. 48
( 68 .)


affected, similar etiologic factors can be assumed, and variable forms of expression
can be identified. This property of within-family studies should be far more widely
applied in other domains of research designed to elucidate etiologic factors.
Although standardized methods for conducting family studies have been well
established, there are still remarkably few family studies that employ these study
principles, including (Weissman et al., 1986b):
(1) recruitment of a well-characterized homogeneous group of probands with a
particular disorder
(2) selection of a control group of individuals who are comparable to the affected
probands on all possible confounding factors except the disorder itself
(3) systematic enumeration of all living and deceased relatives according to the
degree of relationship to the probands and controls
(4) use of structured diagnostic interviews with relatives using predetermined
diagnostic criteria and reliable and valid diagnostic instruments
(5) maintenance of blindness with respect to the diagnostic status of the proband
in collecting diagnostic information from the relatives and formulating the
final diagnostic estimates
(6) collection of information regarding relatives unavailable for interview in a
standardized format from as many informants as possible
(7) inclusion of ancillary information to supplement interview and family history
(8) development of reliable procedures to integrate material from direct inter-
views, family history reports, and ancillary medical or psychiatric information
in deriving the diagnostic assignment of the probands and relatives
(9) application of sophisticated statistical techniques to control for confounding
variables and simultaneously adjust for length of observation of the relatives.
Recent developments in biostatistics now permit simultaneous estimation of
relative risk and familial clustering (Hedeker and Gibbons, 1996)

Review of empirical evidence
Although numerous family studies of both bipolar and MDD have been conducted
during the past 30 years, it is remarkable that only four family studies of bipolar
disorder and five studies of MDD meet the above-cited standards of family study
280 K. Merikangas and K. Yu

methodology including inclusion of control probands and relatives, application of
standardized diagnostic criteria with structured diagnostic instruments, and
blindness with respect to the diagnosis of the probands.
Table 13.1 presents a summary of controlled family studies of probands with
bipolar disorder and MDD. The weighted average risk ratio (comparing the
prevalence of mood disorders among relatives of cases compared to those of
controls) for bipolar disorder among relatives of bipolar probands is 10.3, whereas
the average risk ratio of MDD among relatives of bipolar probands compared to
those of controls is 3.2. This indicates a very high magnitude of familial aggrega-
tion of bipolar disorder, similar to that found for many of the major diseases for
which the genetic basis has been identified. In contrast, the average risk ratio for
MDD among relatives of probands with MDD compared to those of controls was
3.6, indicating only a moderate influence of familial aggregation on non-bipolar
mood disorders. Although sex, age, birth cohort, and socioeconomic status are
differentially associated with MDD (greater risk among females, later birth
cohorts, lower socioeconomic status), there is no evidence that they moderate
the familial recurrence risk of mood disorders (Weissman et al., 1986a; Tsuang and
Faraone, 1990).
Studies of offspring of mood-disorder probands have contributed substantial
information on developmental influences in the expression of mood disorders
among youth. The high-risk studies on bipolar disorder have revealed an increased
risk of mood disorders among offspring of parents with bipolar illness (Decina
et al., 1983; Klein et al., 1986; Radke-Yarrow, 1998). There is now a growing body
of research of the developmental manifestations of bipolar disorder, and several
high-risk studies of bipolar disorder are under way.
In contrast to bipolar disorder, there are far more high-risk studies of MDD,
some of which have now been extended to three generations (Weissman et al.,
1997; Warner et al., 1999). Although there is a consistent association between
mood disorders among parents and offspring (Downey and Coyne, 1990), nearly
all show a lack of familial specificity of mood disorders since the prevalence of
anxiety and behavioral disorders are often equally elevated (Weissman et al.,
1984a; Merikangas and Angst, 1995; Harrington et al., 1997; Rende et al., 1999).
A review of comorbidity of anxiety and depression by Brady and Kendall (1992)
suggests that anxiety and depression may be part of a developmental sequence in
which anxiety is expressed earlier in life than depression.
With respect to subtypes of bipolar disorder, studies of the familial specificity of
the bipolar I“II distinction have yielded inconclusive results. Whereas some
studies suggest a continuum from bipolar I to bipolar II to major depression
based on familial overlap (Gershon et al., 1986; McMahon et al., 1995), others
demonstrate that bipolar II disorder constitutes an independent mood-disorder
Table 13.1 Proportion of first-degree relatives of bipolar (BP) and major depressive disorder probands with a lifetime history of bipolar and major
depressive disorder

Rates/100 subtype in relatives by proband subtype

Study Probands (n) Relatives (n) BP“BP BP“UP UP“BP UP“UP Control“BP Control“UP

Gershon et al. (1975) Bipolar 54 411 19.6 11.5 11.0 20.1 0.2 0.5
Unipolar 16 113
Controls 75 619
Gershon et al. (1982) Bipolar 130 739 21.9 2.6 8.0 3.0 0.4 5.7
Unipolar 30 166
Controls 43 265
Weissman et al. (1984b) Unipolar 133 810 1.5 3.2 1.5 4.1
Controls 82 591
Tsuang et al. (1985) Bipolar 100 230 21.2 1.9 11.9 2.3 0.2 4.8
Unipolar 225 500
Controls 160 543
Heun and Maier (1993) Bipolar 80 504 7.1 17.1 1.3 19.0 0.9 7.7
Unipolar 108 306
Controls 80 221
Total rates/100 6.2 12.2 2.2 13.6 0.6 3.8
Relative risk (versus control) 10.3 3.2 3.5 3.6

UP, unipolar.
Ratio weighted for sample size.
282 K. Merikangas and K. Yu

subtype (Heun and Maier, 1993; Bauer and Dunner, 1996). The lack of adequate
coverage of cases identified in community and primary care settings by the current
diagnostic criteria for depression, as well as the low longitudinal stability of
specific depressive subtypes, has led to a re-evaluation of the classification of
mood disorders (Angst and Merikangas, 1998; Angst, 1999).
Angst and Merikangas (1998) have argued for a more phenomenologic
approach to the classification of depression, based on systematic evaluation of
the thresholds for individual symptoms, specific symptoms included in the diag-
nostic categories, as well as the frequency and duration criteria. A series of
subthreshold definitions of depression and anxiety was developed by Angst et al.
based on a 25-year prospective follow-up of a community sample of a cohort of
young adults from Zurich, Switzerland (1984). Numerous other investigators have
proposed expanding the criteria for mood and anxiety disorders, primarily based
on longitudinal follow-up of clinical and community samples of adults (Judd et al.,
1994) and children (Kovacs and Beck, 1977; Lewinsohn et al., 1999). Several
investigators suggest that a substantial proportion of those with recurrent depres-
sion may actually have a form of bipolar disorder (Pages and Dunner, 1997).
Although the concept of atypical depression has been widely recognized, there
is still no consensus regarding the specific diagnostic criteria or its clinical sig-
nificance. According to Diagnostic and Statistical Manual of Mental Disorders, 4th
edn (DSM-IV), the essential features of the atypical specifier are mood reactivity,
increased appetite or weight gain, hypersomnia, leaden paralysis, and a long-
standing pattern of extreme sensitivity to perceived interpersonal rejection
(Alpert et al., 1997; Merikangas et al., 2002a). Atypical depression is discriminated
from non-atypical depression on the basis of differential clinical features including
an earlier onset, greater chronicity, and more severe symptoms than other depres-
sive subtypes (Angst et al., 2002). Other discriminating factors include differential
association with bipolar II disorder, compared to other depressive subtypes
(Benazzi, 2000), differential treatment response (Stewart et al., 2002), specific
patterns of neuroendocrine challenge response (Schmidt et al., 1997), different
patterns of comorbidity, including stronger associations with social phobia and
migraine, and some degree of specificity in twin (Kendler et al., 1996) and family
studies (Dunner, 1980; Heun and Maier, 1993; Merikangas and Avenevoli, 2002).
Although the validity of expression of atypical features in youth is still under
investigation, studies of depression in children and adolescents confirm that
atypical features do occur in youth as well (Williamson et al., 2000; Angst et al.,
2002). Although there is substantial research on this subtype, to our knowledge
there are no large-scale family studies of atypical depression in the context of the
range of expression of both bipolar and unipolar disorder, as well as anxiety
283 Challenges in the genetics of bipolar disorder

Rapid cycling is defined as the occurrence of four or more discrete episodes of
mood disorders within a 12-month period, with each episode meeting both
duration and symptom criteria for major depressive, manic, mixed, or hypomanic
episodes. Rapid cycling occurs in approximately 5“15% of those with bipolar
disorder, with a higher prevalence in women. Although not widely studied, there
is little evidence for specificity of the familial aggregation of rapid cycling (Post
et al., 1984, 1986; Goodwin and Jamison, 1990). Existing family history data
support a continuum notion in the familial aggregation of bipolar disorder
(Nurnberger et al., 1988; Wehr et al., 1988; Goodwin and Jamison, 1990).
Seasonal affective disorder (SAD) applies explicitly to the major depressive
episode in bipolar I, bipolar II, or MDD at characteristic times of the year,
occurring more commonly in females. Most episodes initiate during the winter
or fall and remit in the spring. They are characterized by hypersomnia, hyper-
phagia, anergia, carbohydrate craving and weight gain, symptoms more com-
monly considered atypical in unipolar depression but common in bipolar
patients. The atypical symptoms occur in 65“85% of study subjects with seasonal
mood disorders, compared to 15“30% of patients unselected for seasonality. Since
50“60% of patients with winter depression have first-degree relatives with MDD,
this form of SAD is likely to be a manifestation of the general spectrum of major
affective disorders that includes manic-depressive illness (Goodwin and Jamison,
1990). Family history data on seasonal depression do not appear to support this
subtype as an independent subtype of major depression (Allen et al., 1993).

Factors associated with familial transmission of mood disorders
The transmission of mood disorders may vary according to polarity, as described
above, degree of relationship to the proband, age of onset of the proband, and sex
of the proband and relative. Many of the above-cited family studies have explored
the relationship between these proband characteristics and those in their relatives.

Relationship to the proband
Numerous studies have presented the rates of mood disorder among both the first-
and second-degree relatives of mood-disorder probands (Gershon, 1989).
According to expectations of traditional genetic models, risks to all classes of
first-degree relatives should be equal for dominant traits, whereas siblings should
have increased rates of disorders for recessive traits. The aggregate data for bipolar
disorder reveal that the risk of bipolar disorder among parents and siblings is
approximately equal for both bipolar disorder and MDD. However, the offspring
tend to have elevated rates of bipolar disorder and equal rates of MDD when
compared to parents and siblings. In the absence of control data, it is difficult to
284 K. Merikangas and K. Yu

interpret the latter finding. However, the elevation in the risk among offspring
could result from comorbidity, recall bias, a cohort effect, or assortative mating in
the parental generation, or genetic anticipation (Tsuang and Faraone, 1990).
Most genetic models described above predict a decrement in risk of disease
according to the degree of relationship to the affected proband. Although a large
number of studies have reported rates of mood disorders among both the first-
and second-degree relatives, there is only a single controlled study in which the
rates of mood disorder were compared between the first- and second-degree
relatives with controls (Gershon et al., 1982). The results of this study revealed
that the rates of bipolar disorder among first-degree relatives of bipolar probands
were approximately twice those of second-degree relatives, which in turn were
greater than those of controls. In contrast, there was no elevation in the rates of
either MDD or bipolar disorder among the second-degree relatives of MDD
probands, nor was there an increased rate of MDD among the second-degree
relatives of bipolar probands.

Age of onset
The effect of age of onset on the familial aggregation of mood disorders was first
described by Stenstedt (1952) and has subsequently been confirmed in several
studies. However, there are numerous possible confounding factors that have not
been adequately addressed in these studies, including: recurrence, comorbidity,
biased recall, and personality factors. Moreover, the conclusions of the above-cited
studies have been based on a dichotomous classification of the age of onset of
probands, rather than significant correlations between the age of onset of pro-
bands and relatives.

Sex of proband
The effect of the sex of the proband and relative has been systematically investi-
gated for both bipolar disorder and MDD. In general, there is little deviation in
family study data from the sex ratio for bipolar and MDD reported in epidemio-
logic studies. In general, the rates of bipolar disorder are nearly equal in male and
female relatives, whereas there is a female preponderance of mood disorders
among the relatives with MDD. However, the transmission of both bipolar dis-
order and MDD has been shown to be unrelated to the sex of the proband, with
equal rates of mood disorders among the relatives of male and female bipolar
disorder and MDD probands (Merikangas et al., 1985; Faraone et al., 1987).
In summary, the family studies of bipolar disorder and MDD demonstrate a
strong degree of familial aggregation of both of these subtypes of mood disorders.
However, the evidence is inconclusive regarding the role of shared underlying
factors in the expression of these subtypes of mood disorders. The transmission of
285 Challenges in the genetics of bipolar disorder

Table 13.2 Twin studies a of mood disorders

Monozygotic Dizygotic

n Concordance (%) n Concordance (%) Relative risk

Bipolar disorder
Rosanoff et al. (1935) 23 70 67 16 4.4
Kallman (1953) 27 93 55 24 3.9
Harvald and Hauge (1965) 15 67 40 5 13.4
Allen et al. (1974) 15 33 34 0
Bertelsen et al. (1977) 55 51 11 14 3.6
Major depression disorders
McGuffin et al. (1991) b 62 53 79 28 1.9
b, c
Kendler et al. (1992) 590 48 440 42 1.2

Studies with !15 twin pairs.
Diagnostic and Statistical Manual of Mental Disorders, 3rd edn revised (DSM-IIIR) criteria.
Females only.

mood disorders appears to be associated with an early age of onset of mood
disorders in probands, the bipolar subtype, and recurrent episodes, but not with
the sex of the affected proband.

Twin studies of mood disorders
There have been numerous studies which compare the rates of mood disorders
among monozygotic and dizygotic twins. The majority of the earlier studies


. 48
( 68 .)