INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

In view of conflicting reports on the depressant, null, or stimulant respiratory effects exerted by benzodiazepines, it remains controversial as to whether benzodiazepines could be or should be prescribed to sleep apnea patients. The matter is further complicated by the fact that distinction should be made between central and obstructive sleep apnea as well as between short- and long-acting benzodiazepines.

Although the conventional approach is that benzodiazepines, particularly long-acting ones, should not be given to patients with obstructive sleep apnea (1), reports from the literature appear to support the use of short-acting benzodiazepines in the management of patients with central sleep apnea (5). This is largely due to findings that short-acting benzodiazepines either produce no respiratory depression (3) or respiratory stimulation (8, 9). Despite the fact that diazepam, one of the most widely prescribed sedative-hypnotics, is a long-acting benzodiazepine, Rao and colleagues (10) felt that respiratory depression associated with its use was comparable to the changes occurring during physiologic sleep.

In this study, we tested the effects of subhypnotic and hypnotic doses of diazepam on respiration, spontaneous (Sp) and post-sigh (PS) apneas, and sleep in an established rat model of central sleep apnea (11). Our results indicate that administration of diazepam to rats stimulated respiration during sleep and consequently led to a decrease of apnea expression.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Nine adult male Sprague-Dawley rats (300 g) were maintained on a 12-h light (8:00 A.M.-8:00 P.M.)/12-h dark (8:00 P.M.-8:00 A.M.) cycle for 1 wk, housed in individual cages and given ad libitum access to food and water. Following 1 wk of adaptation, animals were subjected to a surgical procedure that will be briefly described here.

Rats were anesthetized for the implantation of cortical electrodes for electroencephalogram (EEG) recording and neck muscle electrodes for electromyogram (EMG) recording using a mixture of ketamine (Vetalar 100 mg/ml) and acetylpromazine (10 mg/ml) (4:1, vol/ vol) at a volume of 1 ml/kg body weight. The surface of the skull was exposed and cleaned with a 20% solution of hydrogen peroxide followed by a solution of 95% isopropyl alcohol. Next, a dental preparation of sodium fluoride (Flura-GEL; Saslow Dental, Mt. Prospect, IL) was applied to harden the skull and allowed to remain for 5 min. The fluoride mixture was then removed from the skull above the parietal cortex. A thin layer of Justi resin cement (Saslow Dental) was applied to cover the screw heads and surrounding skull to further promote the adhesion of the implant. Electromyogram electrodes consisted of two ball-shaped wires, which were inserted into the bilateral neck musculature. All leads were soldered to a miniature connector (39F1401, Newark Electronics, Chicago, IL). Last, the entire assembly was fixed to the skull with dental cement.

After the surgery, animals were allowed a 1-wk recovery period before being used in the study. Each rat was recorded in random order, on three occasions: control (saline) and two doses of diazepam (0.05 and 5 mg/kg). Diazepam was administered by intraperitoneal injection at 9:45 A.M. Polygraphic recordings were made from 10:00 A.M.-4:00 P.M. and were separated by at least 3 d.

Respiration was recorded by placing each rat, unrestrained, inside a single chamber plethysmograph (PLYUN1R/U; Buxco Electronics, Sharon, CT; dimension 6 in. W × 10 in. L × 6 in. H) ventilated with a bias flow of fresh room air at a rate of 2 L/min, as we have previously described (14, 18). This bias flow is 2- to 3-fold higher than has been employed by some investigators (c.f. 13), but was chosen after preliminary investigations in 15 animals (not included in the present study) demonstrated that bias flows of 500 ml/min were associated with increased minute ventilation in all sleep/wake states when compared with a bias flow of 2 L/min. This effect may relate to our use of rectangular rather than cylindrical plethysmographs (13). Plethysmograph temperature during each recording was monitored and maintained at 20 ± 1° C. Plethysmograph humidity was controlled by using a bias flow of dry room air from a compressed gas source.

A cable plugged onto the animal's connector and passed through a sealed port was used to carry the bioelectrical activity from the head. Respiration, EEG, and EMG were displayed on a video monitor and simultaneously digitized 100 times per second and stored on computer disk (Experimenter's Workbench; Datawave Technologies, Longmont, CO).

Sleep and waking states were assessed using the biparietal EEG and nuchal EMG signals on 10-s epochs as described by Bennington and colleagues (19). This software discriminated wakefulness (W) as a high-frequency, low-amplitude EEG with a concomitant high EMG tone, non-rapid eye movement (non-REM) sleep by increased spindle and theta activity together with decreased EMG tone, and rapid eye movement (REM) sleep by a low ratio of a delta-to-theta activity and an absence of EMG tone.

As in previous investigations (14), 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: non-REM or REM sleep. The duration requirement of 2.5 s represents at least 2 "missed" breaths, which is therefore analogous to a 10-s apnea duration requirement in humans, which also reflects 2-3 missed breaths. 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. As in previous reports, apneas were observed to occur either as pauses between breaths (Sp) or as periods of respiratory cessation preceded by a sigh (PS). We therefore characterized them as PS apneas and Sp apneas according to the presence or absence of a preceding inspiration at least 150% larger than the average tidal amplitude during regular breathing. An apnea index (AI), defined as apneas per hour in a stage, were separately determined for non-REM and REM sleep. The effects of sleep stage (non-REM versus REM) and dose (saline, 0.05 mg/kg diazepam or 5 mg/kg diazepam) were made using two-way analysis of variance (ANOVA) with repeated measures. Additional one-way ANOVAs were conducted for each main effect: sleep stage and dose. Multiple comparisons were controlled using Fisher's protected least significant difference (PLSD). In addition, the timing and volume of each breath were scored by automatic analysis (Experimenter's Workbench; Datawave Technologies). For each animal the mean inspired minute ventilation (VI) was computed for W during the control recordings and used as a baseline to normalize respiration during sleep and during diazepam administration in that animal. One-way ANOVAs were also performed by nonparametric (Kruskal-Wallis) analysis. Conclusions using parametric and nonparametric ANOVA were identical in all cases.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Figure 1 presents the effects of 0.05 and 5 mg/kg of diazepam on normalized inspiratory minute ventilation (NVI) during W, non-REM, and REM sleep. There was a significant increase of NVI at both doses during W (p = 0.003 for 0.05 mg/kg dose and p = 0.005 for 5 mg/kg dose) and non-REM sleep (p = 0.002 for 0.05 mg/kg dose and p = 0.0005 for 5 mg/kg dose) but only at the 5 mg/kg dose during REM sleep (p = 0.01).

View larger version (49K): [in this window] [in a new window]   Figure 1.   Normalized inspired minute ventilation (NVI; see text for details of the normalization procedure). Each bar represents the group mean ± SE over nine animals for the average NVI observed in a single behavioral state through each 6-h recording. *p Value < 0.05 versus control injection during respective behavioral state.

Figure 2 illustrates the dose-dependent effects of diazepam on Sp and PS apneas during non-REM sleep. Each dose of the drug (0.05 and 5 mg/kg) significantly reduced Sp apneas from approximately 20/h (control value) to 10/h (p = 0.02 and 0.01, respectively). PS apneas were reduced to a similar extent for the 0.05 and 5 mg/kg dose (p = 0.04 and 0.006, respectively). Neither Sp nor PS apneas occurring during REM sleep were reduced by either dose of diazepam (data not shown, p > 0.2 each).

View larger version (11K): [in this window] [in a new window]   Figure 2.   Expression of spontaneous and post-sigh apneas during non-REM sleep following injection of (0) (control), 0.05 or 5 mg/ kg diazepam. Each bar reflects group mean ± SE data. *p Value < 0.05 versus control.

Figure 3 demonstrates that only 5 mg/kg diazepam affected non-REM and REM sleep. While this hypnotic dose increased non-REM sleep (p = 0.03), it suppressed REM sleep (p = 0.009). In addition, only the 5 mg/kg dose reduced the amount of wakefulness (p < 0.05) and the number of awakenings during the 6-h recordings.

View larger version (10K): [in this window] [in a new window]   Figure 3.   Percent recording time (mean ± SE) spent in non-REM and REM sleep. *p Value < 0.05 versus control.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The current study directly supports the findings of three clinical studies where administration of short- or medium-acting benzodiazepines to central sleep apnea patients (5, 6) and normal subjects with altitude associated central apneas (20) improved respiration and reduced central sleep apneas. We obtained similar results in rats with diazepam, which despite having a long plasma half-life in humans, shows a relatively brief duration of central action due to redistribution out of the brain (21) and whose plasma half-life in rats is expected to be much shorter due to the high metabolism of these animals.

As in a clinical study of triazolam where the drug reduced apneas in sleep stage 1 with less impact on stage 2 apneas and without effect on apneas in REM sleep (5), our data show that in rats diazepam also decreased apneas in non-REM but not in REM sleep. It is of interest that triazolam at two dose levels decreased central apneas by about 50%, whereas in our study both doses of diazepam reduced Sp and PS apneas to about half of their baseline value.

The increased minute ventilation in W and sleep obtained in this study following diazepam administration (Figure 1) agrees with findings from a clinical study with conventionally prescribed dosages of triazolam where respiratory rate increased during W (22). The same dosage of triazolam, used in a different clinical study in which respiratory rate was not measured, reduced central apneas in NREM sleep (5). In another human study, triazolam was given to volunteers at two and three times the usual recommended dose and was found to alter respiratory cycle timing, causing an increase in breathing frequency (23). Thus, it appears that triazolam's effects on respiration and suppression of central apneas in humans are comparable to the effects of diazepam on the same parameters in the present study.

In contrast to the increase of NVI (Figure 1), which corresponded with apnea suppression (Figure 2) following each dose of diazepam in non-REM sleep, during REM sleep only the higher dose of the drug increased NVI (Figure 1), an effect which did not correspond with apnea suppression. It is also of interest that although the low dose of diazepam had no effect on any sleep stage volume (Figure 3), it was equipotent to the high dose in reducing non-REM Sp and PS apneas. These observations suggest a differential sensitivity between the respiratory and the sleep effects of diazepam, with significant promotion of NVI and suppression of apneas even at the 0.05 mg/ kg dose, which absolutely had no gross effects on sleep architecture (Figures 1, 2, and 3).

The functional site and mechanism of diazepam's action on apnea expression cannot be directly determined from the present data. Although respiratory stimulation and apnea suppression were clearly associated during non-REM sleep, this association did not persist during REM sleep. Thus, apnea suppression may not result simply from respiratory stimulation.

The sedating and hypnotic effects of benzodiazepines are well known, suggesting that diazepam may suppress apnea by improving sleep consolidation and reducing the number of sleep onsets, a state transition commonly associated with central apnea. However, the 0.05 mg/kg dose of diazepam in the present study was not associated with any change in the amount of wakefulness or number of awakenings with respect to control recordings. Still, non-REM apnea expression was halved by this dose. We cannot rule out the possibility that even 0.05 mg/kg diazepam reduced the number of very brief arousals, thereby stabilizing respiration. However, these arousals would have had to be significantly shorter than 5 s in duration, on average, because we staged sleep on 10-s epochs and the number of awakenings was unchanged by 0.05 mg/kg diazepam.

Finally, the intriguing possibility exists that diazepam reduced apnea expression by improving the coordination of central respiratory drive with the respiratory mechanical apparatus. Nonspecific activation of chest wall muscles may interfere with the ability of these muscles to respond optimally to the respiratory drive signal. If diazepam causes chest wall muscle relaxation during wakefulness and non-REM sleep, this may allow improved efficiency of the respiratory apparatus and contribute to the observed increase in minute ventilation. However, chest wall muscles (other than the diaphragm and parasternal intercostals) are silent during REM sleep, yet minute ventilation was also increased during this sleep state, an effect that cannot be attributed to the muscle relaxant action of diazepam. Overall, it appears most likely that respiratory stimulation contributes to apnea suppression during non- REM sleep, implying that the mechanisms underlying apnea expression differ between non-REM and REM sleep.

We conclude that administration of diazepam to rats induced respiratory stimulation even at doses insufficient to augment non-REM sleep and that this stimulation was associated with suppression of non-REM apneas. However, despite equivalent augmentation of minute ventilation by high-dose diazepam during non-REM sleep and REM sleep, REM apneas were not affected. The discordance of apnea-suppressing effects caused by diazepam during non-REM versus REM sleep in rats, coupled with the identical phenomena produced by another benzodiazepine, triazolam, in humans (5), suggests different mechanisms of apnea genesis during these sleep states, a hypothesis that we have recently advanced (18).