INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Keywords: upper airway; MR imaging

The upper airway undergoes progressive changes during childhood. These include growth of tissues composing the upper airway (1), and functional changes in neuromuscular tone and ventilatory drive (4).

Accelerated growth of the adenoid or tonsils during these years (2, 3) could predispose children to upper airway obstruction, particularly during sleep. Paradoxically, however, given a smaller airway in comparison to adults, most children snore very little and have less obstructive apneas and hypopneas (8). The reasons for these differences are not well understood. Studies assessing the mechanical properties of the upper airway (7) as well as studies assessing ventilatory responses and responses to resistive loading during sleep (5) suggest that the upper airway of children is more resistant to collapse.

In addition to any mechanical or neuromotor factors contributing to this effect, we reason that changes in anatomic dimensions of tissues composing the airway during childhood play an important role in maintaining airway patency. To this end we used magnetic resonance imaging (MRI) to characterize the upper airway and surrounding tissues in normal children during development to study the anatomic factors that may contribute to airway patency during sleep in children. We hypothesized that changes in anatomic dimensions of the tissues comprising the upper airway during development serve to maintain airway patency by one of two mechanisms: first, by modulation of tissue growth rates to offset possible overgrowth of other tissues such as the adenoid or tonsils, or second, by maintaining constant proportional growth of all tissues.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The study was approved by the institutional review board of the Children's Hospital of Philadelphia. Informed consent was obtained from each subject's parent and assent was obtained from children older than 6 yr of age.

Subjects

Ninety-two children with normal growth and development were selected from the population who underwent head or neck MRI during a 1-yr period at the Children's Hospital of Philadelphia. Exclusion criteria included: (1) likelihood of obstructive sleep apnea (OSA) as assessed by a standard questionnaire with apnea scores  1; (2) adenoidectomy or tonsillectomy in the past; (3) evidence of a brain tumor, brain anomaly, or a seizure disorder requiring therapy; (4) genetic disorders associated with any craniofacial anomaly; (5) chronic respiratory disease such as asthma or bronchopulmonary dysplasia.

Sleep Questionnaire

A questionnaire regarding symptoms of sleep-disordered breathing, based on the one developed by Brouillette and coworkers (9), was used to assess the likelihood of OSA in our population. Accordingly, no subject with score < 1 would be expected to have OSA; a score between 1 and 3.5 is considered indeterminate; and a score > 3.5 is considered highly predictive of OSA.

Magnetic Resonance Imaging

MRI studies were performed in the Department of Radiology at the Children's Hospital of Philadelphia. All studies were performed under sedation. Sedation included intravenous pentobarbital of 2 mg/kg increments until sleep was achieved. A maximum of three doses or 200 mg was administered. All subjects were monitored continuously by pulse oximetry and observed until recovery (~ 1 h).

MRI was performed with a 1.5T Siemens Vision system (Iselin, NJ). Images were acquired using a commercially available anterior-posterior volume head coil. The patient's head was positioned supine in the soft tissue Frankfort plane (tragus of the ear to orbital fissure) perpendicular to the table. All images initially included a rapid spin echo sagittal localizing scan to confirm that the field of view (FOV) and centering were appropriate (spin echo repetition time [TR] = 650, echo time [TE] = 14, 192 × 256 matrix, slice thickness 3 mm, 1 acquisition [acq], and FOV = 20 to 24 cm).

Sequential T1-weighted spin echo sagittal sections were obtained spanning bilaterally from the midline. The following parameters were used for the sagittal images: TR = 650, TE = 14, 192 × 256 matrix, slice thickness 3 mm with distance factor 0, 1 acq, FOV = 20 to 24 cm, rectangular field of view (RECFOV) 8   /  8. Similarly, axial sections were obtained, spanning from the orbital cavity to the larynx: TR = 650, TE = 14, 192 × 256 matrix, slice thickness 3 mm with distance factor 0, 1 acq, FOV = 20 to 24 cm, RECFOV 6   / 8.

Image processing and anatomic measurements. Measurements from MRIs were made using image processing software VIDA (Volumetric Image Display and Analysis, Department of Radiology, University of Iowa, Ames, IA) (10).

Sagittal measurements. From a midsagittal T1-weighted image of each subject, we determined the lower face sagittal skeletal growth by measuring the mental spine-clivus length (Figure 1A). Mental spine-clivus length was defined as the distance between mental spine (point of insertion of genioglossus to the mandible) and the clivus passing through the soft palate centroid (the average point of the spatial positions of all pixels within the soft palate region). Along the mental spine-clivus line we performed a series of linear measurements that included tongue oblique width, soft palate oblique width, nasopharyngeal airway oblique width, and adenoid oblique width (Figure 1A). Measurements were later expressed as percentage of mental spine-clivus distance.

View larger version (66K): [in this window] [in a new window]   Figure 1.   (A) Mid-sagittal T1-weighted image with mental spine-clivus oblique line shown (white line). Linear measurements were obtained along this line. (B) Axial T1-weighted image at the level of maximal tonsillar cross-sectional area. Note the transverse intermandibular line (white line) passing the center of each tonsil. Linear measurements were obtained along this axis. Maximal oropharyngeal width was obtained on this plane (dotted white line).

Axial measurements. Lower face axial skeletal growth was determined by the intermandibular length obtained from an axial T1-weighted image of each subject at the level of maximal tonsillar cross-sectional area (Figure 1B). The intermandibular length was defined as the distance between the medial aspects of both mandibular rami along a transverse line passing through each tonsil centroid. Along this line we performed a series of linear measurements that included intertonsillar width, bilateral tonsillar width, bilateral fat pad width, and bilateral pterygoid width (Figure 1B). All measurements were later expressed as percentage of the total intermandibular distance. We later determined the oropharyngeal width on this plane, defined as the maximal oropharyngeal width on a line parallel to the intermandibular line (Figure 1B, dotted line).

Data Analysis

Linear regression analysis was performed to express the relationship between age of subjects and lower face skeletal growth as determined by mental spine-clivus length and intermandibular length obtained from the midsagittal and axial images, respectively. Similar analysis was performed for the various components of the upper airway obtained from our sagittal and axial linear measurements. In addition, we computed the correlation coefficients between subject's height with mental spine-clivus length and intermandibular length. Finally, the correlation coefficient was computed between intertonsillar width and maximal oropharyngeal width, and between maximal oropharyngeal width and age.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

During the 12-mo study period, 92 preadolescent children with normal growth and development were selected from 3,542 patients who underwent head or neck MRI out of a total 4,734 MRI performed at the Children's Hospital of Philadelphia. Ages ranged from 1 to 11 yr with a mean of 5.69 ± 2.51 years. The age distribution and the number of subjects are shown in Figure 2. Forty-three (47%) of the subjects were males. Distribution by ethnic group was 71% white and 29% African American. Mean height was 114.4 ± 18.9 cm (z-score [reference 11], 0.49 ± 1.22, range 2.18 to 3.75); mean weight was 22.7 ± 9.5 kg cm (z-score [11], 0.39 ± 1.00, range 1.51 to 3.49); and mean body mass index was 16.7 ± 2.6 kg/m² (z-score [11], 0.30 ± 1.07, range 2.30 to 3.21). All had normal development and cognitive function, normal brain images, intact tonsils and adenoid, and no respiratory disorders or craniofacial anomalies. Twelve percent of subjects had history of snoring (ranging from rare to frequent to always). However, all had an apnea score less than 1, and as a group had a mean score of 3.2 ± 0.8. The primary indications for head MRI were as follows: migraine/headache (32 subjects), single seizure/febrile convulsion (26 subjects), temper tantrum/ behavioral changes (9 subjects), eye injury/optic atrophy/nystagmus (7 subjects), head concussion (6 subjects), vomiting/ vertigo (5 subjects), tremor (3 subjects), torticollis (3 subject), and scoliosis (1 subject). Thus, none of these clinical indications would be expected to affect the upper airway anatomy.

View larger version (54K): [in this window] [in a new window]   Figure 2.   Age distribution and number of subjects studied.

On the midsagittal plane, skeletal growth determined by mental spine-clivus length related linearly with age (r = 0.86, p < 0.001, Figure 3A). Similar relationships were noted for tongue oblique width (r = 0.78, p < 0.001), soft palate oblique width (r = 0.20, p = 0.05), nasopharyngeal oblique width (r = 0.28, p < 0.01), and adenoid oblique width (r = 0.47, p < 0.001). These tissues maintained proportional growth over the entire age range in relation to skeletal growth (Figure 3B). Accordingly, the following proportions were noted: tongue (57.9 ± 3.2%), adenoids (20.8 ± 3.5%), soft palate (11.3 ± 2.2%), and nasopharyngeal airway (8.4 ± 3.7%).

View larger version (30K): [in this window] [in a new window]   Figure 3.   (A) Correlation for age and mental spine-clivus length for each subject (closed circles), and cumulative correlations for oblique measurements of tongue, soft palate, nasopharyngeal airway, and adenoid. (B) Tongue, soft palate, nasopharyngeal airway, and adenoid lengths presented as percentage of the ratio of mental spine-clivus length. r = 0.86; p < 0.001.

On the axial plane, skeletal growth determined by intermandibular length showed a linear relationship with age (r = 0.78, p < 0.001, Figure 4A). For other tissues, we noted the following correlation with age: intertonsillar width (r = 0.1, p = 0.3), tonsils width (r = 0.46, p < 0.001), parapharyngeal fat pads width (r = 0.14, p = 0.1), and pterygoids width (r = 0.40, p < 0.001). These tissues maintained constant proportion to axial skeletal growth (Figure 4B), tonsils (46.0 ± 5.9%), pterygoids width (35.2 ± 4.4%), intertonsillar distance (10.9 ± 5.4%), and parapharyngeal fat pads (7.9 ± 4.1%).

View larger version (30K): [in this window] [in a new window]   Figure 4.   (A) Correlation between age and intermandibular length for each subject (closed circles), and cumulative correlations for intertonsillar width, bilateral tonsils width, bilateral parapharyngeal fat pads width, and bilateral pterygoid width. (B) Intertonsillar width, bilateral tonsils width, bilateral parapharyngeal fat pads width, and bilateral pterygoid width presented as percentage of the ratio of intermandibular length. r = 0.78; p < 0.001.

A significant linear correlation was noted between subjects' height and mental spine-clivus length (r = 0.89, p < 0.001), and between subjects' height and intermandibular length (r = 0.76, p < 0.001). A significant correlation was noted between intertonsillar width and maximal oropharyngeal width (r = 0.45, p < 0.001). However, we did not note a correlation between maximal oropharyngeal width and age (r = 0.02, p = 0.81).

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The upper airway is a complex structure that is outlined by the soft tissues forming the nasopharynx and oropharynx within the skeletal boundaries of the mid and lower face. It undergoes morphologic change as the result of growth of the skeleton and surrounding tissues as well as functional changes throughout childhood (1). All can impact on the size, shape, and stability of the upper airway, particularly during sleep when a significant decline in ventilatory drive and respiratory muscle tone occur (12).

Our study is the first to use MRI to assess linear dimensions of the upper airway structure during development. We screened our subjects to ensure normal growth and development as well as absence of OSA using a standard questionnaire (9). Although some question the use of this questionnaire in specific populations as OSA subjects (17), we feel confident using it in our subjects as they had no risk factors for OSA and because none had indeterminate scores.

Our data suggest that the mid and lower face skeleton continue to grow linearly along our defined axes in the sagittal and axial planes throughout childhood from 1 yr through 11 yr of age. Moreover, the soft tissues defining the airway, including lymphoid tissue, grow in constant proportion within the skeletal boundary during this period.

Some methodological issues deserve an initial comment. There are no standard MRI measurements to assess the upper airway in children. We wished to assess the airway as well as the adjacent tissues: lymphoid, muscle, fat, and bone that could impact on airway size during development. Therefore, we have chosen two orthogonal planes that clearly allow us to evaluate the various tissues of the nasopharynx and oropharynx, that could be easily identified by anatomic landmarks, and that cut across regions of the airway that are typically restricted by lymphoid tissue.

Measurements of the nasopharynx region were performed on a midsagittal plane in a similar manner to studies in adults (10, 18). This plane is easy to recognize according to midline structures such as pituitary stalk, nasal septum, and epiglottis. Length measurements on this plane were performed along the mental spine-clivus line. We have chosen this line because it approximates the oblique dimension of the mandible and because it was noted to have a strong linear correlation with age and height of our subjects (r = 0.86 and 0.89, respectively).

Measurements of the oropharynx region were performed on an axial plane at the level of maximal tonsillar cross-sectional area. This is in distinction from the adult literature choosing a level at minimal oropharyngeal airway cross-sectional area or retroglossal area (10, 18, 19). We defined the foregoing level because of the speculated role of the tonsils on airway size in children. Our measurements on this plane were performed along the transverse intermandibular line because it crosses the center of the tonsils and the main oropharyngeal structures, and because it is limited on both sides by the rami of the mandible that are clear anatomic landmarks and boundaries for soft tissue growth.

It should be emphasized that our measurements were performed under sedation that could have reduced upper airway muscle activity, thus affecting our anatomic measurements, especially airway dimensions due to muscle relaxation. However, because a similar protocol was used for all subjects, we can assume that measurements would be affected in a similar fashion.

Of the 10 sagittal and axial measurements, 8 showed a significant and positive linear relationship with age; however, the parapharyngeal fat pad width and intertonsillar width did not show such a correlation. It is possible that these measurements do not follow a linear relation with age. Absent or low correlation with age for some of these measurements may also be attributed to our methodology, including the reference planes chosen for our measurements, adding variability to the small magnitude of these measurement combined with low changes over time. For example, the parapharyngeal fat pad width grows from an average of 4.5 mm at 1 yr to 5.9 mm during a 10-yr period, having an average growth rate of 0.14 mm per year. This small increment is masked by the resolution of our MRI technique with a pixel resolution of 0.78 mm.

MRI is accurate and reliable, and in comparison to radiographs that are two-dimensional projections, MRI provides an intrinsically scaled, three-dimensional image of all tissues composing the upper airway structure. Moreover, MRI provides superior resolution for soft tissues compared with other techniques commonly used to assess the upper airway structure in normal children and in children evaluated for OSA (1, 20).

We found that the dimension of the nasopharyngeal airway was maintained at 8.4 ± 3.7% of the sagittal skeletal dimension, and the intertonsillar width that correlated with oropharyngeal airway width (r = 0.45, p = 0.001) was maintained at 10.9 ± 5.4% of the axial skeletal dimension, both measurements being independent of age. Similarly, the thickness of the adenoid and tonsils remain 20.9 ± 3.4% and 46.1 ± 6.0% of the sagittal and axial skeletal dimension, respectively, independent of age. Finally, constant proportional thicknesses of the tongue, soft palate, pterygoid, and parapharyngeal fat pads to the skeletal dimension was noted. These findings suggest that the thicknesses of the tissues defining the upper airway exhibit growth rates that maintain a constant proportion to the increasing mid and lower face skeletal dimensions defined by mental spine-clivus length and intermandibular length.

These findings are intriguing in two aspects. First, we did not find age-related growth rate changes for any tissues forming the airway, particularly the adenoid or tonsils. Second, we noted that airway dimension was maintained proportionally to skeletal growth between 1 and 11 yr of age.

There are few studies in the literature addressing developmental changes of the upper airway. Fujioka and colleagues (1) evaluated the relationship between adenoid and nasopharynx dimension (AN ratio) along the sagittal plane during childhood. They retrospectively analyzed 1,398 random lateral neck radiographs from children who were examined in their hospital and who were not found to have abnormalities in chest X-rays or sinus disease. They noted an AN ratio of 0.33 at 1.5 mo of age that increased to 0.55 at 1.25 yr, and reached its highest value of 0.59 at 4.5 yr, and then declined to 0.52 at 12.5 yr of age. Finally, they noted a further decline at puberty to an AN ratio of 0.38. These findings showing a relatively constant AN ratio between 1 yr of age and puberty are in agreement with the constant proportional adenoid size derived from sagittal images in our subjects, 1 to 11 yr of age.

In another study using lateral neck cephalometry, Jeans and colleagues (2) longitudinally evaluated the areas of the nasopharynx, nasopharyngeal airway, and nasopharyngeal soft tissues of 41 children with no history of otolaryngeal or pulmonary disorders (22 girls, 19 boys). They found a generally steady growth rate of the total nasopharynx area in girls and boys between 3 and 11 yr of age. However, they noted a mild decrease in nasopharyngeal airway area in both groups between 3 and 5 yr that was parallel with mild accelerated growth of the soft tissues between 3 and 5 yr in males and between 3 and 6 yr in females. For both groups, nasopharyngeal airway area continued to increase linearly thereafter, up until 11 yr of age with no further increase in nasopharyngeal soft tissue area.

The findings by Jeans and colleagues suggesting possible overgrowth of the adenoids at the expense of airway area in normal boys and girls between 3 and 5 yr and 3 and 6 yr, respectively, seem reasonable because most children who are evaluated and treated for OSA are in this age group (25). However, these findings are in contrast to ours during the same age range. It is possible that methodological differences in our technique or our measurements are responsible for these differences.

Our findings confirm a recent report by Vogler and coworkers (29) that assessed the thickness of the adenoid pad using a similar methodology to ours. They evaluated a single midline MRI obtained from 189 normal patients (age 1 d to 92 yr) who underwent a brain scan. They found that the adenoid pad continued to grow linearly throughout the first decade of life and is at maximum size between 7 and 10 yr of age, and then progressively diminished until 60 yr of age. Our findings in prepubertal children are similar to their report in the same age range. Moreover, we show in our study that the tonsils exhibit the same pattern of constant proportional growth as the adenoid during this period.

We speculate that growth of the adenoidal and tonsillar tissues in children who present with OSA will not follow the proportional changes we have noted in normal children in the present study. Variation in tissue sizes at the extremes of those found in this study probably predisposes some subjects to development of OSA.

In summary, this study indicates that the mid and lower face skeleton continues to grow linearly along the sagittal and axial planes throughout childhood and that the soft tissues defining the upper airway, including the adenoid and tonsils, grow in constant proportion to the skeletal structures during this period. The maintenance of proportional growth among these tissues ensures airway patency throughout childhood and probably contributes to airway stability in normal children, particularly during sleep.