Reactive atrial‐based antitachycardia pacing therapy reduces atrial tachyarrhythmias

Abstract Background Reactive atrial‐based antitachycardia pacing (rATP) aims to terminate atrial tachyarrhythmia/atrial fibrillation (AT/AF) episodes when they spontaneously organize to atrial flutter or atrial tachycardia; however, its effectiveness in the real‐world has not been studied. We used a large device database (Medtronic CareLink, Medtronic, Minneapolis, MN, USA) to evaluate the effects of rATP at reducing AT/AF. Methods Pacemaker, defibrillator, and resynchronization device transmission data were analyzed. Eligible patients had device detected AT/AF during a baseline period but were not in persistent AT/AF immediately preceding first transmission. Note that 1:1 individual matching between groups was conducted using age, sex, device type, pacing mode, AT/AF, and percent ventricular pacing at baseline. Risks of AT/AF events were compared between patients with rATP‐enabled versus control patients with rATP‐disabled or not available in the device. For matched patients, AT/AF event rates at 2 years were estimated by Kaplan‐Meier method, and hazard ratios (HRs) were calculated by Cox proportional hazard models. Results Of 43,440 qualifying patients, 4,203 had rATP on. Matching resulted in 4,016 pairs, totaling 8,032 patients for analysis. The rATP group experienced significantly lower risks of AT/AF events lasting ≥1 day (HR 0.81), ≥7 days (HR 0.64), and ≥30 days (HR 0.56) compared to control (P < 0.0001 for all). In subgroup analysis, rATP was associated with reduced risks of AT/AF events across age, sex, device type, baseline AT/AF, and preventive atrial pacing. Conclusions Among real‐world patients from a large device database, rATP therapy was significantly associated with a reduced risk of AT/AF. This association was independent of whether the patient had a pacemaker, defibrillator, or resynchronization device.

AF is irregular, typically originates from the pulmonary veins, and, as such, requires cardioversion to terminate persistent episodes. AF is not susceptible to pace-termination; however, slower organized rhythms such as atrial flutter or atrial tachycardia can often be terminated by antitachycardia pacing (ATP). 8 Atrial tachyarrhythmia (AT/AF) episodes can spontaneously transition between AF and atrial flutter/tachycardia, even in patients with chronic AF. 9 Reactive atrialbased ATP therapy (rATP, Medtronic, Minneapolis, MN, USA), found in cardiac implantable electronic devices, aims to terminate AT/AF episodes when they spontaneously organize to atrial flutter or to atrial tachycardia, thereby slowing the AT/AF disease progression from paroxysmal to persistent and permanent forms. The randomized MIN-ERVA trial was the first study to test the effects of rATP, and reported that use of rATP in patients with pacemakers and a history of paroxysmal or persistent AT/AF were associated with a lower incidence of progression to persistent and permanent AF. 10 In addition, use of rATP reduced early recurrence of AF and was associated with more frequent positive atrial remodeling (>10% reduction in left atrial diameter). 11 AF-related hospitalization, AF-related emergency department visits, and cardioversions were also significantly reduced. 10,12 However, MINERVA's population was limited to pacemaker recipients without complete atrioventricular block. Thus, we set out to study the realworld effectiveness of rATP at reducing AT/AF occurrence and subsequently preventing AT/AF progression in patients with pacemakers, implantable cardioverter defibrillators (ICDs), and cardiac resynchronization (CRT) devices using a large nationwide device database.

Study design and database
The study was a retrospective cohort assessment designed to evaluate the effectiveness of rATP therapy at reducing occurrence of AT/AF and subsequently preventing progression of AT/AF in patients with pacemakers, ICDs, and CRT devices. Risks of paroxysmal, persistent, and longstanding-persistent AT/AF events based on device detected daily AT/AF burden were compared between patients with rATP-enabled versus patients with rATP-disabled or not available in the device. Transmission data were extracted from the deidentified CareLink database on August 23, 2016. The database stores pacemaker, ICD, CRT-defibrillator, and CRT-pacemaker data transmitted via the Medtronic remote monitoring server and includes data on patient age, sex, implant date, device make and model, and data parameters (e.g., rATP, pacing mode, preventive atrial pacing), and diagnostics and episodes from the device. Diagnostic data used in this study were daily AT/AF burden and percentage ventricular pacing (VP). These data were continuous as the devices maintain up to 14 months of storage capacity. The device-reported AT/AF burden is determined in a consistent manner across devices and is highly accurate, where appropriate detection of AT/AF episodes and duration sensitivity are >95%, regardless of right atrial lead position. 13 Thus, adjudication of AT/AF episodes was deemed unnecessary. The CareLink database does not collect clinical parameters outside of the aforementioned items (e.g., medications, comorbidities, treatments, outcomes).

Patients
Patients were ≥18 years old and implanted with dual-chamber pacemakers, dual-chamber ICDs, CRT-pacemakers, and CRT-defibrillators.
Devices implanted from January 1, 2012 or later were included, and only data from the patient's first device was used. The date of first CareLink transmission is called Day 0. We use "the first transmission" and "Day 0" interchangeably in this report. In an effort to align with the MINERVA design, 14 eligible patients had a history of AT/AF at baseline but were not in persistent AT/AF immediately preceding Day 0. Specifically, patients had to have at least one day with 5 minutes or more of AT/AF during the baseline period of up to 1 year preceding Day 0 but were excluded if they had >23 hours of AT/AF consecutively from Day minus 7 through Day 0. Further, eligible patients had to have transmitted data beyond Day 0 for up to 2 years. Patients were identified as a rATP subject if they had the rATP feature turned on for all transmissions, or they were identified as the control subjects if they had rATP off for all transmissions or were implanted with devices not capable of delivering atrial-ATP.
To control for imbalances between the two groups, we performed individual matching using baseline data that are known risk factors for AT/AF development. Patients were matched 1:1 for age, sex, device type, pacing mode, percentage VP, and AT/AF at baseline. Patients implanted with CRT devices were also matched on programming of the AdaptivCRT feature.

Mechanism of rATP
The rATP feature is a second-generation algorithm designed to opportunistically terminate AT/AF rhythms when they spontaneously organize to atrial flutter or to an atrial tachycardia, even in persistent episodes. Upon detection of an AT/AF episode, the device monitors rhythm transitions based on regions of atrial cycle lengths and regularity. Up to three ATP pacing therapies (Ramp or Burst+) are delivered for each region that the episode may transition to. Cycle length regions are 50-ms wide for regular rhythms and 100-150-ms wide for irregular rhythms. Furthermore, if the AT/AF episode persists, a time interval feature allows for counters to reset; thus, ATP may be reapplied at programmable durations (e.g., every 7 hours). While we acknowledge that ATP cannot terminate true AF, the strength of the algorithm lies in its ability to redeliver ATP when it detects a change in the patient's rhythm to a moment when the rhythm might be more vulnerable to termination with ATP (e.g., atrial flutter). Performance of rATP, as reported by Padeletti et al., demonstrated a median efficacy of 44.4%, which was higher in episodes of long cycle length and regular rhythm and in episodes with many rhythm transitions. 15

Endpoints
The aim of the study was to demonstrate that rATP was associated with a lower risk of AT/AF events, including paroxysmal, persistent, and longstanding-persistent AT/AF. We used the same AT/AF endpoints from the randomized MINERVA trial. 10 They were time to first AT/AF event lasting ≥1 day (paroxysmal), ≥7 consecutive days (persistent), and ≥30 consecutive days (longstanding-persistent). All the AT/AF events were based on device-detected daily AT/AF burden. In addition, days with >23 hours of AT/AF burden were counted and classified within every 30 days from Day minus 30 to up to 2 years of follow-up, and the effects of rATP on this longitudinal discrete endpoint were evaluated. ATP efficacy was also determined in the rATP group using AT/AF episode data.

Statistical analysis
Individual matching was conducted through a Greedy algorithm. This process ensured that, for each rATP subject, the matched control had For events lasting more than one day, the first day was used as the date of occurrence. Event rates at 2 years were estimated by Kaplan-Meier method.
Effects of rATP on each time-to-event endpoint were evaluated using Cox proportional hazard models. Covariates considered in the models were group, age (<65 and ≥65 years), sex, device type, AT/AF were used. A P-value of < 0.05 was considered statistically significant.
rATP was enabled in 4,203 (9.7%) of the patients.
Baseline characteristics that were considered in the individual matching are shown in Table 1 for the primary analyses.
The individual matching resulted in nearly identical characteristics of matching factors between the matched patient groups (Table 1) Table 1). During the up to 1-year baseline period that preceded the first transmission, 7.9% and 7.5% of patients had >23 hours of daily AT/AF burden lasting for at least 7 and 30 consecutive days, respectively. Differences after matching were observed in the AT/AF burden from Day minus 30 to Day 0 (87.8% rATP vs 85.1% control with 0 days of >23 hours of AT/AF burden), duration from implant to Day 0 (median: 5.9 and 4.3 months in rATP and control subjects, respectively), and preventive atrial pacing therapies at Day 0 (9.3%, 9.7%, and 15.9% in rATP subjects vs 2.7%, 2.0%, and 8.3% in control subjects with APP, ARS, and PMOP being programmed on, respectively).

Time to AT/AF events
The incidence of AT/AF events were significantly reduced in the rATP group compared to control after controlling for other covariates in the main effect Cox models (P < 0.0001 for all,  Table 2). In the final Cox models, rATP effect remained highly significantly (P < 0.0001 for all, data not shown), while ARS and PMOP were not significant and hence removed from the final models and APP was slightly significant only in the final model for AF/AF lasting ≥1 day (P = 0.0469, data not shown).
Patients with rATP had an absolute reduction in the rates of AT/AF events lasting ≥1 day, ≥7 days, and ≥30 days at 2 years by 4.6%, 8.5%, and 7.9%, respectively. In Figure 1 Table S4).

ATP efficacy
In the rATP group, 198,838 out of 451,235 (44.1%) AT/AF episodes were terminated successfully by the last AT/AF therapy; the GEE estimated ATP efficacy on successfully terminating AT/AF episodes was 39.0% (95% CI: 38.1%-39.9%).

Case example
A 48-year-old male was implanted with a Medtronic Protecta XT dualchamber ICD in February 2012; the device was programmed to MVP mode and rATP was enabled. In April 2013, the patient experienced an AT/AF episode (Figure 4). At onset, the rhythm was AF with a variable cycle length of 160-180 ms. First attempts at ATP failed while the rhythm was very irregular with higher variability in cycle length. After 40 hours and 18 minutes in episode duration, the cycle length slowed to 210 ms with the rhythm becoming more regular, and an attempt at ATP was successful at restoring sinus rhythm.

DISCUSSION
In this study, using prospectively acquired data from the deidentified Medtronic CareLink database, 43,440 patients with pacemakers, ICDs, and CRT devices that met eligibility criteria were identified, and rATP therapy was evaluated in an individually matched cohort of 4,016 rATP patients and 4,016 controls. rATP therapy was associated with significant reductions in the risk of having paroxysmal, persistent, and longstanding-persistent AT/AF during 2 years of follow-up. These reductions were consistent in subgroups, and they were independent of whether the patient had a pacemaker, ICD, or CRT device and whether preventive atrial pacing was activated at baseline. Our analysis also indicates reactive ATP is the principal driver of the observed AT/AF risk reductions. To our knowledge, these are the first real-world data on the effectiveness of rATP therapy, and the first data of the therapy in patients with ICD and CRT devices. The data available had the characteristics of "big data," that is, they were related to a large patient sample, although with limited granularity in terms of clinical profile.
Early studies of pacemaker/ICD integrated pacing therapies for AT/AF were promising. Prevention pacing, an atrial overdrive therapy aiming to suppress episode recurrence, was shown in several studies to reduce AT/AF. [16][17][18] However, the positive findings from these trials have been overwhelmed by over a dozen studies showing no difference, the latest being the SAFE and ASSERT studies. 19,20 Firstgeneration atrial ATP therapy was found to be safe and effective at terminating episodes. 21,22 Yet, these effects failed to consistently demonstrate a reduction in AT/AF burden.
The latest trial to test pacing therapies for AT/AF, the randomized , during which first attempt at ATP has no effect. After 40 hours in episode duration, the rhythm shows a slowed and regular cycle length at 210 ms (C), and an attempt at ATP is successful at restoring sinus rhythm (D). AT/AF = atrial tachyarrhythmia; rATP = reactive atrial-based antitachycardia pacing visits were reduced, suggesting significant cost savings. 12 These benefits were largely attributed to rATP therapy. 10,15 One unique aspect of the MINERVA trial was that rather than assessing the efficacy of rATP in terminating single episodes of AT/AF, more clinically valuable arrhythmic endpoints were evaluated, such as prevention of AF episodes of long duration. The novelty of rATP therapy includes the ability to redeliver ATP when the AT/AF rhythm changes cycle length and/or regularity and to retry ATP therapy over extended durations, with the aim to reduce the burden of AF and the occurrence of AF of longer duration. MINERVA showed that the algorithm was more effective when AT/AF episodes transitioned to slower and more regular rhythms. 15  nonparoxysmal AF was associated with significant increases in thromboembolism and death. 23 Recent findings from the ASSERT study reported a strong association between device-detected subclinical AF progression and HF hospitalization. 24 The CASTLE AF study also reported significant risk reductions in HF hospitalization and/or allcause death in the ablation arm along with a reduction in AT/AF burden. 25 By taking into account all of the studies that examined the relationship between type and duration of device-detected AF and risk of stroke, it appears that a dose-response association may exist between AF burden and the subsequent risk of stroke; therefore, reducing AT/AF burden may be indirectly considered as a potentially beneficial endpoint. 26 Our study included patients with advanced AT/AF as 15% in the matched cohort had at least one persistent or longstanding-persistent episode at baseline, yet rATP was effective immediately and effects persisted throughout the 2-year follow-up.
This suggests that use of rATP does not need to be limited to patients with paroxysmal AF only.

Study limitations
The study has its limitations. This was a retrospective study, and similar to other analyses based on "big data," there were uncollected clinical variables that we could not control for that include congestive heart failure, diabetes, hypertension, use of antiarrhythmic drugs, and other therapies for treating AT/AF. However, we controlled for key risk factors at baseline: age, sex, AT/AF, and percentage VP. 27 Also, we controlled for device type and pacing mode which may relate to the patient's underlying arrhythmia condition and comorbidity (e.g., heart failure). There is potential for sampling bias as the data is taken from a single-manufacturer's remote monitoring database largely from patients in the United States. Patients not followed by remote monitoring cannot be accounted for. Finally, patient outcomes such as cardiac function, permanent AF, stroke, and death could not be measured. For example, date of death was not available in the deidentified database either, hence its impact as a potential competing risk on AT/AF incidence could not be determined. Likewise, we did not have data on the use of antiarrhythmic drugs. The MINERVA trial has reported that baseline medications including the use of antiarrhythmic drugs were similar among the study arms, 10 and patients in the DDDRP+MVP arm that had high reactive ATP efficacy experienced a significant reduction in developing permanent or persistent AF after controlling for antiarrhythmic drugs and other baseline medications. 15 Although we cannot rule out all the potential bias with our data, the fact that the reactive ATP benefit that we observed in this observational study is aligned with the findings from the randomized MINERVA trial is reassuring.

CONCLUSION
In conclusion, among real-world patients from a large database, rATP therapy was associated with a lower rate of AT/AF progression. Therapy effects were independent of whether the patient had a pacemaker, defibrillator, or CRT device.