Neuroscience & Biobehavioral Reviews, Vol. 14, pp. 49-63. o Pergamon Press plc, 1990. Printed in the U.S.A.
01.49-7634/90 $3~00--+ -.00
Drug Effects on REM Sleep and on Endogenous Depression G. W. V O G E L , I A. B U F F E N S T E I N , K. M I N T E R A N D A N N H E N N E S S E Y
Sleep Laboratory, Department of Psychiatry, Emory University School of Medicine and The Georgia Mental Health Institute, Atlanta, GA 30306 Received 3 N o v e m b e r 1989
VOGEL, G. W., A. BUFFENSTEIN, K. MINTER AND A. HENNESSEY. Drug effects on REM sleep and on endogenous depression. NEUROSCI BIOBEHAV REV 14(1) 49-63, 1990.--In earlier work REM sleep deprivation (RSD) by arousals improved endogenous depression. This suggested that drugs producing a similar RSD would have antidepressant activity. The arousal RSD was large, persisted for weeks, and was followed by a REM rebound. We call RSD with these properties arousal-type RSD. The present study reviewed literature from 1962 to 1989 on drug REM sleep effects to examine the hypothesis that drugs producing arousal-type RSD improve endogenous depression. The literature reviewed concerned the REM sleep effects of amine precursors, antidepressants, antihistamines, antipsychotics, barbiturates, benzodiazepines, other hypnotics, drugs affecting cholinergic and noradrenergic neurotransmission, ethanol, lithium and narcotics. Four hundred and sixty-eight relevant papers were read and 215 contributed information that could be used in the review. The findings indicated that all drugs producing arousal-type RSD improved endogenous depression. Four drugs that improved endogenous depression did not produce arousal-type RSD. REM sleep
REM sleep deprivation
Antidepressant drugs
ABOUT fifteen years ago REM sleep deprivation (RSD) by arousals was reported to improve endogenous depression but not reactive depression (246). At that time it was also noticed that antidepressant drugs produced REM sleep deprivation (242). Taken together, the two findings suggested that antidepressant drugs improved endogenous depression by REM sleep deprivation (242). Two main objections were raised against the hypothesis. One was that drugs without an antidepressant effect, most noticeably addictive drugs, also produced RSD. These drugs included alcohol, amphetamines, barbiturates and narcotics. The answer to this objection involved salient differences between RSD by arousals and RSD by the addictive nonantidepressant drugs. RSD by arousals had three prominent properties (246). 1. Arousal RSD was initially large, i.e., initially REM sleep was reduced by more than fifty percent of its pretreatment level. 2. Arousal RSD was persistent, i.e., substantial reduction of REM sleep, though often less than its initial reduction, was maintained for several weeks. 3. Arousal RSD was followed by a REM rebound, i.e., immediately after RSD was stopped, REM sleep was significantly greater than before RSD was started. We call RSD with these three properties arousal-type RSD. In 1975 a review found that RSD by early antidepressant drugs produced arousal-type RSD (242). These drugs caused a large persistent RSD that was followed by a REM rebound. The review
Endogenous depression
also found, in a less systematic way, that RSD by nonantidepressant drugs, including amphetamines, barbiturates and narcotics, did not show these properties (242). Their RSD was not large, and/or not persistent, and/or did not produce a REM rebound. This led to a refinement of the original hypothesis. The more specific version was that antidepressant drugs improved endogenous depression by arousal-type RSD rather than by an unspecified kind of RSD. A second early objection to the hypothesis was that not all antidepressants produced arousal-type RSD. For example, among first generation antidepressant drugs iprindole was reported to lack a REM sleep suppressant effect (16). This objection led to a much needed correction and clarification. The finding that RSD by arousals was an antidepressant process did not suggest that all antidepressant drugs "worked" by this process. It was certainly possible, even probable, that some antidepressant drugs improved endogenous depression by processes that did not include RSD. Many individual disorders (e.g., hypertension, heart failure, infectious diseases) are improved by several different drugs that work by different processes. But the finding that RSD by arousals was an antidepressant process did suggest a new hypothesis, viz., that all drugs producing arousal-type RSD would have an antidepressant effect on endogenous depression. Indeed, the hypothesis would be disproved by finding a drug that produced arousal-type RSD and did not improve endogenous depression. Fifteen years ago the effects of drugs on REM sleep were reviewed with this hypothesis in mind. The 1975 review supported the hypothesis (242). Since then, many new antidepressant drugs and many new
tRequests for reprints should be addressed to Gerald W. Vogel, M.D., Sleep Lab, GMHI, 1256 Briarcliff Road, N.E., Atlanta, GA 30306.
50
VOGEL, BUFFENSTEIN, MINTER AND HENNESSEY
nonantidepressant drugs affecting the brain have been introduced. This paper updates and incorporates the earlier review on the effects of these drugs on REM sleep and endogenous depression. The aim is to determine current evidence bearing on the hypothesis that drugs producing arousal-type RSD improve endogenous depression. METHOD The present review concemed the REM sleep effects of amine precursors, antidepressants, antihistamines, antipsychotics, barbiturates, other hypnotics (nonbarbiturate, nonbenzodiazepine), benzodiazepines, drugs affecting cholinergic neurotransmission (cholinergic agonists, cholinesterase antagonists and muscarinic antagonists), drugs affecting noradrenergic transmission (alpha agonists, alpha blockers, beta blockers, sympathomimetic drugs, etc.), ethanol, lithium, and narcotics. Studies on drug REM sleep effects in animals and humans were included in the review. In the following Results section only human studies were displayed in the tables. Where relevant, animal studies were mentioned in the text. A bibliography for drug effects on REM sleep was obtained from the annual sleep literature bibliographies published by the Sleep Research Society (23, 31-38, 202). Begun in 1962, these bibliographies included reasonably complete listings of the world literature on sleep. We reviewed the literature on drug REM sleep effects for the annuals dated years 1962-1989. There were three exceptions to this rule. The sleep literature on antipsychotic drugs was examined for 1975-1989. The benzodiazepines sleep literature was examined for 1985-1989. In both cases these years gave a relatively large sample. The literature on adinazolam and alprazolam was reviewed for 1975-1989 because these unusual benzodiazepines had claims of antidepressant activity. The search began with a list of 646 papers on the effects of the reviewed drugs on sleep. By their titles alone 178 papers were excluded as irrelevant and 468 papers were read. Of the latter, 215 papers contributed information that could be used in the review. The included papers explicitly listed baseline REM sleep levels and one or more of the following: initial drug REM sleep levels, later drug period REM sleep levels, and REM sleep levels after drug discontinuation. This information was used to determine whether the drug caused an initial large REM sleep deprivation, a persistent REM sleep deprivation, and/or a REM rebound. Initial large REM sleep deprivation was present when the first one or two nights of drug administration reduced REM sleep to less than fifty percent of its predrug levels. Initial drug REM sleep reductions to more than one-half predrug levels were called small. Whether initial RSD was large or small, persistent RSD was present if at two or more weeks of nightly drug administration, the REM sleep reduction was more than half the initial reduction. An analogous increase of REM sleep occurred during the treatment of endogenous depression by RSD via arousal (246). In that procedure the number of awakenings required for RSD increased progressively on successive nights of RSD. To prevent the number of awakenings from becoming inordinately large, patients had a single night of no awakenings about every fourth night. On that night they had a REM rebound. This periodic large REM sleep increase during RSD via arousals may be analogous to a much smaller, steady REM sleep increase during drug treatment. In any case, an increase of REM sleep over initial treatment levels, along with the maintenance of a substantial RSD, is consistent with an antidepressant effect by a REM sleep depriving treatment. REM rebound was present if after drug withdrawal REM sleep became significantly greater than its predrug level or was at least twenty percent higher than its predrug level. Very often studies of drug effects on REM sleep reported on only one or two of the three variables that
defined arousal-type RSD. Hence, the determination that a drug did or did not produce arousal-type RSD often required combining the results of several studies. We found no studies that directly examined the relationship between drug arousal-type RSD and drug effect on endogenous depression. Consequently, to determine the relationship between drug effect on REM sleep and drug effect on endogenous depression, we had to combine the results of separate studies on drug REM sleep effects and drug antidepressant effects. The review of drug antidepressant effects on endogenous depression relied on the results of the earlier Morris-Beck review (174) to establish the efficacy of the three standard antidepressant drugs: amitriptyline, desipramine, and imipramine. For newer antidepressant drugs, the efficacy review relied primarily on Excerpta Medica computer bibliographies in the years from 1975 through 1989. For older, first generation antidepressant drugs, we examined the literature back to 1962. All cited English language papers, as well as some foreign language papers, were screened. The Excerpta Medica bibliographies were supplemented by references found in their cited papers that were not listed in the original computer bibliographies. Efficacy papers were included in our review only if they met the following five design and diagnostic criteria. 1) The antidepressant efficacy of a new drug was compared with the efficacy of a placebo or with the efficacy of a standard antidepressant drug (amitriptyline, desipramine or imipramine). 2) Patients were randomly assigned to different treatments. 3) Efficacy was evaluated in a double blind manner. 4) Dropouts from treatment were either included in the drug efficacy analysis or, if not included, consisted of less than twenty-five percent of the initial sample. 5) Treatment duration was at least two weeks. Papers that did not meet all five of these design criteria were excluded from our efficacy analysis. The inclusion of studies that compared new drug with standard drug requires some explanation. The problem here was that placebo response rate was very high in depression (43, 65, 81, 200); e.g., in one study it ranged from thirty-seven percent in manic depressive illness to eighty-three percent in psychoneurotic illness (81). Thus, an equal improvement rate in test drug group and standard drug group might represent a high placebo improvement in both groups rather than a specific improvement by each drug. The best way to overcome this problem was to compare test drug efficacy with placebo efficacy. Unfortunately, the vast majority of drug efficacy trials did not include a parallel placebotreated group. Nevertheless, placebo response rate depended on kind of depression. It was lower in endogenous depression than in nonendogenous depression (81,200). Also the standard antidepressant drugs (amitriptyline, desipramine, and imipramine) were more efficacious than placebo in endogenous depression (174). Thus, a finding that test drug and standard drug had equal efficacy in the treatment of endogenous depression was some evidence that test drug was efficacious, i.e., more effective than placebo. We used this rationale to justify the use of nonplacebo studies that compared the efficacy of test drug and of standard drug. But this, too, presented problems. The method obviously depended on the accuracy of the diagnosis of endogenous depression. Unfortunately, many drug efficacy studies did not use a rigorous endogenous-nonendogenous classification system. We tried to solve this problem as follows: We compared the symptoms of the specified diagnosis with the reliable symptoms of endogenous depression as indicated by symptom consistency across seven empirical factor analytic studies of depression symptoms (164). When most patients with the specified diagnosis probably had many reliable, salient symptoms of endogenous depression, we included the study in our analysis of drug efficacy for endogenous depression. For example, we included as endog-
ANTIDEPRESSIVE DRUGS AND REM SLEEP
51
TABLE 1 ENDOGENOUSAND NONENDOGENOUSDEPRESSIONS Endogenous Depressions
Nonendogenous Depressions
Involutional melancholia (DSM2) (7) Bipolar disorders type I or type 2 (RDC) (236) Major depressive episode with melancholia (DSM3) (8) Bipolar disorder, depressed (DSM3) (8) Primary depression -- endogenous subtype based on symptom profile checklist (58) Major depressive disorder; endogenous subtype (RDC) (236) Manic depressive disorder, circular type, depressed (DSM2) (7) Endogenomorphic depression (126) Endogenous depression (29, 125, 164)
Manic depressive illness, depressed (DSM2) (7) Psychotic depressive reaction (DSM2) (7) Depressive neurosis (DSM2) (7) Reactive, neurotic, or psychoneurotic depression (29, 125, 164) Primary depression (58) Secondary depression (58) Major depressive episode without melancholia (DSM3) (8) Cyclothymic disorder (DSM3) (8) Dysthymic disorder (DSM3) (8) Atypical depression or atypical bipolar disorder (DSM3) (8) Major depressive disorder (RDC) (236)
enous depression studies of patients with major depressive episode with melancholia (DSM 3) (8) because these patients had many of the reliable symptoms of endogenous depression (164,183). And we excluded studies of patients with primary depression because of the report that about fifty percent of patients with primary depression had reactive depression (182). Table 1 shows the diagnoses considered endogenous depressions and included in our review and also shows the diagnoses considered nonendogenous depressions and excluded from our review. RESULTS Drug effects on REM sleep and drug effects on endogenous depression are reported in Tables 2-4. Entries in the tables are the numbers of studies that reported each drug effect listed in the table. Table 2 shows the REM sleep effects of first generation antidepressant drugs. Arousal-type REM sleep deprivation was produced by eight drugs. These were amitriptyline, clomipramine, clorgyline, desipramine, doxepin, imipramine, pargyline and phenelzine. The antidepressant efficacy of six of these drugs was adequately tested. All were efficacious. Three other drugs (nortriptyline, protriptyline and tranylcypromine) had REM sleep profiles consistent with arousal-type REM sleep deprivation but all three properties had not yet been tested in these drugs. Finally two efficacious first generation antidepressant drugs (iprindole and trimipramine) did not produce arousal-type REM sleep deprivation. Table 2 also shows that four second generation antidepressant drugs (amoxapine, butriptyline, viloxazine and zimeldine) produced arousal-type RSD. The four drugs were efficacious in the treatment of endogenous depression. Six other antidepressant drugs had tests for some but not all aspects of arousal-type REM sleep deprivation. The tests were uniformly positive. Three of the six drugs had satisfactory studies of antidepressant efficacy (maprotaline, mianserin and nomifensine) and were found to improve endogenous depression. Two efficacious antidepressant drugs (amineptine and trazodone) did not produce arousal-type RSD. In short, among both first and second generation antidepressant drugs, the findings were consistent with the hypothesis that drugs that produce arousal-type REM sleep deprivation improve endogenous depression. The findings also indicated that not all efficacious antidepressant drugs produced arousal-type REM sleep
deprivation. The effects of nonaddictive, centrally acting drugs on REM sleep are displayed in Table 3. None of these drugs produced unequivocal arousal-type RSD. The only possible exceptions were alprazolam and clonidine. In one of two studies at relatively high doses (>- 3 mg) alprazolam produced large RSD, but no evidence is available on whether such large RSD persisted or was followed by a REM rebound. It is of interest that one (207) of three satisfactory studies (144, 207,217) on alprazolam efficacy in the treatment of endogenous depression found that the drug improved endogenous depression. The other possible exception, clonidine, produced large RSD at higher doses. But again, no evidence was available (from either human or animal studies) [for animal studies see (102, 103, 128, 145, 154, 167, 198,247)] about whether such large RSD was persistent or followed by a REM rebound. We could not find a satisfactory study on the antidepressant effect of clonidine in the treatment of endogenous depression. Scopolamine at high doses also caused a large RSD that was followed by a REM rebound. However, scopolamine did not produce arousal-type RSD because its RSD was not persistent. The other drugs in this group did not produce a large RSD. They either did not reduce REM sleep or had only a small REM suppressant effect. It is to be emphasized that the small RSD produced by these drugs did not come close to the cut off point of a fifty percent REM sleep reduction. Typically in therapeutic doses these drugs reduced REM sleep to eighty-five to ninety percent of its predrug level. Benzodiazepines were among the drugs that produced a small RSD. This was persistent but was not followed by a REM rebound. Thus, none of the nonaddictive drugs produced arousaltype REM sleep deprivation. The several addictive drugs had similar effects on REM sleep (Table 4). Depending on whether their doses were high or low, they produced a large or small RSD. Their RSD was often followed by a REM rebound. High doses almost always produced a REM rebound. They consistently produced a rapid tolerance to their RSD effects. Hence they did not produce arousal-type RSD. It is noteworthy that with somewhat careful criteria for antidepressant efficacy studies, we found only seventy acceptable studies of drug efficacy in the treatment of endogenous depression --and of these only 21 studies included a parallel, placebo-treated control group. Examination of the tables also reveals that for many individual drugs only one or two studies reported the drug effects on REM sleep. In spite of the paucity of REM sleep data on most
52
VOGEL, BUFFENSTEIN, MINTER AND HENNESSEY
TABLE 2 ANTIDEPRESSANT DRUGS
Drug (Therapeutic Dose) First Generation Antidepressant Drugs amitriptyline (150-300 mg) [see review (174)] clomipramine (150-200 mg) clorgyline (20-30 mg) desipramine (150-300 rag) [see review (174)] doxepin (150-300 mg) imipramine (150-300 mg) [see review (174)] iprindole (30-120 mg) nortriptyline (75-100 mg) pargyline (75-90 mg) phenelzine (60-90 mg) protriptyline (40-60 mg) tranylcypromine (20-40 mg) trimipramine (150-300 mg) Totals
Number of Studies Showing: Antidepressant Activity Yes No Large
RSD Small
None
Persistent RSD Yes No
REM Rebound Yes No
1
0
3*
0
0
2
0
1
0
2
0
5
0
0
4
0
3
0
-
-
1
0
0
1
0
1
0
1
0
3*
0
0
3
0
1
0
2
0
2*
0
1
2
0
1
0
1
0
5
0
0
4
0
2
0
2 2
0 0
0 1"
0 0
2 0
. 1
0
-
-
-
-
I
0
0
1
0
1
0
4
0
8
0
0
8
0
4
0
-
-
2*
0
0
2
0
-
-
1
0
.
2
0
2
1
0
1
3
.
18
1
31
1
6
28
16
0
1 5 I
0 0 0 1 0 0 0 0 0 0 1 0 0
2 0 0 0 0 0 0 1 0 0 2 0 0
. 1 1 1 . . . . . . 1 1 2
1 1 -
0 0 -
0 0 0
1 1 1
0 0 0
.
.
.
.
.
.
.
.
.
.
0
Second Generation Antidepressant Drugs amineptine (150-300 mg) amoxapine (150-300 mg) butriptyline (150-200 mg) fluoxetine (60-80 mg) fluvoxamine (150 mg) indalpine (50 mg) maprotiline (150-225 mg) mianserin (30-60 mg) nomifensine (150-300 rag) oxaprotiline (75 mg) trazodone (300-600 rag) viloxazine (200--450 mg) zimelidine (150-300 rag)
10 3 3 10 8 5
0 0 0 0 I 0 2 1 0
Totals
46
4
16
2
5
7
0
5
0
Grand Totals
64
5
47
3
11
35
0
21
0
0 1 1 2* 1 1 1 1 3** 1 0 2 3**
.
.
.
0 0 0 . . . . . .
. . . . . .
. . . . . .
Table entries are numbers of studies showing the drug property listed in the columns. References for the entries in the table are displayed in the supplemental bibliographical table on antidepressant drugs. Large, small, and persistent RSD and REM rebound are defined in the Method section. In a few studies, indicated by asterisks, a large RSD at therapeutic dose was inferred by extrapolation from a dose less than the therapeutic dose. Many studies, involving seventeen distinct antidepressant drugs, indicate that the REM suppressant effect of antidepressant drugs is dose related (6, 16, 48, 91,205,215, 221, 225, 231, 257. 262).
individual drugs, combining the drugs in each group exposed strong differences in the R E M sleep effects o f the different drug groups. Compared with nonaddictive drugs or addictive drugs, the antidepressant drugs showed large RSD more frequently, persistent RSD more frequently and REM rebound more frequently (p
01.49-7634/90 $3~00--+ -.00
Drug Effects on REM Sleep and on Endogenous Depression G. W. V O G E L , I A. B U F F E N S T E I N , K. M I N T E R A N D A N N H E N N E S S E Y
Sleep Laboratory, Department of Psychiatry, Emory University School of Medicine and The Georgia Mental Health Institute, Atlanta, GA 30306 Received 3 N o v e m b e r 1989
VOGEL, G. W., A. BUFFENSTEIN, K. MINTER AND A. HENNESSEY. Drug effects on REM sleep and on endogenous depression. NEUROSCI BIOBEHAV REV 14(1) 49-63, 1990.--In earlier work REM sleep deprivation (RSD) by arousals improved endogenous depression. This suggested that drugs producing a similar RSD would have antidepressant activity. The arousal RSD was large, persisted for weeks, and was followed by a REM rebound. We call RSD with these properties arousal-type RSD. The present study reviewed literature from 1962 to 1989 on drug REM sleep effects to examine the hypothesis that drugs producing arousal-type RSD improve endogenous depression. The literature reviewed concerned the REM sleep effects of amine precursors, antidepressants, antihistamines, antipsychotics, barbiturates, benzodiazepines, other hypnotics, drugs affecting cholinergic and noradrenergic neurotransmission, ethanol, lithium and narcotics. Four hundred and sixty-eight relevant papers were read and 215 contributed information that could be used in the review. The findings indicated that all drugs producing arousal-type RSD improved endogenous depression. Four drugs that improved endogenous depression did not produce arousal-type RSD. REM sleep
REM sleep deprivation
Antidepressant drugs
ABOUT fifteen years ago REM sleep deprivation (RSD) by arousals was reported to improve endogenous depression but not reactive depression (246). At that time it was also noticed that antidepressant drugs produced REM sleep deprivation (242). Taken together, the two findings suggested that antidepressant drugs improved endogenous depression by REM sleep deprivation (242). Two main objections were raised against the hypothesis. One was that drugs without an antidepressant effect, most noticeably addictive drugs, also produced RSD. These drugs included alcohol, amphetamines, barbiturates and narcotics. The answer to this objection involved salient differences between RSD by arousals and RSD by the addictive nonantidepressant drugs. RSD by arousals had three prominent properties (246). 1. Arousal RSD was initially large, i.e., initially REM sleep was reduced by more than fifty percent of its pretreatment level. 2. Arousal RSD was persistent, i.e., substantial reduction of REM sleep, though often less than its initial reduction, was maintained for several weeks. 3. Arousal RSD was followed by a REM rebound, i.e., immediately after RSD was stopped, REM sleep was significantly greater than before RSD was started. We call RSD with these three properties arousal-type RSD. In 1975 a review found that RSD by early antidepressant drugs produced arousal-type RSD (242). These drugs caused a large persistent RSD that was followed by a REM rebound. The review
Endogenous depression
also found, in a less systematic way, that RSD by nonantidepressant drugs, including amphetamines, barbiturates and narcotics, did not show these properties (242). Their RSD was not large, and/or not persistent, and/or did not produce a REM rebound. This led to a refinement of the original hypothesis. The more specific version was that antidepressant drugs improved endogenous depression by arousal-type RSD rather than by an unspecified kind of RSD. A second early objection to the hypothesis was that not all antidepressants produced arousal-type RSD. For example, among first generation antidepressant drugs iprindole was reported to lack a REM sleep suppressant effect (16). This objection led to a much needed correction and clarification. The finding that RSD by arousals was an antidepressant process did not suggest that all antidepressant drugs "worked" by this process. It was certainly possible, even probable, that some antidepressant drugs improved endogenous depression by processes that did not include RSD. Many individual disorders (e.g., hypertension, heart failure, infectious diseases) are improved by several different drugs that work by different processes. But the finding that RSD by arousals was an antidepressant process did suggest a new hypothesis, viz., that all drugs producing arousal-type RSD would have an antidepressant effect on endogenous depression. Indeed, the hypothesis would be disproved by finding a drug that produced arousal-type RSD and did not improve endogenous depression. Fifteen years ago the effects of drugs on REM sleep were reviewed with this hypothesis in mind. The 1975 review supported the hypothesis (242). Since then, many new antidepressant drugs and many new
tRequests for reprints should be addressed to Gerald W. Vogel, M.D., Sleep Lab, GMHI, 1256 Briarcliff Road, N.E., Atlanta, GA 30306.
50
VOGEL, BUFFENSTEIN, MINTER AND HENNESSEY
nonantidepressant drugs affecting the brain have been introduced. This paper updates and incorporates the earlier review on the effects of these drugs on REM sleep and endogenous depression. The aim is to determine current evidence bearing on the hypothesis that drugs producing arousal-type RSD improve endogenous depression. METHOD The present review concemed the REM sleep effects of amine precursors, antidepressants, antihistamines, antipsychotics, barbiturates, other hypnotics (nonbarbiturate, nonbenzodiazepine), benzodiazepines, drugs affecting cholinergic neurotransmission (cholinergic agonists, cholinesterase antagonists and muscarinic antagonists), drugs affecting noradrenergic transmission (alpha agonists, alpha blockers, beta blockers, sympathomimetic drugs, etc.), ethanol, lithium, and narcotics. Studies on drug REM sleep effects in animals and humans were included in the review. In the following Results section only human studies were displayed in the tables. Where relevant, animal studies were mentioned in the text. A bibliography for drug effects on REM sleep was obtained from the annual sleep literature bibliographies published by the Sleep Research Society (23, 31-38, 202). Begun in 1962, these bibliographies included reasonably complete listings of the world literature on sleep. We reviewed the literature on drug REM sleep effects for the annuals dated years 1962-1989. There were three exceptions to this rule. The sleep literature on antipsychotic drugs was examined for 1975-1989. The benzodiazepines sleep literature was examined for 1985-1989. In both cases these years gave a relatively large sample. The literature on adinazolam and alprazolam was reviewed for 1975-1989 because these unusual benzodiazepines had claims of antidepressant activity. The search began with a list of 646 papers on the effects of the reviewed drugs on sleep. By their titles alone 178 papers were excluded as irrelevant and 468 papers were read. Of the latter, 215 papers contributed information that could be used in the review. The included papers explicitly listed baseline REM sleep levels and one or more of the following: initial drug REM sleep levels, later drug period REM sleep levels, and REM sleep levels after drug discontinuation. This information was used to determine whether the drug caused an initial large REM sleep deprivation, a persistent REM sleep deprivation, and/or a REM rebound. Initial large REM sleep deprivation was present when the first one or two nights of drug administration reduced REM sleep to less than fifty percent of its predrug levels. Initial drug REM sleep reductions to more than one-half predrug levels were called small. Whether initial RSD was large or small, persistent RSD was present if at two or more weeks of nightly drug administration, the REM sleep reduction was more than half the initial reduction. An analogous increase of REM sleep occurred during the treatment of endogenous depression by RSD via arousal (246). In that procedure the number of awakenings required for RSD increased progressively on successive nights of RSD. To prevent the number of awakenings from becoming inordinately large, patients had a single night of no awakenings about every fourth night. On that night they had a REM rebound. This periodic large REM sleep increase during RSD via arousals may be analogous to a much smaller, steady REM sleep increase during drug treatment. In any case, an increase of REM sleep over initial treatment levels, along with the maintenance of a substantial RSD, is consistent with an antidepressant effect by a REM sleep depriving treatment. REM rebound was present if after drug withdrawal REM sleep became significantly greater than its predrug level or was at least twenty percent higher than its predrug level. Very often studies of drug effects on REM sleep reported on only one or two of the three variables that
defined arousal-type RSD. Hence, the determination that a drug did or did not produce arousal-type RSD often required combining the results of several studies. We found no studies that directly examined the relationship between drug arousal-type RSD and drug effect on endogenous depression. Consequently, to determine the relationship between drug effect on REM sleep and drug effect on endogenous depression, we had to combine the results of separate studies on drug REM sleep effects and drug antidepressant effects. The review of drug antidepressant effects on endogenous depression relied on the results of the earlier Morris-Beck review (174) to establish the efficacy of the three standard antidepressant drugs: amitriptyline, desipramine, and imipramine. For newer antidepressant drugs, the efficacy review relied primarily on Excerpta Medica computer bibliographies in the years from 1975 through 1989. For older, first generation antidepressant drugs, we examined the literature back to 1962. All cited English language papers, as well as some foreign language papers, were screened. The Excerpta Medica bibliographies were supplemented by references found in their cited papers that were not listed in the original computer bibliographies. Efficacy papers were included in our review only if they met the following five design and diagnostic criteria. 1) The antidepressant efficacy of a new drug was compared with the efficacy of a placebo or with the efficacy of a standard antidepressant drug (amitriptyline, desipramine or imipramine). 2) Patients were randomly assigned to different treatments. 3) Efficacy was evaluated in a double blind manner. 4) Dropouts from treatment were either included in the drug efficacy analysis or, if not included, consisted of less than twenty-five percent of the initial sample. 5) Treatment duration was at least two weeks. Papers that did not meet all five of these design criteria were excluded from our efficacy analysis. The inclusion of studies that compared new drug with standard drug requires some explanation. The problem here was that placebo response rate was very high in depression (43, 65, 81, 200); e.g., in one study it ranged from thirty-seven percent in manic depressive illness to eighty-three percent in psychoneurotic illness (81). Thus, an equal improvement rate in test drug group and standard drug group might represent a high placebo improvement in both groups rather than a specific improvement by each drug. The best way to overcome this problem was to compare test drug efficacy with placebo efficacy. Unfortunately, the vast majority of drug efficacy trials did not include a parallel placebotreated group. Nevertheless, placebo response rate depended on kind of depression. It was lower in endogenous depression than in nonendogenous depression (81,200). Also the standard antidepressant drugs (amitriptyline, desipramine, and imipramine) were more efficacious than placebo in endogenous depression (174). Thus, a finding that test drug and standard drug had equal efficacy in the treatment of endogenous depression was some evidence that test drug was efficacious, i.e., more effective than placebo. We used this rationale to justify the use of nonplacebo studies that compared the efficacy of test drug and of standard drug. But this, too, presented problems. The method obviously depended on the accuracy of the diagnosis of endogenous depression. Unfortunately, many drug efficacy studies did not use a rigorous endogenous-nonendogenous classification system. We tried to solve this problem as follows: We compared the symptoms of the specified diagnosis with the reliable symptoms of endogenous depression as indicated by symptom consistency across seven empirical factor analytic studies of depression symptoms (164). When most patients with the specified diagnosis probably had many reliable, salient symptoms of endogenous depression, we included the study in our analysis of drug efficacy for endogenous depression. For example, we included as endog-
ANTIDEPRESSIVE DRUGS AND REM SLEEP
51
TABLE 1 ENDOGENOUSAND NONENDOGENOUSDEPRESSIONS Endogenous Depressions
Nonendogenous Depressions
Involutional melancholia (DSM2) (7) Bipolar disorders type I or type 2 (RDC) (236) Major depressive episode with melancholia (DSM3) (8) Bipolar disorder, depressed (DSM3) (8) Primary depression -- endogenous subtype based on symptom profile checklist (58) Major depressive disorder; endogenous subtype (RDC) (236) Manic depressive disorder, circular type, depressed (DSM2) (7) Endogenomorphic depression (126) Endogenous depression (29, 125, 164)
Manic depressive illness, depressed (DSM2) (7) Psychotic depressive reaction (DSM2) (7) Depressive neurosis (DSM2) (7) Reactive, neurotic, or psychoneurotic depression (29, 125, 164) Primary depression (58) Secondary depression (58) Major depressive episode without melancholia (DSM3) (8) Cyclothymic disorder (DSM3) (8) Dysthymic disorder (DSM3) (8) Atypical depression or atypical bipolar disorder (DSM3) (8) Major depressive disorder (RDC) (236)
enous depression studies of patients with major depressive episode with melancholia (DSM 3) (8) because these patients had many of the reliable symptoms of endogenous depression (164,183). And we excluded studies of patients with primary depression because of the report that about fifty percent of patients with primary depression had reactive depression (182). Table 1 shows the diagnoses considered endogenous depressions and included in our review and also shows the diagnoses considered nonendogenous depressions and excluded from our review. RESULTS Drug effects on REM sleep and drug effects on endogenous depression are reported in Tables 2-4. Entries in the tables are the numbers of studies that reported each drug effect listed in the table. Table 2 shows the REM sleep effects of first generation antidepressant drugs. Arousal-type REM sleep deprivation was produced by eight drugs. These were amitriptyline, clomipramine, clorgyline, desipramine, doxepin, imipramine, pargyline and phenelzine. The antidepressant efficacy of six of these drugs was adequately tested. All were efficacious. Three other drugs (nortriptyline, protriptyline and tranylcypromine) had REM sleep profiles consistent with arousal-type REM sleep deprivation but all three properties had not yet been tested in these drugs. Finally two efficacious first generation antidepressant drugs (iprindole and trimipramine) did not produce arousal-type REM sleep deprivation. Table 2 also shows that four second generation antidepressant drugs (amoxapine, butriptyline, viloxazine and zimeldine) produced arousal-type RSD. The four drugs were efficacious in the treatment of endogenous depression. Six other antidepressant drugs had tests for some but not all aspects of arousal-type REM sleep deprivation. The tests were uniformly positive. Three of the six drugs had satisfactory studies of antidepressant efficacy (maprotaline, mianserin and nomifensine) and were found to improve endogenous depression. Two efficacious antidepressant drugs (amineptine and trazodone) did not produce arousal-type RSD. In short, among both first and second generation antidepressant drugs, the findings were consistent with the hypothesis that drugs that produce arousal-type REM sleep deprivation improve endogenous depression. The findings also indicated that not all efficacious antidepressant drugs produced arousal-type REM sleep
deprivation. The effects of nonaddictive, centrally acting drugs on REM sleep are displayed in Table 3. None of these drugs produced unequivocal arousal-type RSD. The only possible exceptions were alprazolam and clonidine. In one of two studies at relatively high doses (>- 3 mg) alprazolam produced large RSD, but no evidence is available on whether such large RSD persisted or was followed by a REM rebound. It is of interest that one (207) of three satisfactory studies (144, 207,217) on alprazolam efficacy in the treatment of endogenous depression found that the drug improved endogenous depression. The other possible exception, clonidine, produced large RSD at higher doses. But again, no evidence was available (from either human or animal studies) [for animal studies see (102, 103, 128, 145, 154, 167, 198,247)] about whether such large RSD was persistent or followed by a REM rebound. We could not find a satisfactory study on the antidepressant effect of clonidine in the treatment of endogenous depression. Scopolamine at high doses also caused a large RSD that was followed by a REM rebound. However, scopolamine did not produce arousal-type RSD because its RSD was not persistent. The other drugs in this group did not produce a large RSD. They either did not reduce REM sleep or had only a small REM suppressant effect. It is to be emphasized that the small RSD produced by these drugs did not come close to the cut off point of a fifty percent REM sleep reduction. Typically in therapeutic doses these drugs reduced REM sleep to eighty-five to ninety percent of its predrug level. Benzodiazepines were among the drugs that produced a small RSD. This was persistent but was not followed by a REM rebound. Thus, none of the nonaddictive drugs produced arousaltype REM sleep deprivation. The several addictive drugs had similar effects on REM sleep (Table 4). Depending on whether their doses were high or low, they produced a large or small RSD. Their RSD was often followed by a REM rebound. High doses almost always produced a REM rebound. They consistently produced a rapid tolerance to their RSD effects. Hence they did not produce arousal-type RSD. It is noteworthy that with somewhat careful criteria for antidepressant efficacy studies, we found only seventy acceptable studies of drug efficacy in the treatment of endogenous depression --and of these only 21 studies included a parallel, placebo-treated control group. Examination of the tables also reveals that for many individual drugs only one or two studies reported the drug effects on REM sleep. In spite of the paucity of REM sleep data on most
52
VOGEL, BUFFENSTEIN, MINTER AND HENNESSEY
TABLE 2 ANTIDEPRESSANT DRUGS
Drug (Therapeutic Dose) First Generation Antidepressant Drugs amitriptyline (150-300 mg) [see review (174)] clomipramine (150-200 mg) clorgyline (20-30 mg) desipramine (150-300 rag) [see review (174)] doxepin (150-300 mg) imipramine (150-300 mg) [see review (174)] iprindole (30-120 mg) nortriptyline (75-100 mg) pargyline (75-90 mg) phenelzine (60-90 mg) protriptyline (40-60 mg) tranylcypromine (20-40 mg) trimipramine (150-300 mg) Totals
Number of Studies Showing: Antidepressant Activity Yes No Large
RSD Small
None
Persistent RSD Yes No
REM Rebound Yes No
1
0
3*
0
0
2
0
1
0
2
0
5
0
0
4
0
3
0
-
-
1
0
0
1
0
1
0
1
0
3*
0
0
3
0
1
0
2
0
2*
0
1
2
0
1
0
1
0
5
0
0
4
0
2
0
2 2
0 0
0 1"
0 0
2 0
. 1
0
-
-
-
-
I
0
0
1
0
1
0
4
0
8
0
0
8
0
4
0
-
-
2*
0
0
2
0
-
-
1
0
.
2
0
2
1
0
1
3
.
18
1
31
1
6
28
16
0
1 5 I
0 0 0 1 0 0 0 0 0 0 1 0 0
2 0 0 0 0 0 0 1 0 0 2 0 0
. 1 1 1 . . . . . . 1 1 2
1 1 -
0 0 -
0 0 0
1 1 1
0 0 0
.
.
.
.
.
.
.
.
.
.
0
Second Generation Antidepressant Drugs amineptine (150-300 mg) amoxapine (150-300 mg) butriptyline (150-200 mg) fluoxetine (60-80 mg) fluvoxamine (150 mg) indalpine (50 mg) maprotiline (150-225 mg) mianserin (30-60 mg) nomifensine (150-300 rag) oxaprotiline (75 mg) trazodone (300-600 rag) viloxazine (200--450 mg) zimelidine (150-300 rag)
10 3 3 10 8 5
0 0 0 0 I 0 2 1 0
Totals
46
4
16
2
5
7
0
5
0
Grand Totals
64
5
47
3
11
35
0
21
0
0 1 1 2* 1 1 1 1 3** 1 0 2 3**
.
.
.
0 0 0 . . . . . .
. . . . . .
. . . . . .
Table entries are numbers of studies showing the drug property listed in the columns. References for the entries in the table are displayed in the supplemental bibliographical table on antidepressant drugs. Large, small, and persistent RSD and REM rebound are defined in the Method section. In a few studies, indicated by asterisks, a large RSD at therapeutic dose was inferred by extrapolation from a dose less than the therapeutic dose. Many studies, involving seventeen distinct antidepressant drugs, indicate that the REM suppressant effect of antidepressant drugs is dose related (6, 16, 48, 91,205,215, 221, 225, 231, 257. 262).
individual drugs, combining the drugs in each group exposed strong differences in the R E M sleep effects o f the different drug groups. Compared with nonaddictive drugs or addictive drugs, the antidepressant drugs showed large RSD more frequently, persistent RSD more frequently and REM rebound more frequently (p
