Pharyngeal Function and Snoring Characteristics in Apneic and Nonapneic Snorers 1 , 2
v.
HOFFSTEIN, S. WRIGHT, N. ZAMEL, and T. D. BRADLEY3
Introduction
It is thought that patients with obstructive sleep apnea (OSA) have abnormal pharyngeal structure and function (measured even when these patients are awake), predisposing them toward development of complete airway obstruction during sleep (1-13). Continuing interest in studying pharyngeal properties in patients with sleep apnea is partially stimulated by a desire to find a good screening test for this disorder. Such a test would help us to identify patients at risk and thus reduce the number of expensive and time-consuming sleep studies. Pharyngeal structure is usually assessed by measurements of pharyngeal cross-sectional area using a number of different techniques such as computerized tomographic scans (3, 5, 6), videofluoroscopy (1-4), magnetic resonance imaging (14), X-ray cephalometry (1517),and acoustic reflections (7). Pharyngeal function is usually assessed by some measurements reflecting the dynamic behavior of the pharynx such as pharyngeal resistance (10), change in area in response to collapsing or distending pressure (8, 9, 18), or change in area in response to changes in lung volume (11, 19). The latter two measurements, termed pharyngeal distensibility (Cph) and pharyngeal lung volume dependence (PLVD), respectively(11, 20), may be performed using the acoustic reflection technique for measuring airway area. The equivalence between the above two measures of pharyngeal function has not been demonstrated. From the technical point of view, it would be advantageous to know whether the changes in pharyngeal distensibility parallel changes in PLVD because the measurements of pharyngeal area as a function of pressure are more difficult to make than measurements of pharyngeal area as a function oflung volume. If the two measurements are closely associated, we might be able to replace the technically more difficult measurement (Cph) with the relatively simpler one (PLVD). With this in mind, 1294
SUMMARY Abnormalities in pharyngeal function, manifested even when the patients are awake, are thought to play an important role in the pathogenesis of sleep apnea. Tests of awakepharyngeal function continue to stimulate interest because it is hoped that they may allow physicians to distinguish patients with sleep apnea from those without It, and therefore reduce the number of unnecessary sleep studies. We elected to stUdy two measures of pharyngeal function: changes in pharyngeal area with lung volume (PLVD)and changes in pharyngeal area In response to externally applied positive pressure, t.e., pharyngeal distensibility (Cph). Both measurements have been employed for assessment of pharyngeal function, and both are thought to reflect pharyngeal "floppiness." Measurement of PLVD is technically very simple, whereas the measurement of Cph is technically more complex. If the two measurements are highly correlated, it might be possible to replace the technically more difficult one by the simpler one. Consequently, the purpose of this stUdy was twofold: first, to examine the relationship between pharyngeal distensibility and lung volume depen. dence of pharyngeal area, and second, to compare these parameters In a large group of confirmed snorers with and without obstructive sleep apnea (OSA). Westudied 75 unselected patients referred for the investigation of snoring and suspected sleep apnea. All patients had nocturnal polysomnography, pUlmonary function tests, and measurement of pharyngeal areas at TLC, FRC,and residual volume (RY) employing the acoustic reflection technique. The area measurement at FRC was performed at zero and at 4.1 em H20 positive airway pressure to calculate pharyngeal distensibility. We found no significant correlation between PLVDand Cph (r = 0.15, P = 0.20). Using a stepwise, forward, rnultlple linear regression model, we found that only body mass index and PLVD, but not Cph, were significant determinants of the variability In apnea/hypopnea index (AHI) (multiple r 2 0.33, P < 0.005). Neither PLVD nor Cph correlated with snoring. When we divided our patients Into four groups based on an AHi of less than 10 (controls), between 11 and 30 (mild OSA), between 31 and 50 (moderate OSA), and greater than 50 (severe OSA), we found that PLVD, but not Cph, separated the severe group from the control and the mild groups. We conclude that (1) measurements of PLVD and pharyngeal compliance are not equivalent, and (2) lung-volume-associated changes in pharyngeal area are related to the severity of sleep apnea.
=
AM REV RESPIR DIS 1991, 143:1294-1299
the present study was designed to test the hypothesis that Cph is correlated with PLVD. Methods Patients Westudied 75 unselected consecutive patients referred to our sleep clinic for evaluation of snoring. Each of these subjects admitted to being a long-time heavy snorer; all subjects suffered various consequences of snoring with respect to their marital and/or social relationships, which usually precipitated their referral to the clinic. Although when interviewed some of these patients werealso found to have other complaints such as excessive daytime sleepiness, morning tiredness/fatigue, and restless sleep, the presenting complaint in all of them was snoring. There were 65 male and 10 female patients ranging from 16 to 69 yr of age. Measurements of Pharyngeal Mechanics The area of the pharynx was measured using the acoustic reflection technique, extensively
described previously (7,11,18, 19).This technique allows us to construct an "airway echogram," i.e., a plot of airway area versus distance from the mouth. On this plot we identified the "pharynx" as the segment contained between the teeth and the glottis, and computed the mean area of this segment, referred to in the subsequent discussion as "pharyngeal area." The measurements wereperformed according to three different protocols: (1) patients breathing quietly at FRC, (2) patients breath-
(Received in original/arm December 12, 1989 and in revised form January 22, 1991) 1 From the Department of Medicine, St. Michael's, Mt. Sinai, and The Toronto Hospitals, , Universityof Toronto, Toronto, Ontario, Canada. 2 Correspondence and requests for reprints should be addressed to Dr. Victor Hoffstein, St. Michael's Hospital, 30 Bond Street, Toronto, Ontario, Canada M5B IW8. 3 Career Scientist of the Ontario Ministry of Health.
PHARYNGEAL FUNCTION AND SNORING CHARACTERISTICS IN APNEIC SNORERS
ing at FRC but with 4.1 cm H 2 0 positive pressure applied at the mouth, and (3) measurements taken during a slow quasi-static expiration from TLC to residual volume (RV);this allowed us to obtain a relationship between pharyngeal area and lung volume, i.e., volume-area plot. Positive airway pressures were generated by loading the spirometer, located at the distal end of the wave tube, with different weights. We used 1-, 1.5-, and 2-kg weights, which generated pressures of 4.1, 6.5, and 8.5 cm H 2 0 , respectively. In the first 33 patients, airway areas were obtained at all three pressures. We elected to use only the results at low pressure (4.1 em H 2 0 ) in order to avoid the following problems: (1) large changes in lung volume, which also influence pharyngeal area independently of applied pressure; (2) open velum (21), which leads to overestimation of pharyngeal area; (3) possible nonlinearity of the pressure-area curve of the pharynx at higher pressures. For these reasons, we performed all pressure-area measurements at a pressure of 4.1 em H 20 in the rest of the patients. Measurements of area were performed five times per second, and there were 128 measurements available at FRC, and as many as 24 measurements at TLC and at RV. At each lung volume weaveraged all pharyngeal areas and calculated the mean area for that lung volume. From the measurements of pharyngeal area we derived two indices of pharyngeal func-. tion: PLVD and Cph, PLVD was defined as the difference in pharyngeal area between FRC and RV divided by the expiratory reserve volume (ERV), i.e., (AFRC - ARV)/ERV. Cph was defined as the difference in pharyngeal areas measured at 4.1 ern H 2 0 pressure and at atmospheric pressure (AFRC. and AFRCo, respectively), per unit pressure, i.e., Cph = (AFRC. - AFRCo)/4.1. We also computed specific Cph, which is defined as the Cph normalized by pharyngeal area at FRC
ATLe
p>o
A RV
RV
FRG
FRG
TLG
(0 em H 2 0 ) (4.1 em H 2 0 )
Lung Volume
Fig. 1. Schematic diagram illustrating how to correct pharyngeal area at a pressure P for the effect of lung volume. A. = pharyngeal area at FRC at atmospheric pressure; Ap = pharyngeal area at FRC measured at a positive transpharyngeal pressure; I1A = increment in area; I1V = increase in lung volume; ATLe = pharyngeal area at total lung capacity; ARV = pharyngeal area at residual volume.
at atmospheric pressure, i.e., sCph = Cph/ AFRCo• Because the application of positive airway pressure increases lung volume, which in turn leads to increase in pharyngeal area, it is necessary to correct pharyngeal area for this lung volume effect. This is illustrated schematically in figure 1, which shows two volume-area curves, one obtained during quiet, quasi-static expiration from TLC to RVat the atmospheric pressure, and the other one obtained at a positive pressure P. A, is the pharyngeal area at FRC at atmospheric pressure. Ap is pharyngeal area at FRC measured at a positive transpharyngeal pressure P; it is composed of an increment in area (AA) caused by an increase in lung volume (AV), and an increment in area caused by the application of positive transpharyngeal pressure. Uncorrected Cph is simply (Ap - Ao)/P, whereas the corrected Cph is given by [Ap - A, - AA)/P, where AA = AV multiplied by the slope of the pharyngeal area-lung volume curve. V is taken to be equal to the difference in ERV at atmospheric pressure and at positive pressure P, and the slope is taken to be equal to PLVD. This correction in itself is an approximation because it assumes that (1) pharyngeal area during tidal breathing at FRC is equal to that obtained during slow expiration when passing through the same lung volume, (2) the slope of the pharyngeal area-lung volume curve is constant, and (3) residual volume is not affected by the application of positive transpharyngeal pressure P. We performed all calculations using corrected and uncorrected Cph.
Sleep Studies
1295
and therefore only the spikes in sound intensity exceeding 60 dB were classified as snores. The polysomnographic tracings were analyzed by experienced sleep technologists unaware of the results of the pharyngeal mechanics measurements. Sleep stages were scored according to the criteria of Rechtschaffen and Kales (23). Apneas were determined by the absence of tidal volume excursions on the respiratory inductance plethysmograph of at least 10 s and absence of airflow at the nose and the mouth. Obstructive apneas were scored when the paradoxical chest wall motion on the respitrace was seen. Hypopneas were defined as 50% reduction in tidal volume below the baseline value lasting longer than 10 s (24). Apnea/hypopnea index (AHI) was defined as the number of these events per hour of sleep, and snoring index (SI) was defined as the number of snores per hour of sleep.
Pulmonary Function Tests In all patients we measured flow-volume curves using a dry roIling-seal spirometer (CPI 220; Cardio-Pulmonary Instruments, Houston, TX) and FRC using a constant-volume body plethysmograph (Ohio 3000 system; Ohio Instruments, Madison, WI). TLC was obtained by adding the inspiratory capacity to FRC, RV was obtained by subtracting vital capacity from TLC, and ERV was calculated as the difference between FRC and RV.
Data Analysis We used stepwise, forward, multiple linear regression analysis to examine whether PLVD, Cph, pharyngeal areas, and anthropometric variables were significant determinants of the variability in the AHI and SI. Correlation analysis was used to search for any single relationships between the above variables as well as between PLVD and Cph. The analysis was repeated after correcting Cph for lung volume and also using specific distensibility. To determine if PLVD or Cph can separate apneic from nonapneic patients, and realizing that any predetermined AHI cutoff to define sleep apnea is somewhat arbitrary, we decided to split our group of 75 patients into four subgroups: control (AHI < 10), mild apnea (10 ~ AHI ~ 30), moderate apnea (31 ~ AHI ~ 50), and severe apnea (AHI > 50). We used analysis of variance with Thkey's posthoc tests to compare the anthropometric data, pulmonary function, sleep data, and pharyngea mechanics between these four groups. All statistical analysis was carried out using the SAS statistical software (release 6.03; SAS Institute, Gary, IN). Statistical significance was defined as p < 0.05.
All patients had full nocturnal polysomnography performed in the standard manner, with monitoring of electroencephalogram (EEG), right and left electrooculogram (EOG), submental and anterior tibial electromyogram (EMG), electrocardiogram (ECG), oronasal flow, chest wall and abdominal excursion by impedance plethysmography, oxygen saturation, and snoring. All measurements were displayed on a Grass polygraph (Model 78D;, Grass Instruments, Quincy, MA). EEG, EOG, ECG, and EMG were measured using the Grass amplifiers appropriate for the Model 78D recorder. Oronasal flow was measured using oral and nasal thermistors. Chest wall and abdominal excursions were obtained using a respiratory inductance plethysmograph (Respitraces; Ambulatory Monitoring, Inc., Ardsley, NY). Oxygen saturation was measured using a Biox finger oximeter (Biox 3700; Ohmeda, Boulder, CO). Snoring was measured in a manner similar to our previous study (22) using a microphone and a Results sound meter ~L1220; Pacer Industries, TorThe anthropometric data (table 1)for the onto, Ontario); the microphone-sound meof 75 patients as well as for entire group ter system was calibrated in the range of 40 to 110dB using a I-kHz signal and a sound the different subgroups (control, mild, moderate, and severe) demonstrate that chamber. The microphone was taped to the chest wall band of the Respitrace. With this only weight and BMI were significantly setup, quiet breathing registered below 60 dB, different between the four groups. Post-
1296
HOFFSTEIN, WRIGHT, ZAMEL, AND BRADLEY
TABLE 1 ANTHROPOMETRIC AND PULMONARY FUNCTION DATA Groups All Patients n Age, yr Height, em Weight, kg BMI, kglm' ERV., L 0lb pred ERV., L FRC, L % pred FVC, L % pred FEV" L alb pred
48 172 93 31 0.8 51 1.2 3.4 95 4.4 105 3.4 108
Mild OSA (10 < AHI ~ 30)
Control (AHI ~ 10)
75 ± 11 ± 9 ± 20 ± 6 ± 0.5 ± 33 ± 0.5 ± 0.9 ± 23 ± 1.0 ± 17 ± 0.8 ± 21
45 169 84 29 0.9 63 1.1 3.4 102 4.1 102 3.3 107
23 ± ± ± ± ± ± ±
± ± ± ± ± ±
47 175 92 30 0.9 54 1.3 3.5 94 4.7 107 3.6 108
12 11 20* 5* 0.5 34 0.6 1.1 30 1.1 17 1.0 22
Moderate OSA (30 < AHI ~ 50)
29 ± 12 ± 7 ± 15* ± 5* ± 0.5 ± 34 ± 0.5 ± 0.8 ± 20 ± 0.9 ± 16 ± 0.8 ± 22
53 173 91 30 0.7 43 1.2 3.4 95 4.4 106 3.4 110
11 ± 7 ± 10 ± 14* ± 3* ± 0.6 ± 22 ± 0.6 ± 0.8 ± 19 ± 0.9 ± 21 ± 0.7 ± 24
Severe OSA (AHI> 50) 12 50 172 111 37 0.4 31 0.9 3.1 86 4.4 102 3.5 108
± ± ± ± ± ±
± ± ± ± ± ± ±
10 8 24 7 0.4 33 0.6 0.5 14 1.1 17 0.8 19
Definition of abbreviations: AHI = apnealhypopnea index; BMI = body mass index; ERV. = expiratory reserve volume at atmospheric pressure; ERV, = expiratory reserve volume at 4.1 cm H,O pressure. • Significantly different trom severe OSA group.
TABLE 2 SLEEP DATA Groups All Patients
Control (AHI ~ 10)
75
n AHI SI dBmax Awake O,sat Lowest O,sat Mean O,sat
30 210 84 98 81 93
± 33 ± ± ± ±
255 10 1 12 ± 3
Mild OSA (10 < AHI ~ 30)
3 65 79 98 88 94
23 ± 3 ± 104* ± 11* ± 1 ± 5*t ± 1*
20 226 83 98 83 94
Moderate OSA (30 < AHI ~ 50)
29 ± 6 ± 222 ± 9 ± 1 ± 6* ± 2*
41 220 89 97 76 93
11 ± 5 ± 239 ± 3 ± 1 ± 7* ± 1*
Severe OSA (AHI> 50) 96 424 89 97 62 89
12 ± 27 ± 355 ± 9 ± 2 ± 15 ± 4
Definition of abbreviations: AHI = apnealhypopnea index; SI = snoring index; dBmax = maximal nocturnal sound intensity; O,sat = oxygen saturation. • Significantly different from severe OSA group. t Significantly different from moderate OSA group.
TABLE 3 PHARYNGEAL AREAS, LUNG VOLUME DEPENDENCE, AND DISTENSIBILITY
in lowest nocturnal oxygen saturation when compared with the control group. Pharyngeal areas and distensibility were not significantly different between the four groups (table 3). Betweensubjects variability in pharyngeal areas (between 20 and 30070) greatly exceeded within-subject variability of repeated measurements (between 5 and 10%). PLVD was significantly higher in patients with severeOSA than in control subjects and patients with mild OSA; there was no significant difference in PLVD among the control, mild, and moderate OSA groups. Prior to performing stepwise, forward, multiple regression analysis between AHI or SI (dependent variables) and age, body mass index (BMI), pharyngeal areas at RV and FRC, PLVD, Cph (independent variables), we first performed independent correlations between these variables. The results (table 4) showed that only BMI, pharyngeal area at RV, and PLVD correlated significantly with the AHI (r = 0.50, p < 0.0001; r = - 0.24, P < 0.05; r = 0.50, p = 0.0001, respectively). SI correlated significantly only with BMI (r = 0.39, p < 0.005). Subsequent multiple linear regression analysis using the above independent variables revealedthat the AHI was correlated only with BMI and PLVD, whereas the SI was correlated only with BMI (table 5). These correlations, although weak, were statistically significant. The scatter plot of PLVD versus Cph is shown in figure 2. No significant correlation was found between these two parameters (r = 0.15, p = 0.20).
Groups All Patients n AFRC., em' AFRC., em' ARV, em' ATLC, em' PLVD, em'/L Cph, em'/em H,O
4.3 5.7 3.6 6.0 1.4 0.33
75 ± 1.0 ± 1.4 ± 1.0 ± 1.7 ± 2.2 ± 0.23
Control (AHI ~ 10) 23 4.4 5.9 3.6 5.8 1.1 0.39
± 1.0 ± ± ± ± ±
1.6 1.0 1.9 1.2* 0.26
Mild OSA (10 < AHI -c 30) 29 4.5 ± 1.0 5.5 ± 1.4 3.8 ± 1.2 6.2±1.7 0.7 ± 0.9* 0.26 ± 0.20
Moderate OSA (30 < AHI ~ 50) 4.4 6.0 3.5 6.1 1.3 0.39
11 ± 1.0 ± 1.5 ± 0.6 ± 1.7 ± 1.8 ± 0.21
Severe OSA (AHI> 50) 3.9 5.3 3.0 5.8 3.4 0.34
12 ± 0.8 ± 1.1 ± 0.7 ± 1.7 ± 4.2 ± 0.21
Definition of abbreviations: AFRC. = pharyngeal area at FRC at atmospheric pressure; AFRC, = pharyngeal area at FRC at 4.1 cm H,O pressure; ARV = pharyngeal area at RV; ATLC = pharyngeal area at TLC; PLVD = lung volume dependence ot pharyngeal area; Cph = pharyngeal distensibility. * Significantly different from severe OSA group.
hoc tests showed that only patients with severe OSA were different from the other three groups; there was no significant difference between the control, mild OSA, and moderate OSA groups. Sleep data (table 2) demonstrate that patients with severe OSA snored significantly louder and more frequently than
did control subjects, whereas patients with mild, moderate, and severe OSA were similar with respect to their snoring profile. Mean and lowest nocturnal oxygensaturations also weresignificantly lower in patients with severe OSA than in all other groups. Even patients with moderate OSA had significant reduction
TABLE 4 RESULTS OF CORRELATION ANALYSIS BETWEEN SNORING, APNEA, ANTHROPOMETRIC VARIABLES, AND PHARYNGEAL MECHANICS FOR ALL PATIENTS* AHI
SI
Age, yr
0.13550 0.2464
0.00386 0.9772
BMI
0.49618 0.0001
0.38989 0.0027
AFRC.
-0.10246 0.3817
-0.06187 0.6476
ARV
-0.23764 0.0401
-0.07647 0.5718
0.50133 0.0001
0.23693 0.0760
- 0.01101 0.9253
-0.11845 0.3802
PLVD Cph
For definition of abbreviations, see table 4. • For each variable, the first row gives Pearson's correlation coefficient, and the second row gives the associated p value.
1297
PHARYNGEAL FUNCTION AND SNORING CHARACTERISTICS IN APNEIC SNORERS
UTe 3) is clearly reflected by the difference in distensibilities at each pressure: 0.44 ± 0.29, 0.31 ± 0.19, and 0.32 ± 0.17 cmvcm H 20 at pressures 4.1, 6.5, and 8.5 em H 20, respectively. If the pressure-area curve was linear, then these distensibilities should be the same; however, in this case the analysis of variance showed that Cph at 4.1 em H 20 was significantly (p < 0.05) greater than that at 8.5 em H 20. We found no significant correlation between PLVD and Cph at any pressure in this subset of 33 patients. AHI did not correlate significantly with any of the distensibilities, but it showed a significant correlation with PLVD (r = 0.48, p < 0.005), similar to our finding for the larger group of 75 patients. Correcting the distensibilities for the effect of lung volume did not alter these conclusions. The correction term itself became progressively more important at higher pressures, reflecting larger increments in lung volumes: at 4.1 em H 20 the correction term was 10%, whereas at 8.5 cm H 20 the correction term was 300/0.
also demonstrated that the above two parameters of pharyngeal function do not reliablydistinguish between patients with different degrees of sleep apnea. Cph was not an important determinant of sleep Dependent Independent Partial Model R' R' P Value Variable Variables apnea. PLVD was a little better predictor of sleep apnea, accounting for 25% PLVD 0.2513 0.2513 0.0001 AHI of the variability in the AHI. However, 0.0805 0.3319 0.0043 BMI it distinguished only patients in the ex0.1520 0.1520 0.0027 SI BMI tremes of the apnea range: those with For definition of abbreviations, see tables 1 and 3. AHI less than 10 from those with AHI greater than 50. An additional feature of this study not present in previous investigations is a All of the above analysis was repeated well-characterized group of patients exusing Cph distensibility corrected for the amined here. Not only did they complain change in lung volume caused by appliof snoring, but wealso documented snorcation of positive pressure. The correcing objectively. This is important because tion term was approximately 10%. None snoring has been shown to influence phaof the above conclusions was altered as ryngeal mechanics independently of sleep a result of this correction. Similarly, our apnea; on the basis of our measurements, conclusions remained unchanged when we are certain that all of our subjects did specific distensibility (Cph/AFRCo) was indeed have a substantial degree of used instead of Cph. snoring. We also examined separately a subWe were surprised by the lack of relagroup of 33 patients in whom measurements of pharyngeal area at 4.1, 6.5, and tionship between PLVD and Cph. Our starting hypothesis was that these mea8.5 em H 20 wereavailable (figure 3). Cph surements should be related since both at each of the above pressures were comreflect pharyngeal mechanics. One posputed by subtracting pharyngeal area at Discussion atmospheric pressure from pharyngeal This study indicates that two parameters sible explanation for the lack of relationship may lie in the fact that the two meaarea at each positive pressure and di- describing awake pharyngeal functionviding by the appropriate pressure dif- PLVDand Cph - are independent and do surements represent the final common ference. The nonlinear nature of the not correlate with each other. The study pathway of the interaction of different upper airway muscle groups and intrinpressure-area curve of the pharynx (figsic tissue properties (25). Cph, i.e., change in pharyngeal area in response to externally applied pressure, most likelyreflects the resting, passive tone of the upper ~ 1r airway muscles. PLVD, i.e., lung-volume* * ..0 0.8 L related change in pharyngeal area dur'iii * c: ing active expiration, constitutes a volun.srJl 0.8 tary effort accompanied by the activazs tion.of several muscle groups. Thus, it *4*' Fig. 2. Scalier plot illustrating lack of might be expected that the two measure0.4 relationship between PLVD and Cph. V*",--*.* * * ** * OJ ~ ments do not correlate with each other c: * *JI"'" * >since they both reflect different aspects «; 0.2 * ) : - * * s: of pharyngeal mechanics. a.. '***** * *':....L_~_---'_ _L...--_-'--. o '-----""*"-*l
v.
HOFFSTEIN, S. WRIGHT, N. ZAMEL, and T. D. BRADLEY3
Introduction
It is thought that patients with obstructive sleep apnea (OSA) have abnormal pharyngeal structure and function (measured even when these patients are awake), predisposing them toward development of complete airway obstruction during sleep (1-13). Continuing interest in studying pharyngeal properties in patients with sleep apnea is partially stimulated by a desire to find a good screening test for this disorder. Such a test would help us to identify patients at risk and thus reduce the number of expensive and time-consuming sleep studies. Pharyngeal structure is usually assessed by measurements of pharyngeal cross-sectional area using a number of different techniques such as computerized tomographic scans (3, 5, 6), videofluoroscopy (1-4), magnetic resonance imaging (14), X-ray cephalometry (1517),and acoustic reflections (7). Pharyngeal function is usually assessed by some measurements reflecting the dynamic behavior of the pharynx such as pharyngeal resistance (10), change in area in response to collapsing or distending pressure (8, 9, 18), or change in area in response to changes in lung volume (11, 19). The latter two measurements, termed pharyngeal distensibility (Cph) and pharyngeal lung volume dependence (PLVD), respectively(11, 20), may be performed using the acoustic reflection technique for measuring airway area. The equivalence between the above two measures of pharyngeal function has not been demonstrated. From the technical point of view, it would be advantageous to know whether the changes in pharyngeal distensibility parallel changes in PLVD because the measurements of pharyngeal area as a function of pressure are more difficult to make than measurements of pharyngeal area as a function oflung volume. If the two measurements are closely associated, we might be able to replace the technically more difficult measurement (Cph) with the relatively simpler one (PLVD). With this in mind, 1294
SUMMARY Abnormalities in pharyngeal function, manifested even when the patients are awake, are thought to play an important role in the pathogenesis of sleep apnea. Tests of awakepharyngeal function continue to stimulate interest because it is hoped that they may allow physicians to distinguish patients with sleep apnea from those without It, and therefore reduce the number of unnecessary sleep studies. We elected to stUdy two measures of pharyngeal function: changes in pharyngeal area with lung volume (PLVD)and changes in pharyngeal area In response to externally applied positive pressure, t.e., pharyngeal distensibility (Cph). Both measurements have been employed for assessment of pharyngeal function, and both are thought to reflect pharyngeal "floppiness." Measurement of PLVD is technically very simple, whereas the measurement of Cph is technically more complex. If the two measurements are highly correlated, it might be possible to replace the technically more difficult one by the simpler one. Consequently, the purpose of this stUdy was twofold: first, to examine the relationship between pharyngeal distensibility and lung volume depen. dence of pharyngeal area, and second, to compare these parameters In a large group of confirmed snorers with and without obstructive sleep apnea (OSA). Westudied 75 unselected patients referred for the investigation of snoring and suspected sleep apnea. All patients had nocturnal polysomnography, pUlmonary function tests, and measurement of pharyngeal areas at TLC, FRC,and residual volume (RY) employing the acoustic reflection technique. The area measurement at FRC was performed at zero and at 4.1 em H20 positive airway pressure to calculate pharyngeal distensibility. We found no significant correlation between PLVDand Cph (r = 0.15, P = 0.20). Using a stepwise, forward, rnultlple linear regression model, we found that only body mass index and PLVD, but not Cph, were significant determinants of the variability In apnea/hypopnea index (AHI) (multiple r 2 0.33, P < 0.005). Neither PLVD nor Cph correlated with snoring. When we divided our patients Into four groups based on an AHi of less than 10 (controls), between 11 and 30 (mild OSA), between 31 and 50 (moderate OSA), and greater than 50 (severe OSA), we found that PLVD, but not Cph, separated the severe group from the control and the mild groups. We conclude that (1) measurements of PLVD and pharyngeal compliance are not equivalent, and (2) lung-volume-associated changes in pharyngeal area are related to the severity of sleep apnea.
=
AM REV RESPIR DIS 1991, 143:1294-1299
the present study was designed to test the hypothesis that Cph is correlated with PLVD. Methods Patients Westudied 75 unselected consecutive patients referred to our sleep clinic for evaluation of snoring. Each of these subjects admitted to being a long-time heavy snorer; all subjects suffered various consequences of snoring with respect to their marital and/or social relationships, which usually precipitated their referral to the clinic. Although when interviewed some of these patients werealso found to have other complaints such as excessive daytime sleepiness, morning tiredness/fatigue, and restless sleep, the presenting complaint in all of them was snoring. There were 65 male and 10 female patients ranging from 16 to 69 yr of age. Measurements of Pharyngeal Mechanics The area of the pharynx was measured using the acoustic reflection technique, extensively
described previously (7,11,18, 19).This technique allows us to construct an "airway echogram," i.e., a plot of airway area versus distance from the mouth. On this plot we identified the "pharynx" as the segment contained between the teeth and the glottis, and computed the mean area of this segment, referred to in the subsequent discussion as "pharyngeal area." The measurements wereperformed according to three different protocols: (1) patients breathing quietly at FRC, (2) patients breath-
(Received in original/arm December 12, 1989 and in revised form January 22, 1991) 1 From the Department of Medicine, St. Michael's, Mt. Sinai, and The Toronto Hospitals, , Universityof Toronto, Toronto, Ontario, Canada. 2 Correspondence and requests for reprints should be addressed to Dr. Victor Hoffstein, St. Michael's Hospital, 30 Bond Street, Toronto, Ontario, Canada M5B IW8. 3 Career Scientist of the Ontario Ministry of Health.
PHARYNGEAL FUNCTION AND SNORING CHARACTERISTICS IN APNEIC SNORERS
ing at FRC but with 4.1 cm H 2 0 positive pressure applied at the mouth, and (3) measurements taken during a slow quasi-static expiration from TLC to residual volume (RV);this allowed us to obtain a relationship between pharyngeal area and lung volume, i.e., volume-area plot. Positive airway pressures were generated by loading the spirometer, located at the distal end of the wave tube, with different weights. We used 1-, 1.5-, and 2-kg weights, which generated pressures of 4.1, 6.5, and 8.5 cm H 2 0 , respectively. In the first 33 patients, airway areas were obtained at all three pressures. We elected to use only the results at low pressure (4.1 em H 2 0 ) in order to avoid the following problems: (1) large changes in lung volume, which also influence pharyngeal area independently of applied pressure; (2) open velum (21), which leads to overestimation of pharyngeal area; (3) possible nonlinearity of the pressure-area curve of the pharynx at higher pressures. For these reasons, we performed all pressure-area measurements at a pressure of 4.1 em H 20 in the rest of the patients. Measurements of area were performed five times per second, and there were 128 measurements available at FRC, and as many as 24 measurements at TLC and at RV. At each lung volume weaveraged all pharyngeal areas and calculated the mean area for that lung volume. From the measurements of pharyngeal area we derived two indices of pharyngeal func-. tion: PLVD and Cph, PLVD was defined as the difference in pharyngeal area between FRC and RV divided by the expiratory reserve volume (ERV), i.e., (AFRC - ARV)/ERV. Cph was defined as the difference in pharyngeal areas measured at 4.1 ern H 2 0 pressure and at atmospheric pressure (AFRC. and AFRCo, respectively), per unit pressure, i.e., Cph = (AFRC. - AFRCo)/4.1. We also computed specific Cph, which is defined as the Cph normalized by pharyngeal area at FRC
ATLe
p>o
A RV
RV
FRG
FRG
TLG
(0 em H 2 0 ) (4.1 em H 2 0 )
Lung Volume
Fig. 1. Schematic diagram illustrating how to correct pharyngeal area at a pressure P for the effect of lung volume. A. = pharyngeal area at FRC at atmospheric pressure; Ap = pharyngeal area at FRC measured at a positive transpharyngeal pressure; I1A = increment in area; I1V = increase in lung volume; ATLe = pharyngeal area at total lung capacity; ARV = pharyngeal area at residual volume.
at atmospheric pressure, i.e., sCph = Cph/ AFRCo• Because the application of positive airway pressure increases lung volume, which in turn leads to increase in pharyngeal area, it is necessary to correct pharyngeal area for this lung volume effect. This is illustrated schematically in figure 1, which shows two volume-area curves, one obtained during quiet, quasi-static expiration from TLC to RVat the atmospheric pressure, and the other one obtained at a positive pressure P. A, is the pharyngeal area at FRC at atmospheric pressure. Ap is pharyngeal area at FRC measured at a positive transpharyngeal pressure P; it is composed of an increment in area (AA) caused by an increase in lung volume (AV), and an increment in area caused by the application of positive transpharyngeal pressure. Uncorrected Cph is simply (Ap - Ao)/P, whereas the corrected Cph is given by [Ap - A, - AA)/P, where AA = AV multiplied by the slope of the pharyngeal area-lung volume curve. V is taken to be equal to the difference in ERV at atmospheric pressure and at positive pressure P, and the slope is taken to be equal to PLVD. This correction in itself is an approximation because it assumes that (1) pharyngeal area during tidal breathing at FRC is equal to that obtained during slow expiration when passing through the same lung volume, (2) the slope of the pharyngeal area-lung volume curve is constant, and (3) residual volume is not affected by the application of positive transpharyngeal pressure P. We performed all calculations using corrected and uncorrected Cph.
Sleep Studies
1295
and therefore only the spikes in sound intensity exceeding 60 dB were classified as snores. The polysomnographic tracings were analyzed by experienced sleep technologists unaware of the results of the pharyngeal mechanics measurements. Sleep stages were scored according to the criteria of Rechtschaffen and Kales (23). Apneas were determined by the absence of tidal volume excursions on the respiratory inductance plethysmograph of at least 10 s and absence of airflow at the nose and the mouth. Obstructive apneas were scored when the paradoxical chest wall motion on the respitrace was seen. Hypopneas were defined as 50% reduction in tidal volume below the baseline value lasting longer than 10 s (24). Apnea/hypopnea index (AHI) was defined as the number of these events per hour of sleep, and snoring index (SI) was defined as the number of snores per hour of sleep.
Pulmonary Function Tests In all patients we measured flow-volume curves using a dry roIling-seal spirometer (CPI 220; Cardio-Pulmonary Instruments, Houston, TX) and FRC using a constant-volume body plethysmograph (Ohio 3000 system; Ohio Instruments, Madison, WI). TLC was obtained by adding the inspiratory capacity to FRC, RV was obtained by subtracting vital capacity from TLC, and ERV was calculated as the difference between FRC and RV.
Data Analysis We used stepwise, forward, multiple linear regression analysis to examine whether PLVD, Cph, pharyngeal areas, and anthropometric variables were significant determinants of the variability in the AHI and SI. Correlation analysis was used to search for any single relationships between the above variables as well as between PLVD and Cph. The analysis was repeated after correcting Cph for lung volume and also using specific distensibility. To determine if PLVD or Cph can separate apneic from nonapneic patients, and realizing that any predetermined AHI cutoff to define sleep apnea is somewhat arbitrary, we decided to split our group of 75 patients into four subgroups: control (AHI < 10), mild apnea (10 ~ AHI ~ 30), moderate apnea (31 ~ AHI ~ 50), and severe apnea (AHI > 50). We used analysis of variance with Thkey's posthoc tests to compare the anthropometric data, pulmonary function, sleep data, and pharyngea mechanics between these four groups. All statistical analysis was carried out using the SAS statistical software (release 6.03; SAS Institute, Gary, IN). Statistical significance was defined as p < 0.05.
All patients had full nocturnal polysomnography performed in the standard manner, with monitoring of electroencephalogram (EEG), right and left electrooculogram (EOG), submental and anterior tibial electromyogram (EMG), electrocardiogram (ECG), oronasal flow, chest wall and abdominal excursion by impedance plethysmography, oxygen saturation, and snoring. All measurements were displayed on a Grass polygraph (Model 78D;, Grass Instruments, Quincy, MA). EEG, EOG, ECG, and EMG were measured using the Grass amplifiers appropriate for the Model 78D recorder. Oronasal flow was measured using oral and nasal thermistors. Chest wall and abdominal excursions were obtained using a respiratory inductance plethysmograph (Respitraces; Ambulatory Monitoring, Inc., Ardsley, NY). Oxygen saturation was measured using a Biox finger oximeter (Biox 3700; Ohmeda, Boulder, CO). Snoring was measured in a manner similar to our previous study (22) using a microphone and a Results sound meter ~L1220; Pacer Industries, TorThe anthropometric data (table 1)for the onto, Ontario); the microphone-sound meof 75 patients as well as for entire group ter system was calibrated in the range of 40 to 110dB using a I-kHz signal and a sound the different subgroups (control, mild, moderate, and severe) demonstrate that chamber. The microphone was taped to the chest wall band of the Respitrace. With this only weight and BMI were significantly setup, quiet breathing registered below 60 dB, different between the four groups. Post-
1296
HOFFSTEIN, WRIGHT, ZAMEL, AND BRADLEY
TABLE 1 ANTHROPOMETRIC AND PULMONARY FUNCTION DATA Groups All Patients n Age, yr Height, em Weight, kg BMI, kglm' ERV., L 0lb pred ERV., L FRC, L % pred FVC, L % pred FEV" L alb pred
48 172 93 31 0.8 51 1.2 3.4 95 4.4 105 3.4 108
Mild OSA (10 < AHI ~ 30)
Control (AHI ~ 10)
75 ± 11 ± 9 ± 20 ± 6 ± 0.5 ± 33 ± 0.5 ± 0.9 ± 23 ± 1.0 ± 17 ± 0.8 ± 21
45 169 84 29 0.9 63 1.1 3.4 102 4.1 102 3.3 107
23 ± ± ± ± ± ± ±
± ± ± ± ± ±
47 175 92 30 0.9 54 1.3 3.5 94 4.7 107 3.6 108
12 11 20* 5* 0.5 34 0.6 1.1 30 1.1 17 1.0 22
Moderate OSA (30 < AHI ~ 50)
29 ± 12 ± 7 ± 15* ± 5* ± 0.5 ± 34 ± 0.5 ± 0.8 ± 20 ± 0.9 ± 16 ± 0.8 ± 22
53 173 91 30 0.7 43 1.2 3.4 95 4.4 106 3.4 110
11 ± 7 ± 10 ± 14* ± 3* ± 0.6 ± 22 ± 0.6 ± 0.8 ± 19 ± 0.9 ± 21 ± 0.7 ± 24
Severe OSA (AHI> 50) 12 50 172 111 37 0.4 31 0.9 3.1 86 4.4 102 3.5 108
± ± ± ± ± ±
± ± ± ± ± ± ±
10 8 24 7 0.4 33 0.6 0.5 14 1.1 17 0.8 19
Definition of abbreviations: AHI = apnealhypopnea index; BMI = body mass index; ERV. = expiratory reserve volume at atmospheric pressure; ERV, = expiratory reserve volume at 4.1 cm H,O pressure. • Significantly different trom severe OSA group.
TABLE 2 SLEEP DATA Groups All Patients
Control (AHI ~ 10)
75
n AHI SI dBmax Awake O,sat Lowest O,sat Mean O,sat
30 210 84 98 81 93
± 33 ± ± ± ±
255 10 1 12 ± 3
Mild OSA (10 < AHI ~ 30)
3 65 79 98 88 94
23 ± 3 ± 104* ± 11* ± 1 ± 5*t ± 1*
20 226 83 98 83 94
Moderate OSA (30 < AHI ~ 50)
29 ± 6 ± 222 ± 9 ± 1 ± 6* ± 2*
41 220 89 97 76 93
11 ± 5 ± 239 ± 3 ± 1 ± 7* ± 1*
Severe OSA (AHI> 50) 96 424 89 97 62 89
12 ± 27 ± 355 ± 9 ± 2 ± 15 ± 4
Definition of abbreviations: AHI = apnealhypopnea index; SI = snoring index; dBmax = maximal nocturnal sound intensity; O,sat = oxygen saturation. • Significantly different from severe OSA group. t Significantly different from moderate OSA group.
TABLE 3 PHARYNGEAL AREAS, LUNG VOLUME DEPENDENCE, AND DISTENSIBILITY
in lowest nocturnal oxygen saturation when compared with the control group. Pharyngeal areas and distensibility were not significantly different between the four groups (table 3). Betweensubjects variability in pharyngeal areas (between 20 and 30070) greatly exceeded within-subject variability of repeated measurements (between 5 and 10%). PLVD was significantly higher in patients with severeOSA than in control subjects and patients with mild OSA; there was no significant difference in PLVD among the control, mild, and moderate OSA groups. Prior to performing stepwise, forward, multiple regression analysis between AHI or SI (dependent variables) and age, body mass index (BMI), pharyngeal areas at RV and FRC, PLVD, Cph (independent variables), we first performed independent correlations between these variables. The results (table 4) showed that only BMI, pharyngeal area at RV, and PLVD correlated significantly with the AHI (r = 0.50, p < 0.0001; r = - 0.24, P < 0.05; r = 0.50, p = 0.0001, respectively). SI correlated significantly only with BMI (r = 0.39, p < 0.005). Subsequent multiple linear regression analysis using the above independent variables revealedthat the AHI was correlated only with BMI and PLVD, whereas the SI was correlated only with BMI (table 5). These correlations, although weak, were statistically significant. The scatter plot of PLVD versus Cph is shown in figure 2. No significant correlation was found between these two parameters (r = 0.15, p = 0.20).
Groups All Patients n AFRC., em' AFRC., em' ARV, em' ATLC, em' PLVD, em'/L Cph, em'/em H,O
4.3 5.7 3.6 6.0 1.4 0.33
75 ± 1.0 ± 1.4 ± 1.0 ± 1.7 ± 2.2 ± 0.23
Control (AHI ~ 10) 23 4.4 5.9 3.6 5.8 1.1 0.39
± 1.0 ± ± ± ± ±
1.6 1.0 1.9 1.2* 0.26
Mild OSA (10 < AHI -c 30) 29 4.5 ± 1.0 5.5 ± 1.4 3.8 ± 1.2 6.2±1.7 0.7 ± 0.9* 0.26 ± 0.20
Moderate OSA (30 < AHI ~ 50) 4.4 6.0 3.5 6.1 1.3 0.39
11 ± 1.0 ± 1.5 ± 0.6 ± 1.7 ± 1.8 ± 0.21
Severe OSA (AHI> 50) 3.9 5.3 3.0 5.8 3.4 0.34
12 ± 0.8 ± 1.1 ± 0.7 ± 1.7 ± 4.2 ± 0.21
Definition of abbreviations: AFRC. = pharyngeal area at FRC at atmospheric pressure; AFRC, = pharyngeal area at FRC at 4.1 cm H,O pressure; ARV = pharyngeal area at RV; ATLC = pharyngeal area at TLC; PLVD = lung volume dependence ot pharyngeal area; Cph = pharyngeal distensibility. * Significantly different from severe OSA group.
hoc tests showed that only patients with severe OSA were different from the other three groups; there was no significant difference between the control, mild OSA, and moderate OSA groups. Sleep data (table 2) demonstrate that patients with severe OSA snored significantly louder and more frequently than
did control subjects, whereas patients with mild, moderate, and severe OSA were similar with respect to their snoring profile. Mean and lowest nocturnal oxygensaturations also weresignificantly lower in patients with severe OSA than in all other groups. Even patients with moderate OSA had significant reduction
TABLE 4 RESULTS OF CORRELATION ANALYSIS BETWEEN SNORING, APNEA, ANTHROPOMETRIC VARIABLES, AND PHARYNGEAL MECHANICS FOR ALL PATIENTS* AHI
SI
Age, yr
0.13550 0.2464
0.00386 0.9772
BMI
0.49618 0.0001
0.38989 0.0027
AFRC.
-0.10246 0.3817
-0.06187 0.6476
ARV
-0.23764 0.0401
-0.07647 0.5718
0.50133 0.0001
0.23693 0.0760
- 0.01101 0.9253
-0.11845 0.3802
PLVD Cph
For definition of abbreviations, see table 4. • For each variable, the first row gives Pearson's correlation coefficient, and the second row gives the associated p value.
1297
PHARYNGEAL FUNCTION AND SNORING CHARACTERISTICS IN APNEIC SNORERS
UTe 3) is clearly reflected by the difference in distensibilities at each pressure: 0.44 ± 0.29, 0.31 ± 0.19, and 0.32 ± 0.17 cmvcm H 20 at pressures 4.1, 6.5, and 8.5 em H 20, respectively. If the pressure-area curve was linear, then these distensibilities should be the same; however, in this case the analysis of variance showed that Cph at 4.1 em H 20 was significantly (p < 0.05) greater than that at 8.5 em H 20. We found no significant correlation between PLVD and Cph at any pressure in this subset of 33 patients. AHI did not correlate significantly with any of the distensibilities, but it showed a significant correlation with PLVD (r = 0.48, p < 0.005), similar to our finding for the larger group of 75 patients. Correcting the distensibilities for the effect of lung volume did not alter these conclusions. The correction term itself became progressively more important at higher pressures, reflecting larger increments in lung volumes: at 4.1 em H 20 the correction term was 10%, whereas at 8.5 cm H 20 the correction term was 300/0.
also demonstrated that the above two parameters of pharyngeal function do not reliablydistinguish between patients with different degrees of sleep apnea. Cph was not an important determinant of sleep Dependent Independent Partial Model R' R' P Value Variable Variables apnea. PLVD was a little better predictor of sleep apnea, accounting for 25% PLVD 0.2513 0.2513 0.0001 AHI of the variability in the AHI. However, 0.0805 0.3319 0.0043 BMI it distinguished only patients in the ex0.1520 0.1520 0.0027 SI BMI tremes of the apnea range: those with For definition of abbreviations, see tables 1 and 3. AHI less than 10 from those with AHI greater than 50. An additional feature of this study not present in previous investigations is a All of the above analysis was repeated well-characterized group of patients exusing Cph distensibility corrected for the amined here. Not only did they complain change in lung volume caused by appliof snoring, but wealso documented snorcation of positive pressure. The correcing objectively. This is important because tion term was approximately 10%. None snoring has been shown to influence phaof the above conclusions was altered as ryngeal mechanics independently of sleep a result of this correction. Similarly, our apnea; on the basis of our measurements, conclusions remained unchanged when we are certain that all of our subjects did specific distensibility (Cph/AFRCo) was indeed have a substantial degree of used instead of Cph. snoring. We also examined separately a subWe were surprised by the lack of relagroup of 33 patients in whom measurements of pharyngeal area at 4.1, 6.5, and tionship between PLVD and Cph. Our starting hypothesis was that these mea8.5 em H 20 wereavailable (figure 3). Cph surements should be related since both at each of the above pressures were comreflect pharyngeal mechanics. One posputed by subtracting pharyngeal area at Discussion atmospheric pressure from pharyngeal This study indicates that two parameters sible explanation for the lack of relationship may lie in the fact that the two meaarea at each positive pressure and di- describing awake pharyngeal functionviding by the appropriate pressure dif- PLVDand Cph - are independent and do surements represent the final common ference. The nonlinear nature of the not correlate with each other. The study pathway of the interaction of different upper airway muscle groups and intrinpressure-area curve of the pharynx (figsic tissue properties (25). Cph, i.e., change in pharyngeal area in response to externally applied pressure, most likelyreflects the resting, passive tone of the upper ~ 1r airway muscles. PLVD, i.e., lung-volume* * ..0 0.8 L related change in pharyngeal area dur'iii * c: ing active expiration, constitutes a volun.srJl 0.8 tary effort accompanied by the activazs tion.of several muscle groups. Thus, it *4*' Fig. 2. Scalier plot illustrating lack of might be expected that the two measure0.4 relationship between PLVD and Cph. V*",--*.* * * ** * OJ ~ ments do not correlate with each other c: * *JI"'" * >since they both reflect different aspects «; 0.2 * ) : - * * s: of pharyngeal mechanics. a.. '***** * *':....L_~_---'_ _L...--_-'--. o '-----""*"-*l
