. 49
( 68 .)


selected probands from inpatient settings or treatment registries. Table 13.2
presents the twin studies of probands with bipolar disorder and MDD in which
there were at least 15 twin pairs. The average concordance for mood disorders
among monozygotic twins was 60% and 12% for dizygotic twins. There is a
fivefold greater rate of concordance for mood disorders among monozygotic
than dizygotic twins, thereby indicating the importance of the role of genetic
factors in the familial aggregation of bipolar disorder.
Twin studies of mood disorders reveal that genetic factors have a far greater
etiologic role in bipolar disorder than in MDD. The relative risks comparing
monozygotic and dizygotic twins in two studies were 1.9 (McGuffin and Katz,
1989) and 1.2 (Kendler et al., 1992). Nevertheless, the application of quantitative
models that estimate the relative components of the variance attributable to shared
genes, common environment, or unique non-shared environment yielded signifi-
cant degrees of heritability in both studies (i.e., 0.39 in the former and 0.84 in the
286 K. Merikangas and K. Yu

latter; McGuffin and Katz, 1989; Kendler et al., 1992). Differences in the results of
the two studies could be attributable to difference in sampling (i.e., hospitalized
patients in the former study and inclusion of only females from the general
population in the latter), or to other methodologic differences.
In addition, Tsuang and Faraone (1990) estimated that the heritability of
bipolar depression is 0.59, and a recent meta-analysis of community-based twin
studies of MDD yielded a substantially lower heritability estimate of 0.37 (95%
confidence interval ¼ 0.28“0.42; Sullivan et al., 2000). This latter estimate also
indicates that nearly two-thirds of the liability to MDD cannot be attributed to
genes. The concordance rates were somewhat lower for males than for females: the
range for males was 0.23“0.41 for monozygotic and 0.14“0.34 for dizygotic and the
range for females was 0.47“0.67 for monozygotic and 0.32“0.43 for dizygotic pairs.
Early studies of the specificity of transmission of polarity in twin studies were
reviewed by Zerbin-Rudin (1969). The largest twin study which systematically
investigated differences in concordance among bipolar and MDD twins was
presented by Bertelsen and colleagues in 1977. Studies which examined the con-
cordance rates among twins by polarity support a strong degree of specificity of
transmission of the two subtypes of mood disorders, with little cross-transmission
between bipolar index twins with MDD co-twins, and the converse. The average
relative risk for cross-transmission for probands with either MDD or bipolar
disorder was 1.5. In contrast, bipolar disorder was found to exhibit a strong degree
of specificity, with an eightfold greater risk of bipolar disorder among the co-twins
of bipolar monozygotic probands compared to their dizygotic counterparts.
The major conclusion that can be drawn from the current evidence from twin
studies is that mood disorders are strongly heritable, with bipolar disorder exhi-
biting a much greater degree of involvement of genetic factors in its etiology than
MDD. Moreover, there is little evidence for the cross-transmission of the two
subtypes of mood disorder. Faraone et al. (1987) calculated the aggregate variance
components from the twin studies of MDD then available and found a significant
degree of heritability (i.e., 0.51), a significant contribution of the common envir-
onment of the twins (variance 0.42), and nearly no effect of the unique environ-
ment in the development of mood disorders. Twin studies of milder mood
disorders are difficult to interpret because of differences in diagnostic definitions
and inconsistent application of the criterion of hospitalization for affected status
(Stenstedt, 1966; Shapiro, 1970; McGuffin and Katz, 1989).

Adoption studies of the mood disorders
Adoption studies are the most powerful design to test the relative contributions of
genetic and environmental factors to the etiology of the mood disorders. There are
287 Challenges in the genetics of bipolar disorder

very few adoption studies of mood disorders, and those that are cited were
typically quite small and conducted more than 20 years ago. The chief impediment
to adoption studies of mood disorders is the lack of valid information on depres-
sion in biologic parents, particularly the biologic father.
The aggregate adoption study data on mood disorders reveal a moderate
increase in rates of mood disorders among the biologic compared to adoptive
relatives of adoptees with mood disorders (Tsuang and Faraone, 1990). With
respect to bipolar disorder, there is little evidence for differential risk among
biologic compared to adoptive relatives of adoptees with bipolar disorder.
However, the small numbers of bipolar adoptees who have been studied (i.e.,
fewer than 50) do not provide an adequate test of genetic and environmental
influences (Goodwin and Jamison, 1990). The most compelling finding from
adoption studies, however, is the dramatic increase in completed suicide among
biological relatives of mood-disorder probands (Mendlewicz and Rainer, 1977;
Wender et al., 1986).

Genetic epidemiology of mood disorders in youth

Family studies
Despite the abundance of well-controlled family and genetic studies that have
employed sophisticated methodology to investigate the transmission of mood
disorders among adults, there are only a limited number of controlled family
studies that have focused on the manifestation of mood disorders among adoles-
cents (Merikangas and Swendsen, 1997).
Controlled family studies of adult relatives of children with depression as well as
offspring of adults with depression provide consistent evidence that MDD has a
strong familial component (Weissman, 1987; Neuman et al., 1997; Beardslee et al.,
1998; Kovacs and Devlin, 1998). Children of depressed parents are three times
more likely to have an episode of MDD than children whose parents are not
depressed (Birmaher et al., 1996) and are four times more likely to develop mood
disorders (Lavoie and Hodgins, 1994). By the age of 25, children of affectively ill
parents have a 60% chance of developing MDD (Beardslee et al., 1993). Risk to
children is even greater when both parents exhibit mood disorders (Merikangas
et al., 1988). Studies of parents of children with MDD reveal a strong association
between child and parent MDD (Puig-Antich et al., 1989; Williamson et al., 1995).
Although several studies suggest that early age of onset is associated with increased
familial aggregation of depression, the results of recent family and twin studies of
youth conclude that prepubertal depression is less heritable than postpubertal
depression (Harrington et al., 1997; Silberg et al., 1999).
288 K. Merikangas and K. Yu

Controlled studies of offspring of parents with bipolar disorder exhibit a wide
variation in the frequency of mood disorders among offspring of affected parents
(a range of 23“92%; Hammen et al., 1990; Radke-Yarrow et al., 1992), but
collectively suggest a familial component. Rates of mania and bipolar disorder
are generally low due to the young age of adolescent offspring in these studies;
however, children of bipolar parents show greater specificity of transmission of
mood disorders than do children of parents with unipolar depression (Merikangas
and Angst, 1995).
Although these studies provide evidence of familial influence in the etiology of
mood disorders, they shed little light on possible mechanisms through which such
factors may operate to produce affective psychopathology in children. Familial
aggregation of depression may result from shared genes, common environmental
factors, or a combination thereof.

Twin studies
The role of genetic factors underlying the familial aggregation of depression has been
investigated by several twin studies of depressive symptoms and disorders among
youth. Reports by Thapar and McGuffin (1994, 1995, 1997), Murray and Sines
(1996), Eley and Plomin (1997), and O™Connor et al. (1998) conclude that there is a
modest degree of genetic influence for childhood depressive symptoms, with greater
heritability with age. However, some of these studies suggest that the age-related
increase in heritability is limited to males, whereas the influence of the shared
environment tends to increase with age in females (Eley and Stevenson, 1999).
Genetic factors may also play a role in the recurrence and stability of depression.
The twin study of O™Connor et al. (1998) found that the stability of depressive
symptoms over a 3-year period was primarily explained by genetic influence;
Silberg et al. (1999) found similar results among girls but not boys. Although a
strong genetic component is implicated, O™Connor et al. (1998) warn that it is
premature to accept the conclusion that the identified influence of heritability is,
in fact, purely genetic, given that recent reports suggest strong and pervasive
gene“environment correlations (e.g., evocation of stressful events based upon a
genetic predisposition). Consistent with this, Silberg et al. (1999) found that
individuals who inherited a genetic predisposition for depression also inherited
a tendency to experience negative life events. Finally, adoption studies of depres-
sion symptoms in children and adolescent found only negligible genetic influence
(van den Oord et al., 1994; Eley et al., 1998).
Although evidence suggests that the vulnerability for depression may be inher-
ited, the environmental stressors have also been implicated in the development of
289 Challenges in the genetics of bipolar disorder

mood disorders in youth (Warner et al., 1995; Goldsmith et al., 1997). Indeed, the
family environment of depressed adults is consistently characterized by family and
parental discord, divorce, inattention, rejection, and abuse (Angold, 1988;
Downey and Coyne, 1990; Rutter, 1989). Parker (1979) found that a parental
discipline pattern of affectionless control was strongly associated with depressive
disorders in adolescents; studies of bipolar depressives reveal normal parental
levels on these dimensions (Parker, 1979). Community studies have documented
associations between family dysfunction and depression in children and adoles-
cents (Kandel and Davies, 1982; Garrison et al., 1985; Bird et al., 1988). These
associations appear to be a reciprocal relationship between parental depression
and child maladjustment (Downey and Coyne, 1990).

There is great variability in the estimates of the initial age of onset of depression.
Based on retrospective recall of the onset of depression among adults, the onset of
depression has been previously estimated to occur in the late 20s and early 30s.
However, the results of recent prospective studies reveal that depression often
occurs in childhood. Prior to the recent generation of studies of children and
adolescents, estimates of the age of onset of depression were derived from retro-
spective studies of adults with depression (Angst, 1988) and suggested mid to late
adolescence as the most common age of onset of first episode of MDD (Burke
et al., 1990; Hammen and Rudolph, 1996; Lewinsohn et al., 1998), although the
NCS suggests an average age of onset during early adulthood (24 years for men and
23.5 years for women; Kessler et al., 1993).

Genetic marker studies of mood disorders

Association studies of mood disorders
Association studies investigate the relationship between disease status and a
particular marker or allele across families and individuals. Most association
studies employ the traditional case-control design in which the prevalence of a
putative disease marker is compared among persons with a disorder to persons
without the disorder. The most common methodologic error in association
studies is the lack of equivalence between the cases and controls on factors
which may confound the association between the purported marker and disease.
After exclusion of spurious associations due to methodologic factors or popula-
tion stratification, associations between a disease and a marker could be attributed
to either linkage disequilibrium between genes for the disease and for the marker,
or the effect of a single gene that encodes both the marker and the disease.
290 K. Merikangas and K. Yu

In genetic case-control studies, the most likely source of confounding is ethni-
city because of differential gene and disease frequencies in different ethnic sub-
groups. Aside from confounding, association studies are particularly prone to
false-positive findings due to multiple testing without correction and the low prior
probability of a gene“disease association (Wacholder et al., 2000). In addition,
there is a strong publication bias against reports of negative association studies
(Risch and Merikangas, 1996). The latter problem can be resolved in part by the
use of much higher -levels (i.e., false-positive error rates) in association studies
(Hirschhorn et al., 2002).
The loci for several biochemical parameters that are suspected to be involved
in either etiology or outcome of the psychiatric disorders have been identified. It is
important to note that many of these assignments are based upon a single study,
and replication is clearly necessary. Identification of new loci is occurring at such
a rapid rate that it is necessary to update the human map monthly. Application of
this methodology to psychiatric disorders may be particularly fruitful in identify-
ing major genes that are segregating in informative families.

Review of empirical evidence
Great emphasis has been placed on the association study of various affective
spectrum disorders, focusing mostly on polymorphisms. From these studies, no
functional differences between the alleles have been described (Johansson et al.,
2001). The most noted exception is an insertion/deletion polymorphism located in
the promoter region of the serotonin transporter gene (5-HTTLPR), reported to
affect the expression of the transporter (Lesch et al., 1996), but there are also non-
replications of these results (Rees et al., 1997; Ohara et al., 1998a; Frisch et al., 1999;
Seretti et al., 1999). The 5-HTTLPR polymorphism has also been associated with
clinical subtypes of depression, including SAD and seasonality (Rosenthal et al.,
1998; Sher et al., 1999), but the findings have not been able to be replicated in two
subsequent studies (Johansson et al., 2001). Other polymorphisms include an
amino acid substitution in catechol-O-methyltransferase (COMT) affecting the
enzymatic activity (Lachman et al., 1996a, b), which has been linked to bipolar
disorder, including the rapid-cycling bipolar disorder subtype, and unipolar
depression (Li et al., 1997; Kirov et al., 1998; Ohara et al., 1998b). A repeat
polymorphism in the promoter region of a monoamine oxidase A (MAOA)
gene has also been reported to influence the transcriptional activity (Kunugi
et al., 1999). Unfortunately, conflicting results have been found for both poly-
morphisms (Biomed European Bipolar Collaborative Group, 1997; Frisch et al.,
1999; Kunugi et al., 1999; Schulze et al., 2000).
291 Challenges in the genetics of bipolar disorder

Linkage studies of mood disorders
Linkage is based on the principle that two genes that lie in close proximity on a
chromosome are transmitted to their progeny together. However, if the loci are far
apart, crossing over between the maternal and paternal chromosomes may take
place during meiosis, thereby producing new combinations of alleles. The farther
apart the loci, the greater the probability that crossing over will occur and that the
offspring may inherit a recombinant of the two parental chromosomes. Cross-
overs can be detected by inspecting the maternal and paternal genome; when a
particular chromosome is not identical to the parental chromosome, a cross-over
or recombination between the maternal and paternal chromosomes has occurred.
Linkage studies differ from association studies in that linkage is based on an
association between genetic markers and putative disease genes within families,
whereas association is the co-occurrence of a marker and disease at the level of the
general population. Linkage does not imply that the adjacent gene is etiologically
related to the disease; only that it can be used to track possible genes in families.
Therefore, one allele at a particular locus may be linked to a disease in some
families, whereas the other allele may co-segregate with the same disease in other
families. In contrast, associations are detected in case-control studies that compare
the prevalence of a marker in patients with a particular illness with the proportion
of control subjects who possess the marker. Thus, an association found in patient
samples may not extend to their families.
Two major methods of genetic linkage analysis are the LOD score method
(Morton, 1955) and the affected sib-pair method, derived from Penrose (1935).
The LOD score is defined as the ratio of the log odds of the likelihood of a linkage
between two loci within a pedigree to that of the likelihood of independent
segregation of the two loci, or a recombination frequency of 1/2. A LOD
score > þ3 represents a probability of 1/1000 of falsely concluding that linkage
exists when it is absent, and a LOD score < À2 indicates significant evidence for a
lack of linkage between the putative marker and disease. Scientific evidence for
acceptance of linkage between a disease and genetic marker was described by Risch
(1990), who stated that in addition to a LOD score > 3, a linkage finding should be
replicated in a different sample in a different laboratory.
The affected sib-pair method examines the sharing of marker alleles at a locus
among affected sib pairs. The null hypothesis of no linkage specifies probabilities
of 1/4, 1/2, and 1/4 for sharing 2, 1, and 0 marker alleles among affected sibs. Excess


. 49
( 68 .)