Trimipramine in Primary Insomnia: Results of a Polysomnographic Double-Blind Controlled Study

D. Riemann1 U. Voderholzer1

S. Cohrs2 A. Rodenbeck2

C. Hajak2 E. Riither2

M. H. Wiegand3 C. Laakmann4

T. Baghai4 W. Fischer5

M. Hoffmann6 F. Hohagen7

C. Mayer8 M. Berger1

Trimipramine in Primary Insomnia: Results of a Polysomnographic Double-Blind Controlled Study

In recent years, sedating antidepressants have been increasingly used to treat primary insomnia. Up to now, only one open pilot study with trimipramine and one double-blind placebo-controlled study with doxepin have leant scientific support for this approach in treating primary insomnia. In order to test the hypothesis that sedating antidepressants are useful in the treatment of primary insomnia, the effect of trimipramine on objectively and subjectively measured parameters of sleep was investigated in a double-blind placebo- and lormetazepam-controlled study in a sample of 55 patients with primary insomnia attending outpatient sleep-disorder clinics. Trimipramine was selected since it has shown positive effects on sleep continuity with a lack of REM sleep suppression in studies on depressed patients and in one pilot study on patients with primary insomnia. Trimipramine at an average dose of 100 mg over a period of 4 weeks significantly enhanced sleep efficiency, but not total sleep

time (which had been the primary target variable) compared to placebo as measured by polysomnography. Changes in objective sleep parameters were paralleled by changes in subjective sleep parameters. Trimipramine did not suppress REM sleep. Lormeta- zepam decreased wake time and sleep stage 3 and increased REM sleep compared to placebo. After switching trimipramine to placebo, sleep parameters returned to baseline. There was no evidence of any rebound effect from trimipramine. Side effects from trimipramine were only marginal. This first double-blind placebo-controlled study with trimipramine suggests its efficacy in the treatment of primary insomnia. However, due to the large intra- and interindividual variance in the parameters of interest before and during treatment a larger sample size would have been necessary to strengthen the va- lidity of our findings.

Introduction

Primary insomnia is defined as prolonged sleep latency, disturbance of sleep maintenance, early morning awakening, unrefresh- ing sleep and resulting impairments of daytime professional or

social activities lasting at least for four weeks according to DSM-III-R [1 ] and DSM-IV [2].

In order to diagnose primary insomnia, an underlying psychiatric or organic cause has to be excluded. It is assumed that about 25 to

Affiliation 1 Department of Psychiatry and Psychotherapy, University of Freiburg, Germany

2 Department of Psychiatry and Psychotherapy, University of Gottingen, Germany 3 Department of Psychiatry and Psychotherapy, Technical University of Munich, Germany

4 Department of Psychiatry, University of Munich, Germany 5 Aventis Pharma GmbH, Bad Soden, Germany

6 Department of Psychiatry, University of Erlangen-Nuremberg, Germany 7 Department of Psychiatry and Psychotherapy, University of Lubeck, Germany

8 Neurological Hospital Hephata, Schwalmstadt-Treysa, Germany Correspondence

Prof. Dr. Dieter Riemann • Department of Psychiatry and Psychotherapy • University of Freiburg • Hauptstrasse 5 • 79104 Freiburg • Phone: +49 (761) 270-6919 • Fax: +49 (761) 270-6667 •

E-Mail: [email protected] Funding

The study was supported and sponsored by Aventis Pharma GmbH, Bad Soden, Germany (formerly Rhone Poulenc Rorer GmbH, Cologne, Germany)

Pharmacopsychiatry, 2002, 35, 165-174

35 % of all patients fulfilling the insomnia criteria suffer from a primary insomnia in contrast to an organic or psychiatric insomnia. This equals about 5 to 7% of the total population, for example, in Germany, when considering epidemiological studies, which indicate a prevalence of insomnia according to DSM-III-R or DSM-1V criteria of about 15 to 25% in the general population or in medical settings such as in primary care [17,19,40,42,49,52,53].

Until a decade ago, patients with primary insomnia treated phar- macologically were mainly administered classical benzodiazepine hypnotics, or since their introduction, new benzodiazepine receptor agonists such as zolpidem or zopiclone.

Problems inherent in the application of benzodiazepines such as tolerance, dependence, addiction, withdrawal and rebound phe- nomena (for an overview, see [15]) have led to a steady decline in the prescription of benzodiazepine hypnotics over the last 10 years in Germany [41 ]. Results from the US indicate that the decline in classical benzodiazepine prescriptions have been made up by the increased prescription of new benzodiazepine-receptor agonists and sedating antidepressants [46]. In the US, the rate of sedating antidepressant (such as trazodone) prescriptions for primary insomnia have increased by about 150% over a time span of 10 years. There is no corresponding specific pharmacoe- pidemiological information for Germany, but it seems reasonable to assume that comparable changes concerning the pre- scribing behavior of specialists and general physicians have taken place, leading to increased rates of the prescription of sedating antidepressants, such as doxepin and trimipramine, for primary insomnia.

Up to now, this practice has only scarcely been supported by randomized placebo-controlled clinical trials. For trimipramine, an open pilot study in 17 patients with primary insomnia indicated an increase of one hour in total sleep time after 4 weeks of treatment at an average dose of 160 mg [18]. For doxepin, the first double-blind placebo controlled study was performed in Germa- ny [16]. In this study administration of 25 to 50 mg of doxepin compared to placebo increased total sleep time for an average duration of approximately 60 minutes in 20 patients with primary insomnia.

This phase III study was undertaken to evaluate the effect of trimipramine in comparison to placebo and a classical benzodiazepine hypnotic (lormetazepam) on objectively and subjectively recorded measures of sleep in a sample of primary insomniacs.

Trimipramine was chosen for its strong sedative properties and its positive results from a pilot study [18]. Trimipramine has a tricyclic structure on the one hand, but on the other, it differs strongly from all other tricyclic antidepressants due to its lack of noradrenalin- and 5-HT-reuptake inhibition or downregula- tion of beta-adrenoreceptors. The substance is a strong blocker of histamine 1 and 5-HT2 receptors, and additionally blocks dopamine receptors. The receptor binding affinities for trimipramine are similar to clozapine [7]. From a pharmacological point of view, trimipramine is an atypical substance among the tricyclic antidepressants. Its strong sedative properties are thought to be related to HI-, 5-HT2-, and probably also dopamine-receptor antagonism.

From a clinical point of view, trimipramine exhibits similar side effects to other classical antidepressants, but has only minimal cardiovascular effects [29].

Trimipramine's effects on sleep are especially worth noting. Sev- eral studies on depressed patients have demonstrated its positive effects on sleep continuity, with no REM sleep suppression [34,44,47,51,54]. In a first open pilot study, the sleep inducing effects were confirmed by polysomnography in patients with primary insomnia [18].

Subjects and Methods

Subjects were outpatients attending 12 sleep disorders clinics mainly located at psychiatric university hospitals throughout Germany (see 'trimipramine insomnia study group' in the acknowledgements). Patients had been referred by their primary care providers to the sleep disorder clinics for evaluation and therapy of their insomnia.

The following exclusion criteria were applied: any kind of serious organic disorder, any psychiatric disorder at the time of investigation or life time; pregnancy or risk of pregnancy; sleep apneas (Apnea index > 5/h) or PLMS (periodic leg movements during sleep; PLMS arousal index > 5/h).

In order to confirm exclusion criteria prior to the study, all patients underwent the following procedures: ECG, EEC, routine blood and urine tests, medical and neurological investigation; psychiatric examination to rule out any psychiatric disorder ex- cept primary insomnia; all patients had to score < 20 on the MADRS (Montgomery Asberg Depression Rating Scales [25]).

The following inclusion criteria had to be fulfilled by subjects to be enrolled in the study: Age from 18 to 70 years of both sexes, Caucasian ethnicity, outpatients, suffering from a DSM-III-R sleep disorders fulfilling the criteria for primary insomnia or dys- somnia not otherwise classified (verified by the Structured Inter- view for Sleep Disorders; SIS-D [39]). Patients had to suffer from insomnia for at least one month and had to have a regular bedtime at approximately 11 p.m. (± 60 minutes).

All patients gave their informed written consent prior to inclusion in the study. The study protocol was approved by the ethical committee of the University Hospital of Freiburg and by all local ethical committees of the centers participating in the study.

Altogether, 65 patients were recruited for the study and were randomized. Twenty-one were randomized to placebo, 22 to lormetazepam and 22 to trimipramine. As the intent-to-treat (ITT) population, 55 patients were considered. The 10 patients who were not considered for the ITT population dropped out due to the following reasons: 4 patients quit the study during the wash-out phase (three of these patients withdrew their consent and an intervening illness occurred in one patient, warranting a withdrawal from the study). Further 6 patients (3 placebo; 2 lormetazepam; 1 trimipramine) stopped the intake of study medication prior to day 7 of the study. It turned out after the start that patients fulfilled an exclusion criterion (relevant sleep apnea and hypothyroiditis) in two cases, and patients withdrew their con-


sent in three cases ,and an adverse event (unrelated to the study medication) precluded continuation of the study in one case.

In the ITT population, 18 patients in the placebo group, 18 patients in the lormetazepam group and 19 patients in the trimipramine group were considered. Table 1 shows demographic and psychometric data for this population.

Subjects in the three different groups did not differ from each other concerning any of the listed variables.

Study design Prior to the study, all subjects had to undergo a 7-day wash-out phase from any kind of psychoactive premedication (Days -14 to -8 of the study). From day -7 to day -1, all subjects received placebo. During this phase (placebo run-in), patients were carefully screened to meet the inclusion and exclusion criteria; psychometric interviews and psychometric testing was performed. Subjects underwent an ECG, EEC and routine laboratory blood testing. Dur- ing the last two nights of the placebo run-in phase, two polysom- nographies were registered in the sleep laboratory. The first (night -2) served for adaptation to the sleep laboratory and exclusion of relevant PLMS and apneas, the second sleep laboratory night (night -1) served as baseline night. After that, subjects were randomized to the placebo, trimipramine or lormetazepam treatment.

Patients were seen by the study physician (in general a trained psychiatrist) at 7-day intervals to dispense study medication, to document adverse events and to collect safety data. After 26 days of treatment, subjects returned to the sleep laboratory for 2 nights (night 27 and 28). The first night was an adaptation night without any analysis, and the second night served as the target night to determine effects of the different drugs on polysomnographically recorded sleep. After this laboratory night, drugs were discontinued and exchanged for placebo, which was given for a further 14 days. Two more sleep recordings were performed. The first was 2 days after the last drug intake during the study (night 30) and the second after 12 days (night 42). These two nights served to measure discontinuation effects.

Polysomnography and subjective sleep parameters Sleep was recorded and scored according to standard criteria [33]. Methods were standardized for all participating sleep centers. Medical and technical staff from all participating centers took part in a one-day polysomnography training course before the start of the study.

.Table 1 Description of samples (intent to treat population)

Placebo Lormetazepam Trimipramine

n = 18 n = 18 n = 19 Age (years) 48.8 ± 11.6 45.3 ± 10.3 47.0 ± 10.8

Male : Female 13:5 9:9 10:9 BMI1 (kg:m2) 24.3 ± 3.7 24.5 ± 4.3 23.8 ±3.1 MADRS2 8.7 ± 3.9 10.3 ±3.8 11. 5 ±4.4

Data expressed as the mean ± SD where appropriate.' BMI, Body mass index;2 MADRS, Mont- gomery-Asberg Depression Rating Scale [25]

Polysomnography encompassed EEC (C3-A2, C4-A1), horizontal and vertical eye movements and submental EMG. Musculus anterior tibialis and respiration recording (oral/nasal air flow and thoracal/abdominal respiratory effort and oxygen satura- tion) were only performed in the first night (night -2) in the sleep laboratory.

All recordings were performed from 11 p.m. to 7 a.m. All sleep recordings were scored centrally according to Rechtschaffen and Kales, criteria at the University of Freiburg sleep center by an experienced rater (see acknowledgements) who was unaware of the experimental conditions of the recordings [33]. Recordings were analyzed by epoch (30 s length).

The following sleep parameters were determined: total sleep time (min), sleep efficiency (% = total sleep time:time in bed x 100), latency to stage 2 (sleep latency in min, i.e., time from lights out until the first occurrence of stage 2) and latency to stage REM (min, 1. e. time from the first occurrence of stage 2 until the first occurrence of REM sleep), number of wake periods; wake, stage 1, stage 2, stages 3, 4 and REM expressed as %SPT (Sleep Period Time = time between sleep onset and final awakening).

To determine subjective parameters of sleep the SF-A [14] and the PSQI were used (Pittsburgh Sleep Quality Index [10]; German ver- sion: [35]). The SF-A was used on all mornings in the sleep laboratory to measure the sleep quality of the previous night, whereas the PSQI was used prior to the study, after 4 weeks treatment and after 2 weeks of drug discontinuation to judge several aspects of sleep and overall sleep quality in the preceding 2 weeks.

For the PSQI, the sum score (range: 0-21; 0 = no sleep disturbance at all, 21 = severe sleep disturbance) was included in the data analysis.

For the SF-A the subscales 'sleep quality,' 'feeling refreshed in the morning', 'well-being in the evening', 'exhaustedness in the evening', and 'psychosomatic symptoms during sleep' were analyzed. These scales range from 1 to 5, with 1 denoting impaired quality etc., whereas a score of 5 represents positive estimates ('psychosomatic symptoms' have inverse scaling).

Measurement of side effects/safety data Routine blood tests (hematology, renal and liver-function, glu- cose) were performed on day -1 and on day 28 parallel to a clinical examination including ECG and urine testing.

The Withdrawal Symptom Scale (WSS [24]) was performed on days -1,1,15, 27, 30, 35, 42 by the patients as a self-rating scale. This scale encompasses 20 symptoms that can be rated on 4-point scales (1 = not present; 4 = present always or most of the time).

In order to confirm compliance with the study protocol on day -2, days 14, and day 28, a urine sample was taken to test whether an additional intake of benzodiazepines had occurred in the placebo or trimipramine group (none of which was positive - retrospective analysis after completion of the study).

Study drugs/dosage During the placebo run-in phase (day -7 to -1) and the discontinuation phase (day 29-42), patients received a blister contain-


ing 28 or 56 capsules as applicable, and were instructed to take 4 capsules an hour before bedtime.

Subjects randomized to placebo were given weekly blisters containing 7 x 4 capsules of placebo to take each evening from day 1 to day 28.

Subjects in the lormetazepam condition received the same amount of capsules (with the first capsule containing 1 mg lormetazepam) from day 1 to day 28.

Within the trimipramine group, subjects also received weekly blisters ( 7 x 4 capsules). Flexible dosing of the drug was possible from day 1 to day 28. Dosage steps: day 1-2, 25 mg; day 3-4, 50 mg; day 5 - 6, 75 mg; day 7-9,100 mg; day 10,150 mg; day 11-28; 200 mg. Dosage could be varied starting on day 4 according to the reported effectiveness of the drug. Hence, trimipramine dosage at the end of the study could vary from 50 - 200 mg.

Subjects could also vary their intake of the number of pills upon consultation with the study physician in the placebo and the lormetazepam group. Patients in all conditions were instructed to phone their study physician at 48-hour intervals to report the effectiveness of the drugs on their sleep to adjust the dose. This procedure did not lead to changes in the actual drug dosage in the placebo/lormetazepam group, but only in the trimipramine group.

Blisters were collected by the study physicians at weekly intervals. Retrospective analysis of blisters revealed that all of the patients had complied with the protocol.

Statistics/power calculation The study was aimed at the confirmatory testing of an assumed difference between trimipramine and placebo. Comparisons between trimipramine and lormetazepam and between lormetazepam and placebo were performed on an exploratory basis.

The primary target variable for the confirmatory test to determine differences between trimipramine and placebo was an expected increase of total sleep time for 60 minutes after 4 weeks of treatment with trimipramine compared to baseline values. This was based on our pilot study [18], where an increase of 60 minutes in total sleep time was observed with trimipramine. No mean change of total sleep time was expected for placebo. For a mean difference between trimipramine and placebo of 60 ± 60 min (SD) with a set at 0.05 (two-sided) and /3 set at 0.20, seven- teen patients per group seemed necessary.

For descriptive statistics, calculation of mean ± SD, median, mini- mum and maximum values were performed. For inferential statistics, two-tailed t-tests for dependent or independent samples with a level of significance of p < 0.05 were used. The confirmatory analysis for total sleep time was based on the intent to treat population. For dropouts during the treatment period, the LVCF (Last Value Carried Forward) method was used - values for subjects that did not complete night 28 in the sleep laboratory were substituted with those from the baseline night (indicating a zero therapeutic effect). T-tests (two-tailed) for independent samples were also used for all exploratory analyses without a adjust- ment.

Results

Four-week treatment period Dropouts/'adverse events/safety: Nine out of 55 patients did not complete the 28-day treatment period. The 9 dropouts were dis- tributed evenly to the placebo, lormetazepam, or trimipramine conditions. The reasons for dropping out were: lack of efficiency of study drug (four cases: two patients with lormetazepam, two with placebo), adverse events (two cases: all in the trimipramine group; one patient stopped the study because of strong 'hang- over' effects in the morning; the other patient experienced simul- taneously headache, dizziness and nausea), reversal of initial diag- nosis (two cases: one patient with trimipramine, one with lormetazepam; in both cases detection of clinically relevant periodic leg movements in sleep) and withdrawal of consent (one case: placebo group) without specific reason given. The 9 dropouts were re- tained for the ITTanalysis. Three of the dropouts occurred on day 8 of the study, two on day 15, three on day 22 and one on day 27.

Regarding ECG parameters, no relevant changes were noted from day -1 through day 28. The same applied to the physical examination, vital parameters and laboratory results (hematology and serology, urine parameters).

Concerning adverse events (AE), 14 patients (77.8%) in the placebo group reported 32 AEs, 6 patients (33.3%) in the lormetazepam group reported 13 AEs and 15 patients (78.9%) reported 42 AEs under trimipramine conditions. Adverse events such as dizziness, dry mouth, headache and nausea were more pronounced in the trimipramine group (see Table 2).

The intensity of symptoms was ranked as severe for three events (experienced by two patients) in the placebo group, two events (experienced by one patient) in the lormetazepam group and for seven events (experienced by three patients) in the trimipramine group.

Symptoms were judged as probably due to study medication in 12 events (six patients) in the placebo group, in two events (two patients) in the lormetazepam group and in 10 events (five patients) in the trimipramine group.

Interestingly, five patients from the placebo group already experienced adverse events during placebo run-in period, and six patients each in the lormetazepam and trimipramine groups.

Polysomnography Mean dosage of trimipramine after 4 weeks was 109.4 ± 63.8 mg 5' (16 patients). The final doses were as follows: 50 mg, seven pa-

Table 2 Adverse events during the treatment period (ranked accord- j ing to their frequency) Placebo n Lormetazepam n Trimipramine n

Asthenia 2 Headache 1 Dizziness 9 C

Obstipation 2 Accomodation problems 1 Dry mouth 5 Diarrhea 2 Asthenia 1 Headache 4 Headache ? Cystitis 1 Nausea 4 Nervousness 2 Vertigo 1 Asthenia T Others 22 Others 8 Others 18


tients; lOOmg, three patients; 150mg, two patients; 200 mg, four patients.

Information on polysomnographically recorded sleep variables are shown in Table 3.

Fig.l shows single values for the primary target parameter total sleep time for the three different treatment groups.

With trimipramine, mean total sleep time increased from 349 minutes to 406 minutes. In the placebo group, a mean increase of 9 minutes occurred, which was 18 minutes in the lormetazepam group. Confirmatory testing of the difference between trimipramine and placebo for the increase in total sleep time by t- test (two-tailed, independent samples) showed a non-significant result (p = 0.1007).

Exploratory t-tests for the significance for the other sleep parameters comparing all the three different conditions with each other are shown in Table 4.

With this analysis, a highly significant difference between the effects of placebo and trimipramine on sleep efficiency was observed. The difference between placebo and lormetazepam almost reached significance for this parameter.

Comparing trimipramine vs. placebo, a non-significant trend (p < 0.10) was observed for stage wake %SPT, but for none of the other sleep parameters apart from sleep efficiency.

There was no statistically significant difference between lormetazepam and trimipramine when analyzing differences from

Fig. 1 Single values of total sleep time (min) for placebo, lormetazepam and trimipramine.

night -1 versus 28 in any of the polysomnographic sleep parameters.

For lormetazepam versus placebo the reduction of stage wake %SPT observed on night 28 compared to night -1 was significant. Furthermore, lormetazepam versus placebo led to a significant decrease of stage 3 %SPT and a significant increase of REM sleep %SPT after four weeks.

Subjective sleep variables (SFA/PSQI) Results for the subjective sleep variables concerning means ± SD and the statistical analysis are also included in Tables 3 and 4.

Table 3 Objective and subjective sleep variables at baseline and after 4 weeks treatment (mean ± SD)

Placebo n = 18 Baseline 4-wk treatment (Night -1) (Night 28)

Lormetazepam n = 18 Baseline 4-wk treatment (Night -1) (Night 28)

Trimipramine n = 19 Baseline 4-wk treatment (Night -1) (Night 28)

Polysomnography Total sleep time min. 371. 00 ± 105.61 380.44 ± 89.38 390.31 ± 40.30 408.61 ± 47.29 348.92 ± 84.99 406.1 3 ± 77.25 Sleep efficiency % 80.82 ± 15.20 77.71 ± 17.49 82.43 ± 8.72 86.24 ± 8.05 74.78 ±16.37 84.53 ± 15.20 Sleep latency min. 21.42 ± 23.60 34.03 ± 46.21 25.03 ± 15.77 26.31 ± 33.61 26.08 ±22.11 23.34 ± 24.45 No. wake periods 16.94± 8.56 20.67 ± 12.39 18.72± 11.66 18.17± 8.70 16.74± 7.69 16.74± 8.91 Stage Wake % SPT1 14.09± 14.30 15.54± 14.10 11.54± 7.76 6.79 ± 4.01 18.12 ±15.52 11.19* 13.94 Stage 1 % SPT 7.03 ± 3.27 8.76 ± 3.30 10.04± 4.99 10.29± 5.25 8.91 ± 4.83 11.46 + 7.42 Stage 2% SPT 50.99 ± 11.46 51.26 ± 9.38 54.22 ± 10.16 59.21 ± 7.12 45.29 ± 11.25 48.48 ± 7.58 Stage 3% SPT 2.79 ± 2.35 3.20 ± 2.92 3.74 ± 3.17 2.13± 2.69 5.36 ± 3.54 5.54 ± 4.98 Stage 4% SPT 3.22 ± 5.33 1.72± 2.95 1.78± 4.49 .1.22* 4.99 3.03 ± 5.43 3.35 ± 6.06 Stage REM % SPT 21.37± 5.10 19.01 ± 6.10 18.10 ± 7.29 19.87± 7.84 18.84± 8.37 19.38± 8.60 REM latency min. 88.00 ± 59.63 95.22 ± 60.54 99.11 ± 71.27 82.86 ± 44.14 120.03 ±87.86 125.21 ± 117.23 SFA2 Sleep quality 2.88± 0.85 2.48 ± 0.82 2.89± 0.72 3.19 ± 0.89 2.54 ± 0.81 2.89 ± 0.61 Feeling rested (morning) 3.02 ± 0.95 2.82 ± 1.05 3.10 ± 0.74 2.92 ± 0.87 2.56 ± 0.98 3.08 ± 0.72 Well-being (evening) 3.55± 0.68 3.09 ± 0.99 3.28 ± 0.76 3.41 ± 0.58 3.05 ± 0.69 3.57 ± 0.44 Exhaustedness (evening) 2.57 ± 1.00 2.56± 0.75 2.62 ± 0.73 2.90 ± 0.73 2.73 ± 0.59 2.59 * 0.74 Psychosomatic symptoms 1.89± 0.42 1.99± 0.62 1.93± 0.46 1.71 ± 0.59 1.75± 0.45 1.58 + 0.35 PSQI3 Sum score 11.79± 3.81 12.0 ± 4.04 10.72 ±2.42 8.39 ± 3.36 13.28+ 2.08 9.39 ± 3.35

1 SPT, Sleep Period Time (time from sleep onset until final awakening);2 SFA, Sleep questionnaire according to Cdrfe/meyer (see method section) [14];3 PSQI, Pittsburgh Sleep Quality Index (see method section) [10]


Table 4 Results of statistical evaluation - comparison of Delta-values (night 28 - (-1 )) for all sleep variables between the three conditions (two- tailed t-tests for independent samples)

Placebo vs. lormetazepam Values ofp

Placebo vs. trimipramine Lormetazepam vs. trimipramine

Polysomnography Total sleep time min. 0.7151 0.1007 0.1094 Sleep efficiency^ 0.0610 0.0048" 0.2163 Sleep latency min. 0.3926 0.1726 0.6750 No. wake periods 0.1440 0.1164 0.8396 Stage Wake %SPT' 0.0196* 0.0572 0.6173 Stage 1 %SPT 0.1679 0.6682 0.2390 Stage2%SPT 0.1223 0.3497 0.5779 Stage3%SPT 0.0376* 0.8206 0.0858 Stage 4%SPT 0.4594 0.1398 0.4874 Stage REM %SPT 0.0158" 0.1895 0.5534 REM latency min. 0.1886 0.9369 0.4468 SFA2 Sleep quality 0.0109* 0.0093** 0.8366 Feeling rested (morning) 0.9222 0.0175* 0.0230* Well-being (evening) 0.0198* 0.0001*** 0.0461 * Exhaustedness (evening) 0.3863 0.5969 0.1745 Psychosomatic symptoms 0.1395 0.1778 0.7716 PSQI3 Sum score 0.0096** 0.0001*** 0.1265

Trimipramine compared to placebo significantly improved 'sleep quality', the 'feeling rested in the morning' and 'well-being in the evening' in the SFA and highly significantly decreased the PSQI sum score.

Results for lormetazepam compared to placebo went in the same direction, but were not as pronounced, with a significant increase of 'sleep quality', 'well-being in the evening' and significant decreases in PSQI sum scores.

The direct comparison of lormetazepam vs. trimipramine showed a statistically significant superiority of trimipramine for 'feeling of restedness in the morning' and 'well-being in the evening', but not for any of the other parameters.

Discontinuation period Dropouts/'adverse events/safety: All of the 46 patients who completed night 28 in the sleep laboratory stayed in the study until night 42.

Adverse events were also registered in the discontinuation period with placebo. Three patients in the placebo group, one patient in the lormetazepam group, and six patients in the trimipramine group reported adverse events during this period. With respect to safety measures, no relevant events occurred.

Fig. 2 shows mean values of the WSS (Withdrawal Symptom Scale) during the whole course of the study including the discontinuation period.

Whereas a value of 20 does not denote any kind of withdrawal symptoms, a high value (80) denotes a strong occurrence of withdrawal (especially vegetative) symptoms and negative

well-being. It can be seen that an increase of symptoms occurred in the trimipramine group including all subjects between days 35 and 41. On closer inspection of the data, it is clear that this increase was due to one subject, who experienced symptoms such as nausea with reduced appetite, dizziness and hyperacusis. After excluding this subject from the analysis (see Fig. 2), no increased values in the trimipramine group were visible. Fig. 2 demonstrates that 'withdrawal symptoms' are more pronounced prior to the start of the study compared to the discontinuation period. None of the effects observed were significant, and the mean values from the three groups show that they are always very near the lowest value of the scale, 20.

Polysomnography: Data for total sleep time are depicted in Fig. 3.

Fig. 2 Mean values of the Withdrawal Symptoms Scale (Merz and Ballmer, 1983). The WSS ranges from 20 (no symptoms present) to 80 (symptoms present always or most of the time).

2.

f


Fig. 3 Mean values of Total Sleep Time from night -1 to night 42 including only patients (n = 46) who completed the whole study. Tech- nical failure concerning polysomnography in night 42 led to reduced numbers in the placebo and lormetazepam groups.

Fig. 5 Mean values of PSQI (Pittsburgh Sleep Quality lndex)/sum score from night -1 to night 42 including only patients (n = 46) who completed the whole study. Reduced N in all groups is due to missing data. The PSQI ranges from 0 to 21 (0 = no sleep disturbance at all; 21 = severe sleep disturbance).

In the placebo and trimipramine groups, a significant decrease of total sleep time occurred when comparing night 42 to night 28 (p < 0.05 for placebo and p < 0.001 for trimipramine). No such difference occurred in the lormetazepam group (p = 0.7100). No significant changes indicating a rebound effect occurred for any of the drugs from comparing nights 30 or 42 to the baseline values.

Subjective sleep variables (SFA/PSQI): Data for the SFA and for the PSQI are depicted graphically in Figs. 4 and 5.

This shows that subjectively rated sleep quality, as measured by the SF-A, decreased during the discontinuation period in all of the three groups, but without being significantly lower than during the baseline phase (day 35 vs. -1: placebo, p = 0.4536; trimipramine, p = 0.6287; lormetazepam, p = 0.0848; day 42 vs. -1: placebo, p = 0.0334, trimipramine, p = 0.2609; lormetazepam, p = 0.7532).

The PSQI sum score increased again after patients had discontinued their active medication for 14 days in the lormetazepam and the trimipramine group. Values at day 42 were still lower than at the start of the study.

Fig. 4 Mean values of SF-A/Sleep quality from night -1 to night 42 including only patients (n = 46) who completed the whole study. Re- duced numbers in the placebo and trimipramine group is due to missing data. The scale ranges from 1 to 5 (1 = impaired sleep quality; 5 = excellent sleep quality).

For lormetazepam, the PSQI was still significantly decreased on day 42 compared to day -1 (p = 0.0087). The same applied to trimipramine (p = 0.0010), but not to placebo (p = 0.9166).

Discussion

In the present study, trimipramine in a mean dose of approximately 100 mg per day given over a period o four weeks resulted in an increase of almost 60 minutes in total sleep time as measured polysomnographically. With placebo, total sleep time re- mained almost unaltered, whereas with lormetazepam, total sleep time increased on average for 18 minutes. Though the increase of total sleep time was within the expected range from our pilot study [18], no statistical superiority was observed for trimipramine treatment compared to placebo. This might, on the one hand, be due to the high variance of values for total sleep time observed in the study (see Fig.l). On the other hand, a very conservative two-tailed testing method was applied. Further- more, the substitution technique applied for drop outs (LVCF = Last Value Carried Forward) was very rigorous compared, for example, to a per-protocol analysis.

Thus, it must be concluded that with respect to the primary efficacy parameter of our study, total sleep time, sample size was too small to confirm the statistical superiority of trimipramine compared to placebo.

Interestingly, some of the patients in the trimipramine group dis- played increases of total sleep time much more pronounced than 60 minutes (see Fig.l). This type of strong effect was not observed in any of the other two groups.

In contrast, sleep efficiency, the ratio of total sleep time to time in bed, demonstrated a superiority of trimipramine versus placebo at the p < 0.01 level. Whereas sleep efficiency increased by almost 10% with trimipramine, sleep efficiency only slightly increased for lormetazepam and actually decreased under placebo.

Although the statistical evaluation for the one parameter tested confirmatory, total sleep time, did not show the expected result, it seems reasonable to conclude that trimipramine is effective in


the treatment of primary insomnia. For subjective parameters of sleep, trimipramine demonstrated an impact on the patients' subjective perception of sleep with an overall increase of sleep quality, the feeling of restedness in the morning, well-being in the evening and the total sum score of the PSOJ. Positive effects were less pronounced for lormetazepam and non-existent for placebo.

The criticism is justified that the design, which allowed flexible dosing for trimipramine, might have biased the outcome of the study towards a superiority of trimipramine. It seemed, however, necessary to include flexible dosing for trimipramine; from clinical experience, there are large interindividual differences in the therapeutic dose of trimipramine needed by insomniac patients. The double-blind character of the study was not compromised, as patients in the other groups could also alter the number of pills, but without any impact on the actual dosage ingested.

Trimipramine had no impact on sleep latency or the frequency of nocturnal wake periods. In contrast, benzodiazepine hypnotics significantly reduced sleep latency and the number of wake periods as documented by meta-analysis of the literature [20,30].

The effects of trimipramine on polysomnography in the present study in primary insomniacs are in accord with our pilot study [18] and with other studies on the effects of trimipramine on polysomnography in depressed patients. Wzegand et al. de- scribed a significant increase of sleep efficiency and a 100-min- ute increase in total sleep time in ten depressed inpatients with 200 mg trimipramine [50]. Similar results were reported by Sonntag et al. [44], Ware et al. [47] and by Wo//et al. [54] in ger- iatric depressed patients.

In accordance with these studies, no effects on REM sleep and slow wave sleep occurred in our study in primary insomniacs.

Interestingly, the effects on subjectively experienced parameters on sleep were more pronounced than those documented by objective parameters. In one study on depressed outpatients with 100- 200 mg trimipramine [34], the occurrence of subjectively experienced sleep disturbances decreased from 90% of the patients at baseline to 10% of the patients after four weeks of treatment.

The effects of trimipramine as observed in the present study were similar to the effects observed with doxepin in primary insomnia [16]. In this study, total sleep time after four weeks of treatment with doxepin increased for 60 minutes, and sleep efficiency increased significantly by about 10%.

Other sedating antidepressants (such as trazodone and nefazodone) that are used for primary insomnia have not yet been tested for this indication using any placebo-controlled test design. There are, however, results for the effects of these drugs on the sleep EEC in healthy subjects and depressed patients [3,4, 13,23,48] that reveal a pattern of effects on the sleep EEC comparable to trimipramine with a primary effect on sleep continuity measures but no effect on REM sleep.

Interestingly, a positive effect on primary insomnia has recently been suggested even for paroxetine, which has only weak sedative properties [31].

When judging the effects of trimipramine and other antidepressants in primary insomnia treatment, one needs to compare their effectiveness with a standard treatment for primary insomnia, that is, classical benzodiazepine-hypnotics and benzodiazepine- receptor agonists such as zopiclone or zolpidem. For the benzodiazepines and for zolpidem, two meta-analyses have been published [20,30]. Both meta-analyses come to the surprising con- clusion that effects on subjective and objective sleep have been documented for periods of only 7 to 14 days. Effects of benzodiazepines on total sleep time are comparable to trimipramine, although there is a possible superiority of the benzodiazepines with respect to sleep onset latency. An earlier polysomnographic study with lormetazepam [21 ] for a shorter treatment period indicated a potential for tolerance and signs of rebound insomnia that was not confirmed in another study [45]. Our study did not demonstrate any strong effect of lormetazepam on polysomnography after four weeks, but also no signs of rebound insomnia with the drug. The crucial question is, however, the issue of adverse events and side effects of benzodiazepines like develop- ment of tolerance, abuse and addiction. With benzodiazepines, non-physiological effects on sleep EEC with a slight decrease in REM sleep and an increase in fast frequencies in the EEC are well-documented [8,9,11]. In the present study, trimipramine was superior concerning the effect on total sleep time and sleep efficiency in direct comparison between lormetazepam and trimipramine (exploratory result). Trimipramine did not influence slow-wave sleep or REM sleep, whereas lormetazepam decreased stage 3 sleep and increased REM sleep. The effects on subjective sleep parameters were more favorable with trimipramine compared to lormetazepam.

Another very important treatment strategy for primary insomnia is cognitive-behavioral treatment. Short- and long-term positive effects of psychological treatments have been convincingly demonstrated in three recent meta-analyses of the literature [27,28,43]. In one open pilot study on 20 patients with primary insomnia, 6 weeks of cognitive-behavioral treatment led to an increase in sleep time of over one hour (subjective estimate), which was not only measurable shortly after treatment, but per- sisted for 3 years [5]. Thus, it would be interesting to test a com- bination of sedating antidepressants and cognitive-behavioral therapy for primary insomnia.

An important question to discuss is the issue of adverse events. Adverse events were somewhat more pronounced with trimipramine than with lormetazepam and placebo. Events reported were typical for tricyclic antidepressants such as dry mouth, dizziness, and nausea. It is interesting to note that no clinically relevant changes concerning vital parameters or hematological/sero- logical parameters were observed in any of the groups. The same applied to the ECG. Thus, trimipramine seems to be a safe treatment when adequate precautions are taken.

Effects of discontinuation of the drugs were mild and comparable between trimipramine and lormetazepam. Regarding the Withdrawal Symptoms Scale (see Fig. 2), it must be noted that the highest amount of so called withdrawal symptoms occurred prior to the study during the placebo run-in phase.

Both for lormetazepam and trimipramine after 2 days of discontinuation total sleep time and subjectively measured sleep qual-


ity decreased. At day 42 subjective and objective parameters of sleep had returned to baseline values for trimipramine, with the exception of the PSQI, which still showed some improvement compared to the baseline phase. This indicates that there are no rebound effects with trimipramine on the one hand, but on the other, the effect of the drug does not persist beyond the actual treatment phase.

Summarizing, the present study demonstrates positive effects of trimipramine in primary insomnia concerning sleep efficiency as measured by polysomnography and several subjective parameters of sleep. Thus, the present study is the second study documenting evidence on the level of a randomized placebo controlled trial for sedating antidepressants to be effective in the treatment of primary insomnia. Sedating antidepressants such as trimipramine, doxepin, trazodone, and nefazodone seem to have a positive effect on sleep continuity and no effects on slow-wave or REM sleep. This is different from the effects of other antidepressants, which mostly suppress REM sleep (for overview see Sandor and Shapiro [38]). As primary insomnia is conceptionalized as a psychophysiological disorder [26,32], it seems logical that a positive effect on sleep generally improves the condition of patients afflicted with the disorder. Trimipramine also had positive effects on the subjective experience of feeling refreshed in the morning and well-being in the evening. The present study did not include measurements of hang- over effects in the morning or direct measurements of vigilance parameters or tests for cognitive functioning during the day. This should be included in future studies. Sedating antidepressants may also be an interesting alternative for patients with primary insomnia, as it has been suggested recently [12] that insomnia may be an independent risk factor in developing clinically relevant depression. In the meantime, several studies have been published that emphasize this point of view (for an overview, see [36,37]).

It has to be stressed, however, that sedating antidepressants are afflicted with a higher rate of adverse events and potential risks for some patients, such as ECG or hematological alterations, and less safety in the case of overdose. With respect to these risks, it needs to be emphasized that adequate precautions have to be taken when using sedating antidepressants for the treatment of primary insomnia.

Our study does not prove any effect of trimipramine beyond longer treatment periods of several months or even years. This needs to be investigated in prospective longitudinal designs, not only for trimipramine but also for classical benzodiazepine hypnotics as has been stressed recently by Kripke [22].

Acknowledgements

Study sites were excellently monitored by Dr. Cacilia Haarmann (formerly Rhone Poulenc Rorer GmbH, Cologne, Germany). Statistical data analysis was performed by ANFOMED, Erlangen, Germany. Dr. Preetam Schramm (Freiburg) scored all polysomnographic records of the study. The trimipramine insomnia study group consisted of the following sleep centers and individuals: Berlin: Department of Psychiatry, Free University of Berlin (Prof. Herrmann |)

Bielefeld: Neurological Hospital (Prof. Clarenbach) Bonn: Department of Psychiatry of the University Hospital (Dr. Sobanski, Prof. Maier) Erlangen: Department of Psychiatry of the University Hospital (Dipl.-Psych. Hoffmann, Prof. Barocka) Freiburg: Department of Psychiatry and Psychotherapy of the University Hospital (Prof. Berger, Prof. Hohagen, Prof. Riemann, Dr. Voderholzer) Gottingen: Department of Psychiatry of the University Hospital (Prof. Hajak, Prof. Ruther, Dr. Cohrs, Dr. Rodenbeck) Mannheim: Central Institute of Mental Health, Psychiatric De- partment (Dr. Weber, Prof. Heuser) Munchen: Department of Psychiatry of the University Hospital (Dr. Baghai, Prof. Laakmann) Munchen: Department of Psychiatry of the Technical University (PDDr. Wiegand) Regensburg: Department of Psychiatry of the University Hospital (Prof. Zulley, Dipl.-Psych. T. Cronlein) Schwalmstadt-Treysa: Neurological Hospital Hephata (PD Dr. Mayer) Wurzburg: Department of Psychiatry of the University Hospital (Dr. Vogel, Prof. Strik)

References

1 American Psychiatric Association (APA). Diagnostic and Statistical Manual of Mental Disorders, third edition revised (DSM-III-R). Wash- ington: 1987 Deutsche Bearbeitung: Wittchen, HU, Zaudig, M, Koeh- ler, K, Sal?. H. Weinheim, Beltz: 1989

2 American Psychiatric Association (APA). Diagnostic and Statistical Manual of Mental Disorders, fourth edition. Washington: APA, 1994 Deutsche Bearbeitung: Wittchen, HU, Zaudig, M, Koehler, K, Sal?. H. Weinheim: Beltz, 1994

3 Armitage R, Rush J, Trivedi M, Cain J, Roffwarg HP. The effects of nefazodone on sleep architecture in depression. Neuropsychopharmacolo- gy!994; 10:123-127

4 Armitage R, Yonkers K, Cole D, Rush J. A multicenter, double-blind comparison of the effects of nefazodone and fluoxetine on sleep architecture and quality of sleep in depressed outpatients. J Clin Psycho- pharmacol 1997; 17: 161-168

5 Backhaus J, Hohagen F, Voderholzer U, Riemann D. Long-term effectiveness of a short-term cognitive-behavioral group treatment for primary insomnia. Eur Arch Psychiatry Clin Neurosci 2001; 251: 35 - 41

6 Berger M, Riemann D. REM sleep in depression - an overview. J Sleep Res 1993; 2: 211-223

7 Berger M, Gastpar M. Trimipramine: a challenge to current concepts on antidepressives. Eur Arch Psychiatry Clin Neurosci 1996; 246: 235-239

8 Borbely AA, Loepfe M, Mattmann P, Tober I. Midazolam and triazolam: hypnotic action and residual effects after a single bedtime dose. Arz- neimittelforschung 1983; 33:1500-1502

9 Borbely AA, Mattmann P, Loepfe M, Strauch I, Lehmann D. Effect of benzodiazepine hypnotics on all-night sleep EEC spectra. Hum Neuro- bio!1985;4: 189-194

10 Buysse D, Reynolds CF, MonkTH, Berman SR, Kupfer DJ. The Pittsburgh Sleep Quality Index: A new instrument for psychiatric practice and re- search. Psychiatry Res 1989; 28: 193-213

11 Feige B, Voderholzer U, Riemann D, Hohagen F, Berger M. Independent sleep EEC slow wave and spindle band dynamics associated with 4 weeks of continuous application of short-half-life hypnotics in healthy subjects. Clin Neurophysiology 1999; 110: 1965-1974

12 Ford DE, Kamerow DB. Epidemiologic study of sleep disturbances and psychiatric disorders. An opportunity for prevention. J Am Med Assoc 1989;262: 1479-1484

13 Gillin JC, Rapaport M, Erman MK, Winokur A, Albala BJ. A comparison of nefazodone and fluoxetine on mood and on objective, subjective, and clinician-rated measures of sleep in depressed patients: A double-blind, 8-week clinical trial. J Clin Psychiatry 1997; 58:185-192


Trimipramine in Primary Insomnia: Results of a Polysomnographic Double-Blind Controlled Study

Documents

Transcript of Trimipramine in Primary Insomnia: Results of a Polysomnographic Double-Blind Controlled Study