Background
Obstructive sleep apnea (OSA) is a common condition that is often associated with central obesity [
1]. During sleep, maintenance of upper airway patency is a primary physiologic goal, failure of which causes OSA and its sequelae [
2], with associated cardio-cerebrovascular complications [
3,
4].
Changes in lung volume are well known to affect pharyngeal airway size and stiffness, through thorax caudal traction on the trachea (Ttx) and may predominantly reflect Ttx effects [
5,
6], even though the anatomy narrowing of the upper airway plays an important role in the pathogenesis of OSA. There have been many studies examining the effects of lung volume on collapsibility of the human pharynx that have shown a similar response: using negative extrathoracic pressure (NETP) to elevate lung volume above functional residual capacity (FRC) in OSA and normal subjects during wakefulness or sleeping, is accompanied by improvement in pharyngeal collapsibility [
7‐
11], and decrease in pharyngeal resistance due to increases in pharyngeal size [
12]. These phenomena appear to be more pronounced in obese patients with OSA possibly because they usually respire with low lung volume. Therefore, obese OSA patients would obtain therapeutic benefits: inflation and maintenance of lung volume above FRC or end-expiratory lung volume (EELV) caused by NETP, or continuous positive airway pressure (CPAP) produces marked decreases in the apnea-hypopnea index (AHI) and in the magnitude of “holding pressure” (CPAP is required to eliminate upper airway flow limitation) [
9‐
11]. These studies appear to suggest involvement of lung volume, in terms of FRC or EELV, in the pathogenesis of OSA [
7,
8].
A recent animal study demonstrated that inspiratory resistive breathing (IRB), similar to upper airway obstructed breathing causes significantly increased lung elasticity measured by low-frequency forced oscillation technique (FOT) and downward shift of the pressure-volume curve; as IRB generates large swings in intrathoracic pressures and triggers lung inflammation [
13]. It has been shown by Van De Graff in anesthetized dogs that swings in intrathoracic pressure determine the extent of Ttx, which act independently to either draw the trachea into or push the trachea out of the thorax and that reciprocal movements of the trachea are independent of upper airway muscle activity determining alterations in upper airway patency [
5,
14]. Accordingly, the suggestions of those studies were that lung elasticity properties significantly influence generation of the Ttx, and that lung volume
per se, would not determine the mechanical influence of the thorax on the upper airway [
14]. Onal reported daytime measurements of airway conductance (Gaw) and the reciprocal of airway resistance (Raw), and predicted FRC were inversely associated, to a strong degree, with severity of OSA as defined by AHI in OSA patients without any obstructive lung disease, and suggested that decreased lung volume and increased Raw contribute to the severity of OSA [
15]; Zerah [
16] clearly demonstrated in obese subjects that respiratory system resistance (Rrs) measured by FOT was approximately equal to Raw, and the two parameters increased significantly with the degrees of obesity resulting from the reduction in lung volume. Subsequently, Zerah [
17] analyzed Rrs data obtained in 170 obese OSA patients, back-extrapolated the regression line to 0 Hz, to obtain the total Rrs and its inverse—respiratory conductance (Grs); because Rrs at lower frequencies (4–16 Hz) on FOT was subjected to linear regression analysis over the 4 to 16 Hz frequency range in obese subjects with and without OSA. The study observed that significant increase in the Rrs, and decreases in the Grs as well as in the specific Grs (sGrs: the ratio of Grs over FRC) were independently associated with OSA severity defined by AHI [
17,
18]. The ratio of Gaw over FRC and Gaw were used to reflect a function of the lung elasticity recoil forces, which is an important mechanical property of the respiratory system, as well as determining Raw, and is also sensitive to changes of FRC or EELV [
16]. In fact lung volume is determined by the balance of the elastic recoil forces of lungs (inward recoil) and chest wall (outward recoil) [
19]. From such evidence, we can imagine that Rrs increases and its inverse—Grs decreases more in obese OSA patients than those without OSA may result from increased lung elasticity recoil pressure as well as decreased in lung volume. Although, the studies mentioned previously appear to indicate to us that lung elasticity would be decreased in OSA patients. However, we can expect, despite the paradoxical evidence, that a vicious cycle exists between OSA and lung elasticity properties, which we feel needs further study.
The impulse oscillation system (IOS) is a type of FOT that has been progressively developed for clinical use over the years, as it was thought to provide information related to lung elasticity properties and Rrs during tidal breathing [
20‐
22]. Systematically assessed lung volumes, and respiratory mechanical properties of OSA patients using IOS may provide new insights into the underlying pathophysiology of OSA.
The aim of this study was to investigate the effect of OSA on lung volumes and mechanical properties of the respiratory system without the influence of body mass index (BMI) differences.
Methods
Subjects
In total 290 consecutive subjects (207 males, 83 females) who were obese to various levels without a medical history of lung diseases but who were troubled with snoring or suspicion of OSA were eligible for this cross-sectional study from April 2013 to April 2014. The classification of being pre-obese (overweight) or obese was according to the Chinese criteria [
23]: Subjects with a BMI of 24–27 kg/m
2 were classified as pre-obese; subjects with a BMI of 27.1–40 kg/m
2 and over 40 kg/m
2 were classified as obese and morbidly obese, respectively.
The exclusion criteria were patients previously treated with CPAP or urulo palato pharyhgo plast for snoring or OSA, presence of upper airway disorders, a history and physical examination compatible with cardiopulmonary disease, the presence of airway obstruction (forced expiratory volume in 1 s (FEV1)/forced vital capacity (FVC)) less than 80 % of the predicted value, signs of pulmonary hyperinflation on pulmonary function tests, and curvilinear expiratory flow-volume curves. Subjects with evidence of neuromuscular diseases (such as myasthenia gravis, hypokalemia, or Guillian-Barre syndrome) were also excluded.
Ethics approval
The study was approved by Ethical Committee of the People’s Hospital of Xinjiang Uygur Autonomous Region and informed consent was obtained from all participants.
Study design
Based on overnight polysomnography (PSG), 114 male and 42 female subjects with AHI > 10/h were diagnosed with OSA and formed the OSA group. 8 males and 7 females were excluded for presenting with expiratory flow limitation as detected by pneumotachograph, the FEV1/FVC less than 80 % of the predicted value. Thus 106 middle-aged male (aged: 45 ± 10 years) and 35 female (aged: 50 ± 9 years) subjects with OSA finally remained as the OSA group.
Ninety-three male and forty-one female subjects who fulfilled the same inclusion and exclusion criteria, were identified without OSA as AHI ≤ 10/h on their PSG results and formed the non-OSA group. Twelve (6 males, 6 females) subjects without OSA were excluded due to presenting with expiratory flow limitation. Finally, 87 male and 35 female subjects were distributed into the non-OSA group.
Measurement of lung function and volumes
All participants underwent standard spirometry and lung volume determinations, in line with American thoracic society/European respiratory society guidelines [
24]. Maximal flow-volume loops was conducted for each subject with sitting position using MasterScreen pneumotachograph (Jaeger/Care Fusion, Germany). For each pulmonary function test, three maximal flow-volume loops were taken to determine FVC and FEV
1; the largest one was retained to calculate the ratio of FEV
1 to FVC (FEV
1/FVC). Peak expiratory flow, maximum expiratory flow at 75 % (MEF
75%), at 50 % (MEF
50%), and at 25 % (MEF
25%) of FVC were also measured. Static lung volumes were determined after spirometry using a MasterScreen body plethysmograph (Jaeger/Care-Fusion, Germany) while the subject was sitting in a sealed box. Thoracic gas volume at FRC level was measured while subjects made gentle pants against the shutter at a rate of <1/sec. ERV, inspiratory capacity and vital capacity were measured during the same maneuver. The mean of three technically acceptable FRC measurements was used to calculate total lung capacity (TLC) as FRC + inspiratory capacity and residual volume as TLC − vital capacity.
Sleep studies
All participants underwent an overnight laboratory PSG including electroencephalography, electrooculogram, electromyogram of the chin, electrocardiogram, and recording of snoring and body position, respiratory efforts were detected with ribcage and abdominal piezo belts, oxygen saturation using pulse oximetry, according to the American Academy of Sleep Medicine guidelines [
25]: airflow was monitored by a nasal pressure transducer, supplemented by an oro-nasal thermal sensor. A nasal pressure drop ≥30 % of baseline associated with ≥4 % desaturation and a duration ≥10 s were scored as hypopnea; At least 10 s drop on a thermistor peak signal excursion of ≥90 % from baseline and absence of airflow on a nasal pressure transducer were scored as apnea. In at least 90 % of the events’ duration met the amplitude reduction criteria. Severity of OSA was expressed as the total number of apneas plus hypopneas per hour of sleep (AHI).
Measurement of mechanical properties of the respiratory system
IOS measurements and quality control were conducted in line with European respiratory society recommendations using Masterscreen IOS (Jaeger/Care-Fusion, Germany) [
26]. IOS measurement was performed in each subject sitting in a neutral posture with cheek support while wearing noseclips. Subjects were asked to tightly seal their lips around the mouthpiece and to breathe quietly at FRC level. Continuous rectangular electrical impulse signals were superimposed on the airway via a mouthpiece after stable spontaneous volume and airflow were confirmed and a minimum of 3 consecutive measurements of >30 s were taken. As IOS measures we used respiratory system impedance (Zrs) at 5Hz (Zrs5), mean whole-breath values of Rrs and reactance (Xrs) between 5Hz and 35Hz in 5Hz increments (R5–R35 and X5–X35, respectively) and resonant frequency (Fres). IOS measurements were performed by two experienced technicians who were blinded to the study groupings. Measurements with artifacts, such as irregular breathing, hyperventilation, leakages or swallowing, were discarded. IOS can evaluate Rrs and Xrs at various oscillatory frequencies that are automatically calculated with computer software that uses fast Fourier transform analysis to determine Raw in extrathoracic and intrathoracic airways as well as the elastic properties of lung and chest wall. The Zrs encompasses all forces that hinder air flow into and out of the lung, and includes the resistance, elastance, and inertia of the system. The Rrs is a real part of Zrs, in phase with flow, which reflects energy dissipation due to resistive losses. Rrs is the sum of the extrathoracic airway, intrathoracic airway, and chest wall resistance, all arranged in series. In general, lower frequency data reflect the more peripheral regions of the lung, while higher frequency data are most representative of the central or proximal airways [
20,
22,
26]. The Xrs is an imaginary component of Zrs that comprises out-of-phase lagging flow, which is elastance, and out-of-phase leading flow, which is inertia. Both of these components reflect energy storage. Theoretically, lung elasticity properties are reflected in the lower oscillatory-frequencies, while inertial properties are reflected dominantly in the higher oscillatory-frequencies [
22,
26]. Rrs measured at lower frequencies (from 5 to 15 Hz) on IOS enabled us to obtain the total Rrs (Rrs at 0 Hz; R0) using the linear regression model R(f) = R0 + S × f, where f represents the frequency, S is the slope of the linear relationship of resistance versus frequency, R0 is equivalent to zero-order frequency resistance, namely the intercept). Grs was calculated as the reciprocal of R0 and sGrs was obtained as the ratio of Grs over FRC. In addition; the value of Zrs at 5Hz (Zrs5) yield by IOS is believed to be equivalent to R0, respiratory conductance was also calculated as the reciprocal of Zrs5 and expressed as Gz, the ratio of Gz over FRC as sGz.
Statistical analyses
All data were expressed as mean ± standard deviation. Data for males and females were analyzed separately, because of differences between the genders in lung volume size and airway resistance, as well as the prevalence of OSA which are all well documented. Intergroup comparison was made using Independent-Samples t-test for variables showing normal distribution and homoscedasticity; otherwise, the Mann-Wilcoxon-test was used. Correlations among variables were determined using the least-square linear regression method. Multiple stepwise regression analysis was performed to assess relations between severity of OSA and lung volumes, respiratory mechanical properties and anthropometry. Statistical analysis was performed using SPSS (version 19.0, IBM Corp., Armonk, NY, USA). p-values < 0.05 were considered significant.
Discussion
This study explored the effect of OSA on lung volumes and mechanical properties of the respiratory system in pre-obese/obese subjects. The results showed that OSA impacts upon lung elasticity properties, and they increased with OSA severity. There was also a suggested increased airflow resistance in the extrathoracic and peripheral airways and a decrease in lung volume, in terms of FRC and ERV in obese OSA. The unique and novel aspects of this study were to measure Zrs5 in the morning after an overnight sleep study, and Rrs and Xrs were also measured at different frequencies using IOS with a standardized method. Many previous studies have confirmed that lung function is little changed with class I and II obesity [
27‐
28,
29,
30]. Some research suggests obesity is associated with a higher ratio of FEV
1 to FVC [
28,
30]. Meanwhile, there is general agreement that conventional pulmonary function tests are unhelpful for diagnosis of OSA [
31]. In the present study, conducted on a large number of pre-obese/obese subjects with and without OSA, we were not surprised that there was no significant difference in FEV
1/FVC between obese subjects with and without OSA, although the FEV
1/FVC were slightly higher in the OSA group than non-OSA group.
Hoffstein
et al. demonstrated that the cross-sectional area of the pharyngeal-airway decreases as lung volume decreases from FRC to residual volume, a phenomenon most pronounced in obese OSA patients [
32,
33]. Several studies since then have associated lung volume with severity of OSA. Appelberg
et al. reported ERV was significantly lower in subjects with OSA compared with non-snoring subjects and that ERV was independently correlated to apnea index and oxygen desaturation frequency [
34]. Onal
et al. found predicted FRC was significantly negatively correlated to AHI in OSA patients [
15]. Zehra-Lancner
et al. evaluated pulmonary function and lung volume in 170 obese snorers with or without OSA, and found all of patients had significant decreases in FRC, most markedly decreases in ERV, compared to predicted values [
17]. In the present study, actual ERV significantly decreased with a drop of FRC in both genders of OSA patients compared with non-OSA subjects. However, decreases in those lung volume measurements have poor predicted values for AHI, we reasoned that size of lung volume
per se may play a weak role in the mechanism of upper airway collapse. Our results also indicated that OSA had a negative impact on lung volume, similar to the impact of BMI on lung volume. Obesity is a common feature of OSA [
35], and has been associated with decreased FRC with a drop in ERV [
27‐
28] due to reduced compliance in the respiratory system. Zehra reported that lung volumes showed a marked decrease because of reduced chest wall compliance as the degree of obesity rose [
16]. More recently, Behazin
et al. and Pelosi
et al. [
36,
37] reported that obese subjects breathe at abnormally lower FRC than non-obese subjects due principally to reduced lung compliance rather than chest wall compliance. Our findings in this study appear to indicate that FRC and ERV were decreased more in pre-obese and obese patients with OSA than in those without OSA. The multiple regression analyses demonstrated that the lung volume decreases in obese men with OSA were independent of BMI and negatively correlated with AHI. This result was further confirmed by the female cohort; we classified the cohort according to the presence and absence of OSA, and found OSA group respiration at lower lung volumes than the non-OSA group and the two female groups had similar BMI. Recently, study animals were exposed to several hours of IRB, mimicking upper airway obstructed breathing, the researchers found that lung elasticity significantly increased in the IRB animals compared to the spontaneously breathing animals and found a downward shift of the pressure-volume-curve, suggesting that IRB leads to large negative swings in intrathoracic pressure [
13]. Bijaoui reported that the onset of obstructed breathing was accompanied by an increased lower component of lung elasticity; simultaneously lung resistance increased when obstructed breathing existed in various sleep stages of OSA patients [
38]. Xrs5 is numerically a negative value on IOS and known as lung elasticity properties or its inverse the compliance, so a more negative value indicates increased pulmonary elastance or reduced compliance (the reciprocal of elastance) we infer that this was a result of increased lung elasticity or its inverse, reduced compliance. Interestingly, in this study the Xrs5 was found to be strongly associated not only with the severity of OSA as defined by the AHI or HI, but also with the total airway resistance (R0 or Zrs5) as well as peripheral airways resistance (R5). This result suggested that with increasing severity of OSA pulmonary elastance might be markedly increased or its inverse—namely, the compliance was decreased, which in turn contributed to a significantly decreased FRC and marked increases of airflow resistance in both the upper airway and in peripheral airways in obese OSA patients than those who are solely obese; as it is well documented increased lung elastance or reduced the compliance is associated with low FRC as well as breathing at low lung volumes are related to increases resistance in the structures of upper airway and peripheral airways.
It has been established that a decreasing curvilinear relationship between FRC and total airway resistance or lung resistance in obesity [
36,
39,
40]. This relationship between FRC and respiratory resistance on IOS was also found in our study population (Additional file
2: Figure S2. A, B). Van Noord
et al. initially reported that values of Rrs at lower frequencies (4–16 Hz) measured by FOT are characterized by negative frequency dependence in patients with UA and bronchial obstruction; simultaneously they found that the Xrs values at lower frequencies showed a marked decrease (more negative) with decreased oscillatory frequency in upper airway obstruction patients [
41]. Since Rrs increases linearly with a decrease of oscillatory frequency over lower frequencies as observed by Zerah
et al. in obese subjects and the total airway resistance (R0) increased with level of obesity, this suggests that R0 increased in obese subjects resulting from the reduction in lung volume related to being overweight [
16]. The researchers then demonstrated that with increasing severity of OSA in obese snorers, the R0 was a significantly increased and the reciprocal of R0, the Grs and the sGrs were significantly independent of BMI decreased with increasing AHI [
17]. As result of the sGrs in those patients was found to be highly independently correlated with AHI and had a stronger predictive value for OSA [
17,
18] the researchers suggest that upper airway and peripheral airway obstruction coexisted in obese OSA patients. In this study, we also found a significant increase in R0 and decrease in Grs with severity of OSA in both genders. However, we could not find any correlation between severity of OSA and sGrs in females, and only a weak correlation in males. We further analyzed our data in terms of ratio of Grs to normal predicted FRC, and found a stronger correlation between severity of OSA and sGrs, the result was similar to Zerah
et al. in a previous study [
17]. However, it can be speculated that the connection between OSA and sGrs would be overstated by this method. Small sGrs were obtained in our OSA-patients by dividing Grs with normal predicted FRC which has tendency to be bigger than the actual FRC, therefore, this yielded the small sGrs and a significant decrease of sGrs in OSA patients was found, the connection can be naturally exaggerated, although in fact OSA patients are breathing at low FRC.
A study clearly demonstrated breathing at low lung volume with a less lung compliance produces large swings in intrathoracic pressure in obese subjects and with which airways are prone to close or collapse on exhalation. Similarly, numerous studies have shown especially large swings in intrathoracic pressure exist when OSA occurs in OSA patients. Large swings in intrathoracic pressure means a high driving pressure or lung elasticity recoil pressure is generated at the end of inspiration or start of expiration, as well as a high negative pressure generated at the end of expiration or beginning of inspiration. Application of small sub-atmospheric pressure (from −3 to −10 cmH
2O) at the start of expiration is defined as the negative expiratory pressure (NEP) technique, and is used for early detection of whether subjects have expiratory flow limitation (EFL) in their peripheral airways. However, The association between the degree of EFL induced by NEP and the severity of OSA has been well documented in OSA patients in the past twenty years, and it has been concluded that EFL caused by NEP are common in a population of OSA patients, which can be highly predictive the severity of OSA [
42‐
44].
By putting such evidence together, it can be speculated that small additional NEP applied at the start of expiration for OSA patients, would further elevate intrathoracic driving pressure or lung elasticity recoil pressure [
40,
42], if this already existed, consequently, higher driving pressure or lung elasticity recoil pressure can make alveoli and their opening (peripheral-airway) as well as the extrathoracic airway easily collapse and manifest through markedly increased airflow resistance in those structures. This phenomenon was also shown by many previous studies; for example elevating driving pressure by continuous negative airway pressure (CNAP) or positive extrathoracic pressure for healthy subjects, have always found significant increases in resistance in peripheral airways as well as upper airways that are simultaneously accompanied by slightly decreased lung volume [
9,
40], which are remarkable in OSA patients [
32,
33]. We believe, the converse would also apply, for example, decreasing driving pressure or lung elasticity recoil pressure by NETP or CPAP, improves upper airway collapsibility as well as in the intrathoracic-airway, and manifests as increased lung volume and markedly decreased in resistance in upper airway and intrathoracic airways [
7‐
12]. A study has clearly demonstrated that the CNAP-induced increase in Rrs does not only result from a direct lung volume effect and strongly suggested the involvement of other factors that affect the caliber of intrathoracic and extrathoracic airways [
40]. That was true, in fact, Rrs depends on lung elasticity recoil pressure [
16], and lung resistance always increases with increase in lung elasticity [
38,
37], in the current study we also found Rrs was significantly high correlated with Xrs5 in all study subjects. This evidence makes us conclude that Xrs5 could reflect the lung elasticity recoil pressure in this study. From such evidence, combined with our findings, we infer and conclude that upper airway obstruction together obesity negatively impacts mainly upon lung elasticity or lung compliance, which might be abnormally increased or the compliance decreased, respiration in a less lung compliance lead to increased lung elasticity recoil pressure on exhalation and contribute to a decrease in FRC, which, in turn, aggravate upper airway and intrathoracic airways obstruction or collapsibility and manifest through markedly increased resistance in upper airway and intrathoracic airways.
This study harbors some limitations. To explicate the effect of OSA on lung volumes and mechanical properties of respiratory, with the goal of eliminating a confounding effect of BMI differences on respiratory function tests result, initially, we carried out a case–control study. Although we found that the BMI matched in the female OSA and non-OSA subjects. This was not the case for the male groups. It is a challenge, in the moderately obese male population, to find subjects whose AHI < 5/h or even less than 10/h to match in terms of BMI. Therefore, there was a significant difference in BMI between the male groups. As this study was not intended to evaluate the relationship between BMI and OSA we then had to perform statistical analysis to account for these differences. Secondly, the nature of the study, as a cross-sectional observational one, limits the ability to ascertain the cause of the disease, and does not allow us to draw a causal relationship between respiratory system elastance or compliance and upper airway function in obese OSA patients, thus interventional animal studies are required to further confirm our findings.
Competing interests
The authors declare that they have no competing interests.
Authors’ contributions
AA and NFL conceived and planned the program. YPZ, MHZ and YLDS recorded the patients’ information. YCW and XGY analyzed the data. TY contributed to the writing of the manuscript. All authors read and approved the final manuscript.