Flurazepam Effects on Sleep EEG TI (EEG)

f. airment r results iderable requen- mphren- ice of a alized in bgroups at could 3 Gradi-

)logy has Flurazepam Effects on Sleep EEG TI (EEG) . .[I]t is licate the Visual, Computer, and Cycle Analysis e diagnw Irwin Feinberg, MD; George Fein, PhD James M. Walker, PhD; Leonard J. Price, PhD; Thomas C. Floyd, MA; Jonathan D. March that the erns of )e inter- 0 Analysis of sleep effects of Hurazepam hydrochlorideon four tive to allow firm conclusions to be drawn for any drug mal subjects confirmed that this drug substantially J of the class, the hypothesis of pharmacologic specificity of sleep suppresses both REM and stage 4 sleep. Computer analysls effects remains of considerable interest. (Drugs that reted as disclosed that delta wave amplltude was greatly reduced by Further produce the same behavioral effect sometimes do so fiuratepam. However, low density delta wave activity (le, stage 2 ,through different mechanisms. For this reason, one might analysis sleep, which was increased In duration beyond the reductlon In i of the stage 4), permitted the number of delta waves and the tlme they find occasional agents within a given psychopharmacologic robable occupied per night to remain at baseline levels. Thlr flndlng class that produce effects on sleep different from those c disor- -ts that sedative-hypnotics increase total sleep time by produced by the majority of its members. However, for the slowing the metabolic processes of sleep M that a longer slbp hypothesis to remain tenable, such atypical agents should duraUon Is required for the same biological effects. New obser- not produce the qualitative patterns characteristic of drugs Service of vatlons on the lnductlon times of REM and stage 4 effects are with opposing actions, eg, no analeptic should produce the ed by the also presented. In general, the distortions In sleep EEG effects characteristic of sedative-hypnotics.) If the hypoth- a at Los producad by flurazepam qualitatlvely resemble, but are quantlta- esis of pharmacologic specificity of sleep effects is tively greater than, those produced by barblturates In equivalent confirmed, it would offer the possibility of a drug classifi- hypnotic doses. cation system based on human brain electrophysiology. (Arch Gen Psych 36%-102,1979) A second and perhaps the most common application of logram of the sleep EEG in psychopharmacology thus far has been to bpbalogra- he sleep EEG has been of special interest to those the study of hypnotic efficacy. The therapeutic effects T concerned with drug effects on behavior for several sought with hypnotics are changes in sleep itself more ectomy in reasons. First, it is an empirical fact that most psychoac- rapid onset, decreased awakening, and increased total ica and its tive drugs produce more substantial changes in EEG sleep sleep time. Ideally, these improvements should be achieved EpllepSi4Z patterns than in waking brain waves. It has been hypothe- without distortion of normal physiologic sleep patterns ic system sized that drugs within pharmacologically equivalent (although the functional importance of these patterns is classes, eg, sedative hypnotics, tricyclic antidepressants, or still unknown). Sleep EEG records provide a continuous probabili- antipsychotic drugs, may produce effects that are qualita- measure of brain activity that is closely correlated with 'sychuctry tively similar within classes but differ across classes.' behavioral sleep; and it is, of course, these recordings that There is already some evidence for this hypothesis. For define the normal electrophysiologic anatomy of human analysis. example, benzodiazepines and barbiturates produce similar sleep. Several investigators have concluded, therefore, that ations in effects on both rapid eye movement (REM) and stage 4 the sleep EEG could define insomnia and measure the sleep. Another example is that an$idepressants (both effectiveness of hypnotic drugs."' This conclusion is functional ' sci 3237, monoamine oxidase inhibitors and tricyclic agents) reduce premature. Many subjects who complain of insomnia show REM sleep to extremely low levels while leaving stage 4 sleep patterns within normal limits,"-*although with care- chological intact or somewhat increased.l Although investigations ful matching of insomniacs and controls, some differences sei 3s4, thus far have not been sufficiently systematic or exhaus- can be detected.'" I distribu- One cannot' doubt the importance of the insights into ¶d 50:s Accepted for publieation March 25.1978, hypnotic effects contributed by EEG sleep studies, espe- Fmm the Francisco Veterans Administration Hospital (Drs Fein- Itructluea San cially the demonstration of rapid habituation to the effects berg, Fein, and Price; and Messrs Floyd and March); and the University of JCUPVSUrg California at San Francisa, (Drs Feinberg, Fein, and Price); and Thimann of these drugs on measures of arousal (sleep onset latency Laborptorien, Crone Cdlege, Univemity of California at Santa CNE (Dr and awakenings). Nevertheless, to the degree that the Walker). Reprint requesta to San Francieco Veterans Administration Hospital, correlation between sleep EEG patterns and the complaint 4150 Clement St, San Franci~m,CA 94121 (Dr Feinberg). of insomnia remains imperfect, insomnia must be consid-

Arch 0.n Psychiatfy-Vd 36, Jn1979 Rurazepam and Sleep EEG-Feinberg et at 95 / ered a symptom rather than a sign, and efficacy in its magnitude of the effects produced than do mean differ- treatment must be defined by subjective improvement ences18 A brief report of some of the findings reported rather than electrophysiologic response. here has already appeared.m A third source of interest in drug effects on sleep is that METHODS they might shed light on pharmacokinetics. Whatever the Subjects functional importance of the sleep EEG changes induced by drugs, their substantial nature and relatively easy Subjecta were four male medical students who ranged in age measurement provide the opportunity to use them as from 22.6 to 25.5 years, with a mean age of 23.7 years. None temporal markers, ie, nonintrusive indexes of the time suffered from any sleep disturbance. All were in good health and as gave their informed consent after a full explanation of the required to affect brain function. Such an application procedures. might furnish a useful supplement to other kinetic data, such as plasma clearance rates. Dosage and Recording Schedules Finally, drug effects on sleep can serve as tools to Sleep EEGs had been recorded during five consecutive baseline investigate more basic issues. For example, drugs may be nights two months earlier as part of a previous investigation. used to modify the availability of specific neurotransmit- Nights 2 to 4 provided the baseline values for the present study. ten in an attempt to determine their role in sleep On all drug nights, flurazepam was administered about 15 minutes processes. In addition, drugs may be used to alter sleep before bedtime, which was held constant at 11 PY. On the first two stage distributions and the behavioral significance of such nights in the laboratory, the subjects received 15 mg and 30 mg of changes might be studied. Thus, the fact that monoamine flurazepam hydrochloride, respectively (initial drug). For the next oxidase inhibitors can virtually eliminate REM sleep in three nights, the subjects took 30 mg at home and retired at 11 PN. They then returned to the laboratory for six consecutive nights of human subjects for prolonged periods without detrimental recording-the first three nights taking 30 mg (“short-term” effects on behavior” raises questions regarding the func- drug), and the last three nights without any drug (withdrawal tional importance of the neuronal activity that occurs in condition). Thus, the subjects received 15 mg of flurazepam this state.’* The fact that stage 4 sleep is reduced by hydrachloride on the first drug night and then 30 mg for seven flurazepam to a far greater extent than by barbiturates consecutive nights, followed by three withdrawal nights. Placebos without greater “hangover” or cognitive impairment were not used in baseline or withdrawal conditions. There were raises similar questions regarding this sleep stage. two considerations involved in this decision. The first was that we In the present investigation, we applied computer tech- knew large effects could occur. The second was that we believed it niques to elucidate flurazepam effects on stage 4 EEG and WBB necessary that the subjects who were following their normal we used sleep cycle analysis to determine the time course daytime routines be aware of the possible dangers of drug hangover and we did not wish to extend the restrictions on their of the changes in visually scored stage 4 and REM sleep. activities. However, sleep records were coded and scored without The reduction of stage 4 by flurazepam is of considerable knowledge (“single blind”) of drug condition. interest to those concerned with the psychobiologic importance of the different sleep stages since several facts Data Recording suggest that stage 4 sleep plays an important rple in Sleep recording and visual sleep stage scoring were done human brain function: it is the sleep stage most closely according to methods described previously.z1 Electroencephalo- correlated with duration of previous and grams were recorded on magnetic tape. The computer program we with age.lr.l3Moreover, stage 4 is preserved or increased in used for the analysis of the sleep EEG is presented in detail elsewhere.z2 amount when total sleep time is artificially limited.16.17 Briefly, we visually identify non-REM (NREM), REM, and Visual scoring of stage 4 depends on the occurrence waking stages on the basis of the polygraph record. Segments of within a given temporal epoch of a stipulated density of FM-tape record identified by time code as corresponding to delta waves that exceed an amplitude criterion. A change NREM sleep are then subjected to period and amplitude analysis in amount of visually scored stage 4, therefore, could result with a minicomputer. Analysis is off-line with playback at four from a reduction in any of the following: total number of times the recorded speed. After analog to digital conversion, delta waves, delta wave distribution, or delta wave ampli- waves are classified by duration into one of nine frequency bands. tude. Our method of computer analysis allowed evaluation For each band, four primary measures are obtained integrated of these different possibilities. amplitude, number of zero crossings, time spent in band, and In the present investigation, we studied the initial and length of waves in band (curve length). Subsequent data process- ing (on an IBM 370 system) computes six secondary measures short-term effects of flurazepam hydrochloride on the from the four primary measures and provides results as sums for sleep patterns of four normal subjects. We evaluated the all NREM sleep, for each successive NREM period (NREMP) and, effects of flurazepam on both visual sleep stages and for the primary variables, for the average %second epoch of computer measures with a statistical technique (“protected NREM sleep. t test”)ll(D16*) that reduces the problems caused by multiple tests of significance. Although such multiplicity19is inevit- RESULTS able in drug studies that attempt to assess empirically Statistical Analysis changes in sleep variables over several conditions, this Both the computer-derived and visually scored measures problem is rarely mentioned in the sleep literature. A were analyzed with a multiple regression/correlation second advantage of the statistical analysis we used is that procedure for repeated measures analysis of variance.” it permitted calculation of the percentage of within-subject The advantages of these methods were described previous- variance accounted for by the experimental conditions. ly. Sleep data were compared across experimental condi- These percentages more meaningfully indicate the relative tions using alhight and by-cycle values for the first four

W Arch Gen Psychiatry-Vol36. Jan 1979 Flurazepam and Sleep EEG-Feinberg et al c

* differ- Table 1 .-Mean Values for Vkually Scored Sleep Measures in Baseline and Experimental (;ondltionr.* sported 4 KWM Dlug Vadm * - , Aocou~kr Inltld Short-tm bltxparlm- BawlIlm Dfiro auo Wlthdrawml Condltbn 1 in age Sleep Lalamy and Total Shplimo s. None Sleep latency. min 16.6 13.3 9.7 s.3t 15.4 alth and Total sleep time, min 410.1 430.W 431 .8$ 429.3t 23.2 of the nsu..-- Eye-movement (EM) latency, min 67.3 110.1$ 100.2t 97.9t 26.4 Total stage REM. min 135.7 114.9t 106.7% 127.3 28.0 % stage REM (stage REMltotal sleep time) 33.7 28.6-4 24.86 29.7 36.7 baseline Total EMS (44 epochs) 549.7 346.3% 290.30 457.0 57.0 tigation. EM density (4-s epochs/stage REM [4a epochs]) 0.28 0.20$ 0.185 0.24 37.0 [4-8 0.09 0.055 0.050 0.07t 55.0 it Total EMS (4-s epochsltotal sleep epochs]) study. Burst index (EM,..lEM?',..) 2.69 2.37t 2.33-4 2.6511 34.9 minutes Total NREM interruptions (20a epochs) of REMPs 23.6 14.9t 12.8$ 14.3t 24.5 first two I 30 mg of NREM the next Total NREM, min 274.4 316.06 325.1 6 302.0-4 55.7 Mean duration of NREMP,.,, min 60.3 78.5% 74.25 70.61 48.9 at 11 PY. lights of Total stage 4. min 35.2 32.2 6.411 4.88 1 58.0 rt-term" % stage 4 (stage 4/NREM) 12.7 10.2 1.e$ # 1.75 # 58.2 bdrawal Total stage 3, min 59.1 62.1 50.2 53.4 11.7 razepam % stage 3 (stage 3/NREM) 21.7 19.7 15.6% 17.9 21.8 Total stage 2, min 180.1 221.6% 268.55 # 243.75 65.1 or seven Placebos % stage 2 (stage P/NREM) 65.6 70.0t 82.55 # 80.40 # 59.8 ?re were 'Significance levels of mean differences and proportions of variance accounted for by drug effects are given. Means are based on two nights in initial that we drug condition and three nights in each of the other conditions. mlieved it ?Significantlydifferent from baseline value at .05. $Significantly different from baseline value at .01. L normal &Significantly different from baseline value at ,001. of drug //Significantlydifferent from initial drug condition value at .05. on their fSinificantly different from initial drug condition value at ,001. without 3 Significantly different from initial drug condition value at .01.

NREMPs. For the cycle analyses, the linear and quadratic re done trends within nights were computed and changes across 2401 cephalc- 7 200- experimental conditions in these trends were tested for r gram we significance. (Shown in the appropriate table are P values P0 n detail for statistically significant differences mentioned in the 160- 0 n text.) Table 1gives the results for the visually scored sleep 9 CM, and I variables. .-," 120- nents of .-> iding to Total Sleep Time and Sleep Latency.-There were no -" 80- analysis significant differences in time in bed across conditions L at four (mean for all conditions was 445.7 minutes). Total sleep u iversion, time was significantly elevated above baseline in initial 40- y bands. drug, short-term drug, and withdrawal conditions. Sleep tegrated 0- latency fell from baseline to initial drug to short-term drug I 1 md, and and withdrawal conditions. Only the difference between 1 2 3 4 Process- baseline and withdrawal was statistically significant; REM Period ieasurea however, noted previously, these were normal and not Fig 1.-Eye movement activity (number of four-second epochs iuma for as insomniac subjects. The increase in total sleep time with EM) for successive REMPs during baseline and on the first LIP)and, (15 mg) and second (30 mg) drug nights. Depressant effect of ?phof resulted from an increase in NREM sleep: this increase 15-mg dose of flurazepam becomes apparent in third REMP of consisted of stage 2 EEG and largely occurred in the first first drug night. Earlier onset of drug action on second night may NREMP of the night. The number of sleep cycles was represent carryover or dose effect. unaffected by drug, ie, total sleep time was increased by virtue of increased length, rather than number, of cycles. drawal conditions. There were no significant differences easures REM -.-Rapid eye movement latency, ie, the first between the initial drug and short-term drug conditions or melation NREMP, increased significantly after drug administration between the baseline and withdrawal conditions in REM iance." and remained elevated through short-term and withdrawal duration. Our analysis of sleep cycles permitted us to evious- conditions. The duration and percentage of REM sleep determine the time course of these effects. The 15mg dose, I condi- were significantly depressed in initial drug and short-term given on the first night, did not affect REMP duration Bt. four drug conditions compared with both baseline and with- until the third cycle, and it produced a significant change wg a d Flurazepam and Sleep EEG-Fdnberg et a1 97 # in the trend of REMP durations across the night. The same 351 9 result was found for the changes in eye movement (EM) 7 activity (Fig 1). On the second night, when a 30-mg dose 30 “\\\ was given, all REMP values were lower than baseline but 4 the trend of these values across the night did not differ from baseline. The difference between the two nights L_ could have resulted from either a dose-specific effect - P (night 1) or a carry-over effect (night 2). Whichever the - case, our data indicate that the induction time of the REM sleep effects of an initial 15mg dose of flurazepam hydrochloride is about 5.3 hours (mean onset time of REMP, on the first drug night). 7 The number (total EM) and proportion (density) of four- second epochs of REM sleep containing EM activity were greatly reduced in both initial dose and short-term drug conditions compared with baseline and withdrawal values. - Flurazepam also reduced the tendency for EM to occur in -! bursts and the ratio of EM activity to total sleep time. The I t r 1 2 3 4 effects on total EM accounted for a far greater proportion of within-subject variance (57%)than did effects on REM NREM Period duration (‘28%). There were no significant differences in Fig 2.-Minutes of stage 4 in each sleep cycle plotted by drug condition. Data of two initial drug nights are plotted separately, EM measures between the two drug conditions or between and the mean figures across nights for other conditions are baseline and withdrawal. The linear and quadratic trends shown. First night (15 mg) of flurazepam administration leads to of EM measures across sleep cycles did not differ in an increase in stage 4 in NREMP, and a decrease to very low - levels in NREMP,. Second night of administration shows a similar *! baseline, short-term drug, and withdrawal conditions drui pattern but with stage 4 reduced almost to baseline levels in first t: although, of course, the absolute levels did. This stability of cycle. This finding strikingly resembles the effects of secobarbital within-night trends in the face of strong drug effects f: on stage 4 distribution.n Density of stage 4 within NREMPS 1’ testifies to the fundamental nature of these temporal showed a similar pattern, indicating that increased stage 4 in I( characteristics of sleep. Examination of EM values during NREMP, on initial administration was not simply the consequence 7 3 the three withdrawal nights shows that return to baseline of increased length of NREMP, associated with delayed REM * onset. was not achieved until the third night. The time course of 2: ! the drug effects on EM activity was strikingly different “‘g from that on stage 4 (see the following related section). NREM Interruptions of REM Sleep.-The number of NREM interruptions of REM sleep (the number of epochs requires several nights of administration to become appar- of REM sleep in which a spindle or K-complex occurred in ent. However, the cycle analysis used here demonstrated dr; the absence of EM) was significantly reduced below base- an effect on stage 4 distribution early in the first night. Th line in the three other conditions. It is not clear whether While most reports of flurazepam effects on sleep stages in this unexpected change should be regarded as an effect on indicate that stage 2 EEG is increased, it is the decrease in de NREM or REM sleep (see “Comment”). stage 4 rather than the increase in stage 2 that is usually inc Wets OIL NREM Sleep.-Non-REM duration increased emphasized. As given in Table 1, the increase in stage 2 of significantly in initial dose and short-term drug conditions duration exceeded the combined reductions in stage 4 and fa as compared with baseline. Most of this increase occurred REM duration and entirely accounted for the increase in N in the first cycle. After withdrawal of flurazepam therapy, total sleep time. Experimental effects on stage 2 accounted st NREM duration returned toward, but remained signifi- for 65.1% of the within-S variance, the highest proportion cantly above, baseline levels. Non-REM duration did not of any visually scored sleep variable. b differ significantly in the initial drug and short-term drug There were no significant differences among conditions It conditions. in amount of stage 3 summed over the entire night. ai Stage 4 duration was substantially reduced in short-term However, the distribution of stage 3 across the night was st drug from baseline and initial drug levels. The stage 4 different in initial drug, short-term drug, and withdrawal dc suppression persisted without significant change during conditions from that of baseline (F = 8.11; Pc .01). In ti withdrawal. While total stage 4 during the initial drug both the drug and withdrawal conditions, stage 3 was B condition did not differ significantly from that in baseline, increased in cycle 1 and decreased in cycle 2 as compared .h sleep cycle analysis revealed a major change in its distribu- with baseline. n tion acrow the night. On the first initial dose night, the NREM-REM Relations.-When NREM and REM dura- d amount of stage 4 increased over baseline by 50% in the tions are considered jointly, one might infer that part of a first cyde and demeased by 20%in the second cycle (Fig 2). the ineresse in NREM sleep during drug conditions was Thm effect was not simply due to a lengthened first naecondary” ta the reduced REM duration. However, there E NREMP (increased “REM latency”), since the proportion also seemed to be a ‘primary” effect as well, since NREM as well as the absolute amount of stage 4 within NREMP, sleep increased beyond the decrease in REM and since was increased. A reduction in stage 4 by flurazepam, NREM sleep remained persistently elevated during with-

Arch Gen Psychletry-Vol30, Jan 1979 Flwazepfim and Sleep EEG-Feinberg et al -

- 1 compensated almost perfectly irt. number and duration of amplitude. (2) Andpi$ of visuplly &d data by succes- incm?.! delta waves for the loss of 0- to 3-Hz activity in stage 4. sive NREM/REM perid reveals that the distribution of duuac This compensatory change occurred in NREMP,, which visually mred stage 4 sleep is significantly altered by a that tE normally manifests extremely high levels of stage 4. single 15mg dose of flurazepam hydrochloride, although hrpoth During withdrawal, integrated amplitude of 0- to 3-Hz total stage 4 is not decreased. (3) The REM suppression contd activity remained significantly reduced. Total number and after an initial 15mg dose of flurazepam hydrochloride functic duration of these delta waves also decreased significantly becomes apparent in the third REM period in the night, ie, Alth below baseline. These latter changes were the result of the about 5.3 hours after drug administration. (4) Interrup taking decline in total NREM (stage 2) sleep as REM duration tions of stage REM by brief snatches of stage 2 sleep are amplit returned to baseline levels, ie, the rate of delta wave reduced by flurazepam. In addition to these new findings, sion o production per 20-second epoch (see following related we confirmed previous observations3 J0 that REM and Strstec section) was the same during withdrawal as in the short- stage 4 are both suppressed by flurazepam but with hasth term drug condition. Although stage 2 decreased in with- characteristically different time courses. Rapid eye move- are no drawal, it remained significantly above baseline so that the ment suppression is immediate but dissipates quickly measu reductions in absolute number and duration of 0- to 3-Hz during withdrawal; stage 4 suppression is delayed and includl waves (11% and 13%), while highly significant, were small persists unchanged during a three-day withdrawal period. densit in comparison with the persistent and profound (86%) In what follows, we comment briefly on each of these The reduction of stage 4 in withdrawal. issues. tude a 8 to 3-Hz Density.-Values of 0- to 3-Hz activity for the As noted in the beginning of this article, the profound P average %second epoch of NREM sleep reflect the density suppression of stage 4 EEG by flurazepam, which is deP of delta waves. Since the same total number and duration unaccompanied by substantial behavioral or cognitive Of COK of 0- to 3-Hz waves were distributed over a larger number impairment, raises the possibility that this sleep stage is of comk of epochs, ie, increased total non-REM sleep, in the short- little functional importance. However, the results of the on del term drug condition as compared with baseline, the aver- computer analyses are consistent with an alternative possi- sugge: age levels for these measures declined significantly. Inte- bility. Stages 2, 3, and 4 of NREM sleep may represent REM grated amplitude, which manifested an absolute decrease increasing “intensities” or “rates” of the same qualitative stimul when the subject was receiving the drug although total process.1z If this is true, a longer period of less intense chang sleep time was increased, showed a greater reduction in (stage 2) deep might substitute functionally for a shorter might density than the other measures. The density of delta wave duration of more intense (stage 4) activity. Our finding partic activity was reduced below baseline in the short-term drug that stage 2 sleep under flurazepam is increased in dura- hypno condition and remained essentially unchanged in with- tion well beyond the decrease in stage 4 (47 vs 29 minutes) findin] drawal. The smallest change was shown by the curve is consistent with this possibility. The increased stage 2 tyto length measure, which is subject to opposing effects by produced by this drug, while invariably statistically signif- antico flurazepam, ie, curve length should be reduced when delta icant, has seldom received much emphasis. In the present basis, waves are smaller but augmented by the increased spin- study, flurazepam administration accounted for a higher chang dling and fast activity known to be produced by this proportion of the within-subject variance for stage 2 sleep drug.24.27 28 than for any other visually scored sleep variable. adVXl Secondary Measures of 0 to 3 Hz.-Secondary measures The computer analysis also provided data consistent The are derived from the primary measures.** This analysis with the hypothesis that increased stage 2 EEG compen- of a si revealed that the average frequency of waves within 0 to sates for what is lost with stage 4 suppression by fluraze- to Our 3 Hz was unchanged in the short-term drug condition. pam. This analysis demonstrates that the increased stage 2 intere Average frequency within the 0- to 3-Hz band is a remark- duration provided precisely the number and duration of ably stable individual characteristic, eg, two-night correla- delta waves required to maintain these measures at base- single tion r = .95; N = 20, P c .0001.22This stability permit- line levels. Moreover, this “compensation” did not occur analy ted a rather slight increase over baseline (mean change simply by “tacking on” more sleep at the end of the night: NREl was 0.05 Hz) in frequency of the average wave during with- virtually all of the increase in stage 2 occurred in the first tbefil drawal to reach a high level of statistical significance NREMP, which normally contains more than 50% of the additi (Pc .001). The reduction in integrated amplitude, with night’s stage 4.“ REM number and duration of 0- to 3-Hz waves unchanged, If tke interpretation regarding the relation of stages 2 PdU resulted in significant reductions of the average inte- and 4 is correct, it holds a more fundamental implication Eh grated amplitude/halfwave and of the average sample regarding the mechanism by which hypnotics increase enabl amplitude in both short-term drug and withdrawal condi- total sleep time: this increase may be achieved by a slowing REM tions. of the metabolic activity that underlies the sleep EEG. This RW COMMENT interpretation is consistent with current evidence that night This investigation yielded the following four new obser- sleep is an active process and also with the fact that admi vations regarding flurazepam effects on sleep: (1) Suppres- hypnotics depress neuronal function, reducing brain oxy- least sion of stage 4 during drug administration does not entail a gen uptake and causing coma in bigher dosages. Of course, drug! net, loss in the number of deb waves or in the time hypnotics poesess other effects as well. They decrease &ed occupied by this frequency band; however, total integrated arousal level, reducing the time required for sleep onset, Th amplitude per night vithin the delta band is reduced by and they diminish the number of nocturnal awakenings. effed about 25% zis a result of a decrease in average wave The recent evidence that benzodiazepines and barbiturates WP]

100 Arch Gen Psychiatry-Vol36, Jan 1979 Rwszspam and Sleep EEG-Feinberg et at kch

L* ._--. .%j 4‘ by succea- increase sigma spindles (bursts of 12- to 14-Hz wave8 merit greater attention than they have received. They ibution of characteristic of human sleep) also raisea the possibility raise the possibility that the drug affects more than one Rred by a that they directly stimulate sleep control mechanisms; we receptor and thus may have differential affinities for, or , although hypothesize that sleep spindles are correlated with the quantitatively different metabolic effects on, the neuro- ippression control processes of sleep rather than with its restorative physiologic systems governing the two kinds of sleep. rochloride functions. Alternatively, there may be an effect on but one sleep ? night, ie, Although there is no net loss of delta waves in subjects system (REM or NREM) that induces interaction effects Interrup taking flurazepam, there does occur a decline in the on the other. ! sleep are amplitude of the average wave. Benzodiazepine suppres- A fourth effect of flurazepam (that to our knowledge is I findings, sion of delta amplitude during sleep has been demon- being reported here for the first time) was a reduction in REM and strated with computer techniques by other investigators as the number of interruptions of REM sleep by NREM but with has the reduction in delta wave den~ity.~*.~~However, we activity. Specifically, our measure of “NREM interrup- ?ye move- are not aware of any previous analysis that simultaneously tions” represents the number of 20-second epochs within s quickly measured within NREM sleep a11 of the delta variables stage REM that contain spindles or K-complexes in the ayed and included in this study, ie, number, duration, amplitude, and absence of EM. On its face, this change could result from ral period. density. alteration in either REM or NREM systems. The temporal of these The functional importance of the changes in delta ampli- course of this change may suggest the answer to this tude and density induced by flurazepam are not apparent. question. Whereas no REM measure during withdrawal profound One possibility is that such changes might be related to differed significantly from its corresponding value in the which is degree of drug-induced “hangover.” Obviously, it would be baseline condition, every NREM measure did, except for cognitive of considerable practical value to establish physiologic stage 3. Since NREM interruptions remained significantly stage is of correlates of hangover. However, in addition to the effects below baseline during the three-night withdrawal period, Its of the on delta amplitude and distribution, sleep EEG changes we suggest that the reduction in NREM interruptions .ive possi- suggest several additional candidates, including stage represents an effect on NREM rather than REM represent REM and EM suppression and spindle and beta wave systems. ualitative stimulation. In addition to examining the relation of these * The same logic could be applied to decide whether the s intense changes to the toxic effects or side effects of the drug, one change in REM latency, ie, the duration of the first a shorter might seek to determine whether they are correlated with NREMP, produced by flurazepam is a NREM or REM ir finding particular pharmacologic actions, such as anticonvulsant or effect. We have long emphasized the equivocal meaning I in dura- hypnotic efficacy. Dumermuth et al,33 on the basis of of changes in this measure.34 In the present investigation, minutes) findings with the waking EEG, suggested that the capaci- we found that REM latency remained significantly 3 stage 2 ty to stimulate beta activity may be correlated with increased during withdrawal; as noted previously, changes !ly signif- anticonvulsant effectiveness. However, on an a priori which persisted during flurazepam withdrawal were char- e present basis, the same possibility exists for each of the sleep EEG acteristic of NREM rather than REM measures. The a higher changes produced by flurazepam, taken - severally or persisting increase of REM latency during withdrawal, re 2 sleep together. Studies of these potential relations might therefore, supports the view that this measure-the advance our understanding of hypnotic mechanisms. amount of NREM sleep that precedes the first REM onsistent The changed distribution of stage 4 after administration period-reflects the state of NREM systems rather than compen- of a single 15-mg dose of flurazepam hydrochloride has not being secondary to variations in “REM pressure.” y fluraze- to our knowledge been previously reported. The immediate Finally, the similarity of flurazepam effects on REM d stage 2 interest of this finding is that it strikingly resembles the sleep to those produced by barbiturates merits emphasis in ration of change in stage 4 EEG observed by Lester et aP3 after a view of the persistent misapprehension that flurazepam i at base- single mmgdose of secobarbital. These investigators, does not suppress REM sleep. Here, we confirm previous lot mur analyzing stage 4 by hour of the night rather than by reports of REM sleep reduction and EM suppression by he night: NREM period, found that secobarbital increased stage 4 in flurazepam. The effect on EM is substantial (47%), unequi- the first the first two hours and decreased it subsequently. Thus, in vocal (P< .001), and about twice as large as that on REM 96 of the addition to the similarity of their effects on total stage duration (52%vs 28%of variance). As with barbiturates, it REM and stage 4, barbiturates and benzodiazepines is not clear whether this effect on EM activity represents a stages 2 produce identical initial effects on stage 4 distribution. primary effect on sleep mechanisms or a depression of plication Examination of the temporal pattern of sleep also oculomotor excitability that simply becomes apparent increase enabled us to detect the onset of flurazepam suppression of during This interesting question can best be islowing REM activity. This effect became apparent in the third pursued on the level of basic neurophysiology. EG. This REM period (about 5.3 hours after sleep onset) of the first nee that night. It seems plausible that the intewal between drug This investigation was supported by National Institute on Drug Abuse ‘act that administration and sleep stage effects could serve as at gmnt DA4RGOlZ (F&ese T. Jones, MD, principal investigator) and by =in oxy- least an outer limit of the time required for psychoactive regearcb funds from the Veterans Administration. If course, drugs to penetrate the blood-brain barrier and to exert Cleve Cavness assisted in the collection of the data. decrease effects on neuronal function. ?p onset, Thus, the induction time, as well as the persistence of Nonproprietary Name and Trademark of Drug .kenings. effects after withdrawal, differ for the REM and stage 4 Aturates suppression produced by flurazepam. These differences Flurazepam hydrochloride-Dalmne. _- bergetd Flurazepam end Sleep EEG-Feinberg el el 191 F;r*: 4 * .s “ 6 ’-__.

1. Feinberg I. Wender PH, Koresko RL. et al: Differential effecta of $w tlb Behavioral Schwa. New York; John Wiky & Son8 Inc, 1975. rhbrpromazine and phenobarbital on EEG deep patterns. J P8ychiatr Res 19. Tokey Jw: Some thoughb on clinical trials, especially problems of 7:101-109,1969. multiplicity. SEiaCac 19%619-684, lsn. 2 Kay DC, Blackburn AB, Buckingham JA, et p1: Human pharmacology 20. Feinbetg I, Fain G, Walker JY, et al: Flurazepam effecta on of sleep, in Williams RL, Karatan I (eds):Phunnaedagy ofsleep. New York, slow-wave sleep: Stage 4 suppressed but number of delta waves eonstant. Jdm Why k Son8 Inc, 19’76, pp 83-209. seianu 198:847a4& 1977. Dement WC, Zercone VP, Hoddea E, et JI Sleep laboratory and clinical 21. Feinberg 1, Komsko RL. HelluN: EEG sleep patterns as a function of studies with flurazepam, in Garattini S, Yussini E, Randall LO (eds):!?%e normal and mthological W~UKin man. J Pswhiatr Res 5101-144,1967. Benzodiazepine. New Yo&, Raven mS,1973, pp 599-611. 22 Pein&- I, M-kh J6, gein G, et p1: Period and amplitude analysis of 4. Ifah J, Tan T-L, Swwingen C, et ak Are over-the-counter deep 0.5.3 c/seamd activity in NREM deep of young adults. E&ct?uencephulogr medications effective? All-night EEG studies. Cum ThRes 13148-151, Clin Neutophyskd c1-213, 1978. 1971. 23. Lester BK, Coulter JD, Cowden LC, et al: -barbital and nocturnal 5. Carskadon MA, Dement WC, Mitler MM,et PI: Self-reports M sleep pbysiologid patterns. i?q&phurmacokgia 13:215!2&?6,1968. kbomtory findinp in 122 drug-free subjects with complaints of chronic 24 Johnson LC, Hanson K, Bickford RG Ef€ect of flurazepam on sleep insomnia Am JPsyehiatry 133:13821388,1976. spindles and K complexes. Eleetroeneephalogr Clin Nevrophysid a67-77, 6. Rechtachaf?en A Polygraphic aspects of insomnia, in Gastaut H, et al 1976. (eds):The Abwnnalitics ofSleep in Man: Proceedings of the 15th European 25. Johnson L, Lubin A. Naitoh P, et al: Spectral analysis of the EEG of meeting on electroencephalography. Bologna, North Italy, Aulo Gaggi dominant and nondominant alpha subjects during waking and sleeping. Editore, 1967, pp 109-12.5. Electroeneephubgr Clin Neuroplysior %361-370,1969. 7. Vel& G, Passouant P, Cadilhac J, et al: Donn4es polygraphiques sur les 26. Dumermuth G, Walz W, Soollo-Lavizzari G, et al: Spectral analysis of insomnies. Rev Neud (Paris) 119269-218,1968. EEG activity in different sleep stages in normal adults. Eur Neurol 8. Karacan I, Williams RL, Salis PJ, et al: New approaches to the 7S296,lrn SI evaluation and treatment of insomnia. Psychosomutics 1281-88, 1971. ZI. Itil TM,Saletu B, Akpinar S: Classification of psycbotmpic drugs 9. Karacan I, Williams RL, Littell RC, et ak Insomniacs: Unpredictable based on digital computer sleep prints, in Itil TM (ed): Modern Problems in and idiosyncratic sleepers, in Koella WP, Levin P (eds): Sleep: Physiology, Pharmacopsychiatry, Basel, Switzerland, S Karger, 1974, vol 8. pp 193- bknistry, Ppychdogy, Pharmacology, Clinical Implications. Proceed- 215. Is I ings of the First European Congress on Sleep Research, Badel, Switzerland, 28. Frost JD Jr, Carrie JRG, Borda RP, et al: The effects of Dalmane S Karger, 1973, pp 120-132: (flwazepam hydrochloride) on human EEG characteristics. Electroencepha- 10. Frankel BL, Coursey RD, Buchbinder R, et al: Recorded and reported lagr Clin Neurophysid 34:171-175, 1973. Dank sleep in chronic primary insomnia. Arch Gen Psychiatry 33:615623,1976. 29. Kales A, Kales JD, Scharf MB, et al. Hypnotics and altered sleep- 11. Wyatt RJ, Fmm DH, Kupfer DJ, et al: Total prolonged drug-induced dream patterns: 11. All-night EEG studies of chloral hydrate, flurazepam, REM sleep suppression in anxious-depreased patients. Arch Gen Psychiatry and methaqualone. Arch Gen Psychiatry 23:219-225,1970. !a145155,1971. 30. Gaillard JM,Aubert C: Specificity of benzodiazepine action on human OF 12 Feinberg I: Changes in sleep cycle patterns with age. J Psychiatr Res sleep confirm& Another contribution of automatic analysis of polygraph socic lo-, 1974. rec~nls.Bkd Psychiatry l&lE&l97, 1975. 13. Webb WE, Agnew HW Jr: Stage 4 sleep: Influence of time course 31. Gaillard JM, Tissot R: EEG sleep studies of insomniacs under waul variables. Science 174:1354-1356,1971. flunitmepam treatment. Int Pharmacopsychiatry 10199-2(n, 1975. “ofte 14. Feinberg I, Carlson VR Sleep variables as a function of age in man. 32 Smith JR, Karacan I, Keane BP, et al: Automated sleep EEG analysis Repa Arch Gen Payehiatry 1823%250,1968. applied to the evaluation of drugs: Illustration by study of chlorazepate incre 15. Williams RL, Karaean I, Humh CJ: Eleetmeneephalography(EEG) of dipotassium. Electroencephalogr Clin Neurophyswl41:587-594, 1976. Human Sleep: Clinical Applications. New York, John Wiley Q Sons Inc. 33. Dumermuth G, Gasser T, Hecker A, et ak Exploration of EEG er, m 1974. components in the beta frequency range, in Kellaway P, Petersen I (ed): 2.80 16. Webb WB, Agnew HW Jr: Sleep: Effects of a restricted regimen. Quantitative Analytic Studies in Epilepsy. New Yo*, Raven F’ress, 1976. mpol Seience 150:174&1746,1965. PP533-= Men 17. Dement WC, Greenburg S: Changes in total amount of stage 4 sleep 34. Feinberg I, Braun M,Koresko RL, et al: Stage 4 sleep in schizophre- as a fnnction of partial sleep deprivation. Electroencephalogr Clin Neuh nia Arch Gen Psychiatry 21262-266,1969. abou physiol m.-, 1966. 35. Feinberg I, Hibi S, Cavness C, et al: Absence of REM rebdund after hour 18. Coben J, Coben P Applied Multiple Regresnkm/Correhtim Analysis barbiturate withdrawal. Science 185:534-535,1974. had pill.. by a1 Rah morl (A I 1 I CORRECTION I h in Table.-In the article titled “Life Events and Prisoners,” published in the acrcI”! February ARCHIVES(35:197-203, 1978) an error has occurred in Table 2. The life event hav item “Death of spouse” has been incorrectly ranked for the white middle-class group. cal The frequency should be .04 and the rank should be 36.5. This change in the rank ordering its includes the subtraction of one point in rank from rank item 10 through rank items 35.5. The “Death of spouse” item shares rank 36.5 with item 37. Thus, from “Life Event” am “Personal injury or illness” through to “Foreclosure of mortgage or loan” the entries just under rank for the white middle-class group should read as follows: 7.5; 36.5; 7.5; 9; 10; the 11.5; 11.5; 13.5; 13.5; 15.5; 15.5; 17.5; 17.5; 20,20,20;n; 23; 24; S.5;2Fi.5; 27;28; 29.5; 29.5; 31; ben 32.5; 32.5; 34.5; 34.5; and 36.5. knc 7 I I - L E‘1 Vetc Sim &I R La.

1(LI Arch Gen Psychiatry-Vd 36, Jan 1979 Fkrszepam and Sleef~EEG-Feinberg et al kC

I -