Reduced Clonidine Rapid Eye Movement Sleep Suppression in Patients With Primary Major Affective Illness Michel Schittecatte, MD; G\l=e'\rardCharles, MD; Robert Machowski; Jos\l=e'\Garcia-Valentin, MD;
Julien Mendlewicz, MD; Jean Wilmotte,
hydrochloride, administered intravenously (2 \g=m\g/kg)during the second non\p=m-\rapideye movement period, was significantly less suppressant of rapid eye movement sleep in 10 depressed patients with primary major affective illness, according to Research Diagnostic Criteria, than in three groups of matched subjects (10 normal controls, 10 patients with minor depression, and 10 patients with generalized anxiety). These results suggest that depressed patients with major primary affective illness have downregulated \g=a\2-adrenergic receptors. These findings are consistent with the cholinergic-aminergic balance hypothesis of depression and support the aminergic side of the concept. Finally, the rapid eye movement sleep response to clonidine could provide a new biological marker of affective illness. (Arch Gen Psychiatry. 1992;49:637-642) \s=b\ Clonidine
original catecholamine hypothesis postulated that The depression by deficiency central functional refinement of is characterized
a
of
norepinephrine.1,2 More recent this hypothesis has proposed that a dysregulated noradrenergic (NA) system contributes to the symptoms of af¬ fective disorders.3 One experimental strategy that has been used is the acute challenge with clonidine, a specific and potent agonist of a2-adrenergic receptors and growth hor¬ mone (GH) response. Depressed patients manifest a blunted GH response to clonidine in comparison with control subjects.4-5 The physiological consequences of a2adrenergic receptor occupancy are described for other tar¬ get organs as well. The a2-adrenergic receptors are be¬ lieved to play an important role in the control of wakefulness and sleep, especially rapid eye movement (REM) sleep. Clonidine reduces REM sleep in the rat and in the cat.6,7 A similar decrease has been observed in man with a dose roughly five times smaller than that used in the rat.8-9 No substance other than clonidine, with the possible exception of some peptides, is known to reduce REM sleep at a dose as low as 0.002 mg/kg.
Accepted for publication September 19, 1991. From the Department of Psychiatry, Vincent Van Gogh Hospital, Marchiene-au-Pont, Belgium (Drs Schittecatte, Charles, GarciaValentin, and Wilmotte, and
Mr
Machowski; and the Department of
Hospital, Brussels, Belgium (Dr Mendlewicz). Reprint requests to Department of Psychiatry, H\l=o^\pitalVincent Van Gogh 55, rue de l'h\l=o^\pital6030, Marchienne-Au-Pont, Belgium (Dr
Psychiatry,
Erasmus
Schittecatte).
Downloaded From: by a University of Tennessee User on 12/21/2018
MD
The medial pontine reticular formation (mPRF) is in¬ volved in REM sleep generation because in vivo microin¬ jections of cholinergic compounds into the mPRF can evoke the phenomenology associated with REM sleep.10 However, it has recently been shown that norepinephrine mediates, in vitro and in vivo, the responses of the mPRF. Norepinephrine and clonidine, in rat pontine slice prepa¬ ration, elicit hyperpolarization responses of the mPRF cells. This response persists in the presence of tetrodotoxin, suggesting a direct, nonsynaptic action. Norepinephrine infused into the mPRF of unanesthetized, freely moving cats reduced REM sleep with decreased mPRF cell firing. These results suggest that a2-adrenergic receptors mediate the adrenergic hyperpolarization response observed in the mPRF and the REM sleep suppression observed in vivo." The importance of a2-adrenergic receptors for REM sleep cholinergic stimulation is also supported by the fact that clonidine pretreatment of rats abolished the REM sleep fa¬ cilitation elicited by the injection of carbachol, a cholinergic compound, into the brain stem.12 Gaillard13 had previously proposed that norepinephrine participates in REM sleep generation as a link in a chain of events that ultimately leads to REM sleep production. Rapid eye movement sleep suppression by clonidine suggested a positive relationship between activity of A cells and REM sleep production. However, studies of unit activity of NA cells in the locus coeruleus show a negative relationship with REM sleep.14 This discrepancy is compatible, according to Gaillard, with the idea of a permissive role of NA neurons in REM sleep
production.
Whatever mechanism may be responsible, if a2adrenergic receptors are down regulated in depression, as suggested by several studies,1517 one could expect to see a reduced suppression of REM sleep by clonidine in de¬ pressed patients when compared with control subjects. Therefore, we decided to compare the REM sleep response to clonidine hydrochloride in patients with primary major
affective illness and different groups of controls.
SUBJECTS AND METHODS we used a strategy inspired from the cholinergic REM induction test (CRIT) described by Sitaram et al18,19 and named it by analogy "the clonidine REM suppression test" (CREST). As the interval between the first two REM periods To test
our
hypothesis,
is not different between normal and depressed subjects,20 we in¬ fused intravenous clonidine hydrochloride (2 µg/kg dissolved in
9 mL of saline) or placebo (10 mL of saline) over a 10-minute pe¬ riod timed to end 30 minutes after the completion (last minute) of the first REM period. The time to begin the infusion was cho¬ sen to be sure that the first REM period was terminated, since two REM periods are by definition separated by at least 20 minutes. Patients and controls participated in a 5-consecutive-night protocol. After two habituation nights without perfusion, each subject received one infusion per night and was tested on 3 con¬ secutive nights (baseline night on the third night, clonidine on the fourth night, and placebo on the fifth night). This order was de¬ termined to allow the evaluation of an eventual "rebound" effect of clonidine on sleep. Sleep studies were performed during sub¬ ject's regular sleeping hours, with time of going to bed (11 pm) and getting up (7 am) held constant. The intravenous catheter was in¬ serted into the patient's forearm at 5 PM. The study was performed in the Sleep Laboratory Unit of Vincent Van Gogh Hospital, Charleroi, Belgium, a four-bed unit located in a psychiatric clinic of 100 beds. Subjects were monitored in comfortable private bedrooms. Sleep was recorded in the laboratory on 16-channel polygraphs. The sleep stage was recorded by electroencephalogram, electrooculogram, and submental chin electromyogram. The electro¬ encephalogram consisted of three F4-C4-04 scalp placements ref¬ erenced to tied mastoid. The electro-oculogram consisted of electrode placement at the outer canthi of the eyes, with each eye derivation referenced to tied mastoid. All electrode impedances were determined to be less than 5000 . Filter settings for the electroencephalogram and electro-oculogram were 0.3 to 30 Hz; the submental chin electromyogram was bipolar, with filter set¬ tings of 10 to 90 Hz. Paper speed for all recording was 10 mm/s and a 50-µ signal was calibrated to produce a 10-mm deflection at a sensitivity of 5. All sleep was scored in 30-second epochs fol¬ lowing the criteria of Rechtschaffen and Kales21 by one of us (R.M.), "blind" to the diagnosis. Sleep onset, REM latency, REM latency minus awakening, REM periods, sleep architecture (nonREM stages 1 through 4 and stage REM), and REM activity were defined as described by Knowles et al22 and Kupfer and Heninger.23-24 We calculated an REM1-REM2 interval, which is the time elapsed between the end of the first REM period and the on¬ set of the second REM period or, if no REM sleep further occurs, the end of the recording. We also calculated a REM1-REM2 inter¬ val minus awakening (REM1-REM2.MA), which is the REM1REM2 interval minus any intervening waking time. Finally, blood pressure was automatically determined by a noninvasive elec¬ tronic device (TM 300 MIRA) before sleep onset and 30 minutes after the end of clonidine infusion. In a prospective study, we carefully selected 20 depressed pa¬ tients and 10 anxious patients. They were assessed using the Schedule for Affective Disorders and Schizophrenia.25 Psychiatric diagnoses were determined based on the Research Diagnostic Criteria.26 Severity of depression was evaluated using the Hamil¬ ton Rating Scale for Depression (24-item National Institute of Mental Health version).27 Patients of group 1 (seven women and three men) met Research Diagnostic Criteria for a primary major depressive disorder episode (MDD). The diagnostic distribution of the patients included eight recurrent unipolar (three endoge¬ nous) and two bipolar I depressives. Matched patients of groups 2 and 3 met Research Diagnostic Criteria for a minor depressive episode (mDD) and a generalized anxiety disorder (GAD), respectively. Patients with comorbid conditions or substance abuse were excluded. Patients of group 1 had significantly higher Hamilton scores than patients of the two control groups (group 1 vs group 2 vs group 3, 23.8±8 vs 17.8±6 vs 14+4.7; F=9.07, d/=2,27, P=.001). The mean length of the current episode (15.8± 7.7 vs 10.9±7.9 vs 15±9.8 months; df=2,27, not significant) and the number of prior episodes (1.6±1 vs 1±1 vs 0.6±0.5; d/=2,27, not significant) were not different between patients with MDD, mDD, and GAD. Ten matched controls (group 4) also volunteered to participate in this study. Analysis of variance (ANOVA) did not reveal any significant age difference among the three groups of patients and controls (group 1 vs 2 vs 3 vs 4, 40.5±8.6 vs 37.29±6.38 vs 31.75±6.34 vs
Downloaded From: by a University of Tennessee User on 12/21/2018
33.43+10.49 years; F=1.36, dj=2, not significant). Two patients of group 1 refused to spend the fifth night in the sleep laboratory. All patients with MDD, six with mDD, and five with GAD had received antidepressants in the past but were drug free of benzodiazepines for at least 5 days and of alcohol and other drugs known to affect sleep structure for 2 weeks before the study. The other patients and normal controls had never received antidepressant therapy. Patients who had received neuroleptics, lith¬ ium, monoamine oxidase inhibitors, or electroconvulsive therapy were excluded. None had any medical disorder or was over¬ weight by more than 10%. All subjects underwent physical examinations and laboratory tests (complete blood cell count, chemistry screen, thyroid function). Patients and controls partic¬ ipated in this study after granting informed consent. This inves¬ tigation was approved by the hospital ethics committee.
DATA ANALYSIS Before sample comparisons, we evaluated the distributions of our data with the Kolmogorov-Smirnov goodness-of-fit test for normality. No distribution significantly deviated from normal. Sleep data were first analyzed using a two-factor ANOVA with "group" and "night" as factors. This allowed us to determine ef¬ fects due to clonidine injection (night effects), group effects, and night-by-group interactions. Then we compared the withingroup variation of sleep measures using a one-way ANOVA and a multiple-range Scheffé procedure to determine which pairs of nights significantly differed. Finally, we compared the betweengroup variation of sleep measures using a one-way ANOVA. All statistical procedures were performed with SPSS /PC+ version 2.O.28
RESULTS Patients with MDD had a significantly shorter REM latency minus awakening than the three other groups (P
Julien Mendlewicz, MD; Jean Wilmotte,
hydrochloride, administered intravenously (2 \g=m\g/kg)during the second non\p=m-\rapideye movement period, was significantly less suppressant of rapid eye movement sleep in 10 depressed patients with primary major affective illness, according to Research Diagnostic Criteria, than in three groups of matched subjects (10 normal controls, 10 patients with minor depression, and 10 patients with generalized anxiety). These results suggest that depressed patients with major primary affective illness have downregulated \g=a\2-adrenergic receptors. These findings are consistent with the cholinergic-aminergic balance hypothesis of depression and support the aminergic side of the concept. Finally, the rapid eye movement sleep response to clonidine could provide a new biological marker of affective illness. (Arch Gen Psychiatry. 1992;49:637-642) \s=b\ Clonidine
original catecholamine hypothesis postulated that The depression by deficiency central functional refinement of is characterized
a
of
norepinephrine.1,2 More recent this hypothesis has proposed that a dysregulated noradrenergic (NA) system contributes to the symptoms of af¬ fective disorders.3 One experimental strategy that has been used is the acute challenge with clonidine, a specific and potent agonist of a2-adrenergic receptors and growth hor¬ mone (GH) response. Depressed patients manifest a blunted GH response to clonidine in comparison with control subjects.4-5 The physiological consequences of a2adrenergic receptor occupancy are described for other tar¬ get organs as well. The a2-adrenergic receptors are be¬ lieved to play an important role in the control of wakefulness and sleep, especially rapid eye movement (REM) sleep. Clonidine reduces REM sleep in the rat and in the cat.6,7 A similar decrease has been observed in man with a dose roughly five times smaller than that used in the rat.8-9 No substance other than clonidine, with the possible exception of some peptides, is known to reduce REM sleep at a dose as low as 0.002 mg/kg.
Accepted for publication September 19, 1991. From the Department of Psychiatry, Vincent Van Gogh Hospital, Marchiene-au-Pont, Belgium (Drs Schittecatte, Charles, GarciaValentin, and Wilmotte, and
Mr
Machowski; and the Department of
Hospital, Brussels, Belgium (Dr Mendlewicz). Reprint requests to Department of Psychiatry, H\l=o^\pitalVincent Van Gogh 55, rue de l'h\l=o^\pital6030, Marchienne-Au-Pont, Belgium (Dr
Psychiatry,
Erasmus
Schittecatte).
Downloaded From: by a University of Tennessee User on 12/21/2018
MD
The medial pontine reticular formation (mPRF) is in¬ volved in REM sleep generation because in vivo microin¬ jections of cholinergic compounds into the mPRF can evoke the phenomenology associated with REM sleep.10 However, it has recently been shown that norepinephrine mediates, in vitro and in vivo, the responses of the mPRF. Norepinephrine and clonidine, in rat pontine slice prepa¬ ration, elicit hyperpolarization responses of the mPRF cells. This response persists in the presence of tetrodotoxin, suggesting a direct, nonsynaptic action. Norepinephrine infused into the mPRF of unanesthetized, freely moving cats reduced REM sleep with decreased mPRF cell firing. These results suggest that a2-adrenergic receptors mediate the adrenergic hyperpolarization response observed in the mPRF and the REM sleep suppression observed in vivo." The importance of a2-adrenergic receptors for REM sleep cholinergic stimulation is also supported by the fact that clonidine pretreatment of rats abolished the REM sleep fa¬ cilitation elicited by the injection of carbachol, a cholinergic compound, into the brain stem.12 Gaillard13 had previously proposed that norepinephrine participates in REM sleep generation as a link in a chain of events that ultimately leads to REM sleep production. Rapid eye movement sleep suppression by clonidine suggested a positive relationship between activity of A cells and REM sleep production. However, studies of unit activity of NA cells in the locus coeruleus show a negative relationship with REM sleep.14 This discrepancy is compatible, according to Gaillard, with the idea of a permissive role of NA neurons in REM sleep
production.
Whatever mechanism may be responsible, if a2adrenergic receptors are down regulated in depression, as suggested by several studies,1517 one could expect to see a reduced suppression of REM sleep by clonidine in de¬ pressed patients when compared with control subjects. Therefore, we decided to compare the REM sleep response to clonidine hydrochloride in patients with primary major
affective illness and different groups of controls.
SUBJECTS AND METHODS we used a strategy inspired from the cholinergic REM induction test (CRIT) described by Sitaram et al18,19 and named it by analogy "the clonidine REM suppression test" (CREST). As the interval between the first two REM periods To test
our
hypothesis,
is not different between normal and depressed subjects,20 we in¬ fused intravenous clonidine hydrochloride (2 µg/kg dissolved in
9 mL of saline) or placebo (10 mL of saline) over a 10-minute pe¬ riod timed to end 30 minutes after the completion (last minute) of the first REM period. The time to begin the infusion was cho¬ sen to be sure that the first REM period was terminated, since two REM periods are by definition separated by at least 20 minutes. Patients and controls participated in a 5-consecutive-night protocol. After two habituation nights without perfusion, each subject received one infusion per night and was tested on 3 con¬ secutive nights (baseline night on the third night, clonidine on the fourth night, and placebo on the fifth night). This order was de¬ termined to allow the evaluation of an eventual "rebound" effect of clonidine on sleep. Sleep studies were performed during sub¬ ject's regular sleeping hours, with time of going to bed (11 pm) and getting up (7 am) held constant. The intravenous catheter was in¬ serted into the patient's forearm at 5 PM. The study was performed in the Sleep Laboratory Unit of Vincent Van Gogh Hospital, Charleroi, Belgium, a four-bed unit located in a psychiatric clinic of 100 beds. Subjects were monitored in comfortable private bedrooms. Sleep was recorded in the laboratory on 16-channel polygraphs. The sleep stage was recorded by electroencephalogram, electrooculogram, and submental chin electromyogram. The electro¬ encephalogram consisted of three F4-C4-04 scalp placements ref¬ erenced to tied mastoid. The electro-oculogram consisted of electrode placement at the outer canthi of the eyes, with each eye derivation referenced to tied mastoid. All electrode impedances were determined to be less than 5000 . Filter settings for the electroencephalogram and electro-oculogram were 0.3 to 30 Hz; the submental chin electromyogram was bipolar, with filter set¬ tings of 10 to 90 Hz. Paper speed for all recording was 10 mm/s and a 50-µ signal was calibrated to produce a 10-mm deflection at a sensitivity of 5. All sleep was scored in 30-second epochs fol¬ lowing the criteria of Rechtschaffen and Kales21 by one of us (R.M.), "blind" to the diagnosis. Sleep onset, REM latency, REM latency minus awakening, REM periods, sleep architecture (nonREM stages 1 through 4 and stage REM), and REM activity were defined as described by Knowles et al22 and Kupfer and Heninger.23-24 We calculated an REM1-REM2 interval, which is the time elapsed between the end of the first REM period and the on¬ set of the second REM period or, if no REM sleep further occurs, the end of the recording. We also calculated a REM1-REM2 inter¬ val minus awakening (REM1-REM2.MA), which is the REM1REM2 interval minus any intervening waking time. Finally, blood pressure was automatically determined by a noninvasive elec¬ tronic device (TM 300 MIRA) before sleep onset and 30 minutes after the end of clonidine infusion. In a prospective study, we carefully selected 20 depressed pa¬ tients and 10 anxious patients. They were assessed using the Schedule for Affective Disorders and Schizophrenia.25 Psychiatric diagnoses were determined based on the Research Diagnostic Criteria.26 Severity of depression was evaluated using the Hamil¬ ton Rating Scale for Depression (24-item National Institute of Mental Health version).27 Patients of group 1 (seven women and three men) met Research Diagnostic Criteria for a primary major depressive disorder episode (MDD). The diagnostic distribution of the patients included eight recurrent unipolar (three endoge¬ nous) and two bipolar I depressives. Matched patients of groups 2 and 3 met Research Diagnostic Criteria for a minor depressive episode (mDD) and a generalized anxiety disorder (GAD), respectively. Patients with comorbid conditions or substance abuse were excluded. Patients of group 1 had significantly higher Hamilton scores than patients of the two control groups (group 1 vs group 2 vs group 3, 23.8±8 vs 17.8±6 vs 14+4.7; F=9.07, d/=2,27, P=.001). The mean length of the current episode (15.8± 7.7 vs 10.9±7.9 vs 15±9.8 months; df=2,27, not significant) and the number of prior episodes (1.6±1 vs 1±1 vs 0.6±0.5; d/=2,27, not significant) were not different between patients with MDD, mDD, and GAD. Ten matched controls (group 4) also volunteered to participate in this study. Analysis of variance (ANOVA) did not reveal any significant age difference among the three groups of patients and controls (group 1 vs 2 vs 3 vs 4, 40.5±8.6 vs 37.29±6.38 vs 31.75±6.34 vs
Downloaded From: by a University of Tennessee User on 12/21/2018
33.43+10.49 years; F=1.36, dj=2, not significant). Two patients of group 1 refused to spend the fifth night in the sleep laboratory. All patients with MDD, six with mDD, and five with GAD had received antidepressants in the past but were drug free of benzodiazepines for at least 5 days and of alcohol and other drugs known to affect sleep structure for 2 weeks before the study. The other patients and normal controls had never received antidepressant therapy. Patients who had received neuroleptics, lith¬ ium, monoamine oxidase inhibitors, or electroconvulsive therapy were excluded. None had any medical disorder or was over¬ weight by more than 10%. All subjects underwent physical examinations and laboratory tests (complete blood cell count, chemistry screen, thyroid function). Patients and controls partic¬ ipated in this study after granting informed consent. This inves¬ tigation was approved by the hospital ethics committee.
DATA ANALYSIS Before sample comparisons, we evaluated the distributions of our data with the Kolmogorov-Smirnov goodness-of-fit test for normality. No distribution significantly deviated from normal. Sleep data were first analyzed using a two-factor ANOVA with "group" and "night" as factors. This allowed us to determine ef¬ fects due to clonidine injection (night effects), group effects, and night-by-group interactions. Then we compared the withingroup variation of sleep measures using a one-way ANOVA and a multiple-range Scheffé procedure to determine which pairs of nights significantly differed. Finally, we compared the betweengroup variation of sleep measures using a one-way ANOVA. All statistical procedures were performed with SPSS /PC+ version 2.O.28
RESULTS Patients with MDD had a significantly shorter REM latency minus awakening than the three other groups (P
