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
Department of Biomedicine and Prevention, University of Rome “Tor Vergata”, Rome, ItalyAthinoula A. Martinos Center for Biomedical Imaging and Harvard Medical School, Boston, MA, USA
Department of Child and Adolescent Psychiatry, Psychosomatics and Psychotherapy, University Hospital Frankfurt, Goethe University, Frankfurt am Main, Germany
Department of Child and Adolescent Psychiatry, Psychosomatics and Psychotherapy, University Hospital Frankfurt, Goethe University, Frankfurt am Main, GermanySchool of Psychology, Fresenius University of Applied Sciences, Frankfurt, Germany
Child Neuropsychology Section, Department of Child and Adolescent Psychiatry, Psychosomatics and Psychotherapy, University Hospital, RWTH Aachen, Aachen, GermanyDepartment of Child and Adolescent Psychiatry, Medical Faculty, TU Dresden, Dresden, Germany
Jacobs Center for Productive Youth Development at the University of Zurich, Zurich, SwitzerlandNeuroscience Center Zurich (ZNZ), University and ETH Zurich, Zurich, Switzerland
Child Neuropsychology Section, Department of Child and Adolescent Psychiatry, Psychosomatics and Psychotherapy, University Hospital, RWTH Aachen, Aachen, GermanyJARA-Brain Institute II, Molecular Neuroscience and Neuroimaging, RWTH Aachen and Research Centre Juelich, Juelich, Germany
Department of Child and Adolescent Psychiatry, Psychosomatics and Psychotherapy, University Hospital Frankfurt, Goethe University, Frankfurt am Main, Germany
Childhood maltreatment is common in youths with Conduct Disorder (CD) and both CD and maltreatment have been linked to neuroanatomical alterations. Nonetheless, our understanding of the contribution of maltreatment to the neuroanatomical alterations observed in CD remains limited. We tested the applicability of the ‘ecophenotype’ model to CD, which holds that maltreatment-related psychopathology is (neurobiologically) distinct from psychopathology without maltreatment.
METHODS
Surface-based morphometry was used to investigate cortical volume, thickness, surface area and gyrification in a mixed-sex sample of CD participants (n=114) and healthy controls (n=146), aged 9–18 years. Using vertex-wise general linear models adjusted for sex, age, total intracranial volume, and site, controls were compared with the overall CD group, and the CD subgroups with (n=49) versus without (n=65) maltreatment (assessed by the Children’s Bad Experiences interview). These subgroups were also directly compared.
RESULTS
The overall CD group showed lower cortical thickness in the right inferior frontal gyrus. Maltreated CD youths showed more widespread structural alterations relative to controls, comprising lower thickness, volume and gyrification in inferior and middle frontal regions. Conversely, non-maltreated CD youths only showed greater left superior temporal gyrus folding relative to controls. When contrasting the CD subgroups, those with maltreatment displayed lower right superior temporal gyrus volume, right precentral gyrus surface area, and gyrification in frontal, temporal, and parietal regions.
CONCLUSIONS
Consistent with the ‘ecophenotype’ model, findings indicated that CD youths with versus without maltreatment differ neurobiologically. This highlights the importance of considering maltreatment history in neuroimaging studies of CD and other disorders.
Conduct Disorder (CD) is a common disorder of childhood and adolescence, characterized by persistent aggressive and non-aggressive antisocial behaviors (
American Psychiatric Association (2013): Diagnostic and Statistical Manual of Mental Disorders, 5th ed. Washington, DC: American Psychiatric Publishing.
). CD is associated with various negative outcomes including low educational achievement, poor mental and physical health, and antisocial personality disorder (
) highlights the importance of increasing our understanding of the pathophysiological mechanisms associated with CD to improve its prevention, assessment, and treatment.
In line with neurodevelopmental models, studies of CD have identified structural alterations in brain regions implicated in emotion processing, empathy, decision-making, reinforcement learning and social cognition, which are frequently impaired in CD youth (
). For example, two meta-analyses of voxel-based morphometry (VBM) studies reported lower gray matter volume (GMV) in the amygdala, striatum, insula, precuneus, and frontal and temporal regions in CD youths compared to healthy controls (HCs) (
A systematic review and meta-analysis of neuroimaging in oppositional defiant disorder (ODD) and conduct disorder (CD) taking attention-deficit hyperactivity disorder (ADHD) Into account.
). Overlapping and additional alterations have been reported in surface-based morphometry (SBM) studies. While VBM focuses on GMV as a composite measure, SBM methods assess the components of cortical volume: cortical thickness (CT), surface area (SA) and folding/gyrification. It is important to distinguish between these metrics as they have distinct etiologies and developmental trajectories (
). Amongst the more consistent SBM findings in CD are lower CT in frontal, temporal and parietal regions including the ventromedial prefrontal and orbitofrontal cortex (OFC) (
However, the heterogeneity in findings limits their contribution to our understanding of CD’s pathophysiology. Beyond methodological issues such as small sample sizes (
), childhood maltreatment might constitute a further source of heterogeneity in CD. Childhood maltreatment comprises emotional, physical, and sexual abuse, and neglect (
). However, a more direct pathway has also been proposed, whereby experiences of severe stressors in childhood become ‘biologically embedded’, thereby conferring latent vulnerability for subsequent psychopathology (
Correspondingly, primary and meta-analytical studies have reported associations between maltreatment and cortical structure in children, adolescents, and adults (
). Critically, many of these alterations overlap with those observed in CD, including lower GMV and/or CT in frontal, temporal and parietal regions such as OFC and precuneus (
An examination of the relation between conduct disorder, childhood and adulthood traumatic events, and posttraumatic stress disorder in a nationally representative sample.
), this suggests that it may contribute to structural alterations observed in this population. Accordingly, studies into other disorders frequently associated with maltreatment (e.g., depression (
)) found that at least some of the disorder-related neuroanatomical alterations were specific to patients with maltreatment histories. Based on this and clinical differences between maltreated and non-maltreated patients, Teicher and Samson (
) proposed that maltreatment-related psychopathology might represent a distinct ‘ecophenotype’ of the same disorder.
As maltreatment is a major risk factor for CD and maltreated and CD youths show overlapping structural alterations, testing the applicability of the ‘ecophenotype’ model to CD is important. Clarifying whether CD youths with versus without maltreatment show distinct alterations from controls and differ from each other may help us understand previous inconsistent findings and would strengthen the case for considering maltreatment in future research. Although much remains to be learned regarding how maltreatment increases risk for psychopathology, evidence suggests that the relationship might be mediated by maltreatment-related neurobiological adaptations (
). If CD youths with and without maltreatment history differ in brain structure, this might indicate that maltreatment exposure constitutes a distinct pathway to antisocial behavior (
) and designates a subgroup of CD youths with distinct neurobiology with implications for theory, assessment, and research.
Despite this, our knowledge of the impact of maltreatment on brain alterations in CD remains limited. In a sample of young women with a history of CD, most group differences in CT and SA (
) remained significant when controlling for abuse. In the first study to explicitly compare CD individuals with and without maltreatment history, Gao and colleagues (
) demonstrated that in addition to main effects of diagnosis on STG (CD<HC) and dorsomedial prefrontal cortex (PFC) volume (CD>HC), CD boys with maltreatment showed lower dorsolateral PFC volume, but increased posterior cingulate and putamen volume relative to CD boys without maltreatment. These findings suggest the ‘ecophenotype’ model may be applicable to CD. However, as this study was the first of its kind and recruited a male-only sample, it requires replication and extension to test the robustness and generalizability of findings.
Therefore, we aimed to extend previous research by investigating the structural correlates of CD with versus without maltreatment in a large mixed-sex European sample. Using SBM, we compared controls and CD youths regardless of maltreatment history in terms of cortical structure (assessing cortical volume, CT, SA, and gyrification). We then subdivided the CD group into those with and without maltreatment history and compared each subgroup to controls and each other.
We expected to observe main effects of CD diagnosis in frontal, temporal, and parietal regions, including lower CT in the OFC/ventromedial PFC, STG and precuneus (CD-all<HC). Based on overlapping alterations in CD and maltreated samples, we predicted that some case-control alterations would be specific to the maltreated CD subgroup (e.g., OFC, STG, precuneus (
) study was published after these predictions were formulated but provided evidence for differences between the CD subgroups in the dorsolateral PFC and posterior cingulate.
METHODS AND MATERIALS
Participants
114 youths with CD (32.5% female) and 146 HCs (50.7% female) aged 9-18 years from the European multi-site FemNAT-CD study (
) were included based on availability of manually-edited structural MRI data and maltreatment information (see Figure S1 for a flowchart and Table S1 for site distributions). Participants were recruited through psychiatric clinics, youth offending services, youth welfare institutions, and community outreach. Exclusion criteria included IQ<70, and history of neurological disorders, head trauma, autism, schizophrenia, or bipolar disorder. Standard MRI exclusion criteria were applied. Cases had a DSM-IV-TR diagnosis of CD (or 1-2 current CD symptoms and Oppositional Defiant Disorder) while controls had no current Axis I disorders or history of disruptive behavior disorders. We excluded HCs with maltreatment (n=13) as this group was too small for meaningful statistical analysis. Ethical approvals for the original study and current analyses were acquired (see Supplement). Written informed consent was obtained from all participants or their parents/caregivers, while those aged <16/18 provided assent.
Phenotypic Measures
CD and other psychiatric diagnoses were made using the Kiddie-SADS interview (
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
), which assesses exposure to various negative experiences. The current study focused on the physical and sexual abuse items, based on which maltreatment was categorized as absent, probable, or definite. Consistent with previous studies (
Maternal depression in the intergenerational transmission of childhood maltreatment and its sequelae: Testing postpartum effects in a longitudinal birth cohort.
), we created a dichotomous variable reflecting no maltreatment exposure versus likely exposure (combining probable and definite maltreatment). IQ was estimated using subtests of the Wechsler Intelligence Scales. Socioeconomic status (SES) was calculated based on parental income, education, and occupational status. Attention problems were assessed as a proxy for Attention-Deficit/Hyperactivity Disorder (ADHD) symptoms using the caregiver-report Child Behavior Checklist (CBCL/4-18) attention problems subscale (
Andershed HA, Kerr M, Stattin H, Levander S (2002): Psychopathic traits in non-referred youths: a new assessment tool. Elsevier, pp 131–158. In Blaauw E, Sheridan L, editors. Psychopaths: New international perspectives. The Hague, Netherlands: Elsevier, pp 131–158.
) and are reported for sample/subgroup description purposes only. More information on the phenotypic measures is provided in the Supplement.
MRI Data Acquisition
Structural MRI data were acquired at five sites using Siemens 3T (Tim-Trio and Prisma) or Philips 3T (Achieva) scanners. The supplement provides information on scanner models, head coils and scanning parameters (Table S3) and site qualification procedures undertaken to ensure comparability of data collection across sites. All scans were screened for movement or image artifacts by the MRI operator and repeated as necessary. Additionally, image quality was evaluated prior to analysis using the Backhausen rating system (
CT, SA, volume, and local gyrification index (lGI) were quantified at each vertex using FreeSurfer (v5.3.0, http://surfer.nmr.mgh.harvard.edu), as described in detail elsewhere (
). Surface reconstructions were inspected blind to group status. Segmentation errors and topological defects were manually corrected by deleting/adding gray or white matter and setting control points. CT, SA, and volume were smoothed using a 10-mm and lGI using a 5-mm kernel at full-width/half maximum as lGI is an inherently smooth measure (
R Core Team (2020): R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. URL https://www.R-project.org/.
) using Fisher’s exact tests for categorical variables and Welch’s t-tests and ANOVAs followed by post-hoc pairwise comparisons for continuous variables. Total intracranial volume (TIV) was compared between groups using general linear models (GLMs) adjusted for sex, age, and site.
Group differences in cortical structure were assessed using whole-brain vertex-wise analyses in FreeSurfer. First, to investigate main effects of group (CD-all versus HC), GLMs were fitted per hemisphere and SBM measure. Analyses were adjusted for sex, age, site and TIV (orthogonalized to sex). TIV correction was not applied to CT as it is not related to brain size (
). In response to reviewers’ comments, additional analyses covarying SES and focusing on male participants were performed as CD is associated with lower SES (
All whole-brain analyses were multiple comparison-corrected using Monte Carlo z-field simulations based on vertex- and cluster-wise thresholds of p<0.05. Effect sizes were expressed as Cohen’s d.
RESULTS
Sample Characteristics
The overall CD (CD-all) and HC groups did not differ in age, but the CD-all group included significantly fewer female participants. As expected, the CD-all group had higher levels of CD symptoms, attention problems, psychopathic and callous-unemotional traits, psychotropic medication use, externalizing and internalizing comorbidity, and lower IQs and SES than the HCs (see Table 1). Similar case-control differences emerged when subdividing CD youths into those with (CD/+, n=49) versus without maltreatment (CD/-, n=65). However, only the CD/- group had lower IQs and SES and a smaller proportion of female participants, and only the CD/+ group had higher rates of lifetime Generalized Anxiety Disorder than controls. The CD subgroups were well-matched on demographic and clinical variables, except for higher rates of lifetime Major Depressive Disorder (MDD) in the CD/+ subgroup.
Table 1Demographic and clinical characteristics of the samplea
HC (N=146)
CD-all (N=114)
CD/-(N=65)
CD/+(N=49)
Comparisons
HC vs CD-all
HC vs CD/- vs CD/+
post-hoc
Characteristic
Mean
SD
Mean
SD
Mean
SD
Mean
SD
t
p
F
p
Age (years)
13.87
2.6
13.64
2.46
13.58
2.46
13.71
2.49
-0.73
.467
0.30
.744
IQ
103.47
12.25
96.42
13.71
94.58
10.08
98.86
17.21
-4.31
< .001
9.43
< .001
HC > CD/-
SESb
0.25
1.05
-0.23
0.92
-0.30
0.93
-0.14
0.91
-3.72
<.001
6.96
.001
HC > CD/-
CD symptoms, current
0.04
0.23
4.82
2.64
4.51
2.42
5.22
2.89
19.23
< .001
242.01
< .001
HC < CD/-, CD/+
Attention problems (CBCL)c
2.45
2.65
8.93
4.45
8.57
4.54
9.41
4.34
14.39
< .001
137.00
< .001
HC < CD/-, CD/+
Psychopathic traits (YPI total)
91.52
18.24
109.32
25.46
108.95
26.22
110.9
24.82
6.46
< .001
22.64
< .001
HC < CD/-, CD/+
CU traits (YPI CU subscale)
27.64
5.82
32.42
7.67
32.64
8.17
32.12
7.01
5.53
< .001
16.34
< .001
HC < CD/-, CD/+
CU traits (ICU total)
17.17
7.42
33.86
12.03
33.34
11.44
34.55
12.86
13.01
< .001
94.51
< .001
HC < CD/-, CD/+
N
%
N
%
N
%
N
%
p (Fisher’s exact)
p (Fisher’s exact)
Sex (Female)
74
50.7%
37
32.5%
17
26.2%
20
40.8%
.004
.003
HC > CD/-
Age of onset (CD)
.726d
Childhood
66
57.9%
36
55.4%
30
61.2%
Adolescent
36
31.6%
21
32.3%
15
30.6%
Unspecified
12
10.5%
8
12.3%
4
8.2%
Psychotropic medication
0
0.0%
35
30.7%
20
30.8%
15
30.6%
< .001
< .001
HC < CD/-, CD/+
Lifetime DSM-IV-TR diagnoses
ODD
0
0.0%
97
85.1%
53
81.5%
44
89.8%
< .001
< .001
HC < CD/-, CD/+
ADHD
0
0.0%
64
56.1%
32
49.2%
32
65.3%
< .001
< .001
HC < CD/-, CD/+
MDD
2
1.4%
28
24.6%
9
13.9%
19
38.8%
< .001
< .001
HC < CD/- < CD/+
GAD
0
0.0%
7
6.1%
2
3.1%
5
10.2%
.003
.001
HC < CD/+
PTSD
0
0.0%
13
11.4%
4
6.2%
9
18.4%
< .001
< .001
HC < CD/-, CD/+
Alcohol abuse
0
0.0%
8
7.0%
4
6.2%
4
8.2%
.001
.001
HC < CD/-, CD/+
Substance abuse
1
0.7%
13
11.4%
7
10.8%
6
12.2%
< .001
< .001
HC < CD/-, CD/+
a Post-hoc comparisons were based on Bonferroni-corrected Welch’s t-tests and Fisher’s exact tests comparing two groups at a time. Significant p-values are marked in bold. CD participants were classified as having childhood-onset CD if at least one symptom and functional impairment were reported to have occurred before the age of 10. Otherwise, participants were classified as having adolescent-onset CD. HC=healthy controls; CD=Conduct Disorder; CD/-=Conduct Disorder without maltreatment history; CD/+=Conduct Disorder with maltreatment history; N=sample size; SD=standard deviation; IQ=intelligence quotient; SES=socioeconomic status; CBCL=Child Behavior Checklist; YPI=Youth Psychopathic traits Inventory; CU=callous-unemotional; ICU=Inventory of Callous-Unemotional traits; DSM-IV-TR=Diagnostic and Statistical Manual of Mental Disorders; ODD=Oppositional Defiant Disorder; ADHD=Attention-Deficit/Hyperactivity Disorder; MDD=Major Depressive Disorder; GAD=Generalized Anxiety Disorder; PTSD=Post-Traumatic Stress Disorder.
bMissing for 22 participants (9 HC, 8 CD/-, 5 CD/+)
cMissing for 13 participants (5 HC, 5 CD/-, 3 CD/+)
There were no main effects of group on TIV (see Supplement).
Surface-Based Morphometry results
CD-All versus HC. Relative to controls, the CD-all group showed lower CT in the right pars orbitalis of the inferior frontal gyrus (IFG) extending to the pars triangularis and rostral middle frontal gyrus (MFG; d=-0.40, C1, Figure 1A, Table 2). This cluster remained significant when controlling for IQ, attention problems or SES.
Figure 1Group differences in cortical thickness, surface area, volume and gyrification when controlling for sex, age, site, and total intracranial volumea. A) Relative to HCs, the CD-all group demonstrated reduced cortical thickness in the right pars orbitalis of the inferior frontal gyrus (C1). B) CD participants without a history of maltreatment showed significantly greater gyrification in the left superior temporal gyrus (C2) compared to controls. C) CD youth with a history of maltreatment demonstrated lower cortical thickness in the right pars orbitalis of the inferior frontal gyrus (C3), the right postcentral gyrus (C4) and the left lateral orbitofrontal cortex (C5) relative to controls. They further showed lower volume in the right postcentral gyrus (C6) and left rostral middle frontal gyrus (C7), and lower gyrification in the right rostral middle frontal gyrus (C8). D) Comparing CD participants with versus without maltreatment revealed that the maltreated subgroup displayed lower surface area in the right precentral gyrus (C9) and lower volume in the right superior temporal gyrus (C10). They also showed lower gyrification in a large cluster in the supramarginal gyrus (C11), as well as in the right rostral middle frontal (C12), left fusiform (C13) and left inferior temporal gyri (C14).
a Total intracranial volume was not controlled for in the cortical thickness analyses. CD=Conduct Disorder; HC=Healthy Controls. CD/-=Conduct Disorder without maltreatment history; CD/+=Conduct Disorder with maltreatment history
Inferior temporal gyrus, superior & middle temporal gyrus, temporal pole
lGI
L
2291
2610.25
-52.1
-12.2
-40.9
.013
-3.81
-0.51
0.09
Yes
Yes
Yes
a All analyses controlled for sex, age, site, and total intracranial volume (except thickness). Monte Carlo corrections for multiple comparisons were applied. Cohen’s d was calculated using whole-brain vertex-wise effect size brain maps. Bolded regions represent the location of the peak coordinate. The sensitivity columns present which clusters survived adjustment for IQ or attention problems, respectively. H=hemisphere; NVtxs=number of vertices; MNI=Montreal Neurological Institute; CWP=cluster-wise p-value; Max=maximum -log10(p-value) in the cluster; IQ=intelligence quotient; AP=attention problems (Child Behavior Checklist); SES=socioeconomic status; CD=Conduct Disorder; HC=healthy controls; CD/-=Conduct Disorder without maltreatment; CD/+=Conduct Disorder with maltreatment; CT=cortical thickness, SA=surface area; CV=cortical volume; lGI=local gyrification index.
CD/- versus HC. Relative to controls, CD/- participants showed greater left STG gyrification, extending to transverse temporal, supramarginal and postcentral gyri (d=0.39, C2, Figure 1B, Table 2). This group effect overlapped with a cluster identified in the CD-all versus HC comparison when covarying IQ. It survived adjustment for IQ, attention problems, and SES.
CD/+ versus HC. Relative to controls, CD/+ participants showed lower CT in the right pars orbitalis of the IFG (extending to pars triangularis and rostral MFG, C3), the right postcentral gyrus (extending to precentral gyrus, C4) and the left lateral orbitofrontal cortex (OFC) (extending to rostral MFG, C5, Figure 1C, Table 2). The pars orbitalis cluster (C3) overlapped with the one identified in the CD-all versus HC comparison (C1). CD/+ participants also had lower volume in the postcentral gyrus (extending to precentral gyrus, C6) and left rostral MFG (extending to caudal MFG, C7), and lower gyrification in the right rostral MFG (extending to superior frontal gyrus, C8). These differences had medium effect sizes (ds=0.39-0.43).
The left lateral OFC CT (C5) and left rostral MFG volume (C7) findings did not survive IQ adjustment, while correcting for SES rendered differences in postcentral gyrus CT (C4) and rostral MFG gyrification (C8) non-significant. Only the volumetric differences (C6 and C7) survived adjustment for attention problems.
CD/+ versus CD/-. Relative to CD/- youths, CD/+ participants showed lower SA in the precentral gyrus (extending to superior frontal gyrus, C9) and lower right STG volume (C10, Figure 1D, Table 2). The CD/+ subgroup had lower gyrification in the right supramarginal gyrus extending to pre- and postcentral gyri, inferior parietal lobule, and middle temporal gyrus (C11), the right rostral MFG (C12), left fusiform gyrus (extending to lateral occipital pole, C13), and left inferior temporal gyrus (extending to middle/superior temporal gyrus and temporal pole, C14). The right rostral MFG cluster (C12) overlapped with one identified in the CD/+ versus HC comparison (C8). All effect sizes were medium (ds=-0.43-0.52).
All clusters survived SES and IQ adjustment, except the precentral gyrus SA difference (C9) for IQ. When covarying for attention problems, differences in right precentral/superior frontal gyrus SA (C9) and right supramarginal (C11) and left inferior temporal (C14) gyrification survived. Due to significantly higher rates of MDD in CD/+ youths, we controlled for lifetime MDD (absent/present) in additional analyses. Differences in gyrification, but not volume and SA, survived this adjustment.
Additional group differences emerged in the sensitivity analyses, particularly those adjusting for IQ and SES (see Tables S4-S6 and Figures S2-S4 for details).
When rerunning the main analyses in the male participants only (57.3% of the sample), the location and size of clusters varied, but the pattern of more extensive case-control alterations in the CD/+ group and differences between the CD subgroups was replicated (see supplemental Table S7 and Figure S5).
Results for subcortical regions are presented in the supplement. Briefly, there were no significant group differences after False-Discovery-Rate correction, but uncorrected findings pointed towards greater left accumbens and bilateral pallidum volume in the CD-all and CD (sub)groups relative to controls. Critically, there were no differences in subcortical volumes between subgroups.
DISCUSSION
Using a large mixed-sex sample, this study aimed to investigate cortical structure alterations in CD youths with and without maltreatment and test the ‘ecophenotype’ hypothesis that maltreatment-related CD (CD/+) may be distinct from CD without maltreatment (CD/-) (
In line with our hypotheses and the ‘ecophenotype’ model, maltreated and non-maltreated CD youths demonstrated distinct alterations relative to HCs. Overall, the CD/+ group demonstrated more extensive differences compared with the control group, across multiple measures of cortical structure, including lower CT, volume and gyrification in inferior and middle frontal regions and pre- and postcentral gyri. Conversely, differences between CD/- and HC groups were limited to greater left STG folding. The CD subgroups also differed from each other, with CD/+ youths showing lower right STG volume, right precentral SA and gyrification in frontal, temporal, and parietal regions relative to their CD/- counterparts.
The regions identified when comparing the CD subgroups and controls were largely consistent with previous research. For example, OFC alterations as observed in the CD/+ group were reported in previous studies of CD youths (
). Our findings suggest that OFC alterations may be specific to maltreated CD youths. Likewise, the only difference observed between the CD-all and control groups, lower IFG CT, was only detected in the CD/+ but not the CD/- subgroup. Correspondingly, while alterations in IFG structure have only been reported in a few studies of CD youths and adults with antisocial personality disorder (
), albeit in the opposite (i.e., left) hemisphere. Together with the current findings, this may suggest that structural alterations in the IFG are specific to CD youths with maltreatment history. These alterations may underlie some of the cognitive and social deficits observed in CD youth, such as reduced inhibitory control (
Two systems for empathy: a double dissociation between emotional and cognitive empathy in inferior frontal gyrus versus ventromedial prefrontal lesions.
Conversely, greater left STG folding was only observed in the CD/- subgroup. STG alterations have frequently been reported in CD youths. Whilst most studies demonstrated decreased STG CT (
). Our findings suggest that greater left STG folding may be specific to non-maltreated CD youths. However, the CD/+ subgroup demonstrated lower right STG volume compared to CD/- (but not HC) youths, consistent with reports of STG alterations in maltreated samples (
). This suggests that STG structure might also be influenced by maltreatment but that the measure, direction and/or laterality of effects might differ between maltreated and CD groups.
Further differences between the CD subgroups included lower rostral middle frontal, supramarginal, fusiform and inferior temporal gyrification in the CD/+ group. Although few maltreatment studies have investigated folding, alterations in overlapping regions have been reported, such as lower supramarginal gyrus volume in psychopathology-free children with documented maltreatment exposure (
). Critically, while our findings demonstrate extensive neurobiological differences between CD/+ and CD/- youths, only one region was also identified when comparing each of the subgroups with controls: CD/+ youths showed lower rostral MFG folding compared to both the CD/- and HC groups. This provides strong evidence that lower rostral MFG gyrification is specific to the CD/+ group. This region is implicated in cognitive flexibility, response selection, working memory and decision-making (
) identified lower right rostral/caudal MFG volume in CD/+ relative to CD/- youth. Overlap between the two studies’ findings was otherwise limited, potentially owing to differences in assessment of maltreatment and sample characteristics (males-only versus a mixed-sex sample).
We note that adjusting for IQ, SES, and particularly attention problems affected our findings (e.g., in the CD/+ versus HC comparisons, only volumetric differences in the postcentral gyrus and rostral MFG survived correction for attention problems). Attenuation of brain differences when controlling for ADHD(-like) symptoms and/or IQ is common in the CD literature (
) and adjusting for ADHD may also remove CD-related effects. Similarly, increased rates of ADHD have been reported in individuals who experienced early deprivation (
Early severe institutional deprivation is associated with a persistent variant of adult attention-deficit/hyperactivity disorder: clinical presentation, developmental continuities and life circumstances in the English and Romanian Adoptees study.
). This suggests that maltreatment may constitute one causal pathway to attention problems and that adjusting for the latter could also remove maltreatment-related effects.
Overall, our results indicate that maltreated CD youths showed more extensive alterations in multiple measures of cortical structure than non-maltreated CD youths, despite having similar clinical and demographic profiles. Differences in the CD/+ subgroup might be partly explained by the effects of maltreatment on neurobiological systems, most prominently the hypothalamic-pituitary-adrenal axis (
). Adaptations in the body’s (neuro)hormonal stress system could affect neurodevelopmental processes including neurogenesis, synaptic pruning and myelination, with downstream effects on brain structure and function (
). Importantly, maltreatment-induced alterations may initially reflect adaptive changes, which later become maladaptive in a normative environment, increasing vulnerability to psychopathology (
), our findings of maltreatment-related regional decreases in the CD/+ subgroup (relative to the other groups) may reflect accelerated cortical development. This fits with the stress acceleration model of maltreatment, which postulates that stress-induced neurobiological changes result in accelerated brain maturation, especially in regions involved in emotion processing (e.g., OFC (
)). In line with this and our observation of lower rostral MFG gyrification in the CD/+ group, a recent study found advanced rostral MFG maturation in abused girls, albeit only in those without psychopathology (
). Interestingly, while the CD/+ subgroup differed from controls and CD/- youths on multiple cortical structure metrics, most differences between the CD subgroups were found for gyrification. This is surprising as evidence suggests that sulcal and gyral patterns develop mostly in utero and are under strong genetic influence (
), they suggest that despite having the same diagnosis, CD youths with and without maltreatment differ from each other and show different alterations compared to controls (
An examination of the relation between conduct disorder, childhood and adulthood traumatic events, and posttraumatic stress disorder in a nationally representative sample.
), designates a potentially meaningful ‘ecophenotypic’ (i.e., environmentally-mediated) variant of CD, which differs neurobiologically from non-maltreated CD. This highlights the importance of investigating the utility of subtyping CD youths by maltreatment history and suggests that through its effects on neurobiological systems, maltreatment might constitute a distinct developmental pathway to CD (
). However, further research is needed to substantiate these claims and rule out alternative explanations. For example, given our cross-sectional design and limited information about maltreatment timing, we cannot rule out whether CD/+ youths would have developed CD regardless of maltreatment exposure and/or whether the CD preceded and increased the child's risk of experiencing maltreatment (
). We also note that many regions that differed when comparing the CD subgroups with controls were not identified when directly contrasting the CD subgroups. Hence, alterations that emerged when contrasting CD/+ youth and controls, but not in the other comparison, could reflect quantitative differences between the CD subgroups (i.e., one group shows a stronger effect but it is still present in the same location/direction in the other) rather than qualitative differences (i.e., CD subgroups display alterations in different regions/opposite directions). Further studies including maltreated HCs could help to address these issues by further disentangling maltreatment- and disorder-related effects and establishing specificity through interaction analyses (
). Second, our findings indicate that maltreatment contributes to heterogeneity within CD, at least at a structural level, and may have been an important confound in earlier neuroimaging studies (
). When considering the overall CD group in the current sample (‘enriched’ for maltreatment), we were unable to replicate many of the alterations previously reported in CD (e.g., lower OFC/ventromedial PFC thickness). This highlights the importance of assessing for maltreatment in future studies.
This study had several limitations. First, we used a dichotomous measure of maltreatment, and were unable to explore effects of heterogeneity in maltreatment exposure such as type, timing, or severity (
). Relatedly, our maltreatment measure primarily reflected experiences of physical and sexual abuse and did not capture neglect or other adversities which may also impact the brain and psychopathology (
). Second, we relied on retrospective reports from parents or caregivers who may be unaware that their child has been maltreated or purposefully untruthful. However, the CBE interview aims to increase honesty by focusing on whether maltreatment occurred, rather than the perpetrator’s identity (
). Third, collapsing across sex and a wide age range could have impacted our findings. However, we controlled for both factors in our analyses and re-ran the analyses in a male-only subsample. Lastly, while our sample size was larger than most previous neuroimaging studies of CD, recent evidence (
) suggests that our analyses might not have been adequately powered to detect small effects, highlighting the need for replications in larger samples (see Supplement for further discussion of this issue).
In summary, using a large, mixed-sex sample and sensitive SBM methods to assess multiple cortical properties based on carefully quality controlled and edited data, we found that maltreated and non-maltreated CD subgroups showed distinct neurobiological differences compared to age-matched controls and also differed from each other. Despite similar clinical profiles, maltreated CD youths showed more widespread structural alterations relative to controls, and lower volume, SA and gyrification in frontal, temporal, and parietal regions than non-maltreated CD youths. This supports the ‘ecophenotype’ model, indicating that higher rates of maltreatment in CD youth likely contributed to some of the structural alterations reported in this population and that maltreatment-related and non-maltreatment-related forms of CD might be partly distinct neurobiologically. Findings highlight the need to consider maltreatment in future studies of CD (and other psychiatric disorders) and provide a platform for further research investigating the impact of maltreatment type, timing, and severity/chronicity, and differences in brain activation and neurocognitive functioning between maltreated and non-maltreated CD youths.
American Psychiatric Association (2013): Diagnostic and Statistical Manual of Mental Disorders, 5th ed. Washington, DC: American Psychiatric Publishing.
A systematic review and meta-analysis of neuroimaging in oppositional defiant disorder (ODD) and conduct disorder (CD) taking attention-deficit hyperactivity disorder (ADHD) Into account.
An examination of the relation between conduct disorder, childhood and adulthood traumatic events, and posttraumatic stress disorder in a nationally representative sample.
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
Maternal depression in the intergenerational transmission of childhood maltreatment and its sequelae: Testing postpartum effects in a longitudinal birth cohort.
Andershed HA, Kerr M, Stattin H, Levander S (2002): Psychopathic traits in non-referred youths: a new assessment tool. Elsevier, pp 131–158. In Blaauw E, Sheridan L, editors. Psychopaths: New international perspectives. The Hague, Netherlands: Elsevier, pp 131–158.
R Core Team (2020): R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. URL https://www.R-project.org/.
Two systems for empathy: a double dissociation between emotional and cognitive empathy in inferior frontal gyrus versus ventromedial prefrontal lesions.
Early severe institutional deprivation is associated with a persistent variant of adult attention-deficit/hyperactivity disorder: clinical presentation, developmental continuities and life circumstances in the English and Romanian Adoptees study.
This study was supported in part by grant ES/P000630/1 for the SWDTP awarded to the Universities of Bath, Bristol, Exeter, Plymouth and West of England (UWE) from the Economic and Social Research Council (ESRC)/UKRI and grant MR/N0137941/1 for the GW4 BIOMED MRC DTP awarded to the Universities of Bath, Bristol, Cardiff and Exeter from the Medical Research Council (MRC)/UKRI.
The FemNAT-CD study was funded by the European Commission’s Seventh Framework Programme (FP7/2007-2013, grant number 602407, coordinator Christine M. Freitag).
We thank Sarah Koerner for her help pre-processing the data and the young people and their families for taking part in the FemNAT-CD study.
Christina Stadler receives royalties for a book on aggression. Christine M. Freitag receives royalties for books on Attention-Deficit/Hyperactivity Disorder, Autism Spectrum Disorder, and Depression. Stephane A. De Brito has received speaker fees from the Child Mental Health Centre and the Centre for Integrated Molecular Brain Imaging. All other authors report no biomedical financial interests or potential conflicts of interest.
Preliminary results from this study were presented as a poster at the Annual Meetings of the Society of Biological Psychiatry, April 29 – May 01, 2021, and the Organization for Human Brain Mapping, June 21 – June 25, 2021.
00347-0&id=gr1.jpg){kind=link}