INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

OBSTRUCTIVE SLEEP APNEA (OSA) occurs in infants as well as older children, but for infants in their first year of life OSA is not likely to be attributable to adenotonsillar hypertrophy. The exact prevalence of OSA in this age group (infancy) is not well documented because epidemiological studies have concentrated on older children (10). This may be due to the nonspecific nature of symptoms in infants or to difficulties in parental assessment of symptoms that have been present throughout the life of the infant. These include feeding difficulties, profuse sweating, and breath-holding spells (18). Known consequences of severe OSA in infancy include failure to thrive, chronic respiratory failure, and irreversible developmental delay (13). Recent studies suggesting that OSA may be linked with sudden infant death syndrome (42, 27) emphasize the potential importance of OSA during infancy and the need to recognize and to understand the pathophysiology of the disease.

Previous studies of upper airway obstruction in infants have not considered the palate as a potential site for obstruction (26, 34). However, the retropalatal region is a strong candidate as a site of airway collapse at any age, including infancy. It is a site of airway narrowing with little bony or cartilaginous support that is rendered prone to collapse when muscle tone decreases during sleep. The innervation of the palate is complex, involving at least five different neural centers (32). Palatal function in adults is affected by sleep, phase of respiration, and negative-pressure challenges. These factors may also influence the activity of paltoglossus and levator palatini, which are the muscles primarily responsible for palatal position (16, 30, 36, 37, 40). In adults with OSA, the frequency of palatal obstruction is approximately equal to that of retroglossal obstruction (15). Equivalent, detailed information regarding palatal information is not available in infants.

Early development is a period of rapid maturation in sleep patterns and of many respiratory reflexes. Infants have different reflex responses to many stimuli compared with older age groups, often with a predominance of inhibitory influences. Chemical responses are generally weak at birth and subsequently increase, whereas inhibitory reflexes are strong at birth and subsequently decline (42). The mechanics of spontaneous upper airway obstruction have been studied in adults (15), and many studies of airway collapsibility have been performed in older children (21). Few studies in infancy have, however, included the palate as a potential site of upper airway obstruction, so current literature suggests that the retroglossal region is the most common site of obstruction in this age group (35). Many changes occur in chemical and reflex responses during infancy, so it is likely that infants would exhibit differences in the pathophysiological mechanisms of, and responses to, upper airway obstruction. Examples of inhibitory influences on respiration include the strongly inhibitory laryngeal chemoreflex (4), inhibition of respiration during spontaneous sucking and swallowing (2, 22), and reduced activity of the genioglossus at the onset of obstructive events (44).

We hypothesized that the airway obstruction in infants could predominate at the level of the soft palate. In addition, we postulated that inhibitory responses would cause reduced respiratory effort in response to spontaneous airway occlusion and thus differ from the responses exhibited in older children and adults. Finally, we hypothesized that arousal would occur more commonly after obstructive than central apneas, when continued respiratory efforts during obstructive apnea were defined using an esophageal pressure channel.

To evaluate these hypotheses, a triple-lumen pressure catheter was inserted during overnight sleep studies of infants up to 8 mo of age. This permitted examination of the relative contributions of palatal and the retroglossal regions to obstructive apneas that occurred spontaneously during sleep in infants and permitted a detailed examination of mechanical responses to spontaneous airway obstruction in infants.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Study Design

Evaluation of Oropharyngeal Distance

A regression equation was determined from the measurement of nose to oropharynx on MRIs of a reference group of 70 infants. All digital records of cranial MRI available for infants aged 6 mo or less were recalled from the 4-yr period between 1994 and 1998, inclusive. Separate approval for this aspect of the study was obtained from the Ethics Committee of the Royal Alexandra Hospital for Children (RAHC), and this did not include the infants investigated with manometry.

The MRIs were acquired on a Phillips ACS-NT 1.5-T MRI unit (Powertrak 3000, software version 6.1.2). Images were analyzed using Siemens workstation and MagicView software (VA 31 C). This provided pixel-to-measurement correlations that were read in millimeters. Each MRI study was examined to determine whether a midline sagittal section showing airway structures was available, and, if so, the distances from nares to oropharynx were determined. The distance measured were from the nares to the posterior hard palate, from posterior hard palate to the inferior tip of the soft palate, and from the posterior hard palate to the superior tip of the epiglottis (Fig. 1). Adding these distances gave a minimum and maximum range for the oropharyngeal distance, and the midpoint was taken to be the distance from nares to oropharynx for each infant. The indications for MRI and physical characteristics of the infant at the time of the study were determined from the medical records.

View larger version (142K): [in this window] [in a new window]   Fig. 1.   Midline magnetic resonance image of an infant sagittal head section showing the distances measured for this study. A, tongue; B, hard palate; C, soft palate; D, epiglottis; E, trachea.

The regression curve for age against the MRI measured distance was used to approximate the distance from anterior nares to oropharynx in 19 infants undergoing PSG with concurrent pressure measurements. Final analysis demonstrated that body weight was the best predictor for the propharyngeal distance (Fig. 2). Observation of characteristic "swallowing spikes" on the oropharyngeal pressure channel was taken as confirmation that the catheter was placed in the oropharynx (38).

View larger version (19K): [in this window] [in a new window]   Fig. 2.   Raw data and regression line for body weight vs. distance from nares to oropharynx on magnetic resonance image. See text for details.

Analysis of Obstructive Apnea

Subjects. Infants presenting to the Sleep Unit at the RAHC, at least 36 wk postconceptional age and with a sufficiently normal upper airway to allow pressure recording, were invited to participate in this study. Of 45 infants eligible for study, 30 were recruited and 15 parents declined participation. There were 19 successful studies, 4 infants did not tolerate the catheter, and 7 studies failed because of technical difficulties. The 19 infants with successful studies were 15.9 ± 2.0 (SE) wk of age (range 1-36 wk) and included 10 male infants. Thirteen infants were born at term, and 6 were premature with a mean gestation of 31.7 ± 4.9 (SE) wk. Four infants were being investigated for suspected apnea after an apparent life-threatening event, five because a sibling had died from the sudden infant death syndrome, and five because of witnessed apneic events. The remaining five had clinically suspected OSA, four with failure to thrive (weight <3rd percentile for age and gender) and one with micrognathia. Approval for the study was obtained from the Ethics Committee of the RAHC, and signed parental consent was obtained before each study commenced.

PSG. All infants underwent multichannel overnight PSG, concurrent with airway pressure recordings. Sleep studies were performed in the standard manner for the laboratory. A minimum of five channels were included for sleep staging [2 for electroencephalogram (EEG), 2 for electrooculogram (EOG), and 1 for submental electromyogram (EMG)], and five channels were monitored for respiratory analysis [airflow, respiratory effort from chest and abdomen, arterial O₂ saturation (Sa_O2), and transcutaneous CO₂]. Surface electrodes were used to record EEG, EOG, submental and diaphragm EMGs, and electrocardiogram. Nasal airflow was recorded via nasal cannulas (intermediate infant, no. 1615, Salter Laboratories, Arvin, CA) connected to a low-level pressure transducer referenced to atmospheric pressure (model MP45-4 ± 2 cmH₂O, Validyne, Northridge, CA). Thoracic and abdominal respiratory effort were recorded using inductance plethysmography (Non-Invasive Monitoring Systems, Respitrace, Miami Beach, FL). Sa_O2 was recorded on the hand or foot (Ohmeda Biox 3700e Pulse Oximeter, Datex-Ohmeda, Homebush, Australia), and transcutaneous CO₂ levels were recorded from the upper chest or abdomen (model TINA or TCM3, respectively, Radiometer, Copenhagen, Denmark). All sleep study data were digitally filtered and amplified and were recorded onto a digital data acquisition system (Compumedics, Abbotsford, Victoria, Australia), for later analysis. Nasal pressure signal was recorded on the manometry and the PSG systems to assist in accurate time matching of the respiratory events being analyzed.

Airway manometry. A triple-lumen, saline-filled catheter (Critchley Electrical Products, Sydney, Australia) was used to measure airway pressure fluctuations at the nasopharynx, the oropharynx, and the midthoracic esophagus. Standardized apertures were cut in the separate inner lumens at 1.5, 12.5, and 16.5 cm from the sealed, distal end of the catheter. Each inner lumen was then connected to a separate pressure transducer (Transpac IV, Abbot Critical Care, Dublin, Ireland), and patency was maintained with a total saline flow of 12.5 ml/h to prevent secretions from blocking the apertures.

Evaluation of apnea characteristics. Each apnea was identified on the PSG and then identified and analyzed on the manometry recording. Each respiratory event was characterized independently on the PSG and manometry recordings. For manometric analysis, pressure fluctuations were assumed to cease above the site of obstruction. Thus obstruction at the level of the palate would be indicated by absence of a nasal signal with continuing pressure fluctuations in the esophagus and the oropharynx (Fig. 3A). Retroglossal obstruction was indicated by continued pressure fluctuations only at the level of the esophagus, with no fluctuations above this level, including the oropharyngeal and nasopharyngeal sites (Fig. 3B). The use of a multichannel pressure catheter does not permit separation between obstruction isolated to the retroglossal region and obstruction occurring simultaneously at palatal and retroglossal regions, even when a laryngeal lumen is included (15).

View larger version (23K): [in this window] [in a new window]   Fig. 3.   Manometric recording of obstructive apnea identified 2 ways. A: palatal obstruction terminated by arousal. Pressure fluctuations are absent (or minimal) at nasopharyngeal (NPX) trace but continue at oropharyngeal (OPX) and esophageal (ESO) traces. Uncalibrated nasal airflow (Nasal) indicates the apnea indentified on polysomnography. B: retroglossal obstruction with continuing pressure fluctuations only in the ESO trace and a progressive increase in respiratory effort as the apnea progresses.

Statistical Analyses

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Evaluation of Oropharyngeal Distance

The mean distance from nares to oropharynx was 6.2 ± 0.35 cm. Regression analysis using weight in 55 cases (37 = 67% male infants, and 18 = 33% female infants) provided the best correlation with the oropharyngeal distance (R² = 0.67, P < 0.001; Fig. 2). Corrected age also correlated significantly, but less well (R² = 0.56, P < 0.001), and multiple-regression analysis did not improve the correlation. Using this MRI data, the best correlation for the distance to the oropharynx in infants up to the age of 27 wk postterm was

(Fig. 2). where x = body weight.

The catheter distance to the oropharynx (D_oro) in the 19 study infants fell within the 95% confidence intervals of the weight vs. MRI equation for 79% of cases. This result is consistent with our final analysis demonstrating that weight provided the best estimate of nose-to-oropharyngeal distance. In the four outliers, the regression equation tended to underestimate the distance from nose to oropharynx.

PSG

The RDI was not affected by the presence of the catheter [0.79 ± 0.2 vs. 0.72 ± 0.18 log₁₀ RDI, catheter in place vs. catheter out, respectively; not significant (NS)], nor was the frequency of obstructive events (0.31 ± 0.2 vs. 0.31 ± 0.2 log₁₀ ORDI catheter in place vs. catheter out, respectively; NS).

Site and Characteristics of Upper Airway Obstruction

View larger version (15K): [in this window] [in a new window]   Fig. 4.   Percentage of palatal and retroglossal obstruction between nontreatment (open bars) and treatment (solid bars) groups. Only subjects with 4 or more apneas that could be analyzed for site of obstruction were included in the comparison. Values are means ± SE.

Breathing Efforts During Obstructive Apneas

View larger version (12K): [in this window] [in a new window]   Fig. 5.   A: Amplitude of respiratory efforts during obstructive apnea. The efforts were increasing during successive breaths (P < 0.05). B: mean change in esophageal pressure amplitude compared with the mean of the 5 breaths preceding the apnea (100% amplitude). The pressure swing initially fell but exceeded baseline by the third obstructive effort.

Arousal Frequency After Apneas

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

This study presents new data regarding measurements of the upper airway in infants on the basis of MRI. Triple-lumen pressure catheters were placed according to these measurements, to measure the site and characteristics of airway obstruction in infants. Our results demonstrate that the palate may be the primary site of obstruction in young infants with clinically significant OSA, a site that has not been implicated previously in infants with OSA. Use of this new investigative method also provided information about the responses of infants to airway obstruction, including initial inhibition of respiratory efforts and a tendency to EEG arousal after obstructive but not central apnea.

Catheter Placement

To minimize the need for repositioning and refixing the upper airway catheter, we evaluated a new method of predicting the distance from the external nares to the oropharynx. In our physiological studies, catheter placement was predicted from the MRI data then confirmed by observation of swallowing spikes on the oropharyngeal pressure trace. Thus noninvasive, predictive estimate derived from MRI data achieved the goals of expediting catheter placement and minimizing the handling required for each infant as well as avoiding the need for imaging studies. Body weight is a readily available growth parameter, which is particularly important when proposing tools for clinical studies. If a pressure catheter is being used, it remains advisable to confirm catheter placement by the observation of characteristic swallowing pressure waves, at and below the level of the oropharynx (43).

Site and Response to OSA in Infants

The demonstration that palatal obstruction occurs in infant OSA is supported by adult studies using equivalent techniques showing that obstruction occurs at the level of the palate and at the retroglossal region (15, 19). Most infants in this study had some obstruction at palatal and at retroglossal levels. However, infants also tended to show a predominant site of obstruction, and the incidence of palatal obstruction in the treated group was higher than that reported for adult disease (17, 44). The predominance of palatal obstruction in infants with a higher obstructive RDI suggests that palatal dysfunction plays a role in the increased rate of obstruction experienced by those infants.

Obstructive events with retroglossal obstruction may have included collapse (occlusion) of the palatal region as well. Thus the incidence of palatal obstruction may have been even greater than reported, and such events would involve a larger segment of the upper airway. If this were the case, infants with more severe OSA would be expected to have more retroglossal than palatal events, and intrathoracic pressures at the termination of the obstruction would be expected to be greatest after retroglossal events. Given that neither of these sequelae is supported by our data, we conclude that obstruction tended to occur at one or the other site. More extensive pressure catheters (more pressure sites) may still not help to clarify this issue (15), so that imaging (e.g., rapid MRI; Ref. 38) would be required to distinguish isolated palatal from mixed palatal and oropharyngeal collapse.

The pattern of respiratory efforts by these infants during obstruction suggests that they have less respiratory inhibition in response to spontaneous airway obstruction compared with preterm infants. Preterm infants show reduced amplitude or respiratory efforts throughout spontaneous obstructive episodes (9), but the group in the present study exceeded baseline respiratory efforts by the third attempted breath. Externally imposed upper airway occlusion stimulates a different pattern of respiratory effort as measured by diaphragmatic EMG (9), so it is not clear whether the inhibition seen in older children in response to an external load reflects the same inhibitory reflexes (22). In this study, the first effort made during an obstructive event tended to have reduced amplitude compared with the efforts preceding the apnea (Fig. 5B). However, the infants went on to make substantial, and increasingly greater, efforts against their spontaneous upper airway obstruction (Fig. 5, A and B). Thus postterm infants show active mechanical responses to airway obstruction, whether the obstruction has occurred at the palate and/or the retroglossal region.

It is not clear whether adults and older children show this pattern of initial inhibition in response to airway obstruction, although they do exhibit progressively greater respiratory efforts during externally induced airway occlusion (6). Possible mechanisms for this increase in respiratory effort include hypoxia, hypercarbia, and stimulation of mechanoreceptors by the airway occlusion itself (29, 33, 34). Infants with clinically significant OSA had greater amplitude of respiratory efforts, and infants with externally imposed airway occlusion also demonstrated an increased amplitude of diaphragmatic EMG after the event (9). It remains possible that mechanoreceptor-induced responses were stronger in infants with more severe OSA.

Our study provides new data, quantifying the amplitude of efforts among postterm infants before, during, and after apnea. There was no evidence that the infants had persisting upper airway obstruction ("upper airway resistance") between the discrete obstructive events. Older children clearly demonstrate increased work of breathing between the periods of discrete obstruction (14). The lack of correlation between the amplitude of respiratory efforts at baseline with age, log₁₀ RDI, or log₁₀ ORDI is consistent with data in term infants, suggesting that maturation of the response to later obstructed breaths is due to maturation of chemoreflexes (8, 25) and not to pressure generation or the occurrence of paradoxical respiration. We did not measure acute changes in the Sa_O2 or CO₂, but we found no differences in average Sa_O2 or CO₂ values of the treatment vs. nontreatment groups to explain the increased amplitude of postobstructive efforts (Table 2).

These infants showed few arousals at apnea termination, consistent with other studies examining arousal responses (26). Arousal is so common at apnea termination in adults that the two responses are often dealt with interchangeably (17), and arousal from sleep is deemed an important mechanism in averting potentially dangerous situations, such as hypoxia or hypercapnia (28). We found that obstructive events were more likely to precipitate arousals with EEG disturbance than central apnea. We found a higher arousal index than previous studies, but, importantly, the apneic events in the present study were selected on the basis of the fact that they terminated with arousal or desaturation (rather than duration). In addition, an important new finding was that there was no overall difference in the proportion of central or obstructive apneas terminating in arousal when the definition of arousals included disturbances without EEG shifts. The higher proportion of arousals after respiratory events in this study is almost certainly due to the same methodological difference and inclusion of arousals without EEG shifts. The triggering mechanism for apnea termination may be hypoxia, hypercapnia or airway occlusion (3, 33, 34). Gleeson and associates (11) suggest that the point of apnea termination is determined by the amplitude of the respiratory efforts being made. Our results support the concept that termination of apnea is independent of arousal in infants and confirm that it is unlikely that arousal has a fundamental role in apnea termination in infancy (26). The information regarding respiratory efforts in this study suggests that, after term, apnea termination may be related to the amplitude of the respiratory efforts or other mechanoreceptor pathways.

Small pressure fluctuations were sometimes seen above the site of obstruction, which may have reflected some persisting airflow (see Fig. 1), but because the apnea was defined as a fall to 20% of the baseline this does not preclude definition of the event as an apnea. These and other pressure changes observed during the study period were not due to transmission through the catheter wall between the apertures, confirmed by measurement of catheter compliance and the demonstration that this was negligible up to 100 cmH₂O. In addition, the presence of the catheter did not affect the rate of apnea (including obstructive events). These were measured for the periods with the catheter in place and without the catheter and did not change. Thus we confirmed that the presence of the catheter or of saline flow did not impact on the rate or site of upper airway obstruction observed in this study.

Summary