www.elsevier.

com/locate/ynimg
NeuroImage 28 (2005) 14 – 21

Cerebral correlates of delta waves during non-REM sleep revisited

Thien Thanh Dang-Vu,a,b Martin Desseilles,a Steven Laureys,a,b Christian Degueldre,a
Fabien Perrin,a Christophe Phillips,a Pierre Maquet,a,b and Philippe Peigneuxa,c,*
a
Cyclotron Research Centre, University of Liege, Belgium
b
Neurology Department, CHU Liege, Belgium
c
Neuropsychology Unit, University of Liege, Belgium

Received 25 May 2004; revised 8 April 2005; accepted 20 May 2005

Available online 23 June 2005

We aimed at characterizing the neural correlates of delta activity involved in the generation of cortical delta waves during NREM
during Non Rapid Eye Movement (NREM) sleep in non-sleep- sleep.
deprived normal young adults, based on the statistical analysis of a D 2005 Elsevier Inc. All rights reserved.
positron emission tomography (PET) sleep data set. One hundred
fifteen PET scans were obtained using H215O under continuous Keywords: Delta activity; Non-REM sleep; Brain imaging; Positron
polygraphic monitoring during stages 2 – 4 of NREM sleep. Corre- emission tomography; Statistical parametric mapping
lations between regional cerebral blood flow (rCBF) and delta power
(1.5 – 4 Hz) spectral density were analyzed using statistical para-
metric mapping (SPM2). Delta power values obtained at central
scalp locations negatively correlated during NREM sleep with rCBF
Introduction
in the ventromedial prefrontal cortex, the basal forebrain, the
striatum, the anterior insula, and the precuneus. These regions
embrace the set of brain areas in which rCBF decreases during slow Non Rapid Eye Movement (NREM) sleep is characterized by
wave sleep (SWS) as compared to Rapid Eye Movement (REM) specific oscillations on electroencephalographic recordings (EEG):
sleep and wakefulness (Maquet, P., Degueldre, C., Delfiore, G., Aerts, spindles, delta and slow rhythms. Spindles, a prominent feature of
J., Peters, J.M., Luxen, A., Franck, G., 1997. Functional neuro- light NREM sleep (i.e., sleep stage 2), are defined in humans as
anatomy of human slow wave sleep. J. Neurosci. 17, 2807 – 2812), waxing-and-waning oscillations within the 12 – 15 Hz (sigma band)
supporting the notion that delta activity is a valuable prominent frequency range, lasting at least 0.5 s (Rechtschaffen and Kales,
feature of NREM sleep. A strong association was observed between 1968). During deep NREM sleep (sleep stages 3 and 4), the EEG is
rCBF in the ventromedial prefrontal regions and delta power, in
mainly characterized by a slower oscillation in the delta range
agreement with electrophysiological studies. In contrast to the results
(1.5 – 4 Hz). A slow rhythm (<1 Hz) occurs both during light and
of a previous PET study investigating the brain correlates of delta
activity (Hofle, N., Paus, T., Reutens, D., Fiset, P., Gotman, J., deep NREM sleep and manifests itself, respectively, as the regular
Evans, A.C., Jones, B.E., 1997. Regional cerebral blood flow changes recurrence of spindles every 3 – 10 s, K-complexes, or as slow
as a function of delta and spindle activity during slow wave sleep in waves below 1 Hz (Achermann and Borbely, 1997; Steriade and
humans. J. Neurosci. 17, 4800 – 4808), in which waking scans were Amzica, 1998).
mixed with NREM sleep scans, no correlation was found with The mechanisms that generate spindles are well documented.
thalamus activity. This latter result stresses the importance of an Following a reduction in activating input from brainstem reticular
extra-thalamic delta rhythm among the synchronous NREM sleep formation, repetitive spike-bursts arise from GABAergic thalamic
oscillations. Consequently, this rCBF distribution might preferen- reticular neurons (RE), that generate inhibitory post-synaptic
tially reflect a particular modulation of the cellular processes
potentials (IPSPs) in glutamatergic thalamocortical (TC) neurons.
The latter are entrained to oscillate within the sigma frequency
range (Dijk et al., 1993; Steriade, 1999; De Gennaro and Ferrara,
2003). Although neurophysiological studies in animals suggest
comparable mechanisms for the generation of delta waves, the
process of delta wave generation is less clear; delta oscillations
* Corresponding author. Cyclotron Research Centre, University of Liege, have been studied in vitro or in vivo under anesthesia and two
Sart Tilman, Bat. B30, B-4000 Liege, Belgium. Fax: +32 4 3662946. types of delta activity have been identified. First, when TC neurons
E-mail address: [email protected] (P. Peigneux). reach a sufficient level of hyperpolarization, a clock-like delta
Available online on ScienceDirect (www.sciencedirect.com). rhythm is generated in these neurons by the interplay between two
1053-8119/$ - see front matter D 2005 Elsevier Inc. All rights reserved.
doi:10.1016/j.neuroimage.2005.05.028
 T.T. Dang-Vu et al. / NeuroImage 28 (2005) 14 – 21 15

hyperpolarization-activated currents (McCormick and Bal, 1997). sleeping conditions of the scanner during the two nights before
Second, another delta oscillation is generated within the cortex the experimental night. Polysomnographic recordings were used
because it can be recorded even after extensive thalamectomy to screen for sleep abnormalities, including excessive sleep
(Frost et al., 1966; Ball et al., 1977; Steriade et al., 1993a; Steriade, fragmentation and insufficient total sleep time. Only subjects
2003). The mechanisms of this cortical delta oscillation remain whose sleep duration and quality was sufficient, and who could
poorly understood. maintain 20 min of continuous stage 2, stages 3 – 4 of NREM
In humans, the cerebral correlates of delta activity have been sleep and Rapid Eye Movement (REM) sleep on both acclima-
characterized previously in terms of changes in regional cerebral tization nights were selected for the third, experimental, night.
blood flow (rCBF), using positron emission tomography (PET) During this night, PET scans were performed when the electro-
with oxygen-15-labeled water (H215O) (Hofle et al., 1997). Delta physiological recording showed steady characteristic patterns
activity recorded at the scalp was shown to correlate negatively following standard polysomnographic criteria (Rechtschaffen
with rCBF in the thalamus, brainstem reticular formation, and Kales, 1968), for the sleep stage in which the PET scan
cerebellum, anterior cingulate, and orbitofrontal cortex. According was intended. Waking (W) scans were obtained at rest just before
to the authors, these negative correlations reflected the active and after the sleep episode, with eyes closed in complete
inhibition of thalamocortical relay neurons in association with delta darkness. At least two W, two stage 2, and two stages 3 – 4
waves, and the neural substrates underlying the progressive scans were obtained in each subject, with a total of 161 scans for
attenuation of sensory awareness, motor responsiveness, and all subjects (W = 46, stage 2 = 50, stages 3 – 4 = 65).
arousal that occur during slow wave sleep (SWS). However,
several peculiarities of the experimental design adopted by Hofle et EEG acquisition
al. may have biased their results. First and most importantly, the
analysis looking for delta-related variations of regional brain Polysomnography was recorded using a Synamp (Neuroscan,
activity included scans obtained during wakefulness. Since delta NeuroSoft, Sterling, Virginia) system, at 500 or 1000 Hz sampling
oscillations are much more abundant during NREM sleep than rate, with a bandwidth 0.15 to 100 Hz. EEG on (at least) C3 – A2
during wakefulness, this procedure is confounding the genuine and C4 – A1 channels, electro-oculogram, and chin electromyo-
effect of delta generation with a condition effect (wakefulness gram were recorded on bipolar montages. Polysomnographic
versus NREM sleep). Furthermore, subjects were partially sleep- recordings were scored according to standard international criteria
deprived on the night preceding the PET acquisition, which might (Rechtschaffen and Kales, 1968).
have modified the observed pattern of rCBF. It is known that sleep
deprivation modifies Slow Wave Activity (SWA; 0.25 – 4 Hz) PET acquisition
during ensuing sleep (Borbely et al., 1981; Knoblauch et al., 2002),
and that extended wakefulness leads to rCBF decreases in frontal PET scans were acquired on a Siemens CTI 951 R 16/31
areas (Muzur et al., 2002). Finally, this study was based on a scanner in three-dimensional mode, reconstructed using a Hanning
reduced data sample (32 scans), which considerably weakens the filter (cutoff frequency 0.5 cycles/pixel) and corrected for
statistical power of the experiment. attenuation and background activity. A transmission scan was
In the present paper, our aim was to more specifically performed before the first emission scan of the night to allow a
characterize the cerebral correlates of delta activity during human measured attenuation correction. The subject’s head was stabilized
NREM sleep. To do so, we conducted a meta-analysis on a data set by a thermoplastic facemask secured to the head holder (Truscan
of 115 PET scans acquired during NREM sleep, in a non-sleep Imaging, Annapolis, Maryland), and a venous catheter was inserted
deprived population of normal subjects. In a preliminary analysis, in a left antebrachial vein. When polysomnography showed steady
we additionally included 46 scans acquired during wakefulness to characteristic patterns, cerebral blood flow was estimated, with a
provide a replication of the Hofle et al. study. maximum of 12 scans per subject. Each scan consisted of two
frames: a 30-s background frame and a 90-s active frame. Six
millicuries (mCi) equivalent to 222 megabecquerels (MBq) were
Methods injected for each scan, in 5 cubic centimeters (cc) saline, over a
period of 20 s, starting 10 s before the onset of the active frame.
Subjects and experimental protocol The infusion was totally automated in order not to awake the
subject during the scanning period.
Data were obtained from previous sleep studies conducted in
our center using the H215O technique (Maquet et al., 2000; PET and EEG analysis
Peigneux et al., 2003). These studies were approved by the Ethics
Committee of the Faculty of Medicine of the University of Liège. Raw polygraphic data were considered for the time intervals
All subjects were young, healthy, and right-handed male corresponding to the duration of each PET scan acquisition (90 s).
volunteers (n = 23; mean age = 22.9 years, range = 20.5 – 27, Each time series was visually checked for ocular and muscular
standard deviation = 3.5) who gave their written informed artifacts. Most 90-s EEG epochs were artifact-free; whole epochs
consent. Subjects were recruited after a detailed interview which and corresponding CBF data were eliminated in cases in which
assessed the regularity of their life habits, schedules, and quality artifacts lasted more than 5 s. Delta power spectral density (1.5 – 4
of sleep for the 3-month period before the experiment. They were Hz) was computed on C3 – A2 and C4 – A1 electrodes, using the
asked to keep a regular sleep – wake schedule and to fill in a sleep Welch’s averaged, modified periodogram method, with 4-s
diary for the 2 weeks before the first night in the scanner. They Hamming symmetric windows overlapping by 1 s (Werth et al.,
were also asked to abstain from alcohol and restrict caffeine for 1997). Averaged values of delta power over each active frame were
the week before the experiment. Subjects were accustomed to the used as covariates of interest in the analysis of rCBF modifications.
16 T.T. Dang-Vu et al. / NeuroImage 28 (2005) 14 – 21

PET data were analyzed using statistical parametric mapping were highly correlated to C3 – A2 values (Spearman correlation r >
(SPM2; Wellcome Department of Cognitive Neurology, Institute 0.95, Ps < 0.001) and yielded nearly identical results.
of Neurology, London, UK) implemented in MATLAB (Math- In the first analysis, negative correlations of delta power across
works, Sherborn, Massachusetts). For each subject, all scans NREM sleep and wakefulness were found with rCBF in a set of
were realigned together, then normalized to a standard PET brain areas including the medial frontal cortex, the orbitofrontal
template and smoothed (16 mm full width at half maximum). cortex, the basal forebrain and the anterior hypothalamus, the
Two analyses were separately conducted. The first one repro- striatum (putamen), the thalamus, the anterior part of the insula, the
duced the analysis published by Hofle et al. (1997). It included anterior cingulate gyrus, the posterior cingulate gyrus, and the
observations obtained during wakefulness, light and deep NREM precuneus (Fig. 1, left panel). Positive correlations were found in
sleep. The analysis looked for the brain areas in which rCBF parieto-occipital white matter, primary and secondary visual areas
correlated with delta power density values across these three ( Pscorr < 0.05) (data not shown).
different states. The second analysis aimed at specifying the In the second analysis, after the withdrawal of waking scans
cerebral correlates of delta rhythm specifically and exclusively and corresponding delta power values, negative correlations of
during NREM sleep. It included only data obtained during light delta power during NREM sleep only were found with rCBF in the
and deep NREM sleep and excluded waking scans. The analysis medial frontal cortex, the orbitofrontal cortex, the anterior
looked for the brain areas in which rCBF correlated with delta cingulate gyrus, the basal forebrain and the anterior hypothalamus,
power density values across light and deep NREM sleep only. the striatum (putamen), the anterior part of the insula, and the
The resulting set of voxel values for each analysis constituted a precuneus (Table 1; Fig. 1 middle/right panel, and Fig. 2 left
map of the t statistic [SPM(t)], thresholded at P  0.001 (Z  panel). A positive correlation was found in the white matter of the
3.09). Statistical inferences were then obtained at the voxel level left parietal region ( Pscorr < 0.05) (data not shown). At variance
corrected for multiple comparisons in the whole brain volume with the results of the first analysis, no correlation was detected
( P corr < 0.05). between NREM sleep delta power and rCBF in the thalamus, even
at a very low statistical threshold ( P < 0.05 uncorrected).

Results
Discussion
The results presented in this section are based on correlations
with values of delta power computed from C3 – A2 derivation. NREM sleep rhythms entrain large neuronal populations in
Values from C4 – A1 derivations are not reported here since they synchronous oscillations throughout the entire cerebrum. Accord-

Fig. 1. (Left panel) rCBF decreases as a function of delta power during wakefulness and NREM sleep (stages 2 – 4). Images sections are centered on the
ventromedial prefrontal cortex at the following coordinates: x = 2 mm, y = 48 mm, z = 8 mm ( P corr < 0.05) (Talairach and Tournoux, 1988). (Middle panel)
rCBF decreases as a function of delta power during NREM sleep after exclusion of waking scans and corresponding delta values. Images sections are centered
at the same coordinates ( P corr < 0.05). The color scale between left and middle sections indicates the range of Z values for the activated voxels in both panels.
(Right panel) Plot of the adjusted rCBF responses (arbitrary units) in the ventromedial prefrontal cortex in relation the adjusted delta power values (AV2) during
NREM sleep (corresponding to middle panel pictures): rCBF activity decreases when delta power increases. Each circle/cross represents one scan: green circles
are stage 2 scans, red crosses are stages 3 – 4 scans. The blue line is the linear regression.
 T.T. Dang-Vu et al. / NeuroImage 28 (2005) 14 – 21 17

Table 1 between delta power values and rCBF in the thalamus, the
Negative correlations between rCBF and scalp EEG delta activity during orbitofrontal cortex, and the anterior cingulate cortex (Fig. 1, left
NREM sleep panel) are in agreement with that preceding work.
Side x y z Z However, since delta oscillations are more profuse during
Medial frontal gyrus (BA 9/10) Left 2 48 8 7.36 NREM sleep than during wakefulness in normal human subjects
Medial frontal gyrus (BA 9/10) Right 2 48 8 7.26 and as this study was aimed at exploring the cerebral correlates of
Orbital gyrus (BA 11) Right 6 36 22 7.31 rhythms that characterize NREM sleep, the presence of waking
Orbital gyrus (BA 11) Left 2 60 10 6.42 values of delta power is likely to obscure the interpretation of the
Striatum (putamen) Right 32 0 4 6.70 results. Therefore, in our second analysis, data obtained during
Striatum (putamen) Left 16 8 6 5.89 wakefulness were discarded from the statistical analysis. As
Insula (anterior part) Right 38 6 8 6.41 shown in Fig. 1, this analysis run on data exclusively recorded
Insula (anterior part) Left 34 10 4 5.23
during NREM sleep yielded markedly different results, and
Anterior cingulate gyrus (BA 24) Left 2 26 24 6.35
notably failed to detect any significant correlation of delta activity
Precuneus (BA 31) Left 4 44 42 4.49
Basal forebrain (caudal orbital) Right 14 18 16 5.63 with rCBF in the thalamus. The discrepancy suggests that
Basal forebrain (anterior hypothalamus) Right 6 4 10 5.10 wakefulness data might indeed have played an important
confounding effect in the identification of the cerebral correlates
x, y, and z coordinates (Talairach and Tournoux, 1988) correspond to the
local maxima of significant negative correlation between rCBF and activity of delta waves.
in the delta frequency band (1.5 – 4.0 Hz) during stages 2 – 4 of NREM The present analysis shows, in non-sleep-deprived normal
sleep. x, distance (in millimeters) to right (+) or left ( ) of the midsagittal young adults, that rCBF in a set of brain areas is negatively
line; y, distance anterior (+) or posterior ( ) to the anterior commissure; z, correlated during NREM sleep with delta power values measured
distance above (+) or below ( ) the intercommissural line. Z = statistical Z at the central scalp. These regions include cortical areas (medial
score. For cortical locations, Brodmann’s cytoarchitectonic areas are given frontal and orbitofrontal cortex, anterior cingulate gyrus, anterior
(BA). All reported values are significant at P corr < 0.05 after correction for part of insula, precuneus), the basal ganglia, and the basal forebrain
multiple comparisons in the whole brain volume. Positive correlations are (Fig. 2, left panel). This distribution is actually closely similar to
not shown. the previously published map of brain areas in which rCBF
significantly decreased during NREM sleep as compared to REM
ingly, a global decrease in the cerebral blood flow is reported in sleep and wakefulness, except for the absence of the thalamus
(deep) NREM sleep (Braun et al., 1997; Kajimura et al., 1999). (Maquet et al., 1997) (Fig. 2, right panel). These similarities
Here, our aim was not to confirm this global decrease in CBF with underline the notion of delta power as a prominent feature of
delta power density but to specify the brain areas where the NREM sleep.
regional blood flow is most decreased in relation with values of
delta power, calculated from scalp EEG data at central electrodes. Thalamic versus cortical delta oscillations
First, we replicated the previously described analysis (Hofle et
al., 1997), in which scans and delta power values acquired during As mentioned above, our analysis did not show any significant
stages 2 – 4 of NREM sleep but also during wakefulness were relationship between delta activity and rCBF in the thalamus in
included. The results of this analysis yielding negative correlations NREM sleep, even at low statistical thresholds. In the comparison

Fig. 2. (Left panel) rCBF decreases as a function of delta power during NREM sleep (stages 2 – 4). Image sections are displayed on different levels of the z axis
as indicated on the top of each picture (Talairach and Tournoux, 1988). The color scale indicates the range of Z values for the activated voxels. Displayed
voxels are significant at P < 0.05 after correction for multiple comparisons. (Right panel) Statistical map showing the brain areas in which rCBF decreases
during NREM sleep as compared to wakefulness and REM sleep (Maquet et al., 1997). Note the striking similarity of the regional blood flow distribution
between left and right panels. Copyright 1997 by the Society for Neuroscience.
18 T.T. Dang-Vu et al. / NeuroImage 28 (2005) 14 – 21

between wakefulness and NREM sleep, thalamic deactivation is delta waves, at a frequency of 2 – 4 Hz, generated by both regular-
the most reproducible pattern observed by neuroimaging techni- spiking and intrinsically bursting cortical neurons, are grouped
ques in humans (Maquet, 2000): it was demonstrated in glucose within sequences recurring with the slow rhythm (Steriade et al.,
metabolism studies (Buchsbaum et al., 1989; Maquet et al., 1990) 1993a). Likewise in humans, sleep EEG data have described the
as well as in more recent H215O studies (Braun et al., 1997; Maquet periodic recurrence of sequential mean amplitudes of delta waves
et al., 1997; Andersson et al., 1998). Regional CBF decrease in the with the rhythm of slow oscillation (Steriade et al., 1993b).
thalamus during NREM sleep is explained by the bursting mode of Therefore, intracortical synchronization of cortical delta waves
neuronal firing underlying spindle and delta oscillation generation, appears to be a plausible hypothesis. Our results speak for a
summarized in the Introduction (Steriade and Amzica, 1998; regional modulation of these synchronization processes within the
Steriade, 1999). Due to the temporal resolution of PET scans, the cortex, irrespective of their thalamocortical or genuinely cortical
net effect of the hyperpolarization periods on rCBF exceeds that of origin. Further studies are needed to clarify these issues.
the bursts resulting in a decrease in rCBF in thalamic nuclei
(Maquet, 2000). However, our analysis is methodologically Neocortical correlates of delta waves: a role for ventromedial
different from these previous PET studies. Here, the functional prefrontal areas?
relationship between rCBF and delta power density was assessed
exclusively in NREM sleep. The absence of significant correlation Our results show that delta activity is not homogeneously
between the thalamic rCBF and delta power density suggests that correlated with cortical rCBF but correlates predominantly with
the bursting firing pattern adopted by the thalamic neurons has a rCBF in the medial frontal, anterior cingulate, and orbitofrontal
similar impact on local cerebral blood flow whatever the rhythm regions (Figs. 1 and 2). These areas were commonly referred to as
generated (spindle or delta), and irrespective of the degree of ventromedial prefrontal cortex (VMPFC) (Damasio, 1994).
hyperpolarization or cellular processes involved in the generation A frontal predominance of delta activity during NREM sleep
of these sleep oscillations in TC neurons. In contrast, in some has been outlined in several electrophysiological studies (Buchs-
cortical areas, in the basal forebrain and in the basal ganglia, rCBF baum et al., 1982; Zeitlhofer et al., 1993; Werth et al., 1997). This
does correlate with delta power density. This result shows a EEG frontal (Finelli et al., 2001) or medio-frontal (Happe et al.,
different modulation of rCBF in these areas in response to delta 2002) dominance might thus be linked to structures inside the
power and suggests that this peculiar rCBF distribution is related to VMPFC. However, at present, the implication of the VMPFC in
the regional modulation of delta oscillation within the cortex. the generation of delta waves remains speculative. Detailed
Electrophysiological data in animals have shown the existence of a investigations in animals should particularly focus on the
cortical delta rhythm, distinct from the clock-like delta waves, generation of delta rhythm in cortical areas homologous to
which survive extensive thalamectomy (Frost et al., 1966; Ball et VMPFC.
al., 1977; Steriade et al., 1993a). More recent animal studies have One interpretation for the frontal predominance of rCBF
shown that selective destruction of thalamic RE neurons leads to changes in relation to EEG delta power is that frontal association
reduced delta power values after 1 day. However, there was no areas display higher sleep intensities because of an intensive
complete abolition in most animals and a recovery of delta activity daytime use. Arguments for this theory came from the demon-
was observed after 2 weeks (Marini et al., 2000). The specific stration that the rebound in SWA after extended wakefulness yields
characteristics, intracortical topography, and cellular mechanisms a clear frontal predominance (Cajochen et al., 1999; Finelli et al.,
of this cortical delta rhythm are still unknown. 2000; Knoblauch et al., 2002). These regional changes in EEG
Local field potentials and multi-unit recordings in the cat have been corroborated to different levels of activation by
cerebral cortex have shown a remarkable large-scale synchroniza- neuroimaging PET studies. Frontal association areas including
tion of delta waves during natural SWS (Destexhe et al., 1999). A VMPFC are among the most markedly deactivated brain regions
computational model proposed a mechanism for such large-scale during SWS, potentially reflecting the frontal dominance of delta
synchrony based on thalamocortical loops (Destexhe et al., 1998). activity, but also among the most activated areas during wakeful-
Our results do not allow to discard the involvement of thalamus in ness (Braun et al., 1997; Andersson et al., 1998; Kajimura et al.,
this synchronization mechanism and therefore do not challenge this 1999; Maquet, 2000).
hypothesis. Nevertheless, animal data suggest that delta waves are Also, VMPFC is one of the most active brain areas during the
synchronized by a slow oscillation (usually 0.6 to 1.0 Hz), awake resting state (Gusnard and Raichle, 2001). While the
generated intracortically as it survives extensive thalamic lesions reasons for such a high waking activity are not clearly identified, it
(Steriade et al., 1993a), disappearing in the thalamus of decorti- appears that VMPFC is involved in several important cognitive
cated animals (Timofeev and Steriade, 1996), and whose synchro- processes. For instance, VMPFC has been related to action
nization is disrupted by interruption of intracortical synaptic monitoring (Luu et al., 2000), planning of tasks executed in
circuits (Amzica and Steriade, 1995). This oscillation would be expected sequences (Koechlin et al., 2000), explicit processing of
able to group the other sleep rhythms, including spindles (Steriade, sequential material (Destrebecqz et al., 2003), self-referential
1999). It is known that during the depolarizing phase of the slow mental activity (Gusnard et al., 2001), guessing (Elliott et al.,
oscillation, corticothalamic synaptic volleys succeed in synchro- 1999), holding in mind goals while processing secondary goals
nizing pools of thalamic cells by activating GABAergic thalamic (Koechlin et al., 1999), and decision making (Bechara et al., 1998).
RE neurons that project to thalamic relay cells and hyperpolarize On the other hand, VMPFC is closely related to brain arousal
them (Steriade et al., 1991). This process results in the coherence systems because of its widespread connections with many
of the thalamic component of delta waves. On the other hand, the structures of a distributed ascending activating system (Morecraft
intracortical component of delta waves has not been systematically et al., 1992). Besides, increasing activity in VMPFC was found in
studied at the intracellular level. Although its hypothetical situations of heightened arousal levels associated with some
synchronization is less documented, it has been shown that typical psychiatric conditions such as obsessive – compulsive disorder
 T.T. Dang-Vu et al. / NeuroImage 28 (2005) 14 – 21 19

(Baxter et al., 1987), post-traumatic stress disorder (Rauch and sleep, as seen in other PET studies (Braun et al., 1997; Maquet
Shin, 1997), and depression (Nofzinger et al., 2000). These data et al., 1997; Kajimura et al., 1999), could thus be also related to
consistently support the interpretation that VMPFC is a site of a lower propensity to arousal.
arousal regulation and important higher-order cognitive processes We found a negative correlation of delta activity with rCBF
during wakefulness, which undergoes a deep deactivation during measured in the anterior part of the insula. However, there is no
NREM sleep reflected by the frontal predominance of EEG delta evidence at present for a clear relationship between this structure
activity. and sleep-dependent processes. The anterior insula has been found
deactivated in a study assessing rCBF during SWS (Braun et al.,
Other negative correlations of rCBF with delta 1997). Braun et al. suggested that the anterior insula, as belonging
to the paralimbic structures, would serve as an interface between the
The spatial resolution of PET does not allow to differentiate the external and internal milieus, allowing the limbic core structures
anterior hypothalamus from the basal forebrain itself (Maquet, (hippocampus, amygdala) to be functionally disconnected during
2000). Basal forebrain/anterior hypothalamus (BF/AH) forms a SWS from the brain regions that directly mediate their interactions
functionally and structurally heterogeneous structure (Szymusiak, with the external environment. This disengagement of limbic
1995) implicated in both arousal and sleep generation processes. structures, entrained by the paralimbic deactivation, would be a
GABAergic BF/AH neurons promote delta activity during SWS precondition for their homeostatic restoration (Braun et al., 1997).
(Jones, 2004). They would represent a minority (16%) of BF/AH Finally, the precuneus is known to be deactivated during SWS
neurons. A majority (74%) of BF/AH neurons (including (Braun et al., 1997; Maquet et al., 1997; Andersson et al., 1998)
cholinergic neurons) are involved in cortical activation during but also, intriguingly, during mental states of decreased conscious-
wakefulness and REM sleep (Lee et al., 2004). Thus, the negative ness like pharmacological sedation (Fiset et al., 1999), hypnotic
correlation of rCBF in BF/AH with delta power would be (Maquet et al., 1999) and vegetative states (Laureys et al., 1999).
compatible with a lower activity of these arousal-promoting The precuneus is also a region particularly active during wakeful-
neurons during NREM sleep. ness, which would suggest that its activity reflects more brain
Arguments for a role of striatum in sleep regulation have operations taking place in conscious wakefulness than in sleep
emerged form pathologic conditions involving a dysfunction of (Maquet, 2000). The negative correlation of delta power with
basal ganglia. In particular, changes in the sleep – wake cycle, rCBF in precuneus probably represents a decrease during NREM
including an irregular delta activity and a decrease in SWS, have sleep of waking-dependent processes taking place in the
been reported in patients suffering from Huntington’s disease precuneus, along with mechanisms potentially mediating levels
(Sishta et al., 1974; Wiegand et al., 1991b), a pathology of consciousness and deserving further investigations.
characterized by severe damage in the striatum (Sanberg and
Coyle, 1984). Moreover, a direct association has been outlined
between striatal atrophy and reduced SWS in Huntington’s disease Conclusions
(Wiegand et al., 1991a). However, it is unclear whether these sleep
disturbances are caused by an impaired function of striatum in This analysis of a set of PET scans acquired in non-sleep-
sleep regulation or are simply non-specific consequences of the deprived subjects characterizes the cerebral correlates of delta waves
disease. Several elements support the former hypothesis. It is during NREM sleep with rCBF in a set of brain areas. These regions
known that choreic movements largely cease during sleep in include areas in which rCBF decreases during SWS compared to
patients with Huntington’s disease (Fish et al., 1991). Experimental REM sleep and wakefulness (Maquet et al., 1997), underlining the
models have shown that the sleep disturbances induced by striatal notion of delta activity as a valuable feature of NREM sleep. A
excitotoxic lesions in rats occurred on the 30th day post-lesion, at a strong association was observed between rCBF in the ventromedial
time when motor abnormalities were no longer observed (Mena- prefrontal regions and delta power, in agreement with electro-
Segovia et al., 2002). These data suggest the possible participation physiological studies. Noticeably, the absence of thalamic correla-
of the striatum in the regulation of the sleep – waking cycle, tion stresses the importance of an extra-thalamic delta rhythm
independently of locomotor activity. Two main alternative hypoth- among the synchronous NREM sleep oscillations. Consequently,
eses may be put forward to explain the mechanisms by which the this rCBF distribution might preferentially reflect a particular
striatum would be implicated in sleep – wake regulation and could modulation of the cellular processes involved in the generation of
then be in agreement with the negative correlation of rCBF in this cortical delta waves during NREM sleep. A genuine localization of
area with delta activity. The first one, which integrates the striatum the brain sites of generation for these slow brain rhythms exceeds the
into a cortical – basal ganglia – thalamic – cortical loop, relies on the scope of this meta-analysis. A combination of advanced neuro-
observation that frontal cortex and thalamus are both among the imaging and electrophysiological topographical techniques in
most deactivated brain areas during SWS and major afferents to future, dedicated, studies should enlighten this crucial issue.
basal ganglia (Macchi et al., 1984; Selemon and Goldman-Rakic,
1985; Sadikot et al., 1990). Activity in these regions could entrain
the basal ganglia neuronal population in highly synchronized Acknowledgments
oscillations (Maquet, 2000) with long phases of hyperpolarization
alternating with bursts of discharges (Wilson, 1993). On the other The authors thank the staff of the Cyclotron Research Centre
hand, recent experiments suggest a striatal role in arousal processes for technical professional assistance. The study was supported by
by its connections with the pediculopontine tegmental nucleus FNRS (Fonds National de la Recherche Scientifique), FMRE
(PPT), where afferents arising from the striatum would entrain PPT (Fondation Médicale Reine Elisabeth), Research Fund of ULg, and
disinhibition and result in cortical activation (Mena-Segovia and Interuniversity Attraction Poles Programme—Belgian Science
Giordano, 2003). Decreasing activity in the striatum during NREM Policy. PM, SL, and CP were supported by FNRS.
20 T.T. Dang-Vu et al. / NeuroImage 28 (2005) 14 – 21

References Fiset, P., Paus, T., Daloze, T., Plourde, G., Meuret, P., Bonhomme, V., Hajj-
Ali, N., Backman, S.B., Evans, A.C., 1999. Brain mechanisms of
Achermann, P., Borbely, A.A., 1997. Low-frequency (<1 Hz) oscillations in propofol-induced loss of consciousness in humans: a positron emission
the human sleep electroencephalogram. Neuroscience 81, 213 – 222. tomographic study. J. Neurosci. 19, 5506 – 5513.
Amzica, F., Steriade, M., 1995. Disconnection of intracortical synaptic Fish, D.R., Sawyers, D., Allen, P.J., Blackie, J.D., Lees, A.J., Marsden,
linkages disrupts synchronization of a slow oscillation. J. Neurosci. 15, C.D., 1991. The effect of sleep on the dyskinetic movements of
4658 – 4677. Parkinson’s disease, Gilles de la Tourette syndrome, Huntington’s
Andersson, J.L., Onoe, H., Hetta, J., Lidstrom, K., Valind, S., Lilja, A., disease, and torsion dystonia. Arch. Neurol. 48, 210 – 214.
Sundin, A., Fasth, K.J., Westerberg, G., Broman, J.E., Watanabe, Y., Frost Jr., J.D., Kellaway, P., Gol, A., 1966. Single-unit discharges in
Langstrom, B., 1998. Brain networks affected by synchronized sleep isolated cerebral cortex. Exp. Neurol. 14, 305 – 316.
visualized by positron emission tomography. J. Cereb. Blood Flow Gusnard, D.A., Raichle, M.E., 2001. Searching for a baseline: functional
Metab. 18, 701 – 715. imaging and the resting human brain. Nat. Rev., Neurosci. 2, 685 – 694.
Ball, G.J., Gloor, P., Schaul, N., 1977. The cortical electromicrophysiology Gusnard, D.A., Akbudak, E., Shulman, G.L., Raichle, M.E., 2001. Medial
of pathological delta waves in the electroencephalogram of cats. prefrontal cortex and self-referential mental activity: relation to a default
Electroencephalogr. Clin. Neurophysiol. 43, 346 – 361. mode of brain function. Proc. Natl. Acad. Sci. U. S. A. 98, 4259 – 4264.
Baxter Jr., L.R., Phelps, M.E., Mazziotta, J.C., Guze, B.H., Schwartz, Happe, S., Anderer, P., Gruber, G., Klosch, G., Saletu, B., Zeitlhofer, J.,
J.M., Selin, C.E., 1987. Local cerebral glucose metabolic rates in 2002. Scalp topography of the spontaneous K-complex and of delta-
obsessive – compulsive disorder. A comparison with rates in waves in human sleep. Brain Topogr. 15, 43 – 49.
unipolar depression and in normal controls. Arch. Gen. Psychiatry Hofle, N., Paus, T., Reutens, D., Fiset, P., Gotman, J., Evans, A.C., Jones,
44, 211 – 218. B.E., 1997. Regional cerebral blood flow changes as a function of delta
Bechara, A., Damasio, H., Tranel, D., Anderson, S.W., 1998. Dissociation and spindle activity during slow wave sleep in humans. J. Neurosci. 17,
of working memory from decision making within the human prefrontal 4800 – 4808.
cortex. J. Neurosci. 18, 428 – 437. Jones, B.E., 2004. Activity, modulation and role of basal forebrain
Borbely, A.A., Baumann, F., Brandeis, D., Strauch, I., Lehmann, D., 1981. cholinergic neurons innervating the cerebral cortex. Prog. Brain Res.
Sleep deprivation: effect on sleep stages and EEG power density in 145, 157 – 169.
man. Electroencephalogr. Clin. Neurophysiol. 51, 483 – 495. Kajimura, N., Uchiyama, M., Takayama, Y., Uchida, S., Uema, T., Kato,
Braun, A.R., Balkin, T.J., Wesenten, N.J., Carson, R.E., Varga, M., M., Sekimoto, M., Watanabe, T., Nakajima, T., Horikoshi, S., Ogawa,
Baldwin, P., Selbie, S., Belenky, G., Herscovitch, P., 1997. Regional K., Nishikawa, M., Hiroki, M., Kudo, Y., Matsuda, H., Okawa, M.,
cerebral blood flow throughout the sleep – wake cycle. An H2(15)O Takahashi, K., 1999. Activity of midbrain reticular formation and
PET study. Brain 120 (Pt 7), 1173 – 1197. neocortex during the progression of human non-rapid eye movement
Buchsbaum, M.S., Mendelson, W.B., Duncan, W.C., Coppola, R., sleep. J. Neurosci. 19, 10065 – 10073.
Kelsoe, J., Gillin, J.C., 1982. Topographic cortical mapping of Knoblauch, V., Krauchi, K., Renz, C., Wirz-Justice, A., Cajochen, C., 2002.
EEG sleep stages during daytime naps in normal subjects. Sleep 5, Homeostatic control of slow-wave and spindle frequency activity during
248 – 255. human sleep: effect of differential sleep pressure and brain topography.
Buchsbaum, M.S., Gillin, J.C., Wu, J., Hazlett, E., Sicotte, N., Dupont, Cereb. Cortex 12, 1092 – 1100.
R.M., Bunney Jr., W.E., 1989. Regional cerebral glucose metabolic rate Koechlin, E., Basso, G., Pietrini, P., Panzer, S., Grafman, J., 1999. The
in human sleep assessed by positron emission tomography. Life Sci. 45, role of the anterior prefrontal cortex in human cognition. Nature 399,
1349 – 1356. 148 – 151.
Cajochen, C., Foy, R., Dijk, D.J., 1999. Frontal predominance of a relative Koechlin, E., Corrado, G., Pietrini, P., Grafman, J., 2000. Dissociating the
increase in sleep delta and theta EEG activity after sleep loss in humans. role of the medial and lateral anterior prefrontal cortex in human
Sleep Res. Online 2, 65 – 69. planning. Proc. Natl. Acad. Sci. U. S. A. 97, 7651 – 7656.
Damasio, A., 1994. Descartes’s Error. Emotion, Reason, and the Human Laureys, S., Goldman, S., Phillips, C., Van Bogaert, P., Aerts, J., Luxen, A.,
Brain. Grosset/Putnam, New York, NY. Franck, G., Maquet, P., 1999. Impaired effective cortical connectivity in
De Gennaro, L., Ferrara, M., 2003. Sleep spindles: an overview. Sleep Med. vegetative state: preliminary investigation using PET. NeuroImage 9,
Rev. 7, 423 – 440. 377 – 382.
Destexhe, A., Contreras, D., Steriade, M., 1998. Mechanisms underlying Lee, M.G., Manns, I.D., Alonso, A., Jones, B.E., 2004. Sleep-wake related
the synchronizing action of corticothalamic feedback through inhibition discharge properties of basal forebrain neurons recorded with micro-
of thalamic relay cells. J. Neurophysiol. 79, 999 – 1016. pipettes in head-fixed rats. J. Neurophysiol. 92 (2), 1182 – 1198.
Destexhe, A., Contreras, D., Steriade, M., 1999. Spatiotemporal analysis of Luu, P., Flaisch, T., Tucker, D.M., 2000. Medial frontal cortex in action
local field potentials and unit discharges in cat cerebral cortex during monitoring. J. Neurosci. 20, 464 – 469.
natural wake and sleep states. J. Neurosci. 19, 4595 – 4608. Macchi, G., Bentivoglio, M., Molinari, M., Minciacchi, D., 1984. The
Destrebecqz, A., Peigneux, P., Laureys, S., Degueldre, C., Del Fiore, G., thalamo-caudate versus thalamo-cortical projections as studied in the
Aerts, J., Luxen, A., van der Linden, M., Cleeremans, A., Maquet, P., cat with fluorescent retrograde double labeling. Exp. Brain Res. 54,
2003. Cerebral correlates of explicit sequence learning. Brain Res. 225 – 239.
Cogn. Brain Res. 16, 391 – 398. Maquet, P., 2000. Functional neuroimaging of normal human sleep by
Dijk, D.J., Hayes, B., Czeisler, C.A., 1993. Dynamics of electroencephalo- positron emission tomography. J. Sleep Res. 9, 207 – 231.
graphic sleep spindles and slow wave activity in men: effect of sleep Maquet, P., Dive, D., Salmon, E., Sadzot, B., Franco, G., Poirrier, R., von
deprivation. Brain Res. 626, 190 – 199. Frenckell, R., Franck, G., 1990. Cerebral glucose utilization during
Elliott, R., Rees, G., Dolan, R.J., 1999. Ventromedial prefrontal cortex sleep – wake cycle in man determined by positron emission tomog-
mediates guessing. Neuropsychologia 37, 403 – 411. raphy and [18F]2-fluoro-2-deoxy-d-glucose method. Brain Res. 513,
Finelli, L.A., Baumann, H., Borbely, A.A., Achermann, P., 2000. Dual 136 – 143.
electroencephalogram markers of human sleep homeostasis: correlation Maquet, P., Degueldre, C., Delfiore, G., Aerts, J., Peters, J.M., Luxen, A.,
between theta activity in waking and slow-wave activity in sleep. Franck, G., 1997. Functional neuroanatomy of human slow wave sleep.
Neuroscience 101, 523 – 529. J. Neurosci. 17, 2807 – 2812.
Finelli, L.A., Borbely, A.A., Achermann, P., 2001. Functional topography Maquet, P., Faymonville, M.E., Degueldre, C., Delfiore, G., Franck, G.,
of the human nonREM sleep electroencephalogram. Eur. J. Neurosci. Luxen, A., Lamy, M., 1999. Functional neuroanatomy of hypnotic state.
13, 2282 – 2290. Biol. Psychiatry 45, 327 – 333.
 T.T. Dang-Vu et al. / NeuroImage 28 (2005) 14 – 21 21

Maquet, P., Laureys, S., Peigneux, P., Fuchs, S., Petiau, C., Phillips, C., Selemon, L.D., Goldman-Rakic, P.S., 1985. Longitudinal topography
Aerts, J., Del Fiore, G., Degueldre, C., Meulemans, T., Luxen, A., and interdigitation of corticostriatal projections in the rhesus
Franck, G., Van Der Linden, M., Smith, C., Cleeremans, A., 2000. monkey. J. Neurosci. 5, 776 – 794.
Experience-dependent changes in cerebral activation during human Sishta, S.K., Troupe, A., Marszalek, K.S., Kremer, L.M., 1974. Hunting-
REM sleep. Nat. Neurosci. 3, 831 – 836. ton’s chorea: an electroencephalographic and psychometric study.
Marini, G., Ceccarelli, P., Mancia, M., 2000. Effects of bilateral micro- Electroencephalogr. Clin. Neurophysiol. 36, 387 – 393.
injections of ibotenic acid in the thalamic reticular nucleus on delta Steriade, M., 1999. Coherent oscillations and short-term plasticity in
oscillations and sleep in freely-moving rats. J. Sleep Res. 9, 359 – 366. corticothalamic networks. Trends Neurosci. 22, 337 – 345.
McCormick, D.A., Bal, T., 1997. Sleep and arousal: thalamocortical Steriade, M., 2003. The corticothalamic system in sleep. Front Biosci. 8,
mechanisms. Annu. Rev. Neurosci. 20, 185 – 215. d878 – d899.
Mena-Segovia, J., Giordano, M., 2003. Striatal dopaminergic stimulation Steriade, M., Amzica, F., 1998. Coalescence of sleep rhythms and their
produces c-Fos expression in the PPT and an increase in wakefulness. chronology in corticothalamic networks. Sleep Res. Online 1, 1 – 10.
Brain Res. 986, 30 – 38. Steriade, M., Dossi, R.C., Nunez, A., 1991. Network modulation of a slow
Mena-Segovia, J., Cintra, L., Prospero-Garcia, O., Giordano, M., 2002. intrinsic oscillation of cat thalamocortical neurons implicated in sleep
Changes in sleep – waking cycle after striatal excitotoxic lesions. Behav. delta waves: cortically induced synchronization and brainstem choli-
Brain Res. 136, 475 – 481. nergic suppression. J. Neurosci. 11, 3200 – 3217.
Morecraft, R.J., Geula, C., Mesulam, M.M., 1992. Cytoarchitecture and Steriade, M., Nunez, A., Amzica, F., 1993a. Intracellular analysis of
neural afferents of orbitofrontal cortex in the brain of the monkey. relations between the slow (<1 Hz) neocortical oscillation and
J. Comp. Neurol. 323, 341 – 358. other sleep rhythms of the electroencephalogram. J. Neurosci. 13,
Muzur, A., Pace-Schott, E.F., Hobson, J.A., 2002. The prefrontal cortex in 3266 – 3283.
sleep. Trends Cogn. Sci. 6, 475 – 481. Steriade, M., Nunez, A., Amzica, F., 1993b. A novel slow (<1 Hz)
Nofzinger, E.A., Price, J.C., Meltzer, C.C., Buysse, D.J., Villemagne, V.L., oscillation of neocortical neurons in vivo: depolarizing and hyper-
Miewald, J.M., Sembrat, R.C., Steppe, D.A., Kupfer, D.J., 2000. polarizing components. J. Neurosci. 13, 3252 – 3265.
Towards a neurobiology of dysfunctional arousal in depression: the Szymusiak, R., 1995. Magnocellular nuclei of the basal forebrain:
relationship between beta EEG power and regional cerebral glucose substrates of sleep and arousal regulation. Sleep 18, 478 – 500.
metabolism during NREM sleep. Psychiatry Res. 98, 71 – 91. Talairach, J., Tournoux, P., 1988. Co-Planar Stereotaxic Atlas of the Human
Peigneux, P., Laureys, S., Fuchs, S., Destrebecqz, A., Collette, F., Brain. George Thieme, Stuttgart.
Delbeuck, X., Phillips, C., Aerts, J., Del Fiore, G., Degueldre, C., Timofeev, I., Steriade, M., 1996. Low-frequency rhythms in the
Luxen, A., Cleeremans, A., Maquet, P., 2003. Learned material content thalamus of intact-cortex and decorticated cats. J. Neurophysiol. 76,
and acquisition level modulate cerebral reactivation during posttraining 4152 – 4168.
rapid-eye-movements sleep. NeuroImage 20, 125 – 134. Werth, E., Achermann, P., Borbely, A.A., 1997. Fronto-occipital EEG
Rauch, S.L., Shin, L.M., 1997. Functional neuroimaging studies in power gradients in human sleep. J. Sleep Res. 6, 102 – 112.
posttraumatic stress disorder. Ann. N. Y. Acad. Sci. 821, 83 – 98. Wiegand, M., Moller, A.A., Schreiber, W., Lauer, C., Krieg, J.C., 1991a.
Rechtschaffen, A., Kales, A., 1968. A Manual of Standardized Terminol- Brain morphology and sleep EEG in patients with Huntington’s disease.
ogy, Techniques and Scoring System for Sleep Stages of Human Eur. Arch. Psychiatry Clin. Neurosci. 240, 148 – 152.
Subjects. Brain Information Service/Brain Research Institute, Univer- Wiegand, M., Moller, A.A., Lauer, C.J., Stolz, S., Schreiber, W., Dose, M.,
sity of California, Los Angeles. Krieg, J.C., 1991b. Nocturnal sleep in Huntington’s disease. J. Neurol.
Sadikot, A.F., Parent, A., Francois, C., 1990. The centre median and 238, 203 – 208.
parafascicular thalamic nuclei project respectively to the sensorimotor Wilson, C.J., 1993. The generation of natural firing patterns in neostriatal
and associative-limbic striatal territories in the squirrel monkey. Brain neurons. Prog. Brain Res. 99, 277 – 297.
Res. 510, 161 – 165. Zeitlhofer, J., Anderer, P., Obergottsberger, S., Schimicek, P., Lurger, S.,
Sanberg, P.R., Coyle, J.T., 1984. Scientific approaches to Huntington’s Marschnigg, E., Saletu, B., Deecke, L., 1993. Topographic mapping of
disease. CRC Crit. Rev. Clin. Neurobiol. 1, 1 – 44. EEG during sleep. Brain Topogr. 6, 123 – 129.