INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

NORMAL CHILDREN ARE LESS likely to snore than normal adults and have fewer obstructive apneas during sleep (23). This suggests that the pediatric upper airway is less susceptible to collapse. This difference in upper airway function may be due to structural differences or to alterations in the neuromuscular drive to the upper airway muscles.

A major structural factor that can affect upper airway collapsibility is obesity, which is a well-described risk factor for the development of the obstructive sleep apnea syndrome (OSAS) in children (20) and adults. Because children tend to be thinner than adults (13), differences in body size may partly explain the decreased upper airway collapsibility noted in normal children compared with adults. Previous studies have shown that the structural properties of the upper airway can be evaluated by determining the relationship between nasal pressure (P_N) and inspiratory flow with use of the Starling resistor model of upper airway function (22, 30). According to this model, under conditions of flow limitation, maximal inspiratory flow (I_max) is determined by the pressure changes upstream (nasal) to a collapsible locus of the upper airway and is independent of the downstream (tracheal) pressure generated by the diaphragm. The critical closing pressure (P_crit) occurs at the x-intercept of the pressure-flow curve, i.e., the pressure at which there is zero flow due to upper airway closure. In an isolated Starling resistor model, I_max is determined by P_crit and the resistance of the upper airway upstream to the site of collapse. Therefore, the mechanical properties of the upper airway leading to airflow obstruction can be evaluated by determining P_crit and upstream resistance.

Neuromuscular factors are also important in maintaining upper airway stability. The upper airway muscles are accessory muscles of ventilation and, as such, are affected by changes in ventilatory drive. Because children have a higher ventilatory drive than adults (10, 31, 35), this may partly explain the differences in upper airway function. One of the neuromuscular factors thought to be important in the maintenance of upper airway patency is the reflex response to subatmospheric pressure. Previous studies have shown increases in upper airway neuromuscular activity in response to drops in P_N during wakefulness in adults; this is attenuated during sleep (4). The effects of these reflex responses on upper airway patency have not been systematically examined but can be further evaluated using the Starling resistor model. In an isolated Starling resistor model, the sensitivity of the upstream pressure-flow relationship is represented by the slope of the pressure-flow curve (S_PF). The resistance of the upstream segment of the upper airway is the reciprocal of the S_PF. However, in the living organism the S_PF is affected not only by mechanical and structural factors (the degree of "stiffness" of the upper airway), but also by reflex neural mechanisms. Thus the S_PF is a useful tool for the comprehensive evaluation of upper airway function.

We hypothesized that the pediatric upper airway was better able to compensate for a subatmospheric pressure load than the adult upper airway because of differences in upper airway structure or upper airway neuromuscular responses to the load. We therefore examined the developmental differences in upper airway function among normal individuals during sleep. Specifically, we determined the influence of subatmospheric pressure on the maintenance of upper airway patency across the age spectrum. Responses to subatmospheric pressure were examined by delineating the upstream pressure-flow relationships for each individual.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Nonsnoring subjects were recruited from the general community. All subjects underwent baseline polysomnography to ensure that they were normal. On a separate night, pressure-flow measurements during sleep and occlusion pressure in 100 ms (P_0.1) during wakefulness and sleep were determined.

Study group. The study group consisted of healthy children (old enough to cooperate with testing) and adults. Subjects 16 yr of age were considered to be children. Subjects with habitual (nightly) snoring or obesity [defined as a body mass index (BMI)  30 kg/m²] were excluded. Children with a history of tonsillectomy and/or adenoidectomy were excluded, because OSAS is now a common indication for adenotonsillectomy (27). All children had visible tonsillar tissue on examination. Because tonsillectomies and adenoidectomies were performed so commonly in the past, this exclusion criterion was not applied to the adult group. Informed consent was obtained from each subject, as well as from the parents/legal guardians of the children. The study was approved by the Institutional Review Board of Johns Hopkins University.

Baseline polysomnography. Polysomnographic studies were performed overnight. During polysomnography, the following parameters were measured and recorded continuously using a computerized polysomnography system (Alice 3, Healthdyne, Marietta, GA): electroencephalogram (C3/A2, O1/A2); right and left electrooculogram; submental electromyogram (EMG); tibial EMG; electrocardiogram; chest and abdominal wall motion (piezoelectric transducers); oronasal airflow (3-pronged thermistor); end-tidal PCO₂, measured at the nose by infrared capnometry (model N-1000, Nellcor, Van Nuys, CA); arterial O₂ saturation by pulse oximetry (model N-1000; Nellcor) and oximeter pulse waveform. Subjects were also monitored and recorded on videotape with an infrared videocamera and were continuously observed by a polysomnography technician. Sleep architecture, arousals from sleep, and cardiorespiratory parameters were analyzed using standard techniques (2, 3, 26, 33). Subjects with an apnea index 5/h were excluded.

Pressure-flow measurements. Pressure-flow measurements were obtained during a separate, overnight polysomnogram, as previously described. Routine polysomnographic parameters were measured as described above, except measurements were recorded on a polygraph recorder (model 78E, Grass, Quincy, MA). To measure airflow, the patient breathed through a snug-fitting nasal continuous positive airway pressure (CPAP) mask (Respironics, Murrysville, PA) attached to a heated pneumotachometer (Hans Rudolph, Kansas City, MO) and transducer (Validyne Engineering, Northridge, CA). P_N was measured within the mask with use of a differential pressure transducer referenced to atmosphere. End-tidal PCO₂ was measured via a port on the mask. A thermistor was placed at the mouth to detect oral breathing. P_N was altered in a positive or a negative (subatmospheric) direction by use of CPAP machines (Respironics), one of which had been modified to provide subatmospheric pressure. The subject was allowed to fall asleep on a low level of positive pressure (2-4 cmH₂O), sufficient to abolish inspiratory airflow limitation. Inspiratory airflow limitation was considered to occur when airflow failed to increase, despite increasing respiratory effort, as demonstrated by the characteristic flow waveform. The characteristic waveform pattern consists of increasing inspiratory flow followed by a midinspiratory plateau (7, 22). P_N was then lowered in 2-cmH₂O increments until upper airway obstruction occurred, the patient aroused from sleep, or a maximum pressure of 20 cmH₂O had been applied. Measurements were performed during non-rapid-eye-movement sleep, preferentially during slow-wave sleep (SWS). When measurements during SWS were not possible (because of sleep stage transitions or lack of deep sleep in the laboratory situation), measurements were performed during stage 2 sleep. For purposes of comparison, measurements were performed during SWS and stage 2 sleep in nine subjects; the SWS measurements were used for the final analysis. Pressure-flow curves were constructed by plotting maximal inspiratory airflow (I_max) of flow-limited breaths against P_N. P_N vs. inspiratory airflow curves were fitted by least-squares linear regression. P_crit was defined as the x-axis intercept of the regression line (I_max = 0). Many of the children did not demonstrate flow limitation, even at markedly subatmospheric pressures (Fig. 1B). In these cases, the x-intercept could not be determined without extreme extrapolation, and consequently P_crit could not always be determined. Therefore, S_PF was used to characterize the upper airway response.

View larger version (16K): [in this window] [in a new window]   Fig. 1.   A: maximal inspiratory flow (Imax) vs. nasal pressure (PN) for a 42-yr-old woman. As PN became more negative, Imax decreased. Critical pressure (Pcrit) is 21 cmH2O, and slope of upstream pressure-flow curve is 22 ml · s1 · cmH2O1. B: Imax vs. PN for a 6-yr-old boy. Imax was maintained, despite increasingly subatmospheric PN; thus Pcrit could not be determined. Slope of upstream pressure-flow curve is 2 ml · s1 · cmH2O1.

P_0.1 measurements. P_0.1 was measured during wakefulness at the beginning of the study and then again during stage 2 sleep after the pressure-flow measurements. The respiratory circuit was designed so that the patient could be switched from the P_crit to the P_0.1 circuit by changing a connection at the mask outlet. Because some subjects aroused during this procedure, P_0.1 measurements during sleep could not always be obtained. For P_0.1 measurements the patient wore a nasal mask attached to a pneumotachometer and pressure transducers, as described above. No CPAP or negative pressure was applied; the circuit was exposed to atmospheric pressure only. A balloon valve (Hans Rudolph) was attached to the inspiratory limb of the circuit. A one-way valve was attached to the expiratory limb. During exhalation, the balloon valve was inflated rapidly and quietly via remote control. The mask pressure during the 1st ms of inspiration was recorded; then the valve was opened rapidly to prevent arousal. Five measurements were made for each trial, and the mean value was used.

Statistical analysis. Values are means ± SD unless otherwise specified. The correlation between factors was determined by linear regression. Differences were compared between groups by use of the unpaired t-test (continuous variables) and ² analysis (categorical variables). The difference in S_PF during SWS and stage 2 sleep for the same individual was compared using the paired t-test.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Study population. Twenty-nine subjects were recruited. Two were excluded because of abnormalities on baseline polysomnography (one had an apnea index of 8 and one had an arrhythmia), three declined the second part of the study, and four (all adults) failed to sleep adequately during the pressure-flow measurements. Therefore, 20 subjects (9 children and 11 adults) successfully completed the protocol. Subject characteristics are shown in Table 1. As expected, adults had a greater BMI than children (13). BMI was 25 kg/m² in only two subjects. Although no subject had a history of habitual snoring, mild, intermittent snoring was noted in four subjects (3 children and 1 adult) during polysomnography.

Upstream pressure-flow measurements. Typical pressure-flow curves are shown in detail for an adult (42 yr old; Fig. 1A) and a child (6 yr old; Fig. 1B), and the pressure-flow curves for all the individual subjects are shown in Fig. 2. The configuration of the pressure-flow curve was different for the children. As P_N became increasingly subatmospheric, children were able to maintain upper airway patency and had little decrease in inspiratory flow. Thus the slope of the pressure-flow curve was flat, and P_crit (the x-intercept) could not be determined. Therefore, S_PF was used to characterize upper airway function. In contrast to the children, most adults showed a progressive decline in inspiratory flow with decreasing P_N. The mean S_PF was 8.4 ± 4.9 and 30.0 ± 17.6 ml · s¹ · cmH₂O¹ for children and adults, respectively (P < 0.002). S_PF was <20 ml · s¹ · cmH₂O¹ in all the children but in only three (27%) of the adults. The mean correlation coefficient for P_N vs. I_max was 0.54 ± 0.25 for the children and 0.80 ± 0.23 for the adults (P < 0.05).

View larger version (15K): [in this window] [in a new window]   Fig. 2.   Individual pressure-flow curves for adults (A) and children (B).

1

1

View larger version (17K): [in this window] [in a new window]   Fig. 3.   Slopes of upstream pressure-flow curve plotted against age. There is a significant correlation between slope of upstream pressure-flow curve and age.

View larger version (17K): [in this window] [in a new window]   Fig. 4.   Slopes of upstream pressure-flow curve plotted against body mass index (BMI). , Children; , adults. There is a significant correlation between slope of upstream pressure-flow curve and BMI.

P_0.1 measurements. P_0.1 measurements were obtained in 17 (85%) of the subjects during wakefulness; technically adequate measurements were not obtained in 3 children. There was no correlation between P_0.1 during wakefulness and S_PF (r = 0.28, P = NS). Most subjects aroused from sleep when the respiratory circuit was changed, and therefore P_0.1 measurements during sleep were possible in only eight subjects (6 children and 2 adults). There was a strong inverse correlation between P_0.1 during sleep and S_PF (r = 0.80, P < 0.02; Fig. 5). There was no correlation between age and P_0.1 during wakefulness (r = 0.03, P = NS) or for the smaller group studied during sleep (r = 0.34, P = NS). There was a trend for P_0.1 during sleep to decrease with increasing BMI (r = 0.57, P = NS).

View larger version (17K): [in this window] [in a new window]   Fig. 5.   Slopes of upstream pressure-flow curve plotted against occlusion pressure in 100 ms (P0.1) during sleep. , Children; , adults. There is a significant inverse correlation between upstream pressure-flow curve and P0.1; i.e., subjects with highest ventilatory drive (highest P0.1) were best able to maintain upper airway patency in response to subatmospheric pressure loading (lowest slope of upstream pressure-flow curve).

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

This study examined the inspiratory airflow response to changes in upper airway subatmospheric pressure across the age spectrum. We found a greater change in flow for a given change in pressure in adults. In contrast, children showed little change in flow in response to changes in upper airway pressure; i.e., children were able to maintain upper airway patency, despite increasingly subatmospheric P_N. Thus the S_PF correlated with age. To determine the mechanism of the age-related changes in S_PF, we examined the relationship between S_PF and ventilatory drive and between S_PF and body size. We found a strong inverse correlation between S_PF and P_0.1, suggesting that the central ventilatory drive modulates upper airway muscle function to maintain upper airway patency. However, because P_0.1 during sleep could be measured primarily in the children, it is uncertain to what degree this applies to adults. We also found that S_PF correlated with BMI. In fact, the relationships between S_PF and age and between S_PF and BMI were very similar (Figs. 3 and 4). Because BMI increases with age (13), this increase in S_PF could be due to a BMI effect or other maturational effects, such as changes in structure or in neuromuscular activation. In summary, the response of the upper airway to changes in subatmospheric pressure appears to be fundamentally different between children and adults. Theoretically, it is possible that the increased upper airway protective reflexes present in children are a compensatory mechanism for an anatomically smaller upper airway.

We used techniques similar to those previously used to evaluate children and adults across the spectrum of upper airway obstruction to analyze the upper airway pressure-flow relationship (8, 22, 30, 34). This approach is based on the concept that the upper airway functions as a simple collapsible tube, as predicted by the Starling resistor model (34). According to this model, under conditions of flow limitation, I_max is determined by the pressure changes upstream (nasal) to a collapsible locus of the upper airway and is independent of the downstream (tracheal) pressure generated by the diaphragm. The upper airway can be represented as a tube with a collapsible segment, the resistance of which is zero (Fig. 6). The segments upstream (i.e., nasal) and downstream (i.e., hypopharyngeal) from the collapsible segment have fixed diameters and resistances (R_N and R_HP) and pressures (P_N and P_HP), respectively. R_N can be determined by calculating the reciprocal of the slope of the pressure-flow curve. In this model of the upper airway, inspiratory pressure at the nares is atmospheric and downstream pressure is equal to hypopharyngeal/tracheal pressure. Collapse would occur when the pressure surrounding the collapsible segment of the upper airway (P_crit) becomes greater than the pressure within the collapsible segment of the airway. In the normal subject with low upstream resistance or subatmospheric P_crit, who is breathing at atmospheric pressure, P_HP never drops to P_crit; thus airflow is not limited and is largely determined by negative tracheal (inspiratory) pressure. However, if P_HP falls below P_crit, inspiratory flow reaches a maximum (inspiratory airflow limitation) and becomes independent of downstream pressure swings. Under these circumstances, R_N and P_crit determine I_max, as described by the following equation: I_max = (P_N  P_crit)/R_N. Airflow will become zero (i.e., the airway will occlude) when P_N falls below P_crit. Thus P_crit and R_N can be used to characterize the flow response to changes in P_N. This is analogous to using the slope and x-intercept of the minute ventilation-PCO₂ curve to characterize the sensitivity of the ventilatory response to hypercapnia. In the present study, we found a flattening of the S_PF in children. This precluded the determination of P_crit, because airway collapse did not occur even at maximal subatmospheric P_N. We therefore evaluated the S_PF, which represents the conductance of the upper airway (1/R_N). We found a lower S_PF in children than in adults. This could be due to structural or neuromuscular factors.

View larger version (9K): [in this window] [in a new window]   Fig. 6.   Starling resistor model of upper airway. Upper airway is represented as a tube with a collapsible segment. Segments upstream (nasal) and downstream (hypopharyngeal) from collapsible segment have fixed diameters and resistances (RN and RHP) and pressures (PN and PHP), respectively. Collapse occurs when pressure surrounding airway (Pcrit) becomes greater than pressure within airway.

The finding of a flatter slope in children indicates that the children studied were able to maintain near-constant inspiratory airflow, despite increasing subatmospheric P_N. Thus the pediatric upper airway appeared to dynamically regulate airflow. This maintenance of airflow was not due to increased diaphragmatic effort, because, as discussed above, under conditions of flow limitation, I_max is independent of downstream pressure. The upper airway airflow response may have been due to a decrease in P_crit, but because P_crit was so low, this could not be confirmed experimentally. Alternatively, the response could theoretically be explained by an increase in upstream resistance, because the pediatric airway is smaller than the adult airway. This is unlikely, because there was an ~100-fold difference between the S_PF and the predicted normative values for total pulmonary conductance (36). Although total pulmonary conductance is measured under different conditions (i.e., during wakefulness while the subjects breathes at atmospheric pressure), it seems improbable that technical differences could account for such a huge difference. Therefore, it is likely that dynamic changes in upper airway neuromuscular activation played an important role in the preservation of I_max.

The upper airway has pressure receptors that are sensitive to changes in inspiratory pressure. Previously, it has been shown that the application of subatmospheric pressure during wakefulness results in activation of the upper airway muscles, as demonstrated by EMG (4). However, during sleep in adults, EMG activation often does not occur, and the airway tends to collapse (4, 30). We have now shown that the application of subatmospheric pressure to the pediatric upper airway, in contrast to the adult upper airway, results in preservation of airflow, presumably secondary to upper airway neuromuscular activation.

Neuromuscular factors are known to be important in maintaining upper airway stability. This is demonstrated most obviously by the fact that obstructive apnea occurs only during sleep. If OSAS was due purely to anatomic factors, obstruction would occur during wakefulness and sleep. Studies using denervation of airway dilating muscles or postmortem measurements have shown that, when upper airway muscle function is decreased or absent, the airway is more prone to collapse (5, 38). Thus children with abnormal upper airway neuromotor control, such as children with muscular dystrophy (18) or cerebral palsy (19), often develop OSAS. On the other hand, stimulation of the upper airway muscles decreases collapsibility (31, 32). The present study suggests that the upper airway in children responds to a subatmospheric pressure load by neuromuscular activation, thereby preventing airway collapse. This reflex appears to be more effective in children than in adults.

What causes the upper airway neuromotor activation? Many factors regulate upper airway function, including central ventilatory drive, chemoreceptor afferents, upper airway pressure and flow receptors, pulmonary mechanoreceptors, posture, and sleep state. In the present study, we used P_0.1 as a marker for central ventilatory drive and showed an inverse correlation between P_0.1 and S_PF. This suggests that central drive played a role in preserving upper airway patency. However, the use of P_0.1 as a marker for central ventilatory drive can be criticized. P_0.1 is an index of ventilatory drive to the pump muscles and does not necessarily indicate drive to the upper airway muscles. In fact, animal studies suggest that the diaphragm and upper airway muscles may respond differently to negative pressure (24). The relevance of these animal model studies to humans is unclear, because anesthetized rabbits develop central apnea in response to subatmospheric pressure, whereas this effect was not seen in any of our study subjects. It is known that the upper airway muscles, being accessory muscles of respiration, are affected by ventilatory drive. For example, the upper airway muscles are activated by hypoxemia and hypercapnia. In addition, upper airway muscles are activated by the administration of a pressure load. Previous studies have shown increases in upper airway neuromuscular activity in response to subatmospheric pressure during wakefulness in adults that are attenuated during sleep (4). A number of studies suggest that the upper airway activation in response to pressure loading is centrally mediated. 1) The response of humans to upper airway loading during sleep is very different from the response during wakefulness (4, 14), suggesting a role for the higher central nervous system centers. 2) Functional magnetic resonance imaging studies show activation of central nervous system centers in response to upper airway loading (11). 3) The EMG responses of the upper airway muscles to hypercapnia and inspiratory loading are similar (25). These studies suggest that ventilatory drive plays a role in the upper airway response to subatmospheric pressure. This is confirmed by the present study, in which the P_0.1 during sleep correlated inversely with the degree of inspiratory flow limitation. This finding suggests that the central ventilatory drive plays a key role in maintaining upper airway patency during sleep. In the present study, P_0.1 was measured primarily in children; thus this finding may not apply to adults. However, other studies have shown that children have a higher ventilatory drive than adults (10, 21, 35), which is consistent with their decreased S_PF. The fact that the S_PF correlated with P_0.1 during sleep but not during wakefulness demonstrates that deficits in ventilatory drive may be sleep state specific.

A number of factors may have influenced the age-related changes in S_PF in addition to the ventilatory drive. We found that S_PF correlated with BMI, an index of obesity. Obesity is a common risk factor for OSAS, and weight loss results in a fall in P_crit in obese patients with obstructive apnea (29). Thus structural factors are also determinants of inspiratory airflow during sleep. This is supported by studies demonstrating a lower passive closing pressure of the upper airway in anesthetized children than in adults (15, 16). It is also possible that there is an independent, age-related effect on upper airway function. S_PF correlated with age and BMI. Because BMI is well known to correlate with age (13), multiple linear regression could not be performed, and the relative contribution of each factor could not be determined (1). Thus the study design did not allow us to definitively determine the mechanism for the age-related change in upper airway function. However, the fact that S_PF did not correlate with BMI in the pediatric age group suggests that the role of structural factors in children is limited.

The action of sex hormones may also have influenced the age-related differences in upper airway function. In this study, age, BMI, and P_0.1 played a greater role in determining S_PF in female than in male subjects, suggesting that additional, unstudied factors play a role in men. One such factor may be testosterone-induced changes in the morphology of the male upper airway. In adults, OSAS is twice as common in men as in women (39); the administration of exogenous testosterone may result in OSAS (28) and increase the airway closing pressure (6), and obstructive apnea may recur in adolescent males who were successfully treated during childhood (12). All these facts suggest a role for testosterone in promoting upper airway collapse. This effect would be absent or diminished in prepubertal/pubertal males.

Limitation of methods. We recognize that additional factors could potentially play a role in determining age-related differences in upper airway pressure responses, including developmental changes in upper airway morphology and chest wall mechanics. P_0.1 measurements can be affected by a number of factors, including age-related changes in chest wall compliance and the time constant (37). However, it would not be expected that these mechanical factors would differ significantly between school-aged subjects and adults.

Clinical relevance. In contrast to adults, normal children snore infrequently and rarely have obstructive apneas during sleep (23). This is consistent with the better preservation of upper airway patency in response to subatmospheric pressure noted in the present study. Whereas adults with OSAS tend to have repetitive obstructive apneas, children with OSAS frequently manifest a pattern of persistent, partial upper airway, rather than discrete, apneas (3). Thus the pattern of upper airway muscle recruitment in children with OSAS (17) may be different from that in adults; in children, upper airway muscle activation may be greater, thereby preventing total airway occlusion.

Conclusion. In conclusion, this study has shown that the upper airway response to subatmospheric pressure loads in normal, healthy subjects is modulated by age, ventilatory drive during sleep, and body size. Further studies are needed to delineate the role played by each factor.