Ramipril and cardiovascular events in high risk patients (HOPE)

Effects of an angiotensin-converting-enzyme inhibitor, ramipril, on cardiovascular events in high-risk patients

Authors	The Heart Outcomes Prevention Evaluation (HOPE) study investigators.
Source	New England Journal of Medicine 342:145-53. January 20, 2000.
Institutions	Multi-institutional and international.
Support	Medical Research Council of Canada, Hoechst- Marion Roussel, AstraZeneca, King Pharmaceuticals, Natural Source Vitamin E Association and Negma, and the Heart and Stroke Foundation of Ontario.

Background

Epidemiologic and experimental data suggest that the renin-angiotensin-aldosterone system has a significant role in increasing the risk for cardiovascular events. Although ACE inhibitors have been shown to benefit patients with left ventricular dysfunction, they have not yet been shown to benefit patients with cardiovascular risk factors but without LV dysfunction. The purpose of this trial was to investigate the effect of an ACE inhibitor, ramipril, on cardiovascular events in this group of patients.

Methods

Patients

Eligible: Patients at least 55 years old with a history of: any of the following:

coronary disease
stroke
peripheral vascular disease
diabetes plus another cardiovascular risk factor (hypertension, elevated cholesterol, low HDL, cigarette smoking or microalbuminuria)

Exclusion criteria:

congestive heart failure or a documented ejection fraction less than 40%
patients taking an ACEI or vitamin E
MI or stroke within the previous 4 weeks
uncontrolled hypertension
overt nephropathy

Intervention

During a 3-week run-in period, patients received 2.5 mg of ramipril followed by matching placebo. Approximately 10% were then excluded because of withdrawal of consent, noncompliance, electrolyte abnormalities or side-effects.

The remaining patients were randomized to 10 mg of ramipril once daily (after a brief titration period) or matching placebo. A small subset of patients (5%) was randomized to a low dose of ramipril, 2.5 mg daily.

Because this study also tested the effect of vitamin E supplementation, patients were also randomized to daily 400 IU vitamin E or placebo.

Follow-up

Follow-up visits occurred at one month, 6 months and every 6 months thereafter, with follow-up planned for 5 years.

Analysis

The primary outcome was a composite of cardiovascular death, MI or stroke; each of these outcomes was also analyzed separately. Secondary outcomes were all-cause mortality, hospitalization for CHF or unstable angina, revascularization and the occurrence of diabetic complications. A number of other cardiovascular complications were also recorded.

Results

Patients

Of 10,576 patients who enrolled in the study and participated in the run-in phase, 1,035 were excluded because of withdrawal of consent, noncompliance, electrolyte abnormalities or side-effects. Of the 9,541 remaining patients, 4645 were assigned to receive 10 mg of ramipril, 4652 were assigned to placebo, and 244 were assigned to low dose ramipril (2.5 mg per daily).

Baseline characteristics included (my approximate average of both groups):

Male sex: 73%; age: 66 years
BP: 139/79; BMI: 28
History of coronary disease: 80% (prior MI 53%; CABG 26%; history of angina 55%)
History of stroke or TIA: 11%
Peripheral vascular disease: 43%
Hypertension: 47%; diabetes: 38%; hypercholesterolemia: 66%; current smokers: 14%
Currently on beta-blockers: 39%; aspirin: 76%; lipid agents: 29%; calcium blockers: 47%

Subsequent chart review of randomized patients revealed that 54% had had their LV function determined; it was less than 40% (an exclusion criterion) in about 8% of patients.

Length of treatment and compliance

Patients were recruited between December, 1993 and June, 1995. Length of treatment and follow-up was to be 5 years, but the trial was halted approximately one year early (April, 1999) because of interim analyses showing clear efficacy. Although the mean length of follow up is not given explicitly, it was presumably between 4 and 5 years.

The percentage of patients assigned to ramipril actually receiving 10 mg of ramipril was 83% at one year follow-up and 65% at the final visit. The percentage of patients assigned to ramipril receiving either ramipril or open-label ACEI at the final visit was 79%.

The percentage of patients assigned to placebo receiving an ACEI was 6% at one year follow-up, and 12% at 5 years.

The study drug was discontinued permanently in 29% of patients assigned to ramipril and 27% assigned to placebo. Reasons for discontinuation included cough (7.3% ramipril; 1.8% placebo), and clinical events (6.7% ramipril, 9% placebo).

Main outcomes

Blood pressure at entry was 139/79. At the end of the study it was 136/76 in the ramipril group and 139/77 in the placebo group.

Ramipril Placebo Relative risk p-value

Death from cardiovascular causes 6.1% 8.1% 0.78 <0.001

Death from non-cardiovascular causes 4.3% 4.1%

Death from any cause 10.4% 12.2% 0.84 0.005

Myocardial infarction 9.9% 12.3% 0.80 <0.001

Stroke 3.4% 4.9% 0.68 <0.001

Composite endpoint: MI, stroke or cardiovascular death 14% 17.8% 0.78 <0.001

A number of secondary and other outcomes were significantly lower in the ramipril group compared with placebo, including:

Incidence of CHF: 9.0% vs 11.5% (RR 0.77)
Revascularization: 16% vs 18.3 % (RR 0.85)
Complications of diabetes, including nephropathy, retinopathy requiring laser, dialysis: 6.4% vs 7.6% (RR 0.84)
New development of diabetes: 3.6% vs. 5.4% (RR 0.66)

Worsening of angina was somewhat lower in the ramipril group, but the incidence of and hospitalization for unstable angina was not different in the two groups.

Subgroup analyses and temporal trend

The reduction in the incidence of the composite endpoint (MI, stroke, cardiovascular death) was similar across a broad array of subgroups, including patients with and without diabetes, hypertension, a history of coronary disease, a history of MI, and cerebrovascular disease. It was similar in men and in women and in the patients with a documented ejection fraction of 40% or greater.

The benefit did not appear to reach statistical significance in the 12% of patients who did not have cardiovascular disease at baseline. It was less impressive, though statistically significant, in patients under 65 years of age and in the 56% of patients without peripheral vascular disease.

The benefit became apparent after about one year, and appeared to remain approximately constant over the subsequent years of the study.

Author's discussion

The authors state that the rationale for this study was the hypothesis that ACE inhibition would prevent events related not only to LV dysfunction, but also to ischemia and atherosclerosis. They note that the magnitude of the effect demonstrated here is similar to that seen with other secondary prevention measures such as aspirin, beta-blockade and lipid lowering. The benefit was also appeared to be additive to these other therapies.

They note that blood pressure was reduced slightly in the ramipril group, and that even a small reduction in blood pressure can reduce cardiovascular events, but they feel that the degree of blood pressure reduction seen here is not enough to account for most of the benefit observed.

Since the majority of patients had ejection fractions greater than 40%, the authors do not feel that the benefit observed here is the result of the known beneficial effects of ACEI in patients with LV dysfunction.

They also speculate briefly on the mechanism of the reduction in the incidence of diabetes and complications related to diabetes that were observed here (improved insulin sensitivity, a decrease in hepatic insulin clearance, an antiinflammatory effect, improved pancreatic blood flow, effect on abdominal fat).

Comment

In this controlled clinical trial of an ACE inhibitor, ramipril, in patients with significant risk factors for cardiovascular disease, but mostly without left ventricular dysfunction, the drug reduced a broad array of cardiovascular outcomes and overall mortality by about 20%. This represents a significant benefit, and also represents a fairly large expansion of the population known to benefit from ACEI therapy.

Does this study really apply to patients with cardiovascular risk factors and normal left ventricular function? Perhaps, but the entry criteria of EF > 40% is not exactly a normal ejection fraction (50% or 55% would be normal). Although the authors state that their results were valid in patients whose EF was in fact greater than 40%, they do not give any data on subgroups with truly normal ejection fractions and they do not indicate what the mean baseline ejection fraction was in those patients in whom it was measured. Some of the benefit that was found may have been due to a beneficial effect in patients with mildly impaired left ventricular function.

The authors imply in the results and in the discussion section of the article that their results also applied to patients with diabetes but without overt cardiovascular disease, whereas their figure 2 demonstrates that the benefit in this subgroup did not reach statistical significance. A more honest and direct statement of the results in this subgroup would have been appropriate.

Do the results demonstrated here apply to other ACE inhibitors, or even to angiotensin receptor blockers? The authors refrain from speculating on this issue, but in an accompanying editorial Dr. Gary Francis of the Cleveland Clinic suggests caution when extrapolating to other drugs, and mentions ramipril's "well-known heightened ACE-inhibitory activity in tissue" (I must confess that tissue-ACE inhibition is not so well-known to me). Pharmacologically how far to extrapolate the results of any trial is a question that comes up frequently. All else being equal, it is nice to use the specific agent that has been tested, but all else is seldom equal. There are variations in price, availability and similarity of patients to the study population. How far to extrapolate ends up being a judgement call, to be made by the prescriber.

February 12, 2000

References

References related to this article from the NLM's PubMed database.

Reader Comments

From: samim celebi chalab [samim.celebi@fornet.net.tr]
Sent: Tuesday, February 15, 2000 2:08 AM

They did not discuss the problem of high drop out of patients because of side effects (the major one is cough) and the problems of benefits vs problems of prolonged use to get these benefits. Also, what about the effect of vit E ?

The drop-out rate was almost 30% at the end of the study, which is indeed significant, but probably not unusual for any drug study carried out over several years. The benefits were calculated in an intention-to-treat fashion, so the actual benefit among those actually taking the drug would probably have been even higher. You do need to take the drug for a prolonged period of time, which is similar to other prevention measures such as lipid-lowering therapy or aspirin treatment.

The reduction in the primary endpoint was similar in patients assigned to the vitamin E and vitamin E placebo groups. --mj

June 29, 2000

[The following letter (minimally shortened here) was sent to me in February by Dr. Salim Yusuf, corresponding author of this paper. I have not been updating this website for the past few months, because of time constraints and other activities, thus the delay in posting it. -- mj]

Dear Dr. Jacobson,

Because of space restrictions we could not provide too many subgroup analyses or discuss further details at greater length in the New England Journal paper. You would perhaps be reassured that the mean ejection fraction in the people that we measured was .55 and there was benefit in this subgroup. This will be the basis of a further paper.

I have some reservations about the third paragraph and it perhaps shows some naivete. We have shown the data fully, regarding the subgroup of those with diabetes but without overt cardiovascular disease in figure 2 and it's transparent to the world what the results are. This was a subgroup that was relatively small and had lower event rates. Not surprisingly, although the relative risk reduction was identical to that of the rest of the patients in the trial, it did not reach statistical significance. No trial is designed to show statistical significance in every subgroup examined.

Best wishes,
Yours sincerely,

Salim Yusuf, Dphil, FRCPC
Professor of Medicine
Director, Division of Cardiology
McMaster University

I thank Dr. Yusuf for his clarification. The fact that the overall ejection fraction was 55% is indeed reassuring. I would be curious, however, whether the benefit also extended to the subgroup of patients with EF's greater than 50%, which is still unclear from the paper. -- mj

August 22, 2000
From: Hans Liedholm, MD PhD <hans.liedholm@smi.mas.lu.se>

1. During the last year four important trials have been published regarding pharmacological treatment of cardiovascular disease, three of them dealing with heart failure (bisoprolol, metoprolol, spironolactone). However, the basis for a change in practice must rely on the robustness of the recommendation. The external validity depends on the representativity of included patients. In Evidence-Based-Medicine it is important to translate findings of controlled trials into practice to answer the question if the result is relevant to an individual patient. As was pointed out in a commentary on the MERIT-HF trial, neither of the bisoprolol and metoprolol trials were properly reported according to the CONSORT guidelines for reports of randomised controlled trials. This was not implemented in the presentation of the RALES study and not in the presentation of the HOPE study. How were patients basically recruited and which patients did not take part?

2. In the HOPE study the incidence of the primary outcome and of deaths from any cause were given in Table 3 with estimates of relative risk and 95 % CI. For practical purposes, the table could also have contained information on NNT and corresponding CI. Exploring this opportunity, data could be presented as follows.

Outcome	NNT	95 % CI
Myocardial infarction, stroke, or death from cardiovascular causes	27	19-44
Death from cardiovascular causes	49	32-101
Myocardial infarction	42	27 – 91
Stroke	67	43-144
Death from non cardiovascular causes	(n.a.)	(n.a.)
Death from any cause	54	32 to 176

n.a. = not applicable

We suggest that it would add to the understanding of the outcome of the trial if these figures (with their CI) had been presented in the original paper.

3. Major studies on ACE-inhibitors stress the importance of using target doses of the drugs as some studies have shown a better result of higher doses than of lower doses. In the HOPE study ramipril was also tested at two dose levels, 2.5 mg and 10 mg. For the composite outcome no difference was found between the groups, comprising only 488 patients. This result could of course be due to a large type II error. In fact, the calculated power (post hoc) seem to be only 4 %. What is the dose-response relation of ramipril regarding end points such as those in HOPE?

Hans Liedholm, MD PhD, Agneta Björck Linné, B Pharm
The Drugs and Therapeutics Committee

Juan Merlo, MD PhD
Department of Community Medicine

Malmö University Hospital
S-205 02 Malmö
Sweden.

hans.liedholm@smi.mas.lu.se

I agree that a more complete description of the population being studied is helpful. Sometimes, patients who are excluded are more similar to those we see in practice, raising some doubts about the applicability of the results.

Whether the NNT expression of results will take hold in the reporting of primary research data is not clear, in my opinion. Certainly it is useful in comparing one study to another and one intervention to another.

I didn't see the results of the substudy comparing 2.5 mg of ramipril with 10 mg of ramipril in this article; presumably it was reported elsewhere. -- mj

Addendum (9/5/00). Dr. Liedholm points out that the substudy results are given in a footnote to Table 3 in the NEJM article. Among 244 patients assigned to high dose (10 mg) ramipril, 12.7% reached the composite endpoint; among 244 assigned to low-dose (2.5 mg) ramipril, 13.9% reached the endpoint and among 244 assigned to placebo 16.8% reached the endpoint. Although there was no statistically significant difference, this was probably due to the small sample size and a type II error, since the numbers do suggest a dose-response effect. --mj

Submit a comment

Journal Club homepage