INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Patients with obstructive sleep apnea/hypopnea syndrome (SAHS) suffer repeated episodes of increased upper airway resistance with partial or complete collapse that lead to profound disturbances in arterial blood gases and sleep architecture (1, 2). Repeated inspiratory efforts occur during obstructive events until arousal ensues and airway patency is restored (3). Continuous positive airway pressure (CPAP) therapy applied with a nasal mask is used extensively to compensate for the increased airway collapsibility (4). The level of CPAP is titrated to maintain upper airway patency and alleviate obstructive events. The increase in upper airway resistance that accompanies obstructive events is commonly assessed by indirect signals such as the time-profile of the inspiratory flow signal, or thoracoabdominal motion using strain gauges or inductance plethysmography (5). Airflow obstruction during sleep can also be assessed more directly by recording esophageal pressure (Pes) with an esophageal catheter, but the applicability of this approach in routine clinical studies is limited. The forced oscillation technique (FOT) provides a noninvasive method to directly assess airway obstruction. This technique consists of superimposing on spontaneous breathing a small pressure oscillation through a nasal mask (9). Respiratory impedance (|Z|) is derived from the pressure and flow signals recorded at the nasal mask. The potential applicability of FOT to assessing airway obstruction in SAHS has been confirmed in a model study (10). In that study the amplitude of the |Z| measured by FOT was found to be an accurate index of overall airflow obstruction. We have evaluated the suitability of the technique for diagnostic sleep studies in SAHS where |Z| was found to accurately reflect upper airway obstruction (11). We have also demonstrated the clinical applicability of FOT to assess airflow obstruction in real-time during CPAP treatment in a small number of patients with simultaneous recording of Pes (12). In one recent report from another group, FOT was applied intermittently for brief periods during overnight CPAP titration and appeared to accurately reflect the degree of upper airway obstruction (13). However, the continuous application of FOT throughout the night during CPAP titration has not been evaluated, and there are no published reports on the use of the FOT signal to identify optimal CPAP levels. The aim of the present study was therefore to apply FOT continuously during nasal CPAP titration and assess whether optimal CPAP can be accurately identified using the FOT signal, as well as to perform a descriptive analysis of the |Z| signal and evaluate practical issues surrounding the use of this measurement during CPAP titration.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Subjects

We studied a total of 28 subjects, all male, with a recent diagnosis of SAHS documented by polysomnography that had not been previously treated with CPAP. All patients had moderate to severe SAHS and required CPAP treatment. In 14 subjects (age = 53.2 ± 2.9 yr; BMI = 32.7 ± 1.2 kg/m²; FEV₁ = 77.2 ± 5.3% pred; Epworth sleepiness scale [ESS] = 14.8 ± 1.60, and AHI = 63.7 ± 3.1 events/h) the study was performed during a nap of at least 2 hours of stable sleep to determine the feasibility and accuracy of CPAP titration with this technique. In another 14 patients (age = 54.9 ± 2.49 yr; BMI = 32.5 ± 1.6 kg/m²; FEV₁ = 89.2 ± 3.7% pred; ESS = 13.4 ± 1.4, and AHI = 67.3 ± 2.89 events/h) CPAP titration was carried out during a complete nocturnal study lasting more than 6.5 h of sleep to obtain more information on the behavior of |Z| during CPAP use. Informed written consent was obtained from all subjects. This study was approved by the Human Ethics Committee of the Hospital Clínic.

Settings

CPAP titration was carried out in the setting of complete polysomnography (PSG) in our sleep laboratory with additional on-line measurement of oscillatory impedance using the FOT. PSG was performed in a conventional manner with continuous monitoring of the electroencephalogram (EEG) (O4/A1, C3/A2, Z1/A2), chin electromyogram (EMG), and electrooculogram (EOG) for sleep staging according to standard criteria (14). Sa_O2 was measured continuously with a finger probe using a pulse oximeter (504; Critical Care Systems, Inc., Waukesha, WI). Rib cage and abdominal motion were monitored by bands placed over the thorax and abdomen. Airflow was assessed using a plastic lightweight Fleisch-type pneumotachograph (dead space, 5 ml) constructed in our laboratory. These signals were recorded continuously on a polygraph (SleepLab 1000P; Aequitron, Minneapolis, MN). Respiratory events were scored according to commonly used criteria, with apnea defined as cessation of airflow lasting for 10 s or more and hypopnea as a reduction in airflow or thoracic-abdominal motion lasting 10 s or more, in association with an arousal or with a cyclical dip in Sa_O2. Arousal was defined according to the recommendations of the American Sleep Disorders Association (15). Microarousals were scored following the same criteria, but lasting at least 1.5 s and associated with EMG activity.

CPAP was generated with a conventional device (CP90, Taema; Airliquide, Antony Cedex, France) connected to the patient with a tightly fitted nasal mask. A conventional leak valve was placed at the inlet of the nasal mask to avoid rebreathing. The nasal mask was carefully fitted to minimize leaks. The pneumotachograph was located between the leak valve and the mask and connected to a differential pressure transducer (± 2 cm H₂O, MP54; Validyne, Northridge, CA). Nasal pressure (Pn) was measured with a similar pressure transducer (± 22 cm H₂O) connected to the mask. The assessment of airway obstruction was carried out with the FOT by measuring the amplitude of |Z|, which is the quotient between the amplitudes of oscillatory pressure and flow and represents the total mechanical load of the respiratory system (10). Therefore, increases in airway obstruction are accompanied by an increase in |Z|. Our setup has been extensively described in previous reports (10, 16). Briefly, a loudspeaker generated an oscillating pressure of small amplitude (1 cm H₂O peak-to-peak) at a frequency of 5 Hz. Nasal pressure and flow were low-pass-filtered at 16 Hz cutoff frequency and fed into an analogue circuit that provided continuous estimation of |Z| (17). The resulting analogue signal of |Z| was then fed into a channel of the polysomnographic recorder.

Titration Protocol

CPAP was titrated in the following manner. The study started with the CPAP adjusted to a baseline level of 4 cm H₂O and the patient was allowed to sleep in the position he preferred. After achieving a stable non-REM sleep an experienced sleep technician initiated a stepwise increase in CPAP (steps of 1 cm H₂O every 5 min) from baseline level to the optimal CPAP according to commonly used criteria of absence of respiratory-related arousals and sleep fragmentation associated with snoring, respiratory events, and inspiratory airflow limitation assessed by the morphology of the flow signal (8). When optimal CPAP was reached, an additional increase in the pressure of 2 cm H₂O was maintained for a period of 10 min. The pressure was then reset to the previous optimal pressure for the rest of the night.

The day after the measurement a physician from our sleep unit reviewed the study in order to verify the optimal CPAP defined by conventional PSG criteria. A second physician from our sleep unit, without knowledge of the previous results, reviewed the tracing of |Z| in isolation and determined the optimal CPAP from this signal. FOT was considered "optimal" when persistent or intermittent increases in |Z| lasting more than 10 s and exhibiting phasic changes ceased. Those segments of the record during which the FOT signal was considered optimal were identified, and the CPAP levels associated with these segments were subsequently recalled from the record. The predominant CPAP value for these segments was identified as the FOT-based optimal CPAP.

After each investigator had independently reviewed the study in a blinded mode, and the optimal pressure by conventional criteria and FOT had been defined, all signals were returned to the display and a collective review of the study was performed in order to analyze and describe the behavior and specific characteristics of the FOT signal during CPAP titration.

Statistical Analysis

Data are reported as mean ± SEM. Differences between optimal CPAP defined by both methods were analyzed using a paired t test. Tests were assumed to be significant at p < 0.05.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The FOT was easily applied with CPAP in full-night studies. Our setup does not add substantial complexity to the performance of a conventional PSG and we did not observe tolerance problems or discomfort even in those patients studied all night. This technique allowed continuous monitoring of airflow obstruction during CPAP titration, confirming the findings of previous studies (12, 13). During obstructive apneas the trace obtained by FOT exhibited cycles of persistent increases in |Z|. As treatment pressure was increased, this pattern changed and intermittent increases in respiratory impedance during inspiration accounting for hypopneas were observed. When the optimal pressure was reached |Z| fell to normal values observed during wakefulness and obstructive events were completely alleviated. An example of the behavior of |Z| during this titration procedure is shown in Figure 1.

View larger version (42K): [in this window] [in a new window]   Figure 1.   Compressed polysomnographic tracing showing an example of the titration procedure. During obstructive apneas cycles of persistent increases in |Z| can be seen. As CPAP is increased intermittent increases in respiratory impedance during inspiration are observed. When the optimal pressure is achieved |Z| fell down to normal values and obstructive events were alleviated.

When the optimal treatment pressure obtained by conventional PSG (which we considered to be the current "gold standard" for CPAP titration), and the one defined from the review of the tracing of |Z| were compared, we did not find substantial differences. Individual and group mean values for the 28 patients are shown in Table 1. It can be seen that the optimal CPAP determined from the FOT signal was either identical to or within 1 to 2 cm H₂O of the conventionally determined level in all but one subject for both 2-h nap and overnight studies. There were no significant differences between the group mean values for conventional versus FOT-based optimal CPAP for either the 14 patients studied during naps (p = 0.051) or the 14 patients studied overnight (p = 1.00).

A major aim of our study was to perform a descriptive analysis of the |Z| signal through the course of the night during CPAP titration in order to evaluate the practical issues involved in using this approach. One pertinent observation was the effect of mouth expiration on respiratory impedance (Figure 2). Some patients demonstrated nasal inspiration and mouth expiration as detected by the shape of the pneumotachograph signal and confirmed by visual inspection, which occurred particularly at higher CPAP levels. When this occurred the |Z| tracing showed very prominent increases during expiration. As the forced oscillation is applied via of the nasal route this finding is likely related to a valvelike movement of soft palate structures that blocks the nasal route during mouth expiration. This situation was not infrequent and occurred in eight of the 14 patients studied overnight during periods ranging from 2 to 22% of the total sleep time. This shift in the breathing route from nasal inspiration to mouth expiration could readily be identified from the flow signal. The effect of continuous leaks through the mouth or around the mask must also be considered. Despite taking particular care when fitting the CPAP mask to the patient, significant leaks were not uncommon. We identified periods of continuous leak in seven of the 14 subjects studied overnight, lasting from 4 to 70% of the total sleep time. It is important to detect leaks as they may lead to considerable underestimation of |Z| (Figure 3). Although the correct optimal treatment pressure was able to be determined using the impedance signal in the presence of minor leaks, when large leaks are present FOT does not track airflow obstruction and becomes unreliable. Again, the presence of significant leak was readily detected from the pneumotachograph flow signal.

View larger version (40K): [in this window] [in a new window]   Figure 2.   Effect of mouth expiration on respiratory impedance. This patient demonstrated nasal inspiration and oral expiration as evidenced by the attenuated expiratory flow on the nasal mask pneumotachograph signal. Note that the |Z| tracing (bottom channel ) shows prominent increases during expiration.

View larger version (38K): [in this window] [in a new window]   Figure 3.   Example of the effect of a continuous leak on the values of |Z|. The line below the flow tracing (marked as Flow 2) indicates 0 flow. The flow signal has shifted upward indicating a significant continuous leak. Note that the inspiratory contour of the flow wave exhibits flow limitation. However, in the presence of this leak, the |Z| signal (marked as Gen DC) does not adequately reflect airflow obstruction.

The |Z| tracing could also demonstrate marked fluctuations in relation to arousals or periods of awakening. In some patients we found transient and very irregular increases of |Z| during arousals after obstructive events and also during spontaneous arousals with movement, swallowing, or irregular breathing (Figure 4). This could also be observed in patients with more prolonged awakenings during the night who showed motor activity and irregular breathing before returning to sleep (Figure 5). These increases of |Z| not directly related to obstructive events were found to occur in 12 of the 14 patients studied for more than 6 h of sleep. The number of these episodes varied widely from 10 to 380 during the recording period.

View larger version (49K): [in this window] [in a new window]   Figure 4.   Example showing transient and irregular increases in |Z| during spontaneous arousal.

View larger version (59K): [in this window] [in a new window]   Figure 5.   Irregular increases in respiratory impedance occurring during a prolonged awakening during the night. The patient was moving and breathing irregularly before returning to sleep.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The findings of this study support our previous observations (10) that FOT is well-suited to assessing airflow obstruction during sleep in real-time. In addition, it is a noninvasive technique that does not require patient cooperation. The system to generate forced oscillations was readily implemented in the context of standard PSG and did not result in increased patient discomfort even when applied throughout the night. These observations made over the course of the entire night confirm the findings in a previous report (13) that the intermittent application of FOT for brief intervals during CPAP titration was well tolerated and did not produce sleep disruption.

We studied a substantial number of patients for prolonged periods of sleep and, in all cases, the progressive recovery of airway patency with increasing CPAP was very well detected by FOT. Furthermore, the optimal CPAP treatment pressure could be identified accurately from the FOT signal in that we did not find significant differences in our blinded analysis between the optimal CPAP obtained from |Z| and that defined by conventional polysomnographic criteria. However, our findings indicate that because of the alterations in the |Z| signal that occur in relation to leaks, mouth breathing or movement during arousal/awake periods, the |Z| tracing should be interpreted in conjunction with the flow signal to ensure that such artifacts are identified.

In our experience, the addition of a lightweight pneumotachograph at the nasal mask did not pose any practical difficulty in this setting. However, we have also found that changes in respiratory impedance can be reliably detected from pressure and flow recorded at the CPAP device, thereby simplifying the procedure (18). The key point is that this method depends necessarily on a reliable flow signal in that flow is required both for the determination of |Z| and to detect leaks. The presence of significant leak or mouth expiration can affect the accuracy of the procedure, leading to either underestimation or overestimation of |Z|. Leak and oral expiration were frequent findings in the group of patients studied overnight and are common in routine practice despite improvements in the interfaces available. We believe that these issues are the main limitation to the application of FOT in this setting. However, given that both leak and mouth breathing are readily identified from the flow signal, and given that FOT provides an instantaneous direct measure of respiratory impedance, it would seem eminently feasible to develop algorithms for automated FOT-based CPAP titration based on the |Z| and flow signals. The ease of application of FOT throughout the night in this study suggests that this is an attractive possibility for clinical practice.

We defined optimal CPAP based on our previous work (8) as the pressure level that alleviated respiratory-related arousals associated with frank apneas, hypopneas, snoring, and episodes of inspiratory flow limitation based on the contour of the pneumotachograph flow signal. Although we did not measure esophageal pressure in the present study, we have previously shown that inspiratory flattening of the flow signal is a sensitive marker of flow limitation, and the point at which flattening of the inspiratory flow contour is eliminated is the point at which pleural pressure is minimized and becomes stable (8). As a confirmation that this point represents optimal resolution of upper airway obstruction (12), the respiratory impedance values fell to normal waking levels at the optimal CPAP level, and when the pressure was increased by a further 2 cm H₂O for 10 min, there was no further change in |Z| values. A titration procedure that completely corrects flow limitation therefore seems to be a very solid approach to optimizing titration, and we believe it was an appropriate end point for the purpose of our study. There is evidence suggesting that correcting flow limitation could be associated with a higher CPAP compliance and better improvements in daytime symptoms (19). However, a gold standard has not been defined and evaluation of other end points of CPAP titration procedures such as clinical efficacy in improving quality of life, neuropsychological symptoms, or cardiovascular morbidity requires further investigation.

During conventional CPAP titration, it may be difficult in some cases, particularly at intermediate and higher CPAP levels, to distinguish central events from those caused by residual obstruction. One advantage associated with the use of FOT is that it provides a clear distinction between obstructive and central events. In our patients there were relatively few central apneas observed during CPAP. However, in all cases central apneas occurred without increases in |Z|. The behavior of respiratory impedance was particularly interesting during the onset of REM sleep. Central hypopneas are common in this situation and may be misleading during the titration procedure. The |Z| tracing demonstrated that the majority of these nonapneic events during REM onset occurred without upper airway obstruction, indicating that CPAP would not need to be increased.

In conclusion, FOT is well tolerated when applied continuously during CPAP titration during both daytime nap and whole-night studies. The method provides a direct, real-time measure of upper airway mechanics, and in a blinded analysis accurately identified optimal CPAP. Our results indicate that for adequate interpretation, the tracing and values of respiratory impedance obtained by FOT should be evaluated in conjunction with the flow signal. If this is done, our findings suggest that the |Z| signal may serve as a useful adjunct during manual CPAP titration, and could potentially serve as the basis for automated adjustment of CPAP in patients with SAHS.