Introduction

Obstructive sleep apnea (OSA) is a public health concern associated with behavioral and cognitive dysfunction, cardiovascular risk, and a negative impact on the quality of life [1]. Maxillofacial hypoplasia is a predisposing factor for the development of OSA. A well-known phenotype with transverse maxillary constriction is associated with alteration in tongue posture and high nasal airway resistance [2]. Children with this phenotype have shown OSA treatment success with rapid maxillary expansion (RME) after tonsillectomy and adenoidectomy. Results are limited for adults with a similar phenotype, as techniques for maxillary expansion in adults were previously unreliable [3,4,5,6,7].

With the introduction of mini-implants for the facial bone, we reported Distraction osteogenesis maxillary expansion (DOME) technique to expand the adult maxilla that is similar to pediatric RME [8]. DOME was designed to address the need to expand the adult maxilla rapidly using mini-implant–assisted expander and less invasive osteotomy. DOME changes the palatal vault morphology from a narrow, high-arched shape to a dome shape with increased intraoral volume for the tongue and improved subjective nasal breathing [8, 9].

While transverse maxillary expansion increases nasal cavity volume and decreases nasal airway resistance [4], how this translates to improvement in OSA severity is unknown. We applied computational fluid dynamics (CFD) based on physiologic measurements from rhinomanometry [10, 11] to construct a dynamic airway model involving both the nasal and pharyngeal airways. We hypothesized that a reduction in nasal airflow velocity from DOME results in decreased negative pressure of the pharyngeal airway, leading to less upper airway collapsibility.

Material and methods

Design

We conducted a retrospective cohort study of 20 subjects who underwent DOME from September 2014 to April 2016.

Subjects

Subjects were adults with a diagnosis of OSA based on attended polysomnography, who were intolerant of continuous positive airway pressure (CPAP) or oral appliance therapy. They present with a narrow hard palatal roof, Mallampati class 4 (70%) or 3 (30%), and no palatine or lingual tonsillar hypertrophy. Subjects included must also have nasal rhinomanometry and cone beam computed tomography (CBCT) data available.

Polysomnography

Subjects underwent attended polysomnography (PSG) conducted and scored according to the standards of the American Academy of Sleep Medicine, including electroencephalography, electro-oculography, chin electromyography, and electrocardiography [12]. Subjects had transcutaneous pulse oximetry, with respiratory effort recorded using inductance plethysmography. Apnea was defined by decrease of baseline airflow by more than 90% for at least 10 s as measured by nasal cannula and mouth thermistor. Hypopnea was measured using a nasal pressure cannula and was defined as a partial obstructive event with decrease of airflow by 30% from baseline for at least 10 s associated with either a decrease in oxygen saturation by 3% or inducing an Electro-encephalographic (EEG) arousal for at least 3 s.

Rhinomanometry

Bilateral nasal resistance curves were measured using anterior rhinomanometry (NR-6 Research, GM Instruments). Three measurements of four breaths with at least a 150 Pa pressure drop were acquired [13]. Average Rohrer coefficients for each nasal passage were used to calculate nasal resistance and air pressure in the choanae at peak inspiratory flow.

Morphological evaluation

During the CBCT (i-CAT; Hatfield, PA, USA) examination, subjects were seated in a chair with his or her Frankfort horizontal plane parallel to the floor, and were instructed to maintain head position, not swallow, and remain in centric occlusion with relaxed tongue and lip positions at the end of expiration. Volume-rendering software (INTAGE Volume Editor; Cybernet, Tokyo, Japan) was used to create the three-dimensional (3D) images manually to allow measurements of the intermaxillary molar widths, nasal width, pharyngeal airway cross-section, and airway volume (Fig. 1). Pharyngeal airway cross-sectional measurements [14] included depth (anteroposterior direction) and width (left-right direction). The pharyngeal and intraoral airway volumes were measured between the palatal plane and the base of epiglottis plane, and the palate and tongue, respectively [15].

Fig. 1
figure 1

Measurement of anatomical structure and airway. A, nasal width: the widest distance of the nasal aperture. B, intermaxillary molar width: the distance between the narrowest points of the first maxillary molars. C, Measurement of cross-sections and volumes of the airway: PL plane, a plane passing through the hard plate; EB plane, a plane parallel to the hard plate passing through the base of the epiglottis; RA, retropalatal airway, narrowest cross section was measured parallel to the PL plane; OA, oropharyngeal airway, cross section was measured parallel to the PL plane passing through the anterior-inferior corner of second vertebra; HA, hypopharyngeal airway, cross section was measured along the PL plane passing through the anterior–inferior corner of third vertebra; intraoral airway, airway between the palate and the tongue; pharyngeal airway, airway between the PL and EB planes; D, depth; W, width

Full size image

Modeling nasal airway ventilation (Fig. 2A, B, C)

The 3D nasal airway was generated from the CBCT data by volume-rendering software (INTAGE Volume Editor; Cybernet Systems, Tokyo, Japan) [15]. The airway was segmented primarily on the basis of image intensity with the threshold set midway between the soft tissue and clear airway value. Subsequently, using mesh-morphing software (DEP Mesh Works/Morpher; IDAJ, Kobe, Japan), the 3D model was converted to a smoothed model without losing the patient-specific pattern of the airway shape. The models were exported to CFD software (Phoenics; CHAM Japan, Tokyo, Japan) in stereo lithographic format. CFD of the nasal airway models was analyzed under the following conditions: (1) the volume of air was flowing at a velocity of 300 cm3/s, (2) the wall surface was not slippery, and (3) the simulations were repeated 1000 times to calculate mean values. The simulation estimated airflow pressure, where air flowed from the choanae horizontally and was exhaled through both external nares. The nasal airway resistance model was then conformed to nasal rhinomanometry and calculated from air mass flow with the difference in pressure between external nares and choanae according to Ohm’s law [15]. We standardized the threshold of the nasal airway model such that nasal airway resistance obtained from CFD corresponded to rhinomanometry nasal airway resistance.

Fig. 2
figure 2

Evaluation of the nasal airway model and upper airway pressures during inspiration using computational fluid dynamics. A, Extraction of the nasal airway. B, Construction of the 3D nasal airway model and numeric simulation (expiration air, light yellow arrow). C, Evaluation of pressure of nasal airway. The nasal resistance was calculated by dividing the pressure of the nasal airway by air flow volume. The threshold of the air in the nasal cavity model found that nasal cavity resistance value obtained in CFD corresponded to the nasal cavity resistance value of rhinomanometry that was regulated. D, Extraction of the upper airway. E, Construction of the 3D upper airway model and numeric simulation (Inspiration air, light yellow arrow). F, Evaluation of ventilation condition of upper airway at inspiration (left; velocity, right; pressure)

Full size image

Modeling upper pharyngeal airway ventilation (Fig. 2D, E, F)

We conducted an inspiration simulation (airflow perpendicular to the lower pharyngeal plane at the velocity of 300 cm3/s) similar to what is described for the nasal airway [11]. We estimated maximal negative pressure and maximum velocity in the nasal, retropalatal, oropharyngeal, and hypopharyngeal segments of the upper airway.

Statistical analysis

Continuous data were presented as means ± standard deviation (SD). Paired t tests were conducted to test for significant treatment changes before and after DOME for all study variables. For variables with a non-normal distribution or differing variances, significant treatment changes were determined using the nonparametric Wilcoxon rank test. Intersite differences were determined using the Friedman test with Bonferroni correction. Spearman’s correlation coefficients were calculated to evaluate the relationships between morphological and ventilation parameters and PSG outcome. Statistical significance was defined as p < 0.05. Statistical analyses were conducted using SPSS Ver 24.

Results

Of the 20 subjects enrolled (five females and 15 males), the mean (± SD) age was 29.6 ± 8.3 years. Average height was 176.7 ± 8.5 cm, average weight was 79.7 ± 20.1 kg, and average BMI was 25.8 ± 6.2 kg/m2.

Polysomnography

Mean AHI improved from 17.81 ± 17.56 to 7.82 ± 7.11 events per hour (p < 0.001). Mean lowest oxygen saturation improved from 88.15 ± 7.21 to 90.90 ± 4.23% (p < 0.05). (Table 1). Oxygen desaturation index (ODI) and Apnea index (AI) decreased from 9.7 ± 16 events per hour to 4.9 ± 6 events per hour; and from 5.3 ± 18 events per hour to 0.78 ± 2 events per hour, respectively (p < 0.001).

Table 1. Comparisons of PSG Parameters, intermaxillary molar width, nasal width, pharyngeal airway volume, intraoral airway volume and nasal airway model resistance before DOME and after DOME
Full size table

Morphologic measurements

Intermaxillary molar width increased from 34.04 ± 3.54 to 41.6 ± 3.1 mm (p < 0.001, Table 1) after DOME. The nasal floor width at the level of internal nasal valve increased from 21.0 ± 2.1 to 26.2 ± 2.3 mm (p < 0.001) after DOME.

Pharyngeal airway size

No significant difference in width or depth of the pharyngeal airway was found after treatment (Table 2). The pharyngeal airway volume after DOME was significantly greater (Table 1).

Table 2 Treatment change of pharyngeal airway measurement and correlation between pharyngeal airway measurement and sleep data parameters
Full size table

Rhinomanometry-based computational fluid dynamic modeling (Fig. 2D, E, F)

Nasal airway resistance after DOME (0.48 ± 0.20 Pa/cm3/s) was significantly lower compared to baseline (0.65 ± 0.2 Pa/cm3/s, p < 0.001). The velocity of nasal airflow after DOME (7.41 ± 2.08 m/s) was significantly lower than baseline (15.59 ± 7.29 m/s, p < 0.001, Table 3). The velocity of the retropharyngeal, oropharyngeal, and hypopharyngeal airflow after DOME decreased significantly. Reduction of airflow velocity in the nasal passage was significantly greater than the pharynx.

Table 3 Treatment change of upper airway ventilation condition before and after DOME
Full size table

Correlations with polysomnography

AHI was positively correlated with nasal airflow velocity, and was negatively correlated with the nasal, retropalatal, oropharyngeal, and hypopharyngeal airway pressure. The apnea index (AI) had a similar tendency as the AHI (Table 4). The intermaxillary first molar width, nasal floor width, and pharyngeal airway and intraoral volumes show modest negative correlations with AHI and ODI.

Table 4. Correlation between sleep data and upper airway ventilation condition before DOME, after DOME, and treatment change
Full size table

Discussion

Distraction osteogenesis maxillary expansion (DOME) is a reliable method to expand nasal floor width and hard palatal vault in adults with obstructive sleep apnea (OSA). DOME results in a reduction in the apnea-hypopnea index (AHI) and subjective report of improved nasal breathing [16]. Using rhinomanometry and computational fluid dynamic (CFD) modeling, we propose a mechanism of how DOME reduces upper airway pharyngeal collapse in adults with OSA. Using both radiographic and physiologic parameters before and after DOME, we examine how maxillary constriction serves as a risk factor for adult OSA. Maxillary constriction is an under-recognized finding among adult patients with OSA. For such patients, surgical-assisted rapid maxillary expansion decrease nasal airway resistance (62.9%) [4]. In our study, nasal flow resistance is reduced from 0.53 to 0.16 Pa/cm3/s (30.2%). Of the 20 subjects enrolled, the AHI decreased by 44%, while patients reported significant subjective improvement of nasal breathing. There was no correlation found between degree of expansion and improvement as assessed by AHI.

We modeled nasal and upper airway airflow and pressure using CFD. While previous CFD is mostly imaging-based, ours include physiologic airflow measurements obtained from rhinomanometry. While CFD has limitations in the use of an awake and rigid airway for modeling, our physiologic nasal and oral flow data via rhinomanometry through the nasal and pharyngeal airway significantly improve real-life simulation and approximation.

Based on our findings, we propose the following mechanism of action for maxillary expansion in adults. Prior to DOME, subjects presented with high-flow velocity in the nasal airway. This was significantly reduced after DOME (Fig. 3). Relative to the nasal airway, the pharyngeal airway showed low airflow velocity before and after DOME. While the pharyngeal airway did not show significant changes in airflow velocity after DOME, it did change significantly with regard to negative pressure. Reduction in nasal airflow velocity is associated with reduced negative pressure of the pharyngeal airway (Table 3, Figs. 3 and 4). Concurrently, the reduction of nasal airflow velocity and pharyngeal upper airway negative pressure is associated with improvement of OSA based on AHI, ODI, and lowest oxygen saturation. This hypothesis had been suggested previously with imaging-based CFD [17], but without the physiologic nasal flow measurements which we have addressed with the present study.

Fig. 3
figure 3

Improvement of the upper airway ventilation condition upon inspiration using distraction osteogenesis maxillary expansion in an obstructive sleep apnea patient. A, before DOME, the nasal airway velocity was high (yellow arrow), and the site was considered an OSA causative site. A large negative pressure was observed at the lower part in the upper airway (red arrow) of the site. The large negative pressure in pharyngeal airway was considered easy to collapse during sleep. B, after DOME, the nasal airway velocity slowed (blue arrow), the cause site was improved, and large negative pressure in the pharyngeal airway was reduced (blue arrow). The pharyngeal airway had less negative pressure at inspiration, and became harder to collapse during sleep. Abbreviations: DOME, distraction osteogenesis maxillary expansion; OSA, obstructive sleep apnea

Full size image
Fig. 4
figure 4

The schematic diagram of effect of distraction osteogenesis maxillary expansion on the negative pressure in the pharyngeal airway and the responder and non-responder subjects of distraction osteogenesis maxillary expansion. A, Before DOME, the pharyngeal airway collapses (red arrow) as a result of negative pharyngeal airway pressure (large orange arrow) during inspiration). Dotted line, before inspiration; solid line, during inspiration. B, After DOME, there is a reduction of pharynx airway negative pressure during inspiration (small orange arrow), and less collapsibility of the pharyngeal airway. The pharyngeal airway remains patent during inspiration. Dotted line, before inspiration; solid line, during inspiration. C, C, After DOME, responder, initial pharyngeal airway had adequate width (RA 11.7 mm, OA 14.9 mm, HA 18.9 mm) and small negative pressure (− 108 Pa). Therefore, the airway remains patent and responded to DOME intervention. D, After DOME, non-responder, because hypopharyngeal airway width (red arrow; RA 11.0 mm, OA 7.5 mm, HA 7.2 mm) was narrow, airway remains collapsible despite small negative pressure (− 151 Pa) after DOME intervention

Full size image

Previously, similar types of study without individual-specific nasal airflow measurements suggested that pharyngeal airway volume is the primary determinant of AHI [18,19,20,21]. The rationale is that the smaller the upper airway, the higher the AHI. However, this hypothesis does not account for the nasal airflow which impacts collapsibility of the pharyngeal airway. Using subjects as their own controls before and after DOME, we propose that nasal airflow is a primary determinant of upper airway collapsibility. This is actually further corroborated by two cases that failed to show improvement in AHI after DOME. One subject showed persistent nasal obstruction after DOME due to severe septal deviation. The other subject had the smallest pharyngeal airway of the cohort. Therefore, treatment success from maxillary expansion still needs to take into account two major factors: (1) resolution of nasal obstruction and (2) improvement of nasal airflow may be inadequate to overcome pharyngeal airway collapsibility if the airway volume is excessively reduced. We did not find significant direct correlations between amount of expansion and sleep study parameters. The morphological measurements of the facial skeleton alone are insufficient to capture other elements of sleep including loop gain, arousal threshold, and muscle tone.

There are a number of pertinent limitations to our study. The computed tomography data was not obtained during sleep. It is unethical, however, to expose patients to the degree of radiation required to image upper airway movements during sleep. This is why we included physiologic nasal flow measurements to CFD modeling, which does assume rigid airway morphology. The nasal airway value was similar to the biological data and we considered that the value of the pharyngeal airway was similar to the biological data. However, CFD is a mechanical simulation and should be calibrated in practice. The sample size is not small in comparison to similar types of studies, but it is not large enough for us to make stronger clinical inferences. Lastly, while all subjects presented with transverse maxillary hypoplasia, some also present with anterior-posterior hypoplasia of the upper or lower jaw which may lessen the generalizability of our proposed mechanism.

Conclusion

While previous literature mostly acknowledged that a smaller upper pharyngeal airway is correlated with increased OSA risk and severity, we propose the additional importance of nasal airflow. Based on individual-specific CFD modeling that incorporates both airway morphology and air flow measurements, we propose that: (1) reduction of nasal airflow velocity impacts negative pressure of the pharyngeal airway, which contributes to reduction of OSA severity; (2) DOME demonstrates the efficacy of maxillary expansion in adults for the improvement of nasal breathing and OSA severity.