DSM-5 and Psychiatric Genetics — Round Hole, Meet Square Peg

Abstract

It is useful to consider genetic variation as either common (e.g., an allele found in .5% of the population) or rare because these categories of variation are analyzed using different approaches and have different properties. Most importantly, for neurodevelopmental disorders (NDDs), common genetic variation (e.g., single nucleotide polymorphism [SNP]) is associated with very small effect sizes given evolution constraints on deleterious variation (negative or purifying selection), whereas rare variation can be associated with a much wider range of effect sizes. Chaste et al. (1) look at common variation in autism spectrum disorder (ASD) to answer questions about relationships between clinical and genetic heterogeneity. Enormous advances have been made in understanding of the genomic architecture of ASD, including the role of common and rare variation. There is compelling evidence from multiple studies that common variation represents the major proportion of the genetic risk for ASD (see (2) and citations therein). However, no SNP has been reliably associated with ASD to date because of small effect sizes. Until studies include many thousands of cases, it is exceedingly unlikely that many replicated common variation findings will be made in ASD. There has been much more success with gene discovery in ASD when focusing on rare variation. This increased success is due partly to the fact that there is a significant amount of de novo mutation in ASD: with de novo mutation being quite rare, even a few cases with de novo deleterious variation in a given gene are sufficient to provide statistically significant support for that gene in ASD (see (3) and citations therein). Although the effect size for discoverable rare variation is higher than that of common variation, the total variance explained by rare variation is quite low. Gaugler et al. (2) showed that within a given family with ASD, a rare de novo copy number variant (CNV) or single nucleotide variant can often be the difference between an ASD diagnosis or no ASD diagnosis; however, there must be a “genetic background” in the family, defined by multiple inherited SNPs and other genetic variation, that is a critical part of the architecture in that family. In other words, there appears to be risk in the family in most cases in the form of a multiplicity of common variation, and higher risk rare variation pushes an individual in that family over a liability threshold to manifest with an NDD. This model can explain why risk of family recurrence is high in ASD, while, at the same time, affected sibling or relative pairs within a family may have different rare risk variants, as first shown with CNV and, more recently, with single nucleotide variant (4,5). Chaste et al. study a large (N = 2576 families) and behaviorally well-phenotyped ASD sample (the Simons Simplex Collection) to ask whether subgrouping to enhance phenotypic homogeneity increases the ability to make SNP