INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Keywords: sleep apnea; hypopneas; RERAs; microarousals; scoring of respiratory events

Over the past thirty years there has been an increased recognition of the high prevalence of sleep-disordered breathing, which affects 9% and 4% of middle-age men and women respectively (1). Guilleminault and coworkers using polysomnography have precisely described obstructive apneas and their acute consequences and introduced the obstructive sleep apnea syndrome (OSAS) with its clinical picture of snoring and daytime sleepiness (2). Obstructive hypopneas (OHs) were then secondarily defined as episodic partial upper airway obstructions leading to oxygen desaturations with nearly the same acute consequences and clinical symptoms as for apneas. This led researchers in 1988 to amend OSAS to the obstructive sleep apnea-hypopnea syndrome (OSAHS) (3). Furthermore, even more subtle abnormalities have been described such as progressive increases in respiratory effort, characterized by increased negative esophageal pressure (Pes) during inspiration, reflecting increased upper airway (UA) resistance that terminates owing to an arousal but not oxygen desaturation (i.e., a respiratory effort related arousal [RERA]). In 1993 Guilleminault and associates described a series of patients with typical symptoms of OSAHS who did not demonstrate obstructive apneas or hypopneas on polysomnography but exhibited RERAs and were said to present the upper airway resistance syndrome (UARS) (4).

These obstructive nonapneic respiratory events (ONAREs) (i.e., OHs and RERAs) may be just as clinically important as full-blown apneas in terms of producing sleep fragmentation. The common pathway for the partial UA obstruction occurring in the different types of ONAREs is an increased UA collapsibility during sleep. A given level of respiratory effort, as assessed by Pes measurements, could lead to an hypopnea with a significant flow reduction or to a RERA, depending on the stiffness of the pharyngeal walls, i.e., the level of UA collapsibility. It is not clearly demonstrated whether different levels of collapsibility could exist and whether the level of respiratory effort just preceding the arousal could vary during the night in a given patient.

The polysomnographic features of ONAREs can be summarized by three different patterns: (1) a variable reduction in flow (from inspiratory flow limitation to hypopnea), (2) an increase in respiratory effort, (3) the occurrence of a microarousal ending the respiratory event. These ONAREs (OHs + RERAs) are, however, much more difficult to detect and classify than obstructive apneas unless sensitive and quantitative measures of respiratory effort and flow are employed. Thermistors have been widely used in assessing airflow but are more qualitative than quantitative, leading to inadequate scoring of hypopneas (5). Technology has improved to incorporate methods with better validity and reliability, such as pneumotachography and more recently, nasal pressure measurements (5, 6). This has led the American Academy of Sleep Medicine (AASM) Task Force (7) to propose new definitions for respiratory events and syndromes based on scientifically accepted methodologies.

In the present study we have used the definitions and the reference methods recommended by the AASM Task Force (7) to characterize a representative population of patients with moderate OSAHS or UARS. The aims of the study were: (1) to precisely characterize the proportion of the different types of obstructive nonapneic events (OH and RERAs); (2) to validate the new AASM Task Force definitions as an adequate tool for classifying respiratory events and syndromes; (3) to appreciate whether a given level of respiratory effort (i.e., a value of Pes) is associated with the same type of events during the night for a given patient.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Subjects

Fifteen unselected patients suspected of suffering from UARS or mild to moderate OSAHS were included. They presented clinical complaints of severe snoring and excessive daytime somnolence with limited desaturations on screening with nocturnal oximetry. Therefore, these patients were suspected to demonstrate mainly ONAREs. We used sensitive and quantitative measures of respiratory effort and flow as ONAREs are much more difficult to detect and classify than apneas.

Polysomnography

Sleep analysis. Continuous recordings were taken using an electroencephalogram (EEG) with electrode positions C3/A2-Cz/O1 of the International 10-20 Electrode Placement System, electrooculogram, chin electromyogram, and electrocardiogram (ECG). The polysomnograph was scored manually according to standard criteria for sleep stages (8) and cortical microarousals (9).

Eleven of the 15 patients underwent a noninvasive measurement of sleep fragmentation using pulse transit time (PTT) (10). PTT was calculated as the interval between the ECG R-wave and the point corresponding to 50% of the height of the maximum value on the pulse wave-form detected by photoplethysmography at the finger level. PTT is inversely correlated to blood pressure. The blood pressure surges associated with cortical microarousals can then be detected by PTT (autonomic arousal), thus offering the possibility of estimating sleep fragmentation and erasing the necessity to record EEG. In this study, the beat-to-beat PTT measurements were made using an RM50 recorder (DeVilbiss, Parcay-Meslay, France). The two groups of patients with and without PTT measurement were not different in terms of demographic and polysomnographic data (nonsignificant [NS], Mann-Whitney test).

Respiratory Parameters

Methods of measurement. Airflow was measured using a pneumotachometer (Kontron Instruments, Quentin, France) installed on a full-face mask. Before each study the pneumotachometer was reset to zero and calibrated with a fixed flow generator. The face masks used were those of Respironics (Monroeville, PA). Care was taken with the fit of the mask to the patients' face and a self-made connecting system was developed allowing the esophageal catheter to travel through the mask without any leaks. In the occurrence of significant leaks, which were easily identified by a dip in the baseline of the flow signal, the entire system was repeatedly checked until all leaks were suppressed. When patients did not tolerate the full equipment during the entire night (six patients), oronasal thermistors and a nasal cannula were used, allowing the calculation of an apnea + hypopnea index (AHI) for the whole night. In the present study, only the recording time, including the measurement of airflow with a pneumotachometer and the measurement of respiratory effort with an esophageal catheter, was taken into account for the analysis of ONAREs (mean/median value for the 15 patients = 310/392 min, minimum = 97 min, maximum = 478 min).

Respiratory effort was systematically measured by means of an esophageal catheter (Compliance catheter volgens EGKS-norm, International Medical, Zutphen, Netherlands). Oxygen saturation was measured with a Biox-Ohmeda 3700 oximeter (Ohmeda, Louisville, CO).

Scoring of respiratory events. When patients demonstrated more than 200 respiratory events during the entire night (four patients), only one-fifth of the events, chosen at random, were analyzed.

Definitions of the different types of events. The new AASM Task Force criteria (7) were used to classify the respiratory events as apneas, hypopneas, and RERAs:

Hypopnea corresponded to a transient reduction, but not complete cessation, of breathing of more than 10 s. In the present study, the reduction in flow had to be either more than 50% or less than 50% but associated with either a microarousal or oxygen desaturations of more than 3%. A less than 30% flow reduction was classified as a RERA not a hypopnea. There are some inconsistencies in the task force report (7) regarding the minimum flow reduction amplitude needed to score hypopneas. A "clear amplitude reduction" is initially mentioned although a "discernable but less than 50% decrease" is required in another paragraph. Thus, we chose to use a threshold of 30% flow reduction as it is the most widely used in clinical research. Moreover, we preferred to use flow reduction rather than decrease in tidal volume (i.e., integrated flow) to score hypopneas as this is the parameter generally used in sleep laboratories working with pneumotachometers.

A RERA is an event characterized by increasing respiratory effort of more than 10 s leading to an arousal from sleep but which does not fulfill the criteria for a hypopnea (7).

Indefinite events (IEs). Some obstructive nonapneic events did not fulfill the AASM Task Force definitions for hypopneas or RERAs. This occurred either in the case of a 30 to 50% airflow decrease without arousal or desaturation, or in the case of an airflow decrease less than 30% without arousal.

Scoring of airflow reduction. The percentage of airflow reduction was determined manually for each ONARE. In cases with unstable breathing, as in the example given in Figure E1 in the online data supplement, the reference flow was calculated as the mean inspiratory amplitude of the three largest breaths preceding the respiratory event (a). When stable breathing occurred in the 2 min preceding the onset of the event, the baseline flow corresponded to the mean inspiratory amplitude during this period of stable breathing. The minimal inspiratory flow value, during the ONARE, was identified as (b). The percentage of airflow reduction (%AR) was calculated by the ratio (a  b)/a.

Scoring of respiratory effort (Figure E1). The obstructive pattern was defined by the presence of an increased or sustained respiratory effort during the event. The level of the respiratory effort was calculated by the largest difference between the maximal (expiratory) and minimal (inspiratory) values of Pes (Pes).

Ratio: % AR/Pes. We used this ratio as an indirect assessment of UA collapsibility where Pes measures the level of respiratory effort.

Statistical Analysis

All results are shown as mean ± SD. Different types of events were compared using the Mann-Whitney or the Kruskall-Wallis test. The chi-square test was used for comparison of qualitative variables. Statistical significance was accepted at p < 0.05.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The group of 15 patients (3 males and 2 females) was referred to the sleep laboratory for mild hypersomnia (Epworth Sleepiness Scale: 8.2 ± 5.5). They were middle-aged (48.6 ± 11.8 yr), slightly overweight (body mass index [BMI] = 26.9 ± 4.8 kg/m²), and demonstrated moderate OSAHS (AHI = 30.1 ± 25.8/h of sleep, apnea index [AI] = 3.2 ± 5.2/h of sleep) with a limited amount of desaturation (mean nocturnal Sa_O2 = 94.6 ± 2.2%) (see Table E1 in the online data supplement).

Relative proportion of the different subtypes of respiratory events. Figure E2 in the online data supplement represents the proportion of the three types of ONAREsOHs, RERAs, and IEswhere all 1,061 events, studied in the 15 patients, were taken into account. OHs were prominent (79.9%). RERAs represented only 5.3% of the events; 157 of the ONAREs (14.8%), so-called IEs, did not fulfill the AASM Task Force definitions of either hypopneas or RERAs. In 126 of these 157 events, a 30 to 50% reduction in flow was found without desaturation or cortical arousal.

Both individually and as a whole group (1,061 events), the relative proportion of each type of respiratory event was the same (mean of patients' individual data: OHs = 74.7%, RERAs = 5.8%, IEs = 19.5%; NS, chi-square test; see Figure E3 in the online data supplement).

Respiratory events associated with a 30 to 50% flow reduction (n = 392). In this situation, the identification of a cortical microarousal (CA+) permitted the classification of 246 events as hypopneas (62.8%) (Figure E2). An isolated desaturation equal to 3% without cortical arousal was found in only two of the 392 events where a 30 to 50% flow reduction was observed (0.5%).

Autonomic versus cortical arousals. In a subgroup of 11 patients for whom PTT data were available, the autonomic arousals (AAs) were as accurate as cortical arousals (CAs) in the classification of hypopneas as an event of 30 to 50% flow reduction without desaturation (160 AAs versus 174 CAs, NS). RERAs were defined as an increase in respiratory effort and a flow reduction less than 30% terminated by a microarousal. In this subgroup of patients, 95 events with a flow reduction less than 30%, were scored. Sixty-two events were classified as RERAs owing to the recognition of an AA whereas only 36 CAs were found (p = 0.0004).

% AR/Pes: indirect assessment of upper airway collapsibility. During OHs and RERAs different ranges of airflow reduction (57.9 ± 10.7% versus 21.3 ± 4.9%) were observed although respiratory effort remained similar (Pes = 21.9 ± 5.5 cm H₂O versus 18.9 ± 5.7 cm H₂O, NS). This suggests that these types of respiratory events were mainly determined by the level of UA collapsibility and not by an absolute value of respiratory effort. When looking at individual patient data, the same level of Pes was also found at the end of RERAs and OHs whereas the reduction in flow was, as expected, higher for OHs. (see Figure E4 in the online data supplement) This was true in all but two patients and suggested that different levels of collapsibility could exist during the night in the same subject.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

To our knowledge, this is the first study accurately characterizing the proportion of different types of obstructive nonapneic events (OH and RERAs) using the reference tools for ventilation and respiratory effort measurements. In a population of moderate OSAHS, OHs represented the prominent type of ONAREs when RERAs were limited to 5.3%, and 14.8% of the ONAREs were unclassified (IEs) using the new AASM Task Force definitions. The identification of a microarousal either cortical or autonomic was the key point for classifying, as hypopneas, respiratory events with a limited flow reduction (30 to 50%). For recognition of RERAs, AAs appeared more sensitive than CAs. Finally, we demonstrated that similar levels of Pes could be found at the end of RERAs and OHs although the reduction in flow was higher for OHs.

Guilleminault and colleagues first described UARS in 1991 (11). Women with UARS usually reported daytime fatigue without snoring. Conversely men with UARS mainly complained of snoring. In both sexes, craniofacial abnormalities were present in many of these typically nonobese young patients. These particular clinical features in combination with the demonstration of prominent RERAs on the polygraphic recordings have led the Stanford group to identify UARS as a specific entity (12). In this context specific physiopathology and treatment strategies have been proposed (13, 14). However, the concept of UARS remains controversial. The recent AASM Task Force concluded that there was not enough evidence to support UARS as a distinct syndrome with a specific physiopathology (7). The technical limitations, when monitoring airflow with thermistors and inductance plethysmography, as well as the lack of a precise definition of hypopnea, may have led to the overestimation of UARS as an independent diagnosis (15). Montserrat and Badia (15) firmly promoted the idea that most of the RERAs identified by thermistors with Pes would correspond to hypopneas, if nasal pressure or a pneumotachograph had been used for assessing flow. Thus, these investigators considered RERAs mainly as a measurement artifact and they proposed an alternative name for UARS: the "thermistor syndrome." Our study demonstrates, using accurate measurements, that all types of events exist in the majority of patients with moderate OSAHS and that RERAs are limited to 5% of these events. Thus, these results support the AASM Task Force views considering RERA as a specific but relatively rare respiratory event and thus question whether UARS is a distinct syndrome.

One of the goals of the AASM Task Force was to define all the specific events of sleep-related breathing disorders (7). Thus, the group of experts tried to explain precisely how the whole group of nonapneic obstructive events should be separated into hypopneas or RERAs (7). In our study, we used these new definitions as scoring rules (see METHODS); 14.8% of the events did not fulfill the criteria of hypopneas or RERAs (indefinite events: n = 157/1,061). These IEs may be related either to respiratory events that do not meet the criteria for hypopneas or RERAs, but are still events of physiologic significance, or to true artifacts. The proportion of true artifacts among these IEs is a crucial point. The events were scored according to flow reduction amplitude and the presence of corroborative events. For IEs, desaturations and microarousals were typically not found, hence the events were scored mainly on the basis of flow reduction. However, runs of flow limitation or phase paradox on thorax/abdomen channels were usually identified during these events. Moreover, the scorers were trained to score in the direction of normality when in doubt. Finally we took care, as far as possible, to avoid leaks around the face mask.

Thus, we firmly believe that these IEs did not correspond to true artifacts, but rather to subtle respiratory events not classified by the new AASM definitions. In 126 of these 157 IEs a 30 to 50% reduction of airflow was found without CAs. The insufficient sensitivity of EEG visual analysis to detect microarousals is now well accepted (16). The EEG criteria used for microarousal scoring may underestimate the physiologic changes that allow the termination of a respiratory event. Sforza and coworkers (17) have proposed an integrative hierarchy of arousal responses that begins with autonomic activation and progresses to EEG synchronization, then microarousal and finally to full awakening. Thus, acute rises in pulse rate or blood pressure, shortened R-R intervals as well as PTT dips reflecting brainstem stimulation could all be used as more sensitive non-EEG tools for arousal recognition. A big breath (18) or an abrupt resolution of flow limitation, which actually corresponds to a sudden decrease in UA resistance, could also be considered as an arousal equivalent.

We firmly believe that any reduction in flow or a flow-limited pattern terminated by an arousal, whatever the tool used to recognize brainstem or cortical activation, should be considered as a significant respiratory event. The recent report from the AASM Task Force did not address the major issue of microarousals identification (7). Obviously this is necessary, as the recognition of an arousal, terminating the flow signal modifications, represents the mechanism allowing the separation of a true respiratory event from an artifact. Further studies are needed to validate potentially more sensitive techniques for the arousal-based scoring of ONAREs. Such sensitive tools might reduce the 15% proportion of indefinite events that we found using EEG. Moreover, it would be interesting to compare these noninvasive alternative techniques (nasal cannula and PTT) with the reference methods (pneumotachograph [PNO] and Pes) when scoring hypopneas and RERAs following the AASM rules. To demonstrate a comparable ratio of hypopneas and RERAs when using PNO and Pes or nasal cannula and PTT, in a randomized fashion, during two consecutive nights, would represent a further validation of nasal cannula and PTT as adequate tools to quantitatively measure ventilation during sleep.

The use of arousal-based and desaturation-based scoring criteria has gained considerable popularity in the operational definition of hypopnea (19). Two recent studies using thermistors and inductance plethysmography, which were unable to detect at least some of the subtle respiratory events, suggested that the addition of arousal-based hypopnea criteria had little impact on the hypopnea index (19, 20). Conversely, in these studies, desaturations, which were easily recognized and were associated with pronounced respiratory changes, seemed critical to affirm hypopneas. Tsai and coworkers (19) used relatively insensitive methods to detect flow reduction (i.e., thermistors and inductance plethysmography). On the other hand, clear episodes of desaturation occurred (mean nocturnal Sa_O2 = 89.3%) favored by recruitment of overweight patients (mean BMI = 31.5 kg/m²). In this situation one can understand that only obvious respiratory events, generally associated with desaturations, were detected and therefore the arousal contribution in scoring hypopneas could appear negligible. Thus, the respective contribution of arousal-based or desaturation-based scoring criteria in the operational definition of hypopnea depends, critically, on the population sampled and the technical skills used for respiratory measurements. Our study, using quantitative flow measurements, was addressed to slightly overweight patients with moderate OSAHS (median BMI = 25.8 kg/m²), demonstrating a limited amount of desaturation (mean nocturnal Sa_O2 = 94.6%). In such a sample of patients, the occurrence of microarousals, either cortical or autonomic, became the crucial point for scoring 60 to 70% of the hypopneas (Figure E2). The PTT pattern of AAs is undoubtedly much more simple to visually detect than EEG arousals. In the present study, we demonstrated that AAs were as effective as CAs regarding the arousal-based hypopnea scoring and even more sensitive when classifying RERAs. It remains to be demonstrated, however, that all the arousals recognized on the PTT signal alone correspond to true events. In other words, false-positive microarousals could occur in PTT, related either to artifacts or to physiologic variations in autonomic tone, especially during rapid eye movement (REM) sleep. Thus, the specificity of PTT arousals should be determined in further studies.

It has been demonstrated (21, 22) that snorers could be differentiated from patients with UARS, hypopneics, and patients with OSA by using sleep measurements of pharyngeal critical pressure (Pcrit) as a marker of collapsibility. When looking at these data, there appears to be a continuous spectrum of collapsibility, from snorers to apneics, and some overlap seems to occur in Pcrit values between hypopneic and UARS groups. Some of these patients may behave as hypopneics or as persons with UARS depending on specific conditions such as sleep posture or stages. The data of Boudewyns and coworkers (23) support this hypothesis in demonstrating a more negative value of Pcrit in lateral compared with supine body positions. Furthermore, we found more negative Pcrit values in Stage 3 and REM compared with Stage 1 or 2 (unpublished data). The % AR/Pes ratio, used in our study, basically represents an indirect assessment of collapsibility. We found that patients exhibited both OHs and RERAs and that similar levels of Pes were obtained at the end of RERAs and OHs although the reduction in flow was higher for OHs. These data confirmed nightly spontaneous variations in collapsibility. Further studies should address this issue as those patients with highly variable collapsibility are probably those who could primarily benefit from therapy such as self-controlled continuous positive airway pressure (auto-CPAP).

Conclusions

In a population of patients, demonstrating sleep-disordered breathing with limited desaturation and suspected of having OSAHS or UARS, OH represented the prominent type of ONAREs when RERAs were limited to 5%. This suggested that patients exhibiting mainly RERAs are rare or do not exist. Terminating the respiratory event, the occurrence of a CA or an "arousal equivalent" (i.e., AA, big breath, or transient disappearance of flow limitation) are the key factors allowing the separation of true subtle respiratory events from physiologic changes or artifacts. In the present study, AAs were found to be equivalent to CAs in scoring hypopneas and slightly more sensitive when detecting RERAs.

In a given patient upper airway collapsibility may vary throughout the night, determining the relative proportion of the different types of ONAREs. Whether UARS is a specific condition, in terms of physiopathology and epidemiology, remains to be studied.