INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The obstructive sleep apnea syndrome is a major clinical disorder characterized by recurrent upper airway collapse and obstruction. There is now evidence that mechanical factors influence upper airway collapsibility (1). Two mechanical factors recently recognized to modulate upper airway collapsibility include the level of tongue and trachea displacement (2). Specifically, decreases in collapsibility have been observed with caudal trachea displacement, an effect that has been attributed to changes in airway length and longitudinal tension within the airway wall (2, 3). Tongue displacement has also been found to decrease collapsibility but only when the trachea has already been displaced caudally. The effect of tongue displacement has been attributed to alterations in upper airway dilating forces, which can only decrease collapsibility when tension within the upper airway wall is present (2). Thus, the trachea and tongue displacement represent two distinct mechanical mechanisms by which collapsibility can be altered. It is also widely recognized that neuromuscular activity is an important determinant of upper airway patency (1, 4). Experimentally, one can achieve large changes in upper airway neuromuscular activity by inducing hypercapnia (5), which has been shown to decrease upper airway collapsibility. Several upper airway muscles have been postulated to influence collapsibility and are known to be stimulated by hypercapnia. These muscle groups include the genioglossus and the cervical strap muscles. However, the mechanisms by which these activated muscles stabilize the pharynx have not been extensively investigated. The present study was therefore designed to investigate the mechanisms through which neuromuscular activity alters collapsibility. Specifically, we examined whether hypercapnia, a generalized stimulus of the upper airway musculature, alters collapsibility by a mechanism similar to trachea displacement or tongue displacement.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

This study was approved by the Institutional Animal Use and Care Committee. Two experimental protocols were performed, each performed on six supine male cats (2-2.5 kg). Cats were premedicated with an intramuscular injection of xylazine (3 mg /kg) and then anesthetized with ketamine (50 mg /kg). Arterial blood pressure was monitored through a femoral arterial line and maintained with intravenous normal saline through a femoral vein. Rectal temperature was maintained at 37-38° C. The cranium was exposed and a midcollicular decerebration was performed as previously described (9).

Isolated Upper Airway

An isolated feline upper airway preparation was utilized as previously described (3, 10). In this preparation, the maximal inspiratory airflow (Imax) has been found to be determined by the collapsibility of the pharyngeal airway (Pcrit) at the flow limiting site (FLS) and by the nasal resistance (Rn) upstream from this site. To prepare the isolated upper airway, the cervical trachea was transected at the second or third tracheal ring. Its distal end was cannulated with an endotracheal tube (5 mm ID), and the animal was mechanically ventilated (Model 665; Harvard Apparatus, Dover, MA). A rigid cannula (3 mm ID, 2 cm in length) was inserted into the proximal tracheal stub, through the glottic structures, and fixed in place at the level of the aryepiglottic folds. To monitor respiration, an esophageal balloon was inserted into the lower esophagus, which was then tied below the balloon insertion site. The tongue was allowed to prolapse into the mouth and the lips were sutured shut using a purse-string suture. The cat's head was fixed in place at an angle of ~ 70-80° from the horizontal plane.

A mobile catheter (PE tubing, 1.4 mm ID) with a side hole midway along its length was inserted through one nostril into the pharynx and out the tracheal cannula. This catheter was used to measure the pharyngeal pressure (Pph) at different locations in the airway and served as a reference for measuring changes in airway length. Hypopharyngeal pressure (Php) was measured with a catheter inserted into the tubing connecting the rigid cannula to the pneumotachograph. A pneumotachograph (Flesich 01) was connected in series between the tracheal cannula and a negative pressure source. The pneumotachograph was coupled to a differential pressure transducer (Validyne MP-45; ± 2 cm H₂O) to measure airflow drawn in the inspiratory direction through the isolated upper airway by the negative pressure source. Pharyngeal pressures were monitored with Gould-Statham transducers (P23ID). Pharyngeal secretions were aspirated as needed with a small catheter connected to the tracheal cannula.

Measurement of Tracheal Displacement

Displacement of the proximal tracheal stub was achieved as described previously (3). In brief, the proximal tracheal cannula was mounted on a linear ball-bearing bracket that glided along a rod that limited movement of the cannula to the longitudinal direction. The movement of this bracket represented the amount of caudal tracheal displacement. The amount of displacement was confirmed by measuring the caudal displacement of the mobile pharyngeal catheter during tracheal displacement. We defined the zero tracheal position as the resting baseline tracheal position when secured to the ball-bearing bracket.

Changes in End-tidal CO₂

To achieve large changes in upper airway neuromuscular activity, the arterial tension of carbon dioxide was systematically altered. Hypercapnia is a generalized upper airway neuromuscular stimulus that has been shown to decrease Pcrit (10, 11). Hypocapnia, on the other hand, is associated with a decrease in neuromuscular activity resulting in an airway similar to an airway during generalized paralysis (10). End-tidal CO₂ was measured at the endotracheal tube (LB-2 gas analyzer; Beckman). Arterial blood gases were obtained to confirm that the end-tidal CO₂ and arterial CO₂ tension correlated. Low CO₂ levels (see below for specific levels for each protocol) were achieved by mechanically ventilating the cat with tidal volumes of approximately 30 cc/kg at respiratory rates of ~ 40-50 breaths per minute. The respiratory rate and tidal volume were chosen to be comparable to those present in the hypercapnic condition. High CO₂ levels were achieved by delivering an enriched carbon dioxide and oxygen gas mixture through a T-piece attached to the endotracheal tube. During hypercapnia, arterial oxygen tension levels were monitored and kept > 100 mm Hg. All data were collected once steady-state levels of end-tidal CO₂ were achieved and the respiratory pattern became stable.

Inspiratory airflow (I), Pph, and Php were continuously recorded on a strip chart recorder (Gould Instruments, Cleveland, OH) while Php was lowered in a ramp-like fashion. Analog-to-digital acquisition of the I, Pph, and Php signals was also performed at 128 Hz (Asyst-Keithley Technologies, Rochester, NY). The frequency response characteristics of each catheter-transducer-amplifier-filter system was examined and demonstrated > 95% amplitude at 10 Hz.

Determination of Imax, Pcrit, and Rn

Imax, Pcrit, and Rn were measured as previously described (2, 3, 10). In brief, pressure at the downstream end of the upper airway was rapidly lowered in a ramp-like fashion to approximately 75 to 100 cm H₂O (Figure 1). As Php was lowered (lower panels), I increased and reached a maximum, Imax, at the onset of inspiratory airflow limitation for the isolated upper airway.

View larger version (18K): [in this window] [in a new window]   Figure 1.   Representative pressure and flow recordings obtained in isolated upper airway under hypocapnic (panel A) and hypercapnic (panel B) conditions. Php (lower graphs) is lowered and airflow (upper graphs) rises in the inspiratory direction. Pph (middle graphs) is measured immediately upstream to flow-limiting site. V· Imax is defined as apex in the V· I versus time curve; Pcrit is defined as the nadir in Pph versus time curve.

To determine Pcrit and Rn, the FLS was first delineated by monitoring side-hole pressure at several sites in the pharynx, as previously described (2, 3, 10). In each cat, the FLS was found to occur at a discrete locus ± 1 cm to the rim of the soft palate. After delineation of the FLS, the mobile pharyngeal catheter was maintained at or immediately upstream from this site to measure Pph. Pcrit was defined as the nadir in Pph at the onset of flow limitation as illustrated in Figure 1. Rn was calculated as follows: Rn = (Pn-Pcrit)/Imax, where nasal pressure (Pn) remained atmospheric.

Previous work from our lab has shown that the maximal changes in Imax, Pcrit, and Rn are during the inspiratory phase of the respiratory cycle in spontaneously breathing cats (3, 10). Therefore, while the cats were spontaneously breathing in the hypercapnic condition, all measurements of I, Pph, and Php were obtained during inspiration, as indicated by a negative deflection in the esophageal pressure trace. During the hypocapnic condition, mechanical ventilation eliminated spontaneous respiratory muscle activity, as indicated by the esophageal pressure trace, and I, Pph, and Php were measured irrespective of the respiratory cycle.

Experimental Protocols

Protocol #1: The effect of hypercapnia on trachea displacement. Imax, Pcrit, and Rn were measured before and after 1 cm of trachea displacement at low (1.3 ± 0.3%) and high (10.0 ± 0.3%) levels of CO₂. At each combination of trachea displacement and CO₂, 5 pressure-flow measurements were made to obtain Imax, Pcrit, and Rn. In this protocol, tracheal displacement was maintained for approximately 10 min under each CO₂ condition. Each CO₂ condition was maintained for approximately 20 min. Results were analyzed by a two-factor analysis of variance (ANOVA) (Minitab, State College, PA).

Protocol #2: The effect of cutting the strap muscles and hypoglossal nerve. Imax, Pcrit, and Rn were measured at each of five conditions: (1) Control or low (2.3 ± 0.1%) CO₂; (2) High (9.9 ± 0.3%) CO₂, intact upper airway; (3) High CO₂ after bilateral transection of the strap muscles, the sternohyoid, sternothyroid, and thyrohyoid; (4) High CO₂ after bilateral transection of the hypoglossal nerves; (5) Control or low CO₂ (2.0 ± 0.3%). These conditions were tested sequentially as written. For each condition, 5 pressure-flow measurements were made to obtain Imax, Pcrit, and Rn. Hypercapnia was maintained for approximately 20-30 min while hypocapnia was maintained for 5-10 min before and after the hypercapnic condition. Results were first analyzed using a single-factor ANOVA to determine if there was an effect of condition on Imax, Pcrit or Rn. After the ANOVA showed a significant effect of condition, a post-hoc least-squared test was performed to determine which conditions were significantly different from each other.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

In Figure 1, representative I and pressure recordings are illustrated for the upper airway of one cat under hypocapnic (panel A) and hypercapnic conditions (panel B). As the Php was lowered, airflow rose and plateaued; the plateau was equal to Imax. Pph, measured immediately upstream to the FLS, decreased with Php until flow limitation occurs, at which time it plateaued. Pcrit was the nadir of the Pph versus time curve. In this example, Imax increased from 119 ml/s (panel A, upper graph) to 255 ml/s (panel B, upper graph) when the CO₂ was raised from 1.0% to 9.5%. Pcrit decreased from 2.1 cm H₂O (panel A, middle graph) to 4.8 cm H₂O (panel B, middle graph) with the same change in CO₂.

Protocol #1: The effect of hypercapnia on the trachea displacement. The results of Protocol #1 are shown in Figure 2 where Imax (top), Pcrit (middle), and Rn (bottom) were plotted against trachea displacement at low (open circles) and high (closed circles) CO₂. As has been previously shown, Imax increased with either trachea displacement (from 70 ± 12 ml/s to 151 ± 16 ml/s, p = 0.004; all data presented in this section are mean ± SEM) or hypercapnia (from 70 ± 12 ml/s to 188 ± 31 ml/s, p = 0.002). However, when the trachea was displaced under conditions of hypercapnia, the change in Imax with trachea displacement was smaller than under conditions of hypocapnia (44 ± 12 ml/s versus 81 ± 7 ml/s, p = 0.048).

View larger version (13K): [in this window] [in a new window]   Figure 2.   Protocol #1: The effect of hypercapnia on trachea displacement. V· Imax (upper panel ), Pcrit (middle panel ) and Rn (lower panel ) versus trachea displacement at either low (open circles) or high (closed circles) CO2 levels. Data are means ± SE. V· Imax increases with trachea displacement (p = 0.004) and hypercapnia (p = 0.002) while Pcrit decreases with trachea displacement (p = 0.015) and hypercapnia (p = 0.002). Rn does not change with either condition. The changes in V· Imax, Pcrit, and Rn with trachea displacement under conditions of hypercapnia are attenuated compared with the changes under conditions of hypocapnia (p = 0.048 for V· Imax and Rn, p = 0.001 for Pcrit).

Figure 2 also shows that these changes in Imax were associated with changes in Pcrit and Rn. Pcrit decreased from 1.8 ± 0.2 cm H₂O to 7.0 ± 1.1 cm H₂O with trachea displacement (p = 0.015) and to 6.3 ± 0.5 cm H₂O (p = 0.002) with hypercapnia. The change in Pcrit with trachea displacement under hypercapnic conditions was also attenuated as compared to hypocapnia (2.4 ± 1.1 cm H₂O versus 5.2 ± 1.1 cm H₂O, p = 0.001). Rn did not change with either trachea displacement or hypercapnia (p = NS) but the change in Rn with trachea displacement was also attenuated under hypercapnic conditions (22.6 ± 9.3 cm H₂O/l/s versus 4.5 ± 3.1 cm H₂O/l/s, p = 0.048).

Protocol #2. Figure 3 illustrates the results of Protocol #2. As the CO₂ level was changed from low to high, Imax increased from 89 ± 12 ml/s to 238 ± 40 ml/s (p < 0.001). Imax did not change significantly after the strap muscles were cut (241 ± 45 ml/s, p = NS) or hypoglossal nerves were transected (200 ± 30 ml/s, p = NS). Returning to the baseline hypocapnic condition, Imax decreased back to baseline (77 ± 15 ml/s).

View larger version (17K): [in this window] [in a new window]   Figure 3.   Protocol #2: The effect of cutting the strap muscles and hypoglossal nerves. V· Imax (upper panel ), Pcrit (middle panel ), and Rn (lower panel ) under the following conditions: (1) hypocapnia baseline (open bars); (2) hypercapnia baseline; (3) hypercapnia after cutting the strap muscles; (4) hypercapnia after cutting the strap muscles and hypoglossal nerves; (5) hypocapnia baseline. Data are means ± SE. There was a significant change in V· Imax and Pcrit from baseline hypocapnia to hypercapnia and back (between conditions 1 and 2 and 4 and 5, p < 0.001 for both) but no change in either parameter after cutting either the strap muscles or hypoglossal nerves. There was no change in Rn between any condition.

In this protocol, the changes in Imax were associated with reciprocal changes in Pcrit but no changes in Rn. The changes in Pcrit paralleled those in Imax: a significant decrease in Pcrit going from hypocapnia to hypercapnia (2.7 ± 0.7 cm H₂O to 7.3 ± 0.8 cm H₂O, p < 0.001); no change after cutting the strap muscles (7.4 ± 1.0 cm H₂O) or hypoglossal nerves (5.8 ± 0.8 cm H₂O); a return to baseline (2.1 ± 0.6 cm H₂O) on return to the hypocapnic condition. There were no changes in Rn across all five conditions.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

In examining the effects of tracheal displacement and hypercapnia on upper airway airflow dynamics, we obtained the following results. First, we confirmed the results of previous studies that both hypercapnia and tracheal displacement increase Imax in the isolated upper airway (2, 3, 10, 11). Second, there is a significant and negative interaction between trachea displacement and the level of CO₂ on Imax. In other words, hypercapnia attenuated the change in Imax with trachea displacement (Figure 2), suggesting a similar mechanism for the increase in Imax with CO₂ and tracheal displacement. Third, cutting the hypoglossal nerves and strap muscles when the airway musculature was stimulated by hypercapnia did not lead to any changes in the response of Imax to hypercapnia (Figure 3). These findings suggest that a neuromuscular mechanism accounts for increases in Imax during hypercapnia that is analogous to the action of the trachea on the airway when it is displaced. The effect of hypercapnia, however, is not mediated by either the cervical strap muscles or the tongue.

We have utilized an isolated upper airway model to examine neuromuscular mechanisms modulating maximal airflow through the upper airway (2, 3, 10, 11) for the following reasons. First, it is well recognized that pharyngeal airflow obstruction in the sleeping human is characterized by the development of inspiratory flow limitation. The isolated feline upper airway model reproduces this condition and permits further study of the physiologic basis for alterations in Imax. Second, this model allows a functional approach to the upper airway. Direct measurements of Pcrit and Rn, the physiologic determinants of Imax can be made by partitioning the upper airway functionally into a flow-limiting site, and segments upstream and downstream from this site. In so doing, we have been able to determine how Pcrit, a measure of the collapsibility at the FLS, and Rn modulate Imax when neuromuscular activity is altered. Third, the specific muscles which might mediate these responses could be identified and their mechanism of action could be studied by systematically activating and transecting these muscles, and by elongating the pharyngeal airway, respectively. This approach has helped determine the role played by specific anatomic structures in modulating the degree of pharyngeal airflow obstruction dynamically.

We have previously been able to attribute changes in Imax in the isolated feline upper airway to changes in either Pcrit or Rn (2, 3, 10, 11). In this study, alterations in Imax during tracheal displacement and hypercapnia were associated with inverse changes in Pcrit, reflecting changes in collapsibility at a discrete site of palatal collapse. In addition, neither Imax nor Pcrit changed significantly after transecting either the strap muscles or hypoglossal nerve, suggesting that activation of these muscles did not influence the collapsibility at the palatal rim. On the other hand, hypercapnia did not produce significant changes in Rn, indicating that the segment upstream of the site of collapse was not involved in the neuromuscular modulation of Imax. Therefore, responses in Imax to hypercapnia and tracheal displacement were primarily due to changes in Pcrit, representing alterations in pharyngeal collapsibility.

Mechanism for the Response to CO₂

In this study, we investigated the mechanisms for the response of Pcrit to CO₂. Our study is based on previous work in which we found evidence suggesting that changes in Pcrit could be explained for by either changes in the properties of the airway wall or changes in the pressure surrounding the airway wall (2). Our previous study suggested that caudal tracheal displacement decreases Pcrit by altering the properties of the airway wall; specifically, tracheal displacement increases the longitudinal tension in the airway wall. In contrast, reductions in Pcrit with tongue displacement were attributed to decreases in the pressure surrounding the airway, an effect which varied markedly depending on the level of tracheal displacement. When the effects of tracheal and tongue displacement were compared, we found that the tracheal effect on Pcrit predominated. In the present study, we sought to determine whether CO₂, like tracheal displacement, influenced Pcrit by increasing tension within the airway wall.

In the first protocol, we found that hypercapnia diminished the change in Pcrit to tracheal displacement. An explanation for this finding is that CO₂ activates upper airway muscles that increase the tension within the upper airway wall. Activation of these muscles would diminish the effect of trachea displacement of Pcrit because the CO₂ had already increased the wall tension substantially. Thus, we interpret the results of Protocol #1 as indicating that hypercapnia activates upper airway muscles that decrease pharyngeal wall collapsibility (Pcrit) by increasing tension within the upper airway wall.

Given the large body of evidence indicating a role for the dilator muscles in the upper airway patency (1, 4, 7, 12), we gave serious consideration to the possibility that airway dilators also play a role in the Pcrit response to hypercapnia. However, we could not reconcile this possibility with our own data. Consider the hypothetical findings presented in Figure 4. If the Pcrit response to hypercapnia were solely due to increases in airway wall tension, the response to tracheal displacement at high CO₂ would be flat, as represented in Figure 4A. In this situation, a significant interaction between CO₂ and tracheal position would be found, without any independent effect of hypercapnia on Pcrit. However, we found significant reductions in Pcrit during hypercapnia that were independent of the level of tracheal displacement. We originally thought that this independent effect of CO₂ might be due to activation of muscles that dilate the upper airway. But our previous work examining the effect of tongue position on Pcrit suggests that this effect does not exist (2). Rather, our earlier data show an interaction between tongue and tracheal position such that the change in Pcrit in tracheal displacement is greater if there is concomitant tongue displacement. Thus, these results suggest that activation of airway dilators by CO₂ should produce a greater Pcrit response to tracheal displacement as shown in Figure 4B. Instead, we observed a decrease in the Pcrit response to tracheal displacement during hypercapnia (Figure 2, middle panel, and Figure 4C). Attenuation of this response suggests that dilating mechanisms could have only played a minor role in the Pcrit response to CO₂ (see also subsection THE ROLE OF THE TONGUE MUSCLES below). Thus, while we cannot completely rule out the influence of the airway dilators, they do not appear to play a major role in the mediating dynamic neuromuscular alterations in Pcrit.

View larger version (10K): [in this window] [in a new window]   Figure 4.   Hypothetical results for the interaction between CO2 level and tracheal replacement. (A) Hypothesized result if hypercapnia alters Pcrit solely by increasing the longitudinal tension of the airway wall; note that there is no independent effect of hypercapnia on Pcrit. (B) Hypothesized result if the predominant effect of hypercapnia is activation of upper airway dilators. (C) Hypothesized result if the predominant effect of hypercapnia is activation of muscles that increase longitudinal tension; note that this was the observed result (see Figure 2, middle panel ). Hypocapnic conditions are represented by solid lines; hypercapnic conditions are represented by broken lines. See DISCUSSION for details.

Role of the Strap Muscles

The second protocol was designed to examine the roles of specific muscle groups in mediating the Pcrit response to CO₂. We first examined the role of the cervical strap muscles (sternohyoid, sternothyroid, and thyrohyoid) that provide caudal tug on the upper airway which might have been expected to increase airway wall tension at high CO₂. It is known that these muscles are phasically active during inspiration (13, 14) and are recruited when the experimental animal is breathing hypercapnic air mixtures (5, 6). Despite such recruitment, however, Pcrit was unchanged after transection of the strap muscles while the cats were breathing hypercapnic air mixtures (Figure 3). Thus, in our system the Pcrit response to hypercapnia is not due to activation of the strap muscles.

Nevertheless, several authors (6, 14, 15) have reported that both passive tension and electrical stimulation of the strap muscles decrease the collapsibility of the upper airway in animal preparations. Although we found no role for the strap muscles, there are two explanations for this difference between our results and former work. First, other pharyngeal muscles were quiescent when the strap muscles were either passively stretched (15) or directly stimulated (6, 14) in previous work. In contrast, we examined the role of the strap muscles when there was generalized activation of upper airway muscles. In other words, it is possible that the cervical strap muscles play a greater role in the modulation of Pcrit when the pharyngeal musculature is quiescent than when other pharyngeal muscles are activated by CO₂. Another explanation for differences between our results and other studies may relate to the fact that the tracheal cannula at the lower end of the upper airway in our preparation was fixed in place. This cannula was placed by design in order to maintain the pharyngeal airway at a constant length and longitudinal tension. If airway tension is increased by strap muscle activation, the tracheal cannula may have attenuated the effect of activation of these muscles on Pcrit. Therefore, it is still possible that either passive tension or neuromuscular activation in the strap muscles modulates pharyngeal collapsibility, particularly when other pharyngeal muscles are quiescent.

The Role of the Tongue Muscles

We also investigated the potential contribution of upper airway dilating muscles by investigating the effect of hypoglossal nerve transection on Pcrit. The hypoglossal nerve is the sole innervation of the genioglossus muscle (16), which has been implicated as a major determinant of upper airway patency (1, 4). In particular, it has been shown that genioglossus EMG and the hypoglossal nerve are physically active during inspiration and their activity is augmented by hypercapnia in both experimental animals and humans (5, 8, 17, 18). Moreover, Brouillette and Thach (7) demonstrated an inverse correlation between Pcrit and genioglossus activity when genioglossus activity was varied with the level of anesthesia. These authors did not know, however, whether genioglossus recruitment was required to lower Pcrit. In our study, there was no significant change in the Pcrit response to hypercapnia after hypoglossal nerve transection, although a modest increase in Pcrit in four of the six cats after hypoglossal nerve transection was observed. We conclude that the genioglossus may at best play a small role in modulating the Pcrit response to hypercapnia in our model despite rather substantial recruitment of this muscle.

There are two possible explanations for why transection of the hypoglossal nerve did not change Pcrit in our model. First, it is known that midcollicular decerebration reduces the response of the hypoglossal nerve to hypercapnia (19). Although the effect of decerebration on the activity of other cranial and cervical nerves is not known, there is no reason to suspect differences between the response of these nerves and the hypoglossal nerve to CO₂. Therefore, we do not believe that decerebration accounts for our failure to alter the Pcrit response to hypercapnia with hypoglossal nerve transection.

Second, it is possible that the tongue position minimized the effect of hypoglossal nerve transection on Pcrit. We acknowledge that the tongue was allowed to prolapse into the pharynx in our preparation, thereby placing it at a mechanical disadvantage when the genioglossus is activated by hypercapnia. Such prolapse, however, often occurs in apneics during sleep. Our findings in the isolated feline upper airway suggest that physiologic stimulation may not alter pharyngeal collapsibility if the tongue is prolapsed. Rather, supraphysiologic genioglossal activation may be required to decrease Pcrit substantially. In fact, large reductions in Pcrit have been achieved in the isolated feline upper airway by electrically stimulating the hypoglossal nerve (20), presumably by causing bulk anterior movement of the tongue. Thus, we believe that other muscles play a greater role in the modulation of Pcrit during physiologic chemostimulation of the pharyngeal musculature.

Possible Role of the Intrinsic Pharyngeal Muscles

Since decreases in Pcrit with hypercapnia could not be explained by recruitment of either the cervical strap or tongue muscles, we suggest that other upper airway muscles mediated this response. To determine which muscles were likely to mediate that change in Pcrit, one must consider the anatomy of the upper airway in the context of our experimental setup. In our preparation, the FLS resided in the pharynx at the palatal rim. The walls of the oropharynx are formed by the tongue anteriorly and the pharyngeal constrictors postero-laterally, while the walls of the nasopharynx are formed by the pharyngeal constrictors postero-laterally, the soft palate anteriorly, and muscles connecting the palate and pharynx laterally (21). In cats, the pharyngeal constrictor muscles have been shown to be most active during expiration (22), whereas the largest decreases in Pcrit are seen during inspiration, suggesting that these muscles did not effect the changes in the Pcrit with hypercapnia. In contrast, the palate muscles (tensor veli palatini, levator veli palatini, and musculae uvulae) have been shown to have inspiratory-related phasic activity in canine preparations (23, 24) that is augmented by hypercapnia (24) or negative upper airway pressure (23). While these muscles are not oriented longitudinally, they are intrinsic to the airway wall and their activation likely increases the tension within the airway wall. Other preliminary work in humans indicates that the palatopharyngeus and palatoglossus, which are oriented longitudinally, also have phasic inspiratory activity that is augmented by hypercapnia and negative airway pressure (25, 26). Thus, muscles intrinsic to the soft palate and pharyngeal wall are likely candidates for the upper airway muscles that mediate the Pcrit response to hypercapnia. Future work should be directed at studying the role of the palate muscles and palatopharyngeus in inspiratory airway patency.

Weaknesses of the Model

Several limitations of our isolated feline upper airway preparation should be noted when applying our findings to the human upper airway. First, the effect of midcollicular decerebration and tracheal position on hypoglossal nerve activity and strap muscle function, respectively, is uncertain. Second, large negative suction pressures (40 to 70 cm H₂O) were generated to achieve airflow limitation in our preparation. It is conceivable that these pressures could traumatize the upper airway, leading to changes in collapsibility. These pressures were measured within a narrow cannula considerably downstream to the pharynx, making it unlikely that these markedly negative pressures were transmitted to the pharynx. We also know that the FLS was not actually exposed to these large negative pressures because the airway pressures at this locus (Pcrit) only varied in the range of 2 to 12 cm H₂O. Furthermore, baseline levels of Imax and Pcrit remained stable from the beginning to the end of Protocol #2, indicating no significant deterioration in airway function. Third, we sutured the lips and allowed the tongue to prolapse which might have influenced the activity of the strap and genioglossus muscles. This procedure, however, resembles the human condition in that inspiratory flow limitation is most likely to occur in supine patients when the tongue prolapses most readily into the pharynx.

In conclusion, we have shown that a generalized upper airway neuromuscular stimulus decreases Pcrit in part by altering the tension of the pharyngeal airway wall, similar to that seen with trachea displacement. In addition, we found an effect of CO₂ on Pcrit that was independent of the strap muscles. Our data cannot exclude a role for pharyngeal dilators though it appears the tongue does not contribute to the Pcrit response to hypercapnia. We conclude that an alteration in airway wall tension is a major mechanism by which upper airway neuromuscular activity alters Pcrit and speculate that defects in this mechanism could lead to the pathophysiology that leads to the development of obstructive sleep apnea. This work suggests that further research should be preformed to study the role of the muscles intrinsic soft palate and pharyngeal wall in airway collapsibility and the pathogenesis of obstructive sleep apnea.