1. Multivariate prediction of childhood psychopathology using polygenic scores
- Author
-
Akingbuwa, Wonuola, Hammerschlag, Anke, Allegrini, Andrea, Sallis, Hannah, Kuja-Halkola, Ralf, Rimfeld, Kaili, Lichtenstein, Paul, Lundstrom, Sebastian, Munafo, Marcus, Plomin, Robert, Nivard, Michel, Bartels, Meike, and Middeldorp, Christel
- Abstract
Childhood psychopathology traits are complex, being affected by a large number of genetic variants, each with a small effect. They are also associated with a number of other phenotypes, both psychiatric and non-psychiatric. These associations may be explained by different mechanisms, including pleiotropy, where the same genetic variant(s) influence multiple phenotypes, or shared biological pathways. Either way, polygenic risk scores (PRS) can and have been used to investigate these associations. When PRS of one trait significantly predict another, we can conclude that the traits are genetically correlated. In our previous study, we used PRS of various adult psychopathology, personality and functional outcome traits to predict childhood psychopathology in univariate analyses – using each PRS to predict each childhood psychopathology measure. Performing meta-analyses across 7 large European cohorts, we found that adult educational attainment (EA), major depressive disorder (MDD), schizophrenia, insomnia, subjective wellbeing, neuroticism, and BMI were genetically associated with symptoms of ADHD, internalizing, and social problems in childhood and adolescence. We also found that the association between adult EA PRS and childhood psychopathology became stronger with age. However, given the increased power of our analyses compared to others (sample size over 40,000 + some of the most powerful GWAS of the adult traits we investigated), the predictive power of the scores were still extremely small, accounting for less than 0.01% of variance. This automatically dampens optimism for any use of these PRS in prognostic research, though increased sample sizes in GWAS will eventually increase the predictive accuracy. Methodological advances, particularly the development of multivariate methods means that we can combine various PRS in joint analyses, leveraging the genetic correlation of the trait PRS, and childhood measures with each other, as well as the correlation between PRS and childhood measures, which should enable us to boost the predictive power. Taking a multivariate approach, we hope to understand the patterns of correlation/association between our traits of interest as well as investigate which adult traits show the biggest contribution to the association with childhood psychopathology.
- Published
- 2022
- Full Text
- View/download PDF