Clinical Science (1991) 80, 443-449
443
Effects of treatment with nasal continuous positive airway pressure on atrial natriuretic peptide and arginine vasopressin release during sleep in patients with obstructive sleep apnoea J. KRIEGER*, M. FOLLENIUSt, E. SFORZA*, G. BRANDENBERGERt
AND
J. D. PETERt
'Service d'Explorations Fonctionnelles du Systeme Nerveux and :Departement d'Anesthesiologie, Centre Hospitalier Regional et Universitaire, Strasbourg, France, and tLaboratoire de Physiologie et Psychologie Environnementales, CNRS, Strasbourg, France
(Received 21 May 1990/3 December 1990; accepted 12 December 1990) SUMMARY
1. Patients with obstructive sleep apnoea have increased diuresis during sleep, which decreases with nasal continuous positive airway pressure treatment. These changes have been attributed to an increased release of atrial natriuretic peptide in obstructive sleep apnoea, and its decrease with continuous positive airway pressure treatment. 2. In order to clarify the change in plasma atrial natriuretic peptide level and to investigate the underlying mechanisms, blood samples were taken at 10 min intervals from nine patients with obstructive sleep apnoea during two nights when the patients were either untreated or treated with continuous positive airway pressure. Polysornnographic monitoring, including transcutaneous oximetry, and measurement of oesophageal pressure were performed simultaneously. Plasma arginine vasopressin was also measured. 3. The plasma level of arginine vasopressin did not change. The level of atrial natriuretic peptide was high and exhibited secretion bursts in six out of the nine patients; it drastically decreased with continuous positive airway pressure treatment. 4. Across the patients, the mean plasma levels of atrial natriuretic peptide was correlated with the degree of hypoxaemia and the degree of oesophageal pressure swings during the sleep apnoeas. 5. Within the patients, cross-correlation studies suggested that the atrial natriuretic peptide secretory bursts were related either to the oesophageal pressure swings or to the apnoea-related hypoxaemia. 6. We conclude that release of atrial natriuretic peptide decreases with continuous positive airway pressure treatment in those patients with obstructive sleep apnoea who have increased release of atrial natriuretic peptide before treatment. Correspondence: Dr Jean Krieger, Clinique Neurologique, 67091 Strasbourg Cedex, France.
7. The results are in agreement with the hypothesis that the haemodynamic changes induced by the increased swings in intra-thoracic pressure during ineffective respiratory efforts or by the hypoxia-induced vasoconstriction playa role in these changes. Key words: arginine vasopressin, atrial natriuretic peptide, continuous positive airway pressure, obstructive sleep apnoea. Abbreviations: ANP, atrial natriuretic peptide; AVP, arginine vasopressin; CPAP, continuous positive airway pressure; OSA, obstructive sleep apnoea; Paco z, partial pressure of CO z; Pao z, partial pressure of 0z; Sao z, arterial oxygen saturation.
INTRODUCTION
The obstructive sleep apnoea (OSA) syndrome is due to the occurrence of upper airway occlusions resulting in apnoeas lasting 10-60 s or more, repeated up to several hundred times during a night's sleep; during the ineffective respiratory efforts, the patients develop dramatically increased (more negative) intra-thoracic pressures [1]. The major consequences are repeated asphyxia and sleep fragmentation. Clinical manifestations include daytime somnolence, snoring, and intellectual and sexual deficiency. Obesity, hypertension and polycythaemia are not uncommon [2]. Nasal continuous positive airway pressure (CPAP) has been shown to eliminate sleep apnoeas and apnoea-related hypoxaemia [3]; it also normalizes the sleep pattern, reduces intra-thoracic pressure swings, stabilizes the heart rate and eliminates pulmonary hypertension peaks [4, 4a]. Several recent reports have demonstrated an increase in urine and Na + excretion during sleep in patients with OSA [5, 6], which was normalized when these patients were treated with nasal CPAP. A decrease in the urinary
J. Krieger et al.
444
excretion of cyclic GMP with nasal CPAP [7], as well as measurements of plasma atrial natriuretic peptide (ANP) [8-10], suggested that ANP release might be increased during sleep in patients with OSA and be normalized with nasal CPAP treatment, although this has been questioned [11]. However, in these studies of ANP during sleep only a few blood samples were taken, making an analysis of the secretory profile of ANP impossible. In addition, no data concerning the correlates of changes in ANP were reported, so their mechanisms could not be investigated. It is clearly established that ANP is released in response to atrial stretch [12, 13]. Two conditions associated with the occurrence of OSA may result in atrial distension: repeatedly increased intra-thoracic pressures may act as a pump sucking blood into the thorax and thus increase the cardiac preload [14]; and the hypoxaemia associated with apnoeas may increase the pulmonary artery pressure due to hypoxaemia vasoconstriction, and thus increase the cardiac afterload. The purpose of the present work was to further describe the ANP secretory profile during sleep in patients with OSA, both untreated and treated with nasal CPAP, by sampling blood every 10 min, and to investigate the mechanisms of possible changes in plasma ANP content by simultaneously analysing changes in intrathoracic pressure and arterial oxygen saturation (Sa0 2). In addition, the plasma levels of arginine vasopressin (AVP) were also investigated.
METHODS Patients Nine patients with severe OSA participated in the study. They were selected from the patients diagnosed at our Sleep Disorders Centre on the basis of: (1) their age (less than 45 years), (2) the absence of any drug treatment, and (3) their willingness to participate in the study. After the details of the study had been fully explained, they all gave written consent. The protocol was approved by the Ethics Committee of Strasbourg University Hospital. Patients were aged 41 ± 3 (mean ± SD) years (range 34-46 years) with a body mass index of 35.9 ± 7.3 kg/rn? (range 27-47 kg/m 2 ) . They all had severe OSA, with an
apnoea index of 94 ± 21 apnoeas/h of sleep (range 57-123 apnoeas/h of sleep). They had moderate to severe hypoxaemia during sleep (mean lowest Sa02= 85.2 ±6.9%, range 69.4-91.6%) and high amplitude oesophageal pressure swings during apnoeas (mean maximal apnoeic pressure swings = 59.0 ± 27.5 cmH 20, range 21.8-118.6 cmH 20). None had clinical signs of heart or renal failure; daytime hypoxaemia was absent to moderate [partial pressure of 02 (Pa0 2 ) while awake breathing room air = 75 ± 12 mmHg, range 64-99 mmHg] without CO 2 retention [partial pressure of CO 2 (Pac0 2 ) 38.9 ± 2.0 mmHg, range 36-43 mmHg]. Their mean resting blood pressure was 144 ± 15/92 ± 14 mmHg (range 125/80 to 170/120 mmHg; Table 1). Protocol The patients were studied during two nights, at 7-14 day intervals, one night untreated, the other with nasal CPAP treatment. During each night they underwent polysomnography, using the standard techniques used in our sleep laboratory [15]. These include the recording of the electrooculogram EEG, and electromyogram of chin muscles, according to usual standards [15a]. Breathing was analysed by using a pneumotachograph (Fleisch no. 2) with a Godart-Statham pressure transducer and electronic integrator. Respiratory efforts were measured by using an oesophageal balloon and a Validyne MP 45 pressure transducer. Ear oximetry (Biox Ohmeda III) was used to determine Sa0 2• Nasal CPAP was applied on the first study night in three patients and on the second in six patients. During the treatment night, CPAP was applied via a nasal mask, by using a commercial device (Pression +, SEFAM). Treatment was started at a pressure of 3 cmH 20, and increased until the apnoeas and snoring were eliminated and intra-thoracic pressure swings were minimized. This was obtained within 30-45 min of sleep onset; the CPAP level reached was 12.4 ± 3.3 cmH 20 (range 8-18 cmH 20). During both nights an indwelling catheter was inserted through an antecubital vein 2-3 h before bedtime and connected via a through-the-wall tube to a pump sucking
Table 1. Main characteristics of the patients investigated Abbreviations: BMI, body mass index; AI, apnoea index; A Dur, mean apnoea duration; P os, mean maximal apnoeic oesophageal pressure swing; SBP, systolic blood pressure; DBP, diastolic blood pressure. Patient no.
Age (years)
(km/m")
BMI
AI (no./h)
1 2 3 4 5 6 7 8 9
42 46 42 42 39 38 34 41 42
31 36 27 28 27 45 40 32 37
73 123 90 122 83 95 57 98 103
ADur Mean minimal (s)
(%)
P os (cmH2O)
SBP (mmHg)
DBP (mmHg)
28 23 31 21 25 34 30 25 25
89 83 84 92 85 69 87 78 90
79 22 43 49 42 119 55 63 59
140 130 150 130 160 125 150 170 140
80 80 90 80 100 80 100 120 100
Sa0 2
Peo,
Pac0 2
(mmHg)
(mmHg)
67 69 84 64 71 64
39 39 36 36 41 43 38 38 40
77
78 99
Atrial natriuretic peptide and arginine vasopressin in obstructive sleep apnoea a continuous flow of 0.35 ml/rnin. A 3.5 ml sample was collected every 10 min starting at 22.00 hours. A heparin infusion of 17 i.u.jrnin through a separate vein in the same arm prevented tube clots. Tubes were kept on ice before and during sampling; samples were immediately centrifuged and plasma was frozen at - 20°C until assayed. All urine produced from bedtime until waking the next morning was collected. Assays All blood samples were analysed for their contents of ANP and AVP by r.i.a., using commercially available kits produced from Amersham (Amersham, Bucks, U.K.) for ANP and from Baxter (Maurepas, France) for AVP. ANP and AVP were extracted from the plasma by use of SepPak C I 8 cartridges (Waters, Milford, MA, U.S.A.). The intra-assay precision for duplicate samples was 5% for ANP and 6.2% for AVP. All samples from a given patient were run with the same kit. The urine Na" concentration was assayed by flame photometry (Flame Photometer IL 343, Instrumentation Laboratory, Lexington, MA, U.SA.). Urine flow was computed by dividing the volume by the collection time; Na" excretion was obtained by multiplying the urine flow by the Na" concentration. Data analysis In order to make parallel analyses between plasma samples and polysomnographic data possible, the polysomnographic recordings were divided into consecutive 10 min periods. For each period, the percentage of time awake and of each sleep stage was determined, as well as the number of apnoeas and their cumulative duration. For each period, the degree of respiratory effort was assessed by averaging the values of the highest oesophageal pressure swing during each apnoea, and the degree of hypoxaemia by averaging the lowest Saoz value reached after each apnoea. When a 10 min period was free of apnoeas (during intervening awakenings, or during the CPAP-treated night) an equivalent number of randomly chosen values was averaged over this period. A two-way analysis of variance (with two factors: order and CPAP treatment) for repeated measures on one factor (treatment) was performed first, to look for a possible order effect. Since no order effect could be demonstrated, the data were further analysed considering only the treatment effect. The effects of CPAP treatment was assessed within each patient by comparing plasma ANP and AVP values for the untreated versus the treated night, using the z-test for unpaired values. For the whole group, the mean plasma ANP and AVP values for the entire nights were compared for untreated versus treated nights, using the r-test for paired values. One-tailed tests were used for comparisons of plasma ANP levels where a decrease was expected, whereas two-tailed tests were used for plasma AVP levels, and whenever the direction of the variation had not been anticipated.
445
Correlations were calculated between the mean plasma ANP levels during the untreated nights and various variables depicting the patients' status (i.e. apnoea severity, sleep quality, haemodynamic condition, and urine and Na + excretion). To depict the relationships within the untreated nights between plasma ANP values and intra-thoracic pressures on the one hand, and plasma ANP values hypoxaemia on the other hand, cross-correlation analysis was performed in each patient between plasma ANP and AVP values and the values of the oesophageal pressure and the Sao z. Cross-correlation was used because it was expected that the changes in oesophageal pressure or in Sao z would antedate the changes in hormone secretion by one to several tens of minutes. Ten minute lag blocks of from -10 to - 50 min (lags from -1 to - 5) were analysed in order to cover changes occurring up to 50 min before sampling. Results are given as means ± SD.
RESULTS ANP The plasma ANP level during the untreated night was negatively correlated with the cumulative apnoea duration (r= -0.75, P=0.02) and the mean lowest Sao, (the mean of the minimal Sao z reached after each apnoea; r= - 0.81, P< 0.01), and positively with the mean highest oesophageal pressure swing (the mean of the highest pressure swing during each apnoea, r=0.79, P=O.Ol). No correlation was found with age, body mass index, systemic arterial pressure, daytime Peo; or Paco z, urine or Na + excretion, or sleep efficiency during the untreated night. The mean plasma ANP level before treatment in the whole group was 53.2 ± 23.3 pg/ml; it decreased to 35.8 ± 7.9 pg/rnl with CPAP treatment (P< 0.02, onetailed). Individually, the mean plasma ANP levels during sleep significantly decreased in six patients (Table 2), whereas they did not change in two patients and increased in one patient. These six patients tended to have higher mean plasma ANP values during the untreated nights (62.0±24.1 pg/ml; range 41.7-107.6 pg/rnl) than the three remaining ones (35.6±4.6 pg/ml; range 30.5-39.7 pg/ml; P= 0.10). These six patients had fluctuating plasma ANP levels, whereas ANP was more stable in the remaining three patients (see Fig. 1 for a typical example of a patient belonging to each group). The six patients with a high plasma level of ANP had a longer cumulative apnoea duration (in min spent in apnoea/h of sleep) than the three others (44.7 ± 7.0 versus 35.5 ± 6.9 min/h, P< 0.05). The decrease in mean plasma ANP level from untreated to treated night correlated with the mean plasma ANP level during the untreated night (r= - 0.94, P
443
Effects of treatment with nasal continuous positive airway pressure on atrial natriuretic peptide and arginine vasopressin release during sleep in patients with obstructive sleep apnoea J. KRIEGER*, M. FOLLENIUSt, E. SFORZA*, G. BRANDENBERGERt
AND
J. D. PETERt
'Service d'Explorations Fonctionnelles du Systeme Nerveux and :Departement d'Anesthesiologie, Centre Hospitalier Regional et Universitaire, Strasbourg, France, and tLaboratoire de Physiologie et Psychologie Environnementales, CNRS, Strasbourg, France
(Received 21 May 1990/3 December 1990; accepted 12 December 1990) SUMMARY
1. Patients with obstructive sleep apnoea have increased diuresis during sleep, which decreases with nasal continuous positive airway pressure treatment. These changes have been attributed to an increased release of atrial natriuretic peptide in obstructive sleep apnoea, and its decrease with continuous positive airway pressure treatment. 2. In order to clarify the change in plasma atrial natriuretic peptide level and to investigate the underlying mechanisms, blood samples were taken at 10 min intervals from nine patients with obstructive sleep apnoea during two nights when the patients were either untreated or treated with continuous positive airway pressure. Polysornnographic monitoring, including transcutaneous oximetry, and measurement of oesophageal pressure were performed simultaneously. Plasma arginine vasopressin was also measured. 3. The plasma level of arginine vasopressin did not change. The level of atrial natriuretic peptide was high and exhibited secretion bursts in six out of the nine patients; it drastically decreased with continuous positive airway pressure treatment. 4. Across the patients, the mean plasma levels of atrial natriuretic peptide was correlated with the degree of hypoxaemia and the degree of oesophageal pressure swings during the sleep apnoeas. 5. Within the patients, cross-correlation studies suggested that the atrial natriuretic peptide secretory bursts were related either to the oesophageal pressure swings or to the apnoea-related hypoxaemia. 6. We conclude that release of atrial natriuretic peptide decreases with continuous positive airway pressure treatment in those patients with obstructive sleep apnoea who have increased release of atrial natriuretic peptide before treatment. Correspondence: Dr Jean Krieger, Clinique Neurologique, 67091 Strasbourg Cedex, France.
7. The results are in agreement with the hypothesis that the haemodynamic changes induced by the increased swings in intra-thoracic pressure during ineffective respiratory efforts or by the hypoxia-induced vasoconstriction playa role in these changes. Key words: arginine vasopressin, atrial natriuretic peptide, continuous positive airway pressure, obstructive sleep apnoea. Abbreviations: ANP, atrial natriuretic peptide; AVP, arginine vasopressin; CPAP, continuous positive airway pressure; OSA, obstructive sleep apnoea; Paco z, partial pressure of CO z; Pao z, partial pressure of 0z; Sao z, arterial oxygen saturation.
INTRODUCTION
The obstructive sleep apnoea (OSA) syndrome is due to the occurrence of upper airway occlusions resulting in apnoeas lasting 10-60 s or more, repeated up to several hundred times during a night's sleep; during the ineffective respiratory efforts, the patients develop dramatically increased (more negative) intra-thoracic pressures [1]. The major consequences are repeated asphyxia and sleep fragmentation. Clinical manifestations include daytime somnolence, snoring, and intellectual and sexual deficiency. Obesity, hypertension and polycythaemia are not uncommon [2]. Nasal continuous positive airway pressure (CPAP) has been shown to eliminate sleep apnoeas and apnoea-related hypoxaemia [3]; it also normalizes the sleep pattern, reduces intra-thoracic pressure swings, stabilizes the heart rate and eliminates pulmonary hypertension peaks [4, 4a]. Several recent reports have demonstrated an increase in urine and Na + excretion during sleep in patients with OSA [5, 6], which was normalized when these patients were treated with nasal CPAP. A decrease in the urinary
J. Krieger et al.
444
excretion of cyclic GMP with nasal CPAP [7], as well as measurements of plasma atrial natriuretic peptide (ANP) [8-10], suggested that ANP release might be increased during sleep in patients with OSA and be normalized with nasal CPAP treatment, although this has been questioned [11]. However, in these studies of ANP during sleep only a few blood samples were taken, making an analysis of the secretory profile of ANP impossible. In addition, no data concerning the correlates of changes in ANP were reported, so their mechanisms could not be investigated. It is clearly established that ANP is released in response to atrial stretch [12, 13]. Two conditions associated with the occurrence of OSA may result in atrial distension: repeatedly increased intra-thoracic pressures may act as a pump sucking blood into the thorax and thus increase the cardiac preload [14]; and the hypoxaemia associated with apnoeas may increase the pulmonary artery pressure due to hypoxaemia vasoconstriction, and thus increase the cardiac afterload. The purpose of the present work was to further describe the ANP secretory profile during sleep in patients with OSA, both untreated and treated with nasal CPAP, by sampling blood every 10 min, and to investigate the mechanisms of possible changes in plasma ANP content by simultaneously analysing changes in intrathoracic pressure and arterial oxygen saturation (Sa0 2). In addition, the plasma levels of arginine vasopressin (AVP) were also investigated.
METHODS Patients Nine patients with severe OSA participated in the study. They were selected from the patients diagnosed at our Sleep Disorders Centre on the basis of: (1) their age (less than 45 years), (2) the absence of any drug treatment, and (3) their willingness to participate in the study. After the details of the study had been fully explained, they all gave written consent. The protocol was approved by the Ethics Committee of Strasbourg University Hospital. Patients were aged 41 ± 3 (mean ± SD) years (range 34-46 years) with a body mass index of 35.9 ± 7.3 kg/rn? (range 27-47 kg/m 2 ) . They all had severe OSA, with an
apnoea index of 94 ± 21 apnoeas/h of sleep (range 57-123 apnoeas/h of sleep). They had moderate to severe hypoxaemia during sleep (mean lowest Sa02= 85.2 ±6.9%, range 69.4-91.6%) and high amplitude oesophageal pressure swings during apnoeas (mean maximal apnoeic pressure swings = 59.0 ± 27.5 cmH 20, range 21.8-118.6 cmH 20). None had clinical signs of heart or renal failure; daytime hypoxaemia was absent to moderate [partial pressure of 02 (Pa0 2 ) while awake breathing room air = 75 ± 12 mmHg, range 64-99 mmHg] without CO 2 retention [partial pressure of CO 2 (Pac0 2 ) 38.9 ± 2.0 mmHg, range 36-43 mmHg]. Their mean resting blood pressure was 144 ± 15/92 ± 14 mmHg (range 125/80 to 170/120 mmHg; Table 1). Protocol The patients were studied during two nights, at 7-14 day intervals, one night untreated, the other with nasal CPAP treatment. During each night they underwent polysomnography, using the standard techniques used in our sleep laboratory [15]. These include the recording of the electrooculogram EEG, and electromyogram of chin muscles, according to usual standards [15a]. Breathing was analysed by using a pneumotachograph (Fleisch no. 2) with a Godart-Statham pressure transducer and electronic integrator. Respiratory efforts were measured by using an oesophageal balloon and a Validyne MP 45 pressure transducer. Ear oximetry (Biox Ohmeda III) was used to determine Sa0 2• Nasal CPAP was applied on the first study night in three patients and on the second in six patients. During the treatment night, CPAP was applied via a nasal mask, by using a commercial device (Pression +, SEFAM). Treatment was started at a pressure of 3 cmH 20, and increased until the apnoeas and snoring were eliminated and intra-thoracic pressure swings were minimized. This was obtained within 30-45 min of sleep onset; the CPAP level reached was 12.4 ± 3.3 cmH 20 (range 8-18 cmH 20). During both nights an indwelling catheter was inserted through an antecubital vein 2-3 h before bedtime and connected via a through-the-wall tube to a pump sucking
Table 1. Main characteristics of the patients investigated Abbreviations: BMI, body mass index; AI, apnoea index; A Dur, mean apnoea duration; P os, mean maximal apnoeic oesophageal pressure swing; SBP, systolic blood pressure; DBP, diastolic blood pressure. Patient no.
Age (years)
(km/m")
BMI
AI (no./h)
1 2 3 4 5 6 7 8 9
42 46 42 42 39 38 34 41 42
31 36 27 28 27 45 40 32 37
73 123 90 122 83 95 57 98 103
ADur Mean minimal (s)
(%)
P os (cmH2O)
SBP (mmHg)
DBP (mmHg)
28 23 31 21 25 34 30 25 25
89 83 84 92 85 69 87 78 90
79 22 43 49 42 119 55 63 59
140 130 150 130 160 125 150 170 140
80 80 90 80 100 80 100 120 100
Sa0 2
Peo,
Pac0 2
(mmHg)
(mmHg)
67 69 84 64 71 64
39 39 36 36 41 43 38 38 40
77
78 99
Atrial natriuretic peptide and arginine vasopressin in obstructive sleep apnoea a continuous flow of 0.35 ml/rnin. A 3.5 ml sample was collected every 10 min starting at 22.00 hours. A heparin infusion of 17 i.u.jrnin through a separate vein in the same arm prevented tube clots. Tubes were kept on ice before and during sampling; samples were immediately centrifuged and plasma was frozen at - 20°C until assayed. All urine produced from bedtime until waking the next morning was collected. Assays All blood samples were analysed for their contents of ANP and AVP by r.i.a., using commercially available kits produced from Amersham (Amersham, Bucks, U.K.) for ANP and from Baxter (Maurepas, France) for AVP. ANP and AVP were extracted from the plasma by use of SepPak C I 8 cartridges (Waters, Milford, MA, U.S.A.). The intra-assay precision for duplicate samples was 5% for ANP and 6.2% for AVP. All samples from a given patient were run with the same kit. The urine Na" concentration was assayed by flame photometry (Flame Photometer IL 343, Instrumentation Laboratory, Lexington, MA, U.SA.). Urine flow was computed by dividing the volume by the collection time; Na" excretion was obtained by multiplying the urine flow by the Na" concentration. Data analysis In order to make parallel analyses between plasma samples and polysomnographic data possible, the polysomnographic recordings were divided into consecutive 10 min periods. For each period, the percentage of time awake and of each sleep stage was determined, as well as the number of apnoeas and their cumulative duration. For each period, the degree of respiratory effort was assessed by averaging the values of the highest oesophageal pressure swing during each apnoea, and the degree of hypoxaemia by averaging the lowest Saoz value reached after each apnoea. When a 10 min period was free of apnoeas (during intervening awakenings, or during the CPAP-treated night) an equivalent number of randomly chosen values was averaged over this period. A two-way analysis of variance (with two factors: order and CPAP treatment) for repeated measures on one factor (treatment) was performed first, to look for a possible order effect. Since no order effect could be demonstrated, the data were further analysed considering only the treatment effect. The effects of CPAP treatment was assessed within each patient by comparing plasma ANP and AVP values for the untreated versus the treated night, using the z-test for unpaired values. For the whole group, the mean plasma ANP and AVP values for the entire nights were compared for untreated versus treated nights, using the r-test for paired values. One-tailed tests were used for comparisons of plasma ANP levels where a decrease was expected, whereas two-tailed tests were used for plasma AVP levels, and whenever the direction of the variation had not been anticipated.
445
Correlations were calculated between the mean plasma ANP levels during the untreated nights and various variables depicting the patients' status (i.e. apnoea severity, sleep quality, haemodynamic condition, and urine and Na + excretion). To depict the relationships within the untreated nights between plasma ANP values and intra-thoracic pressures on the one hand, and plasma ANP values hypoxaemia on the other hand, cross-correlation analysis was performed in each patient between plasma ANP and AVP values and the values of the oesophageal pressure and the Sao z. Cross-correlation was used because it was expected that the changes in oesophageal pressure or in Sao z would antedate the changes in hormone secretion by one to several tens of minutes. Ten minute lag blocks of from -10 to - 50 min (lags from -1 to - 5) were analysed in order to cover changes occurring up to 50 min before sampling. Results are given as means ± SD.
RESULTS ANP The plasma ANP level during the untreated night was negatively correlated with the cumulative apnoea duration (r= -0.75, P=0.02) and the mean lowest Sao, (the mean of the minimal Sao z reached after each apnoea; r= - 0.81, P< 0.01), and positively with the mean highest oesophageal pressure swing (the mean of the highest pressure swing during each apnoea, r=0.79, P=O.Ol). No correlation was found with age, body mass index, systemic arterial pressure, daytime Peo; or Paco z, urine or Na + excretion, or sleep efficiency during the untreated night. The mean plasma ANP level before treatment in the whole group was 53.2 ± 23.3 pg/ml; it decreased to 35.8 ± 7.9 pg/rnl with CPAP treatment (P< 0.02, onetailed). Individually, the mean plasma ANP levels during sleep significantly decreased in six patients (Table 2), whereas they did not change in two patients and increased in one patient. These six patients tended to have higher mean plasma ANP values during the untreated nights (62.0±24.1 pg/ml; range 41.7-107.6 pg/rnl) than the three remaining ones (35.6±4.6 pg/ml; range 30.5-39.7 pg/ml; P= 0.10). These six patients had fluctuating plasma ANP levels, whereas ANP was more stable in the remaining three patients (see Fig. 1 for a typical example of a patient belonging to each group). The six patients with a high plasma level of ANP had a longer cumulative apnoea duration (in min spent in apnoea/h of sleep) than the three others (44.7 ± 7.0 versus 35.5 ± 6.9 min/h, P< 0.05). The decrease in mean plasma ANP level from untreated to treated night correlated with the mean plasma ANP level during the untreated night (r= - 0.94, P
