Regular Article

Psychiatry and Clinical Neurosciences

(2003),

57

, 97–104

Blackwell Science, LtdOxford, UKPCNPsychiatric and Clinical Neurosciences1323-13162002 Blackwell Science Pty Ltd571February 2003

1085BDZ effects on sleep EEGX. Tan et al.10.1046/j.1323-1316.2002.01085.x

Regular Article

97104BEES SGML

Correspondence address: Sunao Uchida, Department of Sleep Dis-orders Research, Tokyo Institute of Psychiatry, 2-1-8 Kamikitazawa,Setagaya-Ku, Tokyo 156-8585, Japan. Email: sunao@uchidae.com

Received 19 March 2002; revised 13 June 2002; accepted 17 June2002.

Regular Article

Long-, intermediate- and short-acting benzodiazepine effects on human sleep EEG spectra

XIN TAN,

MD

,

PhD

,

1,2

SUNAO UCHIDA,

MD

,

PhD

,

1

MASATO MATSUURA,

MD

,

PhD

,

2

KYOKO NISHIHARA,

PhD

1

AND TAKUYA KOJIMA,

MD

,

PhD

2

1

Department of Sleep Disorders Research, Tokyo Institute of Psychiatry, Tokyo and

2

Department of Neuropsychiatry, Nihon University, School of Medicine, Tokyo, Japan

Abstract

The effects of 10 mg haloxazolam (HAX), 4 mg flunitrazepam (FNZ), and 0.5 mg triazolam (TRI),each administered for seven consecutive nights were studied in the sleep electroencephalograms(EEG) of 17 (six HAX, five FNZ and six TRI) healthy male student volunteers. Recordings ofC3–A1 EEG data from one baseline night, three drug nights (first, fourth and seventh) and twowithdrawal nights (second and fourth) were analyzed using a fast Fourier transformation method.All three drugs induced similar changes in the 0.5 Hz to 40 Hz power spectrum; namely (i) higherfrequency (including the sigma and beta bands) activity increased and lower frequency activityreduced on the drug nights; (ii) the enhancement of sigma activity peaked during non-rapid eyemovement sleep following the first administration and was maintained at high levels on all drugnights; (iii) beta activity increased through the night after administration of HAX and FNZ, butnot TRI, which suggests a blood concentration level dependent increase of beta activity; and (iv)only HAX showed a residual effect on the fourth withdrawal night. These results indicate that (i)chronic administration of these three benzodiazepine derivatives produce similar profiles in sleepEEG spectral changes, with some differences depending on their half-lives and doses; (ii) themechanism of sigma enhancement is sensitive to even the initial administration night of BDZ; and(iii) frequencies below and above the sigma band are less sensitive to BDZ and also show anincrease through the night after administration, suggesting differences in the mechanisms reflectedby these EEG frequency bands.

Key words

benzodiazepines, pharmacological potency, sleep electroencephalogram, spectral analysis.

INTRODUCTION

Benzodiazepine (BDZ) hypnotics bind to specific high-affinity receptor sites associated with the macromolec-ular gamma-aminobutyric acid (GABA) receptor com-plex and enhance the actions of GABA, which is themajor inhibitory neurotransmitter in the central ner-vous system (CNS). These receptors are located in awide variety of brain areas in conjunction with thesubpopulation of GABA-A receptors. In a mousestudy, the density of BDZ receptors was greatest in thehippocampus and decreased in the following order in

samples from cortex, hypothalamus, striatum, and themidbrain.

1

Clinical differences between BDZ drugsassociate with their pharmacokinetic properties ofabsorption, distribution, elimination and clearance.

2

Elimination half-life appears to be a particularlyimportant factor.

3

The half-lives of BDZ hypnotics dif-fer greatly; ranging from 3 to 100 h.

4

Short-acting hyp-notics lose their effects quickly, and usually producerebound insomnia after discontinuation.

5–7

By contrast,the enduring CNS concentrations of long half-life hyp-notics tend to produce daytime sedation and daytimeperformance decrements.

8,9

The benzodiazepine effectson memory has been also studied.

10

We performed an experiment to study the effects ofthree benzodizepine agents on all-night sleep electro-encephalogram (EEG). The three drugs studied werehaloxazolam (HAX), flunitrazepam (FNZ), and triaz-olam (TRI), which represented long-, intermediate-,

98 X. Tan

et al

.

and short-acting BDZ hypnotics, respectively. Portionsof this study have been published previously, discussingFNZ effects on non-rapid eye movement (NREM) andREM EEG spectra and individual differences,

11

differ-ent FNZ effects on delta-sigma and delta-beta inverserelationships,

12

and comparison of TRI and FNZ.

13

Thepresent study is a systematic comparison of these threeBDZ hypnotics and changes in the sleep EEG spectrawere quantified in detail.

METHODS

The subjects were 17 healthy young male students (agerange 20–25 years old) who were asked to abstain fromalcohol, caffeine and naps for 1 week prior to theexperiment. Each subject was studied on a 14-dayschedule (Table 1) in a double-blind placebo-controlled design. All the data were recorded beforestarting analyses. Written informed consent wasobtained after explaining experimental procedures andpossible risks. Thirty minutes before bedtime on thefourth through tenth nights, six subjects were adminis-tered 10 mg of HAX (half-life 42–123 h); five subjectswere administered 4 mg of FNZ (half-life 9–30 h); andsix subjects were administered 0.5 mg of TRI (half-lifeless than 3 h). Placebo was given on the two initialbaseline nights and four withdrawal nights.

In each laboratory sleep session, which began at23:00 hours and ended at 07:00 hours, scalp EEGsrecorded from F3, F4, C3, C4, O1 and O2 (referencedto linked A1

+

A2 earlobe electrodes; time constant

=

0.3 sec), electrooculographic (EOG) horizontal andvertical eye movements, a chin surface electromyo-gram (EMG), electrocardiographic (ECG) and respi-ratory activities were recorded on magnetic tape foroffline analysis. Fast Fourier transformation (FFT)analyses of the C3 EEG were performed on consecu-tive 20-sec epochs (512 Hz sampling; five 4-sec epochsaverage) from recordings of the second baseline night(2BLN), the initial drug night (IDN), the fourth andseventh drug nights (4DN and 7DN), and the second

and fourth withdrawal nights (2WN and 4WN). Epochsof NREM stages 2, 3, 4 and REM were analyzed usingPASS PLUS (Delta Software, St. Louis, MO, USA;http://www.deltapass.com) sleep analysis software.

Power spectra were summed in 0.5-Hz wide bins forfrequencies from 0.5 to 15 Hz, in 1-Hz wide binsbetween 16 and 30 Hz, and in 2-Hz wide bins between30 and 40 Hz. Epochs containing artifacts were visuallyidentified and removed before statistical analysis.Visual sleep stage scoring was performed on the FFT-analyzed epochs according to standard Rechtschaffenand Kales’ criteria.

14

The sleep stage scorers were blindto experimental conditions.

Feinberg’s

15

definition of NREM–REM cycles wasused except when a ‘skipped’ first REM period wasinferred. Such ‘skipped’ first REM periods were recog-nized when the length of the first NREM periodexceeded 120 min, and a delta (0.3–3 Hz) power troughwas apparent in the computer analysis following deltaduration comparable to that in later NREM–REMsleep cycles. Thus, in the absence of visually identifiedREM, the first NREM period was defined from sleeponset to the first delta trough; and the second NREMperiod extended from the first delta trough to thebeginning of the first REM period.

Statistical comparisons between 2BLN and the othernights were conducted by paired two-way

t

-tests forthose frequencies that showed significant differences ina preliminary two-way

ANOVA

.This protocol was approved by the Tokyo Institute

of Psychiatry Ethics Committee.

RESULTS

Visual assessment of sleep parameters

Visually scored sleep stage parameters and the numberof ‘skipped’ first REM periods in each condition areshown in Tables 2 (HAX), 3 (FNZ) and 4 (TRI). Com-pared to BLN, all three drugs significantly increasedstage 2 sleep on all drug nights, although only HAXeffects persisted during withdrawal, maintaining a

Table 1.

Experimental schedule

Day 1ADN

21BLN

32BLN

4IDN

52 DN

63 DN

74 DN

85 DN

96 DN

107 DN

111WN

122WN

133WN

144WN

Placebo Drug Placebo

P PA

PA

PA

PA

PA

PA

ADN, adaptation night; BLN, baseline night; IDN, initial drug night; DN, drug night; WN, withdrawal night; P,polysomnography; A, analyzed.

BDZ effects on sleep EEG 99

Table 2.

Sleep parameters for baseline, drug and withdrawal nights of haloxazolam (

n

=

6, mean (SD))

Recording night BLN2 IDN 4DN 7DN 2WN 4WN

Sleep efficiency (%) 87.6 (18.2) 95.5 (2.8) 94.5 (4.1) 92.8 (8) 91.7 (8) 92.6 (6.4)Time in bed (min) 485.0 (3.8) 485.4 (2.7) 487.1 (2.7) 483.0 (1.1) 485.8 (2) 484.1 (1.4)Total sleep time 425.0 (88) 463.4 (13.4) 460.2 (20.2) 447.9 (38.3) 445.7 (39.7) 448.0 (30.8)Sleep onset latency 28.2 (23.7) 23.8 (16.1) 18.9 (15.6) 12.4 (3.5) 31.4 (43.9) 20.3 (8.4)REM latency 84.1 (29.8) 116.3 (40.2) 133.1 (47.5) 137.6 (50.2) 117.4 (31.7) 97.0 (31.1)Total wake time 57.6 (88.3) 20.3 (14.3) 24.6 (19.5) 30.8 (33.8) 37.4 (39.1) 33.9 (31.5)‘Skipped REM’ [#] [3] [4] [4] [2]Stage REM (min) 112.3 (36.4) 102.7 (11.5) 92.9 (14.6) 82.5 (10.5) 99.2 (20.3) 112.7 (19.8)Stage 1 18.1 (11.4) 15.3 (9.4) 19.3 (6.8) 26.8 (16.2) 20.8 (8.8) 33.2 (18)Stage 2 206.3 (56.2) 248.2 (35.7)* 290.1 (23.7)* 286.3 (45.5)* 272.6 (27.9)* 252.06 (41.1)*Stage 3

+

4 88.3 (18.5) 97.3 (29) 57.8 (24.3)* 55.0 (20.8)* 53.2 (19)* 50.0 (27.1)*

*

P

<

0.05 by

t

-test.BLN, baseline night; IDN, initial drug night; DN, drug night; WN, withdrawal night.

Table 3.

Sleep parameters for baseline, drug and withdrawal nights of flunitrazepam (

n

=

5, mean (SD))

Recording night 2BLN IDN 4DN 7DN 2WN 4WN

Sleep efficiency (%) 94.56 (0.8) 94.55 (2.5) 96.52 (0.6) 94.69 (1.7) 83.93 (3.2)* 85.45 (4.0)Time in bed (min) 484.73 (1.3) 482.70 (0.4) 482.70 (0.3) 482.50 (0.6) 484.13 (2.0) 483.60 (1.2)Total sleep time 458.33 (11.0) 456.40 (36.0) 456.87 (9.2) 456.93 (24.2) 406.27 (45.8)* 413.27 (58.9)Sleep onset latency 18.26 (5.2) 9.53 (6.1)* 9.60 (3.1)* 20.27 (23.2) 44.87 (28.9) 64.73 (63.1)REM latency 81.27 (22.1) 124.33 (37.3) 174.27 (70.3)* 162.00 (49.2)* 113.67 (26.6) 87.20 (12.3)Total wake time 21.20 (11.5) 22.33 (34.6) 12.87 (8.6) 19.40 (24.7) 72.60 (44.8)* 66.53 (58.3)‘Skipped REM’ [#] [2] [4] [3]Stage REM (min) 116.87 (26.5) 73.40 (22.0)* 94.00 (26.9)* 80.53 (13.0)* 80.93 (29.6) 105.20 (14.0)Stage 1 20.33 (10.2) 9.40 (4.7) 13.20 (5.2) 29.00 (25.4) 29.40 (15.2) 27.93 (13.1)Stage 2 203.00 (14.4) 254.00 (34.0)* 297.00 (24.6)* 293.67 (43.4)* 217.00 (39.3) 180.27 (27.4)Stage 3

+

4 118.13 (38.7) 119.60 (18.2) 61.67 (17.7)* 53.73 (26.4)* 78.93 (32.2)* 99.87 (22.7)

*

P

<

0.05 by

t

-test.BLN, baseline night; IDN, initial drug night; DN, drug night; WN, withdrawal night.

Table 4.

Sleep parameters for baseline, drug and withdrawal nights of triazolam (

n

=

6, mean (SD))

Recording night 2BLN IDN 4DN 7DN 2WN 4WN

Sleep efficiency (%) 94.21 (0.7) 96.54 (0.5)* 95.61 (0.4) 96.14 (0.4)* 87.74 (3.2) 90.33 (2.2)Time in bed (min) 485.17 (3.8) 486.39 (3.6) 486.00 (2.9) 483.78 (2.3) 485.67 (2.0) 485.28 (3.9)Total sleep time 457.06 (10.7) 469.56 (7.6) 464.67 (7.6) 465.11 (6.4) 426.11 (46.8) 438.33 (31.2)Sleep onset latency 16.22 (5.9) 13.44 (3.1) 20.22 (6.3) 14.50 (5.2) 41.50 (51.2) 28.83 (12.9)REM latency 74.61 (39.2) 75.72 (14.4) 129.72 (49.9) 82.44 (8.5) 64.39 (8.1) 82.23 (19.3)Total wake time 25.67 (10.7) 13.28 (6.3)* 18.83 (5.6) 16.11 (4.5)* 52.06 (45.5) 42.56 (32.2)‘Skipped REM’ [#] [1] [1] [1] [1]Stage-REM (min) 129.28 (27.6) 107.94 (10.9) 104.00 (18.2) 99.72 (12.6) 115.56 (38.2) 111.94 (33.6)Stage-1 21.83 (11.7) 15.06 (6.4) 19.80 (6.7) 19.78 (7.8) 20.67 (10.4) 26.72 (8.2)Stage-2 210.50 (12.8) 259.33 (25.4)* 263.06 (20.0)* 277.39 (24.9)* 205.39 (33.4) 222.11 (15.9)Stage-3

+

4 95.44 (28.1) 86.50 (29.8) 77.72 (22.4) 68.22 (24.8) 84.50 (25.9) 77.56 (23.5)

*

P

<

0.05 by

t

-test.BLN, baseline night; IDN, initial drug night; DN, drug night; WN, withdrawal night.

100 X. Tan

et al

.

significant effect through 4WN. Both HAX and FNZsignificantly decreased the amounts of visually scoredsleep stages 3

+

4 on the 4DN and 7DN for FNZ, aneffect that persisted through 2WN and, for HAX,4WN. Only FNZ decreased REM sleep significantly onall the drug nights, beginning on the initial drug night.

Spectral changes in NREM and REM sleep EEG during drug and withdrawal nights

Drug nights

Computer analyses revealed striking EEG spectralchanges during BDZ administration. Figure 1 presentsthe sleep EEG power spectra for each BDZ duringdrug nights IDN, 4DN and 7DN, each relative to BLN.The frequencies that differed significantly from BLNare indicated by bars under each graph. In general,HAX, FNZ and TRI all enhanced frequencies aboveabout 10 Hz and reduced lower frequencies on drugnights.

Following administration of TRI, no significant dif-ferences were found between the drug nights for eitherNREM or REM sleep. By contrast, both HAX andFNZ showed significant changes across drug nights.Compared to their IDNs, HAX and FNZ both signifi-cantly further reduced low frequency activity andincreased beta during NREM and REM sleep(

P

<

0.05). HAX also significantly increased sigmaactivity (11.5–12 Hz,

P

<

0.05) during NREM sleep on4DN and 7DN.

Comparing the average peak values of sigma (11–12.5 Hz) and beta (23–25 Hz in FNZ and TRI, and 26–29 Hz in HAX) frequencies during NREM sleep, HAXproduced stronger effects on sigma than beta on alldrug nights (IDN, 4DN, and 7DN,

P

<

0.024, 0.018 and0.012, respectively), as did TRI (

P

<

0.001, 0.005 and0.006, respectively). By contrast, FNZ produced astronger effect on sigma only on the IDN, but producedbeta enhancements that exceeded the enhancement ofsigma activity on 4DN and 7DN (IDN, 4DN, and 7DN,

P

<

0.004, 0.004 and 0.009, respectively).

Withdrawal nights

Figure 2 shows that the effects of HAX were still evi-dent in EEG spectral power on 4WN. Lower frequen-cies (0.5–9.5 Hz on 2WN; 0.5–8.5 Hz on 4WN) duringNREM sleep were reduced significantly relative toBLN, while higher frequencies (10.5–12 Hz, 14.5–40 Hzon 2WN; 22–25 Hz, 32–40 Hz on 4WN) were enhancedsignificantly. During REM sleep, power in the samelower frequencies was also reduced significantly andhigher frequency power (15–40 Hz on 2WN; 32–40 Hzon 4WN) was significantly enhanced.

The power spectra of NREM sleep after withdrawalfrom both FNZ and TRI had returned to baseline lev-els by 2WN, with the exceptions of small but significantelevations in 6–6.5 Hz, 8–10.5 Hz and 13–13.5 Hz powerduring withdrawal from TRI, and power in 12.5–13 Hz,14–17 Hz and 19–40 Hz after withdrawal from FNZ.During REM sleep, power had returned to baselinelevels by 2WN after withdrawal from TRI. However,during REM sleep a significant reduction in theta (5.5–6 Hz) spectral power and a significant enhancement inthe beta (32–40 Hz) band persisted on the second nightof withdrawal from FNZ.

Comparison of three drugs’ NREM sigma and beta enhancements

As shown in Fig. 3, during NREM sleep, all three BDZdrugs produced marked enhancements in the sigmaband, which were not observed in REM. On the IDN(top panel), there were no significant differencesbetween the three drugs. When the average peak val-ues of sigma (11–12.5 Hz) and beta (23–25 Hz in FNZand TRI, and 26–29 Hz in HAX) band in NREM werecompared on the 4DN and 7DN, FNZ slightly but sig-nificantly increased beta more than HAX (

P

<

0.04)and TRI (

P

<

0.009), while TRI increased sigma morethan FNZ (

P

<

0.005).

11

We did not find any significantdifferences in sigma band enhancement between HAXand TRI, nor between HAX and FNZ.

DISCUSSION

We studied the effects of long-, intermediate- andshort-acting BDZs on the composition of human sleepEEG. All the BDZs in the present study decreasedNREM and REM sleep EEG power in lower frequen-cies and increased power in higher frequencies, whichis consistent with previous studies.

11,13,16–18

In addition,new observations were made in the present study. Dur-ing NREM sleep, the enhancement of beta band activ-ity was found to increase across nights afteradministration of HAX and FNZ but not of TRI. Nosignificant changes in beta activity were seen on thethree TRI drug nights. Unlike the progressive increasesin beta activity across drug nights, sigma enhancementpeaked immediately on the IDN for each BDZ and didnot change significantly across the drug nights, exceptHAX which showed significances when 4DN and 7DNwere compared with IDN. Thus, sigma enhancementappeared less related to plasma accumulations of theseBDZ hypnotics, which is consistent with previousreports.

19,20

Although FNZ produced a distinctlypeaked large enhancement in beta activity duringREM periods as we previously reported

11

it was not

BDZ effects on sleep EEG 101

significantly greater than that produced by HAX orTRI.

The present findings may indicate that the physio-logical system(s) reflected by sigma enhancement man-ifest a particular sensitivity to BDZ derivatives, whichdiffers from the systems underlying the production ofthe lower EEG frequencies (which were uniformlysuppressed) and the beta band (which was enhanced).In our previous report, we suggested that such changes

reflected different mechanisms for the generation ofsigma (spindle waveform) and beta band EEG wave-forms.

12,21

Animal studies have shown that spindle gen-eration originates in the reticular nucleus of thethalamus and that the rhythmic activity is conducted tothalamo-cortical neurons and then to pyramidal neu-rons in the cortex.

22

The present results suggest thisreticulo-thalamo-cortical circuit differs from otherGABA-A related systems in being more sensitive to

Figure 1.

Changes in NREM(left column) and REM (right col-umn) sleep EEG power spectraafter taking haloxazolam (top),flunitrazepam (middle) and triaz-olam (bottom). The ratios of activ-ity relative to baseline night forwhole-night recordings are dem-onstrated. Significant differences(

P

<

0.05) are identified by a barbelow each graph. Paired two-way

t

-tests were performed on the fre-quencies that had shown signifi-cance (

P

<

0.05) in a preliminary

ANOVA

.

X

-axis, frequency (Hz).

Y

-axis, spectral power relative tothe baseline night.

102 X. Tan

et al

.

BDZs and in saturating rapidly, since the spindle activ-ity enhancement peaked following the initial dose ofBDZs. The most probable site of this BDZ actionwould be the GABA-A receptors on thalamic neuronsin the reticulo-thalamic synaptic connections. By con-trast, the systems responsible for the general trend oflow frequency reduction and high frequency enhance-ment could be less sensitive and thus producing dose-dependent responses.

The greatest enhancement of beta band activitywas produced by FNZ. A postmortem study of 3H-Flunitrazepam binding in human brains indicatedthat the 50% inhibitory concentrations (IC

50

) ofnitrazepam, FNZ and TRI in the temporal cortex were

24.8, 5.9 and 1.7 nM, respectively.

23

Since there havebeen no such reports for HAX, we estimate that HAXbinding is similar to that of nitrazepam, given the sim-ilarity of their clinical doses. Thus, it is possible that theCNS effects of 4 mg FNZ were stronger than both10 mg HAX and 0.5 mg TRI. This might explain whyFNZ showed the highest beta peak of the three BDZsin the present study.

On withdrawal nights, visual sleep staging indicatedthat total sleep time and sleep efficiency bothdecreased significantly, while wake time significantlyincreased on 2WN; that ‘rebound insomnia’ was notsignificant during withdrawal from TRI. In computeranalysis, only HAX effects persisted until 4WN; while

Figure 2.

Changes in NREM (left col-umn) and REM (right column) sleepEEG power spectra during withdrawalfrom haloxazolam (top), flunitrazepam(middle) and triazolam (bottom). Theratios of activity relative to baselinenight for whole-night recordings aredemonstrated. Significant differences(

P

<

0.05) are identified by a bar beloweach graph. Paired two-way

t

-testswere performed on the frequenciesthat had shown significance (

P

<

0.05)in a preliminary

ANOVA

.

X

-axis, fre-quency (Hz).

Y

-axis, spectral powerrelative to the baseline night.

BDZ effects on sleep EEG 103

both TRI and FNZ showed minor residual effects by2WN, and those returned to BLN by 4WN. These with-drawal differences could relate primarily to the drughalf-lives, as pointed out by Saletu

et al

.

24

Overall,however, the visually evident rebound insomnia couldbe hardly detected in alterations of the EEG by thepresent analysis. In summary, when continuouslyadministered, HAX, FNZ and TRI increased sigmaand beta in different patterns. Such differences areconsidered to be caused by the pharmacological char-acteristics of each drug and given dose.

Although we still do not know how these patterns ofspectral EEG changes relate to the clinical features ofeach drug, this non-invasive method and these analyses

are potentially useful for the evaluation of new hyp-notics. Considering that most CNS drugs modify neu-ronal activities by acting on neuronal receptors, andthat the EEG represents the sum of electrical activitiesof multiple neurons, sensitive monitoring of pharma-cological effects in the human brain could be achievedusing EEG frequency analysis. However, pharmaco-logical effects on EEG frequency bands differ depend-ing on brain states, as the present and our previousstudy

11

clearly indicate. Thus, spectral alteration byBDZ drugs differs in NREM and REM. In pharmaco-EEG studies, vigilance control has been recognized asan important issue. In sleep recordings, this problemcould be well controlled because subjects are sleeping.In studies of hypnotics, sleep state information is essen-tial because the profiles of hypnotic drugs should beevaluated for the conditions they are intended toaffect. The accumulation of such knowledge wouldcomplement the evaluation of clinical characteristics ofhypnotic drugs.

REFERENCES

1. Duka T, Holt V, Herz A. Occupation by benzodiaz-epines and correlation with pharmaological effect.

BrainRes.

1979;

179

: 147–156.2. Greenblatt DJ, Miller LG, Shader BI. Neurochemical

and pharmacokinetic correlates of the clinical ation ofbenzodiazepine hypnotic drugs.

Am. J. Med.

1990;

88

:18S–24S.

3. Greenblatt DJ. Benzodiazepine hypnotics: Sorting thepharmacokinetic facts.

J. Clin. Psychiatry

1991;

52

(9Suppl.): 4–10.

4. Breimer DD. Pharmacokinetics and metabolism of vari-ous benzodiazepines used as hypnotics.

Br. J. Clin. Phar-macol.

1979;

8

: 7S–13S.5. Roth T, Roehrs TA. Issues in the use of benzodiazepine

therapy.

J. Clin. Psychiatry

1992;

53

: 14–18.6. Hommer D, Weingartner H, Breier A. Disociation of

benzodiazepine induced amnesia from sedation by flu-mazenil pretreatment.

Psychopharmacology

1993;

112

:455–460.

7. Mendelson WB. Clinical distinctions between long-acting and short-acting benzodiazepines.

J. Clin. Psychi-atry 1992; 53: 4–7.

8. Muraoka M, Tada K, Nogami Y, Ishikawa K, Nagoya T.Residual effects of repeated administration of triazolamand nitrazepam in healthy volunteers. Neuropsychobiol-ogy 1992; 25: 134–139.

9. Tazaki T, Tada K, Nogami Y, Takemura N, Ishikawa K.Effects of butoctamide hydrogen succinate andnitrazepam on psychomotor function and EEG inhealthy volunteers. Psychopharmacology (Berl) 1989; 97:370–375.

10. Misaki K, Nakagawa H, Koshino Y et al. Effect of fluni-trazepam on sleep and memory. Psychiatry Clin. Neuro-sci. 1998; 52: 327–332.

Figure 3. Changes in NREM sleep EEG power spectra(ratios relative to BLN) of haloxazolam, flunitrazepam andtriazolam on IDN (top), fourth (middle) and seventh drugnight (bottom). The peaks of sigma and beta band activityamong the drugs, which were significantly different fromother drugs (P < 0.05), are marked by asterisks (*). X-axis,frequency (Hz). Y-axis, spectral power relative to the base-line night. See text for detailed explanation.

104 X. Tan et al.

11. Uchida S, Okudaira N, Nishihara K, Iguchi Y. Fluni-trazepam effects on human sleep EEG spectra: Differ-ences in NREM, REM and individual responses. Life Sci.1996; 58: PL199–205.

12. Uchida S, Okudaira N, Nishihara K, Iguchi Y, Xin T.Flunitrazepam effects on human sleep EEG spectra II.Sigma and beta alterations during NREM sleep. Life Sci.1996; 59: PL117–120.

13. Tan X, Uchida S, Matsuura M, Nishihara K, Iguchi Y,Kojima T. Benzodiazepine effects on human sleep EEGspectra: a comparison of triazolam and flunitrazepam.Life Sci. 1998; 63: 675–684.

14. Rechtschaffen A, Kales A. A Manual of StandardizedTerminology, Techniques and Scoring System for SleepStages of Human Subjects. Washington, DC: PublicHealth Service U.S. Government Printing Office, 1968.

15. Feinberg I. Changes in sleep cycle patterns with age. J.Psychiat. Res. 1974; 10: 283–306.

16. Itil TM, Saletu B, Marasa J. Determination of drug-induced changes in sleep qulity based on digitalcomputer ‘sleep prints’. Pharmakopsychiatr. Neuropsy-chopharmakol. 1974; 7: 265–280.

17. Borbely AA, Mattmann P, Loepfe M, Strauch I,Lehmann D. Effect of benzodiazepine hypnotics on all-night sleep EEG spectra. Hum. Neurobiol. 1985; 4: 189–194.

18. Gevins AS, Stone RK, Ragsdale SD. Differentiating theeffects of benzodiazepines on NREM sleep EEG spec-tra. Neuropsychobiology 1988; 19: 108–115.

19. Johnson LC, Spinweber CL. Effects of short-acting ben-zodiazepine on brain electrical activity during sleep.Electroenceph. Clin. Neurophysiol. 1981; 52: 89–97.

20. Hirshkowitz M, Thornby JI, Karacan I. Sleep spindles:Pharmacological effects in humans. Sleep 1982; 5: 85–94.

21. Uchida S, Maloney T, Feinberg I. Sigma (12–16 Hz) andbeta (20–28 Hz) EEG discriminate NREM and REMsleep. Brain Res. 1994; 659: 243–248.

22. Steriade M, Domich L, Oakson G. The deafferentedreticular thalamic nucleus generates spindle rhythmicity.J. Neurophysiol. 1987; 57: 260–273.

23. Kobayashi T, Kiuchi Y, Shimizu H, Takeuchi J, Ogata H,Toru M. Studies on benzodiazepine recepto in humanpost-mortem brain. Ann. Rep Pharmacopsychiat. Res.Found. 1989; 20: 113–120.

24. Saletu B, Anderer P, Brandstatter N et al. Insomnia ingeneralized anxiety disorder: polysomnographic, psycho-metric and clincal investigations before, during and aftertherapy with a long- versus a short-half-life benzodiaz-epine (quazepam versus triazolam). Neuropyschobiology1994; 29: 69–90.