Advertisement

Adaptive cardiac resynchronization therapy is associated with decreased risk of incident atrial fibrillation compared to standard biventricular pacing: A real-world analysis of 37,450 patients followed by remote monitoring

Open AccessPublished:May 15, 2019DOI:https://doi.org/10.1016/j.hrthm.2019.05.012

      Background

      The AdaptivCRT algorithm (aCRT) automatically adjusts atrioventricular delays each minute to achieve ventricular fusion through left ventricular (LV) or biventricular (BiV) pacing. aCRT is associated with superior clinical outcomes compared to standard BiV pacing, but the association of aCRT and subsequent atrial fibrillation (AF) in a real-world population has not been fully evaluated.

      Objective

      The purpose of this study was to investigate the incidence of AF ≥48 hours with aCRT vs standard BiV pacing after implant.

      Methods

      Patients implanted with a cardiac resynchronization therapy (CRT) device between 2013 and 2016 were studied via the de-identified Medtronic CareLink database. For univariate and multivariate survival analyses, Kaplan-Meier and Cox proportional hazards were used, respectively.

      Results

      Of 37,450 patients (mean age 69.1 ± 11.0 years; 67.9% male) followed for a mean 15.5 ± 9.1 months, 9.7% (n = 3647) developed ≥48 hours of AF. In univariate analysis, compared with standard BiV pacing, the aCRT BiV and LV mode was associated with a 54% lower risk of ≥48 hours of AF (P <.001) at 2 years, which persisted after multivariate adjustment (hazard ratio 0.53; 95% confidence interval 0.49–0.57; P <.001), even when stratified by sensed PR interval ≤200 ms and >200 ms. Higher percentages of LV-only pacing with aCRT were associated with lower incidence of AF (comparing >92% LV-only pacing vs 0%–5% LV-only pacing: HR 0.05; 95% CI 0.04–0.06; P <.001).

      Conclusion

      In a large, real-world population of CRT recipients, aCRT pacing compared to standard BiV pacing was associated with a lower incidence of AF in patients with both long and short PR intervals. A higher percentage of LV-only pacing during aCRT was also associated with lower incidence of AF.

      Keywords

      Introduction

      Atrial fibrillation (AF) is common among patients with heart failure and is associated with an increased risk of all-cause hospitalization, stroke, and death.
      • Hannon N.
      • Sheehan O.
      • Kelly L.
      • et al.
      Stroke associated with atrial fibrillation: incidence and early outcomes in the north Dublin population stroke study.
      • Benjamin E.J.
      • Wolf P.A.
      • D'Agostino R.B.
      • et al.
      Impact of atrial fibrillation on the risk of death: the Framingham Heart Study.
      Approximately 1 in 5 patients with heart failure will develop AF within 4 years.
      • Wang T.J.
      • Larson M.G.
      • Levy D.
      • et al.
      Temporal relations of atrial fibrillation and congestive heart failure and their joint influence on mortality: the Framingham Heart Study.
      Strategies to prevent AF in heart failure patients are, therefore, warranted. Specifically in heart failure patients with evidence of reduced left ventricular (LV) ejection fraction and ventricular dyssynchrony, cardiac resynchronization therapy (CRT) provided by biventricular (BiV) pacing has been shown to improve quality of life as well as heart failure morbidity and mortality,
      • Cleland J.G.
      • Daubert J.C.
      • Erdmann E.
      • et al.
      Longer-term effects of cardiac resynchronization therapy on mortality in heart failure [the CArdiac REsynchronization-Heart Failure (CARE-HF) trial extension phase].
      but the influence of CRT in reducing the risk of incident AF in a heart failure population has not been fully elucidated.
      Recently, an algorithm that adapts CRT timing every minute has been introduced in an attempt to provide superior LV and right ventricular (RV) function compared to standard BiV pacing (AdaptivCRT, Medtronic, Mounds View, MN).
      • Martin D.O.
      • Lemke B.
      • Birnie D.
      • et al.
      Investigation of a novel algorithm for synchronized left-ventricular pacing and ambulatory optimization of cardiac resynchronization therapy: results of the adaptive CRT trial.
      Previous smaller studies have shown that the AdaptivCRT (aCRT) algorithm is noninferior to echo-optimized standard BiV pacing and may be associated with improved clinical outcomes, such as CRT response, decreased risk of incident AF, and reduction in 30-day heart failure readmission.
      • Birnie D.
      • Lemke B.
      • Aonuma K.
      • et al.
      Clinical outcomes with synchronized left ventricular pacing: analysis of the adaptive CRT trial.
      • Birnie D.
      • Hudnall H.
      • Lemke B.
      • et al.
      Continuous optimization of cardiac resynchronization therapy reduces atrial fibrillation in heart failure patients: results of the Adaptive Cardiac Resynchronization Therapy Trial.
      • Starling R.C.
      • Krum H.
      • Bril S.
      • et al.
      Impact of a novel adaptive optimization algorithm on 30-day readmissions: evidence from the Adaptive CRT Trial.
      The aim of the current study was to explore an association of the aCRT algorithm with decreased risk of incident AF, as suggested in smaller cohorts, in an attempt to extend these findings in a larger and more contemporary real-world CRT population. In patients programmed to aCRT pacing, we also sought to evaluate the association of PR interval and percentage of LV-only pacing to evaluate potential underlying mechanisms for the decreased risk of incident AF with use of the aCRT algorithm.

      Methods

       AdaptivCRT algorithm

      The aCRT algorithm adjusts atrioventricular (AV) timing each minute and ventriculoventricular (VV) timing every 16 hours according to measured cardiac intervals. aCRT can be programmed to adaptive BiV and LV pacing (nominal setting) or to adaptive BiV-only pacing mode. In adaptive BiV and LV pacing mode, the algorithm selects adaptive LV-only pacing when intrinsic PR is ≤200 ms (or 250 ms if atrial pacing is occurring) and heart rate is <100 bpm, and selects adaptive BiV pacing otherwise. During adaptive LV-only pacing, the atrial-to-RV sense interval (measured every minute) is used to deliver LV paces at least 40 ms before the anticipated RV sense. During adaptive BiV mode, the AV and VV timing intervals are calculated from atrial-to-RV sense, atrial to P-wave end, and RV sense to QRS end intervals.
      • Krum H.
      • Lemke B.
      • Birnie D.
      • et al.
      A novel algorithm for individualized cardiac resynchronization therapy: rationale and design of the adaptive cardiac resynchronization therapy trial.

       Study design, subject participation, and clinical database

      The de-identified Medtronic CareLink database was used to conduct a retrospective analysis, as has been used in previous reports.
      • Cheng A.
      • Landman S.R.
      • Stadler R.W.
      Reasons for loss of cardiac resynchronization therapy pacing: insights from 32 844 patients.
      All device data were collected by remote transmission to the Medtronic CareLink data server. All patient identification information was removed to protect patient privacy. The database includes age and sex from the device registration database. Patients in the United States, Canada, Australia, and New Zealand with Viva CRT devices and available aCRT (Viva XT and Viva Quad XT) who were implanted between January 29, 2013, and May 1, 2016, and data were transmitted before November 12, 2016, were screened for inclusion. A total of 50,891 patients were considered for analysis. Patients with devices with >5 minutes of AF on any of the first 7 days after implant (n = 8201), devices without an atrial lead (n = 3352), devices with missing atrial sensitivity or maximum programmed atrial sensitivity (n = 1108), devices with any missing days of device-determined AF burden (n = 706), devices with uninterpretable PR interval measurement (n = 8), and patients <18 years old (n = 66) were excluded. After exclusions, 37,450 patients were included in the analysis. The first exclusion criterion was established to avoid including patients with AF immediately after implant (assuming that this represented a previous history of AF) so that the true incidence of new-onset AF after implant could be ascertained. Patients without an atrial lead and those missing atrial sensitivity or maximum programmed atrial sensitivity and thus with unreliable AF diagnostics were excluded.
      Primary analysis evaluated the programmed mode of BiV pacing at the first CareLink upload after implant and incidence of AF during follow-up. In follow-up, 8.2% of patients switched programmed BiV pacing mode and were censored at the first programming change. Possible BiV pacing modes included aCRT BiV and LV pacing, aCRT BiV-only pacing, and standard BiV pacing (aCRT off).
      The primary endpoint for this study was time to first occurrence of 2 consecutive days with at least 23 hours/day of AF (according to device diagnostics), referred to as “48 hours of AF,” as previously published.
      • Chelu M.G.
      • Gunderson B.D.
      • Koehler J.
      • Ziegler P.D.
      • Sears S.F.
      Patient activity decreases and mortality increases after the onset of persistent atrial fibrillation in patients with implantable cardioverter-defibrillators.
      In brief, device-determined AF is the summed duration when the atrial rate exceeded a threshold (nominally 171 bpm) and there was evidence of ≥2:1 AV conduction for at least 3 ventricular intervals. The average burden of AF (ie, percentage of time spent in AF, averaged over the cohort) was also compared over 24 months of follow-up.

       PR interval and incidence of AF

      To determine the association of intrinsic PR interval with the incidence of AF, the sensed PR interval of each patient was measured automatically by the device. We selected the first sensed PR interval measurement occurring within the first 30 days after implant as the PR interval. This analysis excluded patients whose devices did not provide automatic PR interval measurements, that is, patients with complete heart block or sensed PR intervals >300 ms, or devices for which aCRT was never enabled.

       LV-only pacing and incidence of AF

      To evaluate the potential association of percentage of LV-only pacing (ie, percentage of CRT-paced beats that were LV-only paced) and the risk of incident AF, we categorized patients with aCRT programmed to adaptive BiV and LV pacing into quartiles of LV-only pacing percentage. Device data provided the percentage of LV-only pacing and the intrinsic PR interval.

       Statistical analysis

      Normally distributed continuous variables are expressed as mean ± SD and were compared with the Student t test, whereas categorical variables were compared using the χ2 test. For univariate time-to-event survival analyses, cumulative probabilities of the risk of incident AF of 48 hours were displayed according to the Kaplan-Meier method, with comparison of cumulative event rates by the log-rank test. Time 0 was the date of implant, and patients were censored at their final remote monitoring transmission or the date when their aCRT mode was first changed. Although all follow-up data were used in the analysis, curves ended when <20 patients remained at risk or at 36 months. For multivariable survival analyses, Cox proportional hazards regression analyses were performed to estimate the risk of AF of 48 hours, adjusted for age and sex. Additional data regarding cardiovascular comorbidities were not available. All analyses were prespecified a priori. All tests were 2-sided, and P <.05 was considered significant. All patient data were de-identified for analyses. All statistical computations were performed using SAS version 9.4 statistical software (SAS Institute Inc, Cary, NC).

      Results

      The study cohort consisted of 37,450 patients with aCRT-capable devices (mean age 69.1 ± 11.0 years; 67.9% male). Mean follow-up duration of the cohort was 15.5 ± 9.1 months (range 0.2–44.1 months). A total of 3647 patients (9.7%) developed 48 hours of AF during follow-up. Comparison of demographic characteristics between patients who went on to develop 48 hours of AF vs patients who did not is given in Table 1. Patients who went on to develop 48 hours of AF during follow-up were older, more often were men, and had longer average follow-up time. At baseline, 59.8% were programmed to adaptive BiV and LV, 19.8% to adaptive BiV-only, and 20.4% to standard (nonadaptive) BiV pacing (ie, aCRT programmed off).
      Table 1Demographics and characteristics of CRT recipients stratified by 48 hours of AF during follow-up
      CharacteristicAll patients (N = 37,450)Developed 48 h of AF during follow-up (n = 3647)Did not develop 48 h of AF during follow-up (n = 33,803)P value
      Age at implant (y)69.1 ± 11.0 (n = 36,798)72.5 ± 9.7 (n = 3564)68.7 ± 11.1 (n = 33,234)<.001
      Male sex25,131/37,017 (67.9)2955/3615 (81.8)22,176/33,404 (66.4)<.001
      Follow-up time (mo)15.5 ± 9.118.2 ± 9.215.2 ± 9.1<.001
      Categorical variables are given as proportion (%) and continuous variables as mean ± SD.
      AF = atrial fibrillation; CRT = cardiac resynchronization therapy.
      Overall, in univariate survival analyses using Kaplan-Meier curves to represent the cumulative incidence of developing 48 hours of AF (Figure 1), CRT patients programmed to adaptive BiV and LV pacing experienced a 54% lower risk of developing 48 hours of AF compared to those programmed to standard BiV pacing (P <.001). In multivariable analysis adjusting for age and sex, adaptive BiV and LV pacing compared to standard BiV pacing continued to be associated with a lower risk of AF (hazard ratio [HR] 0.53; 95% confidence interval [CI] 0.49–0.57; P <.001). In exploratory analyses using different time duration cutpoints to define incident AF, patients programmed to adaptive BiV and LV pacing compared to those programmed to standard BiV pacing consistently experienced a lower risk of AF, whether the cutpoint definition used was as short as 5 minutes to as long as 1 month of AF (Table 2). However, patients programmed to adaptive BiV-only did not show a significant reduction in AF compared to standard BiV patients (adjusted HR 0.93; 95% CI 0.85–1.01; P = .08).
      Figure thumbnail gr1
      Figure 1Unadjusted Kaplan-Meier estimates of cumulative incidence of developing 48 hours of atrial fibrillation (AF) among 37,450 cardiac resynchronization therapy (CRT) patients followed in the Medtronic CareLink remote monitoring database stratified by programmed mode of biventricular (BiV) pacing (AdaptivCRT algorithm [aCRT] off, green line; adaptive BiV pacing, red line; adaptive BiV and left ventricular (LV) pacing, purple line) over 3 years of follow-up. Adjusted hazard ratio (HR) (95% confidence interval) and P values shown in the inset represent Cox proportional hazards regression estimates for the outcome of developing 48 hours of AF in the 2 aCRT pacing modes vs standard BiV pacing mode, adjusting for age and sex.
      Table 2Risk of developing AF at 24 months of follow-up based on various cutpoint duration definitions of device-detected AF
      Minimum AF durationIncidence at 24 mo (%)Adjusted hazard ratio (95% CI)P value
      Adaptive CRTStandard BiV pacing (n = 15,033)
      aCRT BiV+LV (n = 22,393)
      5 minutes within 1 d31.445.20.66 (0.62–0.69)<.001
      23 h within 1 d11.622.20.55 (0.51–0.59)<.001
      48 h (ie, 2 consecutive days of 23 h)10.120.70.53 (0.49–0.57)<.001
      7 consecutive days of 23 h8.017.40.50 (0.46–0.55)<.001
      30 consecutive days of 23 h5.313.50.47 (0.42–0.52)<.001
      aCRT BiV (n = 7412)
      5 min within 1 d43.945.20.95 (0.90–1.01).08
      23 h within 1 d20.822.20.93 (0.85–1.01).07
      48 h (ie, 2 consecutive days of 23 h)19.220.70.93 (0.85–1.01).08
      7 consecutive days of 23 h15.717.40.91 (0.83–1.01).07
      30 consecutive days of 23 h12.513.50.94 (0.84–1.06).30
      Incidence is based on Kaplan-Meier estimates at 24-month follow-up. Hazard ratios are adjusted for age and sex using a Cox proportional hazards model. P values are based on the adjusted Cox proportional hazards model and compare overall cumulative incidence between aCRT and standard BiV.
      aCRT = AdaptivCRT algorithm; AF = atrial fibrillation; BiV = biventricular; CI = confidence interval; CRT = cardiac resynchronization therapy; LV = left ventricular.
      The average AF burden for patients programmed to aCRT pacing vs standard BiV pacing, plotted by duration of follow-up, is shown in Supplementary Figure S1. For the period 21–24 months after implant, AF burden was 3.3% with aCRT BiV and LV pacing and 8.5% with standard BiV pacing (P <.001).
      Among patients with aCRT programmed to adaptive BiV and LV pacing, higher percentages of LV-only pacing were associated with a markedly decreased risk of incident AF (Figure 2). At 2-year follow-up, the cumulative incidence of AF was 22.4% in the quartile of patients with the least LV-only pacing (0%–5% LV-only pacing), 12.1% in quartile 2 (5%–59% LV-only pacing), 5.1% in quartile 3 (59%–92% LV-only pacing), and 1.3% in quartile 4 (>92% LV-only pacing). The HR for incident AF comparing >92% LV-only pacing and 0%–5% LV-only pacing was 0.05 (95% CI 0.04–0.06; P <.001).
      Figure thumbnail gr2
      Figure 2Unadjusted Kaplan-Meier estimates of cumulative incidence of developing 48 hours of atrial fibrillation (AF) among increasing quartiles of left ventricular (LV)-only pacing in patients programmed to AdaptivCRT algorithm (aCRT) pacing with adaptive biventricular (BiV) and LV pacing, with quartile 1 (0%–5% LV-only pacing; blue line), quartile 2 (5%–59% LV-only pacing; black line), quartile 3 (59%–92% LV-only pacing; green line), and quartile 4 (>92% LV-only pacing; purple line) shown over 3 years of follow-up. Log rank test for comparison, P <.001.
      For patients who had a device-measured PR interval in the first 30 days of implant (ie, excluding patients with complete heart block or PR >300 ms), the risk of developing 48 hours of AF based on aCRT programming (standard nonadaptive BiV pacing, aCRT BiV-only pacing, and aCRT BiV and LV pacing) and stratified by the intrinsic PR interval is shown in Figure 3. Patients with PR interval >200 ms had an increased risk of developing 48 hours of AF compared to patients with PR interval ≤200 ms, with similar relative risk reduction of incident AF associated with aCRT (compared to standard nonadaptive BiV pacing) in both groups. At 2-year follow-up, among patients with PR interval >200 ms, the cumulative incidence of AF was 26.7% in those programmed to standard BiV pacing, 20.4% in those programmed to adaptive BiV-only pacing, and 15.1% in those programmed to adaptive BiV and LV pacing (log rank test for comparison, P <.001). At 2-year follow-up, among patients with PR interval ≤200 ms, the cumulative incidence of AF was 13.5% in those programmed to standard BiV pacing, 7.6% in those programmed to adaptive BiV-only pacing, and 6.1% in those programmed to adaptive BiV and LV pacing (log rank test for comparison, P <.001).
      Figure thumbnail gr3
      Figure 3Influence of AdaptivCRT algorithm (aCRT) programming and PR interval on incident atrial fibrillation (AF). The curves reflect unadjusted Kaplan-Meier estimates of cumulative incidence of developing 48 hours of AF among patients with aCRT-capable devices and a device-measured PR interval >200 ms (A) and measured PR interval ≤200 ms (B), stratified by programmed mode of biventricular (BiV) pacing (aCRT off, green lines; vs adaptive BiV pacing, red lines; vs adaptive BiV and LV pacing, purple lines) shown over 3 years of follow-up. Log rank test for each comparison, P < .001.
      Among patients with aCRT programmed to adaptive BiV and LV pacing, we stratified analyses to 2 cohorts: those with measured PR interval >200 ms and those with PR interval ≤200 ms. In both cohorts, a lower percentage of LV-only pacing was associated with a higher risk of incident AF (Figure 4). At 2-year follow-up, among patients with PR interval >200 ms, the cumulative incidence of AF was 18.8% in patients with 0%–5% LV-only pacing, 9.8% in patients with 5%–59% LV-only pacing, and 2.9% in patients with 59%–92% LV-only pacing (log rank test for comparison, P <.001). At 2-year follow-up, among patients with PR interval ≤200 ms, the cumulative incidence of AF was 27.9% in patients with 0%–5% LV-only pacing, 11.9% in patients with 5%–59% LV-only pacing, 5.1% in patients with 59%–92% LV-only pacing, and 1.3% in patients with >92% LV-only pacing (log rank test for comparison, P <.001). In the cohort of patients with PR interval >200 ms, few patients (n = 21) had >92% LV-only pacing because the aCRT algorithm delivers LV-only pacing during shorter PR intervals and adaptive BiV pacing during longer (>200 ms) PR intervals. Despite this, in both cohorts including those with PR interval >200 ms and PR interval ≤200 ms, the risk of developing AF was higher in patients with lower percentage of LV-only pacing.
      Figure thumbnail gr4
      Figure 4Influence of percent left ventricular (LV)-only pacing and PR interval on incident atrial fibrillation (AF). Curves reflect unadjusted Kaplan-Meier estimates of cumulative incidence of developing 48 hours of AF among patients with AdaptivCRT algorithm (aCRT)-capable devices programmed to adaptive biventricular (BiV) and LV pacing and device-measured PR interval >200 ms (A) and measured PR interval ≤200 ms (B), stratified by increasing quartiles of LV-only pacing, with quartile 1 (0%–5% LV-only pacing; blue line); quartile 2 (5%–59% LV-only pacing; black line); quartile 3 (59%–92% LV-only pacing; green line); and quartile 4 (>92% LV-only pacing; purple line) shown over 3 years of follow-up. There is no purple line in A because, as expected, few patients with PR >200 ms had >92% LV-only pacing. Log rank test for each comparison, P < .001.

      Discussion

      In a large, real-world contemporary population of CRT recipients followed by remote monitoring, aCRT BiV and LV pacing compared to standard BiV pacing was associated with a lower incidence of AF in patients with both long and short PR intervals, with an approximate 54% lower risk of developing of AF with aCRT pacing. A higher percentage of LV-only pacing during aCRT was also associated with lower incidence of AF. This is the largest study of its kind to compare aCRT pacing vs standard BiV pacing with the risk of developing AF. The results provide valuable insight into potential mechanisms behind avoidance of AF with aCRT and provide real-world support for the use of aCRT.
      Previous studies examining the effect of aCRT vs standard BiV pacing on clinical outcomes including incident AF are limited. A single, much smaller study by Birnie et al
      • Birnie D.
      • Hudnall H.
      • Lemke B.
      • et al.
      Continuous optimization of cardiac resynchronization therapy reduces atrial fibrillation in heart failure patients: results of the Adaptive Cardiac Resynchronization Therapy Trial.
      evaluated a cohort of 478 patients randomized 2:1 to receive either aCRT or echo-optimized standard BiV pacing. They reported an unadjusted 46% reduction in the risk of developing 48 hours of AF (HR 0.54; 95% CI 0.31–0.93; P = .03). Importantly, all patients in that study were programmed to aCRT BiV and LV. The unadjusted 54% reduced risk of AF in our study for aCRT BiV and LV is well within the confidence limits of the findings by Birnie et al, thus confirming their observed benefit in a much larger, real-world cohort. In further subgroup analysis, Birnie et al found that patients with PR >200 ms had significant reduction in AF risk with aCRT, whereas patients with PR ≤200 ms did not achieve a statistically significant benefit (HR 0.60; 95% CI 0.19–1.85; P = .37). The broad CIs here suggest a lack of statistical power. Our results expand on their findings and show a statistically significant benefit for aCRT BiV and LV patients with both long and short PR intervals, thus suggesting that study size limited the statistical power of the study by Birnie et al. The greater size and power of our study also showed a significant risk reduction over a wide range of AF duration endpoints (Table 2), whereas Birnie et al only showed a statistically significant benefit at 48 hours of AF duration and borderline benefit at 24 hours of AF (although observed HRs at all but the shortest AF duration were consistent). Therefore, our results present more robust findings over long and short PR intervals along with different definitions of incident AF cutoffs, and add to growing support for the utility of aCRT BiV and LV in reducing the risk of AF.
      It is well known that longer PR intervals are associated with increased risk of AF.
      • Nielsen J.B.
      • Pietersen A.
      • Graff C.
      • et al.
      Risk of atrial fibrillation as a function of the electrocardiographic PR interval: results from the Copenhagen ECG Study.
      • Schnabel R.B.
      • Sullivan L.M.
      • Levy D.
      • et al.
      Development of a risk score for atrial fibrillation (Framingham Heart Study): a community-based cohort study.
      Our results were consistent with these previous findings, as CRT recipients with long PR intervals developed more AF overall than did patients with short PR intervals. However, both cohorts had a reduced incidence of AF with aCRT BiV and LV. Because patients with longer PR intervals receive minimal LV-only pacing from aCRT because of the specific algorithm of BiV pacing when PR interval >200 ms, this suggests that LV-only pacing alone cannot explain the observed benefit; the automatic adjustment of AV delay each minute with aCRT likely also contributed to the reduced risk of AF. This is supported by Figure 3, which shows that aCRT patients with BiV-only programming also had reduced AF compared to those with standard CRT. The observed differences between Figure 1, Figure 3 suggest that aCRT BiV-only patients with normal or moderately prolonged PR intervals also have reduced AF as a result of aCRT, which likely results from the automatic adjustment of AV delay each minute. The benefit of AV interval adjustments with CRT has been supported by previous studies. RAFT (Resynchronization/Defibrillation in Ambulatory Heart Failure Trial) programmed patients to nominal fixed AV delays and found 20% increased risk of developing AF in patients randomized to CRT-D vs ICD, with the authors suggesting that short AV intervals for BiV pacing may have caused the increased risk of AF with CRT.
      • Wilton S.B.
      • Exner D.V.
      • Wyse D.G.
      • et al.
      Frequency and outcomes of postrandomization atrial tachyarrhythmias in the resynchronization/defibrillation in ambulatory heart failure trial.
      In contrast, programming away from the nominal AV delay, suggesting an attempt to optimize AV timing for each subject, was shown to be associated with less incident AF in a study of 5709 CRT recipients followed by remote monitoring in the ALTITUDE database.
      • Steinberg B.A.
      • Wehrenberg S.
      • Jackson K.P.
      • et al.
      Atrioventricular and ventricular-to-ventricular programming in patients with cardiac resynchronization therapy: results from ALTITUDE.
      Although several algorithms for programming patient-specific but nonadaptive AV intervals in CRT recipients have been described, none have addressed subsequent risk of developing AF.
      • Stein K.M.
      • Ellenbogen K.A.
      • Gold M.R.
      • et al.
      SmartDelay determined AV optimization: a comparison of AV delay methods used in cardiac resynchronization therapy (SMART-AV): rationale and design.
      • Abraham W.T.
      • Gras D.
      • Yu C.M.
      • et al.
      Rationale and design of a randomized clinical trial to assess the safety and efficacy of frequent optimization of cardiac resynchronization therapy: the Frequent Optimization Study Using the QuickOpt Method (FREEDOM) trial.
      Data from our study suggest that the patient-specific and adaptive AV interval changes from the aCRT algorithm seem promising for reducing the risk of AF.
      In addition to reducing the risk of AF, a high percentage of LV-only CRT (≥50%) has been associated with a decreased risk of death and heart failure hospitalization, and improved clinical composite score at 6 and 12 months of follow-up.
      • Birnie D.
      • Lemke B.
      • Aonuma K.
      • et al.
      Clinical outcomes with synchronized left ventricular pacing: analysis of the adaptive CRT trial.
      In our study, we saw a clear stepwise benefit of increasing percentage of LV-only pacing associated with less risk of incident AF. The adaptive timing of LV-only pacing to achieve fusion with intrinsic activation of the RV is likely essential for this observed benefit. The exact mechanism of how LV-only pacing with fusion is able to reduce the risk of subsequent AF (eg, unloading of the LV or reduction of mitral regurgitation) should be the focus of future investigations into these findings. Ongoing studies such as the AdaptResponse trial (ClinicalTrials.gov Identifier NCT02205359), which plans to enroll 3800 patients with heart failure, left bundle branch block, and PR interval ≤200 ms and randomize them to aCRT vs standard BiV pacing, will be important to our understanding of the benefit of aCRT and specifically LV-only pacing because of its focus on patients with shorter PR intervals. The risk of AF will be evaluated in the AdaptResponse trial as a secondary endpoint, with the primary endpoint being the time to heart failure event or all-cause mortality, which will help further evaluate the utility of aCRT.

       Study limitations

      First, the de-identified Medtronic CareLink database only contains limited patient demographic information (ie, age and sex) and no information regarding comorbidities, risk factors for AF, medications, or other therapeutic interventions. This may confound the associations evaluated and prohibits adjustments in multivariable modeling. However, the size of the database suggests that this is a real-world population with representative demographics. Second, because this was a retrospective observational study, there was no protocoled assignment of aCRT vs standard BiV pacing programming, and it is not possible to evaluate the reasons why patients with aCRT-enabled devices were initially programmed with aCRT off instead of utilizing the aCRT pacing algorithm. Additionally, 8.2% of patients switched aCRT modes before 48 hours of AF for unknown reasons.

      Conclusion

      In a large, contemporary, real-world population of CRT recipients followed by remote monitoring, use of the aCRT BiV and LV algorithm compared to standard BiV pacing was associated with a significantly reduced risk of developing AF. The benefit of the aCRT algorithm with subsequent reduced incident AF was experienced by patients with both long and short PR intervals and was most profound in patients with higher percentage of LV-only pacing.

      Appendix. Supplementary data

      • Supplementary Figure S1

        The average percent AF burden for patients as a function of time in the study. Patients are included in each 3-month period if they had at least 30 days of follow-up during that period.

      References

        • Hannon N.
        • Sheehan O.
        • Kelly L.
        • et al.
        Stroke associated with atrial fibrillation: incidence and early outcomes in the north Dublin population stroke study.
        Cerebrovasc Dis. 2010; 29: 43-49
        • Benjamin E.J.
        • Wolf P.A.
        • D'Agostino R.B.
        • et al.
        Impact of atrial fibrillation on the risk of death: the Framingham Heart Study.
        Circulation. 1998; 98: 946-952
        • Wang T.J.
        • Larson M.G.
        • Levy D.
        • et al.
        Temporal relations of atrial fibrillation and congestive heart failure and their joint influence on mortality: the Framingham Heart Study.
        Circulation. 2003; 107: 2920-2925
        • Cleland J.G.
        • Daubert J.C.
        • Erdmann E.
        • et al.
        Longer-term effects of cardiac resynchronization therapy on mortality in heart failure [the CArdiac REsynchronization-Heart Failure (CARE-HF) trial extension phase].
        Eur Heart J. 2006; 27: 1928-1932
        • Martin D.O.
        • Lemke B.
        • Birnie D.
        • et al.
        Investigation of a novel algorithm for synchronized left-ventricular pacing and ambulatory optimization of cardiac resynchronization therapy: results of the adaptive CRT trial.
        Heart Rhythm. 2012; 9: 1807-1814
        • Birnie D.
        • Lemke B.
        • Aonuma K.
        • et al.
        Clinical outcomes with synchronized left ventricular pacing: analysis of the adaptive CRT trial.
        Heart Rhythm. 2013; 10: 1368-1374
        • Birnie D.
        • Hudnall H.
        • Lemke B.
        • et al.
        Continuous optimization of cardiac resynchronization therapy reduces atrial fibrillation in heart failure patients: results of the Adaptive Cardiac Resynchronization Therapy Trial.
        Heart Rhythm. 2017; 14: 1820-1825
        • Starling R.C.
        • Krum H.
        • Bril S.
        • et al.
        Impact of a novel adaptive optimization algorithm on 30-day readmissions: evidence from the Adaptive CRT Trial.
        JACC Heart Fail. 2015; 3: 565-572
        • Krum H.
        • Lemke B.
        • Birnie D.
        • et al.
        A novel algorithm for individualized cardiac resynchronization therapy: rationale and design of the adaptive cardiac resynchronization therapy trial.
        Am Heart J. 2012; 163: 747-752 e741
        • Cheng A.
        • Landman S.R.
        • Stadler R.W.
        Reasons for loss of cardiac resynchronization therapy pacing: insights from 32 844 patients.
        Circ Arrhythm Electrophysiol. 2012; 5: 884-888
        • Chelu M.G.
        • Gunderson B.D.
        • Koehler J.
        • Ziegler P.D.
        • Sears S.F.
        Patient activity decreases and mortality increases after the onset of persistent atrial fibrillation in patients with implantable cardioverter-defibrillators.
        JACC Clin EP. 2016; 2: 518-523
        • Nielsen J.B.
        • Pietersen A.
        • Graff C.
        • et al.
        Risk of atrial fibrillation as a function of the electrocardiographic PR interval: results from the Copenhagen ECG Study.
        Heart Rhythm. 2013; 10: 1249-1256
        • Schnabel R.B.
        • Sullivan L.M.
        • Levy D.
        • et al.
        Development of a risk score for atrial fibrillation (Framingham Heart Study): a community-based cohort study.
        Lancet. 2009; 373: 739-745
        • Wilton S.B.
        • Exner D.V.
        • Wyse D.G.
        • et al.
        Frequency and outcomes of postrandomization atrial tachyarrhythmias in the resynchronization/defibrillation in ambulatory heart failure trial.
        Circ Arrhythm Electrophysiol. 2016; 9: e003807
        • Steinberg B.A.
        • Wehrenberg S.
        • Jackson K.P.
        • et al.
        Atrioventricular and ventricular-to-ventricular programming in patients with cardiac resynchronization therapy: results from ALTITUDE.
        J Interv Card Electrophysiol. 2015; 44: 279-287
        • Stein K.M.
        • Ellenbogen K.A.
        • Gold M.R.
        • et al.
        SmartDelay determined AV optimization: a comparison of AV delay methods used in cardiac resynchronization therapy (SMART-AV): rationale and design.
        Pacing Clin Electrophysiol. 2010; 33: 54-63
        • Abraham W.T.
        • Gras D.
        • Yu C.M.
        • et al.
        Rationale and design of a randomized clinical trial to assess the safety and efficacy of frequent optimization of cardiac resynchronization therapy: the Frequent Optimization Study Using the QuickOpt Method (FREEDOM) trial.
        Am Heart J. 2010; 159: 944-948 e941