INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Obstructive sleep apnea (OSA) in children differs from that in adults with regard to clinical manifestations, treatments, and probably the causes. Children with OSA show a variety of daytime neurobehavior abnormalities, such as hyperactivity, aggression, and learning problems; excessive daytime sleepiness, which is common in adults, is observed in only a few (1). Obesity is uncommon and failure to thrive is not unusual. Because adenotonsillectomy usually reverses these symptoms and apnea, enlarged tonsils and adenoids are believed to cause OSA in children (1), although lymphoid tissue sizes are reported not to be proportional to the severity of OSA (2). In addition, some of the children with OSA have persistent apnea postoperatively, and some successfully treated by surgery have recurrence of apnea during adolescence (3). Accordingly, it is reasonable to speculate that, in addition to the adenotonsillar hypertrophy, some children with OSA may have neuromuscular and/or anatomic abnormalities in maintaining patent upper airway. Determination of the precise location of pharyngeal obstruction and collapsibility of each pharyngeal segment in children with OSA will increase knowledge of pathogenesis of pediatric OSA.

Marcus and coworkers compared upper airway collapsibility between children with OSA and those with primary snoring (4). They found that the upper airway collapsibility is higher in children with OSA than those with primary snoring. In their experimental settings, however, due to active contractions of upper airway muscles and probable variances of activity level caused by airway size and negative airway pressure magnitude, it is difficult to estimate the contribution of the anatomic and/or the neuromuscular abnormalities to the increased collapsibility in children with OSA.

As previously reported, we have developed a method to separate the neuromuscular factors from the anatomic factors and to evaluate static mechanical properties of the passive pharynx (5, 6). Neuromuscular factors were completely eliminated by producing total muscle paralysis under general anesthesia. Under such circumstances, the anatomy of the atonic pharynx, that is, the intrinsic mechanical properties of the passive pharynx was evaluated by measuring the cross-sectional area of the pharynx under airway pressure manipulation. The static mechanics of the passive pharynx is graphically expressed by a pressure/area relationship plot, which exhibits the maximum area, the pressure at which the area is zero (closing pressure, P_close), and the passive compliance. Anatomic differences are demonstrated as differences of the static pressure/area relationship.

This approach allowed us to test the anatomic hypothesis that children with OSA have a structurally narrowed and/or a more collapsible pharynx as compared with children without OSA. We have tested a similar hypothesis using this approach in adults (5). This hypothesis was tested by comparing static pressure/area relationships of the passive pharynx of children with sleep-disordered breathing (SDB) with those in normal children. Another purpose of this study was to evaluate the possibility of other anatomic abnormalities besides adenotonsillar hypertrophy in children with SDB.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Subjects

Subjects were 16 children with SDB and 13 normal children. Children with SDB were recruited from those referred to the Department of Otolaryngology at Kaihin Chiba City Hospital for suspected SDB.

Children with SDB consulted our sleep clinic for observed apnea and/or difficulty in breathing during sleep with loud snoring, along with tossing in their sleep as evidenced by the parents. Presence of SDB was confirmed by nocturnal oximetry. Based on the children's history, physical examination, and results of nocturnal oximetry, adenotonsillectomy was scheduled for improvement of SDB.

To obtain age-matched normal control subjects, children who were scheduled to have surgery for optical disorders at Chiba University Hospital were invited to participate in the study. Thirteen children and their families agreed to participate in the study. All of the children had a history absent of cardiac and pulmonary diseases. In addition, according to their parents' observations, the children did not snore nor have apneic events during sleep.

The aim and potential risks of the study were fully explained to each subject and/or the parents, and informed consent was obtained from each. The investigation was approved by the hospital ethics committee.

Nocturnal Oximetry

Sleep-disordered breathing was evaluated by a pulse oximeter (Pulsox-5; Minolta, Tokyo, Japan) before surgery. A finger probe of the oximeter was attached by nurses or their parents immediately after sleep, and removed upon awakening in the morning. Digital readings of arterial oxygen saturation (Sa_O2) and pulse rate were stored every 5 s in a memory card. The stored data were displayed on a computer screen to check quality of the recordings. From the stored data, the computer calculated the oxygen desaturation index (ODI, defined as the number of Sa_O2 events that fell below 4% or lower from baseline per hour), the percent of time that the Sa_O2 was less than 90% saturation (CT₉₀), and the mean nadir of oxygen desaturations for each child. Children with SDB were reevaluated by nocturnal oximetry 6 to 12 mo after adenotonsillectomy.

Pharyngeal Endoscopy under General Anesthesia

Preparation of the subjects. Endoscopic evaluation of the pharynx was performed with the subject placed in the supine position on the operating table, with neck in the neutral position. The subject wore a modified anesthetic nasal mask. The possibility of air leaks between the mask and the face was indicated by inadequate increases in airway pressure, particularly when the airway was pressurized at 20 cm H₂O, and actively investigated by feeling around the mask and corrected. General anesthesia was induced by inhalation of gas mixture of nitrous oxide (33%) and sevoflurane (2 to 5%) or intravenous administration of thiopental (4 mg/kg). After intravenous injection of atropine (0.02 mg/kg), a muscle relaxant (vecuronium 0.2 mg/kg, intravenously) produced complete paralysis throughout the experiment. Anesthesia was maintained by inhalation of 2 to 4% sevoflurane in oxygen while the subject was ventilated with positive pressure using an anesthetic machine. Sa_O2, electrocardiogram, and blood pressure were continuously monitored. Sa_O2 was maintained within normal range throughout the experiment. A slim endoscope (FB 10X; Pentax Inc., Tokyo, Japan, 3 mm outer diameter [OD]) was inserted through the nasal mask and naris to visualize the cross-section of the pharynx. A closed-circuit camera (ETV8; Nisco, Saitama, Japan) was connected to the endoscope and pharyngeal images were recorded on videotape. Surgery was performed after the experiment.

Experimental procedure. After inspection of the entire pharynx, the scope tip was positioned to visualize the narrowest site due to hypertrophied adenoid (AD), and/or margin of the soft palate (SP). AD and SP were included in the velopharyngeal airway (VP; retropalatal space). In order to determine the pressure/area relationship of the pharynx, a set of respiratory T valves were turned so that the patient was switched from the anesthetic machine to a pressure controller system capable of accurately producing a constant, preselected airway pressure (Paw) ranging from 20 cm H₂O to +20 cm H₂O in steps of 1 cm H₂O. A blower produced positive and negative Paw, and the level of the Paw was set manually by changing the current to the blower while measuring the Paw using a water manometer. Apnea followed the cessation of mechanical ventilation owing to complete muscle paralysis. Paw was immediately increased and maintained at 20 cm H₂O. While the subject remained apneic for 2 to 3 min, Paw was slowly reduced from 20 cm H₂O to velopharyngeal closing pressure, i.e., the pressure at which the velopharynx was seen to close completely. Sa_O2 remained above 99% throughout this apneic test in all subjects. The apneic tests were repeated at the narrowest site of tonsil (TN) and at the base of the tongue (TG), which were included in the oropharynx (OP; retroglossal space). This procedure of experimentally induced apnea allowed construction of the pressure/area relationship of the visualized pharyngeal segment. Distance between the tip of the endoscope and narrowing site was measured by a wire passed through the aspiration channel of the endoscope.

Data Analysis

To convert the monitor image to an absolute value of cross-sectional area of the pharynx, magnification of the imaging system was estimated at 1.0-mm interval distances between the endoscopic tip and the object in the range of 10 to 30 mm (Figure 1A). At a defined value of Paw, the image of the pharyngeal lumen was outlined on tracing paper (50 g/m²), cut out and weighed (ER120; A & D, Tokyo, Japan). The area of the paper was converted to pharyngeal cross-sectional area according to the distance/magnification relationship. For a constant distance, the area measurements were validated to be accurate within 8% (0.1 ± 4.6%; range, +6.5 to 7.6%) by known-diameter tubes (4 to 9 mm interior diameter [ID]) (Figure 1B).

View larger version (15K): [in this window] [in a new window]   Figure 1.   (A) A distance/magnification relationship in our experimental setting. This allowed us to obtain an absolute value of cross-sectional area. (B) Accuracy of our area measurement for a constant distance.

The measured luminal cross-sectional area was plotted as a function of Paw. We defined closing pressure (P_close) as the pressure corresponding to the zero area. At high values of Paw, relatively constant cross-sectional areas were revealed; therefore, maximum area (A_max) was determined as the mean value of highest three Paw (18, 19, and 20 cm H₂O). As reported previously (5, 7), the pressure/area relationship of each pharyngeal segment was fitted by an exponential function, A = A_max  B · exp(K · Paw), where B and K are constants. A nonlinear least-squares technique was used for the curve fitting, and the quality of the fitting was provided by coefficient R² (SigmaPlot version 2.0; Jandel Scientific Software, San Rafael, CA). A regressional estimate of closing pressure (P'_close), which corresponds to an intercept of the curve on the Paw axis, was calculated from the following equation for each pharyngeal segment: P'_close = ln(B/A_max)K¹. The shape of the pressure/area relationship was described by the value of K and by the half-dilation pressure (P₅₀), i.e., the pressure above P'_close associated with 0.5A_max [P₅₀ = ln(2)K¹]. P₅₀ equals the increment in Paw, above P'_close required to distend the pharynx to one-half the maximum area. When pressure/area relationship is curvilinear, compliance of the pharynx defined as a slope of the curve varies with changes in Paw. Therefore, a single value of compliance calculated for a given Paw does not represent collapsibility of the pharynx for entire ranges of Paw. By contrast, K value and P₅₀ represent a rate of changes in the slope of the curve. When K value is high, small reduction of Paw results in significant increase in compliance leading to remarkable reduction in cross-sectional area. Accordingly, collapsibility of the pharynx increases with increasing K value. We suggest that both P'_close and K values represent collapsibility of the pharynx, whereby the former determines the position of the exponential curve, and the latter characterizes the shape of the curve.

Statistical Analysis

The groups were compared using an unpaired Student t test. Correlation between the variables was assessed by Pearson correlation coefficients. All values are expressed as means ± SD. A value of p < 0.05 was considered to be significant.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Anthropometric data and results of nocturnal oximetry for normal children and children with SDB are presented in Table 1. Since endoscopic evaluation of the pharynx was unsuccessful in two children with SDB owing to inability to ventilate through the nose, these two were excluded from the analysis. Eleven of 14 children with SDB breathed predominantly through their mouth in the awake state. Breathing difficulty in the supine position was apparent in four and noisy breathing in the awake state in four. Three children with SDB preferred to sleep in the prone position and one in the sitting position. Frequent episodes of pneumonia were present in three, including a patient with cerebral palsy. Four were obese for their age and one failed to thrive probably due to severe dysphagia. Two had an apparent macroglossia associated with Down syndrome and Beckwith-Wiedemann syndrome. Kissing tonsils were observed by visual inspection of the pharynx in five. Lateral neck radiography showed severely narrowed airspace (less than 5 mm) at the adenoid level in seven. Nine had a history of recurrent tonsillitis.

Age, height, weight, and body mass index (BMI) did not statistically differ between the groups. All children with SDB had ODI greater than 2 while normal children had ODI less than 2 (7). None of the normal children had a CT  ₉₀ score greater than 1%. Following adenotonsillectomy, the ODI significantly decreased although the values were still significantly higher than those of normal children, and most of the parents reported that their children breathed considerably less loudly during sleep. It is noteworthy that eight of them had ODI greater than 2.0 even after the surgery.

Seven normal children had moderate-sized tonsils with no apparently enlarged adenoids. Marked adenoid and tonsil enlargements were endoscopically observed in all children with SDB. The enlarged tonsils often ascended above the palatal arch resulting in narrowing of the luminal area at the soft palate edge. Similarly, the retroglossal area was often narrowed by tonsils descending to the base of the tongue.

Closing Pressure and Minimum A_max

The primary site of closure was defined as the segment having the highest observed P_close. Observed P_close and distribution of the primary site of closure for each group are shown in Figure 2. Overlap of P_close between the groups was minimal and none of the normal children had positive values of P_close. Mean values of P_close for children with SDB (3.5 ± 4.3, 1 to 14 cm H₂O) were significantly higher than those of normal children (7.4 ± 4.9, 20 to 2 cm H₂O) (p < 0.01). Half of the normal children primarily closed their airways at the soft palate edges and one-third closed at the tongue base. In contrast, two-thirds of the children with SDB primarily closed their airways at levels of enlarged adenoids or tonsils. It is worth noting that the soft palate was responsible for the primary closure in one-third of the SDB children.

View larger version (21K): [in this window] [in a new window]   Figure 2.   Box plots illustrate observed closing pressure for normal children (normal) and children with SDB. The lower and upper boundaries of the box indicate the 25th and the 75th percentiles. A solid line within the box marks median, and the dotted line represents the mean value. Error bars above and below the box indicate the 90th and 10th percentiles. Open circles are outlying points. Each subject data is illustrated by closed circles. Distribution of pharyngeal segment responsible for the closing pressure is shown in the table. Numbers in the table represent the number of the subjects. **p < 0.01 versus normal.

Among the pharyngeal segments, the narrowest segment at the higher Paw was determined and the cross-sectional area at this segment was defined as minimum A_max. Figure 3 shows this minimum A_max in the pharynx for each group. Mean values of minimum A_max of children with SDB (0.40 ± 0.29, 0.12 to 1.08 cm²) were significantly smaller than those of normal children (0.72 ± 0.29, 0.39 to 1.15 cm²) (p < 0.01). Distribution of the responsible sites for the minimum A_max is also shown in Figure 3 for each group. The retropalatal space was the narrowest in most normal subjects while the adenoid made the luminal area minimum in one-half of children with SDB. Interestingly, the remaining half of the SDB children exhibited minimum A_max at the level of the retropalatal or retroglossal area. Because the pharyngeal size is expected to be a function of age and body size, minimum A_max was plotted as the function of age and height for each group (Figure 4). As clearly demonstrated in Figure 4, minimum A_max was directly related to age and height in normal children (open circles) whereas no such association was observed between the parameters in children with SDB (closed circles). Minimum A_max of SDB children was located below that of normal children regardless of age and height.

View larger version (18K): [in this window] [in a new window]   Figure 3.   Box plots demonstrate difference of minimum Amax, which was defined as the cross-sectional area of the narrowest pharyngeal segment at the higher airway pressure, between normal children and children with SDB. Distribution of pharyngeal segment responsible for the minimum Amax is illustrated in the table. Bars as in Figure 2. **p < 0.01 versus normal.

View larger version (11K): [in this window] [in a new window]   Figure 4.   Minimum Amax as a function of age (left) and height (right) for normal children (open circles) and children with SDB (closed circles). Minimum Amax is directly related to age (R2 = 0.679) and height (R2 = 0.749) in normal children. Note that the minimum Amax of children with SDB locate below that of normal children for all ranges of age and height, and that minimum Amax is not significantly correlated with age (R2 = 0.06) and height (R2 = 0.04) in SDB children.

Static Pressure/Area Relationship of the Passive Pharynx

Figure 5 presents static pressure/area relationship of each pharyngeal segment for each child. At higher airway pressures, maximum cross-sectional area was reached creating a plateau on the curve, whereas with decreasing airway pressure, the cross-sectional area progressively decreased. The shapes of the curves, therefore, were curvilinear and similar to those reported in adults (5, 8). Of interest is that some of the normal children exhibited sigmoid shapes, especially at the soft palate.

View larger version (29K): [in this window] [in a new window]   Figure 5.   Static pressure/area relationships of each pharyngeal segment are presented for each of the children. AD = adenoids; SP = soft palate; TN = tonsils; TG = base of the tongue.

No apparent differences in maximum area of each pharyngeal segment were evident between the two groups (Tables 2 and 3); however, cross-sectional area of each segment at the atmospheric pressure was smaller in children with SDB. From these results, children with SDB appear to have higher collapsibility of the entire pharynx than normal children.

Results of Exponential Curve Fitting

Tables 2 and 3 summarize the results of exponential curve fitting for pressure/area relationship presented in Figure 5. The R² values were acceptably high, indicating a reasonably good fit of the exponential function on the pressure/area data of the pharyngeal segments.

Maximum cross-sectional area (A_max ). Although significant differences in minimum A_max between the groups were evident, A_max of each pharyngeal segment did not statistically differ between the groups.

Estimated closing pressure (P'_close). Figure 6 demonstrates the correlation between P'_close and observed P_close at the primary site of closure for all children. Most data points are located on the identity line. Mean values of P'_close of children with SDB were significantly higher than those of normal subjects at all the pharyngeal segments. Not surprisingly, P'_close values of the hypertrophied adenoid in children with SDB were the highest (5.2 ± 4.8, 0.8 to 14.2 cm H₂O), and SDB children had significantly higher P'_close at the tonsil levels than normal subjects. It should be noted that mean P'_close values of the soft palate and the base of the tongue in SDB children (0.1 ± 1.8, 3.5 to 2.0 cm H₂O, and 2.5 ± 4.6, 7.5 to 8.9 cm H₂O, respectively) were close to atmospheric pressure, and significantly higher than normal children (Tables 2 and 3). Two of the SDB children had remarkably high positive P'_close at the base of the tongue (2.2 and 8.9 cm H₂O). One of the two was a patient with cerebral palsy.

View larger version (16K): [in this window] [in a new window]   Figure 6.   Correlation between estimated closing pressure and observed closing pressure at a primary site of closure for all of the children.

Shape of the curve evaluated by K values and P₅₀. K values of the soft palate and tonsils were significantly larger in children with SDB than normals. P₅₀ values, which are reciprocal numbers of K values, were significantly smaller in SDB children than normals. These data indicate that the shape of the fitted pressure/area curve of the soft palate and tonsils was rounder in children with SDB and flatter in normal children. The rounder pressure/area curve in SDB children suggests a nonlinear progressive reduction in area for a given decrease in Paw and that these segments are more vulnerable to collapse.

Correlation of Mechanical Parameters with Nocturnal Desaturation

Among the mechanical parameters of which significant differences between groups were revealed, only closing pressure was significantly correlated with log(ODI) (Figure 7). Age, height, weight, and BMI had no correlation with ODI.

View larger version (16K): [in this window] [in a new window]   Figure 7.   Correlation of mechanical parameters (observed closing pressure, minimum Amax, K values of a site of primary closure) with nocturnal desaturation. Only the closing pressure is significantly correlated with log(ODI). Minimum Amax = cross-sectional area of the narrowest pharyngeal segment at the higher airway pressure. K is a constant obtained by exponential curve fitting of the pressure/area relationship, and represents shape of the curve.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

We evaluated static pressure/area relationships of the passive pharynx in paralyzed children under general anesthesia. The level of enlarged adenoids and tonsils was the primary site of closure in the majority of children with SDB. Closing pressures of children with SDB were positive and significantly higher than those of normal children. Cross-sectional area of the narrowest segment was significantly smaller in SDB children than normals. In addition, the passive pharynx of children with SDB was more vulnerable to pressure changes near closing pressure independent of the size of the pharynx since P₅₀ was significantly smaller in children with SDB than normal children. These results support the hypothesis that children with OSA have structurally narrowed and/or more collapsible pharynx. More interestingly, we found significant structural differences of the retropalatal and retroglossal segments between children with SDB and normal children in addition to adenotonsillar hypertrophy, which suggests that enlarged adenoids and tonsils are not the sole cause for pediatric obstructive sleep apnea.

Design of the Study

Populations of the study. Nocturnal oximetry alone is not appropriate in distinguishing children with SDB from normal children; therefore, classification of the subjects in the present study would be questionable to some extent. However, the differing selection populations may have assisted in minimizing error in classification. Normal children were selected from a group of children who were going to have ophthalmic surgeries, and had no clinical history suggesting SDB. We selected the children with SDB from a group of children who consulted our sleep clinic because of symptoms suggesting SDB. Although the subjects were not precisely examined for incidence of obstructive apnea or hypopnea during sleep, all had markedly enlarged adenoids and tonsils which were believed to cause pediatric OSA (1). Therefore, prevalence of SDB in the former group is expected to be low and prevalence of the latter high. None of the children in the normal group had ODI greater than 2, whereas all of the children with SDB had ODI greater than 2. Accordingly, we believe that our classification of the subjects is valid although the nature of SDB was not examined in the present study.

Evaluation of static mechanics of the passive pharynx. Pharyngeal patency is determined by a complicated interaction among upper airway muscle activities, airway pressure, and intrinsic mechanical properties of the pharynx (9). Our aim in the present study was to examine whether children with SDB have structurally narrowed and/or collapsible pharynx compared with normal children. Because elimination or regulation of the upper airway muscle activities and airway pressure is necessary in the anatomic evaluation of the pharynx, administration of muscle relaxant under general anesthesia was performed in order to eliminate muscle activities. We therefore consider the measurement of the cross-sectional area of the passive pharynx for each of the given airway pressures during an experimentally induced apnea to reflect the anatomic characteristics of the pharynx. Static pressure/area relationships obtained by this or any other methods that alter airway pressure may be influenced by lung volume which is known to influence collapsibility of the pharynx (10, 11). Lung volume systematically varied with changing airway pressure, possibly resulting in alteration of the pressure/area curve shapes. Despite this limitation, we believe that the influence of lung volume does not change our conclusions since changes of lung volume with Paw may not differ between the age- and BMI-matched groups.

Exponential function was employed to fit pressure/area relationship for each pharyngeal segment in normal children and children with SDB as was done in adults (5, 8). The R² values for the curve fitting of the pressure/area relationship were sufficiently high, and the estimated P'_close values were in good agreement with the P_close at the primary site of closure. However, the shapes of the pressure/area curves in Figure 5 are noticeably different in normal children and children with SDB; normal children have more of sigmoidal shape curve especially at the soft palate, while the curve is exponential in children with SDB. In the sigmoid pressure/area relationship, closing pressure was possibly overestimated by the exponential fitting analysis, indicating that closing pressure in normal children would be more negative than estimated. Because K values represent the shape of the entire curve, slope of the pressure/area relationship at lower Paw tends to be overestimated in the sigmoid curve. Therefore, the flatness of the pressure/area relationship in normal children may have been underestimated. Accordingly, estimation of closing pressure and characterization of shapes of the curve may be misleading particularly in normal children. The differences of the intrinsic mechanical properties of the passive pharynx between the groups may also have been underestimated.

Passive versus Active Pharynx

Our findings agree with those of Marcus and coworkers who demonstrated that critical closing pressure (P_crit) of children with OSA was positive and significantly greater than that of children with primary snoring, where their P_crit is analogous to P_close of our measurements (4). Since upper airway muscle activities were not controlled in their study, their P_crit is considered to reflect collapsibility of the active pharynx. Therefore, difference between their P_crit and our P_close could be primarily attributed to state-dependent changes in neuromuscular activity which contributes to pharyngeal patency. Other factors that may also contribute to a difference in P_crit and P_close could be difference in the methods used obtaining each parameter, and in classification of subjects. The difference between P_crit and P_close of children with SDB is relatively small (1 ± 3 versus 3.5 ± 4.3 cm H₂O), which may indicate that neuromuscular compensation to restore narrowed airway in children with SDB is rather weak and incomplete. In contrast, P_crit of children with primary snoring (20 ± 9 cm H₂O) is considerably lower than P_close of children without snoring (7.4 ± 4.9 cm H₂O). In these children without OSA, neuromuscular mechanisms appear to operate efficiently. However, at present, it is unclear whether the differences of neuromuscular contributions to airway maintenance between normal and abnormal children plays a significant role in the pathogenesis of pediatric OSA.

Influence of Aging on Collapsibility of Passive Pharynx

As we recently reported, mean P_close of adult patients with OSA was positive with 1 to 3 cm H₂O (5), values slightly lower than that of children with SDB presented here (3.5 ± 4.3 cm H₂O). Positive P_close appears to be one essential mechanical parameter in the development of OSA in children and adults, although the primary site of closure in children was the level of adenoids and tonsils whereas it was the velopharynx in adults. Considering the significant difference in the number of apneas per hour between children and adults, this minor difference of P_close between these groups may indicate that neuromuscular compensations function to lower the number of apneas per hour in children with OSA.

In normal adults, we found mean P_close to be 4 cm H₂O on average whereas mean P_close of normal children was 7 cm H₂O. Wilson and coworkers endoscopically examined closing pressure of the passive pharynx of infants without upper airway abnormalities after death (12). They found that the mean P_close was 0.7 cm H₂O which was apparently higher than our findings in normal children and adults although postmortem physiological changes may account for the differences. Developmental aspects of upper airway collapsibility are of greater importance in terms of susceptibility to OSA in each age. Although both neuromuscular and anatomic factors may change with maturation, with these factors interacting to maintain airway patency, evaluation of passive pharynx provides knowledge of developmental changes in the anatomic factors.

Recent advances in the understanding of the mechanical properties of the passive pharynx suggest that the upper airway collapsibility changes with age. A relative collapsible pharynx in infants is likely to gain stability by the first year of life, possibly due to the descent of the larynx during this period. The stiffened upper airway is maintained during childhood regardless of growing adenoids and tonsils. The stability may further increase during adolescence with the decay of these lymphoid tissues since the prevalence of OSA is less common during adolescence (1). The stabilized upper airway during childhood and adolescence, in turn, becomes collapsible in the middle-aged. This increase in upper airway collapsibility may be facilitated in the elderly because the prevalence of snoring and SDB is high in this population (13). These developmental changes in collapsibility of the passive pharynx are speculative at present. Therefore, a well-designed study covering all generations from infants to the elderly, by use of a uniform technique to evaluate the passive pharynx is necessary to address this important issue.

Is OSA in Children a Precursor to Adult OSA?

Clinical evidence suggests that some children with OSA have neuromuscular or anatomic abnormalities in maintenance of airway patency in addition to adenotonsillar hypertrophy (1). Marcus and coworkers provided additional evidence for this hypothesis. In a small group of children with OSA, they found that two responders for adenotonsillectomy decreased P_crit to a greater extent than a nonresponder, and the reduced P_crit was still higher than that of children with primary snoring (4). Our results extended their findings and support the hypothesis. We found that children with SDB have more collapsible retropalatal and retroglossal segments than normal children, and the disordered breathing during sleep did not completely normalize by adenotonsillectomy, suggesting that enlarged adenoids and tonsils may not be a sole cause for pediatric OSA. Although this increased collapsibility of the pharynx can be attributed to abnormal bony structure and/or soft tissue surrounding the pharynx, endoscopic evaluation does not address this important question. A more in-depth study to understand the interaction of the pharyngeal lumen and the surrounding structures, possibly by use of magnetic resonance imaging (MRI) and computed tomographic (CT) scan, is necessary.

Many reports indicate that growing adenoids and tonsils are related to manifestation of OSA in children although the size of these masses is not related to the severity of the disease (1). Assuming that the adenoids and tonsils grow to the same extent in children with predisposing high collapsibility of the pharynx and in children with relatively stiff pharyngeal airway, the collapsibility of the former may be augmented more than the latter. The former would result in manifesting apparent OSA, whereas the latter would remain normal or snore but have no apnea. Even after removing the mass by adenotonsillectomy, or decay of these tissues during adolescence, these children have greater risk for developing OSA with advancing age because of anatomic abnormalities of the pharynx. Although further investigations are necessary to test the hypothesis that OSA in children is a precursor of adult OSA, our results strongly support this hypothesis.

Clinical Implications

Any surgical procedures for OSA must be based on anatomic evaluation of the pharynx while nasal continuous positive airway pressure (CPAP) may reverse OSA regardless of site of pharyngeal obstruction. Static evaluation of the passive pharynx in adult patients with OSA successfully predicted outcome of uvulopalatopharyngplasty (UPPP) (14). Patients who exclusively demonstrated a site of primary narrowing at the level of the velopharynx responded to surgery. Similarly, our approach, which evaluates anatomic abnormalities of the pharynx, is applicable to selection of candidates for surgical corrections other than adenotonsillectomy in pediatric OSA. Because adenotonsillectomy is useful in reversing disordered breathing in the majority of children with OSA, this anatomic evaluation may not be necessary for all children with OSA, but may be particularly useful in children without apparent enlarged adenoids and tonsils, and in failures of adenotonsillectomy. Children with OSA having a primary narrowing at the level of the soft palate may possibly respond to UPPP. Glossectomy or mandibular osteotomy may be indicated in children with a high closing pressure at the base of the tongue.

In conclusion, we demonstrated that primary closure occurred at the level of hypertrophied adenoids and tonsils in children with sleep-disordered breathing. Closing pressure of the site of primary closure was significantly correlated with nocturnal desaturation. In addition to adenoids and tonsils, predisposing anatomic abnormalities of the entire pharynx are likely to contribute to manifestation of nocturnal disordered breathing.