August 20, 2013. People with different psychiatric disorders have similarities in their genomic profiles, according to a study published online August 11 in Nature Genetics. From the Cross-Disorder Group of the Psychiatric Genomics Consortium (PGC) and led by Naomi Wray of the University of Queensland, Brisbane, Australia, and Kenneth Kendler of Virginia Commonwealth University, Richmond, the study surveyed datasets of common variation across the genome for cases of schizophrenia, bipolar disorder, major depressive disorder (MDD), autism spectrum disorder, attention-deficit/hyperactivity disorder (ADHD), and controls. Common variants in genes accounted for 17-28 percent of risk for each disorder, and the data revealed genetic similarities between disorders, including schizophrenia and bipolar disease, and schizophrenia and MDD.
The results put numbers on the genetic overlaps hinted at by previous family studies, which have found increased risk for one disorder in a first-degree relative (parent, sibling, or child) of someone with a different disorder. This could reflect shared genes and/or something about the family environment. The new results, based entirely on genetic data from unrelated people, indicate a clear contribution by genetic factors, specifically common variants. They also reinforce the notion that psychiatric disorders lie on a continuum, with some of the same biological processes contributing to different disorders.
“It could be that what we're picking up is a kind of common vulnerability for risk to psychiatric disorders,” Wray told SRF. “We don’t know if the bits we haven't looked at, for example, the less common variants, are also shared between disorders or are specific to the disorders.”
Earlier this year the same consortium published a genomewide association study (GWAS) on the combined datasets collected by the disorder subgroups of the PGC. This revealed four regions in the genome contributing to risk for the combined psychiatric disorders (see SRF related news story). Though the new study uses the same dataset, which consists of genotypes from 32,298 cases and 46,051 controls, it is not concerned with identifying specific regions. Instead, it estimates the size of the overall contribution of common variants to risk without knowing which variants are doing the contributing. To do this, the study measures the overall pattern of genetic variation—the common sort tagged by nearly one million single nucleotide polymorphisms (SNPs) across the genome—and asks how similar this profile is among people with the same disorder or with different disorders. This allows the researchers to estimate what proportion of risk for each disorder can be explained by common variants.
In essence, the findings quantify what GWAS can deliver. So far, genomewide-significant hits account for a small portion of heritability of a disorder such as schizophrenia—the so-called “missing heritability” problem (Maher, 2008). This has led some to suggest that the genetic culprits are the rarer sort not measured by GWAS. By finding that common variants account for a good-sized chunk of risk for each disorder (17-28 percent), the new study casts the common variant sector as a real player and supports those who advocate for larger sample sizes in GWAS to increase their power to identify the operative variants.
While not denying a role for rare variants, Wray suggests that the proportion of risk explained by common variants may increase with more careful phenotyping. For example, if different subtypes of schizophrenia lurk within the samples, this could dilute any subtype-specific genetic signals.
“So the next step forward is not just about larger sample sizes in GWAS; it’s about larger sample sizes with consistently recorded phenotypic information that we could use to pin that genetic heterogeneity onto,” she said. “I think that's where we want to be going with the psychiatric disorders. Empowered by the fact that this is working, it’s worth the investment.”
Nailing down numbers
First author Sang Hong Lee and colleagues began by asking how much common genetic variation could account for heritability—the genetic factors contributing to risk for each disorder. Using statistical methods applied to human height (Yang et al., 2010) and later to schizophrenia (see SRF related news story), the researchers compared the patterns of SNPs in each individual to get a read on how genetically different cases were from controls. As already reported for schizophrenia, this measure of genomic variance accounted for 23 percent of the variance in risk for schizophrenia (see SRF related news story). Though sizeable, this proportion falls short of 81 percent—the measured heritability for schizophrenia.
The analysis told a similar story for the other disorders, with substantial contributions to risk that did not entirely account for estimates of heritability. For bipolar disorder, common variants accounted for 25 percent of risk compared to its 75 percent heritability; for MDD, this was 21 percent compared to its 37 percent heritability; for autism, 17 percent compared to its 80 percent heritability; and for ADHD, 28 percent compared to 75 percent heritability. These differences could reflect the unmeasured contributions by rare variants and/or diluted common variant signals that could result from grouping together heterogeneous cases of a disorder.
Next, the researchers measured how similar genetic profiles were from one disorder to another disorder. The resulting correlation was highest between schizophrenia and bipolar disorder (r = 0.68 ± 0.04 s.e.) and consistent with previous studies finding genetic overlaps between the two disorders (see SRF related news story and SRF news story). This suggests that a large proportion of the genomic signature for schizophrenia is shared with bipolar disorder as well.
A medium-sized and somewhat unexpected correlation was found between schizophrenia and MDD (r = 0.43 ± 0.06 s.e.); this translated into a 1.6-fold increase in risk for MDD in first-degree relatives of someone with schizophrenia. The researchers verified this effect by doing a meta-analysis of five family studies, which gave a similar 1.5 increase in risk. A small but significant correlation was found between schizophrenia and autism (r = 0.16). This small sign of overlap from the common variant sector complements evidence of shared rare variants, typically copy number variations (Malhotra and Sebat, 2012). The authors also reported moderate correlations between bipolar disorder and MDD (r = 0.47± 0.06 s.e.) and ADHD and MDD (r = 0.32 ± 0.07 s.e.). Significant correlations were not found between other disorder pairings or between the five psychiatric disorders and Crohn’s disease, a non-neural disorder serving as a negative control. The genetic correlations the researchers did find remained even when accounting for potential misdiagnoses.
These shared genetic factors argue that some of the same biological processes go awry in different disorders and highlight a problem with treating disorders as separate entities. The National Institute of Mental Health’s Research Domain Criteria (RDoC) project hopes to remedy this with a classification scheme that will organize psychiatric disorders according to their underlying biology rather than their symptoms. Though the frontiers of causal biology for psychiatric disorders remain largely untraversed, knowing what the common variant approach can provide helps decide how to get there.—Michele Solis.
Cross-Disorder Group of the Psychiatric Genomics Consortium. Genetic relationship between five psychiatric disorders estimated from genome-wide SNPs. Nat Genet. 2013 Aug 11. Abstract