<<

. 50
( 68 .)



>>

sharing of two haplotypes (or conversely, diminished sharing of haplotypes)
provides evidence for linkage. The sib-pair method is a powerful design if the
gene is rare and requires no assumption regarding the mode of inheritance of the
disorder.
292 K. Merikangas and K. Yu


Markers in the candidate region identified by linkage analysis can be used to
narrow the location of the disease gene through linkage disequilibrium analysis.
Linkage disequilibrium is a population association between two alleles at different
loci, and occurs when the same founder mutation exists in a large proportion of
affected subjects in the population studied. Usually, the closer the marker is to the
disease locus, the greater the proportion of affected subjects who carry the identical
allele at the marker (Risch, 2000). However, in measuring the strength of linkage
disequilibrium for a given marker, it is also important to select unaffected control
subjects from the same population, because an allele shared among affected
subjects may also be common in the general population and thus shared by chance
rather than due to proximity to the disease locus (Risch, 2000).


Review of empirical evidence
The initial enthusiasm generated by early claims of linkage between bipolar
disorder and DNA markers including Xq (Baron et al., 1987; Mendlewicz et al.,
1987; Lucotte et al., 1992) and 11p (Egeland et al., 1987) was diminished by
subsequent non-replications (Risch and Botstein, 1996). Numerous linkage stu-
dies of bipolar disorder have subsequently been reported to regions on all but six
chromosomes (Straub et al., 1994; Pekkarinen et al., 1995; Stine et al., 1995;
Blackwood et al., 1996; Detera-Wadleigh et al., 1996, 1999; Freimer et al., 1996;
McMahon et al., 1997; Moldin, 1997; Smyth et al., 1997; Ewald et al., 1998a, b;
Aita et al., 1999; Kelsoe, 1999; Morissette et al., 1999; Kelsoe et al., 2001).
Table 13.3 presents a summary of genome-wide linkage studies of bipolar disorder
updated from a recent review by Prathikanti and McMahon (2001). Based on a
total of 3538 bipolar I disorder scans in affected subjects from 1119 pedigrees
reported in 20 samples, the authors conclude that no two studies conclusively
implicate the same region. However, suggestive findings emerged for two loci
(i.e., 4p12“13 and 13q31“33). The most striking conclusion was that no two studies
employed identical ascertainment procedures and there was substantial diversity
in sampling and methods. In conclusion, sufficient ambiguities exist to give pause
in considering any of these linkage results as unambiguously replicated (Goldin
¨
et al., 1997; Reus and Freimer, 1997; Rice et al., 1997; DeLisi et al., 2000; Altmuller
et al., 2001; Craddock and Jones, 2001). More recently, McMahon et al. (2001)
¨
reported replication of the suggestive finding reported earlier by Nothen et al.
(1999). The demonstration of the low power of existing linkage studies by Risch
and Merikangas (1996) generated a spate of association studies of mood disorders,
particularly those employing within-family controls (Merikangas et al., 2002b).
Some other features of mood disorders that complicate genetic analyses are
described below.
Table 13.3 Summary of recent bipolar linkage studies

Study Sample Pedigrees Affected subjects Genotyped pairs Pair type Number of markers Scanned what? Locus = LOD

Straub Columbia/ 47 490 443 ERP 5“153 Genome-wide 21q22 = 3.4
et al. Hadassah markers per
(1994) University chromosome
Ewald Danish 2 89 87 ERP 12 Chromosome 16p13 = 2.5
et al. 16p13
(1995)
Blackwood Edinburgh 1 27 26 ERP 135 (87 additional Genome-wide 4p12“13 = 4.8
et al. markers on 4p)
(1996)
McMahon Clinic and 23 251 228 ASP; 13 Chromosome18 18q23 = 2.8
et al. inpatient ERP
(1997) from
Baltimore
and Iowa City
Ginns Older-order 5 207 202 ERP 551 Genome-wide 6p25 = 2.5;
et al. Amish 13q13 = 1.4;
(1996) 15q21 = 1.1
Adams Australian 11 224 213 ERP 214 Genome-wide 4q35 = 3.2
et al.
(1998)
Ewald Danish 2 89 87 ERP 16 Chromosome 12q24 = 3.4
et al. 12q22-q24
(1998b)
Ewald Danish 2 89 87 ERP 16 Chromosome 4 4p16 = 2.0
et al.
(1998a)
Table 13.3 (cont.)

Study Sample Pedigrees Affected subjects Genotyped pairs Pair type Number of markers Scanned what? Locus = LOD

Ginns Older-order 4 68 64 ERP 980 Genome-wide 4p12“13 = 4.1;
et al. Amish 4q = 3.3
(1998)
Detera- NIMH-CNG+ 22 159 137 ERP 607 Genome-wide 13q32 = 3.5;
Wadleigh right 1q31“32 = 2.67;
et al. extension 18p11 = 2.32
(1999) of Amish
Morissette Saguenay-Lac 1 53 52 ERP 332 Genome-wide 12q23“24 = 1.3
et al. St. John
(1999)
Aita et al. USA 40 373 333 ERP 31 Chromosome 21 21q22 = 3.4
(1999) (57 pedigrees)
and Israel
(18 pedigrees)
¨
Nothen German 57 176 119 ASP 23 Chromosome 18 18p11.2 = 2.5;
et al. 18q22“23 = 2.1
(1999)
Kelsoe San Diego/ 20 76 56 ERP 443 Genome-wide 22q13 = 3.8
et al. Vancouver,
(2001) British
Columbia
Cichon Germany/Israel 75 206 131 ASP 33 Chromosome 10 10q25 = 2.9
et al. /Italy
(2001)
McMahon Clinic and 58 586 259 NRP 32 Chromosome 18 18q21“23 = 4.7
et al. inpatient
(2001) from
Baltimore
and Iowa City
Liu, C. Caucasian + right 22 371 349 ERP 16 Chromosome 13 13q32 = 3.3
et al. branch of (specific to area
(2001) old-order near 13q32)
Amish pedigree
+ 3 Ashkenazi
Jewish families
Liu, J. USA and Israel 56 range 431 ERP 14 Chromosome 21 21q22 = 3.6
et al. 156“576 (specific to area
(2001) near 21q22)

Adapted from Merikangas et al. (2002b).
ERP, extended pedigree; ASP, affected sib pairs; NIMH-CNG, National Institute of Mental Health, Collaborative genetics study.
296 K. Merikangas and K. Yu



Challenges to the identification of genes for mood disorders
The major problem in defining future strategies for identifying genes for mood
disorders is the lack of consistent findings from existing studies, and uncertainty
regarding appropriate designs and methods for detecting genes for complex
disorders. Although there is a substantial degree of pessimism regarding the
identification of genes underlying mood disorders, the lack of success in identify-
ing major genes for psychiatric disorders in general and mood disorders specifi-
cally using approaches that had been successful for many single-gene disorders is
not altogether unexpected in the light of the complexity of mood disorders.
The conclusion that the lack of consistency in the findings of linkage studies
may be attributed in part to ascertainment and methodological differences across
studies (Prathikanti and McMahon, 2001) suggests that standardization of defini-
tions, analyses, and procedures may enhance the comparability across studies. To
the extent that integration of previous studies is methodologically feasible, pooling
of prior linkage studies of bipolar disorders would increase the power of the
aggregate data to detect linkage.


Psychiatric disorder phenotypes
There is widespread agreement regarding the limitations in applying the current
nomenclature for mental disorders to biologic studies. Psychiatric disorder pheno-
types, based solely on clinical manifestations without pathogenomic markers, still
lack conclusive evidence for validity of classification and reliability of measure-
ment (Kendell, 1989). The lack of specificity of biologic and psychosocial risk
factors and correlates, as well as the lack of longitudinal stability, still suggest
etiologic and phenotypic heterogeneity.
Advances in neuroscience and genetics leading to enhanced understanding of the
structure and function of the human brain, and the role of genetic and environmental
factors involved in cognition, emotion, and behavior will likely have a major impact
on classification. Likewise, advances in our ability to measure and observe the major
components of behavior will also facilitate identification of etiologic mechanisms.


Lack of direct correspondence between the genotype and phenotype
Despite these advances in characterizing human genotypes, application of this
knowledge to human diseases is still limited by the complexity of the process
through which genes exert their influence. A lack of one-to-one correspondence
between the genotype and phenotype is clearly the rule rather than the exception
297 Challenges in the genetics of bipolar disorder


for most human disorders. Phenomena such as penetrance (probability of pheno-
typic expression among individuals with susceptibility gene), variable expres-
sivity (degree to which susceptible individuals express components of genotype),
gene“environment interaction (expression of genotype only in the presence of
particular environmental exposures), pleiotropy (capacity of a gene to manifest
simultaneously several different phenotypes), and genetic heterogeneity (different
genes leading to indistinguishable phenotypes) have been demonstrated for several
human disorders for which susceptibility genes have been identified.
Two of these phenomena that have been of particular concern for genetic
nosology are genetic heterogeneity, or one from many, and pleiotropy, or many
from one. These two situations are reflected in the nosologic tension between
lumping and splitting, as described by Victor McKusick in his review of historical
developments in genetic nosology (1973). With increasing specialization in med-
icine, there has been a tendency to split categories excessively. The lumpers have
corrected the oversplitting; but recent advances in genetics have led to a new wave
of ˜˜better-founded splitting™™ based on an increased ability to detect subtler
phenotypic similarity but genotypically heterogeneous conditions.


Gene“environment interaction
Gene“environment interaction characterizes a broad range of human diseases such
as cancer and birth defects. Not only is the expression of genes modified by the
environment, but there is now also substantial evidence to indicate that numerous
environmental factors may actually alter the genotype. Francis and colleagues (1999)
have shown that maternal behavior mediates stress reactivity in adulthood and is
associated with future maternal behavior among offspring. Genes may also be
involved in the response or resistance to purely environmental agents such as diet,
stress, exercise, drugs, and nutritional deficiencies (Omenn and Motulsky, 1978).
The methods of genetic epidemiology are designed specifically to identify gene“
environment interactions (Ottman, 1995; Yang and Khoury, 1997).
The lack of validity of diagnostic categories is therefore by no means unique to
psychiatry. Similar to other domains of complex disorders, the major impedi-
ments to the establishment of validity of the classification of psychiatric disorders
are: the unreliability of measurement (of both diagnoses and markers); the lack of
specificity of risk factors and biologic markers; and the lack of one-to-one
correspondence between the phenotype and genotype likely attributable to both
etiologic and phenotypic heterogeneity and gene“environment interaction. The
well-known steps for validating phenotypes recently reiterated by Tsuang et al.
(1993) include the following guidelines: specificity, state of independence, herit-
ability, familial association, co-segregation, and biological and clinical plausibility.
298 K. Merikangas and K. Yu


Although there have been numerous controlled family studies of psychiatric
disorders, there are very few that were designed to investigate the specificity of
core components of the phenotypes. Those that have examined subtypes of
psychiatric disorders have not provided sufficient evidence for familial specificity.
The experienced investigators in this field tended to abandon this research because
of the advent of opportunities to identify genes in the early 1990s. This was
unfortunate because this was also the time when there had been numerous
advances in methods and analyses for family studies (Weissman et al., 1986b),
including population-based ascertainment of common disorders and appropriate
selection of controls (Kendler, 1990; Klein, 1993; Hill and Neiswanger, 1997) and
random-effects regression models that incorporate familial clustering.
The recent shift in psychiatric genetics to identify endophenotypes, or underlying
biologic factors, that explain familial recurrence is an important step in moving
from broad phenotypes to specific components of disorders. Recent advances in
neuroscience and the behavioral sciences, not available to the pioneers in family
study research in psychiatry, will be important tools in enhancing this process.
Substantial effort should be devoted to the application of genetic epidemiologic
studies that are designed to define more homogeneous components of mood
disorders and associated biologic markers that may yield higher familial relative
risk than the heterogeneous category of major depression. Ironically, however,
genetic mapping strategies may also assist in defining subtypes (Reus and Freimer,
1997). Nevertheless, lessons from other disorders have demonstrated that, even after
the identification of the gene for single-gene disorders, the classification still requires
additional testing to identify sources of heterogeneity in phenotypic expression.
For example, despite the identification of the actual gene for Marfan™s syndrome, the
checklist criteria appear to be remarkably similar to those within the realm of
DSM-IV. Likewise, recent studies of neurofibromatosis have examined familial
specificity of diverse clinical manifestations of the same genetic mutations.
Two directions of research that are particularly promising for informing pheno-
typic validity include: (1) genetic epidemiologic strategies and prospective longi-
tudinal studies for phenotype refinement; and (2) studies designed to identify
endophenotypes of mood disorders, informed by advances in neuroscience.


Future directions
The tremendous progress in molecular biology, neuroscience, and related fields is
likely to lead to new approaches to the genetics of psychiatric disorders of human
diseases. Identification of all of the genes in the human genome and their func-
tional variation will provide opportunities for studying the impact of those
variants on phenotypic outcomes of interest. Studies of gene regulation,
299 Challenges in the genetics of bipolar disorder


microarray techniques that can detect tissue-specific expression, transgenic mice,
expression arrays, and proteomics, allowing identification of gene function, will
enhance our knowledge of the biologic significance of gene markers. Parallel
research across numerous species will provide important information regarding
gene expression, phenotypic models, and gene“environment interaction (Peltonen
and McKusick, 2001). Knowledge of gene function and regulation is likely to lead to
a shift from reverse-genetic approaches (linkage and linkage disequilibrium analysis)
to forward-genetic approaches (Risch, 2000).
Similar progress in neuroscience, particularly developmental neuroscience, which
investigates molecular, cellular, and integrative brain functions involved in the devel-
opment of mental disorders, will advance our understanding of the complex biologic
processes underlying mental disorders. The tools of neuroimaging, psychophysiology,
and preclinical models of emotion are likely to provide information on etiologic
pathways to mental disorders. Likewise, the impact of environmental exposures on

<<

. 50
( 68 .)



>>