492
Electroencephalography and clinical Neurophysiology, 1990, 75: 492-499 Elsevier Scientific Publishers Ireland, Ltd.
EEG 89528
Effect of partial sleep deprivation on sleep stages and E E G power spectra: evidence for n o n - R E M and R E M sleep homeostasis Daniel P. Brunner, Derk-Jan Dijk, Irene Tobler and Alexander A. Borbrly Institute of Pharmacology, University of Zurich, Zurich (Switzerland)
(Accepted for publication: 8 October 1989)
Summary The effect of repeated partial sleep deprivation on sleep stages and sleep EEG parameters was investigated in young subjects. After 2 baseline nights (B1, B2) of 7.5 h, sleep was restricted for 2 nights (D1, D2) to the first 4 h of the habitual bedtime period. Two recovery nights (R1, R2) with 7.5 h sleep followed. During the deprivation nights, stages 1 and 2 and REM sleep were reduced, while slow wave sleep (SWS; stages 3 and 4) was not significantly affected. However, the time integral of EEG power density in the range of 0.75-4.5 Hz (slow wave energy) was reduced. In the recovery period, SWS showed an enhancement in R1, and REM sleep showed a rebound in R1 and R2. An increase of REM sleep in the early part of the sleep period was evident in R1. Sleep latency was reduced in D2, R1 and R2. In accordance with the 2-process model of sleep regulation, EEG power density in non-REM sleep in the range of 0.75-4.5 Hz (slow wave activity) was only slightly higher in D2 and R1 than in baseline. An enhancement of slow wave activity in REM sleep was present in D2. Power density in the frequency range of 13-16 Hz was reduced in non-REM sleep (R1), SWS (R2) and stage 2 (R1). The results show (1) that the moderate reduction of slow wave energy in the deprivation nights induces only a minor enhancement of slow wave activity during recovery sleep; and (2) that a REM sleep deficit gives rise to an immediate rebound when 'slow wave pressure' is low. A comparison of the changes in the EEG power spectrum in non-REM sleep with published data on the effects of total sleep deprivation suggests that an increase of ' REM sleep pressure' has repercussions on the EEG in non-REM sleep. Key words: Sleep deprivation; Sleep homeostasis; REM sleep; Sleep EEG; EEG spectral analysis T o t a l sleep d e p r i v a t i o n e n h a n c e s the p r o p e n s i t y for sleep i n i t i a t i o n a n d n o n - R E M sleep i n t e n s i t y . T h e f o r m e r effect is e v i d e n t f r o m the r e d u c t i o n of sleep l a t e n c y after a n i g h t w i t h o u t sleep ( C a r s k a d o n a n d D e m e n t 1979; B o r b r l y et al. 1985), the l a t t e r effect f r o m the i n c r e a s e o f slow w a v e sleep ( S W S ) ( W i l l i a m s et al. 1964; A g n e w et al. 1967; K a l e s et al. 1970; W e b b a n d A g n e w 1971; M o s e s et al. 1975) a n d E E G slow w a v e a c t i v i t y ( B o r b r l y et al. 1981; ,~tkerstedt a n d G i l l b e r g 1986; D i j k et al. 1987; B r u n e t et al. 1988) d u r i n g r e c o v e r y sleep. I n c o n t r a s t to n o n - R E M sleep, R E M sleep is little a f f e c t e d w h e n sleep has b e e n d e p r i v e d for n o t
Correspondence to: Prof. A. Borbrly, Institute of Pharmacology, University of Zurich, Gloriastrasse 32, CH-8006 Zurich (Switzerland).
m o r e t h a n a single n i g h t ( N a k a z a w a et al. 1978; B o r b r l y et al. 1981). A f t e r a l o n g e r p e r i o d w i t h o u t sleep a R E M sleep r e b o u n d is e v i d e n t (Berger a n d O s w a l d 1962; W i l l i a m s et al. 1964; G u l e v i c h et al. 1966; K a l e s et al. 1970). W h e n sleep d u r a t i o n was r e s t r i c t e d for several nights, the a m o u n t of S W S r e m a i n e d u n c h a n g e d or e v e n i n c r e a s e d , w h e r e a s R E M sleep was d r a s t i cally r e d u c e d ( W e b b a n d A g n e w 1965, 1974; Dem e n t a n d G r e e n b e r g 1966). T h e s e r e s u l t s d e m o n strate the h i g h p r i o r i t y of S W S i n c o m p a r i s o n to R E M sleep. It has b e e n c o n c l u d e d t h a t S W S is m o r e f i n e l y r e g u l a t e d t h a n R E M sleep ( B o r b r l y 1982), a n d t h a t it is the i n c r e a s e d ' s l o w w a v e p r e s s u r e ' w h i c h p r e v e n t s the r e b o u n d of R E M sleep after sleep d e p r i v a t i o n ( W i l h a m s et al. 1964; W e b b a n d A g n e w 1965; T i l l e y a n d W i l k i n s o n 1984). T h u s , a c c o r d i n g to this i n t e r p r e t a t i o n , the non-REM-REM sleep i n t e r a c t i o n o b s c u r e s the
0013-4649/90/$03.50 © 1990 Elsevier Scientific Publishers Ireland, Ltd.
PARTIAL SLEEP DEPRIVATION AND SLEEP HOMEOSTASIS manifestation
of REM
sleep homeostasis
has been demonstrated by selective REM d e p r i v a t i o n e x p e r i m e n t s ( D e m e n t e t al.
493
which
Methods
sleep 1966;
Subjects
A g n e w e t al. 1967). T h e a i m o f t h e p r e s e n t s t u d y was to gain further insight into the processes underlying the regulation of the two major sleep s t a t e s a n d t h e i r i n t e r a c t i o n s . T o m i n i m i z e t h e inh i b i t o r y e f f e c t o f S W S o n R E M sleep, a n e x p e r i mental protocol was selected in which SWS was reduced by only a small extent whereas REM sleep was markedly curtailed. In contrast to previous studies, we did not restrict the analysis to v i s u a l l y s c o r e d s l e e p stages, b u t q u a n t i f i e d t h e sleep EEG by spectral analysis.
Sleep records were obtained from 9 paid, young, m a l e s u b j e c t s ( 2 2 - 2 7 y e a r s , m e a n 24.1) o n 7 c o n secutive nights. The subjects were in good health and did not report any sleep disorders. They were requested to refrain from sleeping during the day (which was verified by wrist activity monitoring), and from using alcohol or drugs throughout the experiment. The protocol consisted of an adaptat i o n n i g h t ( A D ) , 2 b a s e l i n e n i g h t s (B1, B2), 2 partial sleep deprivation nights (D1, D2), and 2 r e c o v e r y n i g h t s ( R 1 , R 2 ) (see Fig. 1). T h e s u b j e c t s
TABLE I Sleep parameters in minutes (mean with S.E.M. in parentheses) for the baseline nights (B1, B2), the nights with sleep restriction to 4 h (D1, D2) and the recovery nights (R1, R2). Sleep latency, latency from lights off to stage 2; REM sleep (REMS) latency, latency of REMS from stage 2. Slow wave energy, i.e., EEG power density (0.75-4.5 Hz), integrated over time, is expressed as a percentage of the baseline mean (100% = (B1 + B2)/2). Night
F
B1
Sleep latency REMS latency REMS in first 2 h
B2
12.5 (1.7) 67.3 (3.5) 7.6 (1.6)
12.1 (2.4) 70.3 (11.2) 11.4 (2.5)
D1
D2
R1
12.7 (3.6) 75.4 (10.0) 12.1 (4.7)
8.1 (1.9) * * 61.7 (3.3) 9.9 (2.4)
R2
6.2 (1.3) * * 58.2 (8.1) a 20.5 (4.2) * *
8.8 (1.3) * * 60.6 (5.3) 12.9 (2.1)
18.19 4.79 13.68
Entire sleep period Total sleep time Waking after ~ieep onset Stage 1 Stage 2 Stage 3 Stage 4 REMS SWS (stage 3 + 4) Slow wave energy
430.8 (3.3)
433.1 (3.4)
240.3 (1.5) * *
239.4 (0.4) * *
434.1 (3.7)
3.9 (0.8) 39.5 (3.4) 216.1 (9.7) 47.1 (4.7) 33.8 (5.5) 94.4 (5.7) 80.8 (4.7) 103.8 (2.9)
3.1 (1.6) 40.7 (3.1) 216.1 (8.4) 41.1 (4.0) 30.6 (6.0) 104.7 (4.3) 71.7 (6.9) 96.2 (2.9)
2.7 (1.0) 18.1 (1.4) 114.1 (6.7) 36.6 (4.7) 33.7 (8.1) 37.8 (4.7) 70.3 (6.6) 80.2 (3.3)
0.7 (0.5) 15.3 (2.0) 109.2 (7.5) 44.8 (2.7) 37.9 (7.4) 32.2 (3.2) 82.7 (5.8) 87.3 (3.3)
0.6 (0.5) 29.4 (2.2) 184.2 (9.3) 53.4 (4.9) 41.8 (8.1) 125.2 (5.2) 95.2 (8.1) 104.6 (7.1)
17.7 (1.3) 115.2 (4.2) 35.6 (2.9) 30.2 (5.9) 39.2 (4.8) 65.8 (5.9) 96.2 (3.7)
17.9 111.4 36.4 33.7 37.6 70.1 103.9
** **
** **
** ** **
** **
** ** ** * ** *
433.6 (3.0)
32.10
3.9 (0.3) 39.0 (3.3) 188.1 (6.5) * * 51.9 (3.9) 34.5 (8.1) 120.1 (4.1) * * 86.4 (8.2) 99.6 (3.8)
15.19 33.70 39.20 13.79 2.02 39.67 17.25 22.90
16.0 100.7 47.2 32.6 41.7 79.8 103.2
11.47 18.99 10.11 3.19 9.04 16.41 8.05
First 4 h of sleep period Stage 1 Stage 2 Stage 3 Stage 4 REMS SWS (stage 3 + 4) Slow wave energy
20.7 109.9 41.3 31.2 33.6 72.6 103.8
(2.9) (5.2) (4.7) (4.4) (4.8) (2.2) (3.7)
(1.4) (6.4) (4.6) (8.1) (4.7) (6.6) (3.4)
15.3 (2.0) 108.8 (7.6) 44.4 (2.8) 37.9 (7.4) 31.4 (3.3) 82.3 (5.6) * * 114.2 (4.6)
11.4 (1.7) * * 92.9 (6.2) * * 44.8 (5.2) 40.4 (7.6) 48.8 (4.6) 85.3 (5.0) * * 106.8 (6.5)
(1.6) (4.9) * (4.6) (7.5) (4.3) (6.6) (4.5)
F = Friedman 2-way non-parametric ANOVA for repeated measures: P < 0.05 for F>11.1; P < 0.01 for F>15.1. * P < 0.05; * * P < 0.01; significant difference from baseline mean (Wilcoxon matched-pairs signed-rank test, 2-tailed). Between B1 and B2 no significant differences were found. a One sleep onset REM episode included.
494 went to bed at their habitual bedtime (23-24 h). They were awakened after 7 h or 7.5 h sleep on the AD, B and R nights (the sleep duration was individually adjusted to the habitual sleep duration), and after a sleep duration of 4 h on DI and D2. Sleep duration was determined on-line from the polygraph records. After the partial sleep deprivation nights, the subjects were asked to maintain their usual day-time activities, but to avoid sleep-inducing situations.
D.P. BRUNNER ET AL.
Bq
!
m
m
B2
Recording and data analysis EEG ( C 3 / F z , and C4/Fz), E O G and E M G were recorded continuously on polygraph paper (Grass 78D; paper speed 10 mm/sec). The highpass filter setting of the EEG channels was 0.1 Hz which corresponds to a time constant of 0.6 sec. The amplifier output of all signals was digitized on-line (sampling rate 128 Hz) by a signal processor card (Texas Instruments, TMS-320-10) installed in a personal computer (Olivetti M24). The EEG signals were low-pass filtered (25 Hz, 24 d B / o c t ) prior to A / D conversion, and then subjected to spectral analysis by a Radix 8 Fast Fourier Transform (FFT) routine implemented on the signal processor card. Power spectra were computed for consecutive 4 sec epochs by applying a rectangular window. If the EEG signal reached the maximum or minimum level of the A / D converter, the corresponding 4 sec spectrum was excluded. This routine, which eliminated 0.4% of all 4 sec epochs, served to prevent contamination of the data set by spurious data arising from movement artifacts or from saturation of the amplifier. Mean values over adjacent frequencies were computed to collapse the 0.25 Hz data into 0.5 Hz bins for frequencies between 0.25 and 5.0 Hz, and into 1 Hz bins for frequencies between 5.25 and 25.0 Hz. One minute mean spectra were computed from 15 consecutive 4 sec spectra, and stored on the hard disk of the personal computer. Sleep stages were scored for 30 sec epochs according to the criteria of Rechtschaffen and Kales (1968). Since the bipolar derivations used in the present study probably yielded signals with smaller amplitudes than the recommended C 3 / A 2 , C 4 / A ] derivations, the amounts of stages 3 and 4 may have been slightly underestimated. However,
m
m
•
•
m
m
DI
!
m
D2
I
m
I
R1
I
23
m
•
m
•
I
24
I
01
m
m
m 1
I
02 03 Time o¢ doy
m
m I
04
m I
05
I
06
Fig. 1. Time course of EEG slow wave activity (power density in the 0.75-4,5 Hz band) on 6 consecutive nights of 1 subject: 2 baseline nights (B1, B2), 2 partial sleep deprivation nights (D1, D2) and 2 recovery nights (R1, R2). Data are plotted from lights off until the termination of sleep. REM sleep episodes are indicated by horizontal bars. Calibration (entire scale) of power density: 66.3/~V2/Hz.
495
P A R T I A L SLEEP D E P R I V A T I O N A N D SLEEP HOMEOSTASIS 10
the effects on relative E E G spectra are largely independent of the distance between the electrodes, since similar results were obtained with a C 3 / A 2 derivation (Dijk and Beersma 1989). The scores were entered into the computer. Two consecutive 30 sec sleep scores corresponded to a 1 min EEG spectrum. For the computation of power density within a specific sleep stage (Fig. 3; SWS was considered as a single stage), only 1 min epochs with 2 identical scores were used. Thus 30 sec epochs scored as movement time were not incorporated in the spectral data. R E M - n o n - R E M sleep cycles were defined according to the criteria of Feinberg and Floyd (1979), R E M sleep episodes according to Czeisler et al. (1980).
Level of S
1158~
lOOX 05
I O0 B1
160
B2
D1
% of baseline mean
D2
R1
R2
(first 4h)
150 140
SWA
130
(model)
SWA (data)
120
REM m sleep
J,
110 100 90
Results
Sleep parameters Restricting sleep to 4 h reduced the sleep latency in the following night (D2, R1). A significant reduction of sleep latency was still present in the second recovery night (Table I). During the 2 short nights (D1, D2), the amounts of stage 1, stage 2 and R E M sleep were markedly below baseline, whereas stages 3 and 4 were little changed (Table I). In comparison to the first 4 h of baseline sleep, a significant increase of SWS was seen in the second deprivation night. The first recovery night was characterized by a reduction in waking, stage 1 and stage 2, and an increase in stage 3, SWS and R E M sleep. The decrease in stages 1 and 2 and the increase in SWS were significant also in the first 4 h of sleep. R E M sleep in the first 2 h was massively increased. The reduction of stage 2 and the increase of R E M sleep were still present in the second recovery night (Table I, entire sleep period).
Spectral analysis of the sleep EEG Fig. 1 illustrates for 1 subject the time course of E E G power density in the 0.75-4.5 Hz range (slow wave activity) on 6 consecutive nights. R E M sleep episodes are indicated by horizontal bars. Slow wave activity is typically high in the first 1 - 2 n o n - R E M sleep episodes, and lower in subsequent
80 70
BL D1 D2 R1 R2
BL D1 D2 I~1 R2
BL D1 D2 R1 R2
Fig. 2. Upper panel: time course of process S during the experimental protocol as simulated with the 2-process model of sleep regulation. Sleep episodes are indicated by horizontal bars and shading. Lower panel: sleep parameters in the first 4 h of sleep. Left: prediction of slow wave activity (SWA) by the model; middle: recorded SWA in n o n - R E M sleep; right: a m o u n t of REM sleep. BL, mean of the 2 baseline nights; other abbreviations as in Fig. 1.
episodes. For this reason, the shortening of sleep duration to 4 h reduced slow wave energy (the time integral of slow wave activity) only to 80.2% (D1) and to 87.3% (D2) of baseline (Table I). The changes of slow wave activity throughout the experiment were simulated by the 2-process model (Daan et al. 1984). As shown in Fig. 2 (upper panel) it is assumed in the model that process S, the marker of n o n - R E M sleep homeostasis derived from slow wave activity, declines exponentially during sleep (shaded area) and increases according to a saturating exponential function during waking. The model predicts that after reducing sleep to 4 h, the initial level of S at sleep onset is raised in D2 to 115.8% and in R1 to 117.8% of control. However, the first recovery night is sufficient to bring back the initial level of S in R2 close to the control level. It should be noted that, because of the exponential function,
496
D.P. B R U N N E R ET AL. 215
BL
................... D 2
2.10,
t,, /' ',,,/,,,~/%,,
- - R 1 ......
~
R2
,/,/*"',,,,
205.
"6 2.00-
195"
REM sleep -
o0 c'(D "0
-
i
•
r
i
-
•
r
D2
D2
R1
2.05"
0 ("3
......./'"'..,,.
.(--., _.~#~=;, .............. ..""" ',,.....,....
, j~'=~: .......... .,..
2.00"
(.9 III
LU
195
SWS 190
R1 R2
"
~-/
i
-
5 •
p
,0
"
"
"
i
~
•
tage .
2o HZ
R 1
•
R2
.
e-o
.
~
.
.
.
.
~
.
.
.
~ ~
2o =
Hz
=
=
Fig. 3. EEG power density for the first 4 h of sleep in deprivation night 2 (D2), recovery night 1 (RI) and recovery night 2 (R2). For each subject and each frequency bin the mean value of the 2 baseline nights (first 4 h) is defined as 100%. The values are expressed relative to the corresponding baseline value. The data represent means (N = 9) of individual log-transformed values and are plotted at the upper fimit of each frequency bin. The frequency range in which the values differ significantly from the baseline mean (BL) are indicated below the abscissae ( P < 0.05; Wilcoxon matched-pairs signed-ranks test).
the change in the peak value of S at sleep onset is identical to the value integrated over the first 4 h of sleep. The lower panel of Fig. 2 contrasts the predictions of the model with the data. In accordance with the model, slow wave activity in nonR E M sleep is increased in D2 and R1. The empirical values are somewhat lower than the simulated values for D2 and R1, and somewhat higher for D1 and R2. However, the differences between the data and the simulations were not statistically significant ( P > 0.1 in all cases; paired t test). Taken together, the general time course of slow wave activity conforms to the prediction of the model. In the lower panel of Fig. 2, the different time courses of slow wave activity and visually scored R E M sleep are illustrated. Fig. 3 shows the effect of partial sleep deprivation on E E G power density in n o n - R E M sleep, R E M sleep, stages 3 and 4 and stage 2. The values are shown for the first 4 h of D2, R1 and R2, and are expressed relative to the corresponding mean baseline value. Power density in the low frequency
range tended to be above the baseline level in total non-REM sleep and R E M sleep. However, a significant elevation was present in n o n - R E M sleep only in the 4 - 5 Hz band of D2 and R1, and in R E M sleep in some low frequency bands of D2. In SWS the low frequency values were close to the baseline level or below it; in stage 2 the values in the frequency band of 0.75-2.0 Hz in R1 were significantly reduced. A depression in the 13.2516.0 Hz range was present in total n o n - R E M sleep and stage 2 of R1, and in the 17.25-20.0 Hz range in stage 2 of R2.
Discussion
N o n - R E M sleep homeostasis It has been demonstrated in previous experiments that in sleep restriction protocols the level of SWS is preserved, whereas R E M sleep is reduced (Webb and Agnew 1965, 1974; Dement and Greenberg 1966). The present experiment con-
PARTIAL SLEEP DEPRIVATION AND SLEEP HOMEOSTASIS
497
firms these observations and provides further evidence for the high priority of SWS in sleep regulation. However, in addition to determining SWS by visual scoring of the polygraph records, the E E G was subjected to all-night spectral analysis to obtain a quantitative measure of the sleep-related E E G changes. These measures were also used to investigate the homeostatic mechanisms underlying n o n - R E M sleep. The 2-process model of sleep regulation (Borbrly 1982; Daan et al. 1984) postulates that a sleep/wake-dependent process (process S) increases during waking and declines during sleep. The time course of S had been derived from the changes of E E G slow wave activity. Simulations of the present experimental schedule predicted a minor increase in slow wave activity during the second deprivation night and the first recovery night (Fig. 2 upper panel). The experimental data were in agreement with the prediction, since only a minor and non-significant enhancement of slow wave activity was present in the nights D2 and R1 (Fig. 2 lower panel). Spectral analysis by individual sleep stages showed that in contrast to the recovery from total sleep deprivation (Borbrly et al. 1981), slow wave activity within SWS was not enhanced (Fig. 3). Thus, the present partial sleep deprivation schedule did not intensify this sleep stage. At first sight it may be puzzling that only a small rise or even a reduction of slow wave activity is seen in stage 2 and SWS, whereas an increase is present in the entire non-REM sleep (Fig. 3). This apparent contradiction is resolved by recognizing that the absolute level of slow wave activity is much higher in SWS than in stage 2 (Borbrly et al. 1981). Thus, an increase in the amount of SWS enhances slow wave activity within non-REM sleep even though slow wave activity within SWS may be decreased. The increase of SWS in D2, R1 and R2 relative to stage 2 (Table I) therefore resulted in an enhanced slow wave activity in n o n - R E M sleep, although no such increase was present within the individual n o n - R E M sleep stages.
1978; Borbrly et al. 1981), it has been concluded that the regulation of n o n - R E M sleep is more finely ' t u n e d ' than that of R E M sleep (Borbrly 1982). However, total sleep deprivation experiments are not ideal for investigating the regulation of R E M sleep, because the high propensity for SWS may impede the manifestation of R E M sleep. The present experimental protocol was designed to minimize this interference since the limitation of sleep to 4 h was expected to have only a minor effect on SWS. The experiment revealed that a R E M sleep deficit of 131% incurred during 2 nights induces a potent compensatory response in the 2 recovery nights. In addition to the increase of R E M sleep over the entire night, the heightened ' R E M sleep pressure' is reflected also by the increase of this sleep stage in the first part of the night (Fig. 2). This effect was not yet present after the first deprivation night, indicating that a buildup of ' R E M sleep pressure' or a shift in the ' R E M / s l o w wave balance' is necessary for its manifestation. It should be noted that the R E M sleep deficit incurred in the 2 deprivation nights was not fully compensated by the excess in the 2 recovery nights, and it is possible that the enhancement of R E M sleep persisted in the following nights. A prolonged R E M sleep rebound has been observed also in other studies (Berger and Oswald 1962; Dement et al. 1966; Agnew et al. 1967) and could represent a typical feature of R E M sleep regulation. The prolonged compensatory response is in accordance with the hypothesis that a R E M sleep deficit is made up mainly in terms of R E M sleep time, whereas a non-REM sleep deficit can be compensated by an increase in n o n - R E M sleep intensity (Borbrly and Neuhaus 1979; Borbrly 1982). Furthermore, it is interesting to note that sleep latency was still significantly shortened in the second recovery night at a time when SWS no longer differed significantly from baseline. Thus, the processes controlling sleep latency may be more closely related to R E M sleep than to SWS.
R E M sleep homeostasis Since following 1 night of total sleep deprivation, SWS exhibits a marked rebound, whereas R E M sleep is little affected (Nakazawa et al.
Further effects on EEG spectra A detailed analysis of the E E G spectra revealed that sleep restriction induced more subtle changes than just a general rise of slow wave activity in
498
n o n - R E M sleep. It is informative to compare the effects within individual sleep stages (Fig. 3) with those observed after total sleep deprivation (Borb61y et ah 1981). As in the previous study, power density in the 13-16 Hz band was depressed in n o n - R E M sleep, SWS and stage 2. This effect was most prominent in the first recovery night, but tended to persist in the subsequent night. These E E G changes were also observed in recent sleep deprivation experiments (Dijk and Beersma 1989). Moreover, it was shown in a nap study that EEG activity in the 15 Hz band decreases as a function of prior waking (Dijk et ah 1987). Taken together these observations are consistent with the assumption that this EEG parameter is related to sleep homeostasis. However, unlike slow wave activity, it does not exhibit a clear trend across the night (Borb61y et al. 1981), and its changes appear to persist for a longer time after sleep deprivation. Activity in the 13 16 Hz range may therefore reflect long-term aspects of recovery which are not formalized by process S. As has been observed after total sleep deprivation (Borb61y et al. 1981), an increase in slow wave activity was present also in REM sleep (Fig. 3). It is important to note that the absolute level of slow wave activity is much lower in R E M sleep than in non-REM sleep and that only relative changes are shown in Fig. 3. Interestingly, the largest increase was present in D2 at a time when 'slow wave pressure' was enhanced whereas ' R E M sleep pressure' had only moderately risen. The further increase of the REM sleep deficit from D2 to R1 was not accompanied by a further increase of slow wave activity in R E M sleep. Thus, the changes of slow wave activity in R E M sleep and in n o n - R E M sleep were rather similar, and different from the changes in the amount of R E M sleep and the R E M sleep deficit. These observations, in conjunction with previous results (Borb61y et al. 1981; Dijk and Beersma 1989), indicate that a common homeostatic process modulated the slow wave activity in non-REM sleep and R E M sleep. When comparing the effects of total and partial sleep deprivation on slow wave activity in nonR E M sleep, it is evident that not only the extent but also the pattern of enhancement differs. In contrast to the marked peak in the 1 2 Hz band
D.P. BRUNNIER ET AL.
after total sleep deprivation (Borb61y et ah 1981; Dijk et al. 1987) and selective slow wave deprivation (Dijk and Beersma 1989), the rise induced by partial sleep deprivation encompasses rather uniformly a broad frequency range (Fig. 3). This difference in pattern may be simply due to a lower 'sleep pressure.' However, this is unlikely to be the case because the 1 - 2 Hz peak is present even in the cycle-to-cycle changes of a baseline night. Alternatively, the predominant REM sleep deprivation and the activation of R E M sleep homeostasis may exert a depressing influence on power density in the lowest frequency range and thus eliminate the typical 1 - 2 Hz peak. This interpretation could also apply to a recent observation based on period-amplitude analysis in which, during recovery from a partial sleep deprivation schedule (i.e., sleep restriction to the first 100 min of sleep), the wave density in the 0 - 3 Hz range was increased, whereas the EEG amplitude was not affected (Feinberg et ah 1988). Our own preliminary analysis of the EEG amplitude is consistent with the assumption that an enhanced ' R E M sleep pressure' exerts a depressing influence on maximal E E G amplitude. The depression of activity at the lowest end of the stage 2 spectrum in R1 could also be explained along these lines. Although this assumption could account for the changes in the 1 - 2 Hz range of the power spectra, firmer evidence is needed in support.
Concluding comments In studies of sleep regulation, non-REM sleep has been in the focus of interest for several reasons: (1) it represents 75-80% of total sleep time; (2) the proportion of slow wave activity within n o n - R E M sleep appears to be an easily accessible measure of n o n - R E M sleep intensity; and (3) slow wave activity is homeostatically regulated. These considerations have led to the formulation of the 2-process model of sleep regulation where one of the constituent processes is directly derived from slow wave activity. R E M sleep regulation was not incorporated into the model except for a qualitative description of its circadian c o m p o n e n t (Borb61y 1982). The present study provides further evidence for the homeostatic component underlying REM sleep regulation. Moreover, it supports
PARTIAL SLEEP DEPRIVATION AND SLEEP HOMEOSTASIS the assumption that 'slow wave pressure' exerts an inhibitory effect on sleep (Webb
and
the
Agnew
manifestation of REM 1965;
Dement
et
al.
1966; Borb61y 1982). I n a n a t t e m p t to a c c o u n t f o r the REM-non-REM sleep cyclicity, we p r o p o s e d in t h e m o d e l a r e c i p r o c a l i n t e r a c t i o n b e t w e e n p r o cess S and a REM sleep controlling process, analo g o u s to t h e n e u r o p h y s i o l o g i c a l h y p o t h e s i s o f M c C a r l e y a n d H o b s o n (1974). T h e p r e s e n t r e s u l t s l e n d f u r t h e r c r e d e n c e to a n i n t e r a c t i o n b e t w e e n t h e 2 s l e e p s t a t e s . It is e v e n p o s s i b l e t h a t t h e rise in ~ R E M s l e e p p r e s s u r e ' h a s r e p e r c u s s i o n s o n t h e E E G in n o n - R E M sleep. A n e x t e n s i o n o f t h e 2 - p r o c e s s m o d e l will b e r e q u i r e d t o a c c o m m o d a t e these new aspects.
This study was supported by the Swiss National Science Foundation, Grants 3.234-0.85 and 31.25634.88.
References Agnew, HW., Webb, W.B. and Williams, R.L. Comparison of stage four and 1-REM sleep deprivation. Percept. Motor Skills, 1967, 24: 851-858. ,~kerstedt, T. and Gillberg, M. Sleep duration and the power spectral density of the EEG. Electroenceph. clin. Neurophysiol., 1986, 64: 119-122. Berger, R.J. and Oswald, 1. Effects of sleep deprivation on behaviour, subsequent sleep, and dreaming. J. Ment. Sci., 1962, 108: 457-465. Borb61y, A.A. A two-process model of sleep regulation. Hum. Neurobiol., 1982, 1:195 204. Borb61y, A.A. and Neuhaus, H.U. Sleep deprivation: effects on sleep and EEG in the rat. J. Comp. Physiol., 1979, 133: 71 87. Borb61y, A.A., Baumann, F., Brandeis, D., Strauch, !. and Lehmann, D. Sleep deprivation: effect on sleep stages and EEG power density in man. Electroenceph. clin. Neurophysiol., 1981, 51:484 493. BorbSly, A.A., Balderer, G., Trachsel, L. and Tobler, I. Effect of midazolam and sleep deprivation on day-time sleep propensity. Arzneimittel-Forsch., 1985, 35: 1696-1699. Brunet, D., Nish, D., MacLean, A.W., Coulter, M. and Knowles, J.B. The time course of 'process S': comparison of visually scored slow wave sleep and power spectral analysis. Electroenceph. clin. Neurophysiol., 1988, 70: 278280. Carskadon, M.A. and Dement, W.C. Effects of total sleep loss on sleep tendency. Percept. Motor Skills, 1979, 48: 495-506. Czeisler, C.A., BorbSly, A.A., Hume, K.I., Kobayashi, T., Kronauer, R.E., Schulz, H., Weitzman, E.D., Zimmerman, J.C. and Zulley, J. Glossary of standardized terminology
499 for sleep-biological rhythm research. Sleep, 1980, 2: 287288. Daan, S., Beersma, D.G.M. and Borb61y, A.A. Timing of human sleep: recovery process gated by a circadian pacemaker. Am. J. Physiol., 1984, 246:RI61 R178. Dement, W. and Greenberg, S. Changes in total amount of stage four sleep as a function of partial sleep deprivation. Electroenceph. clin. Neurophysiol., 1966, 20: 523-526. Dement, W., Greenberg, S. and Klein, R. The effect of partial REM sleep deprivation and delayed recovery. J. Psychiat. Res., 1966, 4: 141-152. Dijk, D.J. and Beersma, D.G.M. Effects of SWS deprivation on subsequent EEG power density and spontaneous sleep duration. Electroenceph. clin. Neurophysiol., 1989, 72: 312-320. Dijk, D.J., Beersma, D.G.M. and Daan. S. EEG power density during nap sleep: reflection of an hourglass measuring the duration of prior wakefulness. J. Biol. Rhythms, 1987, 2: 207-219. Feinberg, I. and Floyd, T.C. Systematic trends across the night in human sleep cycles. Psychophysiology. 1979, 16: 283--291. Feinberg, I., Baker, T., Leder, R. and March, J.D. Response of delta (0 3 Hz) EEG and eye movement density to a night with 100 minutes of sleep. Sleep, 1988, 11: 473-487. Gulevich, G., Dement, W. and Johnson, L. Psychiatric and EEG observations on a case of prolonged (264 hours) wakefulness. Arch. Gen. Psychiat., 1966, 15: 29-35. Kales, A., Tan T.-L., Kollar, E.J., Naitoh, P., Preston, T.A. and Malmstr/3m, E.J. Sleep patterns following 205 hours of sleep deprivation. Psychosom. Med., 1970, 32: 189-200. McCarley, R.W. and Hobson, J.A. Neuronal excitability modulation over the sleep cycle: a structural and mathematical model. Science, 1974, 189: 58-60. Moses, J.M., Johnson, L.C., Naitoh. P. and Lubin, A. Sleep stage deprivation and total sleep loss: effects on sleep behavior. Psychophysiology, 1975, 12:141 146. Nakazawa, Y., Kotorii, M., Ohshima, M., Kotorii, T. and Hasuzawa, H. Changes in sleep pattern after sleep deprivation. Folia Psychiat. Neurol. Jpn., 1978, 32: 85-93. Rechtschaffen, A. and Kales, A. (Eds.). A Manual of Standardized Terminology, Techniques, and Scoring System for Sleep Stages of Human Subjects. National Institutes of Health, Publication No. 204. U.S. Government Printing Office, Washington, DC, 1968. Tilley, A.J. and Wilkinson, R.T. The effects of a restricted sleep regime on the composition of sleep and on performance. Psychophysiology, 1984, 21:406 412. Webb, W.B. and Agnew, H.W. Sleep: effects of a restricted regime. Science, 1965, 150:1745 1747. Webb, W.B. and Agnew, H.W. Stage 4 sleep: influence of time course variables. Science, 1971. 174: 1354-1356. Webb, W.B. and Agnew, H.W. The effects of a chronic limitation of sleep length. Psychophysiology, 1974, 11: 265-274. Williams, H.L., Hammack, J.T., Daly, R.L., Dement, W.C. and Lubin, A. Responses to auditory stimulation, sleep loss and the EEG stages of sleep. Electroenceph. clin. Neurophysiol., 1964, 16: 269-279.
Electroencephalography and clinical Neurophysiology, 1990, 75: 492-499 Elsevier Scientific Publishers Ireland, Ltd.
EEG 89528
Effect of partial sleep deprivation on sleep stages and E E G power spectra: evidence for n o n - R E M and R E M sleep homeostasis Daniel P. Brunner, Derk-Jan Dijk, Irene Tobler and Alexander A. Borbrly Institute of Pharmacology, University of Zurich, Zurich (Switzerland)
(Accepted for publication: 8 October 1989)
Summary The effect of repeated partial sleep deprivation on sleep stages and sleep EEG parameters was investigated in young subjects. After 2 baseline nights (B1, B2) of 7.5 h, sleep was restricted for 2 nights (D1, D2) to the first 4 h of the habitual bedtime period. Two recovery nights (R1, R2) with 7.5 h sleep followed. During the deprivation nights, stages 1 and 2 and REM sleep were reduced, while slow wave sleep (SWS; stages 3 and 4) was not significantly affected. However, the time integral of EEG power density in the range of 0.75-4.5 Hz (slow wave energy) was reduced. In the recovery period, SWS showed an enhancement in R1, and REM sleep showed a rebound in R1 and R2. An increase of REM sleep in the early part of the sleep period was evident in R1. Sleep latency was reduced in D2, R1 and R2. In accordance with the 2-process model of sleep regulation, EEG power density in non-REM sleep in the range of 0.75-4.5 Hz (slow wave activity) was only slightly higher in D2 and R1 than in baseline. An enhancement of slow wave activity in REM sleep was present in D2. Power density in the frequency range of 13-16 Hz was reduced in non-REM sleep (R1), SWS (R2) and stage 2 (R1). The results show (1) that the moderate reduction of slow wave energy in the deprivation nights induces only a minor enhancement of slow wave activity during recovery sleep; and (2) that a REM sleep deficit gives rise to an immediate rebound when 'slow wave pressure' is low. A comparison of the changes in the EEG power spectrum in non-REM sleep with published data on the effects of total sleep deprivation suggests that an increase of ' REM sleep pressure' has repercussions on the EEG in non-REM sleep. Key words: Sleep deprivation; Sleep homeostasis; REM sleep; Sleep EEG; EEG spectral analysis T o t a l sleep d e p r i v a t i o n e n h a n c e s the p r o p e n s i t y for sleep i n i t i a t i o n a n d n o n - R E M sleep i n t e n s i t y . T h e f o r m e r effect is e v i d e n t f r o m the r e d u c t i o n of sleep l a t e n c y after a n i g h t w i t h o u t sleep ( C a r s k a d o n a n d D e m e n t 1979; B o r b r l y et al. 1985), the l a t t e r effect f r o m the i n c r e a s e o f slow w a v e sleep ( S W S ) ( W i l l i a m s et al. 1964; A g n e w et al. 1967; K a l e s et al. 1970; W e b b a n d A g n e w 1971; M o s e s et al. 1975) a n d E E G slow w a v e a c t i v i t y ( B o r b r l y et al. 1981; ,~tkerstedt a n d G i l l b e r g 1986; D i j k et al. 1987; B r u n e t et al. 1988) d u r i n g r e c o v e r y sleep. I n c o n t r a s t to n o n - R E M sleep, R E M sleep is little a f f e c t e d w h e n sleep has b e e n d e p r i v e d for n o t
Correspondence to: Prof. A. Borbrly, Institute of Pharmacology, University of Zurich, Gloriastrasse 32, CH-8006 Zurich (Switzerland).
m o r e t h a n a single n i g h t ( N a k a z a w a et al. 1978; B o r b r l y et al. 1981). A f t e r a l o n g e r p e r i o d w i t h o u t sleep a R E M sleep r e b o u n d is e v i d e n t (Berger a n d O s w a l d 1962; W i l l i a m s et al. 1964; G u l e v i c h et al. 1966; K a l e s et al. 1970). W h e n sleep d u r a t i o n was r e s t r i c t e d for several nights, the a m o u n t of S W S r e m a i n e d u n c h a n g e d or e v e n i n c r e a s e d , w h e r e a s R E M sleep was d r a s t i cally r e d u c e d ( W e b b a n d A g n e w 1965, 1974; Dem e n t a n d G r e e n b e r g 1966). T h e s e r e s u l t s d e m o n strate the h i g h p r i o r i t y of S W S i n c o m p a r i s o n to R E M sleep. It has b e e n c o n c l u d e d t h a t S W S is m o r e f i n e l y r e g u l a t e d t h a n R E M sleep ( B o r b r l y 1982), a n d t h a t it is the i n c r e a s e d ' s l o w w a v e p r e s s u r e ' w h i c h p r e v e n t s the r e b o u n d of R E M sleep after sleep d e p r i v a t i o n ( W i l h a m s et al. 1964; W e b b a n d A g n e w 1965; T i l l e y a n d W i l k i n s o n 1984). T h u s , a c c o r d i n g to this i n t e r p r e t a t i o n , the non-REM-REM sleep i n t e r a c t i o n o b s c u r e s the
0013-4649/90/$03.50 © 1990 Elsevier Scientific Publishers Ireland, Ltd.
PARTIAL SLEEP DEPRIVATION AND SLEEP HOMEOSTASIS manifestation
of REM
sleep homeostasis
has been demonstrated by selective REM d e p r i v a t i o n e x p e r i m e n t s ( D e m e n t e t al.
493
which
Methods
sleep 1966;
Subjects
A g n e w e t al. 1967). T h e a i m o f t h e p r e s e n t s t u d y was to gain further insight into the processes underlying the regulation of the two major sleep s t a t e s a n d t h e i r i n t e r a c t i o n s . T o m i n i m i z e t h e inh i b i t o r y e f f e c t o f S W S o n R E M sleep, a n e x p e r i mental protocol was selected in which SWS was reduced by only a small extent whereas REM sleep was markedly curtailed. In contrast to previous studies, we did not restrict the analysis to v i s u a l l y s c o r e d s l e e p stages, b u t q u a n t i f i e d t h e sleep EEG by spectral analysis.
Sleep records were obtained from 9 paid, young, m a l e s u b j e c t s ( 2 2 - 2 7 y e a r s , m e a n 24.1) o n 7 c o n secutive nights. The subjects were in good health and did not report any sleep disorders. They were requested to refrain from sleeping during the day (which was verified by wrist activity monitoring), and from using alcohol or drugs throughout the experiment. The protocol consisted of an adaptat i o n n i g h t ( A D ) , 2 b a s e l i n e n i g h t s (B1, B2), 2 partial sleep deprivation nights (D1, D2), and 2 r e c o v e r y n i g h t s ( R 1 , R 2 ) (see Fig. 1). T h e s u b j e c t s
TABLE I Sleep parameters in minutes (mean with S.E.M. in parentheses) for the baseline nights (B1, B2), the nights with sleep restriction to 4 h (D1, D2) and the recovery nights (R1, R2). Sleep latency, latency from lights off to stage 2; REM sleep (REMS) latency, latency of REMS from stage 2. Slow wave energy, i.e., EEG power density (0.75-4.5 Hz), integrated over time, is expressed as a percentage of the baseline mean (100% = (B1 + B2)/2). Night
F
B1
Sleep latency REMS latency REMS in first 2 h
B2
12.5 (1.7) 67.3 (3.5) 7.6 (1.6)
12.1 (2.4) 70.3 (11.2) 11.4 (2.5)
D1
D2
R1
12.7 (3.6) 75.4 (10.0) 12.1 (4.7)
8.1 (1.9) * * 61.7 (3.3) 9.9 (2.4)
R2
6.2 (1.3) * * 58.2 (8.1) a 20.5 (4.2) * *
8.8 (1.3) * * 60.6 (5.3) 12.9 (2.1)
18.19 4.79 13.68
Entire sleep period Total sleep time Waking after ~ieep onset Stage 1 Stage 2 Stage 3 Stage 4 REMS SWS (stage 3 + 4) Slow wave energy
430.8 (3.3)
433.1 (3.4)
240.3 (1.5) * *
239.4 (0.4) * *
434.1 (3.7)
3.9 (0.8) 39.5 (3.4) 216.1 (9.7) 47.1 (4.7) 33.8 (5.5) 94.4 (5.7) 80.8 (4.7) 103.8 (2.9)
3.1 (1.6) 40.7 (3.1) 216.1 (8.4) 41.1 (4.0) 30.6 (6.0) 104.7 (4.3) 71.7 (6.9) 96.2 (2.9)
2.7 (1.0) 18.1 (1.4) 114.1 (6.7) 36.6 (4.7) 33.7 (8.1) 37.8 (4.7) 70.3 (6.6) 80.2 (3.3)
0.7 (0.5) 15.3 (2.0) 109.2 (7.5) 44.8 (2.7) 37.9 (7.4) 32.2 (3.2) 82.7 (5.8) 87.3 (3.3)
0.6 (0.5) 29.4 (2.2) 184.2 (9.3) 53.4 (4.9) 41.8 (8.1) 125.2 (5.2) 95.2 (8.1) 104.6 (7.1)
17.7 (1.3) 115.2 (4.2) 35.6 (2.9) 30.2 (5.9) 39.2 (4.8) 65.8 (5.9) 96.2 (3.7)
17.9 111.4 36.4 33.7 37.6 70.1 103.9
** **
** **
** ** **
** **
** ** ** * ** *
433.6 (3.0)
32.10
3.9 (0.3) 39.0 (3.3) 188.1 (6.5) * * 51.9 (3.9) 34.5 (8.1) 120.1 (4.1) * * 86.4 (8.2) 99.6 (3.8)
15.19 33.70 39.20 13.79 2.02 39.67 17.25 22.90
16.0 100.7 47.2 32.6 41.7 79.8 103.2
11.47 18.99 10.11 3.19 9.04 16.41 8.05
First 4 h of sleep period Stage 1 Stage 2 Stage 3 Stage 4 REMS SWS (stage 3 + 4) Slow wave energy
20.7 109.9 41.3 31.2 33.6 72.6 103.8
(2.9) (5.2) (4.7) (4.4) (4.8) (2.2) (3.7)
(1.4) (6.4) (4.6) (8.1) (4.7) (6.6) (3.4)
15.3 (2.0) 108.8 (7.6) 44.4 (2.8) 37.9 (7.4) 31.4 (3.3) 82.3 (5.6) * * 114.2 (4.6)
11.4 (1.7) * * 92.9 (6.2) * * 44.8 (5.2) 40.4 (7.6) 48.8 (4.6) 85.3 (5.0) * * 106.8 (6.5)
(1.6) (4.9) * (4.6) (7.5) (4.3) (6.6) (4.5)
F = Friedman 2-way non-parametric ANOVA for repeated measures: P < 0.05 for F>11.1; P < 0.01 for F>15.1. * P < 0.05; * * P < 0.01; significant difference from baseline mean (Wilcoxon matched-pairs signed-rank test, 2-tailed). Between B1 and B2 no significant differences were found. a One sleep onset REM episode included.
494 went to bed at their habitual bedtime (23-24 h). They were awakened after 7 h or 7.5 h sleep on the AD, B and R nights (the sleep duration was individually adjusted to the habitual sleep duration), and after a sleep duration of 4 h on DI and D2. Sleep duration was determined on-line from the polygraph records. After the partial sleep deprivation nights, the subjects were asked to maintain their usual day-time activities, but to avoid sleep-inducing situations.
D.P. BRUNNER ET AL.
Bq
!
m
m
B2
Recording and data analysis EEG ( C 3 / F z , and C4/Fz), E O G and E M G were recorded continuously on polygraph paper (Grass 78D; paper speed 10 mm/sec). The highpass filter setting of the EEG channels was 0.1 Hz which corresponds to a time constant of 0.6 sec. The amplifier output of all signals was digitized on-line (sampling rate 128 Hz) by a signal processor card (Texas Instruments, TMS-320-10) installed in a personal computer (Olivetti M24). The EEG signals were low-pass filtered (25 Hz, 24 d B / o c t ) prior to A / D conversion, and then subjected to spectral analysis by a Radix 8 Fast Fourier Transform (FFT) routine implemented on the signal processor card. Power spectra were computed for consecutive 4 sec epochs by applying a rectangular window. If the EEG signal reached the maximum or minimum level of the A / D converter, the corresponding 4 sec spectrum was excluded. This routine, which eliminated 0.4% of all 4 sec epochs, served to prevent contamination of the data set by spurious data arising from movement artifacts or from saturation of the amplifier. Mean values over adjacent frequencies were computed to collapse the 0.25 Hz data into 0.5 Hz bins for frequencies between 0.25 and 5.0 Hz, and into 1 Hz bins for frequencies between 5.25 and 25.0 Hz. One minute mean spectra were computed from 15 consecutive 4 sec spectra, and stored on the hard disk of the personal computer. Sleep stages were scored for 30 sec epochs according to the criteria of Rechtschaffen and Kales (1968). Since the bipolar derivations used in the present study probably yielded signals with smaller amplitudes than the recommended C 3 / A 2 , C 4 / A ] derivations, the amounts of stages 3 and 4 may have been slightly underestimated. However,
m
m
•
•
m
m
DI
!
m
D2
I
m
I
R1
I
23
m
•
m
•
I
24
I
01
m
m
m 1
I
02 03 Time o¢ doy
m
m I
04
m I
05
I
06
Fig. 1. Time course of EEG slow wave activity (power density in the 0.75-4,5 Hz band) on 6 consecutive nights of 1 subject: 2 baseline nights (B1, B2), 2 partial sleep deprivation nights (D1, D2) and 2 recovery nights (R1, R2). Data are plotted from lights off until the termination of sleep. REM sleep episodes are indicated by horizontal bars. Calibration (entire scale) of power density: 66.3/~V2/Hz.
495
P A R T I A L SLEEP D E P R I V A T I O N A N D SLEEP HOMEOSTASIS 10
the effects on relative E E G spectra are largely independent of the distance between the electrodes, since similar results were obtained with a C 3 / A 2 derivation (Dijk and Beersma 1989). The scores were entered into the computer. Two consecutive 30 sec sleep scores corresponded to a 1 min EEG spectrum. For the computation of power density within a specific sleep stage (Fig. 3; SWS was considered as a single stage), only 1 min epochs with 2 identical scores were used. Thus 30 sec epochs scored as movement time were not incorporated in the spectral data. R E M - n o n - R E M sleep cycles were defined according to the criteria of Feinberg and Floyd (1979), R E M sleep episodes according to Czeisler et al. (1980).
Level of S
1158~
lOOX 05
I O0 B1
160
B2
D1
% of baseline mean
D2
R1
R2
(first 4h)
150 140
SWA
130
(model)
SWA (data)
120
REM m sleep
J,
110 100 90
Results
Sleep parameters Restricting sleep to 4 h reduced the sleep latency in the following night (D2, R1). A significant reduction of sleep latency was still present in the second recovery night (Table I). During the 2 short nights (D1, D2), the amounts of stage 1, stage 2 and R E M sleep were markedly below baseline, whereas stages 3 and 4 were little changed (Table I). In comparison to the first 4 h of baseline sleep, a significant increase of SWS was seen in the second deprivation night. The first recovery night was characterized by a reduction in waking, stage 1 and stage 2, and an increase in stage 3, SWS and R E M sleep. The decrease in stages 1 and 2 and the increase in SWS were significant also in the first 4 h of sleep. R E M sleep in the first 2 h was massively increased. The reduction of stage 2 and the increase of R E M sleep were still present in the second recovery night (Table I, entire sleep period).
Spectral analysis of the sleep EEG Fig. 1 illustrates for 1 subject the time course of E E G power density in the 0.75-4.5 Hz range (slow wave activity) on 6 consecutive nights. R E M sleep episodes are indicated by horizontal bars. Slow wave activity is typically high in the first 1 - 2 n o n - R E M sleep episodes, and lower in subsequent
80 70
BL D1 D2 R1 R2
BL D1 D2 I~1 R2
BL D1 D2 R1 R2
Fig. 2. Upper panel: time course of process S during the experimental protocol as simulated with the 2-process model of sleep regulation. Sleep episodes are indicated by horizontal bars and shading. Lower panel: sleep parameters in the first 4 h of sleep. Left: prediction of slow wave activity (SWA) by the model; middle: recorded SWA in n o n - R E M sleep; right: a m o u n t of REM sleep. BL, mean of the 2 baseline nights; other abbreviations as in Fig. 1.
episodes. For this reason, the shortening of sleep duration to 4 h reduced slow wave energy (the time integral of slow wave activity) only to 80.2% (D1) and to 87.3% (D2) of baseline (Table I). The changes of slow wave activity throughout the experiment were simulated by the 2-process model (Daan et al. 1984). As shown in Fig. 2 (upper panel) it is assumed in the model that process S, the marker of n o n - R E M sleep homeostasis derived from slow wave activity, declines exponentially during sleep (shaded area) and increases according to a saturating exponential function during waking. The model predicts that after reducing sleep to 4 h, the initial level of S at sleep onset is raised in D2 to 115.8% and in R1 to 117.8% of control. However, the first recovery night is sufficient to bring back the initial level of S in R2 close to the control level. It should be noted that, because of the exponential function,
496
D.P. B R U N N E R ET AL. 215
BL
................... D 2
2.10,
t,, /' ',,,/,,,~/%,,
- - R 1 ......
~
R2
,/,/*"',,,,
205.
"6 2.00-
195"
REM sleep -
o0 c'(D "0
-
i
•
r
i
-
•
r
D2
D2
R1
2.05"
0 ("3
......./'"'..,,.
.(--., _.~#~=;, .............. ..""" ',,.....,....
, j~'=~: .......... .,..
2.00"
(.9 III
LU
195
SWS 190
R1 R2
"
~-/
i
-
5 •
p
,0
"
"
"
i
~
•
tage .
2o HZ
R 1
•
R2
.
e-o
.
~
.
.
.
.
~
.
.
.
~ ~
2o =
Hz
=
=
Fig. 3. EEG power density for the first 4 h of sleep in deprivation night 2 (D2), recovery night 1 (RI) and recovery night 2 (R2). For each subject and each frequency bin the mean value of the 2 baseline nights (first 4 h) is defined as 100%. The values are expressed relative to the corresponding baseline value. The data represent means (N = 9) of individual log-transformed values and are plotted at the upper fimit of each frequency bin. The frequency range in which the values differ significantly from the baseline mean (BL) are indicated below the abscissae ( P < 0.05; Wilcoxon matched-pairs signed-ranks test).
the change in the peak value of S at sleep onset is identical to the value integrated over the first 4 h of sleep. The lower panel of Fig. 2 contrasts the predictions of the model with the data. In accordance with the model, slow wave activity in nonR E M sleep is increased in D2 and R1. The empirical values are somewhat lower than the simulated values for D2 and R1, and somewhat higher for D1 and R2. However, the differences between the data and the simulations were not statistically significant ( P > 0.1 in all cases; paired t test). Taken together, the general time course of slow wave activity conforms to the prediction of the model. In the lower panel of Fig. 2, the different time courses of slow wave activity and visually scored R E M sleep are illustrated. Fig. 3 shows the effect of partial sleep deprivation on E E G power density in n o n - R E M sleep, R E M sleep, stages 3 and 4 and stage 2. The values are shown for the first 4 h of D2, R1 and R2, and are expressed relative to the corresponding mean baseline value. Power density in the low frequency
range tended to be above the baseline level in total non-REM sleep and R E M sleep. However, a significant elevation was present in n o n - R E M sleep only in the 4 - 5 Hz band of D2 and R1, and in R E M sleep in some low frequency bands of D2. In SWS the low frequency values were close to the baseline level or below it; in stage 2 the values in the frequency band of 0.75-2.0 Hz in R1 were significantly reduced. A depression in the 13.2516.0 Hz range was present in total n o n - R E M sleep and stage 2 of R1, and in the 17.25-20.0 Hz range in stage 2 of R2.
Discussion
N o n - R E M sleep homeostasis It has been demonstrated in previous experiments that in sleep restriction protocols the level of SWS is preserved, whereas R E M sleep is reduced (Webb and Agnew 1965, 1974; Dement and Greenberg 1966). The present experiment con-
PARTIAL SLEEP DEPRIVATION AND SLEEP HOMEOSTASIS
497
firms these observations and provides further evidence for the high priority of SWS in sleep regulation. However, in addition to determining SWS by visual scoring of the polygraph records, the E E G was subjected to all-night spectral analysis to obtain a quantitative measure of the sleep-related E E G changes. These measures were also used to investigate the homeostatic mechanisms underlying n o n - R E M sleep. The 2-process model of sleep regulation (Borbrly 1982; Daan et al. 1984) postulates that a sleep/wake-dependent process (process S) increases during waking and declines during sleep. The time course of S had been derived from the changes of E E G slow wave activity. Simulations of the present experimental schedule predicted a minor increase in slow wave activity during the second deprivation night and the first recovery night (Fig. 2 upper panel). The experimental data were in agreement with the prediction, since only a minor and non-significant enhancement of slow wave activity was present in the nights D2 and R1 (Fig. 2 lower panel). Spectral analysis by individual sleep stages showed that in contrast to the recovery from total sleep deprivation (Borbrly et al. 1981), slow wave activity within SWS was not enhanced (Fig. 3). Thus, the present partial sleep deprivation schedule did not intensify this sleep stage. At first sight it may be puzzling that only a small rise or even a reduction of slow wave activity is seen in stage 2 and SWS, whereas an increase is present in the entire non-REM sleep (Fig. 3). This apparent contradiction is resolved by recognizing that the absolute level of slow wave activity is much higher in SWS than in stage 2 (Borbrly et al. 1981). Thus, an increase in the amount of SWS enhances slow wave activity within non-REM sleep even though slow wave activity within SWS may be decreased. The increase of SWS in D2, R1 and R2 relative to stage 2 (Table I) therefore resulted in an enhanced slow wave activity in n o n - R E M sleep, although no such increase was present within the individual n o n - R E M sleep stages.
1978; Borbrly et al. 1981), it has been concluded that the regulation of n o n - R E M sleep is more finely ' t u n e d ' than that of R E M sleep (Borbrly 1982). However, total sleep deprivation experiments are not ideal for investigating the regulation of R E M sleep, because the high propensity for SWS may impede the manifestation of R E M sleep. The present experimental protocol was designed to minimize this interference since the limitation of sleep to 4 h was expected to have only a minor effect on SWS. The experiment revealed that a R E M sleep deficit of 131% incurred during 2 nights induces a potent compensatory response in the 2 recovery nights. In addition to the increase of R E M sleep over the entire night, the heightened ' R E M sleep pressure' is reflected also by the increase of this sleep stage in the first part of the night (Fig. 2). This effect was not yet present after the first deprivation night, indicating that a buildup of ' R E M sleep pressure' or a shift in the ' R E M / s l o w wave balance' is necessary for its manifestation. It should be noted that the R E M sleep deficit incurred in the 2 deprivation nights was not fully compensated by the excess in the 2 recovery nights, and it is possible that the enhancement of R E M sleep persisted in the following nights. A prolonged R E M sleep rebound has been observed also in other studies (Berger and Oswald 1962; Dement et al. 1966; Agnew et al. 1967) and could represent a typical feature of R E M sleep regulation. The prolonged compensatory response is in accordance with the hypothesis that a R E M sleep deficit is made up mainly in terms of R E M sleep time, whereas a non-REM sleep deficit can be compensated by an increase in n o n - R E M sleep intensity (Borbrly and Neuhaus 1979; Borbrly 1982). Furthermore, it is interesting to note that sleep latency was still significantly shortened in the second recovery night at a time when SWS no longer differed significantly from baseline. Thus, the processes controlling sleep latency may be more closely related to R E M sleep than to SWS.
R E M sleep homeostasis Since following 1 night of total sleep deprivation, SWS exhibits a marked rebound, whereas R E M sleep is little affected (Nakazawa et al.
Further effects on EEG spectra A detailed analysis of the E E G spectra revealed that sleep restriction induced more subtle changes than just a general rise of slow wave activity in
498
n o n - R E M sleep. It is informative to compare the effects within individual sleep stages (Fig. 3) with those observed after total sleep deprivation (Borb61y et ah 1981). As in the previous study, power density in the 13-16 Hz band was depressed in n o n - R E M sleep, SWS and stage 2. This effect was most prominent in the first recovery night, but tended to persist in the subsequent night. These E E G changes were also observed in recent sleep deprivation experiments (Dijk and Beersma 1989). Moreover, it was shown in a nap study that EEG activity in the 15 Hz band decreases as a function of prior waking (Dijk et ah 1987). Taken together these observations are consistent with the assumption that this EEG parameter is related to sleep homeostasis. However, unlike slow wave activity, it does not exhibit a clear trend across the night (Borb61y et al. 1981), and its changes appear to persist for a longer time after sleep deprivation. Activity in the 13 16 Hz range may therefore reflect long-term aspects of recovery which are not formalized by process S. As has been observed after total sleep deprivation (Borb61y et al. 1981), an increase in slow wave activity was present also in REM sleep (Fig. 3). It is important to note that the absolute level of slow wave activity is much lower in R E M sleep than in non-REM sleep and that only relative changes are shown in Fig. 3. Interestingly, the largest increase was present in D2 at a time when 'slow wave pressure' was enhanced whereas ' R E M sleep pressure' had only moderately risen. The further increase of the REM sleep deficit from D2 to R1 was not accompanied by a further increase of slow wave activity in R E M sleep. Thus, the changes of slow wave activity in R E M sleep and in n o n - R E M sleep were rather similar, and different from the changes in the amount of R E M sleep and the R E M sleep deficit. These observations, in conjunction with previous results (Borb61y et al. 1981; Dijk and Beersma 1989), indicate that a common homeostatic process modulated the slow wave activity in non-REM sleep and R E M sleep. When comparing the effects of total and partial sleep deprivation on slow wave activity in nonR E M sleep, it is evident that not only the extent but also the pattern of enhancement differs. In contrast to the marked peak in the 1 2 Hz band
D.P. BRUNNIER ET AL.
after total sleep deprivation (Borb61y et ah 1981; Dijk et al. 1987) and selective slow wave deprivation (Dijk and Beersma 1989), the rise induced by partial sleep deprivation encompasses rather uniformly a broad frequency range (Fig. 3). This difference in pattern may be simply due to a lower 'sleep pressure.' However, this is unlikely to be the case because the 1 - 2 Hz peak is present even in the cycle-to-cycle changes of a baseline night. Alternatively, the predominant REM sleep deprivation and the activation of R E M sleep homeostasis may exert a depressing influence on power density in the lowest frequency range and thus eliminate the typical 1 - 2 Hz peak. This interpretation could also apply to a recent observation based on period-amplitude analysis in which, during recovery from a partial sleep deprivation schedule (i.e., sleep restriction to the first 100 min of sleep), the wave density in the 0 - 3 Hz range was increased, whereas the EEG amplitude was not affected (Feinberg et ah 1988). Our own preliminary analysis of the EEG amplitude is consistent with the assumption that an enhanced ' R E M sleep pressure' exerts a depressing influence on maximal E E G amplitude. The depression of activity at the lowest end of the stage 2 spectrum in R1 could also be explained along these lines. Although this assumption could account for the changes in the 1 - 2 Hz range of the power spectra, firmer evidence is needed in support.
Concluding comments In studies of sleep regulation, non-REM sleep has been in the focus of interest for several reasons: (1) it represents 75-80% of total sleep time; (2) the proportion of slow wave activity within n o n - R E M sleep appears to be an easily accessible measure of n o n - R E M sleep intensity; and (3) slow wave activity is homeostatically regulated. These considerations have led to the formulation of the 2-process model of sleep regulation where one of the constituent processes is directly derived from slow wave activity. R E M sleep regulation was not incorporated into the model except for a qualitative description of its circadian c o m p o n e n t (Borb61y 1982). The present study provides further evidence for the homeostatic component underlying REM sleep regulation. Moreover, it supports
PARTIAL SLEEP DEPRIVATION AND SLEEP HOMEOSTASIS the assumption that 'slow wave pressure' exerts an inhibitory effect on sleep (Webb
and
the
Agnew
manifestation of REM 1965;
Dement
et
al.
1966; Borb61y 1982). I n a n a t t e m p t to a c c o u n t f o r the REM-non-REM sleep cyclicity, we p r o p o s e d in t h e m o d e l a r e c i p r o c a l i n t e r a c t i o n b e t w e e n p r o cess S and a REM sleep controlling process, analo g o u s to t h e n e u r o p h y s i o l o g i c a l h y p o t h e s i s o f M c C a r l e y a n d H o b s o n (1974). T h e p r e s e n t r e s u l t s l e n d f u r t h e r c r e d e n c e to a n i n t e r a c t i o n b e t w e e n t h e 2 s l e e p s t a t e s . It is e v e n p o s s i b l e t h a t t h e rise in ~ R E M s l e e p p r e s s u r e ' h a s r e p e r c u s s i o n s o n t h e E E G in n o n - R E M sleep. A n e x t e n s i o n o f t h e 2 - p r o c e s s m o d e l will b e r e q u i r e d t o a c c o m m o d a t e these new aspects.
This study was supported by the Swiss National Science Foundation, Grants 3.234-0.85 and 31.25634.88.
References Agnew, HW., Webb, W.B. and Williams, R.L. Comparison of stage four and 1-REM sleep deprivation. Percept. Motor Skills, 1967, 24: 851-858. ,~kerstedt, T. and Gillberg, M. Sleep duration and the power spectral density of the EEG. Electroenceph. clin. Neurophysiol., 1986, 64: 119-122. Berger, R.J. and Oswald, 1. Effects of sleep deprivation on behaviour, subsequent sleep, and dreaming. J. Ment. Sci., 1962, 108: 457-465. Borb61y, A.A. A two-process model of sleep regulation. Hum. Neurobiol., 1982, 1:195 204. Borb61y, A.A. and Neuhaus, H.U. Sleep deprivation: effects on sleep and EEG in the rat. J. Comp. Physiol., 1979, 133: 71 87. Borb61y, A.A., Baumann, F., Brandeis, D., Strauch, !. and Lehmann, D. Sleep deprivation: effect on sleep stages and EEG power density in man. Electroenceph. clin. Neurophysiol., 1981, 51:484 493. BorbSly, A.A., Balderer, G., Trachsel, L. and Tobler, I. Effect of midazolam and sleep deprivation on day-time sleep propensity. Arzneimittel-Forsch., 1985, 35: 1696-1699. Brunet, D., Nish, D., MacLean, A.W., Coulter, M. and Knowles, J.B. The time course of 'process S': comparison of visually scored slow wave sleep and power spectral analysis. Electroenceph. clin. Neurophysiol., 1988, 70: 278280. Carskadon, M.A. and Dement, W.C. Effects of total sleep loss on sleep tendency. Percept. Motor Skills, 1979, 48: 495-506. Czeisler, C.A., BorbSly, A.A., Hume, K.I., Kobayashi, T., Kronauer, R.E., Schulz, H., Weitzman, E.D., Zimmerman, J.C. and Zulley, J. Glossary of standardized terminology
499 for sleep-biological rhythm research. Sleep, 1980, 2: 287288. Daan, S., Beersma, D.G.M. and Borb61y, A.A. Timing of human sleep: recovery process gated by a circadian pacemaker. Am. J. Physiol., 1984, 246:RI61 R178. Dement, W. and Greenberg, S. Changes in total amount of stage four sleep as a function of partial sleep deprivation. Electroenceph. clin. Neurophysiol., 1966, 20: 523-526. Dement, W., Greenberg, S. and Klein, R. The effect of partial REM sleep deprivation and delayed recovery. J. Psychiat. Res., 1966, 4: 141-152. Dijk, D.J. and Beersma, D.G.M. Effects of SWS deprivation on subsequent EEG power density and spontaneous sleep duration. Electroenceph. clin. Neurophysiol., 1989, 72: 312-320. Dijk, D.J., Beersma, D.G.M. and Daan. S. EEG power density during nap sleep: reflection of an hourglass measuring the duration of prior wakefulness. J. Biol. Rhythms, 1987, 2: 207-219. Feinberg, I. and Floyd, T.C. Systematic trends across the night in human sleep cycles. Psychophysiology. 1979, 16: 283--291. Feinberg, I., Baker, T., Leder, R. and March, J.D. Response of delta (0 3 Hz) EEG and eye movement density to a night with 100 minutes of sleep. Sleep, 1988, 11: 473-487. Gulevich, G., Dement, W. and Johnson, L. Psychiatric and EEG observations on a case of prolonged (264 hours) wakefulness. Arch. Gen. Psychiat., 1966, 15: 29-35. Kales, A., Tan T.-L., Kollar, E.J., Naitoh, P., Preston, T.A. and Malmstr/3m, E.J. Sleep patterns following 205 hours of sleep deprivation. Psychosom. Med., 1970, 32: 189-200. McCarley, R.W. and Hobson, J.A. Neuronal excitability modulation over the sleep cycle: a structural and mathematical model. Science, 1974, 189: 58-60. Moses, J.M., Johnson, L.C., Naitoh. P. and Lubin, A. Sleep stage deprivation and total sleep loss: effects on sleep behavior. Psychophysiology, 1975, 12:141 146. Nakazawa, Y., Kotorii, M., Ohshima, M., Kotorii, T. and Hasuzawa, H. Changes in sleep pattern after sleep deprivation. Folia Psychiat. Neurol. Jpn., 1978, 32: 85-93. Rechtschaffen, A. and Kales, A. (Eds.). A Manual of Standardized Terminology, Techniques, and Scoring System for Sleep Stages of Human Subjects. National Institutes of Health, Publication No. 204. U.S. Government Printing Office, Washington, DC, 1968. Tilley, A.J. and Wilkinson, R.T. The effects of a restricted sleep regime on the composition of sleep and on performance. Psychophysiology, 1984, 21:406 412. Webb, W.B. and Agnew, H.W. Sleep: effects of a restricted regime. Science, 1965, 150:1745 1747. Webb, W.B. and Agnew, H.W. Stage 4 sleep: influence of time course variables. Science, 1971. 174: 1354-1356. Webb, W.B. and Agnew, H.W. The effects of a chronic limitation of sleep length. Psychophysiology, 1974, 11: 265-274. Williams, H.L., Hammack, J.T., Daly, R.L., Dement, W.C. and Lubin, A. Responses to auditory stimulation, sleep loss and the EEG stages of sleep. Electroenceph. clin. Neurophysiol., 1964, 16: 269-279.
