The highs and lows of bipolar disorder can be considered a metaphor for the exciting and sometimes frustrating search for genes for this complex disorder.
Despite some early starts and stops, this field is now poised to make significant advances regarding the biology of bipolar disorder. This review will summarize the background, technology and recent history of the search for genes and describe the exciting promise of its future impact on clinical practice. Over many decades, numerous epidemiological studies have pointed to the likely role of genetic factors in the predisposition to bipolar disorder Tsuang and Faraone.
These data come primarily from three types of studies: In a family study, the familial nature of the disorder is explored by determining the rate of illness among the first-degree relatives of a bipolar proband. Though these data indicate that bipolar disorder is familial, they do not prove that it is genetic.
To separate nature from nurture, twin studies and adoption studies have been employed. In twin studies, both monozygotic MZ and same-sex dizygotic DZ twin pairs are identified in which at least one twin has bipolar disorder.
The portion of twin pairs in which the co-twin is also affected is referred to as the concordance rate. In the most common strategy, the twin pairs have been raised together, so that presumably environmental factors are the same. The role of genes is indicated by a higher concordance rate among MZ twins, who share all their genes, compared to DZ twins. This represents an over four-fold increase in risk and is the strongest evidence for the role of genes.
Adoption studies separate nature from nurture by examining the rates of illness in the biological parents of probands with bipolar disorder who were adopted away at birth.
Though fewer data are available for this more difficult strategy, these studies also provide some support for the role of genetics. Though the epidemiological data indicating the role of genes have been available for some time, only recently have methods been developed to identify the specific genes involved. This process, referred to as positional cloning, includes two steps: In the genetic mapping or linkage mapping stage, the approximate chromosomal location of a disease susceptibility gene is determined by identifying a DNA marker whose alleles are transmitted along with the illness to the affected members in a family.
It is the development of DNA marker technology that made genetic mapping possible. These markers are highly variable between individuals and have been mapped to precisely known chromosomal locations. Thousands of such markers have now been developed for this purpose, making it possible to systematically search the genome for disease genes.
Once a disease locus has been mapped by genetic methods, its location has been narrowed from the 3 billion base pairs of DNA which make up the entire genome to a specific region on a chromosome typically 5 to 10 million base pairs in size. Physical mapping methods, such as cloning and sequencing large regions of DNA, are then required to identify the disease gene. Ultimately, one or more mutations are found in the disease gene, which cause it to malfunction, resulting in the disorder.
The first reported success using these molecular genetic methods in psychiatry was a study by Egeland and others of bipolar disorder in a large Old Order Amish kindred. These authors examined two markers on the end of chromosome 11 and found strong evidence of genetic linkage. The story was made more enticing by the recent mapping of the gene for tyrosine hydroxylase, which controls catecholamine biosynthesis, to the same chromosomal region Meloni and others.
This finding resulted in a great deal of excitement about the prospects of applying positional cloning methods to psychiatric disorders. Unfortunately, a number of subsequent studies in numerous other populations failed to replicate the finding. Furthermore, a subsequent study of an expanded version of the same Amish pedigree failed to find significant evidence of linkage Kelsoe and others A similar course of events occurred at about the same time regarding a report of linkage on the X chromosome.
A variety of epidemiological data had for many years suggested the possibility of a gene for bipolar disorder on the X chromosome. For example, some studies indicated a lower rate of father-to-son transmissions of illness than would be expected. This is consistent with an X-linked gene.
In , Baron and colleagues examined several conventional genetic markers including color blindness and glucosephosphate hydrogenase deficiency in a set of Israeli families. The researchers found strong evidence of linkage to the Xq28 region. In similar fashion to the report in the Amish, this finding caused considerable excitement and stimulated several efforts to replicate the finding in other populations. Unfortunately, no other study could replicate the linkage result. Subsequently, a reexamination of the same families using molecular genetic markers resulted in significantly diminished evidence for linkage Baron and others It is in those two findings that the course of genetic research seems to mirror the highs and lows of the illness itself.
Investigators in psychiatry and genetics were left disappointed and puzzled by those events. They were also left with the question of how those reversals could occur given the strength of the original findings. Though the answer to this question is still not clear, a likely culprit is the genetic heterogeneity of bipolar disorder.
Epidemiological data suggest that many genes are involved in the genetic transmission of bipolar disorder. They may be a complex combination of genes of large effect and small effect, as well as autosomal dominant, recessive and X-linked transmission. In many families the susceptibility to illness may require the interaction of multiple genes. Other factors also complicate the problem.
The twin studies indicate that some individuals may inherit genes for the disorder, yet never display symptoms. This reduced penetrance of the genes makes it difficult to assess the status of unaffected individuals in families. Further problems are added by the difficulties in diagnosis, the range of manifestations of the illness and the resulting uncertainties in the optimum definition of illness.
The initial response to these dramatic reversals was a conservative one. Many investigators assumed that false positives would be the most serious problem confronting such studies.
Many, therefore, proposed that much higher statistical thresholds lod scores must be achieved before linkage could be confidently declared. Subsequent experience seemed to defy this expectation.
As more investigators surveyed more of the genome in more families, they began to fear they would reach the end of the genome with no linkages rather than finding frequent high lod scores. The field, therefore, went through a discouraging period of several years when no positive results were reported meeting the high threshold assumed necessary for statistical significance.
After this period of scientific depression, the outlook has recently begun to improve based on several new developments. Stimulated at least partly by the discouraging results in psychiatry, geneticists developed alternative, more sophisticated statistical methods for complex heterogeneous disorders.
Some of these methods termed nonparametric methods avoided the problems resulting from assuming a given mendelian mode of transmission and instead focused only on the sharing of marker alleles between affected individuals. Other approaches have involved developing more complex models of the interaction of multiple genes in the same families.
Also, methods were developed to determine how sensitive a given linkage result is to the diagnosis in a given individual. This would prevent the dramatic change in linkage statistics resulting from a new onset of illness, as occurred in the original Amish chromosome 11 finding.
Another important change in the field has been a move to greater caution regarding both false positives and false negatives.
In other words, strong linkage results should not be too readily embraced, and mildly suggestive results not too readily rejected. Rather, support for a finding should come from multiple families, multiple markers, a variety of statistical methods and ultimately independent replication in different family collections. Such support seems to be developing for several different regions of the genome.
One of the most promising is the peri-centromeric region of chromosome This finding was originally reported by Berrettini and colleagues, who found suggestive evidence of linkage using conventional model-based methods in several families at several markers on chromosome However, using nonparametric methods, such as the affected sibling pair method, they found highly significant evidence for linkage.
Subsequently, several other groups have reported independent replication of these results, though some found the strongest evidence for linkage at markers somewhat distant from those originally reported. In their replication of the chromosome 18 linkage, Stine and others found the strongest evidence of linkage in families in which illness was transmitted through the fathers rather than the mothers.
Such paternal transmission suggests a parent-of-origin effect as seen in other genetic disorders. It may indicate genetic imprinting in the transmission of bipolar disorder. Chromosome 21 Another promising region is chromosome Straub and colleagues have reported and Detera-Wadleigh and coworkers confirmed evidence of linkage to several markers on chromosome 21 21q Though the strongest evidence for linkage was found in one large American family, nonparametric statistical methods indicated strong evidence for linkage in their entire family collection.
This linkage has been replicated recently by British investigators who found evidence for linkage to both chromosome 21 and the chromosome 11 region originally reported in the study of the Amish Gurling and others. They employed a statistical method that examined the effect of both these loci together in their family collection.
Further support for the original chromosome 11 region comes from an association study of the tyrosine hydroxylase gene Meloni and others; however, other investigators have not replicated these results. Several other regions of the genome appear as promising spots for bipolar loci.
Dawson and others observed co-segregation of bipolar disorder and a rare skin disease, Darier's disease, in one Welsh family. This autosomal dominant skin disorder has been mapped to 12qq The researchers then examined markers near the Darier's locus in several bipolar families not affected with Darier's disease and found suggestive evidence of linkage to bipolar disorder.
The X Chromosome A decade ago, Mendlewicz and colleagues focused attention on X chromosomal markers for manic depression. Now, a recent report by a Finnish group of linkage to X chromosome markers brings renewed interest to this chromosome Pekkarinen and others , as has the work of a French group Lucotte and coworkers.
This writer's laboratory has recently reported suggestive evidence of linkage to the locus for the dopamine transporter gene on chromosome 5 Kelsoe and others. As the site of action of amphetamine and cocaine, this is a very interesting candidate gene for bipolar disorder. Another group of investigators have also recently reported suggestive evidence for linkage to loci on chromosome 16 Ewald and others. What lies ahead in the search for bipolar genes? These promising results suggest that the approximate chromosomal location of several genes for bipolar disorder may have already been identified.
For those loci already replicated, further work needs to be done to both more securely confirm the result, and to more finely map the implicated chromosomal region.
Of the more preliminary results, some will likely be confirmed through replication, while others may prove to be false positives. Soon, it will be very exciting to see some of these loci move to the next stage of the process, physical mapping. Here a task equally arduous to that of genetic mapping will be faced. It will likely produce its share of frustration and false positives, but ultimately will lead to the identification of specific susceptibility genes and the mutations affecting them.
The benefit of identifying such specific genes will be great and manifold. First and most importantly will be elucidation of the basic brain pathophysiology of the disorder. As many of these genes will likely be previously unknown, they may point to errors in previously unknown neuronal physiology or new aspects of known neurotransmitter systems.
This may lead to a better understanding of the mechanism of action of current therapeutic agents, and may identify targets for the development of new drugs.