INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Obstructive sleep apnea syndrome (OSA) is a common and debilitating disorder characterized by sleep-induced upper airway collapse. A combination of an anatomically small airway and loss of upper airway dilator muscle activity during sleep are likely the principal causes of OSA (1). The genioglossus muscle (GG) is phasically active with the breathing cycle, with initiation of activity immediately before inspiratory airflow, peak activity in early to midinspiration, and lowest ("tonic") activity during expiration (2). This spontaneous phasic GG activity emanates from the brainstem respiratory pattern generator and serves to dilate or stabilize the upper airway in preparation for and throughout airway negative pressure changes generated by the respiratory pump muscles (1). This spontaneous phasic GG activity is modulated by reflex mechanoreceptive influences. For example, reflex GG activation has been demonstrated in awake humans during imposed upper airway negative pressure (3, 4). Furthermore, we have shown that both upper airway anesthesia and breathing via a tracheal stoma (bypassing the upper airway) substantially reduce GG activity during spontaneous respiration, suggesting that upper airway mucosal receptors are involved in GG activation during spontaneous breathing in humans (5). Thus, such receptors may play a role in the maintenance of adequate pharyngeal patency. The reflex GG activation to negative pressure also may drive the augmented upper airway dilator muscle activity that has been observed in patients with apnea during wakefulness, thereby serving to compensate for the diminished upper airway patency in these individuals (8). This negative pressure reflex is greatly diminished during sleep (4, 9). Hence, loss of the reflex during non-rapid eye movement (NREM) sleep could help explain the diminished muscle activity that leads to obstructive apneas.

In addition to mechanoreceptive influences on GG activity, it is also well established that GG activity is modulated by changes in arterial blood gases, with increased activity at times when respiratory drive is increased (e.g., References 10-12). Hypercapnia and hypoxia during an obstructive apnea increase GG activity and may help to reopen the airway to terminate the apnea (13). On the other hand, the hyperventilation in association with arousal at the termination of an apnea can lead to hypocapnia that may decrease GG activity, possibly predisposing the airway to collapse and a subsequent apnea. Thus, it is important to understand the interaction between mechanoreceptive and chemoreceptive influences on GG activity.

It seems likely that the magnitude of the GG reflex to negative pressure would be influenced by increased respiratory drive as hypercapnia and hypoxia likely cause GG motoneurons to be closer to their firing threshold, such that the same mechanical stimulus would produce an increased GG response. Similarly, it is likely that under conditions of reduced respiratory drive (hypocapnia) GG motoneurons would be relatively hyperpolarized, such that a standard mechanical stimulus would produce a decreased GG response. Thus, the current study was designed to assess in awake humans the effect of arterial blood gases, in the range encountered during mild obstructive apnea-arousal cycles, on phasic GG activity and the magnitude of the GG response to a standard negative pressure stimulus.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Subjects

The protocol was approved by the Human Subjects Committee at Brigham and Women's Hospital. All subjects provided written informed consent before participation in this study. None of the subjects had symptoms of neurological, cardiovascular, pulmonary, or sleep-related breathing disorders or reported snoring. We did not perform overnight polysomnography to rule out obstructive sleep apnea syndrome as none of the subjects had symptoms suggestive of obstructive sleep apnea and none were obese. We studied 20 healthy adult volunteers, with adequate data being collected from 18 subjects (10 men, 8 women; age range, 23 to 38 yr; body mass index [BMI] range, 19.0 to 25.6). Women were studied between Days 5 and 11 of their menstrual cycle.

Measurements and Recording

To assess the function of a representative upper airway dilator muscle, the activity of the GG electromyogram (EMG) was measured with two 36-gauge Teflon-coated stainless steel intramuscular wires. Each wire was passed through a 25-gauge needle. Each needle was inserted into the floor of the mouth at a location 3-5 mm on either side of the frenulum and 15-20 mm into the body of the GG near its insertion in the mandible. After insertion, the needles were extracted, leaving the intramuscular electrodes in place. These wires were referred to a ground electrode on the forehead. The EMG signal was amplified (model 7P122G; Grass, W. Warwick, RI; filter characteristics, 50 Hz-5 kHz), rectified, and "integrated" on a moving time average basis with a 100-ms time constant (model MA-821-4; CWE, Ardmore, PA). Since the baseline level of EMG can be different between individuals, the GG data were normalized to the percentage of the maximal activation observed during three maximal maneuvers performed at the beginning of the study: inspiratory negative pressure generated against an occluded nose mask, lingual protrusive force against upper incisor teeth, and a voluntary swallow. To ensure relative passivity of ventilation during the negative pressure ventilator condition, diaphragmatic electromyogram (DIA EMG) was obtained from a pair of surface electrodes that were placed in the right sixth, seventh, or eighth intercostal spaces adjacent to the costal margin. The pair of electrodes that gave the best phasic inspiratory signal during voluntary inspiration was chosen in each subject before data collection and this pair was used throughout the experiment. Airway pressures were recorded at the level of the choanae (Pchoa) and in the hypoglossal airspace at the level of the epiglottis (Pepi) (transducer-tipped catheters; Millar, Houston, TX). Before insertion of the catheters into the nose, one nostril was decongested (two or three inhalations of 0.05% oxymetazoline hydrochloride [Afrin]) and lightly anesthetized (approximately 1 ml of 4% lidocaine hydrochloride topical spray). After placement, both catheters were taped to the nose to ensure stability. In almost all subjects there was occasional loss of an upper airway pressure signal during the study, likely due to build-up of secretions on a catheter. When this occurred the catheters were repositioned, or removed, cleaned, and reinserted. Subjects breathed through a sealed nose mask (Healthdyne Technologies, Marietta, GA; dead space approximately 50 ml). To detect expiratory leaks, CO₂ was monitored around the periphery of the mask. The subjects also were monitored via closed-circuit television using a low-light camera (model WV-CU-101; Panasonic, Secaucus, NJ) to ensure that the mouth was closed and subjects breathed through the nose during recordings. Inspiratory flow was measured with a pneumotachometer (No. 2; Fleisch, Lausanne, Switzerland) and pressure transducer (± 2 cm H₂O differential amplifier; Validyne, Northridge, CA). End-tidal PCO₂ (PET_CO2) was measured as an estimate of arterial PCO₂ from expired air sampled within the mask with a calibrated infrared analyzer and arterial oxygen saturation (Sa_O2) was measured with a pulse oximeter attached to an earlobe (capnograph/oximeter monitor; BCI, Waukesha, WI). To verify that subjects remained awake, they were instrumented with two channels of electroencephalography (EEG; C4/A1 and C4/O2) and one channel of electrooculography (EOG; LOC/ROC).

Signals were recorded on a 16-channel polygraph (Grass model 78E; paper speed, 10 mm/s). Airway pressures and flow plus the rectified, integrated GG EMG were also digitized at 200 Hz for assessment of phasic GG activity and at 1,000 Hz for the assessment of the magnitude of the GG reflex response to a pressure pulse using signal averaging software (Sigavg; CED 1401 digitizer; Cambridge Electronic Design, Cambridge, UK).

Protocol

Throughout the study, subjects lay supine, with their bodies within and their heads outside a negative pressure ventilator (Iron Lung, series J; Emerson, Cambridge, MA). Subjects were studied while awake with their eyes open throughout five "steady-state" arterial blood gas conditions. Each condition lasted 15-20 min and was presented in randomized order among subjects. The blood gas conditions were selected to represent typical changes in arterial blood gases encountered during mild obstructive apnea-arousal cycles and were as follows: (1) baseline; (2) hypercapnia; (3) hypocapnia; (4) hypoxia; and (5) hypercapnia combined with hypoxia. The mean PET_CO2 and Sa_O2 under the five conditions are displayed in Table 1. PET_CO2 and Sa_O2 were controlled by manually altering the inspired gas mixtures via four rotameters from calibrated gas sources including air; 10% O₂, balance N₂; 100% O₂; and 25% CO₂, 21% O₂, balance N₂. The ventilator was switched on only during the hypocapnic condition. A nylon wrap was comfortably secured around the neck of the subject to minimize leak from the ventilator while avoiding uncomfortable pressure on the neck. When the ventilator was switched on it produced an approximate sine wave of pressure around the subject's body (excluding the head). To enable passive mechanical ventilation, the frequency of ventilation was set at a rate that was slightly higher than the subject's own breathing frequency, and hyperventilation was brought about by altering the peak negative pressure in the ventilator. Depending on the subject's respiratory system compliance and the degree of leak around the neck seal, peak negative pressures of 15 to 20 cm H₂O were required to produce tidal volumes of approximately 1 L along with pressure swings of approximately 5 cm H₂O in the epiglottic region of the airway. Most subjects required some initial coaching to enable passive mechanical ventilation as judged from the consistent timing and shapes of the pressure and flow traces, as well as reduced phasic DIA EMG (in the subset of subjects in whom a detectable phasic DIA EMG was evident from the surface electrodes during the baseline condition).

Spontaneous phasic GG activity was recorded for approximately 5 min under each condition. Thereafter, under each condition, the magnitude of the GG reflex to negative pressure was assessed by the application of negative upper airway pressure pulses having characteristics that have previously been shown to reliably activate the GG in healthy awake subjects (4, 5, 14). Each stimulus was applied in early inspiration by activation of a solenoid valve (MarcValve Corporation, Tewksbury, MA). When activated, this valve connected the inspiratory tubing to a vacuum source and resulted in a brief pulse of negative pressure to the upper airway. The time from onset of pressure generation to maximal pressure was approximately 80 ms. Thereafter the pressure decayed exponentially to baseline within 350 ms. The target pressure for each pressure pulse was 12 cm H₂O at the level of the choanae under each condition. Between 30 and 50 negative pressure pulses were applied at irregular intervals (every 2-7 breaths) to each subject under each of the five conditions.

Data Analysis

All unusual breaths such as swallows and coughs were excluded from analysis. Time-appropriate signal-averaged waveforms were produced for each variable for each individual under each condition (Sigavg software; Cambridge Electronic Design). For assessment of spontaneous phasic respiratory activity, time "zero" was considered to be the onset of inspiratory airflow and these data were used to quantify mean PET_CO2, Sa_O2, tidal volume, breath duration, pharyngeal resistance [(Pchoa  Pepi)/flow; calculated at an inspiratory flow of 0.2 L · s¹], nasal resistance [(Pmask  Pchoa)/flow; calculated at 0.2 L · s¹], and various measures of GG EMG activity calculated from the rectified and integrated GG EMG and expressed as a percentage of maximal contraction. These GG EMG measurements included the tonic activity (minimum EMG during expiration), peak phasic activity (absolute peak EMG during inspiration), and phasic activity (peak EMG expressed as a percentage change from tonic EMG). DIA EMG was used only for coaching of passivity under the iron lung (hypocapnic) condition, and was not quantified (due to the existence of an EKG artifact on the recordings).

For the pressure pulse periods, time zero was considered to be the onset of the negative pressure stimulus at the choanae. After the onset of the stimulus, the time to the first detectable change in GG EMG and the time to the maximal GG EMG response were both used as measures of response latency. From each subject's normalized data the peak GG EMG within 250 ms of the onset of the stimulus were measured. Only changes within 250 ms of the stimulus onset were analyzed in order to avoid volitional responses to the stimulus. For meaningful comparisons of the GG reflex between conditions it was necessary to compare both the absolute change in GG EMG (peak EMG  EMG before stimulus, expressed in units of percent maximal contraction) and the percentage change {100 × [(peak EMG  EMG before stimulus)/EMG before stimulus]}, as previously described (5). The difference in pressure between the choanae and the epiglottis during the pressure pulse was quantified in each state as an index of airway collapsibility, as previously described (4). This was accomplished by subtracting the peak epiglottic negative pressure from the peak choanal negative pressure on the signal-averaged waveform.

For each variable, comparisons were made between baseline (eucapnia and euoxia) and the other four conditions by applying paired t tests to the individual's means within each condition (two-tailed,  = 0.05). In addition, the relationship between epiglottic pressure and GG EMG was assessed among the five conditions by using the Pearson correlation coefficient.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Sufficient data were collected from all 18 subjects under baseline, hypercapnic, and hypocapnic conditions, from 15 subjects under hypoxia, and from 14 subjects under the combined hypoxic and hypercapnic conditions.

Effect of Arterial Blood Gas Changes on GG EMG

The mean results for the group for all variables, as well as the results of the statistical comparisons, are shown in Table 2. During baseline breathing, the mean tonic (expiratory) GG EMG was 4.3% maximum and increased to a peak of 6.6% maximum during inspiration (i.e., a 60% increase above tonic level). Under each of the other four conditions (hypercapnia, hypoxia, hypocapnia, and combined hypoxia plus hypercapnia), tidal volume, ventilation, peak inspiratory flow, peak negative Pepi, and peak negative Pchoa were all significantly increased relative to the baseline condition, with the greatest values occuring under the hypoxic plus hypercapnic condition. Along with these significant increases in respiratory variables relative to baseline, there was a significant increase in phasic GG EMG, again with the greatest values occurring during the combined hypoxic plus hypercapnic condition. Despite these increases in phasic upper airway dilator muscle activity along with increased ventilation, there were also increases in both pharyngeal and nasal resistance (most comparisons were significant). There was no significant change in mean breath duration between conditions.

Traces of the raw signals from one subject are shown in Figure 1. It can be seen that GG EMG increased as Pepi became more negative during baseline spontaneous breathing, during passive mechanical ventilation, (with reduced DIA EMG), and during stimulated breathing (with increased DIA EMG). Mean Pepi and GG EMG results from another typical subject are displayed in Figure 2. It can be seen from the shapes and amplitudes of the Pepi and GG EMG traces in Figure 2 that there is a robust inverse relationship between the magnitudes of Pepi and GG EMG. This relationship was consistent within each condition (notice mirror images of shapes of Pepi and GG EMG within each condition). In addition, this inverse relationship between Pepi and GG EMG was relatively stable across conditions within this subject (Figure 2) and for the group as a whole (Figure 3). In Figure 3 peak negative Pepi is plotted against the peak GG EMG and there is an obvious relationship (Pearson r = 0.87; n = 5; p = 0.06). This correlation was even more robust for peak negative Pchoa versus peak GG EMG (Pearson r = 0.97, n = 5; p < 0.01). Thus, there was a robust inverse relationship between upper airway pressure and GG EMG throughout conditions including baseline spontaneous breathing, breathing stimulated by chemoreceptors, and even passive ventilation.

View larger version (28K): [in this window] [in a new window]   Figure 1.   Individual raw data traces for one typical subject under three of the five conditions: Baseline (left), hypocapnia (middle), and hypercapnia combined with hypoxia (right). Shown from top to bottom in each panel are diaphragmatic EMG (raw signal, arbitrary units; note EKG artifact), GG EMG (raw signal, arbitrary units), moving time average (MTA) of rectified GG EMG (arbitrary units, with a downward deflection indicating increased GG EMG), epiglottic pressure (Pepi, with an upward deflection indicating negative pressure), and inspiratory airflow (expiratory airflow was not recorded). Note that, for ease of comparison among conditions, the scales are the same in each panel even for the variables reported in arbitrary units. Under the baseline condition (left) there was phasic DIA EMG and phasic GG EMG. Under the hypocapnic condition (middle) there were relatively large phasic epiglottic pressure swings and phasic GG EMG despite diminished DIA EMG. Under the hypercapnic plus hypoxic condition (right) there was increased phasic activity in both DIA EMG and GG EMG along with large phasic epiglottic pressure swings. Note that, under this condition, the peak inspiratory airflow signal was "off scale" on the chart recorder (but was not off scale on the computer recording).

View larger version (20K): [in this window] [in a new window]   Figure 2.   Effect of arterial blood gas changes on GG EMG and Pepi in a typical subject. A single subject's integrated GG EMG and Pepi signal-averaged waveforms under five conditions of altered arterial blood gases are presented. There were between 30 and 50 breaths averaged under each condition. The GG EMG is normalized to the percentage of maximal EMG. The mean tidal volume under each condition is indicated in each Pepi plot. The shapes and amplitudes of the Pepi and GG EMG traces suggest that there is a robust inverse relationship between the magnitudes of Pepi and GG EMG under each condition and across conditions within this subject.

View larger version (15K): [in this window] [in a new window]   Figure 3.   Relationship between peak upper airway pressures and peak GG EMG among arterial blood gas conditions. Peak negative Pepi and peak negative Pchoa are both plotted against peak GG EMG. Group mean data are shown under the five blood gas conditions. The Pearson correlation coefficients and probabilities are shown. The data indicate a robust inverse relationship between upper airway pressure and GG EMG among conditions including baseline spontaneous breathing, chemoreceptor-stimulated breathing, and even passive mechanical ventilation. (For Pepi the points from left to right represent the following conditions: spontaneous breathing; low O2; high CO2; high CO2 plus low O2; and low CO2. For Pchoa the points from left to right represent the following conditions: spontaneous breathing; low O2; high CO2; low CO2; and high CO2 plus low O2.)

Overall there was a trend toward an inverse relationship between supraglottic resistance and the degree of phasic GG activity although this relationship did not reach significance for the group (nasal resistance versus phasic GG EMG, Pearson r = 0.21, n = 5; p = 0.74; pharyngeal resistance versus phasic GG EMG, Pearson r = 0.78, n = 5; p = 0.12).

Effect of Arterial Blood Gas Changes on GG Reflex to Negative Airway Pressure and Airway Collapsibility

The group mean integrated GG EMG and choanal pressure waveforms under each of the five conditions are shown in Figure 4 and the derived mean data are presented in Table 3. During baseline, 12 cm H₂O in the upper airway caused a rapid doubling in GG activity (mean response, 108% increase). (Note: Small differences occur between the values in Table 3 and the impression from Figure 4 because Table 3 was derived from analysis of individual subject waveforms, whereas the group average waveform loses definition as peaks and troughs occur at slightly different times between individuals.) The magnitude of this reflex to virtually identical pressure stimuli was not significantly increased under the stimulated breathing conditions (hypercapnia, 81%; hypoxia, 98%; combined hypoxia plus hypercapnia, 101%), and this reflex was not significantly decreased under the hypocapnic condition (130%). There were no significant differences between baseline and the other four conditions when the size of the reflex was expressed either as the percentage change or as the absolute change in GG EMG from the level that immediately preceded pressure pulse application (Table 3). In addition, there were no differences between baseline and the other four conditions in the latency of the GG EMG response from the onset of the stimulus to the first GG EMG change or to the maximal change in GG EMG, with the single exception that the initial GG EMG latency was increased from 33 to 49 ms under the combined hypoxic plus hypercapnic condition. Finally, our index of airway collapsibility (the difference between Pepi and Pchoa during the pressure pulse) also indicated no systematic changes between baseline and three of the four conditions. The exception was slightly reduced collapsibility under the hypercapnic condition.

View larger version (19K): [in this window] [in a new window]   Figure 4.   Effect of arterial blood gas changes on GG reflex to negative airway pressure. Shown is the group mean integrated Pchoa and GG EMG waveforms under baseline spontaneous breathing and the four other blood gas conditions. During each condition the GG EMG signal is normalized to the activity immediately preceding the pressure stimulus during baseline spontaneous breathing (100%). The vertical dotted line in each plot represents the onset of the pressure stimulus in the choanae. It can be seen that during baseline spontaneous breathing there was a brisk GG EMG activation induced by negative pressure. Virtually identical stimuli were presented under each condition. There were no significant differences between baseline and the other four conditions when the size of the reflex was expressed either as the percentage change or as the absolute change in GG EMG from the level that immediately preceded pressure pulse application (also see Table 3). Note that small differences occur between the values in Table 3 and the impression from Figure 4 because Table 3 was derived from the analysis of individual subject waveforms, whereas the group average waveform (Figure 4) loses definition as peaks and troughs occur at slightly different times among individuals.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

This study was designed to assess in awake healthy humans the effect of changing arterial blood gases, in the range encountered during mild obstructive apnea-arousal cycles, on the phasic GG activity and on the magnitude of the GG response to a standard negative pressure stimulus. Our results demonstrate a robust inverse relationship between upper airway pressure and GG EMG. This relationship has previously been demonstrated for rapid pulses of pressure but is now extended to the slow, phasic changes in pressure occurring within breaths and among conditions that included baseline spontaneous breathing, chemoreceptor-stimulated breathing, and even passive negative pressure ventilation. In addition, we found that GG activation in response to rapid pulses of upper airway negative pressure was not related to the magnitude of the respiratory drive (as assessed by the level of ventilation and DIA EMG). This result implies that this protective mechanoreceptive reflex would not strengthen throughout an obstructive apnea as arterial blood gases worsen. This result also indicates that the protective mechanoreceptive reflex would not be lost during the hypocapnia occurring between apnea cycles. Finally, our results suggest that mechanoreceptive control of GG EMG can fully explain all changes in GG activity in the range of blood gases seen in mild obstructive apnea- arousal cycles. Thus, one interpretation of our findings is that chemoreceptive control of GG activity is minimal or absent within the arterial blood gas ranges studied. An alternative interpretation is that chemoreceptive influences on GG activity are present but under these conditions they do not summate or facilitate the mechanoreceptive control of GG activity, suggesting redundancy among mechano- and chemoreceptive mechanisms in the control of upper airway muscles.

Effect of Arterial Blood Gas Changes on Phasic GG Activity

A principal finding was the relationship between phasic changes in upper airway pressure and GG EMG within breaths and among all conditions. This result presents a number of possible interpretations. First, we acknowledge that it is likely that chemoreceptors increase respiratory drive leading directly to recruitment of upper airway dilator muscle activity. However, our finding of increased phasic GG activity during passive mechanical ventilation suggests that the same GG activity can occur when only mechanoreceptive stimuli are present. Thus, it is plausible that chemoreceptors drive phasic GG activity indirectly via upper airway negative pressure swings. These direct and indirect pathways are not mutually exclusive.

Effect of Arterial Blood Gas Changes on GG Reflex to Negative Airway Pressure

We found no significant change in GG activation in response to a standard upper airway negative pressure stimulus during arterial blood gas changes encompassing those experienced during mild obstructive apneas, and involving more than a doubling of ventilation. We had reasoned that under conditions of increased respiratory drive (hypercapnia, hypoxia, and combined hypercapnia plus hypoxia) it is likely that more GG motoneurons would be closer to their threshold for firing, such that the same mechanical stimulus would produce an increased GG response. Similarly, under conditions of reduced respiratory drive (hypocapnia produced via passive mechanical ventilation) it appeared likely that GG motoneurons would be relatively hyperpolarized such that the same mechanical stimulus would produce a decreased GG response. Yet our results indicate that the GG reflex was greatest under the hypocapnic condition (although there was no significant difference between hypocapnia and baseline). Other workers have found in tracheotomized, anesthetized cats that hypoxia (Pa_O2 = 34 mm Hg) accentuates the GG response to upper airway occlusion (11) and that hypercapnia increases the response of the GG to upper airway negative pressure only when PET_CO2 rises above ~ 57 mm Hg (12). The different results of the current study and the previous study by Gauda and coworkers (12) may lie in the possibility that awake humans behave differently from tracheotomized, anesthetized cats, or that in our study, the hypoxic and hypercapnic conditions were below the chemoreceptor thresholds at which such an interaction between mechano- and chemoreceptor effects occurs.

In an analogous way, it has been documented that the amplitude of the human soleus H reflex can be modulated while running when compared with walking (analogous to stimulated versus baseline breathing), and that this modulation can be different during the stance phase versus the swing phase of the gait (analogous to expiration versus inspiration) (15). In addition, Tantucci and coworkers (16) found that the GG response to negative pressure could be reliably elicited during early expiration but not during late expiration. Thus, it seems possible that any mechanoreceptive interaction with chemoreceptive influences on GG activity may be nonlinear (involving chemoreceptive thresholds) and this interaction may also be different at different phases of the respiratory cycle. We have studied this mechanical stimulus during mild changes in arterial blood gases and only during early inspiration. Nonetheless, we believe that these stimuli as we presented them are the most physiologically relevant since the levels of arterial blood gases encompass those changes occurring throughout a mild obstructive apnea-arousal cycle, and the pressure stimulus on inspiration matches when negative pressure naturally occurs within the upper airway during both normal and obstructed breathing.

Effect of Arterial Blood Gas Changes on Pharyngeal Resistance and Airway Collapsibility

We found that the conditions of stimulated breathing (hypercapnia, hypoxia, and combined hypercapnia plus hypoxia) were associated with significant increases from baseline in supraglottic resistance (i.e., nasal resistance, pharyngeal resistance, or both resistances) (Table 2). Furthermore, supraglottic resistance also tended to increase during passive hyperpnea in the hypocapnic condition (not significant). Overall there was a trend toward an inverse relationship between supraglottic resistance and the degree of phasic GG activity although this relationship did not reach significance for the group (see RESULTS). Leiter and coworkers (17) also found increases in nasal resistance, with virtually no change in pharyngeal resistance during hypercapnic stimulated breathing and in association with increases in GG activity. These authors suggested that GG activity increased to stabilize pharyngeal resistance under conditions of high nasal resistance. (The reason for the increased nasal resistance during stimulated breathing is unknown, but could involve turbulent airflow, or increases in blood volume in the tissues surrounding the airway.) Certainly in our data nasal resistance generally increased more than pharyngeal resistance. However, we did find significant increases in pharyngeal resistance during hypercapnia and combined hypercapnia plus hypoxia, suggesting that this proposed compensatory mechanism to stabilize pharyngeal resistance was not completely effective at high ventilatory drives under the conditions of our experiment.

There was little change in the degree of upper airway collapse (i.e., presumed narrowing of the airway) during application of the pressure stimulus among these conditions, with the exception of slightly decreased collapsibility under the hypercapnic condition. Similar to our findings, other workers have found that hypercapnia decreases upper airway collapsibility during applied negative pressure in cats (12). However, since changes in the current study were relatively small and did not persist under either the hypoxic condition or the combined hypoxic plus hypercapnic conditions, we believe that this isolated finding in hypercapnia is likely a chance observation (Type I statistical error). An alternative interpretation is that hypoxia causes an increase in collapsibility while hypercapnia causes a decrease in collapsibility. We believe that this latter interpretation is unlikely because hypoxia clearly stimulates the upper airway dilator muscles (e.g., current study; and Refs. 10 and 11) and we found no increase in collapsibility during hypocapnia in the current study.

Potential Weaknesses of Study

Our data must be viewed in the context of several caveats to our interpretations. First, we assessed the responsiveness of only one of the many upper airway muscles. Mechano- and chemoreceptive related alterations in the responsiveness of palatal muscles, or of the whole ensemble of upper airway muscles, may prove to be more relevant to the pathogenesis of OSA than assessment of GG activity alone. Second, the applied mechano- and chemoreceptive stimuli may not ideally emulate normal physiological or pathophysiological stimuli. For instance, we chose to study steady-state arterial blood gas conditions as these are easier to control and interpret than the dynamic changes that occur throughout an apnea. Third, we did not assess diaphragmatic EMG in all subjects so we cannot be certain that the respiratory pump muscles were totally quiescent under the iron lung ventilation condition. Each subject was coached to become passive at the beginning stages of this iron lung ventilation, based on breath-to-breath uniformity of airflow and airway pressure profiles. Although all subjects managed a high degree of uniformity in these variables, and despite the induced hypocapnia, it remains possible that the respiratory pump muscles were entrained to the ventilator in the same way that GG activity became entrained. Such entrainment could occur via a mechanoreceptive reflex mechanism as we have proposed, but a remote possibility exists that subjects voluntarily entrained their pump and airway dilator muscle activity. Future similar studies during sleep would likely resolve this matter. A fourth concern is that the responses were quite variable among subjects leading to relatively low statistical power, despite studying 20 subjects (with sufficient data collected from 18 subjects). For example, the mean increase in GG reflex between baseline and the hypocapnic condition was 0.33% of maximal GG activity, and the standard deviation of this difference was 3.87% of maximal GG activity. Given this variability among subjects, we have calculated that it would require studying 1,082 subjects to find the same observed mean difference to be significant at the 0.05 probability level (two-tailed test), using a statistical power of 80%. Similar calculations for the other conditions in comparison with baseline indicate the need to study between 166 and 1,512 subjects to find the same observed mean differences to be significant. Thus, while there is clearly the possibility of a Type II statistical error, the results indicate that any real group mean physiological effect is either small or nonexistent.

A final caveat to our interpretations must concern the relevance of this study to obstructive sleep apnea syndrome. We studied the mechanoreceptive reflex in healthy awake subjects rather than in sleeping apneic patients. Certainly there may be differences brought about by the disease or the sleep state. For instance, awake patients with OSA have relatively increased GG activity, possibly driven by the negative pressure reflex to counteract the effects of an innately smaller upper airway (8). Also, sleep affects the reflex activation of upper airway muscles (4, 9, 14) as well as the level of arterial blood gases and the ventilatory chemosensitivity (18). Thus, further studies of the interaction between mechano- and chemoreceptive influences on upper airway muscles in sleeping apneic patients would be worthwhile.

Summary

We believe that understanding the changes in upper airway reflexes under conditions of altered arterial blood gases is germane to understanding the mechanisms that initiate and terminate obstructive apneas. We found in healthy awake humans that changes in arterial blood gases that span those encountered in mild obstructive apnea-arousal cycles do not affect the GG reflex to negative pressure, that phasic GG EMG increases as ventilation increases, and that these changes can be fully explained by mechanoreceptive control of GG EMG. This result implies that this protective mechanoreceptive reflex would not strengthen throughout an obstructive apnea as arterial blood gases worsen. This result also indicates that the protective mechanoreceptive reflex would not be lost during the hypocapnia occurring between apnea cycles. Finally, our results suggest that mechanoreceptive control of GG EMG can fully explain all changes in GG activity in the range of blood gases seen in mild obstructive apnea-arousal cycles. One interpretation of our findings is that chemoreceptive control of GG activity is minimal or absent within the arterial blood gas ranges studied. An alternative interpretation is that chemoreceptive influences on GG activity are present but under these conditions they do not summate or facilitate the mechanoreceptive control of GG activity, suggesting redundancy among mechano- and chemoreceptive mechanisms in the control of upper airway muscles.