INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Approximately 30% of patients with arterial hypertension may be expected to have obstructive sleep apnea (OSA) (1- 3). Conversely, the prevalence of hypertension among patients with OSA has been estimated to vary around 50% (4, 5). Although it remains a matter of debate whether OSA per se causes hypertension, effective reduction of apneas during sleep, through the use of continuous positive airway pressure (CPAP), has become an important component of hypertension management when OSA is present. However, elimination or reduction of apneas cannot be expected to normalize hypertension in all of these patients (6). Moreover, long-term treatment of apnea cannot be implemented in the whole OSA population, but only in a limited proportion (7). Consequently, drug therapy remains a mainstay of hypertension management in OSA.

Specific (known or unknown) physiologic mechanisms related to apneas during sleep may be responsible for the development of hypertension in OSA (8). It may therefore be speculated that antihypertensive drugs with different pharmacodynamic modes of action might perform differently in lowering blood pressure (BP) in hypertensive OSA patients. For example, if overactivity of the sympathetic nervous system, a condition closely linked to OSA (9), is thought to result in hypertension, treatment with a sympathicolytic substance may be expected to be particularly effective. However, since the pathophysiology of hypertension in OSA is still incompletely understood, superior effectiveness of other pharmacodynamic principles cannot be ruled out. Therefore, we compared the BP-lowering effect of substances representing five different classes of antihypertensive agents: atenolol (a ₁-selective -blocker), amlodipine (a calcium-channel blocker), enalapril (an angiotensin-converting enzyme [ACE] inhibitor), hydrochlorothiazide (a diuretic), and losartan (an angiotensin receptor antagonist) in patients with OSA and hypertension.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Study Population

Participants were selected from a population of males aged 25 to 70 yr and with both arterial hypertension and OSA. Arterial hypertension was defined as ongoing antihypertensive treatment or at least three office BP measurements (supine position) yielding a systolic BP (SBP) of 160 mm Hg or more or diastolic BP (DBP) of 95 mm Hg or more after a minimum of 15 min of undisturbed rest. For randomization, a stable sitting DBP of between 95 and 115 mm Hg at the end of the run-in period was also required. The assessment of disordered breathing during sleep was based on at least one overnight ambulatory monitoring of ventilation (Eden Trace [Edentec, Eden Prairie, MN] or in-house equipment), including recording of oxygen saturation (Sa_O2) (by pulse oximetry), respiratory movements (impedance), heart rate (HR), nasal and oral air flow (by thermistor), and body position. Patients with an oxygen desaturation index or an apnea-hypopnea index (AHI)  10/h (based on self-reported sleep duration) were considered eligible for participation in the study (nine patients with a mean desaturation index of 18.7 ± 7.9/h [mean ± SD], and 31 patients with a mean AHI of 40.7 ± 47.5/h, respectively). The cutoff value of 10/h for the desaturation index or AHI was chosen to minimize the probability of erroneous inclusion of subjects without OSA as a result of low diagnostic precision. Baseline measurements made after enrollment into the study (Table 4) confirmed that only subjects with OSA had been included. (More recently established clinical criteria for the diagnosis of OSA have been reported [13].) We excluded patients with myocardial infarction or symptoms of cerebrovascular disease occurring within 3 mo before enrollment, as well as those with treatment-demanding coronary heart disease, bronchial asthma or chronic obstructive pulmonary disease, insulin-treated diabetes, cardiac failure, cardiac arrhythmia (except atrioventricular-block or single bundle-branch block), known allergy to any of the study drugs, current alcohol or drug abuse, and other diseases judged by the investigators to threaten patient safety or compliance with the study protocol. Use of continuous positive airway pressure (CPAP) was not allowed during the study. Before inclusion in the study, all patients received written and oral information and gave written informed consent.

Outcome Variables

The primary outcome variable of the study was the change in daytime office DBP during treatment, on the basis of the: (1) proven validity of office DBP as a surrogate marker of cardiovascular morbidity and mortality; (2) extensive use of office BP for the diagnosis of hypertension and evaluation of treatment effects in clinical and research practice; and (3) high reliability of DBP in terms of patient compliance and technical solidity. As secondary outcomes, we measured office SBP and HR, as well as ambulatory daytime and nighttime SBP, DBP, and HR. In order to detect clinically important adverse effects of individual drugs on AHI, plasma electrolytes, blood glucose, or plasma lipids, we monitored nocturnal ventilation and took blood samples during treatment. Similarly, use of questionnaires for the assessment of well-being was intended to disclose evident beneficial effects or adverse events of a particular treatment.

Study Design

The study was approved by the Ethics Committee of the Medical Faculty of Gothenburg University. After screening and enrollment, patients underwent a single-blind placebo washout period lasting 3 wk. Thereafter, each of the 40 patients who fulfilled the study inclusion criteria was randomized to one of 20 possible treatment sequences, each comprising two successive treatment periods, during each period of which one of the five study drugs was given (Table 1). Thus, two patients were randomized to each possible treatment sequence. The two double-blind treatment periods of 6 wk each were separated by a single-blind washout period lasting 3 wk. This balanced incomplete block design was selected for its effectiveness (i.e., the ability to provide relatively precise comparisons among treatments with a given number of subjects [14]).

Medications

Study medications for once-daily intake were prepared by the hospital pharmacy, using commercially available tablets in the following dosages for once-daily use: atenolol 50 mg (Tenormin; Zeneca AB, Gothenburg, Sweden), amlodipine 5 mg (Norvasc; Pfizer AB, Täby, Sweden), enalapril 20 mg (Renitec; Merck, Sharp & Dohme Sweden AB, Sollentuna, Sweden), hydrochlorothiazide 25 mg (Esidrex; Novartis Sverige AB, Täby, Sweden), and losartan 50 mg (Cozaar; Merck, Sharp & Dohme Sweden AB). The dosages for amlodipine, hydrochlorothiazide, and losartan corresponded to defined daily dosages obtained from the Swedish Medical Products Agency, and to recommended dosages for maintenance treatment from the respective Swedish manufacturers. On the basis of previously reported comparative studies (15), the atenolol dosage (50 mg) was selected to be smaller than the defined daily dosage (75 mg), whereas the enalapril dosage (20 mg) was selected to exceed the defined daily dosage (10 mg). Patients were instructed to take their study medications at home, between 7:00 and 9:00 A.M.

For blinding, tablets were given in opaque capsules. Moreover, investigators and nurses were not allowed to open a patient's tablet container before the end of the study. In order to prevent possible identification of patients receiving -adrenoceptor antagonists through a decreased HR, no comparisons with baseline values were made during the study. For determining compliance at the end of the treatment periods, returned tablets were counted.

Clinical Assessments

Screening for enrollment in the study included a medical history as well as routine clinical and laboratory examinations. At the ends of the run-in period, the two active treatment periods, and the washout period, office and ambulatory BPs, and HR were measured, blood samples for clinical-chemical analysis were taken, general well-being was screened with visual analogue scales, and nocturnal breathing was monitored. For safety reasons and assessment of adverse events (through exact open questions), visits for measurement of office BP and patient interviews were scheduled every second week.

Trough DBP and SBP, taken with the patient in the sitting position at 24 h after the last intake of study medication, were averaged from three measurements obtained at 1-min intervals after at least 5 min of undisturbed rest in the sitting position (cuff method; Omega 1400; Invivo Research Laboratories Inc., Orlando, FL). Ambulatory BP was recorded in the nondominant arm at 20-min intervals in a single 24-h cycle, with a portable BP monitor (Model 90207; SpaceLabs, Redmond, WA) (16). Patients were instructed to cease activity and keep the measurement arm still during readings. Mean values of DBP, SBP, and HR were calculated for predefined daytime (9:00 A.M. to 9:00 P.M.) and nighttime (11:30 P.M. to 5:30 A.M.) periods, using interval-weighted measurement values (17).

Blood samples were obtained after an overnight fast and were immediately sent to the hospital laboratory for routine analysis of blood glucose and the plasma contents of sodium, potassium, creatinine, alanine aminotransferase, aspartate aminotransferase, alkaline phosphatase, and total and high-density lipoprotein (HDL)-cholesterol.

The Minimal Symptoms Evaluation (MSE) questionnaire (18) was used to quantify selected aspects of general well-being through visual analogue scales (VAS). The extent of fatigue, daytime sleepiness, morning sleepiness, morning headache, nocturnal sweating, and irritability were monitored with six additional VAS.

During the study, ventilation during sleep was recorded with an ambulatory system (Eden Trace), which plots Sa_O2, respiratory movements (impedance), HR, nasal and oral air flow (thermistor), breathing sounds, and body position during a one-night sleep period. Apneas and hypopneas were defined as episodes lasting 10 s or longer with airflow cessation or reductions of at least 50% in the thermistor signal amplitude, respectively. The AHI (events per hour) was estimated by a blinded observer (L.G.) counting all apneas and hypopneas during one night and dividing the number by the patient's self-estimated sleep time.

Analysis of Data

After testing each variable for the absence of a difference between corresponding run-in and washout measurements, we used the average of these two values as a baseline estimate of the primary study outcome, office DBP, and secondary outcomes, respectively. Differences between measurements obtained during active treatment and the corresponding baseline value were expressed as 95% confidence intervals (CIs) and were compared through analysis of variance (ANOVA), with allowance for treatment order and subject (SPSS version 8.0 software; SPSS, Inc., Chicago, IL [14]). If the p value associated with the main factor (treatment) fell below 0.05, single contrasts between the five treatments were tested with the Newman-Keuls test. Differences with values of p < 0.05 were considered statistically significant.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The age and body mass index of the 40 randomized patients were 56.9 ± 7.0 yr and 29.5 ± 3.0 kg/m² (mean ± SD), respectively. On average, hypertension had been diagnosed 6.4 ± 6.2 yr before enrollment in the study. The onset of typical OSA symptoms causing the patients to seek medical aid occurred 2.8 ± 4.2 yr before the study. Eight patients had previously tried but voluntarily ended CPAP therapy. The remaining participants had not yet commenced CPAP therapy. At inclusion, 10 patients took no antihypertensive drugs, 22 patients were treated with a single medication (six with a -antagonist, six with a calcium-channel antagonist, four with an ACE-inhibitor, three with a diuretic, and three with an angiotensin receptor antagonist), seven patients were treated with a combination of two antihypertensive agents (two with a -antagonist, four with a calcium-channel antagonist, three with an ACE-inhibitor, three with a diuretic, and two with an angiotensin-receptor antagonist), and one patient took a combination of three antihypertensive agents (a -antagonist, calcium-channel antagonist, and ACE inhibitor). This last patient withdrew his consent because of headache after the first treatment period (despite well-controlled BP with enalapril). Another patient, whose DBP exceeded the safety limit of 115 mm Hg, was excluded during the first treatment period. The remaining 38 subjects finished both treatment periods. Other missing results for 24-h BP measurements, clinical chemical analyses, or nocturnal monitoring of ventilation (numbers of valid measurements specified in Tables 234) had technical explanations not linked to specific treatments.

At the end of the run-in period, office SBP, DBP, and HR averaged, respectively, 161.3 ± 11.8 mm Hg, 102.5 ± 5.2 mm Hg, and 74.6 ± 12.2 beats/min. Mean daytime ambulatory SBP, DBP, and HR were 154.2 ± 15.4 mm Hg, 100.51 ± 9.9 mm Hg, and 81.4 ± 11.9 beats/min, respectively. Mean nighttime ambulatory SBP, DBP, and HR were 131.9 ± 14.3 mm Hg, 81.4 ± 10.9 mm Hg, and 67.8 ± 9.4 beats/min, respectively. Office DBP, our primary outcome variable, was most effectively lowered by atenolol (Table 2), with all four post hoc differences between atenolol and the remaining treatment agents being significant. Office SBP reductions, in contrast, did not differ significantly among the five treatments. There were no significant differences between the reductions in mean daytime ambulatory DBP or SBP with any of the treatments (Table 3). However, atenolol reduced mean nighttime ambulatory DBP and SBP more effectively than did amlodipine, enalapril, or losartan (but not hydrochlorothiazide). The most marked characteristic of atenolol was its dampening effect on office, daytime, and nighttime HR, which were virtually uninfluenced by the other four substances (Tables 2 and 3).

There was no evidence of a differential influence of the five treatments on AHI, which averaged 43.0 ± 27.4/h at randomization. A weak decrease in AHI found during treatment with amlodipine, and a slight increase in AHI with losartan, did not reach statistical significance (Table 4). (A Newman-Keuls test based on the nearly significant ANOVA result yielded a significant difference between amlodipine and losartan, but insignificant differences for all remaining comparisons.)

The following clinical-chemical results were obtained at randomization: potassium, 4.25 ± 0.34 mmol/L; glucose, 4.74 ± 1.11 mmol/L; total cholesterol, 5.90 ± 0.99 mmol/L; HDL-cholesterol, 1.22 ± 0.33 mmol/L; and triglycerides, 2.16 ± 1.52 mmol/L. Hydrochlorothiazide but none of the remaining substances (data not shown), lowered the plasma potassium concentration significantly (0.32 mmol/L; 95% CI: 0.47 to 0.16 mmol/L; p = 0.001 for treatment effect in ANOVA). The remaining clinical-chemical variables were not influenced in a significant manner by any of the treatments (data not shown).

The most common adverse events (two or more reports) during treatment with atenolol included dyspepsia (two reports) and tiredness (two reports). During treatment with amlodipine, no adverse event was reported more than once. Patients treated with enalapril reported headache (four reports), cough (three reports), dyspepsia (two reports) and obstipation (two reports). During treatment with hydrochlothiazide and losartan, headache (two reports each) was the most common adverse event. Neither MSE profiles nor the extent of fatigue, daytime sleepiness, morning sleepiness, morning headache, nocturnal sweating or irritability (visual analogue scales) were significantly influenced by any of the treatments (data not shown).

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Our results suggest that in male patients with hypertension and OSA, 50 mg of the ₁-selective antagonist atenolol, given once daily, is more effective in lowering office DBP, our main outcome, than amlodipine, enalapril, hydrochlorothiazide, or losartan in their recommended maintenance dosages. Moreover, atenolol may have a stronger effect on nighttime ambulatory DBP and SBP.

The difference between atenolol and the remaining four treatment substances in reducing office DBP was not corroborated by our results of daytime ambulatory BP measurements. Other investigators have previously observed this discrepancy (19), and have suggested that office BP levels contain a larger alerting component than do ambulatory measurements (19). As a consequence, an antihypertensive agent with a more powerful effect on the alerting response in the office might surpass other substances in reducing office but not ambulatory BP. So far, however, we have no data to support such explanation for our findings.

Differences among the investigated substances in office but not daytime ambulatory BPs might also be expected to result from differences in their elimination rates, since office measurements were made when trough concentrations of the given substances were reached, whereas daytime ambulatory measurements reflect average values during the hours after drug intake. However, since atenolol (together with the active losartan metabolite E-3174) has the shortest plasma half-life (between 6 and 9 h) of the compounds studied, differences in pharmacokinetic properties are not a plausible explanation for the more potent effect of atenolol on trough DBP. Consequently, the prolonged effect of atenolol over the circadian cycle, which is corroborated by its stronger effect on nocturnal SBP and DBP, may rather be explained by its pharmacodynamic characteristics. However, because no patients without OSA were included in our study, we were not able to determine whether the extended effect half-life of atenolol is specifically augmented in patients with OSA.

The current study was designed to detect differences between treatment effects on BP in OSA patients, but did not provide sufficient statistical power to allow a comparison of drug effects on breathing during sleep in a confirmatory sense. Since only small differences were expected between treatments in their effect on the AHI (22, 23), we estimated that such an objective would have required more precise methods of nocturnal monitoring of ventilation, as well as the inclusion of many more patients. Therefore, the absence of significant changes in AHI and subjective symptoms, as well as the lack of reliable differences between the five treatments, may primarily reflect a lack of test power, although we expected clinically relevant changes in OSA severity during treatment to be detectable with our study design.

The treatment agents and dosages used in our study have previously been compared in hypertensive patients not selected according to an OSA criterion. The majority of the studies in which this was done concluded that these substances show similar effectiveness in lowering BP (15, 19, 24), and in particular found no superiority of atenolol over amlodipine, enalapril, hydrochlorothiazide, or losartan. In the current study, in contrast, only patients with OSA were selected from a population of hypertensive patients. Therefore, our findings suggest that atenolol may be more effective than the four reference substances in patients with OSA but not in unselected populations of hypertensive patients, where this distinction would be diluted. Nevertheless, varying dosages of the compounds in question will have to be compared in order to confirm a distinctive susceptibility of OSA-related hypertension to one specific substance. Moreover, even if an exceptional intrinsic activity could be assumed for one compound (e.g., atenolol), generalization to a whole pharmacodynamic principle (e.g., -blockade), would require consideration of several representatives of the same substance class. Moreover, if it is hypothesized that sympathicolytic substances are particularly effective in OSA, -adrenoceptor antagonists and compounds reducing sympathetic discharge may deserve attention in future investigations.

The first comparative study of antihypertensive agents for reduction of BP in OSA patients addressed the effects on continuously measured BP of metoprolol (100 mg once daily) and cilazapril (2.5 mg once daily), each given for 1 wk to groups of six patients with OSA and heart failure (New York Heart Association Class 1 or 2), using a parallel-groups design (30). Although the authors concluded that cilazapril had distinct advantages, no statistical analysis of between-group differences was provided. In a repeated trial involving two groups of 12 patients each, no significant difference was found in the reduction of daytime office BP with metoprolol and cilazapril (31). Kantola and coworkers (32), studying parallel groups of six OSA patients each, compared metoprolol (50 or 100 mg twice daily) with israpidine (1.25 or 2.5 mg twice daily), but found no reliable reduction in office BP in either group after 8 wk of treatment. Using a randomized crossover design with 8-wk treatment periods, the same investigators analyzed the effects of atenolol (50 mg), israpidine (2.5 mg), hydrochlorothiazide (25 mg), and spirapril (6 mg) on office BP in 18 hypertensive and predominantly obese OSA patients (33). No statistical analysis of contrasts between treatments was provided, but the figures presented indicated no difference in BP reduction. More recently, Kantola and coworkers repeated their study with the same design, but analyzed 24-h BP profiles of 15 obese hypertensive patients with sleep apnea (23). Without statistical analysis of between-treatment differences, atenolol was reported to reduce daytime ambulatory SBP and DBP most effectively, whereas none of the four treatments reduced BP to a considerable extent during the night. When summarized, the available data from previous studies comparing the effectiveness of antihypertensive agents in reducing BP in OSA patients show no statistically supported evidence for differences between the substances considered. That the current study was able to provide some evidence for a greater effectiveness of atenolol in such patients may have mainly been due to the use of a distinct study design, a larger study sample, and inclusion of patients with different characteristics (particularly a lower mean BMI) than in these other studies.

Repetitive activation of the sympathetic nervous system during the night, and increased sympathetic tone during the day, are important characteristics of OSA (9). It has been suggested that this sympathetic overactivity may be an important mechanism behind nocturnal, apnea-related increases in BP, as well as sustained daytime hypertension, in patients with sleep apnea (8). This hypothesis gained support from a study in rats which showed that hypertension induced by long-term intermittent hypoxia may be abolished by destruction of the sympathetic nervous system with 6-hydroxydopamine (34). If sustained overactivity of the sympathetic nervous system plays a dominant part in the physiology of hypertension in OSA, adrenoceptor blockade might be expected to be particularly effective for BP reduction in patients with this condition. The current study, indicating that ₁-blockade with atenolol might be more effective than other pharmacologic approaches in reducing office DBP as well as nighttime SBP and DBP in hypertensive patients with OSA, provides some evidence for this concept. It may therefore motivate more specific inquiry into the mechanisms of action and benefit of sympathicolytic treatment of hypertension in OSA.