Introduction

Central hemodynamic variables, including aortic pulse wave velocity (PWV), aortic augmentation index (AIx) and central blood pressure (BP) are independently associated with organ damage, incident cardiovascular disease and cardiovascular events in a variety of populations, including individuals with essential hypertension.1, 2, 3, 4, 5, 6, 7, 8, 9

When the aorta is stiff, PWV increases, accelerating the incident and reflected waves. Thus, the reflected wave merges with the incident wave in systole and augments AIx and aortic systolic pressure rather than diastolic pressure.7

A variety of cardiovascular risk factors can contribute to significant aortic stiffness.9, 10 In addition, BP-lowering drugs can have substantially different effects on central aortic pressures and hemodynamics despite similar impacts on brachial BP.1, 4, 11, 12, 13

Although BP lowering per se is the major determinant of cardiovascular event reduction, during the last decade, important multicenter trials have prompted the hypothesis that new antihypertensive drugs, such as renin–angiotensin system (RAS) blockers, may reduce cardiovascular outcomes beyond (peripheral) BP control.7

Both the Heart Outcomes Prevention Evaluation study14 and the Losartan Intervention For Endpoint Reduction in Hypertension (LIFE) study15 demonstrated that blockade of the RAS with angiotensin-converting enzyme inhibitors or angiotensin II receptor blockers results in benefits that are not explained by reduction in the brachial artery BP alone. The reason behind such outcomes may be that conventional brachial artery hemodynamic measurements underestimate the true central hemodynamic measurements that are directly related to the loads imposed on the heart and brain.7, 12

However, those results might be attributed to the comparator drugs used, such as the traditionally used beta (β)-blocker atenolol, which detrimentally affects myocardial contractility, vascular resistance, pressure wave reflection to augment the central systolic pressure wave, and carbohydrate and lipid metabolism.16 In contrast to classic β-blockers, newer agents with vasodilating properties, such as carvedilol and nebivolol, result in improved hemodynamic and metabolic profiles.16, 17

On the basis of the above considerations, we performed the present study to compare the effects of losartan and carvedilol on central hemodynamic indices in patients with mild-to-moderate hypertension. In addition, the effects on peripheral arterial pressure, metabolic profile and inflammatory parameters were also compared between the two treatments.

Methods

Study subjects

We included patients aged over 18 years with previously treated or untreated hypertension. Patients with the following conditions were excluded: (1) secondary hypertension; (2) myocardial infarction or stroke within the previous 6 months; (3) percutaneous coronary intervention or coronary artery bypass surgery within the previous 6 months; (4) angina pectoris requiring treatment with β-blockers or calcium antagonists; (5) a disorder that required treatment with losartan or another angiotensin II receptor blocker, carvedilol or another β-blocker, hydrochlorothiazide (HCTZ) or angiotensin-converting enzyme inhibitor, (6) atrial fibrillation; (7) contraindication for the use of losartan (bilateral renal artery stenosis, serum potassium >5.0 mmol l−1, serum creatinine >2.5 mg dl−1, severe aortic stenosis) or carvedilol (asthma, second- or third-degree heart block, sick sinus syndrome (in the absence of a permanent pacemaker), sinus bradycardia (<50 beats per minute; b.p.m.)); (8) a disorder other than hypertension requiring use of other agents affecting BP; or (9) concomitant continuing use of the following agents: (a) lithium, psychotropic drugs, antidepressants, anxiolytics or sleeping pills; (b) oral corticosteroid, mineralocorticoid or adrenocorticotropic hormone; (c) ephedrine, pseudoephedrine; or (d) non-steroidal anti-inflammatory drugs including COX-2 inhibitors. All subjects provided written informed consent and the institutional review board of each participating center, in accordance with the World Medical Association’s Declaration of Helsinki, approved the study protocol. This study was registered as NCT00496834 at ClinicalTrials.gov.

Study design

This was a 24-week multicenter, prospective, randomized, open, blinded end point study. Seven centers in Korea participated in the study. It was estimated that 78 patients per treatment group would provide 80% power at the 2.5% level (one-sided) to show noninferiority of losartan versus carvedilol regarding the primary efficacy variable, the difference in the carotid-femoral pulse wave velocity (cfPWV) change between the treatment groups. A predefined noninferiority margin was set at 1.5 m s−1 based on available historical data from a cfPWV18 and brachial PWV study19 and an assumed s.d. of 3.34 m s−1. The first objective of this study was to reject the hypothesis that losartan is inferior in efficacy to carvedilol (null hypothesis). The primary population used in this assessment was the modified intention-to-treat population. Losartan was considered noninferior to carvedilol regarding the primary end point if the entire 95% confidence interval of the difference in cfPWV change between the losartan and carvedilol groups was higher than −1.5 m s−1. As 20% of the randomized patients were expected to drop out during the study, we planned to recruit more than 196 patients.

After written informed consent was obtained, BP medications (if previously used) were stopped for a washout, run-in phase of 2 weeks, after which eligibility criteria were reconfirmed. At the end of the run-in period, patients with brachial systolic BP ⩾140 mm Hg or diastolic BP ⩾90 mm Hg (for diabetes patients, systolic BP ⩾130 mm Hg or diastolic BP ⩾80 mm Hg) were randomized to receive losartan (50 mg per day) or carvedilol (12.5 mg per day). Randomization was performed centrally using a validated system that automated the random assignment of treatment groups to randomization numbers. Patients were requested to take their study medication with water once in the morning and at approximately the same time each day.

If the brachial BP goal (systolic BP <140 mm Hg and diastolic BP <90 mm Hg; systolic BP <130 mm Hg and diastolic BP <80 mm Hg in diabetic patients) was not achieved at each follow-up visit (weeks 4, 8 and 16), losartan and carvedilol were up-titrated to 100 and 25 mg, respectively, and HCTZ was commonly added to each drug sequentially (12.5 and 25 mg). If the BP goal was successfully achieved, the medication regimen prescribed at the previous visit was maintained in both groups. However, if brachial systolic BP and diastolic BP exceeded the safety parameters of >180 and >110 mm Hg, respectively, at any point during the study, patients were withdrawn from the study and treated accordingly. Similarly, if systolic BP <100 mm Hg and diastolic BP <50 mm Hg, patients were withdrawn for safety reasons. Patients were reviewed at screening (week−2) and at randomization (week 0). Baseline assessments and measurements were performed at the randomization visit and at week 24. If the patients were taking antidiabetic or antidyslipidemic drugs at randomization, those drugs were maintained without changes in regimen or dose.

Measurements

All measurements and procedures were performed in the morning with the patients in a fasting state apart from the study medication, which was taken as a single-morning dose (2–4 h earlier), and having refrained from nicotine, alcohol and caffeine for at least the previous 10 h. Brachial and central BP parameters, including aortic AIx and cfPWV measurements, were taken as described previously.20, 21 Aside from baseline and study-end measurements, brachial BP was measured at each follow-up visit (weeks 4, 8 and 16), and central BP, cfPWV and AIx were measured at week 8.

Brachial BP measurement

After at least 10 min of rest, sitting systolic and diastolic brachial BP was recorded from the nondominant arm in an office twice on a 2-min basis using an automated sphygmomanometer (Omron Digital Blood Pressure Monitor HEM907, Bannockburn, IL, USA). The recorded BPs were averaged for use in the analysis. Patients with systolic pressure >140 mm Hg or diastolic pressure >90 mm Hg were defined as having hypertension.

Measurement of PWV

All procedures were conducted in a controlled environment by an experienced technician who was blinded to the patients’ clinical information. The cfPWV was determined using the PP-1000 semiautomatic aortic PWV analyzer (Hanbyul Meditech, Jeonju, Korea).20 Briefly, the left common carotid artery, radial artery, femoral artery and dorsal artery pressure waveforms were recorded noninvasively using a mechanotransducer. An electrocardiogram, phonocardiogram and four-channel pressure waveforms were simultaneously measured. The distance (D) traveled by the pulse wave was automatically obtained from the age-, gender- and height-based distance of the upper and lower extremity arteries of Koreans provided by the Korea Research Institute of Standards and Science. The pulse transit time (t) measured between the feet of the pressure waveforms that were recorded at two different recording points (the foot-to-foot method) was automatically determined. The cfPWV was automatically calculated as D/t. The Pearson’s correlation coefficient for intraobserver reproducibility was 0.99 (P<0.001). In the Bland–Altman plot of intraobserver measurements, the mean difference in the repeated measurements was −0.056±0.223 m s−1, and most of the values were within the mean±1.96 s.d.

Measurement of the pulse wave analysis

The AIx was calculated from the right radial artery pulse waves using applanation tonometry (Gaon21A System; Hanbyul Meditech).22, 23 Previous studies21, 23, 24, 25 showed that applanation tonometry provides a good indicator of ascending aortic pressure wave contour both in the control state and after acute therapeutic intervention, although the Gaon21A System lacks validation studies. Data that met the automatic quality controls specified by the integrated software were used to derive central aortic pressure waveforms by a generalized transfer function, from which aortic systolic BP and diastolic BP values were obtained. Brachial arterial BP was used to calibrate the radial pressure pulse. Aortic pressure waveforms were subjected to further analysis by integrated software to calculate the aortic AIx. The systolic part of the arterial waveform is characterized by two pressure peaks. The first systolic shoulder (inflection point) is caused by the left ventricular ejection, whereas the second peak is a result of the wave reflection. AIx is defined as the ratio of the augmented pressure peak (height of second peak minus height of the first systolic shoulder) to central pulse pressure; this parameter is expressed as a percentage and provides a quantitative measure of central BP augmentation.21 As AIx strongly depends on heart rate (HR), AIx corrected for a HR of 75 b.p.m. (AIx@HR75) was also obtained.26

Laboratory measurement

All blood samples were obtained in the morning in a fasting state, and the samples were immediately stored at −80 °C for subsequent assays. Laboratory data, except for serum high-sensitivity C-reactive protein and tumor necrosis factor-alpha, were analyzed in the central core lab of Korea University Guro Hospital. Serum total cholesterol, triglycerides, high- and low-density lipoprotein cholesterol, and uric acid were determined enzymatically using a model 747 chemistry analyzer (Hitachi, Tokyo, Japan). A glucose oxidase method was employed to measure the plasma glucose, and an electrochemiluminescence immunoassay (Roche Diagnostics, Indianapolis, IN, USA) was used to measure the insulin levels. Insulin resistance was calculated by the homeostasis model assessment.27 Glycated hemoglobin was measured by high performance liquid chromatography using a Varian II apparatus (Bio-Rad, Hercules, CA, USA). Serum high-sensitivity C-reactive protein and tumor necrosis factor-alpha were measured using an enzyme-linked immunosorbent assay (Samkwang Medical Laboratories, Seoul, Korea).

Statistical analyses

Data are summarized as the mean±s.d. for continuous variables and frequency (percent) for categorical variables. Differences in characteristics between groups were compared using Student’s t-test and Pearson’s χ2-test. Differences between the baseline and post-treatment values were analyzed using the paired t-test. The differences in cfPWV changes from baseline between the treatment groups were further analyzed with analysis of covariance, in which the possible confounding effects, including the baseline value, change in HR, BP changes, use of HCTZ, age and gender, were included in the model as covariates.

A P-value<0.05 was considered statistically significant. All statistical results were based on two-sided tests. The data were analyzed using SPSS (Statistical Package for the Social Sciences) for Windows (version 12.0; SPSS, Chicago, IL, USA).

Results

Subject characteristics

A total of 210 patients were screened, and 201 were eligible to enter the study. These patients were randomized to the losartan (n=101) and carvedilol (n=100) groups. One hundred and eighty-two patients (90.5%; losartan group=88 and carvedilol group=94) were included in the modified intention-to-treat population, defined as all randomized patients who provided baseline and ⩾1 follow-up data, even if they did not complete the entire study (Supplementary Figure S1). The study subjects consisted of 108 men (59.3%) and 74 women (40.7%). The mean age was 49.6±9.9 years. The baseline characteristics of the study subjects are shown in Tables 1 and 2. There were no significant baseline differences in anthropometric, clinical or biochemical variables.

Table 1 Baseline characteristics of the study subjects
Full size table
Table 2 Changes of laboratory data from baseline to week 24 in the losartan and carvedilol groups
Full size table

The mean daily doses were 81.8±24.2 mg for losartan and 21.8±5.5 mg for carvedilol. The frequency of additional HCTZ use (39.8% for the losartan group, 42.6% for the carvedilol group, P=0.7033) and mean HCTZ doses among the users (17.1±6.1 mg for losartan group, 17.8±6.3 mg for carvedilol group, P=0.6420) did not differ between groups.

Brachial BP and pulse rate

From the 4th week, systolic and diastolic BP were significantly reduced (both P<0.0001) in both groups (Figure 1). At the end of the study, the systolic/diastolic BPs were 135.9±11.9/87.7±8.8 mm Hg in the losartan group and 137.7±13.8/87.5±10.8 mm Hg in the carvedilol group. The mean reductions in systolic/diastolic BPs were 15.1±14.6/8.4±9.3 mm Hg for losartan and 14.8±15.4/7.8±9.6 mm Hg for carvedilol, which were not significantly different between the two groups. Although the 24-week treatment did not change the pulse rate in the losartan group (71.3±8.7 b.p.m. at week 24), the pulse rate was significantly reduced by 6.4±8.6 b.p.m. (P<0.0001) in the carvedilol group (65.0±8.5 b.p.m. at week 24, P<0.0001 versus losartan group).

Figure 1
figure 1

Changes in systolic (a) and diastolic (b) brachial blood pressure (BP) before and after antihypertensive treatment in the losartan and carvedilol groups. P-values of all follow-up BPs were <0.0001 compared with baseline.

Full size image

Central hemodynamic variables

After 24 weeks of treatment, the mean decline in central systolic BP (losartan versus carvedilol, −14.3±15.2 (from 140.7±12.0 to 126.7±14.4 mm Hg) versus −13.6±16.3 mm Hg (from 142.0±12.4 to 128.9±14.8 mm Hg)) and central diastolic BP (−7.6±10.0 (from 96.5±10.3 to 88.9±9.7 mm Hg) versus −9.0±11.5 mm Hg (from 96.1±10.7 to 87.6±11.1 mm Hg)) were similar between the groups (Figure 2a). The mean decline in central pulse pressure (−6.7±10.2 mm Hg (from 44.1±10.3 to 37.6±9.5 mm Hg) in losartan versus −4.6±11.5 mm Hg (from 45.9±12.2 to 41.2±10.5 mm Hg) in carvedilol, P=0.2216) was not different.

Figure 2
figure 2

Changes in aortic blood pressure (BP; a) and carotid-femoral pulse wave velocity (cfPWV) (b) from baseline to 24 weeks after treatment in the losartan and carvedilol groups. *P-value for the comparison of the changes from baseline to 24 weeks between the losartan and carvedilol groups.

Full size image

There was a significant increase in cfPWV (0.28±1.29 m s−1 (from 7.52±1.30 to 7.80±0.90 m s−1), P=0.042) in the losartan group, whereas there was a nonsignificant decrease in cfPWV (−0.12±1.55 m s−1 (from 7.68±1.42 to 7.56±1.05 m s−1)) in the carvedilol group. However, the difference in the cfPWV changes from baseline to the study end point between carvedilol and losartan was not significant (−0.41±1.43 m s−1, 95% confidence interval (−0.8–0.01), P=0.0565; Figure 2b) even after adjustment for the baseline value, change in HR, changes in BPs, use of HCTZ, age and gender (Table 3).There was a significant reduction in AIx (−5.0±16.3% (from 26.2±10.9 to 21.5±17.1%), P=0.009) in the losartan group, whereas there was a nonsignificant increase in AIx (0.18±12.0% (from 29.1±14.4 to 30.0±12.2%)) in the carvedilol group. The difference in the AIx changes was significant (P=0.024) between the groups (Figure 3a). In addition, there was a significant decrease in AIx corrected for a HR of 75 b.p.m. (AIx@HR75) by 5.3±15.8% (from 22.8±10.2 to 17.9±15.9%, P=0.005) in the losartan group, whereas there was a nonsignificant reduction in AIx@HR75 by 1.25±12.3% (from 25.4±14.2 to 24.8±11.5%) in the carvedilol group. However, we found no difference in AIx@HR75 changes between the groups (Figure 3b).

Table 3 Comparison of the changes of cfPWV between two groups using ANCOVAa with each covariate
Full size table
Figure 3
figure 3

Changes in augmentation index (Aix; a) and AIx@HR75 (b) from baseline to 24 weeks after treatment in the losartan and carvedilol groups. *P-value for the comparison of the changes from baseline to 24 weeks between the losartan and carvedilol groups.

Full size image

Metabolic profile and inflammatory parameters

Losartan and carvedilol administration for 24 weeks did not result in any significant changes in glycemic profile, and there were no between-group differences in the glycemic profile changes. Losartan showed a neutral effect on the lipid profile, whereas carvedilol significantly increased triglyceride levels and decreased high-density lipoprotein cholesterol levels (Table 2). However, no between-group differences in lipid profiles were evident. Losartan demonstrated a beneficial effect on uric acid, but carvedilol displayed a negative effect on uric acid. The difference in uric acid changes was significant between the groups (P=0.001). Both treatments showed a neutral effect on high-sensitivity C-reactive protein and tumor necrosis factor-alpha, and those changes were not significantly different (Table 2).

Safety data

There were no deaths during the study, and the adverse event profile was similar between the groups. There were a total of 11 serious adverse events (five in the losartan group and six in the carvedilol group), of which two in the losartan group (traumatic hematoma resulting in hospitalization and new development of variant angina) and one in the carvedilol group (back pain due to acute transverse myelitis resulting in hospitalization) resulted in withdrawal. The other reported serious adverse events, which did not result in discontinuation, were in the losartan group (car accident, overdose, oral cavity bleeding; one case each) and the carvedilol group (chronic sinusitis and five cases of overdose). Three patients in the losartan group (one headache, one abdominal discomfort and one chest discomfort) and two in the carvedilol group (one facial edema and one nausea/dizziness) had nonserious adverse events that led to withdrawal. There were no reported cases of serious laboratory adverse events.

Discussion

We compared the effects on central hemodynamic indices, peripheral arterial pressure and metabolic profile as well as inflammatory parameters between losartan and carvedilol in a 24-week treatment trial in patients with mild-to-moderate hypertension. There were no between-group differences in central BP, brachial BP, cfPWV or AIx@HR75 changes, but losartan showed a more favorable effect than carvedilol for AIx. The changes in all measured metabolic and inflammatory parameters were not significantly different between the two groups, except for uric acid.

It has been shown that the most reliable measure of arterial stiffness is aortic PWV (cfPWV),5, 9, 28 which independently predicts outcomes in a variety of populations, including individuals with essential hypertension.2, 9, 28 In contrast to most previous studies, there was no significant decrease in cfPWV even after similar significant BP reductions with both treatments in the present study. However, we previously demonstrated that the cfPWV change with antihypertensive drugs is largely affected by the baseline cfPWV.29 The baseline PWV was also the strongest predictor of PWV regression with treatment in the recent EXPLOR trial.30 Protogerou et al.31 reported similar results as well. In their post hoc analysis of the REASON (Preterax in Regression of Arterial Stiffness in a Controlled Double-Blind) study, the latter authors divided 375 patients into three tertiles according to the baseline cfPWV. The cfPWV decline (−2.2 m s−1) after a 12-month treatment was significantly greater in the highest tertile (mean (s.e.), 15.1 m s−1 (13.1–24.1)), whereas the cfPWV did not decrease, and was even increased by 0.2 m s−1 in the lowest tertile (mean (s.e.), 9.3 m s−1 (5.5–10.65)) even after BP reduction.31 The baseline cfPWV of the lowest tertile in that study was within the reference range (6–11 m s−1)5 even after considering age and BP categories.32 The baseline cfPWV of our study subjects (7.52±1.30 m s−1 for the losartan group, 7.68±1.42 m s−1 for the carvedilol group) was also within the reference range5, 32 and was lower than that of the lowest tertile reported previously.31 A recent study in which the study subjects’ baseline cfPWV was similar to that of ours showed no significant decrease in cfPWV even after 1 year of antihypertensive treatment as well.33

The collective results suggest that BP reduction with antihypertensive drugs might have minimal effects on cfPWV when the baseline cfPWV is not very high, as in our study. A baseline cfPWV within the normal range even when hypertension is present suggests that the aorta continues to function normally to tone down the PWV using its potential elasticity to buffer stroke from the heart. Therefore, ‘physiologically normal’ baseline cfPWV under control of the aorta, we believe, would not decrease enough to be ‘pathologically low’, although high BP decreased to normal with treatment. The aorta’s burden could be alleviated and the aorta would extend less than before to sustain physiologic PWV. Moreover, our results can also be explained by the relatively young age of the study subjects when we consider changes in cfPWV are more marked in older individuals (>50 years) in contrast to changes in AIx (more marked in younger individuals, <50 years).5

AIx was significantly decreased only in the losartan group and the difference in AIx changes between the treatments was significant. AIx is influenced not only by large artery stiffness but also by the magnitude of the wave reflected at peripheral sites, which is affected by the vascular tone of peripheral muscular arteries.9, 28, 34 A recent study35 separately examined the relationship between two major AIx components, including the first systolic shoulder in the arterial pulse waveform (P1), and augmentation pressure (ΔPaug), or the height of central systolic pressure above P1, and PWV. P1, the outgoing component of the pressure wave, was highly correlated with PWV, whereas ΔPaug did not independently correlate with cfPWV. These findings suggest that P1 is determined by PWV but PWV is not a major determinant of ΔPaug. These observations are consistent with other studies,12, 19, 33, 36, 37 including our results showing dissociation between AIx and PWV during interventions that influence vasomotor tone. Thus, the peripheral vasodilatation induced by antihypertensive medications may affect AIx by reducing the magnitude of wave reflections, independent of the medications’ effects on aortic PWV.36

In the carvedilol group, however, the medication’s potentially beneficial vasodilating property could be counterbalanced to some extent by reduced HR. Reducing HR prolongs cardiac ejection duration, which causes the peak of the forward wave to appear later and AIx to be higher, but reduced HR has no major effect on pulse velocity.1, 4, 34, 38

As AIx strongly depends on HR, AIx@HR75 is also used.5 In our study, AIx@75 was also significantly decreased only in the losartan group, but the difference in AIx@75 changes with either treatment was not significant. Our results agree with a study in which losartan showed a more favorable effect on AIx than atenolol, but the beneficial effect was no longer significant when HR was controlled.19 It is a debatable issue whether HR should be controlled when assessing AIx,39, 40 as some believe that AIx itself is important regardless of the HR.19

Central hemodynamic variables are now well accepted, not only as surrogate markers for subclinical cardiovascular disease but also as intermediate end points for cardiovascular events. Therefore, our study showing comparable effects between losartan and carvedilol on the central hemodynamics except for AIx suggests the possibility that clinical outcomes might also be similar between the two treatments. In addition, there were no between-group differences in inflammatory parameters or the metabolic profile, except for uric acid, in the present study.

However, several issues need to be discussed. First, our study subjects were relatively young (49.6±9.9 years), low-risk (diabetes 4%, dyslipidemia 16%) patients with mild-to-moderate hypertension without significant aortic stiffness. Thus, our findings might not be applicable to older, high-risk hypertensive patients. Second, because the de-stiffening process, such as regression of arterial hypertrophy, occurs in hypertensive patients after 1 year of treatment with RAS blockers,37 a 24-week treatment may not be sufficient to make a difference between the two treatments. Third, although there were no between-group differences in triglyceride and high-density lipoprotein cholesterol changes, both of these parameters became significantly worse in the carvedilol group. If the study had been maintained longer, those negative effects of carvedilol on the lipid profile could have affected the clinical outcome. Fourth, the population size in the present study was calculated based on the hypothesis that losartan is not inferior to carvedilol regarding cfPWV. Therefore, the priority of other study variables was not as high as that of cfPWV. Fifth, there has been no large clinical trial evaluating differences between losartan and carvedilol in cfPWV changes as a primary end point. Therefore, this study, a pilot project, was performed more conservatively with a wide noninferiority margin and a small sample size, although we referred to other studies.18, 19 In this context, we believe the results might be underpowered to demonstrate noninferiority and must be interpreted with caution. Finally, this study was an open-label study, which has innate limitations in some between-group comparisons.

This is the first study to directly compare the effectiveness of losartan, a prototype angiotensin II receptor blocker, and carvedilol, a newer vasodilating noncardioselective β-blocker, for central hemodynamic indices in patients with mild-to-moderate hypertension. In this 24-week, prospective, randomized, open, blinded end point study, losartan and carvedilol showed generally comparable effects on central hemodynamic indices, metabolic profile, inflammatory parameters and peripheral arterial pressure when they were used once daily.