Objectives To determine whether large (>100 000 base pairs) and rare (found in <1% of the population) CNVs are associated with susceptibility to bipolar disorder and to compare with findings in schizophrenia.
Design A genome-wide survey of large, rare CNVs in a case-control sample using a high-density microarray.
Setting The Wellcome Trust Case Control Consortium.
Participants There were 1697 cases of bipolar disorder and 2806 nonpsychiatric controls. All participants were white UK residents.
Main Outcome Measures Overall load of CNVs and presence of rare CNVs.
Results The burden of CNVs in bipolar disorder was not increased compared with controls and was significantly less than in schizophrenia cases. The CNVs previously implicated in the etiology of schizophrenia were not more common in cases with bipolar disorder.
Conclusions Schizophrenia and bipolar disorder differ with respect to CNV burden in general and association with specific CNVs in particular. Our data are consistent with the possibility that possession of large, rare deletions may modify the phenotype in those at risk of psychosis: those possessing such events are more likely to be diagnosed as having schizophrenia, and those without them are more likely to be diagnosed as having bipolar disorder.