INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

In view of their potential to stimulate respiration and upper airways motor outputs, serotonin-enhancing drugs have been tested as pharmacologic treatments for sleep apnea syndrome (SAS). One early report suggested that L-tryptophan, a serotonin precursor, may have a beneficial effect on SAS (1). More recently, fluoxetine (2) and paroxetine (3), selective serotonin reuptake inhibitors (SSRIs), were demonstrated to benefit some but not all patients with SAS. In addition, combinations of serotonin precursors and reuptake inhibitors reduced sleep-disordered respiration in the English bulldog model (4). However, despite ongoing investigations, these encouraging early results with serotonin enhancing drugs have not been reproduced.

At least two factors may contribute to the inconsistent impact of serotonin-enhancing drugs on sleep apnea. First, SSRIs are associated with reduced sleep efficiency, increased arousals from sleep, and reduced slow-wave sleep (5). Current theories of sleep apnea pathogenesis suggest a vicious cycle between apnea and arousal, with each behavior perpetuating the other (see Reference 6). In principle, the cycle may be broken by blocking either arousal or apnea. Indeed, patients with SAS sometimes express periods of deep slow-wave sleep without arousal. Apneas are rarely observed in this state (7). Conversely, evoking transient arousals destabilizes autonomic regulation and can increase apnea expression (8, 9). Thus, any benefits to patients with SAS deriving from respiratory stimulation may be offset by the sleep fragmentation produced by SSRIs.

Second, systemically administered serotonergic compounds have complex and poorly understood effects on respiratory and upper airway motor outputs. Although some studies suggest that activation of endogenous serotonin within the central nervous system stimulates phrenic and upper airway motoneurons (10, 11), others conclude that brainstem serotonergic systems are important inhibitors of chemoreflex ventilatory patterns (12). Intracarotid administration of serotonin typically produces a biphasic response, including both excitation and inhibition of the respiratory system (13). Injection of serotonin into the venous circulation, heart, pulmonary artery, or laryngeal artery produces immediate dose-dependent apnea, an effect mediated by 5-HT₂ and 5-HT₃ receptors in or on the nodose ganglia (14).

Taken together these findings suggest that global activation of serotonergic systems will have an unpredictable effect on respiratory stability and may even exacerbate apnea. In support of this view, we demonstrated that systemic administration of serotonin increased (15), whereas administration of a specific 5-HT₃ receptor antagonist reduced (16), apnea expression in the rat model of sleep-related central apnea. Thus, focused serotonergic interventions may be more effective in treating apnea than the global manipulations produced by serotonin precursors or SSRIs. In fact buspirone, a specific 5-HT_1A agonist that stimulates respiration (17), has recently been shown to reduce apnea index in four of five patients with SAS (18).

Mirtazapine (as Remeron, the +/ racemate of mirtazapine) is an antidepressant with a pharmacologic profile relevant to the above considerations. Mirtazapine enhances serotonin neurotransmission at 5-HT₁ receptors in the brain but acts as a specific antagonist at 5-HT₂ and 5-HT₃ receptors in the central and peripheral nervous systems (5, 19); effects that, based on the above evidence, may be expected to reduce apnea expression. Moreover, in contrast to the sleep disruption associated with SSRIs, mirtazapine produces increased sleep efficiency, augmented slow-wave sleep, and improved sleep consolidation (5, 20). The objective of the present study was to determine whether mirtazapine can alleviate sleep apnea in the normal rat model of sleep-disordered breathing.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Nine adult male Sprague-Dawley rats weighing 200 to 300 g were included in this study. Rats were anesthetized intraperitoneally (ketamine, 80 mg/kg, and xylazine, 10 mg/kg) and a surgical incision of the scalp was made to allow bilateral implantation of stainless steel screws into the frontal and parietal bones of the skull for electroencephalogram (EEG) recording. Bilateral wire electrodes were placed into the nuchal muscles for electromyogram (EMG) recording. The EEG and EMG leads were soldered to a miniature connector and fixed to the skull with cranioplastic cement. The skin was then sutured and the rats were allowed at least 7 d for surgical recovery. Throughout the surgical and experimental period, rats were maintained on a 12-h light and 12-h dark cycle in a fixed environment at 20° C with 40% humidity. Food and water were available ad libitum.

Respirations were recorded by placing each rat inside a single-chamber plethysmograph (PLYUN1R/U; Buxco Electronics, Sharon, CT) 6 inches wide by 10 inches long by 6 inches high. Thermal fluctuations associated with tidal respiration induce changes in pressure within the plethysmograph which, under appropriate conditions, are proportional to tidal volume (21, 22). Plethysmograph pressure was transduced using a Validyne DP45-14 differential pressure transducer (± 2 cm H₂O) (Validyne Engineering Corp., Northridge, CA). Plethysmograph pressure was referenced to a low-pass filtered (5-s time constant) version of itself in order to minimize the effects of any drift in temperature or ambient pressure during a recording. To minimize any possible artifact related to asymmetry or nonuniformity of pressure within the rectangular chamber, the transducer was mounted to and centered on the lid of the plethysmograph.

The plethysmograph chamber was flushed with room air at a constant regulated flow rate of 2 L/min. This flow was approximately one order of magnitude greater than the rat's expired minute ventilation and was thus sufficient to ensure that carbon dioxide rebreathing did not occur. The room and box temperature were measured and maintained at 20° ± 0.5° C throughout. Animal core temperature was not continuously monitored, but individual measurements at ambient temperature = 20° C revealed a mean of (37.4° C ± 1.0° C SD) for nine animals. Using these values, tidal volume was calibrated using the formula of Epstein and colleagues (21). Ambient temperature and pressure were measured immediately prior to each recording, and between-study calibrations were performed for repeated measurements in individual animals. Inspiratory minute ventilation ( I) was defined as the product of breath inspiratory tidal volume and breath respiratory rate (RR).

EEG and EMG activity was carried from the connector plug on the rat head by a cable and passed through a sealed port in the plethysmograph. EEG, EMG and respirations were continuously digitized (100/s), displayed on a computer monitor (Experimenter's Workbench; Datawave Systems, Longmont, CO), and stored on disk. All signals were low-pass filtered (50 Hz corner frequency, 6-pole Butterworth filter) to prevent aliasing.

All polygraphic recordings were 6 h in length and were made between 10:00 A.M. and 4:00 P.M. Each rat was recorded on four occasions, in random order, after intraperitoneal injection with one of the following: (1) saline (control), (2) 0.1 mg/kg mirtazapine, (3) 1.0 mg/ kg mirtazapine, or (4) 5.0 mg/kg mirtazapine. Recordings for an individual animal were separated by at least 3 d.

Polygraphic recordings of sleep and wakefulness were assessed by computer algorithm using the bifrontal EEG and nuchal EMG signals on 10-s epochs as we have previously described (15, 16). This software discriminates wakefulness (W) as a high frequency, low amplitude EEG with concomitant high EMG tone, NREM sleep by increased spindles and theta EEG together with decreased EMG, and REM sleep by a low ratio of delta to theta band EEG activity and an absence of EMG tone.

Throughout each 6-h recording, each breath was detected by an adaptive threshold algorithm (DataWave Systems), and the values of RR and VI for each breath were extracted. Normalized RR (NRR) and I (N I) were computed by dividing the appropriate value for each breath by the mean value recorded during wakefulness throughout the 6-h control recording for that animal. Sleep apneas, defined as cessation of respiratory effort for at least 2.5 s, were scored for each recording session and were associated with the stage in which they occurred: W, NREM, or REM sleep. The duration requirement of 2.5 s was arbitrarily chosen, but it reflects at least two "missed" breaths, as we have previously described (15, 16, 22). The events detected represent central apneas because decreased ventilation associated with obstructed or occluded airways would generate an increased plethysmographic signal rather than a pause. We characterized apneas as post-sigh (PS) or spontaneous (SP) according to the presence or absence of a preceding inspiration at least 150% larger than the average amplitude during regular breathing. This classification was performed because the mechanisms underlying SP and PS apneas may be distinct. Previous studies have demonstrated that these types of apnea may respond differently to certain experimental interventions. Apnea indexes (AI), defined as apneas per hour in stage, were separately determined for NREM and REM sleep.

The major effects of recording hour, sleep state, and mirtazapine dose were assessed using separate one-way analyses of variance (ANOVAs) with repeated measures. Interaction terms were evaluated using multiway ANOVAs. Multiple comparisons between means were controlled using Fisher's protected least-significant difference (PLSD) or by use of specific paired t tests, as indicated.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The effects of mirtazapine on the rate of apnea expression (PS + SP apneas) during NREM sleep are demonstrated in Figure 1. Two-way analysis of variance identified significant effects of time (p = 0.01) and mirtazapine (p < 0.0001) on NREM apnea index: AI increased significantly during the 6-h polygraphic recordings, and mirtazapine reduced AI by more than 50%. The effects of time and mirtazapine were independent (p = 0.92 for the interaction term by ANOVA). Post-hoc testing (Fisher's PLSD) revealed that the apnea suppressant effect of mirtazapine during NREM sleep was equivalent at all doses tested (p < 0.001 for each dose versus control, with no significant interdose differences). Similar effects of time and mirtazapine were observed when PS and SP apneas were examined separately (data not shown).

View larger version (17K): [in this window] [in a new window]   Figure 1.   Effect of mirtazapine on apnea index during NREM sleep. Each point represents mean ± SE for nine rats. Each dose of mirtazapine significantly reduced apnea index with respect to control (p < 0.0001 for each).

Apnea expression (PS + SP) during REM sleep is depicted in Figure 2. REM-related AI was reduced from 31.6 per hour to 12.5 per hour (60% reduction) by mirtazapine (p < 0.0001 for major effect of mirtazapine by ANOVA) with all doses having equivalent efficacy (no significant interdose differences). Again, the effects of mirtazapine were equivalent for PS and SP apneas (data not shown).

View larger version (19K): [in this window] [in a new window]   Figure 2.   Effect of mirtazapine on apnea index during REM sleep. Each dose of mirtazapine significantly reduced apnea index with respect to control (p < 0.0001 for each).

In concert with suppression of NREM and REM apneas, mirtazapine produced increased inspiratory minute ventilation (Figure 3). The significant increase in N I (p = 0.0003) was observed in all sleep/wake states (p < 0.001 for each) and was equivalent for all mirtazapine doses tested (no significant interdose differences by Fisher's PLSD). The significant progressive reduction in N I with transitions from wake to NREM to REM sleep in control recordings (p < 0.0001 for major effect of state) was preserved after mirtazapine administration (p = 0.35 for state * mirtazapine interaction term).

View larger version (41K): [in this window] [in a new window]   Figure 3.   Effect of mirtazapine on normalized inspiratory minute ventilation (see text for details). Each bar represents mean ± SE over all six recording hours with all animals pooled. Minute ventilation increased significantly after all doses of mirtazapine in each wake/sleep state (p < 0.001 for each).

By contrast to respiratory pattern, mirtazapine had relatively minor effects on sleep architecture. The %NREM sleep and %REM sleep are shown, respectively, in Figures 4 and 5. Mirtazapine had no effect on NREM sleep expression during the 6-h polygraphic recordings (p = 0.42) (Figure 4). Additional analysis revealed that all doses of mirtazapine were associated with increased EEG delta-band (1 to 4 Hz) power (mean increase of 31%; p = 0.04), suggesting increased deep NREM sleep and improved sleep consolidation. Mirtazapine reduced %REM sleep by an average of 35% (p = 0.03), with all doses tested having an equivalent effect (Figure 5). This REM suppressing effect persisted for at least 6 h (time *mirtazapine interaction term not significant).

View larger version (18K): [in this window] [in a new window]   Figure 4.   Effect of mirtazapine on the percentage of total recording time spent in NREM sleep. Mirtazapine had no significant effect at any dose.

View larger version (19K): [in this window] [in a new window]   Figure 5.   Effect of mirtazapine on the percentage of total recording time spent in REM sleep. Mirtazapine reduced REM sleep percentage by an average of 35% (p = 0.03) with all doses tested having an equivalent effect.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The findings from this study demonstrate that acute treatment with the mixed profile serotonergic antidepressant mirtazapine improves sleep-disordered breathing in the rat by reducing expression of apneic events during NREM and REM sleep. Apnea suppression was accompanied by increased minute ventilation and improved sleep consolidation.

The findings detailed above document that mirtazapine, over a 50-fold dose range, elicits a 50% to 60% reduction in central apnea expression during NREM and REM sleep in rats. The fact that all doses tested produced a similar effect on apnea expression suggests that greater apnea suppression cannot be elicited by this pharmacologic approach. These data also suggest that even lower doses of mirtazapine may be effective in suppressing apnea; a contrast to the selective 5-HT3 receptor antagonist GR38032F, for which the minimum effective dose to suppress apnea falls between 0.1 and 1.0 mg/kg (15, 16).

One may well question the relevance of this finding to the mechanisms and management of human sleep apnea syndromes. It is our view that both central and obstructive apnea reflect, at least in part, dysregulation of central neural motor output patterning to the respiratory system. In humans with upper airways predisposed to collapse by anatomic, mechanical, or muscular factors, this dysregulation may be manifest primarily by obstructive apneas. In humans or rats with mechanically stable upper airways, dysregulation of respiratory motor output patterning may be expressed primarily by central apneas or hypopneas.

Indirect support for our view comes from several lines of investigation. Most patients with sleep apnea syndrome exhibit a combination of central, mixed, and obstructive apneas in a single sleep period, leading to the suggestion that any factor that destabilizes respiratory drive during sleep promotes apnea genesis. Önal and Lopata (23) demonstrated that patients with sleep apnea exhibited obstructive apneas when breathing through their own upper airways, but central apneas when breathing through a tracheostomy. These investigators concluded that obstructive apnea reflects unstable central respiratory drive in patients with upper airways predisposed to collapse by anatomic or neuromuscular defects. Furthermore, in some cases, continuous positive airway pressure converts obstructive apneas to central apneas, again supporting the conclusion that unstable central respiratory motor patterning contributes to the pathogenesis of obstructive sleep apnea syndrome.

If apnea reflects unstable respiratory motor patterning, interventions stabilizing respiratory drive during sleep may reduce or eliminate apnea. Indeed, inspired carbon dioxide, used to elevate respiratory drive, reduced the expression of both central (24, 25) and obstructive (24, 26) apnea in man. Conversely, supplemental inspired oxygen that raises mean arterial oxygen saturation is often associated with longer or more frequent apneas in humans (26). The above human findings suggest that central and obstructive apnea during sleep share common central neural pathogenic mechanisms.

In testing the validity of the normal rat model of sleep-disordered breathing we have demonstrated that central apneas in rats are expressed in similar patterns and are influenced by interventions in a fashion similar to human central and obstructive apnea. In patients, both central and obstructive apnea are most severe during REM sleep (7). In the rat, central apnea is two to 10 times more frequent during REM than during non-REM sleep (15, 16, 22, 27). In both humans and rats, inspired hypercapnia decreases, whereas hyperoxia increases, the severity of apnea (22). Essential hypertension appears to increase the risk for sleep apnea syndrome (28), and effective treatment of hypertension can ameliorate sleep apnea (29). Genetic hypertension in rats is associated with a 2- to 5-fold increase in apnea, and pharmacologic treatment significantly reduces apnea expression (27).

The above evidence depicts similar patterns of expression and responses to intervention for central and obstructive apnea in humans and central apnea in rats. Thus, the significant mirtazapine-induced suppression of apnea in all sleep stages documented by the present investigation is expected to be of relevance to the mechanisms and management of human sleep-related apnea. Still, limitations of the model must be considered before extrapolating the present results to human sleep-disordered respiration.

First, the impact of 2.5-s respiratory pauses on gas exchange in the rat has not been directly demonstrated. However, mathematical modeling suggests that during a brief apnea, alveolar and arterial carbon dioxide partial pressure will rise exponentially toward the mixed venous partial pressure, with a time constant that scales according to the ratio of mlv/ Q_c, where mlv is the mean lung volume and Q_c is the cardiac output (30). Normal values of these parameters yield approximate time constants of 9.9 s for an adult man (30) and 1.7 s for a 500-g rat (31, 32). Assuming equivalent arteriovenous differences in carbon dioxide partial pressure in humans and rats, a 1.7-s apnea in rats should produce equivalent arterial hypercarbia to a 9.9-s apnea in humans. On this basis, we believe that by analogy to human expression of 2.5-s apneas does represent "sleep-disordered breathing" in the rat.

Second, neither clinical nor behavioral sequelae of elevated apnea expression have been documented in the rat. We have now recorded several hundred "normal" rats under baseline physiologic conditions. The range of apnea indexes among individual animals is very large. Large cross-sectional studies have not been attempted in rats to determine if a threshold apnea index can be identified above which behavioral or physiologic morbidity results. Similar studies in humans have proven difficult to standardize and replicate.

The rationale for using SSRIs such as fluoxetine or paroxetine to treat SAS rests on their ability to stimulate respiration and upper airway motor outputs. In accordance with this rationale, systemic administration of 5-HT₂ receptor antagonists to English bulldogs (an animal model of obstructive sleep apnea) reduced electrical activation of upper airway muscles, diminished upper airway cross-sectional area, and promoted obstructive apnea (33). Application of 5-HT to the floor of the fourth ventricle (34) produced upper airway motor activation in cats, an effect that appeared to be mediated predominantly by 5-HT₂ receptors. These observations provide a likely but unconfirmed explanation for the improvements in sleep-disordered breathing observed in some patients after SSRI treatment (2, 3). Even in patients with a positive response to SSRI treatment, however, the benefit was restricted to NREM sleep, with no reduction in REM-related apneas (2, 3).

The inconsistent and state-specific impact of SSRIs on SAS may relate to the fact that these agents nonspecifically augment serotonergic function throughout the body. At least 14 distinct serotonin receptor subtypes have been characterized. In view of the functional diversity of these receptors, even within the respiratory system (10, 17), it is to be expected that the net effects of global serotonin-enhancing drugs may be inconsistent or difficult to predict in individual patients.

Consideration of the novel pharmacologic profile of mirtazapine reveals a plausible explanation for the significant mirtazapine-induced suppression of NREM and REM apnea documented by the present study. Mirtazapine blocks presynaptic ₂-adrenoreceptors as well as postsynaptic 5-HT₂ and 5-HT₃ receptors (5, 19). Antagonism of presynaptic ₂-receptors located on serotonergic neurons (heteroreceptors) enhances serotonin release. Because the affinity of mirtazapine for central ₂-receptors is 10 times higher than for peripheral ₂- receptors, central serotonin release is increased, with minimal adrenergic side effects such as hypertension. Because mirtazapine is a high affinity antagonist at 5-HT_2A, 5-HT_2C, and 5-HT₃ receptors, the net effect is increased postsynaptic 5-HT₁ activity within the brain and reduced 5-HT₂ and 5-HT₃ postsynaptic activity in the central and peripheral nervous systems. Each of these pharmacologic effects may serve to stimulate respiration and suppress apnea.

The net 5-HT₁ agonist activity of mirtazapine in the central nervous system may have contributed to NREM apnea suppression. Buspirone, a 5-HT_1A agonist associated with respiratory stimulation (17), decreased overall apnea index by 35% in a group of five male patients with SAS (18). This effect of buspirone was almost identical to the NREM-specific reductions in apnea index attributed to fluoxetine (2) and paroxetine (3). The combination of 5-HT₂ and 5-HT₃ antagonist effects also may have promoted NREM apnea suppression in the present study. It is not possible to distinguish the importance of 5-HT₁ agonist from 5-HT₂ plus 5-HT₃ antagonist effects on NREM apnea suppression based on the present data.

Another mechanism that may have contributed to NREM apnea suppression relates to the impact of mirtazapine on sleep architecture. Mirtazapine increased EEG delta power during NREM sleep, suggesting improved sleep consolidation and deeper NREM sleep. This shift in sleep architecture may have contributed to the NREM apnea suppression observed in the present study (8, 9). Although full antidepressant responses require several weeks of treatment to develop, similar improvements in sleep architecture have been observed in human trials of mirtazapine, even after single dose administrations (20).

Suppression of REM-related apnea in the present study most likely resulted from the 5-HT₃ antagonist activity of mirtazapine. We have previously demonstrated that GR38032F, a specific 5-HT₃ antagonist, yields powerful suppression of REM-related apneas in rats, with little or no impact on NREM apneas (16). In further support of this interpretation, peripherally administered serotonin exacerbated REM-related apnea without influence on NREM apnea expression, an effect that was completely blocked by pretreatment with a low dose of GR38032F (15). The fact that serotonin does not cross the blood brain barrier argues that activity at 5-HT₃ receptors in the peripheral nervous system strongly promotes REM-related apnea. Yoshioka and colleagues (14) demonstrated that intravenously by administered serotonin-induced central apnea in anesthetized rats was blocked by 5-HT₂ or 5-HT₃ antagonists acting primarily at the nodose ganglia. On this basis, the combined 5-HT₂ and 5-HT₃ antagonist activity of mirtazapine in the peripheral nervous system may have contributed to the REM apnea suppression demonstrated here. Still, the specific sites and mechanisms by which mirtazapine reduced apnea expression cannot be determined from the present data.

The impact of mirtazapine on upper airway motor activity cannot easily be inferred. While application of serotonin to the brainstem augmented upper airway motor outputs (34), it is uncertain whether this is the basis for therapeutic responses produced by SSRIs in some patients with sleep apnea syndrome. For example, when administered to healthy adults, fluoxetine did not increase genioglossus muscle activity or decrease pharyngeal resistance during NREM sleep (35).

In summary, the present study demonstrates that mirtazapine, over a 50-fold dose range, significantly reduces central apnea expression during NREM and REM sleep in the rat. This apnea suppression is associated with respiratory stimulation and improved NREM sleep consolidation. The efficacy of this compound in all sleep stages most probably arises from its mixed agonist/antagonist profile at serotonin receptors. The implications of these findings for the management of sleep apnea syndrome must be verified by appropriate clinical trials.