INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULT DISCUSSION REFERENCES

Sleep deprivation is a common feature of modern life. Sleep deprivation is exacerbated by shift work and jet lag. In addition, many clinical conditions result in disrupted sleep that worsens the degree of sleep deprivation, including chronic and acute respiratory disorders and primary sleep disorders (e.g., insomnia and obstructive sleep apnea). If sleep deprivation alters aspects of respiratory control this could have important implications for the management of patients with respiratory disease and sleep disorders. For example, it has been proposed that the sleep disruption in patients with severe chronic obstructive airways disease, often in association with breathlessness and nocturnal cough, can further impair the ventilatory reserve of these patients and contribute to the development of respiratory failure (1). Furthermore, sleep disruption has been shown to increase the severity of disordered breathing during subsequent nights of sleep in adults and infants with sleep- related breathing disorders (2).

A number of studies have examined the effect of sleep deprivation on respiratory control during wakefulness in healthy adults (1, 7). Three of five studies found that one night of sleep deprivation significantly reduces the awake hypercapnic ventilatory response (HCVR) by 17 to 24% (1, 7, 8); the other studies found no significant change (9, 10). Similarly, a reduction in HCVR was found after one night of sleep deprivation in patients with chronic obstructive pulmonary disease (COPD) (11). In addition, sleep deprivation caused significant reductions in the arousability from sleep during induced hypercapnia, hypoxia, laryngeal stimulation, and upper airway obstruction in dogs (12, 13) and during bronchoconstriction in subjects with asthma (14).

Although it seems clear that sleep deprivation reduced both ventilatory chemosensitivity and arousal responses in the above-mentioned studies, the underlying mechanisms are unknown. One possibility is that the altered responses are caused by sleep deprivation itself (such as the gradual build-up of a sleep-promoting factor affecting the brainstem respiratory complex). However, it is equally plausible that the altered respiratory and arousal responses in these earlier studies are an artifact of the experimental protocols, with these changes being caused by the altered behaviors that occur as a consequence of imposed nighttime wakefulness, such as increased activity, stress, social interaction, food intake, ambient light and noise, or altered posture. For example, the increased light exposure during the sleep deprivation period will likely shift the phase of the circadian rhythm (with unknown consequences on respiratory control) because subjects are exposed to light during their usual sleep hours when the resetting of the endogenous circadian pacemaker is most sensitive to the effects of even low-level room light (15, 16). In addition, it should be noted that even without sleep deprivation, there is substantial between-day variability in HCVR measurements (the source of this variability is not known). For example, Sahn and coworkers (17) found the within-day coefficient of variation in HCVR to be 17.9% (range 8.3-26.3%), and that the between-day variability was as much as five times greater than the within-day variability in more than half of the subjects. The existence of this degree of between-day variability makes it difficult to be certain that any changes observed from single measurements before and after sleep deprivation in the above-mentioned studies (1, 7) are attributable to the effects of sleep deprivation alone.

The aim of the current study was to determine the "pure" effect of sleep deprivation on respiratory control without the confounding effects of altered behaviors and environment during the imposed wakefulness. This was achieved by ensuring that the subjects maintained a constant posture (semirecumbent, without gross body movements), ate regular small meals (uniform snacks every 2 h), had constant interaction with experimenters, and stayed in an environment with constant, very low light (10 lux) and constant room temperature. Additional improvements of the current protocol relative to studies mentioned above include a repeated measurements design (six measurements performed both before and after 24 h of sleep deprivation), measurements made across a substantial portion of the circadian cycle instead of at only one point, electroencephalogram (EEG) confirmation of wakefulness throughout the protocol, and assessment of resting breathing and metabolism in addition to ventilatory chemosensitivity.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULT DISCUSSION REFERENCES

The study was approved by the internal review board of Brigham and Women's Hospital (Boston, MA). Written informed consent was obtained from each subject. Ten healthy, male subjects participated. They were 23.7 ± 3.9 (mean ± SD) yr old, 176.3 ± 6.4 cm in height, and weighed 73.9 ± 11.3 kg. Their lung function was normal: vital capacity 5.5 ± 0.9 L, forced expiratory volume in 1 s 4.5 ± 0.8 L, and peak flow 10.2 ± 1.9 L · s¹. One week before the study, subjects abstained from caffeinated drinks, smoking, and any medication. Individuals with evidence of significant psychopathology were excluded. For at least 2 wk before each study, subjects were required to establish a regular sleep-wake cycle with their habitual bedtimes and waketimes varying by no more than 1 h each day. For at least 3 d before the start of the laboratory phase of the study, subject activity was monitored by a wrist-worn activity monitor (Ambulatory Monitoring, Ardsley, NY). During the inpatient portion of the study, subjects were isolated from external time cues, including clocks, radios, television, visitors, mail, and sunlight, but maintained contact with staff members, who were trained to avoid communicating the time of day. Environmental temperature was maintained at approximately 23° C.

Protocol

The subjects spent 4 days and nights in the laboratory. These consisted of two adaptation days and nights, 41 h of "constant routine" (see below), and a recovery night. For analysis of blood cortisol concentration, 1.75-ml venous blood samples were collected every half-hour via an indwelling forearm catheter. The samples were immediately centrifuged at 4° C, and the plasma was pipetted into polystyrene tubes and frozen at 20° C. Cortisol analysis was performed with a paramagnetic, chemoluminescent immunoassay (Beckman Coulter, Miami, FL). Core body temperature (CBT) was recorded every minute by means of a rectal temperature thermistor (Yellow Springs Instrument Company, Yellow Springs, OH). Food was calculated to have the same composition of 25% fat, 50% carbohydrate, and 25% protein every day [calculated by the Harris-Benedict formula with an activity factor of 1.4 (18)]. The subjects also received 3.5 L of fluids per day.

Adaptation Days and Nights

Day 1. Food and drink were provided in four portions: breakfast, lunch, dinner, and a snack. During this first day, the subjects were familiarized with each of the respiratory tests that they were to perform repetitively during the "constant routine" (see below); spirometric tests were performed to assure normal lung function. The light level was 150 lux for this day.

Night 1 (adaptation night). Surface electrodes (Beckman Instrument Company, Schiller Park, IL) were applied at bedtime for recording of two EEGs, two electrooculograms, and a submental electromyogram. Also, subjects were equipped with two plethysmographic belts around the chest (Respitrace; Ambulatory Monitoring), a gas-sampling tube in the nostril for O₂ and CO₂ analyses, and fingerclip to monitor oxygen saturation (Cardiocap II; Datex Medical Instruments, Tewksbury, MA). The light level was < 0.02 lux. All signals were recorded on a Nihon-Kohden electroencephalograph (Nihon-Kohden Instrument Company, Irvine, CA). These measurements were taken to assure the absence of any primary sleep disorders.

Day 2 and Night 2. Day 2 and Night 2 were scheduled in a similar way as Day 1 and Night 1.

Constant Routine Protocol

Day 3, Night 3, and Day 4 ("constant routine"). After being woken up at their habitual time, the subjects stayed in bed and awake in a semirecumbent position for 41 h. The light level was 10 lux. Food and drink were provided in identical 2-hourly snacks. At all times, an experimenter was in the laboratory room to ensure that the subject did not fall asleep. All necessary adjustments were performed during the first hour of wakefulness (e.g., change to semirecumbent posture, apply or reapply electrodes, etc.). Thereafter, the subjects performed the following tests at 2-hourly intervals:

Resting O₂ consumption, CO₂ production, and ventilation. Subjects wore a noseclip and breathed through a mouthpiece for 10 min. Ventilatory variables (E [minute ventilation], VT [tidal volume], and fR [respiratory frequency]), as well as O₂ consumption (CO₂) and CO₂ production (CO₂) were calculated breath by breath, using a calibrated metabolic cart that compensated for any changes in barometric pressure (OxyconAlpha system; Erich Jaeger, Würzburg, Germany). This system used fast-responding gas analyzers (paramagnetic for O₂ and infrared for CO₂) and a turbine for volume measurement. To ensure a relatively constant level of relaxed wakefulness throughout this test, subjects were asked to move as little as possible, to gaze at a picture in front of them, and to memorize the number of short "beep" sounds that were programmed to occur randomly at 10 to 15-s intervals. To ensure a steady state, only breath-by-breath data from the 5th to the 9th min were averaged and used in comparisons.

Hypercapnic ventilatory response. Subjects again wore a noseclip and breathed through a mouthpiece. After 2 min, a valve was turned close to the mouthpiece to switch the subjects into a spirometry breathing circuit (Warren E. Collins, Braintree, MA). This is a modification of the Read rebreathing technique (19). The spirometer was initially filled with 7% CO₂ and 93% O₂ to a volume of 1 L above the vital capacity of the subject. The high starting CO₂ ensured that inhaled CO₂ was close to equilibrium with mixed venous PCO₂. The high inhaled O₂ prevented hypoxia, ensuring that Pa_CO2 was the only chemosensitive stimulus to breathe throughout the HCVR.

E and end-tidal CO₂ pressure (PET_CO2) were determined for every breath. Values were corrected for humidity and barometric pressure, and converted to BTPS. For determination of the CO₂ sensitivity, a linear regression was performed on breath-by-breath values of E and PET_CO2. Since the relationship between E and PET_CO2 is linear only above a threshold PET_CO2 (20) an objective method was used to determine the point on the E/PET_CO2 relationship above which there was a positive-going slope (21). In short, sequential breaths were added to the linear regression until the slope of the regression line reached and then stayed above +0.15 for at least five consecutive breaths. To determine the HCVR slope, a linear regression was calculated using all data collected at a PET_CO2 above this threshold.

Analysis

The data of all 10 subjects were aligned with respect to their wake-up time and then averaged. Measurements were taken in 2-hourly intervals, starting 1 h after waking up, for 40 h, resulting in 20 measurements per subject. The data obtained during the first 5 h of the constant routine protocol were excluded from the analysis to eliminate any residual masking effects of sleep (15). To assess the effect of sleep deprivation, paired measurements needed to be 24 h apart. This enabled us to compare six pairs of measurements, the first six measurements were between 5 and 17 h after awakening and the second six measurements were between 29 to 41 h after awakening (i.e., encompassing approximately 1:00 P.M. through 1:00 A.M. on each day, depending on the habitual wake time of each subject). Thus, for group analysis, two-way repeated measures analyses of variance were performed on all variables, comparing six pairs of measurements with each pair separated by 24 h of sleep deprivation (main effect = sleep deprivation) (SAS statistical software; SAS Institute, Cary, NC). In addition, for within-subject comparisons of the effect of sleep deprivation, the six measurements before and after 24 h of sleep deprivation were compared by paired t tests. Significance was noted if p < 0.05 in two-tailed tests.

	RESULT

TOP ABSTRACT INTRODUCTION METHODS RESULT DISCUSSION REFERENCES

All subjects remained awake throughout the protocol. This was verified by direct observation and interaction with the subject throughout the protocol, and by EEG recordings during the periods of measurement.

The mean values of each of the variables derived from the various tests are shown in Table 1, separated into averages before and after sleep deprivation. In contrast to previous results from other studies, in this strictly controlled protocol, 24 h of sleep deprivation failed to cause a significant decrease in HCVR. Indeed, mean HCVR increased by 17% after 24 h of sleep deprivation (nonsignificant; p = 0.252). Likewise, sleep deprivation did not cause a significant change in PET_CO2, ventilation, breathing pattern variables, or metabolism. Overall, the repeated measures analyses of variance detected only two significant effects of sleep deprivation: a 20% increase in plasma cortisol level (p = 0.017) and a 2% increase in systolic blood pressure (p = 0.046) (Table 1).

The comparisons of individual changes revealed that 3 of the 10 subjects had a significant increase in HCVR after 24 h of sleep deprivation (Figure 1), whereas 7 of the 10 subjects had no significant change (and none of the 10 subjects had a significant reduction in HCVR after sleep deprivation). The number of subjects who had significant changes in other variables with 24 h of sleep deprivation were as follows: PET_CO2, 2 subjects significantly increased, 1 subject significantly decreased (7 subjects had no significant change); E (10, no change); VT, 1 increased (9, no change); fR, 2 increased, 1 decreased (7, no change); O₂, 1 increased, 3 decreased (6, no change); CO₂, 2 increased (8, no change); cortisol, 3 increased (7, no change); systolic blood pressure, 3 increased (7, no change); diastolic blood pressure, 1 increased (9, no change); heart rate, 1 decreased (9, no change); body temperature, 1 increased, 3 decreased (6, no change). Thus, for each variable at least 60% of the subjects had no discernible change with sleep deprivation. This is relatively consistent with the results of the group analyses.

View larger version (14K): [in this window] [in a new window]   Figure 1.   Effect of 24 h of sleep deprivation on hypercapnic ventilatory response (HCVR). Individual data are shown before and after 24 h of sleep deprivation (means of six measurements before and six measurements after 24 h of sleep deprivation are connected by lines). The six individuals who increased HCVR after 24 h of sleep deprivation are indicated by circles (asterisk indicates significant increase); the four individuals with decreased HCVR are indicated by squares, and the group mean values are shown as short lines.

Since cortisol and systolic blood pressure increased after sleep deprivation, we performed additional correlation analyses to determine the relationship between these markers of stress and the HCVR. Across the 12 measurements, we found no significant correlation between HCVR and cortisol (r = 0.35, n = 12, p = 0.28), or between HCVR and systolic blood pressure (r = 0.27, n = 12, p = 0.40). Furthermore, using the individual data, we found no significant correlation between mean HCVR (averaged within subjects over six measurements before sleep deprivation) and the individual mean changes in cortisol over the 24 h of sleep deprivation (r = 0.27, n = 10, p = 0.46), and no significant correlation between mean HCVR and the individual mean changes in systolic blood pressure over the 24 h of sleep deprivation (r = 0.03, n = 10, p = 0.93).

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULT DISCUSSION REFERENCES

We found no significant change in HCVR with 24 h of sleep deprivation in contrast to three of five previous studies, which had found significant decreases in HCVR with 24 h of sleep deprivation ranging from 17 to 24%; weighted mean change was 19% (1, 7, 8). Moreover, with 24 h of sleep deprivation none of our 10 subjects had a significant decrease in HCVR, 3 subjects significantly increased their HCVR, and there was an overall group mean change of +17%. Thus, our results were different from previous results. Since we found a mean increase in HCVR with sleep deprivation, it is extremely unlikely that we would have made a Type II error of failing to find a significant decrease in HCVR when one truly exists. Our most conservative conclusion is that there is no significant reduction in HCVR after 24 h of sleep deprivation when studied under strictly controlled behavioral and environmental conditions in the current experiment. We found no significant effects of sleep deprivation on resting ventilation, basal breathing pattern, metabolism, or PET_CO2, which is consistent with most other reports (8, 14, 22) [although one study found that resting ventilation increased after 30 h of sleep loss (23)].

Because of the difference between the HCVR results of the current study and the HCVR results of previous studies mentioned above, it is important to examine differences in protocol that could explain this discrepancy. Insufficient information was given in these other reports to determine the true extent of the behavioral or environmental conditions during the period of sleep deprivation, although subjects certainly worked throughout the night of sleep deprivation in some studies (1, 7, 9). Our study had much stricter control of all behavioral and environmental conditions than the other studies, including posture, diet, physical activity, social interaction, light level, sound level, and temperature. All of these conditions could affect breathing and possibly the HCVR during the specific activities. For instance, reading has been shown to increase the HCVR by 9% (24). There are profound effects of food intake and physical activity on breathing, and cognitive activity and merely opening the eyes have been shown to significantly affect breathing (25). However, as behavioral and environmental conditions were likely similar before and after sleep deprivation in all studies, if the behavioral and environmental conditions were the reason for the differences in the results among studies, we would have to assume that such effects on ventilatory control were relatively long lasting (akin to a long term potentiation). Such a long-lasting effect could occur, for example, with changes in diet, because of the specific dynamic action of food (26). Similarly, the increased light exposure during the sleep deprivation period will likely shift the phase of the circadian rhythm because subjects are exposed to light during their usual sleep hours when the resetting of the endogenous circadian pacemaker is most sensitive to the effects of even low-level room light (15, 16). It is likely that the numerous physiological changes that occur across the circadian cycle would have some effect on ventilatory control. For instance, core body temperature varies with the circadian cycle (15) and CO₂ sensitivity is affected by changes in core body temperature (27, 28). In our study identical snacks were served at 2-hourly intervals, and the subjects were in dim light, ensuring that the circadian phase was not altered, and we found no effect of sleep deprivation on core body temperature. Also, in the current study measurements before and after sleep deprivation spanned a 12-h period, thereby encompassing approximately half of the circadian cycle, whereas other studies made a maximum of two HCVR measurements within 1 h of each other both before and after sleep deprivation at undetermined phases of the circadian cycle (1, 7, 9). Further differences among protocols included the fact that the current study incorporated a 2-d laboratory acclimation period, data of the first 5 h after awakening were excluded from analysis to ensure that there was no residual effect of sleep, postural change, or the nocturnal fast on the results, and wakefulness was confirmed by constant interaction with the subjects during the sleep deprivation periods, and this was verified during each measurement period by analysis of EEG [only one of the other studies recorded EEG during the HCVR tests (8)]. Last, similar techniques to provide the CO₂ stimulus were used in each study (i.e., rebreathing in hyperoxia) so it is unlikely that this accounted for the different results (1, 7). Nonetheless, we used an objective technique to derive the PET_CO2 threshold point beyond which the HCVR slope was determined, whereas the other studies did not fully define whether or not they used a threshold or how the PET_CO2 threshold was defined. We believe that an objective determination of PET_CO2 threshold is imperative because of various behavioral influences on breathing (e.g., the wakefulness drive to breathe) that maintain ventilation even at a PCO₂ level below the central chemoreceptor threshold for activation of ventilation (reviewed in References 20 and 25). With all of these additional precautions, we believe that we have been able to demonstrate quite conclusively that sleep deprivation per se does not reduce hypercapnic ventilatory chemosensitivity.

Two caveats to this interpretation are worth mentioning. First, there was a significant increase in plasma cortisol and a small increase in systolic blood pressure after 24 h of sleep deprivation in the current study. It is possible that our "constant routine" procedure involves a higher degree of stress than other sleep deprivation protocols that allow subjects more control over their behaviors. Such stress could affect HCVR and counteract any effect of sleep deprivation on HCVR. This concern cannot be answered with the available data because the effect of stress on HCVR has not been established, and other studies generally did not measure these other physiological variables. Cortisol has not been shown to affect respiration directly and changes in blood pressure were small (mean increase of systolic blood pressure by 2.5 mm Hg). Nonetheless, we found no correlation between these markers of stress and the HCVR either on a group or individual basis. Moreover, we found no change in heart rate before and after sleep deprivation, suggesting that stress was minimal. Also, our subjects had the greatest trouble staying awake, and presumably the greatest amount of stress, in the early morning hours on the night of sleep deprivation, which is the time span excluded from analysis between the two 12-h periods compared in this study. Finally, our subjects had little difficulty remaining awake throughout the daytime before and after sleep deprivation, presumably because of the circadian rhythm of alertness (29). Thus, we do not believe that stress significantly contributed to our results any more than would have occurred in the other studies. The second caveat is that repetitive CO₂ challenges every 2 h throughout the present study may have affected subsequent measurements, for example, via long-term facilitation. While this certainly remains a possibility, we believe that such an effect is unlikely to have affected our interpretation because other authors have found no significant mean change in HCVR during six sequential CO₂ challenges separated by 5-min breaks (30), or during four sequential CO₂ challenges separated by 30-min breaks (17).

Summary

Our results strongly suggest that 24 h of sleep deprivation per se does not result in a significant reduction in HCVR in humans. We were surprised by these results because of the weight of evidence presented in the other similar, although less well controlled studies of sleep deprivation. Thus, we believe that further controlled studies are necessary to completely understand this issue. Last, we need to be aware that our research has studied only short-term total sleep deprivation in healthy subjects. The implications of this research in terms of chronic sleep loss, sleep fragmentation, and the impact of sleep loss in patients with respiratory disorders and/or sleep disorders are still unknown, but may be less important than hitherto suggested.