Intraindividual Blood Pressure Responses to Angiotensin-Converting Enzyme Inhibition and Angiotensin Receptor Blockade
George S. Stergiou, MD, Hypertension Center, Third University Department of Medicine, Sotiria Hospital, 152 Mesogion Avenue, Athens 11527, Greece
E-mail: E-mail: gstergi@@med.uoa.gr
This study aims to test the hypothesis that in some hypertensive subjects the blood pressure (BP) response to angiotensin-converting enzyme inhibition differs from that to angiotensin receptor blockade (ARB); a responder to angiotensin-converting enzyme inhibition may not respond to ARB or the opposite. A randomized, open-label, crossover, comparative trial of lisinopril 20 mg compared with telmisartan 80 mg (5 weeks per treatment period) was conducted in 32 untreated hypertensives using 24-hour ambulatory BP monitoring. Subjects were classified as “responders” and “nonresponders” using an arbitrary threshold of ambulatory BP response (≥10 mm Hg systolic or ≥5 diastolic) or the median response achieved by each drug. No difference was detected between the drugs in their effect on ambulatory BP (mean difference 1.2±7.1/0.7±5.1 mm Hg, systolic/diastolic). Significant correlations were found between the antihypertensive responses to the two drugs (r=0.77, p<0.001). Using the arbitrary response criterion, there was a difference between the drugs in the responses in 28%/13% of subjects (9/4 patients) for systolic/diastolic BP (19%/25% using the median response criterion).
These data suggest that in some hypertensive patients the BP response to angiotensin-converting enzyme inhibition may fail to predict the response to ARB. It appears that there are differences in the antihypertensive action of angiotensin-converting enzyme inhibitors and ARBs that may be clinically important.
Recent guidelines by the European Society of Hypertension-European Society of Cardiology1 and by the United Statess' Seventh Report of the Joint National Committee on Prevention, Detection, Evaluation, and Treatment of High BP2 recommend the use of two different blockers of the renin-angiotensin system, the angiotensin-converting enzyme (ACE) inhibitors and the angiotensin II type 1 receptor (AT1R) blockers, as possible initial therapy for hypertension, together with thiazide-type diuretics, β blockers, and calcium antagonists.
At present, the only clinically important difference between the two drug classes in the management of hypertension is the absence of cough with the AT1R blockers as compared with an ACE inhibitor.1,3,4 There are, however, several theoretical differences between these two drug classes because they act on different sites of the renin-angiotensin system.3,4 Experimental evidence suggests that their mechanisms of antihypertensive action may differ in several respects. ACE inhibition (ACEI) is associated with a decrease in the breakdown of kinins (and thereby accumulation of bradykinin)5 and an increase in the levels of angiotensin 1-7,6 both of which may contribute to the antihypertensive effect of these drugs.3,4 On the other hand, in contrast to ACEI, which blocks both the AT1R and the angiotensin II type 2 receptor, AT1R blockade is associated with activation of the angiotensin II type 2 receptor, which may, via vasodilation, act synergistically with the AT1R blockade in reducing BP (BP).7,8 It is not clear, however, whether, from the clinical point of view, any of the above differences are associated with intraindividual variation in the antihypertensive effects of these two drug classes.
Two previous studies have suggested that, in a significant proportion of hypertensive subjects, the antihypertensive response to one of these drug classes fails to predict the response to the other and vice versa.9,10 Although these studies had the appropriate design, there were methodological problems, and, therefore, the validity of the findings is questionable. The objective of this work is to compare the intraindividual responses to ACEI and AT1R blockade using two effective and longacting drugs in a randomized crossover trial with 24-hour ambulatory BP (ABP) monitoring.
Study Design and Participants
This is a randomized, open-label, crossover comparative study of the ACE inhibitor lisinopril compared with the AT1R blocker telmisartan using clinical and ABP measurements. Untreated subjects with essential hypertension and a diastolic BP of 90-110 mm Hg on at least two clinic visits were assessed with 24-hour ABP monitoring. Subjects with an average awake diastolic ABP >85 mm Hg were randomized to receive open treatment with lisinopril at 20 mg o.d. or telmisartan at 80 mg o.d. for 5 weeks. Then patients were crossed over to the alternative treatment for a second 5-week period. Reasons for exclusion were electrocardiographic left ventricular hypertrophy; diabetes mellitus; known cardiovascular, renal, or liver disease; repeated clinical BP >200/115 mm Hg (systolic/diastolic) before or during the study; evidence of secondary hypertension; and an unwillingness to participate in the study. The protocol and the process for obtaining informed consent were approved by the Hospitals's Scientific Committee.
BP Measurements and Data Management
Clinic BP was measured before randomization and on the final day of each treatment period (at trough) by one physician (SE), who fulfilled the requirements of the British Hypertension Society for BP measurement.11 Triplicate measurements were taken at each visit after 5 minutes of sitting rest using a standard mercury sphygmomanometer (bladder size 12×22 cm, 15×35 cm or 18×38 cm where appropriate, Korotkoff phase V for diastolic BP). The average of the second and third clinical BP of each visit was used in the analysis.
ABP was monitored for 24 hours before randomization and on the final day of each randomized treatment period using noninvasive oscillometric devices SpaceLabs 90207 (SpaceLabs Inc., Redmond, WA) bladder size 12×23 cm or 14×30 cm where appropriate). The recorders were programmed to measure BP at 20-minute intervals for 24 hours and were always applied on a routine workday at the time of the morning dose. Participants were instructed to carry out their usual daily activities but to remain still, with the forearm extended, during each reading. They were also asked to keep a brief diary specifying the time when they were in and out of bed. The average awake, asleep, and 24-hour ABP was calculated using individual, patient-reported, sleeping hours. Recordings with <30 valid measurements during the awake period and <10 during the asleep period were rejected. BP readings flagged by the software of the monitors as technically erroneous and measurements with systolic BP <70 mm Hg or >260 mm Hg or with diastolic BP <40 mm Hg or >150 mm Hg (24% of the total readings) were discarded. Readings taken in the first 20 minutes after the monitors were attached to patients were also discarded as they were taken in the clinical environment. All clinic and ABP measurements in individual patients were performed on the same (nondominant) arm.
Comparison of BP Responses
The agreement in BP responses between the two drugs was assessed by the following criteria.
Arbitrary Response Criterion. Subjects were classified as responders and nonresponders according to the magnitude of the antihypertensive response to each treatment. Responders were defined as subjects with a ≥5 mm Hg decrease in 24-hour diastolic or ≥10 mm Hg in 24-hour systolic ABP. Agreement between the two drugs in the responders-nonresponders classification was regarded as the proportion of responders to both drugs plus the proportion of nonresponders to both drugs, whereas disagreement was considered as the proportion of subjects whose BP responded to only one of the drugs.
Median Response Criterion. To correct for any difference in the antihypertensive efficacy between the two drugs and to select the same proportion of responders and nonresponders in each of the two study groups, an alternative definition of respond-ers was applied using as a threshold for response the median BP response of all patients achieved by each of the drugs. Levels of agreement and disagreement in the BP response between the two drugs were calculated as they were done for the arbitrary response criterion.
Correlation Criterion. The antihypertensive responses to the two drugs were treated as continuous variables; the association between them was examined by performing linear correlations.
Statistical analysis was performed using the Statistical Package for the Social Sciences software (SPSS Inc., Chicago, IL, release 10.0). The possibility of a period effect or a treatment period interaction was tested using two-sample t tests.12 The differences in treatment-induced changes in clinical BP and ABP were analyzed. The treatment effects were compared by performing a one-sample t test on all within subject differences between telmisartan and lisinopril.12 Ninety-five percent confidence intervals (CIs) were calculated for each treatment-effect estimation. The Oldham correction (actual BP change divided by the average of pretreatment and posttreatment BP)13 was applied in the analysis of BP responses to each treatment, the association of which was investigated by Pearson correlations. A probability value p<0.05 was considered statistically significant.
A total of 36 untreated subjects with elevated clinic and awake ambulatory diastolic BPs were randomized. Four patients were withdrawn (three because of protocol violation and one because of a persistent dry cough on lisinopril). Thirty-two subjects with complete data were included in the analysis (26 men, mean age 49.5±7.1 [SD] years, body mass index 28.8±3.9 kg/m2, 13 smokers). A total of 84.1±7.8 readings were obtained with 24-hour ABP monitoring. Of the total, 20±8 were discarded (as they were considered erroneous according to the software criteria for the device or the study criteria); 3±8 of time points were not represented in the ABP profile because both the initial and the automatically repeated readings 2 minutes later were considered erroneous. There was no difference in the number of acceptable BP readings obtained with 24-hour ABP monitoring before randomization and on the final day of each treatment period. No evidence of a period effect or a treatment period interaction was observed for systolic and diastolic BP.
Data on the BP and pulse pressure lowering effects, trough:peak ratio, and smoothness index of telmisartan compared with lisinopril in this study have been published.14 Clinic and average 24-hour ABP at baseline were 149.1±10.6/99.3±6.5 mm Hg (systolic/diastolic) and 138.4±9.7/89.3±7.2 mmHg, respectively. Using both measurement techniques, no difference in antihypertensive efficacy was detected between the two drugs (average clinic BP difference −0.1±11.1 mm Hg, 95% CI, −4.1, 3.9/0.7±6.3 mm Hg, 95% CI, −1.6, 3.0 [systolic/ diastolic] and 24-hour ABP difference 1.2±7.1 mm Hg, 95% CI, −1.4, 3.8/0.7±5.1 mm Hg, 95% CI, 1.2, 2.5 [systolic/diastolic]).
An analysis of responders using the median response criterion showed disagreement in the antihypertensive response between ACEI and AT1R antagonism in six subjects (19%) for 24-hour systolic and in eight subjects (25%) for 24-hour diastolic ABP (Figure). Using the arbitrary response criterion, the corresponding proportions of disagreement were 28% (nine patients) and 13% (four patients). Subjects in whom there was disagreement in the response to the two drugs did not differ from those in whom there was agreement in terms of age, sex, and initial clinic and ABP before randomization. Strong correlations were found between the BP responses to lisinopril and telmisartan for both clinical (r=0.50/0.52 [systolic/diastolic], p<0.01) and 24-hour ABP (r=0.77/0.78, p<0.001).
This study suggests that a proportion of hypertensive patients may experience a good antihypertensive response to ACEI but not to AT1R blockade, or the opposite. Several studies comparing AT1R blockers with ACE inhibitors have shown similar antihypertensive efficacy with 50% of the subjects responding to each drug class3,4; however, these studies had a parallel group design, which does not allow for an intraindividual comparison of BP responses. Thus the similarity in the average antihypertensive response to these different agents for renin-angiotensin-system blockade does not preclude that the same subjects respond differently to treatment with the different drugs. A crossover design, where all patients are exposed to all treatments, is appropriate to address this issue.
Two randomized crossover studies that investigated the intraindividual BP responses to ACEI and to AT1R blockade in hypertensive patients have been reported.9,10 The first study compared the ACE inhibitor lisinopril (20 mg o.d.) with the AT1R blocker candesartan (16 mg o.d.) and showed disagreement in the antihypertensive responses between the drugs in 31% of patients.9 That study, however, assessed only clinic BP measurements, which are known to have poor reproducibility15 and to be affected by the placebo effect and the white-coat effect.16 The second study compared the ACE inhibitor lisinopril (20 mg o.d.) with the AT1R blocker losartan (50 mg o.d.) and again showed disagreement in the antihypertensive responses between the drugs in 36% of patients.10 There were significant differences in the antihypertensive efficacy of the comparator drugs, however, probably because of the low dose of losartan that was used in that study (which was the recommended dose for clinical use in hypertension when the study was designed). Thus the interpretation of the findings of this study is problematic.
The present study compared two of the more effective and long-acting agents among AT1R blockers and ACE inhibitors.17-23 Previous comparative trials have shown telmisartan to be as effective as other antihypertensive drugs, such as calcium antagonists,17β blockers,18 and ACE inhibitors.19 Moreover, telmisartan has been shown to be more effective than the AT1R blockers losartan20 and valsartan21 and as effective as a fixed-dose combination of medium doses of losartan with hydro-chlorothiazide.22 Lisinopril has also proved to be among the more effective antihypertensive drugs in comparative trials with other drug classes.23 The initial report from this study showed a similar antihypertensive efficacy, homogeneity (smoothness index), and duration of action (trough:peak ratio) of telmisartan and lisinopril.14 In fact, 95% CIs excluded any difference in antihypertensive efficacy of the two drugs >3.8/2.5 mm Hg (average 24-hour ABP).14 Despite the open design of the study, a selection bias was prevented by randomizing patients to the two treatment arms; a placebo effect was avoided by using ABP monitoring.16
Most of the previous studies have defined responders on the basis of an arbitrary BP decline, e.g., ≥10 mm Hg systolic or ≥5 mm Hg diastolic.24-28 In this study, in addition to using an arbitrary BP criterion for responders, the median BP response achieved by each of the drugs was applied as a threshold for response. This approach allows for the selection of the same proportion (50%) of responders to the two treatments. Consistent with the previous studies,9,10 both criteria for responders applied in this study confirmed the existence of disagreement in the antihypertensive responses between the two drugs in a proportion of hypertensive patients. These studies suggest that 20%-30% of hypertensives with a poor BP response to treatment with one of these drug classes will respond well to the other.
An alternative approach that avoids the use of arbitrary thresholds and the classification of patients into responders and nonresponders was performed. The BP responses to the two drugs were treated as continuous variables (after correcting for the BP level)13; the association between them was examined using linear correlations. The strong correlation found between the responses to the drugs in this study is not surprising given that the main mechanism of action of both drugs is regarded as the blockade of the renin-angiotensin system.3,4 Previous crossover studies that examined intraindividual responses to different antihypertensive drug classes showed weak correlations (correlation coefficients r=0.15−0.46) between drugs known to have different mechanisms of action (ACE inhibitors vs. calcium antagonists or diuretics, β blockers vs. calcium antagonists or diuretics).24-27 On the other hand, strong correlations (r=0.61−0.75) were found between the antihypertensive responses to drugs known to have common mechanisms of action (ACE inhibitors vs. β blockers, calcium antagonists vs. diuretics).24-26,28 In the present study, the association found between the responses to ACEI and AT1R antagonism was as strong as that reported between the responses of ACE inhibitors and β blockers or between calcium antagonists and diuretics. In other words, the similarity between the antihypertensive responses to ACEI and AT1R antagonism appears to be at the same level as it is between ACE inhibitors and β blockers or between calcium antagonists and diuretics.
When arbitrary thresholds are being used for the comparison of BP responses, a 1-2 mm Hg difference in response might switch patients from one category to the other, however, alternative analyses in the present study (median response analysis, awake and asleep ABP analysis, and correlation analysis) lead to the same conclusion in regard to the agreement in the responses between the two drugs. Taken together, the above findings suggest that the conclusion of the study in regard to the intraindividual responses to the two drugs is based on a true phenomenon. It might be argued that the intrinsic variability of ABP may be responsible, at least in part, for the intraindividual differences in the responses to the two drugs. To assess the impact of this factor on the study findings, we evaluated the proportion of subjects with a difference in 24-hour ABP achieved by the two drugs greater than the intrinsic variability of 24-hour ABP. For this purpose, the variability yielded by a previous study15 in which the average 24-hour ABP difference between two occasions was 0.5±8.3/0.3±5.6 mm Hg (systolic/diastolic) was used. Using as a threshold the variability + 1 SD (8.8/5.9 mm Hg for systolic/diastolic BP), the proportion of subjects with a difference in 24-hour ABP between the two drugs that exceeded the threshold was 15.6%/34.4% (for systolic/diastolic BP), corroborating the conclusions derived from our initial analyses.
This study suggests that 20%-30% of hypertensive patients may have a good BP response to ACEI but not to AT1R antagonism, or the opposite. These data imply that there are clinically important differences in the antihypertensive effects of these two drug classes. This view is also supported by data showing additive antihypertensive effects by combined ACEI and AT1R antagonism.29-31 This information should be taken into account when using ACE inhibitors and AT1R blockers in the management of hypertension in clinical practice.
Acknowledgment: The authors thank Stamatis P. Efstathiou for clinical research in preparing this manuscript.