Journal of Applied Physiology

Effects of nasal continuous positive airway pressure and oxygen supplementation on norepinephrine kinetics and cardiovascular responses in obstructive sleep apnea

Paul J. Mills, Brian P. Kennedy, Jose S. Loredo, Joel E. Dimsdale, Michael G. Ziegler


Obstructive sleep apnea (OSA) is characterized by noradrenergic activation. Nasal continuous positive airway pressure (CPAP) is the treatment of choice and has been shown to effectively reduce elevated norepinephrine (NE) levels. This study examined whether the reduction in NE after CPAP is due to an increase in NE clearance and/or a decrease of NE release rate. Fifty CPAP-naive OSA patients with an apnea-hypopnea index >15 were studied. NE clearance and release rates, circulating NE levels, urinary NE excretion, and blood pressure and heart rate were determined before and after 14 days of CPAP, placebo CPAP (CPAP administered at ineffective pressure), or oxygen supplementation. CPAP led to a significant increase in NE clearance (P ≤ 0.01), as well as decreases in plasma NE levels (P ≤ 0.018) and daytime (P < 0.001) and nighttime (P < 0.05) NE excretion. NE release rate was unchanged with treatment. Systolic (P ≤ 0.013) and diastolic (P ≤ 0.026) blood pressure and heart rate (P ≤ 0.014) were decreased in response to CPAP but not in response to oxygen or placebo CPAP treatment. Posttreatment systolic blood pressure was best predicted by pretreatment systolic blood pressure and posttreatment NE clearance and release rate (P < 0.01). The findings indicate that one of the mechanisms through which CPAP reduces NE levels is through an increase in the clearance of NE from the circulation.

noradrenergic activation, as indicated by augmented sympathetic neural activity, elevated circulating norepinephrine (NE) levels, and elevated urinary NE excretion, is a hallmark of obstructive sleep apnea (OSA) (3, 4, 8). Cardiovascular consequences of this activation include high blood pressure and enhanced cardiac contractility, as well as adrenergic receptor desensitization (8, 10, 1618). The elevated NE levels in OSA could be the result of an enhanced NE release rate and/or decreased NE clearance rate. We previously reported that OSA patients have an enhanced NE release rate under hypoxic conditions and a tendency toward decreased NE clearance (27).

Treatment of OSA with nasal continuous positive airway pressure (CPAP) reduces circulating levels of norepinephrine, as well as NE excretion (2, 8, 9, 11). The CPAP-induced reduction in NE is associated with a reduction in blood pressure, cardiac contractility, and restoration of desensitized β-adrenergic receptors (18, 19, 25).

The purpose of this study was to determine whether CPAP reduces NE levels by altering daytime release and/or clearance rate and whether such effects would be related to reductions in blood pressure. Considering prior observations that patients with OSA have a tendency toward diminished NE clearance and increased NE levels and that CPAP reduces NE levels, we hypothesized that CPAP treatment would lead to an increase in NE clearance and/or a decrease in NE release rate.

We compared CPAP to a placebo CPAP treatment and to an oxygen supplementation treatment. Nocturnal supplemental oxygen has been suggested by some as an alternative therapy in the nonsomnolent or the CPAP-noncompliant OSA patient (14, 21). In some OSA patients who cannot tolerate CPAP and are not candidates for a surgical procedure, supplemental oxygen therapy is associated with a reversal of OSA-related nocturnal hypoxemia but does not affect respiratory disturbance index (RDI) (14). The effect of supplemental oxygen on sympathetic activation in OSA has not been rigorously studied.


Patients and screening.

Patients were recruited from the community through advertisement and referral from sleep-disorder clinics. Screening criteria included clinical suspicion of OSA as per loud snoring with or without excessive daytime sleepiness. Because apneic patients frequently have hypertension (6), and because antihypertensive treatments have noticeable effects on the sympathetic nervous system (24), all hypertensive patients receiving such treatment were tapered off medication for at least 2 wk before participation in the study. Other inclusion criteria included weight being between 1.0 and 2.0 times the ideal body weight (15) and age between 30 and 65 yr. Patients were excluded if they were receiving medications known to affect sleep. Patients with other major medical disorders, including congestive heart failure, symptomatic obstructive pulmonary disease, history of narcolepsy, prior surgery for treatment of OSA, current alcohol or drug abuse, or psychosis were excluded.

Screening included a history and physical examination and complete blood count, chemistry panel, and electrocardiogram. Potentially eligible candidates underwent an unattended overnight home screening sleep study using the Stardust (Respironics, Marietta, GA) home sleep recording system. Subjects with an apnea-hypopnea index (AHI) >15 were then admitted for 3 nights to the University of California, San Diego (UCSD) General Clinical Research Center (GCRC) Gillin Laboratory of Sleep and Chronobiology, where full polysomnography confirmed the diagnosis of OSA (2, 11). Subjects with periodic limb movement >15 on baseline polysomnography were excluded from further participation. The same team of nighttime technicians and daytime technicians performed and scored the polysomnograms.

Sleep was recorded using the Grass Heritage (model PSG36-2, West Warwick, RI) sleep recording system using a standard montage. All records were scored according to Rechtshaffen and Kales criteria (22). Percent time less than 90% arterial oxygen saturation (SaO2<90%) was determined and averaged over the nights. Apneas were defined as a decrement in airflow >90% from baseline for a period ≥10 s. Hypopneas were defined as a decrement in airflow ≥50% but <90% from baseline for a period ≥10 s. The number of apneas and hypopneas per hour of sleep were determined to obtain the AHI.

Fifty men and women completed screening and were admitted to the study. All patients gave signed consent to the protocol, which was approved by the UCSD Human Subjects Institutional Review Board.

Treatment design and randomization.

Randomization took place on the second night of admission. Patients were randomized to a 2-wk therapeutic trial of nasal CPAP (n = 17), placebo CPAP (n = 17), or nocturnal oxygen (n = 17). In the therapeutic CPAP group, conventional manual overnight CPAP titration was performed in increasing steps of 1–2 cmH2O until unequivocal obstructive apneas or hypopneas were controlled in the second or third rapid eye movement sleep period. All patients randomized to CPAP had an effective titration as defined by an AHI <10. Patients randomized to placebo CPAP or supplemental oxygen underwent a mock titration. On the third night of admission, patients slept with their assigned treatment.

Equipment for the three treatment arms was similar and consisted of a CPAP generator (Aria LX CPAP System, Respironics, Murrysville, PA), CPAP mask (Profile Light, Respironics), tubing, heated humidifier (Fisher and Pykel HC199, Aukland, New Zealand), and oxygen concentrator (Alliance, Healthdyne Technologies model 505, Marietta, GA). The concentrator could be switched to produce room air. The supplemental gas (room air or oxygen) was introduced into the CPAP system at the level of the humidifier. To maintain the blind to treatment, subjects randomized to therapeutic CPAP received active CPAP plus an oxygen concentrator providing room air. Subjects randomized to placebo CPAP received subtherapeutic CPAP (<1 cmH2O at the mask) plus an oxygen concentrator providing room air. The placebo CPAP consisted of a CPAP mask with 10 quarter-inch drill holes for adequate room air exchange with pressure set at a constant 3 cmH2O. (25). A pressure reducer was placed in the tubing between the CPAP unit and the modified mask. With this system, the pressure at the mask was 0.5 cmH2O at end expiration and 0 cmH2O during inspiration, and the patients felt a gentle breeze at the nose. Those assigned to nocturnal oxygen received placebo CPAP plus an oxygen concentrator delivering oxygen at 3 l/min (inspired oxygen fraction of 32–34% at the mask).

The morning after the third night, patients were instructed and discharged with their assigned home treatment. Research staff had frequent phone calls with patients to answer questions about the equipment and to encourage compliance with the therapy. All CPAP units had a hidden compliance clock allowing measurement of the nightly time the unit was switched on at pressure. After 2 wk of home treatment, patients were readmitted to the GCRC for another full overnight polysomnography with their assigned treatment.

NE clearance and release rate.

Before randomization, and after 2 wk of treatment, patients underwent NE kinetics testing at the UCSD GCRC. Patients refrained from drinking caffeinated beverages since the preceding evening and consumed a light breakfast before testing. Blood samples were collected by using an indwelling catheter placed early in the morning in a forearm vein contralateral to the arm used for the infusion. NE clearance and release rates were measured in supine subjects as previously described (27) using [3H]NE of >98% radiochemical purity (New England Nuclear, Boston, MA). [3H]NE was infused into an antecubital vein at 1.5 μCi/min for 10 min. The infusion rate was then decreased to 0.78 μCi/min. Plateau levels were obtained in <1 h with this technique. Blood samples were drawn from the antecubital vein of the contralateral arm to measure plasma NE and [3H]NE levels during the infusion and for 16 min after the infusion was terminated. Clearance of NE measured in arterial and venous blood differs slightly. We chose to measure clearance of [3H]NE on venous samples because all reports of changes in NE levels among patients with sleep apnea have used venous samples, so elucidation of the effects of clearance on NE levels is most appropriately carried out on venous blood samples. The [3H]NE content of the infusate and plasma samples was measured after alumina chromatography by scintillation spectroscopy. O-methylated metabolites of NE do not adhere to alumina. We checked for the presence of deaminated metabolites of NE by solvent extraction of plasma samples and found no measurable levels of [3H]NE metabolites after alumina chromatography. Plasma NE levels were measured by the radioenzymatic method of Kennedy and Ziegler (13).

The rate of NE clearance from plasma was calculated by the formula: Math The rate of NE release was calculated by the formula: Math Volume of distribution was calculated from the half-life and clearance by the equation: volume = clearance/rate constant of decay (26). The plasma NE half-lives were derived by using a standard curve-fitting program assuming biexponential decay of the [3H]NE.

NE excretion.

While at the GCRC, patients had urine collected from 1700 to 2200, 2200 to 0600, and 0600 to 1700. We evaluated the adequacy of urine collection by measures of volume and creatinine excretion. Urine NE was measured by a catechol-O-methyltransferase-based radioenzymatic procedure with a purification step before assay (13). Urinary NE excretion is expressed in micrograms excreted per hour during wake (16 h, 0600–2200) and sleep (8 h, 2200–0600).

Blood pressure, heart rate, and plasma NE.

Blood pressure and heart rate were assessed at a different time by using an automated monitor while the subject was supine. The means of three consecutive measurements were taken as the blood pressure and heart rate values. Under these supine conditions, blood was drawn for plasma NE levels and assayed as previously described (13).

Data analysis.

Data were analyzed by two-way (group × time) ANOVA and hierarchical multivariable regression. Separate ANOVAs were run to test whether treatment led to a change in NE clearance and to test whether treatment led to a change in NE release rate. Distribution of all variables was evaluated by Levene's test of equality of variance; parametric tests were used only in case of normal distribution. Initial relationships among variables were explored by Pearson correlations. Significance level was set at P < 0.05 for two-sided tests. Because NE clearance and release rate are not independent variables, these ANOVA tests were alpha adjusted. Data analyses were performed using the SPSS 12.0 programming package (Chicago, IL). All variables are presented as means ± SE.



Table 1 presents subject characteristics and treatment compliance data. Body mass index (BMI) was computed as the ratio of body weight in kilograms divided by the square of height in meters (kg/m2). Before treatment, subjects randomized to the three treatment groups were similar in age, BMI, gender distribution, and hypertension distribution. Compliance with treatment was based on the average number of hours of use per night during the 2-wk treatment period. There was no significant difference among the three groups in treatment compliance (F = 1.693, P = 0.195).

View this table:
Table 1.

Subject characteristics and treatment compliance data

Before treatment, among all patients, blood pressure was related to both day and night NE excretion rates (P < 0.05) (Table 2). AHI was related to NE release rate (P < 0.01) and day and night NE excretion rates (P < 0.05). SaO2<90% correlated with NE release rate (P < 0.01), supine plasma NE levels (P < 0.01), and day and night NE excretion rates (P < 0.01).

View this table:
Table 2.

Pretreatment correlation coefficients among all subjects


Table 3 presents the effects of treatment on AHI, SaMath<90%, blood pressure, and heart rate. There were no significant pretreatment differences among the groups for any of these variables. AHI was significantly reduced in the CPAP condition (F = 28.9, P < 0.001) but not in the oxygen or placebo CPAP conditions. SaMath<90% was significantly reduced in the CPAP and the oxygen conditions (F values >6.3, P < 0.05) but not in the placebo CPAP condition. Systolic (F = 7.8, P ≤ 0.013) and diastolic (F = 6.1, P ≤ 0.026) blood pressure and heart rate (F = 7.7, P ≤ 0.014) were decreased in response to CPAP but not the oxygen or placebo CPAP.

View this table:
Table 3.

Treatment effects on apnea-hypopnea index, SaO2 < 90%, and blood pressure and heart rate

Two weeks of CPAP led to a significant increase in NE clearance (F = 9.53, P ≤ 0.01) (Table 4 and Fig. 1). The enhanced clearance of NE was due to an expanded volume of distribution (F = 7.7, P ≤ 0.017), not to a shortened half-life (P = 0.06) (Table 4). NE clearance, volume of distribution, and half-life were unchanged after either oxygen or placebo CPAP. Supine plasma NE levels were reduced after CPAP (F = 7.9, P ≤ 0.018) but unchanged after either oxygen or placebo CPAP. Daytime NE excretion was reduced after 2 wk of CPAP treatment (F = 20.8, P < 0.001) and oxygen treatment (F = 9.9, P < 0.01) but unchanged after placebo CPAP. Nighttime NE excretion was reduced after CPAP treatment only (F = 5.6, P < 0.05). NE release rate was unchanged with any treatment (Fig. 1).

Fig. 1.

Plasma norepinephrine clearance (A) and release rate (B) in patients with sleep apnea following 14 days of continuous positive airway pressure (CPAP) (n = 17), oxygen (O2) (n = 17), or placebo CPAP (n = 16). CPAP led to a significant increase in norepinephrine clearance (*P ≤ 0.01), whereas there was no change following O2 supplementation or placebo CPAP. Norepinephrine release rate was unchanged with any treatment. Pre, before treatment; Post, after treatment.

View this table:
Table 4.

Treatment effects on norepinephrine kinetics, plasma levels, and urinary excretion

In an effort to better understand the role of NE in blood pressure and heart rate responses to treatment, we conducted a series of multiple regression analyses examining possible predictors of posttreatment levels of systolic and diastolic blood pressure and heart rate. Dependent variables were entered into the regression in blocks as follows: block 1: age, BMI, gender, and diagnosis of hypertension; block 2: pretreatment AHI, SpMath<90%, and the respective pretreatment blood pressure or heart rate; block 3: pretreatment NE clearance, NE release rate, supine plasma NE, and daytime and nighttime NE excretion; block 4: posttreatment AHI, SaMath<90%; block 5: posttreatment NE clearance, NE release rate, supine plasma NE, and daytime and nighttime NE excretion. Posttreatment systolic blood pressure was predicted by pretreatment systolic blood pressure (β = 0.450, P = 0.005), posttreatment NE clearance (β = −0.836, P = 0.001), and posttreatment NE release rate (β = 0.717, P = 0.005), yielding the full regression model of r2 = 0.687, F = 16.1, P < 0.001, with all other predictor variables dropping out as not significant. Posttreatment diastolic blood pressure was predicted by pretreatment diastolic blood pressure (full regression model: r2 = 0.284, F = 10.5, P < 0.01). Posttreatment heart rate was predicted by pretreatment heart rate (β = 0.706, P = 0.000), pretreatment NE clearance (β = 0.253, P = 0.008), and pretreatment SaMath<90% (β = 0.293, P = 0.004), yielding the full regression model of r2 = 0.788, F = 35.7, P < 0.001.


The impetus for this study arose from observations that CPAP treatment successfully reduces plasma and urine NE in OSA. We wondered whether a mechanism of this effect was CPAP-induced alterations in NE clearance or release. We observed that in patients treated with CPAP, in addition to reductions in circulating NE and NE excretion, daytime NE clearance was increased. We did not observe a change in daytime NE release rate, although prior studies have shown that OSA patients have increased sympathetic nerve activity in the daytime and that CPAP leads to a reduction in sympathetic neural activity during sleep. Our measures help specify where CPAP induces changes in daytime sympathetic nerve activity. There was a clear reduction in urine NE with CPAP. NE in the urine comes from the blood and from NE released by sympathetic nerves in the kidney. Our findings suggest that CPAP causes a reduction in renal sympathetic neuronal activity, which might diminish sodium retention and renin release. CPAP is reported to diminish daytime muscle sympathetic nerve activity to the leg (23), so there appears to be localized changes in nerve activity to the kidney and some muscle vasculature, although we did not find an overall decrease in NE release rate. CPAP also led to an increase in the volume of distribution for NE. This could be due to more active diffusion of NE out of the bloodstream, enhanced reversible transport of NE by uptake-1 and uptake-2 mechanisms, or enhanced binding of NE by plasma proteins and cellular structures. Our study does not indicate which of these potential mechanisms predominates, but we have previously found a suggestion of enhanced binding to receptors after CPAP (25).

We compared CPAP against another treatment for OSA, namely oxygen supplementation. Although both treatments successfully improved oxygen saturation, only CPAP reduced AHI. Only CPAP significantly lowered daytime blood pressure; its effect was about twice as large as the nonsignificant decrease seen with oxygen. Both CPAP and oxygen lowered daytime urine NE; the effect of CPAP was about twice as large as oxygen. CPAP also lowered nighttime urine NE, but oxygen failed to do so. This is not surprising, because oxygen failed to alter the number of nighttime apneas, and sympathetic nerve activity increases markedly after apneic episodes (25).

We previously reported that CPAP reduces daytime ambulatory mean arterial blood pressure to the same extent as placebo CPAP but that CPAP leads to a much greater decrease in nighttime ambulatory mean arterial blood pressure compared with placebo CPAP (5). This previous study was of 1-wk duration as opposed to 2-wk duration herein. Studies by Becker et al. (1) and Pepperrell et al. (20) were of longer duration and found blood pressure to decrease with CPAP. It is likely that the beneficial effects of CPAP on blood pressure are more evident with longer treatment.

OSA is associated with hypertension. The proposed mechanism for the association is sympathetic activation (3, 4, 8). Intermittent hypoxia causes hypertension in rats, but chronic hypoxia does not usually cause hypertension in humans (7). On the other hand, recovery from individual episodes of apnea increases sympathetic nerve activity and blood pressure. Our study suggests that apneic episodes are more important than hypoxia in increasing plasma and urine NE and blood pressure.

We also examined possible predictors of posttreatment blood pressure and heart rate. We found that posttreatment systolic blood pressure was best predicted by a combination of pretreatment systolic blood pressure and posttreatment NE clearance and release rates (e.g., a lower posttreatment blood pressure was associated with a lower pretreatment blood pressure, a higher posttreatment NE clearance, and a lower posttreatment NE release rate). We are not aware of evidence for a direct link between a change in NE clearance and cardiovascular consequences. Although increased clearance decreases plasma NE levels, the decrease is too small to have direct measurable consequences. An indirect link between NE clearance and cardiovascular consequences is that NE clearance is useful in calculating NE release rate. NE release rate did not change with CPAP therapy despite the decrease in plasma NE levels. In contrast, urine NE levels did decrease with CPAP. An unchanged NE release rate and a diminished urine NE level after CPAP imply that CPAP decreases renal sympathetic nerve activity. Decreased renal sympathetic nerve activity might decrease sodium retention and renin, with lower blood pressure as a consequence. We did find lower blood pressure after CPAP.

As described in methods, we took the opportunity to assess plasma NE at two different time points in the study. The general magnitude and direction of change of the two plasma NE levels in Table 4 were very consistent, although only the supine plasma NE level taken with the blood pressure and heart rate assessment was significantly lower at posttreatment in the CPAP group.

Our randomization of apneic patients to treatment was successful in that there were no significant pretreatment differences among the three groups in terms of apnea severity, age, weight, or blood pressure. Before treatment, among all patients, apnea severity, as indicated by AHI and SaMath<90%, was related to higher plasma NE levels and higher NE excretion, confirming prior observations of increasing noradrenergic activation with increasing apnea severity (3, 4, 8).

In summary, 2 wk of CPAP treatment for OSA resulted in increased NE clearance rate. Posttreatment systolic blood pressure was related to pretreatment systolic blood pressure and posttreatment NE clearance and NE release rate. CPAP lowered urinary NE and blood pressure more effectively than oxygen supplementation, suggesting that apneic episodes might be more important than hypoxia in determining sympathetic activity and blood pressure.


This work was supported by grants HL-073355, HL-40102, and HL-57265 from the National Heart, Lung, and Blood Institute and the UCSD General Clinical Research Center (MO1RR-00827).


  • The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.


View Abstract