INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Obstructive sleep apnea/hypopnea syndrome (SAHS) is a respiratory disorder characterized by recurrent airflow obstruction caused by total or partial collapse of the upper airway (1). These obstructive events can be caused by an increase in the collapsibility of the upper airway (2). In these circumstances, the negative transluminal pressure generated during inspiratory efforts induces the collapse of the upper airway. Continuous positive airway pressure (CPAP) therapy applied with a nasal mask is extensively used to compensate for the increased airway collapsibility (3). The level of CPAP therapy is titrated to maintain upper airway patency minimizing the obstructive events. Therefore, monitoring airflow obstruction during CPAP is helpful in assessing airway collapsibility and in adjusting the CPAP level.

Airflow obstruction during sleep can be assessed from the relationship between nasal airflow (V') and esophageal pressure (P_es). However, recording P_es requires the introduction of an esophageal catheter, which limits its clinical applicability in routine studies. By contrast, measuring respiratory resistance (R_rs) by means of the forced oscillation technique (FOT) is a noninvasive approach potentially useful in assessing airflow obstruction in SAHS. The method is based on superimposing a small-amplitude pressure oscillation onto the nasal mask of the patient. R_rs is computed from the oscillatory pressure and flow signals recorded at the nasal mask. We recently showed in a model study (4) that FOT can be applied during CPAP for continuously monitoring the degree of airflow obstruction of collapsible airways. Therefore, the aim of the present study was to test the clinical suitability of FOT for assessing inspiratory and expiratory airflow obstruction during nasal CPAP and to investigate the CPAP dependence of R_rs in patients with SAHS. The forced oscillation was applied at the nose during nasal CPAP. R_rs was continuously measured at increasing CPAP levels. Assessment of airflow obstruction by means of FOT was compared with esophageal pressure measurements.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Six male patients with normal spirometry and severe SAHS documented by previous diagnostic polysomnography were studied. They had 44.2 ± 3.4 yr (mean ± SEM), a body mass index of 34.7 ± 2.0 kg/ m², and an apnea-hypopnea index of 70.3 ± 4.4 events/h (range 52 to 85). The measurements were carried out during a period of stepwise, full-night, polysomnography-controlled CPAP titration. Electroencephalogram (EEG) (C4/A1, C3/A2), chin electromyogram (EMG), and electro-oculogram (EOG) for sleep staging according to standard criteria (5) were recorded. Sa_O2 was measured continuously with a finger probe (504; Critical Care Systems, Inc., Waukesha, WI). Rib cage and abdominal motion were monitored by bands placed over the thorax and abdomen. The parameters were recorded continuously on a polygraph (SleepLab 1000P; Aequitron, MN). Respiratory events were scored according to commonly used criteria of airflow cessation lasting 10 s or more for apneas and a 10-s or more period of discernible airflow reduction associated with an arousal and/or a 4% Sa_O2 dip for hypopneas. Arousals were scored according to the American Sleep Disorders Association recommendations (6).

CPAP was generated with a conventional device (CP90; Taema, Airliquide, France) connected to the patient with a tightly fitted nasal mask (Figure 1). A conventional leak valve was placed at the inlet of the nasal mask to avoid rebreathing. Care was taken to minimize flow leaks through the mask. V' was measured with a Fleisch no. 2 pneumotachograph (Sibel, Spain) located between the leak valve and the mask. The pneumotachograph was connected to a differential pressure transducer (± 2 cm H₂O, MP54; Validyne, Northridge, CA). The common mode rejection ratio of the flow measuring system was greater than 60 dB at 30 Hz. Nasal pressure (P_n) was measured with a similar pressure transducer (± 22 cm H₂O) connected to the mask. P_es was recorded with a piezoresistive transducer (± 71 cm H₂O, 176PC28HD2; Honeywell, Freeport, IL) connected to a balloon-tipped catheter placed in the mid-esophagus (7). The frequency responses at 5 Hz of the flow and pressure transducers were matched within 1% in gain and 1° in phase.

View larger version (14K): [in this window] [in a new window]   Figure 1.   Schematic diagram of the circuit used to apply forced oscillation during CPAP. LS = loudspeaker; PN = pneumotachograph; LV = leak valve; EB = esophageal balloon; PC = personal computer; V' = nasal flow; Pn = nasal pressure; Pes = esophageal pressure.

Small-amplitude pressure oscillation ( 1 cm H₂O) of 5 Hz was superimposed to the CPAP level by means of a loudspeaker (JBL-800GTI, 8-in subwoofer, 600 W; JBL, Vitoria, Spain) connected in parallel to the CPAP device (4). The rear part of the loudspeaker was attached to a 2-L closed chamber to withstand continuous positive pressures up to  20 cm H₂O. The raw flow and pressure recordings were analogically low-pass filtered (Butterworth 8-poles, 32 Hz) and digitized at 100 Hz with a microcomputer (486-type PC; 2831 Data Translation A/D-D/A board). The breathing components of these recordings were obtained by applying a moving average filter of 0.2 s (8). The 5-Hz oscillatory nasal flow and pressure components were computed by subtracting the breathing flow and pressure components from the respective raw recordings. R_rs was computed on a cycle-by-cycle basis from the Fourier transform of the oscillatory components of flow and pressure (4). Finally, R_rs was low-pass filtered with a moving average of 0.4 s.

The patient slept with the CPAP adjusted to a baseline level of 4 cm H₂O. After initiation of stable non-REM sleep, the forced oscillation was applied. Then, CPAP was increased approximately every 5 min in steps of 2 cm H₂O from the baseline level to optimal CPAP. In this study optimal CPAP was defined by the absence of arousals, snoring and respiratory events, and normal esophageal pressure swings.

Mid-inspiratory (R_rs,i) and mid-expiratory (R_rs,e) resistances were computed as the average over 0.4 s of the R_rs values centered at mid-inspiration and mid-expiration, respectively. Inspiratory pressure swings (P_es) were calculated as the difference in P_es between end-expiratory and peak inspiratory values. For each CPAP level, results of R_rs,i, R_rs,e, and P_es were computed as the average of the last two- three apnea or hypopnea cycles before an arousal. In the absence of arousals, three cycles at the end of the CPAP step were averaged. Data are reported as mean ± SEM. Differences between data obtained at baseline and at optimal CPAP and between inspiration and expiration were compared by means of paired t tests. The relationship between R_rs,i and P_es was analyzed by means of linear correlation. Tests were assumed to be significant at p < 0.05.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

All the patients exhibited obstructive events at baseline CPAP (4 cm H₂O). Three of the patients (numbers 2, 4, and 6) exhibited apneas at baseline CPAP. In Patient 4 apneas evolved to hypopneas when CPAP was increased to 6 cm H₂O. In Patient 2 and 6 apneas remained up to CPAP of 6 cm H₂O and evolved to hypopneas at CPAP of 8 cm H₂O. The other three patients presented hypopneas at baseline CPAP. The pattern of breathing of all the patients normalized progressively with increasing CPAP. Optimal CPAP was 11.3 ± 0.4 (range 10 to 12 cm H₂O). Four patients achieved a normal inspiratory flow pattern at optimal CPAP. By contrast, two patients (numbers 1 and 3) presented partial inspiratory flow limitation at optimal CPAP as indicated by relatively high values of R_rs,i and P_es.

Figure 2 shows representative changes in R_rs found in Patient 2 when CPAP increased. At baseline CPAP the patient exhibited periods of apnea with large breathing efforts (P_es = 36.2 cm H₂O) associated with a marked rise in oscillatory resistance (R_rs,i = 47.6 cm H₂O · s/L; R_rs,e = 45.4 cm H₂O · s/L). It is interesting to note that R_rs varied little during the apnea, thus indicating that airway closure remained during the expiratory phase. The apnea finished with an arousal which caused a sudden airway opening. Apneas were still present at CPAP = 6 cm H₂O, again with occluded airway both during inspiration (R_rs,i = 47.9 cm H₂O · s/L) and expiration (R_rs,e = 42.9 cm H₂O · s/L). At 8 cm H₂O airway obstruction evolved to hypopneas associated with substantial variations in resistance during the breathing cycle. R_rs rose sharply when the inspiratory flow reached a plateau with increasing P_es, indicating the beginning of inspiratory flow limitation. It is worth noting that resistance reached values (R_rs,i = 38.4 cm H₂O · s/L) close to those observed in the apnea during inspiration but fell sharply during expiration (R_rs,e = 10.0 cm H₂O · s/L). When CPAP was increased to the optimal level for this patient (12 cm H₂O) normal tidal flow was achieved with small breathing efforts (P_es = 9.3 cm H₂O) indicating restored airway patency. Consistently, R_rs exhibited a regular pattern with low values and minor changes during the breathing cycle (R_rs,i = 9.7 cm H₂O · s/L and R_rs,e = 5.9 cm H₂O · s/L).

View larger version (23K): [in this window] [in a new window]   Figure 2.   Breathing flow (V'), esophageal pressure (Pes), and respiratory resistance (Rrs) measured in a patient at different levels of CPAP.

Average values of R_rs,i and R_rs,e found at different CPAP levels are shown in Figure 3. R_rs,i decreased markedly and significantly from 36.0 ± 4.0 cm H₂O · s/L to 13.1 ± 2.8 cm H₂O · s/L when CPAP was raised from baseline to optimal level. Resistance was systematically lower during expiration but the difference did not reach statistical significance at CPAP  6 cm H₂O. R_rs,e decreased more sharply than R_rs,i reaching normal values about 4 cm H₂O below optimal CPAP.

View larger version (15K): [in this window] [in a new window]   Figure 3.   CPAP dependence of mid-inspiratory resistance (Rrs,i) and mid-expiratory resistance (Rrs,e). Values are mean ± SEM.

Phasic changes in resistance depended dramatically on the degree of flow obstruction. Figure 4 shows the average values of R_rs,i and R_rs,e of the recorded events pooled according to the degree of flow amplitude (V') defined as the difference between mid-inspiratory and mid-expiratory flows. In all the recorded apneas (V' < 0.05 L/s) resistance was high during inspiration and decreased very little during expiration. By contrast, in marked hypopneas (0.05 L/s < V' < 0.1 L/s) resistance remained high during inspiration but fell sharply during expiration. These huge phasic changes progressively disappeared as breathing was normalized with increased CPAP.

View larger version (29K): [in this window] [in a new window]   Figure 4.   Average mid-inspiratory resistance (Rrs,i) and mid-expiratory resistance (Rrs,e) as a function of flow amplitude (V') scored as the difference between mid-inspiratory and mid-expiratory flows.

Changes in the inspiratory pressure swings paralleled swings in R_rs,i. P_es exhibited high variability and fell significantly from 44.8 ± 11.4 cm H₂O at baseline to 13.8 ± 6.8 cm H₂O at optimal CPAP. Figure 5 demonstrates a strong correlation (r² = 0.94, p < 0.01) between inspiratory resistance and breathing effort.

View larger version (17K): [in this window] [in a new window]   Figure 5.   Scatter plot of mid-inspiratory resistance (Rrs,i) and esophageal pressure swings (Pes) of the events recorded at different CPAP levels. Values are mean ± SEM. Labels are CPAP levels in cm H2O. Dashed line: linear regression.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

FOT has been extensively used for assessing airway obstruction in spontaneously breathing awake patients. However, as far as we know, this is the first application of the technique to monitor inspiratory and expiratory airway obstruction during CPAP in patients with sleep apnea. In this work we found that the beginning of apnea or hypopnea events was associated with a marked increase in inspiratory resistance which correlated with the degree of inspiratory effort. Expiratory resistance remained high during apneas but decreased sharply during hypopneas. The reduction in the severity of obstructive events as CPAP rose resulted in a progressive normalization in both inspiratory and expiratory resistances. Nevertheless, achievement of airway patency required about 4 cm H₂O less in expiration than in inspiration.

FOT appears to be very well suited for assessing airflow obstruction during sleep. It is a noninvasive technique which does not require patient cooperation. Computations can be carried out automatically in real time. Furthermore, it is the only technique which allow us to measure airflow resistance in the absence of breathing flow. In this study, the oscillation frequency was set in such a way as to balance sensitivity and time discrimination. On the one hand, using low oscillatory frequency minimizes the shunt of the upper airway wall, improving the sensitivity of oscillatory resistance to changes in airway patency (4). On the other hand, low frequency increases the length of the time window of spectral computations, worsening time discrimination. The 5-Hz oscillation allowed us to compute resistance with enough time discrimination ( 0.2 s) to precisely detect the beginning and end of airway occlusions and to track changes in airway patency during the breathing cycle.

Applying forced oscillation during CPAP requires a generator able to withstand the positive pressure of the circuit. We generated the oscillation by means of a high-power loudspeaker with the rear side of the cone enclosed in a small-volume chamber (4). The gas in the rear chamber acts as a high elastance which helps to compensate the CPAP pressure applied to the front side of the loudspeaker's cone. With this method we were able to superimpose forced oscillation to clinical levels of CPAP. The low-amplitude pressure oscillations superimposed in the circuit did not interfere with the servocontrol system of the CPAP device. In the present study we applied forced oscillation with a loudspeaker placed in parallel with the CPAP device. We recently designed a new device to generate CPAP and forced oscillation simultaneously (9). This new approach simplifies the technique and could facilitate the use of FOT for automatic CPAP titration.

Oscillatory resistance was computed from airflow measured at the entrance of the mask. Therefore, a possible leak in the mask or through the mouth acts as a shunt resistance resulting in an underestimation of respiratory resistance. Although this shunt resistance does not affect the assessment of qualitative changes in airway patency it is important to fit the mask tightly to minimize leaks and optimize sensitivity. However, leak artefacts, which are common to other techniques that compute resistance from nasal flow recordings, can be easily detected as a drift in the volume signal.

We did not observe apparent changes in the neurological (EEG, EMG) or respiratory (V', P_es) variables when we connected the 5-Hz forced oscillation. By contrast, Henke and Sullivan (10) reported an increase in the activity of upper airway and inspiratory muscles associated with partial or complete reversal of the upper airway obstruction when pressure oscillation of 30 Hz was applied to sleep apnea patients. These different oscillatory responses could be explained by the higher frequency used in their study, which was within the frequency range of snoring vibration. Nevertheless, this discrepancy will have to be further addressed in future studies.

The progressive recovery of inspiratory airway patency with increasing CPAP was very well detected by FOT. Oscillatory resistance computed from nasal pressure and flow corresponds to the resistance of the total respiratory system. Because chest wall and lung tissue resistances decrease inversely with frequency (11, 12), R_rs at 5 Hz mainly accounts for airway resistance. Therefore, the rise in R_rs found during apneas and hypopneas can be mostly attributed to changes in upper airway patency. We recently showed in a model study (4) that oscillatory resistance of a collapsible airway achieves a high but finite value during total occlusion. This maximum value is determined by the shunt compliance of the upper airway walls and the possible leak resistance in the mask. We also demonstrated that oscillatory resistance measured during flow limitation can be interpreted as an index for quantifying the degree of airflow obstruction. This contrasts with lung resistance computed as the quotient between nasal flow and pressure drop [R_l = V' / (P_n  P_es)] which is an effort-dependent magnitude in flow limitation conditions (13) and is, therefore, meaningless during apneas and hypopneas. At baseline CPAP (4 cm H₂O), the patients exhibited apneas or hypopneas characterized by large inspiratory pressure swings (Figure 2) with null and low flows, respectively. Consistently with the model study, we found very high values of R_rs,i at baseline CPAP indicating a dramatic reduction in airway patency. As CPAP increased the patients achieved larger inspiratory flows with lower pressure efforts showing less inspiratory dynamic obstruction. Also in agreement with the model study (4), R_rs,i decreased with increasing CPAP indicating progressive recovery of airway patency. Our mean R_rs,i (13.1 cm H₂O · s/L) at optimal CPAP was in the range of R_l reported in normal subjects during sleep (14). Nevertheless, the two patients (numbers 1 and 3) with higher R_s,i at optimal CPAP exhibited a plateau at mid-inspiratory flow, suggesting that a slight flow limitation (4, 13, 17) still remained. This indicates that these patients normalized sleep before airflow resistance and that CPAP was titrated slightly below the level required to completely avoid inspiratory dynamic obstruction.

Phasic changes in resistance within the breathing cycle may provide information about upper airway collapsibility. The high values of resistance with small phasic variations observed in the apneas recorded at low CPAP indicate that the collapse of the upper airways persisted in the absence of inspiratory effort (Figure 2). This static occlusion suggests that the CPAP level was below the critical opening pressure (P_crit) of the upper airway (4, 18). On the other hand, the marked phasic variations in resistance, with high values during inspiration and lower values during expiration, observed in hypopneas with severe flow limitation suggest that CPAP was raised just above P_crit. In these circumstances, the inspiratory efforts diminish intraluminal upper airway pressure (P_in) below P_crit, then upper airway partially collapses and inspiratory flow becomes limited. During expiration, P_in is slightly higher than P_crit distending the upper airway as reflected by the fall in R_rs,e. As CPAP is raised sufficiently above P_crit to compensate for the dynamic inspiratory pressure drop, the upper airway remains distended during inspiration (P_in > P_crit). Then, R_rs,i approaches R_rs,e and normal tidal flow is achieved with small breathing efforts indicating restored inspiratory airway patency. This suggests that in order to avoid inspiratory dynamic collapse of the upper airway, CPAP should be raised until minimizing phasic changes of R_rs. Our findings reveal that patients with severe obstructive SAHS require lower CPAP pressure to normalize R_rs,e than R_rs,i. Therefore, the assessment of these phasic changes could be useful for setting different inspiratory and expiratory nasal pressures in patients with severe obstructive SAHS.

This study demonstrates the clinical applicability of FOT in assessing airflow obstruction during CPAP in patients with SAHS. This technique allowed us to monitor airflow obstruction with a dynamic response fast enough to detect sudden alterations in airway and to track phasic changes within the breathing cycle. Our findings suggest that FOT can be used as a noninvasive alternative to the esophageal balloon for CPAP titration. Furthermore, since oscillatory resistance can be automatically computed in real time, FOT is potentially useful for automatic CPAP titration.