. 23
( 68 .)


Disorders: A Comparative Long-Term Study.] Berlin: Springer.
Meier, R. (1985). Katamnese von 40 jugendlichen Patienten nach einem Suizidversuch bei
verschiedenen Behandlungsmoglichkeiten im Anschluss an eine somatische Klinik. Inaugural
dissertation, Med. Diss. Zurich: University of Zurich.
Merikangas, K., Wicki, W., and Angst, J. (1990). Combined Depression. Royal College of
Psychiatrists Annual Meeting 1990. Birmingham, p. 54.
Montgomery, S. A. (1997). Suicide and antidepressants. Ann. N. Y. Acad. Sci., 836, 329“38.
Montgomery, S. A., Montgomery, D., and Baldwin, D. (1989). Intermittent 3-day depressions
and suicidal behaviour. Neuropsychobiology, 22, 128“34.
Montgomery, D. B., Roberts, A., and Green, M. (1994). Lack of efficacy of fluoxetine in
recurrent brief depression and suicidal attempts. Eur. Arch. Psychiatry Clin. Neurosci. 244,
Paskind, H. A. (1929). Brief attacks of manic-depressive depressions. Arch. Neurol. Psychiatry,
22, 123“34.
Paskind, H. A. (1930). Manic-depressive psychosis as seen in private praxis. Length of the attack
and length of the interval. Arch. Neurol. Psychiatry, 23, 789“94.
Pazzaglia, P. J., Post, R. M., and Ketter, T. A. (1993). Preliminary controlled trial of nimodipine
in ultra-rapid cycling affective dysregulation. Psychiatry Res., 49, 257“72.
Pazzaglia, P. J., Post, R. M., Ketter, T. A. et al. (1998). Nimodipine monotherapy and carbfama-
zepine augmentation in patients with refractory recurrent affective illness. J. Clin.
Psychopharmacol., 18, 404“13.
Pezawas, L., Stamenkovic, M., and Aschauer, N. (2002a). Successful treatment of recurrent brief
depression with reboxetine “ a single case analysis. Pharmacopsychiatry, 35, 75“6.
Pezawas, L., Stamenkovic, M., and Jagsch, R. (2002b). A longitudinal view of triggers and
thresholds of suicidal behavior in depression. J. Clin. Psychiatry, 63, 866“73.
Pezawas, L., Wittchen, H. U., and Pfister, H. (2003). Recurrent brief depressive disorder
reinvestigated: a community sample of adolescents and young adults. Psychol. Med., 33,
Pfortmuller, J. (1983). Beitrag zur Validierung des Depressionsratings im Fragebogen SPIKE IV an
psychiatrisch hospitalisierten Patienten. Inaugural dissertation, Med. Diss. Zurich: University
of Zurich.
Post, R. M., L™Herrou, T., and Luckenbaugh, D. A. (1998). Statistical approaches to trial dura-
tions in episodic illness. Psychiatry Res., 78, 71“87.
Stamenkovic, M., Pezawas, L., and De Zwaan, M. (1998). Mirtazepine in recurrent brief
depression. Int. Clin. Psychopharmacol., 13, 39“40.
Stamenkovic, M., Blasbichier, T., Riederer, F. et al. (2001). Fluoxetine treatment in patients with
recurrent brief depression. Int. Clin. Psychopharmacol., 16, 221“6.
Staner, L., De La Fuente, J. M., and Kerkhofs, M. (1992). Biological and clinical features of
recurrent brief depression: a comparison with major depressed and healthy subjects. J. Affect.
Disord., 26, 241“6.
130 J. Angst et al.

Verkes, R. J., Van Der Mast, R. C., and Hengeveld, M. W. (1998). Reduction by paroxetine of
suicidal behavior in patients with repeated suicide attempts but not major depression. Am. J.
Psychiatry., 155, 543“7.
Wittchen, H.-U., Zaudig, M., and Fydrich, T. (1998). Strukturiertes Klinisches Interview fur
¨ ¨
DSM“IV (SKID-I und SKID-II). Gottingen: Hogrefe-Verlag fur Psychologie.
World Health Organization (1992). The ICD-10 Classification of Mental and Behavioural
Disorders. Clinical Descriptions and Diagnostic Guideline. Geneva: World Health

Atypical depression and its relation to
bipolar spectrum
Franco Benazzi
Hecker Psychiatry Research Center, University of California in San Diego (USA) collaborating center, Ravenna, Italy;
University of Szeged (Hungary); National Health Service, Forli, Italy

Introduction: the relationship of atypical depression to bipolar II disorder
The focus of this chapter on the relationship between atypical depression (AD)
(different definitions, including Diagnostic and Statistical Manual of Mental
Disorders, 4th edn (DSM-IV): (American Psychiatric Association, 1994) defini-
tion) and bipolar (BP) spectrum is the relationship between BP-II and AD,
because BP-II is the most common and best-studied disorder of the BP spectrum
disorders. BP-II was recently found to be very common in the community (11.0%
BP-II versus 11.4% unipolar (UP): Angst et al., 2003) and in major depressive
episode (MDE) outpatients (up to 60%: Cassano et al., 1992; Angst, 1996;
Benazzi, 1997a, 2001a; Hantouche et al., 1998; Perugi et al., 1998; Akiskal et al.,
2000; Benazzi and Akiskal, 2003a). But nevertheless BP-II is still underdiagnosed
(Ghaemi et al., 2000). Lumping bipolar-I (BP I) and BP-II together is not
supported by the BP-II strong diagnostic stability (Coryell et al., 1995), different
family history (more BP-II than BP-I in first-degree relatives of BP-II) (Goodwin
and Jamison, 1990; Coryell, 1999), and by linkage studies (McMahon et al., 2001).
The most recent definitions of BP spectrum come from Akiskal and Pinto
(1999), Ghaemi et al. (2002), and Angst et al. (2003). Akiskal and Pinto™s definition
includes BP-I, BP-II (hypomania and MDE Æ cyclothymic disorder), BP-III
(antidepressant and stimulant-associated hypomania), and BP-IV (depressive
mixed state, that is, a MDE plus some concurrent hypomanic symptoms). The
definition of Angst et al. (2003) includes BP-II, minor bipolar disorders (hypo-
mania and mild depressions), and single hypomania (with no depression). The
diagnostic criteria of Ghaemi et al. (2002) for bipolar spectrum disorder include a
UP MDE plus signs of bipolarity, such as bipolar family history, antidepressant-
induced mania/hypomania, and AD. The bipolar spectrum disorder of Ghaemi
et al. (2002) links pure UP with BP. Also depressive mixed state (Benazzi and
Cambridge University Press, 2005.
132 F. Benazzi

Akiskal, 2001; Akiskal and Benazzi, 2003;) was found to be a link between pure UP
and BP. In the American Psychiatric Association, DSM-IV (1994) bipolar disorders
are divided into:
(1) BP-I disorder
(2) BP-II disorder
(3) cyclothymic disorder
(4) bipolar disorder not otherwise specified
(5) manic or hypomanic episodes due to a general medical disorder or substance-
DSM-IV BP-II criteria for hypomania require elevated or irritable mood, lasting
at least 4 days, plus at least three (four, if mood is irritable) hypomanic symptoms,
an observable change in functioning, a mild episode not due to substances,
antidepressants, or medical disorders, and not superimposed on psychotic dis-
orders. DSM-IV BP-II criteria have some problems:
(1) No data support the cut-off of 4 days (Dunner, 1998), while a cut-off of 2 days
is supported by data (Akiskal et al., 2000)
(2) There are no clear boundaries between mania and hypomania
(3) Antidepressant-associated hypomania is not classified as BP-II, while follow-
up studies found that antidepressant-associated hypomania will have sponta-
neous hypomania in many cases (Akiskal and Pinto, 1999)
(4) Hypomanic mood is the first criterion, while recent studies (Akiskal et al.,
2001; Angst et al., 2003; Benazzi and Akiskal, 2003a, b) found that overactive
behavior is at least as important as hypomanic mood
(5) The Structured Clinical Interview for DSM-IV Axis I Disorders“Clinician Version
(SCID-CV; First et al., 1997) skip-out stem question on mood does not allow the
assessment of the other hypomanic symptoms if it is negative, while Dunner and
Tay (1993) and Benazzi and Akiskal (2003a) found that systematic assessment of
all past hypomanic symptoms increased the frequency of BP-II diagnoses.

Recent literature review
A comprehensive review of the literature on AD until the early 1990s can be found
in Rabkin et al. (1996), and an updated review in Angst et al. (2002). The
diagnostic validity of AD is mainly based on its better response to monoamine
oxidase inhibitors (MAOI) rather than to tricyclic antidepressants (TCA) (Rabkin
et al., 1996), and on latent class analysis (Kendler et al., 1996; Sullivan et al., 1998).
An important limitation of these studies is the inclusion of mainly UP samples.
The current definition of AD is mostly based on the Columbia group definition of
AD (Rabkin et al., 1996).
133 Atypical depression and bipolar spectrum

In DSM-IV, AD is not classified as a distinct mood disorder, but as a subtype
(specifier) of BP MDE and depressive (UP) disorders. DSM-IV criteria for the AD
specifier always require mood reactivity, plus at least two symptoms, including
weight gain or overeating, hypersomnia, leaden paralysis, and interpersonal rejec-
tion sensitivity (plus criteria for concurrent melancholic or catatonic features not
met). There are a lack of data supporting the inclusion of mood reactivity (Posternak
and Zimmerman, 2002; Rabkin et al., 1996). AD™s better response to MAOI than to
TCA was not related to any one AD symptom (McGrath et al., 1992), and when only
mood reactivity was present, MAOI were not more effective than TCA (Quitkin
et al., 1989). Angst et al. (2002) found support for a definition of AD where mood
reactivity has no priority over the other symptoms. Posternak and Zimmerman
(2002) found that mood reactivity was not correlated with the other AD symptoms,
while Angst et al. (2002) found that it was associated with other AD symptoms.
Benazzi (2002a) found that mood reactivity was associated with the other AD
symptoms in BP-II, but not in UP (the previous studies combined in the analysis
BP-II and UP, and had very different proportions of BP-II).
Surprisingly little research has been done on AD in BP. The DSM-IV-TR 2000
(American Psychiatric Association, 2000) literature review concluded that AD
was more common in BP. There is some similarity between the Columbia group
definition, and antidepressant response, of AD and the definition of Himmelhoch
et al. (1991) and antidepressant response of anergic BP depression. Anergic BP
depression of Himmelhoch et al. (1991) included loss of energy, psychomotor
retardation, hypersomnia, and weight gain. Its frequency was 73% among 77 BP
outpatients (57.1% were BP-II). Himmelhoch (1999) also reported that melancholic
features were present in only 12 of 1100 BP patients. Data on non-
melancholic depression in BP can give indirect information about AD, as at least
some of the non-melancholic depressions can be AD. Parker et al. (2000) found
that BP (I þ II) versus UP depression was more likely to be melancholic. However,
the studies of Parker et al. were based on mixed in-/outpatient samples (melan-
cholic depression was reported to be more common in inpatients: American
Psychiatric Association, 2000), and BP-I and BP-II were lumped together.
Serretti et al. (1998) found more AD symptoms in BP-I versus UP inpatients.
Mitchell et al. (2001) found that BP-I versus UP depression (matched for age) had
significantly more hypersomnia and retardation. More hypersomnia in BP depres-
sion versus UP depression is reported in standard textbooks (Akiskal, 2000;
Goodwin and Jamison, 1990). In Baldessarini™s review (2000), anergia and hyper-
somnia were reported to be typical bipolar features. More AD in BP-II versus UP
was found by Angst (1998), Perugi et al. (1998), Agosti and Stewart (2001), Angst
et al. (2002, 2003). No more or less AD in BP-II versus UP was found by McGrath
et al. (1992), Robertson et al. (1996), and Posternak and Zimmerman (2002).
134 F. Benazzi

Agosti and Stewart (2001) found more BP-II in AD versus non-AD, but the
frequency of BP-II found was very low (12%).
A prospective study found that AD often progressed to BP spectrum (Ebert et al.,
1993). Akiskal et al. (2000) reported that hypersomnic-retarded depression had 88%
specificity for predicting BP outcome. Interpersonal sensitivity (an AD symptom)
was found to predict the switching of UP to BP-II (Akiskal et al., 1995). Perugi et al.
(1998) found that BP-II and BP spectrum were present in 72% of 86 AD outpatients,
that BP-II AD (n ¼ 28) versus UP (n ¼ 24) AD had higher family history of BP, but
similar female frequency, age of onset, and MDE recurrences. The power of the
sample was however low. Hantouche et al. (1998) found that hypersomnia was more
common in BP-II MDE (n ¼ 100) versus UP MDE (n ¼ 113). Angst (1998) found
AD more common in the soft BP spectrum versus UP (28.6% versus 6.8%).
Cassano et al. (1992) found a similar high frequency of melancholic features
between BP-II and UP (AD was not assessed, but, according to DSM-IV, AD
diagnosis cannot be made if the criteria of melancholic features are met). In the
community study of Levitan et al. (1997), MDE with some AD symptoms (over-
eating, weight gain, hypersomnia) was found to have higher mania frequency versus
MDE without these symptoms (BP-II was not assessed). In the community study of
Sullivan et al. (1998), AD (defined by overeating, weight gain, and hypersomnia)
had similar BP-I comorbidity and age of onset versus non-AD (BP-II was not
assessed). In the community study of Horwath et al. (1992), first onset of mania
at 1-year follow-up in UP AD (defined by overeating and oversleeping) versus non-
AD was similar and very low. McGrath et al. (1992) reported that, among 401 AD,
only 10% had BP-II. Robertson et al. (1996), comparing 79 UP with 30 BP (I þ II),
found similar frequency of AD (28% versus 30%), but only 10 BP-II were included.
Posternak and Zimmerman (2002) did not find more BP-II in AD versus non-AD,
but, in the 579 sample, almost all were UP (28 patients were BP-II (4.8%)). The
community study of Angst et al. (2003) found that AD versus non-AD had earlier
onset, more BP spectrum, females, recurrences, and chronicity. Angst et al. (2002)
found that BP-II (n ¼ 89) had more AD than UP (n ¼ 101: 49.5% versus 29.6%).

The author™s studies
In the author™s studies AD means DSM-IV ˜˜atypical features specifier.™™ The
studies try to answer the following questions.

Is AD more common in BP-II versus UP?
Frequency of AD was significantly much higher (more than 40%) in BP-II MDE
versus UP MDE. In BP-II (n ¼ 251) versus UP (n ¼ 306) MDE, AD was present in
135 Atypical depression and bipolar spectrum

45.4% versus 25.4% (P ¼ 0.0000). In another study, prevalence of BP-II in AD was
64.2% (n ¼ 140). These findings are in line with previous reports (Perugi et al.,
1998; Angst et al., 2003). Factor analysis of the Montgomery Asberg Depression
Rating Scale (which has only two items negatively related to sleep and eating) in
251 BP-II MDE and 306 UP MDE outpatients found three factors in BP-II, one
including reduced sleep (negative) and reduced appetite (negative). This factor
was not found in UP (Benazzi, 1997a, 1999a, 2000a, 2001a, b; Benazzi and Akiskal,
2003a). Two studies, including only 10 BP-II and 28 BP-II, found no AD frequency
difference between BP and UP (Posternak and Zimmerman, 2002; Robertson
et al., 1996).

Is BP-II versus UP difference in AD frequency age-related?
UP MDE versus BP-II MDE had significantly higher age. BP-II MDE (n ¼ 187)
versus UP MDE (n ¼ 126) had a significantly higher frequency of AD (49.7%
versus 18.2%), persisting when controlled for age (Benazzi, 2003c). Findings
suggest that age may not be important for the BP-II versus UP difference in AD.
Mitchell et al. (2001) suggested instead that BP versus UP depression differences
could be related to an age difference.

Is there a difference in AD frequency in BP-II samples when probing for
past hypomania focused on overactivity?
Frequency of BP-II MDE versus UP MDE increased when probing for past
hypomania, with the SCID-CV focused more on overactivity than on mood.
This BP-II sample (n ¼ 103) had the same frequency of AD (47.5% versus
45.4%) found in a previous BP-II sample interviewed strictly following the
SCID-CV (n ¼ 251) (in the same setting and by the same interviewer), and a
significantly higher frequency of AD versus UP MDE (n ¼ 65, 16.9%: Benazzi and
Akiskal, 2003a). Findings support the usefulness of probing for overactivity when
assessing past hypomania, as found by Angst et al. (2003) and Akiskal et al. (2001).
The focus on past overactivity was also supported by a factor analysis study of past
hypomania using the Mood Disorder Questionnaire (MDQ) (Hirschfeld et al.,
2000) in 181 remitted BP-II MDE and UP MDE, which found two factors, one of
which had only overactivity items (Benazzi and Akiskal, 2003b).

Is AD frequency still higher in BP-II versus UP when BP-II had a short hypomania?
In BP-II MDE with a history of hypomania lasting less than 4 days, frequency of
AD was significantly higher in comparison with UP MDE (Benazzi, 2001c),
136 F. Benazzi


. 23
( 68 .)