Mechanisms of Death in the CABG Patch Trial

abstract & commentary

Synopsis: In this trial, ICD therapy did reduce arrhythmic death but had no significant effect on nonarrhythmic deaths. Total mortality was not significantly reduced.

Source: Bigger JT, et al. Circulation 1999;99: 1416-1421.

The cabg patch trial was a randomized, controlled clinical trial evaluating the efficacy of implantable cardioverter defibrillator (ICD) implantation at the time of coronary artery bypass surgery in patients with low ejection fractions and a positive signal averaged ECG. In the CABG Patch trial, all patients undergoing coronary revascularization surgery without concomitant valve surgery or aneurysm resection were screened for participation in the trial. Patients who had an ejection fraction less than 36% had a signal averaged ECG performed. If the signal averaged ECG was positive and no other exclusions were present, they were invited to participate in the trial. Patients with a previous history of sustained ventricular tachycardia or fibrillation were excluded. A high proportion (approximately 70%) of eligible patients participated. Although patients gave consent before their operation, the actual randomization occurred as the patient was coming off cardiopulmonary bypass. This randomization timing was chosen to allow the surgeon to only randomize patients in whom no early major complication had occurred and in whom they felt implantation of an epicardial ICD and its testing would be safe. The results of the trial have been previously reported (Bigger JT Jr. N Engl J Med 1997;337:1569-1575). No difference in survival was noted between the ICD and control groups.

During the study, there were 198 deaths among the 900 patients randomized in the trial. This paper describes the mechanisms of death and the significance of the observed pattern.

Data concerning each death were reviewed by an independent events committee. Deaths were classified as either arrhythmic or nonarrhythmic cardiac deaths or noncardiac deaths. A primary arrhythmic death was defined as sudden and unexpected death within five minutes of acute cardiac symptoms in patients without preceding active symptoms and/or signs of cardiac failure or prior NYHA functional class IV. Deaths in patients who were previously well and died during sleep were classified as primary arrhythmic deaths. Secondary arrhythmic/mechanical deaths were those where preceding acute symptoms and/or signs of heart failure were present but there was no evidence of myocardial pump failure before death. Nonarrhythmic cardiac deaths included deaths due to myocardial pump failure with circulatory collapse, despite a maintained rhythm, and deaths due to cardiac procedures. The location of death was also determined.

During average follow-up of 32 ± 16 months, 198 (22%) of the 900 randomized patients died. For both groups combined, 130 deaths occurred in-hospital, 25 occurred in a nonhospital medical setting, and 29 occurred outside a medical facility. Thirteen deaths occurred in a hospital emergency room department and the location of one death was unknown. Of these deaths, 73% were witnessed. Most of the deaths in both the control group (82%) and the ICD group (75%) were due to cardiac causes. ICD therapy decreased the proportion of arrhythmic deaths in comparison to that in controls (15% vs 29%; P = 0.24). However, there was a slight excess of nonarrhythmic cardiac mortality in the ICD group, particularly during the first two years of follow-up. In this trial, relatively few of the devices implanted had the ability to store electrograms. A change in symptom status during the last seven days of life was noted in a large proportion of deaths. Angina was noted in 10% of patients who died, myocardial infarction occurred in 13%, and 68% had overt heart failure.

Bigger and colleagues conclude that in the CABG Patch trial, ICD therapy did reduce arrhythmic death but had no significant effect on nonarrhythmic deaths. However, since the majority (71% of the deaths) were nonarrhythmic, total mortality was not significantly reduced.

Comment by John P. DiMarco, MD, PhD

Three reasonably large randomized, controlled trials on the efficacy of ICD implantation for primary prophylaxis of sudden death have been published. The Multi-Center Automatic Defibrillator Implantation Trial (MADIT) performed electrophysiologic studies in coronary artery disease patients with low ejection fractions and nonsustained VT. Patients with inducible ventricular tachycardia were randomized to either ICD implantation or "conventional therapy." MADIT reported a large and highly significant reduction in total mortality and across-the-board mortality reductions were seen in arrhythmic, nonarrhythmic, cardiac, and noncardiac mortality. The CABG Patch trial found no difference in total mortality since, as shown in this paper, a slight advantage in arrhythmic mortality was offset by a slight disadvantage in nonarrhythmic mortality. The latter category constituted the largest proportion of deaths. Preliminary data from the Multi-Center Unsustained Tachycardia Trial (MUSTT) have shown no difference in mortality with electrophysiologically guided drug therapy but a major advantage with ICD therapy in low-ejection-fraction patients with inducible ventricular tachycardia. We should ask ourselves why the results of CABG Patch are different from those of MUSTT and MADIT. This paper goes a long way to explain that.

In the CABG Patch trial, about 25% of the deaths occurred during the in-hospital period after the coronary revascularization procedure. It would not be expected that the ICD would make a major impact on deaths in that period since the patient would be monitored and resuscitation facilities would be immediately available. In addition, the stress of cardiac surgery may have precipitated ventricular arrhythmias, which, if successfully terminated, would have led to the institution of appropriate additional therapy in both ICD and control group patients. Even after the early post-CABG period, many of the deaths in CABG Patch occurred in the setting of progressive heart failure. Once again, one would not expect ICD implantation to affect these deaths favorably.

These data pose a problem for electrophysiologists interested in the primary prevention of sudden death. The highest risk groups are almost always defined by severe depression of left ventricular function. However, these groups also have high mortality rates that are not directly due to a primary arrhythmia. In lower risk groups, a greater proportion of deaths may be due to arrhythmia without preceding progressive heart failure, but the overall incidence of sudden arrhythmic death would be too low to make ICD implantation an economically attractive option. It is also possible that by performing revascularization, the CABG Patch trial changed the mechanism of death. ICD implantation should favorably influence ischemic sudden deaths as well as nonischemic sudden deaths. Revascularization itself may have reduced the incidence of ischemically mediated arrhythmias and therefore caused the CABG Patch population to differ in a critical way from the populations in both MUSTT and MADIT.

Finally, we should also note that the data presented here are relevant toward the potential benefits of individual use of automatic external defibrillators (AEDs). These devices are now available by prescription for individual patient use. However, as seen in this study, relatively few of the deaths in high-risk populations occur outside the hospital in a witnessed setting. Thus, it would be anticipated that an AED prescription would have relatively little impact on overall mortality in these groups.


Readers are Invited . . .

Readers are invited to submit questions or comments on material seen in or relevant to Clinical Cardiology Alert. Send your questions to: Michelle Moran — Reader Questions, Clinical Cardiology Alert, c/o American Health Consultants, P.O. Box 740059, Atlanta, GA 30374. Or, you can reach the editors and customer service personnel for Clinical Cardiology Alert via the Internet by sending e-mail to michelle.moran@medec.com. We look forward to hearing from you.