- Email: [email protected]
Shorter REM latency associated with more sleep cycles of a shorter duration in healthy humans
Psychiatry Research 104 Ž2001. 75᎐83
Shorter REM latency associated with more sleep cycles of a shorter duration in healthy humans Olivier Le Bona,U , Luc Staner b, Guy Hoffmanna , Monique Kentos a , Isidore Pelc a , Paul Linkowski c a
Sleep Center, Centre Hospitalier Uni¨ ersitaire Brugmann S48, Uni¨ ersite´ Libre de Bruxelles, Place Van Gehuchten 4, B-1020 Brussels, Belgium b Sleep Laboratory, FORENAP, Centre Hospitalier de Rouffach, Rouffach, France c Cliniques Uni¨ ersitaires Erasme, Uni¨ ersite´ Libre de Bruxelles, Brussels, Belgium Received 21 November 2000; received in revised form 9 July 2001; accepted 6 August 2001
Abstract A significant association between rapid eye movement ŽREM. sleep latency and the number of non-REMrREM sleep cycles was found 15 years ago in a large retrospective study. The present prospective study further explored this intra-sleep relationship and analyzed the links between these two variables and the mean cycle duration. It was based on a carefully selected group of healthy control subjects whose sleep was polysomnographically recorded at home for 4 sequential nights. The latency of REM sleep was inversely correlated with the number of cycles and positively correlated with the mean cycle duration, both in individual nights and on means of 4 nights. The present study demonstrated that variations in the number of cycles or the mean cycle duration between the nights are far less important than the substantial differences observed between subjects. Present outcomes support the study of sleep cycle periods and frequencies in those psychiatric disorders where REM sleep latencies have been found to be shorter, and they suggest that these variables be included in sleep studies in which cycles are compared with each other. 䊚 2001 Elsevier Science Ireland Ltd. All rights reserved. Keywords: REM sleep; REM latency; Non-REM sleep; Polysomnography
U
Corresponding author. Tel.: q32-2-477-25-54; fax: q32-2-477-25-50. E-mail address: [email protected] ŽO. Le Bon..
0165-1781r01r$ - see front matter 䊚 2001 Elsevier Science Ireland Ltd. All rights reserved. PII: S 0 1 6 5 - 1 7 8 1 Ž 0 1 . 0 0 2 9 5 - 5
76
O. Le Bon et al. r Psychiatry Research 104 (2001) 75᎐83
1. Introduction The latency of rapid eye movement ŽREM. sleep is the time elapsed between sleep onset and the first occurrence of REM sleep. This variable has received considerable attention in psychiatric research since shortened REM latency was found to be associated with depression ŽKupfer and Foster, 1972.. Subsequent studies have shown, however, that short REM latency lacks sensitivity and specificity Žsee Benca et al., 1992, for a meta-analysis .. Also, several methodological difficulties remain unresolved, such as the choice of criteria for sleep onset and REM onset, the choice of cut-offs, and the relative advantages and disadvantages of averaging outcomes or using the ‘worst night’ when consecutive recordings are performed. It is also not clear whether short REM latencies represent affective traits, states or sequelae from previous episodes Žsee Buysse and Kupfer, 1990; Le Bon, 1992, for a discussion.. Nevertheless, short REM latency remains a frequent and puzzling phenomenon in a broad spectrum of affective and anxiety disorders, and it merits our continued interest for diagnostic and clinical matters. The search for other sleep-related indicators of psychiatric disorders has been extensive. REM density was found to be associated with depression ŽFoster et al., 1976.. Lower frequencies of electroencephalographic ŽEEG. spectral power were reduced in depressive subjects in one study ŽBorbely ´ et al., 1984., but the reduction was only apparent in males in another study ŽArmitage et al., 2000.. The first cycle was reduced in comparison with consecutive ones in depressive patients ŽArmitage et al., 1992.. Higher EEG spectral power was associated with better outcome after prescription of antidepressant drugs ŽLuthringer et al., 1995.. However, further studies, notably on depression subgroups, are needed to demonstrate the clinical usefulness of REM density or spectral power analyses in psychiatry. Another way to find predictors is to search for correlations with previously established ones, as these could open new areas of exploration. Yet there is in general little redundancy between the large series of variables measured in polysomno-
graphic studies, and REM latency is apparently no exception. Only one study on a large sample of normal subjects ŽMerica and Gaillard, 1985. found a positive correlation of REM latency with REM sleep and a negative correlation with the number of cycles per night. However, this retrospective study provided limited information about subjects’ characteristics, included recordings of nights made at different moments within different protocols, involved a percentage of subjects treated with placebo, and used automatic sleep scoring without visual correction. To our knowledge, no study has yet confirmed these links. We were interested to re-examine this intrasleep relationship and to analyze the potential links between REM latency and mean cycle duration. The study was carried out in a group of carefully selected healthy subjects who were recruited prospectively for an ongoing study on healthy controls’ sleep. We tested here the associations between two definitions of REM latency and a series of sleep criteria including the number of cycles as well as two measures of mean cycle duration, in four sequential full polysomnographic sleep studies carried out at the subjects’ homes. A secondary objective was to establish the relative importance of the between-nights variation in the total variability of potential predictors, as large between-tests components would leave them little potential as diagnostic tools.
2. Methods 2.1. Subjects Eighty-four volunteers, aged 15᎐45 Žmean 27.8, S.D. 9.7, 47 females., were recruited by advertisement and paid for their participation. A comprehensive screening was carried out to ensure selection of individuals with no known existing or previous condition that might correlate with abnormal sleep. Volunteers first answered a detailed questionnaire designed to elicit sleep history by phone. Those meeting questionnaire-based criteria were then given a structured interview Žby O.L. and G.H.., using the ASDA Ž1990. criteria for sleep disorders and axis I DSM-IV ŽAmerican
O. Le Bon et al. r Psychiatry Research 104 (2001) 75᎐83
Psychiatric Association, 1994. criteria for psychiatric diagnoses Žexcept for sleep disorders.. Inclusion criteria were: regular sleep schedules; absence of sleep-related complaints or regular naps; regular weekday work schedules or no employment; no previous polysomnography; and completion of informed consent. Exclusion criteria were: DSM-IV axis I disorder; personal or first-degree familial history of affective disorder wbecause of potential influence on REM latency ŽGiles et al., 1989.x; significant somatic condition; excessive daytime sleepiness; report by significant other of periodic limb movements, snoring or sleep apnea; sleep-apnea index ŽAHI. of 5 or above on the initial night of monitoring; periodic limb movement episodes on the initial night of recording; routine consumption of more than 10 alcoholic beverages Ž10 g units. per week or consumption of illicit drugs; use of psychotropic drugs within 3 weeks before the study; and transmeridian flights or shift work within 4 weeks preceding the study. Subjects were requested not to drink alcohol for a week before entering the protocol and to change their life habits as little as possible during the time of the study. The hospital’s ethics committee approved the protocol, and subjects provided written informed consent. The study was conducted in accordance with the rules and regulations for the conduct of clinical trials stated by the World Medical Assembly ŽHelsinki, Tokyo, Venice and Hong Kong.. 2.2. Procedures Four sequential nights of sleep studies with full polysomnography were carried out in the subjects’ homes Žsleep analyzer Alice; Respironics, Pittsburgh, PA, USA.. Traditional scoring criteria were used ŽRechtschaffen and Kales, 1968.. Visual scoring was in three steps: Ža. determination of sleep stages; Žb. detection and quantification of respiratory sleep events and periodic limb movements; Žc. detection and quantification of microarousals. The inter-rater reliability was measured in another recent protocol and exceeded 0.90 for all variables ŽLe Bon et al., 1997a.. Sleep onset latency ŽSOL. was defined as the time between lights out and the first epoch of stage 2
77
sleep. Intermittent wake time represented the time spent awake after sleep onset ŽWASO.. As REM latency ŽRL. has received several operational definitions in the past ŽKnowles et al., 1982; Reynolds et al., 1983., both a more sensitive ŽRLy A. and a more specific ŽRLy B. definition were used here. RLy A was defined as the time between the first epoch of stage 2 sleep and the first epoch of stage REM sleep; RLy B was defined as the time between the first epoch of 10 min of consecutive non-REM ŽNREM. sleep not interrupted by more than 1 min of stage 1 or wake and the first epoch of 3 consecutive min of stage REM. NREMrREM sleep cycles were defined as each REM sleep episode and the NREM sleep immediately preceding it, going back to the sleep onset Ž1st NREMrREM sleep cycle. or to the limit of another REM sleep episode Žfrom the 2nd NREMrREM sleep cycle to the end of the night.. The first NREM sleep episode began with the first epoch of stage 2 and ended with the first REM sleep episode. Each REM sleep episode began with the first epoch of REM sleep and ended after the last epoch of REM sleep that was followed by at least 15 min of NREM sleep ŽFeinberg and Floyd, 1979; Merica and Gaillard, 1991.. In a modification of these criteria, no other minimal duration was, however, required for REM or NREM sleep episodes. The cycle durations were the time lengths of the NREMrREM sleep cycles, starting from sleep onset for cycle 1 and the end of the preceding cycles in all other cases. The mean cycle duration was the average duration of individual NREMrREM sleep cycles per night, thus excluding potential NREM sleep after the last cycle; total sleep time ŽTST.rnumber of cycles corresponded to the same measure but included potential residual NREM sleep. 2.3. Statistics All the variables presented here followed a normal distribution, as measured by Kolmogoroff᎐Smirnoff tests. Small degrees of skewness were found in RLy A4, RLy B3, mean cycle durations 2 and 4, TSTrnumber of cycles 2, 3 and 4 but in no case on the means over 4 nights Žsee Merica and Gaillard, 1985; Merica et al.,
78
O. Le Bon et al. r Psychiatry Research 104 (2001) 75᎐83
1997, on skewness and RL.. No usual transformation was able to normalize skewness in all cases, and the analyses were performed on actual values, so that some caution should be used in reading the results including one of these variables. Hypothesis tests were two-sided and carried out at the 5% significance level. Between-group comparisons were computed using Student’s t-test for unpaired series. Correlations were established using Pearson’s product moment. Time measures, such as bedtime and wake time, were transformed in decimal and positive variables by computing the number of minutes between them and the preceding 21.00 h. Repeated measures analyses of covariance ŽRM-ANCOVAs., with age as a covariate, were used to quantify the importance of the variability between the nights in regard to the total variability. Levene tests for homogeneity of variances were not significant, whereas Mauchly tests for abnormal sphericity were slightly significant, so that the Geisser᎐Greenhouse correction was used. The intraclass correlation coefficient was computed as: sum of square between subjectsr Žsum of square within subjects q sum of square between subjects q sum of square of the errors.4 . All statistics were computed with SPSS 9.0 ŽSPSS, Inc., Chicago, IL, USA.. Graph 1 was performed using Statview 5.0 ŽSAS Institute, GA, USA..
3. Results Eighty-four subjects responded to our advertisement Žmean age 27.8, range 15᎐45 years, S.D. 9.7, 47 females.. Results from telephone questionnaires and physician interviews were causes for exclusion of 47 individuals Žfive parasomnias, five irregular sleep schedules, seven restless legs or suspicion of periodic limb movements, 10 snoring problems, five excessive daytime sleepiness, nine anxiety disorders, six affective disorders.. First night polysomnography resulted in the exclusion of an additional six subjects Žtwo periodic limb movement and four apneicrhypopneic indices over five.. Thirty-one subjects Ž36.9%. met inclusion criteria and were considered to be normal control subjects. Data from
five subjects had to be excluded because of technical problems Žtwo 800-Mb optical diskettes seriously damaged during storage, for unknown reasons.. Twenty-six subjects Žmean age 26.7, range 15 to 45 years, S.D. 9.8, 12 females. completed all aspects of the study, and no missing polysomnographic epochs were observed. The index of sleep respiratory disorders in the final group was 2.8rh ŽS.D. 1.49., and no episodes of periodic limb movements were observed. No relationship was established between the number of cycles, mean cycle duration, TSTrnumber of cycles or RLy A and RLy B and gender or age. No difference was observed between bedtimes on the 4 nights, as was shown in another article describing the same group of patients ŽLe Bon et al., 2001.. No association could be made between bedtime or waking time and RLy A, RLy B, number of cycles, mean cycle duration and TSTrnumber of cycles Ždata not shown.. Table 1 describes the RL and cycle durations as split by the number of cycles in the nights. For this description, the 4 nights were pooled, in order to maximize the number of nights with few or many cycles Ž26 = 4 s 104 nights.. No correlation was found between mean RLy A or RLy B and the following variables as averaged over 4 nights: age; TST; SPT; SOL; NREM sleep; REM density or microarousal index. A trend was observed between REM sleep and RL y A Ž r s y0.350, Ps 0.079., but not RLy B. In contrast, both definitions of RL showed negative correlations with the number of cycles, and positive correlations with mean cycle duration and TSTrnumber of cycles. These associations were significant on individual nights, with a few exceptions on the first 2 nights, and highly significant on the means of four nights as a whole and in the male group. The links were weaker in the female group, with the link between RLy A and mean cycle duration not reaching r s 0.05 ŽTable 2.. These computations were performed using TST as a covariate, in order to eliminate potential influences of the night durations, except for the comparisons with TSTrnumber of cycles, which includes TST in its definition. Comparisons using
Number of cycles Žno..
n
REM latency A Žmin.
REM latency B Žmin.
2 3 4 5 6
6 15 46 34 3
185.2 Ž12.8. 121.9 Ž38.7. 96.4 Ž40.9. 69.0 Ž22.4. 66.5 Ž15.0.
177.3 Ž14.5. 129.1 Ž63.7. 108.7 Ž44.1. 76.8 Ž40.4. 60.6 Ž12.4.
Cycle 1 duration Žmin.
Cycle 2 duration Žmin.
Cycle 3 duration Žmin.
129.5 Ž29.2. 104.2 Ž25.7. 90.5 Ž26.4. 75.1 Ž19.8. 53.0 Ž12.6.
135.6 Ž15.2. 138.8 Ž52.9. 112.1 Ž31.9. 88.9 Ž16.7. 68.5 Ž14.7.
77.0 Ž21.9. 110.6 Ž24.9. 99.2 Ž21.1. 84.5 Ž12.1.
Cycle 4 duration Žmin.
62.8 Ž26.1. 104.6 Ž24.1. 92.8 Ž33.8.
Cycle 5 duration Žmin.
46.8 Ž20.3. 91.8 Ž16.9.
Cycle 6 duration Žmin.
Mean cycle duration Žmin.
41.5 Ž11.4.
132.5 Ž22.2. 106.6 Ž33.5. 94.0 Ž27.4. 82.9 Ž20.4. 72.0 Ž16.9.
The 4 nights were grouped for the description. RLy A : REMS latency with more sensitive definition; RLy B: REMS latency with more specific definition. Cycle duration: duration of corresponding NREMSrREMS cycle in a night with 2᎐6 cyclesrnight.
O. Le Bon et al. r Psychiatry Research 104 (2001) 75᎐83
Table 1 REM latency and cycle duration as a function of the number of cycles per night Ž n s 104.
79
O. Le Bon et al. r Psychiatry Research 104 (2001) 75᎐83
80
Table 2 Correlations of RLy A and RL y B with number of cycles, mean cycle duration and TSTrnumber of cycles Individual nights
RL y A and number of cycles RL yB and number of cycles RL y A and mean cycle duration RL y B and mean cycle duration RL y A and TSTrnumber of cycles RL y B and TSTrnumber of cycles
Means of 4 nights
N1 Ž n s 26.
N2 Ž n s 26.
N3 Ž n s 26.
N4 Ž n s 26.
Males Ž n s 14.
Females Ž n s 12.
Whole group Ž n s 26.
r s y0.646 Ps 0.000 Ž r s y0.387. Ž Ps 0.056. Ž r s 0.391. Ž Ps 0.053. r s 0.398 Ps 0.049 r s 0.716 Ps 0.000 r s 0.490 Ps 0.011
r s y0.617 Ps 0.001 r s y0.461 Ps 0.020 Ž r s 0.294. Ž Ps 0.153. r s 0.501 Ps 0.011 r s 0.579 Ps 0.002 r s 0.547 Ps 0.004
r s y0.682 Ps 0.000 r s y0.543 Ps 0.005 r s 0.607 Ps 0.001 r s 0.549 Ps 0.004 r s 0.808 Ps 0.000 r s 0.679 Ps 0.000
r s y0.793 Ps 0.000 r s y0.617 Ps 0.001 r s 0.686 Ps 0.000 r s 0.532 Ps 0.006 r s 0.818 Ps 0.000 r s 0.526 Ps 0.006
r s y0.954 P s 0.000 r s y0.754 P s y0.003 r s y0.808 P s y0.001 r s y0.708 P s y0.007 r s y0.903 P s y0.000 r s 0.748 P s 0.002
r s y0.627 Ps 0.039 r s y0.663 Ps 0.026 Ž r s 0.356. Ž P s 0.282. r s 0.550 Ž P s 0.079. r s 0.511 Ž P s 0.063. r s 0.701 Ps 0.011
r s y0.813 Ps 0.000 r s y0.701 Ps 0.000 r s 0.607 Ps 0.001 r s 0.624 Ps 0.001 r s 0.755 Ps 0.000 r s 0.719 Ps 0.000
Correlations with number of cycles and mean cycle duration were partialled for TST.
sleep period time as a covariate produced similar but slightly less significant results Ždata not shown.. When means over 4 nights were used, the correlations between number of cycles and, respectively, mean cycle duration and TSTrnumber of cycles were: r s y0.823 and r s y0.906. Fig. 1 shows the bivariate scattergram of the relationship between the mean RLy A and the mean number of cycles across the four nights. Intraclass correlations were, respectively: num-
Fig. 1. Bivariate scattergram: mean RLy A and mean number of cycles.
ber of cycles: 0.95; mean cycle duration: 0.97; TSTrnumber of cycles: 0.95; RLy A: 0.82 and RLy B: 0.81.
4. Discussion In this prospective study of four consecutive nights of polysomnographic studies recorded at home in young healthy subjects, two measures of RL were shown to be negatively associated with the number of cycles and positively with two measures of mean cycle duration. In contrast, no links were established between RLy A or RLy B and TST, SPT, SOL, NREM sleep, REM density or microarousal index, and only a trend was observed in the relationship between RLy A and REM sleep. The high intraclass correlations that were found indicate, at least over four nights and in the absence of pathological conditions, that the number and the duration of sleep cycles are robust alternative measures of REM sleep characteristics. Since TST was used as a covariate, the associations between RL and number of cycles or mean cycle duration were not merely due to differences in sleep duration, and since no associations were observed between these variables and bedtime or
O. Le Bon et al. r Psychiatry Research 104 (2001) 75᎐83
wake time, they were not due to differences in circadian phases. The description of cycle durations as split by number of cycles per night ŽTable 1. showed that the cycle durations varied in a relatively homogeneous way as a function of the number of cycles. That finding was supported by the strong associations that emerged between both of the variables related to mean cycle duration and both measures of RL. More sleep cycles of a shorter duration were thus observed when the RL was short. These results were obtained in healthy normal subjects, who are somewhat easier to study than depressive patients waiting for their treatment, and are an excellent sample for a study of the internal relationships between sleep components. The present findings should now be replicated in patients suffering from disorders in the affective spectrum, first to confirm the links observed here and second to compare affective disorders with control subjects, and diagnostic subclasses between them. In fact, one study has already shown that major depressive and dysthymic patients have more sleep cycles than controls ŽMerica et al., 1993.. Another study has also shown that the temporal coherence of ultradian rhythms was perturbed in depression ŽArmitage et al., 1999.. Periods and frequencies of sleep cycles also deserve to be analyzed in detail in cases of narcolepsy, where an ultradian hypothesis has been raised ŽNobili et al., 1996.. The reduction of RL in clinical conditions has been the subject of intense speculation and three main theories compete to explain what has been observed: Ž1. homeostatic hypotheses ŽBorbely, ´ 1982. consider short RL to appear when the NREM sleep pressure Ž‘process S’. is reduced, and NREM sleep reduction is seen here as the primum movens in affective disorders; Ž2. the reciprocal interaction hypothesis ŽHobson et al., 1975; McCarley and Massaquoi, 1992. advocates a skewing of the balance between adrenergic and cholinergic influences on brainstem cells that start and interrupt REM sleep; Ž3. circadian hypotheses associate short RL with defects in cycling regulation: free-running cycles ŽHalberg, 1968., phase advance ŽPapousek, 1975; Wehr and WirzJustice, 1982., or cycle amplitude reduction ŽSchulz and Lund, 1985.. None of these theories
81
have been supported by unequivocal evidence to date, and circadian theories have lost much of their appeal Žsee Le Bon et al., 1997b, for a discussion.. The present results suggest that the RL may merely represent an epiphenomenon of frequencies and periods of NREMrREM sleep cycles. Our findings also suggest that such variables should be introduced as covariates in sleep studies in which cycles are compared with each other. In many studies, four cycles are used: potential cycles 5 and 6 are dismissed, whereas nights with fewer cycles are not included. Nights with four cycles thus include on average longer cycles than the first four cycles of nights with six cycles. Since depressive and dysthymic patients were shown to have more cycles than controls ŽMerica et al., 1993., merely comparing the cycles and their content with those of healthy controls on a cycle-tocycle basis could be biased by the different mean cycle durations. Also, since it has been shown ŽPreud’homme et al., 2000. that the number of cycles is not correlated with the total spectral power per night, we can expect more cycles to be associated with more reduced spectral power per cycle. Confirmations of the link between depression and reduced spectral power per cycle, when compared with findings in controls, are thus needed. Regarding the choice of definitions, RLy A showed stronger associations than RLy B in most cases. Also, the variable including residual NREM sleep after the last cycle ŽTSTrnumber of cycles. provided stronger correlations with RL than the corresponding measure excluding it Žmean cycle duration. in all cases. If these definitions prove to be more heuristic, they should be favored in future comparisons. Limitations to this study may include: Ž1. Data are lacking on caffeine or tobacco consumption, which may have some impact on sleep, though it probably did not influence the general results in a significant manner. Ž2. In the protocol, subjects determined the sleep recording duration. Although this may seem more open to uncontrolled factors, it was considered important to respect the usual timing of sleep, as entraining sleep to common schedules may prove more favorable to some
82
O. Le Bon et al. r Psychiatry Research 104 (2001) 75᎐83
subjects than to others. Ž3. No data were available on the menstrual cycle in female subjects, which could prove to be important, as outlined by Armitage et al. Ž1995, 2001.. The present results showed the correlations of RL with number of cycles and mean duration variables to be stronger in the male group, which was only marginally larger than the female group. It is thus likely that factors such as menstrual cycles interfere with that relationship, even in healthy control groups, and that they should be carefully examined in further studies. In conclusion, this study re-explored and extended previous findings of strong inverse correlations between RL and the number of NREMr REM sleep cycles, and positive relationships with measures of cycle durations. These variables, which have received little attention in the literature to date, proved to be robust and deserve to be studied more extensively, especially in research on anxiety, depression, and narcolepsy disorders.
Acknowledgements The authors thank Philippe Dupont, Anita Bessemans and Marleen Bocken, for their meticulous work, constant help and availability. This work was supported by SOMALCPE ŽBrussels ., a private organization dedicated exclusively to research in psychiatry. References American Psychiatric Association, 1994. DSM-IV: Diagnostic and Statistical Manual of Mental Disorders, 4th ed. Author, Washington, DC. Armitage, R., Calhoun, J.S., Rush, A.J., Roffwarg, H.P., 1992. Comparison of the delta EEG in the first and second non-REM periods in depressed adults and normal controls. Psychiatry Research 41 Ž1., 65᎐72. Armitage, R., Hudson, A., Trivedi, M., Rush, A.J., 1995. Sex differences in the distribution of EEG frequencies during sleep: unipolar depressed outpatients. Journal of Affective Disorders 34 Ž2., 121᎐129. Armitage, R., Hoffmann, R.F., Rush, A.J., 1999. Biological rhythm disturbance in depression: temporal coherence of ultradian sleep EEG rhythms. Psychological Medicine 29 Ž6., 1435᎐1448.
Armitage, R., Hoffmann, R., Fitch, T., Trivedi, M., Rush, A.J., 2000. Temporal characteristics of delta activity during NREM sleep in depressed outpatients and healthy adults: group and sex effects. Sleep 23 Ž5., 607᎐617. Armitage, R., Emslie, G.J., Hoffmann, R.F., Rintelmann, J., Rush, A.J., 2001. Delta sleep EEG in depressed adolescent females and healthy controls. Journal of Affective Disorders 63 Ž1-3., 139᎐148. Benca, R., Obermeyer, W.H., Thisted, R.A., Gillin, J.C., 1992. Sleep and psychiatric disorders. A meta-analysis. Archives of General Psychiatry 49, 561᎐568. Borbely, ´ A.A., 1982. A two-process model of sleep regulation. Human Neurobiology 1, 195᎐204. Borbely, ´ A.A., Tobler, I., Loepfe, M., Kupfer, D.J., Ulrich, R.F., Grochocinski, V., Doman, J., Matthews, G., 1984. All-night spectral analysis of the sleep EEG in untreated depressives and normal controls. Psychiatry Research 12 Ž1., 27᎐33. Buysse, D.J., Kupfer, D.J., 1990. Diagnostic and research applications of electroencephalographic sleep studies in depression. Conceptual and methodological issues. Journal of Nervous and Mental Disease 178 Ž7., 405᎐414. Feinberg, I., Floyd, T.C., 1979. Systematic trends across the night in human sleep cycles. Psychophysiology 16 Ž3., 283᎐291. Foster, F.G., Kupfer, D.J., Coble, P., McPartland, R.J., 1976. Rapid eye movement sleep density. An objective indicator in severe medical-depressive syndromes. Archives of General Psychiatry 33 Ž9., 1119᎐1123. Giles, D., Etzel, B., Reynolds, C.F., Kupfer, D.J., 1989. Stability of polysomnographic parameters in unipolar depression: a cross-sectional report. Biological Psychiatry 27, 807᎐810. Halberg, F., 1968. Physiological considerations underlying rhythmometry with special reference to emotional illness. In: de Ajuriaguerra, J. ŽEd.., Cycles Biologiques et Psychiatrie. Masson, Paris, pp. 73᎐126. Hobson, J.A., McCarley, R., Wyzinski, P.W., 1975. Sleep cycle oscillations: reciprocal discharge by two brainstem neuronal groups. Science 189, 55᎐58. Knowles, J.B., MacLean, A.W., Cairns, J., 1982. Definitions of REM latency: some comparisons with particular reference to depression. Biological Psychiatry 17 Ž9., 993᎐1002. Kupfer, D.J., Foster, F.G., 1972. Interval between onset of sleep and rapid eye movement sleep as an indicator of depression. Lancet 2, 684. Le Bon, O., 1992. Is REM latency a dying concept? Acta Psychiatrica Belgica 92 Ž3., 131᎐150. Le Bon, O., Verbanck, P., Hoffmann, G., Murphy, J.R., Staner, L., De Groote, D., Mampunza, S., Den Dulk, A., Vacher, C., Kornreich, C., Pelc, I., 1997a. Sleep in detoxified alcoholics: impairment of most standard sleep parameters and increased risk for sleep apnea, but not for myoclonias ᎏ a controlled study. Journal of Studies on Alcohol 58 Ž1., 30᎐36. Le Bon, O., Staner, L., Murphy, J.R., Hoffmann, G., Pull, C., Pelc, I., 1997b. Critical analysis of theories advanced to
O. Le Bon et al. r Psychiatry Research 104 (2001) 75᎐83 explain short REM sleep latencies and other sleep anomalies in several psychiatric conditions. Journal of Psychiatric Research 31 Ž4., 433᎐450. Le Bon, O., Staner, L., Hoffmann, G., Dramaix, M., San Sebastian, I., Murphy, J.R., Kentos, M., Pelc, I., Linkowski, P., 2001. The first-night effect may last more than one night. Journal of Psychiatric Research 35 Ž3., 165᎐172. Luthringer, R., Minot, R., Toussaint, M., Calvi-Gries, F., Schaltenbrand, N., Macher, J.P., 1995. All-night EEG spectral analysis as a tool for the prediction of clinical response to antidepressant treatment. Biological Psychiatry 15 Ž38Ž2.., 98᎐104. McCarley, R., Massaquoi, S., 1992. Neurobiological substrates of the revised limit cycle reciprocal interaction model of REM sleep control. Journal of Sleep Research 1, 126᎐131. Merica, H., Gaillard, J.M., 1985. Statistical description and evaluation of the interrelationships of standard sleep variables for normal subjects. Sleep 8 Ž3., 261᎐273. Merica, H., Gaillard, J.M., 1991. A study of the interrupted REM episode. Physiology and Behavior 50, 1153᎐1159. Merica, H., Blois, R., Bovier, Ph., Gaillard, J.M., 1993. New variables for defining sleep continuity. Physiology and Behavior 54, 825᎐831. Merica, H., Blois, R., Fortune, R.D., Gaillard, J.M., 1997. Evolution of delta activity within the nonREM sleep
83
episode: a biphasic hypothesis. Physiology and Behavior 62 Ž1., 213᎐219. Nobili, L., Ferrillo, F., Besset, A., Rosadini, G., Schiavi, G., Billiard, M., 1996. Ultradian aspects of sleep in narcolepsy. Neurophysiologie Clinique 26 Ž1., 51᎐59. Papousek, M., 1975. Chronobiologische aspekte der zyklothymie. Fortschrift fur ¨ Neurologie und Psychiatrische Grenzgebied 43, 381᎐440. Preud’homme, X., Lanquart, J.P., Mendlewicz, J., Linkowski, P., 2000. Characteristics of spontaneous sleep with varying NREMS episodes in healthy men: implication for delta activity homeostasis. Sleep 23 Ž2., 193᎐203. Rechtschaffen, A., Kales, A., 1968. A Manual of Standardized Terminology, Techniques and Scoring System for Sleep States of Human Subjects. Superintendent of Documents, US Government Printing Office, Washington, DC. Reynolds, C.F., Taska, L.S., Jarrett, D.B., Coble, P.A., Kupfer, D.J., 1983. REM latency in depression: is there one best definition? Biological Psychiatry 18 Ž8., 849᎐863. Schulz, H., Lund, R., 1985. On the origin of early REM episodes in the sleep of depressed patients: a comparison of three hypotheses. Psychiatry Research 16, 65᎐77. Wehr, T.A., Wirz-Justice, A., 1982. Circadian rhythms mechanisms in affective illness and in antidepressant drug action. Pharmacopsychiatry 15, 31᎐39.
Shorter REM latency associated with more sleep cycles of a shorter duration in healthy humans Olivier Le Bona,U , Luc Staner b, Guy Hoffmanna , Monique Kentos a , Isidore Pelc a , Paul Linkowski c a
Sleep Center, Centre Hospitalier Uni¨ ersitaire Brugmann S48, Uni¨ ersite´ Libre de Bruxelles, Place Van Gehuchten 4, B-1020 Brussels, Belgium b Sleep Laboratory, FORENAP, Centre Hospitalier de Rouffach, Rouffach, France c Cliniques Uni¨ ersitaires Erasme, Uni¨ ersite´ Libre de Bruxelles, Brussels, Belgium Received 21 November 2000; received in revised form 9 July 2001; accepted 6 August 2001
Abstract A significant association between rapid eye movement ŽREM. sleep latency and the number of non-REMrREM sleep cycles was found 15 years ago in a large retrospective study. The present prospective study further explored this intra-sleep relationship and analyzed the links between these two variables and the mean cycle duration. It was based on a carefully selected group of healthy control subjects whose sleep was polysomnographically recorded at home for 4 sequential nights. The latency of REM sleep was inversely correlated with the number of cycles and positively correlated with the mean cycle duration, both in individual nights and on means of 4 nights. The present study demonstrated that variations in the number of cycles or the mean cycle duration between the nights are far less important than the substantial differences observed between subjects. Present outcomes support the study of sleep cycle periods and frequencies in those psychiatric disorders where REM sleep latencies have been found to be shorter, and they suggest that these variables be included in sleep studies in which cycles are compared with each other. 䊚 2001 Elsevier Science Ireland Ltd. All rights reserved. Keywords: REM sleep; REM latency; Non-REM sleep; Polysomnography
U
Corresponding author. Tel.: q32-2-477-25-54; fax: q32-2-477-25-50. E-mail address: [email protected] ŽO. Le Bon..
0165-1781r01r$ - see front matter 䊚 2001 Elsevier Science Ireland Ltd. All rights reserved. PII: S 0 1 6 5 - 1 7 8 1 Ž 0 1 . 0 0 2 9 5 - 5
76
O. Le Bon et al. r Psychiatry Research 104 (2001) 75᎐83
1. Introduction The latency of rapid eye movement ŽREM. sleep is the time elapsed between sleep onset and the first occurrence of REM sleep. This variable has received considerable attention in psychiatric research since shortened REM latency was found to be associated with depression ŽKupfer and Foster, 1972.. Subsequent studies have shown, however, that short REM latency lacks sensitivity and specificity Žsee Benca et al., 1992, for a meta-analysis .. Also, several methodological difficulties remain unresolved, such as the choice of criteria for sleep onset and REM onset, the choice of cut-offs, and the relative advantages and disadvantages of averaging outcomes or using the ‘worst night’ when consecutive recordings are performed. It is also not clear whether short REM latencies represent affective traits, states or sequelae from previous episodes Žsee Buysse and Kupfer, 1990; Le Bon, 1992, for a discussion.. Nevertheless, short REM latency remains a frequent and puzzling phenomenon in a broad spectrum of affective and anxiety disorders, and it merits our continued interest for diagnostic and clinical matters. The search for other sleep-related indicators of psychiatric disorders has been extensive. REM density was found to be associated with depression ŽFoster et al., 1976.. Lower frequencies of electroencephalographic ŽEEG. spectral power were reduced in depressive subjects in one study ŽBorbely ´ et al., 1984., but the reduction was only apparent in males in another study ŽArmitage et al., 2000.. The first cycle was reduced in comparison with consecutive ones in depressive patients ŽArmitage et al., 1992.. Higher EEG spectral power was associated with better outcome after prescription of antidepressant drugs ŽLuthringer et al., 1995.. However, further studies, notably on depression subgroups, are needed to demonstrate the clinical usefulness of REM density or spectral power analyses in psychiatry. Another way to find predictors is to search for correlations with previously established ones, as these could open new areas of exploration. Yet there is in general little redundancy between the large series of variables measured in polysomno-
graphic studies, and REM latency is apparently no exception. Only one study on a large sample of normal subjects ŽMerica and Gaillard, 1985. found a positive correlation of REM latency with REM sleep and a negative correlation with the number of cycles per night. However, this retrospective study provided limited information about subjects’ characteristics, included recordings of nights made at different moments within different protocols, involved a percentage of subjects treated with placebo, and used automatic sleep scoring without visual correction. To our knowledge, no study has yet confirmed these links. We were interested to re-examine this intrasleep relationship and to analyze the potential links between REM latency and mean cycle duration. The study was carried out in a group of carefully selected healthy subjects who were recruited prospectively for an ongoing study on healthy controls’ sleep. We tested here the associations between two definitions of REM latency and a series of sleep criteria including the number of cycles as well as two measures of mean cycle duration, in four sequential full polysomnographic sleep studies carried out at the subjects’ homes. A secondary objective was to establish the relative importance of the between-nights variation in the total variability of potential predictors, as large between-tests components would leave them little potential as diagnostic tools.
2. Methods 2.1. Subjects Eighty-four volunteers, aged 15᎐45 Žmean 27.8, S.D. 9.7, 47 females., were recruited by advertisement and paid for their participation. A comprehensive screening was carried out to ensure selection of individuals with no known existing or previous condition that might correlate with abnormal sleep. Volunteers first answered a detailed questionnaire designed to elicit sleep history by phone. Those meeting questionnaire-based criteria were then given a structured interview Žby O.L. and G.H.., using the ASDA Ž1990. criteria for sleep disorders and axis I DSM-IV ŽAmerican
O. Le Bon et al. r Psychiatry Research 104 (2001) 75᎐83
Psychiatric Association, 1994. criteria for psychiatric diagnoses Žexcept for sleep disorders.. Inclusion criteria were: regular sleep schedules; absence of sleep-related complaints or regular naps; regular weekday work schedules or no employment; no previous polysomnography; and completion of informed consent. Exclusion criteria were: DSM-IV axis I disorder; personal or first-degree familial history of affective disorder wbecause of potential influence on REM latency ŽGiles et al., 1989.x; significant somatic condition; excessive daytime sleepiness; report by significant other of periodic limb movements, snoring or sleep apnea; sleep-apnea index ŽAHI. of 5 or above on the initial night of monitoring; periodic limb movement episodes on the initial night of recording; routine consumption of more than 10 alcoholic beverages Ž10 g units. per week or consumption of illicit drugs; use of psychotropic drugs within 3 weeks before the study; and transmeridian flights or shift work within 4 weeks preceding the study. Subjects were requested not to drink alcohol for a week before entering the protocol and to change their life habits as little as possible during the time of the study. The hospital’s ethics committee approved the protocol, and subjects provided written informed consent. The study was conducted in accordance with the rules and regulations for the conduct of clinical trials stated by the World Medical Assembly ŽHelsinki, Tokyo, Venice and Hong Kong.. 2.2. Procedures Four sequential nights of sleep studies with full polysomnography were carried out in the subjects’ homes Žsleep analyzer Alice; Respironics, Pittsburgh, PA, USA.. Traditional scoring criteria were used ŽRechtschaffen and Kales, 1968.. Visual scoring was in three steps: Ža. determination of sleep stages; Žb. detection and quantification of respiratory sleep events and periodic limb movements; Žc. detection and quantification of microarousals. The inter-rater reliability was measured in another recent protocol and exceeded 0.90 for all variables ŽLe Bon et al., 1997a.. Sleep onset latency ŽSOL. was defined as the time between lights out and the first epoch of stage 2
77
sleep. Intermittent wake time represented the time spent awake after sleep onset ŽWASO.. As REM latency ŽRL. has received several operational definitions in the past ŽKnowles et al., 1982; Reynolds et al., 1983., both a more sensitive ŽRLy A. and a more specific ŽRLy B. definition were used here. RLy A was defined as the time between the first epoch of stage 2 sleep and the first epoch of stage REM sleep; RLy B was defined as the time between the first epoch of 10 min of consecutive non-REM ŽNREM. sleep not interrupted by more than 1 min of stage 1 or wake and the first epoch of 3 consecutive min of stage REM. NREMrREM sleep cycles were defined as each REM sleep episode and the NREM sleep immediately preceding it, going back to the sleep onset Ž1st NREMrREM sleep cycle. or to the limit of another REM sleep episode Žfrom the 2nd NREMrREM sleep cycle to the end of the night.. The first NREM sleep episode began with the first epoch of stage 2 and ended with the first REM sleep episode. Each REM sleep episode began with the first epoch of REM sleep and ended after the last epoch of REM sleep that was followed by at least 15 min of NREM sleep ŽFeinberg and Floyd, 1979; Merica and Gaillard, 1991.. In a modification of these criteria, no other minimal duration was, however, required for REM or NREM sleep episodes. The cycle durations were the time lengths of the NREMrREM sleep cycles, starting from sleep onset for cycle 1 and the end of the preceding cycles in all other cases. The mean cycle duration was the average duration of individual NREMrREM sleep cycles per night, thus excluding potential NREM sleep after the last cycle; total sleep time ŽTST.rnumber of cycles corresponded to the same measure but included potential residual NREM sleep. 2.3. Statistics All the variables presented here followed a normal distribution, as measured by Kolmogoroff᎐Smirnoff tests. Small degrees of skewness were found in RLy A4, RLy B3, mean cycle durations 2 and 4, TSTrnumber of cycles 2, 3 and 4 but in no case on the means over 4 nights Žsee Merica and Gaillard, 1985; Merica et al.,
78
O. Le Bon et al. r Psychiatry Research 104 (2001) 75᎐83
1997, on skewness and RL.. No usual transformation was able to normalize skewness in all cases, and the analyses were performed on actual values, so that some caution should be used in reading the results including one of these variables. Hypothesis tests were two-sided and carried out at the 5% significance level. Between-group comparisons were computed using Student’s t-test for unpaired series. Correlations were established using Pearson’s product moment. Time measures, such as bedtime and wake time, were transformed in decimal and positive variables by computing the number of minutes between them and the preceding 21.00 h. Repeated measures analyses of covariance ŽRM-ANCOVAs., with age as a covariate, were used to quantify the importance of the variability between the nights in regard to the total variability. Levene tests for homogeneity of variances were not significant, whereas Mauchly tests for abnormal sphericity were slightly significant, so that the Geisser᎐Greenhouse correction was used. The intraclass correlation coefficient was computed as: sum of square between subjectsr Žsum of square within subjects q sum of square between subjects q sum of square of the errors.4 . All statistics were computed with SPSS 9.0 ŽSPSS, Inc., Chicago, IL, USA.. Graph 1 was performed using Statview 5.0 ŽSAS Institute, GA, USA..
3. Results Eighty-four subjects responded to our advertisement Žmean age 27.8, range 15᎐45 years, S.D. 9.7, 47 females.. Results from telephone questionnaires and physician interviews were causes for exclusion of 47 individuals Žfive parasomnias, five irregular sleep schedules, seven restless legs or suspicion of periodic limb movements, 10 snoring problems, five excessive daytime sleepiness, nine anxiety disorders, six affective disorders.. First night polysomnography resulted in the exclusion of an additional six subjects Žtwo periodic limb movement and four apneicrhypopneic indices over five.. Thirty-one subjects Ž36.9%. met inclusion criteria and were considered to be normal control subjects. Data from
five subjects had to be excluded because of technical problems Žtwo 800-Mb optical diskettes seriously damaged during storage, for unknown reasons.. Twenty-six subjects Žmean age 26.7, range 15 to 45 years, S.D. 9.8, 12 females. completed all aspects of the study, and no missing polysomnographic epochs were observed. The index of sleep respiratory disorders in the final group was 2.8rh ŽS.D. 1.49., and no episodes of periodic limb movements were observed. No relationship was established between the number of cycles, mean cycle duration, TSTrnumber of cycles or RLy A and RLy B and gender or age. No difference was observed between bedtimes on the 4 nights, as was shown in another article describing the same group of patients ŽLe Bon et al., 2001.. No association could be made between bedtime or waking time and RLy A, RLy B, number of cycles, mean cycle duration and TSTrnumber of cycles Ždata not shown.. Table 1 describes the RL and cycle durations as split by the number of cycles in the nights. For this description, the 4 nights were pooled, in order to maximize the number of nights with few or many cycles Ž26 = 4 s 104 nights.. No correlation was found between mean RLy A or RLy B and the following variables as averaged over 4 nights: age; TST; SPT; SOL; NREM sleep; REM density or microarousal index. A trend was observed between REM sleep and RL y A Ž r s y0.350, Ps 0.079., but not RLy B. In contrast, both definitions of RL showed negative correlations with the number of cycles, and positive correlations with mean cycle duration and TSTrnumber of cycles. These associations were significant on individual nights, with a few exceptions on the first 2 nights, and highly significant on the means of four nights as a whole and in the male group. The links were weaker in the female group, with the link between RLy A and mean cycle duration not reaching r s 0.05 ŽTable 2.. These computations were performed using TST as a covariate, in order to eliminate potential influences of the night durations, except for the comparisons with TSTrnumber of cycles, which includes TST in its definition. Comparisons using
Number of cycles Žno..
n
REM latency A Žmin.
REM latency B Žmin.
2 3 4 5 6
6 15 46 34 3
185.2 Ž12.8. 121.9 Ž38.7. 96.4 Ž40.9. 69.0 Ž22.4. 66.5 Ž15.0.
177.3 Ž14.5. 129.1 Ž63.7. 108.7 Ž44.1. 76.8 Ž40.4. 60.6 Ž12.4.
Cycle 1 duration Žmin.
Cycle 2 duration Žmin.
Cycle 3 duration Žmin.
129.5 Ž29.2. 104.2 Ž25.7. 90.5 Ž26.4. 75.1 Ž19.8. 53.0 Ž12.6.
135.6 Ž15.2. 138.8 Ž52.9. 112.1 Ž31.9. 88.9 Ž16.7. 68.5 Ž14.7.
77.0 Ž21.9. 110.6 Ž24.9. 99.2 Ž21.1. 84.5 Ž12.1.
Cycle 4 duration Žmin.
62.8 Ž26.1. 104.6 Ž24.1. 92.8 Ž33.8.
Cycle 5 duration Žmin.
46.8 Ž20.3. 91.8 Ž16.9.
Cycle 6 duration Žmin.
Mean cycle duration Žmin.
41.5 Ž11.4.
132.5 Ž22.2. 106.6 Ž33.5. 94.0 Ž27.4. 82.9 Ž20.4. 72.0 Ž16.9.
The 4 nights were grouped for the description. RLy A : REMS latency with more sensitive definition; RLy B: REMS latency with more specific definition. Cycle duration: duration of corresponding NREMSrREMS cycle in a night with 2᎐6 cyclesrnight.
O. Le Bon et al. r Psychiatry Research 104 (2001) 75᎐83
Table 1 REM latency and cycle duration as a function of the number of cycles per night Ž n s 104.
79
O. Le Bon et al. r Psychiatry Research 104 (2001) 75᎐83
80
Table 2 Correlations of RLy A and RL y B with number of cycles, mean cycle duration and TSTrnumber of cycles Individual nights
RL y A and number of cycles RL yB and number of cycles RL y A and mean cycle duration RL y B and mean cycle duration RL y A and TSTrnumber of cycles RL y B and TSTrnumber of cycles
Means of 4 nights
N1 Ž n s 26.
N2 Ž n s 26.
N3 Ž n s 26.
N4 Ž n s 26.
Males Ž n s 14.
Females Ž n s 12.
Whole group Ž n s 26.
r s y0.646 Ps 0.000 Ž r s y0.387. Ž Ps 0.056. Ž r s 0.391. Ž Ps 0.053. r s 0.398 Ps 0.049 r s 0.716 Ps 0.000 r s 0.490 Ps 0.011
r s y0.617 Ps 0.001 r s y0.461 Ps 0.020 Ž r s 0.294. Ž Ps 0.153. r s 0.501 Ps 0.011 r s 0.579 Ps 0.002 r s 0.547 Ps 0.004
r s y0.682 Ps 0.000 r s y0.543 Ps 0.005 r s 0.607 Ps 0.001 r s 0.549 Ps 0.004 r s 0.808 Ps 0.000 r s 0.679 Ps 0.000
r s y0.793 Ps 0.000 r s y0.617 Ps 0.001 r s 0.686 Ps 0.000 r s 0.532 Ps 0.006 r s 0.818 Ps 0.000 r s 0.526 Ps 0.006
r s y0.954 P s 0.000 r s y0.754 P s y0.003 r s y0.808 P s y0.001 r s y0.708 P s y0.007 r s y0.903 P s y0.000 r s 0.748 P s 0.002
r s y0.627 Ps 0.039 r s y0.663 Ps 0.026 Ž r s 0.356. Ž P s 0.282. r s 0.550 Ž P s 0.079. r s 0.511 Ž P s 0.063. r s 0.701 Ps 0.011
r s y0.813 Ps 0.000 r s y0.701 Ps 0.000 r s 0.607 Ps 0.001 r s 0.624 Ps 0.001 r s 0.755 Ps 0.000 r s 0.719 Ps 0.000
Correlations with number of cycles and mean cycle duration were partialled for TST.
sleep period time as a covariate produced similar but slightly less significant results Ždata not shown.. When means over 4 nights were used, the correlations between number of cycles and, respectively, mean cycle duration and TSTrnumber of cycles were: r s y0.823 and r s y0.906. Fig. 1 shows the bivariate scattergram of the relationship between the mean RLy A and the mean number of cycles across the four nights. Intraclass correlations were, respectively: num-
Fig. 1. Bivariate scattergram: mean RLy A and mean number of cycles.
ber of cycles: 0.95; mean cycle duration: 0.97; TSTrnumber of cycles: 0.95; RLy A: 0.82 and RLy B: 0.81.
4. Discussion In this prospective study of four consecutive nights of polysomnographic studies recorded at home in young healthy subjects, two measures of RL were shown to be negatively associated with the number of cycles and positively with two measures of mean cycle duration. In contrast, no links were established between RLy A or RLy B and TST, SPT, SOL, NREM sleep, REM density or microarousal index, and only a trend was observed in the relationship between RLy A and REM sleep. The high intraclass correlations that were found indicate, at least over four nights and in the absence of pathological conditions, that the number and the duration of sleep cycles are robust alternative measures of REM sleep characteristics. Since TST was used as a covariate, the associations between RL and number of cycles or mean cycle duration were not merely due to differences in sleep duration, and since no associations were observed between these variables and bedtime or
O. Le Bon et al. r Psychiatry Research 104 (2001) 75᎐83
wake time, they were not due to differences in circadian phases. The description of cycle durations as split by number of cycles per night ŽTable 1. showed that the cycle durations varied in a relatively homogeneous way as a function of the number of cycles. That finding was supported by the strong associations that emerged between both of the variables related to mean cycle duration and both measures of RL. More sleep cycles of a shorter duration were thus observed when the RL was short. These results were obtained in healthy normal subjects, who are somewhat easier to study than depressive patients waiting for their treatment, and are an excellent sample for a study of the internal relationships between sleep components. The present findings should now be replicated in patients suffering from disorders in the affective spectrum, first to confirm the links observed here and second to compare affective disorders with control subjects, and diagnostic subclasses between them. In fact, one study has already shown that major depressive and dysthymic patients have more sleep cycles than controls ŽMerica et al., 1993.. Another study has also shown that the temporal coherence of ultradian rhythms was perturbed in depression ŽArmitage et al., 1999.. Periods and frequencies of sleep cycles also deserve to be analyzed in detail in cases of narcolepsy, where an ultradian hypothesis has been raised ŽNobili et al., 1996.. The reduction of RL in clinical conditions has been the subject of intense speculation and three main theories compete to explain what has been observed: Ž1. homeostatic hypotheses ŽBorbely, ´ 1982. consider short RL to appear when the NREM sleep pressure Ž‘process S’. is reduced, and NREM sleep reduction is seen here as the primum movens in affective disorders; Ž2. the reciprocal interaction hypothesis ŽHobson et al., 1975; McCarley and Massaquoi, 1992. advocates a skewing of the balance between adrenergic and cholinergic influences on brainstem cells that start and interrupt REM sleep; Ž3. circadian hypotheses associate short RL with defects in cycling regulation: free-running cycles ŽHalberg, 1968., phase advance ŽPapousek, 1975; Wehr and WirzJustice, 1982., or cycle amplitude reduction ŽSchulz and Lund, 1985.. None of these theories
81
have been supported by unequivocal evidence to date, and circadian theories have lost much of their appeal Žsee Le Bon et al., 1997b, for a discussion.. The present results suggest that the RL may merely represent an epiphenomenon of frequencies and periods of NREMrREM sleep cycles. Our findings also suggest that such variables should be introduced as covariates in sleep studies in which cycles are compared with each other. In many studies, four cycles are used: potential cycles 5 and 6 are dismissed, whereas nights with fewer cycles are not included. Nights with four cycles thus include on average longer cycles than the first four cycles of nights with six cycles. Since depressive and dysthymic patients were shown to have more cycles than controls ŽMerica et al., 1993., merely comparing the cycles and their content with those of healthy controls on a cycle-tocycle basis could be biased by the different mean cycle durations. Also, since it has been shown ŽPreud’homme et al., 2000. that the number of cycles is not correlated with the total spectral power per night, we can expect more cycles to be associated with more reduced spectral power per cycle. Confirmations of the link between depression and reduced spectral power per cycle, when compared with findings in controls, are thus needed. Regarding the choice of definitions, RLy A showed stronger associations than RLy B in most cases. Also, the variable including residual NREM sleep after the last cycle ŽTSTrnumber of cycles. provided stronger correlations with RL than the corresponding measure excluding it Žmean cycle duration. in all cases. If these definitions prove to be more heuristic, they should be favored in future comparisons. Limitations to this study may include: Ž1. Data are lacking on caffeine or tobacco consumption, which may have some impact on sleep, though it probably did not influence the general results in a significant manner. Ž2. In the protocol, subjects determined the sleep recording duration. Although this may seem more open to uncontrolled factors, it was considered important to respect the usual timing of sleep, as entraining sleep to common schedules may prove more favorable to some
82
O. Le Bon et al. r Psychiatry Research 104 (2001) 75᎐83
subjects than to others. Ž3. No data were available on the menstrual cycle in female subjects, which could prove to be important, as outlined by Armitage et al. Ž1995, 2001.. The present results showed the correlations of RL with number of cycles and mean duration variables to be stronger in the male group, which was only marginally larger than the female group. It is thus likely that factors such as menstrual cycles interfere with that relationship, even in healthy control groups, and that they should be carefully examined in further studies. In conclusion, this study re-explored and extended previous findings of strong inverse correlations between RL and the number of NREMr REM sleep cycles, and positive relationships with measures of cycle durations. These variables, which have received little attention in the literature to date, proved to be robust and deserve to be studied more extensively, especially in research on anxiety, depression, and narcolepsy disorders.
Acknowledgements The authors thank Philippe Dupont, Anita Bessemans and Marleen Bocken, for their meticulous work, constant help and availability. This work was supported by SOMALCPE ŽBrussels ., a private organization dedicated exclusively to research in psychiatry. References American Psychiatric Association, 1994. DSM-IV: Diagnostic and Statistical Manual of Mental Disorders, 4th ed. Author, Washington, DC. Armitage, R., Calhoun, J.S., Rush, A.J., Roffwarg, H.P., 1992. Comparison of the delta EEG in the first and second non-REM periods in depressed adults and normal controls. Psychiatry Research 41 Ž1., 65᎐72. Armitage, R., Hudson, A., Trivedi, M., Rush, A.J., 1995. Sex differences in the distribution of EEG frequencies during sleep: unipolar depressed outpatients. Journal of Affective Disorders 34 Ž2., 121᎐129. Armitage, R., Hoffmann, R.F., Rush, A.J., 1999. Biological rhythm disturbance in depression: temporal coherence of ultradian sleep EEG rhythms. Psychological Medicine 29 Ž6., 1435᎐1448.
Armitage, R., Hoffmann, R., Fitch, T., Trivedi, M., Rush, A.J., 2000. Temporal characteristics of delta activity during NREM sleep in depressed outpatients and healthy adults: group and sex effects. Sleep 23 Ž5., 607᎐617. Armitage, R., Emslie, G.J., Hoffmann, R.F., Rintelmann, J., Rush, A.J., 2001. Delta sleep EEG in depressed adolescent females and healthy controls. Journal of Affective Disorders 63 Ž1-3., 139᎐148. Benca, R., Obermeyer, W.H., Thisted, R.A., Gillin, J.C., 1992. Sleep and psychiatric disorders. A meta-analysis. Archives of General Psychiatry 49, 561᎐568. Borbely, ´ A.A., 1982. A two-process model of sleep regulation. Human Neurobiology 1, 195᎐204. Borbely, ´ A.A., Tobler, I., Loepfe, M., Kupfer, D.J., Ulrich, R.F., Grochocinski, V., Doman, J., Matthews, G., 1984. All-night spectral analysis of the sleep EEG in untreated depressives and normal controls. Psychiatry Research 12 Ž1., 27᎐33. Buysse, D.J., Kupfer, D.J., 1990. Diagnostic and research applications of electroencephalographic sleep studies in depression. Conceptual and methodological issues. Journal of Nervous and Mental Disease 178 Ž7., 405᎐414. Feinberg, I., Floyd, T.C., 1979. Systematic trends across the night in human sleep cycles. Psychophysiology 16 Ž3., 283᎐291. Foster, F.G., Kupfer, D.J., Coble, P., McPartland, R.J., 1976. Rapid eye movement sleep density. An objective indicator in severe medical-depressive syndromes. Archives of General Psychiatry 33 Ž9., 1119᎐1123. Giles, D., Etzel, B., Reynolds, C.F., Kupfer, D.J., 1989. Stability of polysomnographic parameters in unipolar depression: a cross-sectional report. Biological Psychiatry 27, 807᎐810. Halberg, F., 1968. Physiological considerations underlying rhythmometry with special reference to emotional illness. In: de Ajuriaguerra, J. ŽEd.., Cycles Biologiques et Psychiatrie. Masson, Paris, pp. 73᎐126. Hobson, J.A., McCarley, R., Wyzinski, P.W., 1975. Sleep cycle oscillations: reciprocal discharge by two brainstem neuronal groups. Science 189, 55᎐58. Knowles, J.B., MacLean, A.W., Cairns, J., 1982. Definitions of REM latency: some comparisons with particular reference to depression. Biological Psychiatry 17 Ž9., 993᎐1002. Kupfer, D.J., Foster, F.G., 1972. Interval between onset of sleep and rapid eye movement sleep as an indicator of depression. Lancet 2, 684. Le Bon, O., 1992. Is REM latency a dying concept? Acta Psychiatrica Belgica 92 Ž3., 131᎐150. Le Bon, O., Verbanck, P., Hoffmann, G., Murphy, J.R., Staner, L., De Groote, D., Mampunza, S., Den Dulk, A., Vacher, C., Kornreich, C., Pelc, I., 1997a. Sleep in detoxified alcoholics: impairment of most standard sleep parameters and increased risk for sleep apnea, but not for myoclonias ᎏ a controlled study. Journal of Studies on Alcohol 58 Ž1., 30᎐36. Le Bon, O., Staner, L., Murphy, J.R., Hoffmann, G., Pull, C., Pelc, I., 1997b. Critical analysis of theories advanced to
O. Le Bon et al. r Psychiatry Research 104 (2001) 75᎐83 explain short REM sleep latencies and other sleep anomalies in several psychiatric conditions. Journal of Psychiatric Research 31 Ž4., 433᎐450. Le Bon, O., Staner, L., Hoffmann, G., Dramaix, M., San Sebastian, I., Murphy, J.R., Kentos, M., Pelc, I., Linkowski, P., 2001. The first-night effect may last more than one night. Journal of Psychiatric Research 35 Ž3., 165᎐172. Luthringer, R., Minot, R., Toussaint, M., Calvi-Gries, F., Schaltenbrand, N., Macher, J.P., 1995. All-night EEG spectral analysis as a tool for the prediction of clinical response to antidepressant treatment. Biological Psychiatry 15 Ž38Ž2.., 98᎐104. McCarley, R., Massaquoi, S., 1992. Neurobiological substrates of the revised limit cycle reciprocal interaction model of REM sleep control. Journal of Sleep Research 1, 126᎐131. Merica, H., Gaillard, J.M., 1985. Statistical description and evaluation of the interrelationships of standard sleep variables for normal subjects. Sleep 8 Ž3., 261᎐273. Merica, H., Gaillard, J.M., 1991. A study of the interrupted REM episode. Physiology and Behavior 50, 1153᎐1159. Merica, H., Blois, R., Bovier, Ph., Gaillard, J.M., 1993. New variables for defining sleep continuity. Physiology and Behavior 54, 825᎐831. Merica, H., Blois, R., Fortune, R.D., Gaillard, J.M., 1997. Evolution of delta activity within the nonREM sleep
83
episode: a biphasic hypothesis. Physiology and Behavior 62 Ž1., 213᎐219. Nobili, L., Ferrillo, F., Besset, A., Rosadini, G., Schiavi, G., Billiard, M., 1996. Ultradian aspects of sleep in narcolepsy. Neurophysiologie Clinique 26 Ž1., 51᎐59. Papousek, M., 1975. Chronobiologische aspekte der zyklothymie. Fortschrift fur ¨ Neurologie und Psychiatrische Grenzgebied 43, 381᎐440. Preud’homme, X., Lanquart, J.P., Mendlewicz, J., Linkowski, P., 2000. Characteristics of spontaneous sleep with varying NREMS episodes in healthy men: implication for delta activity homeostasis. Sleep 23 Ž2., 193᎐203. Rechtschaffen, A., Kales, A., 1968. A Manual of Standardized Terminology, Techniques and Scoring System for Sleep States of Human Subjects. Superintendent of Documents, US Government Printing Office, Washington, DC. Reynolds, C.F., Taska, L.S., Jarrett, D.B., Coble, P.A., Kupfer, D.J., 1983. REM latency in depression: is there one best definition? Biological Psychiatry 18 Ž8., 849᎐863. Schulz, H., Lund, R., 1985. On the origin of early REM episodes in the sleep of depressed patients: a comparison of three hypotheses. Psychiatry Research 16, 65᎐77. Wehr, T.A., Wirz-Justice, A., 1982. Circadian rhythms mechanisms in affective illness and in antidepressant drug action. Pharmacopsychiatry 15, 31᎐39.