Brain Structure in Youth Antisocial Behaviour: Exploring Sources of Heterogeneity and Categorical Versus Dimensional Approaches Through Large-Scale Analyses

: (Alternative Format Thesis)

Abstract

Antisocial behaviour comprises aggression, violence, rule-breaking, and delinquent behaviours and is central to the diagnosis of Conduct Disorder (CD) in youth. Although CD is a debilitating disorder associated with adverse outcomes and the highest burden of any mental disorder in children aged 0-14 years, CD continues to be under-recognised and our understanding of its aetiology remains incomplete. Neurocognitive models propose that dysfunction in several key brain regions (e.g., amygdala) may confer risk for CD. However, while neuroimaging research indicates that youths with CD show alterations in brain structure compared to healthy controls, evidence is inconsistent and suffers from multiple limitations. These include: 1) small samples and heterogeneous data acquisition and analysis methods, 2) insufficient consideration of participant characteristics (e.g., maltreatment, sex, age-of-onset), and 3) limited exploration of dimensional approaches to CD and related phenotypes.

To address these limitations, a major objective of my PhD was to collate the largest harmonised neuroimaging dataset on youths with CD and/or conduct problems by combining data from research groups across the world in the context of the ENIGMA Antisocial Behavior working group. The specific aims of this thesis were to 1) identify the robust structural brain alterations in CD by performing well-powered analyses based on an unprecedented sample size, 2) increase our understanding of the influence of participant characteristics/clinical heterogeneity on brain structure in CD by assessing the impact of sex and age, and testing current and novel subtyping approaches, and 3) investigate dimensional approaches to conceptualising CD by comprehensively exploring associations between conduct problems and brain structure in the largest sample assembled to date.

In the first study of this thesis, we investigated the impact of childhood maltreatment – a largely neglected risk factor – on brain structure differences in CD using data from 260 children and adolescents from a mixed-sex European multi-site cohort (FemNAT-CD). Adopting a vertex-wise, whole-brain approach and assessing multiple measures of cortical structure, we found that CD youths with a history of maltreatment (n=49) showed more extensive alterations in cortical structure relative to controls (n=146) than their non-maltreated counterparts (n=65). These included lower volume, gyrification, and cortical thickness in inferior and middle frontal regions. The maltreated CD group also showed lower gyrification, surface area, and volume in frontal, temporal, and parietal brain regions compared to the non-maltreated CD group. This study highlighted that maltreatment experiences designate a potentially meaningful ‘ecophenotypic’ (i.e., environmentally-mediated) variant of CD, which differs neurobiologically from non-maltreated CD, and that maltreatment is an important variable to consider in (neuroimaging) studies of CD.

For Study 2 of this thesis, we pooled data from 15 international cohorts from the ENIGMA Antisocial Behavior working group to compare regional cortical thickness, surface area, and subcortical volumes between 1,185 youths with CD and 1,253 typically-developing controls, a sample more than 20-times larger than the average CD neuroimaging study to date. Youths with CD showed small but widespread reductions in surface area (in 26 out of the 34 cortical regions assessed and total surface area) and lower volume in limbic and striatal subcortical regions. Only limited differences in cortical thickness were identified. There was no evidence for moderation by sex and DSM-5 defined subtypes of CD based on age-of-onset (childhood-onset versus adolescent-onset) and callous-unemotional traits (high versus low) showed similar alterations. These findings confirm previously identified brain alterations (e.g., lower amygdala volume), but they also provide novel evidence that brain alterations in CD are far more widespread than previously reported. In contrast to results from small-scale studies and theory, they also suggest that CD is associated with brain structural alterations regardless of sex, age-of-onset, or level of callous-unemotional traits.

Finally, complementing the above case-control analyses, in Study 3 we examined the dimensional relationship between continuous conduct problems and brain structure in 14,160 participants from 18 case-control and community-/population-based cohorts contributing to the ENIGMA Antisocial Behavior working group. Similar to findings in CD youths, we observed small but widespread negative associations between conduct problems and surface area (in 22 of the 34 regions assessed and total surface area) as well as negative associations with amygdala and hippocampal volume. However, in contrast to the case-control study, we also found widespread associations with cortical thickness (in 15 of the 34 investigated regions and mean cortical thickness) and sex was identified as a key moderator. Many of the negative associations between conduct problems and surface area were observed in male youths but not in female youths. These findings provide strong evidence that brain alterations occur on a continuum and are not specific to youths with clinically-elevated conduct problems (or CD). As dimensional analyses in representative samples are less subject to some of the conceptual and methodological limitations of case-control studies (e.g., confounding through systematic differences between cases and controls), the findings from Study 3 support overlapping alterations in youths with CD and suggest that these are unlikely to merely be a consequence of having CD or lifestyle differences between groups.

Overall, the current thesis provides novel and robust evidence of small but widespread brain alterations in CD and overlapping associations between conduct problems and brain structure, especially in surface area. These results confirm the involvement of regions implicated in the leading neurocognitive models of CD/conduct problems, but simultaneously indicate that these models should be extended to include a wider range of brain regions. Our findings also indicate that sex and current diagnostic subtypes might not map onto very distinct neurobiological profiles in youths with CD, but that further, larger studies are needed to investigate the impact of childhood maltreatment as a potential subtyping factor. These findings clarify and extend our knowledge of the brain structural correlates of CD and conduct problems and provide a robust foundation for future research focused on subtyping, multivariate investigations, and exploring whether CD is better conceptualised as a neurodevelopmental than a behavioural disorder. In this context, the ENIGMA Antisocial Behavior dataset – collated as part of this PhD – will be an invaluable resource to the research community and should continue to contribute substantially to advancing our understanding of the role of the brain in youth antisocial behaviour, including for the purposes of prevention and intervention.

Date of Award	26 Jun 2024
Original language	English
Awarding Institution	University of Bath
Supervisor	Graeme Fairchild (Supervisor) & Esther Walton (Supervisor)