Citation
Sasia, A. B.; Lu, Q.; Wu, Y.; & Li, J. J. (2022). Estimating “genetic nurture” effects in polygenic scores for trajectories of antisocial behavior.. 2022 Add Health Users Conference. Chapel Hill, NC.Abstract
Antisocial behaviors (ASB) are characterized by aggressive and non-aggressive rule-breaking behaviors. Despite being highly heritable, little is known about how genetic influences impact development of ASB over time. For example, polygenic scores (PGS) for ASB, which characterize aggregate effects of genetic risks underlying externalizing behaviors (EXT; e.g., risky behaviors, substance use, hyperactivity), have been shown toexplain 10% of the variance in EXT across various population-based datasets (Karlsson Linnér et al., 2021). However, a substantial proportion of this effect may be operating through the environment, a phenomenon known as genetic nurture (Kong et al., 2018). Thus, it is possible that PGS effects for heritable complex traits like ASB may be overestimated. Additionally, while some studies have examined ASB over discrete periods,
examining ASB from a truly developmental perspective is important because not everyone exhibits ASB to the same degree or severity. Examining ASB developmentally also matters in the context of genetics, as heritability of ASB can differ based on trajectory. Our study employs a longitudinal approach to examine the extent to which genetic nurture effects play a role in PGS associations for ASB. We used five Waves of data from Add
Health, spanning ages 13 to 41, using five ASB items measured at each wave. We identified 4 trajectories using growth mixture modeling: Moderate (18.9% in this group), Low (67.0%), High Decline (3.6%), and Adolescence-Peaked (10.6%). Then, we examined associations between PGS for EXT and membership in each ASB trajectory using multinomial logistic regression, where the Low group served as the reference. PGS were associated with increased risk of membership into all groups, but primarily the High Decline, followed by Moderate, and then the AdolescencePeaked trajectory. Then, we conducted tests of mediation by examining the extent to which associations between PGS and High Decline and Moderate trajectories were mediated by supportive parenting (e.g., closeness, warmth), which we used as a proxy variable for genetic nurture effects. We found evidence of mediation in these models, suggesting genetic nurture is likely contributing PGS associations. These preliminary results provide confidence in the success for the next step of our research, which is to estimate both direct and indirect (i.e., “genetic nurture”) effects in the PGS by leveraging family data (i.e., parents, siblings, and self) in the UK Biobank. We will employ a statistical framework called DONUTS (Wu et al., 2021) which leverages family data by generating two sets of GWAS for EXT: one for offspring (GWAS-O), one for parents (GWAS-MP). These data will then be modeled using multinomial logistic regression and mediation models to test the hypotheses that the indirect effects PGS should be mediated by supportive parenting (as it should reflect genetic nurture effects), while the direct effects PGS should have reduced mediation (given it reflects inherited genetic influences). Upon completion of this work, we expect to produce novel PGS for ASB that disentangle environmental signals from genetic ones. We will conduct parallel analyses to validate these PGS. Accounting for environmental signals in PGS will have powerful consequences for developmental psychopathology research, as studies of gene-environment interactions have long been confounded by the always present influence of gene-environment correlations. We will employ a statistical framework called DONUTS (Wu et al., 2021) which leverages family data by generating two sets of GWAS for EXT: one for offspring (GWAS-O), one for parents (GWAS-MP). These data will then be modeled using multinomial logistic regression and mediation models to test the hypotheses that the indirect effects PGS should be mediated by supportive parenting (as it should reflect genetic nurture effects), while the direct effects PGS should have reduced mediation (given it reflects inherited genetic influences). Upon completion of this work, we expect to produce novel PGS for ASB that disentangle environmental signals from genetic ones. We will conduct parallel analyses to validate these PGS. Accounting for environmental signals in PGS will have powerful consequences for developmental psychopathology research, as studies of gene-environment interactions have long been confounded by the always present influence of gene-environment correlations.
URL
https://addhealth.cpc.unc.edu/wp-content/uploads/2022/07/2022-Abstract-Document_AH-Users-Conference.pdfReference Type
Conference proceedingBook Title
2022 Add Health Users ConferenceAuthor(s)
Sasia, A. B.Lu, Q.
Wu, Y.
Li, J. J.