delta sleep ratio as a predictor of sleep deprivation response in major depression

Delta sleep ratio as a predictor of sleepdeprivation response in major depression

Christoph Nissena, Bernd Feigea, Almut Konigb,Ulrich Voderholzera, Mathias Bergera, Dieter Riemanna,*

aDepartment of Psychiatry and Psychotherapy of the University Hospital of Freiburg, Hauptstrasse 5, D-79104 Freiburg, GermanybDepartment of Psychiatry, Offenburg, Germany

Received 20 November 2000; received in revised form 3 April 2001; accepted 30 April 2001

Abstract

The fast but short-lasting improvement of depressive symptoms by sleep deprivation (SD) in about 60% of patients with a majordepressive disorder is well established, but the mechanisms of action are still not clear. Recent studies suggest that changes in nonrapid eye movement (NREM) sleep, especially in slow wave activity (SWA), could be associated with the therapeutic outcome ofSD. In the current study, spectral analysis of NREM sleep EEG directly prior to SD was performed to determine if automatically

derived sleep parameters predict SD response. Sixteen pair matched and drug free patients with a major depressive disorder, 8 SDresponders and 8 non-responders (response criterion: 50% reduction on the 6-item HAMD score), were included. Average EEGspectral power was calculated for the whole night before SD and for single NREM episodes. While whole-night averages of spectral

power did not differ significantly between subgroups, SD responders showed a steady decrease of SWA across successive NREMepisodes, whereas in non-responders an increase from the first to the second episode was observed. The different distribution ofSWA was significantly expressed in the delta sleep ratio (quotient of SWA in the first to the second NREM episode). In conclusion,

a high delta sleep ratio is a positive predictor for SD response. Referred to psycho- and pharmacotherapeutic results it is hypo-thesized that low and high values of the delta sleep ratio characterize subgroups of depressed patients with different neurobiologicalalterations, which could be relevant for further scientific and therapeutic approaches. # 2001 Elsevier Science Ltd. All rights

reserved.

Keywords: Electroencephalography; Spectral analysis; Sleep deprivation; Major depressive disorder; Non rapid eye movement sleep; Slow wave sleep

1. Introduction

Therapeutic sleep deprivation (SD) is an effectivenonpharmacological method of treatment for depressedpatients: approximately 60% of patients with majordepressive disorder experience an acute beneficialresponse (SD responders), whereas 40% do not (SDnon-responders). Unfortunately 50–80% of drug-freeSD responders relapse into depression following con-secutive nocturnal sleep (overview see Wu and Bunney,1990). Even brief daytime sleep periods can reverse thetherapeutic effect (Wiegand et al., 1987; 1993; Riemannet al., 1993; Hemmeter et al., 1998).

With regard to the clinical utility of sleep deprivation,several studies have focused on the advantages of treat-ment variants like total or partial sleep deprivation orselective REM deprivation and possibilities to stabilizethe therapeutic effect, including additional anti-depressant medication and light therapy (overview seevan den Hoofdakker, 1997; Wirz-Justice and van denHoofdakker, 1999). In our department it was shownthat a sleep phase advance design following SD stabi-lized 60 to 75% of sleep deprivation responders for atleast one week (Vollmann and Berger, 1993; Berger etal., 1997; Riemann et al., 1999). This finding was alsoconfirmed in an independent study (Albert et al., 1998).

Besides clinical issues, SD offers an interesting para-digm to study biological processes during intensivemood changes in short time-spans. The elucidation ofthe underlying biological mechanism(s) could gainnew perspectives for understanding and treatment of

0022-3956/01/$ - see front matter # 2001 Elsevier Science Ltd. All rights reserved.

PI I : S0022-3956(01 )00021-8

Journal of Psychiatric Research 35 (2001) 155–163

www.elsevier.com/locate/jpsychires

* Corresponding author. Tel.: +49-761-270-6919; fax: +49-761-

270-6523.

E-mail address: [email protected]

(D. Riemann).

affective illnesses. One possible strategy to shed light onthese mechanisms is to identify biological parameterswhich differentiate SD responders from non-responders.

Summarizing previous studies, some patient char-acteristics seem to be of major importance for theresponsiveness to SD: from a clinical point of view, themost consistently positive predictors of SD are a more‘‘melancholic’’ type of depression, a high level of vigi-lance and the propensity to produce diurnal mood var-iations (see cf. Kuhs and Tolle, 1991; Gordijn et al.,1994, 1995). Other variables, such as sex, age, age atfirst onset of illness, duration or severity of the acuteepisode have not consistently been related to respon-siveness (for overview see Kuhs and Tolle, 1991).

Four studies using the functional imaging techniquesSPECT (single photon emission computer tomography:Ebert et al., 1991; 1994; Volk et al., 1992; 1997) and fourPET (positron emission tomography) studies (Wu et al.,1992, 1994, 1999; Smith et al., 1999) have shown anenhanced glucose metabolism in limbic areas of thebrain at baseline in responders, but not in non-respon-ders to SD. Consistently, SPECT as well as PET studiesdemonstrated significant decreases of metabolism in theventral inferior parts of the cingulate only after suc-cessful SD.

With regard to polysomnographic sleep research, ashort REM latency has previously been described as apositive predictor (Riemann et al., 1991) with a strongertrend for normalization in SD responders compared tonon-responders (Riemann and Berger, 1990). However,other studies did not report a predictive value of REMlatency for SD response (Gerner et al., 1979; Duncan etal., 1980).

Spectral parameters as a more physiological anddynamic approach to the sleep EEG have been descri-bed as predictors for pharmaco- and psychotherapeutictreatment outcome in depressed patients (cf. Frank etal., 1984; Kupfer et al., 1984; 1990; 1993; Jarrett et al.,1990; Thase et al., 1998). Comparable studies concern-ing the outcome of therapeutic sleep deprivation have notbeen published yet and will be presented in this paper.

For three reasons the electroencephalographic activityof the slowest frequencies [Slow Wave Activity (SWA)]of NonREM sleep (NREM) is of special interestregarding the close relationship between sleep anddepression: First, the SWA of NREM sleep is reducedin depressed patients relative to controls (cf. Kupfer etal., 1984; Armitage, 1995). Second, SWA has beenshown to increase as a function of prior waking time(Borbely, 1982). Third, according to previous studies,changes in NREM sleep, especially in SWA, — ratherthan changes in REM sleep — may be associated withthe clinical effect of SD (Beersma and Van den Hof-dakker, 1992).

In addition to the average SWA across NREM sleepof a whole night, another focus of research in recent

years has been the distribution of SWA across succes-sive NREM episodes, which appears to be changed indepressed patients: unlike healthy subjects, who typi-cally demonstrate a steady decrease of SWA throughoutthe night (Borbely, 1982), depressed patients displayedlower SWA during the first NREM episode comparedto the second one (Kupfer et al., 1984; 1986).

Recently, the distribution of SWA across the first twoNREM sleep episodes, expressed as the ratio of SWA inthe first to the second NREM episode (termed deltasleep ratio), has been shown to be relevant for clinicaloutcome of pharmaco- (Ehlers et al., 1996) and psy-chotherapy in depressed patients (Kupfer et al., 1990;Buysse et al., 1992 a,b; Thase et al., 1998). In thoseinvestigations a higher delta sleep ratio was associatedto a more favourable therapeutic outcome, i.e. longerclinical remission or better response. Data for deltasleep ratio concerning the outcome of sleep deprivationtherapy have not been published yet.

In the present study, spectral analysis of the sleepEEG in drug free inpatients with major depressive dis-order was carried out the night before therapeutic sleepdeprivation. We hypothesized that SD responderswould differ from non-responders with respect to auto-matically computed sleep parameters, like delta sleepratio.

2. Subjects and methods

2.1. Patient sample

Sixteen hospitalized patients with a primary majordepressive disorder without acute suicidal or psychoticfeatures according to DSM-III-R-criteria (AmericanPsychiatric Association, 1987) were included in the pre-sent analysis. The analysis was part of a clinical andpolysomnographical study on the stabilization of theantidepressant effect of SD by shifting the sleep phase(advance vs. delay) after SD response (complete studypublished elsewhere: Riemann et al., 1999)

All patients included were free of any psychoactivemedication for at least 7 days prior to the beginning ofthe study (fluoxetine and neuroleptics: 21 days). Beforeentering the protocol, all participants underwent anextensive medical examination. Subjects with any ser-ious somatic disorder were excluded.

Of the 57 patients who took part in a polysomno-graphic study with at least an adaptation and a baselinenight in the sleep laboratory and a trial of therapeuticsleep deprivation, n=16 were retained for the presentspectral data analysis. This reduction of N was due tothe following methodical criteria: limitation to a singletype of signal recording hardware, exclusion of patientsshowing artifacts in spectral analysis, strict SD responsecriteria and precise matching (see below).

156 C. Nissen et al. / Journal of Psychiatric Research 35 (2001) 155–163

Because the sleep EEG spectral power has beenshown to depend on age and sex (Astrom & Trojaborg,1992; Landoldt et al., 1996; Ehlers & Kupfer, 1997), wecarefully pair-matched the patient groups (respondersvs. non-responders) with respect to these parameters.The resulting close correspondence of demographicaland clinical characteristics of our samples is shown inTable 1.

All patients were informed in detail about the experi-mental procedures and gave their informed writtenconsent. The study had been approved by the local ethiccommittee.

2.2. Experimental design

All patients slept 2 nights in the sleep laboratory priorto SD. After an adaptation night, a baseline night wasrecorded for which the spectral analysis is presented inthis paper. Before adaptation, habitual bed times wereensured by the inpatient conditions (about 23:00 to06:00–07:00). During the study, bed times were strictlycontrolled under sleep laboratory conditions from 23:00to 06:00. The sleep recordings after SD, following asleep phase advance/delay design, were also investigatedpolysomnographically; the results for the conventionalRechtschaffen and Kales (1968) scored parameters forthe whole sample are presented elsewhere (Riemann etal. 1999).

To measure psychopathology the 21-item HAMDwas performed prior to the first investigation in thesleep laboratory and repeated the day before SD.Patients were required to have a score 5 18 to be eligi-ble. The 6-item HAMD, an abbreviated version of the21-item HAMD, is a suitable instrument for repeatedmeasurements (Bech et al., 1975) and was used tomeasure depressive mood daily throughout the study at09:00 and 16:00. For the present analysis, patientsshowing at least a 50% improvement of the 6-HAMDvalues (averaged 09:00/16:00 h values) comparing theday after the sleepless night to the day directly beforewere defined as SD responders, patients showing animprovement <50% as SD non-responders. The chosen50% response criterion guarantees a clear improvement

of depressive symptoms by SD. In responders, averagedecrease on the 6-HAMD was 74.8% (paired samples t-test: P=0.000), whereas in non-responders no sig-nificant decrease of the 6-HAMD occurred (12.9%;P=0.174).

2.3. Sleep recordings and spectral analysis

Sleep EEG recordings were performed and scoredaccording to standard procedures. During the night,continuous EEG (C3 referenced to linked mastoids) wasamplified with a time constant of 0.3 s and a lowpass at70 Hz, digitized at 102.4 Hz and stored on disk. Due tothe sampling theorem, this led to an aliasing of fre-quencies above 51.2 Hz into the frequency rangebetween 32.4 and 51.2 Hz. Only frequencies up to 25 Hzwere analyzed in this study. The sleep polysomno-graphic data were scored off-line in epochs of 30 s byexperienced raters according to the criteria ofRechtschaffen and Kales (1968). REM latency wasdefined as the time interval from sleep onset (firstoccurrence of NREM stage 2,3,4 or REM sleep) untilthe first occurrence of REM sleep. Two variants wereused: the ‘‘lenient’’ definition included no furtherrequirement, while the ‘‘strict’’ definition required thefirst occurrence of REM sleep to last for at least 3 con-secutive minutes.

An all-night spectral analysis was performed on thesame 30 s epochs for which sleep stages had been deter-mined (for further details see Feige et al., 1999). Spectralestimates for each epoch were obtained as the averageof 46 128-point FFT windows overlapping by half,resulting in a spectral resolution of 0.8 Hz. The spectralpower values were then log-transformed (base e) andcontinuously stored on disk. All subsequent analysissteps including statistical analysis were performed onthese logarithmic values, which have more symme-trically distributed errors than raw spectral power(Blackman & Tukey, 1958; Feige, 1999).

The goal of the further analysis was to minimize theeffects of confounding variables on the spectra averagedacross epochs, such as the number of movements orarousals and other sleep parameters that can be

Table 1

Demographic and clinical characteristics of investigated samples (mean�standard deviation)

S.D.

responders

S.D. non-

responders

Significance of

the differencea

n 8 8 –

Male/female (No.) 5/ 3 5/ 3 –

Age (years) 37.8 (� 11.5) 37.1 (� 13.8) 0.923

Age at first depressive episode (years) 28.9 (� 8.4) 29.9 (� 14.1) 0.866

Duration of affective illness (years) 8.9 (� 11.2) 7.3 (� 7.2) 0.735

Duration of acute episode (weeks) 40.9 (� 96.3) 35.4 (� 34.6) 0.881

Severity of acute episode (21-Hamilton rating depression score at baseline) 24.9 (� 5.4) 24.9 (� 3.7) 1.000

a t-Test for independent samples.

C. Nissen et al. / Journal of Psychiatric Research 35 (2001) 155–163 157

analyzed separately. This was done by excluding deviantepochs from the average. Deviant epochs were thosecontaining movements or arousals as determined duringstaging; furthermore, the total (0.8–48 Hz) and gamma-band (32–48 Hz) power of each epoch was related to thecorresponding median-filtered value (the median ofvalues in the 5 min preceding and 5 min following theepoch) and an epoch was excluded if the deviation waslarger than the difference between the median and thefirst quartile of all median-filtered values across thenight. In this way, artifacts mainly restricted to low fre-quencies (such as EOG events) as well as those occur-ring mainly in higher frequencies (such as EMGcontamination) were eliminated in a data-driven way.The log spectra for accepted epochs rated as NREMsleep were averaged across each complete night and alsoseparately for each NREM episode. The first NREM

episode was defined as NREM sleep from sleep onsetuntil the first 5 6 epochs of REM sleep, the followingNREM episodes as the NREM sleep between the con-secutive REM episodes. Spectral band power was cal-culated for the following frequency ranges: delta (0.1–3.5 Hz), theta (3.5–8 Hz), alpha (8–12 Hz), sigma (12–16 Hz), beta 1 (16–20 Hz) and beta2 (20–24 Hz). Totalspectral power (0.1–24 Hz) represents the sum of thepower of all included frequency ranges, relative spectralpower the power of a frequency range referred to thetotal spectral power.

2.4. Statistics

For descriptive purposes, means and standard devia-tions (SDs) were calculated. To compare the twogroups, student’s t-test for independent samples and a

Fig. 1. (a) Absolute spectral power in S.D. responders and non-responders. (b) Relative spectral power in S.D. responders and non-responders.


2-factorial ANOVA with factors NREM episode andresponse to sleep deprivation was used. The level ofsignificance was set at P<0.05 (two-tailed).

3. Results

As can be seen in Fig. 1a,b, SD responders and SDnon-responders demonstrated no significant differencesin averaged absolute and relative spectral power thenight before total sleep deprivation.

Regarding the distribution of SWA across NREMsleep episodes (Fig. 2), SD responders showed a steadydecrease while in non-responders an increase in the sec-ond NREM episode was observed, although these dif-ferences were not statistically significant using t-testsand a 2-factorial ANOVA (factors: NREM episode andresponse status; F(5; 64)=5.6; significance of responsestatus: P=0.055). The amount of SWA at the beginningand at the end of the night in both groups was remark-ably similar.

The different distribution of SWA is reflected by theratio of SWA in the first to the second NREM episode(Fig. 3): SD responders showed a significantly higherdelta sleep ratio than SD non-responders under baselinecondition (SD responders: 1.377�0.362, SD non-responders: 0.887�0.311, significance of the differenceusing t-test for independent samples: P=0.024).

Polysomnographic parameters are listed in Table 2:SD responders presented a significantly longer REMlatency than SD non responders according to the ‘‘leni-ent’’ REM latency definition. Using the ‘‘strict’’ REMlatency definition, no differences were observed. Otherconventional sleep parameters did not differ sig-nificantly between the groups.

4. Discussion

The immediate albeit short-lasting therapeutic effectof sleep deprivation (SD) in approximately 60% ofpatients with a major depressive disorder has become afascinating approach in research on biological processesin depressive disorders. Polysomnographic sleepresearch of the last two decades showed sleep peculia-rities in depressed patients and interesting relations toneurobiological hypotheses of affective disorders, butno entirely convincing concept of the mechanisms ofaction of SD. Recent studies suggested that mainlychanges in NREM sleep, especially in slow wave activity(SWA), could be associated with the therapeutic effectof SD.

In the present study we evaluated the all-night sleepEEG in the night before SD by spectral analysis, whichis well suited to detect fine-grained differences in NREMsleep, including slow wave activity. Between SD

Table 2

Polysomnographic sleep parameters of investigated samples (mean�standard deviation), REM latency definitions see methods

SD responders SD non-responders Significance of the differencea

n 8 8 –

Time in bed (min) 424.7 (� 23.9) 416.9 (� 4.5) 0.382

Sleep latency (min) 17.6 (� 6.5) 18.3 (� 9.6) 0.869

Sleep period time (min) 399.1 (� 22.6) 394.0 (� 8.6) 0.563

Total sleep time (min) 354.4 (� 43.2) 358.9 (� 28.6) 0.807

Sleep efficiency (%) 83.8 (� 12.4) 86.1 (� 7.1) 0.656

REM latency: lenient definition (min) 71.6 (� 16.7) 44.2 (� 28.1) 0.033

REM latency: strict definition (min) 92.13 (� 48.9) 69.8 (� 58.4) 0.420

Total REM density (%) 35.5 (� 10.2) 34.0 (� 11.3) 0.778

First NREM sleep episode

Total duration (min)=REM latency (lenient definition)

Waking time (min) 4.9 (� 6.9) 0.9 (� 1.7) 0.134

Stage 2 (min) 43.8 (� 13.3) 28.2 (� 20.1) 0.089

Stage 3 and 4 (min) 19.6 (� 22.5) 12.9 (� 13.6) 0.484

First REM sleep episode

Total duration (min) 18.9 (� 10.7) 21.7 (� 14.3) 0.663

Second NREM sleep episode

Total duration (min) 73.9 (� 20.3) 73.3 (� 14.0) 0.938

Waking time (min) 6.5 (� 11.2) 3.0 (� 4.7) 0.429

Stage 2 (min) 54.8 (� 11.1) 51.3 (� 24.0) 0.714

Stage 3 and 4 (min) 9.8 (� 11.9) 21.4 (� 17.6) 0.144

a t-Test for independent samples.


responders and non-responders age and gender werecarefully pair-matched in order to avoid their knownstrong influences on the spectrally analyzed sleep(Astrom & Trojaborg, 1992; Landoldt et al., 1996;Ehlers & Kupfer, 1997). Also, clinical parameters didnot differ between the groups under baseline conditions.Furthermore, all participants were free of any psy-

choactive drugs for at least 1 week prior to the sleeprecordings.

Whole-night absolute and relative NREM sleep spec-tral power did not differ significantly between pro-spective SD responders and non-responders in the nightbefore SD. In so far, spectral power in all frequencybands, including SWA, did not characterize subgroupsof therapeutic outcome. This finding corresponds toprevious polysomnographic studies, which were notable to establish predictors for SD response regardingNREM sleep, including slow wave sleep stages (Rie-mann et al., 1991).

Regarding the distribution of SWA across the night,SD responders showed a constant decline across NREMsleep episodes under baseline conditions, which is char-acteristic for sleep homeostasis in healthy controls(Borbely, 1982; Landoldt et al., 1996; Preudhomme etal., 1997). Interestingly, SD non-responders showedSWA values similar to the responders at the beginningand the end of the night, but an increase in SWA fromthe first to the second NREM episode. This deviantSWA distribution has previously been described fordepressed patients (Kupfer et al., 1984, 1986). However,SWA within any single NREM episode did not differsignificantly between SD responders and non-respon-ders (t-tests and ANOVA), probably due to large inter-individual variations.

To describe the distribution of SWA in the first twoNREM episodes, Kupfer et al. (1990) introduced theratio between the delta power in the first and secondNREM episode, termed delta sleep ratio (DSR); values>1 for the DSR indicate a decline in delta power fromthe first to the second NREM episode, whereas values <1indicate an inverse distribution. Using spectral analysis of

Fig. 2. Slow wave activity for responders and non-responders to SD over the different NREM episodes (2-factorial ANOVA with factors NREM

episode and response status).

Fig. 3. Delta sleep ratio for responders and non-responders to S.D.


all-night sleep EEG of 74 remitted patients with recur-rent depressive disorders following discontinuation ofdrug treatment, Kupfer and colleagues showed that thisparameter predicted clinical outcome for maintenanceinterpersonal psychotherapy (IPT): Individuals with ahigh DSR (i.e. a ‘‘normal’’ distribution of SWA)achieved significantly longer clinical remission thanthose with a lower DSR. The major finding of the cur-rent study was a correspondingly higher DSR inresponders than in non-responders to sleep deprivation.

With regard to polysomnography, SD responderspresented a significantly longer REM latency using the‘‘lenient’’ definition, but not according to the ‘‘strict’’definition of REM latency. The inconsistency to otherfindings (reduced REM latency as a positive predictor:Riemann et al., 1991; no association of REM latencyand SD response: Gerner et al., 1979; Duncan et al.,1980) may be due to intervening variables, like drugstatus, age and gender, which were not as strictly con-trolled in these studies. Moreover, investigations ofREM latency raise the question on the relation betweenspectral and polysomnographic sleep parameters: actu-ally it is not clear in which way the average SWA withinthe first two NREM sleep episodes is related to theduration of these episodes. The answer to this questioninvolves the problem of the ‘‘two roads to REMlatency’’ (Kupfer and Ehlers, 1989), since it dependsupon whether REM sleep occurs passively, unmaskedby the homeostatic NREM background, or to somedegree actively interrupts NREM sleep, i.e. has its ownand independent homeostasis. However, in our studyREM latency and averaged SWA in the first NREMepisode were not correlated (Pearson correlation coeffi-cient: 0.2611; P=0.179). Differences in SWA were pro-minent in the second NREM episode in whichpolysomnographic parameters were very similar, indi-cating an independent importance of SWA distributionfor the therapeutic outcome of SD.

Beyond the obvious differences between our approachand the mentioned investigations of recurrence underIPT (Kupfer et al., 1990), in both studies a low DSRcharacterized a subgroup of depressed patients having amore unfavorable therapeutic outcome concerning SDresponse respectively ‘‘survival’’, i.e. length of clinicalremission following IPT.

Previous studies of patients with recurrent depressivedisorder described that slow wave sleep abnormalities,including a decreased DSR, were not limited to theacute episode, but persisted as a trait marker intoremission (Buysse & Kupfer, 1993; Thase & Howland,1995; Thase et al., 1998). Polysomnographic findings ofSWS disturbances, like decreased slow wave sleep,based on longitudinal follow-up (Buysse et al., 1997)and family studies (Giles et al., 1989; Linkowsky et al.,1991; Lauer et al., 1995), were in favour of a geneticdetermination and against the hypothesis of reduced

SWS as due to a biological ‘‘scar’’ at the beginning ofdepressive illness.

Based on its putative trait character and the negativepredictive value for antidepressant treatment of a DSR< 1 (SD response in the present paper, length ofremission following IPT in the Kupfer et al. study, 1990)we hypothesize that a decreased DSR could indicate astronger neurobiological predisposition for depressivedisorders. Following this hypothesis, SD responders couldbe viewed as a subgroup of patients less biologicallyfixed in their depressed state. At the clinical level thisassumption fits to the frequent observation of diurnalmood variation as a positive predictor for SD response.Diurnal variation of mood could thus be interpreted asa sign of spontaneous flexibility of the affective stateand also probably its underlying biological substrate.

With regard to the clinical utility, our findings sup-port the proposition of Kupfer et al. (1990): a deviantDSR may indicate the need of a therapeutic conceptincluding higher dosed and longer antidepressant phar-macotherapy to compensate the assumed increased bio-logical predisposition. The possible shift of SWAtowards the first NREM episode by antidepressantdrugs, which was significantly and positively correlatedto later clinical benefit (Reynolds et al., 1991; Ehlers etal., 1996), could point in this direction.

Functional interpretations of the different SWA dis-tributions dependent on SD response are difficult.Mainly triggered by pace-making thalamocortical neu-rons (Steriade et al., 1991) the SWA has been shown tobe controlled by various influences. Porrka-Heiskanenet al. (microdialysis measurements in cats, 1997) showedthat the extracellular concentration of the neuromodu-lator adenosine in the cholinergic region of the basalforebrain increased during wakefulness and declinedslowly during recovery sleep as contrasted with slowwave homeostasis, but presently a relevance of adenosinefor depressive disorders has not yet been demonstrated.The relationship between SWA and neurotransmittersrelated to pathophysiological hypothesis of affectivedisorders, for example enhancement of SWA by sero-tonin receptor antagonists (Landoldt et al., 1999) and closepositive correlation between growth hormone releaseand SWA (Gronfier et al., 1996; Seifritz et al., 1996; forreview of neuropeptides and sleep see: Steiger & Holsboer,1997), may indicate the direction of further studies.

In conclusion, no SD predictor was found in theaveraged absolute and relative spectral power withinany accessible NREM sleep EEG frequency band.However, we found that prospective SD responders andnon-responders were characterized by a different dis-tribution of SWA across NREM sleep episodes in thenight before sleep deprivation, expressed in a sig-nificantly lower delta sleep ratio in the non-responders.This finding may contribute to new therapeutic and sci-entific approaches towards depressive illness.


References

Albert R, Merz A, Schubert J, Ebert D. Sleep deprivation and sub-

sequent sleep phase advance stabilizes the positive effect of sleep

deprivation in depressive episodes. Nervenarzt 1998;69:66–9.

American Psychiatric Association. Diagnostic and statistical manual

of mental disorders, third ed. rev. Washington DC: American Psy-

chiatric Association, 1987.

Armitage R. Microarchitectural findings in sleep EEG in depression:

diagnostic implications. Biological Psychiatry 1995;37:72–84.

Astrom C, Trojaborg W. Relationship of age to power spectrum ana-

lysis of EEG during sleep. J Clinical Neurophysiology 1992;9:424–30.

Bech P, Gram LF, Dein E, Jacobson O, Vitger J, Bohoys TG. Quan-

titative ratings of depressive states. Acta Psychiatrica Scandinavica

1975;61:161–70.

Beersma DG, Van den Hofdakker RH. Can non-REM sleep be

depressogenic? Journal of Affective Disorders 1992;24:101–8.

Berger M, Vollmann J, Hohagen F, Konig A, Lohner H, Voderholzer

U, Riemann D. Sleep deprivation combined with consecutive sleep

phase advance as a fast-acting therapy in depression: an open pilot

trial in medicated and unmedicated patients. American Journal of

Psychiatry 1997;154:870–2.

Blackman RB, Turkey JW. The measurement of power spectra. New

York: Dover, 1958.

Borbely AA. A two process model of sleep regulation. Human Neu-

robiology 1982;1:195–204.

Buysse DJ, Kupfer DJ. Sleep disorders in depressive disorders. In:

Mann JJ, Kupfer DJ, editors. Biology of depressive disorders Part

A: a systems perspective. New York: Plenum Publishing Corp, 1993.

p. 123–54.

Buysse DJ, Kupfer DJ, Frank E, Monk TH, Ritenour A, Ehlers CL.

Electroencephalographic sleep studies in depressed outpatients trea-

ted with interpersonal psychotherapy: I. Baseline studies in respon-

ders and nonresponders. Psychiatry Research 1992a;42:13–26.

Buysse DJ, Kupfer DJ, Frank E, Monk TH, Ritenour A. Electro-

encephalographic sleep studies in depressed outpatients treated with

interpersonal psychotherapy: II. Longitudinal studies at baseline

and recovery. Psychiatry Research 1992b;42:27–40.

Buysse DJ, Frank E, Lowe KK, Cherry CR, Kupfer DJ. Electro-

encephalographic sleep correlates of episode and vulnerability to

recurrence in depression. Biological Psychiatry 1997;41:406–18.

Duncan WC, Gillin JC, Post RM, Gerner RH, Wehr TA. Relation-

ship between EEG sleep patterns and clinical improvement in

depressed patients treated with sleep deprivation. Biological Psy-

chiatry 1980;15:879–89.

Ebert D, Feistel H, Barocka A. Effects of sleep deprivation on the lim-

bic system and the frontal lobes in affective disorders: a study with

Tc-99–HMPAO-SPECT. Psych Res Neuroimag 1991;40:247–51.

Ebert D, Feistel H, Barocka A, Kaschka W. Increased limbic blood

flow and total sleep deprivation in major depression with melanch-

olia. Psych Res Neuroimag 1994;55:101–11.

Ehlers CL, Kupfer DJ. Slow-wave sleep — do young adult men and

women age differently. Journal of Sleep Research 1997;6:211–5.

Ehlers CL, Havstad JW, Kupfer DJ. Estimation of the time course of

slow-wave sleep over the night in depressed patients: effects of clo-

mipramine and clinical response. Biological Psychiatry 1996;39:171–

81.

Feige B. Oscillatory brain activity and its analysis on the basis of MEG

and EEG. Munster/New York/Munchen/Berlin: Waxmann , 1999.

Feige B, Voderholzer U, Riemann D, Hohagen F, Berger M. Inde-

pendent sleep EEG slow-wave and spindle band dynamics asso-

ciated with 4 weeks of continuous application of short-halflife

hypnotics in healthy subjects. Clin Neurophysiol 1999;110:1965–74.

Frank E, Jarrett DB, Kupfer DJ, Grochocinski VJ. Biological and

clinical predictors of response in recurrent depression: a preliminary

report. Psychiatry Research 1984;13:315–24.

Gerner RH, Post RM, Gillin JC, Bunney WE. Biological and beha-

vioral effects of one night’s sleep deprivation in depressed patients

and normals. Journal of Psychiatry Research 1979;15:21–40.

Giles DE, Kupfer DJ, Roffwarg HP, Rush AJ, Biggs MM, Etzel BA.

Polysomnographic parameters in first-degree relatives of unipolar

probands. Psychiatry Research 1989;27:127–36.

Gordijn MCM, Beersma DGM, Bonhuys NL, Reinink E, Van den

Hofdakker RH. A longitudinal study of diurnal mood variation in

depression; characteristics and significance. Journal of Affective

Disorders 1994;31:261–73.

Gordijn MCM, Beersma DGM, Bonhuys NL, Korte HJ, Van den

Hofdakker RH. A longitudinal study of sleep deprivation response

in depression; the variability is highly related to diurnal mood

variability. Acta Neuropsychiatrica 1995;7:58–60.

Gronfier C, Luthringer R, Follenius M, Schaltenbrand N, Macher JP,

Muzet A, Brandenberger G. A Quantitative evaluation of the rela-

tionships between growth hormone secretion and delta wave elec-

troencephalographic activity during normal sleep and after

enrichment in delta waves. Sleep 1996;19:817–24.

Hemmeter U, Bischof R, Hatzinger M, Seifritz E, Holsboer-Trachsler

E. Microsleep during partial sleep deprivation in depression. Biol

Psychiatry 1998;43(11):829–39.

Jarret DB, Rush AJ, Khatami M, Roffwarg HP. Does the pretreat-

ment of polysomnogram predict response to cognitive therapy in

depressed outpatients? Psychiatry Research 1990;33:285–99.

Kuhs H, Tolle R. Sleep deprivation therapy. Biological Psychiatry

1991;29:1129–48.

Kupfer DJ, Ehlers CL. Two roads to REM latency. Archives of Gen-

eral Psychiatry 1989;46:945–8.

Kupfer DJ, Ulrich RF, Coble P, Jarret DB, Grochocinski VJ, Doman

J, Matthews G, Borbely AA. Application of an automated REM

and slow wave sleep analysis: testing the assumptions of the two-

process model of sleep regulation in normal and depressed subjects.

Psychiatry Research 1984;13:335–43.

Kupfer DJ, Reynolds CF, Ulrich RF, Grochocinski VJ. Comparison

of automated REM and slow-wave sleep analysis in young and

middle-aged depressed subjects. Biological Psychiatry 1986;21:189–

200.

Kupfer DJ, Frank E, Mc Eachran A, Grochocinski VJ. Delta sleep

ratio. Archives of General Psychiatry 1990;47:1100–5.

Kupfer DJ, Ehlers CL, Frank E, Grochocinski VJ, McEachran AB,

Buhari A. Electroencephalographic sleep studies in depressed

patients during long-term recovery. Psychiatry Research

1993;49:121–38.

Landoldt HP, Dijk DJ, Achermann P, Borbely AA. Effect of age on

the sleep EEG: slow-wave activity and spindle frequency activity in

young and middle-aged men. Brain Research 1996;738:205–12.

Landoldt HP, Meier V, Burgess HJ, Finelli LA, Cattelin F, Acher-

mann P, Borbely AA. Serotonin-2 receptors and human sleep: effect

of a selective antagonist on EEG power spectra. Neuropsycho-

pharmacology 1999;21:455–66.

Lauer CJ, Schreiber W, Holsboer F, Krieg JC. In quest of identifying

vulnerability markers for psychiatric disorders by all-night poly-

somnography. Arch Gen Psychiatry 1995;52(2):145–53.

Linkowski P, Kerkhofs M, Hauspie R, Mendlewicz J. Genetic deter-

minants of EEG sleep: a study in twins living apart. Electro-

encephalogr Clin Neurophysiol 1991;79:114–8.

Porkka-Heiskanen T, Strecker RE, Thakkar M, Bjorkum AA, Greene

RW, McCarley RW. Adenosine: a mediator of the sleep-inducing

effects of prolonged wakefulness. Science 1997;276:1265–8.

Preudhomme XA, Lanquart JP, Mendlewicz J, Linkowski P. Dis-

tribution of delta activity across non-rapid eye movement sleep epi-

sodes in healthy young men. Sleep 1997;20:313–20.

Rechtschaffen A, Kales A. A manual of standardized terminology,

techniques and scoring system for sleep stages of human subjects.

Washington DC: US Government Printing Office, 1968.

Reynolds CF, Hoch CC, Buysse DJ, George CJ, Houck PR, Mazum-

dar S, Miller M, Pollock BG, Rifai H, Frank E. Sleep in late-life


recurrent depression. Changes during early continuation therapy

with nortriptyline. Neuropsychopharmacology 1991;5:85–96.

Riemann D, Berger M. The effects of total sleep deprivation and sub-

sequent treatment with clomipramine on depressive symptoms and

sleep electroencephalography in patients with a major depressive

disorder. Acta Psychiatrica Scandinavica 1990;81:24–31.

Riemann D, Wiegand M, Berger M. Are there predictors for sleep

deprivation response in depressed patients? Biological Psychiatry

1991;29:707–10.

Riemann D, Wiegand M, Lauer C, Berger M. Naps after total sleep

deprivation in depressed patients: are they depressiogenic? Psy-

chiatry Research 1993;49:109–20.

Riemann D, Konig A, Hohagen F, Kiemen A, Voderholzer U, Back-

haus J, Bunz B, Wesiack B, Hermle L, Berger M. How to preserve

the antidepressant effect of sleep deprivation: a comparison of sleep

phase advance and sleep phase delay. European Archives of Psy-

chiatry and Clinical Neurosience 1999;249:231–7.

Seifritz E, Mueller MJ, Trachsel L, Lauer CJ, Hemmeter U, Hatzinger

M, Moore P, Holsboer-Trachsler E. Revisiting the Ehlers and

Kupfer hypothesis: the growth hormone cortisol secretion ratio

during sleep is correlated with electroencephalographic slow wave

activity in normal volunteers. Biol Psychiatry 1996;39(2):139–42.

Smith GS, Reynolds CF, Pollock B, Derdyshire S, Nofzinger E, Dew

MA, Houck PR, Milko D, Metzler CC, Kupfer DJ. Cerebral glu-

cose metabolic response to combined sleep deprivation and anti-

depressant treatment in geriatric depression. American Journal of

Psychiatry 1999;156:683–9.

Steiger A, Holsboer F. Neuropeptides and human sleep. Sleep

1997;20(11):1038–52.

Steriade M, Dossi R, Curro Nunez A. Network modulation of a slow

intrinsic oscillation of cat thalamocortical neurons implicated in

sleep delta waves: cortically. J Neurosci 1991;11:3200–17.

ThaseME, Howland RH. Biological processes in depression: an updated

review and integration. In: Beckham EE, Leber WR, editors. Hand-

book of depression. New York: Guilford Press, 1995. p. 213–79.

Thase ME, Fasiczka AL, Berman SR, Simons AD, Reynolds CF.

Electroencephalographic sleep profiles before and after cognitive

behavior therapy of depression. Archives of General Psychiatry

1998;55:138–44.

Van den Hofdakker RH. Total Sleep Deprivation: Clinical and theo-

retical Aspects. In: Depression: Neurobiological, Psychopathologi-

cal and Therapeutic Advances, John Wiley & Sons, 1997, p. 563-

589.

Volk S, Kaendler S, Weber R, Georgi K, Maul F, Hertel A, Pflug B.

Evaluation of the effects of total sleep deprivation on cerebral blood

flow using single photon emission computerized tomography. Acta

Psychiatrica Scandinavica 1992;86:478–83.

Volk S Kaendler S, Weber R Georgi K, Maul F Hertel A, Pflug B Hor

G. Can response to partial sleep deprivation in depressed patients be

predicted by regional changes of cerebral blood flow? Pychiatry

Research Neuroimaging 1997;75:67–74.

Vollmann J, Berger M. Sleep deprivation with consecutive sleep phase

advance therapy in patients with major depression. A pilot study.

Biological Psychiatry 1993;33:54–7.

Wiegand M, Berger M, Zulley J, Lauer C, von Zerssen D. The influ-

ence of daytime naps on the therapeutic effect of sleep deprivation.

Biological Psychiatry 1987;22:389–92.

Wiegand M, Riemann D, Schreiber W, Lauer CJ, Berger M. Effect of

morning and afternoon naps on mood after total sleep deprivation

in patients with major depression. Biological Psychiatry

1993;33:467–76.

Wirz-Justice A, Van den Hofdakker RH. Sleep deprivation in depres-

sion: what do we know, where do we go? Biological Psychiatry

1999;46:445–53.

Wu JC, Bunney WE. The biological basis of an antidepressant

response to sleep deprivation and relapse: review and hypothesis.

American Journal of Psychiatry 1990;147:14–21.

Wu JC, Gillin JC, Buchsbaum MS, Hershey T, Johnson JC, Bunney

WE. Effect of sleep deprivation on brain metabolism of depressed

patients. American Journal of Psychiatry 1992;149:538–43.

Wu JC, Gillin JC, Buchsbaum MS, Cadwell S, Lottenberg S, Bunney

WE. PET studies of sleep deprivation in unipolar depression. Neu-

ropsychopharmacology 1994;10:876.

Wu JC, Buchsbaum MS, Gillin JC, Tang C, Cadwell S, Wiegand M,

Najafi A, Klein E, Hazen K, Bunney WE. Prediction of anti-

depressant effects on sleep deprivation by metabolic rates in the

ventral anterior cingulate and medial prefrontal cortex. American

Journal of Psychiatry 1999;156:1149–58.


delta sleep ratio as a predictor of sleep deprivation response in major depression

Documents