- Email: [email protected]
Rhythm control in atrial fibrillation: endpoints for device-based trials
Heart Rhythm (2004) 1, B52–B57
www.heartrhythmjournal.com
Rhythm control in atrial fibrillation: Endpoints for device-based trials Anne M. Gillis, MD From the Cardiovascular Research Group, the University of Calgary, and Division of Cardiology, Calgary Health Region, Calgary, Alberta, Canada. KEYWORDS Atrial fibrillation; Clinical outcomes; Pacemaker diagnostics
Newer generations of pacemakers and implantable cardioverter defibrillators have enhanced diagnostic features that permit detection and storage of information about the frequency, duration, time of onset, and rate characteristics of multiple consecutive episodes of atrial tachycardia and atrial fibrillation (AF). Parameters available in implantable devices that might serve as outcome measures to assess the impact of pharmacologic and nonpharmacologic therapies for prevention of AF include time to first recurrence of AF, intervals between successive episodes of AF, AF frequency, AF duration, AF burden (amount/ unit time), and episodes of symptomatic AF. Each parameter has advantages and limitations. The characteristics/temporal patterns of AF may determine the ideal study outcome measure to address a specific hypothesis. Clinical studies of therapies for prevention of AF need to be directed to populations with the arrhythmia substrate that will be targeted by the therapy. Device-based metrics are suitable to assess the impact of pacing or pharmacologic therapies for prevention of AF. However, it is important to establish that suppression of AF is associated with an improvement in some clinical outcome. Relevant outcomes might include improvement in symptoms, quality of life, or functional capacity; reduction in hospitalization for cardiovascular causes, stroke death, or overall mortality; or a composite of some of these parameters. © 2004 Heart Rhythm Society. All rights reserved.
Atrial fibrillation (AF) occurs frequently in the pacemaker population, particularly in patients with sinus node dysfunction as the primary indication for pacing.1,2 Recent prospective randomized clinical trials have demonstrated that atrial-based pacing prevents or reduces the frequency of paroxysmal and persistent AF in the general pacemaker population and in patients with sinus node dysfunction.3–5 Newer pacemakers have incorporated novel atrial pacing algorithms designed to modify the atrial substrate that pre-
disposes to AF or to identify and suppress triggers of AF.6 –12 Some devices have incorporated atrial antitachycardia pacing (ATP) algorithms for termination of atrial tachycardias (ATs) that might degenerate into AF. A number of randomized clinical trials have tested these different algorithms, with seemingly divergent outcomes.7–12
Supported by the Canadian Institutes for Health Research and the Heart and Stroke Foundation of Alberta. Dr. Gillis is a Scientist of the Alberta Heritage Foundation for Medical Research. Address reprint requests and correspondence: Dr. Anne M. Gillis, Faculty of Medicine, University of Calgary, HSC Room 1634, 3330 Hospital Drive NW, Calgary, Alberta, Canada T2N 4N1. E-mail address: [email protected].
Newer generations of pacemakers and cardioverter defibrillators have enhanced diagnostic features that permit detection and storage of information about the frequency, duration, time of onset, and rate characteristics of multiple consecutive episodes of AT and AF. The specificity of the detection algorithms has been validated by review of intra-
Device AF diagnostics
1547-5271/$ -see front matter © 2004 Heart Rhythm Society. All rights reserved. doi:10.1016/j.hrthm.2004.03.075
Gillis
Rhythm Control in Atrial Fibrillation
Figure 1 Temporal patterns of atrial fibrillation (AF). Open bars represent time; solid bars represent episodes of AF in time. Some episodes of AF are sporadic, separated by years between recurrent events. Some episodes of AF are persistent, requiring pharmacologic or electrical cardioversion to restore sinus rhythm. Many episodes of AF are paroxysmal, spontaneously terminating but varying in duration and intervals between episodes.
atrial electrograms stored during many of these episodes.13–15 Some devices provide trend information on the frequency and amount of AF, expressed as %AF/day for up to 14 months. Because many episodes of AT/AF are not associated with clinical symptoms, the ability to reliably collect information on AF frequency, AF duration, and ventricular rate during AF over the long term has proven valuable for the study of temporal patterns of AT/AF recurrence2,16 –19 and for measuring clinical outcomes in randomized clinical trials in pacemaker and implantable defibrillator populations.1,8 –11
Metrics of rhythm control Temporal patterns of AF AF may be paroxysmal, characterized by episodes of spontaneously terminating AF of varying duration and varying time intervals between successive episodes. Left untreated, the natural progression of paroxysmal AF is for episodes to increase in frequency and duration over time and, in some, for AF to become more persistent, ie, fewer but longer episodes (Figure 1).2,17 The appropriate metric to assess the impact of a therapy for prevention of AF may be influenced by these different temporal patterns of AF.19 A number of parameters available in implantable devices might serve as outcome measures to assess the impact of pharmacologic and nonpharmacologic therapies for prevention of AF. These are summarized in Table 1.
Time to first recurrence of AF The time to first recurrence of symptomatic AF has frequently been used to assess antiarrhythmic drug efficacy, particularly in pharmacologic trials.20 –22 This endpoint as-
B53 sumes that recurrence rates are random and independent of a preceding event. However, episodes of paroxysmal AF frequently are nonrandomly distributed over time in patients who receive a pacemaker for treatment of symptomatic bradycardia in the setting of sinus node dysfunction.16 –18 Episodes of paroxysmal AF frequently occur in clusters, although the proportion of episodes that cluster may change significantly over time.16 The time to first recurrence of AF may provide some information about therapeutic efficacy. However, it is more likely that if intervals between successive episodes of AF prolong substantially or if episodes of AF shorten significantly, these measures may be linked to improved clinical outcomes. Thus, recurrence of one episode of AF should not necessarily be considered a failure of therapy. The time point of entry into a study may be critical to using the time to first event recurrence as an outcome measure. If the patient has been in the midst of a cluster of AF episodes, the probability of early AF recurrence is likely high. In the Atrial Pacing Peri-Ablation for Prevention of Atrial Fibrillation (PA3) study, time to first recurrence of AF based on device detection of AF was used as the primary outcome measure.23 Secondary outcomes included the interval between first and second episodes of AF, and AF burden defined as the quantity of AF in hours per day calculated from the duration of episodes of AF stored in the device counters. The interval between first and second episodes of AF was substantially shorter than the time to first AF consistent with clustering of AF23 (Figure 2).We have consistently observed this pattern in other patient populations, including those with sinus node dysfunction and patients with coexisting AF and ventricular tachycardia in the setting of left ventricular dysfunction.15,16 Given that electrical remodeling secondary to frequent AF occurs, it may be important to allow a therapy “run in” or “stabilization” phase to allow the effects of a therapy to be established in order to determine long-term treatment effects.
AF frequency AF frequency can be used as an index of therapy efficacy. In newer implantable pacemakers and defibrillators, it
Table 1 Potential surrogate endpoints available in implantable devices for atrial fibrillation therapy studies ● Time to AF ● Intervals between successive episodes of AF ● AF frequency ● AF duration ● AF burden (amount of AF in time or %/day) ● Symptomatic AF AF ⫽ atrial fibrillation.
B54
Figure 2 Time intervals to first and second recurrence of sustained atrial fibrillation (AF) detected using pacemaker diagnostics in the Atrial Pacing Peri-Ablation for Prevention of Atrial Fibrillation (PA3) study. Note that the interval between first and second episodes of AF was substantially shorter compared to the interval to first AF in both the atrial pacing and no pacing groups.
Heart Rhythm, Vol 1, No 2, July Supplement 2004 is possible to track changes in AF frequency over time. However, this parameter, used in isolation, is unlikely to be an adequate outcome measure because some patients may continue to experience frequent but shorter episodes of AF (Figure 3A), or, as AF becomes more persistent, the frequency may diminish but episodes may last substantially longer (Figure 3B). Furthermore, device-based counts of AF frequency may differ between devices because criteria for onset and termination of AF may differ. As well, transient undersensing during AF may occur because episodes fall in the device blanking period or because of reduction in signal amplitude resulting in inappropriate device classification of episode termination and detection of a new episode of AF when atrial electrograms are redetected. Another approach is to measure the frequency of episodes of symptomatic AF over time. This approach was used by the ADOPT (Atrial Dynamic Overdrive Pacing Trial) investigators.10 The number of days spent in symptomatic AF was compared in patients randomized to dynamic atrial overdrive therapy for suppression of AF to
Figure 3 A: Atrial fibrillation (AF) burden (hr/day) and AF frequency (episodes/day) over time in a patient who received a Medtronic AT501 dual-chamber pacemaker. Note that AF burden decreases following programming on atrial antitachycardia pacing therapies (ATP). Compared to AF burden measured 1 month after pacemaker implantation, AF burden decreased by 76%, whereas AF frequency increased by 171%. B: AF burden and frequency over time in a patient after implantation of a Medtronic AT501 pacemaker. Compared to AF burden measured 1 month after pacemaker implantation, AF burden increased over time by 79%, whereas AF frequency decreased by 77%, reflecting longer episodes of paroxysmal AF.
Gillis
Rhythm Control in Atrial Fibrillation
B55
Figure 4 Atrial fibrillation (AF) burden and frequency measured over time. Atrial antitachycardia pacing (ATP) was programmed ON 1 month after device implantation, and AF burden decreased from 46% to 16%. AF frequency was unchanged. Following addition of sotalol, further reductions in AF burden and frequency were maintained over long-term follow-up.
those randomized to no therapy following dual-chamber pacemaker implantation. The ADOPT investigators reported a modest reduction in symptomatic AF associated with this overdrive pacing algorithm and concluded that this therapy had efficacy for suppression of AF. The pacemakers implanted did not have very sophisticated atrial tachyarrhythmia detection and data storage features. However, the quantity of AF was estimated from the atrial high rate histogram counters. Using this metric, the amount of AF increased substantially over time in both groups, and overall AF was not reduced by the overdrive pacing algorithm. Some implantable devices have patient-activated data storage counters that permit tracking of symptomatic AF over time.
AF burden Monitoring the quantity of AF over time, ie, AF burden expressed as time per day, as an assessment of therapeutic efficacy has appeal. PA3 was the first study to use AF burden determined from device counters as a study outcome measure.23 In this population, AF burden increased significantly over time following AV junction ablation.24 The devices used in the PA3 study had limited memory and, in many patients with frequent AF, the counters usually were filled early during follow-up. Furthermore, intraatrial electrograms were not collected to enable more AF events to be stored in the device data counters. The expanded memory of newer devices, with the ability to store information on multiple consecutive episodes of AF and to verify appropriate classification of AF from review of intraatrial electrograms during episodes, has provided us with a powerful tool to assess the effects of drug therapy and newer pacing algorithms for prevention of AF. An example device diag-
nostics used to track response to atrial pacing therapy and antiarrhythmic drug therapy over time is shown in Figure 4.
Clinical trials using device metrics to assess AF prevention therapy A number of atrial pacing algorithms have been designed for the prevention of AF. Studies have been conducted to evaluate these algorithms, with divergent outcomes. The ADOPT investigators randomized 388 patients with sinus node dysfunction and paroxysmal AF to dual-chamber pacing versus dual-chamber pacing plus dynamic atrial overdrive (DAO) pacing.10 Patients were assessed 1, 3, and 6 months after pacemaker insertion. The investigators reported the DAO algorithm was associated with a very modest, but statistically significant, reduction in symptomatic AF during (2.6% in the control group vs .9% in the treatment group, P ⫽ 0.008). The absolute risk reduction of DAO pacing decreased over time (1.25% at 1 month vs 0.36% at 6 months). In the Atrial Septal Pacing Clinical Efficacy Trial (ASPECT), three atrial pacing algorithms were evaluated for prevention of AF.9 After a 1-month stabilization period, these therapies were programmed ON or OFF for 3 months, then patients were crossed over to the opposite pacing strategy and followed for another 3 months. Neither AF frequency nor AF burden was reduced by the combined pacing algorithms (atrial pacing preference, atrial rate stabilization, and post mode switch overdrive pacing) despite a significant reduction in supraventricular premature beat frequency.9 In the Pacing in Prevention of AF study, 192 patients with bradycardia and AF were randomized to a 6-month crossover trial of three atrial pacing algorithms for prevention of AF.11 The study endpoint, total mode switch
B56
Heart Rhythm, Vol 1, No 2, July Supplement 2004
duration, was similar when the pacing prevention algorithms were programmed ON (11.9 ⫾ 27.7 days) compared OFF (11.6 ⫾ 26.5, P ⫽ NS). The hypothesis that successful pace termination of atrial flutter/AT prevents the development of AF and reduces overall AF burden was tested in the Atrial Therapy Efficacy and Safety Trial (ATTEST).8 Patients (n ⫽ 370) with paroxysmal AF received a Medtronic AT500 and were randomized to DDDR pacing or DDDR pacing with atrial ATP therapies and atrial pace prevention therapies programmed ON. Over 3-month follow-up, ⬎15,000 episodes of AT were treated by atrial ATP therapies, and device-classified efficacy was 41%. However, the atrial ATP therapies in combination with the three atrial AF prevention therapies did not reduce AF burden. These studies suggest that selective atrial pacing algorithms designed for prevention of AF or termination of ATs have modest incremental benefit compared to AV sequential pacing for prevention of AF.
Figure 5 Atrial fibrillation (AF) burden and clinical outcomes. It will be important to demonstrate in clinical trials that suppression of AF is associated with improvement in one or more clinical outcomes.
Lessons learned from device clinical trials
death, or overall mortality; or a composite of some of these parameters.
Many patients do not experience AF following pacemaker implantation. Accordingly, future trial designs should identify patients with AF following device implantation prior to enrollment into studies assessing the effects of a new AF therapy. The frequency, duration, and overall burden of AF vary substantially among individual patients. Accordingly, patient inclusion criteria or stratification criteria based on AF characteristics should be considered in future clinical criteria assessing the effects of a new AF therapy. The electrophysiologic substrate for AF likely differs among patients. Frequency of AF, prior duration of AF (years since first onset), history of persistent AF requiring cardioversion, coexisting heart disease, and coexisting drug therapy may influence outcomes. The design of clinical trials evaluating device therapies should be based on presumed mechanism of effect, and patient inclusion criteria should be fairly rigid to identify and include patients with the presumed AF substrate that will be target by the therapy. Future trial designs should consider potential proarrhythmic effects of new therapies and consider methods to identify therapy responders and exclude or minimize proarrhythmic potential.
Are device-based metrics adequate to assess AF therapy efficacy? Device-based metrics are suitable to assess the impact of pacing or pharmacologic therapies for prevention of AF. However, it is important to establish that suppression of AF is associated with an improvement in some clinical outcome (Figure 5).Relevant outcomes might include improvement in symptoms, quality of life, or functional capacity; reduction in hospitalization for cardiovascular causes, stroke
References 1. Gillis AM. Clinical trials of pacing for maintenance of sinus rhythm. J Interv Card Electrophysiol 2004;10(Suppl 1):55– 62. 2. Gillis AM, Morck M. Atrial fibrillation following DDDR pacemaker implantation. J Cardiovasc Electrophysiol 2002;13:542–547. 3. Andersen HR, Nielsen JC, Thomsen PE, Thuesen L, Mortensen PT, Vesterlund T, Pedersen AK. Long-term follow-up of patients from a randomized trial of atrial versus ventricular pacing for sick-sinus syndrome. Lancet 1997;350:1210 –1216. 4. Connolly SJ, Kerr CR, Gent M, Roberts RS, Yusef S, Gillis AM, Sami MH, Talajic M, Tang AS, Klein GJ, Lau C, Newman DM, for the CTOPP Investigators. Effects of physiologic pacing versus ventricular pacing on the risk of stroke and death due to cardiovascular causes. N Engl J Med 2000;342:1385–1391. 5. Lamas GA, Lee KL, Sweeney MO, Silverman R, Leon A, Yee R, Marinchak RA, Flaker G, Schron E, Orav EJ, Hellkamp AS, Greer S, McAnulty J, Ellenbogen K, Ehlert F, Freedman RA, Estes NA 3rd, Greenspon A, Goldman L. Ventricular pacing or dual-chamber pacing for sinus-node dysfunction. N Engl J Med 2002;346:1854 –1862. 6. Gillis AM, Morck M, Fitts S. Antitachycardia pacing therapies and arrhythmia monitoring diagnostics for the treatment of atrial fibrillation. Can J Cardiol 2002;18:992–995. 7. Israel CW, Hugl B, Unterberg C, Lawo T, Kennis I, Hettrick D, Hohnloser SH. Pace-termination and pacing for prevention of atrial tachyarrhythmias: results from a multicenter study with an implantable device for atrial therapy. J Cardiovasc Electrophysiol 2001;12:1121–1128. 8. Lee MA, Weachter R, Pollak S, Kremers MS, Naik AM, Silverman R, Tuzi J, Wang W, Johnson LJ, Euler DE, ATTEST Investigators. The effect of atrial pacing therapies on atrial tachyarrhythmia burden and frequency: results of a randomized trial in patients with bradycardia and atrial tachyarrhythmias. J Am Coll Cardiol 2003;41:1926 –1932. 9. Padeletti L, Purerfellner H, Adler SW, Waller TJ, Harvey M, Horvitz L, Hollbrook R, Kempen K, Mugglin A, Hettrick DA, Worldwide ASPECT Investigators. Combined efficacy of atrial septal lead placement and atrial pacing algorithms for prevention of paroxysmal atrial tachyarrhythmia. J Cardiovasc Electrophysiol 2003;14:1189 –1195.
Gillis
Rhythm Control in Atrial Fibrillation
10. Carlson MD, Ip J, Messenger J, Beau S, Kalbfleisch S, Gervais P, Cameron DA, Duran A, Val-Mejias J, Mackall J, Gold M, Atrial Dynamic Overdrive Pacing Trial (ADOPT) Investigators. A new pacemaker algorithm for the treatment of atrial fibrillation: results of the Atrial Dynamic Overdrive Pacing Trial (ADOPT). J Am Coll Cardiol 2003;42:627– 633. 11. Mabo P, Funck R, De Roy L, Poezevara Y, Anselme F, Di Donna P. Impact of ventricular pacing on atrial fibrillation prevention. Pacing Clin Electrophysiol 2002;24:II– 621. 12. Camm J, for the AF Therapy Investigators. AF therapy study: prevention pacing for paroxysmal atrial fibrillation. Pacing Clin Electrophysiol 2002;24:554. 13. Fitts SM, Hill MRS, Mehra R, Gillis AM. High rate atrial tachyarrhythmia detections in implantable pulse generators: low incidence of false positive detections. Pacing Clin Electrophysiol 2000;23:1080–1086. 14. Purerfellner H, Gillis AM, Holbrook R, Hettrick DA. Accuracy of atrial tachyarrhythmia detection in implantable devices with arrhythmia therapies. Pacing Clin Electrophysiol (in press). 15. Gillis AM, Unterberg-Buchwald C, Schmidinger H, Massimo S, Wolfe K, Kavaney DJ, Otterness MF, Hohnloser SH, for the GEM III AT Worldwide Investigators. Safety and efficacy of advanced atrial pacing therapies for atrial tachyarrhythmias in patients with a new implantable dual chamber cardioverter-defibrillator. J Am Coll Cardiol 2002;40: 1653–1659. 16. Rose MS, Gillis AM, Sheldon RS. Evaluation of the bias in using the time to first event when the inter-event intervals have a Weibull distribution. Stat Med 1999;18:139 –154. 17. Gillis AM, Rose MS. Temporal patterns of paroxysmal atrial fibrillation following DDDR pacemaker implantation. Am J Cardiol 2000; 85:1445–1450. 18. Kaemmerer WF, Rose S, Mehra R. Distribution of patients’ paroxysmal atrial tachyarrhythmia episodes: implications for detection of treatment efficacy. J Cardiovasc Electrophysiol 2001;12:121–130. 19. Wyse DG. Selection of endpoints in atrial fibrillation studies. J Cardiovasc Electrophysiol 2002;13:S47–S52. 20. Clair WK, Wilkinson WE, McCarthy EA, Page RL, Pritchett EL. Spontaneous occurrence of symptomatic paroxysmal atrial fibrillation and paroxysmal supraventricular tachycardia in untreated patients. Circulation 1993;87:1114 –1122. 21. Roy D, Talajic M, Dorian P, Connolly S, Eisenberg MJ, Green M, Kus T, Lambert J, Dubuc M, Gagne P, Nattel S, Thibault B. Amiodarone to prevent recurrence of atrial fibrillation. Canadian Trial of Atrial Fibrillation Investigators. N Engl J Med 2000;342:913–920. 22. Page RL, Tilsch TW, Connolly SJ, Schnell DJ, Marcello SR, Wilkinson WE, Pritchett EL, Azimilide Supraventricular Arrhythmia Program (ASAP) Investigators. Asymptomatic or “silent” atrial fibrillation: frequency in untreated patients and patients receiving azimilide. Circulation 2003;107:1141–1145. 23. Gillis AM, Wyse DG, Connolly SJ, Dubuc M, Philippon F, Yee R, Lacombe P, Rose MS, Kerr CD. Atrial pacing periablation for prevention of paroxysmal atrial fibrillation. Circulation 1999;99:2553–2558. 24. Gillis AM, Connolly SJ, Lacombe P, Philippon F, Dubuc M, Kerr CR, Yee R, Rose MS, Newman D, Kavanagh KM, Gardner MJ, Kus T, Wyse DG. Randomized crossover comparison of DDDR versus VDD pacing after atrioventricular junction ablation for prevention of atrial fibrillation. The atrial pacing peri-ablation for paroxysmal atrial fibrillation (PA3) study investigators. Circulation 2000;102:736 –741.
Questions and Answers Newman: AF burden measures are obviously useful, but they may not easily correlate with clinical measures, and I am not sure if we necessarily yet know if it correlates with symptoms. I think we all agree that it may be useful for an
B57 initial proof of concept, but at the end of the day, if it does not help symptoms, it does not translate to measures related to efficacy, do you think it is going to be something useful? Gillis: It is a useful metric to measure the response of the therapy, but as I tried to point out, you then have to link it to some relevant outcome. Crijns: Two points. First, burden combines both frequency and duration of the arrhythmias. If you want to tease out why does the treatment work or why does the treatment fail, you have to differentiate between them. Second, for patient selection, I think we are primarily talking about paroxysmal AF patients, or do you also suggest the use of burden on persistent patients? Gillis: The major focus of the device-based therapies and endpoints right now is geared to patients with paroxysmal AF and the ability to prevent the development of persistent AF in these patients. We are currently using previous cardioversion as an exclusion criterion for some of our clinical trials. These patients may be less likely to respond to some of the therapies. Crijns: Could you envision that before implanting a device, you would do an EP (electrophysiologic) test to look at some measure of prevention or some measure of termination? Gillis: I do not at this point in time because I do not know what I would look for. I guess one could try to induce atrial flutter or AT as an indicator of patients who may respond to ATP. If you are looking at a therapy that only has the ability to terminate atrial tachyarrhythmias, then you need to identify the patient population that has those arrhythmias so that they will enjoy the benefit of that therapy. It is clear that none of the pacing therapies that we have terminates AF, but we do recognize that patients transition back and forth between atrial fibrillation and atrial tachycardia. Mehra: I am curious about the percentage of patients who have AF and flutter. Ed mentioned 15% earlier, what is the perception around the room? Gillis: In our population, I think it is around about 30%. Wyse: I do not have a clue. My question/comment about this whole thing that is much more interesting to me than what the therapies do is what we are learning from just monitoring these patients. I am confused about a lot of things that I thought I knew. I still think that for a lot of patients, true AF gets more frequent and it gets longer in duration. However, some of the data suggested that this might not be true for all patients. Zipes: Some of the apparent paradoxes of AF regression may relate to medications given for heart failure. We think that these medications are not really impacting atrial fibrillation and yet they may well be. Gillis: We have actually looked at that in our patient population this summer and, much to my surprise, I did not see a difference in AF burden. Future AF studies should ensure that patients are treated with appropriate therapeutic doses of ACE (angiotensin-converting enzyme) inhibitors. You could be really aggressive about blood pressure control to see if that is of benefit as well.
www.heartrhythmjournal.com
Rhythm control in atrial fibrillation: Endpoints for device-based trials Anne M. Gillis, MD From the Cardiovascular Research Group, the University of Calgary, and Division of Cardiology, Calgary Health Region, Calgary, Alberta, Canada. KEYWORDS Atrial fibrillation; Clinical outcomes; Pacemaker diagnostics
Newer generations of pacemakers and implantable cardioverter defibrillators have enhanced diagnostic features that permit detection and storage of information about the frequency, duration, time of onset, and rate characteristics of multiple consecutive episodes of atrial tachycardia and atrial fibrillation (AF). Parameters available in implantable devices that might serve as outcome measures to assess the impact of pharmacologic and nonpharmacologic therapies for prevention of AF include time to first recurrence of AF, intervals between successive episodes of AF, AF frequency, AF duration, AF burden (amount/ unit time), and episodes of symptomatic AF. Each parameter has advantages and limitations. The characteristics/temporal patterns of AF may determine the ideal study outcome measure to address a specific hypothesis. Clinical studies of therapies for prevention of AF need to be directed to populations with the arrhythmia substrate that will be targeted by the therapy. Device-based metrics are suitable to assess the impact of pacing or pharmacologic therapies for prevention of AF. However, it is important to establish that suppression of AF is associated with an improvement in some clinical outcome. Relevant outcomes might include improvement in symptoms, quality of life, or functional capacity; reduction in hospitalization for cardiovascular causes, stroke death, or overall mortality; or a composite of some of these parameters. © 2004 Heart Rhythm Society. All rights reserved.
Atrial fibrillation (AF) occurs frequently in the pacemaker population, particularly in patients with sinus node dysfunction as the primary indication for pacing.1,2 Recent prospective randomized clinical trials have demonstrated that atrial-based pacing prevents or reduces the frequency of paroxysmal and persistent AF in the general pacemaker population and in patients with sinus node dysfunction.3–5 Newer pacemakers have incorporated novel atrial pacing algorithms designed to modify the atrial substrate that pre-
disposes to AF or to identify and suppress triggers of AF.6 –12 Some devices have incorporated atrial antitachycardia pacing (ATP) algorithms for termination of atrial tachycardias (ATs) that might degenerate into AF. A number of randomized clinical trials have tested these different algorithms, with seemingly divergent outcomes.7–12
Supported by the Canadian Institutes for Health Research and the Heart and Stroke Foundation of Alberta. Dr. Gillis is a Scientist of the Alberta Heritage Foundation for Medical Research. Address reprint requests and correspondence: Dr. Anne M. Gillis, Faculty of Medicine, University of Calgary, HSC Room 1634, 3330 Hospital Drive NW, Calgary, Alberta, Canada T2N 4N1. E-mail address: [email protected].
Newer generations of pacemakers and cardioverter defibrillators have enhanced diagnostic features that permit detection and storage of information about the frequency, duration, time of onset, and rate characteristics of multiple consecutive episodes of AT and AF. The specificity of the detection algorithms has been validated by review of intra-
Device AF diagnostics
1547-5271/$ -see front matter © 2004 Heart Rhythm Society. All rights reserved. doi:10.1016/j.hrthm.2004.03.075
Gillis
Rhythm Control in Atrial Fibrillation
Figure 1 Temporal patterns of atrial fibrillation (AF). Open bars represent time; solid bars represent episodes of AF in time. Some episodes of AF are sporadic, separated by years between recurrent events. Some episodes of AF are persistent, requiring pharmacologic or electrical cardioversion to restore sinus rhythm. Many episodes of AF are paroxysmal, spontaneously terminating but varying in duration and intervals between episodes.
atrial electrograms stored during many of these episodes.13–15 Some devices provide trend information on the frequency and amount of AF, expressed as %AF/day for up to 14 months. Because many episodes of AT/AF are not associated with clinical symptoms, the ability to reliably collect information on AF frequency, AF duration, and ventricular rate during AF over the long term has proven valuable for the study of temporal patterns of AT/AF recurrence2,16 –19 and for measuring clinical outcomes in randomized clinical trials in pacemaker and implantable defibrillator populations.1,8 –11
Metrics of rhythm control Temporal patterns of AF AF may be paroxysmal, characterized by episodes of spontaneously terminating AF of varying duration and varying time intervals between successive episodes. Left untreated, the natural progression of paroxysmal AF is for episodes to increase in frequency and duration over time and, in some, for AF to become more persistent, ie, fewer but longer episodes (Figure 1).2,17 The appropriate metric to assess the impact of a therapy for prevention of AF may be influenced by these different temporal patterns of AF.19 A number of parameters available in implantable devices might serve as outcome measures to assess the impact of pharmacologic and nonpharmacologic therapies for prevention of AF. These are summarized in Table 1.
Time to first recurrence of AF The time to first recurrence of symptomatic AF has frequently been used to assess antiarrhythmic drug efficacy, particularly in pharmacologic trials.20 –22 This endpoint as-
B53 sumes that recurrence rates are random and independent of a preceding event. However, episodes of paroxysmal AF frequently are nonrandomly distributed over time in patients who receive a pacemaker for treatment of symptomatic bradycardia in the setting of sinus node dysfunction.16 –18 Episodes of paroxysmal AF frequently occur in clusters, although the proportion of episodes that cluster may change significantly over time.16 The time to first recurrence of AF may provide some information about therapeutic efficacy. However, it is more likely that if intervals between successive episodes of AF prolong substantially or if episodes of AF shorten significantly, these measures may be linked to improved clinical outcomes. Thus, recurrence of one episode of AF should not necessarily be considered a failure of therapy. The time point of entry into a study may be critical to using the time to first event recurrence as an outcome measure. If the patient has been in the midst of a cluster of AF episodes, the probability of early AF recurrence is likely high. In the Atrial Pacing Peri-Ablation for Prevention of Atrial Fibrillation (PA3) study, time to first recurrence of AF based on device detection of AF was used as the primary outcome measure.23 Secondary outcomes included the interval between first and second episodes of AF, and AF burden defined as the quantity of AF in hours per day calculated from the duration of episodes of AF stored in the device counters. The interval between first and second episodes of AF was substantially shorter than the time to first AF consistent with clustering of AF23 (Figure 2).We have consistently observed this pattern in other patient populations, including those with sinus node dysfunction and patients with coexisting AF and ventricular tachycardia in the setting of left ventricular dysfunction.15,16 Given that electrical remodeling secondary to frequent AF occurs, it may be important to allow a therapy “run in” or “stabilization” phase to allow the effects of a therapy to be established in order to determine long-term treatment effects.
AF frequency AF frequency can be used as an index of therapy efficacy. In newer implantable pacemakers and defibrillators, it
Table 1 Potential surrogate endpoints available in implantable devices for atrial fibrillation therapy studies ● Time to AF ● Intervals between successive episodes of AF ● AF frequency ● AF duration ● AF burden (amount of AF in time or %/day) ● Symptomatic AF AF ⫽ atrial fibrillation.
B54
Figure 2 Time intervals to first and second recurrence of sustained atrial fibrillation (AF) detected using pacemaker diagnostics in the Atrial Pacing Peri-Ablation for Prevention of Atrial Fibrillation (PA3) study. Note that the interval between first and second episodes of AF was substantially shorter compared to the interval to first AF in both the atrial pacing and no pacing groups.
Heart Rhythm, Vol 1, No 2, July Supplement 2004 is possible to track changes in AF frequency over time. However, this parameter, used in isolation, is unlikely to be an adequate outcome measure because some patients may continue to experience frequent but shorter episodes of AF (Figure 3A), or, as AF becomes more persistent, the frequency may diminish but episodes may last substantially longer (Figure 3B). Furthermore, device-based counts of AF frequency may differ between devices because criteria for onset and termination of AF may differ. As well, transient undersensing during AF may occur because episodes fall in the device blanking period or because of reduction in signal amplitude resulting in inappropriate device classification of episode termination and detection of a new episode of AF when atrial electrograms are redetected. Another approach is to measure the frequency of episodes of symptomatic AF over time. This approach was used by the ADOPT (Atrial Dynamic Overdrive Pacing Trial) investigators.10 The number of days spent in symptomatic AF was compared in patients randomized to dynamic atrial overdrive therapy for suppression of AF to
Figure 3 A: Atrial fibrillation (AF) burden (hr/day) and AF frequency (episodes/day) over time in a patient who received a Medtronic AT501 dual-chamber pacemaker. Note that AF burden decreases following programming on atrial antitachycardia pacing therapies (ATP). Compared to AF burden measured 1 month after pacemaker implantation, AF burden decreased by 76%, whereas AF frequency increased by 171%. B: AF burden and frequency over time in a patient after implantation of a Medtronic AT501 pacemaker. Compared to AF burden measured 1 month after pacemaker implantation, AF burden increased over time by 79%, whereas AF frequency decreased by 77%, reflecting longer episodes of paroxysmal AF.
Gillis
Rhythm Control in Atrial Fibrillation
B55
Figure 4 Atrial fibrillation (AF) burden and frequency measured over time. Atrial antitachycardia pacing (ATP) was programmed ON 1 month after device implantation, and AF burden decreased from 46% to 16%. AF frequency was unchanged. Following addition of sotalol, further reductions in AF burden and frequency were maintained over long-term follow-up.
those randomized to no therapy following dual-chamber pacemaker implantation. The ADOPT investigators reported a modest reduction in symptomatic AF associated with this overdrive pacing algorithm and concluded that this therapy had efficacy for suppression of AF. The pacemakers implanted did not have very sophisticated atrial tachyarrhythmia detection and data storage features. However, the quantity of AF was estimated from the atrial high rate histogram counters. Using this metric, the amount of AF increased substantially over time in both groups, and overall AF was not reduced by the overdrive pacing algorithm. Some implantable devices have patient-activated data storage counters that permit tracking of symptomatic AF over time.
AF burden Monitoring the quantity of AF over time, ie, AF burden expressed as time per day, as an assessment of therapeutic efficacy has appeal. PA3 was the first study to use AF burden determined from device counters as a study outcome measure.23 In this population, AF burden increased significantly over time following AV junction ablation.24 The devices used in the PA3 study had limited memory and, in many patients with frequent AF, the counters usually were filled early during follow-up. Furthermore, intraatrial electrograms were not collected to enable more AF events to be stored in the device data counters. The expanded memory of newer devices, with the ability to store information on multiple consecutive episodes of AF and to verify appropriate classification of AF from review of intraatrial electrograms during episodes, has provided us with a powerful tool to assess the effects of drug therapy and newer pacing algorithms for prevention of AF. An example device diag-
nostics used to track response to atrial pacing therapy and antiarrhythmic drug therapy over time is shown in Figure 4.
Clinical trials using device metrics to assess AF prevention therapy A number of atrial pacing algorithms have been designed for the prevention of AF. Studies have been conducted to evaluate these algorithms, with divergent outcomes. The ADOPT investigators randomized 388 patients with sinus node dysfunction and paroxysmal AF to dual-chamber pacing versus dual-chamber pacing plus dynamic atrial overdrive (DAO) pacing.10 Patients were assessed 1, 3, and 6 months after pacemaker insertion. The investigators reported the DAO algorithm was associated with a very modest, but statistically significant, reduction in symptomatic AF during (2.6% in the control group vs .9% in the treatment group, P ⫽ 0.008). The absolute risk reduction of DAO pacing decreased over time (1.25% at 1 month vs 0.36% at 6 months). In the Atrial Septal Pacing Clinical Efficacy Trial (ASPECT), three atrial pacing algorithms were evaluated for prevention of AF.9 After a 1-month stabilization period, these therapies were programmed ON or OFF for 3 months, then patients were crossed over to the opposite pacing strategy and followed for another 3 months. Neither AF frequency nor AF burden was reduced by the combined pacing algorithms (atrial pacing preference, atrial rate stabilization, and post mode switch overdrive pacing) despite a significant reduction in supraventricular premature beat frequency.9 In the Pacing in Prevention of AF study, 192 patients with bradycardia and AF were randomized to a 6-month crossover trial of three atrial pacing algorithms for prevention of AF.11 The study endpoint, total mode switch
B56
Heart Rhythm, Vol 1, No 2, July Supplement 2004
duration, was similar when the pacing prevention algorithms were programmed ON (11.9 ⫾ 27.7 days) compared OFF (11.6 ⫾ 26.5, P ⫽ NS). The hypothesis that successful pace termination of atrial flutter/AT prevents the development of AF and reduces overall AF burden was tested in the Atrial Therapy Efficacy and Safety Trial (ATTEST).8 Patients (n ⫽ 370) with paroxysmal AF received a Medtronic AT500 and were randomized to DDDR pacing or DDDR pacing with atrial ATP therapies and atrial pace prevention therapies programmed ON. Over 3-month follow-up, ⬎15,000 episodes of AT were treated by atrial ATP therapies, and device-classified efficacy was 41%. However, the atrial ATP therapies in combination with the three atrial AF prevention therapies did not reduce AF burden. These studies suggest that selective atrial pacing algorithms designed for prevention of AF or termination of ATs have modest incremental benefit compared to AV sequential pacing for prevention of AF.
Figure 5 Atrial fibrillation (AF) burden and clinical outcomes. It will be important to demonstrate in clinical trials that suppression of AF is associated with improvement in one or more clinical outcomes.
Lessons learned from device clinical trials
death, or overall mortality; or a composite of some of these parameters.
Many patients do not experience AF following pacemaker implantation. Accordingly, future trial designs should identify patients with AF following device implantation prior to enrollment into studies assessing the effects of a new AF therapy. The frequency, duration, and overall burden of AF vary substantially among individual patients. Accordingly, patient inclusion criteria or stratification criteria based on AF characteristics should be considered in future clinical criteria assessing the effects of a new AF therapy. The electrophysiologic substrate for AF likely differs among patients. Frequency of AF, prior duration of AF (years since first onset), history of persistent AF requiring cardioversion, coexisting heart disease, and coexisting drug therapy may influence outcomes. The design of clinical trials evaluating device therapies should be based on presumed mechanism of effect, and patient inclusion criteria should be fairly rigid to identify and include patients with the presumed AF substrate that will be target by the therapy. Future trial designs should consider potential proarrhythmic effects of new therapies and consider methods to identify therapy responders and exclude or minimize proarrhythmic potential.
Are device-based metrics adequate to assess AF therapy efficacy? Device-based metrics are suitable to assess the impact of pacing or pharmacologic therapies for prevention of AF. However, it is important to establish that suppression of AF is associated with an improvement in some clinical outcome (Figure 5).Relevant outcomes might include improvement in symptoms, quality of life, or functional capacity; reduction in hospitalization for cardiovascular causes, stroke
References 1. Gillis AM. Clinical trials of pacing for maintenance of sinus rhythm. J Interv Card Electrophysiol 2004;10(Suppl 1):55– 62. 2. Gillis AM, Morck M. Atrial fibrillation following DDDR pacemaker implantation. J Cardiovasc Electrophysiol 2002;13:542–547. 3. Andersen HR, Nielsen JC, Thomsen PE, Thuesen L, Mortensen PT, Vesterlund T, Pedersen AK. Long-term follow-up of patients from a randomized trial of atrial versus ventricular pacing for sick-sinus syndrome. Lancet 1997;350:1210 –1216. 4. Connolly SJ, Kerr CR, Gent M, Roberts RS, Yusef S, Gillis AM, Sami MH, Talajic M, Tang AS, Klein GJ, Lau C, Newman DM, for the CTOPP Investigators. Effects of physiologic pacing versus ventricular pacing on the risk of stroke and death due to cardiovascular causes. N Engl J Med 2000;342:1385–1391. 5. Lamas GA, Lee KL, Sweeney MO, Silverman R, Leon A, Yee R, Marinchak RA, Flaker G, Schron E, Orav EJ, Hellkamp AS, Greer S, McAnulty J, Ellenbogen K, Ehlert F, Freedman RA, Estes NA 3rd, Greenspon A, Goldman L. Ventricular pacing or dual-chamber pacing for sinus-node dysfunction. N Engl J Med 2002;346:1854 –1862. 6. Gillis AM, Morck M, Fitts S. Antitachycardia pacing therapies and arrhythmia monitoring diagnostics for the treatment of atrial fibrillation. Can J Cardiol 2002;18:992–995. 7. Israel CW, Hugl B, Unterberg C, Lawo T, Kennis I, Hettrick D, Hohnloser SH. Pace-termination and pacing for prevention of atrial tachyarrhythmias: results from a multicenter study with an implantable device for atrial therapy. J Cardiovasc Electrophysiol 2001;12:1121–1128. 8. Lee MA, Weachter R, Pollak S, Kremers MS, Naik AM, Silverman R, Tuzi J, Wang W, Johnson LJ, Euler DE, ATTEST Investigators. The effect of atrial pacing therapies on atrial tachyarrhythmia burden and frequency: results of a randomized trial in patients with bradycardia and atrial tachyarrhythmias. J Am Coll Cardiol 2003;41:1926 –1932. 9. Padeletti L, Purerfellner H, Adler SW, Waller TJ, Harvey M, Horvitz L, Hollbrook R, Kempen K, Mugglin A, Hettrick DA, Worldwide ASPECT Investigators. Combined efficacy of atrial septal lead placement and atrial pacing algorithms for prevention of paroxysmal atrial tachyarrhythmia. J Cardiovasc Electrophysiol 2003;14:1189 –1195.
Gillis
Rhythm Control in Atrial Fibrillation
10. Carlson MD, Ip J, Messenger J, Beau S, Kalbfleisch S, Gervais P, Cameron DA, Duran A, Val-Mejias J, Mackall J, Gold M, Atrial Dynamic Overdrive Pacing Trial (ADOPT) Investigators. A new pacemaker algorithm for the treatment of atrial fibrillation: results of the Atrial Dynamic Overdrive Pacing Trial (ADOPT). J Am Coll Cardiol 2003;42:627– 633. 11. Mabo P, Funck R, De Roy L, Poezevara Y, Anselme F, Di Donna P. Impact of ventricular pacing on atrial fibrillation prevention. Pacing Clin Electrophysiol 2002;24:II– 621. 12. Camm J, for the AF Therapy Investigators. AF therapy study: prevention pacing for paroxysmal atrial fibrillation. Pacing Clin Electrophysiol 2002;24:554. 13. Fitts SM, Hill MRS, Mehra R, Gillis AM. High rate atrial tachyarrhythmia detections in implantable pulse generators: low incidence of false positive detections. Pacing Clin Electrophysiol 2000;23:1080–1086. 14. Purerfellner H, Gillis AM, Holbrook R, Hettrick DA. Accuracy of atrial tachyarrhythmia detection in implantable devices with arrhythmia therapies. Pacing Clin Electrophysiol (in press). 15. Gillis AM, Unterberg-Buchwald C, Schmidinger H, Massimo S, Wolfe K, Kavaney DJ, Otterness MF, Hohnloser SH, for the GEM III AT Worldwide Investigators. Safety and efficacy of advanced atrial pacing therapies for atrial tachyarrhythmias in patients with a new implantable dual chamber cardioverter-defibrillator. J Am Coll Cardiol 2002;40: 1653–1659. 16. Rose MS, Gillis AM, Sheldon RS. Evaluation of the bias in using the time to first event when the inter-event intervals have a Weibull distribution. Stat Med 1999;18:139 –154. 17. Gillis AM, Rose MS. Temporal patterns of paroxysmal atrial fibrillation following DDDR pacemaker implantation. Am J Cardiol 2000; 85:1445–1450. 18. Kaemmerer WF, Rose S, Mehra R. Distribution of patients’ paroxysmal atrial tachyarrhythmia episodes: implications for detection of treatment efficacy. J Cardiovasc Electrophysiol 2001;12:121–130. 19. Wyse DG. Selection of endpoints in atrial fibrillation studies. J Cardiovasc Electrophysiol 2002;13:S47–S52. 20. Clair WK, Wilkinson WE, McCarthy EA, Page RL, Pritchett EL. Spontaneous occurrence of symptomatic paroxysmal atrial fibrillation and paroxysmal supraventricular tachycardia in untreated patients. Circulation 1993;87:1114 –1122. 21. Roy D, Talajic M, Dorian P, Connolly S, Eisenberg MJ, Green M, Kus T, Lambert J, Dubuc M, Gagne P, Nattel S, Thibault B. Amiodarone to prevent recurrence of atrial fibrillation. Canadian Trial of Atrial Fibrillation Investigators. N Engl J Med 2000;342:913–920. 22. Page RL, Tilsch TW, Connolly SJ, Schnell DJ, Marcello SR, Wilkinson WE, Pritchett EL, Azimilide Supraventricular Arrhythmia Program (ASAP) Investigators. Asymptomatic or “silent” atrial fibrillation: frequency in untreated patients and patients receiving azimilide. Circulation 2003;107:1141–1145. 23. Gillis AM, Wyse DG, Connolly SJ, Dubuc M, Philippon F, Yee R, Lacombe P, Rose MS, Kerr CD. Atrial pacing periablation for prevention of paroxysmal atrial fibrillation. Circulation 1999;99:2553–2558. 24. Gillis AM, Connolly SJ, Lacombe P, Philippon F, Dubuc M, Kerr CR, Yee R, Rose MS, Newman D, Kavanagh KM, Gardner MJ, Kus T, Wyse DG. Randomized crossover comparison of DDDR versus VDD pacing after atrioventricular junction ablation for prevention of atrial fibrillation. The atrial pacing peri-ablation for paroxysmal atrial fibrillation (PA3) study investigators. Circulation 2000;102:736 –741.
Questions and Answers Newman: AF burden measures are obviously useful, but they may not easily correlate with clinical measures, and I am not sure if we necessarily yet know if it correlates with symptoms. I think we all agree that it may be useful for an
B57 initial proof of concept, but at the end of the day, if it does not help symptoms, it does not translate to measures related to efficacy, do you think it is going to be something useful? Gillis: It is a useful metric to measure the response of the therapy, but as I tried to point out, you then have to link it to some relevant outcome. Crijns: Two points. First, burden combines both frequency and duration of the arrhythmias. If you want to tease out why does the treatment work or why does the treatment fail, you have to differentiate between them. Second, for patient selection, I think we are primarily talking about paroxysmal AF patients, or do you also suggest the use of burden on persistent patients? Gillis: The major focus of the device-based therapies and endpoints right now is geared to patients with paroxysmal AF and the ability to prevent the development of persistent AF in these patients. We are currently using previous cardioversion as an exclusion criterion for some of our clinical trials. These patients may be less likely to respond to some of the therapies. Crijns: Could you envision that before implanting a device, you would do an EP (electrophysiologic) test to look at some measure of prevention or some measure of termination? Gillis: I do not at this point in time because I do not know what I would look for. I guess one could try to induce atrial flutter or AT as an indicator of patients who may respond to ATP. If you are looking at a therapy that only has the ability to terminate atrial tachyarrhythmias, then you need to identify the patient population that has those arrhythmias so that they will enjoy the benefit of that therapy. It is clear that none of the pacing therapies that we have terminates AF, but we do recognize that patients transition back and forth between atrial fibrillation and atrial tachycardia. Mehra: I am curious about the percentage of patients who have AF and flutter. Ed mentioned 15% earlier, what is the perception around the room? Gillis: In our population, I think it is around about 30%. Wyse: I do not have a clue. My question/comment about this whole thing that is much more interesting to me than what the therapies do is what we are learning from just monitoring these patients. I am confused about a lot of things that I thought I knew. I still think that for a lot of patients, true AF gets more frequent and it gets longer in duration. However, some of the data suggested that this might not be true for all patients. Zipes: Some of the apparent paradoxes of AF regression may relate to medications given for heart failure. We think that these medications are not really impacting atrial fibrillation and yet they may well be. Gillis: We have actually looked at that in our patient population this summer and, much to my surprise, I did not see a difference in AF burden. Future AF studies should ensure that patients are treated with appropriate therapeutic doses of ACE (angiotensin-converting enzyme) inhibitors. You could be really aggressive about blood pressure control to see if that is of benefit as well.