The relationship between maxillary constriction and the etiology of obstructive sleep apnea (OSA) is not clear. This prospective case-control study compared maxillary morphology in 94 dentate subjects (47 OSA and 47 control subjects), using upright lateral cephalograms and study models. Each subject had height, weight, and neck circumference measurements recorded and underwent an orthodontic examination. An upright lateral cephalogram and dental impressions were obtained. All data were analyzed using the SPSS statistical package applying nonparametric tests at the 5% level of significance. Male and female subjects were examined separately, and statistically significant differences were found between the cephalometric measurements for OSA and the control subjects. The palatal angle was more obtuse in male OSA subjects (P < .05). The PNS-posterior pharyngeal wall was shorter (P < .05) and the soft palate longer in female OSA subjects (P < .05). Minimum palatal airway widths were significantly reduced in both male (P < .01) and female (P < .001) subjects. In the comparison of study model measurements, palatal heights in OSA subjects were greater (P < .05). Thus, maxillary morphological differences do exist between OSA and control subjects, supporting their role as a etiological factor.
Obstructive sleep apnea (OSA) is a potentially life-threatening disorder, estimated to affect 3.9% of men and 1.2% of women.1 OSA is diagnosed using a combination of a sleep history, supporting questionnaires, a clinical examination of the upper airway, and overnight sleep monitoring. Snoring and excessive daytime sleepiness are the most common presenting complaints. The Epworth sleepiness scale (ESS) questionnaire elicits the likelihood of falling asleep in eight different situations.2 Overnight polysomnography is considered the “gold standard” for the diagnosis of OSA.
Guilleminault et al3 studied the relationship between maxillary constriction and the etiology of OSA and reported a familial tendency of narrow, high palates in the relatives of OSA patients. Cistulli and Sullivan4 have shown a high prevalence of OSA among patients with Marfan's syndrome.
Continuous positive nasal airway pressure has traditionally been the choice of treatment for moderate to severe OSA.5 Mandibular advancement splints are used in the management of subjects with mild to moderate OSA.6 Recently, the use of rapid maxillary expansion (RME) as a treatment modality for OSA has been put forward.6,7
RME increases the width of the maxilla, reduces nasal resistance,7 and increases the intranasal capacity. Hershey et al8 reported a 45–55% reduction in nasal airway resistance, which was maintained after removal of the appliance. Wertz9 concluded that RME cannot be justified for the purposes of increased nasal permeability unless there is a relative maxillary arch width deficiency and the obstruction lies in the lower portion of the nasal cavity. Cistulli et al10 investigated the effect of RME as a treatment in six young adults who underwent RME and elective surgical assistance and in four who were treated with RME alone. In seven cases, the apnea-hypopnea index (AHI) was reduced to normal values, and the authors attributed this to an improvement in nasal airflow, tongue posture, and soft palate function.
The role of the maxilla in the etiology of OSA is not well described, and this deficiency needs to be addressed to develop evidence-based practice. The aims of this study were to evaluate the role of the maxilla in the etiology of OSA and to determine any differences between sexes.
MATERIALS AND METHODS
The test sample consisted of randomly selected Caucasian adults, referred from the Royal National Throat, Nose and Ear Hospital to the Royal London Dental Hospital (RLH). The control group consisted of Caucasian adults recruited from the Restorative Department waiting room of the RLH, accompanying parents or grandparents of patients attending the Orthodontic Department (RLH) and partners of the test group.
The inclusion criteria for the test group were over 18 years of age with a diagnosis of OSA confirmed by overnight polysomnography (AHI > 5 per hour of sleep) and the presence of at least six teeth in the maxillary arch. The same criteria were applied to the control group except they had to have an ESS score <9 (exclude daytime sleepiness) and no reported snoring. The exclusion criteria for both groups were edentulous or partially dentate in the buccal segments, a history of orthodontic treatment, or a history of reconstructive/orthognathic surgery to the head and neck or pharyngeal surgery.
Ethical approval was obtained from the East London and City Health Authority research Ethics Committee. When all subjects were fully informed of the purpose and nature of the study, written consent was obtained.
Data collection was divided into four sections:
Medical and sleep history including an ESS (Appendices 1 and 2).
Clinical examination with recordings of the subject's height, weight, and neck circumference. For each subject, their body mass index (BMI) was calculated (kg/ m2).11
Study cast analysis: measurements were taken from the maxillary cast using a pair of digital calipers (Mitutoyo Corporation, Kawasaki, Japan code number 500-321), with a resolution of 0.01 mm (Figures 1 and 2). The presence of a posterior transverse discrepancy was recorded as a buccal crossbite.12
Lateral cephalogram analysis: before the radiograph was taken, a thin layer of barium sulfate paste was applied to the dorsum of the tongue to enhance soft tissue identification. To standardize hyoid position, radiographs were exposed at the end of expiration, in the natural head position with the teeth in light occlusion.
A single operator traced the radiographs in a random order. The radiographs were orientated with the maxillary plane horizontal, and the landmarks used are shown in Figures 3 through 5. Each tracing was randomly digitized sequentially twice to a tolerance of 0.2 mm, using a GTCO Accutab digitizer (GTCO Corporation, Columbia, Md) and a customized software program. All data were converted to life-size using the known magnification for each radiograph. Linear, angular, and area measurements of the oropharynx, soft palate, and tongue were calculated.
Method error was calculated by repeat measurement of the 20 randomly selected models and cephalograms after an interval of two weeks.13 Random error was calculated using Dahlberg's equation14 and by determining a reliability index.13 Systematic error (bias) was assessed using the method of Houston.13
All data were analyzed using the SPSS statistical package (version 7.51 SPSS Inc., Chicago, Ill). Nonparametric tests were applied. For cephalometric and study model linear and angular measurements, the Mann-Whitney U-test was used, with statistical significance stated at the 5% level or less. For skeletal pattern and buccal crossbite parameters, the Fisher's exact test was used.
The model measurement error study produced an index of reliability generally greater than 93%. The error study revealed that 12 lateral cephalometric measurements had a Dahlberg value >1. Seven of these related to gonion, a point with a large envelope of error.15 No systematic error (bias) was detected.
The male OSA and control subjects were well matched for age (Table 1), but the female OSA subjects were approximately seven years older than the control group (Table 2). All OSA subjects exhibited a higher BMI score, which was statistically significant (P < .05) in female subjects. Neck circumference, which reflects obesity, was statistically greater in both male (P < .001) and female (P < .05) subjects. The median ESS score for both male and female OSA subjects was statistically significant (P < .001), confirming that OSA subjects alone demonstrated excessive daytime sleepiness.
Measurements reflecting the anteroposterior (AP) and vertical skeletal dimensions of the maxilla showed the palatal angle to be more obtuse in the OSA group (P < .05), and the PNS-posterior pharyngeal wall distance significantly reduced in female OSA subjects (Figure 4, Tables 3–4).
Although not statistically significant, Sella-Nasion, SNA, and upper posterior facial height were reduced, whereas measurements for basion-PNS, upper anterior facial height, and palatal angle were all increased in the OSA subjects. No significant differences were found in the distribution of skeletal classification for OSA and control subjects (Table 5).
In the OSA groups, the minimum palatal width was smaller both in men (five mm; P < .01) and in women (3.38 mm; P < .001). In female OSA subjects, the soft palate was longer (4.02 mm; P < .05), whereas in male subjects, the difference was not statistically significant. The thickness of the soft palate was marginally greater in OSA subjects than in controls; however, this did not reach statistical significance (Tables 6 and 7).
Study model measurements
In the OSA subjects, the palatal height was larger (P < .05) at the level of the first premolar, second premolar, and first molar (Figure 2). There were no significant differences in maxillary depth or intertooth distances (Tables 8 and 9).
Posterior transverse discrepancy
No statistically or clinically significant transverse dental differences were present between the groups (Table 10).
Age and sex distribution
The prevalence of OSA within the sample size compared favorably with published studies.1 OSA and control subjects were well matched for age and sex. Male and female subjects were reported separately, unlike previous studies that have either combined the sexes or described results from male subjects only. Female OSA subjects were approximately seven years older than their control group. This supported the finding that the prevalence of OSA increases with age to approximately 50–55 years in men and 60–65 years in women, after which it appears to decline.1,16,17
The male subjects were classified as overweight and the female subjects were classified as obese based on the BMI and neck circumference. These findings are consistent with other studies.18–21 In the present study, however, a greater frequency of obesity and lower prevalence of OSA were present in women, which is in agreement with the findings of Mohsenin.22 Walker et al23 concluded that OSA severity was less weight dependent in women than in men.
The neck circumference recorded was statistically greater in both male (P < .001) and female (P < .05) OSA subjects, and other studies have reported similar find-ings.18,24–26 This is consistent with the findings of Whittle et al,26 who demonstrated that normal men have a greater proportion of neck fat than normal women. These highlight the need to separate the sexes when evaluating the oropharyngeal dimensions in OSA subjects.
In the current study, the OSA group was obese and the control group classified as overweight. Predictive models of OSA have been proposed in which the degree of skeletal abnormality in a patient is inversely related to the degree of obesity required to cause OSA.27–30
Lateral cephalometric measurements
Traditionally, OSA has been associated with mandibular retrusion and hence a skeletal Cl II pattern.31 However, in the current study, there was no significant differences in the distribution of skeletal classification (SNA, SNB) in OSA or control subjects (Table 5) in agreement with several previous studies.18,20,32 Analyses did not reveal a significant difference in the AP or vertical planes (or both) for the maxilla between OSA and control subjects, nor between the sexes, confirming the findings of Seto et al.18
In the AP plane, the PNS-posterior pharyngeal wall measurements were reduced in all OSA subjects as previously reported by Battagel and Johal,27 Battagel et al,33 and Seto et al.18 The anterior cranial base was shorter, resulting in a more retrusive facial complex and reduced pharyngeal airway.28,32,34,35 Seto et al18 reported a statistically significant shorter ANS-PNS in OSA subjects. However, in their study, over 25% of the OSA subjects failed to have radiographs taken.
The palatal angle (ANS-PNS-uvula) was significantly more obtuse in male OSA subjects (P < .05). Because all other maxillary skeletal measurements detected no significant differences, the discrepancy appeared to be with the soft palate and its orientation. Tsuchiya et al28 proposed OSA subtypes based on the degree of obesity, the severity of OSA, and craniofacial data. In the current study, both the OSA and control subjects had BMI values classifying them as overweight, and significant differences were detected in the soft, but not hard tissue, measurements. Seto et al18 found significant skeletal discrepancies, which may be explained by the difference in BMI values between their OSA and control groups, a smaller sample size, or incomplete data analyses (or all).
In the present study, the soft palate of all OSA subjects was larger and occupied significantly more of the upper pharyngeal space than in the control subjects. Soft palate length was increased in OSA subjects, which is in agreement with other studies.18,27,36 Furthermore, soft palate length increases with age,37 and thus studies must match control subjects for age. Soft palate area was increased in all OSA subjects, but the values were less than reported by Battagel and L'Estrange.20 The latter study only investigated men, with a significantly younger control group and consequent lower BMI values. An increase in soft palate thickness in all OSA subjects was not statistically significant but was in agreement with the studies of Battagel and Johal27 and Battagel et al.33 This increase in thickness may be attributed to an extension of adipose tissue from accumulations in the neck, although this is debatable.38 Palatal airway width was significantly reduced in OSA subjects, confirming the findings of previous studies.20,33,35
Study model measurements
The relationship between maxillary constriction and the etiology of OSA is not clear. To date, only three studies have used models to assess the transverse arch dimensions of the maxilla in OSA subjects: Cistulli et al,39 Kushida et al,30 and Seto et al.18 Cistulli et al39 examined the influence of maxillary morphology in a sample of 13 patients with Marfan's syndrome. However, the suitability of Marfan's patients as OSA subjects is debatable.
The present study found a statistically significant difference in palatal heights between OSA and control subjects at the level of the first premolar, second premolar, and molar. However, these results were not in agreement with those of Cistulli et al39 and Seto et al,18 who did not find differences in palatal heights between OSA and control subjects. This was surprising in the study of Cistulli et al39 because a high vaulted palate is very characteristic of Marfan's syndrome. This was confirmed by Seto et al,18 who reported that palatal height measurement alone was not a reliable indicator of maxillary constriction. Kushida et al30 included palatal height in their predictive morphometric model for OSA, but they defined palatal height as the distance from the dorsum of the tongue to the highest point of the palate.
The present study could find no significant differences in maxillary depth or intertooth distances. Again, this was not in agreement with previous studies.18,39 The predictive morphometric model for OSA uses maxillary and mandibular intermolar distances measured from the second molar tooth,30 with no allowances made for this tooth being absent. Seto et al18 also found that the maxillary depth was significantly reduced in OSA subjects.
Posterior transverse discrepancy
In this study, no statistical significance was detected between OSA and control subjects in transverse dental dimensions. However, Seto et al18 in their study, reported statistically significant posterior transverse discrepancies but did not clarify whether they were skeletal or dental. Previous studies have shown that OSA is not exclusive to one skeletal classification.20,32 Thus, it remains very questionable as to whether maxillary constriction can be a primary etiological factor in OSA and more importantly, whether treatment directed at this is evidence based. The present study, therefore, cannot support the use of RME as a treatment modality for subjects with OSA in the absence of any obvious maxillary constriction. Furthermore, such practices must engage in objective outcome measures, such as follow-up sleep studies to determine their therapeutic value.
Maxillary morphological differences exist between OSA and control subjects, identifying a potential etiological role in OSA.
Statistically significant differences exist between OSA and control subjects, in both maxillary skeletal morphology and oropharyngeal dimensions.
Study model analyses demonstrated that OSA subjects differ significantly from control subjects in palatal height measurements.
We thank Mr A Ferman, Biometrics Laboratory Supervisor, for his assistance with the radiographic digitizing program and Dr W Marcenes, Statistics Department, Royal London Hospital, for his help with translation skills in the language of statistics.
Questionnaire for OSA and control subjects
Name: Height (m):
Sex: Weight (kg):
Group (OSA or control): BMI:
Date of Birth: Neck circumference (cm):
Upper airway obstruction:
Cardiac anomalies (angina, myocardial infarction, cardiac failure)
Preexisting breathing disorders:
Teeth present: Molar relationship:
Overjet: Posterior buccal crossbite:
Overbite: Scissor bite:
The Epworth sleepiness scale
How likely are you to doze off or fall asleep in the following situations (in contrast to feeling just tired)?
Even if you have not been in some of these situations recently, try to work out how they would have affected you.
Use the scale to choose the most appropriate number for each situation:
0 = would never doze
1 = slight chance of dozing
2 = moderate chance of dozing
3 = high chance of dozing
Situation Chance of dozing
Sitting and reading ………
Watching TV ………
Sitting, inactive in a public place
(eg, a theatre or a meeting) ………
As a passenger in a car for an hour
without a break ………
Lying down to rest in the afternoon
when circumstances permit ………
Sitting and talking to someone ………
Sitting quietly after a lunch without
In a car, while stopped a few
minutes in traffic ………
Total score ………
Corresponding author: Ama Johal, BDS Hons, MSc, Department of Orthodontics, Queen Mary's College, University of London, New Road, Whitechapel, London, UK E1 1BB (firstname.lastname@example.org)