Genetics of Bipolar Disorder: Where Do We Stand?

In this issue are several articles on bipolar disorder. As a cause of human suffering and lives disrupted or lost, the importance of bipolar disorder is very well established, as is its underrecognition, undertreatment, and associated high costs (1–3). The practical benefits of successfully identifying the genes underpinning bipolar disorder could be in making its diagnosis surer and faster and its treatment rational and much more effective.

So, 17 years after the first map of the human genome was published (4), where do we stand in terms of the genetics of bipolar disorder? The primary strategy for locating genes in diseases of unknown etiology has been linkage analysis, employing data from genome maps and DNA samples from families with multiple affected members. Studies of common diseases of unknown etiology, including asthma, diabetes mellitus, and most psychiatric disorders, have proven that this is a difficult exercise. On the other hand, recent genetic discoveries in schizophrenia that began with linkage analyses of genome scans are so encouraging that we have every reason to expect that the identities of genes underlying bipolar disorder will follow.

Where are the bipolar disorder genes? There are many positive linkage findings, although only five of these have met the stringent criteria set years ago for significant linkage (5). Nonetheless, debates have continued about whether any of these linkages are proven and whether our current linkage and association study paradigms will work to find genes for bipolar disorder. There are about 16 favored chromosomal regions of the genome, but the 1,000 or so original reports and the more than 40 review papers plus two published meta-analyses (6, 7) reach sometimes overlapping but also differing conclusions about the location of bipolar disorder genes. The greatest surprise for psychiatric genetics as a whole has been the remarkable success in locating and identifying genes associated with schizophrenia (8, 9) while simultaneously debating the linkage results. In the bipolar disorder studies, a truly notable surprise has been the overlap of genomic regions linked to or associated with both schizophrenia and bipolar disorder (10).

Despite the linkage debates and admitting that further studies are needed, evidence is now mounting for declaring the gene complex G72/G30 as the first confirmed gene related to bipolar disorder. Statistically significant associations of single nucleotide polymorphism (SNP) alleles in this gene complex with bipolar disorder have been observed in four independent samples (11–13). The gene was first isolated in DNA on chromosome 13q33 (14), where there was an early linkage finding in schizophrenia (15) that was patented to a French company (Genset S.A.) 3 years ago. This result has directed our attention to the other genes associated with schizophrenia, especially those residing in chromosomal regions where both bipolar disorder and schizophrenia pedigrees show evidence favoring linkage (8p22, 10p14, 18p11, and 22q11), old paradigms not withstanding (16).

How should we proceed? From the 17-year perspective, the scientific resources produced by the Human Genome Project (large, publicly available genetic databases, mass genotyping services, etc.) and by the NIMH Human Genetics Initiative (DNA and clinical datasets for disease-specific clinical samples [17]) are moving our field forward, albeit in fits and starts. Call these “Big Science” resources.

For complex phenotypes like bipolar disorder, the “Big Science” resources are necessary but not sufficient. Hypothesis-driven departures from the conventional assumptions are essential. Since these usually come from individual or small groups, call them “Boutique” innovations. The departures (innovations) might be changing the operational definition of bipolar disorder (i.e., the phenotype) by clinical or by laboratory means. Familial rather than individual phenotype approaches, devised 40 years ago to divide the mucopolysaccharidoses (18), are now being tested in bipolar disorder studies. Studies of families in which there is clustering of milder bipolar disorder forms point with linkage evidence to one chromosomal region 18q22 (19), whereas studies of families with clustering of psychotic or severe forms of bipolar disorder point to chromosomes 13q31 and 22q11 (20).

Creating new statistical methods for finding genes is also a work in progress. Linkage statistics combine the phenotype and genotype data to predict where genes reside. Association statistics combine the data to predict which DNA sequence variation (in a gene or its regulatory components) contributes to the risk of disease. Quantitative Trait Loci analyses are powerful linkage methods that use more information than the standard binary phenotype. They have helped to identify the genes responsible for the size of tomatoes and are currently being adapted to human studies.

In this issue, Faraone and colleagues employ “Big Science” resources and test an innovative (dimensional) definition of the bipolar disorder phenotype in a linkage study of bipolar disorder. Their study sample, derived from the NIMH Genetics Initiative Bipolar collection, is the largest publicly available bipolar disorder sample in existence. Their genotype data was generated at the NIH Center for Inherited Disease Research. Their boutique innovations were 1) the use of the age at onset of bipolar disorder, not the diagnosis of bipolar disorder itself, as the phenotype and 2) the use of a new statistical method that favors a quantitative rather than categorical definition of the phenotype (21). Their tantalizing results point to 12p, 14p, and 15p as chromosomal regions that could harbor genes related to bipolar disorder. The three chromosomal regions cannot be proclaimed confidently as established sites for bipolar disorder genes. Rather, these regions must stand in line, like the other 16 “favored regions,” for attempts at replication or refutation using other samples. With persistence and innovative approaches, several of these regions will yield important genes. The studies exploring the age at onset, familial phenotypes, and the overlapping schizophrenia and bipolar disorder findings also have the attractive potential to redefine the scourge that, at least for today, we call bipolar disorder.