Association of subcortical gray-matter volumes with life-course-persistent antisocial behavior in a population-representative longitudinal birth cohort

Neuropsychological evidence supports the developmental taxonomy theory of antisocial behavior, suggesting that abnormal brain development distinguishes life-course-persistent from adolescence-limited antisocial behavior. Recent neuroimaging work confirmed that prospectively-measured life-course-persistent antisocial behavior is associated with differences in cortical brain structure. Whether this extends to subcortical brain structures remains uninvestigated. This study compared subcortical gray-matter volumes between 672 members of the Dunedin Study previously defined as exhibiting life-course-persistent, adolescence-limited or low-level antisocial behavior based on repeated assessments at ages 7–26 years. Gray-matter volumes of 10 subcortical structures were compared across groups. The life-course-persistent group had lower volumes of amygdala, brain stem, cerebellum, hippocampus, pallidum, thalamus, and ventral diencephalon compared to the low-antisocial group. Differences between life-course-persistent and adolescence-limited individuals were comparable in effect size to differences between life-course-persistent and low-antisocial individuals, but were not statistically significant due to less statistical power. Gray-matter volumes in adolescence-limited individuals were near the norm in this population-representative cohort and similar to volumes in low-antisocial individuals. Although this study could not establish causal links between brain volume and anti-social behavior, it constitutes new biological evidence that all people with antisocial behavior are not the same, supporting a need for greater developmental and diagnostic precision in clinical, forensic, and policy-based interventions.


Attrition analysis of childhood SES comparing age-45 study members with original cohort
No significant differences in childhood IQ were found between the full cohort, those still alive, those seen at age-45 or those scanned at age-45. Those who were deceased by the age-45 data collection had significantly lower childhood IQ's than those who were still alive (t=2.09, p=0.04).
Attrition analysis of childhood IQ comparing age-45 study members with original cohort eFigure 1. FreeSurfer aseg results overlaid on one Dunedin Study member's T1-weighted structural scan eTable 1. Demographic characteristics and co-occurring experiences and conditions associated with antisocial behavior. All measures are described in the main text.
A similar version of this table including the same groups of Study members is presented in Carlisi et al., 2020 Values for between-group comparisons are reported as odds ratios (OR) (95% confidence intervals), or t-statistics (degrees of freedom), p-values. a Growth-mixture modelling was applied to derive developmental subtypes of antisocial behavior, within sex, as previously published in Carlisi et al., 2020 b SES, socio-economic status. The scale ranges from 1 (low) to 6 (high). c ACE, adverse childhood experience. The scale ranges from 0 (none) to 4+.

. Implications of statistically controlling for co-occurring experiences and conditions on comparisons of subcortical grey-matter volume between lifecourse-persistent and low-antisocial trajectory groups
All analyses controlled for sex, and group comparisons were corrected for multiple testing using a false discovery rate (FDR) procedure. Results in the first column, "sex only", are those presented in the main text, Table 1. P-values for significant differences (p<.05, FDR corrected) between groups are presented in bold. ACEs = adverse childhood experiences, CI = confidence interval, SES = socioeconomic status. Regression coefficients (betas) are standardized coefficients. .00 (-.08-.08) p=.95 .03 (

. Implications of statistically controlling for co-occurring experiences and conditions on comparisons of subcortical grey-matter volume between adolescence-limited and low-antisocial trajectory groups
All analyses controlled for sex, and group comparisons were corrected for multiple testing using a false discovery rate (FDR) procedure. Results in the first column, "sex only", are those presented in the main text,

. Implications of statistically controlling for co-occurring experiences and conditions on comparisons of subcortical grey-matter volume between lifecourse-persistent and adolescence-limited trajectory groups
All analyses controlled for sex, and group comparisons were corrected for multiple testing using a false discovery rate (FDR) procedure. Results in the first column, "sex only", are those presented in the main text,  All volumes in the left-hand section of the table are reported in cubic millimeters (mm 3 ). All analyses controlled for sex. Regression coefficients (betas) are standardized coefficients.

eAppendix 2. Correlations among global and regional MRI measures
It is expected that subcortical grey-matter volumes will be intercorrelated with one another, and also that these regional volumes will be correlated with total brain volume. eFigure 2 presents heatmaps of the correlations among regional subcortical grey-matter volumes and total brain volume, stratified by sex (as all primary analyses control for sex). Correlation coefficients show that correlations among subcortical regional volumes were varied, ranging from 0.2-0.7. Moreover, correlations of regional subcortical volumes with total brain volume suggest a degree of independence of these regional findings from an overall global effect of total brain volume. Lastly, regional subcortical volume correlations with surface-based cortical measures similarly suggest that, although the general pattern of subcortical grey-matter findings is in line with patterns observed in our previous study of cortical thickness and surface area (8), there is a degree of independence of the subcortical analyses presented here. eFigure 2. Correlations among regional subcortical grey-matter volumes and total brain volume, global surface area and average cortical thickness A) Correlation matrix for men, B) correlation matrix for women. Pearson's correlation coefficients are presented for each test. Surface area corresponds to total surface area, cortical thickness corresponds to average cortical thickness.