Definitions and endpoints for device clinical trials in atrial fibrillation--a pressing need.

Abstract

A new generation of implantable antiarrhythmic devices suitable for use in patients with atrial ~brillation (AF) is now in clinical trials [1,2]. Prior efforts to apply device therapy to supraventricular arrhythmias were con~ned to patients with paroxysmal supraventricular tachycardia and atrial _utter [3–5]. While important concepts with respect to arrhythmia detection, preventative pacing, and pacing therapy for tachycardia termination were developed, the technique was rapidly supplanted by catheter ablation methods. Device clinical trials in AF faded from view and discussion of the endpoints of such studies no longer appeared germane to the development of investigative efforts. This was indeed shortsighted because such endpoints were as relevant to radiofrequency catheter ablation as implantable device therapy [6,7]. The recent and expanding interest in drug and device clinical trials in patients with AF is forcing rethinking in this area. In cardiac pacing trials assessing the impact of pacing mode on patient survival in the sick sinus syndrome, a secondary endpoint is the recorded occurrence of AF [8]. Further considerations arose in the development of primary AF trials such as the Atrial Fibrillation Follow-Up: Investigation of Rhythm Management (AFFIRM) [9]. Simple yet critical issues, such as de~ning an episode of AF (as diffrentiated from atrial tachycardia or atrial _utter), arose. The debate over classi~cation of AF based on arrhythmia recurrence patterns continues, with even terms such as paroxysmal and chronic lacking consistent de~nitions [10]. The limitations of the surface ECG in de~ning the mechanism of underlying atrial arrhythmia have become increasingly obvious in clinical electrophysiologic studies, but methods to reclassify these rhythms are not readily apparent [11]. The clinical impact of the duration of a single AF episode is also unclear. While it is known that patient symptoms can emerge immediately at onset or after a highly variable interval, the occurrence of arrhythmia complications such as thrombosis and embolism are even less well de~ned. The AFFIRM investigators chose 1 hour of electrocardiographic AF as their de~nition. In early data from the Assessment of Cardioversion Using Transesophageal Echocardiography [ACUTE] study, speci~c echocardiographic abnormalities predict higher thromboembolic event rates but do not de~ne it in terms of AF duration [12]. For many of our patients with implantable devices, it is possible to recognize episodes of exact duration in minutes or even seconds with device datalogs. In the Dual Site Atrial Pacing study for Prevention of Atrial Fibrillation [DAPPAF], AF episodes are required to be symptomatic and to have electrocardiographic documentation. The former feature and its impact on AF duration are unclear, and the latter is highly dependent on the availability of device electrogram logs. With the availability of such logs, instantaneous electrocardiographic corroboration exists, whereas in its absence access to surface electrocardiogram recordings in_uences the trial endpoint. A standard is needed for each of these methodologic aspects of AF trials. De~ning primary trial endpoints that are clinical relevant is another challenge in such trials. In AFFIRM, which includes patients with new-onset AF of both paroxysmal or chronic patterns, the primary hypothesis focuses on the morbidity conferred by AF in terms of disabling stroke and death. Despite the anticipated relative risk reduction of 30%, the low annual event rate expected in such a study mandates a large study sample of 5000 patients followed for several years. In sick sinus syndrome populations, trials such as the Pacemaker Selection in the Elderly study [PASE] and the Mode Optimization Survival Trial [MOST] evaluated the effect on AF as a secondary endpoint. In such trials, this is described largely in terms of AF incidence and the recorded occurrence of stroke. Mortality endpoints are still the primary endpoints, but the dominant mechanism of bene~t, if any, is hypothesized to be due to improvement in ventricular function rather than AF prevention. Due to slightly higher event rates, the sample size may be anticipated to be somewhat smaller but may still involve up to 2000 patients being followed for an intermediate period. In contrast, in studies of paroxysmal AF such as DAPPAF, quantitative prevention of high-density AF events and symptomatic bene~t as assessed by quality-of-life measures form the dominant endpoints. The high-density event rate limits sample size to 120 patients and follow-up duration to 18 months. An even briefer period is required in a mode comparison study such as SYMBIAPACE, with 3month active treatment arms. Finally, a device trial designed for regulatory evaluation of device performance may opt simply for demonstration of arrhythmia detection and reversion over a 6-month follow-up period. Important questions remain regarding studies using survival endpoints, arrhythmia event rates, and quality-of-life measures. While the best projections of mortality rates are based on other large AF trials and Journal of Interventional Cardiac Electrophysiology 1997;1:173–174 The Editor’s Forum © Kluwer Academic Publishers. Boston. Printed in U.S.A.