INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Keywords: upper airway; apnea; breath; negative pressure; genioglossus

The mechanisms controlling genioglossal (GG) activation are clearly important in understanding the pathogenesis of obstructive sleep apnea syndrome (OSA) (1). The prevailing hypothesis regarding the control of pharyngeal dilator muscles suggests that input from both a central pattern generator in the brainstem as well as feedback from local phenomena in the upper airway influence muscle activity. The presence of "preactivation" (hypoglossal nerve firing 50-100 ms prior to phrenic) supports the presence of premotor inputs to the hypoglossal motor nucleus in the medulla (6). In addition, substantial evidence supports a contribution of local upper airway mechanisms in modulating pharyngeal dilator muscle activation, although the precise stimulus and receptors involved remain unclear. We have recently reported an iron lung ventilation model in humans that can substantially attenuate or eliminate the preactivation of pharyngeal dilator muscles. Therefore, this model can serve as a useful tool to investigate the relative influences of intrapharyngeal stimuli while minimizing the influences of central pattern generator activity through passivity and/or entrainment.

Negative (suction, subatmospheric) pressure is considered the most likely stimulus to phasic pharyngeal dilator activation although most studies assessed this using nonphysiological pulses of negative pressure (7). However, other stimuli may also be important (15). In theory, pharyngeal airflow could also be a stimulus to pharyngeal dilator muscle activation based on a number of previous observations. First, the demonstration that airflow can influence ventilation/ventilatory control suggests that flow receptors are indeed present and could potentially modulate upper airway dilator muscle activity (20, 21). Second, the greater strength of the negative pressure reflex observed when negative pressure was applied with the glottis open as compared with the glottis closed in previous studies could be explained by increased flow developing in the former condition. Third, inspiratory resistive loading has previously been demonstrated to have a rather modest effect on pharyngeal dilator muscle activation during wakefulness, despite substantial intrapharyngeal negative pressure (22). The diminished inspiratory flow associated with inspiratory loading could be one explanation for the minimal pharyngeal dilator activation. To dissociate the relative influences of airway pressure and airflow, different gas densities (Heliox versus air) can be employed.

Purpose

By using variable gas densities during spontaneous and iron lung ventilation, we sought to determine if negative pressure (and/or flow, resistance, CO₂) is the principal stimulus to pharyngeal dilator muscle activation on a within-breath and between-breath basis. By using the iron lung ventilation model, we attempted to assess these local stimuli while minimizing central pattern generator influences.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Subjects

We enrolled 18 normal volunteers (11 males) who were historically healthy and had no sleep complaints. Each underwent a history and physical examination by one of the investigators. The mean age of the subjects was 26.0 ± 5.1 (SD) yr and mean body mass index was 21.7 ± 1.7 kg/m² (see Table E1 in the online data supplement). Informed consent was obtained from each subject, with the protocol having the prior approval of the Human Subjects Committee of the Brigham and Women's Hospital.

Equipment and Techniques

All studies were performed during wakefulness (eyes open as confirmed by video camera) in the supine posture. Subjects were studied lying in the iron lung, which was turned on only for the iron lung conditions (see below).

Subjects wore a nasal mask (Healthdyne Technologies, Marietta, GA) connected to a two-way valve partitioning inspiration and expiration. Inspiratory flow was determined with a pneumotachometer (Fleish, Inc., Lausanne, Switzerland) and differential pressure transducer (Validyne Corp., Northridge, CA), calibrated with a rotameter. For flow measurements during heliox (80% helium/20% oxygen) conditions a correction quotient of 1.09 was used. This value is based on published gas viscosity correction factors making the assumption that flow through the pneumotachometer is largely laminar (23). Because only inspiratory flow was measured, temperature and humidity values for dry room air gas were used in deriving this value. Subjects were instructed to breathe exclusively through the nose and were carefully monitored by video camera to ensure that the mouth was completely closed. The standard techniques of our laboratory were used to measure end-tidal CO₂ pressure (PET_CO2), mask leak, mask pressure (Pmask), choanal pressure (Pcho), epiglottic pressure (Pepi), and intramuscular genioglossus electromyography (GGEMG, as a percent of maximum activity). All signals were sampled by computer at 125 Hz and demonstrated to be without phase lags at over 50 Hz.

Iron lung ventilationSubjects were studied while supine with the head outside and body within a negative pressure ventilator (Iron Lung; Series J; Emerson, MA). The ventilator was switched on only for specific parts of the experiment (see PROTOCOL below). This device was lightly sealed around the neck with a flexible twisted nylon sheet while an external piston created negative pressure around the chest wall and abdomen thus assisting each breath. The iron lung could be adjusted to achieve the desired upper airway pressure and breathing frequency, so that passivity/ entrainment was achieved. All subjects required some initial coaching to enable entrainment. This involved asking the subjects to remain completely relaxed while the investigators provided feedback on a breath-by-breath basis to achieve consistent timing and shape of the pressure and flow traces. Recordings were stopped when there was departure from this passive pattern until adequate passivity could be achieved or the experiment was terminated. Measurements were recorded only during steady-state conditions.

Heliox administrationA 50-L meteorological balloon was attached to the inspiratory line of the breathing apparatus connected by a three-way stopcock. When breathing room air, the device was open to atmosphere. While breathing heliox, the valve was open to the 50-L balloon that had been previously filled with a mixture of helium (80%) and oxygen (20%). Following each change in inspired gases, 3 to 5 min passed before data collection to allow full equilibration.

Protocol

Each subject reported to the laboratory having fasted for at least 4 h. After obtaining informed consent, the pressure catheters and intramuscular EMG wires were inserted. Subjects then assumed the supine posture in the iron lung, and the nasal mask was attached. Subjects subsequently lay with eyes open in this posture and were allowed to acclimate to the equipment. Determination of maximal EMG was then performed as described by our laboratory previously (1). Subjects were subsequently recorded during both spontaneous breathing and during iron lung breathing. Subjects initially placed in the iron lung were instructed to "relax and let the machine breathe for them." It required up to 10-20 min with steady feedback on a breath-by-breath basis until passivity/entrainment was achieved as evidenced by the visual appearance of "synchronization with the ventilator." The amplitude of the iron lung excursions was adjusted as necessary to achieve synchrony. An additional post hoc test of passivity/entrainment was performed based on the loss of preactivation (i.e., no detectable rise in GG activity prior to the onset of inspiratory airflow). When stable iron lung ventilation was achieved, data were recorded for 5 min. In random order, each subject was recorded under a number of conditions: (1) during spontaneous basal breathing (air and heliox) and (2) during passive/entrained ventilation: eucapnia (both air and heliox) and hypocapnia (PET_CO2 = 32 mm Hg), both air and heliox at negative pressure 1, and hypocapnia (PET_CO2 = 32 mm Hg), negative pressure 2 heliox only. Two pressures were chosen to allow for more range in the assessment of GGEMG versus epiglottic pressure. We did not choose precise pressure goals for either assessment as this is difficult with iron lung ventilation due to variable upper airway resistance. However, we did have a higher and lower pressure for each participant. Carbon dioxide was added to the inspiratory line when necessary to achieve the desired end-tidal CO₂ concentration. The observed PET_CO2 levels were thus a function of both the inspired CO₂ and the respiratory system mechanics of the subject. Therefore, each subject was studied under seven different conditions in random order.

Data Recordings and Analyses

All signals (GGEMG [raw and moving time average], airway pressure [mask, choanal, epiglottic], and inspiratory flow) were recorded on a 16-channel Grass model 78 polygraph (Grass Instruments, Astro-Med, Inc., West Warwick, RI). Certain signals (GGEMG MTA, airway pressures, and inspiratory flow) were also recorded onto computer using signal-averaging software (Spike 2; Cambridge Electronic Design, Ltd, Cambridge, UK). Sampling frequency was 125 Hz.

For each condition, one buffered breath was generated by signal averaging all breaths in that particular condition. This was accomplished by aligning each breath to the onset of inspiration and then averaging the value for each channel. For each buffered breath the following variables were determined: peak negative pressure (at the levels of choanae and epiglottis), peak flow, tonic GGEMG (minimum level of activation during expiration), and peak phasic GGEMG (peak activation during inspiration). Phasic EMG was defined as the difference between peak phasic and tonic EMG. Pharyngeal resistance (Rpha, choanae to epiglottis) was calculated at both peak inspiratory flow and at 0.2 L/s. Supraglottic resistance was also calculated as the sum of nasal and pharyngeal resistance at both peak flow and 0.2 L/s (SGRPF and SGR0.2, respectively). The relationships between inspiratory GGEMG and both epiglottic pressure and flow were determined across the buffered breath (within-breath analysis) as a linear regression. Between-breath analysis was also performed (in addition to within-breath analysis) by examining the relationships between peak genioglossus EMG, peak flow, nadir epiglottic pressure, and pharyngeal resistance at peak flow.

Statistical Analyses

All statistical analyses were performed with commercially available software (Excel 97, Microsoft; and SigmaStat + Sigmaplot, SPSS, Chicago, IL). Standard linear regression analyses were performed to examine the slope and correlation between negative epiglottic pressure and flow and GGEMG on a within-breath basis from the signal averaged breath in each condition. Two-tailed paired t tests were performed to compare both basal values and the slopes of relationships between conditions. Whenever the data were not normally distributed, Wilcoxon signed rank tests were performed. Where specified in the RESULTS, ANOVA was used for repeated measures if normally distributed and ANOVA by ranks (Kruskal-Wallis) if not normally distributed. Results are presented as means ± standard deviations. For all analyses,  was set at 0.05.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Complete data sets were obtained for each of the 18 subjects (see raw data examples, Figure 1). As detailed below, 2 of the 18 subjects (3 and 13) did not lose preactivation during iron lung ventilation, therefore the results of the remaining 16 subjects are presented. Across all conditions, the genioglossus/inspiratory negative pressure (GG/Pepi) relationship within breaths remained robust, with statistically significant R values observed across all conditions (R > 0.7, p < 0.05) (see Figure 2). Furthermore, the slope of this GG/Pepi relationship was not statistically different in any of the conditions, regardless of the induced pressure, flow, CO₂ level, or gas density (one-way ANOVA, p = 0.74).

View larger version (34K): [in this window] [in a new window]   Figure 1.   A figure illustrating the raw signals during spontaneous breathing and during iron lung ventilation. For each tracing, the signals include the flow, mask pressure, epiglottic pressure, and tidal volume. Only inspiration is seen due to the one-way valve present in the breathing apparatus.

View larger version (15K): [in this window] [in a new window]   Figure 2.   A representative plot of genioglossus (GG) versus epiglottic pressure (Pepi) in a single subject during inspiration. The data are from a buffered breath (signal averaged) from a single condition, with only inspiration shown. The extremely tight correlation was observed for the majority of subjects in most conditions.

Spontaneous Breathing

During spontaneous breathing, the epiglottic pressure was significantly less negative during heliox breathing compared with air breathing as expected (1.35 ± 0.32 versus 1.73 ± 0.38 cm H₂O, p < 0.01). Moreover, the peak flow increased from spontaneous air breathing to spontaneous heliox breathing (0.44 ± 0.08 to 0.55 ± 0.12 L/s, p < 0.05). Therefore, heliox did provide the anticipated dissociation of pressure and flow stimuli. The slope of the GG/Pepi (within breath) relationship did not change significantly (1.59 ± 2.09 versus 0.89 ± 3.11 %max/ cm H₂O, p = NS for air versus heliox) between conditions (see Table 1). Therefore, the inspiratory phasic activity of the genioglossus muscle tended to fall (4.85 ± 5.9 versus 4.06 ± 4.4 %maximum, p = 0.13 air versus heliox) with the change in gas density. On the other hand, the GG/inspiratory flow relationship during spontaneous breathing did change with altered gas density. Although, as anticipated, the R value remained robust (R > 0.78 for both), the slope of the GG/flow curve tended to change between air and heliox breathing (7.48 ± 9.33 versus 5.57 ± 6.0 %max/L/s, p = 0.06) (see Table 1).

Iron Lung Ventilation

During iron lung ventilation, we were successful in attenuating the preactivation of the genioglossus muscle, as in previous studies, using this model (10). That is, we did not observe an increase in genioglossus activation prior to the onset of inspiratory flow in the buffered breaths in 16 of 18 subjects (see example, Figure 3). In the remaining two subjects, there was some measurable increase in GGEMG prior to the onset of inspiratory flow. Only the results of the 16 subjects who lost preactivation during iron lung breathing are presented.

View larger version (24K): [in this window] [in a new window]   Figure 3.   A raw data example illustrating the presence and absence of preactivation. During spontaneous breathing, there is a clear increase in the genioglossus muscle activation prior to the onset of inspiratory flow, whereas during iron lung ventilation, there is minimal observed increase in genioglossus activation before the onset of inspiratory flow. The vertical dotted line through flow = 0 represents the onset of inspiratory flow. The peak phasic genioglossal activation is greater in the iron lung condition, because more negative epiglottic pressure was present than during spontaneous breathing.

In addition to the stable correlation between GG and Pepi, the slope of the GG/Pepi relationship remained stable under iron lung ventilation conditions despite altered gas density, PET_CO2 levels, and negative epiglottic pressures (one-way ANOVA, p = NS). Although the correlations between flow and genioglossus activation were also robust, there were clear changes in the slope of this relationship with altered gas density. During eucapnic passive ventilation, while the GG/Pepi slope was unchanged between air and heliox (1.04 ± 1.3 versus 1.14 ± 1.25 %max/cm H₂O, p = 0.9), there was a significant change in the slope of the GG/flow relationship (7.0 ± 7.8 versus 4.8 ± 5.3 %max/L/s, p = 0.01) (see Table 2). Although similar trends were observed during hypocapnic ventilation, the change in slope of the GG/flow relationship failed to reach statistical significance (6.0 ± 7.4 versus 4.3 ± 4.07 %max/L/s, p = NS). Therefore, these data suggest that genioglossus activity tracks pressure more closely than flow, when the two stimuli are dissociated (see Figure 4).

View larger version (37K): [in this window] [in a new window]   Figure 4.   The influence of gas density on the genioglossal/epiglottic pressure (GG/ Pepi) and genioglossal/flow (GG/flow) relationships. All of these data are representative breaths from a single subject, derived from buffered breaths (signal averaged) under various different conditions. The left two graphs are buffered breaths during air breathing, while the right two graphs are during heliox breathing. The top two curves are GG/Pepi curves while the lower two graphs are GG/flow relationships. The figure illustrates that the GG/Pepi curves have identical slopes for both air and heliox breathing for similar levels of stimulus. On the other hand, there is a distinct change in the slope of the GG/flow relationship with the change in gas density.

Influence of Carbon Dioxide

Carbon dioxide level did not have a major independent influence on the inspiratory GG/Pepi relationship. During iron lung ventilation, peak phasic genioglossus activity was not different despite altered CO₂ levels during either air or heliox breathing (for air: 11.8 ± 11.0 versus 12.8 ± 11.7 %maximum hypocapnia versus eucapnia, p = NS; for heliox 11.3 ± 8.9 versus 11.5 ± 9.9 %maximum hypocapnia versus eucapnia, p = NS). In addition, the slope of the GG/Pepi relationship was unchanged despite substantial changes in end tidal CO₂ levels (0.9 ± 1.45 %max/cm H₂O during hypocapnia and 1.04 ± 1.3 %max cm H₂O during eucapnia [p = 0.44]). Similar findings occurred during iron lung ventilation with heliox. Chemistry did not appear to have an independent effect on phasic pharyngeal dilator activation or responsiveness that could not be explained by negative pressure.

Other Analyses

In addition to the variables described above, the correlations between genioglossus activation and resistance, as well as GGEMG versus rate of change of flow, were also examined. The rate of change of flow was calculated as the first derivative of the flow value with respect to time. For both resistance and rate of change of flow, no significant correlations with GG activity were observed, precluding analysis of slopes (p = NS for both). Therefore, genioglossus activity does not appear to respond to either of these variables on a moment-to-moment basis during inspiration within breaths.

Finally, to further assess this relationship between epiglottic pressure and genioglossus activity further, we plotted the group mean average epiglottic pressure versus the group mean average peak genioglossus activity for each of the seven conditions (between-breath analysis, see Figure 5). As the figure illustrates, an extremely robust GG/Pepi relationship was observed for the group averages (R² = 0.98). Other variables such as flow and pharyngeal resistance showed less robust correlations with peak phasic GGEMG for the group mean average analysis. Therefore, these data also support the importance of local factors, negative epiglottic pressure in particular, in determining genioglossus activity across the various experimental conditions.

View larger version (23K): [in this window] [in a new window]   Figure 5.   (a) Group mean epiglottic pressure (Pepi) versus group mean peak genioglossus activation (GGpeak) for all seven conditions. These data were obtained for each of the seven conditions by obtaining a mean value for all of the 18 subjects. The figure illustrates an extremely tight relationship between epiglottic pressure and genioglossus activation for the group mean averages (R2 = 0.98). (b) The group mean relationship between genioglossus peak phasic activity and mean peak flow across the seven conditions. The correlation is less strong than for epiglottic pressure using this same type of analysis. (c) The group mean relationship between genioglossus peak phasic activity and mean pharyngeal resistance across the seven conditions. Again, the correlation was less strong than for epiglottic pressure using the same analysis.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The observations of this study improve our understanding of pharyngeal motor control in normal human subjects. The correlations between epiglottic pressure and genioglossus activation remained robust despite manipulations of gas density, chemistry, central drive, etc. This relationship persisted across the various conditions when analyzed both between breaths and within breaths. The results suggest that independent of central drive, negative pressure is the most crucial stimulus to pharyngeal dilator activation, rather than chemistry, resistance, airflow, or rate of change of airflow. Genioglossus activity therefore appears to be closely regulated by intrapharyngeal pressure on a moment-by-moment basis within inspiration, and similarly changes activity between breaths with alterations in this variable.

The relationship between airflow and GG activity was assessed under a variety of conditions. The observed change in the slope of the GG/flow relationship across these conditions suggests that other variables (likely negative pressure) are more important in the modulation of pharyngeal dilator activity in humans. For example, during spontaneous breathing the reduction in gas density was associated with reduced negative epiglottic pressure (less negative) but increased flow. Despite this observed increase in flow, the phasic genioglossus activity fell, implying that the loss of the negative pharyngeal pressure stimulus was more important in driving pharyngeal dilator muscle activity. Therefore, the flow receptors that have been demonstrated to be of importance in the modulation of ventilatory control are apparently not as directly involved in driving genioglossus activity.

The absence of an independent influence of chemistry on muscle activity is consistent with previous observations (24, 25). Although both hypoxia and hypercapnia have been shown to activate pharyngeal dilator muscles during wakefulness, these stimuli are generally associated with changes in intrapharyngeal pressure due to increases in minute ventilation (24). We have recently reported no significant influence of carbon dioxide levels on pharyngeal dilator activation, independent of intrapharyngeal pressure during wakefulness. Therefore, the results of the present study are consistent with the available literature in this regard.

We failed to show a significant relationship between pharyngeal resistance or rate of change of flow and genioglossal activation. For resistance to be detected or monitored, presumably two pressure receptors would need to be present in the human upper airway (e.g., larynx and/or nose), as well as some ability to measure airflow (or potentially temperature). This is an appealing hypothesis as we have recently observed that substantial manipulations of both pressure and flow in the upper airway have a rather modest influence on pharyngeal resistance, suggesting that resistance is carefully protected (26). However, based on the lack of correlation observed in the present study, pharyngeal resistance itself does not appear to drive pharyngeal dilator activation during inspiration. Rapidly adapting receptors in the upper airway have also been demonstrated in animal experiments and could potentially modulate genioglossus activity. However, again the lack of correlation between the rate of change of airflow and GGEMG implies that the genioglossus is responding to other stimuli.

This study has a number of limitations. First, all subjects may not have become completely passive during iron lung ventilation. Although the absence of pharyngeal dilator activation prior to the onset of airflow (preactivation) in most individuals is reassuring, whether iron lung ventilation completely eliminates brainstem pattern generator activity is unclear. In previous studies using this model, we have observed reduced surface diaphragm EMG activity during passive breathing, also supporting reduced brainstem output during iron lung ventilation. However, we cannot definitely prove passivity in our model and therefore have referred to it as "entrainment." In either case, we believe that premotor output to the genioglossus muscle (and the variability of this output) to be substantially reduced, facilitating assessment of local stimuli on muscle activation. Second, to measure airflow with variable gas densities, we have used the assumption that flow through the pneumotachometer is purely laminar. During laminar flow, gas viscosity is the primary determinant of airflow rather than gas density. Although some turbulence may exist in the Fleisch pneumotachometer within the range of flows measured in the present study, we believe that any errors introduced by this assumption are small and unlikely to influence the conclusions of the present study. Third, to match levels of end-tidal CO₂ across conditions, we administered inhaled CO₂ into the inspiratory line. As a result, the partial pressure of CO₂ present at local laryngeal receptors may have been different between conditions. Thus, although the chemical stimulation of the systemic chemoreceptors (carotid, CNS) should have been the same for equivalent PET_CO2 s, the possibility exists that slight differences in laryngeal stimulation were present (27, 28). We believe that any errors introduced by this assumption would be small and unlikely to influence the conclusions of this study. Furthermore, as stated, PET_CO2 is not a purely independent variable as the observed value is a function of the inspired concentration as well as the extent of pharyngeal collapse. However, minimal pharyngeal collapse occurred during these studies, as evidenced by resistance measurements. Thus, the influence of pharyngeal collapse on ventilation level and thereby PET_CO2 is likely quite small. Finally, as our ultimate objective from these studies is to gain insight into the pathogenesis of obstructive sleep apnea, one could argue that the study of normal subjects during wakefulness across a modest range of pharyngeal pressures says little about disease. However, we believe that a more precise understanding of normal upper airway physiology is required to obtain major insights into the pathophysiology of disease.

In conclusion, negative pressure itself is the major stimulus to genioglossal activation during inspiration. Using both within-breath and between-breath analysis, negative pressure consistently tracked closely with genioglossus activity, whereas other potential stimuli (flow, resistance) were dissociated from genioglossus activity across conditions. These observations do not provide anatomical specificity for the local mechanoreceptors important in mediating pharyngeal motor control.