Polygenic risk scores and brain structures both contribute to externalizing behavior in childhood - A study in the Adolescent Brain and Cognitive Development (ABCD) cohort

Abstract

Introduction: Externalizing behaviors are defined as behaviors violating social norms and can be harmful to self and others. Predicting the escalation of externalizing behaviors in children would allow for early interventions to prevent the occurrence of antisocial and criminal acts. Externalizing behaviors are heritable traits, and have been associated with structures of the brain. Brain structure, in turn, is also influenced by genetics. Here, we investigated the association of genetic and brain structural variation with externalizing behaviors in late childhood, and we assessed potential mediating effects.

Methods: Data was collected for 11,878 children aged 9-10 years old from the Adolescent Brain Cognitive Development (ABCD) cohort. We extracted data on externalizing behaviors measured by the parent-reported Child Behavior Checklist (CBCL), brain volumes and white matter integrity measured by magnetic resonance imaging (MRI), and polygenic risk scores (PRS) for (i) antisocial behaviors, (ii) attention-deficit/hyperactivity disorder comorbid with disruptive behavior disorder (ADHD ​+ ​DBD), (iii) irritability, and (iv) traits related to self-regulation & addiction. We examined the associations between brain structures, PRS, and externalizing behavior, and to what extent brain structures mediate the association between PRS and externalizing behavior. Phenotypic associations between brain structures and externalizing behaviors were validated in an independent cohort of 150 adolescents aged 12-21 years enriched for individuals with antisocial behavior.

Results: Increasing levels of externalizing behaviors were associated with reduced total brain and focal gray matter volumes, but not with white matter integrity. These results could not be validated in the independent cohort, except for a good correlation of several effect sizes between the cohorts. Higher PRS for externalizing behaviors were associated with lower cortical gray matter volume, larger subcortical gray matter volume, larger white matter volume, and reduced global white matter fractional anisotropy. Genetic and brain structural variation, combined with sociodemographic factors, explained up to 7% of variation in externalizing behaviors in late childhood; brain structures and PRS each explained up to ∼0.5% of variation. A multivariate model with all sociodemographic factors, brain structures and PRS combined explained up to 11.9% (+5%). Total cortical gray matter volume mediated the association between PRS for ADHD ​+ ​DBD and externalizing behavior in late childhood. However, a large proportion of individual variation in externalizing behavior remained unidentified (∼90%). Brain function and interaction effects with the environment are surmised as potential sources of additional variation.

Keywords: Adolescence; Externalizing behaviors; Gray matter volume; Late childhood; Magnetic resonance imaging; Polygenic risk scores; White matter integrity.

Fig. 1

Association of polygenic risk scores (PRS) based on the (A) attention-deficit/hyperactivity disorder comorbid with disruptive behavior disorders (ADHD ​+ ​DBD), (B) antisocial behavior (ASB), (C) irritability, and (D) traits related to self-regulation & addiction GWAS with scores on the externalizing behavior scale of the Child Behavior Checklist (Ext-CBCL) in 4201 children from the Adolescent Brain and Cognitive Development (ABCD) cohort. The bar plots (top-row) display the linear regression results of the PRS computed across a range of P-value thresholds (x-axis), with the variance explained (R²) indicated on the y-axis and uncorrected P-values displayed on top of each bar. The best-fit P-value threshold (and corresponding number of SNPs included in that threshold and empirical P-value) were 0.1 (35664 SNPs; empirical P ​= ​0.0002), 1 (207771 SNPs; empirical P ​= ​0.003), 0.3 (133697 SNPs; empirical P ​= ​9.999E-05), and 0.1 (47059 SNPs; empirical P ​= ​9.999E-05) for ADHD ​+ ​DBD-PRS, ASB-PRS, irritability-PRS, and traits related to self-regulation and addiction-PRS, respectively. Quantile plots (bottom-row) show how the Ext-CBCL scores vary with increasing PRS (using the identified best-fit PRS). The x-axis shows five equally sized quantiles, and the y-axis shows the difference in the externalizing behavior scores (in terms of regression coefficient and standard error bars) when comparing with the median (reference) quantile.

Fig. 2

Mediation of the association between polygenic risk scores (PRS) for (A) ADHD ​+ ​DBD, (B) Irritability, and (C) Externalizing with parent-reported externalizing behaviors on the Child Behavior Checklist (CBCL) through global brain structures. Significant effects at uncorrected p_unc < 0.05 are printed in boldface. The total effect (c) of PRS on externalizing behaviors was decomposed into a direct effect (c’) and an indirect effect (a∗b), where the indirect path was mediated through a brain structure. Reported values on the paths of the mediation analysis are the effect size and bootstrapped 95% confidence intervals.