- Email: [email protected]
Does more sleep matter? Differential effects of NREM- and REM-dominant sleep on sleepiness and vigilance
Neurophysiologie Clinique/Clinical Neurophysiology (2015) 45, 167—175
Disponible en ligne sur
ScienceDirect www.sciencedirect.com
ORIGINAL ARTICLE/ARTICLE ORIGINAL
Does more sleep matter? Differential effects of NREM- and REM-dominant sleep on sleepiness and vigilance Est-ce mieux de dormir plus ? Effets différentiels du sommeil lent et du sommeil paradoxal sur la somnolence et la vigilance D. Neu a,∗,b,c, O. Mairesse a,b,d,e, J. Newell a, P. Verbanck a,b, P. Peigneux f, G. Deliens a,f a
Brugmann University Hospital, U.L.B./V.U.B, Sleep Laboratory & Unit for Chronobiology U78, Arthur Van Gehuchten Square, Building Hh, 1020 Brussels, Belgium b UNI, ULB Neurosciences Institute, Faculty of Medicine, Laboratory for Medical Psychology ULB312, Université Libre de Bruxelles (U.L.B.), Brussels, Belgium c Center for the Study of Sleep Disorders, Edith Cavell Medical Institute, CHIREC, Brussels, Belgium d Department of Experimental and Applied Psychology (EXTO), Vrije Universiteit Brussel (V.U.B.), Brussels, Belgium e Royal Military Academy (R.M.A.), Department LIFE, Brussels, Belgium f UR2NF, Neuropsychology and Functional Neuroimaging Research Group at CRCN - Center for Research in Cognition and Neurosciences, Université Libre de Bruxelles (ULB) and UNI - ULB Neurosciences Institute, Brussels, Belgium Received 12 April 2014; accepted 25 October 2014 Available online 15 April 2015
KEYWORDS NREM sleep; REM sleep; Sleepiness; Psychomotor vigilance test; Visual analog scale; Split-night
∗
Summary We investigated effects of NREM and REM predominant sleep periods on sleepiness and psychomotor performances measured with visual analog scales and the psychomotor vigilance task, respectively. After one week of stable sleep-wake rhythms, 18 healthy sleepers slept 3 hours of early sleep and 3 hours of late sleep, under polysomnographic control, spaced by two hours of sustained wakefulness between sleep periods in a within subjects split-night, sleep interruption protocol. Power spectra analysis was applied for sleep EEG recordings and sleep phase-relative power proportions were computed for six different frequency bands (delta, theta, alpha, sigma, beta and gamma). Both sleep periods presented with similar sleep duration and efficiency. As expected, phasic NREM and REM predominances were obtained for early and late sleep conditions, respectively. Albeit revealing additive effects of total sleep duration, our results showed a systematic discrepancy between psychomotor performances and sleepiness
Corresponding author. Tel.: +2 0 2 4772554; fax: +32 0 2 4772162. E-mail address: [email protected] (D. Neu).
http://dx.doi.org/10.1016/j.neucli.2014.10.004 0987-7053/© 2015 Elsevier Masson SAS. All rights reserved.
168
D. Neu et al. levels. In addition, sleepiness remained stable throughout sustained wakefulness during both conditions, whereas psychomotor performances even decreased after the second sleep period. Disregarding exchanges for frequency bands in NREM or stability in REM, correlations between outcome measures and EEG power proportions further evidenced directional divergence with respect to sleepiness and psychomotor performances, respectively. Showing that the functional correlation pattern changed with respect to early and late sleep condition, the relationships between EEG power and subjective or behavioral outcomes might however essentially be related to total sleep duration rather than to the phasic predominance of REM or NREM sleep. © 2015 Elsevier Masson SAS. All rights reserved.
MOTS CLÉS Sommeil lent (SL) ; Sommeil paradoxal (SP) ; Somnolence ; Test psychomoteur de vigilance ; Échelle visuelle analogique ; Nuit coupée
Résumé Nous avons étudié les effets de 2 périodes de sommeil, composées majoritairement de sommeil lent (SL) ou de sommeil paradoxal (SP), sur la somnolence et la vigilance en utilisant des échelles visuelles analogiques et une tache de vigilance psychomotrice. Dans le cadre d’un protocole intra-sujet d’interruption de sommeil, des 18 enregistrements polysomnographiques ont été effectués chez 18 sujets sains pendant 2 périodes de 3 heures en début et en fin de nuit entrecoupées par 2 heures d’éveil maintenu. Nous avons réalisé une analyse des puissances spectrales de l’EEG de sommeil et calculé les proportions de puissance relative par phase de sommeil pour 6 différentes bandes de fréquence (delta, theta, alpha, sigma, beta, et gamma). La durée et l’efficacité de sommeil étaient similaires dans les deux conditions. Nos résultats confirment la prédominance de SL et de SP en début et fin de nuit, respectivement et montrent des effets additifs de la durée totale de sommeil ainsi qu’une divergence systématique entre les performances psychomotrices et les niveaux de somnolence. En effet, la somnolence reste stable au cours de la veille soutenue dans les deux conditions alors que les performances psychomotrices diminuent après le second épisode de sommeil. Abstraction faite des changements de distribution des puissances spectrales en SL ou de la stabilité en SP, les corrélations entre les résultats et les proportions de puissance EEG montrent une divergence directionnelle entre la somnolence et les performances psychomotrices. En montrant que le pattern de corrélations fonctionnelles change en début et fin de nuit, la relation entre les puissances EEG et les données subjectives et comportementales semble essentiellement liée à la durée totale de sommeil plutôt qu’à la prédominance phasique de SP ou de SL. © 2015 Elsevier Masson SAS. Tous droits réservés.
Introduction Total sleep deprivation is known to raise subjective feelings of sleepiness and markedly impact on psychomotor performance. However, effects of partial sleep deprivation (PSD) are far less consistent. Discrepancies between PSD paradigms might stem from various factors including the type of sleep restriction and its duration, potential circadian biases and population sampling [1,13]. Sleep restriction to 4 hours for two consecutive nights during either the first or the second part of the night resulted in higher amounts of slow wave sleep (SWS) and rapid eye movement (REM) sleep and lowered amounts of light (N2) sleep after late than early sleep during the recovering night’s sleep recording [11]. A similar sleep restriction regimen was associated with greater increases in the percentage of SWS than of REM sleep, irrespective of sleep restriction taking place the first or the second half of the night. Still, sleep resistance was less affected when sleep was allowed in the second half than in the first half of the night [6]. Finally, sleep restriction to 3 hours for 4 consecutive days mainly resulted in increased SWS percentage and associated decreases in light sleep (N1 and N2) both after the first and the second part of the night, whereas REM sleep decreased more during an early than in a late night sleep restriction [13].
At the behavioral level, mild to moderate sleep restriction for 7 consecutive days during the second part of the night eventually led to reduced vigilance (as assessed using the psychomotor vigilance task, PVT) and stabilization below baseline level for the last 2 days, without a concomitant increase in subjective sleepiness levels [2]. Furthermore, deterioration of daytime vigilance on the PVT was found more pronounced in a late night sleep restriction condition [13]. Hence, available data suggest that non-REM (NREM) sleep, and SWS in particular, are more closely related than REM sleep to PSD-associated changes in vigilance and sleepiness parameters. These results are in line with models of vigilance stating that objective vigilance (as measured using the PVT) is mostly regulated by homeostatic sleep pressure, respectively related to SWS restriction or recovery, and not specifically related to REM sleep. Considering that the proportion of NREM and REM sleep dynamically vary across the night, SWS predominating during the first part of a normal night of sleep and REM propensity increasing later on to the expense of SWS [11], it is conceivable that vigilance and sleep resistance are differentially impacted by early vs late sleep restriction paradigms. However, it remains unclear how subjective sleepiness levels and psychomotor performance will be differentially modulated by NREM vs REM predominance within a single
Early and late sleep impact on sleepiness and vigilance night in the absence of a prior sleep restriction. Using a within-subject design, in which participants get first NREMthen REM-dominant sleep periods, with an intermediate wakefulness period, we thus propose to investigate the potential additive and/or interactive effects of sleep phases on sleepiness and vigilance. In this sleep interruption protocol, young healthy sleepers with neutral chronotype sleep first for 3 hours (‘‘early sleep’’ condition) followed by a period of 2 hours of sustained wakefulness, then sleep again for another 3 hours (‘‘late sleep’’ condition) with subsequent final awakening. Psychomotor vigilance and sleepiness were assessed immediately and one hour after each period of sleep. We hypothesized that (1) sleepiness levels and vigilance performance will improve differently with sustained wake duration after each period of sleep and (2) that neurobehavioral stability and sleepiness will further improve after fulfillment of both sleep periods. Hence, we surmised here that even if early sleep might substantially differ from late sleep with respect to the recovery of sleepiness and vigilance, ‘‘more’’ sleep also matters.
Methods Subjects Twenty young healthy participants gave their written informed consent to participate in a previously published sleep and memory study [3] approved by the local ethics committee. They were required to keep regular sleep patterns during two weeks before and throughout the experiment, remain free from neuropharmacological treatments and psychotropics, and refrain from alcohol and stimulant drinks. Regularity of sleep habits was controlled using actigraphic recording (Daqtometer, Daqtix GbR, Oetzen, Germany) and the completion of daily sleep logs. Two participants were excluded from the analyses due to irregular sleep/wake schedules. The final population included 18 subjects (mean age 26.2 ± 4.7 years, 11 males) who met the following criteria: non-smokers or fewer than 10 cigarettes per day, no medical history of neurological or sleep disorders, no current or past affective disorders (Structured Mini International Neuropsychiatric Interview for DSM-IV, American Psychiatric Association, 2004) and intermediate or neutral chronotype [7].
Psychomotor vigilance test (PVT) The PVT is a simple reaction time (RT) task aimed at to evaluating behavioral vigilance [4]. Participants are instructed to respond as fast as possible by pressing a button as soon as a digital counter starts on the computer’s screen. The stimulus appears at random intervals between 2000 and 10,000 milliseconds, for 10 minutes. After each trial, a visual feedback is provided showing reaction times (RTs) in milliseconds (ms). Whereas mean RTs mainly reflect psychomotor speed, quantitative analyses have previously shown that slowest (90th percentile) and fastest (10th percentile) RTs are sensitive to changes in homeostatic sleep pressure [5], and that lapses (number of RTs > 500 ms) reflect neurobehavioral stability [5,12]. In addition, the interpercentile range between
169 fastest and slowest RTs (difference 10th—90th percentile) is particularly sensitive to sleep resistance [5,12].
Procedure After at least one week of actigraphic and sleep diaries recordings, in order to assure stable sleep-wake patterns preceding the experiment, participants slept in the laboratory during one night for 2 consecutive periods of 3 hours, interspersed with 2 hours of sustained wakefulness (from 23:30 h to 02:30 h [early sleep] and from 04.30 h to 07.30 h [late sleep]). The psychomotor vigilance test (PVT, 10 min) and visual analog scales for sleepiness (VAS-S) were administered at baseline (22.30 h) and after early and late sleep at two time points spaced by a 60 minutes interval (immediately at wake-up and 1 hour later, see Fig. 1).
EEG recordings and spectral analysis Polygraphic recordings (Fig. 1) comprised 3 EEG leads (C3-A2, O1-A2, FP1-A2 derivations, according to the International 10—20 system), electromyography (EMG), and electro-oculograms (EOG). Effective recording sampling rate was 2000 Hz (Alice5® , Philips Respironics IncTM , Philips HealthcareTM , Eindhoven, The Netherlands, European Union). For all recordings, a low bandpass filter at 45 Hz and a high pass filter at 0.75 Hz were applied to EEG channels. According to the criteria of the American Academy of Sleep Medicine, the 36 sleep recordings were visually scored a posteriori as wake, NREM 1 (N1), 2 (N2), SWS (N3) and REM stages by 30 seconds (s) epochs on 21-inch screens. Sleep onset latency (SOL) was computed as the time from lightsoff to the first 30 s epoch of sleep, sleep period time (SPT) as the time interval from sleep onset to final awakening, total sleep time (TST) as SPT minus the total duration of cumulated intra-sleep awakenings (wake time after sleep onset, WASO), and sleep efficiency (SEI) as TST divided by time in bed (TIB = SPT + SOL), expressed in percent. Spectral analysis was computed on the central derivation using the PRANA® software package (PhiTools© , Strasbourg, France). Artifacts due to eye and muscle movements were automatically detected and removed prior to the spectral analysis. The EEG signal was then visually checked for the accuracy of the computerized artifact detection. Fast Fourier transformation was applied on 4 s intervals (frequency resolution of 0.25 Hz). Power spectra were then averaged over 30 s intervals to match the sleep stage scoring, and the subsequent power of standard frequency bands was computed (␦: 0.75—4.5 Hz; : 4.5—8.5 Hz; ␣: 8.5—12.5; : 12.5—15.5; : 15.5—22.5; ␥: 22.5—45.5). Intervals containing more than 50% of artifacts were considered as missing data.
Statistical analyses Analyses were conducted using paired-samples t-tests or repeated-measures analyses of variance (ANOVAs). Violations of sphericity were assessed using Mauchly’s tests. Greenhouse-Geisser corrections are reported when sphericity is violated. Inspection of higher order polynomials was performed to reveal shape differences in power spectrum
170
D. Neu et al.
Figure 1 Study protocol. Arrows indicate measurements and assessments of the psychomotor vigilance test (PVT) and of visual analog scales (VAS). PSG: polysomnographic recordings during early and late sleep conditions.
distribution within each sleep phase. Relationships between variables were evaluated using Pearson’s coefficients of correlation. Two-tailed significance was set at P < .05 (unless otherwise stated). All statistical analyses were performed using SPSS 22® (Industrial Business Machines© , SPSS© Inc., Chicago, IL, USA).
Results Standard sleep variables Recording time (time in bed, TIB), sleep period time (SPT) and total sleep time (TST) were similar between early and late sleep. However, wake time after sleep onset (WASO) was significantly higher in the early than in the late sleep condition (Table 1). As expected, standard PSG derived parameters disclosed significant differences between sleep conditions in NREM and REM proportions (and SWS in particular; Table 1). Hence, according to standard sleep variables’ results, the early sleep condition can be depicted as NREM predominant and the late sleep condition as REM predominant.
Sleepiness and psychomotor vigilance Fig. 2 shows factorial plots of sleepiness (VAS-S) and PVT outcomes immediately after and one-hour after forced awakening in the early and late sleep conditions,
Table 1
respectively. Our prediction of a further improvement onehour after the second period of sleep is depicted in Fig. 2 a. Data inspection reveals an apparent discrepancy between sleepiness and psychomotor performance (Fig. 2 b—e). Regarding sleepiness levels, the observed parallelism in the factorial plots reveals an additive effect of successive early then late sleep period. The ANOVA disclosed a main effect of the sleep period (early vs late sleep; F(1,17) = 9.929, P < .01, p2 = .37), a non-significant effect of measurement time (i.e. upon forced awakening vs after one hour of wakefulness; F(1,17) = 3.064, P = .098, p2 = .15) and a non-significant interaction effect between the two factors (F(1,17) = .256, P = .620, p2 = .02). Regarding psychomotor speed, reaction times were similar after early vs late sleep (F(1,16) = 2.856, P = .110, p2 = .15) and upon forced awakening vs one hour later (F(1,16) = .287, P = .600, p2 = .02). The interaction between sleep period and measurement time was non-significant (F(1,16) = 1.006, P = .331, p2 = .06). There was a main effect of the sleep period for sleep resistance (i.e. RT difference between 10th—90th percentile; F(1,17) = 11.849, P < .005, p2 = .411), but not of measurement time (forced awakening vs one hour later; F(1,17) = .014, P = .909, p2 = .01). Again, the interaction between the sleep period and the measurement time was not significant (F(1,17) = .006, P = .940, p2 = .01). Similarly, neurobehavioral stability (i.e. number of lapses) was lower after late than early sleep (F(1,16) = 7.611, P < .05, p2 = .32), similar upon forced awakening vs one hour later (F(1,16) = .441,
Sleep variables. Early
TIB (min) SPT (min) TST (min) SOL (min) WASO (min) SEI (%) NREMS (min) REMS (min) N1 (%) N2 (%) N3 (%) REM (%)
202.42 188.47 160.19 13.94 28.28 80.73 160.19 11.53 5.66 50.77 36.51 7.06
Late (34.64) (36.02) (14.58) (8.56) (33.03) (12.27) (14.58) (9.30) (3.22) (6.83) (10.54) (5.63)
189.06 174.50 163.86 13.56 10.64 87.12 117.78 46.08 5.10 51.00 15.95 28.18
(3.18) (12.15) (13.64) (11.54) (9.31) (6.96) (12.90) (11.18) (2.16) (6.35) (9.74) (6.18)
t (17)
P
1.695 1.740 1.333 .196 2.119 —1.981 11.871 11.024 .673 —.165 7.692 —24.974
NS NS NS NS .049 (.064) < .001 < .001 NS NS < .001 < .001
Early and late sleep conditions are depicted in Fig. 1. TIB: time in bed; SPT: sleep period time; TST: total sleep time = SPT − wake after sleep onset (WASO); SOL: sleep onset latency; SEI: sleep efficiency index = TST/TIB, expressed in percent; REMS: rapid eye movement sleep; non-REMS (NREMS) stages N1, N2 and N3 (slow wave sleep); values are presented as means ± standard deviation. Scores are presented as mean scores in percent of total sleep time (TST) ± standard deviation. Ratios are presented as means ± standard deviation and expressed as percentages. Trends are marked between brackets.
Early and late sleep impact on sleepiness and vigilance
171
Figure 2 Factorial plots of psychomotor vigilance and sleepiness after sleep periods. Early and late sleep conditions are depicted in Fig. 1 and present measurements at 02.30 h and 03.30 h for the early condition and 07.30 h and 08.30 h for the late condition, respectively. Prediction illustration upon hypothesis: a. Experimental results: b—e. Sleepiness visual analog scale (VAS) expressed in centimeters; mean reaction time (RT) is the average RT drawn from the psychomotor vigilance test (PVT); slowest reaction times (SRT) are the average of the 10% slowest RTs; fastest reaction times (FRT) are the average of the 10% fastest RTs; SRT − FRT = PVT range; PVT lapses: number of RTs > 500 ms. Error bars indicate standard deviation.
P = .516, p2 = .03), and the interaction was non-significant (F(1,16) = .565, P = .463, p2 = .03).
Spectral analysis Given the significant differences of REM and NREM proportions between conditions (Table 1), spectral power in
each defined frequency band was assessed as the relative power proportion of the total power within each sleep phase (REM and NREM) in order to control for differences of respective sleep phase durations [9] in each condition. Fig. 3 shows early and late sleep differences in sleep phaserelative spectral power distributions for the six computed frequency bands [e.g. NREM phase-relative delta power = V2 delta NREM/ V2 delta NREM+V2 theta NREM+V2 alpha
172
D. Neu et al.
Figure 3
Normalized spectral power proportions in each frequency band per total power of each sleep phase 2 [(fV )2a ]N/R ˙a (V ) N/R N/R reads as the considered sleep phase (NREM or REM), ‘‘a’’ to ‘‘f’’ read as the 6 considered frequency bands and (V2 ) as the computed power within each frequency band. (a) and (b) shows standardized phase-relative power distributions of early and late sleep conditions. Comparisons of relative power between early and late sleep were controlled for sleep phase durations and carried out within subjects as repeated ANOVA. While power distribution within frequency bands during REM sleep remains stable across both conditions, NREM sleep displays a consistent power exchange from slower to faster frequencies (all P < .0001) between both conditions.
NREM + V2 sigma NREM + V2 beta NREM + V2 gamma NREM)]. Phase-relative spectral powers were standardized using a linear z-transformation to allow between-conditions comparisons. While power distribution within frequency bands during REM sleep remains similar across both conditions (F(5,70) = 2.710, P = .064, p2 = .162), there was a significantly different power band distribution in NREM sleep between conditions (F(5,85) = 38.539, P < .0001, p2 = .694). Visual inspection of the factorial plots suggests a difference in curvature of the power spectra distributions between early and late conditions for NREM sleep. Statistically, we find the latter being mainly located in the quadratic term of the interaction between conditions and power distribution (F(1,17) = 58.861, P < .0001, p2 = .776). The late sleep period presents here with a consistent power exchange from slower to faster frequency bands.
Functional correlations of sleep EEG power proportions and outcome measures Correlations between sleep EEG power bands’ proportions and behavioral outcomes (PVT and VAS) are depicted for all conditions (early/late, time awake) Fig. 4. The apparent discrepancy between subjective (VAS) and behavioral measurements (PVT) is evidenced here by directional differences in their correlations with the phase-relative EEG power spectra. Upon forced awakening, after the first sleep period (early condition, 02:30 h), higher sleepiness was associated with decreased NREM delta power proportions and increased proportions of NREM theta power. Similarly, lower proportions of REM delta power were associated with higher
sleepiness, as were higher proportions of REM alpha, sigma, beta and gamma. With respect to psychomotor vigilance, the decrease in sleep resistance was associated with lower power proportions of the sigma and beta frequency bands during NREM. The latter was also associated with an increase in attentional lapses (all P’s < .05, 1-tailed). We further found no significant associations between PVT outcomes and proportions of REM power spectra. When measured one hour after the awakening (at 03:30 h), higher sleepiness levels were associated with lower proportions of delta power and higher proportions of alpha power during REM sleep. For all PVT outcomes (mean RT, lapses and interpercentile range; higher scores signify worse psychomotor performance), we found an inverse relation with theta, beta and gamma proportions in NREM sleep. Additionally, NREM alpha power proportions were negatively correlated with mean RTs (all P’s < .05, 1-tailed). During REM sleep, lower gamma power proportions were associated with a general decrease in psychomotor performance (all P’s < .01, 1-tailed). Immediately upon forced awakening (at 07:30 h) after the second sleep period (late sleep condition), higher delta and lower beta power proportions during NREM were associated with a decrease in sleep resistance. Lower proportions of NREM alpha power were in turn associated with an increased number of attentional lapses, as were lower proportions of alpha and sigma during NREM. Measured one hour later (08:30 h), the PVT interpercentile range was significantly correlated with NREM delta power and negatively related with NREM alpha power proportions (all P’s < .05, 1-tailed). Subjective sleepiness was not significantly correlated to any EEG power spectrum proportion when measured immediately or one hour later, after late sleep.
Early and late sleep impact on sleepiness and vigilance
173
Figure 4 Functional correlations between power proportions of sleep EEG frequency bands and VAS/PVT outcome measures. Functional correlation illustrates Pearson’s moment r for each psychomotor vigilance and sleepiness measure in relation to sleep phase-relative EEG power across six different frequency bands. In order to allow visual comparison between both conditions (early and late) and sleep phases, the plots are organized in measurements of the early conditions (02.30 h and 03.30 h) shown in the first two horizontal rows of plots and relations for NREM in the first vertical row. Dotted lines represent significance thresholds at the P < .05 level.
174
Discussion After one week of stable, monitored sleep-wake rhythms, participants entered a within-subjects sleep interruption protocol comprising sustained wakefulness for 120 minutes in between a NREM- and a REM-sleep predominant sleep period of fixed duration (3 hours), aimed at assessing the potentially additive impact of these sleep periods on psychomotor performance and subjective vigilance measurements at forced awakening and one hour later. Despite higher WASO during the first sleep period, PSG recordings evidenced similar sleep efficiency, total sleep time duration and light sleep proportions. Both early and late sleep conditions showed the expected significant phasic predominance for NREM and REM sleep, respectively. With respect to within-subjects comparisons between behavioral outcome measures (VAS and PVT), and in line with previous studies reporting dissociations between subjective and objective alertness measures [8,12], our results suggest a systematic discrepancy between psychomotor performance and sleepiness levels. Although an additive effect of successive sleep periods is found for sleepiness, that improved after ‘‘more’’ sleep, sleepiness levels remained stable throughout sustained wakefulness during both conditions. The more tight relation between homeostatic drives and slow wave activity (i.e. delta power during SWS) during early sleep and their impact on objective vigilance levels (PVT) is confirmed here. Given that the divergence between subjective and objective measurements exists, the latter indicates that perception may also be related to more (REM) sleep rather than the sole resolution of homeostatic pressure (SWS and delta power during the early sleep condition). It has previously been suggested that different neurophysiological mechanisms might indeed underlie sleepiness and neurobehavioral responses to sleep loss, potentially reflective of the activity of different underlying brain regions [10]. In the present study, sleep resistance and neurobehavioral stability even decreased after the second sleep period, despite the accumulated sleep duration. In order to investigate whether the power band distribution of brain activity may account for these findings, we assessed phase-relative EEG power proportions, thus controlling for durations of respective sleep phases. In line with homeostatic regulation models, our results showed a significant power shift of frequency bands towards more rapid frequencies during the second NREM period (i.e. in the late sleep condition), but a consistent stability for the REM sleep power distribution. Therefore, we computed pairwise correlations between behavioral outcome measures and the proportions of phaserelative spectral power. The apparent divergence between psychomotor performance and subjective sleepiness was again evidenced here by directional discrepancies in their correlation patterns. In the present study, more (REM) sleep improved subjective sleepiness levels, on the one hand, but also more surprisingly decreased neurobehavioral stability and resistance to sleep. To disentangle the effects of sleep duration and REM/NREM sleep predominance, we investigated the relationship between subjective and behavioral vigilance and EEG spectral power proportional to each sleep phase, independently of their occurrence early or late in the night. As shown in functional correlation plots, the
D. Neu et al. relationships between outcome measures and EEG power are similar for NREM and REM sleep phases, irrespective of their occurrence during early or late sleep. However, correlational patterns do change with respect to the early or late sleep condition (i.e. subjective sleepiness and behavioral outcomes diverge, albeit to a lesser extent after late sleep). To sum up, our results indicate that (1) neurobehavioral and subjective vigilance measures bear differential relationships with early and late sleep periods, maybe due to different underlying neurophysiological mechanisms, and that (2) the relationships between brain oscillation patterns and each of these measures, taken individually, remain similar irrespective of the sleep phase predominance. Therefore, our results reinforce prior observations and dissociations between sleepiness and vigilance measures in a single, interrupted night comprising a sufficient sustained wakefulness period. Finally, the observed relationships between EEG power and subjective or behavioral outcomes appear to be essentially related to the total sleep duration rather than to the predominance of either REM or NREM sleep.
Disclosure of interest The authors declare that they have no conflicts of interest concerning this article.
References [1] Åkerstedt T, Kecklund G, Ingre M, Lekander M, Axelsson J. Sleep homeostasis during repeated sleep restriction and recovery: support from EEG dynamics. Sleep 2009;32:217—22. [2] Belenky G, Wesensten NJ, Thorne DR, Thomas ML, Sing HC, Redmond DP, et al. Patterns of performance degradation and restoration during sleep restriction and subsequent recovery: a sleep dose-response study. J Sleep Res 2003;12:1—12. [3] Deliens G, Neu D, Peigneux P. Rapid eye movement sleep does not seem to unbind memories from their emotional context. J Sleep Res 2013;22:656—62. [4] Dinges DF, Powell JW. Microcomputer analyses of performance on a portable, simple visual RT task during sustained operations. Behav Res Methods 1985;17:652—5. [5] Graw P, Kräuchi K, Knoblauch V, Wirz-Justice A, Cajochen C. Circadian and wake-dependent modulation of fastest and slowest reaction times during the psychomotor vigilance task. Physiol Behav 2004;80:695—701. [6] Guilleminault C, Powell NB, Martinez S, Kushida C, Raffray T, Palombini L, et al. Preliminary observations on the effects of sleep time in a sleep restriction paradigm. Sleep Med 2003;4:177—84. [7] Horne JA, Ostberg O. A self-assessment questionnaire to determine morningness-eveningness in human circadian rhythms. Int J Chronobiol 1976;4:97—110. [8] Leproult R, Colecchia EF, Berardi AM, Stickgold R, Kosslyn SM, Van Cauter E. Individual differences in subjective and objective alertness during sleep deprivation are stable and unrelated. Am J Physiol Regul Integr Comp Physiol 2003;284:280—90. [9] Neu D, Mairesse O, Verbanck P, Linkowski P, Le Bon O. NonREM sleep EEG power distribution in fatigue and sleepiness. J Psychosom Res 2014;76:286—91. [10] Rupp TL, Wesensten NJ, Balkin TJ. Trait-like vulnerability to total and partial sleep loss. Sleep 2012;35:1163—72.
Early and late sleep impact on sleepiness and vigilance [11] Tilley AJ, Wilkinson RT. The effects of a restricted sleep regimen on the composition of sleep and on performance. Psychophysiology 1984;21:406—12. [12] Van Dongen HP, Maislin G, Mullington JM, Dinges DF. The cumulative cost of additional wakefulness: dose-response effects on neurobehavioral functions and sleep physiology from
175 chronic sleep restriction and total sleep deprivation. Sleep 2003;26:117—26. [13] Wu H, Stone WS, His X, Zhuang J, Huang L, Yin Y, et al. Effects of different sleep restriction protocols on sleep architecture and daytime vigilance in healthy men. Physiol Res 2010;59: 821—9.
Disponible en ligne sur
ScienceDirect www.sciencedirect.com
ORIGINAL ARTICLE/ARTICLE ORIGINAL
Does more sleep matter? Differential effects of NREM- and REM-dominant sleep on sleepiness and vigilance Est-ce mieux de dormir plus ? Effets différentiels du sommeil lent et du sommeil paradoxal sur la somnolence et la vigilance D. Neu a,∗,b,c, O. Mairesse a,b,d,e, J. Newell a, P. Verbanck a,b, P. Peigneux f, G. Deliens a,f a
Brugmann University Hospital, U.L.B./V.U.B, Sleep Laboratory & Unit for Chronobiology U78, Arthur Van Gehuchten Square, Building Hh, 1020 Brussels, Belgium b UNI, ULB Neurosciences Institute, Faculty of Medicine, Laboratory for Medical Psychology ULB312, Université Libre de Bruxelles (U.L.B.), Brussels, Belgium c Center for the Study of Sleep Disorders, Edith Cavell Medical Institute, CHIREC, Brussels, Belgium d Department of Experimental and Applied Psychology (EXTO), Vrije Universiteit Brussel (V.U.B.), Brussels, Belgium e Royal Military Academy (R.M.A.), Department LIFE, Brussels, Belgium f UR2NF, Neuropsychology and Functional Neuroimaging Research Group at CRCN - Center for Research in Cognition and Neurosciences, Université Libre de Bruxelles (ULB) and UNI - ULB Neurosciences Institute, Brussels, Belgium Received 12 April 2014; accepted 25 October 2014 Available online 15 April 2015
KEYWORDS NREM sleep; REM sleep; Sleepiness; Psychomotor vigilance test; Visual analog scale; Split-night
∗
Summary We investigated effects of NREM and REM predominant sleep periods on sleepiness and psychomotor performances measured with visual analog scales and the psychomotor vigilance task, respectively. After one week of stable sleep-wake rhythms, 18 healthy sleepers slept 3 hours of early sleep and 3 hours of late sleep, under polysomnographic control, spaced by two hours of sustained wakefulness between sleep periods in a within subjects split-night, sleep interruption protocol. Power spectra analysis was applied for sleep EEG recordings and sleep phase-relative power proportions were computed for six different frequency bands (delta, theta, alpha, sigma, beta and gamma). Both sleep periods presented with similar sleep duration and efficiency. As expected, phasic NREM and REM predominances were obtained for early and late sleep conditions, respectively. Albeit revealing additive effects of total sleep duration, our results showed a systematic discrepancy between psychomotor performances and sleepiness
Corresponding author. Tel.: +2 0 2 4772554; fax: +32 0 2 4772162. E-mail address: [email protected] (D. Neu).
http://dx.doi.org/10.1016/j.neucli.2014.10.004 0987-7053/© 2015 Elsevier Masson SAS. All rights reserved.
168
D. Neu et al. levels. In addition, sleepiness remained stable throughout sustained wakefulness during both conditions, whereas psychomotor performances even decreased after the second sleep period. Disregarding exchanges for frequency bands in NREM or stability in REM, correlations between outcome measures and EEG power proportions further evidenced directional divergence with respect to sleepiness and psychomotor performances, respectively. Showing that the functional correlation pattern changed with respect to early and late sleep condition, the relationships between EEG power and subjective or behavioral outcomes might however essentially be related to total sleep duration rather than to the phasic predominance of REM or NREM sleep. © 2015 Elsevier Masson SAS. All rights reserved.
MOTS CLÉS Sommeil lent (SL) ; Sommeil paradoxal (SP) ; Somnolence ; Test psychomoteur de vigilance ; Échelle visuelle analogique ; Nuit coupée
Résumé Nous avons étudié les effets de 2 périodes de sommeil, composées majoritairement de sommeil lent (SL) ou de sommeil paradoxal (SP), sur la somnolence et la vigilance en utilisant des échelles visuelles analogiques et une tache de vigilance psychomotrice. Dans le cadre d’un protocole intra-sujet d’interruption de sommeil, des 18 enregistrements polysomnographiques ont été effectués chez 18 sujets sains pendant 2 périodes de 3 heures en début et en fin de nuit entrecoupées par 2 heures d’éveil maintenu. Nous avons réalisé une analyse des puissances spectrales de l’EEG de sommeil et calculé les proportions de puissance relative par phase de sommeil pour 6 différentes bandes de fréquence (delta, theta, alpha, sigma, beta, et gamma). La durée et l’efficacité de sommeil étaient similaires dans les deux conditions. Nos résultats confirment la prédominance de SL et de SP en début et fin de nuit, respectivement et montrent des effets additifs de la durée totale de sommeil ainsi qu’une divergence systématique entre les performances psychomotrices et les niveaux de somnolence. En effet, la somnolence reste stable au cours de la veille soutenue dans les deux conditions alors que les performances psychomotrices diminuent après le second épisode de sommeil. Abstraction faite des changements de distribution des puissances spectrales en SL ou de la stabilité en SP, les corrélations entre les résultats et les proportions de puissance EEG montrent une divergence directionnelle entre la somnolence et les performances psychomotrices. En montrant que le pattern de corrélations fonctionnelles change en début et fin de nuit, la relation entre les puissances EEG et les données subjectives et comportementales semble essentiellement liée à la durée totale de sommeil plutôt qu’à la prédominance phasique de SP ou de SL. © 2015 Elsevier Masson SAS. Tous droits réservés.
Introduction Total sleep deprivation is known to raise subjective feelings of sleepiness and markedly impact on psychomotor performance. However, effects of partial sleep deprivation (PSD) are far less consistent. Discrepancies between PSD paradigms might stem from various factors including the type of sleep restriction and its duration, potential circadian biases and population sampling [1,13]. Sleep restriction to 4 hours for two consecutive nights during either the first or the second part of the night resulted in higher amounts of slow wave sleep (SWS) and rapid eye movement (REM) sleep and lowered amounts of light (N2) sleep after late than early sleep during the recovering night’s sleep recording [11]. A similar sleep restriction regimen was associated with greater increases in the percentage of SWS than of REM sleep, irrespective of sleep restriction taking place the first or the second half of the night. Still, sleep resistance was less affected when sleep was allowed in the second half than in the first half of the night [6]. Finally, sleep restriction to 3 hours for 4 consecutive days mainly resulted in increased SWS percentage and associated decreases in light sleep (N1 and N2) both after the first and the second part of the night, whereas REM sleep decreased more during an early than in a late night sleep restriction [13].
At the behavioral level, mild to moderate sleep restriction for 7 consecutive days during the second part of the night eventually led to reduced vigilance (as assessed using the psychomotor vigilance task, PVT) and stabilization below baseline level for the last 2 days, without a concomitant increase in subjective sleepiness levels [2]. Furthermore, deterioration of daytime vigilance on the PVT was found more pronounced in a late night sleep restriction condition [13]. Hence, available data suggest that non-REM (NREM) sleep, and SWS in particular, are more closely related than REM sleep to PSD-associated changes in vigilance and sleepiness parameters. These results are in line with models of vigilance stating that objective vigilance (as measured using the PVT) is mostly regulated by homeostatic sleep pressure, respectively related to SWS restriction or recovery, and not specifically related to REM sleep. Considering that the proportion of NREM and REM sleep dynamically vary across the night, SWS predominating during the first part of a normal night of sleep and REM propensity increasing later on to the expense of SWS [11], it is conceivable that vigilance and sleep resistance are differentially impacted by early vs late sleep restriction paradigms. However, it remains unclear how subjective sleepiness levels and psychomotor performance will be differentially modulated by NREM vs REM predominance within a single
Early and late sleep impact on sleepiness and vigilance night in the absence of a prior sleep restriction. Using a within-subject design, in which participants get first NREMthen REM-dominant sleep periods, with an intermediate wakefulness period, we thus propose to investigate the potential additive and/or interactive effects of sleep phases on sleepiness and vigilance. In this sleep interruption protocol, young healthy sleepers with neutral chronotype sleep first for 3 hours (‘‘early sleep’’ condition) followed by a period of 2 hours of sustained wakefulness, then sleep again for another 3 hours (‘‘late sleep’’ condition) with subsequent final awakening. Psychomotor vigilance and sleepiness were assessed immediately and one hour after each period of sleep. We hypothesized that (1) sleepiness levels and vigilance performance will improve differently with sustained wake duration after each period of sleep and (2) that neurobehavioral stability and sleepiness will further improve after fulfillment of both sleep periods. Hence, we surmised here that even if early sleep might substantially differ from late sleep with respect to the recovery of sleepiness and vigilance, ‘‘more’’ sleep also matters.
Methods Subjects Twenty young healthy participants gave their written informed consent to participate in a previously published sleep and memory study [3] approved by the local ethics committee. They were required to keep regular sleep patterns during two weeks before and throughout the experiment, remain free from neuropharmacological treatments and psychotropics, and refrain from alcohol and stimulant drinks. Regularity of sleep habits was controlled using actigraphic recording (Daqtometer, Daqtix GbR, Oetzen, Germany) and the completion of daily sleep logs. Two participants were excluded from the analyses due to irregular sleep/wake schedules. The final population included 18 subjects (mean age 26.2 ± 4.7 years, 11 males) who met the following criteria: non-smokers or fewer than 10 cigarettes per day, no medical history of neurological or sleep disorders, no current or past affective disorders (Structured Mini International Neuropsychiatric Interview for DSM-IV, American Psychiatric Association, 2004) and intermediate or neutral chronotype [7].
Psychomotor vigilance test (PVT) The PVT is a simple reaction time (RT) task aimed at to evaluating behavioral vigilance [4]. Participants are instructed to respond as fast as possible by pressing a button as soon as a digital counter starts on the computer’s screen. The stimulus appears at random intervals between 2000 and 10,000 milliseconds, for 10 minutes. After each trial, a visual feedback is provided showing reaction times (RTs) in milliseconds (ms). Whereas mean RTs mainly reflect psychomotor speed, quantitative analyses have previously shown that slowest (90th percentile) and fastest (10th percentile) RTs are sensitive to changes in homeostatic sleep pressure [5], and that lapses (number of RTs > 500 ms) reflect neurobehavioral stability [5,12]. In addition, the interpercentile range between
169 fastest and slowest RTs (difference 10th—90th percentile) is particularly sensitive to sleep resistance [5,12].
Procedure After at least one week of actigraphic and sleep diaries recordings, in order to assure stable sleep-wake patterns preceding the experiment, participants slept in the laboratory during one night for 2 consecutive periods of 3 hours, interspersed with 2 hours of sustained wakefulness (from 23:30 h to 02:30 h [early sleep] and from 04.30 h to 07.30 h [late sleep]). The psychomotor vigilance test (PVT, 10 min) and visual analog scales for sleepiness (VAS-S) were administered at baseline (22.30 h) and after early and late sleep at two time points spaced by a 60 minutes interval (immediately at wake-up and 1 hour later, see Fig. 1).
EEG recordings and spectral analysis Polygraphic recordings (Fig. 1) comprised 3 EEG leads (C3-A2, O1-A2, FP1-A2 derivations, according to the International 10—20 system), electromyography (EMG), and electro-oculograms (EOG). Effective recording sampling rate was 2000 Hz (Alice5® , Philips Respironics IncTM , Philips HealthcareTM , Eindhoven, The Netherlands, European Union). For all recordings, a low bandpass filter at 45 Hz and a high pass filter at 0.75 Hz were applied to EEG channels. According to the criteria of the American Academy of Sleep Medicine, the 36 sleep recordings were visually scored a posteriori as wake, NREM 1 (N1), 2 (N2), SWS (N3) and REM stages by 30 seconds (s) epochs on 21-inch screens. Sleep onset latency (SOL) was computed as the time from lightsoff to the first 30 s epoch of sleep, sleep period time (SPT) as the time interval from sleep onset to final awakening, total sleep time (TST) as SPT minus the total duration of cumulated intra-sleep awakenings (wake time after sleep onset, WASO), and sleep efficiency (SEI) as TST divided by time in bed (TIB = SPT + SOL), expressed in percent. Spectral analysis was computed on the central derivation using the PRANA® software package (PhiTools© , Strasbourg, France). Artifacts due to eye and muscle movements were automatically detected and removed prior to the spectral analysis. The EEG signal was then visually checked for the accuracy of the computerized artifact detection. Fast Fourier transformation was applied on 4 s intervals (frequency resolution of 0.25 Hz). Power spectra were then averaged over 30 s intervals to match the sleep stage scoring, and the subsequent power of standard frequency bands was computed (␦: 0.75—4.5 Hz; : 4.5—8.5 Hz; ␣: 8.5—12.5; : 12.5—15.5; : 15.5—22.5; ␥: 22.5—45.5). Intervals containing more than 50% of artifacts were considered as missing data.
Statistical analyses Analyses were conducted using paired-samples t-tests or repeated-measures analyses of variance (ANOVAs). Violations of sphericity were assessed using Mauchly’s tests. Greenhouse-Geisser corrections are reported when sphericity is violated. Inspection of higher order polynomials was performed to reveal shape differences in power spectrum
170
D. Neu et al.
Figure 1 Study protocol. Arrows indicate measurements and assessments of the psychomotor vigilance test (PVT) and of visual analog scales (VAS). PSG: polysomnographic recordings during early and late sleep conditions.
distribution within each sleep phase. Relationships between variables were evaluated using Pearson’s coefficients of correlation. Two-tailed significance was set at P < .05 (unless otherwise stated). All statistical analyses were performed using SPSS 22® (Industrial Business Machines© , SPSS© Inc., Chicago, IL, USA).
Results Standard sleep variables Recording time (time in bed, TIB), sleep period time (SPT) and total sleep time (TST) were similar between early and late sleep. However, wake time after sleep onset (WASO) was significantly higher in the early than in the late sleep condition (Table 1). As expected, standard PSG derived parameters disclosed significant differences between sleep conditions in NREM and REM proportions (and SWS in particular; Table 1). Hence, according to standard sleep variables’ results, the early sleep condition can be depicted as NREM predominant and the late sleep condition as REM predominant.
Sleepiness and psychomotor vigilance Fig. 2 shows factorial plots of sleepiness (VAS-S) and PVT outcomes immediately after and one-hour after forced awakening in the early and late sleep conditions,
Table 1
respectively. Our prediction of a further improvement onehour after the second period of sleep is depicted in Fig. 2 a. Data inspection reveals an apparent discrepancy between sleepiness and psychomotor performance (Fig. 2 b—e). Regarding sleepiness levels, the observed parallelism in the factorial plots reveals an additive effect of successive early then late sleep period. The ANOVA disclosed a main effect of the sleep period (early vs late sleep; F(1,17) = 9.929, P < .01, p2 = .37), a non-significant effect of measurement time (i.e. upon forced awakening vs after one hour of wakefulness; F(1,17) = 3.064, P = .098, p2 = .15) and a non-significant interaction effect between the two factors (F(1,17) = .256, P = .620, p2 = .02). Regarding psychomotor speed, reaction times were similar after early vs late sleep (F(1,16) = 2.856, P = .110, p2 = .15) and upon forced awakening vs one hour later (F(1,16) = .287, P = .600, p2 = .02). The interaction between sleep period and measurement time was non-significant (F(1,16) = 1.006, P = .331, p2 = .06). There was a main effect of the sleep period for sleep resistance (i.e. RT difference between 10th—90th percentile; F(1,17) = 11.849, P < .005, p2 = .411), but not of measurement time (forced awakening vs one hour later; F(1,17) = .014, P = .909, p2 = .01). Again, the interaction between the sleep period and the measurement time was not significant (F(1,17) = .006, P = .940, p2 = .01). Similarly, neurobehavioral stability (i.e. number of lapses) was lower after late than early sleep (F(1,16) = 7.611, P < .05, p2 = .32), similar upon forced awakening vs one hour later (F(1,16) = .441,
Sleep variables. Early
TIB (min) SPT (min) TST (min) SOL (min) WASO (min) SEI (%) NREMS (min) REMS (min) N1 (%) N2 (%) N3 (%) REM (%)
202.42 188.47 160.19 13.94 28.28 80.73 160.19 11.53 5.66 50.77 36.51 7.06
Late (34.64) (36.02) (14.58) (8.56) (33.03) (12.27) (14.58) (9.30) (3.22) (6.83) (10.54) (5.63)
189.06 174.50 163.86 13.56 10.64 87.12 117.78 46.08 5.10 51.00 15.95 28.18
(3.18) (12.15) (13.64) (11.54) (9.31) (6.96) (12.90) (11.18) (2.16) (6.35) (9.74) (6.18)
t (17)
P
1.695 1.740 1.333 .196 2.119 —1.981 11.871 11.024 .673 —.165 7.692 —24.974
NS NS NS NS .049 (.064) < .001 < .001 NS NS < .001 < .001
Early and late sleep conditions are depicted in Fig. 1. TIB: time in bed; SPT: sleep period time; TST: total sleep time = SPT − wake after sleep onset (WASO); SOL: sleep onset latency; SEI: sleep efficiency index = TST/TIB, expressed in percent; REMS: rapid eye movement sleep; non-REMS (NREMS) stages N1, N2 and N3 (slow wave sleep); values are presented as means ± standard deviation. Scores are presented as mean scores in percent of total sleep time (TST) ± standard deviation. Ratios are presented as means ± standard deviation and expressed as percentages. Trends are marked between brackets.
Early and late sleep impact on sleepiness and vigilance
171
Figure 2 Factorial plots of psychomotor vigilance and sleepiness after sleep periods. Early and late sleep conditions are depicted in Fig. 1 and present measurements at 02.30 h and 03.30 h for the early condition and 07.30 h and 08.30 h for the late condition, respectively. Prediction illustration upon hypothesis: a. Experimental results: b—e. Sleepiness visual analog scale (VAS) expressed in centimeters; mean reaction time (RT) is the average RT drawn from the psychomotor vigilance test (PVT); slowest reaction times (SRT) are the average of the 10% slowest RTs; fastest reaction times (FRT) are the average of the 10% fastest RTs; SRT − FRT = PVT range; PVT lapses: number of RTs > 500 ms. Error bars indicate standard deviation.
P = .516, p2 = .03), and the interaction was non-significant (F(1,16) = .565, P = .463, p2 = .03).
Spectral analysis Given the significant differences of REM and NREM proportions between conditions (Table 1), spectral power in
each defined frequency band was assessed as the relative power proportion of the total power within each sleep phase (REM and NREM) in order to control for differences of respective sleep phase durations [9] in each condition. Fig. 3 shows early and late sleep differences in sleep phaserelative spectral power distributions for the six computed frequency bands [e.g. NREM phase-relative delta power = V2 delta NREM/ V2 delta NREM+V2 theta NREM+V2 alpha
172
D. Neu et al.
Figure 3
Normalized spectral power proportions in each frequency band per total power of each sleep phase 2 [(fV )2a ]N/R ˙a (V ) N/R N/R reads as the considered sleep phase (NREM or REM), ‘‘a’’ to ‘‘f’’ read as the 6 considered frequency bands and (V2 ) as the computed power within each frequency band. (a) and (b) shows standardized phase-relative power distributions of early and late sleep conditions. Comparisons of relative power between early and late sleep were controlled for sleep phase durations and carried out within subjects as repeated ANOVA. While power distribution within frequency bands during REM sleep remains stable across both conditions, NREM sleep displays a consistent power exchange from slower to faster frequencies (all P < .0001) between both conditions.
NREM + V2 sigma NREM + V2 beta NREM + V2 gamma NREM)]. Phase-relative spectral powers were standardized using a linear z-transformation to allow between-conditions comparisons. While power distribution within frequency bands during REM sleep remains similar across both conditions (F(5,70) = 2.710, P = .064, p2 = .162), there was a significantly different power band distribution in NREM sleep between conditions (F(5,85) = 38.539, P < .0001, p2 = .694). Visual inspection of the factorial plots suggests a difference in curvature of the power spectra distributions between early and late conditions for NREM sleep. Statistically, we find the latter being mainly located in the quadratic term of the interaction between conditions and power distribution (F(1,17) = 58.861, P < .0001, p2 = .776). The late sleep period presents here with a consistent power exchange from slower to faster frequency bands.
Functional correlations of sleep EEG power proportions and outcome measures Correlations between sleep EEG power bands’ proportions and behavioral outcomes (PVT and VAS) are depicted for all conditions (early/late, time awake) Fig. 4. The apparent discrepancy between subjective (VAS) and behavioral measurements (PVT) is evidenced here by directional differences in their correlations with the phase-relative EEG power spectra. Upon forced awakening, after the first sleep period (early condition, 02:30 h), higher sleepiness was associated with decreased NREM delta power proportions and increased proportions of NREM theta power. Similarly, lower proportions of REM delta power were associated with higher
sleepiness, as were higher proportions of REM alpha, sigma, beta and gamma. With respect to psychomotor vigilance, the decrease in sleep resistance was associated with lower power proportions of the sigma and beta frequency bands during NREM. The latter was also associated with an increase in attentional lapses (all P’s < .05, 1-tailed). We further found no significant associations between PVT outcomes and proportions of REM power spectra. When measured one hour after the awakening (at 03:30 h), higher sleepiness levels were associated with lower proportions of delta power and higher proportions of alpha power during REM sleep. For all PVT outcomes (mean RT, lapses and interpercentile range; higher scores signify worse psychomotor performance), we found an inverse relation with theta, beta and gamma proportions in NREM sleep. Additionally, NREM alpha power proportions were negatively correlated with mean RTs (all P’s < .05, 1-tailed). During REM sleep, lower gamma power proportions were associated with a general decrease in psychomotor performance (all P’s < .01, 1-tailed). Immediately upon forced awakening (at 07:30 h) after the second sleep period (late sleep condition), higher delta and lower beta power proportions during NREM were associated with a decrease in sleep resistance. Lower proportions of NREM alpha power were in turn associated with an increased number of attentional lapses, as were lower proportions of alpha and sigma during NREM. Measured one hour later (08:30 h), the PVT interpercentile range was significantly correlated with NREM delta power and negatively related with NREM alpha power proportions (all P’s < .05, 1-tailed). Subjective sleepiness was not significantly correlated to any EEG power spectrum proportion when measured immediately or one hour later, after late sleep.
Early and late sleep impact on sleepiness and vigilance
173
Figure 4 Functional correlations between power proportions of sleep EEG frequency bands and VAS/PVT outcome measures. Functional correlation illustrates Pearson’s moment r for each psychomotor vigilance and sleepiness measure in relation to sleep phase-relative EEG power across six different frequency bands. In order to allow visual comparison between both conditions (early and late) and sleep phases, the plots are organized in measurements of the early conditions (02.30 h and 03.30 h) shown in the first two horizontal rows of plots and relations for NREM in the first vertical row. Dotted lines represent significance thresholds at the P < .05 level.
174
Discussion After one week of stable, monitored sleep-wake rhythms, participants entered a within-subjects sleep interruption protocol comprising sustained wakefulness for 120 minutes in between a NREM- and a REM-sleep predominant sleep period of fixed duration (3 hours), aimed at assessing the potentially additive impact of these sleep periods on psychomotor performance and subjective vigilance measurements at forced awakening and one hour later. Despite higher WASO during the first sleep period, PSG recordings evidenced similar sleep efficiency, total sleep time duration and light sleep proportions. Both early and late sleep conditions showed the expected significant phasic predominance for NREM and REM sleep, respectively. With respect to within-subjects comparisons between behavioral outcome measures (VAS and PVT), and in line with previous studies reporting dissociations between subjective and objective alertness measures [8,12], our results suggest a systematic discrepancy between psychomotor performance and sleepiness levels. Although an additive effect of successive sleep periods is found for sleepiness, that improved after ‘‘more’’ sleep, sleepiness levels remained stable throughout sustained wakefulness during both conditions. The more tight relation between homeostatic drives and slow wave activity (i.e. delta power during SWS) during early sleep and their impact on objective vigilance levels (PVT) is confirmed here. Given that the divergence between subjective and objective measurements exists, the latter indicates that perception may also be related to more (REM) sleep rather than the sole resolution of homeostatic pressure (SWS and delta power during the early sleep condition). It has previously been suggested that different neurophysiological mechanisms might indeed underlie sleepiness and neurobehavioral responses to sleep loss, potentially reflective of the activity of different underlying brain regions [10]. In the present study, sleep resistance and neurobehavioral stability even decreased after the second sleep period, despite the accumulated sleep duration. In order to investigate whether the power band distribution of brain activity may account for these findings, we assessed phase-relative EEG power proportions, thus controlling for durations of respective sleep phases. In line with homeostatic regulation models, our results showed a significant power shift of frequency bands towards more rapid frequencies during the second NREM period (i.e. in the late sleep condition), but a consistent stability for the REM sleep power distribution. Therefore, we computed pairwise correlations between behavioral outcome measures and the proportions of phaserelative spectral power. The apparent divergence between psychomotor performance and subjective sleepiness was again evidenced here by directional discrepancies in their correlation patterns. In the present study, more (REM) sleep improved subjective sleepiness levels, on the one hand, but also more surprisingly decreased neurobehavioral stability and resistance to sleep. To disentangle the effects of sleep duration and REM/NREM sleep predominance, we investigated the relationship between subjective and behavioral vigilance and EEG spectral power proportional to each sleep phase, independently of their occurrence early or late in the night. As shown in functional correlation plots, the
D. Neu et al. relationships between outcome measures and EEG power are similar for NREM and REM sleep phases, irrespective of their occurrence during early or late sleep. However, correlational patterns do change with respect to the early or late sleep condition (i.e. subjective sleepiness and behavioral outcomes diverge, albeit to a lesser extent after late sleep). To sum up, our results indicate that (1) neurobehavioral and subjective vigilance measures bear differential relationships with early and late sleep periods, maybe due to different underlying neurophysiological mechanisms, and that (2) the relationships between brain oscillation patterns and each of these measures, taken individually, remain similar irrespective of the sleep phase predominance. Therefore, our results reinforce prior observations and dissociations between sleepiness and vigilance measures in a single, interrupted night comprising a sufficient sustained wakefulness period. Finally, the observed relationships between EEG power and subjective or behavioral outcomes appear to be essentially related to the total sleep duration rather than to the predominance of either REM or NREM sleep.
Disclosure of interest The authors declare that they have no conflicts of interest concerning this article.
References [1] Åkerstedt T, Kecklund G, Ingre M, Lekander M, Axelsson J. Sleep homeostasis during repeated sleep restriction and recovery: support from EEG dynamics. Sleep 2009;32:217—22. [2] Belenky G, Wesensten NJ, Thorne DR, Thomas ML, Sing HC, Redmond DP, et al. Patterns of performance degradation and restoration during sleep restriction and subsequent recovery: a sleep dose-response study. J Sleep Res 2003;12:1—12. [3] Deliens G, Neu D, Peigneux P. Rapid eye movement sleep does not seem to unbind memories from their emotional context. J Sleep Res 2013;22:656—62. [4] Dinges DF, Powell JW. Microcomputer analyses of performance on a portable, simple visual RT task during sustained operations. Behav Res Methods 1985;17:652—5. [5] Graw P, Kräuchi K, Knoblauch V, Wirz-Justice A, Cajochen C. Circadian and wake-dependent modulation of fastest and slowest reaction times during the psychomotor vigilance task. Physiol Behav 2004;80:695—701. [6] Guilleminault C, Powell NB, Martinez S, Kushida C, Raffray T, Palombini L, et al. Preliminary observations on the effects of sleep time in a sleep restriction paradigm. Sleep Med 2003;4:177—84. [7] Horne JA, Ostberg O. A self-assessment questionnaire to determine morningness-eveningness in human circadian rhythms. Int J Chronobiol 1976;4:97—110. [8] Leproult R, Colecchia EF, Berardi AM, Stickgold R, Kosslyn SM, Van Cauter E. Individual differences in subjective and objective alertness during sleep deprivation are stable and unrelated. Am J Physiol Regul Integr Comp Physiol 2003;284:280—90. [9] Neu D, Mairesse O, Verbanck P, Linkowski P, Le Bon O. NonREM sleep EEG power distribution in fatigue and sleepiness. J Psychosom Res 2014;76:286—91. [10] Rupp TL, Wesensten NJ, Balkin TJ. Trait-like vulnerability to total and partial sleep loss. Sleep 2012;35:1163—72.
Early and late sleep impact on sleepiness and vigilance [11] Tilley AJ, Wilkinson RT. The effects of a restricted sleep regimen on the composition of sleep and on performance. Psychophysiology 1984;21:406—12. [12] Van Dongen HP, Maislin G, Mullington JM, Dinges DF. The cumulative cost of additional wakefulness: dose-response effects on neurobehavioral functions and sleep physiology from
175 chronic sleep restriction and total sleep deprivation. Sleep 2003;26:117—26. [13] Wu H, Stone WS, His X, Zhuang J, Huang L, Yin Y, et al. Effects of different sleep restriction protocols on sleep architecture and daytime vigilance in healthy men. Physiol Res 2010;59: 821—9.