- Email: [email protected]
Alpha activity and cardiac correlates: three types of relationships during nocturnal sleep
Clinical Neurophysiology 111 (2000) 940±946 www.elsevier.com/locate/clinph
Alpha activity and cardiac correlates: three types of relationships during nocturnal sleep J. Ehrhart a,*, M. Toussaint b, C. Simon a, C. Gron®er a, R. Luthringer b, G. Brandenberger a a
Laboratoire des Regulations Physiologiques et des Rythmes Biologiques chez l'Homme, 4, rue Kirschleger, 67085 Strasbourg Cedex, France b Forenap, Centre Hospitalier de Rouffach 68250 Rouffach, France Accepted 29 December 1999
Abstract Objective: We examined simultaneously alpha activity and cardiac changes during nocturnal sleep, in order to differentiate non-rapid eye movement (NREM) sleep, REM sleep, and intra-sleep awakening. Methods: Ten male subjects displaying occasionally spontaneous intra-sleep awakenings underwent EEG and cardiac recordings during one experimental night. The heart rate and heart rate variability were calculated over 5 min periods. Heart rate variability was estimated: (1) by the ratio of low frequency (LF) to high frequency (HF) power calculated from spectral analysis of R-R intervals; and (2) by the interbeat autocorrelation coef®cient of R-R intervals (rRR). EEG spectral analysis was performed using a fast Fourier transform algorithm. Results: Three types of relationships between alpha waves (8±13 Hz) and cardiac correlates could be distinguished. During NREM sleep, alpha activity and cardiac correlates showed opposite variations, with high levels of alpha power associated with decreased heart rate, rRR and LF/HF ratio, indicating low sympathetic activity. Conversely, during REM sleep, alpha activity was low whereas heart rate, rRR, and the LF/HF ratio peaked, indicating high sympathetic activity. During intra-sleep awakenings, alpha activity and cardiac correlates both increased. No difference in time-course between alpha 1 (8±10 Hz) and alpha 2 (10±13 Hz) activity could be shown. Alpha waves occurred in fronto-central areas during slow wave sleep (SWS), migrated to posterior areas during REM sleep, and were localized in occipital areas during intra-sleep awakenings. Conclusions: These results suggest that alpha waves are not simply a sign of arousal, as is commonly thought. Fronto-central alpha waves, associated with decreased heart rate, possibly re¯ect sleep-maintaining processes. q 2000 Elsevier Science Ireland Ltd. All rights reserved. Keywords: EEG activity; Alpha power; Heart rate; Heart rate variability; Sleep; Human
1. Introduction Studies on cardiac activity have described systematic variations of heart rate during sleep, with an overall slowing during the night and ultradian variations associated with rapid eye movement (REM)-non(N)REM sleep cycles. Heart rate is lowest during slow wave sleep (SWS) and higher and more labile, with abrupt ¯uctuations, during REM sleep (Aserinsky and Kleitman, 1953; Snyder et al., 1964; Khatri and Freis, 1967; Welch and Richardson, 1972; Aldredge and Welch, 1973). Not only heart rate but also heart rate variability differs according to sleep stages. Studies using Poincare plots have demonstrated distinctive and characteristic patterns depending on sleep stages, REM sleep being characterized by wider overall variations than those observed during NREM sleep (Zemaityte et al., 1984; * Corresponding author. Tel.: 133-3-8824-3558; fax: 133-3-88243334. E-mail address: [email protected] (J. Ehrhart)
Kamen and Tonkin, 1995; Vaughn et al., 1995; Kamen et al., 1996). Studies using spectral analysis of R-R intervals to assess heart rate variability reported a lower ratio of low frequency (LF) to high frequency (HF) components during NREM sleep than during REM sleep, which suggests a sympathetic dominance in the autonomic system during REM sleep and a predominant vagal in¯uence during NREM sleep (Berlad et al., 1993; Bootsma et al., 1994; Baharav et al., 1995; Vanoli et al., 1995; Bonnet et al., 1997; Scholz et al., 1997), although this interpretation has been questioned (Eckberg, 1997; Houle and Billman, 1999). All these results were based on conventional scoring of sleep stages. Electroencephalographic (EEG) spectral analysis makes possible a more detailed and dynamic description of sleep processes and, in recent years, time course of power in the different frequency bands has been precisely de®ned (Aeschbach and Borbely, 1993; Merica and Blois, 1997; Achermann and Borbely, 1998). However, only a few studies have examined overnight cardiac changes
1388-2457/00/$ - see front matter q 2000 Elsevier Science Ireland Ltd. All rights reserved. PII: S13 88-2457(00)0024 7-9
CLINPH 99694
J. Ehrhart et al. / Clinical Neurophysiology 111 (2000) 940±946
together with sleep EEG activity. It has been reported that heart rate displays ultradian variations which are not closely correlated with changes in slow wave activity (Cajochen et al., 1994). In a recent study, we found a relationship between overnight pro®les of EEG mean frequency and the interbeat autocorrelation coef®cient of R-R intervals (rRR) (Otzenberger et al., 1997). We also reported that rRR is related to overnight variations in the LF/HF ratio (Otzenberger et al., 1998). Very few studies have examined EEG activity in the alpha band in human sleep. Alpha activity has a frequency of 8±13 Hz and is generally accepted as an indicator of relaxed wakefulness (IFSECN, 1974). The desynchronization of alpha activity is employed in visual sleep stage scoring systems as a sign of sleep onset. During sleep, visually detected alpha waves have been thought to re¯ect arousal processes and a shift toward wakefulness (ASDA, 1992). However, more recent investigations, based on spectral analysis of sleep EEG, reported that alpha waves are also present during NREM and REM sleep, which suggests possible functional differences for alpha waves during sleep (Cooley and Tukey, 1965; Hauri and Hawkins, 1973; Pivik and Harman, 1995). This prompted us to examine the time course of alpha waves during NREM sleep, REM sleep, and intra- sleep awakening in relation to other signs of physiological arousal, such as heart rate and heart rate variability, which may differentiate sleep-maintaining from sleep-disturbing processes. Further analyses were conducted in order to test whether the differences observed in these relationships could be attributed to different spectral contributions in the alpha band, or whether they re¯ect topographical differences of alpha generators. 2. Methods and materials Ten male subjects (21±28 years old) displaying occasionally spontaneous awakenings during nocturnal sleep were included in the study. They gave their written consent to participate. They had regular sleep-wake habits and did not take any medication. The study was approved by the local Ethics Committee. The experiments were carried out in a soundproof, air-conditioned sleep room. After an habituation night, the subjects underwent one experimental session during which EEG and cardiac recordings were carried out. Electrodes for polysomnographic recordings were applied 2 h before the beginning of the experiment. Lights were switched off at 23:00 h and the subjects were awakened at 07:00 h. 2.1. Sleep analysis Sleep recordings were performed using 4 EEG leads (F3, C3, P3 vs. A2 and C4 vs. A1), one chin electromyographic (EMG) and one diagonal electrooculographic (EOG) lead (upper canthus of one eye vs. lower canthus of the other
941
eye). The recordings were visually scored at 30 s intervals using standardized criteria (Rechtschaffen and Kales, 1968). For all-night spectral analysis, the EEG signal (C3-A2 or C4-A1) was high-pass (0.3 Hz) and low-pass (35 Hz) ®ltered before conversion from analog to digital with a sampling frequency of 128 Hz. Subsequently, spectra were computed for consecutive 2 s periods with a fast Fourier transform algorithm (Bootsma et al., 1994). In order to yield one spectrum every 5 min, a median ®lter was applied for 150 consecutive 2 s periods. The spectral parameter studied was the absolute power in the alpha frequency band (8±13 Hz); for further analysis, the total alpha frequency band was subdivided into alpha 1 (8±10 Hz) and alpha 2 (10±13 Hz) frequency bands, which have been shown to differ according to the physiological state of the subject (Cantero et al., 1999). 2.2. Heart rate analysis The electrocardiogram signal was fed into a generator that produced a pulse at the rising phase of each R wave. The trigger event times were recorded with a precision of ^1 ms, and the R-R intervals were calculated on a computer equipped with a data acquisition control board including a timer. Computers and polygraphs were synchronized. Ectopic beats were identi®ed and replaced with interpolative RR interval data. Each R-R interval was plotted against the previous R-R interval to produce a 5 min Poincare plot (RRn 1 l vs. R-Rn). The interbeat autocorrelation coef®cient of R-R intervals (rRR; i.e. Pearson's correlation coef®cient between R-Rn and R-Rn 1 l) was calculated over 5 min periods. Power spectral analysis of each consecutive 5 min recording was performed in a sequential fashion with the use of a fast Fourier transform (based on a non-parametric algorithm using a Welsh window) after the ectopic-free data were detrended and resampled. A ®xed resampling frequency of 1024 equally spaced points per 5 min period was used. The power in the LF band (0.04±0.15 Hz) and in the HF band (0.15±0.50 Hz) was calculated for each 5 min density spectrum by integrating the spectral density in the respective frequency bands. The LF/HF ratio was calculated using the power in each band. 2.3. Statistical analysis To assess the relationship between the time-courses of alpha activity, heart rate and the measures of heart rate variability, we selected for each subject one uninterrupted REM-NREM-REM sleep episode containing SWS, one NREM-REM-NREM sleep episode, and one NREM-intrasleep awakening-NREM episode. In each case, NREM, REM and intra-sleep awakening lasted at least 10 min, and the immediately preceding and immediately following sleep episode lasted at least 10 min. In order to compensate for the individual differences in the duration of the NREM sleep episodes, REM sleep
942
J. Ehrhart et al. / Clinical Neurophysiology 111 (2000) 940±946
episodes and intra-sleep awakenings, the individual pro®les were normalized according to a method derived from Achermann et al. (1993). Individual data were subdivided into equal parts within each period: (1) into 16 parts from the beginning to the end of NREM sleep; (2) into 8 parts from the beginning to the end of REM sleep; (3) into 16 parts from the beginning to the end of intra-sleep awakening; (4) into 4 parts for the sleep episodes surrounding the three states of vigilance. Data were averaged over subjects and the standard error was calculated for each interval. Mean changes in alpha activity, in heart rate and in the measures of heart rate variability were analyzed by one-way ANOVA for repeated measures with time as a factor. Results are expressed in percentages of the mean overnight values, which normalizes the data across subjects. The difference was considered to be signi®cant if P , 0:05. 2.4. Topographical analysis An additional group of 8 subjects (4 males, 4 females; 20±31 years old) participated in a second study in which topographical distribution of alpha activity was investigated. Twenty-eight EEG channels were recorded from all subjects, each channel referred to a common linked-ear reference with an impedance lower than 5 KOhm. Electrode placement followed the 10±20 system (Jasper, 1958) and the additional electrodes were each placed at the center of a space formed by 4 standard electrodes. Two EOG leads and one chin-EMG were added to allow visual sleep stage scoring (30 s epochs) according to the criteria of Rechtschaffen and Kales (1968). The ®lter settings were, respectively, 0.5±30 Hz for EEGs, 0.5±20 Hz for EOGs,
and 5±70 Hz for EMG. All recordings were sampled at 128 Hz. The spectral analysis was performed by the fast Fourier transform algorithm on consecutive 2 s epochs (256 points) with a 0.5 Hz frequency resolution, and truncating error was reduced by applying a Hamming window. In addition, the values for 15 adjacent 2 s epochs were averaged to yield spectral density values for 30 s periods computed for each of the 28 EEG channels. Thus, sleep visual scores of each 30 s period were synchronized with spectral density values. To reduce the noise generated by the relatively short epoch length, a median ®lter was applied with a window of 15 points (i.e. 30 s). Channels along the antero-posterior axis (Fz, Cz, Pz, Oz) were subjected to a one-way ANOVA for repeated measures with electrode as factor. Contrasts were assessed by two tailed paired Student t tests. To reduce interindividual variability, data were expressed as percentages of the mean overnight values.
3. Results Table 1 gives the mean (^SE) changes in alpha activity and in cardiac correlates during NREM sleep, REM sleep, and during intra- sleep awakenings, in 10 subjects. As expected, alpha activity showed large interindividual differences. During NREM sleep, individual basal levels ranged from 1.57 to 6.05 mV 2 and individual peak values ranged from 3.08 to 22.16 mV 2. Therefore, levels are given in percentages of the mean overnight value, which normalizes the interindividual differences in the raw data. Fig. 1 illustrates the mean (^SE) time-courses of alpha
Table 1 Mean (^SE) levels of alpha activity (8±13 Hz), alpha 1 activity (8±10 Hz), alpha 2 activity (10±13 Hz), of heart rate, rRR, and of the LF/HF ratio a NREM sleep Levels Alpha activity (8±13 Hz) mV 2 % Alpha 1 activity (8±10 Hz) mV 2 % Alpha 2 activity (10±13 Hz) mV 2 % Heart rate Beats/min % rRR % LF/HF %
Basal
REM sleep Peak/trough
Basal
4.37 ^ 0.49 54.0 ^ 4.6
9.80 ^ 1.71 115.2 ^ 12.7
8.37 ^ 1.3 103.0 ^ 8.1
2.25 ^ 0.29 55.9 ^ 5.9
4.57 ^ 0.99 105.3 ^ 4.3
1.78 ^ 0.23 50.6 ^ 14.6 55.1 ^ 1.4 101.2 ^ 1.7 0.54 ^ 0.06 115.5 ^ 14.7 1.64 ^ 0.25 121.0 ^ 9.1
Intra-sleep awakenings Peak/trough
Basal
Peak/trough
4.35 ^ 0.54 56.9 ^ 5.5
8.10 ^ 2.43 90.5 ^ 9.4
14.53 ^ 4.06 151.0 ^ 20.6
3.91 ^ 0.67 100.3 ^ 9.5
2.24 ^ 0.58 60.4 ^ 8.7
3.40 ^ 0.97 83.8 ^ 13.6
7.45 ^ 1.36 163.7 ^ 7.7
4.37 ^ 0.66 121.4 ^ 10.2
3.80 ^ 0.31 105.6 ^ 5.8
1.84 ^ 0.23 54.0 ^ 5.2
4.17 ^ 2.32 98.4 ^ 28.5
6.25 ^ 2.38 132.0 ^ 17.8
51.1 ^ 0.9 94.0 ^ 1.0 0.16 ^ 0.09 26.2 ^ 20.4 0.55 ^ 0.07 43.9 ^ 6.5
54.2 ^ 1.8 98.5 ^ 2.0 0.35 ^ 0.09 64.9 ^ 13.9 0.86 ^ 0.15 62.4 ^ 8.3
57.9 ^ 1.3 105.3 ^ 1.2 0.64 ^ 0.03 136.8 ^ 13.0 2.16 ^ 0.34 155.7 ^ 18.2
54.4 ^ 2.5 93.6 ^ 1.1 0.52 ^ 0.09 82.6 ^ 11.8 2.84 ^ 1.34 86.9 ^ 11.8
61.8 ^ 2.6 106.6 ^ 2.6 0.82 ^ 0.04 135.3 ^ 8.7 4.66 ^ 2.12 NS 139.0 ^ 14.7
a rRR, interbeat autocorrelation coef®cient of R-R intervals; LF/HF, ratio of low frequency to high frequency power calculated from spectral analysis of R-R intervals. Values are expressed in raw data, and in percentages of the mean overnight values for each individual. Basal levels: highest (or lowest) values before the onset of the two states of sleep and of the intra- sleep awakenings. Peak (or trough) levels: highest (or lowest) values during these three vigilance states. P , 0:01, unless otherwise indicated.
J. Ehrhart et al. / Clinical Neurophysiology 111 (2000) 940±946
activity and of heart rate during NREM sleep, REM sleep, and during intra-sleep awakenings. Three types of relationships could be distinguished. During NREM sleep, alpha activity and cardiac correlates showed opposite variations with signi®cantly (P , 0:0001) increased alpha activity associated with signi®cantly (P , 0:0001) decreased heart rate. Conversely, during REM sleep, alpha activity decreased (P , 0.0001), whereas heart rate increased (P , 0:0001). During intra-sleep awakenings, both alpha activity and heart rate signi®cantly increased (P , 0:0001). Fig. 2 represents the mean (^SE) time-courses of alpha activity and of two measures of heart rate variability, rRR and the LF/HF ratio. The time-courses of rRR and of the LF/ HF ratio paralleled that of heart rate, with declining (P , 0:001) levels during NREM sleep, and increasing (P , 0:001) levels during REM sleep and intra-sleep awakenings. Again, alpha activity and both rRR and the LF/HF ratio were in phase opposition during NREM sleep and REM sleep. In contrast, they were in phase relationship during intra-sleep awakenings. Fig. 3 illustrates the mean (^SE) time-courses of alpha activity subdivided into two bands: alpha 1(8±10 Hz) and alpha 2 (10±13 Hz) during NREM sleep, REM sleep, and during intra-sleep awakenings. There were no signi®cant differences in the time-courses between alpha 1 and alpha 2 activity which paralleled total alpha activity, with increases during NREM sleep and during intra-sleep awakenings, and decreases during REM sleep. Topographical aspects of alpha activity in SWS, REM sleep and intra-sleep awakenings were displayed using brain-mapping techniques (Fig. 4), and tested statistically along the antero-posterior axis (Fig. 5). During SWS, alpha activity progressively increased from occipital to frontocentral areas (Fig. 4). During REM sleep, alpha activity was distributed from occipital to fronto-central areas, showing progressive decline as alpha activity became more anterior over the scalp. As expected, during intra-sleep awakenings, alpha waves were localized over occipital areas. Fig. 5 shows the distribution of alpha activity (mean ^
943
SE) along the antero-posterior axis averaged over the group of 8 subjects. During SWS, alpha activity signi®cantly (P , 0:0001) decreased along the antero-posterior axis. During REM sleep and during intra-sleep awakenings, alpha activity signi®cantly (P , 0:0001) increased. Posthoc Student t tests showed that alpha activity was signi®cantly (P , 0:01) reduced between Cz and Pz and between Pz and Oz leads during SWS, whereas it was signi®cantly (P , 0:01) increased during REM sleep and during intrasleep awakenings. No signi®cant modi®cation could be observed between Fz and Cz leads in any of the three states of vigilance.
4. Discussion Changes in EEG amplitude and frequency are the major discriminating markers between sleep and wakefulness, and between NREM sleep and REM sleep. Until now, attention has focused mainly on the relationships between the overnight time-courses in the different frequency bands, and delta wave activity has proved to be a valuable index of sleep deepening and lightening. Few studies, however, have explored alpha activities and their signi®cance during human sleep. The present study revealed that reciprocal changes in alpha power and in cardiac activity distinguish the two states of sleep and intra-sleep awakenings. Thus, alpha activity increased during NREM sleep, decreased during REM sleep, and dramatically increased during intra-sleep awakening. These changes were accompanied by changes in heart rate and in heart rate variability, which reveal changes in autonomic nervous system activity, distinctive of each state of vigilance. During NREM sleep, together with the increase in alpha activity, there was a decrease in heart rate, rRR, and in the LF/HF ratio, indicating prominent vagal activity. Conversely, REM sleep and intra-sleep awakening were associated with increases in heart rate, in rRR, and in the LF/HF ratio, indicating sympathetic activation. Thus,
Fig. 1. Mean (^SE) time-courses of alpha activity (8±13 Hz) and heart rate during NREM sleep, REM sleep, and during intra-sleep awakenings, in 10 subjects. Individual data were subdivided and averaged according to a method derived from Achermann et al. (1993). Values are expressed as percentages of the mean overnight values in each subject. The x axis gives the real time of sleep episodes, averaged for 10 subjects.
944
J. Ehrhart et al. / Clinical Neurophysiology 111 (2000) 940±946
Fig. 2. Mean (^SE) time-courses of alpha activity (8±13 Hz) and of measures of heart rate variability during NREM sleep, REM sleep, and during intra- sleep awakenings, in 10 subjects. rRR, interbeat autocorrelation coef®cient of R-R intervals; LF/HF, ratio of low frequency to high frequency power calculated from spectral analysis of R-R intervals. Individual data were averaged as in Fig. 1.
together with low sympathetic in¯uences, alpha waves could re¯ect sleep-maintaining processes, as is the case during NREM sleep, or, conversely, they could be consid-
ered as a sign of arousal, when sympathetic activity increased, as is the case during REM sleep and during intra-sleep awakenings.
Fig. 3. Mean (^SE) time-courses of alpha 1 activity (8±10 Hz) and alpha 2 activity (10±13 Hz) during NREM sleep, REM sleep, and during intra-sleep awakenings, in 10 subjects. Individual data were averaged as in Fig. 1.
J. Ehrhart et al. / Clinical Neurophysiology 111 (2000) 940±946
945
Fig. 4. Topographical patterns of absolute alpha activity (8±13 Hz) during SWS, REM sleep, and during intra-sleep awakenings. Absolute power brain maps correspond to a selected 30 s epoch in each state for one representative subject.
Studies reported that alpha activity re¯ects attentional processes during wakefulness (Ray and Cole, 1985; Mulholland, 1996). The alpha attenuation test (AAT) has been developed as a method of quantifying variations in physiological sleepiness (Stampi et al., 1993). It has been found also that different frequency bands within the alpha frequency range re¯ect quite different cognitive processes: upper alpha desynchronization (10.5±12.5 Hz) is selectively associated with the processing of sensory-semantic information, whereas desynchronization in the lower range (6.5± 10.5 Hz) re¯ects attentional processes (Klimesch et al., 1998). Such differences in the frequency range of alpha activity were not observed in the present study, in which a similar time-course of alpha 1(8±10 Hz) and alpha 2 (10±13 Hz) activity was found. In contrast, brain mapping revealed topographical differences in alpha distribution during the different sleep states. Con®rming results from previous studies (Broughton et al., 1994; Cantero et al., 1999) we found that the alpha waves occur during wakefulness over occipital areas and increase along the antero-posterior axis during REM sleep (Roth et al., 1999). Alpha waves occurred in fronto-central areas during SWS, which is less well documented (Pivik and Harman, 1995). A large intrusion of alpha waves, called `EEG NREM sleep anomaly' (Moldofsky et al., 1975), often related to musculoskeletal
Fig. 5. Distribution of alpha activity (8±13 Hz) along the antero-posterior axis (Fz, Cz, Pz, Oz) during SWS, REM sleep, and during intra-sleep awakenings, in 8 subjects. Mean (^SE) values are plotted.
symptoms, has been attributed to impaired sleep function and considered as a sign of sleep lightening and of increased vigilance (Moldofsky, 1989). This interpretation is inconsistently supported by the literature and does not account for the presence of alpha activity in normal subjects with no evidence of sleep disturbance. A more recent report describing the effects of endotoxin administration revealed signi®cant increases in the alpha waves during sleep, which, in the absence of sleep disruption, has been associated with sleep maintenance (Trachsel et al., 1994). The present data further support the theory that the alpha waves do not only represent an arousing, disturbing in¯uence, as is commonly thought. When alpha waves occur in fronto-central areas, they possibly re¯ect sleep maintaining processes. This ®nding could have important clinical and pharmacological implications. Acknowledgements We thank Daniel Joly and MicheÁle SimeÂoni for technical assistance and data treatment, and Dr Paul Bailey for his English language assistance. References Achermann P, Borbely AA. Coherence analysis of the human sleep electroencephalogram Neuroscience 1998;85:1195±1208. Achermann P, Dijk DJ, Brunner DP, Borbely AA. A model of human sleep homeostasis based on EEG slow wave activity: quantitative comparison of data and simulations. Brain Res Bull 1993;31:97±113. Aeschbach D, Borbely AA. All-night dynamics of the human sleep EEG. J Sleep Res 1993;2:70±81. Aldredge JL, Welch AJ. Variations of heart rate during sleep as a function of the sleep cycle. Electroenceph clin Neurophysiol 1973;35:193±198. ASDA (American Sleep Disorders Association and Sleep Research Society). Atlas Task Force Report EEG arousals: scoring rules and examples. Sleep 1992;15:173-184. Aserinsky E, Kleitman N. Regularly occurring periods of eye motility, and concomitant phenomena, during sleep. Science 1953;118:273±274. Baharav A, Kotagal S, Gibbons V, Rubin BK, Pratt G, Karin J, Akselrod S. Fluctuations in autonomic nervous activity during sleep displayed by power spectrum analysis of heart rate variability. Neurology 1995;45: 1183±1187. Berlad I, Shlitner A, Ben Haim S, Lavie P. Power spectrum analysis and heart rate variability in stage 4 and REM sleep: evidence for statespeci®c changes in autonomic dominance. J Sleep Res 1993;2:88±90.
946
J. Ehrhart et al. / Clinical Neurophysiology 111 (2000) 940±946
Bonnet MH, Arand DL. Heart rate variability: sleep stage, time of night, and arousal in¯uences. Electroenceph clin Neurophysiol 1997;102: 390±396. Bootsma M, Swenne CA, Bolhuis HH, Chang PC, Cats VM, Bruschke AV. Heart rate and heart rate variability as indexes of sympathovagal balance. Am J Physiol (Heart Circ Physiol 35) 1994;266:H1565± H1571. Broughton R, Hasan J, Dunham W. Anterior slow alpha of drowsiness: topography, source dipole analysis. Sleep Res 1994;23:5±10. Cajochen C, Pischke J, Aeschbach D, Borbely AA. Heart rate dynamics during human sleep. Physiol Behav 1994;55:769±774. Cantero JL, Atienza M, Gomez CM, Salas RM. Spectral structure and brain mapping of human alpha activities in different arousal states. Neuropsychobiology 1999;39:110±116. Cooley JW, Tukey JW. An algorithm for machine calculation of complex Fourier series. Math Comput 1965;19:297±301. Eckberg D. Sympathovagal balance. A critical appraisal. Circulation 1997;96:3224±3232. Hauri P, Hawkins DR. Alpha-delta sleep. Electroenceph clin Neurophysiol 1973;34:233±237. Houle MS, Billman GE. Low-frequency component of the heart rate variability spectrum: a poor marker of sympathetic activity. Am J Physiol (Heart Circ Physiol 45) 1999;276:H215±H223. IFSECN (International Federation of Societies for Electroencephalography and Clinical Neurophysiology). A glossary of terms most commonly used by clinical electroencephalographers. Electroenceph clin Neurophysiol 1974;37:538-548. Jasper WH. The ten twenty electrodes system of the International Federation. Electroenceph clin Neurophysiol 1958;10:371±375. Kamen PW, Tonkin AM. Application of the Poincare plot to heart rate variability: a new measure of functional status in heart failure. Aust NZ J Med 1995;25:18±26. Kamen PW, Krum H, Tonkin AM. Poincare plot of heart rate variability allows quantitative display of parasympathetic nervous activity in humans. Clin Sci 1996;91:201±208. Khatri IM, Freis ED. Hemodynamic changes during sleep. J Appl Physiol 1967;22:867±873. Klimesch W, Doppelmayr M, Russegger H, Pachinger T, Schwaiger J. Induced alpha band power changes in the human EEG and attention. Neurosci Lett 1998;244:73±76. Merica H, Blois R. Relationship between the time courses of power in the frequency bands of human sleep EEG. Neurophysiol Clin 1997;27:116± 128. Moldofsky H. Nonrestorative sleep and symptoms after a febrile illness in patients with ®brositis and chronic fatigue syndromes. J Rheumatol 1989;16:150±153. Moldofsky H, Scarisbrick P, England R, Smythe H. Musculoskeletal symp-
toms and non-REM sleep disturbance in patients with `®brositis syndrome' and healthy subjects. Psychomatic Med 1975;37:341±351. Mulholland T. The concept of attention and the electroencephalographic alpha rhythm. In: Evans CR, Mulholland TB, editors. Attention in Neurophysiology, London: Butterworths, 1996. pp. 100±127. Otzenberger H, Simon C, Gron®er C, Brandenberger G. Temporal relationship between dynamic heart rate variability and electroencephalographic activity during sleep in man. Neurosci Lett 1997;229:173± 176. Otzenberger H, Gron®er C, Simon C, Charloux A, Ehrhart J, Piquard F, Brandenberger G. Dynamic heart rate variability: a tool for exploring sympathovagal balance continuously during sleep in men. Am J Physiol (Heart Circ Physiol 44) 1998;275:H946±H950. Pivik RT, Harman K. A reconceptualization of EEG alpha activity as an index of arousal during sleep: all alpha activity is not equal. J Sleep Res 1995;4:131±137. Ray WY, Cole HW. EEG alpha activity re¯ects attentional demands, and beta activity re¯ects emotional and cognitive processes. Science 1985;228:750±752. Rechtschaffen A, Kales A, editors. A manual of standardized terminology, techniques and scoring system for sleep stages of human subjects. National Institute of Health Publication 204 Washington, DC: U.S. Government Printing Of®ce, 1968. Roth C, Achermann P, Borbely AA. Alpha activity in the human REM sleep EEG: topography and effect of REM sleep deprivation. Clin Neurophysiol 1999;110:632±635. Scholz UD, Bianchi AM, Cerutti S, Kubicki S. Vegetative background of sleep: spectral analysis of the heart rate variability. Physiol Behav 1997;62:1037±1043. Snyder F, Hohson JA, Morrison DF, Golpank F. Changes in respiration, heart rate and systolic blood pressure in human sleep. J Appl Physiol 1964;19:417±422. Stampi C, Stone P, Michimori A. The alpha attenuation test: a new quantitative method for assessing sleepiness and its relationship to the MSLT. Sleep Res 1993;22:115±119. Trachsel L, Schreiber W, Holsboer F, PollmaÈcher T. Endotoxin enhances EEG alpha and beta power in human sleep. Sleep 1994;17:132±139. Vanoli E, Adamson PB, Ba-Lin S, Pinna GD, Lazzara R, Orr WC. Heart period variability during speci®c sleep stages. A comparison of healthy subjects with patients after myocardial infarction. Circulation 1995;91: 1918±1922. Vaughn BV, Quint SR, Messenheimer JA, Robertson KR. Heart period variability in sleep. Electroenceph clin Neurophysiol 1995;94:155±162. Welch AJ, Richardson PC. Computer sleep stage classi®cation using heart rate data. Electroenceph clin Neurophysiol 1972;34:145±152. Zemaityte D, Varoneckas G, Sokolov E. Heart rhythm control during sleep. Psychophysiology 1984;21:279±289.
Alpha activity and cardiac correlates: three types of relationships during nocturnal sleep J. Ehrhart a,*, M. Toussaint b, C. Simon a, C. Gron®er a, R. Luthringer b, G. Brandenberger a a
Laboratoire des Regulations Physiologiques et des Rythmes Biologiques chez l'Homme, 4, rue Kirschleger, 67085 Strasbourg Cedex, France b Forenap, Centre Hospitalier de Rouffach 68250 Rouffach, France Accepted 29 December 1999
Abstract Objective: We examined simultaneously alpha activity and cardiac changes during nocturnal sleep, in order to differentiate non-rapid eye movement (NREM) sleep, REM sleep, and intra-sleep awakening. Methods: Ten male subjects displaying occasionally spontaneous intra-sleep awakenings underwent EEG and cardiac recordings during one experimental night. The heart rate and heart rate variability were calculated over 5 min periods. Heart rate variability was estimated: (1) by the ratio of low frequency (LF) to high frequency (HF) power calculated from spectral analysis of R-R intervals; and (2) by the interbeat autocorrelation coef®cient of R-R intervals (rRR). EEG spectral analysis was performed using a fast Fourier transform algorithm. Results: Three types of relationships between alpha waves (8±13 Hz) and cardiac correlates could be distinguished. During NREM sleep, alpha activity and cardiac correlates showed opposite variations, with high levels of alpha power associated with decreased heart rate, rRR and LF/HF ratio, indicating low sympathetic activity. Conversely, during REM sleep, alpha activity was low whereas heart rate, rRR, and the LF/HF ratio peaked, indicating high sympathetic activity. During intra-sleep awakenings, alpha activity and cardiac correlates both increased. No difference in time-course between alpha 1 (8±10 Hz) and alpha 2 (10±13 Hz) activity could be shown. Alpha waves occurred in fronto-central areas during slow wave sleep (SWS), migrated to posterior areas during REM sleep, and were localized in occipital areas during intra-sleep awakenings. Conclusions: These results suggest that alpha waves are not simply a sign of arousal, as is commonly thought. Fronto-central alpha waves, associated with decreased heart rate, possibly re¯ect sleep-maintaining processes. q 2000 Elsevier Science Ireland Ltd. All rights reserved. Keywords: EEG activity; Alpha power; Heart rate; Heart rate variability; Sleep; Human
1. Introduction Studies on cardiac activity have described systematic variations of heart rate during sleep, with an overall slowing during the night and ultradian variations associated with rapid eye movement (REM)-non(N)REM sleep cycles. Heart rate is lowest during slow wave sleep (SWS) and higher and more labile, with abrupt ¯uctuations, during REM sleep (Aserinsky and Kleitman, 1953; Snyder et al., 1964; Khatri and Freis, 1967; Welch and Richardson, 1972; Aldredge and Welch, 1973). Not only heart rate but also heart rate variability differs according to sleep stages. Studies using Poincare plots have demonstrated distinctive and characteristic patterns depending on sleep stages, REM sleep being characterized by wider overall variations than those observed during NREM sleep (Zemaityte et al., 1984; * Corresponding author. Tel.: 133-3-8824-3558; fax: 133-3-88243334. E-mail address: [email protected] (J. Ehrhart)
Kamen and Tonkin, 1995; Vaughn et al., 1995; Kamen et al., 1996). Studies using spectral analysis of R-R intervals to assess heart rate variability reported a lower ratio of low frequency (LF) to high frequency (HF) components during NREM sleep than during REM sleep, which suggests a sympathetic dominance in the autonomic system during REM sleep and a predominant vagal in¯uence during NREM sleep (Berlad et al., 1993; Bootsma et al., 1994; Baharav et al., 1995; Vanoli et al., 1995; Bonnet et al., 1997; Scholz et al., 1997), although this interpretation has been questioned (Eckberg, 1997; Houle and Billman, 1999). All these results were based on conventional scoring of sleep stages. Electroencephalographic (EEG) spectral analysis makes possible a more detailed and dynamic description of sleep processes and, in recent years, time course of power in the different frequency bands has been precisely de®ned (Aeschbach and Borbely, 1993; Merica and Blois, 1997; Achermann and Borbely, 1998). However, only a few studies have examined overnight cardiac changes
1388-2457/00/$ - see front matter q 2000 Elsevier Science Ireland Ltd. All rights reserved. PII: S13 88-2457(00)0024 7-9
CLINPH 99694
J. Ehrhart et al. / Clinical Neurophysiology 111 (2000) 940±946
together with sleep EEG activity. It has been reported that heart rate displays ultradian variations which are not closely correlated with changes in slow wave activity (Cajochen et al., 1994). In a recent study, we found a relationship between overnight pro®les of EEG mean frequency and the interbeat autocorrelation coef®cient of R-R intervals (rRR) (Otzenberger et al., 1997). We also reported that rRR is related to overnight variations in the LF/HF ratio (Otzenberger et al., 1998). Very few studies have examined EEG activity in the alpha band in human sleep. Alpha activity has a frequency of 8±13 Hz and is generally accepted as an indicator of relaxed wakefulness (IFSECN, 1974). The desynchronization of alpha activity is employed in visual sleep stage scoring systems as a sign of sleep onset. During sleep, visually detected alpha waves have been thought to re¯ect arousal processes and a shift toward wakefulness (ASDA, 1992). However, more recent investigations, based on spectral analysis of sleep EEG, reported that alpha waves are also present during NREM and REM sleep, which suggests possible functional differences for alpha waves during sleep (Cooley and Tukey, 1965; Hauri and Hawkins, 1973; Pivik and Harman, 1995). This prompted us to examine the time course of alpha waves during NREM sleep, REM sleep, and intra- sleep awakening in relation to other signs of physiological arousal, such as heart rate and heart rate variability, which may differentiate sleep-maintaining from sleep-disturbing processes. Further analyses were conducted in order to test whether the differences observed in these relationships could be attributed to different spectral contributions in the alpha band, or whether they re¯ect topographical differences of alpha generators. 2. Methods and materials Ten male subjects (21±28 years old) displaying occasionally spontaneous awakenings during nocturnal sleep were included in the study. They gave their written consent to participate. They had regular sleep-wake habits and did not take any medication. The study was approved by the local Ethics Committee. The experiments were carried out in a soundproof, air-conditioned sleep room. After an habituation night, the subjects underwent one experimental session during which EEG and cardiac recordings were carried out. Electrodes for polysomnographic recordings were applied 2 h before the beginning of the experiment. Lights were switched off at 23:00 h and the subjects were awakened at 07:00 h. 2.1. Sleep analysis Sleep recordings were performed using 4 EEG leads (F3, C3, P3 vs. A2 and C4 vs. A1), one chin electromyographic (EMG) and one diagonal electrooculographic (EOG) lead (upper canthus of one eye vs. lower canthus of the other
941
eye). The recordings were visually scored at 30 s intervals using standardized criteria (Rechtschaffen and Kales, 1968). For all-night spectral analysis, the EEG signal (C3-A2 or C4-A1) was high-pass (0.3 Hz) and low-pass (35 Hz) ®ltered before conversion from analog to digital with a sampling frequency of 128 Hz. Subsequently, spectra were computed for consecutive 2 s periods with a fast Fourier transform algorithm (Bootsma et al., 1994). In order to yield one spectrum every 5 min, a median ®lter was applied for 150 consecutive 2 s periods. The spectral parameter studied was the absolute power in the alpha frequency band (8±13 Hz); for further analysis, the total alpha frequency band was subdivided into alpha 1 (8±10 Hz) and alpha 2 (10±13 Hz) frequency bands, which have been shown to differ according to the physiological state of the subject (Cantero et al., 1999). 2.2. Heart rate analysis The electrocardiogram signal was fed into a generator that produced a pulse at the rising phase of each R wave. The trigger event times were recorded with a precision of ^1 ms, and the R-R intervals were calculated on a computer equipped with a data acquisition control board including a timer. Computers and polygraphs were synchronized. Ectopic beats were identi®ed and replaced with interpolative RR interval data. Each R-R interval was plotted against the previous R-R interval to produce a 5 min Poincare plot (RRn 1 l vs. R-Rn). The interbeat autocorrelation coef®cient of R-R intervals (rRR; i.e. Pearson's correlation coef®cient between R-Rn and R-Rn 1 l) was calculated over 5 min periods. Power spectral analysis of each consecutive 5 min recording was performed in a sequential fashion with the use of a fast Fourier transform (based on a non-parametric algorithm using a Welsh window) after the ectopic-free data were detrended and resampled. A ®xed resampling frequency of 1024 equally spaced points per 5 min period was used. The power in the LF band (0.04±0.15 Hz) and in the HF band (0.15±0.50 Hz) was calculated for each 5 min density spectrum by integrating the spectral density in the respective frequency bands. The LF/HF ratio was calculated using the power in each band. 2.3. Statistical analysis To assess the relationship between the time-courses of alpha activity, heart rate and the measures of heart rate variability, we selected for each subject one uninterrupted REM-NREM-REM sleep episode containing SWS, one NREM-REM-NREM sleep episode, and one NREM-intrasleep awakening-NREM episode. In each case, NREM, REM and intra-sleep awakening lasted at least 10 min, and the immediately preceding and immediately following sleep episode lasted at least 10 min. In order to compensate for the individual differences in the duration of the NREM sleep episodes, REM sleep
942
J. Ehrhart et al. / Clinical Neurophysiology 111 (2000) 940±946
episodes and intra-sleep awakenings, the individual pro®les were normalized according to a method derived from Achermann et al. (1993). Individual data were subdivided into equal parts within each period: (1) into 16 parts from the beginning to the end of NREM sleep; (2) into 8 parts from the beginning to the end of REM sleep; (3) into 16 parts from the beginning to the end of intra-sleep awakening; (4) into 4 parts for the sleep episodes surrounding the three states of vigilance. Data were averaged over subjects and the standard error was calculated for each interval. Mean changes in alpha activity, in heart rate and in the measures of heart rate variability were analyzed by one-way ANOVA for repeated measures with time as a factor. Results are expressed in percentages of the mean overnight values, which normalizes the data across subjects. The difference was considered to be signi®cant if P , 0:05. 2.4. Topographical analysis An additional group of 8 subjects (4 males, 4 females; 20±31 years old) participated in a second study in which topographical distribution of alpha activity was investigated. Twenty-eight EEG channels were recorded from all subjects, each channel referred to a common linked-ear reference with an impedance lower than 5 KOhm. Electrode placement followed the 10±20 system (Jasper, 1958) and the additional electrodes were each placed at the center of a space formed by 4 standard electrodes. Two EOG leads and one chin-EMG were added to allow visual sleep stage scoring (30 s epochs) according to the criteria of Rechtschaffen and Kales (1968). The ®lter settings were, respectively, 0.5±30 Hz for EEGs, 0.5±20 Hz for EOGs,
and 5±70 Hz for EMG. All recordings were sampled at 128 Hz. The spectral analysis was performed by the fast Fourier transform algorithm on consecutive 2 s epochs (256 points) with a 0.5 Hz frequency resolution, and truncating error was reduced by applying a Hamming window. In addition, the values for 15 adjacent 2 s epochs were averaged to yield spectral density values for 30 s periods computed for each of the 28 EEG channels. Thus, sleep visual scores of each 30 s period were synchronized with spectral density values. To reduce the noise generated by the relatively short epoch length, a median ®lter was applied with a window of 15 points (i.e. 30 s). Channels along the antero-posterior axis (Fz, Cz, Pz, Oz) were subjected to a one-way ANOVA for repeated measures with electrode as factor. Contrasts were assessed by two tailed paired Student t tests. To reduce interindividual variability, data were expressed as percentages of the mean overnight values.
3. Results Table 1 gives the mean (^SE) changes in alpha activity and in cardiac correlates during NREM sleep, REM sleep, and during intra- sleep awakenings, in 10 subjects. As expected, alpha activity showed large interindividual differences. During NREM sleep, individual basal levels ranged from 1.57 to 6.05 mV 2 and individual peak values ranged from 3.08 to 22.16 mV 2. Therefore, levels are given in percentages of the mean overnight value, which normalizes the interindividual differences in the raw data. Fig. 1 illustrates the mean (^SE) time-courses of alpha
Table 1 Mean (^SE) levels of alpha activity (8±13 Hz), alpha 1 activity (8±10 Hz), alpha 2 activity (10±13 Hz), of heart rate, rRR, and of the LF/HF ratio a NREM sleep Levels Alpha activity (8±13 Hz) mV 2 % Alpha 1 activity (8±10 Hz) mV 2 % Alpha 2 activity (10±13 Hz) mV 2 % Heart rate Beats/min % rRR % LF/HF %
Basal
REM sleep Peak/trough
Basal
4.37 ^ 0.49 54.0 ^ 4.6
9.80 ^ 1.71 115.2 ^ 12.7
8.37 ^ 1.3 103.0 ^ 8.1
2.25 ^ 0.29 55.9 ^ 5.9
4.57 ^ 0.99 105.3 ^ 4.3
1.78 ^ 0.23 50.6 ^ 14.6 55.1 ^ 1.4 101.2 ^ 1.7 0.54 ^ 0.06 115.5 ^ 14.7 1.64 ^ 0.25 121.0 ^ 9.1
Intra-sleep awakenings Peak/trough
Basal
Peak/trough
4.35 ^ 0.54 56.9 ^ 5.5
8.10 ^ 2.43 90.5 ^ 9.4
14.53 ^ 4.06 151.0 ^ 20.6
3.91 ^ 0.67 100.3 ^ 9.5
2.24 ^ 0.58 60.4 ^ 8.7
3.40 ^ 0.97 83.8 ^ 13.6
7.45 ^ 1.36 163.7 ^ 7.7
4.37 ^ 0.66 121.4 ^ 10.2
3.80 ^ 0.31 105.6 ^ 5.8
1.84 ^ 0.23 54.0 ^ 5.2
4.17 ^ 2.32 98.4 ^ 28.5
6.25 ^ 2.38 132.0 ^ 17.8
51.1 ^ 0.9 94.0 ^ 1.0 0.16 ^ 0.09 26.2 ^ 20.4 0.55 ^ 0.07 43.9 ^ 6.5
54.2 ^ 1.8 98.5 ^ 2.0 0.35 ^ 0.09 64.9 ^ 13.9 0.86 ^ 0.15 62.4 ^ 8.3
57.9 ^ 1.3 105.3 ^ 1.2 0.64 ^ 0.03 136.8 ^ 13.0 2.16 ^ 0.34 155.7 ^ 18.2
54.4 ^ 2.5 93.6 ^ 1.1 0.52 ^ 0.09 82.6 ^ 11.8 2.84 ^ 1.34 86.9 ^ 11.8
61.8 ^ 2.6 106.6 ^ 2.6 0.82 ^ 0.04 135.3 ^ 8.7 4.66 ^ 2.12 NS 139.0 ^ 14.7
a rRR, interbeat autocorrelation coef®cient of R-R intervals; LF/HF, ratio of low frequency to high frequency power calculated from spectral analysis of R-R intervals. Values are expressed in raw data, and in percentages of the mean overnight values for each individual. Basal levels: highest (or lowest) values before the onset of the two states of sleep and of the intra- sleep awakenings. Peak (or trough) levels: highest (or lowest) values during these three vigilance states. P , 0:01, unless otherwise indicated.
J. Ehrhart et al. / Clinical Neurophysiology 111 (2000) 940±946
activity and of heart rate during NREM sleep, REM sleep, and during intra-sleep awakenings. Three types of relationships could be distinguished. During NREM sleep, alpha activity and cardiac correlates showed opposite variations with signi®cantly (P , 0:0001) increased alpha activity associated with signi®cantly (P , 0:0001) decreased heart rate. Conversely, during REM sleep, alpha activity decreased (P , 0.0001), whereas heart rate increased (P , 0:0001). During intra-sleep awakenings, both alpha activity and heart rate signi®cantly increased (P , 0:0001). Fig. 2 represents the mean (^SE) time-courses of alpha activity and of two measures of heart rate variability, rRR and the LF/HF ratio. The time-courses of rRR and of the LF/ HF ratio paralleled that of heart rate, with declining (P , 0:001) levels during NREM sleep, and increasing (P , 0:001) levels during REM sleep and intra-sleep awakenings. Again, alpha activity and both rRR and the LF/HF ratio were in phase opposition during NREM sleep and REM sleep. In contrast, they were in phase relationship during intra-sleep awakenings. Fig. 3 illustrates the mean (^SE) time-courses of alpha activity subdivided into two bands: alpha 1(8±10 Hz) and alpha 2 (10±13 Hz) during NREM sleep, REM sleep, and during intra-sleep awakenings. There were no signi®cant differences in the time-courses between alpha 1 and alpha 2 activity which paralleled total alpha activity, with increases during NREM sleep and during intra-sleep awakenings, and decreases during REM sleep. Topographical aspects of alpha activity in SWS, REM sleep and intra-sleep awakenings were displayed using brain-mapping techniques (Fig. 4), and tested statistically along the antero-posterior axis (Fig. 5). During SWS, alpha activity progressively increased from occipital to frontocentral areas (Fig. 4). During REM sleep, alpha activity was distributed from occipital to fronto-central areas, showing progressive decline as alpha activity became more anterior over the scalp. As expected, during intra-sleep awakenings, alpha waves were localized over occipital areas. Fig. 5 shows the distribution of alpha activity (mean ^
943
SE) along the antero-posterior axis averaged over the group of 8 subjects. During SWS, alpha activity signi®cantly (P , 0:0001) decreased along the antero-posterior axis. During REM sleep and during intra-sleep awakenings, alpha activity signi®cantly (P , 0:0001) increased. Posthoc Student t tests showed that alpha activity was signi®cantly (P , 0:01) reduced between Cz and Pz and between Pz and Oz leads during SWS, whereas it was signi®cantly (P , 0:01) increased during REM sleep and during intrasleep awakenings. No signi®cant modi®cation could be observed between Fz and Cz leads in any of the three states of vigilance.
4. Discussion Changes in EEG amplitude and frequency are the major discriminating markers between sleep and wakefulness, and between NREM sleep and REM sleep. Until now, attention has focused mainly on the relationships between the overnight time-courses in the different frequency bands, and delta wave activity has proved to be a valuable index of sleep deepening and lightening. Few studies, however, have explored alpha activities and their signi®cance during human sleep. The present study revealed that reciprocal changes in alpha power and in cardiac activity distinguish the two states of sleep and intra-sleep awakenings. Thus, alpha activity increased during NREM sleep, decreased during REM sleep, and dramatically increased during intra-sleep awakening. These changes were accompanied by changes in heart rate and in heart rate variability, which reveal changes in autonomic nervous system activity, distinctive of each state of vigilance. During NREM sleep, together with the increase in alpha activity, there was a decrease in heart rate, rRR, and in the LF/HF ratio, indicating prominent vagal activity. Conversely, REM sleep and intra-sleep awakening were associated with increases in heart rate, in rRR, and in the LF/HF ratio, indicating sympathetic activation. Thus,
Fig. 1. Mean (^SE) time-courses of alpha activity (8±13 Hz) and heart rate during NREM sleep, REM sleep, and during intra-sleep awakenings, in 10 subjects. Individual data were subdivided and averaged according to a method derived from Achermann et al. (1993). Values are expressed as percentages of the mean overnight values in each subject. The x axis gives the real time of sleep episodes, averaged for 10 subjects.
944
J. Ehrhart et al. / Clinical Neurophysiology 111 (2000) 940±946
Fig. 2. Mean (^SE) time-courses of alpha activity (8±13 Hz) and of measures of heart rate variability during NREM sleep, REM sleep, and during intra- sleep awakenings, in 10 subjects. rRR, interbeat autocorrelation coef®cient of R-R intervals; LF/HF, ratio of low frequency to high frequency power calculated from spectral analysis of R-R intervals. Individual data were averaged as in Fig. 1.
together with low sympathetic in¯uences, alpha waves could re¯ect sleep-maintaining processes, as is the case during NREM sleep, or, conversely, they could be consid-
ered as a sign of arousal, when sympathetic activity increased, as is the case during REM sleep and during intra-sleep awakenings.
Fig. 3. Mean (^SE) time-courses of alpha 1 activity (8±10 Hz) and alpha 2 activity (10±13 Hz) during NREM sleep, REM sleep, and during intra-sleep awakenings, in 10 subjects. Individual data were averaged as in Fig. 1.
J. Ehrhart et al. / Clinical Neurophysiology 111 (2000) 940±946
945
Fig. 4. Topographical patterns of absolute alpha activity (8±13 Hz) during SWS, REM sleep, and during intra-sleep awakenings. Absolute power brain maps correspond to a selected 30 s epoch in each state for one representative subject.
Studies reported that alpha activity re¯ects attentional processes during wakefulness (Ray and Cole, 1985; Mulholland, 1996). The alpha attenuation test (AAT) has been developed as a method of quantifying variations in physiological sleepiness (Stampi et al., 1993). It has been found also that different frequency bands within the alpha frequency range re¯ect quite different cognitive processes: upper alpha desynchronization (10.5±12.5 Hz) is selectively associated with the processing of sensory-semantic information, whereas desynchronization in the lower range (6.5± 10.5 Hz) re¯ects attentional processes (Klimesch et al., 1998). Such differences in the frequency range of alpha activity were not observed in the present study, in which a similar time-course of alpha 1(8±10 Hz) and alpha 2 (10±13 Hz) activity was found. In contrast, brain mapping revealed topographical differences in alpha distribution during the different sleep states. Con®rming results from previous studies (Broughton et al., 1994; Cantero et al., 1999) we found that the alpha waves occur during wakefulness over occipital areas and increase along the antero-posterior axis during REM sleep (Roth et al., 1999). Alpha waves occurred in fronto-central areas during SWS, which is less well documented (Pivik and Harman, 1995). A large intrusion of alpha waves, called `EEG NREM sleep anomaly' (Moldofsky et al., 1975), often related to musculoskeletal
Fig. 5. Distribution of alpha activity (8±13 Hz) along the antero-posterior axis (Fz, Cz, Pz, Oz) during SWS, REM sleep, and during intra-sleep awakenings, in 8 subjects. Mean (^SE) values are plotted.
symptoms, has been attributed to impaired sleep function and considered as a sign of sleep lightening and of increased vigilance (Moldofsky, 1989). This interpretation is inconsistently supported by the literature and does not account for the presence of alpha activity in normal subjects with no evidence of sleep disturbance. A more recent report describing the effects of endotoxin administration revealed signi®cant increases in the alpha waves during sleep, which, in the absence of sleep disruption, has been associated with sleep maintenance (Trachsel et al., 1994). The present data further support the theory that the alpha waves do not only represent an arousing, disturbing in¯uence, as is commonly thought. When alpha waves occur in fronto-central areas, they possibly re¯ect sleep maintaining processes. This ®nding could have important clinical and pharmacological implications. Acknowledgements We thank Daniel Joly and MicheÁle SimeÂoni for technical assistance and data treatment, and Dr Paul Bailey for his English language assistance. References Achermann P, Borbely AA. Coherence analysis of the human sleep electroencephalogram Neuroscience 1998;85:1195±1208. Achermann P, Dijk DJ, Brunner DP, Borbely AA. A model of human sleep homeostasis based on EEG slow wave activity: quantitative comparison of data and simulations. Brain Res Bull 1993;31:97±113. Aeschbach D, Borbely AA. All-night dynamics of the human sleep EEG. J Sleep Res 1993;2:70±81. Aldredge JL, Welch AJ. Variations of heart rate during sleep as a function of the sleep cycle. Electroenceph clin Neurophysiol 1973;35:193±198. ASDA (American Sleep Disorders Association and Sleep Research Society). Atlas Task Force Report EEG arousals: scoring rules and examples. Sleep 1992;15:173-184. Aserinsky E, Kleitman N. Regularly occurring periods of eye motility, and concomitant phenomena, during sleep. Science 1953;118:273±274. Baharav A, Kotagal S, Gibbons V, Rubin BK, Pratt G, Karin J, Akselrod S. Fluctuations in autonomic nervous activity during sleep displayed by power spectrum analysis of heart rate variability. Neurology 1995;45: 1183±1187. Berlad I, Shlitner A, Ben Haim S, Lavie P. Power spectrum analysis and heart rate variability in stage 4 and REM sleep: evidence for statespeci®c changes in autonomic dominance. J Sleep Res 1993;2:88±90.
946
J. Ehrhart et al. / Clinical Neurophysiology 111 (2000) 940±946
Bonnet MH, Arand DL. Heart rate variability: sleep stage, time of night, and arousal in¯uences. Electroenceph clin Neurophysiol 1997;102: 390±396. Bootsma M, Swenne CA, Bolhuis HH, Chang PC, Cats VM, Bruschke AV. Heart rate and heart rate variability as indexes of sympathovagal balance. Am J Physiol (Heart Circ Physiol 35) 1994;266:H1565± H1571. Broughton R, Hasan J, Dunham W. Anterior slow alpha of drowsiness: topography, source dipole analysis. Sleep Res 1994;23:5±10. Cajochen C, Pischke J, Aeschbach D, Borbely AA. Heart rate dynamics during human sleep. Physiol Behav 1994;55:769±774. Cantero JL, Atienza M, Gomez CM, Salas RM. Spectral structure and brain mapping of human alpha activities in different arousal states. Neuropsychobiology 1999;39:110±116. Cooley JW, Tukey JW. An algorithm for machine calculation of complex Fourier series. Math Comput 1965;19:297±301. Eckberg D. Sympathovagal balance. A critical appraisal. Circulation 1997;96:3224±3232. Hauri P, Hawkins DR. Alpha-delta sleep. Electroenceph clin Neurophysiol 1973;34:233±237. Houle MS, Billman GE. Low-frequency component of the heart rate variability spectrum: a poor marker of sympathetic activity. Am J Physiol (Heart Circ Physiol 45) 1999;276:H215±H223. IFSECN (International Federation of Societies for Electroencephalography and Clinical Neurophysiology). A glossary of terms most commonly used by clinical electroencephalographers. Electroenceph clin Neurophysiol 1974;37:538-548. Jasper WH. The ten twenty electrodes system of the International Federation. Electroenceph clin Neurophysiol 1958;10:371±375. Kamen PW, Tonkin AM. Application of the Poincare plot to heart rate variability: a new measure of functional status in heart failure. Aust NZ J Med 1995;25:18±26. Kamen PW, Krum H, Tonkin AM. Poincare plot of heart rate variability allows quantitative display of parasympathetic nervous activity in humans. Clin Sci 1996;91:201±208. Khatri IM, Freis ED. Hemodynamic changes during sleep. J Appl Physiol 1967;22:867±873. Klimesch W, Doppelmayr M, Russegger H, Pachinger T, Schwaiger J. Induced alpha band power changes in the human EEG and attention. Neurosci Lett 1998;244:73±76. Merica H, Blois R. Relationship between the time courses of power in the frequency bands of human sleep EEG. Neurophysiol Clin 1997;27:116± 128. Moldofsky H. Nonrestorative sleep and symptoms after a febrile illness in patients with ®brositis and chronic fatigue syndromes. J Rheumatol 1989;16:150±153. Moldofsky H, Scarisbrick P, England R, Smythe H. Musculoskeletal symp-
toms and non-REM sleep disturbance in patients with `®brositis syndrome' and healthy subjects. Psychomatic Med 1975;37:341±351. Mulholland T. The concept of attention and the electroencephalographic alpha rhythm. In: Evans CR, Mulholland TB, editors. Attention in Neurophysiology, London: Butterworths, 1996. pp. 100±127. Otzenberger H, Simon C, Gron®er C, Brandenberger G. Temporal relationship between dynamic heart rate variability and electroencephalographic activity during sleep in man. Neurosci Lett 1997;229:173± 176. Otzenberger H, Gron®er C, Simon C, Charloux A, Ehrhart J, Piquard F, Brandenberger G. Dynamic heart rate variability: a tool for exploring sympathovagal balance continuously during sleep in men. Am J Physiol (Heart Circ Physiol 44) 1998;275:H946±H950. Pivik RT, Harman K. A reconceptualization of EEG alpha activity as an index of arousal during sleep: all alpha activity is not equal. J Sleep Res 1995;4:131±137. Ray WY, Cole HW. EEG alpha activity re¯ects attentional demands, and beta activity re¯ects emotional and cognitive processes. Science 1985;228:750±752. Rechtschaffen A, Kales A, editors. A manual of standardized terminology, techniques and scoring system for sleep stages of human subjects. National Institute of Health Publication 204 Washington, DC: U.S. Government Printing Of®ce, 1968. Roth C, Achermann P, Borbely AA. Alpha activity in the human REM sleep EEG: topography and effect of REM sleep deprivation. Clin Neurophysiol 1999;110:632±635. Scholz UD, Bianchi AM, Cerutti S, Kubicki S. Vegetative background of sleep: spectral analysis of the heart rate variability. Physiol Behav 1997;62:1037±1043. Snyder F, Hohson JA, Morrison DF, Golpank F. Changes in respiration, heart rate and systolic blood pressure in human sleep. J Appl Physiol 1964;19:417±422. Stampi C, Stone P, Michimori A. The alpha attenuation test: a new quantitative method for assessing sleepiness and its relationship to the MSLT. Sleep Res 1993;22:115±119. Trachsel L, Schreiber W, Holsboer F, PollmaÈcher T. Endotoxin enhances EEG alpha and beta power in human sleep. Sleep 1994;17:132±139. Vanoli E, Adamson PB, Ba-Lin S, Pinna GD, Lazzara R, Orr WC. Heart period variability during speci®c sleep stages. A comparison of healthy subjects with patients after myocardial infarction. Circulation 1995;91: 1918±1922. Vaughn BV, Quint SR, Messenheimer JA, Robertson KR. Heart period variability in sleep. Electroenceph clin Neurophysiol 1995;94:155±162. Welch AJ, Richardson PC. Computer sleep stage classi®cation using heart rate data. Electroenceph clin Neurophysiol 1972;34:145±152. Zemaityte D, Varoneckas G, Sokolov E. Heart rhythm control during sleep. Psychophysiology 1984;21:279±289.