Review of the CAPRICORN Trial

Effect of carvedilol on outcome after myocardial infarction in patients with left-ventricular dysfunction: the CAPRICORN randomized trial

Lancet 2001;357:1385-90.

Background While beta blockers were frequently used in patients with acute myocardial infarction complicated by heart failure, at the time the CAPRICORN trial was undertaken, supportive evidence from contemporary, large scale randomized trials was lacking.

The BHAT and ISIS-1 trials, reviewed in this section, were published in 1982 and 1986, respectively; before contemporary therapies like thrombolysis/revascularization and ACE-inhibition had been established. These early trials also excluded patients with overt heart failure or those who had concerning signals, such as soft hemodynamics.

The CAPRICORN trial sought to test the hypothesis that early initiation of carvedilol in patients with AMI complicated by heart failure and LV dysfunction would reduce morbidity and mortality compared to placebo.

Patients Eligible patients were 18 years or older with a stable, definite myocardial infarction occurring 3-21 days prior to randomization with either a left ventricular ejection fraction of ≤40% or a wall motion index score of ≤1.3 and who were receiving an ACE inhibitor for at least 48 hr and were on a stable dose for at least 24 hr.

Patients could have received IV diuretics during the acute phase but were excluded if they continued to require IV diuretics or inotropes or had uncontrolled heart failure. Other reasons for exclusion included unstable angina, SBP <90 mmHg, heart rate <60 bpm or unstable insulin dependent diabetes. Patients with any other indication to receive beta blockers were also excluded as were individuals on inhaled beta 2 agonists or steroids.

Baseline characteristics The average age of patients was 63 years and over 70% were men. Approximately one third of patients had a prior MI, more than 20% had diabetes and 50% had hypertension. The average ejection fraction was 32% and over half of patients had an anterior MI. Patients were hemodynamically stable with an average blood pressure slightly above 120/70 mmHg and heart rate of 77 beats per minute. Nearly 50% of patients underwent thrombolysis or angioplasty for the primary MI and, at the time of randomization, 98% of patients were on an ACE inhibitor and 86% were on aspirin.

Procedures Patients received either carvedilol or an identical placebo. Study medication was titrated up to the highest tolerated dose for each patient, to a maximum of 25 mg twice daily. The initial dose was 6.25 mg twice daily. If not tolerated, it was re-administered or reduced by half. If that dose was not tolerated the patient received no study medication but was followed up anyway. After successful initial dosing the patient returned as an outpatient every 3-10 days for assessment of tolerability and further up-titration. In the absence of adverse events or clinical heart failure and if the heart rate was >50 bpm and the systolic BP was >80 mmHg, the dose was increased to the next level. The patient remained in the outpatient department for 2hr to ensure that no side effects ensued.

During the maintenance period patients were reviewed every 3 months during the first year and every 4 months thereafter. Investigators were encouraged to review the dose of study medication at every visit and to ensure that doses of other drugs, especially ACE inhibitors, were adjusted accordingly to achieve optimum dose levels.

Endpoints The original primary endpoint was all-cause mortality and the trial was intended to run until a total of 633 primary endpoint events occurred. However, during a masked analysis the data and safety monitoring committee noted that overall mortality was lower than predicted and that the study could not be completed with the original sample size and power as planned. The steering committee, therefore, designated co-primary endpoints. The first was a new composite primary endpoint of all-cause mortality or cardiovascular hospital admission and the second was all-cause mortality.

The trial was still intended to run until 633 primary endpoint events occurred with the original sample size. The alpha (false-positive) was divided between the 2 primary endpoints so that the composite endpoint of all-cause mortality and cardiovascular hospitalization was tested at 0.045 and all-cause mortality was tested at 0.005. Secondary endpoints included sudden death, hospitalization for heart failure, recurrent nonfatal MI, and all-cause mortality or nonfatal MI.

Results 1,959 patients were included in the final analysis; 984 in the placebo group and 975 in the carvedilol group. The mean follow-up was 1.3 years. Compared to placebo, carvedilol did not significantly reduce the composite primary endpoint of all-cause death or cardiovascular hospitalization (HR 0.92; 35% vs 37%; 95% CI 0.80-1.07) or all-cause death alone (HR 0.77; 12% vs 15%; 95% CI 0.6-0.98).

The authors present the all-cause mortality result as positive; however, it is NOT. In table 1, the p for all-cause mortality is 0.031, which is >0.005. Thus, there is a good chance that the mortality result represents a false positive finding.

For the secondary endpoints, there were reductions in nonfatal MI and the combined endpoint of all-cause death or nonfatal MI for patients in the carvedilol group but these results are based on a relatively small number of events.

Of the 940 patients who entered the maintenance phase in the carvedilol group, 74% reached the maximum dose of 25 mg bid while 11% and 7% reached 12.5 mg bid and 6.25 mg bid.

There is no subgroup data or safety data presented in the main paper.   

Conclusions In patients with AMI complicated by heart failure and significant LV dysfunction, carvedilol did not significantly reduce all-cause death or cardiovascular hospitalizations compared to placebo.

This trial is often presented and discussed as a positive trial but it is not. The appropriately powered endpoint in this case was the composite primary endpoint, which was clearly negative.

The difference in all-cause mortality is based on a small number of events with a post-hoc power of 19.3% at the prespecified alpha of 0.005. The reported p-value for all-cause mortality was 0.031, which was much higher than the prespecified p value of ≤ 0.005 to declare a positive result. Due to the underpowered nature of all-cause death, the investigators chose to make the alpha as low as it was to avoid making false positive claims but then they did so anyway!

While power is generally taught to infer a trial’s strength to avoid false negative results, low power also increases a trial’s susceptibility to false positive results. We have discussed this with small trials involving magnesium and nitrates, which were highlighted by the meta-analyses presented in the ISIS 4 publication.

In conclusion, CAPRICORN was a negative trial. Its primary results were negative and differences in secondary endpoints, including the reduction in death, should be viewed as “hypothesis generating” only. It does not clearly establish a beneficial role for beta blockers in post-MI patients with significant LV dysfunction and heart failure.