Obstructive sleep apnoea (OSA) is a well known and common sleep-related breathing disorder characterised by upper airway resistance leading to obstruction and repeated sleep fragmentation. Due to its impact on quality of life and its association with many comorbidities, OSA is recognised as a serious public health issue by governments worldwide.1 As the aetiology of OSA is multifactorial, there is broad interest across many health professions.
Of the professions, orthodontics is well suited to managing OSA because a large component of practice revolves around treatment using dento-facial orthopaedics. Craniofacial disproportion in skeletal retrognathia of one or both jaws, dolicocephalic headform, constricted maxillae and open bite tendencies are among the reported characteristics that can predispose to OSA. Alarmingly, a large proportion of patients exhibiting the characteristics of OSA remain undiagnosed.2,3 Given the focus on the craniofacial complex and the routine diagnostic processes undertaken during treatment planning, orthodontists are primed to identify these factors.
One area of interest is the assessment of airway morphology and OSA severity based on the apnoea-hyponea index (AHI). Many studies have investigated the association between airway shape, volume, cross-sectional area and diameter with AHI based on 2D and 3D radiographic imaging. Despite this, the reported associations are wide-ranging from little to statistical significance.4–10 Common areas of focus include the assessment of the posterior airway space, tongue size, adenoid mass, minimum cross-sectional area and soft palate length from cephalometric and cone beam analysis. However, once these parameters were identified, there is less guidance regarding the best course of action. This may be partly due to a paucity of research correlating morphology with actual treatment requiring a multidisciplinary emphasis. The deficiency in co-ordinated care is a problem identified by all health professions involved with OSA.
The problem has recently been identified by the American Association of Orthodontists (AAO) citing a lack of consolidated guidelines. The AAO therefore established a task force to review protocols for OSA and resulted in the publication of the AAO whitepaper as an authoritative guide for orthodontists.11 The underpinning AAO philosophy is the multidisciplinary management of OSA under the guidance of a medical specialist.11
Among the medical specialties, a significant proportion of an ENT surgeon’s clinical practice is devoted to patients presenting with sleep-related breathing issues with the most common being OSA.12 Kizirian et al. estimated that almost one-third of referred patients to an ENT surgeon exhibited sleep disorders.13 With many options in the ENT’s treatment armamentarium, clinical judgement is required to determine whether surgery and its nature, is appropriate. The challenge in managing OSA is due to the heterogeneous nature of the disease as the differential diagnosis of airway obstruction is not singular but rather diverse and complex.
In line with AAO’s recently reported principles of OSA management, there is a benefit in identifying those factors which might influence a medical specialist to consider surgical intervention. Awareness of these factors might ultimately improve communication, triage, consent, and the management of OSA between health professionals in a co-ordinated effort. Therefore, the present study has a multidisciplinary focus and a primary aim to assess the relationship between morphology and surgical intervention in a cohort of patients presenting with an increased BMI and a confirmed diagnosis of OSA. A secondary aim is to revisit the relationship between morphology and OSA severity.
This was a retrospective study focusing on patients previously diagnosed with moderate to severe OSA and who had seen and received surgical treatment by an Ear Nose Throat (ENT) specialist practising in Perth, Western Australia. Ethics approval for the study was obtained from the Human Research Ethics Committee of The University of Western Australia (RA/4/1/5392).
The present study consisted of consecutive records of the most recent 32 patients (26 males and 6 females) with a diagnosis of moderate (apnea-hypopnea index (AHI) ≥ 15) to severe OSA (AHI ≥30) presenting and receiving surgical treatment by an ENT surgeon. All patients were within the age range of 30 to 63 years; had body mass index (BMI) in the overweight range (BMI ≥ 25); exhibited an AHI ≥15 per hour of sleep; and had a pre-surgical low dose CT scan. The mean BMI was 30.7 and 29.7 for males and females, respectively. Patient-specific data concerning polysomnography, age, gender, height, BMI and the type of ENT procedure were obtained from the surgeon.
For each patient, the pre-treatment 3D-CT scans obtained from the same radiological clinic were sourced and a lateral composite image was generated. The images were digitally stored in DICOM (Diagnostic Imaging and Communications in Medicine) format.
The lateral head images were imported into
DICOM CT images for each patient were loaded into 3dMDVultus (3dMDVultus software, 3dMD, Atlanta, GA), and a semi-assisted software program provided 3D visualisation of the airway (Figure 2). This formed the basis of the morphological analysis. The patient’s orientation was corrected prior to airway analysis, with the head centred and positioned so that the palatal plane was parallel to the horizontal plane. The airway was outlined by a measurement grid, from the lowermost border of C4 extending to the palatal plane (ANS-PNS), defined visually by the investigator.14 A 2 mm distance slice along the airway was determined, and measurements of cross-sectional areas and linear dimensions in two directions (anteroposterior and mediolateral) were calculated by the software for each slice from the axial view. Although maximum and minimum cross-sectional areas were identified, only the minimum value was recorded. In addition, measurements of airway length and total airway volume was determined by the software.
The accuracy and reliability of 3D airway analysis using 3dMDVultus software were previously documented by Schendel and Hatcher.15 In the present study, to test the reliability of measurements, 6 patient’s images were randomly selected and measurements were repeated by the same operator 3 times at two-week intervals. The intraclass correlation coefficient (ICC) was used to assess the consistency of quantitative measurements. The results were highly reproducible, with the lowest ICC of 0.883 for airway length, and the remaining measurements ranged from 0.97 to 0.999.
Statistical analyses were performed using the R statistical computer software (R version 3.0.0). Descriptive statistics were analysed for each variable and sample mean values were determined for gender (Table I). The relationships between AHI and BMI with other variables were tested using simple linear regression. A multiple regression analysis was applied to determine the association of volumetric and morphological findings with the severity of OSA and prediction of BMI. Additionally, the connection between the types of treatment proposed with other variables was evaluated using logistic regression. An accepted
Descriptive statistics.
Total | Male | Female | |||||||
---|---|---|---|---|---|---|---|---|---|
N | Mean | SD | N | Mean | SD | N | Mean | SD | |
BMI (kg/m2) | 22 | 30.4 | 3.8 | 16 | 30.7 | 3.4 | 6 | 29.7 | 5.2 |
Cross SA (mm2) | 32 | 51.2 | 25 | 25 | 56 | 24.4 | 6 | 34.7 | 22.1 |
ML Diameter | 32 | 14.2 | 8.97 | 25 | 13.25 | 4.96 | 6 | 19.31 | 18.3 |
AP Diameter | 32 | 3.48 | 3.6 | 25 | 3.72 | 3.84 | 6 | 2.81 | 2.79 |
Volume (cm3) | 32 | 12.8 | 7.3 | 25 | 13.6 | 7.6 | 6 | 9.9 | 6 |
Length (mm) | 32 | 83.4 | 7.9 | 25 | 84.2 | 7.1 | 6 | 80.3 | 11.3 |
SNA (deg) | 32 | 81.9 | 5.1 | 25 | 81.8 | 4.5 | 6 | 80.1 | 5.7 |
SNB (deg) | 32 | 78.9 | 5.2 | 25 | 78.9 | 5.2 | 6 | 77.3 | 4.3 |
ANB (deg) | 32 | 3.6 | 2 | 25 | 3.8 | 2.1 | 6 | 2.6 | 1.4 |
PAS-NL (mm) | 32 | 23.8 | 3.8 | 25 | 24.2 | 3.9 | 6 | 21.8 | 3.8 |
PAS-UP (mm) | 32 | 3.4 | 2.4 | 25 | 3.6 | 2.6 | 6 | 2.5 | 1.3 |
PAS-UT (mm) | 32 | 7.6 | 3.7 | 25 | 7.9 | 4 | 6 | 5.8 | 2.1 |
PAS-TP (mm) | 32 | 9.7 | 5.5 | 25 | 9.8 | 5 | 6 | 9.7 | 7.9 |
Length BN (mm) | 32 | 42.5 | 4.9 | 25 | 42.9 | 5.1 | 6 | 41.3 | 4.6 |
ANS-PNS (mm) | 32 | 51.6 | 6.5 | 25 | 53.2 | 6.1 | 6 | 45.4 | 4 |
HYOID C3 (mm) | 32 | 37.2 | 6.6 | 25 | 38.5 | 6.7 | 6 | 32.8 | 3.9 |
GT HYOID (mm) | 32 | 39.3 | 5.3 | 25 | 40 | 5.4 | 6 | 36.6 | 4.8 |
Length SP (mm) | 32 | 39.7 | 8.3 | 25 | 40.5 | 8.2 | 6 | 36.6 | 9.6 |
SP thick (mm) | 32 | 11.2 | 1.9 | 25 | 11.4 | 1.5 | 6 | 10.3 | 3 |
All measurements were represented in millimetres, angles between lines in degrees and individual areas in square millimetres. Descriptive results were depicted as mean ±SD.
Table I outlines key descriptive statistics relating to airway measurements of the cohort of 32 obese patients with a history of OSA.
The relationship between AHI with other variables was tested using simple linear regression. Of the variables which included age, gender and airway measurements (linear, angular, cross-sectional area and volumetric), only age was significantly associated with AHI (
Relationship between OSA, BMI and other variables.
Obstructive Sleep Apnoea | Body Mass Index | |||
---|---|---|---|---|
Variable Category | Coefficient | p-value | Coefficient | p-value |
Sex | -0.607 | 0.9555 | 1.022 | 0.591 |
Age | 1.183 | -0.017 | 0.8637 | |
BMI | 0.954 | 0.5035 | 0.024 | 0.5035 |
Cross-sectional Area (mm2) | -0.113 | 0.5172 | 0.001 | 0.9802 |
ML Diameter (mm) | 0.032 | 0.9476 | -0.11 | 0.1924 |
AP Diameter (mm) | -0.149 | 0.9001 | 0.339 | 0.153 |
Volume (cm3) | -0.614 | 0.3156 | -0.036 | 0.7335 |
Length (mm) | -1.039 | 0.0751 | -0.152 | 0.1603 |
SNA | 0.957 | 0.3126 | 0.158 | 0.3699 |
SNB | 0.983 | 0.2754 | 0.268 | 0.1006 |
ANB | -0.6 | 0.8023 | -0.332 | 0.4512 |
PAS-NL | -0.738 | 0.5182 | 0.132 | 0.5572 |
PAS-UP | -1.408 | 0.4375 | 0.63 | |
PAS-UT | 0.17 | 0.8852 | 0.35 | 0.1352 |
PAS-TP | -0.037 | 0.9632 | 0.37 | |
Length of BN | -0.941 | 0.3443 | 0.244 | 0.207 |
ANS-PNS | -0.389 | 0.6317 | 0.082 | 0.5711 |
HY0ID-C3 | 0.023 | 0.9751 | 0.355 | |
GT-HYOID | 0.623 | 0.4508 | 0.357 | |
Length SP | -0.435 | 0.4788 | 0.242 | |
SP thick | 1.939 | 0.4016 | 1.939 | 0.4016 |
A multiple regression was fitted using the volumetric and morphological measurements of the airway (Table III). Using backwards selection and a variable selection cut-off of
Predicting key variables using multiple regression analysis.
Obstructive Sleep Apnoea (Moderate to Severe) | ||||
---|---|---|---|---|
Coefficient | Std. Error | t value | p-value | |
(Intercept) | -159.404 | 90.859 | -1.754 | 0.0933 |
AP Diameter | 2.655 | 1.596 | 1.663 | 0.1104 |
SNA | 2.99 | 1.114 | 2.685 | 0.0135 |
ANB | -3.729 | 2.492 | -1.496 | 0.1488 |
PAS-UP | -4.916 | 2.526 | -1.946 | 0.0645 |
Length of BN | -1.778 | 1.012 | -1.757 | 0.0928 |
SP thick | 4.072 | 2.291 | 1.778 | 0.0893 |
R2=0.3182, p-value: 0.1655 |
Body Mass Index | ||||
---|---|---|---|---|
Coefficient | Std. Error | t value | p-value | |
(Intercept) | -3.05286 | 5.09058 | -0.6 | 0.5577 |
Cross-sectional Area | -0.14134 | 0.02645 | -5.344 | <0.0001 |
ANB | -0.77577 | 0.2616 | -2.965 | 0.0096 |
PAS-NL | 0.46682 | 0.14898 | 3.134 | 0.0068 |
PAS-UT | 0.57712 | 0.14678 | 3.932 | 0.0013 |
HYOID-C3 | 0.18169 | 0.08468 | 2.146 | 0.0487 |
GT-HYOID | 0.52891 | 0.10859 | 4.871 | 0.0002 |
R2=0.8166, p-value: <0.0001 |
Despite all patients exhibiting a high BMI, there was no correlation with BMI and severity of disease based on AHI (
The relationship between BMI and other variables was tested using simple linear regression (Table II). Cross-sectional linear measurements at different levels of the airway, PAS-UP (
A multiple regression was also fitted to all volumetric and morphological measurements (Table III). Variables were selected using a backward variable selection process and a cut-off
A list of treatments performed for each patient is shown in Table IV. Table V outlines the relationship between two common procedures, uvulopalatopharyngoplasty (UPPP) and tongue channelling (TC), with airway morphology.
List of treatment for each patient.
Treatment done | Patients (N) |
---|---|
CPAP only | 3 |
CPAP + Septoplasty | 3 |
Maxillo advancement surgery | 1 |
Septoplasty | 1 |
UPPP + Septoplasty | 1 |
UPPP | 3 |
UPPP + lingual tonsilectomy | 1 |
UPPP + Transpalatal advancement | 6 |
UPPP + tongue channelling + transpalatal advancement | 10 |
UPPP + tongue channelling | 3 |
Relationship between surgical procedure and other variables.
Uvulopalatopharyngoplasty | Tongue Channelling | |||
---|---|---|---|---|
Variable Category | Odds Ratio | p-value | Odds Ratio | p-value |
Sex | 2.111 | 0.4664 | 1 | 1 |
Age | 0.91 | 0.1003 | 1.017 | 0.7039 |
BMI | 1.049 | 0.7136 | 0.889 | 0.3506 |
AHI/RDI | 0.97 | 0.137 | 0.987 | 0.4605 |
Cross-sectional Area (mm2) | 1.007 | 0.7036 | 0.964 | 0.1086 |
ML Diameter (mm) | 1.03 | 0.7204 | 0.939 | 0.3984 |
AP Diameter (mm) | 0.991 | 0.9342 | 0.758 | 0.1395 |
Volume (cm3) | 1.206 | 0.1341 | 1.073 | 0.2565 |
Length (mm) | 1.163 | 0.0515 | 1.219 | |
SNA | 0.923 | 0.341 | 0.965 | 0.6372 |
SNB | 0.97 | 0.6967 | 0.937 | 0.39 |
ANB | 0.789 | 0.2732 | 0.922 | 0.6657 |
PAS-NL | 0.862 | 0.2475 | 1.079 | 0.484 |
PAS-UP | 1.097 | 0.6425 | 0.756 | 0.2213 |
PAS-UT | 1.159 | 0.256 | 0.765 | |
PAS-TP | 1.188 | 0.0994 | 0.956 | 0.5316 |
Length of BN | 0.915 | 0.2936 | 1.047 | 0.5454 |
ANS-PNS | 0.977 | 0.7112 | 0.96 | 0.4946 |
HY0ID-C3 | 0.967 | 0.595 | 0.896 | 0.1047 |
GT-HYOID | 0.929 | 0.3763 | 0.899 | 0.1545 |
Length SP | 1.025 | 0.6604 | 1.01 | 0.8414 |
SP thick | 0.91 | 0.6893 | 0.776 | 0.2509 |
No variables were statistically associated with receiving UPPP. Airway length, PAS-UT was associated with TC surgery (
Volumetric and morphologic variables and tongue channelling surgery.
Coefficient | Std. Error | t value | p-value | |
---|---|---|---|---|
(Intercept) | -21.33241 | 8.85015 | -2.41 | 0.0159 |
Cross-sectional Area | -0.05826 | 0.02799 | -2.081 | 0.0374 |
Length | 0.28107 | 0.10884 | 2.582 | 0.0098 |
Coefficient | Std. Error | t value | p-value | |
---|---|---|---|---|
(Intercept) | 8.37026 | 3.92575 | 2.132 | 0.033 |
PAS-UT | -0.36943 | 0.17189 | -2.149 | 0.0316 |
GT-HYOID | -0.15917 | 0.08435 | -1.887 | 0.0592 |
The association between increased age and AHI is consistent with previous published studies.16–18 Although the mechanisms are not fully understood, the most widely accepted proposition suggests that aging impacts on airway anatomy, by an increase in pharyngeal fat deposition and a reduced control of the pharyngeal dilator muscles. Malhotra et al. examined a group of mixed subjects with normal BMI (<30) separated by age (35 year group vs 65 year group) and demonstrated that fat deposition, lengthening of the soft palate and a reduction in airway dimensions occurred as a function of increased age.17 These outcomes occurred independent of BMI, suggesting a process related to physiological maturation. Other studies associated longer airway length with OSA in older individuals.10,19,20 An additional factor is the aging effects on sensorimotor impairment, a reduction in neuromuscular control and the degradation of protective airway reflexes.21 Anatomically, there are reports linking an age-related reduction in genioglossus activity as a contributory factor to pharyngeal collapse.17
In concert, many studies have demonstrated a wide range of age-related effects and their association with OSA. While the impact and contribution of each of these factors are not well understood, aging is foremost a non-modifiable risk factor that can affect the success of short to medium-term treatments.
Of all known risk factors for OSA, obesity is perhaps the most investigated. Many cross-sectional studies have consistently demonstrated a relationship between increased body mass and OSA.18 Estimates place significant OSA in 40% of obese individuals and approximately 70% of patients with OSA are obese.22 In a prospective study by Peppard et al., a 10% weight gain was associated with a 6-fold increase in the probability of developing OSA.23 Other studies reported similar characteristics including a positive correlation between fat distribution in the neck and increased AHI.24–26
Specific morphological traits in the airway have been associated with increased BMI. Several investigations, including Rodenstein et al.7 and Schwab et al.,27 demonstrated that patients with OSA had a predominant pharyngeal dimension in the anterior-posterior axis as opposed to control subjects without a sleep disorder exhibiting a predominant transverse axis.28 At the very least, this provides a foundation for lateral cephalometric analysis as a screening tool. The underlying mechanisms involved have been attributed to the pattern of fat infiltration around the airway and the influence of the dilator reflexes in awake patients.29 Despite the increased BMI of patients examined in the present study, BMI was not correlated with the severity of OSA (
It is worthwhile to speculate why BMI was not correlated with OSA severity in the present study. A possible explanation for this outcome was that all patients participated in weight loss programs required by the ENT surgeon before surgical intervention, underscoring the significance of weight loss in combatting OSA.32 Additionally, the pathogenesis of OSA is multifactorial and its expression in certain individuals is highly variable.33 Finally, Mayer demonstrated that the architecture of the pharyngeal lumen was influenced by a combination of BMI and reflexes while patients were awake including the time of CBCT exposure.28
Measurements of the upper airway were consistent with published research investigating potential correlations with OSA. The calculated mean of the velopharyngeal airway [PAS – UP] in the present study was smaller compared to controls of similar non-OSA patients suggesting that airway narrowing may be a risk factor for OSA.34,35 Reports of increased length and width of the soft palate leading to reduced minimal cross-sectional airway in OSA patients was also featured in the present study and consistent with Walsh et al.9 These authors concluded that the minimal cross-sectional area in the velopharyngeal region was significantly reduced in OSA patients compared with matched controls.9 Similarly, Johal et al. compared patients with OSA and a control group and found that soft palate length was significantly increased (
Despite these similarities, the present study showed no statistically significant correlation with airway measurement and increased AHI. This may be attributed to the study methodology and the multifaceted nature of OSA. Reports consistently show significant variation in upper airway shape and soft tissue coverage between individuals.28 Stratemann et al.38 using 3D airway analysis demonstrated significant variability in upper airway morphology, luminal shape and soft tissue outline within the same individual. Exposing CBCT in the supine position in which a lateral cephalogram was constructed does not physiologically replicate the actual condition of the upper airway and associated reflexes, or lack of, during sleep.39 Static images derived from CBCT scans do not realistically represent the dynamic and dysfunctional nature of OSA. Non-anatomical parameters such as neuromuscular function, ventilatory response and arousal thresholds are potential implicating factors that are not quantified from morphological assessment.
Patients involved in the present study had undergone not only one but a combination of different surgical procedures. PAS-UT and airway length were shown to be significantly associated with patients who received tongue channelling surgical treatment; and some measurements were found to be significant predictors of tongue channelling including airway length, PAS-UT and cross-sectional area. This was based on a comprehensive ENT examination including the evaluation of the airway using optic nasendoscopy enabling a detailed subjective assessment of the airway and surrounding soft tissues. These factors likely led to the decision for surgical intervention. Despite this, the findings of the present study do not support advocating treatment based on morphology alone.
A limitation of the present study was the small sample size, a lack of objective parameters leading to the decision to provide a surgical procedure and a non-OSA control group. BMI data from the surgeon consisted only of those with high BMI and the data were not amenable to division into age groups or based on severity. This would have permitted a more meaningful analysis against other factors. Furthermore, soft tissue characteristics would be more accurate if the airway was captured while patients were asleep to represent the expected reduction in pharyngeal muscle activity and reflexes. Additionally, attempts to standardise breathing during the procedure is a consideration. A study with these features would be a closer representation of the dynamic and dysfunctional nature of OSA. A greater qualitative focus involving more specialists would provide additional insight for referring health professionals to identify factors which might predispose the prescription of a particular treatment modality. An ideal situation would be for the primary clinician to screen for risk factors and to better predict and appreciate more targeted strategies for the individual. This would improve the initial examination, communication, informed consent, triage, and referrals and represents the AAO’s ultimate philosophy of managing OSA.11
In principle, airway assessment from 2D and 3D radiographic imaging obtained as part of orthodontic treatment planning is an opportunistic and invaluable screening tool for OSA. However, other clinically related factors need to be considered before a treatment modality is proposed. The judicial use of 3D imaging is recommended rather than the blanket prescription of high-dose imaging for the sole purpose of diagnosing OSA. The study is confirmational in that morphology does not reflect OSA severity based on AHI. Additionally, there is little correlation between the range of treatments provided by a medical specialist and morphological assessment. Therefore, potential risk factors identified from imaging should raise the index of suspicion and prompt referral to a medical specialist for multidisciplinary co-ordinated care, in line with the AAO’s principles of OSA management.
The aetiology of OSA is complex and multifactorial in which known risk factors and their relative contribution to this sleep disorder vary between individuals. In a group diagnosed with increased BMI and moderate to severe OSA, airway measurements and morphology had little correlation to OSA severity based on AHI even though some characteristics were observed. There was also a limited correlation between actual treatment provided by a medical specialist and airway assessment from 2D and 3D imaging. Therefore, additional factors need to be considered before a treatment modality is proposed. Of all variables assessed, the present study confirms the significant association between increased age and OSA severity. The judicial use of radiographs is recommended and, when available, is an invaluable screening tool to identify individuals at risk of OSA which should prompt referral to a medical specialist for multidisciplinary co-ordinated care.