NEJM 1991;324:781-788
Background A hallmark of post-myocardial infarction (MI) care in the 1980’s was the monitoring and suppression of premature ventricular contractions (PVCs) via use of antiarrhythmic drugs. The practice was based on pathophysiologic rationale that PVC burden is a strong risk factor for sudden and non-sudden cardiac death following MI and thus, suppression must reduce death.
PVC reduction was a seductive surrogate endpoint that was easy to measure and declare victory on, but it had never been tested in a proper RCT.
The Cardiac Arrhythmia Suppression Trial (CAST) was sponsored by the National Heart, Lung and Blood Institute (NHLBI) and sought to test the hypothesis that suppression of asymptomatic or mildly symptomatic PVCs with antiarrhythmic therapy with encainide, flecainide, or moricizine after MI would reduce death due to arrhythmia.
Patients Patients were eligible for enrollment 6 days to 2 years post MI with an average of ≥6 PVCs per hour on ambulatory monitoring of at least 18 hours duration, and no runs of VT of ≥15 beats at a rate of ≥120 bpm. An ejection fraction (EF) of ≤55% was required within 90 days of MI or ≤40% if recruited after 90 days.
There was a run-in phase. Patients were only enrolled in the main trial if they had at least 80% suppression of PVCs and at least 90% suppression of runs of VT during an initial, open-label titration period. Initial open-label drug assignment was based, in part, on the EF. Flecainide was not given to patients with an EF of ≤30%. Moricizine was only used as a second line drug in patients with an EF of ≥30%.
Baseline characteristics Baseline characteristics of the patients enrolled in the trial are not provided in the main manuscript and cannot be inferred from the results, tables or figures presented.
Procedures Patients in whom arrhythmias were suppressed were randomly assigned to receive either the effective drug or its matching placebo. A detailed description of study procedures is not presented in the main manuscript. Compliance with the study drug was assessed in follow-up visits and based on pill counts of tablets returned but the schedule of these visits is not provided. Concomitant drug therapy was assessed at the time of the last visit, according to a standardized checklist.
During the trial, patients could be instructed to discontinue the study drug based on the occurrence of the following events: ventricular tachycardia, significant increase in arrhythmia burden, disqualifying ECG changes including significant QT prolongation or bradycardia, new or worsened congestive heart failure, the need for treatment with an antiarrhythmic agent outside the entry criteria for the study, or any number of other adverse medical events divided into cardiovascular or non-cardiovascular events.
Endpoints The primary endpoint of the study was death or cardiac arrest with resuscitation due to arrhythmia. The site PI was responsible for classifying each death without knowledge of the patient’s assigned treatment. Secondary endpoints included cardiovascular and non-cardiovascular causes of death, disqualifying ventricular tachycardia without arrest, syncope, pacemaker implantation, recurrent MI, congestive heart failure, angina pectoris or coronary artery revascularization.
Results Observation began on the day of randomization to blinded therapy and was censored on April 18, 1989, the date when the use of encainide and flecainide was discontinued by the Data and Safety Monitoring Board because the data indicated it was unlikely that benefit could be demonstrated, and it was likely that the drugs were harmful. The original CAST trial manuscript reports data on patients assigned to the encainide and flecainide groups. Moricizine use was continued and would be reported separately in the revised CAST II trial.
1498 participants were randomized to receive either encainide, flecainide or their matching placebo and followed for an average of 10 months. Compliance with the assigned treatments was estimated to be >90% in 70% of all patients and was similar in the active-drug and placebo groups. Antiarrhythmic therapy significantly increased the relative risk of the primary endpoint of death or cardiac arrest due to arrhythmia (RR 2.64; 5.7% vs 2.2%; p=0.0004) and was associated with a number needed to harm (NNH) of approximately 29. It also increased the risk of all deaths and cardiac arrests (RR 2.38; 8.3% vs 3.5%; p=0.0001; NNH = 20); even those not associated with arrhythmia (2.3% vs 0.7%; p=0.01).
Conclusions The CAST trial unexpectedly demonstrated that treatment of asymptomatic or mildly symptomatic PVCs in post-MI patients, with encainide and flecainide, increased death and cardiac arrests.
From a chronological standpoint, it is the first major trial in cardiovascular medicine (perhaps all of medicine) that “reversed” a standard medical practice. In this case, one that was instituted and broadly adopted on the basis of pathophysiologic reasoning and one that targeted a surrogate endpoint.
Thus, more than anything it highlights the importance of testing interventions in properly conducted RCTs prior to adoption and basing the analysis on hard outcomes that are meaningful to patients and society.
How many practices in modern medicine are supported by high quality RCTs? It may be as low as 30-40%.