OUP user menu

The consistency of the treatment effect of an ACE-inhibitor based treatment regimen in patients with vascular disease or high risk of vascular disease: a combined analysis of individual data of ADVANCE, EUROPA, and PROGRESS trials

Jasper J. Brugts, Toshiharu Ninomiya, Eric Boersma, Willem J. Remme, Michel Bertrand, Roberto Ferrari, Kim Fox, Stephen MacMahon, John Chalmers, Maarten L. Simoons
DOI: http://dx.doi.org/10.1093/eurheartj/ehp103 1385-1394 First published online: 4 April 2009

Abstract

Aims Angiotensin-converting enzyme (ACE) inhibitors have been shown to reduce cardiovascular risk in different groups of patients. Whether these effects can be generalized to the broad group of patients with vascular disease is unknown. Therefore, we undertook a combined analysis using individual data from ADVANCE, EUROPA, and PROGRESS to determine the consistency of the treatment effect of perindopril-based regimen in patients with vascular disease or at high risk of vascular disease.

Methods and results We studied all-cause mortality and major cardiovascular outcomes during a follow-up of about 4 years in the 29 463 patients randomly assigned a perindopril-based treatment regimen or placebo. The perindopril-based regimens were associated with a significant reduction in all-cause mortality [hazard ratio (HR) 0.89; 95% confidence interval (CI) 0.82–0.96; P = 0.006], cardiovascular mortality (HR 0.85; 95% CI 0.76–0.95; P = 0.004), non-fatal myocardial infarction (HR 0.80; 95% CI 0.71–0.90; P < 0.001), stroke (HR 0.82; 95% CI 0.74–0.92; P = 0.002), and heart failure (HR 0.84; 95% CI 0.72–0.96; P = 0.015). Results were consistent in subgroups with different clinical characteristics, concomitant medication use, and across all strata of baseline blood pressure.

Conclusion This study provides strong evidence for a consistent cardiovascular protection with an ACE-inhibitor treatment regimen (perindopril–indapamide) by improving survival and reducing the risk of major cardiovascular events across a broad spectrum of patients with vascular disease.

Keywords
  • Perindopril
  • Coronary artery disease
  • Diabetes
  • Stroke
  • Vascular disease
  • Prevention
See page 1307 for the editorial comment on this article (doi:10.1093/eurheartj/ehp169)

Introduction

Clinical trials have demonstrated the efficacy of angiotensin-converting enzyme (ACE)-inhibitors in specific groups of patients at risk of cardiovascular events.19 Nowadays, the use of ACE-inhibitors is recommended in guidelines on the management of hypertension, stable coronary artery disease (CAD), myocardial infarction (MI), and heart failure.1015 The beneficial effect of ACE-inhibitors is related, at least in part, to the blood pressure (BP)-lowering effects. However, ACE-inhibitors may have many valuable protective properties quite apart from BP lowering. Especially, perindopril has been extensively investigated in this regard and has been shown to improve endothelial function and neurohumoral balance, and inhibit remodelling of the coronary arteries.1619 Since, the BPLTTC has demonstrated that BP-independent effects of ACE-inhibitor-based regimens do contribute to the reduction in CAD,20,21 it becomes more important to base decisions on the treatment of individual patients, on the assessment of total cardiovascular risk, rather than on arbitrary cut-off points for single risk factors such as BP. Most of the evidence currently available has been obtained from studies conducted in patient populations with vascular disease in a single vascular territory or with a metabolic disorder such as diabetes. There is, therefore, a need for a broadly based study pooling individual patient data from patient populations with a wider spectrum of vascular disease. For this purpose, we undertook a combined analysis of the individual data from trials of perindopril-based regimens in patients with diabetes, CAD, and cerebrovascular disease.79 We investigated the treatment effect on clinical endpoints and their consistency in patient subgroups.

Methods

The methodological principles that lie behind a combined analysis of randomized clinical trials based on data from individual patients have been described in detail.22 We therefore only briefly describe the applied methods of trial selection, data-management, endpoint definitions, and statistical analysis.

Trial selection

We obtained data from the ADVANCE, EUROPA, and PROGRESS studies that are the three main large trials with a regimen based primarily on the ACE-inhibitor perindopril. The ASCOT-BPLA was not selected because the combined treatment regimen was based on amlodipine with the addition of perindopril so that it was impossible to make unbiased estimates for the treatment effect of perindopril.23 Since all trials studied a regimen based on the same agent, perindopril, we had the opportunity to include individual data in this combined analysis, which made important subgroup analyses possible at the patient level. The types of patients included in these studies were different in their primary diagnoses, but since atherosclerosis and vascular disease is not restricted to a single vascular bed, we conclude that these patients are at least homogenous in having vascular disease or being at a high risk of vascular disease. The combined data set consisted of 29 463 patients, who were followed for on average 4 years. We had full access to all individual data of the trials. After data-merging, data were tested carefully for completeness, internal consistency of patients' records, and consistency with the published reports. Table 1 shows the main inclusion and exclusion criteria, outcomes, recruitment, and follow-up of the three trials.

View this table:
Table 1

Main features of the ADVANCE, EUROPA, and PROGRESS trials

ADVANCE (n = 11 140)EUROPA (n = 12 218)PROGRESS (n = 6105)
Main inclusion criteria
 Age at entry (years)≥55≥18Not specified
 Type of patients (entry)Diabetes mellitusCoronary artery diseaseStroke or TIA
Main exclusion criteria
 Known congestive heart failureNoNoNo
 No. of patients with LVEF <40%Not assessed18/7096 assessedNot assessed
 Myocardial infarctionWithin 3 months
 StrokeWithin 5 years
 Percutaneous coronary interventionWithin 6 months
 Coronary artery bypass graftWithin 6 months
ACE-inhibitor and target daily dosePerindopril 4 mg/indapamide 1.25 mgPerindopril 8 mgPerindopril 4 mg/indapamide 2.5 mg (except Japan 2 mg)
Main outcomes
 Primary (composite)Major macro- or micro-vascular eventsCardiac death, MI, or CA with successful resuscitationFatal or non-fatal stroke
 Secondary and othersAll-cause mortality, cardiovascular mortality, non-fatal MI, fatal- and non-fatal stroke, revascularization, heart failure admissions, new or worsening nephropathyAll-cause mortality, cardiovascular mortality, non fatal MI, fatal- and non-fatal stroke, UAP, revascularization, heart failure admissionsAll-cause mortality, cardiovascular mortality, non-fatal MI, fatal- and non-fatal stroke, revascularization, heart failure admissions
Recruitment
 Recruitment periodJune 2001 to March 2003October 1997 to June 2000May 1995 to November 1997
 Mean follow-up duration4.3 years4.2 years3.9 years
  • MI, myocardial infarction; UAP, unstable angina pectoris; LVEF, left ventricular ejection fraction; TIA, transient ischaemic attack.

Outcomes

In this analysis, we assessed all-cause mortality, cardiovascular mortality, fatal and non-fatal stroke, revascularization, non-fatal MI, hospital admission for heart failure, and major cardiovascular endpoints as a composite of cardiovascular mortality, MI, and stroke. The definitions of endpoints in these trials were carefully checked and were not essentially different across the trials. Where there were slight differences in endpoints or thresholds, we did not try to match everything, since that this is often impossible retrospectively. Moreover, this is not necessary since in each study the treatment effects were compared within rigorously randomized and well-balanced treatment groups. Since we know that heterogeneity in an endpoint definition will not lead to invalid results, we applied the trial-specific definition of MI for practical reasons.24 Several endpoints were not presented in the main papers of the ADVANCE and PROGRESS trials, for example, the data for revascularization. For a comprehensive and comparable analysis, these endpoints were included.

Subgroups

We tested the consistency of the treatment effect of perindopril-based regimens in relation to baseline clinical characteristics [gender, age-categories (<60; 60–70; >70 years), mean age (<63; >63 years) hypertension, diabetes, cerebrovascular accident (CVA)/transient ischaemic attack (TIA), and revascularization] and in relation to concomitant medication (anti-platelet agents, lipid-lowering agents, β-blockers, diuretics, and calcium antagonists). We investigated the treatment effect of perindopril-based regimen on the outcomes according to BP levels at baseline (first screening visit). Ordinal categories of systolic BP and diastolic BP were defined as follows: systolic BP <120, 120–139, 140–159, and ≥160 mmHg; diastolic BP <80, 80–90, 90–100, and ≥100 mmHg. Further, we investigated whether the treatment effect was independent of the level of BP-reduction by active treatment. All patients were treated during run-in period of 4 weeks. Blood pressure reduction was calculated as the difference between the baseline measurement and the end of the run-in period. We accounted for the difference in dosage of perindopril (EUROPA 4–8 mg, ADVANCE and PROGRESS 2–4 mg) and combination with indapamide (ADVANCE and PROGRESS 2.5 mg) in all analyses. We emphasize that for analyses of BP reduction during run-in, baseline BP was added as continuous variable in the model to adjust for the starting level of BP before initiation of ACE-inhibitor treatment (avoids regression to the mean).

Statistical analysis

The statistical principles used in these analyses have been described previously in the main papers of the trials for the assessment of reduction in outcomes. Although baseline characteristics are well balanced between the randomized groups for each trial, we choose to perform a multivariate Cox regression analysis because of the difference between the studies themselves and because adjustment for baseline characteristics, even when randomized, is recommended in clinical trials.25 By multivariate analysis, confidence intervals (CIs) will widen slightly (although compensated by the large number of patients) but the validity will increase which is most important. In multivariate Cox regression analysis, we adjusted for age, gender, hypertension, diabetes, smoking, prior MI, prior percutaneous coronary intervention/coronary artery bypass grafting, prior CVA/TIA, medication use (anti-platelet agents, β-blockers, diuretics, calcium-antagonists, and lipid-lowering agents), indapamide use, and active treatment (perindopril) dosage. We further adjusted for differences in baseline risk across trials by adding dummy study variables to the model. Tests for heterogeneity in the treatment effects were performed by including interaction terms [treatment * characteristic] in the multivariate model for each covariate separately. Test for heterogeneity among trials was by including interaction terms of dummy study variables with treatment, and we calculated the difference in −2 log likelihood of two models. The difference χ2 (Q-test) between the two −2 log likelihoods follows a χ2 distribution. If P < 0.05, then there is evidence of heterogeneity between trials. Survival analysis was performed by the Kaplan–Meier analysis. Hazard ratios (HR) and 95% CIs are presented with corresponding two-sided P-values. A P-value <0.05 was considered significant. In all trials, analysis was by intention-to-treat principle.

Results

The baseline characteristics of the total study population (n = 29 463) are summarized in Table 2. The mean (SD) age was 63.0 (8.8) years, 28.4% were female, 54.1% hypertensives, 45.5% diabetics, and 32.8% experienced a previous MI. Mean BP was 142/82 mmHg, 70.8% were taking anti-platelet agents, 38.8% β-blockers, 39.5% lipid-lowering agents, and 33.3% calcium antagonists. Of the 29 463 patients, 14 730 (50.0%) were randomized to active treatment (perindopril-based regimen) and 14 733 (50.0%) received matched placebo. Indapamide was used in 14 684 patients (49.8%), and other diuretics in 15.1%.

View this table:
Table 2

Baseline characteristics patients in the ADVANCE, EUROPA, and PROGRESS trials

ADVANCE (n = 11 140)EUROPA (n = 12 218)PROGRESS (n = 6105)Total study population (n = 29 463)
Characteristics
 Age (years), mean (SD)66 (6)60 (9)64 (10)63 (9)
 Female (%)42.514.630.328.4
 Previous MI (%)12.064.87.032.8
 Previous PCI/CABG (%)8.554.92.726.6
 Previous CVA/TIA (%)12.93.499.9a27.0
 Previous PAD (%)2.47.34.14.8
 Current smokers (%)15.115.220.016.2
 Diabetes (%)100.012.312.545.5
 Hypertension (%)68.727.147.854.1
 Hypercholesterolaemia (%)58.963.361.2b
 Systolic blood pressure (mmHg)145 (21)137 (15)147 (9)142 (19)
 Diastolic blood pressure (mmHg)81 (11)82 (8)86 (11)82 (10)
Medications
 Anti-platelet agents (%)46.792.372.370.8
 β-Blockers (%)24.561.717.038.8
 Lipid-lowering agents (%)35.355.914.139.5
 Calcium antagonists (%)30.831.439.933.3
 Diuretics use (%)c9.223.711.515.1
  • Summary statistics for continuous variables are presented as mean (SD). Categorical data are summarized as percentages. The definition of hypertension was unified in all trials, with EUROPA as template, as a blood pressure of 160/95 mmHg or use of antihypertensives. MI, myocardial infarction; PCI, percutaneous coronary intervention; CABG, coronary artery bypass grafting; CVA, cerebrovascular disease; TIA, transient ischaemic attack; PAD, peripheral arterial disease.

  • aProtocol violation for 7/6105 patients.

  • bHypercholesterolaemia data were not present in PROGRESS, percentage was 61.2% (14 294) based on 23 358 patients (ADVANCE and EUROPA).

  • cDiuretics use, not indapamide study medication.

Clinical endpoints

The perindopril-based regimen was associated with a significantly lower cumulative incidence of all-cause mortality as well as major cardiovascular events when compared with placebo in the Kaplan–Meier analysis (both log ranks P-value <0.001) as presented in Figure 1. During a mean follow-up of 4.0 years (SD 0.8), there were 1089 deaths (7.4%) in the active treatment group and 1210 (8.2%) in the placebo group (HR 0.89; 95% CI 0.82–0.96; P-value = 0.006). No heterogeneity between the studies was observed (P for interaction 0.56). Results for the other endpoints were comparable and are presented in Table 3. Significant heterogeneity in the treatment effect between the studies was observed for heart failure admission (no effect in ADVANCE) and for fatal and non-fatal stroke (no effect in ADVANCE or EUROPA). The perindopril-based regimen did not significantly affect subsequent revascularizations (HR 0.92; 95% CI 0.84–1.01; P-value = 0.092), with 852 revascularizations (5.8%) in treatment group and 920 revascularizations (6.2%) in placebo group.

Figure 1

Kaplan–Meier analysis of cumulative incidence of all-cause mortality and major cardiovascular events (cardiovascular mortality, myocardial infarction, and stroke) in 29 463 patients. The x-axis corresponds to the duration of follow-up in years. The y-axis to the cumulative incidence of all-cause mortality (A) and major cardiovascular events in percentages (%) (B). The percentages along the lines correspond to the Kaplan–Meier estimates at 1, 2, 3, and 4 years of follow-up in patients allocated the perindopril-based regimen (red line) or placebo (black line). Below the graph, the number of events and the number of patients at risk during follow-up (per year) are presented.

View this table:
Table 3

Data for total mortality and cardiovascular events

Active treatment based on perindopril events/patients (%)Placebo events/patients (%)HR (95% CI)P-valueP-value for interaction
All-cause mortality
 ADVANCE (n = 11 140)408/5569 (7.3)471/5571 (8.5)0.86 (0.75–0.98)0.0260.56
 EUROPA (n = 12 218)375/6110 (6.1)420/6108 (6.9)0.89 (0.77–1.02)0.098
 PROGRESS (n = 6105)306/3051 (10.0)319/3054 (10.4)0.96 (0.82–1.12)0.596
 Total (n = 29 463)1089/14 730 (7.4)1210/14 733 (8.2)0.89 (0.82–0.96)0.006
Cardiovascular mortality
 ADVANCE (n = 11 140)211/5569 (3.8)257/5571 (4.6)0.82 (0.68–0.98)0.0280.58
 EUROPA (n = 12 218)215/6110 (3.5)249/6108 (4.1)0.86 (0.72–1.03)0.001
 PROGRESS (n = 6105)181/3051 (5.9)198/3054 (6.5)0.91 (0.75–1.12)0.380
 Total (n = 29 463)607/14 730 (4.1)704/14 733 (4.8)0.85 (0.76–0.95)0.004
Cardiovascular mortality, myocardial infarction
 ADVANCE (n = 11 140)320/5569 (5.7)370/5571 (6.6)0.86 (0.74–1.00)0.0450.82
 EUROPA (n = 12 218)484/6110 (7.9)596/6108 (9.8)0.80 (0.71–0.91)<0.001
 PROGRESS (n = 6105)230/3051 (7.5)279/3054 (9.1)0.82 (0.69–0.96)0.025
 Total (n = 29 463)1034/14 730 (7.0)1245/14 733 (8.5)0.82 (0.76–0.88)<0.001
Cardiovascular mortality, myocardial infarction, stroke
 ADVANCE (n = 11 140)480/5569 (8.6)520/5571 (9.3)0.91 (0.80–1.03)0.1350.10
 EUROPA (n = 12 218)552/6110 (9.0)664/6108 (10.9)0.83 (0.74–0.92)0.001
 PROGRESS (n = 6105)458/3051 (15.0)604/3054 (19.8)0.74 (0.66–0.84)<0.001
 Total (n = 29 463)1490/14730 (10.1)1788/14 733 (12.1)0.82 (0.76–0.87)<0.001
Fatal and non-fatal stroke
 ADVANCE (n=11 140)215/5569 (3.9)218/5571 (3.9)0.98 (0.82–1.19)0.7550.02*
 EUROPA (n = 12 218)98/6110 (1.6)102/6108 (1.7)0.96 (0.73–1.27)0.774
 PROGRESS (n = 6105)307/3051 (10.1)420/3054 (13.8)0.72 (0.62–0.83)<0.001
 Total (n = 29 463)620/14 730 (4.2)740/14 733 (5.0)0.82 (0.74–0.92)0.002
Non-fatal myocardial infarction
 ADVANCE (n = 11 140)136/5569 (2.4)135/5571 (2.4)1.01 (0.80–1.29)0.9130.06
 EUROPA (n = 12 218)295/6110 (4.8)378/6108 (6.2)0.77 (0.66–0.90)0.001
 PROGRESS (n = 6105)60/3051 (2.0)96/3054 (3.1)0.62 (0.44–0.86)0.003
 Total (n = 29 463)491/14 730 (3.3)609/14 733 (4.1)0.80 (0.71–0.90)<0.001
Hospital admission for heart fail
 ADVANCE (n = 11 140)197/5569 (3.5)199/5571 (3.6)0.98 (0.81–1.20)0.8560.04
 EUROPA (n = 12 218)63/6110 (1.0)103/6108 (1.7)0.61 (0.44–0.83)0.002
 PROGRESS (n = 6105)75/3051 (2.5)93/3054 (3.1)0.80 (0.59–1.09)0.159
 Total (n = 29 463)335/14 730 (2.3)395/14 733 (2.7)0.84 (0.72–0.96)0.015
  • Cox regression multivariate analysis was used to calculate HRs and 95% CI adjusted for age, gender, hypertension, diabetes mellitus, smoking, history of MI, prior revascularization, prior CVA/TIA, use of β-blockers, lipid-lowering agents, anti-platelet agents, diuretics, calcium antagonists, indapamide, and perindopril dosage. The multivariate Cox model was applied for the results of the individual trials as well as the pooled analysis of individual data presented as ‘total’ (n = 29 463) adjusted for the three trials. HR, hazard ratio; CI, confidence interval; MI, myocardial infarction.

  • *In single therapy analysis, P for heterogeneity >0.10.

Blood pressure

The perindopril-based treatment regimen was associated with lower major cardiovascular event rates across all strata of systolic BP as well as diastolic BP, although in the lower systolic BP categories, the treatment effect estimates did not reach significance. Also, the relative treatment benefit was somewhat larger at higher diastolic BP levels (P for interaction <0.001) (Figure 2A). During the run-in period of 4 weeks, all patients were treated with perindopril (n = 29 463) and BP decreased from 142/82 to 134/78 mmHg. The average BP reduction during run-in was 8.3/3.8 mmHg and during follow-up 5.4/2.3 mmHg. The BP reduction during run-in was comparable in each of the three trials (ADVANCE: 8/3 mmHg; EUROPA 9/4 mmHg; and PROGRESS 9/5 mmHg). The perindopril-based regimen reduced the incidence of major cardiovascular events in all categories of BP reduction during run-in (Figure 2B). However, the relative treatment benefit was somewhat larger in patients with a higher diastolic BP reduction during run-in (P for interaction 0.05).

Figure 2

(A) Treatment effect of perindopril-based regimen in relation to major cardiovascular diseases (cardiovascular mortality, MI, and stroke) according to baseline blood pressure (BP) strata. Forest plot of the treatment effect of the perindopril-based regimen at categories of baseline systolic BP (<120, 120–139.9, 140–159.9, and ≥160 mmHg) and diastolic BP (<80, 80–89.9, 90–99.9, and ≥100 mmHg). Next to the forest plot, the HR and 95% CI are presented as well as the P-values for interaction. Cox regression multivariate analysis was used to calculate HRs and 95% CI adjusted the full model. Additionally, P-values for interaction were calculated by including an interaction term in the model using systolic and diastolic BP levels as continuous variable * treatment. HR, hazard ratio; CI, confidence interval. (B) Treatment effect of perindopril-based regimen in relation to major cardiovascular diseases (cardiovascular mortality, MI, and stroke) according to BP reduction levels. Forest plot of the treatment effect of the perindopril-based regimen at tertiles of systolic and diastolic BP reduction during the run-in period. Cox regression multivariate analysis was used to calculate HRs and 95% CI. We emphasize that these analyses were also corrected for the starting systolic and diastolic BP values before application of treatment in all patients. Additionally, P-values for interaction were calculated by including an interaction term in the model using systolic and diastolic BP reduction levels as continuous variable * treatment. MI, myocardial infarction; HR, hazard ratio; CI, confidence interval.

Subgroups

The treatment effect of the perindopril-based regimen was independent of baseline clinical characteristics and concomitant medication (Figure 3). In subgroups of age and diabetes mellitus, P for interaction was, respectively, 0.07 and 0.06, but the perindopril-based regimen was associated with a significant positive treatment effect in patients with or without diabetes, and in patients aged <60, 60–70, or >70 years. A dichotomized age variable, at the median age of the study population (63.0 years), resulted in similar treatment effect estimates (age ≤63: HR 0.78; 95% CI 0.69–0.87 and age >63: HR 0.85; 95% CI 0.78–0.93). For combinations of concomitant medication use compromising platelet-inhibitors, lipid-lowering agents, calcium antagonists, β-blockers, and diuretics use (none, one, two, three, four, or all of the mentioned medication use), HRs were similar as shown for the individual medications and no heterogeneity in the treatment effect was observed (Figure 3, data not shown). Results for other study endpoints were similar and no heterogeneity was observed.

Figure 3

Treatment effect of perindopril-based regimen in reducing cardiovascular mortality, MI, and stroke according to baseline characteristics. Forest plot of the treatment effect of the perindopril-based regimen vs. placebo in subgroups of baseline clinical characteristics. Cox regression multivariate analysis was used to calculate HRs and 95% CI. *P-values for interaction were calculated for each covariate individually by including an interaction term with treatment and the covariate [categorical or continuous (age)] in the multivariate model. HR, hazard ratio; CI, confidence interval; MI, myocardial infarction; CVA, cerebrovascular accident; TIA, transient ischaemic attack.

Significant heterogeneity between the studies was observed for fatal and non-fatal stroke, non-fatal MI, and hospital admission for heart failure (Table 3). In a separate analysis of single therapy by perindopril (n = 14 779), by excluding all indapamide use, heterogeneity vanished (P > 0.10). Single therapy was no longer associated with a significant treatment benefit on fatal and non-fatal stroke (HR 0.96; 95% CI 0.82–1.12) (Figure 4). In contrast, the effect on non-fatal MI and heart failure admissions increased to, respectively, HR 0.76 (95% CI 0.66–0.89) and HR 0.71 (95% CI 0.55–0.92).

Figure 4

The treatment effect of the perindopril-based regimen and single therapy on the reduction of fatal/non-fatal stroke, non-fatal MI, and hospitalization for heart failure (*). *Presented results for non-fatal/fatal stroke are: perindopril-based regimen (HR 0.82; 95% CI 0.74–0.92) and single therapy (HR 0.96; 95% CI 0.82–1.12). Results for non-fatal MI were perindopril-based regimen (HR 0.80; 95% CI 0.71–0.90) and single therapy (HR 0.76; 95% CI 0.66–0.89). Results for hospitalization for heart failure: perindopril-based regimen (HR 0.84; 95% CI 0.72–0.96) and single therapy (HR 0.71; 95% CI 0.55–0.92). In all single therapy analyses, P for heterogeneity was >0.10. NFMI, non-fatal MI.

Discussion

The present analysis, based on individual data from 29 463 patients, demonstrates that a perindopril-based regimen reduces the risk of mortality and major cardiovascular events among patients with various levels of cardiovascular risk. Furthermore, the benefit was observed across a wide spectrum of baseline BP levels. Therefore, the use of ACE-inhibitors should be considered in all patients with established vascular disease or at high risk of vascular disease even in patients with normal BP levels.

A prior study, a meta-analysis, which studied three trials with similar type of patients with CAD (HOPE, EUROPA, and PEACE) showed a consistent treatment effect of ACE-inhibitors in reducing cardiovascular events.26 The current analysis demonstrates clear evidence for a consistent cardiovascular protection in a broader range of patients with vascular disease including clear benefits in reducing death. These results confirm the generalizability of prior studies in separate patients’ groups and support the provision of treatment, not on the basis of arbitrary cut-off points for BP, but rather on the assessment of absolute or total vascular risk, which is raised in patients with stable CAD, diabetes, and stroke as shown in EUROPA, ADVANCE, and PROGRESS. These results are consistent with the findings of the HOPE trial which studied the ACE-inhibitor ramipril.5

Significant interaction was observed for fatal and non-fatal stroke as a separate outcome (Table 3). Such heterogeneity can be related to differences across trials in the study medication, patient selection, or concomitant medication (Table 2). While ADVANCE and EUROPA showed low stroke rates and no treatment benefit for stroke, PROGRESS showed a significant reduction of 28% in stroke in patients with cerebrovascular disease. Therefore, the ACE-inhibitor-based treatment regimen was more effective in reducing recurrent stroke (PROGRESS) than in reducing incident stroke in patients without a history of cerebrovascular disease (EUROPA and ADVANCE). These benefits were contingent on the BP-lowering effect with additional indapamide use. Single therapy with perindopril was not associated with a significant reduction in the risk of stroke. Additional interaction was observed on the separate outcome of non-fatal MI and hospitalization for heart failure (Table 3), mainly caused by the lack of benefit in ADVANCE conducted in patients with DM but most without a history of macrovascular disease. In both cases, interaction was observed for cause-specific outcomes, however, there was no heterogeneity in reducing the over-arching endpoints of all-cause mortality or the composite endpoint of cardiovascular mortality, MI, and stroke. In the analysis of single perindopril therapy, heterogeneity vanished and the risk reduction by single therapy improved for both outcomes.

Regarding BP, the findings from the BPLTTC already suggested some BP-dependent and some BP-independent effects of ACE-inhibitors on the risk of major coronary disease events by using tabular data.20,21,27,28 Now, we can confirm these findings in a large combined patient-level data analysis with access to individual BP measurements. We showed a lower major cardiovascular event rate across all strata of systolic and diastolic BPs, although not statistically significant for lower systolic BP strata, as well as across the tertiles of BP reduction during run-in. Patients with the highest diastolic BP or highest diastolic BP reduction experienced a somewhat larger relative treatment effect. Also the patients with the highest BP reduction remained on average also at lower BP during follow-up.

The current study is unique and important because it studies the individual data of large trials across a broad range of vascular risk and is focused on one ACE-inhibitor. The use of individual data makes the results robust and provides a unique opportunity to detect the subgroups of patients who might (or might not) benefit the most from treatment. Furthermore, the ACE-inhibitor perindopril has several properties different from other agents in its class, being a long-acting (24 h), once-daily lipophilic ACE-inhibitor with high affinity for both tissue and circulating ACE. Tissue ACE affinity is related to specific anti-atheroscletic and anti-thrombotic effects, as well as improvements in endothelial function.29 However, we acknowledge that the exact clinical meaning of these properties remains unknown, since ACE-inhibitors have not been prospectively compared with each other in studies.

Several limitations of this analysis can be noted. The trials differed in the patient selection, adjunctive therapy, and drug dosage. Therefore, we adjusted for perindopril dosage and indapamide use in all analyses and checked for interaction or heterogeneity among trials. This is unlikely to have affected our conclusions as discussed above. Furthermore, we acknowledge that the accuracy of the BP measurements in large trials and different hospital settings may be questioned. However, any inaccuracy would be equally distributed among randomized treatment groups and unlikely to affect our conclusions. Although, the patient populations in the three trials studied were different, it is well established that atherosclerotic vascular disease is a generalized process not restricted to a single vascular bed. Therefore, we believe that our pooled patient sample is broadly representative of patients with established vascular disease at high risk of vascular disease across the world. Despite the differences between the trials, the outcomes are consistent with findings from other different meta-analyses on patients without heart failure26,28,30,31 and patients with heart failure.32

The currents study demonstrates that perindopril-based regimens (perindopril in part of the patients combined with indapamide) reduce death in patients with vascular disease. The consistency of the relative effects across subgroups indicates that the absolute benefits conferred by treatment will be determined mainly by each patient's future risk of vascular complications, rather than their initial level of BP alone or other risk factors. These results support the provision of ACE-inhibitor-based treatment, not on the basis of arbitrary cut-off points for BP, but rather on the assessment of vascular risk, which is raised in patients with stable CAD, diabetes, and stroke. This approach, based on total cardiovascular risk, should form the basis for the recommendations for treatment in the revision of the major national and international guidelines.

Funding

This combined analysis was initiated by the authors and was designed, conducted, interpreted, and reported independently of the original sponsor. The current study had no funding source or any with a participating role in data collection, outcome assessment, or writing of the manuscript. All authors had joint responsibility for the decision to submit for publication.

Conflict of interest: J.J.B., T.N., and E.B. have declared no conflict of interest. J.C., S.M., K.F., R.F., and M.L.S. have received research grants and fees from Servier.

Acknowledgements

All authors contributed to the analysis of the data and writing of the report. All authors approved the final version of the manuscript. Dr J.J.B. and Dr T.N. merged individual data from the trials and had full access to the total data sets of all three trials. Dr J.J.B. performed the statistical analyses.

References

View Abstract