Analyses of aerodynamic characteristics of the oropharynx applying CBCT: obstructive sleep apnea patients versus control subjects

Abstract

Objectives:

To determine the most relevant aerodynamic characteristic of the oropharynx related to the collapse of the upper airway in obstructive sleep apnea (OSA) patients; and to determine the correlation between the most relevant aerodynamic characteristic(s) of the oropharynx and anatomical characteristics of the oropharynx in OSA patients.

Methods:

31 mild to moderate OSA patients (mean ± SD age = 43.5 ± 9.7 years) and 13 control subjects (mean ± SD age = 48.5 ± 16.2 years) were included in this prospective study. The diagnosis of OSA patients was based on an overnight polysomnographic recording. To exclude the presence of OSA in the control subjects, they were asked to fill out a validated questionnaire to determine the risk of OSA. NewTom5G cone beam CT (CBCT) scans were obtained from both OSA patients and control subjects. Computational models of the oropharynx were reconstructed based on CBCT images. The aerodynamic characteristics of the oropharynx were calculated based on these computational models. Pearson correlation analysis was used to analyse the correlation between the most relevant aerodynamic characteristic(s) and anatomical characteristics of the oropharynx in OSA patients.

Results:

Compared with controls, the airway resistance during expiration (R_ex) of the OSA patients was significantly higher (p = 0.04). There was a significant negative correlation between R_ex and the minimum cross-sectional area (CSA_min) of the oropharynx (r = −0.41, p = 0.02), and between R_ex and the volume of the oropharynx (r = −0.48, p = 0.01) in OSA patients. After excluding an outlier, there is only significant correlation between R_ex and the CSA_min of the oropharynx (r = −0.45, p = 0.01).

Conclusions:

Within the limitations of this study, we concluded that the most relevant aerodynamic characteristic of the oropharynx in the collapse of the upper airway in OSA patients is R_ex. Therefore, the repetitive collapse of the upper airway in OSA patients may be explained by a high R_ex, which is related to the CSA_min of the oropharynx.

INTRODUCTION

Obstructive sleep apnea (OSA) is a sleep-related breathing disorder, characterized by recurrent obstructions of the airflow in the upper airway,^1,2 often associated with compromised upper airway space and an increase in upper airway collapsibility.³ The most common complaints of OSA patients are excessive daytime sleepiness, unrefreshing sleep, poor concentration and fatigue.⁴ The real pathogenesis of OSA is still unknown.⁵ However, both anatomical and functional abnormalities of the upper airway may play an important role in the repetitive collapse of the upper airway.⁶

Various imaging techniques have been used for upper airway analysis, such as multidetector row CT⁷ and MRI.⁸ In recent decades, the use of cone beam CT (CBCT) in dentistry has increased considerably. Due to its high spatial resolution, adequate contrast between the soft tissue and empty space and the relatively low radiation dose compared to multidetector row CT, CBCT can be used to analyse the upper airway anatomy three-dimensionally.⁹

Computational fluid dynamics (CFD) is an engineering tool used to analyse problems involving fluid or air flow, by simulating the way that the fluid or air flow is going through a specific tube or conduit.¹⁰ For this reason, CFD has been introduced to simulate the air flow in the upper airway of OSA patients based on three-dimensional (3D) datasets from CBCT.^11–13 CFD can provide a qualitative insight into how the air flow is travelling through the upper airway in an intuitive way, by showing the contours of the airflow characteristics, such as air velocity, pressure and wall shear stress.^12,14,15 CFD also allows quantitative assessment of the air velocity, wall shear stress, wall static pressure and airway resistance.^16–19 Comparison of the aerodynamic characteristics of the airflow within the oropharynx using CBCT, between OSA patients and control subjects, can add insight into the recurrent obstructions of the airflow in the oropharynx of OSA patients from the perspective of aerodynamics. So far, only one study compared the aerodynamic characteristics of the oropharynx in OSA patients and control subjects, however with a relatively small sample size (n = 8).¹³ Based on that study, it is unclear which aerodynamic characteristic of the oropharynx is the most relevant one in the collapse of the upper airway. In this study, we will compare the aerodynamic characteristics of the oropharynx in OSA patients and controls within a relatively large sample size based on CBCT images.

The primary aim of this study was to determine the most relevant aerodynamic characteristic of the oropharynx related to the collapse of the upper airway in OSA patients. The secondary aim was to assess the correlation between the most relevant aerodynamic characteristic(s) of the oropharynx and the anatomical characteristics of the oropharynx in OSA patients.

METHODS AND MATERIALS

Recruitment process

OSA patients were recruited from a prospective study designed to compare two different mandibular advancement device (MAD) therapies in mild and moderate OSA patients (ClinicalTrials.gov.identifier: NCT02724865). The inclusion criteria were: (1) 18 years and older; (2) ability to speak, read and write Dutch; (3) ability to follow-up; (4) ability to use a computer with Internet connection for online questionnaires; (5) diagnosis with symptomatic mild or moderate OSA [5 ≤ apnea-hypopnea index (AHI) <30] and (6) Expected to maintain current lifestyle (sports, medicine, diet, etc.).

The exclusion criteria were: (1) untreated periodontal problems, dental pain and a lack of retention possibilities for a MAD; (2) Medication used/related to sleeping disorders; (3) evidence of respiratory/sleep disorders other than OSA (e.g. central sleep apnea syndrome); (4) systemic disorders (based on medical history and examination, e.g. rheumatoid arthritis); (5) temporomandibular disorders (based on the function examination of the masticatory system); (6) medical history of known causes of tiredness by day, or severe sleep disruption (Insomnia, PLMS, Narcolepsy); (7) known medical history of mental retardation, memory disorders, or psychiatric disorders; (8) reversible morphological upper airway abnormalities (e.g. enlarged tonsils); (9) inability to provide informed consent; (10) simultaneous use of other modalities to treat OSA and (11) previous treatment with a MAD.

17 control subjects were prospectively recruited from among those who were referred for various diagnostic reasons to the department of Oral and Maxillofacial Radiology of Academic Centre for Dentistry Amsterdam, Netherlands. The inclusion criteria were age >18 years and CBCT images covering the entire upper airway from the level of the hard palate to the base of the epiglottis. The exclusion criteria were edentulousness, presence of a palatal cleft, presence of a craniofacial syndrome and upper airway surgery in the past.

This study was approved by the Medical Ethics Committee of Academic Medical Centre Amsterdam, protocol number: NL44085.018.13. Written informed consent was obtained from all participants.

Polysomnography (PSG)

For the diagnosis of OSA, all 31 patients included in this study underwent an overnight PSG recording (SOMNOscreenTM Plus PSG, Randersacker, Germany) at Onze Lieve Vrouwe Gasthuis West in Amsterdam. PSG included the following variables: electroencephalogram, electro-oculogram, leg and chin electromyograms, electrocardiogram, pulse oximetry, body position, neck microphone, nasal cannula pressure transducer and inductive plethysmography by means of thoracic and abdominal bands. The PSG recordings were scored manually in a standard fashion.²⁰ Apnea was defined as cession of airflow ≥90% for at least 10 s. Hypopnea was defined as a decrease in airflow of more than 30% for at least 10 s, and an oxygen desaturation greater than 4%.²⁰ The mean AHI of the OSA group, defined as the number of apneas and hypopneas per hour of sleep, is shown in Table 1.

Table 1.

Baseline demographic characteristics of OSA patients and control subjects

	OSA (31) Mean ± SD	Control (13) Mean ± SD	T/X2	p
Age (years)	43.5 ± 9.7	48.5 ± 16.2	−1.28(T)	0.21
Gender	32% (F)	69% (F)	5.1(X2)	0.02a
BMI (kg m−2)	26.4 ± 3.0	24.7 ± 2.1	1.93(T)	0.06b
AHI (time/hour)	15.0 ± 6.8	N.A.
Risk (%)	N.A.	11.7 ± 8.6

AHI, apnea-hypopnea index; BMI, body mass index; N.A., not applicable; OSA, obstructive sleep apnea.

Risk (%): risk (%) of having OSA, obtained by completing Philips questionnaire.

^ap<0.05.

^bTendency to significance.

Questionnaire

To evaluate their risk (%) of having OSA, each control subject was asked to complete a validated questionnaire on sleep apnea.¹⁶ This questionnaire was specifically developed to calculate the risk (%) of having OSA. It consists of three sections with a total of 23 questions on personal characteristics, sleep behaviour and health condition. On the basis of their answers to the questionnaire, only 13 controls with a low risk (%) of having OSA were included in this study.

Cone beam CT (CBCT)

The CBCT data sets of the OSA patients and control subjects were obtained using a NewTom 5G CBCT system (QR systems, Verona, Italy), according to the standard imaging protocol of the department of Oral and Maxillofacial Radiology of Academic Centre for Dentistry Amsterdam, Netherlands. During the imaging procedure, automatic exposure control was applied, and the patients were positioned in a supine position with the Frankfort horizontal (FH) plane perpendicular to the floor.¹⁷ They were instructed to maintain light contact between the molars in natural occlusion, to keep quiet breathing and to avoid swallowing and other movements during the scanning period. The exposure settings were 110 kV, 4 mA, 0.3 mm voxel size, 3.6 s exposure time (pulsed radiation) and 18–36 s scanning time, depending on the size of the patient.¹⁷ The CBCT scans were imported into NNT software to obtain a standard head orientation. The re-orientation was performed by using the palatal plane as a reference (anterior nasal spine -posterior nasal spine) parallel to the global horizontal plane in the sagittal view and perpendicular to the global horizontal plane in the axial view.¹⁸ For further analysis, the images were saved as digital imaging and communications in medicine files. All images were presented to the observers in a room with dimmed light.

Anatomical modelling of the upper airway

Using Amira^® (v. 4.1, Visage Imaging Inc., Carlsbad, CA), the automatic process of the oropharynx segmentation was performed following the same protocol as in a previous study.¹⁷ First, a voxel set was built to include all of the information of the oropharynx; second, a new mask was built with its thresholds ranging from −1000 to −400; and third, the superior boundary (i.e. the plane across the posterior nasal spine parallel to the FH plane) and the inferior boundary (i.e. the plane across the base of the epiglottis parallel to the FH plane) of the oropharynx were selected in the corresponding axial planes and put into the voxel set. Finally, all of the slices between the upper and lower boundaries were selected and put into the voxel set. The minimum cross-sectional area (CSA_min), the anteroposterior dimension of CSA_min, the lateral dimension of CSA_min, the volume of the oropharynx and the length of the oropharynx were calculated by an orthodontist (HC) and confirmed by an oral radiologist (PvdS) based on the segmented oropharynx (Figure 1).¹⁷ By surface triangulation, all the segmented oropharynx models were subsequently converted into 3D standard tessellation language models.

Figure 1.

(a) The segmented oropharynx. V, volume of the oropharynx; L, length of the oropharynx. (b) The minimum cross-sectional area (CSA_min) on the axial slice of the CBCT image. AP, anteroposterior dimension of CSA_min; lateral, lateral dimension of CSA_min. CBCT, cone beam CT; CSA, cross-sectional area.

Aerodynamic modelling of the oropharynx

The segmented standard tessellation language models of the oropharynx were exported into ANSYS ICEM CFD 17.0 (ANSYS, Inc., Canonsburg, Pennsylvania) to generate tetrahedral volume meshes. Depending on the complexity of the oropharynx model, a typical grid consisted of about 1,000,000 tetrahedral cells. ANSYS Fluent (ANSYS, Inc.) was used to conduct flow simulation within the oropharynx. The steady-state Reynolds Averaged Navier-Stokes formulation with the κ-ω shear stress transport turbulence model was used to model aerodynamic characteristics within the oropharynx.¹⁹ The air within the oropharynx was considered adiabatic.²¹ Least squares cell-based gradient was used for spatial discretization.^22,23 Second-order discretization schemes were used for the pressure and momentum equations. The coupling between the velocity and pressure fields was realized using the SIMPLE algorithm.^13,21,24 The density of the air within the oropharynx is set as 1.225 kg m³ and the viscosity of the air is set as 1.79E-05 kg/m/s. One boundary was set at the hard palate plane, and another boundary was set at the base of the epiglottis. The boundary condition consisted of axial velocity at the inlet plane, and no-slip boundary conditions for the oropharynx wall. An inlet volume flow rate of 166 ml s⁻¹ (10 L min^–1) was used in the flow simulation.^15,25

The aerodynamic characteristics, viz., velocity, wall shear stress and wall static pressure, were calculated by an orthodontist (HC) and confirmed by a medical engineer (YL) in each oropharynx model of OSA patients and controls during both inspiration and expiration. The inspiration phase was simulated by setting the inlet plane at the hard palate level and the outlet plane at the base of epiglottis. Vice versa, the expiration phase was simulated by setting the inlet plane at the base of epiglottis, and the outlet plane at the hard palate level.

Based on CFD calculations, the airway resistance (R) was determined using the following formulation:

$${\displaystyle \text{R}=\mathrm{\Delta }\text{P/Q}}$$

where, ΔP is the total pressure drop between the inlet and outlet boundaries of the oropharynx, and Q is the volume flow rate within the oropharynx.

Statistical analysis

Whether the data are normally distributed was tested by the Shapiro-Wilk W Test. The Mann–Whitney- U test (for non-normally distributed variables) or Chi-squared test (for categorical variables) and the independent t-test (for normally distributed variables) were used to compare the differences in the demographic characteristics between the OSA patients and their controls. Patient characteristics that were significantly different between the two groups were used as covariate(s) in the following between-group analysis. One-way multivariate analysis of covariance was used to compare the differences in aerodynamic and anatomical characteristics between the OSA patients and their controls. In the OSA group, Pearson correlation analysis was used to analyse whether significant aerodynamic characteristic(s) are correlated with the anatomical characteristics of the oropharynx. A significance level was set at p < 0.05.

RESULTS

Contours of the velocity (m s^–1), wall shear stress (Pa) and wall static pressure (Pa) for an OSA patient and a control subject during inspiration are shown in Figure 2. Contours of the velocity (m s^–1), wall shear stress (Pa) and wall static pressure (Pa) for an OSA patient and a control subject during expiration are shown in Figure 3.

Figure 2.

Contours of the velocity (m s^–1), wall shear stress (Pa) and wall static pressure (Pa) of a typical obstructive sleep apnea patient (a) and control subject (b) during inspiration.

Figure 3.

Contours of the velocity (m s^–1), wall shear stress (Pa) and wall static pressure (Pa) of a typical obstructive sleep apnea patient (a) and control subject (b) during expiration.

The demographic characteristics of the OSA patients and the control subjects are shown in Table 1. 31 OSA patients and 17 control subjects were recruited in the study, but four control subjects were excluded due to their high risk of having OSA. There was no significant difference in age between OSA patients and control subjects (p = 0.21). The OSA patients tended to have a higher body mass index (BMI) than their controls (p = 0.06). There was a significant difference in the gender distribution between OSA patients and control subjects (X² = 5.1, p = 0.02): there were fewer females in the OSA group (32%) than in the control group (69%). Therefore, BMI and gender were entered as covariates in the below-described analyses of covariance.

The aerodynamic characteristics within the oropharynx of OSA patients and control subjects are shown in Table 2. There was a significant difference in the airway resistance during expiration (R_ex) between OSA patients and control subjects (F = 2.90, p = 0.04). Compared with control subjects, the R_ex of OSA patients was significantly higher. The other aerodynamic characteristics in OSA patients were not significantly different from those in control subjects (p = 0.08–0.77).

Table 2.

Results of computational fluid dynamics analysis of OSA patients and control subjects during respiration, with gender and body mass index as covariates

Subject	OSA mean ± SD/median (interquartile range)	Control mean ± SD/median (interquartile range)	F	p
Maximum velocity during inspiration (m s–1)	7.1 ± 5.3	5.1 ± 1.4	1.24	0.31
Maximum wall shear stress during inspiration (Pa)	2.4 ± 2.3	2.4 ± 1.4	0.37	0.77
Minimum wall static pressure during inspiration (Pa)	−5.0 (-17.2,–3.4)	−5.7 (-7.4,–1.9)	1.21	0.31
Airway resistance during inspiration (Rin) (Pa/L/min)	1.2 (0.9, 2.5)	0.96 (0.69, 1.1)	1.95	0.13
Maximum velocity during expiration (m s–1)	7.7 ± 5.9	4.9 ± 2.7	2.27	0.09
Maximum wall shear stress during expiration (Pa)	1.26 (0.7, 3.3)	0.83 (0.6, 1.71)	2.25	0.09
Minimum wall static pressure during expiration (Pa)	−21.2 ± 34.3	−8.0 ± 11.3	2.37	0.08
Airway resistance during expiration (Rex) (Pa/L/min)	1.5 (0.7, 2.8)	0.73 (0.42, 1)	2.90	0.04 a

OSA, obstructive sleep apnea.

^ap < 0.05.

The anatomical measurements of the oropharynx of OSA patients and control subjects are shown in Table 3. There were significant differences in the minimum cross-sectional area (CSA_min) and the length of the oropharynx between OSA patients and control subjects. The CSA_min of the oropharynx of OSA patients, 65.5 (SD 30.7) mm², was significantly smaller than that of control subjects, 97.2 (SD 56.5) mm² (F = 4.26, p = 0.01). The length of the oropharynx of OSA patients, 65.2 (SD 8.1) mm, was significantly longer than that of control subjects, 56.1 (SD 8.1) mm (F = 19.19, p = 0.00).

Table 3.

The mean (±SD) of the oropharynx morphology of OSA patients and control subjects, with gender and body mass index as covariates

Variable	OSA	Control	F	p
Minimum cross-sectional area (CSAmin) (mm2)	65.5 ± 30.7	97.2 ± 56.5	4.26	0.01 a
Anteroposterior dimension of the CSAmin (mm)	5.1 ± 2.2	6.6 ± 2.1	1.39	0.26
Lateral dimension of the CSAmin (mm)	13.4 ± 4.4	15.3 ± 4.7	4.13	0.01 a
Volume of the oropharynx (cm3)	11.2 ± 3.7	10.8 ± 4.1	2.10	0.12
Length of the oropharynx (mm)	65.2 ± 8.1	56.1 ± 8.1	19.2	<0.001 a

^ap<0.05. OSA, obstructive sleep apnea. CSA, cross-sectional area.

The correlation between R_ex and anatomical characteristics of the oropharynx in OSA patients is shown in Table 4. There were significant negative correlations between R_ex and the CSA_min of the oropharynx (r = −0.41, p = 0.02) and between R_ex and the volume of the oropharynx (r = −0.48, p = 0.01) (Figure 4a and b). After removing an outlier, there was still significant negative correlation between R_ex and the CSA_min of the upper airway (r = −0.45, p = 0.01) (Figure 4c).

Table 4.

Correlation between airway resistance during expiration (R_ex) and oropharynx characteristics in OSA patients

	r	p
Minimum cross-sectional area (CSAmin) (mm2)	−0.41	0.02 a
Anteroposterior dimension of the CSAmin (mm)	−0.25	0.18
Lateral dimension of the CSAmin (mm)	−0.24	0.19
Volume of the oropharynx (cm3)	−0.48	0.01 a
Length of the oropharynx (mm)	−0.23	0.21

CSA, cross-sectional area; OSA, obstructive sleep apnea.

^ap < 0.05.

Figure 4.

(a) Correlation between airway resistance during expiration (R_ex) and minimum cross-sectional area (CSA_min) of the oropharynx; (b) correlation between R_ex and volume of the oropharynx; and (c) correlation between R_ex and CSA_min after removing an outlier.

DISCUSSION

The aerodynamic characteristics within the oropharynx of OSA patients and control subjects were compared based on their CBCT images. The most relevant aerodynamic characteristic of the oropharynx was the airway resistance during expiration (R_ex). Subsequently, the R_ex was correlated with anatomical characteristics of the oropharynx in OSA patients. The R_ex was related to the CSA_min of the oropharynx.

Limitations

One limitation is that we have no PSG recordings of the control subjects, because there was no medical need for these recordings. All control subjects, however, have completed a validated questionnaire to certify their low risk of having OSA.¹⁶ Besides, severe OSA patients (AHI > 30) were not included in this study. It is hypothesized that the R_ex of severe OSA patients will be much higher than the R_ex of controls due to their smaller CSA_min.²⁶ This hypothesis needs to be investigated in a future study. Except the oropharynx, the upper airway also includes the nasopharynx and hypopharynx. Another limitation of this study is that only the airway resistance (R) in the oropharynx was investigated. Obstruction of the nasal cavity could also play a role in the pathogenesis of OSA.^27,28 Therefore, it could be interesting to set up a model of the entire upper airway to fully understand the pathogenesis of the OSA from the perspective of aerodynamics. However, CFD analysis of the entire upper airway could be time-consuming due to the complex anatomical structure of the nasal cavity, which needs further investigation.

Confounders

There are many risk factors of OSA, such as higher age, elevated BMI and male gender.²⁹ The prevalence of OSA increases with age, with a peak between the ages of 55 and 64 years.³⁰ It is suggested that BMI is the most important risk factor for OSA.³¹ In this study, there was no significant difference in age between OSA patients and control subjects. In addition, BMI of OSA patients tended to be higher than that of control subjects. In this study, 32% of the OSA patients were female, which is in accordance with the prevalence of OSA in the population.³² In the control group, however, 69% of the patients were female. For that reason, in this study, the possible confounding effects of BMI and gender were taken into consideration in the statistical analyses.

Aerodynamic characteristics of the oropharynx

In this study, we simulated the aerodynamic characteristics of OSA patients and control subjects. We found that the airway resistance of OSA patients is higher than that of control subjects during expiration, which suggests that R_ex may be the most relevant aerodynamic characteristic of the oropharynx in the repetitive collapse of the upper airway in OSA patients. A previous study by Powell et al also found that the airway resistance of the OSA patients is higher than that of control subjects.¹³ However, only four controls and four OSA patients were included in their study, and they did not perform between-group statistical analysis, because this was not the aim of their study.¹³ Further, there is no consensus in literature if the collapse of the upper airway is more likely to occur during inspiration^33,34 or at the end of the expiration.^35–38 Based on the present study, we concluded that the collapse of the upper airway in OSA patients probably occurs during expiration.

Correlation between aerodynamic characteristics and anatomical characteristics of the oropharynx in OSA patients

In the current study, we found that OSA patients have a smaller CSA_min than control subjects, which is consistent with the conclusion of a systematic review wherein the most relevant anatomical characteristic of the upper airway related to the pathogenesis of OSA was a small CSA_min.²⁶ Further, we found a negative correlation between the CSA_min of the oropharynx and the R_ex of the oropharynx (Figure 4c). To the best of our knowledge, no studies have been performed on the correlation between the anatomical characteristics and aerodynamic characteristics of the oropharynx in the OSA patients. However, one study reported that the change in the anatomical characteristics of the upper airway after treatment is related to the change in the airway resistance in OSA patients.¹³ Powell et al found that the airway resistance was significantly decreased after upper airway surgery, which was in agreement with an increase of the oropharynx volume by 120 ± 70%.¹³ Therefore, it is hypothesized that in OSA patients, the smaller CSA_min and volume of the oropharynx will result in a higher R_ex of the oropharynx and therefore to a higher risk of collapse during sleep.

Clinical relevance

Previous studies have concluded that various OSA therapies (viz., MAD, and upper airway surgery) change the airflow characteristics within the upper airway of the OSA patients.^{10–12, 35–43} In responders, the airway resistance reduced significantly as a result of treatment, which shows that the repetitive collapse of the upper airway may indeed be explained by the high R of the oropharynx.^{11–14, 44, 45} However, there is still a lack of knowledge on the comparison of the aerodynamic characteristics of the oropharynx between responders and non-responders at baseline, which can help to recognize the non-responders before starting a treatment. These kinds of studies may improve our selection of OSA patients for a certain OSA treatment. Due to the advantages of non-invasiveness, low costs and the often limited amount of time to perform the examination, CFD analysis has gained widespread attention in airway research recently.⁴⁶ Based on CBCT imaging, CFD analysis permits the spatial visualization of the 3D airflow in the oropharynx during both inspiration and expiration.^{13, 46} The specific aerodynamic characteristics of the airflow, such as a higher R_ex, could be used to recognize OSA patient and as an estimator for selecting the most appropriate treatment modality for patients with OSA problems.^13,14,39 However, there is still not enough evidence to support CFD analysis as a clinical routine to predict the treatment outcome before starting these therapies, which needs further investigation.

CONCLUSION

The airway resistance during expiration (R_ex) is the most relevant characteristic of the airflow in the oropharynx related to the upper airway collapse in OSA patients. Therefore, the repetitive collapse of the upper airway may be explained by the high R_ex, which is related to the CSA_min of the oropharynx.