14 June 2007. Two recent genome-wide association (GWA) studies support the notion that many genetic variants each contribute a little to the risk of developing bipolar disorder. The largest of the new studies, published in Nature on June 7, comes from The Wellcome Trust Case Control Consortium, a collaboration of genetics researchers in the United Kingdom, including lead author Peter Donnelly of the University of Oxford, England, and Nick Craddock of Cardiff University, Wales. It comes on the heels of another study, directed by Francis J. McMahon of the National Institute of Mental Health (NIMH), that appeared online in Molecular Psychiatry on May 8.
Some of the markers implicate genes or chromosomal regions that have previously been tied to schizophrenia, which will add grist to the argument that schizophrenia and bipolar disorder share too many symptoms and genes to justify defining them as distinct syndromes (see SRF Live Discussion led by Nick Craddock and Mike Owen; Maier et al., 2006).
In their study, McMahon, first author Amber E. Baum of NIMH, and colleagues in Germany and the United States searched for single-nucleotide polymorphisms (SNPs) linked to the risk of having bipolar l disorder. This “classic” form of bipolar disorder features one or more manic or manic-depressive episodes, and sometimes psychosis or episodes of major depression (see NIMH information about the different forms of bipolar disorder). Baum and colleagues recruited 461 unrelated people with bipolar l disorder who had affected siblings, and 563 controls without major depression or a history of bipolar disorder or psychosis; all had solely European ancestry. As a check against the false positives that can compromise GWA studies, the researchers replicated their findings in a German sample of 772 people with bipolar l disorder, identified through hospital admissions and linkage studies, and 876 controls with no history of affective disorder or schizophrenia.
To control costs, the investigators used pooled DNA from many subjects to detect genes related to bipolar disorder and used individual genotyping to confirm the associations. This process detected 10 genes that were associated with bipolar disorder in both samples when analyzed separately and an extra 15 in the combined samples. Several mapped to areas previously tied to bipolar disorder or schizophrenia. Controls as well as cases had risk alleles, but cases typically carried nine or more.
Calling the effect sizes “modest,” Baum and colleagues write that the highest odds ratio based on individual genotyping, 1.67 (95 percent CI 1.32-2.13), was for a SNP in SORCS2, “which maps to a region on chromosome 4p that has been widely linked to bipolar disorder.” Three SNPs in SORCS2 and three in the gene encoding diacylglycerol kinase eta (DGKH) showed significant associations in both samples. The enzyme DGKH is notable in that it acts in a lithium-sensitive pathway.
The Wellcome Trust study searched for genetic ties to seven diseases in residents of Great Britain and addressed methodological issues that affect GWA studies. It enrolled 2,000 cases each for bipolar disorder, coronary artery disease, Crohn’s disease, hypertension, rheumatoid arthritis, type 1 diabetes, and type 2 diabetes. Bipolar disorder cases had contacted mental health services and met Research Diagnostic Criteria for a lifetime diagnosis of the disorder. Half of the 3,000 controls were born during one week in 1958; the others gave blood anonymously for the study. Analyses omitted people of non-Caucasian descent.
Case-control comparisons found 24 independent associations that met the consortium’s strictest criteria (P <5 x 10-7), including one each for bipolar disorder and coronary artery disease, nine for Crohn’s disease, three for rheumatoid arthritis, seven for type 1 diabetes, and three for type 2 diabetes. According to the researchers, many coincide with prior findings, and others have been backed by later studies, validating the GWA approach.
As Craddock, who led the bipolar part of the study, tells SRF via e-mail, “The data suggest that, at least as currently defined, there are fewer susceptibility genes of relatively large effect in bipolar disorder than for several of the other diseases studied, but more genes of smaller effect.” One SNP on chromosome 16p12 stands out for its strong evidence of a link with bipolar disorder, though this finding did not receive additional support in comparisons that used an expanded reference group of nearly 15,000 subjects, which was created by adding cases from the other six disease groups to the controls. While acknowledging the need for replication, the authors note that several genes at that site could affect bipolar disorder, including DCTN5 or dynactin 5, which “encodes a protein involved in intracellular transport that is known to interact with the gene ‘disrupted in schizophrenia 1’ (DISC1).” Other genome regions topping the list as possibly important for bipolar disorder (P values in the 10-5 range) contain genes affecting voltage-gated potassium channels and synaptic function, as well as GABA and glutamate neurotransmission. The marker in SYN3, the gene for synapsin 3, is notable in this regard, as synapsin 3 has been linked to schizophrenia in association and expression studies.
At first glance, it may seem that the NIMH and Wellcome papers implicate different genes in bipolar disorder. However, Craddock explains that the data sets have yet to be perused in enough detail to warrant that conclusion, since the papers focused on the “top hits.”
To Baum and colleagues, GWA studies serve as “a powerful alternative to genetic linkage studies, which are often underpowered to detect genes contributing to complex phenotypes, and to candidate gene association studies, which are biased by the choice of genes included.” Even so, Craddock warns, “There are opportunities for problems at every stage from sample preparation, through running chips, to cleaning and managing data. With such large data sets, small errors can generate highly significant differences between cases and controls. Such spurious positives must be weeded out ruthlessly before the data set is analyzed.”
Despite the large samples needed to adequately power GWA studies, methods used in the NIMH and Wellcome studies, such as pooling DNA and sharing control groups, can make them more manageable. Researchers intrigued by the possibilities can dive into the data sets themselves: just surf to the NIMH or WTCCC websites and apply for access to compare and contrast the results.—Victoria L. Wilcox.
Baum AE, Akula N, Cabanero M, Cardona I, Corona W, Klemens B, Schulze TG, Cichon S, Rietschel M, Nöthen MM, Georgi A, Schumacher J, Schwarz M, Abou Jamra R, Höfels S, Propping P, Satagopan J, Detera-Wadleigh SD, Hardy J, McMahon FJ. A genome-wide association study implicates diacylglycerol kinase eta (DGKH) and several other genes in the etiology of bipolar disorder. Molecular Psychiatry. May 8, 2007. Advance online publication. Abstract
The Wellcome Trust Case Control Consortium. Genome-wide association study of 14,000 cases of seven common diseases and 3,000 shared controls. Nature. June 7, 2007;447(7145):661-678. Abstract
Q&A with Nick Craddock. SRF questions by Hakon Heimer and Victoria L. Wilcox.
Q: What do you see as the most important findings of the Wellcome Trust study regarding the genetic underpinnings of bipolar disorder?
A: This data set, and others that will become available in the near future, open up the possibility for systematic identification of the biological systems influencing bipolar and related disorders. The most significant association signals are obviously of great interest, and many are likely to implicate specific genes in pathogenesis. However, the great strength is having genotypes available on all individuals for all SNPs because this allows (a) sets of genes to be examined for association (rather than just one SNP at a time), and (b) subsets of patients to be examined according to their clinical characteristics (rather than just treating them as a “case”).
Q: How do the findings relate to schizophrenia and other psychoses?
A: We have also examined an additional set of 700 mood-psychosis spectrum cases using the same approach (not yet published) and are currently undertaking analyses that will inform understanding of the relationship among bipolar disorder, schizophrenia and related psychoses.
Q: Do the Wellcome Trust results alter or strengthen ideas about breaking down the “Kraepelinian divide”?
A: The published findings related to a specific definition of bipolar disorder and used only a case-control analysis and so, themselves, do not directly address the issue of the relationship between mood disorders and psychosis. Our ongoing analyses will.
Q: The study included patients with bipolar disorder classified by “Research Diagnostic Criteria.” How would this group differ from patients diagnosed in clinics in the UK or in the U.S.? Is this group of patients likely to be different from the Baum study population?
A: The vast majority of the sample also meets DSM-IV criteria for bipolar l disorder. The main difference is that about 9 percent of the sample meets criteria for bipolar II disorder. (Research Diagnostic Criteria were developed in the U.S., and DSM-III was largely based on them.)
Q: It seems that the Wellcome study and the Baum study differed in the genes they found linked to bipolar disorder. What might account for that?
A: That is a premature conclusion. The important question is whether, when the two data sets are compared in detail, there is evidence of support for loci across the studies. I think it likely that when this is done there will be genes that receive support from both data sets.
It is wrong to focus just on top hit(s) because in complex genetics you do not expect the same top signals to arise in each data set examined. This is the experience in diseases, such as type 2 diabetes, where several susceptibility genes are robustly known. Even in very large data sets (larger than that of Baum et al.), a particular robustly known gene may not show a strong signal.
Q: When will the data become available to other researchers?
A: Researchers can apply for access to the data, and details are given in the paper. The process should not take long.