Clinical Neurophysiology 120 (2009) 248256

Contents lists available at ScienceDirect

Clinical Neurophysiology
journal homepage: www.elsevier.com/locate/clinph

All-night EEG power spectral analysis of the cyclic alternating pattern

at different ages
Oliviero Bruni a,*, Luana Novelli a, Elena Finotti b, Anna Luchetti c, Giordana Uggeri a, Debora Aric d,
Raffaele Ferri d
a
Pediatric Sleep Center, Department of Developmental Neurology and Psychiatry, Sapienza University, Rome, Italy
b
Department of Pediatrics, University of Padua, Padua, Italy
c
Child Neuropsychiatry Division, 2nd School of Medicine, Sapienza University, Rome, Italy
d
Sleep Research Centre, Department of Neurology I.C., Oasi Institute (IRCCS), Troina, Italy

a r t i c l e i n f o a b s t r a c t

Article history: Objective: To analyze in detail the frequency content of the different EEG components of the Cyclic Alter-
Accepted 3 November 2008 nating Pattern (CAP) in the whole sleep of pre-school and school age children compared to normal young
Available online 24 December 2008 adults.
Methods: Fourteen pre-school age and 18 school age children and 16 adults were included in this study.
Keywords: Each participant underwent a polysomnographic overnight recording, after an adaptation night; sleep
Sleep stages and CAP were scored following standard criteria. Average spectra were obtained for each CAP con-
Sleep stages
dition from the signal recorded from C3/A2 or C4/A1, separately in sleep stage 2 and slow-wave sleep
Cyclic alternating pattern
Fast Fourier transform
(SWS), for each subject.
EEG power spectra Results: The analysis of the relative power density in the three groups showed that in sleep stage 2 and in
Pre-school age SWS, CAP A1, A2, A3 subtypes had a signicantly higher power in all frequency ranges in pre-school chil-
School age dren than in adults, while school children differed mainly for the lower frequencies (<7 Hz). For non-CAP,
Adults pre-school and school children differed from adults at almost all frequencies analyzed. Generally, A1, A2
and A3 showed clear spectral differences in the three different groups of subjects with pre-school age
children showing slightly less evident differences.
Conclusions: CAP subtypes are characterized by clearly different spectra at different ages and also the
same subtype shows a different power spectrum, during sleep stage 2 or SWS. This study shows that
pre-school children have a different structure of sleep, especially from the microstructural (CAP) point
of view: the differences are evident for all the CAP components and for non-CAP in almost all the fre-
quency bands. This nding might be associated to the age-related delta decline in the 03 Hz frequency
reported in children of the same age.
Signicance: Our data seem to provide information not available before and useful for the understanding
of the impact of CAP on the sleep EEG neurophysiological dynamics at different ages. This type of infor-
mation is crucial for a more adequate interpretation of data provided by a growing number of studies
analyzing CAP in groups of pediatric patients.
2008 International Federation of Clinical Neurophysiology. Published by Elsevier Ireland Ltd. All rights
reserved.

1. Introduction
Non-REM sleep is characterized by the recurring appearance of
EEG transients that translate the sleep state stability or instability.
These EEG patterns are dened by phasic events, such as sleep
spindles, K-complexes, vertex waves, and short-lasting arousals
and are coded phases A of the Cyclic Alternating Pattern or CAP
* Corresponding author. Address: Center for Pediatric Sleep Disorders, Depart-
ment of Developmental Neurology and Psychiatry, University of Rome La Sapienza,
Via dei Sabelli 108, 00185 Rome, Italy. Tel.: +39 0644712257; fax: +39 06 4957857.
E-mail address: oliviero.bruni@uniroma1.it (O. Bruni).
(Terzano et al., 1985, 1988; Halsz, 1998). CAP is therefore a peri-
odic EEG activity of non-REM sleep characterized by sequences of
transient electro-cortical activations (phase A of the cycle) that
are distinct from the background EEG activity (phase B of the
cycle). Sequences of CAP are orderly distributed in NREM sleep,
and the percentage of CAP time to NREM sleep time (CAP rate) is
considered to be a physiologic marker of NREM sleep instability.
CAP A phases have been subdivided into different subtypes: A1,
A2 and A3 based on the frequency content (Terzano et al., 2001;
Parrino et al., 2001). A1 subtypes show a power spectrum charac-
terized by a predominant peak in the frequency range of 0.25
2.5 Hz (De Carli et al., 2004; Ferri et al., 2000, 2005a,b); these

1388-2457/$34.00 2008 International Federation of Clinical Neurophysiology. Published by Elsevier Ireland Ltd. All rights reserved.
doi:10.1016/j.clinph.2008.11.001
 O. Bruni et al. / Clinical Neurophysiology 120 (2009) 248256
frequencies, during CAP events, are likely to be generated at the le-
vel of the frontal areas of the brain and the scalp topographic map-
ping and localization of their eventual scalp generators by means
of the LORETA method (Pascual-Marqui et al., 1994, 1999) conrms
this idea (Ferri et al., 2005b). On the contrary, the high-frequency
components characterizing CAP A2 and A3 phases have been
shown to involve both midline and hemispheric areas within the
parietal and occipital areas (Ferri et al., 2005b).
Both visual and automatic detection of CAP are based mostly on
the frequency content of A phases and on that of the ongoing back-
ground activity, together with their amplitude. Ferri et al. (2005a)
analyzed in detail the spectral frequency content of the different
CAP EEG components in normal young adults in order to obtain
information functional for their detection, showing that CAP sub-
types are characterized by different spectra and also the same sub-
type shows a different power spectrum, during sleep stage 2 or
slow-wave sleep (SWS).
However, all these studies are based on adult samples and no
spectral EEG data are available in children, although the specic
features of CAP in this age range have been visually characterized
and clearly shown to be somewhat different from adults (Bruni
et al., 2002, 2005; Parrino et al., 1998).
The aim of the present study was that of analyzing in detail the
frequency content of the different CAP EEG components, taking
into account the ongoing EEG background and the non-CAP (NCAP)
periods in polysomnographic (PSG) recordings of pre-school and
school age children also in comparison to adults. Finally, we ana-
lyzed the spectral differences between these components in order
to obtain information useful for their detection and the quantica-
tion of their impact on the sleep EEG neurophysiological dynamics.
2. Methods
2.1. Subjects
Forty-eight healthy subjects, 14 pre-school age children (seven
males, seven females, mean age 5.0, SD 1.19, range 2.96.5 years),
18 school age children (11 males, seven females, mean age 7.8, SD
1.20, range 6.610.3) and 16 adults (ve females and 11 males,
mean age 30.7 years, SD 3.57, range 24.539 years), were included
in this study. The group of adults was the same that was enrolled in
the study by Ferri et al. (2005a). Children were recruited in the
community to participate to the study. Parents signed an informed
consent. Inclusion criteria were: normal healthy pre-pubertal chil-
dren attending local school, with normal sleep habits. These crite-
ria were conrmed by parental and child interviews, sleep habits
and sleep disorders questionnaires, sleep diaries obtained for 15
days before the experiment and complete physical examination
that included determination of Tanner stage. None had any serious
physical or neurological or psychiatric disorder nor history of ma-
jor sleep problems and none was taking medication at the time of
the recording.
2.2. PSG recording
For this study each participant underwent a PSG overnight
recording, after an adaptation night, in a standard sleep laboratory
with controlled sound (noise level to a maximum of 30 dB SPL).
Subjects were not allowed to have drinks containing caffeine dur-
ing the afternoon preceding the recording and were allowed to
sleep until their spontaneous awakening in the morning.
The PSG montage included: EEG (at least three channels, F3 or
F4, C3 or C4, and O1 or O2, referenced to the contralateral mastoid;
electrodes placed following the 1020 International System); left
and right electrooculogram (electrodes placed 1 cm above the right
249
outer cantus and 1 cm below the left outer cantus and referred to
A1), electromyogram (submental EMG), EMG of the right and left
tibialis anterior muscle, and ECG (one derivation). Sleep signals
were sampled at 200 or 256 Hz and stored on hard disk in Euro-
pean data format (EDF, see Kemp et al., 1992 for details) for further
analysis. EEG signals, in particular, were rst acquired with a wide
band analog lter (0.00170 Hz) and then digitally band-pass l-
tered at 0.150 Hz.
2.3. Sleep stage scoring
Sleep stages were scored following standard criteria (Rechts-
chaffen and Kales, 1968) on 30-s epochs, and the following conven-
tional sleep parameters were evaluated:
Time in bed (TIB).
Sleep period time (SPT): time from sleep onset to sleep end.
Total sleep time (TST): the time from sleep onset to the end
of the nal sleep epoch minus time awake.
Sleep latency (SOL): time from lights out to sleep onset,
dened as the rst of two consecutive epochs of sleep stage
1 or one epoch of any other stage, in minutes.
REM latency (FRL): time from sleep onset to the rst REM
sleep epoch.
Number of stage shifts/hour (SS/h).
Number of awakenings/hour (AWN/h).
Sleep efciency (SE%): the percentage ratio between total
sleep time and time in bed (TST/TIB*100).
Percentage of SPT spent in wakefulness after sleep onset
(WASO%), i.e., the time spent awake between sleep onset
and end of sleep.
Percentage of SPT spent in sleep stages 1 (S1%), 2 (S2%),
slow wave sleep (SWS%), and REM sleep (REM%).
2.4. Cyclic alternating pattern (CAP) scoring
CAP A phases were detected in each recording (on the C3/A2 or
C4/A1 derivation), during NREM sleep, and classied into three
subtypes (A1, A2, and A3), according to the rules dened by Terz-
ano et al. (2001).
CAP A phases have been subdivided into a 3-stage hierarchy of
arousal strength:
A1: A phase with synchronized EEG patterns (intermittent
alpha rhythm in stage 1; sequences of K-complexes or delta
bursts in the other NREM stages), associated with mild or
trivial polygraphic variations.
A2: A phase with desynchronized EEG patterns preceded by
or mixed with slow high-voltage waves (K-complexes with
alpha and beta activities, k-alpha, arousals with slow wave
synchronization), linked with a moderate increase of mus-
cle tone.
and/or cardiorespiratory rate.
A3: A phase with desynchronized EEG patterns alone (tran-
sient activation phases or arousals) or exceeding 2/3 of the
phase A length, and coupled with a remarkable enhance-
ment of muscle tone and/or cardiorespiratory rate (Terzano
et al., 2001).
These criteria were applied also with the subsequent modica-
tions proposed by Bruni et al. (2005) for pre-school age children.
CAP sequences are dened as three or more A phases separated
from each other by no more than 60 s. The percentage of NREM
occupied by CAP sequences denes the CAP rate. All the remaining
NREM sleep, not occupied by CAP sequences is called NCAP.
250 O. Bruni et al. / Clinical Neurophysiology 120 (2009) 248256
The detection of CAP was carried out by means of the sleep
analysis software Hypnolab 1.2 (SWS Soft, Italy) which allows
computer-assisted detection of CAP A phase subtypes. With this
software, the automatic detection of CAP is performed by analyzing
the power course of two frequency bands, one for the detection of
candidates for A1 CAP subtypes and one for candidates of A3 CAP
subtypes. Periods in which both A1 and A3 candidates coexist are
assigned to the A2 CAP subtype. The automatic approach is super-
vised by the scorer who can change all the detection parameters.
The performances of this system have been evaluated and vali-
dated (Ferri et al., 2005c); additionally, for the present study, after
the supervision of the detection, the scorer visually edited carefully
the detections proposed by the automatic analysis, before the com-
putation of the various CAP parameters, which were automatically
generated by the same software and were used for statistical
analysis.
The following CAP parameters were measured:
CAP rate (percentage of total NREM sleep time occupied by
CAP sequences).
Percentage and duration of each A phase subtype.
A1 index (number of phases A1 per hour of NREM sleep,
and of S1, S2, and SWS sleep stage).
A2 index (number of phases A2 per hour of NREM sleep,
and of S1, S2, and SWS sleep stage).
A3 index (number of phases A3 per hour of NREM sleep,
and of S1, S2 and SWS sleep stage).
Duration of B phases.
Number and duration of CAP sequences.
2.5. Selection of EEG mini-epochs and power spectra computation
For this study, we utilized the same approach that was used in
the previous study on adults by Ferri et al. (2005a). The same chan-
nel used for the detection of CAP A phases (C3/A2 or C4/A1) was
subdivided into 2-s mini-epochs. Each mini-epoch was assigned
to a sleep stage, based on the sleep scoring previously performed
- only mini-epochs from sleep stage 2 and sleep stages 3 or 4
(SWS) were considered in this study - and to a CAP condition
(based on the CAP detection carried out previously); in particular,
the following CAP conditions were assigned: A1, A2, A3, and NCAP
(periods of absence of CAP). Additionally, also CAP B phases were
indicated as B1 (B phase following A1), B2 (B phase following
A2), and B3 (B phase following A3). In order to be assigned one
of the conditions listed above, mini-epochs needed to be occupied
for more than half of their length (>1 s) by the same condition. All
mini-epochs belonging to these different conditions were ana-
lyzed; thus practically all NREM was included in this study, in each
subject. Epochs with artifacts and with a signicant amount of con-
tamination by electromyographic activity were carefully detected
and excluded from the analysis.
Power spectra were calculated for each mini-epoch using the
sleep analysis software Hypnolab 1.2 (SWS Soft, Italy), after Welch
windowing, in order to minimize the truncation error and reduce
spectral leakage by suppressing sidelobes (Press et al., 1989), by
means of the fast Fourier transform (Cooley and Tukey, 1965); in
order to calculate the FFT on 2-s epochs of recordings sampled at
200 Hz, a zero-padding was performed in order to reach the re-
quired samples for this calculation (i.e. 512 samples) followed by
a linear interpolation of the spectral values in order to obtain
power spectrum data for frequencies between 0.5 and 25 Hz with
a frequency step of 0.5 Hz. The linear interpolation was not needed
for recordings sampled at 256 Hz. Average spectra were obtained
for each CAP condition, separately in sleep stage 2 and SWS, for
each subject.
2.6. Statistical analysis
For the statistical analysis, individual average spectra were
used, for each CAP and sleep stage condition, so that each subject
contributed equally to the group average.
The comparison between the different CAP subtypes and
between them and NCAP periods during sleep stage 2 or SWS
was carried out by means of the Students t-test for paired data
sets which was computed for each frequency bin in the spectra.
For these comparisons, the Bonferroni correction for multiple
statistical tests was applied and differences were considered as
statistically signicant at p < 0.025, because two simultaneous
comparisons were carried out. This analysis was repeated for each
of the three groups. Statistical estimators in the context of this
study were used essentially as descriptors. The comparisons be-
tween sleep or CAP parameters obtained in each group (pre-school
age children, school age children, adults) was conducted using the
one-way ANOVA followed by post-hoc comparisons, when
appropriate.
The commercially available Statistica software package (Stat-
Soft, Inc., 2001. STATISTICA data analysis software system, version
6. www.statsoft.com) was used for this statistical analysis.
3. Results
Table 1 shows the statistical comparison between the sleep
architecture parameters obtained in the three groups. As expected,
adults showed a reduction of SPT and TST and a signicantly higher
number of stage shifts per hour vs. school and pre-school age
children.
The evaluation of CAP parameters (Table 2) revealed several sig-
nicant differences with the pre-school age group showing the big-
gest differences vs. the other two groups: lower A1%, higher A2%
and A2 index and shorter A2 mean duration. School age children
showed only few differences vs. adults: shorter mean A1 duration
and a slight increase in number of CAP sequences.
Fig. 1 shows the comparison between the relative power den-
sity of the three groups for each CAP phase. Each frequency bin
of pre-school and school age children for each CAP subtype is ex-
pressed as a percentage of the corresponding bin of the adult
group. The dotted line indicates the baseline value of adults, while
black bars at the bottom indicate frequency bins in which the rel-
ative power density of pre-school and school age children groups
differ signicantly from adults.
In NREM S2 sleep stage, CAP A1, A2, A3 and NCAP showed sta-
tistically signicant differences in all frequency ranges between
pre-school children and adults while school children differed
mainly for low frequencies (<7 Hz).
In SWS, statistically signicant differences were found in all fre-
quency ranges for CAP A1, A2 and A3 between pre-school and
adults. School aged children showed differences vs. adults for
CAP A1, A2 and A3 mainly in the low-frequency range (<7 Hz)
and in the high frequency range (>20 Hz). For NCAP, pre-school
children differed from adults at all frequencies analyzed while
school children differed for frequency 0.519 Hz and 2125 Hz.
Fig. 2 shows the comparison between the different CAP sub-
types (A1, A2, A3) in NREM S2 (left column) and in SWS (right col-
umn), within the same group of subjects. The dotted line indicates
A1 as the baseline value. As expected adults showed signicant dif-
ferences in A2 and A3 for almost all the frequency bins except for
35 Hz for A2 and 58 Hz for A3, in NREM S2 while in SWS the dif-
ferences were found only above 5 Hz for both A2 and A3.
School age children showed a different trend; in NREM S2 the
A2 differed from A1 in frequency bins 411 Hz and 14.525 Hz,
while A3 phases in 0.54 Hz, 811 Hz and 1425 Hz; in SWS
Table 1
Comparison between the different sleep scoring parameters obtained in the three groups of subjects.

1. Adults 2. School-age children 3. Pre-school-age children ANOVA Post-hoc

1 vs. 2 1 vs. 3 2 vs. 3
Mean Standard 95% Condence Mean Standard 95% Condence Mean Standard 95% Condence p< p< p< p<
deviation interval deviation interval deviation interval
TIB, min 419.4 32.93 401.9437.0 547.1 37.43 528.5565.7 559.3 46.91 532.2 586.4 0.000001
SPT, min 398.1 37.47 378.1418.1 499.3 45.06 476.8521.7 533.3 60.51 498.3 568.2 0.000001 0.000001 0.000001 NS
TST, min 378.9 34.87 360.3397.5 484 46.05 461.1506.9 521.9 69.38 481.8 561.9 0.000001 0.000001 0.000001 0.05
SOL, min 18.9 15.6 10.627.2 36.3 24.97 23.948.8 22.4 20.88 10.334.4 0.05 0.02 NS NS
FRL, min 65.5 28.85 50.280.9 122.9 53.52 96.3149.5 96.1 35.52 75.6116. 6 0.001 0.00025 NS NS
SS/hour 8.9 1.79 8.09.9 7 2.57 5.78.2 6.7 2.42 5.38.1 0.02 0.02 0.015 NS
AWN/hour 1.7 1.26 1.12.4 1 1.19 0.41.6 1 1.21 0.31.7 NS
MT/hour 1 0.59 0.71.3 1.8 1.6 1.02.6 2.3 1.13 1.73.0 0.02 NS 0.01 NS
SE% 90.5 6.28 87.193.8 88.6 7.17 85.092.1 93 5.94 89.696.4 NS
WASO, % 4.7 4.35 2.47.0 3.7 3.25 1.95.5 2.7 2.72 1.04.5 NS

O. Bruni et al. / Clinical Neurophysiology 120 (2009) 248256

S1, % 0.9 0.67 0.51.2 2.9 2.2 1.84.0 3.1 2.07 1.94.3 0.002 0.02 0.015 NS
S2, % 46.8 7.02 43.150.6 44.6 7.2 41.048.1 44.8 7.07 40.748.9 NS
SWS, % 23.8 6.54 20.327.3 27.6 4.88 25.230.0 26.7 6.42 23.030.4 NS
REM, % 23.8 4.89 21.226.4 21.9 5.23 19.324.5 23.1 4.92 20.325.9 NS

TIB, time in bed; SPT, sleep period time; TST, total sleep time; SOL, sleep onset latency; FRL, 1st REM latency; SS, stage shifts; AWN, awakenings; MT, movement time; SE%, sleep efciency; WASO, wakefulness after sleep onset; S1,
NREM sleep stage 1; S2, NREM sleep stage 2; SWS, slow-wave sleep; REM, rapid-eye-movement sleep.

Table 2
Comparison between the different CAP parameters obtained in the three groups of subjects.

1. Adults 2. School-age children 3. Pre-school-age children ANOVA Post-hoc

1 vs. 2 1 vs. 3 2 vs. 3
Mean Standard 95% Condence Mean Standard 95% Condence Mean Standard 95% Condence p< p< p< p<
deviation interval deviation interval deviation interval
CAP rate, total 34.8 8.59 30.339.4 37.5 9.58 32.742.2 34 17.96 23.644.4 NS
CAP rate, S2 18.8 7.5 14.822.8 23.4 8.18 19.327.4 21.2 12.66 13.928.5 NS
CAP rate, SWS 66.1 17.92 56.575.7 60.4 19.78 50.670.3 55.3 25.95 40.370.3 NS
A1, % 79.2 8.1 74.983.5 81.4 5.18 78.884.0 65.9 10.74 59.772.1 0.000005 NS 0.00005 0.000002
A2, % 10 4.36 7.712.3 10.3 5.56 7.513.0 21 6.97 16.925.0 0.000002 NS 0.000003 0.000003
A3, % 10.8 5.33 8.013.6 8.4 3.38 6.710.0 13.2 6.38 9.516.8 0.04 NS NS 0.011
A1 duration, min 9.2 1.65 8.310.1 7.3 2.76 6.08.7 7.5 2.27 6.28.8 0.05 0.03 0.05 NS
A2 duration, min 11.5 1.78 10.612.5 10.5 1.87 9.611.4 8.9 1.87 7.810.0 0.0014 NS 0.00032 0.02
A3 duration, min 15.3 2.93 13.716.8 18.5 4.37 16.320.6 16.6 3.61 14.518.7 NS
A1 index 36.7 9.48 31.741.8 42.2 9.22 37.646.8 34 21.19 21.846.3 NS
A2 index 3.9 1.97 2.84.9 5.2 4.31 3.17.4 8.5 4.93 5.611.3 0.008 NS 0.0026 0.026
A3 index 3.8 2.54 2.45.1 3 1.47 2.33.7 4.4 1.99 3.25.5 NS
B duration, min 22.4 2.74 20.923.8 22 3.49 20.323.8 22.3 3.94 20.024.5 NS
Sequence 243.1 83.08 198.8287.3 276 114 219.3332.7 231.2 110.11 167.6294.8 NS
duration, min
Sequence number 25.8 7.2 21.929.6 32.7 7.01 29.236.2 34.6 8.18 29.839.3 0.005 0.01 0.0025 NS

251
252 O. Bruni et al. / Clinical Neurophysiology 120 (2009) 248256

Fig. 1. Comparison between the relative power density of the three groups for each CAP phase. Each frequency bin of pre-school and school age children for each CAP subtype
is expressed as a percentage of the corresponding bin of the adult group. The dotted line indicates the baseline value of adults, while black bars at the bottom indicate
frequency bins in which the relative power density of pre-school and school age children groups differ signicantly from adults.

differences were found in the frequency ranges 36 Hz and 15
24 Hz for A2 and 46 Hz, 8.511.5 Hz and 1225 Hz for A3.
Pre-school age children showed more similarities between A1
and A2 with scattered differences in the whole frequency range
in sleep stage 2 while A3 showed differences mainly in the higher
frequency range. In SWS the differences were more evident and A3
differed in all frequency bins analyzed while A2 differed for fre-
quencies above 4 Hz.
Fig. 3 shows the different CAP subtypes and NCAP periods dur-
ing NREM S2. Each frequency bin of CAP A subtypes and of their
following B phases is expressed as a percentage of the correspond-
ing bin of NCAP. The dotted line indicates the baseline value of
NCAP, while black bars at the bottom of each panel indicate fre-
quency bins in which the relative power density of the CAP sub-
type or the B phase differs signicantly from NCAP.
A1 showed a clear difference from NCAP in all frequency ranges
and with a similar trend in all age groups; also the phase B after A1
was clearly different from NCAP in the frequency range below
15 Hz.
Similarly to A1, A2 showed differences in all frequency bands in
all the three age groups considered vs. NCAP. The phase B following
A2 showed differences in frequencies below 10 Hz in pre-school
and school age children, while it could be considered similar to
NCAP in adults.
Again, A3 showed a difference from NCAP in all frequency
ranges and in the three age groups while the phase B after A3
was virtually equivalent to NCAP.
Fig. 4 shows the different CAP subtypes and NCAP periods dur-
ing NREM SWS. In SWS, CAP A1 showed very few differences from
NCAP (in frequencies below 7 Hz) and the phase B following A1 is
practically no different from NCAP in all the three groups.
Phases A2 and A3, on the other hand, showed a clearly different
spectrum from NCAP while the respective phases B were very sim-
ilar to NCAP.
 O. Bruni et al. / Clinical Neurophysiology 120 (2009) 248256
Fig. 2. Comparison between the different CAP subtypes (A1, A2, A3) in NREM S2 (left column) and in SWS (right column), within the same group of subjects. The dotted line
indicates A1 as the baseline value. Black bars indicate frequency bins in which absolute power density of CAP subtypes A2 or A3 differed signicantly from A1.
4. Discussion
Relatively recent previous studies on CAP in pre-school and
school age children have shown clear developmental changes
characterized mainly by the distribution and percentage of A
phases (Bruni et al., 2002, 2005; Lopes et al., 2005). These changes
seem to reect the maturational course of the sleep EEG in the
rst years of life and are also strictly related to the modication
of the sleep structure and to the redistribution of NREM sleep.
EEG frequencies are overall somewhat slower in children and
gradually get faster until they reach adult values, while the
amplitude of waveforms at all frequencies increases gradually
reaching a peak in the school age period and then declines. These
developmental changes occur also in several PSG measures, par-
ticularly during transition from pre-school to early school age
(Montgomery-Downs et al., 2006); in fact the process of sleep
maturation determines a change in proportion of NREM stages
during pre-school years, related to napping (Kahn et al., 1973;
Smith et al., 1977; Sheldon, 1996; Curzi Dascalova and Challamel,
2000). Also during the transition from preadolescence to adoles-
cence, this process determines a steep decline in NREM sleep
and in the slow-wave (delta) activity (Coble et al., 1987; Feinberg
et al., 1990a) related to the waning synaptic connectivity that de-
creases the intensity of waking brain activity (Feinberg et al.,
1990a,b). Moreover, a recent longitudinal study reported a dra-
matic change in power density from preadolescence to adoles-
cence beginning at age 915 year (Campbell et al., 2007) and
showed that the power density decline is not limited to the delta
253
frequencies but involves all frequency bands between 0.3 and
30 Hz, although it is most evident in the theta and delta bands.
Recently, signicant differences in the EEG spectra have been
reported in adults related to the presence or absence of CAP, even
in the same sleep stage (Ferri et al., 2005a). Starting from this pio-
neering study we extended the same analysis to other two younger
age groups. Therefore, the present study compared three different
age groups and is the rst which included in its analysis the com-
plete NREM sleep EEG and computed separate power spectra for
the different CAP phases and NCAP. As expected, the evaluation
of CAP parameters revealed several signicant differences with
the pre-school age group showing the biggest differences vs. the
other two groups, while school age children showed only few dif-
ferences vs. adults:
Our analysis demonstrated that pre-school children have a dif-
ferent structure of sleep, especially from the microstructural (CAP)
point of view. The differences are evident for all the CAP compo-
nents and for NCAP in almost all the frequency bands. These data
are in agreement with a recent report that showed that develop-
mental changes occur in several PSG measures particularly during
transition from pre-school to early school age (Montgomery-
Downs et al., 2006).
CAP A1, A2 and A3 phases in pre-school and school aged chil-
dren showed differences vs. adults in all frequencies with the big-
gest divergence in the delta range, in both stage 2 and SWS. This
nding might be associated to the age-related delta decline in
the 03 Hz frequency reported by Feinberg et al. (1990) and Coble
et al. (1987) in children of the same ages. Also NCAP was
254 O. Bruni et al. / Clinical Neurophysiology 120 (2009) 248256
Fig. 3. Different CAP subtypes and NCAP periods during NREM S2. Each frequency bin of CAP A subtypes and of their following B phases is expressed as a percentage of the
corresponding bin of NCAP. The dotted line indicates the baseline value of NCAP, while black bars at the bottom of each panel indicate frequency bins in which the relative
power density of the CAP subtype or the B phase differs signicantly from NCAP.
completely different in the three age groups showing that NREM
sleep undergoes profound modication in its structure during
development.
Analyzing the different power spectrum of the CAP A phases, we
found a similar trend in the three age groups: CAP A2 and A3
phases are clearly different from A1 mostly with respect to the
higher frequencies. This is clearly evident in adults and the differ-
ence tends to be smaller in school and pre-school children. Fur-
thermore, CAP A phases are characterized by clearly different
spectra also during sleep stage 2 or SWS, in all groups and this nd-
ing underlines a probable different functional meaning of the same
CAP subtype during different sleep stages, independently from age.
Compared to NCAP, CAP A phases are characterized by clearly
different spectral patterns: the A1 subtype is characterized by a
prominent peak in the delta range evident in stage 2 but not so
clear in SWS; the power spectrum of A1 is signicantly higher than
that of NCAP over the whole frequency range studied during sleep
stage 2, but only up to 7.5 Hz during SWS, in all age groups.
Also A2 and A3 subtypes show a signicant increase in power,
with respect to NCAP, over the whole frequency range studied;
they show an prominent increase in the high-frequency part of
the spectrum, during both sleep stage 2 and SWS. This increase
in the high frequencies has been already reported (De Carli et al.,
2004; Ferri et al., 2005a) and was more evident in adults than in
the younger groups.
Analyzing Figs. 3 and 4, we can observe that the trend of the
spectral differences between CAP phases and NCAP in stage 2
was similar in the three age groups indicating that the structural
component of CAP is very stable over time and is not greatly inu-
enced by age. In SWS, although the dissimilarities are less marked,
we could observe again a similar trend of the slope of power den-
sity in the pre-school, school and adult subjects. This results is in
part conrmed by the stability of the time structure of CAP at dif-
ferent ages as we demonstrated in previous studies (Bruni et al.,
2005).
Regarding the B phase, we can state that this component of
CAP seems not to be a simple return to the ongoing baseline
activity but shows a small and signicant increase in power in
the delta-theta frequencies and a decrease in power in the sigma
frequency range. The difference is more marked in stage 2 than
in SWS, reecting the difculty of scoring CAP visually in some
epochs of SWS.
As stated in the previous paper (Ferri et al., 2005b), the A
components of CAP might correspond to periods in which the
very-slow delta activity of sleep groups a range of different EEG
activities, including slow waves, and the sigma and beta bands,
while the B phase of CAP might correspond to a period in which
this activity is quiescent or inhibited.
While some results were expected because the denition of
CAP, even if developed for visual detection, is also based on the
EEG spectral content, our analysis was able to disclose additional
ne aspects which were not only the effect of the initial denition
of the CAP components and were only detected because of the de-
tailed quantitative analysis performed.
Our analysis clearly conrmed that age is a potent modulator of
the power spectra of sleep EEG. There are clear differences be-
tween adults and children in the EEG spectra that are strictly re-
lated to age. This nding has been conrmed by previous studies
on the spectral components of sleep in children. Feinberg et al.
(1990) in a group of subjects aged 4.223.7 years showed that
 O. Bruni et al. / Clinical Neurophysiology 120 (2009) 248256
Fig. 4. Different CAP subtypes and NCAP periods during NREM SWS. Each frequency bin of CAP A subtypes and of their following B phases is expressed as a percentage of the
corresponding bin of NCAP. The dotted line indicates the baseline value of NCAP, while black bars at the bottom of each panel indicate frequency bins in which the relative
power density of the CAP subtype or the B phase differs signicantly from NCAP.
amplitude and incidence of delta waves decline at roughly the
same rate from the middle of the rst decade through the second
decade of life, in agreement with the previous report by Coble
et al. (1987).
However, we should also consider that the homeostatic mecha-
nism of sleep is active since the early period of life and remains sta-
ble until elderly, considering NREM sleep and mainly SWS. In the
same fashion, the intrinsic oscillatory mechanism of slow EEG
activity (Achermann and Borbly, 1997; Amzica and Steriade,
1997) remains constant. These ndings are supported by our pre-
vious studies on CAP in children (Bruni et al., 2002, 2005) showing
that the periodicity of the very low-frequency periodic brain activ-
ities (CAP A1 component) is very stable during development. Also
Ferri et al. (2005a) demonstrated that a signicant inuence of
sleep homeostatic mechanisms is still evident mostly for the delta
band of CAP A phases during sleep stage 2. Previous studies in in-
fants and children supported these results: SWS recurrence time
remains of similar length in infants, children, and adults (Bes
et al., 1991) and the typical declining trend of the delta activity
in the course of the night was reported for infants aged 26
months, indicating that the mechanisms underlying sleep homeo-
stasis emerge early and are constant for all life through (Jenni et al.,
2004).
In conclusion, the present study is the rst which has included
in its analysis the complete NREM sleep EEG and has computed
separate power spectra for the different CAP subtypes, the inter-
vening B phases, and the remaining NCAP periods in different
age groups. Our data seem to provide information not available be-
fore and useful for the understanding of the impact of CAP on the
sleep EEG neurophysiological dynamics at different ages. This type
255
of information is crucial for a more adequate interpretation of data
provided by a growing number of studies analyzing CAP in groups
of pediatric patients (Bruni et al., 2002, 2005, 2007, 2008; Lopes
et al., 2005; Guilleminault et al., 2005; Kheirandish-Gozal et al.,
2007).
References
Achermann P, Borbly AA. Low frequency (<1 Hz) oscillations in the human sleep
electroencephalogram. Neuroscience 1997;81:21322.
Amzica F, Steriade M. The K-complex: its slow (<1-Hz) rhythmicity and relation to
delta waves. Neurology 1997;49:9529.
Bes F, Schulz H, Navelet Y, Salzarulo P. The distribution of slow-wave sleep across
the night: a comparison for infants, children, and adults. Sleep 1991;14:512.
Bruni O, Ferri R, Miano S, Verrillo E, Vittori E, Della Marca G, et al. Sleep cyclic
alternating pattern in normal school-age children. Clin Neurophysiol
2002;113:180614.
Bruni O, Ferri R, Miano S, Verrillo E, Vittori E, Farina B, et al. Sleep cyclic alternating
pattern in normal pre-school-age children. Sleep 2005;28:22030.
Bruni O, Ferri R, Vittori E, Novelli L, Vignati M, Porrio MC, et al. Sleep architecture
and NREM alterations in children and adolescents with Asperger syndrome.
Sleep 2007;30:157785.
Bruni O, Ferri R, Novelli L, Finotti E, Miano S, Guilleminault C. NREM sleep instability
in children with sleep terrors: the role of slow wave activity interruptions. Clin
Neurophysiol 2008;119:98592.
Campbell IG, Higgins LM, Trinidad JM, Richardson P, Feinberg I. The increase in
longitudinally measured sleepiness across adolescence is related to the
maturational decline in low-frequency EEG power. Sleep 2007;30:167787.
Coble PA, Reynolds CF, III, Kupfer DJ, Houck P. Electroencephalographic sleep of
healthy children. Part II. Findings using automated delta and REM sleep
measurement methods. Sleep 1987;10:55162.
Cooley JW, Tukey OW. An algorithm for the machine calculation of complex Fourier
series. Math Comput 1965;19:297301.
Curzi Dascalova L, Challamel MJ. Neurophysiological basis of sleep development. In:
Loughlin GM, Carroll JL, Marcus CL, editors. Sleep and breathing in children. A
developmental approach. New York: Marcel Dekker; 2000.
256 O. Bruni et al. / Clinical Neurophysiology 120 (2009) 248256
De Carli F, Nobili L, Beelke M, Watanabe T, Smerieri A, Parrino L, et al. Quantitative
analysis of sleep EEG microstructure in the time-frequency domain. Brain Res
Bull 2004;63:399405.
Feinberg I, March JD, Flach K, Maloney T, Chern WJ, Travis F. Maturational changes
in amplitude, incidence and cyclic pattern of the 03 Hz (Delta)
electroencephalogram of human sleep. Brain Dysfunction 1990a;3:18392.
Feinberg I, Thode HC, Chugani HT, March JD. Gamma distribution model describes
maturational curves for delta wave amplitude, cortical metabolic rate and
synaptic density. J Theoret Biol 1990b;142:14961.
Ferri R, Parrino L, Smerieri A, Terzano MG, Elia M, Musumeci SA, et al. Cyclic
alternating pattern and spectral analysis of heart rate variability during normal
sleep. J Sleep Res 2000;9:138.
Ferri R, Bruni O, Miano S, Plazzi G, Terzano MG. All-night EEG power spectral
analysis of the cyclic alternating pattern components in young adult subjects.
Clin Neurophysiol 2005a;116:242940.
Ferri R, Bruni O, Miano S, Terzano MG. Topographic mapping of the spectral
components of the cyclic alternating pattern (CAP). Sleep Med 2005b;6:2936.
Ferri R, Bruni O, Miano S, Smerieri A, Spruyt K, Terzano MG Inter-rater reliability of
sleep Cyclic Alternating Pattern (CAP) scoring, et al. Clin Neurophysiol
2005c;116:696707.
Guilleminault C, Lee JH, Chan A, Lopes MC, Huang YS, da Rosa A. Non-REM-sleep
instability in recurrent sleepwalking in pre-pubertal children. Sleep Med
2005;6:51521.
Halsz P. Hierarchy of micro-arousals and the microstructure of sleep. Neurophysiol
Clin 1998;28:46175.
Jenni OG, Borbely AA, Achermann P. Development of the nocturnal sleep
electroencephalogram in human infants. Am J Physiol Regulatory Integrative
Comp Physiol 2004;286:52838.
Kahn E, Fischer C, Edwards A, Davis DM. 24-hour sleep patterns: a comparison
between 23 years old and 46 years old children. Arch Gen Psychiatry
1973;29:3805.
Kemp B, Varri A, Rosa AC, Nielsen KD, Gade J. A simple format for exchange of
digitized polygraphic recordings. Electroencephalogr Clin Neurophysiol
1992;82:3913.
Kheirandish-Gozal L, Miano S, Bruni O, Ferri R, Pagani J, Villa MP, et al. Reduced
NREM sleep instability in children with sleep disordered breathing. Sleep
2007;30:4507.
Lopes MC, Rosa A, Roizenblatt S, Guilleminault C, Passarelli C, Tuk S, et al. Cyclic
alternating pattern in peripubertal children. Sleep 2005;28:2159.
Montgomery-Downs HE, OBrien LM, Gulliver TE, Gozal D. Polysomnographic
characteristics in normal preschool and early school-aged children. Pediatrics
2006;117:74153.
Parrino L, Boselli M, Spaggiari MC, Smerieri A, Terzano MG. Cyclic alternating
pattern (CAP) in normal sleep: polysomnographic parameters in different age
groups. Electroencephalogr Clin Neurophysiol 1998;107:43950.
Parrino L, Smerieri A, Rossi M, Terzano MG. Relationship of slow and rapid EEG
components of CAP to ASDA arousals in normal sleep. Sleep 2001;24:
8815.
Pascual-Marqui RD, Michel CM, Lehmann D. Low resolution electromagnetic
tomography: a new method for localizing electrical activity in the brain. Int J
Psychophysiol 1994;18:4965.
Pascual-Marqui RD, Lehmann D, Koenig T, Kochi K, Merlo MCG, Hell D, et al. Low
resolution brain electromagnetic tomography (LORETA) functional imaging in
acute, neuroleptic-naive, rstepisode, productive schizophrenia. Psychiatry Res
1999;90:16979.
Press WH, Flannery BP, Teukolsky SA, Vetterling WT. Numerical recipes. The art of
scientic computing. Cambridge: Press Syndicate of the University of
Cambridge; 1989.
Rechtschaffen A, Kales A. A manual of standardized terminology, techniques and
scoring system of sleep stages of human subjects. USA: Washington Public
Health Service, US Government Printing Ofce; 1968.
Sheldon SH. Development of sleep structure. In: Sheldon SH, editor. Evaluating sleep
in infants and children. Philadelphia (PA): LippincottRaven; 1996. p. 1120.
Smith JR, Karacan I, Yang M. Ontogeny of delta activity during human sleep.
Electroencephalogr Clin Neurophysiol 1977;43:22937.
Terzano MG, Mancia D, Salati MR, Costani G, Decembrino A, Parrino L. The cyclic
alternating pattern as a physiologic component of normal NREM sleep. Sleep
1985;8:13745.
Terzano MG, Parrino L, Spaggiari MC. The cyclic alternating pattern sequences in the
dynamic organization of sleep. Electroencephalogr Clin Neurophysiol
1988;69:43747.
Terzano MG, Parrino L, Smerieri A, Chervin R, Chokroverty S, Guilleminault C, et al.
Consensus report. Atlas, rules, and recording techniques for the scoring of cyclic
alternating pattern (CAP) in human sleep. Sleep Med 2001;2:53753.