The effects of flurazepam hydrochloride on brain electrical activity during sleep

Electroencephalography and Chnical Neurophysmlogy , 1979, 47. 309--321 © Elsevier/North-Holland Scientific Publishers, Ltd.

309

THE E F F E C T S OF F L U R A Z E P A M H Y D R O C H L O R I D E ON BRAIN ELECTRICAL ACTIVITY D U R I N G SLEEP l L.C. JOHNSON, D.M SEALES 2, p. NAITOH, M.W. CHURCH 3 and M. SINCLAIR Naval Health Research Center, P O. Box 85122, San Dzego, Cahf 92138 (U.S.A.) (Accepted for publicatLon. December 14, 1978)

One of the well-established effects of flurazepam (Dalmane) on sleep EEG is a reduction of delta activity and, consequently, a reduction in stage 4 sleep (Greenblatt et al. 1975; Kay et al. 1976). Using a modified period analysis, Feinberg et al. (1977) found that nightly administration of 30 mg flurazepam m 4 subjects over 7 nights resulted in significant decreases in integrated amplitude, number of delta waves, and delta time when averaged over 20 sec epochs. They reported, however, that the total number of delta waves/night was n o t reduced by flurazepam because of increased delta activity in stage 2. As anticipated, flurazepam reduced stage 4 time. They also observed a significant increase in the rate of sleep spindle bursts, which has been previously noted by Johnson et al. (1976}. Feinberg and his associates did n o t report the changes in delta waves from one night to the next as flurazepam administration continued. As a part of a larger study on the effects of

Subjects

1 This study was supported in part by Hoffmann-La Roche Inc , Nutley, N.J., and by Department of the Navy, Bureau of Medicine and Surgery, under Work Unit M0096-PN.001-1029. The views presented in this paper are those of the authors No endorsement by the Department of the Navy has been given or should be mferred. 2 Present address: Naval Aerospace Medmal Research Laboratory Detachment, New Orleans, La., U S.A. 3 Present address: Oklahoma Center for Alcohol and Drug Related Studms, Umverslty of Oklahoma Health Sciences Center, Oklahoma City, Okla., U.S A

Subjects were 12 male poor sleepers (mean age 2 1 . 3 + 1 . 0 years) and 12 male good sleepers (mean age 21.2 + 1.2 years) selected on the basis of EEG and subjective criteria. Subjective criteria included responses to a questionnaire designed to evaluate an individual's estimate of his sleep quality, followed by personal interview. To qualify as a poor sleeper, subjects had to rate their sleep quality as 'poor' or 'very p o o r ' and indicate a usual sleep latency greater than 30 min, or more than 30 min of awake time after sleep onset. To meet EEG sleep criteria, poor sleepers had

flurazepam on sleep, arousal thresholds, m o o d and performance, this study examined the changes in delta activity of poor sleepers over 10 nightly administrations of 30 mg flurazepam. This study also closely re-examined Feinberg et al.'s observation that flurazepam did not change the total number of delta waves over a night. Though the focus of the study was on delta activity, sleep spindle activity and the evoked K-complex were also examined. Skinner and Shimota (1975) have provided suggestive evidence that flurazepam reduces the amplitude of the cortical P2, N2 and P3 components evoked by tone-burst stimulation during stage 2 and stage 3 sleep. No confidence intervals were reported m association with their findings.

Methods

310 to exhibit sleep latencles of 30 min or more on each of 2 consecutive screening nights or 30 min of awake time after sleep onset. All poor sleepers met the sleep latency criterion but no subject met the awake time after sleep onset criterion. During screening nights, average sleep latency was 57.9 + 26.1 min. Average sleep latency for good sleepers was 11.1 ± 3.6 min. All subjects had more than 5% of their total sleep time (TST) in stages 3 + 4. Total bed time was 7.5 h. Subjects were screened for possible psychiatric conditions, sensitivity to benzodiazepines, alcohol or drug abuse, and recent illnesses. A few subjects were not accepted because of a recent illness requiring medication. All subjects were in good health and denied use of any type of sleep medication or other drugs. On screening nights, to insure that poor sleep was not due to nocturnal myoclonic jerks or sleep apnea, electromyogram (EMG) recordings were made from the left tibialis and respiration rate was recorded from an abdominally placed strain gauge. No potential subject was observed to have nocturnal myoclonus or sleep apnea. All subjects were informed about the general nature o f the experiment and willingly signed Informed Consent and Privacy Act statements. All subjects were asked to refrain from napping and taking drugs or alcohol durmg the course of the study. Breath analyzer and urine tests, used aperiodically, indicated no significant use of alcohol or other drugs during the study. During screening mghts, 21 possible poor sleepers were rejected because of sleep latencies less than 30 min. One subject was dropped from the study due to complaints of excessive drowsiness.

Procedure Following the 2 screening nights, subjects received placebos in a single-blind paradigm for 7 consecutive baseline nights. The good sleepers were terminated at this point. Following the baseline nights, 6 poor sleepers received 30 mg flurazepam for 10 additional

L.C JOHNSON ET AL nights while the other 6 continued to receive placebo in a double-blind paradigm. The placebo or drug capsules were given at 21.45 h each night. Subjects were put to bed at 22.00 and awakened at 05.30. Only the first 2 drug nights were not recorded. Although subjects did not sleep in the laboratory on these 2 nights, they were mstructed to take the medication 15 min before bedtime and to keep the same 2 2 . 0 0 - 0 5 . 3 0 sleep schedule. Two to 3 weeks after the final drug mght, 3 placebo follow-up mghts were recorded on all subjects. Each subject slept in an electrically shmlded, air-conditioned room with soundproofing. All electrophysiological variables were recorded on a 12-channel Beckman Type R dynograph. The electro-oculogram (EOG) was recorded from biopotential electrodes placed on the outer canthus of each eye and referenced to linked mastoids (A~ + A2). The EEGs were obtained by use of silver chlorided disc electrodes from C3 and C4 electrode placements referenced to linked mastoids. Both EOG and EEG time constants were 0.3 sec. Sleep stages were determined according to standard criteria (l~echtschaffen and Kales 1968).

Off-hne delta analysis The EEG activity analyzed for delta was recorded on both paper and FM instrumentah o n tape, the latter on a Hewlett-Packard 380 FM recorder at 1~ in./sec. The EEG recorded tape was band-passed filtered at 0.2--50 c/sec. Data were obtained on the 2nd and 3rd nights of the baseline phase, the 3rd, 5th, 7th, and 9th nights of the treatment phase (drug or placebo nights), and the 2nd follow-up night. During these nights, the subject's sleep was undisturbed. During the other nights, the subjects were either aroused from sleep to measure arousal threshold with or without flurazepam (Johnson et al. 1979), or exposed to soft clicks to obtain event-related auditory potentials (see below). Analysis of delta activity was done on a PDP-12 computer using a modified period

FLUAZEPAM AND EEG ACTIVITY DURING SLEEP

311

(zero~ross) program. The off-line c o m p u t e r analysis was limited to the first 4 h of sleep. This period was chosen because most slow wave sleep (SWS} occurs during the first half of sleep, and all of this sleep was on one FM tape and could be easily synchronized with paper write-out for editing of b o d y movements and categorization by sleep stages. The taped EEG data were played back 8 times faster than real time. During playback, the EEG was filtered with an analog filter that passed all frequencies above 0 . 5 c / s e c to remove DC offset. The program detected time of up~rossing (time when EEG voltage changed from positwlty to negativity). The period of one full cycle wave was time elapsed between t w o up~rossings and zero-voltage baseline. When the waves fell within the period from 500 msec (2 c/sec) to 2000 msec (0.5 c/sec) with a peak-to-trough amplitude of at least 10 #V, they were tagged as delta waves. For each delta wave, peak amplitude (difference between peak and trough} was measured. Church et al. (1975) as well as Johnson et al. (1969) found that most of the variation in SWS, as a subject goes from awake to asleep and through the stages of sleep, occurs in the 0.5--2 c/sec band. The o u t p u t was grouped into an analysis epoch of 1 min. For each 1 min epoch, 4 measures of delta activity were calculated: (1) average peak amplitude in pV, (2) total number of delta waves, (3) total time occupied by the delta waves in seconds, and (4) average frequency. The average frequency was calculated by dividing the total number of delta waves/epoch by total time occupied b y delta waves/epoch (Smith et al. 1975). These data were stored in digital format on magnetic tape. To synchronize stage information with period-analysis output, the first stage scored by human scorer for each night was matched to the first period-analysis o u t p u t of that night. Since the period analysis was based on 1 min epochs while staging of sleep data was for every 30 sec epoch, each period-analysis o u t p u t was labeled by t w o stages, the first

one to correspond to the first 30 sec, and the second one to the second 30 sec. For example, one period-analysis o u t p u t could be labeled by stages 2 and 3, 2 and REM (rapid eye m o v e m e n t sleep}, or 3 and 3. Only those period-analysis outputs which had the same stage (e.g., 3 and 3) were retained for further analysis. To edit o u t b o d y movements, the average delta amplitude of a given subject for a given sleep stage over a night was converted to a standard score, and epochs whose average delta amplitude had a standard score of 3 or greater were identified as outliers and removed from further statistical analysis. No attempt was made to differentiate K-complexes from delta waves.

On-hne EEG analysis Detection of delta half-waves (0.5--2 c/sec), alpha bursts (8--12 c/sec), and sleep spindle bursts (11.75--15 c/sec) was accomplished on-line by means of the Smith phasic EEG detector (Smith et al. 1975}. The detector o u t p u t was counted by the detector's digital counters and printed out in 2 min epochs b y a printer. At the end of the sleep period, the total n u m b e r of delta half-waves, and the number of sleep spindles and alpha bursts were available both from the counters and the printed output. Procedure for evoked K-complex Poor sleepers received click stimulation on the last night of the baseline, drug, and follow-up portions of the experiment (nights 7, 17, and 20). G o o d sleepers received click stimulation on the 2nd and 7th placebo baseline nights. In the group of good sleepers, on 1 of these 2 nights (order counterbalanced) clicks were presented in synchrony with spindle bursts. Spindle-synchronous averaged evoked K-complex (AEK) data are discussed elsewhere (Church et al. 1978). Following sleep onset, clicks (10 msec duration) were presented once every 32 sec, except when the EEG indicated transitory waking or stage 1. Click intensity was kept

312

relatively low (44 dB SPL) so as not to disturb sleep. Because our subjects had recently passed naval entrance hearing examinations, hearing was tested only in terms of subjective threshold for a 4 sec, 1000 c/sec tone presented on other nights of the study. Presleep thresholds varied between 36--38 dB SPL. Clicks were generated and timed by means of BRS Fonnger Logic equipment, and were presented through a University Sound, Model MCL, speaker located approximately 46 cm over the subject's head as the subject lay in bed. EEG was recorded from C3 referenced to linked mastoids, using the same electrodes, dynograph, filters, and tape system described earlier. Off-line analysis consisted first of sleep-stage scoring according to the m e t h o d of Rechtschaffen and Kales (1968). Then, AEKs were obtained for each consecutive period of stage 2 and stage 3--4 sleep for an entire mght. This allowed visual inspection of wave form stability in AEKs comprised of different numbers of trials and collected over the course of the 7.5 h sleep time. Finally, for each night an overall stage 2 AEK and an overall stage 3--4 AEK were derived by means of a weighted combination of the consecutive stage 2 and stage 3--4 AEKs. AEKs used in statistical comparisons contained relatively equivalent numbers of trials (see legend, Fig. 3). AEKs were quantified in terms of peak latencies of major components and peak-totrough amplitude of the major negative-topositive biphasic complex. AEKs of 6 randomly selected good sleepers and 11 poor sleepers were compared by means of betweensubjects t-tests. This comparison involved AEKs from night 7 (baseline) of the poor sleepers and from either night 2 or 7 of the good sleepers, whichever night was periodic (as opposed to spindle-synchronous) stimulation. Both within- and between-subjects analyses were carried out on AEKs of poor sleepers to evaluate the effect of flurazepam. Nights 7 and 17, and nights 7 and 20 were compared within subjects. A score indicating percentage change in AEK amplitude from night 7 to night 17 was derived and con-

L(" JOHNSONETAI,

trasted between 'placebo' (N = 5) and "drug' (N -- 6) poor sleepers. One placebo record was technically inadequate. All statistical tests for delta, spmdle, and AEK analyses were evaluated at the 0.05 level and were one-tailed, unless otherwise stated. Significant results were checked by means of comparable non-parametric stahstics

Results On-line analysis results indicated there were no significant differences in delta, alpha or spindle activity between good and poor sleepers during baseline sessions. Analysis by t-tests also indicated no significant difference in baseline activity between subjects who later received flurazepam and those who continued to receive placebo capsules, for on-line alpha, delta, or spindle counts, or for any of the delta values obtained off-line by computer analysis. The 2 baseline nights also did n o t differ significantly for any measure for either within- or between-groups comparisons. The average of these 2 nights was thus used for the baseline value, B. The recovery night values were similar to baseline mean values but were not added to the baseline values because 2 subjects, 1 drug and 1 placebo, had unusually poor sleep on their recovery night.

Effect of flurazepam Between groups. For this comparison, the data over the 4 treatment nights (3rd, 5th, 7th, and 9th) were averaged over each subject for each measure. These averaged values were then used for comparison by one-tailed t-tests with the averaged baseline values. Since there was no change in the period (frequency) of delta during the study for either group, changes in delta time and delta count were highly correlated. To reduce the n u m b e r of statistical comparisons reported, delta time was omitted. Also, delta count was obtained by both on-line and off-line analysis techniques.

FLUAZEPAM AND EEG ACTIVITY DURING SLEEP

313

TABLE I Treatment-basehne difference scores for flurazepam (N = 6) and placebo (N -- 6) groups. Negative means indicate a decrease from baseline Variable

Off-hne computer analysis Delta amphtude (pV) SWS Stage 2 REM Delta count/min epoch SWS Stage 2 REM

Flurazepam

Placebo

Mean diff.

Mean diff

S.D.

--27 0-+8.2 --17.0 -+ 6.9 --7.8 -+ 7.2 --4.5 -+ 1.2 --4.4 + 2 6 --1 8-+4.0

t (10)

P

5 0 + 8.2 --2.6 + 8.5 --2 6 -+ 11 1

6257 3 318 0 966

<0.001 <0.005 N.S.

2.0 + --1.0 + 12-+

6.806 2.063 1.754

<0.001 <0.05 N.S

S.D.

2.0 30 1.4

to analyze these data by sleep stages. As in the off-line analysis, the average baseline/ treatment between-groups difference scores were tested for significant differences by onetailed t-tests. The results indicated a significant decrease in delta count and a significant increase in spindle bursts in the flurazepam group. The baseline, treatment, and follow-up data are presented in Table II. Since the off-line c o m p u t e r analysis of delta activity examined only the first half of sleep, the number of delta half-waves tallied b y the on-line detector was divided into the first and second halves of sleep and then compared to appropriate baseline values. Compari-

The results of the between-groups baseline/ treatment difference score analyses are presented in Table I for the off-line technique. The off-line computer analysis indicated a significant decrease in delta amplitude and in delta c o u n t during both stage 2 and SWS. For the flurazepam group, though there was a decrease in delta activity during REM sleep, there was no significant change in either amplitude or c o u n t during REM sleep. The most robust changes were seen during SWS. On-line analysis. By means of the Smith phasic EEG detector, delta half-waves, sleep spindles, and alpha waves were counted for the total sleep period. No a t t e m p t was made

TABLE II Mean sleep spindle and delta counts during sleep for basehne, treatment, and follow-up (from on-line Smith phasic EEG detector) * Variable

Spindles (no. of bursts) Delta (no. of half-waves)

Basehne

Flurazepam Placebo Flurazepam Placebo

Treatment

Follow-up

Mean

S.D

Mean

S.D.

Mean

S.D

894 1,055 14,315 9,273

(382) (343) (5,187) (3,637)

1,538 ** 1,011 9,977 ** 9,731

(554) (349) (4,716) (3,496)

908 1,148 11,821 8,317

(357) (380) (7,567) (4,130)

* R E M time was comparable for the two groups. ** Flurazepam-induced change significant at 0.001 level, one-tailed.

314

L.C J O H N S O N E T A L

son of between-groups difference scores indicated there was a significant decrease in delta count for both halves of the night for the flurazepam group. The increase in sleep spindles was also significant for each half of the night for the subjects receiving flurazepam. Within-group off-hne analysis. Visual inspection of the data indicated that delta amplitude decreased more rapidly during the first half of the 10 day drug period than during the latter nights. In contrast, delta count decreased at a slower but more even rate (see Figs. 1 and 2). To investigate these differing trends in the off-line data, the baseline and drug night SWS and stage 2 data for both amplitude and count were analyzed by oneway ANOVAs, followed by Duncan's multiple range test. The ANOVA results showed significant linear trends• There were also s]gnifiI10 ~ws I0o

90

@

40

'~

BO

"~.

-o

c~ "-...

a6

:E

7O

3O

\

Q.

E

35

eO c

5lOli 2

o ~,J

7O

~o

\

o

25

60

\

o

o

•--.~.

-¢D

5O

+ 5°f 40

::'o

;'5 ~

;'0

f

15

30

. . . . . . II

D|

D5 D7

,~ D9

. . . . . . It

I

D$

DS D7

,~ D9

R

N~ghl

Fig 1. Average delta actwity during first half of night for REM, stage 2 and SWS for flurazepam subjects. A linear decrease in delta amplitude occurred over flurazepam nights for all stages and in delta counts for stage 2 and SWS. B, baseline; D3--Dg, drug nights; R, recovery (follow-up)

CL

c \

\ 3U

400 "0

o u

£

Z5

~,

c3

u'l Q"

\

a ~0

300 ZOO

, e

,

,

,

, ,,ipi

D3 DS D~ D9

N~ght

R

g

D3 0 5

D7 D9

R

Night

Fzg 2 Average count of delta half-waves and sleep spindles during all-mght sleep for flurazepam subjects A hnear decrease in delta waves and a hnear increase in sleep spindles occurred over flurazepam nights B, baseline; D3--D9, drug nights, R, recovery (followup)

cant quadratic components in the decrease in delta amplitude but they accounted for only 3% of the variance during SWS and 5% during stage 2. A non-parametric rank-order correlation procedure (Lubin 1961) also indicated a significant linear trend for both delta amplitude and delta count during stage 2 and SWS. Because of the small number of points and the unequal intervals between points between baseline and the 3rd drug night, the change over nights was further analyzed by the Duncan multiple range test. Delta counts had n o t decreased significantly until the 7th and 9th drug nights for SWS and stage 2, respectively. The multiple range test indicated that, for both stage 2 and SWS, delta ampli, tude had decreased significantly from baseline by the 3rd drug night• So, even though flurazepam decreased both delta amplitude and number with repeated use during both stage 2 and SWS, the drug's suppression of delta amplitude appeared to be the more dramatic and rapidly occurring of the two and appeared to be leveling off by the end of the 10 day period. On-line analysis• For the on-line detector data, parametric and non-parametric statistics indicated a significant linear decrease in delta count and a significant linear increase in spindle count for the total sleep period across the 10 day drug period (see Fig. 2). Duncan's

FLUAZEPAM AND EEG ACTIVITY DURING SLEEP

315 Night 7 - Night 17 . . . . . .

multiple range test indicated that the number of delta half-waves had decreased significantly from baseline by drug night 5. Sleep spindles had significantly increased above baseline by the 5th drug night.

Stage 2

Evoked K-complex There were no significant differences between AEKs of good and poor sleepers with respect to peak-to-trough amplitude or peak latencies of major components. Though no statistical evaluation was done, visual inspection of the all-night sleep records indicated no difference in number or amplitude of spontaneous K-complexes for the two types of sleepers.

l

~~

Stage 3-4 I

A

5uV

B

Fig. 3. Composite AEKs during stages 2 and 3--4 from nights 7 and 17 for those poor sleepers who received placebo on both nights (section A), and for those who recewed flurazepam on night 17 (section B). The average number of trials per AEK was 1308 and 1504 for mghts 7 and 17 for placebo poor sleepers stage 2 , 5 1 0 and 540 for placebo stage 3--4; 1650 and 2689 for flurazepam stage 2 , 9 9 1 and 610 for flurazepam stage 3--4

Flurazepam Within group. In Fig. 3 are composite AEKs from nights 7 (baseline) and 17 (treatment) for those poor sleepers who received only placebo (section A), and those who received flurazepam on treatment nights {section B). Statistical analysis is presented in Table III. In the placebo group of poor sleepers, AEK amplitude (peak to trough) did not differ on nights 7 and 17. In the flurazepam group, however, AEK amplitude was significantly reduced on night 17 (10th drug night). AEK amplitude recovered on night 20, 2--3 weeks post-treatment for the flurazepam poor sleepers. Flurazepam did not perceptibly affect latencies of AEK components. Between groups. Percentage changes in

AEK amplitude from nights 7 to 17 were calculated during stages 2 and 3--4 for both the placebo and flurazepam groups of poor sleepers. The flurazepam group showed significantly greater percentage decrements (stage 2 : - - 4 1 + 20%; stage 3 - - 4 : - - 3 1 + 26%) than did the placebo group (stage 2 : + 1 0 + 23%; stage 3--4: --1 + 25%) (t(9) = 3.936; 1.952; respectively). Further visual analysis of the EEG traces from the 6 poor sleepers who received flurazepam indicated that the decrease in amplitude of AEKs resulted from

TABLE III The effect of flurazepam mean (+ S.D.) amplitudes (#V) of AEKs (peak to trough) from nights 7 and 17 of poor sleepers, both placebo (N = 5) and drug (N = 6) groups. Group

Placebo t (4) Flurazepam t(5)

Stage 2. Night

Stage 3--4. Night

7

17

7

17

69 + 18

75 -+ 20 1.310 44 -+ 40 --2.759 *

112 + 33

108 +- 30 ---0.344 62 -+ 45 --2.639 *

78 + 63

94 -+ 63

* Significantly different from night 7 at or beyond 0.05

316

L.C JOHNSON ET AL_

Placebo

Discussion ~ ,,

Jj

ib

Flurazopam B

o t~4

3 sec Fng. 4 K-complexes during stage 2 elicited by consecutive click stimuli (arrows) at comparable times during the placebo (upper) and flurazepam (lower) nights of one poor sleeper

decreases in both the amplitude and the rate of elicitation of K-complexes in response to individual click stimuli. Rate of elicitation dropped from 50 to 33% in stage 2 (t(5) = 4.213; two-tailed) and from 55 to 39% in stage 3--4 (t(5) = 4.489; two-tailed). In Fig. 4 are stage 2 K-complexes from consecutive stimuli at comparable times of the night during the placebo and flurazepam nights of one poor sleeper. K-complexes observed in association with the administration of flurazepam were usually smaller in size and fewer in number than those observed on placebo nights. The reduction in amplitude of K~omplexes was more pronounced in some subjects than in others, and was more evident during the latter half of the night. Since flurazepam increased the arousal thresholds of our poor sleepers, in relation to the placebo group of poor sleepers (Johnson et al. 1979), the relation of AEK to arousal threshold and to increase in arousal threshold was investigated. Correlations of overall baseline arousal levels with baseline AEK amplitudes were close to zero (stage 2: 0.06; stage 3--4: 0.06; N = 17). In addition, correlations of percentage change in arousal threshold from nights 7 to 17 with percentage change of AEK amplitude also were non-significant stage 2: --0.46; stage 3--4: - 0 . 3 7 ; N = 6).

The results of this study indicate that the reduction of stage 4 during the early nights of sleep with flurazepam is due primarily to the decrease in delta amplitude, but this decrease is not limited to SWS as a decrease m delta amplitude was also significant during stage 2 sleep for both between- and within-group analyses. There was also a decrease in delta amplitude during REM sleep, but this decrease was not significant. Reduction in delta activity was not confined to SWS. While the SWS delta amplitude decrease was the most marked response on early drug mghts, with continued drug use (over 5--7 mghts), a reduction in the number of SWS delta waves occurred and this reduction may also have contributed to the reduced scoring of stage 4 sleep. The decrease in delta count across consecutive drug mghts was more gradual than that for amplitude and did not show the same leveling off seen with amplitude during the latter drug nights. By use of the Smith phasic EEG detector, the number of delta waves was found to decrease during the total sleep period and, by the 5th drug night, this total nightly decrease was significantly different from baseline. Inspection of the changes in delta waves by stages of sleep during the first 4 h of sleep indicated that for SWS the number of delta w a v e s / l m i n epoch had decreased significantly by the 7th drug night. The on-line detector delta count decrease for the first half of the night was also significant on the 7th drug night. Computer analysis of stage 2 slow waves indicated that for the first 4 h of sleep the delta count/1 mln epoch decrease was significant but n o t as marked as that seen in SWS. The on-hne detector indicated that the delta count was also significantly reduced during the second half of the night which is composed primarily of REM and stage 2. Feinberg et al. {1977) examined the effect of flurazepam on delta activity over 7 nights in 4 medical students and also found that flurazepam decreased integrated amplitude,

FLUAZEPAM AND EEG ACTIVITY DURING SLEEP number of baseline crossings, and time in band when these measures were averaged for 20 sec epochs for NREM sleep (combined stages 2, 3 and 4). With respect to number of delta waves, their report is confusing. The nightly total number of delta waves and the time they occupied did not change with flurazepam, although their density (e.g. the number of slow waves/20 sec epochs) fell. The apparent conflict in the two statements by Feinberg and his colleagues might be due to their failure to control for TST. Thus, the increase in TST (an average of a b o u t 22 min) on drug nights could account for the failure to find a decrease in the total number of slow waves over the night. Feinberg et al. note that TST was significantly increased on drug nights. Femberg et al. (1977), however, used this apparent constancy of the total number of delta waves/night with flurazepam to account for a lack of substantial behavioral deficit after experimental stage 4 deprivation and flurazepam medication. They maintained that the decrease in the number of delta waves with stage 4 suppression by flurazepam was 'compensated by delta activity in increased stage 2 sleep' (p. 848). The results of our study do n o t agree with their interpretation. The total delta count/night was found to be significantly lower with flurazepam by means of the Smith phasic EEG detector and, also, the delta c o u n t for stage 2 was reduced for the first 4 h of sleep and probably during the entire night of sleep. When TST is comparable for baseline and drug conditions, the total number of delta waves/night will be reduced. Feinberg, noting the lack of research findings supporting the 'biological importance' of stage 4 sleep, suggests that their m e t h o d of analysis may help explain the discrepancy between the purported importance of stage 4 and the negative data (Johnson 1973). They imply that the total number of delta waves is the crucial factor. Other data from this research project do n o t support Feinberg's assumption that the number of delta waves, and n o t a m o u n t of stage 4 per se, has biologi-

317 cal significance. First, the decrease m stage 4 is highly correlated with decrease in delta count, r = 0.71, as well as delta amplitude, r = 0.77. Second, even though the per cent decrease in delta c o u n t ranged from almost no change, 0.2--53.1% over the total drug period, we found no relation between decrease in delta count and change in sleep quality or change in mood. There was also no significant relation between change in delta and the performance after morning awakening on tests of reaction time, cognitive activity, and immediate m e m o r y (Church and Johnson 1979). Also, even though stage 4 (the period of sleep generally referred to as deep sleep) was decreased, arousal threshold was found to be increased during flurazepam-induced sleep (Bonnet et al. 1979; Johnson et al. 1979). The biological function of stage 4 and the delta activity necessary to score it remain unknown. In addition to reduction in delta activity, there was also a decrease in number and amplitude of evoked K-complexes during NREM sleep in subjects receiving flurazepam. Flurazepam-induced reduction of AEK amplitude was predicted on the basis of evidence that the K-complex is a sign of transient cortical arousal (Roth et al. 1956; Sassin and Johnson 1968) and that flurazepam lowers cortical arousal or responsivity to stimulation. The observations of an increase in the intensity of stimulation required to arouse the poor sleepers who received flurazepam, and a decrease in AEK amplitude, lend credence to the view that the K-complex, in some manner, reflects the degree of transient cortical arousal or the responsivity of the cortex to stimulation. However, other data suggest that there is n o t a simple linear relation between changes in arousal threshold and magnitude of the Kcomplex. There was no correlation between arousal threshold and amplitude of AEK. Furthermore, the characteristic observation of larger amplitude AEKs in stage 3--4 where arousal thresholds are often highest in humans is contrary to the idea that the K-complex reflects the degree to which the cortex and

318 assocmted structures are 'aroused' by stimulation. These decreases in AEK by flurazepam are probably not due to depressed auditory neural transmission since the brain stem auditory evoked response is not affected by flurazepam in man or cat receiving therapeutic or toxin doses, respectively (Stockard et al. 1977). In evaluating the significance of the reduction of AEK amplitude following flurazepam administration, it is important to keep in mind that flurazepam also signifmantly reduces the amplitude of delta activity during sleep stages 2, 3 and 4. This is most evident in stage 4 sleep where delta is the predominant feature of the EEG. Since flurazepam markedly affects 'background' EEG, the question arises whether the lowering of AEK amplitude is merely a result of lowered delta activity. Though the decrease in delta amplitude is probably a factor m the smaller AEK, the likelihood that this mechanism accounts for the entire drop in AEK amplitude is slight. On the average, AEK amplitude decreased more in stage 2 where delta is less prevalent. In addition, there is evidence from other studies (Bond and Lader 1973; Hablltz and Borda 1973) that flurazepam reduces the amplitude of averaged evoked responses during waking when a reduction of delta amplitude would be of negligible influence. Finally, the correlation between percentage change in stage 3--4 delta amplitude from nights 7 to 17 and percentage change of stage 3--4 AEK amplitude was --0.87 ( N = 6 ) , indicating that during stage 3--4 those poor sleepers who showed the greatest decrement in delta amplitude showed the least decrement in AEK amplitude. The same correlation in stage 2 was +0.57 (N = 6). This significant and positive correlation was not surprising since K-complexes probably constituted a part of the computer-analyzed stage 2 delta activity. AEKs of good and poor sleepers did not differ significantly. Originally, we expected that poor sleepers might have larger amplitude AEKs since we were examining the K-complex as a sign of transient cortical arousal and

LC JOHNSONETAL, smce poor sleepers m general complmn of being 'light' sleepers easily awakened by noise. Many of our poor sleepers complained of a sensitivity to noise, particularly while trying to fall asleep. Objective evidence from our larger study {Johnson et al. 1979) verified the difficulty of our poor sleepers in falling asleep, but indicated also that, once asleep, sleep patterns and auditory arousal threshold levels of good and poor sleepers were indistinguishable. It is thus not surprising, in retrospect, that the AEKs of our poor sleepers did not differ from those of the good sleepers. The electrophysiological mechanisms or changes which result in the recorded decrease in delta and in the AEK are unknown. Assuming the delta waves result from synchronous, slow-potential fluctuations in neurons or dendrites, there is no firm evidence whether these potentials are purely cortical in origin and not controlled by subcortical pacemakers, as suggested by Nmtoh et al. (1971), or under the control of subcortmal pacemakers as suggested by Elul (1972). Feinberg et al. (1978) suggest that the decline in delta amplitude with age may reflect a decrease in the size of the neuronal pool, or a decrease in the magnitude of the average potential change of the individual generators within the pool. If such is the case, the changes seen with short-term use of flurazepam reflect a drug-induced inhibition rather than permanent change. Benzodiazepines have a marked inhibitory effect on neuronal activity (Sherwm 1971; Schallek et al. 1972}, and benzodlazepine receptor sites have been demonstrated to have their highest concentration in the cerebral cortex (Mohler and Okada 1977). The increase in spindles could be a consequence of the cortical inhibition or due to a separate action of flurazepam involving the 'spindle generator'. Whether this decrease in delta activity and increase in spindles would be irreversible with prolonged intake or with intermittent use of flurazepam remains to be determined.

FLUAZEPAM AND EEG ACTIVITY DURING SLEEP

319

Summary

Rdsumd

To further evaluate the effects of flurazepam on EEG during sleep, following 7 nights of placebo baseline, flurazepam (30 mg) was administered to 6 young adult poor sleepers for 10 additional nights while 6 other young adult poor sleepers continued to receive placebo capsules in a double-blind paradigm. Three placebo follow-up nights were recorded 2--3 weeks post-treatment. Twelve good sleepers received only placebo capsules for the first 7 nights. Delta waves, 0.5--2 c/sec, and sleep spindles were counted on-line by a phasic detector. Delta activity was also analyzed off-line by PDP-12 computer for only the first 4 h of sleep and involved a comparison over stages of sleep. Click-evoked K-complexes during NREM sleep were analyzed for 6 good sleepers and 11 poor sleepers. Repeated use of flurazepam caused a gradual decrease in delta amplitude and count, and a gradual increase in sleep spindle rate. The decrease in delta amplitude was seen in all sleep stages, b u t the decrease was significant only during SWS and stage 2. The decrease in delta amplitude was significant by the 3rd drug night, b u t the rate of amplitude decrease tended to slow with continued treatment. The decrease in delta count was less pronounced and more gradual over drug nights than the rate of decrease in amplitude. Flurazepam also significantly reduced evoked K-complex amplitude b u t did n o t affect latency. Sleep spindle rate was significantly increased by drug night 5. Results of this study indicate that the reduction of SWS with flurazepam during the initial drug nights is due primarily to the decrease in delta amplitude, but, with continued use, the decrease in delta count also contributes to the decrease in stage 4 sleep.

Les effets du chlorhydrate de flurazdpam sur l'activitd dlectrique cdrdbrale au cours du sommeil

Pour mmux dvaluer les effets du fluraz~pam sur I'EEG au cours du sommeil, 6 jeunes adultes classes comme mauvais dormeurs ont dt~ enregistr~s pendant 7 nuits de contrSle sous placebo, puis pendant 10 nuits additionnelles o~ 30 mg de fluraz~pam leur ont ~t~ administr~s alors que 6 autres adultes jeunes ~galement mauvais dormeurs continuaient ~ recevoir des capsules de placebo dans un paradlgme en double-aveugle. Trois nuits successires sous placebo ont ~t~ enregistr~es 2 et 3 semaines apr~s le traitement. Douze bons dormeurs n ' o n t requ que des capsules de placebo pendant les 7 premieres nuits. Les ondes delta, de 0,5--2 c/sec, et les spindles de sommell ont dt~ compt~s en temps r~el par un ddtecteur de phase. L'actlvit~ delta est ~galement analysde en off-line par un ordinateur PDP 12 pendant les 4 premieres heures seulement du sommeil et ont entrain~ une comparaison entre les stades du sommeil. Les Kcomplexes dvoquds par clics au cours du sommeil lent ont ~td analys~s chez 6 bons dormeurs et 11 mauvais dormeurs. L'utilisation r~p~t~e du fluraz~pam provoque une diminution graduelle de l'amplitude delta et du nombre d'ondes, et une augmentation graduelle du taux de spindles de sommeil. La diminution d'amplitude delta s'observe t o u s l e s stades, mais cette diminuhon n'est significative qu'au cours du sommell ~ ondes lentes et du stade 2. La diminution d'amphtude delta est significative dbs la 3~me nuit sous m~dicaments, mais le taux de dimmution d'amplitude tend ~ se ralentir lorsque le traitement continue. La diminution du nombre d'ondes delta est moins prononc~e, plus graduelle d'une nuit ~ l'autre que n'est la diminution d'amplitude. Le fluraz~pam r~duit ~galement de faqon significative l'amplitude des Kcomplexes ~voquds mais n'affecte pas leur latence. Le taux de spindles de sommeil est

320

slgmfmativement augmentO lors de la 5Ome nult sous mOdicaments. Les r~sultats de cette ~tude md lq u en t que la r~ductmn du sommefl ondes lentes avec fluraz6pam au cours des premiOres nults sous drogue est prmcipalem e n t due ~ la diminution de l'amplitude delta rams, apr~s usage durable de la drogue, le nombre d'ondes delta contrlbue 6galement la diminution du stade IV de sommeil.

The authors appreciate the programming and statistical assistance of R P. Hilbert and S A Sunderman, as well as the technical assistance of Valerie S Rosslter, M.T. Austin, D.A. Irwin, V H. Peace and R Mulhs

References Bond, A.J. and Lader, M H The residual effects of flurazepam. Psychopharmacologia (Ber].), 1973, 32_ 223--235. Bonnet, M.H., Webb, W B. and Barnard, G Effect of flurazepam, pentobarbital, and caffeine on arousal threshold. Sleep, 1979, 1: 271--279. Church, M W. and Johnson, L C Mood and performance of poor sleepers during repeated use of flurazepam Psychopharmacology, 1979, 61: 309-316_ Church, M_W, March, J . D , Hibl, S., Benson, K , Cavness, C and Felnberg, I Changes in frequency and amplitude of delta activity during sleep Electroenceph clin. Neurophysiol , 1975, 39 1--7 Church, M.W, Johnson, L.C and Seales, D.M. Evoked K-complexes and cardiovascular responses to spindle-synchronous and spindle-asynchronous stimulus clicks during NREM sleep Electroenceph clin Neurophysiol , 1978, 45 443--453. Elul, R The genesis of the EEG In C.C. Pfeiffer and J R. Smythies (Eds), International Review of Neurobiology, Vol_ 15 Academic Press, New York, 1972 228--272 Feinberg, I , Fein, G , Walker, J.M , Price, L.J., Floyd, T C and March, J,D. Flurazepam effects on slowwave sleep stage 4 suppressed but number of delta waves constant. Science, 1977, 198- 847--848 Feinberg, I , March, J . D , Fein, G , Floyd, T C , Walker, J.M and Price, L Period and amplitude analysis of 0.5--3 c/sec activity in NREM sleep of young adults Electroenceph chn Neurophysiol., 1978, 44 202--213_ Greenblatt, D J., Shader, R I. and Koch-Weser, J Flurazepam hydrochloride Clin Pharmacol T h e r , 1975, 17 1--14

[, (' JOHNSON ET AL tiablitz, J J_ and Borda, R P The elfects ol Dalmane ~"3 (flurazepam hydroehloride) on the contlngent negative variation In W C. McCallum and J R Knott (Eds), Event-related Slow Potentials of the Brain their Relations to Behavior Electroenceph clin Neurophyslol, 1973, Suppl 33 317-320 Johnson, L C Are stages of sleep related to waking behavior 9 Amer Scl, 1973, 61 326--338 Johnson, L , Lubin, A , Naitoh, P , Nute, C and Austin, M Spectral analysis of the EEG of dominant and non-dominant alpha subjects during waking and sleeping Electroenceph elm Neurophysiol., 1969, 26 361--370 Johnson, L C , Hanson, K and Blckford, R G Effect of flurazepam on sleep spindles and K-complexes Electroenceph. clln Neurophysiol , 1976, 40: 67-77 Johnson, L C., Church, M W., Seales, D M and Rossiter, V S Auditory arousal thresholds of good sleepers and poor sleepers with and without flurazepam. Sleep, 1979, l 259--270 Kay, D C , Blackburn, A B., Buckingham, J.A. and Karacan, I Human pharmacology of sleep In R L Wdhams and I Karacan (Eds.), Pharmacology (if Sleep, Wiley, New York, 1976' 83--210 Lubin, A L'utlhzatlon des correlations par rang pour 6prouver une tendance dins un ensemble de moyennes. Bull C E R P , 1 9 6 1 , 1 0 : 4 3 3 - - 4 4 4 Mohler, H and Okada, T Benzodlazepine receptor demonstration m the central nervous system Science, 1977, 198. 849--851 Naitoh, P , Johnson, L C , Lubln, A. and Wyborney, G Brain wave 'generating' processes during waking and sleeping Electroenceph clin Neurophysiol, 1971, 31 294 Rechtschaffen, A and Kales, A_ ( Ed s) A Manual of Standardized Terminology, Techniques and Scoring System for Sleep Stages of Human Subjects Public Health Service Publ 204, U.S. Government Printing Office, Washington, D C , 1968 Roth, M_, Shaw, J and Green, J The form, voltage distribution and physiological sigmflcance of the K-complex Electroenceph clin Neurophysiol , 1956, 8_ 385--402 Sassin, J.F and Johnson. L C Body motihty during ~leep and its relation to the K-complex Exp Neur o 1 , 1 9 6 8 , 2 2 133--144 Schallek, W , Schlosser, W. and Randall, L O Recent developments in the pharmacology of the benzo diazepines Advanc Pharmacol C h e m o t h e r , 1972, 10 119--183. Sherwin, I Differential action of diazepam on evoked cerebral responses Electroenceph clin Neurophysiol ~ 1971, 30- 445--452. Skinner, P. and Shimota, J A comparison of the effects of sedatives on the auditory evoked cortical

FLUAZEPAM

AND EEG ACTIVITY DURING

SLEEP

response_ J. Amer audiol. Soc., 1975, 1 71--78. Smith, J.R., Funke, W . F , Yeo, W.C. and Ambuehl, R A. Detection of human sleep EEG waveforms. Electroenceph. clin. Neurophysiol., 1975, 38 435---437

321 Stockard, J J., Rossiter, V.S., Jones, T A. and Sharbrough, F.W. Effects of centrally acting drugs on brainstem auditory responses. Electroenceph clin. Neurophysiol., 1977, 43: 550--551.