28 December 2011. Bipolar disorder has joined the ranks of brain disorders associated with spontaneously occurring mutations, according to a study published in Neuron on December 22. Led by Jonathan Sebat of the University of California in San Diego, the study finds a higher frequency of non-inherited de-novo copy number variants (CNVs), the wholesale loss or gain of chromosomal segments, in bipolar disorder and in schizophrenia compared to controls. Though the CNV locations were not specific to bipolar disorder, these de novo events seem to contribute to the risk for this disease.
The study also bolsters the role of de novo CNVs in increasing risk for schizophrenia, a story that continues to gain momentum (see SRF related news story). The new findings implicate some familiar loci in schizophrenia, including genes involved in synapses and neurodevelopment.
“The study highlights the importance of a genetic model involving rare and disruptive variants to further our understanding of complex neuropsychiatric traits,” write Santhosh Girirajan and Evan Eichler of the University of Washington in Seattle in an accompanying perspective piece.
Since rare CNVs arrived on the scene as players in the genetic etiology of autism and schizophrenia, bipolar disorder has watched from the sidelines, with mixed evidence for CNV involvement (Priebe et al., 2011; and see SRF related news story). As SRF reported earlier this year from the World Congress on Psychiatric Genetics (see SRF related news story), Sebat and colleagues decided to try the de novo route in bipolar disorder because de novo CNVs had delivered some of the original evidence of CNV involvement in autism and schizophrenia, giving greater effect sizes than those obtained from inherited CNVs (Sebat et al., 2007; Xu et al., 2008). Similarly, in the new study Sebat’s group found that de novo CNVs increased the risk for bipolar disorder with an odds ratio of about 4—several times that reported for inherited CNVs (~1.3). This supports the idea that de novo CNVs are more damaging than those passed from generation to generation, and can contribute to the risk for bipolar disorder.
First author Dheeraj Malhotra and colleagues screened the genomes of 788 subject-mother-father trios, with 185 subjects diagnosed with bipolar disorder, 177 with schizophrenia, and 426 healthy controls. Using a comparative genomic hybridization (CGH) array with higher-than-usual density of 2.1 million probes that can detect CNVs down to 10 kb, the researchers turned up 23 de novo CNVs—14 deletions and nine duplications—in the subjects but not in their parents. Size-wise, these ranged from 15.1 kb to 7,178 kb, with a median of 112 kb, and contained a median of two genes.
These de novo CNVs occurred more frequently in bipolar disorder and schizophrenia compared to controls. Specifically, 10 of the CNVs were found in eight bipolar subjects (i.e., 4.3 percent of bipolar subjects had de novo CNVs; p = 0.009 compared to controls), nine of the CNVs were found in eight schizophrenia subjects (4.5 percent; p = 0.007 compared to controls), and four were found in four controls (0.9 percent).
The frequency of de novo CNVs was also associated with age of illness onset in bipolar disorder, but not in schizophrenia. Subjects with early-onset bipolar disorder, becoming ill at 18 years old or younger, had more de novo CNVs (odds ratio = 6.3) than those who became ill later (odds ratio = 2.9). In contrast, family history of mental illness, defined as a first-degree relative diagnosed with bipolar disorder, major depression, schizophrenia, schizoaffective disorder, autism, or intellectual disability, did not influence de novo CNV frequency. Individuals with bipolar disorder or schizophrenia without a family history of mental illness did not have a higher frequency of de novo CNVs compared to those with a positive family history. This contrasts with previous findings of a higher proportion of de novo CNVs in sporadic cases of schizophrenia (see SRF related news story) and autism (see SRF related news story), and the discrepancies may lie in how family history is determined by different studies.
The researchers also looked at inherited CNVs to see if they contributed any risk for bipolar disorder or schizophrenia. Among the gene-hitting CNVs larger than 100 kb, they found an enrichment for inherited duplications in cases of familial bipolar disorder (OR = 1.77; p = 0.03). All other permutations of CNV type and disorder did not find evidence of larger inherited CNV burdens in familial or sporadic cases.
Location, location, location
Returning to the de novo CNVs, the researchers did not find that the chromosomal locations of those found in bipolar disorder were specific to the disorder. For example, three CNVs landed in regions already implicated in other disorders (3q29, 9p23, 16p11.2), and a follow-up analysis of the 23 de novo CNV locations in genome data from the Bipolar Genome Study (BiGS) did not find an association among these regions and bipolar disorder. In contrast, three regions were significantly associated with schizophrenia in a follow-up analysis of data from the Molecular Genetics of Schizophrenia Study (MGS): 3q29, 7q36.3, and 16p11.2, which have all been previously implicated in schizophrenia.
Pathway enrichment analysis helped to glean some insight into the functions potentially disrupted by these CNVs. The functional categories impacted by the de novo CNVs found in bipolar disorder included cell proliferation, cell shape, and phospholipid metabolism, with the authors noting that the biological relevance of these categories is “far from obvious.” In contrast, the functional categories impacted by the schizophrenia de novo CNVs fit with the familiar themes of neural development and synapses.
But what to make of the significant increase of de novo CNVs in bipolar disorder that don’t seem all that specific to the disease? Girirajan and Eichler attempt an explanation in their perspective article, in which they suggest that the ultimate phenotype of a CNV associated with different conditions depends on the genomic context in which the CNV finds itself. When a CNV occurs alongside other genetic “hits”—be they inherited or de novo variants—this results in a more severe phenotype than when occurring alone, and the authors have turned up evidence for such an additive effect (Girirajan et al., 2011). This view puts bipolar disorder on the less severe end of a neuropsychiatric phenotype continuum, which includes schizophrenia, autism, and, at the most severe end, intellectual disability. The frequency of duplication CNVs, as opposed to more drastic deletions, found in bipolar disorder in this study may fit with this.
If true, then finding the full complement of rare variants either through arrays that can detect ever-smaller CNVs or sequencing to find point mutations (see SRF related news story) will be important in understanding how genetic variants combine to influence risk for neuropsychiatric disorders.—Michele Solis.
Malhotra D, McCarthy S, Michaelson JJ, Vacic V, Burdick KE, Yoon S, Cichon S, Corvin A, Gary S, Gershon ES, Gill M, Karayiorgou M, Kelsoe JR, Krastoshevsky O, Krause V, Leibenluft E, Levy DL, Makarov V, Bhandari A, Malhotra AK, McMahon FJ, Nöthen MM, Potash JB, Rietschel M, Schulze TG, Sebat J. High Frequencies of De Novo CNVs in Bipolar Disorder and Schizophrenia. Neuron. 2011 Dec 22; 72: 951-963. Abstract
Girirajan S, Eichler EE. De Novo CNVs in Bipolar Disorder: Recurrent Themes or New Directions? Neuron. 2011 Dec 22; 72: 885-887. Abstract