7 March 2013. The first genomewide search of five psychiatric disorders together finds they share some genetic roots, reports a study published online February 28 in The Lancet. Orchestrated by the Cross-Disorder Group of the Psychiatric Genomics Consortium (PGC), with Jordan Smoller of Massachusetts General Hospital in Boston at the helm, the study pinpointed four genomewide significant signals in a sample consisting of people with schizophrenia, autism, attention deficit-hyperactivity disorder (ADHD), bipolar disorder, and major depressive disorder. The results suggest some of the same biological processes are perturbed in these disorders, and particularly highlight calcium channel signaling.
The findings bring to light overlapping genetic risk factors that have been hinted at by family and twin studies. For example, relatives of someone with bipolar disorder are at increased risk for the disorder, but also for schizophrenia (see SRF related news story). Similarly, having a first-degree relative with schizophrenia or bipolar disorder increases risk for autism (see SRF related news story). More recently, this idea has been vividly illustrated by studies of rare copy number variants (CNVs), in which the same deletions or duplications are associated with different disorders (Malhotra and Sebat, 2012; and see SRF related news story).
The new study looks for any contributions by common variants to risk for all five disorders. Previous genomewide association studies (GWAS) have combined pairs of disorders before, like schizophrenia and bipolar disorder (International Schizophrenia Consortium, 2009; see SRF related news story), but the new study is the first time five different disorders have been mixed together. Though mixing disorders like this would likely dilute any disorder-specific signal, it will increase the power—through bigger sample sizes—to find any signals that contribute to psychiatric disease in general.
To assemble this melting pot, the researchers drew from samples of the other disorder-specific PGC groups. This amounted to 33,332 cases and 27,888 controls, all of European ancestry but coming from more than 19 countries. Among the cases, 4,949 were diagnosed with autism, 2,787 with ADHD, 6,990 with bipolar disorder, 9,227 with major depressive disorder, and 9,379 with schizophrenia. The researchers then tallied over one million single nucleotide polymorphisms (SNPs) to see if any were overrepresented in the psychiatric disorder group compared to controls.
This revealed four regions of the genome with signals reaching genomewide significance (p <5 x 10-8): one on chromosome 3p21.1; one at 10q24; one within calcium channel gene CACNB2, also on chromosome 10; and another within another calcium channel gene, CACNA1C, on chromosome 12. As has been the case for common variants, each of these increased risk only slightly, with odds ratios between 1.07 and 1.13. For three of these regions, largely similar effects were seen when breaking down the cases by disorder, which suggests that these three signals in the mixed sample were not driven by a select few disorders. The exception was the CACNA1C signal, which seemed largely driven by schizophrenia and bipolar samples. This is consistent with the evidence for CACNA1C’s involvement in each of these disorders (see SRF related conference story; SRF related news story; SRF news story).
Some of the genes of interest were hard to resolve. The signal on chromosome 3 emanated from an intron of ITIH3, but given its tight linkage with nearby SNPs, it implicates up to 35 other genes in the region. Similarly, the SNP at 10q24 lies within an intron of AS3MT, but casts suspicion on another 25 genes in the area. In contrast, the SNPs within the subunits for the same type of voltage-gated calcium channels send a clear message about the importance of calcium signaling.
To get a more fine-grained look at the genetic overlaps among these five disorders, the researchers made pairwise comparisons among them. Specifically, they asked how well the risk variants associated with one disorder could account for risk of another disorder. For example, the researchers took the SNPs associated with schizophrenia in a previous GWAS and measured their combined contributions to risk with a polygenic risk score (International Schizophrenia Consortium, 2009). Then, they computed the polygenic risk score derived from these schizophrenia-associated variants for each of the other four diseases. This accounted for some variance in risk for bipolar disorder and for major depressive disorder, but less so for autism or ADHD. The other pairwise analyses revealed a similar picture, with greater overlaps among schizophrenia, bipolar disorder, and major depressive disorder, which appear in adolescence or early adulthood, than between the childhood-onset disorders of autism and ADHD.
The evidence for genetic overlap does not deny the uniqueness of each disorder, and future research will have to separate the generic risk factors from disease-specific ones, as a recent gene network analysis attempts to do for autism, ADHD, schizophrenia, and X-linked intellectual disability (Cristino et al., 2013). The genomewide significant signals found in the new study indicate that common variants may set a common stage upon which other disease-specific risk factors act. These genetic overlaps also somewhat blur the categorical distinctions among psychiatric disorders, and conjure up a more enlightened future when these conditions may be defined by their causes rather than their outward signs.—Michele Solis.
Cross-Disorder Group of the Psychiatric Genomics Consortium. Identification of risk loci with shared effects on five major psychiatric disorders: a genomewide analysis. Lancet. 2013 Feb 27. pii: S0140-6736(12)62129-1. Abstract