INTRODUCTION

Top Abstract Introduction Methods Results Discussion References

VENTILATION DECREASES and arterial PCO₂ increases during non-rapid-eye-movement (NREM) sleep in normal healthy humans (2, 19). One mechanism that contributes to this phenomenon is a sleep-related loss of drive to the upper airway muscles, resulting in an increase in resistance and decrease in ventilation (5, 11). An important aspect of this mechanism is that compensatory responses to increased resistive loads are impaired during sleep, allowing the decrease in ventilation. Thus, during wakefulness, an increase in airway resistance is immediately compensated for, resulting in a maintenance in the level of ventilation (13). During sleep, however, this immediate reflexive component of resistive load compensation is effectively absent (4, 10, 13, 23, 24). Furthermore, the sensitivity of chemical responsiveness is reduced (19). Thus, after application of a load, ventilation decreases and then gradually recovers, but not to preload levels.

A second mechanism that contributes to the decrease in ventilation during sleep is a loss of central drive to the respiratory pump muscles. The independence of this mechanism from that of airway resistance has been demonstrated by a number of procedures. For example, ventilation has been shown to decrease during sleep in the absence of an increase in resistance, either in patients with tracheostomies (17) or by use of continuous positive airway pressure (18). Furthermore, the PCO₂ threshold is higher during sleep than during wakefulness when the effects of airway resistance are removed by mechanical ventilation (20).

The relative contribution of the increase in airway resistance and loss of central drive to the respiratory pump muscles to the sleep-related decrease in ventilation and increase in PCO₂ remains unclear. It is possible that their relative importance differs between individuals (14, 15) and differs as a function of the development of sleep within a particular sleep period (16). During sleep onset, ventilation is critically dependent on fluctuations in sleep-wake state, with large decreases in ventilation occurring abruptly at the loss of alpha activity in the electroencephalogram (EEG) (3, 22). In contrast, although sleep-related changes in airway resistance are initiated at the loss of alpha activity, the changes are initially small and increase gradually as a function of the development of sleep (15, 16). Thus, changes in ventilation early in sleep onset appear less dependent on airway resistance than they appear later in sleep onset.

A related issue, which remains to be resolved, is whether the small changes in resistance that occur early in sleep onset elicit compensatory responses. That is, is reflex compensation to resistive loading lost abruptly with the loss of alpha EEG activity, permitting the rise in airway resistance to affect ventilation, or does the loss of compensation occur gradually as sleep develops? This study aimed to assess the effect of state, as defined by the presence vs. absence of alpha EEG activity during sleep onset, on the presence of reflexive load compensation.

	METHODS

Top Abstract Introduction Methods Results Discussion References

Subjects and Design

Procedures

General laboratory procedures. The study was approved by the University of Melbourne's Human Ethics Committee, and subjects gave written informed consent before participation. Subjects were asked to refrain from consuming caffeine or alcohol on the day before each sleep session. Because the study was concerned with the sleep-onset period, each experimental session involved use of a multiple sleep-onset technique. After the attainment of ~10 min of stable stage 2 sleep, subjects were awakened and kept awake until alert. Subjects were then requested to go to sleep again. A 3- to 4-h recording session resulted in three to eight sleep onsets per subject. During data collection, subjects were required to maintain a supine position. They were instructed as to the detection of mask leaks and were asked to alert the experimenter if these occurred. The experimenter also checked the mask for leaks during the period of wakefulness between each sleep onset.

Measurement of ventilation. Subjects wore a face mask used for anesthesia (size 6; Commonwealth Industrial Gases). The mask was tightly secured by using a head strap to cover the nose and mouth, and it was attached to a two-way nonrebreathing valve (Hans Rudolph series 2600). The mask and breathing valve had a dead space of ~120 ml. Inspiratory airflow was measured by using a heated Morgan pneumotachograph that was placed in the inspiratory line and connected to a differential pressure transducer (Validyne DP45-14). The output was converted to a voltage signal by using a carrier demodulator (Validyne CD15). Without an added external load, the inspiratory circuit had an internal resistance of 2.6 cmH₂O · l¹ · s at a flow rate of 15 l/min. Airflow was calibrated before each session by using a flowmeter (model 1355 Shorate). Inspiratory volume was calculated by computer integration of the flow signal. Automated off-line analysis was used to calculate tidal volume (VT), minute inspiratory ventilation (I), inspiratory duration (TI), and total cycle duration (TT).

Measurement of sleep state. Sleep state was determined by using gold cup surface electrodes to record central (C₃-A₂) and occipital (O₁-A₂) EEGs, electrooculogram, and submental electromyogram (EMG) activity. On the basis of these recordings, each sleep onset was divided into three phases. The rationale and scoring procedures for the subclassification of sleep onset have been previously presented (14) and is briefly described below.

Automated EEG scoring. An automated period analysis of the occipital EEG was first used to classify the EEG frequency associated with each breath. A ratio of EEG alpha activity (8 Hz) to theta activity (<8 Hz) was calculated for each breath by using a peak-to-peak period analysis. The obtained ratio was then compared with a criterion ratio to determine whether its value indicated predominantly alpha or theta EEG activity. The criterion ratio was computed separately for each subject because of individual variability of alpha in the waking state. To calculate the criterion ratio, samples of 100-150 breaths of unambiguous alpha and theta EEG activity were visually identified by an experienced scorer. The alpha-to-theta ratio for each of these breaths was calculated, and a signal-detection method (equal likelihood ratio) was used to determine the value (criterion) that discriminated between unambiguous alpha and unambiguous theta breaths. These methods have been described in detail in a previous publication (14). Visual analysis was then used to identify arousals not defined by alpha activity and to identify periods of theta associated with sleep spindles and K-complexes.

Application of external resistive loads. To apply a load, the inspiratory side of the breathing circuit was connected to a two-way tap which allowed for inspiration either with or without an additional external resistive load. The loading circuit was constructed by using layers of fine wire mesh that produced an added external resistance of 3.4 cmH₂O · l¹ · s. Thus, the breathing apparatus plus the additional resistive load totaled 6 cmH₂O · l¹ · s. The additional load was barely perceptible to subjects and was selected to be approximately the same size as average physiological resistive loads that occur in response to upper airway changes during alpha-to-theta transitions in normal sleep (14).

Data Reduction

Control data were obtained for the same states and phases as loaded trials. The definition of the beginning of a control trial in the continuous alpha and continuous theta conditions was the sixth breath after termination of a previous loaded trial. Where this was not possible to obtain because of a movement, rearousal, or application of another load, control trials were obtained unsystematically from unloaded periods when the experimenter was blinded to ventilation variables. Control trials after an alpha-to-theta transition were obtained from transitions in which loads were not applied. Analysis of control data was identical to that of loaded data.

Because of the difficulty of rapid visual classification of breaths as alpha or theta during the experiment, it was not always possible to apply the load on the first theta breath of an alpha-to-theta transition. Off-line analysis indicated the load was applied on either the first, second, or third theta breath (40, 44, or 16% of trials, respectively) after an alpha-to-theta transition. This resulted in a maximum of two unloaded theta breaths that were after a transition but preceding a load. These breaths were included in the analysis, because we felt it was important to maintain each breath's particular position with respect to the transition.

Data Analysis

The second analysis compared loaded and control trials during theta as a function of whether the load was administered during continuous theta or immediately after a transition from alpha to theta. For the purpose of this analysis, the two levels of state refer to periods of continuous theta and transitions from alpha to theta EEG activity. In addition, this analysis determined whether there was an effect of phase during sleep onset [phase 2 (early in sleep onset) vs. phase 3 (late in sleep onset)] in response to the load. A 2 (state: continuous theta vs. after transition into theta) × 2 (condition: load vs. control) × 2 (phase 2 vs. phase 3) × 5 (breath position) ANOVA, with repeated measures on each factor, was conducted.

	RESULTS

Top Abstract Introduction Methods Results Discussion References

The average number of trials obtained for subjects in each of the loaded and control conditions was 52 (range, 40-78) in continuous alpha, 17 (range, 10-25) in continuous theta (phase 2), 58 (range, 47-78) in continuous theta (phase 3), 16.33 (range, 10-33) at alpha-to-theta transitions (phase 2), and 26.67 (range, 15-45) at alpha-to-theta transitions (phase 3).

Continuous Alpha vs. Continuous Theta EEG Activity

View larger version (22K): [in this window] [in a new window]   Fig. 1.   Average inspiratory profiles as %preload baseline for respiratory variables during loaded () and control () trials in continuous alpha electroencephalogram (EEG) activity. A: minute ventilation; B: tidal volume; C: inspiratory duration; D: total cycle duration. Mean values for each breath position were obtained by averaging trials across nights, and then averaging over subjects. Raw mean values were then expressed as %preload/control baseline. Bars, SD values obtained by first expressing each postload breath as %preload/control baseline and then calculating SD of difference between load and control values over subjects. * Significant difference between load and control values. P < 0.05.

View larger version (24K): [in this window] [in a new window]   Fig. 2.   Average inspiratory profiles as %preload baseline for respiratory variables during loaded () and control () trials in continuous theta EEG activity (phases 2 and 3 combined). A: minute ventilation; B: tidal volume; C: inspiratory duration; D: total cycle duration. See Fig. 1 legend for further explanation.

View this table: [in this window] [in a new window]   Table 1.   Absolute I and TI values for load and control breaths during each state and phase of sleep onset

The maintenance of ventilation in the alpha state was associated with an immediate prolongation in TI on the first loaded alpha breath (Fig. 1C), the effect being significant for breaths 1 and 2. During theta, there was a more gradual increase in TI over loaded theta breaths (Fig. 2C). The significance of these effects was indicated by significant state-by-breath position [F(4,20) = 41.47, P < 0.05] and state-by-condition-by-breath position [F(4,20) = 8.23, P < 0.001] interactions for TI. Thus, it appears that the gradual return of ventilation toward preload levels during loaded theta trials occurred because of a progressive increase in TI. No variation in TT was detected for loaded or control conditions in either state.

The results of this analysis were consistent with the hypothesis that immediate compensation in response to application of an external resistive load occurs during wakefulness but not during sleep. Furthermore, the results indicate that these state differences in inspiratory control change rapidly with changes in state. The immediate reflex response is present during brief periods of alpha activity interposed between periods of theta but not during the brief periods of theta.

Transitions From Alpha to Theta vs. Continuous Theta EEG Activity

View larger version (23K): [in this window] [in a new window]   Fig. 3.   Average inspiratory profiles as %preload baseline for respiratory variables during loaded () and control () trials at alpha-to-theta EEG activity transitions (phase 2). A: minute ventilation; B: tidal volume; C: inspiratory duration; D: total cycle duration. See Fig. 1 legend for further explanation.

View larger version (23K): [in this window] [in a new window]   Fig. 4.   Average inspiratory profiles as %preload baseline for respiratory variables during loaded () and control () trials at alpha-to-theta EEG activity transitions (phase 3). A: minute ventilation; B: tidal volume; C: inspiratory duration; D: total cycle duration. See Fig. 1 legend for further explanation.

View larger version (24K): [in this window] [in a new window]   Fig. 5.   Average inspiratory profiles as %preload baseline for respiratory variables during loaded () and control () trials in continuous theta EEG activity (phase 2). A: minute ventilation; B: tidal volume; C: inspiratory duration; D: total cycle duration. See Fig. 1 legend for further explanation.

View larger version (23K): [in this window] [in a new window]   Fig. 6.   Average inspiratory profiles as %preload baseline for respiratory variables during loaded () and control () trials in continuous theta EEG activity (phase 3). A: minute ventilation; B: tidal volume; C: inspiratory duration; D: total cycle duration. See Fig. 1 legend for further explanation.

Because of the confounding effect of the state transition on I, the critical comparison in determining the effect of the experimental load was whether the difference between the loaded and control condition differed between continuous theta and theta immediately after a transition. This comparison held constant the effect of the alpha-to-theta transition. It indicated that during phase 2, the effect of the experimental load was smaller in theta immediately after a transition compared with continuous theta. Inspection of Table 1 indicates that in phase 2 the decrease in I was 0.77 l/min greater in the loaded than in the control condition for continuous theta, but only 0.31 l/min greater in theta after an alpha-to-theta transition. The effect can also be observed by comparing the difference between load and control for Fig. 3, A and B, with Fig. 5, A and B. These results suggest relatively good compensation in phase 2 soon after the transition into theta.

In phase 3, however, the decrease in I was 1.15 l/min greater for the loaded condition in continuous theta and 0.87 l/min greater in theta after an alpha-to-theta transition, indicating poor compensation soon after the transition into theta in phase 3. This can also be seen by comparing Fig. 4, A and B, with Fig. 6, A and B. Indeed, phase had only a small effect on the difference between load and control conditions during continuous theta but had a substantial effect at transitions. Statistically, this pattern was supported by the significant phase [F(1,5) = 16.68, P < 0.01], state-by-phase [F(1,5) = 7.88, P < 0.05], but not state-by-condition-by-phase [F(1,5) = 1.06, P > 0.05] effects for I. In summary, late in sleep onset (phase 3), the experimental load had the same effect on ventilation whenever it was applied during theta and was additive with physiological changes at alpha-to-theta transitions. However, early in sleep onset (phase 2), the experimental load had a smaller effect when applied immediately after the alpha-to-theta transition than when applied during continuous theta.

Consideration of the pattern of change in ventilation after load application indicated differences between continuous theta and theta after a transition. After a transition into theta, with the application of a load, ventilation decreased progressively until the third breath in the series and then recovered slightly on the last two breaths. This effect was particularly marked in phase 3 (Fig. 4, A and B). After a transition without application of a load (control trials), ventilation did not recover over subsequent breaths. In contrast, the decrease in ventilation during loaded periods of continuous theta was at a maximum on the first breath, recovering slightly for each subsequent breath in the series (Fig. 6, A and B). These observations were confirmed by statistical analysis that revealed a significant main effect of breath position for I and VT [F(4,20) = 5.55, P < 0.01; and F(4,20) = 3.43, P < 0.05, respectively], and a significant interaction among state, phase, and breath position for I [F(4,20) = 3.7, P < 0.05], and a condition-phase-breath position interaction for VT [F(4,20) = 3.04, P < 0.05]. The differences in the pattern of decrease in ventilation after load application most likely reflects two factors: the influence of the transition itself and the fact that the load could not always be applied on the first posttransitional breath.

Inspiratory duration during loaded compared with control trials increased progressively from the second loaded breath during alpha-to-theta transitions in phase 2 (Fig. 3C) and increased progressively to a smaller extent during loaded breaths in continuous theta (Fig. 5C). This was confirmed by a significant state-by-breath position [F(4,20) = 5.42, P < 0.01] and condition-by-breath position [F(4,20) = 9.42, P < 0.001] interaction for TI. Although a progressive increase in TI was observed during loaded transitions in phase 3, a similar but slightly smaller progressive increase was also observed for control trials (Fig. 6C). With respect to TT, no variation between loaded and control conditions was observed at alpha-to-theta transitions except for a fourth breath increase during loaded phase 2 (Fig. 3D) and a third breath increase during loaded phase 3 (Fig. 4D) transitions.

Absolute Ventilation Values

	DISCUSSION

Top Abstract Introduction Methods Results Discussion References

The results of the study indicate that the presence vs. absence of the load-compensating reflex is tightly controlled by state. Thus, during sleep onset, when the sleep-wake state fluctuates between brief periods of EEG alpha activity (wakefulness) and EEG theta activity (sleep), the compensation reflex is intact during the brief periods of wakefulness but not during brief periods of sleep. This is consistent with, and adds to, previous resistive- and elastic-loading studies that demonstrate immediate compensation in relaxed wakefulness and an absence of compensation during stable NREM sleep (9, 10, 13, 23, 24). The mechanisms responsible for immediate load compensation during wakefulness are likely to involve reflex actions mediated by stimulation of airway, lung, and chest wall mechanoreceptors and by the intrinsic properties of the respiratory muscles.

The loss of reflex compensation was, in general, very rapid after the loss of alpha activity in the EEG. Thus, late in sleep onset (phase 3), the loss of reflex compensation was within a breath of the transition. However, this loss was less rapid early in sleep onset, as loading had no appreciable effect on ventilation after an alpha-to-theta transition in phase 2. Nevertheless, this effect was restricted to the first few breaths after the loss of alpha activity, because loading in continuous theta in phase 2 was not compensated for. It should be noted that the difference between the theta state at alpha-to-theta transitions and during continuous theta was that in the former the load was applied within the first three theta breaths after a transition, while in the later it was applied after a minimum of five continuous theta breaths. It is unclear why reflex load compensation does not drop out very early in sleep onset. We speculate that this may be caused by a loose coupling between sleep and respiratory mechanisms and between the different respiratory components, early as opposed to later in sleep onset. Thus, early in sleep onset, although ventilation is affected immediately, the influence of sleep on reflex compensation is delayed. However, as sleep develops, the influence of sleep mechanisms on respiratory activity is more effectively coordinated, such that with each reentry into sleep, reflex load compensation is lost immediately.

The decrease in ventilation during alpha-to-theta transition control trials, where no additional load was applied, occurred rapidly (within 1-3 breaths) at the transition in both phases 2 and 3, with the decrease in ventilation being greater in phase 3 than in phase 2. This pattern was consistent with previous research investigating ventilation changes during sleep onset (22). Although airway resistance was not measured in this study, previous work has shown small increases in resistance at the alpha-to-theta transition (14-16). However, as noted above, the lack of any appreciable reduction in ventilation for loaded compared with control trials at phase 2 alpha-to-theta transitions indicates that compensation was still present for theta breaths immediately after alpha breaths in phase 2. This suggests that in the absence of the experimental load, the normal decrease in ventilation at the alpha-to-theta transition was not caused by an increase in airway resistance. Rather, by exclusion, the initial sleep-induced decrease in ventilation appears to be largely caused by reduced central drive of the respiratory pump muscles. However, the larger reduction in ventilation, observed from addition of a load during phase 3 alpha-to-theta transitions, indicates the relative absence of a load-compensating response. This suggests that the increase in airway resistance plays a greater role in the decrease in ventilation at normal alpha-to-theta transitions as sleep onset develops. Thus, consistent with a previous study (16), the data indicate that the increase in airway resistance and loss of central drive to the respiratory pump muscles both contribute to the sleep-induced decrease in ventilation, but the timing of the two effects differs over the progression of sleep onset.

TI showed a gradual increase after external loading in both continuous theta and in theta after an alpha-to-theta transition. The gradual increase is likely to caused by stimulation of chemoreceptors in response to hypoventilation produced from the increased mechanical load. However, it is of interest to note that although ventilation decreased during normal alpha-to-theta transitions, there was no associated change in TI. This finding is consistent with previous research (15, 21, 22). The difference in the response of TI to external resistive loads as opposed to the physiological load associated with alpha-to-theta transitions may be caused by qualitative differences in the physiological response to internal vs. external loads. Alternatively, it may indicate that the decrease in ventilation reflects a decrease in ventilatory drive at alpha-to-theta transitions, either as a consequence of a loss of wakefulness input or as a reduction in the threshold or slope of ventilatory responses to chemical stimuli. According to this view, TI would not increase after the transition, because ventilation would be appropriate to the level of drive.

In assessing the findings of this experiment, several methodological issues must be considered. The state instability which characterizes sleep onset (phases 2 and 3) is associated with substantial fluctuations in respiratory activity (22), a pattern reflected in the present data (see Table 1). Thus, transitions from alpha to theta are associated with hypoventilation and increased upper airway resistance in theta, whereas theta-to-alpha transitions are associated with hyperventilation and normal waking upper airway resistance during alpha (12, 22). Furthermore, the ventilatory instability results in fluctuations in chemical drive (7, 22). It is possible that the effect of a load might differ as a function of the conditions present at load application. It was for this reason that the effect of the experimental load was assessed by comparison with a control condition so that respiratory drive, both central and chemical, would have been equivalent in the loaded and control conditions. Thus, within a state, the comparison between a load and control would reflect the presence vs. absence of reflex load compensation. However, it remains possible that the magnitude of effects between states and phases may have been modified by preload differences in ventilation and chemical drive. Nevertheless, it should be noted that the relevant comparisons in this study were between the load and control conditions, and thus the general thrust of the results remains valid.

The response to an added respiratory load may also be different if presented at different lung volumes. During stable stage 2 sleep, functional residual capacity (FRC) has been found to decrease by ~200 ml (12). It is theoretically possible that the reduced ventilatory response to the added respiratory load during sustained theta and after phase 3 transitions may be affected by the lower FRC observed during sleep. However, VT has been shown to be maintained over a wide range of increased FRC in sleep (1). It is therefore unlikely that our results were caused by changes in FRC associated with sleep.

As noted in METHODS, it was not always possible to apply the experimental load on the first theta breath in alpha-to-theta transition trials. Although most of the transition trials were obtained from loads applied on the first or second theta breaths, several unloaded theta breaths would still have been included in the analysis of loaded trials. The addition of unloaded trials would have dampened the effects of loading on the first two breaths in the series. Pure loaded trials would probably have resulted in a steeper reduction in ventilation on initial loaded breaths.

Although compensation in response to the external resistive load was impaired relatively early during sleep onset, this impairment may not be indicative of a complete loss of compensation. To be certain of a complete loss of neural compensation, it would have been necessary to measure the inspiratory muscle response to the external load, so that the complete absence of a compensatory response would be indicated by the absence of an increase in inspiratory EMG activity in response to the load. However, the results suggest that the loss of compensation was largely complete, because the decrease in ventilation in response to the addition of an external load of similar magnitude to known physiological loads had a similar effect on ventilation.

The rapid changes in ventilation and airway resistance, and lack of compensation observed during periods of theta in sleep onset, are potentially significant to the development of respiratory instability during sleep. Pathological respiratory events, such as apneas, hypopneas, and periodic breathing, have been found to be positively related to the decrease in ventilation at alpha-to-theta transitions (21). The lack of adequate reflexive load compensation mechanisms during sleep onset permits increases in airway resistance to adversely affect ventilation and may therefore play a significant role in the development of these pathological respiratory events.

In conclusion, the effectiveness of the reflex compensation response to resistive loads is critically dependent on sleep-wake state. Thus, with the exception of a brief period after the loss of alpha activity early in sleep onset, the reflex response is present during the EEG alpha state but not during theta EEG activity. Nevertheless, because compensation is present briefly in phase 2 when ventilation decreases abruptly, the decrease in ventilation at this time may be primarily caused by withdrawal of drive to the inspiratory pump muscles, whereas the decrease in ventilation at alpha-to-theta transitions later in sleep onset appears to be caused additionally by the increase in airway resistance and lack of immediate load compensation. It is suggested that this lack of compensation, in association with resistance and ventilation changes observed during sleep onset, is critical to the development of respiratory instability during sleep.