INTRODUCTION

Top Abstract Introduction Materials & Methods Results Discussion References

HYSTERESIS OF THE INSPIRATORY pressure-flow relationship of the nasal airway has previously been observed during partial nasal obstruction (14, 16). The hysteresis demonstrated a smaller transnasal pressure difference during the increasing-flow portion of inspiration and a larger pressure difference during the decreasing-flow portion. When transnasal pressure was plotted against inspiratory flow, this resulted in an anticlockwise hysteresis. However, previous reports of nasal pressure-flow hysteresis (9, 14, 16) have not systematically identified the hysteresis, located the site of hysteresis in the nasal passages, or attempted to elucidate the physiological mechanisms responsible for the pressure-flow hysteresis. In addition, these reports have not totally excluded the possibility of measurement error as a cause of hysteresis, as there may have been slight phase mismatch of the pressure and flow signals. Therefore, it remains unclear whether nasal pressure-flow hysteresis is a normal physiological phenomenon during nasal hyperpnea.

Previously, other workers (5, 8) have demonstrated that the major site of resistance in the nose is within the nasal vestibule at the nasal valve. This segment of the nasal passage is not as rigid as the rest of the bony nasal cavum and is supported by cartilage and muscles. Thus the stability of the nasal vestibule during inspiration is dependent on both cartilaginous support and muscular activity of the nasal dilator muscles, predominantly the alae nasi (AN) (21). The electromyographic (EMG) activity of the AN muscles is well described (1, 11, 12, 15, 19, 20, 22). The AN demonstrate phasic inspiratory activity that commences before the onset of inspiratory flow, peaks before maximum inspiratory flow, and then progressively declines over the rest of inspiration, reaching tonic baseline levels during midexpiration (4, 17, 18, 22). Therefore, we hypothesize that inspiratory pressure-flow hysteresis of the nasal airway will occur because of changes in size or flow regime within the nasal vestibule caused by decreasing levels of nasal dilator muscle activity during inspiration. These changes will result in an increase in nasal resistance (for any given flow) over the course of an inspiration (i.e., nasal pressure-flow hysteresis). Furthermore, these changes are most likely to occur when negative intranasal pressure is greatest (at peak flow during hyperventilation) and when AN muscle activity is at reduced levels (during the decreasing-flow portion of inspiration).

Therefore, we studied the pressure-flow relationships of both the entire nasal airway and the nasal vestibular airway in normal subjects during both voluntary and hypercapnic hyperventilation. Specific aims were to identify the nasal pressure-flow hysteresis present during either quiet tidal breathing or hyperpnea and to demonstrate that the hysteresis predominantly arises in the nasal vestibule. In addition, we have examined the normal pattern of AN muscle activation as a mechanism involved in producing nasal pressure-flow hysteresis during hyperpnea and the role of voluntary nasal flaring in reducing the hysteresis.

	MATERIALS AND METHODS

Top Abstract Introduction Materials & Methods Results Discussion References

We studied six healthy male subjects [age 32.0 ± 6.4 (SD) yr, height 172 ± 9 cm, weight 65 ± 11 kg] during short voluntary hyperventilation runs while recording the pressure-flow relationships of both one entire nasal passage and the ipsilateral nasal vestibule airway. All the subjects who participated in this study were normal laboratory personnel and had no history of chronic nasal disease or recent upper airway infection.

During short periods of voluntary hyperventilation, the effect of AN activation on the nasal pressure-flow relationship was assessed by voluntary flaring of both nostrils and by comparing this to nonflaring hyperventilation. In addition, two of the six subjects performed the same voluntary flaring maneuvers during hypercapnic hyperventilation. The protocol was approved by the institutional Ethics Committee.

In the upright posture, subjects breathed via one nostril through a nasal mask with the mouth closed. All studies were performed by using a nasal continuous positive airway pressure mask (Respironics) with a dead space of 75-100 ml. The mask was modified to permit measurements of nasal flow, nasal airway pressures, CO₂ concentration, and AN EMG. Care was taken with mask placement to avoid any pressure on the nares or any obstruction to nasal airflow.

Flow was measured with a heated pneumotachograph (Fleisch no. 2) coupled to a differential pressure transducer (Validyne MP 45; ±5 cmH₂O) and connected to the mask. Flow was calibrated with a rotameter. Measured flow was bulk flow and did not provide regional information. Pressures in the mask (Pmask), ipsilateral nasal vestibule (Pvest) and posterior nasal choanae (Ppc) were measured simultaneously with differential pressure transducers (Validyne MP 45; ±100 cmH₂O). Pvest was measured by using a side-holed catheter located inside the ipsilateral patent nostril at 1 cm from the external naris (catheter tip was just proximal to the nasal valve region) (Fig. 1). Ppc was measured by using a side-holed catheter located >1 cm inside the contralateral nasal passage, with that external nostril then occluded by dental impression material (Coltene President Heavy Body) to provide an airtight seal. The pressure measured in the occluded nasal passage was then equivalent to Ppc. The pressure transducers were calibrated by using a water manometer. The pressure and flow signals were phase matched up to 7 Hz by oscillation with a sinusoidal pressure waveform from a loudspeaker. The EMG of the AN muscle of the unoccluded nostril was recorded from bipolar surface electrodes placed ~1 cm apart over the lateral wall of the external naris, with a reference electrode placed on the forehead. The raw EMG was displayed on an oscilloscope and amplified, band-pass filtered between 100 and 1,000 Hz, rectified, and passed through a "leaky integrator" with a time constant of 100 ms (Neotrace NT 1900) to produce a moving time average (MTA) EMG. The concentration of CO₂ in the mask was monitored by using an infrared CO₂ analyzer (Datex Normocap).

View larger version (20K): [in this window] [in a new window]   Fig. 1.   Parasagittal view of patent nasal cavity with nasal mask in place. N, nasal flow; Pmask, nasal mask pressure; Pvest, nasal vestibular pressure; Ppc, posterior choanal pressure; Ptv, transvestibular pressure (Pmask  Pvest); Ptn, transnasal pressure (Pmask  Ppc); NV, approximate region of nasal valve; HP, hard palate; IT, inferior turbinate; MT, middle turbinate; ST, superior turbinate. Ppc was measured with a catheter inserted past NV via the occluded nostril. Pressure measured in occluded nasal passage was then equivalent to Ppc in the patent nasal cavity.

Nasal flow and the three pressures, AN MTA EMG, and CO₂ level were simultaneously recorded on a strip-chart recorder (Hewlett-Packard 7758B). In addition, all signals were recorded together with the raw AN EMG onto videotape by using an analog-to-digital converter (Medical Systems, Greenvale, NY) for later off-line analysis.

Protocol 1: Voluntary hyperventilation. Six subjects performed voluntary hyperventilation runs while breathing room air. Each run consisted of three consecutive breaths where peak flow was doubled, with a 2- to 3-min rest between each run to avoid significant hypocapnia. Each voluntary hyperventilation run was performed either with voluntary constant nostril flaring or without nostril flaring. Between 7 and 10 runs were performed in a random fashion for both flaring and nonflaring maneuvers. During hyperventilation, the subjects were aided by feedback of a visual display of both nasal flow and AN MTA EMG signals on an oscilloscope (Tektronix 5B10N). Subjects were instructed to attempt to maintain similar peak inspiratory flows for both flaring and nonflaring runs. For each subject, the AN MTA EMG signal was calibrated to produce a full-scale deflection on the oscilloscope screen, with a maximum voluntary nasal flare. During flaring maneuvers, the subject was instructed to keep the AN EMG signal as high as possible throughout the entire inspiration. No special maneuvers were required during the nonflaring runs, and spontaneous EMG activity was recorded.

Protocol 2: Hypercapnic hyperventilation. In two of the six subjects, hyperventilation was induced by inspiring a mixture of air and between 4 and 8% CO₂ through a one-way valve (Hans Rudolph) connected to the pneumotachograph. The inspired CO₂ concentration was increased until the peak inspiratory AN MTA EMG was >80% of the activity seen during a maximum voluntary nasal flare. Each hypercapnic hyperventilation run (3 consecutive breaths) was performed either with voluntary constant nostril flaring or without nostril flaring (as for protocol 1). Between 7 and 10 runs were performed in a random fashion for both flaring and nonflaring maneuvers. The subjects were again instructed to maintain similar peak inspiratory flows for both flaring and nonflaring runs (by using visual feedback from the oscilloscope). Instructions for voluntary nasal flaring and nonflaring runs were the same as for protocol 1.

Data analysis. After the study, data were digitized with a sampling rate of 100 Hz (ADC 488/164, Iotech) and stored on a Macintosh IIcx computer. Transnasal (Ptn) and transvestibular (Ptv) pressures were calculated by subtraction of Ppc and Pvest (respectively) from Pmask (Ptn = Pmask  Ppc; Ptv = Pmask  Pvest). Tidal volume was determined by integration of the flow signal, and the inspiratory minute ventilation was calculated from the tidal volume and breathing frequency data.

The AN MTA EMG was quantified in arbitrary units above electrical zero and expressed as a percentage of the peak activity seen during maximum voluntary nostril flaring. For a flaring breath during hyperventilation to be included in the mean data, the minimum AN MTA EMG throughout an entire inspiration had to be >35% of maximum activity. For a nonflaring breath during voluntary hyperventilation, the peak AN MTA EMG during inspiration had to be <20% of maximum. Breaths that did not meet these criteria were excluded. AN MTA EMG activity for a breath was then calculated as the median value during the inspiratory phase of each breath.

Both transnasal and transvestibular inspiratory pressure-flow plots were then constructed for each hyperventilation run. Where hysteresis of the pressure-flow relationship was apparent, it was anticlockwise in direction with an ascending and descending limb. The inspiratory transnasal and transvestibular resistances were calculated from the ascending limb of the pressure-flow curves both at a flow rate of 0.2 l/s and at peak flow for both voluntary hyperventilation and hypercapnic hyperventilation.

Inspiratory hysteresis for both the nasal and vestibular airways was quantified as the ratio of the area under the descending limb of the pressure-flow curve over the area under the ascending limb (between zero and peak flow) (Fig. 2). Areas were measured from single-breath pressure-flow plots by use of a planimeter.

View larger version (37K): [in this window] [in a new window]   Fig. 2.   Representative inspiratory Ptn-flow curve for 1 breath (solid line) during hyperventilation, demonstrating presence of hysteresis, which loops in an anticlockwise direction (arrows). Vertical dashed line drawn at peak inspiratory flow. Hysteresis was quantified as ratio of area under descending pressure curve (A + B) to area under ascending pressure curve (A). This ratio of areas was used to normalize for variation in the absolute values of peak flows and pressures.

Statistical analysis was done by using Students t-test for paired samples. Data are means ± 1 SE.

	RESULTS

Top Abstract Introduction Materials & Methods Results Discussion References

Voluntary hyperventilation. For the group, between 7 and 21 breaths for each subject met the AN EMG criteria for nostril flaring (see METHODS) and between 15 and 21 breaths for each subject met the criteria for nonflaring. Data that met the criteria were meaned separately to give flaring and nonflaring values in each subject. The median AN MTA EMG activity during flaring runs (62.6 ± 4.4%max) was eight times greater than during nonflaring runs (7.8 ± 2.4%max; P < 0.01; Fig. 3).

View larger version (13K): [in this window] [in a new window]   Fig. 3.   Alae nasi moving time average (MTA) EMG (median inspiratory values) measured under both nonflaring and voluntary flaring conditions during hyperventilation. Each symbol represents an individual subject. Horizontal bars, mean values. * P < 0.001 relative to nonflaring. Note the consistent large increase in EMG activity during voluntary flaring.

Voluntary hyperventilation without nostril flaring resulted in significant anticlockwise hysteresis of the inspiratory pressure-flow relationship in all subjects, for both the vestibular and total nasal airways. There was no hysteresis at zero flow or during expiration in any subject. This is illustrated for the nasal airway of one subject (Fig. 4), in whom the direction of the hysteresis is anticlockwise and the AN MTA EMG activity is very low. During flaring, there was a dramatic increase in the AN MTA EMG throughout the inspiration (Fig. 4A), which was associated with marked decreases in both the pressure-flow hysteresis and the nasal resistance at higher inspiratory flows (Fig. 4B). Similar results were seen for the vestibular airway pressure-flow relationships.

View larger version (23K): [in this window] [in a new window]   View larger version (14K): [in this window] [in a new window]   Fig. 4.   Representative single breath from 1 subject during voluntary hyperventilation for both nonflaring and voluntary nasal flaring maneuvers. A: strip-chart record of nasal flow, Ptn, and alae nasi MTA EMG. Time intervals are 1 s. Vertical dashed lines mark end of inspiration. During flaring, note marked sustained increase in alae nasi EMG associated with an overall decrease in Ptn while nasal flow remains similar. B: corresponding Ptn-flow curves from data in A. Arrows show direction of hysteresis. Note 1) absence of hysteresis at 0 flow; and 2) decrease in Ptn and marked decrease in hysteresis of the pressure-flow relationship during voluntary nasal flaring. At higher flows, nasal resistance during flaring is substantially less than during nonflaring.

Inspiratory hysteresis was quantified as a ratio (see METHODS) such that a value of 1 indicated no hysteresis, and values >1 corresponded to increasing pressure-flow hysteresis. For both the vestibular and nasal airways, all subjects demonstrated decreased hysteresis during flaring compared with nonflaring hyperventilation (Fig. 5). Flaring reduced the vestibular airway hysteresis ratio from 1.96 ± 0.06 to 1.15 ± 0.06 (P < 0.01) and the total nasal airway hysteresis ratio from 2.05 ± 0.13 to 1.28 ± 0.06 (P < 0.01). Inspiratory hysteresis during nonflaring hyperventilation was similar for the vestibular and total nasal airways (P > 0.4) and decreased during flaring maneuvers by a similar amount for both airways (0.79 ± 0.08 and 0.70 ± 0.07, respectively, P > 0.2).

View larger version (18K): [in this window] [in a new window]   Fig. 5.   Inspiratory hysteresis (expressed as a ratio, see Fig. 2) of both vestibular and nasal pressure-flow relationships for nonflaring and flaring maneuvers during voluntary hyperventilation. Each symbol represents an individual subject. Horizontal bars, mean values. * P < 0.001 relative to nonflaring. Horizontal dotted line = no hysteresis (ratio 1.0). Note that the marked decrease in hysteresis during voluntary flaring is similar for both vestibular and nasal pressure-flow relationships.

Although we attempted to keep peak inspiratory flow constant between the flaring and nonflaring runs, the mean peak flows during flaring (1.0 ± 0.1 l/s) were slightly greater than those during nonflaring hyperventilation (0.8 ± 0.1 l/s; P < 0.01). Consequently, inspiratory ventilation during flaring (15.6 ± 1.5 l/min) was also greater than during nonflaring hyperventilation (11.4 ± 1.7 l/min; P < 0.01). We measured both nasal and vestibular resistances at a low flow (0.2 l/s) to document the nasal resistance at the onset of inspiration. This demonstrated that nasal and vestibular resistances (ascending limb) at the onset of flaring runs were similar to those at the onset of nonflaring runs (Table 1; P > 0.05). However, at peak inspiratory flow, both Ptn and Ptv values during flaring (5.2 ± 1.2 and 2.2 ± 0.9 cmH₂O, respectively) were less than the corresponding pressures during nonflaring hyperventilation (9.6 ± 1.9 and 6.2 ± 4.3 cmH₂O, respectively; both P < 0.01). Thus, when measured at peak flow, both vestibular and total nasal resistances during flaring were <45% of values obtained during nonflaring hyperventilation (Table 1). Nasal resistances were also measured at 0.2 l/s from the descending limb of the hysteresis loop, and the ascending limb resistances were less than the descending limb resistances during nonflaring runs but not during flaring runs (Table 1).

Hypercapnic hyperventilation. In the two subjects in whom hyperventilation was induced by hypercapnia, there were 21 and 23 breaths each that met the nostril flaring AN EMG criteria. Both the median and minimum AN MTA EMG values during nonflaring runs were less than during flaring runs in both subjects (Table 2). However, the peak inspiratory AN MTA EMG did not differ between flaring and nonflaring hyperventilation for either subject (Table 2). Thus the AN MTA EMG activity during a nonflaring inspiration demonstrated a rapid initial increase to a high peak value at the onset of inspiration, followed by a rapid decrease around the middle of the inspiration to relatively low values (Fig. 6). As the AN EMG activity fell, there was a corresponding decrease in nasal flow that was not associated with a comparable decrease in the Ptn (Fig. 6A). This resulted in substantial anticlockwise hysteresis of the nonflaring nasal airway inspiratory pressure-flow relationship (Fig. 6B). There was no significant hysteresis at zero flow or during expiration in either subject.

View larger version (20K): [in this window] [in a new window]   View larger version (12K): [in this window] [in a new window]   Fig. 6.   Representative single breath from 1 subject during hypercapnic hyperventilation for both nonflaring and voluntary flaring maneuvers. A: stripchart record of nasal flow, Ptn, and alae nasi MTA EMG. Time intervals are 1 s. Vertical dashed lines mark end of inspiration. During flaring, note marked sustained increase in alae nasi EMG compared with rapid fall off in EMG activity before midinspiration for the nonflaring maneuver. Initial peak levels of EMG activity are similar. B: corresponding Ptn-flow curves from data in A. Arrows show direction of hysteresis. Note 1) absence of hysteresis at 0 flow; and 2) marked decrease in hysteresis of pressure-flow relationship during voluntary nasal flaring when high levels of AN EMG activity are present throughout inspiration.

The AN MTA EMG activity during a flaring inspiration demonstrated a rapid initial increase to a high peak value that was relatively well maintained throughout the inspiration (Fig. 6A). In addition, nasal flow was relatively higher throughout the middle and latter part of inspiration compared with the nonflaring inspiration (Fig. 6A). For flaring breaths, decreases in the Ptn corresponded well with decreases in flow. Consequently, there was a marked reduction in the hysteresis of the nasal airway inspiratory pressure-flow relationship (Fig. 6B). From inspection of the flaring and nonflaring pressure-flow relationships, it was evident that the ascending flow limb of the relationship was relatively unchanged by the flaring maneuver, whereas a shift in the descending flow limb was primarily responsible for the decreased hysteresis during flaring (Fig. 6B). The descending flow limb during nonflaring corresponded to the time when the AN EMG activity was rapidly decreasing (Fig. 6A). Thus it appeared that nasal airway pressure-flow hysteresis occurred at a time when AN EMG activity fell to low levels during the inspiration.

Similar amounts of inspiratory hysteresis were present for both the nasal and vestibular airways under nonflaring conditions (Table 2). During flaring, the hysteresis decreased substantially for both the nasal and vestibular airways (Table 2). In subject one, the hysteresis was nearly abolished during flaring (Table 2 and Fig. 6B).

During hypercapnic hyperventilation, inspiratory ventilation was well matched between flaring and nonflaring runs (Table 3). However, peak inspiratory flow, peak Ptn, and peak Ptv were all reduced during the flaring runs in both subjects (Table 3). This resulted in lower nasal and vestibular resistances (at peak flow) during flaring, compared with nonflaring, hypercapnic hyperventilation (Table 3). Vestibular and nasal resistances at the onset of inspiration (at 0.2 l/s) tended to be similar for flaring and nonflaring hyperventilation (Table 3).

	DISCUSSION

Top Abstract Introduction Materials & Methods Results Discussion References

This study demonstrates that during modest nasal hyperventilation there was a similar degree of inspiratory hysteresis of the pressure-flow relationships for both the vestibular and total nasal airways in all normal subjects. In addition, voluntary nostril flaring throughout inspiration increased AN EMG activity and substantially decreased the pressure-flow hysteresis of both the vestibular and total nasal airways. However, high peak levels of AN EMG activity present only in early inspiration did not prevent nasal pressure-flow hysteresis. Rather, high levels of AN EMG activity that were maintained throughout inspiration (during nostril flaring) were associated with reduced nasal pressure-flow hysteresis.

The major finding of this study was the presence of nasal inspiratory pressure-flow hysteresis during increased nasal ventilation, produced either voluntarily or with hypercapnia in all normal subjects. We were confident that the hysteresis was not an artifact or methodological error for the following reasons. First, we were very careful to check that there was no phase mismatch of the pressure and flow signals during recording and subsequent analysis (see METHODS). Our signals were phase matched up to 7 Hz, which was well beyond the breathing frequency that was observed in this study. Second, no hysteresis was observed at the zero-flow points between expiration and inspiration. If pressure and flow signals were out of phase, hysteresis should have been present at zero flow. The absence of hysteresis at zero flow is important in regard to ensuring pressure catheter fidelity. Finally, no significant hysteresis was observed during expiration, which argues against a constant methodological error, as the hysteresis should have been present throughout the whole respiratory cycle.

Previous investigators (14, 16) have also reported nasal pressure-flow hysteresis during partial nasal obstruction. Solomon and Stohrer (16) noted hysteresis of the pressure-flow curves during nasal hyperventilation, with separation of the ascending and descending flow limbs around zero flow. This was different from the hysteresis observed in our study. Subsequently, Hey and Price (9) described nasal pressure-flow hysteresis that had similar characteristics to those seen in the present investigation. However, they only gave a qualitative description of hysteresis in one subject breathing a gas mixture of sulfurhexafluoride and O_2. No other investigators have reported nasal pressure-flow hysteresis during increased ventilation, which is somewhat surprising given the relative ease of making the pressure-flow measurements. Therefore, our study is the first to conclusively demonstrate that nasal pressure-flow hysteresis occurs during modest hyperventilation in normal subjects and is not an artifact due to measurement error.

In attempting to understand the cause of the hysteresis, we first considered the fluid dynamic principles of airflow through a tube. In applying these principles to the nasal passages, pressure may be utilized to 1) overcome airflow resistance (or frictional and viscous forces); 2) collapse elastic structures; and 3) overcome inertia to gas flow. Thus hysteresis of the nasal pressure-flow relationship may imply that some of the total Ptn is utilized to collapse elastic structures or to overcome inertia during acceleration of gases. To first consider inertial forces, acceleration of air in the nasal passages may cause inertial pressure losses, but this is generally unmeasurable at normal breathing frequency (10), and the pattern of hysteresis should result in maximal pressure-flow hysteresis around zero flow. Therefore, we consider it unlikely that the hysteresis seen in our study is due to pressure losses in overcoming inertia to gas flow. Alternatively, pressure may be utilized in the collapse or distension of elastic structures, such as occurs in the lung during inspiration (2, 7, 13). The only compliant portion of the nose is the nasal vestibule, the lateral walls of which are composed of fibrocartilaginous tissue together with discrete muscular attachments. Negative pressure applied to the nasal vestibule airway during inspiration may cause narrowing or collapse of the nasal vestibule (3, 8). In addition, the absence of hysteresis at zero flow and the anticlockwise direction of the hysteresis suggest that there are no tissue elastic effects in the nasal airway, as opposed to the lung. Thus the majority of inspiratory pressure-flow hysteresis observed in our study is unlikely to be due to pressure dissipation in collapsing the elastic structures of the nasal vestibule.

Therefore, the observed nasal pressure-flow hysteresis must be due predominantly to pressure expended in overcoming frictional and viscous forces related to airflow resistance. The frictional pressure losses produced by flow in a tube are determined by flow rate, fluid physical properties, and tube geometry (23). We discounted the gas physical properties as a cause of hysteresis as they did not change over an inspiration. To consider the role of flow rate during inspiration, identical flow rates occurred at a higher Ptn later in inspiration than occurred earlier in the same inspiration (i.e., pressure-flow hysteresis). If the identical flow rate required an increased driving pressure during the same inspiration, then either a change in nasal airway geometry or a change in flow regime must have occurred during the inspiration to explain the increased pressure losses. Thus, from a theoretical point of view, the increased Ptn during the descending flow limb of the nasal pressure-flow relationship (in the absence of flow limitation) would be consistent with narrowing of the nasal airway during inspiration. In support of this, nasal resistance values at low flows (0.2 l/s) are likely to be in the range where laminar flow occurs. Although measurements at low flow and pressure are more difficult to make accurately, our data suggest that the Ptn (or transnasal resistance) was higher for the descending flow limb compared with the ascending flow limb at low flows (Table 1). If the flow regime remains truly laminar for both these measurements, this suggests that the cross-sectional area of the vestibule must decrease during inspiration. However, the bulk of our hysteresis was observed at higher flows, where turbulent and transitional bulk flow regimes would occur. Therefore, in the present study, we cannot exclude the possibility that changes in flow regime may be responsible for a substantial proportion of the observed hysteresis. Thus the precise mechanisms responsible for the hysteresis in the present study remain undefined.

In this study, we measured the pressure-flow relationships of both one entire nasal cavity and the major portion of the corresponding nasal vestibule. Importantly, our results demonstrated that the amount of hysteresis that occurred in the nasal vestibule was virtually identical to that for the whole nasal cavity. This suggested that the segment of the nasal cavity that caused the pressure-flow hysteresis was predominantly located within the nasal vestibule. This agreed well with anatomic studies that demonstrated that the nasal vestibule was the only compliant part of the nasal cavity (6). Furthermore, the narrowest segment of the nasal cavity occurs in the region of the nasal valve, which is within the nasal vestibule (3). Bridger and Proctor (3) first reported a critical inspiratory transmural pressure at which the vestibular walls in the nasal valve region collapsed against the septum and created a Starling valve effect. Thus the nasal vestibule clearly has been demonstrated to be the most compliant section of the nasal cavity, which may collapse in the face of negative transmural pressures during increased ventilation. This is in agreement with the present study, in which the majority of the hysteresis arises in the nasal vestibule.

Contraction of the AN muscles can dilate the nasal vestibules and prevent their collapse during inspiration (17). Thus the size and shape of the nasal vestibule airway during increased negative transmural pressures (as may occur during nasal hyperventilation), will be partially dependent on the level of activity and resultant muscle contraction of the AN muscles. AN EMG activity has an inspiratory onset that precedes that of diaphragm electrical activity and inspiratory airflow (17). By the time inspiratory airflow starts, AN EMG activity has already reached approximately one-third of its maximum value. Activity reaches a peak in early or midinspiration and then decreases over the remainder of inspiration. Recently, Wheatley et al. (22) observed an increase in the lead time by which AN EMG activity preceded the onset of inspiratory flow during hypercapnic hyperventilation compared with quiet tidal respiration. However, the time from the onset of EMG activity to peak activity remained unchanged. This demonstrated that during hypercapnic hyperventilation, AN EMG activity did not remain at high levels throughout the whole of inspiration. This was similar to the nonflaring hyperventilation runs in the present study where AN activity was not maintained in the decreasing-flow portion of inspiration and pressure-flow hysteresis occurred. However, the nasal- flaring maneuvers had the effect of prolonging high levels of AN EMG activity throughout the inspiration and decreasing pressure-flow hysteresis. Therefore, the AN EMG data from the present study suggested that if adequate levels of AN activity were maintained throughout the whole inspiration then nasal vestibule pressure-flow hysteresis was substantially prevented, presumably by either decreasing the compliance of the nasal vestibule walls or altering flow regimes within the nasal vestibule. Conversely, inadequate levels of AN activity during the decreasing flow portion of inspiration were associated with the development of pressure-flow hysteresis.

Our observations may have clinical importance during maneuvers that increase ventilation (such as exercise) where patterns of AN activation are likely to be similar to those during hypercapnic hyperventilation. Based on our results, nostril flaring during exercise will substantially decrease hysteresis, nasal resistance, and the resultant work of breathing. This may allow trained athletes to breathe only nasally for longer periods while exercising before switching to oronasal breathing, with resultant benefits in filtration and humidification of the inspired air. It may also increase the maximum ventilation able to be achieved during oronasal breathing at high exercise work rates. However, training of athletes to achieve voluntary nasal flaring throughout each inspiration during exercise is likely to be difficult. An alternative means to stabilize the nasal vestibule, such as with an adhesive external nasal dilator strip, may be equally effective in preventing hysteresis and is worthy of further investigation.

In conclusion, we have confirmed that true hysteresis of the inspiratory pressure-flow relationship for both the vestibular and total nasal airways does occur in normal subjects during modest nasal hyperventilation. The hysteresis arises almost entirely in the compliant vestibule segment of the nasal airway, most likely because of either progressive narrowing of the nasal vestibule or a change in vestibular flow regimes during inspiration. Thus a real increase in nasal airflow resistance during inspiration (for a given flow) leads to the pressure-flow hysteresis, but the precise mechanism responsible for the hysteresis remains undefined. In addition, voluntary nostril flaring can largely prevent the pressure-flow hysteresis of the nasal vestibule airway. Our findings suggest that the prevention of hysteresis is achieved by stabilizing the nasal vestibule airway with continuous AN muscle contraction maintained throughout inspiration. We speculate that failure to adequately stabilize the nasal vestibule airway during increased ventilation may result in an increase in the mechanical work of breathing.