Body Temperature and Mood Variations during Forced
Desynchronization in Winter Depression:
A Preliminary Report
Kathelijne M. Koorengevel, Domien G.M. Beersma, Marijke C.M. Gordijn,
Johan A. den Boer, and Rutger H. van den Hoofdakker
Background: It has been suggested that certain abnormalities (e.g., in phase or amplitude) of the circadian
pacemaker underlie seasonal affective disorder.
Methods: One male seasonal affective disorder patient
(blind to the study design) participated in two 120-hour
forced desynchrony experiments and was subjected to six
20-hour days, once during a depressive episode and once
after recovery. Core body temperature was continuously
measured. During wakefulness, the Adjective Mood Scale
was completed at 2-hour intervals.
Results: Sleep–wake as well as pacemaker-related variations of mood were found, both when the subject was
depressed and when he was euthymic. Compared with
recovery, during the depressive episode the circadian
temperature minimum and the circadian mood variation
showed phase delays of approximately 1 and 2 hours,
respectively.
Conclusions: The data of this first seasonal affective
disorder patient, participating in forced desynchrony experiments, may indicate a phase delay of the circadian
pacemaker during a seasonal affective disorder episode.
Biol Psychiatry 2000;47:355–358 © 2000 Society of Biological Psychiatry
Key Words: Seasonal affective disorder, mood, circadian
rhythms, phase shift, forced desynchronization, light
therapy
Introduction
T
he circadian pacemaker, or biological clock, is localized in the suprachiasmatic nucleus of the brain and is
known to regulate seasonal changes in a variety of animal
behaviors (e.g., reproduction and hibernation). It serves as
From the Department of Biological Psychiatry, Psychiatric University Clinic
(KMK, DGMB, JAdB, RHvdH) and Zoological Laboratory, University of
Groningen (DGMB, MCMG), Groningen, The Netherlands.
Address reprint requests to Kathelijne M. Koorengevel, Department of Biological
Psychiatry, Room 1.33, Psychiatric University Clinic, P.O. Box 30.001, 9700
RB Groningen, The Netherlands.
Received February 1, 1999; revised May 20, 1999; revised August 2, 1999;
accepted August 10, 1999.
© 2000 Society of Biological Psychiatry
an internal timing mechanism with a period of approximately 24 hours. The environmental light– dark cycle is
the most important cue for synchronization of the circadian system to the habitual 24-hour sleep–wake cycle.
Therefore, the seasonal manifestation of depressive symptoms and the efficacy of light therapy might suggest
involvement of the human circadian pacemaker in seasonal affective disorder (SAD). Core body temperature
exhibits a pronounced 24-hour variation and is often the
object of study in human circadian rhythm research;
however, the effects of daily activities on core body
temperature mask its overt circadian rhythm. There are
two ways to measure the characteristics of the circadian
pacemaker, by either experimentally controlling or mathematically correcting for masking effects on core body
temperature. First, under the controlled conditions of
wakefulness with the subject in supine posture, low light
intensity, and isocaloric snacks of the constant routine
protocol, circadian amplitude and phase can be accurately
determined. Such studies have shown that the dim-light
melatonin onset and certain characteristics of the circadian
temperature rhythm were phase delayed in female SAD
patients compared with those in controls (Dahl et al 1993;
Wirz-Justice et al 1995). A disadvantage of constant
routine protocols is that subjects are deprived of sleep as
they have to stay awake for more than 24 hours. As in
nonseasonal major depressive disorder, in SAD this deprivation may lead to changes in mood (Graw et al 1998),
which in turn can have an influence on overt rhythms.
Therefore, the second method of forced desynchronization
could be more appropriate. When subjects are forced to be
awake and sleep on an artificial day length (e.g., 20 or 28
hours), outside the range of entrainment of the human
circadian system, the circadian pacemaker characteristics
can be separated from those dependent on sleep or time
awake, without sleep deprivation interfering. Healthy
subjects subjected to a forced desynchrony protocol show
a complex and nonadditive influence of circadian phase
and duration of prior wakefulness on subjective mood
(Boivin et al 1997). We thus subjected one male SAD
0006-3223/00/$20.00
PII S0006-3223(99)00225-5
356
K.M. Koorengevel et al
BIOL PSYCHIATRY
2000;47:355–358
patient to a forced desynchrony protocol to study the
pacemaker characteristics and the circadian and sleep–
wake-dependent variations of mood, during a depressive
episode and after good response to light treatment in the
winter of 1997–1998. To the best of our knowledge, this is
the first application of this technique to depression.
Methods and Materials
The subject, who met the DSM-IV criteria for recurrent major
depression with seasonal pattern (American Psychiatric Association 1994) and the Rosenthal criteria for SAD (Rosenthal et al
1984), was a 54-year-old male outpatient, known to respond
favorably to light treatment. He was in good physical health, did
not use any medication at least 1 month before the experiments
(especially no psychotropic medication in the previous 6 months)
and smoked approximately 15 cigarettes a day. The subject gave
written informed consent to the protocol, approved by the
Medical Ethics Committee of the Academic Hospital Groningen.
During the winter season, the Beck Depression Inventory
(BDI) (Beck et al 1961) and the Structured Interview Guide for
the Hamilton Rating Scale of Depression (HAM-D), Seasonal
Affective Disorder self-rating version (SIGH-SAD-SR; Williams
et al 1992) consisting of the 21-item HAM-D and an additional
eight-item atypical symptom scale (ATYP) were completed
weekly. A BDI score of $16 was required for inviting the patient
to participate during the depressive episode. A BDI score of ,6
was required for participation following response to light treatment. After 4 baseline days at home, in which sleep was
scheduled between midnight and 8:00 AM (verified by wrist
actometry), the patient entered a temporal isolation unit in which
no information on the time of day was available. Following one
habituation night in this temporal isolation unit, he was subjected
to a 120-hour forced desynchrony protocol (six 20-hour “days”;
see Hiddinga et al 1997). The subject, blind to clock time and the
timing of the sleep–wake schedule, lived under an artificial
20-hour light– dark schedule consisting of 13.5 hours of wakefulness in dim light (,10 lux) and 6.5 hours of darkness, during
which he had to be in bed. Core body temperature data were
stored at 1-min intervals. Every 2 hours the Adjective Mood
Scale (AMS; Von Zerssen 1986)—ranging from 0, not depressed, to 56, severely depressed—was completed to monitor
mood. The circadian variation of mood was computed by
calculating the mean score of all AMS questionnaires completed
at specific real clock times. To adjust the computed circadian
mood variation for the sleep–wake-related variation of mood, the
mean sleep–wake-dependent variation at specific times of the
subjective day was subtracted from the raw AMS scores first. In
a similar way, the sleep–wake-related variation of mood was
obtained by calculating the mean scores of AMS ratings obtained
per specific time of the subjective day. In turn, before computing
the sleep–wake-related variation of mood and SDs, the mean
circadian variation at specific real clock times was subtracted
from the raw AMS data to adjust for the circadian-related mood
variation. During the 4 baseline days and in temporal isolation, a
subjective sleep quality questionnaire, with scores ranging from
Figure 1. Double plot of the circadian variation of subjective
mood (mean Adjective Mood Scale [AMS] score at specific
clock times and SD), monitored every 2 hours during wakefulness, in one male seasonal affective disorder patient subjected to
two 120-hour forced desynchrony experiments, once during a
depressive episode (black circles) and once 6 weeks after
recovery upon 1 week of light therapy (open squares). Linear
regression analysis, using a 24-hour sine function, revealed a
significant circadian variation for both curves (in the depressive
state R 5 .445, p 5 .014; in the remitted state R 5 .394, p 5
.037).
0 (high quality) to 14 (low quality) was completed after each
period of sleep (Mulder-Hajonides van der Meulen et al 1980).
Results
The first experiment was conducted in November 1997.
At baseline, the subject’s BDI score was 15; after the
experiment the score was 18 (SIGH-SAD-SR 24 and 32;
HAM-D 17 and 22, ATYP 7 and 10). The BDI score
was 7 (SIGH-SAD-SR 11; HAM-D 8, ATYP 3) 7 days
after finishing subsequent light treatment (5 consecutive
days of 30-min 10,000-lux light in the morning). The
second experiment was conducted in January 1998,
when the BDI score was 5. After the experiment, it was
3 (SIGH-SAD-SR 9 and 18; HAM-D 5 and 11, ATYP 4
and 7). Until the end of winter, the BDI score remained
5 or less (SIGH-SAD-SR # 10). The temperature data
were analyzed by an iterative method (Hiddinga et al
1997). It showed an endogenous circadian rhythm
period (t) of 23 hours 38 min during the depressive
episode and one of 24 hours 4 min after recovery. The
circadian temperature minimum, defined as the midpoint between the upward and downward crossings
through the average value of the temperature curve
smoothed by spline approximation, was reached at 5:13
AM and 4:16 AM on the first day of the forced desynchrony protocol, respectively. The absolute AMS scores
Forced Desynchronization in SAD
BIOL PSYCHIATRY
2000;47:355–358
357
Discussion
Figure 2. Double plot of the sleep–wake-related variation of
subjective mood (mean Adjective Mood Scale [AMS] score and
SD), monitored every 2 hours during 13.5 hours of wakefulness
of the 20-hour subjective day, in one male seasonal affective
disorder patient, subjected to two 120-hour forced desynchrony
experiments, once during a depressive episode (black circles) and
once 6 weeks after recovery upon 1 week of light therapy (open
squares).
(mean 6 SD) during the experiment were 24 6 15 in
the depressed state and 8 6 7 after recovery, with
ranges of 1–53 and 0 – 40. The circadian variation of
mood (mean and SD) is double plotted (Figure 1). In
both experiments, the t values (as determined from
body temperature data) were close to 24 hours. Over the
entire protocol, the difference with a 24-hour period did
not accumulate to more than 2 hours, which is the
interval between successive mood ratings. A linear
regression analysis, using a 24-hour sine function, was
thus applied to the circadian variation of mood observed
in both conditions. This analysis revealed a significant
circadian variation, both in the depressed state (R 5
.445, p 5 .014) and in the remitted state (R 5 .394, p 5
.037). Both curves reveal peaks in depression scores in
the early morning hours. Cross-correlation suggests that
the circadian variation of mood during depression is
phase delayed by approximately 2 hours as compared
with recovery. Figure 2 presents the sleep–wake-related
variation of mood as a function of time since awakening
(mean and SD). In both conditions, a steady deterioration of mood is apparent, without indications of a phase
difference. During the 4 baseline days, the sleep quality
score (mean 6 SD) was 6 6 2 at times of the depressive
episode and 2 6 2 in the remitted state, with ranges of
4 – 8 and 0 –5, respectively. In temporal isolation, the
sleep quality score (mean 6 SD) was 6 6 5 (range
1–13) in the depressed and 6 6 4 (range 2–13) in the
remitted state.
Before and after the forced desynchrony experiments, the
BDI scores did not reveal systematic trends in depressed
mood in both conditions. The observed changes in SIGHSAD-SR scores during the experiment are primarily a
consequence of an accompanied increased appetite and a
low sleep quality.
The t values obtained from the two experiments with
this SAD patient lie within the range of t values found in
healthy subjects (Hiddinga et al 1997). Czeisler et al
(1999) recently reported t values of 24.18 6 0.04 hours in
both young and older subjects participating in a forced
desynchrony study. A comparison with Hiddinga et al
(1997) and Czeisler et al (1999) reveals that the difference
in t values found in the depressed and remitted state is
likely due to the less accurate 120-hour forced desynchrony protocol and probably not so much to intra- or
interindividual differences. As a consequence, a larger
group of subjects is required to demonstrate systematic
differences between SAD patients and healthy controls.
We observed circadian and sleep–wake-related variations of mood. The magnitudes of the variations are
similar for the two conditions. The demonstrated nadir of
circadian mood modulation in the early morning hours and
the deterioration of mood during cumulative wakefulness
are consistent with observations in healthy subjects
(Boivin et al 1997). Compared with recovery, there is
support for the existence of a phase delay of the circadian
temperature minimum and of subjective mood during the
depressive episode. The subject received 1 week of light
treatment after participation in the first experiment. The
second experiment was conducted 6 weeks after finishing
this treatment week. Thus, the results cannot be considered
as a masking effect of light per se; however, in temporal
isolation, subjective sleep quality showed a wide variation
in both experiments. This may have influenced the computed periods of the pacemaker, the observed phase of the
circadian temperature minimum, and the sleep–wake- as
well as the pacemaker-related mood variation. More data
are presently being collected to clarify whether the manifestation of SAD and the efficacy of its treatment are
associated with phase shifts of the circadian pacemaker.
The funding of this study was provided by Nederlandse Gasunie B.V.,
Medical Faculty of the University of Groningen, Academic Hospital
Groningen, Ministry of Health, and National Fund of Mental Health.
References
American Psychiatric Association (1994): Diagnostic and Statistical Manual of Mental Disorders, 4th ed. Washington, DC:
American Psychiatric Assocation Press.
358
BIOL PSYCHIATRY
2000;47:355–358
Beck AT, Rush AJ, Shaw BF, Emergy G (1961): An inventory
for measuring depression. Arch Gen Psychiatry 4:561–571.
Boivin DB, Czeisler CA, Dijk DJ, Duffy JF, Folkard S, Minors
DS, et al (1997): Complex interaction of the sleep-wake cycle
and circadian phase modulates mood in healthy subjects. Arch
Gen Psychiatry 54:145–152.
Czeisler CA, Duffy JF, Shanahan TL, Brown EN, Mitchell JF,
Rimmer DW, et al (1999): Stability, precision, and near-24hour period of the human circadian pacemaker. Science
284:2177–2181.
Dahl K, Avery DH, Lewy AJ, Savage MV, Brengelmann GL,
Larsen LH, et al (1993): Dim light melatonin onset and
circadian temperature during a constant routine in hypersomnic winter depression. Acta Psychiatr Scand 88:60 – 66.
Graw P, Haug H-J, Leonhardt G, Wirz-Justice A (1998): Sleep
deprivation response in seasonal affective disorder during a
40-h constant routine. J Affect Disord 48:69 –74.
Hiddinga AE, Beersma DG, van den Hoofdakker RH (1997):
Endogenous and exogenous components in the circadian
variation of core body temperature in humans. J Sleep Res
6:156 –163.
K.M. Koorengevel et al
Mulder-Hajonides van der Meulen WREH, Wijnberg JR, Hollander
JJ, De Diana IPF, van den Hoofdakker RH (1980): Measurement
of subjective sleep quality. Abstract presented at the Fifth
European Sleep Congress of the ESRS, Amsterdam.
Rosenthal NE, Sack DA, Gillin JC, Lewy AJ, Goodwin FK,
Davenport Y, et al (1984): Seasonal affective disorder. A
description of the syndrome and preliminary findings with
light therapy. Arch Gen Psychiatry 41:72– 80.
Von Zerssen D (1986): Clinical self-rating scales of the Munich
Psychiatric Information System. In: Sartorius, Ban TA, editors. Assessment of Depression. Berlin: Springer Verlag,
270 –303.
Williams JBW, Link MJ, Rosenthal NE, Amira L, Terman M
(1992): Structured Interview Guide for the Hamilton Rating
Scale-Seasonal Affective Disorder Version (SIGH-SAD;
SIGH-SAD-SR, Self-Rating Version). New York: New York
State Psychiatric Institute.
Wirz-Justice A, Kräuchi K, Brunner DP, Graw P, Haug H-J,
Leonhardt G, et al (1995): Circadian rhythms and sleep
regulation in seasonal affective disorder. Acta Neuropsychiatr
7:41– 43.