After the HOPE (Heart Outcomes Prevention Evaluation) trial (1), angiotensin-converting enzyme inhibitors (ACE-Is) have been recommended for reduction of cardiovascular (CV) events in patients at high CV risk without heart failure (HF) (2). The results of the HOPE study, in which a substantial reduction of major CV events (CV death, myocardial infarction [MI], and stroke) was reported, were confirmed in the PROGRESS (Perindopril pROtection aGainst REcurrent Stroke Study) (3) and EUROPA (EURopean trial On reduction of cardiac events with Perindopril in stable coronary Artery disease) (4) trial but not in other trials comparing ACE-Is with placebo in patients at high CV risk ((5),(6),(7),(8),(9),(10),(11),12). However, a meta-analysis by Dagenais et al. (13), collecting 3 major randomized placebo-controlled studies on ACE-Is in patients without HF, showed favorable effects of ACE-Is on CV events.

The rationale for ACE-I therapy in patients without HF relies on the effects of vascular angiotensin II or bradykinin/prostaglandin on the progression of atherosclerosis (14). However, it is well known that during ACE-I therapy, angiotensin II synthesis may shift to alternative ACE-independent enzymatic pathways, which could reduce the efficacy of therapy (15). The unfavorable effects of angiotensin II on atherosclerosis progression are mediated through stimulation of angiotensin II receptor 1. Angiotensin receptor blockers (ARBs) prevent angiotensin II receptor 1 stimulation without direct effects on bradykinin/prostaglandin, which improves their adverse effect profile. Although ARBs reduce diabetic retinopathy and nephropathy ((16),(17),18), and CV morbidity and mortality in patients with HF (19), their effects in patients without HF are less certain because major clinical trials comparing ARBs with placebo reported conflicting results ((20),(21),(22),(23),(24),(25),(26),(27),(28),29).

The aim of the current study was to assess, in a meta-analysis, the effects of ACE-I and ARB therapy on the composite outcome of CV death, MI, and stroke as well as on all-cause death, new-onset HF, and new-onset diabetes mellitus (DM) in high-risk patients without HF.

MEDLINE, Cochrane Database, ISI Web of Sciences, and SCOPUS were searched for articles with no language restrictions until June 2012. Studies were identified by the following headings: angiotensin receptor blocker, antagonist of angiotensin II receptor 1, ARB, angiotensin-converting enzyme, ACE, randomly, random, randomized controlled trial, and clinical trial. We used reference lists of the retrieved articles as well as information from colleagues to identify additional eligible studies.

This study was designed according to the PRISMA (Preferred Reporting Items for Systematic reviews and Meta-Analyses) statement ((30),31). Only randomized, double-blind, clinical trials comparing either an ARB or an ACE-I with placebo, excluding patients with systolic or diastolic HF and reporting clinical events (including all-cause and CV death, MI, stroke, new-onset HF, and new-onset DM), were considered for the analysis.

Two reviewers independently selected potentially eligible trials. Discrepancies were resolved by discussion and consensus. Two reviewers independently read the full text of retained studies, which were checked to avoid inclusion of data published in duplicate. Data on baseline characteristics, presence of DM, hypertension, coronary artery disease, and pre-specified outcomes, including all-cause and CV death, MI, stroke, new-onset HF, and new-onset DM, were abstracted. The first objective of the study was to assess the effect of treatments on the composite outcome (CV death, MI, and stroke) and on all-cause death. In addition, the effects of treatments on the risk of each component of the composite outcome and on new-onset HF and new-onset DM were also explored.

The quality of each trial was evaluated by giving a score for each study using the Detsky method (32) (Table 1). Of 25,661 articles identified by the initial search, 43 were retrieved for more detailed evaluation, and 25 (corresponding to 26 trials) were included in the study (Figure 11_gr1). Included trials and population details are listed in Tables (Table 1) to (Tables 2, 3). Thirteen trials compared ACE-Is with placebo ((1),(3),(4),(5),(6),(7),(8),(9),(10),(11),(12),(),), and 13 trials compared ARBs with placebo ((17),(18),(20),(21),(22),(23),(24),(25),(26),(27),(28),29). Characteristics of patient populations enrolled in ACE-I and ARB trials were compared by using the Student t test for unpaired samples or the chi-square test, as appropriate.

To assess the level of risk of patients included in trials, the incidence rate (IR; expressed as events per 1,000 patient-years) of each outcome analyzed was calculated for the placebo patients enrolled in ACE-I trials and for the placebo patients enrolled in ARB trials, by using the following formula: IR = (number of placebo patients with event × 1,000)/(number of placebo patients × years of follow-up) (35).

Odds ratios (ORs) of the effect of randomized treatments were calculated by using the metan routine (STATA version 11.0, StataCorp, College Station, Texas). ORs and 95% confidence intervals (CIs) for each outcome were separately calculated for each trial, with grouped data, by using the intention-to-treat principle ((36),37). The choice to use ORs was driven by the retrospective design of the meta-analysis, based on published studies that vary in design, subjects' population, treatment regimens, primary outcome measures, and quality ((38),39). Overall estimates of effect were calculated with a random effects model (40). The assumption of homogeneity between the treatment effects in different trials was tested with the Q and I² statistics. Pooled ORs were logarithmically transformed and weighted for the inverse of variance. The significance level for the overall estimates of effect and for meta-regression analyses was set at p ≤ 0.05. Only a single, first event could contribute to each outcome measure.

To explore the influence of potential effect modifiers on outcomes, a meta-regression analysis was performed with the metareg command (41) (STATA version 11.0) to test demographic characteristics of the study population, body mass index, percentage of patients with coronary artery disease, percentage of patients with DM, percentage of patients with hypertension, blood pressure values and blood pressure differences from start to end of each study, current therapy, length of follow-up, and quality of trials (32). For all meta-regression analyses, a random effects model was used to take into account the mean of a distribution of effects across studies. In fact, a random effects model more appropriately provides wider CIs for the regression coefficients than a fixed effect analysis, if residual heterogeneity exists (42). The weight used for each trial was the inverse of the sum of the within-trial variance and the residual between trial variance, to correspond to a random effects analysis. To estimate the additive (between-study) component of variance tau-2, the restricted maximum likelihood method was used to take into account the occurrence of residual heterogeneity not explained by the potential effect modifiers (42). To verify the consistency of the results, individual meta-analyses were performed for single drugs (candesartan, olmesartan, telmisartan, irbesartan, perindopril, ramipril, and enalapril) against each outcome when a drug was used in at least 2 trials.

To evaluate potential publication bias, a weighted linear regression was used, with the natural log of the OR as the dependent variable and the inverse of the total sample size as the independent variable. This is a modified Macaskill's test, which gives more balanced type I error rates in the tail probability areas compared with other publication bias tests (43).

Baseline characteristics of the 26 trials included in the meta-analysis are shown in Tables (Table 1) through (Tables 2, 3). Of 108,212 patients, a total of 53,791 were enrolled in ACE-I trials and 54,421 in ARB trials. Duration of follow-up ranged from 2 to 6.5 years (3.68 ± 1.11 years). The overall mean age of subjects was 58 ± 11 years, and 35% were women. Mean age was 58.3 ± 8.3 years in ACE-I trials and 57.7 ± 13.1 years in ARB trials (p = NS); 26% of patients enrolled in ACE-I trials were women compared with 44% in ARB trials (p < 0.05). Length of follow-up was not different between ACE-I (3.66 ± 0.90 years) and ARB (3.69 ± 1.32 years) (p = NS) trials. Placebo patients enrolled in ACE-I trials compared with placebo patients enrolled in ARB trials showed nonsignificant differences in the IR of composite outcome (27 vs. 25 events per 1,000 patient-years; p = 0.608) as well as all-cause death (21 vs. 19 events per 1,000 patient-years; p = 0.898), new-onset HF (7 vs. 6 events per 1,000 patient-years; p = 0.784), and new-onset DM (15 vs. 24 events per 1,000 patient-years; p = 0.176).

ACE-Is significantly reduced the risk of the composite outcome by 14.9% compared with placebo (OR: 0.830 [95% CI: 0.744 to 0.927]; comparison p = 0.001, heterogeneity p = 0.002) (Figure 11_gr2A). When components of the composite outcome were considered separately, the 10% reduction of CV death did not achieve statistical significance (OR: 0.896 [95% CI: 0.783 to 1.026], comparison p = 0.112, heterogeneity p = 0.087) (6), whereas ACE-Is significantly reduced the risk of MI by 17.7% (OR: 0.811 [95% CI: 0.748 to 0.879], comparison p < 0.001, heterogeneity p = 0.438) (6) and of stroke by 19.6% (OR: 0.796 [95% CI: 0.682 to 0.928], comparison p = 0.004, heterogeneity p = 0.115) (6). In addition, ACE-Is reduced the risk of all-cause death by 8.3% (OR: 0.908 [95% CI: 0.845 to 0.975], comparison p = 0.008, heterogeneity p = 0.368) (Figure 11_gr2B), new-onset HF by 20.5% (OR: 0.789 [95% CI: 0.686 to 0.908], comparison p = 0.001, heterogeneity p = 0.252) (6), and new-onset DM by 13.7% (OR: 0.851 [95% CI: 0.749 to 0.965], comparison p = 0.012, heterogeneity p = 0.069) (6). Significant heterogeneity among trials was found for the composite outcome but not for all other outcomes explored in the analysis.

ARBs significantly reduced the risk of the composite outcome by 7.0% compared with placebo (OR: 0.920 [95% CI: 0.869 to 0.975], comparison p = 0.005, heterogeneity p = 0.686) (Figure 11_gr2A). When components of the composite outcome were analyzed separately, ARBs did not reduce the risk of CV death (OR: 1.033 [95% CI: 0.847 to 1.260], comparison p = 0.748, heterogeneity p = 0.012) (6), whereas the 9.5% reduction of MI risk approximated statistical significance (OR: 0.903 [95% CI: 0.803 to 1.015], comparison p = 0.086, heterogeneity p = 0.420 (6). ARBs significantly reduced the risk of stroke by 9.1% (OR: 0.900 [95% CI: 0.830 to 0.977], comparison p = 0.011, heterogeneity p = 0.469) (6). No significant effect was found on the risk of all-cause death (OR: 1.006 [95% CI: 0.941 to 1.075], comparison p = 0.866, heterogeneity p = 0.368) (Figure 11_gr2B) and of new-onset HF (OR: 0.892 [95% CI: 0.761 to 1.046], comparison p = 0.159, heterogeneity p = 0.188) (6). ARBs significantly reduced the risk of new-onset DM by 10.6% (OR: 0.855 [95% CI: 0.798 to 0.915], comparison p < 0.001, heterogeneity p = 0.819) (6). Significant heterogeneity among trials was found only for CV death. ORs for the effects of ACE-Is and ARBs are reported in (Figure 11_gr3).

Results were confirmed when potential effect modifiers were introduced as covariates in the meta-regression analysis (6). Moreover, similarly to the overall results, telmisartan (OR: 0.915 [95% CI: 0.853 to 0.982], p = 0.013), perindopril (OR: 0.768 [95% CI: 0.679 to 0.869], p < 0.001), and enalapril (OR: 0.580 [95% CI: 0.370 to 0.908], p = 0.017) reduced the risk of the composite outcome, whereas olmesartan (OR: 1.063 [95% CI: 0.748 to 1.509], p = 0.733), candesartan (OR: 0.728 [95% CI: 0.357 to 1.482], p = 0.381), irbesartan (OR: 0.949 [95% CI: 0.597 to 1.508], p = 0.824), and ramipril (OR: 0.861 [95% CI: 0.661 to 1.122], p = 0.268) did not (6). Results for other outcomes are reported in (Table 4).

Macaskill's modified test did not show publication bias for any outcome.

This meta-analysis confirmed that, compared with placebo, ACE-Is substantially reduced the composite of CV death, MI, and stroke as well as all-cause death, new-onset HF, and new-onset DM in high-risk patients without HF, mostly with coronary or other vascular diseases. ARBs, in high-risk patients mostly with DM or impaired glucose tolerance without HF, reduced the composite outcome and new-onset DM but did not seem to reduce rates of all-cause death or new-onset HF.

Our findings confirm and extend a previous meta-analysis of ACE-I trials reported by Dagenais et al. (13), adding an additional 10 ACE-I trials (corresponding to 23,986 additional patients) ((3),(5),(6),(7),(10),(11),(12),(33),(),) to HOPE (1), EUROPA (4), and PEACE (Prevention of Events with Angiotensin Converting Enzyme inhibition) (8) included in that previous analysis. In particular, consistent with the study by Dagenais et al, our analysis demonstrates that ACE-Is, compared with placebo, substantially reduced all-cause death by 8.3%, CV death by 10%, MI by 17.7%, and stroke by 19.6%, as well as new-onset HF by 20.5%. In addition, the current analysis reported a significant 13.7% reduction of the risk of new-onset DM.

The effects on major clinical events of ARBs in patients without HF have been evaluated in several trials reporting conflicting results ((17),(18),(20),(21),(22),(23),(24),(25),(26),(27),(28),29). In the RENAAL (Reduction in ENdpoints with the Angiotensin Antagonist Losartan) study (21), losartan, compared with placebo, failed to reduce the secondary composite CV outcome, including CV death, MI, stroke, coronary or peripheral revascularization, and new HF, in diabetic patients with nephropathy. Similarly, in high-risk diabetic patients without (20) and with (22) overt nephropathy, irbesartan did not reduce CV events. In the SCOPE (Study on Cognition and Prognosis in the Elderly) (24), a candesartan-based therapy in hypertensive elderly patients failed to reduce the primary composite outcome of CV death, MI, and stroke, whereas it significantly decreased the risk of new-onset DM. In contrast, in a small study enrolling high-risk patients (23), candesartan, compared with placebo, significantly reduced the risk of a composite outcome including CV death, MI, and coronary revascularization. Subsequently, however, high-dose candesartan did not reduce CV events in patients with type 1 or type 2 DM with no previous CV events ((17),18). In the TRANSCEND (Telmisartan Randomised AssessmeNT Study in ACEiNtoleran subjects with cardiovascular Disease) trial (27), ACE-I–intolerant patients were enrolled and assigned to either telmisartan or placebo. Although the primary endpoint of CV death, MI, stroke, and new-onset HF was not significantly reduced by telmisartan, a 13% reduction in the secondary combined endpoint, including CV death, MI, and stroke was reported (p = 0.05), whereas new-onset DM was not reduced by telmisartan. However, in the PROFESS (Prevention Regimen For Effectively Avoiding Second Strokes) study (26), which enrolled patients with a recent stroke, telmisartan, compared with placebo, failed to reduce the recurrence of stroke, all-cause mortality, DM, or any CV endpoint. More recently, in the NAVIGATOR (Nateglinide And Valsartan in Impaired Glucose Tolerance Outcomes Research) trial (28), which enrolled stable patients with impaired glucose tolerance with or at high risk of developing CV disease, valsartan reduced the occurrence of DM but failed to reduce CV morbidity or mortality. Finally, the recently concluded ROADMAP (Randomized Olmesartan And Diabetes MicroAlbuminuria Prevention) (29) trial and the ORIENT (Olmesartan Reducing Incidence of End stage renal disease in diabetic Nephropathy Trial) (25) trial reported an increase of the prespecified secondary endpoint of CV mortality in diabetic patients receiving olmesartan, despite a favorable effect on the primary renal endpoint.

The current meta-analysis demonstrates that ARBs significantly reduce the composite outcome of CV death, MI, and stroke by 7.0%, as well as stroke by 9.1% and new-onset DM by 10.6% in high-risk, mostly diabetic or glucose-intolerant patients without HF enrolled in randomized clinical trials. However, no significant effects could be found on the risk of all-cause death, MI, and new-onset HF.

To our knowledge, no previous meta-analysis investigated the effects of ARBs compared with placebo, on top of concomitant therapy, in patients without HF. In fact, in a previous meta-analysis by Baker et al. (44), only the TRANSCEND (27) study, among ARB trials reported, was included, whereas the meta-analysis by Bangalore et al. (45), that mainly focused on the risk of MI in patients receiving ARBs, reported also trials in patients with HF. More recently, van Vark et al. (35) reported a meta-analysis of randomized clinical trials of renin-angiotensin-aldosterone system inhibitors. Compared with the current analysis, only hypertension trials were included in the study by van Vark et al. (35), and only CV and all-cause death were analyzed. The study demonstrated a significant benefit on all-cause and CV death, significantly associated with blood pressure reduction and almost entirely driven by ACE-I effects, but it could not assess the effects of ARBs on high-risk, nonhypertensive patients. Finally, McAlister et al. (46) reported a meta-analysis of ACE-Is and ARBs in normotensive atherosclerotic patients showing a favorable effect on major clinical outcomes. In this study, however, ACE-I and ARB trials were grouped into the same analysis, and no distinction was made between ACE-I and ARB trials. In addition, trials also enrolling patients with HF were included. Thus, the CV effects of ARBs in high-risk patients without HF were not adequately investigated in previous meta-analyses.

First, we used summary rather than individual patient data. In addition, the characteristics of populations enrolled in ACE-I and ARB trials were different. ACE-I trials were mostly conducted in patients with coronary or other vascular atherosclerotic disease, and ARB trials were mostly conducted in patients with DM or impaired glucose intolerance. Although the level of risk in the placebo arms of the ACE-I and ARB trials was not significantly different for any outcome analyzed, and the characteristics of study populations did not influence the findings of the study in sensitivity analyses, the differences between ACE-I and ARB populations encourage caution regarding the generalizability of the results. In addition, inherent limitations of meta-regression analysis should be considered when interpreting the results of sensitivity analysis (47). Thus, this study does not represent a direct comparison between ACE-Is and ARBs that can only be adequately assessed in ad hoc trials. In fact, the ONTARGET (ONgoing Telmisartan Alone and in combination with Ramipril Global End-point Trial) (48) study observed no significant difference between telmisartan and ramipril on major CV outcomes, although no placebo arm was available in that trial, preventing a thorough evaluation of the net benefits of the 2 treatments on top of concomitant CV therapy. A substantial number of patients discontinued ARBs in clinical trials, and this action may have attenuated the effect of therapy by an intention-to-treat analysis. In addition, a potentially greater benefit from ARBs in particular subgroups of patients, such as those with organ damage, cannot be excluded from the current analysis. Finally, the current study considered ACE-Is and ARBs as 2 classes of drugs, although both include several agents with different pharmacokinetic and pharmacodynamics properties.

In high-risk patients without HF, ACE-Is, as a class, reduced all-cause and CV death, as well CV morbidity and new-onset DM. ARBs reduced the composite outcome of CV death, MI, and stroke as well the risk of new-onset DM. Thus, ARBs represent a viable option for high-risk patients who do not tolerate ACE-I therapy.