| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
1 INSERM U 492 et Service de Physiologie, Hôpital Henri Mondor, AP-HP, et Université Paris XII, Créteil, France. 2 Service de Physiologie, Hôpital Raymond Poincaré, AP-HP, Garches, France
CORRESPONDENCE: A.M. Lorino, Service de Physiologie – Explorations Fonctionnelles, Hôpital Henri Mondor, 94010, Créteil, France. Fax: 33 149812698
The aim of the study was to determine whether the response of respiratory impedance (Zrs) to decreasing levels of continuous negative airway pressure (CNAP) during wakefulness, differs in controls and subjects with obstructive sleep apnoea syndrome (OSAS).
Zrs was measured by the forced oscillation technique (4–32 Hz) in 15 controls and 21 patients with OSAS (apnoea/hypopnoea index >20 per sleep hour) with normal lung function, in the basal state and with application of decreasing CNAP of –5, –10, and –15 hPa. Respiratory resistance was extrapolated to 0 Hz (R0) and estimated at 16 Hz (R16) by linear regression analysis of respiratory resistive impedance versus frequency. Respiratory elastance (Ers) and inertance (Irs) were estimated by multilinear regression analysis of respiratory reactance versus frequency, and resonance frequency (RF) was determined as RF=(1/2
In both groups, R0, R16, Ers and RF significantly increased as the CNAP level decreased (p<0.0001 for all). R0, Ers, and RF increased significantly more in OSAS than in controls (p<0.01, 0.001, and 0.0001, respectively), independently of the severity of obesity. Receiver operator characteristic curves showed that the parameter which best detected OSAS was RF, with a sensitivity of 81% and 93% specificity for the 13.6 Hz cut-off point.
The results of the present study suggest that the response of respiratory impedance to decreasing continuous negative airway pressure levels, might allow detection of obstructive sleep apnoea syndrome in subjects with normal lung function.
)(Ers/Irs)0.5.
Obstructive sleep apnoea/hypopnoea syndrome (OSAS) is characterized by the occurrence of repetitive upper airway (UA) narrowing and/or closure during sleep. Obstructive respiratory events are generally explained by the inefficiency of the dilating muscles in counterbalancing the tendency of the pharyngeal airway to collapse, due to the inspiratory decrease in UA transmural pressure 1. The main factors predisposing subjects to OSAS appear to be: 1) a reduction of UA calibre 2; 2) an increase in UA resistance, as well as in pulmonary and respiratory resistances 3–5; 3) an increase in UA compliance 4, 6, 7; and 4) dysfunction of the UA dilating muscles 8.
The application of continuous negative airway pressure (CNAP) at the airway opening generates a subatmospheric intraluminal UA pressure that simulates the conditions promoting the occurrence of obstructive respiratory events. The effect of CNAP on pharyngeal cross-sectional area and resistance have been widely investigated in control and OSAS subjects 4, 8–11, but, to the authors' knowledge, CNAP-induced changes in the respiratory mechanical parameters remain poorly documented 12. The aim of the present study was, therefore, to investigate whether the response of respiratory mechanics to decreasing CNAP levels differs between control and OSAS subjects, and might provide a predictive index of OSAS. To this end, the forced oscillation technique (FOT) was used, which allows easy measurement of respiratory impedance (Zrs) during CNAP application.
 |
Materials and methods |
|---|
 TOP Materials and methods ResultsDiscussion |
|---|
29 kg·m–2 (mean=32.7±3.9 kg·m–2). All patients had normal lung volumes and forced expiratory flow rates (80% of predicted values), and no history of cardiopulmonary disease. Their apnoea/hypopnoea index, established previously on the basis of overnight polysomnography, ranged from 20–87 per sleep hour (mean 44±25 per sleep hour). The experimental protocol had the prior approval of the authors' hospital Ethics Committee, and all subjects gave informed consent.
Respiratory resistance measurement
Zrs was measured at the mouth by FOT. The pseudo-random forced flow used in this study was composed of 29 harmonics (4–32 Hz), with enhanced amplitudes at the lower frequencies, to limit the influence of spontaneous breathing. The forced signal, generated by a digital-to-analogue converter, excited, through a power amplifier, a 50-W loudspeaker (Audax HM 130 XO, Château du Loir, France) enclosed in a 25-litre rigid chamber and placed in parallel with the CNAP device, as previously described 13. Mouth flow was measured with a screen pneumotachograph (resistance=0.35 hPa·L–1·s; Jaeger, Wurzburg, Germany) connected to a differential pressure transducer (±70 hPa; Sensym SCX 01D, Sunnyvale, CA, USA), and mouth pressure by a similar pressure transducer referenced to the atmosphere. The pneumotachograph and the tubing were flushed by a constant bias flow (0.5 L·s–1). Pressure and flow data were collected over 16 s periods and high-pass filtered (3rd order, cut-off frequency 3.5 Hz) to eliminate the breathing noise. A fast Fourier transform algorithm was applied to adjacent 4 s periods. Impedance data were calculated from the auto- and cross-spectra of pressure and flow, averaged over 3–4 manoeuvres, and retained for analysis when coherence was 0.9 14. Respiratory resistance (Rrs) was submitted to linear regression analysis versus frequency over the 4–16 and 17–32 Hz frequency ranges. Rrs extrapolated to 0 Hz (R0) was derived from the first linear regression analysis, and Rrs estimated at 16 Hz (R16), from the second one 15. The frequency dependence of Rrs was assessed by the difference 
Rrs=R0–R16. Respiratory elastance (Ers) and inertance (Irs) were estimated by multilinear regression analysis of respiratory reactance versus frequency. Resonance frequency (RF) was determined as the frequency corresponding to the zero value of respiratory reactance, i.e. as RF=(1/2)(Ers/Irs)0.5.
Experimental protocol
Each subject was studied in the sitting position under four successive conditions: in the basal state (CNAP0), and with decreasing levels of CNAP of –5, –10, and –15 hPa (CNAP–5, CNAP–10, and CNAP–15, respectively) applied at the mouth. CNAP was generated by an adjustable vacuum source, set at the negative pressure required before the subject was connected to the breathing circuit, and continuously monitored with a manometer. The subject, wearing a nose clip and firmly supporting his cheeks with his hands, was then connected to the CNAP device, and asked to breathe quietly and not to fight the applied pressure. After a 30 s period of adaptation to CNAP, forced oscillation was applied for 16 s, and the subject was disconnected from the CNAP device for 2 min. This sequence was repeated 3–4 times at each CNAP level, and a 5 min period was allowed to pass after the last release from one CNAP level before the first application of the next CNAP level.
Data analysis
Values are mean±sem, except when otherwise indicated. Statistical analysis of data was performed using linear regression analysis, analysis of variance, paired t-tests, and unpaired t-tests. A value of p<0.05 was considered significant.
The sensitivity and specificity of possible cut-off points for R0, Ers, and RF, to discriminate between control subjects and OSAS patients, were studied using the receiver operator characteristic (ROC) curve 16. The ROC curve allowed the visualization of the true positive rate (sensitivity) as a function of the false positive rate (1–specificity) for the different values of the respiratory parameters, and the determination of the cut-off point corresponding to the best compromise between sensitivity and specificity, defined as the closest point of the curve to the upper left hand corner.
|
Results |
|---|
|
TOPMaterials and methodsResultsDiscussion |
|---|
With CNAP0, R0 and R16 were significantly higher in groups OSAS1 and OSAS2 than in controls, but the small difference between OSAS1 and OSAS2 was not significant (fig. 1
). Within each group, no significant difference was observed between R0 and R16. Ers and RF were slightly but significantly higher in group OSAS2 than in controls and OSAS1 (fig. 2), and Irs was slightly but significantly lower in OSAS2 than in controls (0.012±0.001 versus 0.014±0.001 hPa·s2·L–1).
|
|
. Interestingly, the differences observed for the respiratory mechanical parameters at CNAP0 between OSAS1 and OSAS2, disappeared as the CNAP level decreased, and no significant differences were observed for the mechanical parameter responses to decreasing CNAP levels between OSAS1 and OSAS2 (figs. 1 and 2
). Therefore, in what follows, the results will be presented for controls and OSAS, without making a distinction between OSAS1 and OSAS2.
|
Rrs=R0–R16 increased significantly as the CNAP level decreased (p<0.0001 for all, fig. 4). The CNAP-induced increases in R0 and Rrs were significantly greater in patients than in controls (p<0.01 and 0.001, respectively). Ers and RF significantly increased as the CNAP level decreased (p<0.0001 for both) (fig. 5), whereas no significant change was observed for Irs. The CNAP-induced increases in Ers and RF were significantly greater in patients than in controls (p<0.001 and 0.0001, respectively).
|
|
).
|
). OSAS patients were detected optimally by the cut-off points of 6.6 hPa·L–1·s for R0 with a 67% sensitivity and a 93% specificity, 66.7 hPa·L–1 for Ers with a 90% sensitivity and a 67% specificity, and 13.6 Hz for RF with an 81% sensitivity and a 93% specificity.
|
|
Discussion |
|---|
|
TOPMaterials and methodsResultsDiscussion |
|---|
Initially, the authors divided the patients into two groups, in order to dissociate the respective influences of OSAS and BMI on the mechanical respiratory parameter values. That way, any difference between controls and OSAS1 could be attributed to the presence of OSAS, whereas any difference between OSAS1 and OSAS2 could be attributed to the severity of obesity.
It would have been more relevant to the clinical situation of OSAS if impedance measurements had been made in the supine position and via the nose. However, both the supine position and application of CNAP tend to promote nasal congestion, which would have resulted in unstable Rrs values. Furthermore, it may be assumed that the influence of nasal congestion on Rrs would have depended on the CNAP level and thereby biased the authors' results. In addition, neck position is more difficult to control when supine, and neck extension or flexion is known to affect UA geometry 17, and consequently Rrs. Lastly, compared to the sitting position, the supine position induces an immediate decrease in end-expiratory lung volume (EELV), and consequently increases in Rrs, Ers Irs, and RF 18, which might have affected the responses of these parameters to decreasing CNAP levels.
The FOT was used, as opposed to the classical oesophageal balloon technique for assuming respiratory mechanics, because: 1) FOT is a noninvasive method, i.e. it does not necessitate the placement of a nasopharyngeal catheter which may influence UA resistance during CNAP application 19; 2) FOT provides respiratory parameters, i.e. parameters which describe the mechanical behaviour of both the pulmonary component and the chest wall component, including respiratory muscles; and 3) with appropriate frequencies, FOT allows measurement of Newtonian respiratory resistance, and delayed respiratory resistance not including tissue viscoelasticity 15, 20, as well as respiratory reactance and its derived parameters. The authors, therefore, chose pseudorandom forced flow rather than a monosinusoidal oscillation, in order to study the frequency dependence of respiratory resistance, and to determine the parameters characterizing respiratory reactance, namely respiratory elastance, inertance, and resonance frequency. However, the pseudorandom forced flow did not allow partitioning of airway resistance into inspiratory and expiratory resistances.
The authors chose two complementary indices of respiratory resistance, namely R0 and R16. Indeed, Rrs, which was roughly constant over 4–32 Hz in subjects with normal lungs, showed negative frequency dependence when gas redistribution within the airways occurred, as a result of series or parallel inhomogeneities 21. As this frequency dependence occurs mainly below 16 Hz, R16 can be considered to reflect Newtonian respiratory resistance, and R0, total respiratory resistance, i.e. Newtonian resistance plus the delayed airflow resistance resulting from gas redistribution, when present 15, 20. It also appeared interesting to consider the parameters derived from respiratory reactance. Indeed, parallel variations of R0 and RF have been reported in the course of bronchial challenge tests to inhaled methacholine 22, and any increase in tissue elastance is associated with an increase in Ers, and thereby in RF.
Basal respiratory impedance
The basal R0, Ers, and Irs values are in the range of those previously reported in normal subjects with or without minimal obesity 18, 23. The fact that basal R0, and R16 values were significantly higher in OSAS1 and OSAS2 groups than in controls, may be attributed to the OSAS factor. This result is in accordance with previous reports that for a comparable BMI, OSAS subjects have smaller pharyngeal cross-sectional areas 2 and higher Rrs values than non-OSAS subjects 5. The fact that both basal R0 and R16 were not significantly different in OSAS1 and OSAS2 groups must be interpreted with caution, due to the small numbers in the groups. However, it suggests that airway geometry might be less influenced by the obesity factor than by the OSAS factor in subjects with normal lung volumes during wakefulness.
No significant difference was observed between basal R0 and R16 values in controls, OSAS1, and OSAS2, which illustrates the fact that Rrs was not frequency dependent. This result reflects homogeneous distribution of gas flow within the respiratory system 21, and would be expected in subjects with no detectable airway obstruction.
Only the mechanical parameters derived from respiratory reactance, namely Ers, Irs, and RF, which were significantly different in OSAS1 and OSAS2, were influenced by obesity (fig. 2). These results are in accordance with those previously reported by Zerah et al. 23 who attributed obesity-related increases in Ers mainly to those in chest wall elastance.
Respiratory response to decreasing continuous negative airway pressure levels
The CNAP-induced increases in R0 and R16 presently observed in controls (figs. 1 and 4) are comparable to those previously reported in a similar study 12, in which it was demonstrated that a direct lung volume effect on airway resistance could not totally explain the increase in Rrs 12, and that other factors were probably involved. These may include: 1) a narrowing of extrathoracic airways directly resulting from the CNAP-induced decrease in UA transmural pressure, since, for a given transrespiratory pressure, supralaryngeal resistance increases more during CNAP than during continuous positive extrathoracic pressure 9; and 2) an indirect effect of lung deflation, i.e. an additional increase in oropharyngeal resistance possibly due to an increase in UA collapsibility involving mechanical linkages between the thorax and the UA 24. However, it is likely that the Rrs response to CNAP also resulted partly from an increase in tissue resistance (Rti) due to the CNAP-induced decrease in lung volume 25.
The CNAP-induced increases in Ers reflect a decrease in respiratory compliance, i.e. in gas, and/or lung, and/or chest wall compliance. Indeed, CNAP application induces significant decreases in EELV 12, and consequently in gas compliance. Furthermore, these decreases in EELV have already been reported to result in reduced lung compliance under CNAP, due to the sigmoid shape of the lung volume/transpulmonary pressure relationship 26. Besides, decreases in respiratory reactance have previously been observed during resistive or elastic loading, and attributed to an increase in tissue elastance resulting from increased respiratory muscle activity 27. As it has been reported that CNAP application results in an increase in diaphragmatic and parasternal intercostal electromyographic activity 10, 28, the Ers response to CNAP presently observed may be explained by the enhanced activity of inspiratory muscles. As Irs remained unchanged when the CNAP level was decreased, the significant increases observed in RF directly result from those in Ers.
The fact that Rrs increased significantly as the CNAP level decreased, i.e. that Rrs became more and more frequency dependent, may be attributed to: 1) the increase in Rti resulting from reduced lung volume, which has been reported to be more pronounced at the lower frequencies 25; and 2) the occurrence and development of series and/or parallel gas redistribution 21. Series gas redistribution is mainly due to UA shunt compliance, whose influence on Rrs was probably reduced under CNAP, due to the increased activity of UA dilating muscles 10. Consequently, it is more probable that parallel gas redistribution developed within intrathoracic airways, as a result of the pulmonary inhomogeneities promoted by the CNAP- induced decrease in lung volume.
An interesting result is that none of the responses of the respiratory mechanical parameters to decreasing CNAP levels were significantly different in groups OSAS1 and OSAS2. This suggests that, whereas in the basal state, Ers, Irs, and RF were slightly influenced by the obesity level in subjects with normal lung volumes, at low CNAP levels they depend mainly on the physiological conditions of the ventilation, namely decreased lung volume and increased respiratory muscle activity. Consequently, any difference between the respiratory responses to CNAPs in control and OSAS subjects can be considered independent of the obesity level.
The improvement in the quality of the linear relationship between Ers and R0 when decreasing the CNAP level may be explained as follows: R0 represents the resistance to be overcome by the respiratory muscles, to the exclusion of the delayed resistance due to the tissue viscoelastic properties 15; Ers partly reflects respiratory muscle activity; both R0 and Ers increase as the CNAP level is lowered. As Irs remained unchanged, the improvement in the quality of the linear relationship between RF and R0 when decreasing the CNAP level may be similarly explained.
Comparisons between control and obstructive sleep apnoea syndrome subjects
Surprisingly, no significant difference was observed between the two groups regarding the response of R16 to decreasing CNAP levels. An enhanced response of UA resistance to negative transmural pressures might be expected in OSAS patients whose UAs are narrower and more collapsible 4, 8. Therefore, it could be speculated that UA patency was maintained at the cost of a greater increase in UA muscle electromyographic activity in OSAS than in normal subjects, as previously observed in patients during sleep 29. The CNAP-induced increase in UA muscle activity appears to result from the simultaneous occurrences of the decrease in lung volume, the low intraluminal pressure, and the increase in diaphragmatic activity, since it has been reported that: 1) the active reduction of EELV results in increased UA muscle activity 30; 2) when the CNAP-induced decrease in EELV is prevented by applying identical pressure to the body surface, UA muscle activity is enhanced, despite the absence of change in diaphragmatic activity 11; and 3) UA muscle activity responds to CNAP in a manner similar to diaphragmatic activity 10. The difference between the increases in Ers observed in the two groups strongly suggests that the CNAP-induced increase in diaphragmatic activity was greater in OSAS than in control subjects. In this connection, it has been reported that the activation of UA dilators is reflected by indices that quantify the central inspiratory drive, and that for the same increase in the central inspiratory drive, the decrease in pharyngeal resistance is more marked in OSAS patients than in normal subjects 31. It is worth noting that, during inspiration, the compensatory increase in activity of the airway dilator muscles minimizes the suction effect of CNAP on extrathoracic airways. Consequently, the CNAP-induced increase in UA resistance was probably less than during expiration. The resistance values were, however ,determined over the entire ventilatory cycle, and did not allow partitioning of airway resistance into inspiratory and expiratory resistances. Although no significant difference was observed between OSAS patients and control subjects regarding the responses of R16 to decreasing CNAP levels, may it be speculated that the respective contributions of inspiratory and expiratory resistances to R16 might have been different in OSAS and normal subjects.
The fact that the CNAP-induced increases in R0 and Rrs were significantly greater in OSAS than in controls, might be due to a greater increase in Rti and/or delayed airflow resistance resulting from intrapulmonary gas redistribution at low lung volume. OSAS may be an independent risk factor for small airway disease as assessed by abnormally low expiratory flow rates at 50 and 25% of the vital capacity 5. Subclinical abnormalities of distal airways might thus be revealed with application of CNAP in OSAS subjects with normal lung function. Besides, the fact that the decreases in EELV were more marked in the patients than in the controls, cannot be excluded.
When comparing patients to controls, the CNAP-induced increases in Ers and RF were significantly more marked than those in R0, probably because R0 does not take into account the resistive component due to the tissue viscoelastic properties 15. This component, which varies with lung volume 32, is partly responsible for the respiratory effort that Ers, and consequently RF, may be assumed to reflect.
When comparing the ROC curves for R0, Ers, and RF at CNAP–15, to discriminate between control and OSAS patients (fig. 7), it may be observed that the best results were obtained for RF which, for a given specificity, identified OSAS patients with the highest sensitivity. This shows that although highly correlated (fig. 6), the different indices derived from respiratory resistance and reactance do not provide redundant information.
In conclusion, this study demonstrated that the responses of respiratory impedance to decreasing continuous negative airway pressure levels differ in obstructive sleep apnoea syndrome and control subjects during wakefulness, and that this difference cannot be attributed to the obesity level. The results suggest that the response of respiratory impedance to decreasing continuous negative airway pressure, and more particularly the response of resonance frequency, which probably reflects the increase in respiratory muscle activity, might allow discrimination between control and obstructive sleep apnoea syndrome subjects with normal lung function. Thus, the application of subatmospheric pressures might be useful as a screening test for obstructive sleep apnoea syndrome during wakefulness, although additional studies including a larger number of subjects are needed before it can be recommended for this purpose.
Received: March 14, 2000
Accepted September 16, 2000
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
) with body mass index (BMI)<29 kg·m–2; OSAS2 patients (
) with BMI
). Data are presented as mean±sem. *: p<0.05 versus controls; **: p<0.01 versus controls; +: p<0.005 versus controls.
: CNAP 0 hPa; 
: CNAP –5 hPa; 
: CNAP –10 hPa; 
: CNAP –15 hPa.
: respiratory elastance; and 
: resonance frequency. The closed symbols represent optimal cut-off points.