Abstract
The aim of the present study was to examine the relationships between the responses to progressive isocapnic hypoxia and hypoxic withdrawal test in patients with obstructive sleep apnoea-hypopnoea syndrome (OSAHS) and to analyse the determinants of carotid body sensitivity in OSAHS.
Nineteen consecutive OSAHS patients and 13 healthy subjects were selected. Ventilatory (ΔV′I/Sa,O2/BSA) and inspiratory neural drive (ΔP0.1/Sa,O2) responses to progressive isocapnic hypoxia were determined. Peripheral chemosensitivity was evaluated by the hypoxic withdrawal test, which measures the decrease in ventilation caused by two breaths of 100% oxygen (%ΔV′I).
Withdrawal response and ventilatory and inspiratory neural drive responses to hypoxia were lower in OSAHS patients than in control subjects. In patients with OSAHS, %ΔV′I correlated significantly with ΔV′I/Sa,O2/BSA and with ΔP0.1/Sa,O2. On stepwise multiple linear regression analysis, a strong correlation between %ΔV′I and ΔP0.1/Sa,O2 was found. Moreover, %ΔV′I, ΔV′I/Sa,O2/BSA and ΔP0.1/Sa,O2 were significantly correlated with minimum arterial oxygen saturation and with arousal index.
Obstructive sleep apnoea-hypopnoea syndrome patients have a strong relationship between peripheral chemosensitivity and respiratory response to hypoxia, suggesting that hypoxic stimulation of central chemoreceptors is minimally relevant in obstructive sleep apnoea-hypopnoea syndrome. Moreover, sensitivity of the carotid body in patients with obstructive sleep apnoea-hypopnoea syndrome is related to sleep disruption and to nocturnal hypoxia.
This study was supported by FIS (96/1280 and 99/0252) and Neumomadrid (2000) Grants.
The study of peripheral chemosensitivity in patients with obstructive sleep apnoea-hypopnea syndrome (OSAHS) has been of considerable interest over several years 1–7. The peripheral chemoreflexes are an important mechanism for regulation of both breathing and autonomic cardiovascular function 8. In fact, abnormalities in chemoreflex mechanisms have been implicated in the increased cardiovascular stress in patients with OSAHS. The relationship between peripheral chemosensitivity and blood pressure profile suggests that recurrent obstructive apnoeas may reset the peripheral chemoreceptor output to a higher level, causing a chronic increase in sympathetic tone and initiating hypertension 9, 10. Intersubject variation of heart rate changes during sleep apnoea could also be due to variations in response to hypoxia 11.
In spite of these pathophysiological implications, the role of peripheral chemoreception in OSAHS has not been adequately evaluated. The nature of the respiratory response to chemical stimuli in awake patients with OSAHS is still unclear. In these patients, depression of peripheral chemosensitivity has been reported 7. In contrast, other investigators have concluded that response to hypoxia in OSAHS patients is normal 2–6 or increased 1.
These discrepancies in findings concerning chemical sensitivity in OSAHS could be due in part to confounding factors, such as obesity, age, sex, smoking habit, metabolic acidosis, hypercapnia, resting metabolic rate, alcohol abuse and genetic factors 12–15. But the stimulation method employed in each study also seems to be responsible for some differences. The best established techniques for assessment of hypoxic response are progressive isocapnic hypoxic stimulation 16 and the hypoxic withdrawal test 17. It has been demonstrated that progressive hypoxia does not adequately evaluate chemosensitivity of the carotid body because it stimulates peripheral and central chemoreceptors together 8. Since the time required for central nervous system stimulation is considered to be about 20 s 18, the hypoxic withdrawal test eliminates peripheral chemoreceptor activity but leaves the humoral environment of the central respiratory system unchanged 17.
The aim of this study was to evaluate the relationship between responses to progressive isocapnic hypoxia and the hypoxic withdrawal test in OSAHS patients. The determinants of carotid body sensitivity in patients with OSAHS were also studied.
Materials and methods
Study subjects
Nineteen consecutive OSAHS patients and 13 healthy subjects were selected to be studied. Patients were excluded from the study for the following reasons: unwillingness or inability to perform the testing procedure; obstructive or restrictive lung disease demonstrated by pulmonary function testing; known valvular heart disease; current drug or mechanical treatment for sleep apnoea; known neuromuscular disease; abnormal thyroid function; morbid obesity (body weight >150% ideal); and recent (<3 months) myocardial infarction or cerebrovascular accident. Control subjects were judged healthy by history, physical examination, electrocardiogram (ECG), spirometry and chest radiography.
Subjects were asked not to eat for 4 h before the study and they were also asked to refrain from using coffee, tea and alcohol for ≥12 h, and tobacco for ≥2 h before each study. The study was approved by the Institutional Ethics Committee at the hospital. All subjects gave their written informed consent prior to enrolment.
Methods
Polysomnography
Healthy subjects and OSAHS patients underwent polysomnography from 23:00–07:00 h. Electroencephalogram (C3-A2, C4-A1), electrooculogram, chin electromyogram, electromyograms of the tibialis anterior of both legs, and ECG were continuously recorded. Breathing was monitored using nasal cannulas, oronasal thermistors and thoracoabdominal stain gauges. Simultaneously, arterial oxygen saturation (Sa,O2) was monitored with a pulse oximeter (Pulsox DP-8, Minolta, Osaka, Japan). Sleep was analysed using the standard criteria 19 for epochs of 20 s and the following sleep variables were calculated: total sleep time, wake time after sleep onset and sleep efficiency, defined as the ratio of total sleep time to sleep episode duration. Micro-arousals were scored according to the American Sleep Disorders Association (ASDA) definition 20. An obstructive apnoea/hypopnoea event was characterised by a >50% decrease from baseline in the amplitude of breathing for ≥10 s associated with either O2 desaturation of >3% or an arousal in the presence of continued respiratory efforts 21. The apnoea/hypopnoea index (AHI) was established as the number of apnoeas/hypopnoeas per hour of sleep. The number of arousals per hour of sleep was expressed as the arousal index (ARI). OSAHS was defined as excessive daytime sleepiness unexplained by other factors plus five or more obstructed breathing events per hour during sleep 21. As indices of nocturnal O2 saturation, the mean Sa,O2 throughout the night, the mean low Sa,O2 (mean of the minimum value for Sa,O2 in each 30-s epoch) and the minimum Sa,O2 (lowest values recorded during sleep) were computed.
Respiratory function
Immediately after awakening, pulmonary function tests were performed as previously described 22, with subjects seated, and always in the same order allowing enough rest between each manoeuvre. All procedures were performed by the same technician, blinded to the results.
Arterial blood gas values were measured with subjects in a seated position, while they breathed room air. Spirometry was performed by means of a pneumotachograph and static lung volumes were measured with a constant-volume body plethysmograph (MasterLab Body, Erich Jaeger GmbH, Würzburg, Germany), according to European Respiratory Society standardisation 23. Resting O2 uptake and carbon dioxide (CO2) output were measured over 5 min, using an automated ergometry set up (Oxycon Alpha, Jaeger). Mean values of the last 4 mins were taken for analysis.
Maximal static inspiratory pressure (PI,max) was measured using a differential pressure transducer (M-163; Sibelmed, Barcelona, Spain). Patients, comfortably seated and wearing a noseclip, performed maximal respiratory efforts either at residual volume or at total lung capacity against an obstructed mouthpiece with a small leak (internal diameter, 0.7 mm) to minimise oral pressure artifacts. The manoeuvres were repeated until three measurements sustained for ≥3 s and with <5% variability were recorded. The highest value obtained was used for analysis.
Mouth occlusion pressure at 0.1 s after the beginning of inspiration (P0.1) was measured by the Whitelaw method 24. Mouth pressure was recorded with a differential pressure transducer (Model DWD, Jaeger). Approximately every 15 s the inspiratory line was occluded without the subject's knowledge for <0.5 s by means of a pneumatic inflatable balloon (Series 9327; Hans-Rudolph, St. Louis, MO, USA). The mean of five or more measurements was determined. The values for dead space and resistance of the system up to a flow of 100 mL were 173 mL and 0.1 kPa·s·L−1, respectively.
Ventilatory and P0.1 responses to progressive isocapnic hypoxia were determined using the rebreathing method of Rebuck and Campbell 17. Sa,O2 was measured continuously with a finger-pulse oximeter (model Oscar II, Datex, Helsinki, Finland). In the seated position with noseclips applied, subjects breathed room air through a mouthpiece via a three-way valve while expired gas was continuously sampled at the mouthpiece using a rapidly responding infrared CO2 analyser (model Oscar II, Datex). The gas analyser was calibrated with gases previously analysed by the Scholander technique. After a stable end-tidal CO2 concentration was achieved, subjects rebreathed through a 7-l bag containing the initial gas mixture: 21% O2 and 7% CO2 in nitrogen (N2). CO2 was held constant (end-tidal CO2 tension (Pet,CO2)±1 mmHg) at the resting end-tidal level (“mixed-venous”) using a variable CO2 absorber bypass, containing soda lime CO2 absorbent and a variable fan. Inspiratory minute ventilation (V′I) was measured by electrically integrating the inspiratory flow signal obtained with a heated (37°C) pneumotachograph (Screenmate Box, Jaeger). Approximately every 15 s without the subject's knowledge, P0.1 was recorded as indicated previously. V′I, P0.1, Sa,O2 and Pet,CO2 were displayed on a 12-bit analogue digital board and a personal computer running LabVIEW software (National Instruments, Austin, Texas, USA). Signals were sampled at 100 Hz. P0.1 was measured from each tracing. V′I and P0.1 were plotted against Sa,O2 on linear coordinates and the slopes were calculated by least-squares linear regression. The procedure was terminated when the Sa,O2 reached 80%.
Peripheral chemosensitivity was detected as a fall of ventilation following sudden elimination of mild hypoxia 7, 25. At the beginning of the test, V′I and Pet,CO2 were measured while the subject was breathing room air from a rubber bag. N2 and CO2 were then added to obtain a end-tidal O2 tension (Pet,O2) of 60 mmHg and Pet,CO2 5 mmHg higher than the control. Two breaths of O2 were then given by turning a three-way stopcock near the inlet of the respiratory valve, to raise Pet,O2 >200 mmHg. Pet,CO2 was also decreased by 2–3 mmHg because inspired CO2 pressure decreased to zero at this time 25. After two breaths of 100% O2, the inspiratory gas was switched back to the hypercapnic hypoxic gas. The V′I during room-air breathing was defined as V′I,N. The V′I before breathing 100% O2 during the mildly hypercapnic hypoxic state was defined as V′I,0. The V′I between 5 and 20 s after changing the inspiratory gas was defined as V′I,5–20. The difference between V′I,0 and V′I,5–20 was defined as the withdrawal response (ΔV′I) and %ΔV′I (ΔV′I/V′I,0×100) was used as an index of the peripheral chemoreceptor activity. The withdrawal test was performed three or more times at intervals of 20 min. The subject breathed room air between tests to avoid the effects of hypoxic ventilatory depression 7. To eliminate the effects of body size and sex, the indices of each ventilatory response were corrected by body surface area (BSA).
Analysis
The comparisons between the patients with OSAHS and the control subjects were performed by the Mann-Whitney U-test. Coefficient of variability was computed as 100× standard deviation (sd) of the repeated determinations divided by the mean value (100×sd·mean−1). Correlations between respiratory responses to progressive isocapnic hypoxic stimulation and the hypoxic withdrawal test were analysed by linear regression analysis, using Spearman's rank correlation coefficient (r). In order to determine which independent variables were correlated with hypoxic withdrawal response, stepwise multiple linear regression analysis was performed 26. Independent variables entered into the regression included AHI, ARI, mean lowest Sa,O2, minimum Sa,O2, O2 uptake and CO2 production. Stepwise criteria were a probability of F-Snedecor test to enter <0.05 and a probability of F-Snedecor test to remove >0.10. In all cases, p-values of <0.05 were considered to be significant. Data are expressed as mean±SD.
Results
The anthropometric characteristics of the patients with OSAHS and the control subjects are shown in table 1⇓. There was no significant difference in sex, age, body mass index or smoking habit between both groups.
Anthropometric characteristics, lung function and sleep architectures in control subjects and patients with obstructive sleep apnoea-hypopnoea syndrome (OSAHS)
The mean values of lung volumes, PI,max, arterial blood gases, O2 uptake and CO2 production were similar in both groups. Of the basal lung function tests, only P0.1 was significantly higher (p<0.001) in the OSAHS patients than in the control subjects.
The respiratory responses to hypoxia are shown in table 2⇓. Compared with control subjects, patients with OSAHS had lower ventilatory and inspiratory neural drive responses to progressive isocapnic hypoxic stimulation (p<0.001). There was no significant difference in V′I during room air breathing. The mean values of V′I during mildly hypercapnic hypoxic state (ΔV′I,0/BSA) and withdrawal response (ΔV′I/BSA and %ΔV′I) for patients with OSAHS were lower than those for the control subjects.
Respiratory responses to chemical stimuli in control subjects and patients with obstructive sleep apnoea-hypopnoea syndrome (OSAHS)
In control subjects, withdrawal response was correlated with ventilatory and P0.1 responses to progressive isocapnic hypoxic stimulation (fig. 1⇓). As shown in figure 2⇓, patients with OSAHS had significant correlations between ΔV′I/Sa,O2/BSA and %ΔV′I, and between ΔP0.1/Sa,O2 and %ΔV′I. On stepwise multiple linear regression analysis, %ΔV′I significantly correlated with ΔP0.1/Sa,O2 (multiple r2=0.970, p=0.000) in patients with OSAHS.
a) Ventilatory and b) inspiratory neural drive responses to progressive isocapnic hypoxic stimulation plotted against hypoxic withdrawal in control subjects. ΔV′I: withdrawal response; %ΔV′I: ΔV′I/V′I,0×100, where V′I,0 is V′I before breathing 100% O2 during the mildly hypercapnic hypoxic state. Sa,O2: arterial oxygen saturation; BSA: body surface area; ΔV′I/Sa,O2/BSA: ventilatory response to progressive isocapnic hypoxic stimulation; P0.1: mouth occlusion pressure; ΔP0.1/Sa,O2: central inspiratory drive response to progressive isocapnic hypoxic stimulation. (a) r=0.685, p=0.010; b) r=0.961, p=0.000).
a) Ventilatory and b) inspiratory neural drive responses to progressive isocapnic hypoxic stimulation plotted against hypoxic withdrawal in patients with obstructive sleep apnoea-hypopnoea syndrome (OSAHS). ΔV′I: withdrawal response; %ΔV′I: ΔV′I/V′I,0×100, where V′I,0 is V′I before breathing 100% O2 during the mildly hypercapnic hypoxic state. Sa,O2: arterial oxygen saturation; BSA: body surface area; ΔV′I/Sa,O2/BSA: ventilatory response to progressive isocapnic hypoxic stimulation; P0.1: mouth occlusion pressure; ΔP0.1/Sa,O2: central inspiratory drive response to progressive isocapnic hypoxic stimulation. a) r=0.976, p=0.000; b) r=0.926; p=0.000.
Within-day coefficients of variation for chemosensitivity indices as assessed in all OSAHS patients for three study sessions were 6.1% (range, 2.0–8.5%) for ΔV′I/Sa,O2 /BSA, 5.8% (range, 1.7–8.0%) for ΔP0.1/Sa,O2, and 8.3% (range, 3.2–12.4%) for %ΔV′I. Day-to-day coefficients of variation as assessed in eight patients with OSAHS on three different days, were 9.1% (range, 3.3–13.5%) for ΔV′I/Sa,O2/BSA, 8.9% (range, 4.1–12.6%) for ΔP0.1/Sa,O2, and 12.6% (range, 5.2–18.1%) for %ΔV′I.
In OSAHS patients, ΔV′I/Sa,O2/BSA correlated with mean low Sa,O2 (r=0.732, p=0.002), minimum Sa,O2 (r=0.697, p=0.004), AHI (r=0.649, p=0.003) and ARI (r=0.864, p=0.000) (fig. 3⇓). ΔP0.1/Sa,O2 also correlated with mean low Sa,O2 (r=0.824, p=0.000), minimum Sa,O2 (r=0.771, p=0.001), AHI (r=0.650, p=0.003) and ARI (r=0.902, p=0.000) (fig. 4⇓). Indeed, %ΔV′I correlated with mean low Sa,O2 (r=0.816, p=0.000), minimum Sa,O2 (r=0.768, p=0.001), AHI (r=0.676, p=0.001) and ARI (r=0.913, p=0.000) (fig. 5⇓). O2 uptake and CO2 production were related with ΔP0.1/Sa,O2 (r=0.582, p=0.009 and r=0.576, p=0.010, respectively) and with %ΔV′I (r=0.553, p=0.014 and r=0.560, p=0.013, respectively), but they did not correlate with ΔV′I/Sa,O2/BSA (r=0.375, p=0.114 and r=0.386, p=0.103, respectively). No correlations were found between the withdrawal response and mean nocturnal Sa,O2 (r=−0.238, p=0.455), pH (r=−0.234, p=0.336), CO2 tension in arterial blood (Pa,CO2) (r=0.086, p=0.726), functional residual capacity (r=0.290, p=0.242), PI,max (r=0.175, p=0.473) or central inspiratory drive (r=−0.192, p=0.430). On stepwise multiple linear regression for %ΔV′I as dependent variable, the variables entered into the more significant model (r=0.882, adjusted r2=0.740) were ARI (standardised coefficient beta=0.551, p=0.009) and minimum Sa,O2 (standardized coefficient beta=0.422, p=0.034).
Relationship between a) mean nocturnal arterial oxygen saturation (Sa,O2), b) minimum Sa,O2, c) apnoea/hypopnoea index and d) arousal index with ventilatory response to progressive isocapnic hypoxia (ΔV′I/Sa,O2/BSA). BSA: body surface area; ΔV′I: withdrawal response. a) r=0.732, p=0.002; b) r=0.697, p=0.004 c) r=0.649, p=0.003; d) r=0.864; p=0.000.
Relationship between a) mean nocturnal arterial oxygen saturation (Sa,O2), b) minimum Sa,O2, c) apnoea/hypopnoea index and d) arousal index with inspiratory neural drive response to progressive isocapnic hypoxia (ΔP0.1/Sa,O2). P0.1: mouth occlusion pressure. (a) r=0.824, p=0.000; b) r=0.771, p=0.001; c) r=0.650, p=0.003; d) r=0.902, p=0.000).
Relationship between a) mean nocturnal arterial oxygen saturation (Sa,O2), b) minimum Sa,O2, c) apnoea/hypopnoea index and d) arousal index with hypoxic withdrawal response (ΔV′I). (a) r=0.816, p=0.000; b) r=0.768; p=0.001; c) r=0.676; p=0.001; d) r=0.913, p=0.000).
Discussion
The main results of the present study are the following: peripheral chemosensitivity is reduced in OSAHS patients in comparison with control subjects; the sensitivity of carotid body is strongly related with inspiratory neural drive response to progressive isocapnic hypoxic stimulation; and peripheral chemosensitivity of OSAHS patients is related to sleep disruption and to nocturnal hypoxaemia.
There are several factors to be considered when interpreting the decreased peripheral chemosensitivity of the patient group. Differences in age, body size and lung volumes could modify responses to chemical stimuli 27. However, none of these varied between the OSAHS and control subjects. It is also known that chemosensitivity is highly dependent on the acid-base status of the patient. The respiratory response to hypoxia is sensitive to variations in arterial pH induced by partial pressure of CO2 changes. Specifically, acidosis and hypercapnia increase the response to hypoxia 27. Only one of 19 OSAHS patients had a Pa,CO2 >45 mmHg (46.8 mmHg). ΔV′I/Sa,O2/BSA, ΔP0.1/Sa,O2 and ΔV′I of this subject (0.27 L/min/%/m2, 0.0250 kPa/% and 24.2%, respectively) were higher than mean values of the OSAHS group (table 2⇑). A hypercapnia-related increase in peripheral chemosensitivity could not be excluded in this patient. Thus, it is possible that the results slightly underestimate the true depression of peripheral chemosensitivity of OSAHS patients. The contribution of metabolic rate to the decrease in peripheral chemosensitivity should also be evaluated. The present results, which show similar levels of O2 consumption and CO2 production between OSAHS patients and control subjects, suggest that changes in the sensitivity of the carotid body of these patients are not due to lower basal metabolism.
A strong relationship between peripheral chemosensitivity and respiratory responses to progressive isocapnic hypoxic stimulation in OSAHS patients was found. These findings suggest that hypoxic stimulation of central chemoreceptors is minimally relevant in OSAHS patients. Thus, inspiratory neural drive response to hypoxia could be used to estimate peripheral chemosensitivity in these patients. Furthermore, a limitation of the withdrawal test should be considered. In the withdrawal test, the level of the stimulus varies with the tidal volume and frequency of the hyperoxic breaths, the initial alveolar O2 pressure, and the distribution of ventilation in the lungs 8. Clearly, it would be difficult to apply this test in OSAHS patients in whom maldistribution of ventilation would prevent the relatively abrupt institution of a hyperoxic state, such as would occur in normal subjects.
This limitation of the withdrawal test in OSAHS patients could explain the slightly higher coefficients of variation for %ΔV′I than for P0.1 or ventilatory response to hypoxia found in this study. However, variability of both tests is moderate. Within-subject variability of hypoxic withdrawal response in the OSAHS patients (12.6%) was equal to spontaneous variation estimated by Osanai et al. 7 in healthy subjects. The coefficient of variability for P0.1 response to hypoxia found in the OSAHS patients (8.9%) was also similar to the coefficients of variability described by White et al. 28 and the present authors' 29 in healthy subjects.
The mechanism of the reduced peripheral chemosensitivity in OSAHS patients is not well established. It seems probable that the diminished sensitivity of the carotid body represents a specific adaptation to the repeated hypoxia induced during apnoeas 30. It is known that a reduction in ventilatory response to hypoxia occurs during both short- and long-term hypoxic exposure 31. Although only very limited data exist on ventilatory changes during repeated hypoxia 30, the reduced peripheral chemosensitivity of OSAHS patients could represent an adaptive response to the hypoxic environment. Previous studies have shown that adaptation to sustained hypoxia may result from changes in either carotid chemoreceptors or central hypoxic sensitivity. The latter may be due either to altered central processing of afferent carotid body stimuli or to a change in direct central nervous system sensitivity to hypoxia 31. The results of the present study do not exclude alterations in central processing of afferent stimuli, but show that central hypoxic stimulation is scarcely relevant.
In accordance with the role of hypoxia as an inducer of peripheral chemosensitivity alterations, a significant relationship between sensitivity of the carotid body and minimum nocturnal O2 saturation was found. Previously, it has been demonstrated that ventilatory responses to hypoxia are negatively correlated with the degree of hypoxaemia during sleep in OSAHS patients 32. Osanai et al. 7 reported that the hypoxic withdrawal response showed negative correlations with 4%- and 10%-desaturation ratios. Contrary to these authors 7, the present authors found that peripheral chemosensitivity is also positively correlated with AHI. The importance of sleep structure as a contributing factor to peripheral chemosensitivity changes in OSAHS patients is not well known. Despite a reduction of ventilatory response to hypoxia after sleep deprivation having been demonstrated in healthy males 28, no relationship between diurnal sleepiness and chemosensitivity has been established. Previously, the present authors have proposed that repetitive abrupt arousals from sleep could be important contributors to the increase of the peripheral chemosensitivity in OSAHS patients 9. The relationship between ARI and hypoxic withdrawal response found in the OSAHS patients in the present study might suggest a contribution of sleep disruption to changes in peripheral chemosensitivity. In consequence, it could be hypothesised that the sensitivity of the carotid body in OSAHS patients might be determined by a balance between hypoxic depression and arousal stimulation.
The sequence of events leading to a decrease in peripheral chemosensitivity in OSAHS patients is unresolved, but it has been proposed that these subjects have an abnormality of dopaminergic mechanisms in the peripheral chemoreceptors 7. Since dopamine appears to be an inhibitory transmitter in mammalian carotid bodies, the effects of dopamine on the carotid body of patients with OSAHS might be increased 7. On the other hand, familial aggregation of blunt ventilatory responses to chemical stimuli has been reported for healthy family members of OSAHS patients 33. Thus a contribution of genetic factors to the changes in peripheral chemosensitivity observed in patients with OSAHS cannot be excluded.
To conclude, this study shows a strong relationship between peripheral chemosensitivity and the respiratory response to hypoxia in obstructive sleep apnoea-hypopnoea syndrome patients, suggesting that hypoxic stimulation of central chemoreceptors is minimally relevant in these patients. Moreover the sensitivity of the carotid body in patients with obstructive sleep apnoea-hypopnoea syndrome is related to the apnoea-hypopnoea index and with nocturnal oxygenation.
Acknowledgments
The investigators acknowledge the excellent technical assistance provided by A. Alvarez, P. Librán, A. Pérez and C. Suárez.
- Received June 1, 2001.
- Accepted April 11, 2002.
- © ERS Journals Ltd
References
