Effects of Lormetazepam and of Flurazepam on Sleep

Br. J. clin. Pharmac. (1984), 17, 531-538 Effects of lormetazepam and of flurazepam on sleep

KIRSTINE ADAM & I. OSWALD University Department of Psychiatry. Royal Edinburgh Hospital, Edinburgh EH 10 5HF

1 Nine poor sleepers of mean age 61 years took part in a double-blind, balanced order study in which, during three periods of 3 weeks, each took lormetazepam 1 mg, lormetazepam 2.5 mg, and flurazepam 30 mg. 2 Using electrophysiological measures, sleep was found to increase by 0.75 h with each treatment condition, mainly through more of stage 2 sleep. The treatments reduced the delay to sleep and led to fewer and shorter awakenings, with little difference among the three treatments. Slow-wave sleep was reduced by flurazepam and by lormetazepam 2.5 mg. 3 After flurazepam intake ceased, there was evidence of persisting drug effects for as long as 7 nights. In contrast, when lormetazepam 2.5 mg ceased, there was significant rebound reduction of sleep duration below baseline for up to 3 withdrawal nights, and there was a similar though non-significant trend after lormetazepam 1 mg had ceased. 4 Wakefulness in the final 2 h of nocturnal recording during the third week of drug intake was significantly reduced below baseline by flurazepam, but was little affected by lormetazepam. 5 The differences among the treatment conditions could be attributed to the long- persistence of flurazepam vs the more rapid elimination of lormetazepam. Keywords lormetazepam flurazepam sleep

Introduction Lormetazepam and flurazepam are benzodia- design included 3 weeks of regular intake, to see zepine derivatives in use as hypnotics, and if tolerance and withdrawal effects could be differing sharply in their persistence in the body. measured. Lormetazepam has an elimination half-life of Seven women (aged 54, 54, 60, 60, 63, 65, 66 about 10 h in young adults, and 20 h in the elderly years) and two men (57, 62 years), mean age 61 (Humpel et al., 1979, 1980). Flurazepam, years, chosen because they considered them- through its principal active metabolite, has an selves to be poor sleepers, and to be of the age elimination half-life of the order of 100 h even in and sex distribution of those who commonly take young adults (Breimer & Jochemsen, 1983). We hypnotic drugs, took part. They had taken no sought to measure and to compare their effects CNS drugs in the preceding months, they were on sleep. asked not to take other drugs or alcohol. The It may be assumed that an hypnotic should study was approved by the Royal Edinburgh increase the time spent asleep, and reduce the Hospital Ethics Committee and each subject amount of wakefulness interrupting sleep, gave informed consent. though it is uncertain to what degree there is a Each subject participated in three sequences, correlation between duration of sleep, its sub- each of 6 weeks, with 4-week intervals between jective quality and its presumptive restorative the first and second and between the second and value (Adam, 1979). Hypnotic drugs are third. Throughout each sequence the subjects commonly taken for weeks or more and so the took matching capsules every night at bedtime 531 532 Kirstine Adam & I. Oswald and during the first 2 weeks these were placebo used, and this was used for sleep latency, capsules, during the next 3 weeks they contained measures of wakefulness and amounts of slow active drug, and in the sixth week placebos again wave sleep. Where analysis of variance proved were taken. During each 3-week period of active significant, correlated t-tests or the Wilcoxon drug, the same preparation was taken each signed-ranks test were used. In the t-tests night, and during one 3-week period the subject degrees of freedom were 8, and in the Wilcoxon took lormetazepam 1 mg, in another 3-week generally 9, depending on the number of non- period the subject took lormetazepam 2.5 mg zero differences. A further analysis of variance and during the other 3-week period the subject was carried out to determine whether there were took flurazepam 30 mg. The order of admin- any differences among the drugs during drug- istration of drug was by a latin square design, taking periods. The data used in these analyses with 'single-blind' conditions for the placebo were the differences from the baseline means. A periods and 'double-blind' conditions for the drug term (df = 2,16) referred to differences active drug periods. Subjects attended the sleep among the three drugs, additionally a night term laboratory on a total of 14 nights during each 6 (df = 1,8), indicated any consistent difference weeks. In the first week of placebos there were 2 between the early and late drug periods nights at the sleep laboratory for adaptation and irrespective of the drug being taken and a drug x in the second week 2 nights for the recording of night interaction term revealed any difference baseline values. The first and fourth nights of the among the treatments in the pattern of change subsequent week gave 'early drug' data. At the between the early and late drug periods. There- end of the fourth week there was a further after, correlated t-tests between pairs of con- adaptation night and in the fifth week (the third ditions were carried out to compare the drugs, week of active drug) the nineteenth and twenty- particularly during the early and late drug first nights on the drug gave 'late drug' data. In periods, using the differences from the baseline the final week the first, second, third, fifth and means. seventh nights provided withdrawal data. To see if the withdrawal of any one of the On all nights the electroencephalogram, eye drugs led to greater disturbance of sleep, further movements and submental muscle tone were analysis of variance was carried out in which recorded. Lights-out was at approximately 22.30 three different withdrawal periods (drug term) at h, and 8 h 45 min was recorded each night. five levels (night term) were compared, using Subjects slept in comfortable, air-conditioned differences from the baseline means. The drug bedrooms. Ultimately the records were coded term could thereby reveal differences among the and scored 'blind' into the different stages of drugs upon withdrawal. The night term could sleep and wakefulness (Rechtschaffen & Kales show if there was a significant change across the 5 1968). Thereafter the code was broken, and the individual nights following withdrawal and the raw data analyzed by a computer programme, drug x night interaction term could reveal any and finally by use of the BMDP statistical inter-dependence. When appropriate, correlated packages of the Health Sciences Computing t-tests were subsequently employed using Facility, University of California. differences from baseline means to compare, for In the statistical analysis, the mean of the 2 example, the effect of withdrawal of flurazepam baseline nights, the mean of the 2 early drug 30 mg with the effect of withdrawal of one of the nights, and the mean of the 2 late drug nights doses of lormetazepam on a specified night. were determined for each individual for each of 33 different sleep measures and these means have been used in the analysis of results, Results together with the individual values for each of In Tables 1, 2 and 3 are summarized the data of the 5 withdrawal nights. In the first place, each of principal interest and we elaborate below the the three drug sequences was treated separately main findings. by an analysis of variance with repeated measures to determine whether any of the Sleep duration increased individual drugs had an effect on the particular measure of sleep. Here the degrees of freedom Subjects slept about 0.75 h longer with any of the were always 7,56. The individual withdrawal three treatments, mainly through increased nights were used rather than the means because duration of stage 2 sleep. Making comparisons mean values could conceal systematic changes with the baseline means for total sleep, and using with time following the withdrawal. Where, for t-tests, for all three treatments the early drug any measure, the co-efficient of variation was mean and the late drug mean were both signifi- consistently above 0.50, the non-parametric cantly lengthened at the P < 0.05 criterion or Friedman's analysis of variance by ranks was better. The same was true when sleep duration Lormetazepam, flurazepam and sleep 533

.t st el*|tf kt^°o N b $zQ 00 0 0 £ b ^ N 0 11o _ o

cq en worl W)On 0e 0 0 e 0

1. cq (ON tn en 'ICNo0 n C1* 'I cq}N _ 00x oo 't cn 00 +1N N '+l +l eq +l +l

00 O' 00 a) r- en en 'I - 00 en t,oo00 oio 0; C£c4£ en 25e n " W) tz +l N~~~~~~~ +l " +l X ttoo-^N+l +l

O r0 00 0N Nt r1 'I W) cq r-t O 00'I 'I 08t U) tnO 'I r1 N t cq+ +l +l cq +l +l W

Q m b -0s 0 sO ^ F t - o O N - CsO e0rIC .Y; ~~~~~en+1 +1 rq +1 +1 m

ON Q _ sCM oy~~~-I - oen 'ItU'I CN W) cn ;~~~~~~~~~~~~~~~~t cn °NN+C) un 1£0 t-N = <~~~~~~'~ +l +l rq +l +l

= z z o ts xo 'It£ enrl o00 't ON No Y < > l- N N ~~~~~00st-1-01en <2-0 +1 L_ +I +n1o^O.Ot+1 en +1 mt

Q~~~~~~~r CZ 00 ON IC't_- 'I It W-'I +l

W) q QE CC t v} +l) *e t- -) +l st +l tj Oo +1 +1 'rz

X N r a

E9 ;2 _ _ _ _ A=tt_* 534 Kirstine Adam & l. Oswald

\C rC^,~~~~~~~~~~~~~0

z i~ri,Ei- Ii-oo oNI5- O0-0 C- N. (N - (Na - NI

Ct W) XXCi)XiX

It~O~00 It~NI\C

-CW) m £ o ooW) N q X - t

s oN00 r 00Fnc W)(N en 00 m enI 00 l r4 N+r-- or-C-+1 ++l +l - +l +1I

ri (NC-00~~~~~~~o>ON'C 0\'C 00 oo CDom>W) ~ F-- CO -~~~~~~~~0 tn'C- nW066066oeq oen rce' ~~~~+i~~~~~~++ +1

C) Qm00t OC o ~ _I t)- (NJ _I t

._~~~~~~~~~~~r +1 f) -t+1 +1 +1° ) X+ O\IC 00 00 r 1 0r It) 'ICr- Cl) 00_0r'rI C-"J 1ot) 00 ri +1 Q~~~~~~~~~~~~~ro0o -z" oo .I -=£ W~~~~~~~~~~~0_NN mt t rq oo 0:en o

In +1 CZ +1 +1 +1

00 C) v ~' ri00 cn 'I 00 00 r t k ~~O\C~~~ r- ~~~~oO--O~~It oIt ro - It) °LQ > o +1 N - o)-Nm~~~~~~~~~+1mP 1 °°+1++1~~~~~+

N N CO X £00 '!t 0~00~C'-Cto t 0-oO ..~ ~ ~ r~~ 00*_) - e (,(SIt^vo ~NJ N +l E~~~~~~~~~~1-+ +E ^It)

rI e= *=-.gs C)~~~~~~~~~~~~~~~~~~~~~~~~~C It)- Lormetazepam, flurazepam and sleep 535

WI a, m 8 m OM N°°m°M°t °

II II

c£ o6 oo6oiF N f +1 tl++lI>+I tl - "

+1 +1 ~~~+1 0 vt 0 -4 < _ m _ <,1-400 °-4 ° --

<,rm < r-.t, oo~_ N 11 tltelt

ort+I N , _£ b6 __o N "o+I r4-o6 - i tt - +ltl 6 <

+1 +1 ~~~~+1 oo - -oo r -- oo 'O - _l av d-O0 drtl tl ^ .*~~ tl-I tn O ~~ti~~00 cq "0~ +I

t e £ ^-o £-N oo £ X_ °°i N NN 00e e 4+1

n Oz0....~0e en _ e

d tl4 tlr.-r^-r ti f0 tl c~r - > 0t0, q -nt +1 °tC+1~~~~~~+

0~ ~ ~~ ~ ~ ~ ~ ~ ~ ~~~~~~0

00o cto tw d- tl N > > tl n ° -

LQ 00 +I +I 0O cu t :t 3~~CY t-- 0 ON C )tX+l qtN

E~~~0E ~ .c

Eu *9 -v -sG -, - -0EW =='EoH' 536 Kirstine Adam & I. Oswald was treated as a proportion of the total duration Sleep stages of nocturnal recording time. When the three Administration of all three treatments, as treatment conditions were themselves com- mentioned, increased total stage 2 sleep. The pared, using differences from the baseline duration of REM sleep in the whole night was means, there were no significant differences little affected, though in the first 4 h of cumu- among the treatments. lative sleep, REM sleep tended to be reduced by flurazepam and by lormetazepam 2.5 mg, with Quicker to fall asleep which latter the reduction was significant on t- The latency to the onset of sleep, always showing test, both for early and late drug periods at the P a high degree of scatter among poor sleepers, = 0.001 criterion. was reduced on average by about 10 min under Slow-wave sleep (stages 3 + 4) was little each of the three drug conditions. In the case affected by lormetazepam 1 mg, but was reduced of lormetazepam 1 mg, non-parametric analysis by flurazepam during early drug (P = 0.04) and of variance just missed significance, but for late drug (P = 0.02) and during late intake of lormetazepam 2.5 mg analysis of variance was lormetazepam 2.5 mg (P = 0.02). The amount of significant and subsequent Wilcoxon tests, time spent in stage 1 (drowsiness) was reduced making comparison with the baseline mean, significantly only by flurazepam: early drug, P = revealed significance during early drug intake (P 0.02; late drug, P = 0.01. There were no signifi- = 0.01), but a non-significant difference (P = cant differences among the treatments in effects 0.07) for late drug intake. The effect of on slow wave sleep or on stage 1. flurazepam was non-significant during early drug (P = 0.09) but significant during late drug intake Thefinal 2 h (P = 0.03). Using differences from baseline We examined the final two recorded hours of means, there were no significant differences each night during baseline and during late drug. among the three treatments. The number of minutes of wakefulness was highly variable from night to night and subject to Lessfrequent awakenings subject. We took the mean of the two nights for Throughout the study no subject slept less than a each subject and then the group means were cumulative total of 4 h on any one night (though used and revealed: lormetazepam 1 mg, baseline sometimes less than 5 h) and so, keeping total 21.5 min, late drug 18.1 min; lormetazepam 2.5 sleep constant, we considered how often subjects mg, baseline 20.9 min, late drug 14.0 min; awakened in the course of accumulating their flurazepam 30 mg, baseline 18.2 min, late drug first 4 h of sleep. Analysis of variance was signifi- 5.5 min. Thus the only substantial difference was cant for each treatment and inspection of the for flurazepam and, on the Wilcoxon test, only Tables reveals a reduction in the frequency of this reduction was significant, P = 0.03. awakenings associated with all three active treatments. Wilcoxon tests confirmed that with Changes during drug administration lormetazepam 1 mg subjects wakened less often There were a few trends to increasing effect of during both early drug (P = 0.025) and late drug flurazepam within the two early drug nights, and (P = 0.04). The lesser frequency of awakenings between the early and the late drug periods in, did not reach significance with the larger dose of for example, slow wave sleep, but these did not lormetazepam, nor during early intake of reach significance. flurazepam, though it did during late flurazepam Tolerance, evidenced as a lesser effect of drug intake (P = 0.01). There were no significant administration during late drug intake compared differences among the three treatments. with that seen in the early days of intake, was suggested by a number of measures in the case of Less intervening wakefulness lormetazepam 2.5 mg; for example, mean total sleep duration decreased by 15 min (P = 0.06). The cumulative duration of wakefulness that intervened in the course of accumulating the first Drug persistence 4 h of sleep was reduced by all three treatments. Using Wilcoxon tests the reduction was signifi- After flurazepam intake had ceased there was cant with lormetazepam 1 mg, both early (P = evidence of persisting drug. Thus, on the first 0.01) and late (P = 0.02); with lormetazepam 2.5 withdrawal night, total sleep duration was still mg for early drug only (P = 0.03); and with significantly elevated (P = 0.01) and the mean flurazepam both early (P = 0.04) and late (P = figures suggested a continuing effect for another 0.01). There were no significant differences two nights. Stage 2 sleep continued significantly among the treatments. elevated compared with baseline into the second Lormetazepam, flurazepam and sleep 537 withdrawal night (P = 0.02). Persistence was among groups, but, like ourselves, the authors even clearer in the reduced frequency of awaken- concluded that there was little greater efficacy to ings in the first 4 h of accumulated sleep, which be achieved from the higher dosages. In their effect continued as long as the fifth withdrawal data, too, there was a suggestion that higher night (P = 0.03), while intervening wakefulness in doses of lormetazepam were associated with the first 4 h of sleep was still low compared with some tolerance by a second week of intake, baseline on the seventh withdrawal night (P = though, as in our study, such trends were not 0.05). In contrast. when either dose of significant. lormetazepam ceased to be taken, withdrawal Rebound phenomena upon withdrawal of phenomena tended to appear. hypnotics are to be expected and we have, for example, found rebound reduction of sleep Withdrawal effects duration below baseline subsequent to regular intake of nitrazepam 5 mg or lormetazepam 2 mg Withdrawal of lormetazepam 1 mg was associated et Such with shorter sleep and greater variability of (Adam et al., 1976; Oswald al., 1982). rebound phenomena were not manifest in the duration among subjects, but on no measure did present study after flurazepam intake, and Kales comparisons of baseline mean values with indivi- et al. (1982b) have similarly reported absence of dual withdrawal nights reach significance, rebound following flurazepam. Presumably this though for total sleep on the second withdrawal is because it is impossible abruptly to withdraw night, P = 0.07. flurazepam. What we find are effects that must Withdrawal of lormetazepam 2.5 mg was associated with diminished total sleep and be attributed to prolonged persistence of active greater variability compared with baseline; for metabolite for as many as 7 days. Kales et al. (1982b) imply that it is a good feature of total sleep on the first withdrawal night, P = flurazepam that there is no rebound, but we have 0.004 and on the third withdrawal night, P = to emphasize the corollary of flurazepam's 0.007. persistence, namely drug accumulation and Comparisons were made among the treat- and ments using differences from baseline for the five impairment of social judgement of psycho- motor skill by day, as we have found to be withdrawal nights and analysis of variance associated with regular flurazepam intake by the provided a significant drug term (F = 5.30, P = middle-aged, in contrast to the absence of 0.017) and a significant drug x night interaction measured impairments with lormetazepam 1 mg, term (F = 2.38, P = 0.026). Correlated t-tests and if any, impairments revealed no significant differences between the only minimal, daytime 2.5 mg et al. two doses of but sleep was from lormetazepam (Oswald 1979). lormetazepam, The clinical choice is thus: daytime dopiness and significantly shortened on the first withdrawal no rebound, or daytime clear-headedness but night after lormetazepam 1 mg when comparison bad for some when the drug stops was made with the first night after flurazepam sleep nights following regular intake. In our view the second intake ceased, (t = 3.30, P = 0.01). The same of these offers the lesser disadvantage, was true of the first withdrawal night of options as a small dose of the more rapidly- lormetazepam 2.5 mg, compared with flurazepam particularly eliminated lormetazepam is an effective (t = 4.74, P = 0.002) and on the third withdrawal night there was still a difference between hypnotic. Although lormetazepam is an example of a flurazepam and lormetazepam 2.5 mg, (t = 4.25, fairly short-life hypnotic, others are more rapidly P = 0.003). eliminated, for example, triazolam, which has an Discussion elimination half-life of about 3 h (Jochemsen et al., 1983). There may be disadvantages from a Both lormetazepam and flurazepam in the doses drug that can lead to abruptly-varying tissue used were effective hypnotics. The Tables concentrations, and withdrawal effects during indicate that in, for example, augmenting sleep the day. Morgan & Oswald (1982) reported in- duration and diminishing awakenings, there was creasing daytime anxiety from regular triazolam little advantage in the larger dose of lornetazepam 0.5 mg at night, a finding since confirmed by compared with the smaller dose, and indeed the Kales et al. (1983), and consistent with more larger dose was associated with significant extreme psychological symptoms associated with rebound phenomena. Using a population 20 larger doses (Van Der Kroef, 1979). On the years younger, Kales et al., (1982a) gave doses of other hand, 24 weeks of intake of lormetazepam lormetazepam 0.5 mg, 1 mg, 1.5 mg and 2 mg to 2 mg nightly had no effect on daytime anxiety different groups of poor sleepers. There were (Oswald et al., 1982). The critical factor is only six subjects in each group and inter-group presumably the difference in elimination time comparisons were constrained by the variability and it was with this in mind that we examined the 538 Kirstine Adam & I. Oswald last 2 h of the night to see if any withdrawal triazolam. The amount of wakefulness was rebound effects could be discerned as the day similar to baseline, and would suggest relative was about to begin. Kales et al. (1983) had re- freedom from residual drug effects as the time ported 'early morning insomnia' as a late-night for rising approached. In contrast, because withdrawal phenomenon during the second flurazepam was still having strong effects just week of intake of 'rapidly eliminated benzodia- before the time for rising, freedom from residual zepines' and similar late-night broken sleep from effects at breakfast time could not be expected. accustomed sodium amylobarbitone had pre- If a very rapidly-eliminated drug such as tria- viously been reported from our laboratory zolam may bring disadvantages of one kind, and (Ogunremi et al., 1973). Although lormeta- the long-acting flurazepam bring disadvantages zepam is fairly rapidly eliminated, there was no of an opposite nature, an optimal compromise evidence in our data of the early morning may perhaps be found in an hypnotic, such as insomnia (Kales et al., 1983) reported for lormetazepam, which avoids either extreme.

References Adam, K. (1979). Do drugs alter the restorative value Kales, A., Bixler, E. O., Soldatos, C. R., Vela- of sleep? In Pharmacology of the states of alert- Bueno, A., Jacoby, J. & Kales, J. D. (1982b). ness, eds Passouant, P. & Oswald, I., pp 105-111. Quazepam and flurazepam: long-term use and Oxford: Pergamon Press. extended withdrawal. Clin. Pharmac. Ther., 32, Adam, K., Adamson, L., Brezinova, V., Hunter, 781-788. W. M. & Oswald, I. (1976). Nitrazepam: lastingly Kales, A., Soldatos, C. R., Bixler, E. 0. & Kales, effective but trouble on withdrawal. Br. med. J., 1, J. D. (1983). Early morning insomnia with rapidly 1558-1560. eliminated benzodiazepines. Science, 220, 95-97. Breimer, D. D. & Jochemsen, R. (1983). Pharmaco- Morgan, K. & Oswald, I. (1982). Anxiety caused by a kinetics of hypnotic benzodiazepines in man. In short-life hypnotic. Br. med. J., 284, 942. Sleep 1982, ed Koella, W. P., pp 89-104. Basel: Ogunremi, 0. O., Adamson, L., Brezinovi, V., Karger. Hunter, W. M., Maclean, A. W., Oswald, I. & Church, M. W. & Johnson, L. C. (1979). Mood and Percy-Robb, 1. W. (1973). Two anti-anxiety drugs: performance of poor sleepers during repeated use a psychoneuroendocrine study. Br. med. J., 2. of flurazepam. Psychopharmac., 61, 309-316. 202-205. Humpel, M., Illi, V., Milius, W., Wendt, H. & Oswald, I., Adam, K., Borrow, S. & Idzikowski, C. Kurowski, M. (1979). The pharmacokinetics and (1979). The effects of two hypnotics on sleep, biotransformation of the new benzodiazepine subjective feelings and skilled performance. In lormetazepam in humans. Eur. J. Drug Metab. Pharmacology of the states of alertness, eds Pharmacokin., 4, 237-243. Passouant, P. & Oswald, I., pp 51-63. Oxford: Humpel, M., Nieuweboer, B., Milius, W., Hanke, H. Pergamon Press. & Wendt, H. (1980). Kinetics and biotransfor- Oswald, I., French, C., Adam, K. & Gilham, J. mation of lormetazepam. Clin. Pharmac. Ther., (1982). Benzodiazepine hypnotics remain effective 28,673-679. for 24 weeks. Br. med. J., 284, 860-863. Jochemsen, R., Van Boxtel, C. J. & Breimer, D. D. Rechtschaffen, A. & Kales, A. (1968). A manual of (1983). Pfiarmacokinetics of 5 benzodiazepine standardized terminology, techniques and scoring hypnotics in the same panel of healthy subjects. In system for sleep stages of human subjects. Wash- Clinical pharmacokinetics of benzodiazepine ington, D.C.: U.S. Government Printing Office, hypnotics, ed Jochemsen, R., pp 81-94. 'S- NIH Publication, No. 204. Gravenhage: J. H. Pasmans. Van Der Kroef, C. (1979). Reactions to triazolam. Kales, A., Bixler, E. O., Soldatos, C. R., Mitsky, Lancet, ii, 526. D. J. & Kales, J. D. (1982a). Dose-response studies of lormetazepam: efficacy, side effects, and (Received October 13, 1983, rebound insomnia. J. clin. Pharmac., 22, 520-530. accepted February 3, 1984)