Boston, MA - Rates of appropriate therapy delivered by implantable cardioverter defibrillators (ICDs) in clinical trials apparently far exceed what would be expected from control-group outcomes, a difference that could have important clinical implications, according to a report in the April 15, 2006 American Journal of Cardiology [1]. Perhaps it is the devices themselves that are responsible for some of the difference, propose the authors.

Explanations for the apparent discrepancy have been advanced, but "the finding that appropriate ICD therapies outnumber control-group arrhythmic events by such a wide margin in randomized studies is difficult to fully explain by . . . previously described mechanisms," write Dr Joseph J Germano (Beth Israel Deaconess Medical Center, Boston, MA) and associates. "Another possibility that must be considered is that ICDs and [defibrillating cardiac-resynchronization] devices are themselves proarrhythmic."

The occurrence of appropriate therapies and effective shocks does not constitute proof that ICDs are superior to alternative management strategies.

The group examined seven randomized controlled trials of transvenous-lead ICDs that allowed differentiation between appropriate and inappropriate device therapy—that is, shocks or antitachycardia pacing. Methodologic differences among the studies—AVID, MUSTT, AMIOVIRT, MADIT-2, DEFINITE, DINAMIT, and SCD-HeFT—prevented a conventional meta-analysis, according to the authors.

Over 20 to 45 months of follow-up, rates of inappropriate therapy ranged from 10% to 24% and appropriate therapy from 17% to 64%. In six of the seven trials, appropriate-therapy rates "equaled or exceeded control-group all-cause mortality," write Germano et al. "In all studies that classified deaths, appropriate ICD therapies outnumbered control-group sudden death by a factor of two to three."

Plausible mechanisms behind any device-related proarrhythmia, they write, could include "local effects from leads, direct and indirect effects of ventricular pacing, and problems associated with improper device function."

"I buy that as a hypothesis, but you'd have to prove it to me," Dr Gerald V Naccarelli (Pennsylvania State University, Hershey, PA) told heartwire. He is "not surprised" at the trials' apparent excess of "appropriate" device therapies, explaining that an ICD-group may receive, for example, 30 shock or pacing therapies that are usually for nonlethal runs of VT or other arrhythmias. And a non-ICD control group may experience 10 sudden deaths but also a lot of unobserved nonlethal arrhythmias that would have triggered a device response. "To say that 'every time an ICD fires you save a life' is a little bit overstated."

The numbers are on the side of the patients with the devices.

Beyond that, he said, any discrepancy could be explained by unintentional differences between the device-therapy and control groups and similar factors. "That's a bias that any trial has." And even if the hypothesis is somewhat true, Naccarelli suggested, there's little evidence that it's a pervasive problem. "The numbers are on the side of the patients with the devices." Even if 1% of delivered therapies in the studies were a response to device proarrhythmia, he said, "people still lived longer by over 25%."

Germano and associates agree that there are big differences among development of arrhythmias, device-delivered therapy, and aborted sudden death. They describe two studies in which more than a third of detected VF episodes terminated spontaneously before ICDs could discharge and how variations in device programming parameters can profoundly influence whether treatment is actually delivered. But such explanations, they write, don't go far enough to explain all of the apparent arrhythmia-rate discrepancy between device and control groups. "The occurrence of appropriate therapies and effective shocks does not constitute proof that ICDs are superior to alternative management strategies, because not all VT and VF treated by ICDs would have resulted in sudden death."

"We believe that these data have . . . practical implications for ICD programming," according to the authors. For patients with no history of symptoms from slow VT, they write, detection intervals can be optimized to avoid its treatment. Antitachycardia pacing should be used to reduce shock rates. And some inappropriate therapies, according to the group, may be prevented by the use of dual-chamber devices and those that can distinguish supraventricular tachycardia from more threatening arrhythmias.