Increasing Trans-Ethnic Diversity In Genome-Wide Association Meta-Analysis Of ADHD

Abstract Attention Deficit Hyperactivity Disorder(ADHD) is a highly heritable childhood behavioural disorder affecting 5% of school-age children and 2.5% of adults worldwide. Recently, the Psychiatric Genomics Consortium and iPSYCH successfully identified 12 independent loci significantly associated with ADHD in genome-wide meta-analysis of 20,183 ADHD cases and 35,191 controls of predominantly European ancestry (Demontis et al., in preparation). We expand upon this result by incorporating additional samples of non-European ancestry in order to advance genome-wide variant discovery, improve fine-mapping, and evaluate transferability of ADHD risk prediction across ancestries. To date, most genome-wide association studies (GWAS) have focused on samples of European descent. As a result, current GWAS results are overly enriched for alleles that are common in European populations, which can yield unexpected biases when trying to evaluate those risk variants in other populations (Martin et al., 2017). Incorporating diverse populations in GWAS can not only address this bias, but improve the resolution of fine-mapping for GWAS effects by allowing comparison of the effects across different LD structures in samples of different ancestry (Zaitlen et al., 2010). Therefore, the current analysis focuses on GWAS of ADHD in non-European ancestry samples for comparison to the existing European-ancestry GWAS results. In particular, we perform GWAS for more than 4,000 ADHD cases and ancestry-matched controls, including cohorts from China and Brazil, and individuals of admixed and predominantly non-European ancestry from large population samples of Denmark and United Kingdom. Logistic mixed models (Chen et al., 2016) are used to analyze the diverse admixed samples while correctly controlling for sample ascertainment. These results are then combined with the European-ancestry GWAS in order to identify additional ADHD-associated loci through trans-ethnic meta-analysis. Fine-mapping is also performed for the 12 previously identified loci. In addition to the trans-ethnic meta-analysis, we also evaluate the transferability of the European ancestry results to other populations using polygenic scores computed from the European ancestry GWAS results as the discovery sample. We quantify the reduced performance of the polygenic scores in non-European ancestry individuals compared to other independent European-ancestry cohorts.