So Where Are the Genes for Bipolar Disorder?

Over the past several years, large-scale genome association studies of schizophrenia have been notably successful in identifying a number of statistically significant risk loci in a variety of genes, many of which have been replicated in independent samples and meta-analyses (1). The effect sizes for individual genes, however, remain small, and how they might interact in a given individual or family to produce the syndrome remains obscure. Rather surprisingly, this progress has not been matched in the results reported for comparable efforts to identify risk variants in bipolar mood disorder. Why this should be the case is not entirely clear, and it is somewhat counterintuitive, given the strong heritability of bipolar mood disorder and its more specifically defined clinical phenotype.

One contributing reason may simply be the numbers of people studied; the cohorts accrued in the international consortia to date and the merged data sets are significantly larger for schizophrenia, and it has become clear that sample sizes of tens of thousands, and maybe more, are necessary for clear and replicable signals to emerge in the genetic analysis of complex diseases when heterogeneous populations are studied and the effect sizes of individual gene mutations small.

Another issue, not unique to bipolar disorder, is that the clinical phenotype, although more distinctly defined than many other psychiatric conditions, may still suffer from etiologic heterogeneity and the inclusion of phenocopies. To this end, numerous groups have begun to rely on quantitative assessment of relevant component traits of bipolar disorder, such as activity-level parameters, neurocognitive function and changes in brain structure, and dimensions of temperament, in the hope that defining genetic factors contributing to individual endophenotypic components may be more productive than continuing reliance on categorical diagnosis alone (2). The genetic exploration of lithium responsivity is yet another example of a strategy that may provide a key to unique insights of pathophysiology (3).

One alternative approach emerges from the finding that some of the risk variants that have been reported for schizophrenia have also been found in bipolar disorder, as well as some other psychiatric conditions. This may be indicative of a shared pathophysiology and reflected in components of a shared intermediate phenotype—for example, psychosis. To this end, some research groups have reported that both bipolar disorder and schizophrenia can be reclassified into differing subgroups encompassing both disorders, on the basis of biomarkers such as brain structure, cognition, and sensorimotor profiles (4).

Which of these approaches—brute force through greatly increased numbers of persons studied, or more selective dissection of the unique or shared phenotypes involved—will be more likely to achieve greater clarity in understanding the relation between genetic risk and our most serious psychiatric illnesses remains unresolved.