cigarettes, drank alcohol, or used drugs). Drugs included cannabis, glue, tranquilizers, amphetamines, LSD, mushrooms, cocaine, crack cocaine, heroin, narcotics, MDMA, ketamine, GHB, and anabolic steroids.
Emotional Reactivity fMRI Task
The task used in the study was adapted from Grosbras and Paus (26). Participants watched 18-second blocks of either a face movie (depicting anger or neutrality) or a control stimulus. Each face movie showed black-and-white video clips (200–500 ms) of male or female faces. Five blocks each of angry and neutral expressions were interleaved with nine blocks of the control stimulus. Each block contained eight trials of six face identities (three of them female). The same identities were used for the angry and neutral blocks. The control stimuli were black-and-white concentric circles expanding and contracting at various speeds that closely matched the contrast and motion characteristics of the face clips (see Figure S1 in the data supplement).
Although some groups report significant activation in neural structures involved in threat detection, such as the amygdala (11, 27), findings have been mixed regarding neural reactivity to neutral stimuli in people with and without mental health problems. Therefore, we explored neural reactivity associated with the neutral stimuli in our task and found that while our target region of interest (the amygdala) was significantly activated in two of the contrasts (angry faces versus control stimulus and neutral faces versus control stimulus), there was no significant activation of the amygdala in the angry-versus-neutral faces contrast (see Table S2A in the data supplement). As a result, we proceeded with the analysis in the angry faces-versus-control and neutral faces-versus-control contrasts, acknowledging that the analysis would be of the neural response to an angry or neutral face as a whole, rather than specifically isolating the emotion.
fMRI Acquisition and Processing
Structural and functional MRI data were acquired with 3-T MRI scanners (Siemens, Philips, GE, and Bruker). Four sites (using GE and Philips scanners) used an eight-channel coil, and four sites (using Siemens scanners) used a 12-channel coil. All sites used the same scanning protocol. High-resolution T1-weighted three-dimensional structural images were acquired for anatomical localization and registration with the functional time series. Data were preprocessed centrally (at NeuroSpin-CEA, Gif-sur-Yvette, France) using SPM8 (http://www.fil.ion.ucl.ac.uk/spm/); see the data supplement for further information. Individuals with anatomical abnormalities or poor realignment (e.g., greater than 3 mm of head motion in at least one of the translations; N=20) did not pass quality control and were not included in these analyses.
Statistical Analysis
Behavioral analysis.
We used separate multiple regression models to establish the relationship between each of the symptoms and the number of stressful life events. Sex, study site, verbal IQ, socioeconomic status, and pubertal status were included as covariates of no interest in all analyses.
fMRI analysis.
fMRI data were analyzed using SPM8, revision 6313. To define a functional amygdala region of interest, we used a separate group of IMAGEN participants (N=326) for whom full phenotypic data were not available (for demographic characteristics, see Table S2B in the data supplement). There was robust bilateral activation (see the data supplement). Using the peak Montreal Neurological Institute coordinates, we created an amygdala region of interest with an 8-mm sphere using MarsBaR (28; http://marsbar.sourceforge.net/) and extracted summarized beta values in the full sample for both contrasts for analysis in SPSS.
Using separate regression models, we explored the extent to which the psychosocial stress score and symptom count scores for each of the three symptoms (conduct, hyperactivity/inattention, and emotional) were associated with fMRI activation in the amygdala as main effects. We also explored the extent to which psychosocial stress moderated the relationship between each symptom and amygdala activation as an interaction. Independent variables (conduct, hyperactivity/inattention, and emotional symptoms and psychosocial stress score) were mean centered. Interaction terms were calculated in SPSS for input into the second-level regression models. To adjust for multiple testing, we applied a Bonferroni correction threshold of p<0.05/12 (two amygdala regions of interest, two fMRI contrasts, and three symptom domains).
To account for comorbidity among the three symptoms, we carried out post hoc regression models for each symptom while controlling for the other two symptoms. We also carried out post hoc regression models to examine the effects of site, sex, and substance use on the results.
We conducted exploratory psychophysiological interaction (PPI) functional connectivity analyses to investigate potentially distinct amygdala networks involved in the symptom-by-stress results. PPI analyses compute the interaction between the seed blood-oxygen-level-dependent (BOLD) time series and a chosen condition-specific interaction factor when predicting each voxel BOLD time series. Generalized PPI regression analyses were carried out via the SPM-based CONN toolbox (29; http://www.nitrc.org/projects/conn/) using the same amygdala region of interest as our a priori seed region (see the data supplement) and the same covariates of no interest. To adjust for multiple testing, we applied the same Bonferroni correction (p<0.05/12).
We conducted exploratory whole brain multiple regression analyses of the same main and interaction effects, using the same covariates of no interest as in the region-of-interest and PPI analyses. To adjust for multiple testing we applied the same Bonferroni correction (p<0.05/12).
Results
Behavioral Analysis
We used regression models to establish the relationship between conduct, hyperactivity/inattention, and emotion symptoms and the psychosocial stress score. Among these three symptom domains, psychosocial stress had the strongest association with conduct symptoms (t=10.55; model r2=0.12, p=5.51×10−25), followed by hyperactivity/inattention (t=8.36; model r2=0.11, p=1.59×10−16) and emotional symptoms (t=4.7; model r2=0.08, p=3.0×10−6). Other significant predictors for psychopathology symptoms included verbal IQ for all symptom domains, male sex for conduct and hyperactivity/inattention symptoms, and socioeconomic status and female sex for emotional symptoms (see Table 1). Descriptive statistics and differences by site are provided in the data supplement (see the Supplemental Results section and Table S3A,B).
| Test Statistics for Symptom Domains | |||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|
| Descriptive Statistics | Conduct Symptoms | Hyperactivity/Inattention Symptoms | Emotional Symptoms | ||||||||
| Variable | Mean | SD | β | t | p | β | t | p | β | t | p |
| Sex | –0.12 | –3.70 | 2.25×10–4 | –0.06 | –1.99 | 0.047 | 0.17 | 5.09 | 4.03×10–7 | ||
| Stressful life event frequency | 3.35 | 2.08 | 0.29 | 10.55 | 5.51×10–25 | 0.23 | 8.36 | 1.59×10–16 | 0.13 | 4.70 | 3.0×10–6 |
| Verbal IQ | 111.98 | 14.82 | –0.10 | –3.50 | 4.83×10–4 | –0.17 | –5.90 | 4.53×10–9 | –0.10 | –3.52 | 4.53×10–4 |
| Socioeconomic status | 0.66 | 1.06 | 0.04 | 1.50 | 0.13 | 0.01 | 0.36 | 0.72 | 0.11 | 3.82 | 1.4×10–4 |
| Puberty development stage | 3.64 | 0.70 | 0.01 | 0.25 | 0.81 | –0.03 | –0.79 | 0.43 | –0.05 | –1.55 | 0.12 |
| Conduct problems | 2.49 | 1.58 | |||||||||
| Hyperactivity/inattention problems | 4.34 | 2.17 | |||||||||
| Emotional problems | 2.24 | 2.30 | |||||||||
TABLE 1. Demographic and Clinical Symptom Characteristics of 1,288 Adolescents From the IMAGEN Studya
fMRI Region-of-Interest Analysis
To understand the relationship between the three symptom domains, psychosocial stress, and amygdala reactivity to our face stimuli, we conducted region-of-interest fMRI regression analyses in the left and right amygdalae. We first examined these relationships in all participants and then examined sex differences post hoc. As sex was a covariate in the regression models, we also checked whether there were any main effects of sex on amygdala activation but found none. We examined sex differences in amygdala activation during emotional face processing (30) and found that males had greater right amygdala activation during the angry-versus-control contrast (t=2.82, family-wise error-corrected p=0.005, Bonferroni-corrected p=0.04). Including substance use in the regression models did not change the results (see the Method section and the data supplement). As no clusters survived whole brain correction, analyses were thresholded at p<0.001 (voxel-level uncorrected), and statistically significant clusters were reported at p<0.05 (family-wise error corrected).
All main effect and interaction test statistics are shown in Table 2. There were significant main effects neither of conduct, hyperactivity/inattention, and emotional symptoms nor of psychosocial stress on amygdala activation in either the angry-versus-control or neutral-versus-control fMRI contrasts. We did, however, find several interactions, described below.
| Angry Faces Versus Control Stimulus, Left Amygdala | Angry Faces Versus Control Stimulus, Right Amygdala | Neutral Faces Versus Control Stimulus, Left Amygdala | Neutral Faces Versus Control Stimulus, Right Amygdala | |||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Measure | β | t | p | β | t | p | β | t | p | β | t | p |
| Conduct problems | ||||||||||||
| Main effect | 0.008 | 0.272 | 0.786 | –0.048 | –1.59 | 0.111 | –0.015 | –0.517 | 0.605 | 0.003 | 0.109 | 0.913 |
| Stress main effect | –0.001 | –0.027 | 0.978 | 4×10–4 | –0.013 | 0.99 | –0.037 | –1.22 | 0.224 | –0.032 | –1.04 | 0.297 |
| Conduct × stress | 0.081 | 2.86 | 0.0044 | 0.089 | 3.11 | 0.0019 | 0.093 | 3.27 | 0.001 | 0.081 | 2.84 | 0.005 |
| Hyperactivity/inattention problems | ||||||||||||
| Main effect | –0.004 | –0.125 | 0.901 | 0.007 | 0.224 | 0.823 | –0.047 | –1.60 | 0.110 | 0.019 | 0.638 | 0.523 |
| Stress main effect | 0.009 | 0.315 | 0.753 | –0.005 | –0.167 | 0.867 | –0.020 | –0.664 | 0.507 | –0.026 | –0.887 | 0.375 |
| Hyperactivity × stress | 0.091 | 3.28 | 0.0011 | 0.062 | 2.23 | 0.026 | 0.070 | 2.49 | 0.013 | 0.070 | 2.52 | 0.012 |
| Emotional problems | ||||||||||||
| Main effect | 0.016 | 0.542 | 0.588 | –0.001 | –0.045 | 0.964 | 0.035 | 1.20 | 0.230 | 0.055 | 1.88 | 0.060 |
| Stress main effect | 0.008 | 0.262 | 0.793 | –0.003 | –0.092 | 0.927 | –0.034 | –1.16 | 0.246 | –0.031 | –1.05 | 0.296 |
| Emotional × stress | 0.045 | 1.57 | 0.117 | 0.032 | 1.14 | 0.254 | 0.034 | 1.18 | 0.240 | 0.052 | 1.82 | 0.070 |
TABLE 2. Results of Amygdala Region-of-Interest Regression Analyses
Conduct symptoms.
We found a significant interaction between conduct symptoms and psychosocial stress score in the right amygdala (t=3.11, family-wise error-corrected p=0.002, Bonferroni-corrected p=0.024) in the angry-versus-control contrast (Figure 1A). The greater the number of stressful life events experienced by individuals with severe conduct symptoms (see Table S1 in the data supplement), the greater the amygdala activation. The interaction in the left amygdala was not significant (family-wise error-corrected p=0.053). We also found an interaction effect in the neutral-versus-control contrast in the left amygdala (t=3.24, family-wise error-corrected p=0.0012, Bonferroni-corrected p=0.014) (Figure 1B), but not in the right amygdala (family-wise error-corrected p=0.06). Controlling for hyperactivity and emotional symptoms did not affect the result in either the angry-versus-control contrast (t=3.12, family-wise error-corrected p=0.002, Bonferroni-corrected p=0.024) or the neutral-versus-control contrast (t=3.21, family-wise error-corrected p=0.0014, Bonferroni-corrected p=0.017).
FIGURE 1. Scatterplots Depicting Interaction Effects Between Conduct and Hyperactivity/Inattention Symptoms and Stress on Contrast Estimates of Angry or Neutral Faces Versus Control Stimulus in the Amygdala in Adolescentsa
a BOLD=blood-oxygen-level-dependent.
To explore sex differences in these results, we split the sample and compared the interaction of conduct symptoms and psychosocial stress on amygdala activation in males and females. Males had a stronger symptom-by-stress interaction on amygdala activation (right amygdala, angry-versus-control contrast: r=0.133, family-wise error-corrected p=0.001, Bonferroni-corrected p=0.024; left amygdala, neutral-versus-control contrast: r=0.146, family-wise error-corrected p=0.0005, Bonferroni-corrected p=0.012) than females (right amygdala, angry-versus-control contrast: r=0.022, family-wise error-corrected p=0.558; left amygdala, neutral-versus-control contrast: r=0.030, family-wise error-corrected p=0.424). These differences between the sexes were significant (right amygdala, angry-versus-control contrast: two-tailed Fisher’s Z=1.99, p=0.047; left amygdala, neutral-versus-control contrast: Z=2.09, p=0.037).
Hyperactivity/inattention symptoms.
We identified the same interaction effect between hyperactivity symptoms and psychosocial stress in the left amygdala (t=3.28, family-wise error-corrected p=0.0011, Bonferroni-corrected p=0.013) in the angry-versus-control contrast (Figure 1C). The interaction in the right amygdala was not significant (family-wise error-corrected p=0.312). There was no significant interaction in the neutral-versus-control contrast after correction for multiple testing (Table 2). Controlling for conduct and emotional symptoms did not affect the result (t=3.18, family-wise error-corrected p=0.002, Bonferroni-corrected p=0.024). There were no sex differences.
Emotional symptoms.
We found no significant symptom-by-stress interactions in amygdala activation in either fMRI contrast (Table 2). The main effects and interactions remained nonsignificant when controlling for conduct and hyperactivity symptoms. There were no sex differences.
PPI Functional Connectivity Analysis
We carried out exploratory PPI functional connectivity analysis to explore potential amygdala networks related to conduct, hyperactivity/inattention, and emotional symptoms, psychosocial stress, and angry/neutral face stimuli. We found a positive relationship between hyperactivity/inattention symptoms and connectivity between the right amygdala and the left ventral precuneus in the angry condition compared with the control condition (family-wise error-corrected p=0.0027, Bonferroni-corrected p=0.032, nonparametric-permuted (5,000 permutations) p=0.026; coordinates: −8, −56, 18; ke=150). There were neither main effects nor any significant symptom-by-stress interactions in the hyperactivity/inattention or emotional symptoms or in the neutral-versus-control contrast (Table 3).
| Contrast and Effect | Positive or Negative PPI | Region | Side | Coordinates of Peak Activationa | t | β | Cluster Size (k) | p |
|---|---|---|---|---|---|---|---|---|
| Angry faces versus control stimulus, left amygdala | ||||||||
| Main effect: conduct symptoms | Positive | Frontal pole | Left | –38, –10, –18 | 5.22 | 0.043 | 117 | 0.016 |
| Angry faces versus control stimulus, right amygdala | ||||||||
| Main effect: hyperactivity/inattention symptoms | Positive | Precuneus | Left | –8, –56, 18 | 4.10 | 0.034 | 150 | 0.0027b |
| Neutral faces versus control stimulus, left amygdala | ||||||||
| Main effect: hyperactivity/inattention symptoms | Positive | Caudate | Left | –6, 10, 0 | 4.51 | 0.042 | 96 | 0.044 |
| Neutral faces versus control stimulus, right amygdala | ||||||||
| Main effect: stress in conduct symptoms model | Positive | Angular gyrus | Right | 52, –50, 28 | 4.99 | 0.040 | 152 | 0.004 |
| Main effect: stress in hyperactivity/inattention symptoms model | Positive | Angular gyrus | Right | 52, –50, 28 | 4.79 | 0.037 | 132 | 0.008 |
| Interaction effect: hyperactivity/inattention symptoms × psychosocial stress frequency | Positive | Frontal pole | Right | 0, 64, 6 | 3.73 | 0.023 | 86 | 0.066 |
| Main effect: stress in emotional symptoms model | Positive | Angular gyrus | Right | 52, –50, 28 | 4.66 | 0.036 | 118 | 0.015 |
TABLE 3. Results of Psychophysiological Interaction (PPI) Functional Connectivity Analyses
fMRI Whole Brain Analysis
We conducted exploratory whole brain regression analyses of the same main and interaction effects as in the region-of-interest and PPI analyses. As no clusters survived whole brain correction, analyses were thresholded at p<0.001 (voxel-level uncorrected) and statistically significant clusters reported at p<0.05 (family-wise error corrected).
Conduct symptoms.
Adolescents with more conduct symptoms showed significantly larger BOLD responses when they had also experienced a greater number of stressful life events. We found a significant interaction between conduct symptoms and stress frequency in the angry-versus-control contrast in the superior temporal gyrus, thalamus, anterior cingulate cortex, superior frontal gyrus, and inferior frontal gyrus (Figure 2A; see also Table S3C in the data supplement). We also found a significant main effect for conduct symptoms in the precuneus and postcentral gyrus but no significant main effect for psychosocial stress score.
FIGURE 2. Statistical Parametric Map Overlaid on a T1-Weighted Structural Brain Imagea
a In panel A, youths with conduct symptoms show increased fMRI blood-oxygen-level-dependent (BOLD) responses with greater psychosocial stress (angry faces versus control stimulus). The image is centered at the middle cingulate cluster (coordinates: 3, 5, 37), and only clusters showing a spatial extent of at least 38 contiguous voxels are shown for visualization purposes. In panel B, youths with hyperactivity/inattention symptoms show increased fMRI BOLD responses with greater psychosocial stress (angry faces versus control stimulus). The image is centered at the middle cingulate cluster (coordinates: −3, 2, 37), and only clusters showing a spatial extent of at least 48 contiguous voxels are shown for visualization purposes. Panel C illustrates the overlap between the two significant midcingulate clusters common to both hyperactivity/inattention and conduct symptoms. The image is centered at coordinates 0.8, 3.5, 37, and ke is 272 voxels.
Hyperactivity/inattention symptoms.
We found a significant interaction between hyperactivity/inattention symptoms and psychosocial stress score in the mid and anterior cingulate cortex. Youths with more symptoms showed larger BOLD responses when they had also experienced more stress (Figure 2B; see also Table S3C in the data supplement).
Emotional symptoms.
There were no significant main or interaction effects for emotional symptoms in either fMRI contrast.
We compared the size and location of the significant clusters from the whole brain analyses of conduct and hyperactivity/inattention symptoms and found an overlapping portion of the midcingulate common to both conduct and hyperactivity symptoms (Figure 2C).
Discussion
In order to comprehensively characterize the relationship between adolescent mental health and psychosocial stress, we examined three behavioral symptom domains in a large sample of community-recruited adolescents. Consistent with previous reports, we found evidence to suggest that psychosocial stress is associated with greater conduct, hyperactivity/inattention, and emotional symptoms. Our primary research objective was to explore how the relationship between these symptoms and neural reactivity to emotional stimuli is influenced by the experience of psychosocial stress. Using an fMRI paradigm designed to target stress-related neural systems, we found that the degree to which adolescent brains respond to social-emotional stimuli depends on the type and severity of conduct and hyperactivity/inattention symptoms and also the amount of stress adolescents had experienced.
Our region-of-interest analyses showed that heightened amygdala response to anger was related to severe conduct and hyperactivity/inattention symptoms only when adolescents also had a higher psychosocial stress score. We observed this finding in the angry-versus-control contrasts and to a lesser extent in the neutral-versus-control contrasts, suggesting that the interaction of conduct symptoms or hyperactivity/inattention and stress not only may be related to affective stimuli, but also may involve processing of social stimuli. This finding may also suggest that previously observed altered response to social-emotional stimuli in adolescents with conduct disorder (11, 31) and ADHD (32) may be related to increased psychosocial stress, and perhaps not to the externalizing symptoms per se. For example, increased amygdala activation is found in stressed youths when they view angry and neutral faces (33).
Our finding of a stress-dependent interaction of attention deficit/hyperactivity and conduct symptoms on social-emotional processing is not limited to heightened amygdala activation alone but, as shown in the results of our whole brain analysis, extends to other brain regions related to the behavioral deficits observed in externalizing behavior. We found increased brain activity related to psychosocial stress and conduct disorder in a network of regions that influence cognitive and emotional processes, including perception (thalamus, superior temporal gyrus, middle temporal gyrus), interpretation (anterior cingulate cortex, insula, superior frontal gyrus), and inhibitory control (inferior frontal gyrus). Furthermore, the midcingulate cortex, an area identified in the promotion of aggressive behavior in response to angry emotional expressions (34), was commonly activated in stressed youths with conduct symptoms and stressed youths with hyperactivity/inattention symptoms. These findings suggest that the impaired social functioning and emotional regulation observed in youths with externalizing symptoms may be regulated by networks of brain activity related to hyperactivity/inattention symptoms and conduct symptoms that are both common and distinct for these symptom domains. These networks involve both cortical and subcortical structures, which is consistent with the behavioral complexity of externalizing symptoms.
In the exploratory PPI analyses, we observed greater functional connectivity between the amygdala and the ventral precuneus in the angry-versus-control condition with increasing hyperactivity/inattention (but not with conduct symptoms) that was independent of psychosocial stress. The precuneus has negative connectivity with the amygdala (35), and increased functional coupling between these regions is important for emotion regulation, particularly distraction (i.e., shifting attention away from emotional stimuli) (36). Our finding is counterintuitive considering that individuals with hyperactivity/inattention symptoms are described as having emotion regulation deficits (32). However, in the context of hyperactivity/inattention symptoms, adolescents may encounter angry faces more frequently, as stress levels are high in parents of ADHD children (37, 38). Therefore, it may be that individuals with hyperactivity/inattention symptoms are more adept at distraction as a means to regulate emotion.
Although psychosocial stress frequency was associated with emotional symptoms, the magnitude of this association was less than for conduct and hyperactivity/inattention symptoms. We found neither main effects of emotional symptoms nor of emotional symptom-by-stress interaction on amygdala activation or connectivity. Considering that the task used in this study has an emotional component and targets stress-related neural systems, the absence of significant findings was unexpected. One reason for the disparity between conduct and hyperactivity/inattention symptoms, stress, and neural reactivity compared with emotional symptoms may lie in the nature of the emotional items from the Strengths and Difficulties Questionnaire and the specific fMRI contrasts used. The emotional subscale items of the Strengths and Difficulties Questionnaire are more reflective of anxiety and irritability than of depression (e.g., “I worry a lot”; “I am nervous in new situations”). Of interest, youths with high anxiety tend to orient away from emotional faces (39, 40), which may help explain our null findings; participants with emotional symptoms may not have been engaged with the fMRI task.
The study was limited by the fMRI paradigm in terms of the range of emotional stimuli available for analysis. Future studies may expand these findings to explore reactivity in response to different emotional expressions. Although it might have been informative, we did not examine or account for potential within-session amygdala habituation effects (41). Although behavioral data during the viewing of angry and neutral faces were not collected, participants were asked after the scanning session to specify whether or not they had seen a set of faces; 99% were found to have good reliability. Site differences were observed in key demographic covariates, such as verbal IQ and socioeconomic status. These may reflect differences in recruitment strategies or in specific cultural attitudes and environment. While we controlled for the effect of site on our results and did not detect any systematic bias, we acknowledge that multisite studies add heterogeneity to the data. Despite these limitations, this study was strengthened by the use of quantitative analysis of clinical symptoms and psychosocial stress frequency, allowing us to explore how all adolescents responded, including those with few or no pathological symptoms, rather than simply those at the high end of the distribution. The exploratory analyses, while not hypothesis driven, are hypothesis generating and in need of further exploration and eventual replication.
Our results highlight the importance of studying how environmental stress affects functional brain responses to social-emotional stimuli, particularly in adolescents with externalizing symptoms. The observed heightened amygdala activation, and associated networks implicated in top-down control of emotion regulation, may affect an individual’s ability to effectively assess risk and may contribute to aggressive or fearful behavioral responses to incoming stimuli. Improved understanding of how stress and externalizing symptoms influence social-affective neurobiological processes may inform the development of therapies that enhance emotional awareness and reduce disproportionate neural reactions in challenging social situations.
1 : The British Child and Adolescent Mental Health Survey 1999: the prevalence of DSM-IV disorders. J Am Acad Child Adolesc Psychiatry 2003; 42:1203–1211Crossref, Medline, Google Scholar
2 : Outcomes of conduct problems in adolescence: 40 year follow-up of national cohort. BMJ 2009; 338:a2981Crossref, Medline, Google Scholar
3 : Prospective effects of attention-deficit/hyperactivity disorder, conduct disorder, and sex on adolescent substance use and abuse. Arch Gen Psychiatry 2007; 64:1145–1152Crossref, Medline, Google Scholar
4 : Characteristics, correlates, and outcomes of childhood and adolescent depressive disorders. Dialogues Clin Neurosci 2009; 11:45–62Medline, Google Scholar
5 : Economic outcomes in adulthood and their associations with antisocial conduct, attention deficit, and anxiety problems in childhood. J Ment Health Policy Econ 2011; 14:137–147Medline, Google Scholar
6 : Common stressful life events and difficulties are associated with mental health symptoms and substance use in young adolescents. BMC Psychiatry 2012; 12:116Crossref, Medline, Google Scholar
7 : Differences in early childhood risk factors for juvenile-onset and adult-onset depression. Arch Gen Psychiatry 2002; 59:215–222Crossref, Medline, Google Scholar
8 : Deficits in facial expression recognition in male adolescents with early-onset or adolescence-onset conduct disorder. J Child Psychol Psychiatry 2009; 50:627–636Crossref, Medline, Google Scholar
9 : Facial expression recognition, fear conditioning, and startle modulation in female subjects with conduct disorder. Biol Psychiatry 2010; 68:272–279Crossref, Medline, Google Scholar
10 : Adolescents’ ability to read different emotional faces relates to their history of maltreatment and type of psychopathology. Clin Child Psychol Psychiatry 2009; 14:237–250Crossref, Medline, Google Scholar
11 : Neural abnormalities in early-onset and adolescence-onset conduct disorder. Arch Gen Psychiatry 2010; 67:729–738Crossref, Medline, Google Scholar
12 : Emotional processing in male adolescents with childhood-onset conduct disorder. J Child Psychol Psychiatry 2008; 49:781–791Crossref, Medline, Google Scholar
13 : Abnormal amygdalar activation and connectivity in adolescents with attention-deficit/hyperactivity disorder. J Am Acad Child Adolesc Psychiatry 2011; 50:828–837.e3Crossref, Medline, Google Scholar
14 : Amygdala activation during emotion processing of neutral faces in children with severe mood dysregulation versus ADHD or bipolar disorder. Am J Psychiatry 2010; 167:61–69Link, Google Scholar
15 : Adolescents with major depression demonstrate increased amygdala activation. J Am Acad Child Adolesc Psychiatry 2010; 49:42–51Crossref, Medline, Google Scholar
16 : Amygdala activation in maltreated children during pre-attentive emotional processing. Br J Psychiatry 2013; 202:269–276Crossref, Medline, Google Scholar
17 : Limbic scars: long-term consequences of childhood maltreatment revealed by functional and structural magnetic resonance imaging. Biol Psychiatry 2012; 71:286–293Crossref, Medline, Google Scholar
18 : Enhanced amygdala reactivity to emotional faces in adults reporting childhood emotional maltreatment. Soc Cogn Affect Neurosci 2013; 8:362–369Crossref, Medline, Google Scholar
19 : The IMAGEN study: reinforcement-related behaviour in normal brain function and psychopathology. Mol Psychiatry 2010; 15:1128–1139Crossref, Medline, Google Scholar
20 : The Strengths and Difficulties Questionnaire: a research note. J Child Psychol Psychiatry 1997; 38:581–586Crossref, Medline, Google Scholar
21 : A multidimensional assessment of stressful life events among adolescents. J Health Soc Behav 1981; 22:400–415Crossref, Google Scholar
22 : A self-report measure of pubertal status: reliability, validity, and initial norms. J Youth Adolesc 1988; 17:117–133Crossref, Medline, Google Scholar
23 : The Development and Well-Being Assessment: description and initial validation of an integrated assessment of child and adolescent psychopathology. J Child Psychol Psychiatry 2000; 41:645–655Crossref, Medline, Google Scholar
24 : The Wechsler Intelligence Scale for Children, 4th ed. San Antonio, Tex, Psychological Corp, 2004Google Scholar
25 : The 1995 ESPAD report: alcohol and other drug use among students in 26 European countries. Stockholm, Swedish Council for Information on Alcohol and Other Drugs, 1997Google Scholar
26 : Brain networks involved in viewing angry hands or faces. Cereb Cortex 2006; 16:1087–1096Crossref, Medline, Google Scholar
27 : Amygdala activation in the processing of neutral faces in social anxiety disorder: is neutral really neutral? Psychiatry Res 2006; 148:55–59Crossref, Medline, Google Scholar
28 : Region of interest analysis using an SPM toolbox. Eighth International Conference on Functional Mapping of the Human Brain; Sendai, Japan, June 2–6, 2002Google Scholar
29 : CONN: a functional connectivity toolbox for correlated and anticorrelated brain networks. Brain Connect 2012; 2:125–141Crossref, Medline, Google Scholar
30 : Boys do it the right way: sex-dependent amygdala lateralization during face processing in adolescents. Neuroimage 2011; 56:1847–1853Crossref, Medline, Google Scholar
31 : Atypical neural responses during face processing in female adolescents with conduct disorder. J Am Acad Child Adolesc Psychiatry 2014; 53:677–687.e5Crossref, Medline, Google Scholar
32 : Emotion dysregulation in attention deficit hyperactivity disorder. Am J Psychiatry 2014; 171:276–293Link, Google Scholar
33 : Brain activation to facial expressions in youth with PTSD symptoms. Depress Anxiety 2012; 29:449–459Crossref, Medline, Google Scholar
34 : Orbitofrontal cortex reactivity to angry facial expression in a social interaction correlates with aggressive behavior. Cereb Cortex 2015; 25:3057–3063Crossref, Medline, Google Scholar
35 : Functional connectivity mapping of the human precuneus by resting state fMRI. Neuroimage 2012; 59:3548–3562Crossref, Medline, Google Scholar
36 : Emotion regulation and amygdala-precuneus connectivity: focusing on attentional deployment. Cogn Affect Behav Neurosci 2016; 16:991–1002Crossref, Medline, Google Scholar
37 : The adolescent outcome of hyperactive children diagnosed by research criteria, III: mother-child interactions, family conflicts, and maternal psychopathology. J Child Psychol Psychiatry 1991; 32:233–255Crossref, Medline, Google Scholar
38 : A follow-up study of maternal expressed emotion toward children with attention-deficit/hyperactivity disorder (ADHD): relation with severity and persistence of ADHD and comorbidity. J Am Acad Child Adolesc Psychiatry 2014; 53:311–319.e1Crossref, Medline, Google Scholar
39 : Preliminary evidence for an association between social anxiety symptoms and avoidance of negative faces in school-age children. J Clin Child Adolesc Psychol 2006; 35:431–439Crossref, Medline, Google Scholar
40 : Ventrolateral prefrontal cortex activation and attentional bias in response to angry faces in adolescents with generalized anxiety disorder. Am J Psychiatry 2006; 163:1091–1097Link, Google Scholar
41 : Amygdala habituation: a reliable fMRI phenotype. Neuroimage 2014; 103:383–390Crossref, Medline, Google Scholar
