Advances in the Genetics of Attention-Deficit/Hyperactivity Disorder

Abstract

One of the most replicated findings in studies of attentiondeficit/hyperactivity disorder (ADHD) is its very high heritability, which averages approximately 75% across 20 twin studies conducted on three continents (1). ADHD’s high heritability kick-started a search for DNA variants with the hope that their discovery would lead to advances in diagnosis and treatment. During the dark decades of linkage and candidate gene studies, progress was slow, but in the past few years, boosted by genome-wide association studies (GWAS), the ADHD research community has made real breakthroughs. GWAS taught us that 25% to 30% of ADHD’s heritability could be accounted for by a polygenic liability consisting of many common DNA variants (2). GWAS also gave us replicated discoveries of rare deletions and insertions known as copy number variants (CNVs) (3). These GWAS data confirmed what many had suspected from epidemiologic data, that ADHD’s genetic liability consists of a complex mixture of common and rare DNA variants. In this issue of Biological Psychiatry, two articles address ADHD’s complex inheritance. Martin et al. (4) illustrate the use of the recently developed “polygenic paradigm,” which, remarkably, was conceptually foreshadowed by Gottesman and Shields (5) more than 40 years ago. The polygenic paradigm uses a discovery sample (in this case ADHD cases and controls) to discover and define a polygenic liability score indexing the number of ADHD risk alleles carried by an individual. Each risk allele or risk variant is the nucleotide that is more frequently observed among ADHD cases, using an arbitrary (and typically liberal) level of statistical significance. When defined like this, many of the risk variants comprising the polygenic liability score will be true risk variants, but many will be false positives. It is the statistical significance of the polygenic score that assures us that it has captured many true positives in the mix. ADHD’s polygenic liability score is a molecular genetic tease. On the one hand, it confirms that many common DNA variants (defined by single nucleotide polymorphisms) are associated with ADHD, but it cannot tell us if “many” means hundreds or thousands or even more, and it cannot tell us which of the many variants comprising the score are truly associated with the disorder. Despite this limitation, polygenic scores provide a valuable tool. For example, when the Psychiatric Genomics Consortium applied this method to schizophrenia, bipolar disorder, depression, ADHD, and autism spectrum disorders (ASDs), a polygenic background defined by single nucleotide polymorphisms accounted for 17% to 29% of the variance in liability to these disorders (http://www.med.unc.edu/pgc/) (2). Moreover, for some disorders, the correlation among their polygenic liabilities was