Journal of Affective Disorders 85 (2005) 135 – 145
www.elsevier.com/locate/jad
Research report
The cyclothymic temperament in healthy controls and familially
at risk individuals for mood disorder: endophenotype
for genetic studies?
Pierre Chiaroni a,*, E.-G. Hantouche b, J. Gouvernet c, J.-M. Azorin a, H.S. Akiskal d
a
Service de Psychiatrie, Hôpital Sainte Marguerite, Bd. Sainte-Marguerite, 13009 Marseille, France
b
Université Paris VI, Hôpital Pitié-Salpétrière, Paris, France
c
Service de l’Information Médicale, Hôpital de la Timone, Marseille, France
d
International Mood Center, University of California at San Diego, San Diego, USA
Received 12 March 2003; accepted 18 December 2003
Abstract
Background: The modern concept of affective disorders focuses increasingly on the study of subthreshold conditions on the
border of manic or depressive episodes. Indeed, a spectrum of affective conditions spanning from temperament to clinical
episodes has been proposed by the senior author. As bipolar disorder is a familial illness, an examination of cyclothymic
temperament (CT) in controls and relatives of bipolar patients is of major relevance. Methods: We recruited a total sample of
177 healthy symptom-free volunteers. These controls were divided into three groups. The first one is comprised of 100 normal
subjects with a negative familial affective history (NFH); the second of 37 individuals, with positive affective family history
(PFH); and a third of 40 subjects, with at least one sib or first-degree kin with bipolar disorder type I according to the DSM-IV
(BPR). The last two groups defined at risk individuals. We interviewed all subjects with CT, as described by the senior author.
Results: We found a statistically significant difference in the rates of CT between the subjects in BPR versus others. CT was also
more prevalent in the PFH compared with NFH. Additionally, the simple numeration of the CT traits exhibited gradation in the
distribution of individuals inside the NFH, PFH and BPR. Finally, categorically defined CT and CT traits predominated in
females. Limitation and conclusion: Although not all relatives of bipolar probands were studied, our results exhibit an
aggregation of CT in families with affective disorder—and more specifically those with bipolar background. These results allow
us to propose the importance of including CT for phenotypic characterization of bipolar disorder. Furthermore, our results
support a spectrum concept of bipolar disorder, whereby CT is distributed in ascending order in the well-relatives of those with
depressive and bipolar disorders. We submit that this temperament represents a behavioral endophenotype, serving as a link
between molecular and behavioral genetics.
D 2004 Elsevier B.V. All rights reserved.
Keywords: Affective temperament; Cyclothymic temperament; Cyclothymia; Bipolar disorder; Psychiatric genetics
1. Introduction
* Corresponding author. Tel.: +33-4-9138-5568; fax: +33-49138-5085.
E-mail address: pierre.chiaroni@wanadoo.fr (P. Chiaroni).
0165-0327/$ - see front matter D 2004 Elsevier B.V. All rights reserved.
doi:10.1016/j.jad.2003.12.010
During the last three decades, modern bipolar
concepts of a broader nature developed under the
136
P. Chiaroni et al. / Journal of Affective Disorders 85 (2005) 135–145
influence of many authorities (Akiskal et al., 1983;
Angst et al., 1980; Goodwin and Jamison, 1990;
Klerman, 1981). Beyond the classical manic-depressive illness leading to the proposal of a broad bipolar
spectrum, which at present encompasses cyclothymia, hypomania, sporadic brief hypomania, recurrent
brief hypomania and hypomania associated to other
episodic disorders. In bipolar patients, the clinical
focus has clearly shifted from the study of isolated
episodes to attention to the entire lifetime, the
personality and, especially, the temperament which
is more and more part of this sub-affective spectrum
definition.
The temperament concept appeared with French
and German classical authors at the beginning of
this century, such as Kraepelin, Falret, and Baillarger (for review, see Brieger and Marneros, 1997;
Kretschmer, 1930; Schneider, 1973). More recently,
within this framework, the senior author (HAS)
operationalized four temperaments, namely hyperthymic, depressive, irritable and cyclothymic, with
empirical criteria to define them (Akiskal and Mallya, 1987). In general, temperament is considered as
the biological and genetic base of the individual
and, in particular, these different temperaments are
described as trait sub-syndromic conditions of the
affective pathology. They are likely to represent
liability factors in the subsequent development of
affective disorders (Akiskal, 1995).
At present, the term cyclothymia defines a subaffective disorder (Diagnostic and Statistical Manual
of Mental Disorders IV: American Psychiatric Association, 1995) as well as a temperament (Akiskal et
al., 1979; Kraepelin, 1921; Kretschmer, 1930). The
DSM and ICD classifications do not include a
cyclothymic pattern in the personality disorders. This
condition was removed from the rubric of personality
disorder of the DSM-II (American Psychiatric Association, 1968) and classified in chronic mood disorders in DSM-III (American Psychiatric Association,
1968). Epidemiological investigations showed its
prevalence in the general population to be from 3%
to 6%, which is substantial when compared to the
accepted 1% prevalence for manic depressive illness
(Depue et al., 1981; Placidi et al., 1998a, b). However, some authors argued that this condition is
heterogeneous and that its link with mood disorders
must be clarified (Angst, 1978; Schneider, 1973;
Wetzel et al., 1980). Some studies suggested a
genetic association between cyclothymia and bipolar
disorder (Akiskal et al., 1977; Depue et al., 1981;
Dunner et al., 1982; Klein et al., 1986; Klein et al.,
1988). Akiskal (1995) hypothesized that cyclothymia
could evolve into bipolar disorder and that the same
familial antecedents are found in subjects with cyclothymia as well as in patients suffering from manicdepressive illness.
This is of major relevance because the manicdepressive condition is supposed to be at least
partially of genetic etiology. Indeed cumulative data
from traditional pedigree, twin, adoption studies and
segregation analyses provided strong evidence that
manic-depressive illness runs in families (Kelsoe,
2003). Further, the closer the kinship between the
affected individuals and their relatives, the higher the
risk of becoming affected to the latter. These observations strongly suggest a genetic liability in the
etiology of bipolar disorders. However, the degree
and the kind of this genetic contribution remain
unknown until now, and no clear familial transmission patterns have emerged.
In psychiatric genetics, studies generally employ
either linkage or association analyses. Both are considered powerful and useful methods, although many
obvious methodological differences separate these
approaches. Briefly, the linkage approach, on the
one hand, is strongly dependent on a priori hypotheses, especially the mechanism of inheritance. On the
other hand, the association study is based on the
‘‘candidate gene’’ assumption, which postulates that
the studied gene potentially participates in the physiopathology of the disorder and that its locus is
assigned to a region in linkage with the condition.
In both cases, however, two methodological contingencies are necessary. The first one is the clear
definition of the phenotype; from this point of view,
many clinical criteria are used to delineate symptoms
and patterns of putative genetic relevance. The second
is the strict distinction between affected and nonaffected individuals, especially under suspected polygenic hypotheses underlying processes of susceptibility. Indeed, discerning mental health from mental
pathology, in terms of the absence or the existence
of morbid episodes, could prove insufficient. So, this
aspect is particularly important as many authors
consider that temperamental characteristics are sub-
P. Chiaroni et al. / Journal of Affective Disorders 85 (2005) 135–145
served by a genetic component and constitute a
putative early phenotypic condition (Akiskal et al.,
1983, 1985; Cassano et al., 1992; Cloninger et al.,
1993). But, the main argument against this concept is
the difficulty of linking the primacy of the temperamental characteristics to the illness (Hirschfeld et al.,
1983, 1989).
The aim of the present study is to compare the
rates of the cyclothymic temperament (CT) in ‘‘supernormal’’ individuals (i.e. free of symptoms and
without any familial history of affective disorders)
and in genetically at risk subjects (i.e. individuals
with affective disorders antecedents or with bipolar
patients in their kin). As this temperament is evaluated as an independent variable, a positive result
could constitute a new, indirect and strong argument
for the hypothesis of an association between mood
disorders and this temperament.
137
and/or had been specifically treated by psychotropic
drugs for affective disorder. We thereby excluded
eight subjects with a history of personal affective
disorder (n = 7) or a family history of psychotic
disorder (n = 1). We separated the remaining subjects
into two subgroups (n = 137; 61 men and 76 women;
mean age = 40.1 F 9.52 years, range = 22– 71; without
a personal affective history, psychotropic treatment or
psychiatric hospitalization)
– the first one with negative affective disorder family
history (NFH, n = 100; 48 men and 52 women;
mean age = 38.3 F 9.1 years; range = 23 –60),
– the other one with positive affective disorder family
history (PFH, n = 37; 13 men and 24 women; mean
age = 37.8 F 11.2 years; range = 22– 71).
Further diagnostic information was obtained, as far
as possible, from hospital records or from the physician who took charge of the patient.
2. Methods
2.1. Ethical considerations
This study received the approbation of the local
ethics committee [CCPPRB-Marseilles 1]. All participants were given a description of the aim and purpose
of the investigation, and written informed consents
were obtained from every subject included.
2.2. Populations selection
2.2.1. Controls
Healthy unrelated Caucasian volunteers, recruited
from the hospital staff in Marseilles, participated in
this study (n = 145). For each subject, we conducted a
semi-structured interview by the Schedules for Clinical Assessment in Neuropsychiatry (SCAN, PSE-10,
Wing et al., 1990) to detect a lifetime process of
psychotic or affective disorder. The clinician and
principal investigator of this study (PIS: Dr. P.C.) is
a senior experienced psychiatrist trained in such
interviewing. A pedigree was assessed by at least
two informative members of the family and limited
to the first and second degree (Thompson et al., 1979).
A psychiatric family history was considered positive
each time a subject’s relative was hospitalized in a
psychiatric service or consulted with a psychiatrist
2.2.2. The selection of bipolar patients’ relatives
We matched to these controls a comparison group
of individuals who had a positive family history of
bipolar disorder in first degree. We also interviewed in
the same manner 40 unrelated Caucasian subjects with
kin with bipolar patients who participated in another
genetic study (Chiaroni et al., 2000).
Briefly, 60 unrelated Caucasians participated in
this study. A consensus diagnosis for all was established by two experienced senior psychiatrists (including at all times the PIS and the physician who
was in charge of the patient). All were Bipolar, type
I, according to the Diagnostic and Statistical Manual
of Mental Disorders (American Psychiatric Association, 1995). Likewise, the following items, concerning the development and course were recorded:
age of illness onset (defined by the age of first
hospitalization), recurrent episodes, polarity, and
medications.
From these 60 families, we excluded the families
with one member and those who refused the interview
for personal or material reasons. Then, among the
families, we excluded the individuals with a diagnosis
of affective disorder. Finally, among the kin (n = 92),
only one relative in each family was assessed blindly.
So, 40 were assessed for the present study (15 males
and 25 females; mean age = 48.2 F 16.2 years; range =
138
P. Chiaroni et al. / Journal of Affective Disorders 85 (2005) 135–145
21 – 79). To date, none have developed an acute
episode of affective disorder.
significance was classically defined at the 0.05 level,
two-tailed.
2.3. Data collection
3. Results
In this study, the PIS interviewed all subjects,
demographic data were systematically recorded. After
a standardized explanation, each subject had to answer a questionnaire for the CT. The construction and
validation of this instrument is described in Akiskal et
al. (1979, 1998); Akiskal and Mallya (1987); Placidi
et al. (1998a).
This scale has eight items, scored by dichotomous
‘‘TRUE/FALSE’’ answers, exploring two clusters:
– cluster A: multiple subjective manifestations
including
1. lethargy and somatic discomfort versus eutonia,
2. dulling of senses versus keen perceptions,
3. slow-witted versus sharpened thinking,
4. shaky self-esteem alternating between low selfconfidence and overconfidence;
– cluster B: behavioral manifestations including
1. hypersomnia versus decreased need for sleep,
2. introverted self-absorption versus uninhibited
people seeking,
3. taciturn versus talkative,
4. unexplained tearfulness versus buoyant
jocularity.
A diagnosis of CT includes the following requirements: that these manifestations must be present since at least age 21, evolve in intermittent
cycles with rapid oscillations, and characterize the
habitual functioning of the individual. A defined CT
requires at least two items in cluster A and three in
cluster B.
2.4. Statistical procedures
Results were analyzed using descriptive statistics.
To compare groups for continuous variables, we used
parametric tests, when possible (Student’s t-test,
Fisher’s test and Pearson’s C v2)] or, conversely,
non-parametric tests (Mann – Withney, M – W, or
Kruskall– Wallis, K – W). For the qualitative variables, we used the v2 test or Fisher’s test. Statistical
3.1. Socio-demographic and clinical data
The demographic data are summarized in Table 1.
As regards gender, we observe a large prominence of
females in the total sample: from the 177 recruited
subjects, there are 101 women and 76 men. But this
sex distribution does not significantly differ between
the three subgroups (v2 = 2.45; P>0.29), even though
the rates of women are higher in the PFH and BPR
groups (62.5 and 64.8%, respectively) than in the
NFH (52%). Also, the mean ages in the different
subgroups are significantly different (K – W; P =
0.003): in comparison to the both controls groups,
the BPR are older, on an average. Further, the range of
ages indicates that a lot of people are in an age bracket
where they are still at risk for bipolar disorder, and
especially the bipolar type II, and could so be exposed
to the risk of developing an affective episode.
Concerning PFH subjects, when the diagnostic
distribution is checked in each category of affective
disorders, we find that 24.8% of the control sample
(n = 34) have only depressive family history, and only
2.9% (n = 4) have a bipolar disorder familial history.
Table 1
Socio-demographic characteristics of controls subgroups and
bipolar Patients’ relatives
NFH
Females
n
%
52
52
PFH
24
64.9
BPR
25
62.5
TOTAL
101
57.62
Males
n
48
13
15
76
%
48
35.1
37.5
42.93
n
100
37
40
177
%
100
100
100
100
Mean age [F] 38.3 [9.1] 37.8 [11.2] 48.2 [16.2]
Min
23
22
21
Max
60
71
79
Median
38
35
49
A non-significant predominance of females vs. males [v2 = 2.45;
P = 0.29], a statistically significant difference in mean ages [K – W;
P = 0.003].
139
P. Chiaroni et al. / Journal of Affective Disorders 85 (2005) 135–145
3.2. Cyclothymic temperament in controls versus
bipolar relatives
Table 3
Distribution of CT in NFH controls, PFH controls and BPR
3.2.1. Rates of cyclothymic temperament
In the study population, 11 subjects have a CT, as
strictly defined by the Akiskal and Mallya’s (1987)
criteria; data are presented in Table 2. Inter-group
comparison of the CT frequencies between the controls and the BPR exhibits a significant difference,
with respectively 3.6 versus 15% (v2 = 6.84;
P = 0.017). The calculated odd ratio is also statistically significant (OR = 4.66; confidence interval:
95%; 1.17 < OR <18.98). Further, when we examine
the controls group in more detail (or more closely) as
a function of the family history (Table 3), the
distribution of the CT frequencies is radically different in NFH controls (1%) and PFH controls (10.8%).
In this last group, when checking the typology of
affective disorder in the familial history, we notice
that all subjects have no more than recurrent depressive antecedents. This result actually indicates the
very low CT rate in the ‘‘super normal’’ controls,
who are not only free of affective symptoms but also
free of familial mood disorder antecedents. So, when
we compare the NFH versus the group that is at risk
for affective disorder (PFH and BPR), we notice a
significant difference (v2 = 10.72; P Fisher = 0.001)
with an OR equal to 14.78 (confidence interval:
95%; 1.88 < OR < 315). Conversely, a tendency towards the same rates of CT are observed in the PFH,
compared to the BPR and so, there is no statistical
difference between the two groups (v2 = 0.30 with P
Fisher = 0.74; and OR: 1.46; 0.32 < OR < 6.87). Cumulative data of the total sample of bipolar (controls
and relatives) exhibit a frequency of 6 CT subjects in
Non-CT
n
%
99
99
33
89.18
34
85
166
93.78
CT
n
%
Total n
Total %
1
1
100
100
4
10.81
37
100
6
15
40
100
11
6.21
177
NFH
Table 2
Rates of definite CT, showing a statistically significant difference
[v2 = 6.84; P = 0.017]
Controls
BPR
Non-CT
n
%
132
96.4
34
85
CT
n
%
n
5
3.64
137
6
15
40
PFH
BPR
Total
NFH vs. PFH vs. BPR, v2 = 11.30; P = 0.001; NFH vs. [PFH and
BPR] = 10.72; P Fisher = 0.001; OR = 14.78; PFH vs. BPR:
v2 = 0.30; P Fisher = 0.74, non-significant; OR = 1.46.
44 (13.6%). On the other hand, 11.7% of the subjects with depressive antecedents have CT.
3.2.2. Socio-demographic characteristics of individual cyclothymic temperaments
Looking at gender, comparison of the CT frequencies shows a higher rate in females: 72.3 versus
27.2% in males. But according to the overrepresentation of the females in the total sample, we observe a
non-significant excess towards this prevalence (respectively 8/101 vs. 3/76, i.e. 7.9 vs. 3.9%: v2 = 1.17;
P Fisher = 0.35; NS). Further, this result tends to be
confirmed when we segregate the groups as a function of the family antecedents. So we observe respectively 12.5% females versus 7.6% males in PFH and
20 versus 6.6% in BPR. In the NFH, we count only
one male.
3.2.3. Temperamental frequencies of individual
cyclothymic items
In Table 4 we summarize the distribution of the
participants as a function of their scores on the
cyclothymic scale. In the total sample, we previously
found 11 individuals with CT and, conversely, only
90 responded negatively to all items, corresponding
respectively to 63% of the NFH, 43.2% of the PFH
and only 29.7% of the BPR. So all of these participants could be considered as strictly non-cyclothymic (non-CT).
In addition to these two subgroups considering the
strictly CT and non-CT individuals, we do observe
that cyclothymic traits are quite widespread to some
extent in the other subjects. Indeed, we count in total
140
P. Chiaroni et al. / Journal of Affective Disorders 85 (2005) 135–145
Table 4
Scores of the CT scale in different groups NFH, PFH and BPR
Table 5
Socio-demographic characteristics of CT scale score subgroups
CT scale score
0
From 1 up to 8
CT
NFH
n
%
CT scale
score
63
63
36
36
1
1
PFH
n
%
16
43.24
17
45.94
4
10.81
BPR
n
%
n
%
11
27.5
90
50.84
23
57.5
76
42.93
6
15
11
6.23
NFH vs. PFH vs. BPR: [ P < 0.001], NFH vs. subjects at high risk
[PFH and BPR]: P < 0.001.
42.9% of the controls versus 57.5% of the BPR who
exhibit some positive responses to the scale (with total
scores between 1 and 8), some exceeding, others
without being strictly defined as CT according to the
criteria of Akiskal and Mallya (1987).
With respect to the familial antecedents, the distribution of these individuals differs significantly in the
different subgroups NFH, PFH and BPR: respectively,
36 versus 45.9 versus 54.0% ( P exact < 0.001). Further, the NFH also differ significantly from the highrisk subjects ( P exact < 0.001). We notice, then, that
one from the 76 subjects reaches the score threshold
but cannot be considered as cyclothymic, because the
distribution of these positive items do not respect the
threshold of each cluster.
As shown in Table 5, the mean ages in each group
are very close to each other and do not exhibit a
statistically significant difference (K – W = 0.389;
P = 0.823; NS). Compared to the bipolar relatives
without CT, we note that the mean age of the CT
BPR is younger (41.5 F 14.1 vs. 49.2 F 16.3). The
latter result raises the possibility that CT could evolve
to affective disorder over the course of time. As regards
gender, however, we show a statistically significant
difference between the groups ( P = 0.02). As opposed
to the NFH group, which shows a slight frequency of
males, the females are largely more numerous in the
subjects with CT or with cyclothymic scores between 1
and 8: respectively 51 versus 25 and 8 versus 3. So we
found a large prevalence of female individuals with 59
subjects out of 87, on the whole (i.e. 67.8%), who
0
from 1
up to 8
CT
Total
Females
n
%
42
46.7
51
67.1
8
72.7
101
57.1
Males
n
%
n
%
48
53.3
90
100
25
32.9
76
100
3
27.3
11
100
76
42.9
177
100
40.6 [11.7]
39.8 [12.3]
41.5 [14.1]
40.3 [12.0]
Age
Mean years
[F]
The mean age are not significantly different [K – W = 0.389;
P = 0.823]. The gender are not significantly different [v2 = 1.17;
P Fisher = 0.35].
exhibit cyclothymic features to some extent. This result
is, in our opinion, of major relevance as, we previously
noticed the CT rates increase in the same subgroups.
In Tables 6 and 7 we present respectively the
recruitment of subjects as a function of number of
positive responses to the cyclothymic scale and the
rates of responses to each item. In a dimensional
approach no symptomatic pattern is likely to emerge
from these distributions.
3.3. Comparison of cyclothymic versus non-cyclothymic temperaments in bipolar families
When we compared the clinical and socio-demographic characteristics of the probands’ bipolar disorder as a function of the temperament of their relatives,
no statistical difference was observed on the following
items: sex, polarity of the first episode, age of the first
Table 6
Frequency of non-cyclothymic individuals as a function of number
of positive response to constituent items or traits of the cyclothymic
scale
Non-CT score
[range 1 – 8] [n = 76]
1
2
3
4
5
item
items
items
items
items
29
24
13
9
1
P. Chiaroni et al. / Journal of Affective Disorders 85 (2005) 135–145
Table 7
Item rates in individuals with positive response to the cyclothymic
scale
Cluster A
Lethargy and somatic discomfort vs. eutonia
Dulling of senses vs. keen perceptions
Slow-witted vs. sharpened thinking
Shaky self-esteem alternating between low
Self-confidence and overconfidence
Cluster B
Hypersomnia vs. decreased need for sleep
Introverted self-absorption vs.
uninhibited people seeking
Taciturn vs. talkative
Unexplained tearfulness vs. buoyant jocularity
n
%
25
26
15
32.8
34.2
19.7
22
28.9
26
14
34.2
18.4
16
8
21.1
10.5
episode and number of episodes, therapeutic history
(with special regards to antidepressives and normothymics), and alcohol familial history (data not shown).
4. Discussion
4.1. Major findings and methodologic limitations
Compared to the prevalence of bipolar disorders,
our study clearly shows that the CT rate in our total
control sample is significant. This finding is in line
with the previously reported data in the literature
(ranging from 0.4 up to 6.3% (Depue et al., 1981;
Kraepelin, 1921; Weissman and Myers, 1978; Wetzel
et al., 1980)). Placidi et al. (1998a, b), with almost the
same methodology as ours, found the highest prevalence (6.3%) in 1010 Italian students who were much
younger (between 14 and 26) than our subjects.
We observe a dramatically higher rate in subjects
who are at risk for mood disorders when compared to
the subjects without any positive familial history of
mood disorder; that is to say, the controls who are first
degree relatives of unipolar depressive patients and
the subjects with bipolar kin. So, the CT is likely to
aggregate in families with a history of affective
disorder. Moreover, the CT rates increase from the
depressive family group to the bipolar family group
(kin). But an important bias deserves to be mentioned
as well. Indeed, our interviewed BPR group is only a
part of the total patients’ sibship and perhaps our
141
result could not be generalized to the entire family as
this sample represents 36.8% of the total kin (data
counted from the pedigrees). Thus, our percentage
probably indicates a surprisingly high proportion.
Nevertheless, as expected, this estimation in BPR is
a lower than previously reported data on this temperament in bipolar patients themselves (Akiskal et al.,
2003; Hantouche et al., 1998).
Before discussing these specific findings, we will
consider other methodological aspects that could
affect our results. Of course, there are some inevitable
limitations due to the design of this study. And so, our
findings must of course be replicated in independent
and larger samples (see Evans et al., 2004, this issue).
First, the TRUE/FALSE answer format can habitually result in the collection of insufficient information. But in the present case, it is easily conceivable
that this procedure tends to maximize the variance of
the information, which could be strongly dependent
on the subjects’ insight and on the degree of lived
functional inconvenience and social impairment during the lifetime. Further, the ‘‘pessimistic’’ or ‘‘optimistic’’ state of the mind of the subject in CT (Bromet
et al., 1986; Kandel and Davies, 1986) could largely
influence the kind of given response studied, for
example, by Platman et al. (1969) with the perception
of self across the affective states. From this point of
view, our 3 months’ test – retest procedure may not
represent an adequate time frame to provide accurate
results, but does require longer intervals of testing.
However, in the Placidi et al. (1998b) study, cyclothymia was the most robust temperament over a 2year period of prospective follow-up.
Also, the problem of the definition of the CT
remains open. This problem is certainly of major
relevance for at least two reasons. The first one is
derived from the widespread scores, from 0 to 8, in
the total sample individuals. Also, the distributions of
these scores do not seem to be homogenous in the
different subgroups. In a dimensional approach, a
threshold for the global score of this questionnaire
defines CT in subjects who never had psychiatric
hospitalization nor consulted a psychiatrist. Therefore,
we have to attribute to this definition an identical
weighting to each item. In this hypothesis one could
consider the CT to be an exaggeration of normal traits
in the general population and we could think of a
continuum from the non-CT to the CT. This is exactly
142
P. Chiaroni et al. / Journal of Affective Disorders 85 (2005) 135–145
in line with the typology of Kretschmer (1930);
Schneider (1973) that considered a hard core and
some doubtful cases for each described temperament
or abnormal personality. So the accurate value of this
threshold requires some prospective studies to validate its pertinence. But, even upon the seemingly
straight-forward criteria of the evolution course, such
as the appearance of pathological manifestation, we
know perfectly that this design is methodologically
really hard to develop, as the duration of such a
prospective study should be equivalent to the investigator’s lifetime. But in an alternative categorical
approach, we could hypothesize that the different
items are not all equivalent to each other. The CT
item frequencies we observed are likely to weaken
this affirmation but a multivariate analysis, which we
could not perform, would be more informative. Finally, a third way is to consider the CT as a heterogeneous condition (Howland and Thase, 1993).
4.2. Cyclothymic temperament and gender
The CT seems to predominate in females. Furthermore, females present more cyclothymic traits in the
non-CT subgroups. Higher prevalence of CT in female
controls has also been observed by others in the
general population (Depue et al., 1981; Placidi et
al., 1998a) and among clinical cohorts (Akiskal et
al., 1977; Alnaes and Torgensen, 1989; Dunner et al.,
1982). In contrast, the patients’ gender received much
attention in temperamental studies (Akiskal et al.,
1998; Cassano et al., 1992; Dell’Osso et al., 1991;
Nolen-Hoeksema et al., 1993; Perugi et al., 1990,
1997). But, to our knowledge, the CT was not
reported to have distribution as a function of gender.
Especially, Cassano and colleagues showed in 538
patients that the hyperthymic temperament was observed more commonly in men and that the depressive
temperament was more prevalent in women. All these
results can be contrasted with the equal sex distribution in bipolar disorder. In balance, it is uncertain if
the gender effect represents a substantial bias for the
generalization of our results.
4.3. Cyclothymic temperament and affective illness
The results described herein indicate that most of
the participants (90.9%) with a defined CT have a
familial history of affective disorders, divided in
36.3% for the depressive disorder and 54.5% for the
bipolar type I disorder, that is to say in subjects really
at risk for mood disorders. When replicated and
confirmed, these results could indicate that CT runs
frequently in families with mood disorder history and
especially in a statistically significant manner. When
compared to NFH controls, they indicate furthermore
an aggregation of CT and affective disorders, bipolar
as well as unipolar, depressive disorders. This finding
is similar, yet independently made discovery that goes
beyond the study by Mendlowicz et al. (2004) (this
issue) in showing an excess of CT in subjects who are
actually symptom free as far as clinical episodes are
concerned.
But concerning specifically cyclothymic disorder,
many authors subsequent to Jelliffe (1911), reported a
frequent coexistence of bipolar and cyclothymic cases
in families. In the studies of either cyclothymic or
bipolar probands, all the conclusions are consistent
with a strong association between the two conditions
(Akiskal et al., 1977, 2003; Depue et al., 1981;
Dunner et al., 1976, 1982; Endicott et al., 1985;
Gershon et al., 1982; Klein et al., 1985, 1986). Many
authors also reported that cyclothymia and CT
evolved frequently into manic-depressive illness, especially at the beginning of this century (Jelliffe,
1911; Kraepelin, 1921; Kretschmer, 1930), and more
recently (Akiskal et al., 1977, 1985; Klein and Depue,
1984; Waters, 1979).
Concerning the problem of unipolar depressive
disorder, the findings are more conflicting. The majority of authors reported no differences in the rate of
cyclothymia between unipolar probands and controls
(Dunner et al., 1976; Gershon et al., 1975, 1982;
Klein et al., 1988; Weissman et al., 1984). But others
have reported opposite findings (Depue et al., 1981;
Dunner et al., 1982). Hantouche et al. (1998) recently
showed in the EPIMAN study of 250 affective
patients that CT was less frequent in pseudo-unipolar
depressives. But these discrepancies are perhaps in
relation to the degree of lesser certainty in the diagnosis of unipolar depression compared to that of frank
bipolar disorder. In light of these results, we would
hypothesize that some depressive relatives of the
cyclothymic subject could be undoubtedly pseudounipolar patients or BP II, which could change the
nature of our conclusions. So, given all the above, it is
P. Chiaroni et al. / Journal of Affective Disorders 85 (2005) 135–145
likely to further hypothesize a common shared pathway between these CT individuals and subjects suffering from full-blown mood disorders.
4.4. Cyclothymic temperament and genetics of
bipolar disorder
From the differences in the rate between NFH
controls, on the one hand, and at risk individuals, on
the other, we could hypothesize that CT is unlikely to
breed true as a genetic trait in subjects, but apparently
segregates in families in which affective antecedents
are present. Moreover, a familial aggregation of Cyclothymic temperamental features and acute clinical
episodes could suggest a partially shared contribution
of common genetic factors. More provocatively, the
problem of temperaments could be of major relevance, especially concerning the problem of a genetic
liability that could underlie the pathophysiology of
bipolar illness.
Indeed, a genetic component is relatively well
established in bipolar illness, although the genes and
the underlying pathogenic mechanisms remain mysterious. Several methodological limitations confront
us. First, our results emphasize the importance of a
careful selection of control groups. Of course, the
identification of affected subjects has to be consistently related to a precise psychiatric illness and
discerned from other diagnoses or social maladjustment. But, even in terms of the factor of predisposition, the distinction between screened and unscreened
individuals, from the temperamental point of view,
could be very important in genetic studies because it
could delineate a clear distinction between subjects
actually free of symptoms in ‘‘super normal’’ or at risk
individuals as a function of the family history. This is
of major relevance. So we join Tsuang et al. (1988) in
proposing that both screened and unscreened controls
must be included in the design of genetic studies to be
compared to patients.
Since absolutely no biological marker has been
proven so far, it is clear that whatever the methodological approach used—association or linkage methods, for instance—an accurate definition of the
phenotype is crucial in the design of genetic studies.
So, in familial studies, do we have to consider
‘‘affected’’ the individuals with affective temperaments, and with especially CT?
143
Actually with our results, we have reasonable
arguments to submit that CT could contribute to a
broad phenotypic definition of the bipolar condition.
This is in line with Kretschmer (1930) and subsequent
reformulation by the senior author (Akiskal, 1977;
Akiskal, 1995; Akiskal and Akiskal, 1996), or CT
could merely play pathoplastic role in forecasting
disability; it could actually be a valuable psychological marker for different methodologies in molecular
biology. The focus in the field of genetic research may
shift from the study of episodes, as it is unlikely that
patients inherit the illness itself, but more probably
some general predisposition to become affectively ill.
Such an approach was fruitful when Ebstein et al.
(1997); Staner et al. (1998) reported an association
between human personality traits—respectively the
reward dependence and the novelty seeking—and
chromosomal loci, which encode for the 5-HT2C
Receptor, the Dopamine D3 or D4 Receptor proteins.
From this point of view, it could be informative to
evaluate the different temperaments (depressive and
hyperthymic as well as the cyclothymic) among all the
kin of bipolar probands (Chiaroni et al., in press).
In conclusion, such a perspective is a potentially
productive approach in the investigation of genetic
and environmental influences in the occurrence of
acute episodes from a pathogenic as well as therapeutic point of view. Even when the CT could appear as a
likely precursor of manic-depressive illness (Akiskal
et al., 1979, 1985; Kraepelin, 1921)—especially bipolar disorder II with its unstable course (Akiskal et
al., 2003; Dunner et al., 1976)—it may not necessarily
be a sufficient factor for the development of a bipolar
disorder (Wetzel et al., 1980).
Surprisingly, temperament has received little attention in genetic studies. Danger exists today that
the development of sophisticated molecular genetic
methods may elude behavioral genetics. This special
issue dedicated to temperament is an antidote to this
problem.
Acknowledgements
Pierre Chiaroni was funded by a Research
Fellowship from the ‘‘Assistance Publique de Marseille’’. The authors are very grateful to Dr. L.
Chatenêt-Duchêne for her support.
144
P. Chiaroni et al. / Journal of Affective Disorders 85 (2005) 135–145
References
Akiskal, H.S., 1983. The bipolar spectrum: new concepts in classifications and diagnosis. In: Greenspoon, L.Psychiatry Update.
American Psychiatric Association Annual Review, vol. II.
American Psychiatric Press Inc, pp. 271 – 292.
Akiskal, H.S., 1995. Toward a temperament-based approach to depression: Implications for neurobiologic research. Adv. Biochem. Psychopharmacol. 49, 99 – 112.
Akiskal, H.S., Akiskal, K., 1996. The temperamental foundations of
mood disorders. In: Mundt, C.H.Interpersonal Factors in the
Origin and Course of Affective Disorders. Gaskell, London,
pp. 3 – 30.
Akiskal, H.S., Mallya, G., 1987. Criteria for the ‘‘soft’’ bipolar
spectrum: treatment implications. Psychopharmacol. Bull. 23,
68 – 73.
Akiskal, H.S., Djenderedjian, A.H., Rosenthal, R.H., Khani, M.K.,
1977. Cyclothymic disorder. Validity for inclusion in the bipolar
affective group. Am. J. Psychiatry 134, 1227 – 1233.
Akiskal, H.S., Khani, M.K., Scott-Strauss, A., 1979. Cyclothymic
temperamental disorders. Psychiatr. Clin. North Am. 2, 527 – 554.
Akiskal, H.S., Downs, J., Jordan, P., Watson, S., Daugherty, D.,
Pruitt, D.B., 1985. Affective disorders in referred children and
younger siblings of manic-depressives: modes of onset and prospective course. Arch. Gen. Psychiatry 42, 996 – 1003.
Akiskal, H.S., Placidi, G.F., Signoretta, S., Liguori, A., Gervasi, R.,
Maremmani, I., Mallya, G., Puzantian, V.R., 1998. TEMPS-I:
delineating the most discriminant traits of cyclothymic, depressive, irritable and hyperthymic temperaments in a nonpatient
population. J. Affect. Disord. 51, 7 – 19.
Akiskal, H.S., Hantouche, E.G., Allilaire, J.-F., 2003. Bipolar II
with and without cyclothymic temperament: ‘‘dark’’ and ‘‘sunny’’ expressions of soft bipolarity. J. Affect. Disord. 73, 49 – 57.
Alnaes, R., Torgensen, S., 1989. Characteristics of patients with
major depression in combination with dysthymic or cyclothymic
disorders. Acta Psychiatr. Scand. 79, 11 – 18.
American Psychiatric Association, 1968. Diagnostic and Statistical
Manual of Mental Disorders, 2nd ed. Washington, DC.
American Psychiatric Association, 1995. Diagnostic and Statistical
Manual of Mental Disorders, 4th ed. Washington, DC.
Angst, J., 1978. The course of affective disorders: II. Typology of
bipolar manic-depressive illness. Arch. Psychiatr. Nervenkr.
226, 65 – 73.
Angst, J., Frey, R., Lohmeyer, B., Zerbin-Rudin, E., 1980. Bipolar
manic-depressive psychoses: results of a genetic investigation.
Hum. Genet. 55, 237 – 254.
Brieger, P., Marneros, A., 1997. Dysthymia and cyclothymia: historical origins and contemporary development. J. Affect. Disord.
45, 117 – 126.
Bromet, E.J., Dunn, L.O., Connell, M.M., Dew, M.A., Schulberg,
H.C., 1986. Long-term reliability of diagnosing lifetime major
depression in a community sample. Arch. Gen. Psychiatry 43,
435 – 440.
Cassano, G.B., Akiskal, H.S., Perugi, G., Musetti, L., Savino, M.,
1992. The importance of measures of affective temperaments
in genetic studies of mood disorders. J. Psychiatr. Res. 26,
257 – 268.
Chiaroni, P., Azorin, J.-M., Dassa, D., Henry, J.-M., Giudicelli, S.,
Malthiéry, Y., Planells, R., 2000. Association study between
bipolar affective disorder and dopamine D3 receptor gene polymorphisms which could involve this locus as a minor effect
gene. Psychiatr. Genet. 10, 43 – 49.
Chiaroni, P., Hantouche, E.G., Gouvernet, J., Azorin, J.M., Roure,
L., Akiskal, H.S. Étude psychometrique des échelles tempéramentales dépressive, hyperthymique et irritable chez 164 sujets
contrôles et apparentés de patients souffrant de troubles de l’humeur. Encéphale (in press).
Cloninger, C.R., Svrakic, D.R., Przybeck, T.R., 1993. A psychobiological model of temperament and character. Arch. Gen. Psychiatry 50, 975 – 990.
Dell’Osso, L., Placidi, G.F., Nassi, R., Freer, P., Cassano, G.B.,
1991. The manic depressive mixed state: familial, temperamental and psychopathologic characteristics in 108 females inpatients. Eur. Arch. Clin. Neurosci. 240, 234 – 239.
Depue, R.A., Slater, J.F., Wolfstter-Kausch, H., Klein, D.,
Goplerud, E., Farr, D., 1981. A behavioral paradigm and bipolar
for identifying persons at risk for bipolar depressive disorder: a
conceptual framework and five validation studies. J. Abnorm.
Psychol. 90, 381 – 437.
Dunner, D.L., Gershon, E.S., Goodwin, F.K., 1976. Heritable factors
in the severity of affective illness. Biol. Psychiatr. 11, 31 – 42.
Dunner, D.L., Russek, F.D., Russek, B., Fieve, R.R., 1982. Classification of bipolar affective disorder subtypes. Compr. Psychiatry 18, 561 – 566.
Ebstein, R.P., Segman, R., Benjamin, J., Osher, Y., Nemanov, L.,
Belmaker, R.H., 1997. 5-HT2C serotonin receptor gene polymorphism associated with the human personality trait of reward
dependence: interaction with dopamine D4 receptor (D4DR)
and dopamine D3 receptor (D3DR) polymorphisms. Am. J.
Med. Genet. 74, 65 – 72.
Endicott, J., Nee, J., Andreassen, N., Clayton, P., Keller, M., Coryell, W., 1985. Bipolar II: combine or keep separate? J. Affect.
Disord. 8, 17 – 28.
Evans, L., Akiskal, H.S., Keck Jr., P.E., McElroy, S.L., Sadovnick,
A.D., Remick, R.A., Kelsoe, J.R., 2004. Familiality of temperament in bipolar disorder: support for a genetic spectrum.
J. Affect. Disord. 85, 152 – 168.
Gershon, E.S., Mark, A., Cohen, N., Belizon, N., Baron, M.,
Knobe, K.E., 1975. Transmitted factors in the morbid risk of
affective disorders: a controlled study. J. Psychiatr. Res. 12,
283 – 299.
Gershon, E.S., Hamovit, J., Guroff, J.J., Dibble, E., Leckman, J.F.,
Sceery, W., Targum, S.D., Nurnberger, J.I., Goldin, L.R., Bunney, W.E., 1982. A family study of schizoaffective, bipolar I,
bipolar II, unipolar, and normal controls probands. Arch. Gen.
Psychiatry 39, 1159 – 1167.
Goodwin, F., Jamison, K., 1990. Manic-Depressive Illness, Oxford
University Press, New York.
Hantouche, E., Akiskal, H.S., Lancrenon, S., Allilaire, J.-F.,
Sechter, D., Azorin, J.-M., Bourgeois, M., Fraud, J.-P., Châtenet-Duchêne, L., 1998. Systematic clinical methodology for
validating bipolar-II disorder: data in mid-stream from a French
National Multisite Study (EPIDEP). J. Affect. Disord. 50,
163 – 173.
P. Chiaroni et al. / Journal of Affective Disorders 85 (2005) 135–145
Hirschfeld, R.M.A., Klerman, G.L., Clayton, P.J., Keller, M.B.,
1983. Personality and depression. Arch. Gen. Psychiatry 40,
993 – 998.
Hirschfeld, R.M.A., Klerman, G.L., Lavori, P., Keller, M.B., Griffith, P., Coryell, W., 1989. Premorbid personality assessments of
first onset of major depression. J. Affect. Disord. 46, 345 – 350.
Howland, R.H., Thase, M.E., 1993. A comprehensive review of
cyclothymic disorder. J. Nerv. Ment. Dis. 181, 485 – 493.
Jelliffe, S.E., 1911. The mild forms of manic-depressive psychoses
and the manic-depressive constitution. Am. J. Insanity 67,
661 – 676.
Kandel, D.B., Davies, M., 1986. Adult sequelæ of adolescent depressive symptoms. Arch. Gen. Psychiatry 43, 255 – 262.
Kelsoe, J.R., 2003. Arguments for the genetic basis of the bipolar
spectrum. J. Affect. Disord. 73, 183 – 197.
Klein, D.N., Depue, R.A., 1984. Continued impairment in persons
at risk for bipolar affective disorder: results of a 19 month
follow-up study. J. Abnorm. Psychol. 93, 345 – 347.
Klein, D.N., Depue, R.A., Slater, J.F., 1985. Cyclothymia in the
adolescent offspring of parents with bipolar affective disorder.
J. Abnorm. Psychol. 94, 115 – 127.
Klein, D.N., Depue, R.A., Slater, J.F., 1986. Inventory identification of cyclothymia: IX. Validation in offspring of bipolar I
patients. Arch. Gen. Psychiatry 43, 441 – 445.
Klein, D.N., Clark, D.C., Dansky, L., Margolis, E.T., 1988. Duysthymia in the offspring of parents with primary unipolar affective disorder. J. Abnorm. Psychol. 97, 265 – 274.
Klerman, G.L., 1981. The spectrum of mania. Comp. Psychiatry 22,
11 – 20.
Kraepelin, E., 1921. Manic-Depressive Insanity and Paranoia. Edinburgh, Livingston (ed.).
Kretschmer, E., 1930. La structure du corps et le caractère. Payot
(éd.), Paris.
Mendlowicz, M.V., Jean-Louis, G., Kelsoe, J.R., Akiskal, H.S.,
2004. A comparison of recovered bipolar patients, healthy relatives of bipolar probands, and normal controls using the short
TEMPS-A. J. Affect. Disord. 85, 147 – 151.
Nolen-Hoeksema, S., Morrow, J., Fredrickson, B.L., 1993. Response styles and the duration of episodes of depressed mood.
J. Abnorm. Psychol. 102, 20 – 28.
Perugi, G., Musetti, L., Simonini, E., Piagentii, F., Cassano, G.B.,
Akiskal, H.S., 1990. Gender mediated clinical features of depressive illness: the importance of temperament differences. Br.
J. Psychiatry 157, 835 – 841.
Perugi, G., Akiskal, H.S., Micheli, C., Musetti, L., Paiano, A.,
145
Quilici, C., Rossi, L., Cassano, G.B., 1997. Clinical subtypes
of bipolar mixed states: validating a broader European definition
in 143 cases. J. Affect. Disord. 43, 169 – 180.
Placidi, G.F., Signoretta, S., Liguori, A., Gervasi, R., Maremmani,
I., Akiskal, H.S., 1998a. The semi-structured affective temperament interview (TEMPS-I). Reliability and psychometric properties in 1010 14 – 26-year old students. J. Affect. Disord. 47,
1 – 10.
Placidi, G.F., Maremmani, I., Signoretta, S., Liguori, A., Akiskal,
H.S., 1998b. A prospective study of stability and change over
2 years of affective temperaments in 14 – 18 year old Italian
high school students. J. Affect. Disord. 51, 199 – 208.
Platman, S.R., Plutchik, R., Fieve, R.R., Lawlor, W.G., 1969. Emotion profiles associated with mania and depression. Arch. Gen.
Psychiatry 20, 210 – 214.
Schneider, K., 1973. Psychopathologie clinique, 4th ed. Maloine
éd, Paris.
Staner, L., Hilger, C., Hentges, F., Monreal, J., Hoffmann, A., Couturier, M., Le Bon, O., Stefos, G., Souery, D., Mendlewicz, J.,
1998. Association between novelty-seeking and the dopamine
D3 receptor gene in bipolar patients: a preliminary report. Am.
J. Med. Genet. 81, 192 – 194.
Thompson, W.D., Kidd, J.R., Weissman, M.M., 1979. A procedure
for the efficient collection and processing of pedigree data suitable for genetics analysis. J. Psychiatr. Res. 15, 291 – 303.
Tsuang, M.T., Fleming, J.A., Kendler, K.S., Gruenberg, A.S., 1988.
Selection of controls for families studies. Arch. Gen. Psychiatry
45, 1006 – 1008.
Waters, B.G.H., 1979. Early symptoms of bipolar affective psychosis. Can. Psychiatr. Assoc. J. 2, 55 – 60.
Weissman, M.M., Myers, J.K., 1978. Rates and risks of depressive
symptoms in a United States urban community. Acta Psychiatr.
Scand. 57, 219 – 321.
Weissman, M.M., Gershon, E.S., Kidd, K.K., Prusoff, B.A., Leckman, J.F., Dibble, E., Hamovit, J., Thompson, W.D., Pauls,
D.L., Guroff, J.J., 1984. Psychiatric disorders in the relatives of
probands with affective disorders. Arch. Gen. Psychiatry 41,
13 – 21.
Wetzel, R.D., Cloninger, C.R., Hong, B., Reich, T., 1980. Personality as a subclinical expression of the affective disorders.
Compr. Psychiatry 21, 197 – 205.
Wing, J.K., Babor, T., Brugha, T., Burke, J., Cooper, J.E., Giel, R.,
Jablenski, A., Regier, D., Sartorius, N., 1990. SCAN. Schedules
for Clinical Assessment in Neuropsychiatry. Arch. Gen. Psychiatry 47, 589 – 593.