Subclinical Atrial Fibrillation Burden and Adverse Clinical Outcomes in Patients With Permanent Pacemakers

Clinical, electrocardiographic, PPM-related, and echocardiographic variables in all patients undergoing PPM implantation at our center were entered prospectively into the database for our PPM registry. Between January 2005 and December 2016, a total of 1359 consecutive patients underwent PPM implantation. Inclusion criteria for the present study were as follows: age ≥18 years, de novo PPM implantation based on current guidelines, and PPM with atrial sensing capability. Patients who met any of the following criteria were excluded: previously documented clinical AF or atrial flutter, atrial lead fracture or malfunction making the detection of a true atrial signal impossible or unreliable, lack of or insufficient data on SCAF burden, or follow-up loss after PPM implantation. Significant portion of patients in our PPM registry, mostly referred from far distant regional hospitals, have been followed-up irregularly. Thus, device analysis was sometimes performed in our center but in other cases at their own local hospitals. This was one of the main reasons of lack of or insufficient data. The details of the study population are presented in Figure 1. The data supporting the findings of this study are available from the corresponding author on reasonable request. This study was approved by the Institutional Review Board of our hospital, and the requirement for written informed consent was waived.

Patients underwent implantation of pacemakers with atrial sensing capability, such as the DDD(R), VDD(R), or AAI(R) pacing systems. Patients then underwent device interrogation at 3 months and then every 6 months after PPM implantation. PPM interrogation data, including SCAF burden and 12-lead surface ECG, were collected at every device clinic visit. The presence of SCAF was defined as any atrial high rate episode lasting for ≥6 minutes at a rate >170 bpm without prior documentation of AF on a surface ECG, including 12-lead standard ECGs, Holter ECGs, or telemonitoring.^5,13 The SCAF burden was defined as the total amount of time spent in atrial high rate episode during the follow-up interval. The SCAF burden was reported either as the total number of hours or percentages of time in SCAF according to different versions or manufacturers of PPMs. When the SCAF burden was initially reported as the percentage of time instead of the total number of hours, we converted the percentage into hours by multiplying the percentage by 24 hours and the specific follow-up interval (in days) for our analyses. When the cumulative SCAF burden was reported only as <0.1%, we assumed it was 0.1%. Device follow-up for almost all PPM patients at our center is typically performed every 6 months, and we, therefore, calculated the 6-month SCAF (6Mo-SCAF) burden in hours (Figure in the Data Supplement).

Previously, SCAF >24 hours was closely associated with an increased risk of ischemic stroke or systemic embolism or HF-related hospitalization.^14,15 Accordingly, patients with at least one report indicating their 6Mo-SCAF was greater than 24 hours were considered to have a high-burden of SCAF. All patients were classified into 3 groups according to their 6Mo-SCAF burden: no SCAF (no SCAF episode was detected in any device analysis), low-burden SCAF (6Mo-SCAF <24 hours in at least 1 device analysis), or high-burden SCAF (6Mo-SCAF ≥24 hours in at least 1 device analysis). Additionally, the greatest value of all 6Mo-SCAF burdens, that is, the maximal 6Mo-SCAF, was obtained in each patient.

Transthoracic echocardiography was performed before PPM implantation using commercially available equipment (Vivid 9 or 7 GE Healthcare, Chicago, IL). Echocardiographic parameters were obtained from all patients as recommended by the American Society of Echocardiography.¹⁶ Two-dimensional guided M-mode imaging was used to measure the dimensions of the left ventricle (LV) and left atrium (LA). The LV ejection fraction was calculated using a modified biplane Simpson method.

The primary outcome of this study was composite adverse events incorporating progression to clinical AF, ischemic stroke, MI, HF hospitalization, and cardiac death. Each component of the primary outcomes was evaluated individually as a secondary outcome. Clinical AF was defined as any AF episode, whether symptomatic (overt) or asymptomatic (silent), documented on surface 12-lead ECGs, Holter ECGs, or telemonitoring.¹⁷ An ischemic stroke was counted only when all the following criteria were met: the presence of neurological deficits, findings on brain imaging studies that were compatible with ischemic stroke, and a neurologist’s confirmation. MI was defined by a rise of cardiac enzyme with at least one of the following: symptom of myocardial ischemia, new-onset significant change in ST-segment/T-wave, new-onset left bundle branch block, development of pathological Q-wave, new loss of viable myocardium or new regional wall motion abnormality on imaging studies, or intracoronary thrombus identified on coronary angiography.¹⁸ HF hospitalization was defined according to the 2016 European Society of Cardiology guidelines following careful evaluation of HF symptoms or signs, pulmonary congestion on a chest radiography, objective findings of cardiac dysfunction by echocardiography, and cardiac biomarkers.¹⁹ All deaths were considered to be cardiac unless a definitive noncardiac cause could be identified.

Continuous variables were expressed as the median with interquartile range or mean with SD, as appropriate. To test differences in continuous variables among the 3 groups, a nonparametric Kruskal-Wallis test was applied and followed by a Tukey test using ranks. Categorical variables are presented as counts and percentages and were compared using the Fisher exact test. Predictors for primary outcomes were evaluated using univariate and multiple logistic regression analyses. The variables that attained a P<0.2 in the univariate analysis were entered into a multiple logistic regression model. SCAF burden was incorporated into the model as a continuous or categorical variable. Event-free survival following PPM implantation was estimated by the Kaplan-Meier analyses, and the log-rank test was applied to evaluate differences among survival curves of the no, low-burden, and high-burden SCAF groups. In the low-burden and high-burden SCAF groups, we reconstructed and compared the survival curves with the initial detection time of a SCAF episode used as a landmark. All tests were 2-tailed, and a P<0.05 was considered statistically significant. Statistical analyses were performed using SPSS version 23 for Windows (IBM, Armonk, NY).

A total of 1379 patients underwent PPM implantations at our institute during the study period. Of these, 751 patients were excluded for any of the following reasons: PPM without atrial lead (n=560), preexisting AF or atrial flutter (n=163), loss to follow-up after PPM implantation (n=8), lack of or insufficient data regarding the SCAF burden (n=128), or atrial lead malfunction (n=4), as shown in Figure 1. The remaining 497 patients were included in the final analyses. The mean (±SD) age of our patients was 65.6±12.9 years, and 233 (47.0%) patients were male. The mean (±SD) CHA₂DS₂-VASc score was 2.5±1.5, and 42 (8.5%) patients had a history of stroke. Baseline mean (±SD) LV ejection fraction and LA diameter were 64±9% and 41±6 mm, respectively. The DDD(R) (n=481, 97.2%), VDD(R) (n=10, 2.0%), or AAI(R) (n=4, 0.8%) PPMs were implanted for the indications of sinus node dysfunction (n=267, 53.8%) or atrioventricular block (n=229, 46.2%).

A total of 5452 interrogations were performed in 496 patients during the median (interquartile range) follow-up of 5.2 (2.9−8.3) years after PPM implantation. Most of the interrogations (4972/5452, 91.2%) were at 6±1-month intervals with median (interquartile range) device follow-up intervals of 179 (175–182) days. A total of 496 patients were classified into one of 3 groups according to their 6Mo-SCAF burden (Table 1): no SCAF (n=152, 31%), low-burden SCAF (n=287, 58%), and high-burden SCAF (n=57, 11%) groups. The median value (interquartile range) of the maximal 6Mo-SCAF in each group was 0, 4.3 (4.3−4.3), and 362.9 (172.8−747.4) hours (P<0.001), respectively.

There were no significant differences among the 3 groups regarding their baseline characteristics, including sex, prevalence of HF, hypertension, or diabetes, history of stroke, MI, or valve surgery, and CHA₂DS₂-VASc score, except that age was inversely correlated with the degree of SCAF burden (P=0.011). Similarly, the LV dimensions and ejection fraction, LA size, types of PPM, and discharge medications (including antiplatelet agents and anticoagulants) were not significantly different among the 3 groups. DDD(R)-mode PPM was implanted in most of the patients (≥95%) in all 3 groups. Class Ic antiarrhythmic drugs had been prescribed for 2 patients in the high-burden and one in the low-burden SCAF group to treat frequent symptomatic ventricular premature contraction before PPM implantation. These medications were also included in their discharge medications.

During the entire follow-up, the primary composite adverse outcome occurred in 49 (9.8%) of 496 patients (Figure 2, Table in the Data Supplement). The risk of this composite adverse outcome was significantly greater in the high-burden (52.6%, 30/57) than in the no (6.6%, 10/152) or low-burden (3.1%, 9/287) SCAF groups (P<0.001). In particular, the high-burden SCAF group exhibited a much higher risk of progression to clinical AF (P<0.001), ischemic stroke (P<0.001), and cardiac death (P=0.018) than did the other 2 groups. Additionally, MI and HF hospitalization were also more likely to occur in the high-burden SCAF group than in the other 2 groups. Accordingly, the high-burden SCAF group had the worst prognosis following the PPM implantation (P<0.001) among all 3 groups (Figure 3A). In contrast, there were no significant differences between the no and low-burden SCAF groups with respect to their composite event-free survival rates.

Even if the Kaplan-Meier curves were reconstructed using the first detection time of SCAF as the landmark, the risk of progression to clinical AF was consistently higher in the high-burden group than in the low-burden group (Figure 3B). The average time for progression to clinical AF in the high-burden group (8.9±13.4 months) was significantly shorter than the value (32.6±50.4 months) in the low-burden group (P=0.032). With regard to the non-AF progression composite outcome, which comprised only ischemic stroke, HF hospitalization, MI, and cardiac death, the high-burden SCAF group also showed a worse prognosis than did the low-burden group (P<0.001; Figure 3C).

Detailed information for the patients with ischemic stroke is provided in Table 2. Overall, ischemic stroke occurred in 6 patients, comprising 1 from the no SCAF, 1 from the low-burden SCAF, and 4 from the high-burden SCAF groups. The mean CHA₂DS₂-VASc score of the 4 stroke patients in the high-burden group (5.8±2.2) was greater than the values of stroke patients in the no and low-burden groups. In 5 of 6 stroke patients with SCAF, the SCAF events always preceded their ischemic strokes. The average time from PPM implantation to the first detection of a SCAF episode was 38.7±25.0 months. Subsequently, ischemic strokes were diagnosed 12.0±10.0 months after the first SCAF detection. In the 4 patients with ischemic stroke in the high-burden group, the intervals from the first SCAF detection to ischemic stroke events were always shorter than that in the stroke patients from the low-burden group (Table 2).

To assess the independent predictors of the composite adverse outcomes and progression to clinical AF, multivariate logistic regression models were constructed with risk factors attaining P values <0.2 in the univariate analyses. Among these, the SCAF burden, even if incorporated into multivariate analysis models as a continuous (the maximal 6Mo-SCAF in hours) or categorical variable (the high-burden SCAF group versus no SCAF group), was identified as an independent predictor of composite adverse outcomes (odds ratio, 20.1 [95% CI, 7.60−52.7], P<0.001) and progression to clinical AF (odds ratio, 6.2 [95% CI, 15.9−87.8], P<0.001; Table 3). LA diameter was also independently associated with composite adverse outcomes (odds ratio, 1.08 [95% CI, 1.01−1.17], P=0.048).

Our main findings are as follows: (1) approximately two thirds of 496 PPM patients demonstrated SCAF episodes in ≥1 device interrogation during post-PPM follow-up (median, 5.2 years; interquartile range, 2.9−8.3 years), (2) 57 of 344 (16.6%) PPM patients with SCAF were classified into the high-burden group whereas the remaining 287 (83.4%) were classified into the low-burden group, (3) the risk of composite adverse outcomes was greatest in the high-burden SCAF group and was primarily driven by progression to clinical AF and ischemic stroke, and (4) the SCAF episodes always preceded the ischemic stroke events, and moreover, the presence of high-burden SCAF was identified as an independent predictor for composite adverse outcomes and progression to clinical AF in PPM patients without evidence of clinical AF.

Previously, device-detected AF episodes were reported in 10% to 70% of CIED patients when those with preexisting AF were included.9–12 In contrast, the prevalence of SCAF, which represents only device-detected AF episodes in patients without preexisting AF, ranged from 10% to 30% of the patients.^5–7,13 In our data, the incidence of SCAF, defined using a similar criteria to previous studies (an atrial high rate episode ≥175−180 bpm lasting >5−6 minutes), was 69% (344/497). One reason why our prevalence of SCAF was higher than in previous studies may be that our follow-up duration was much longer than that in any of the previous reports (mostly 1−2.5 years). In the REVEAL AF study, SCAF detection rates using implantable loop recorders at 30 days and 6, 12, 24, and 30 months were 6.2%, 20.4%, 27.1%, 33.6%, and 40.0%, respectively.²⁰

Additionally, consistent with previous studies,^5,7,14 the SCAF duration was <24 hours in the majority (83.4%) of our patients with SCAF. In a retrospective observational study including 394 biventricular pacemaker patients, only 7% of patients had SCAF episodes ≥24 hours within the first 6 months of follow-up.⁷ In a substudy of ASSERT (Asymptomatic Atrial Fibrillation and Stroke Evaluation in Pacemaker Patients and the Atrial Fibrillation Reduction Atrial Pacing Trial), SCAF ≥24 hours was detected in 10.7% of 2455 PPM or ICD patients during mean follow-up of 2.5 years.¹⁴

Unexpectedly, baseline age showed an inverse relationship with the degree of SCAF burden. Interestingly, several studies also revealed that CIED patients with SCAF were likely to be younger than those without.^6,7,21 However, the inverse relationship probably happened by chance with their relatively small sample sizes. In a large-scale study like the ASSERT subanalysis, there was a graded increase in the mean age with the duration of the longest SCAF.¹⁴ Therefore, if our sample size had been large enough as in the ASSERT study, a positive correlation might have been found in our results as well.

Patients in the low-burden group revealed no significant differences in overall prognosis compared with those in the No SCAF group (Figure 2), suggesting their low-burden of SCAF would not have been enough to induce adverse clinical outcomes. In contrast, the high-burden SCAF group experienced adverse outcomes much more frequently, which were primarily driven by progression to clinical AF and ischemic strokes (Figure 2). Approximately half of the patients in the high-burden group developed overt clinical AF during the follow-up. Interestingly, several previous studies also demonstrated that presence of SCAF lasting ≥24 hours of duration was associated with a significantly increased risk of clinical AF, thromboembolic events, and HF admission in CIED patients.^7,14,15

However, in most of the previous studies, the SCAF burden was not determined by the total cumulative time spent in SCAF but rather by the longest SCAF episode within a given follow-up interval. Accordingly, the longest SCAF episode alone might not sufficiently reflect the actual SCAF burden. Moreover, temporal precedence of the longest SCAF episode before stroke events was only proven in 20% to 50% of patients in previous reports.^7,22,23 Not only the longest SCAF episode but also a cumulative sum of multiple SCAFs (with various durations) may contribute to the aggravation of LV function, atrial cardio(myo)pathy, subsequent thrombus formation and stroke, and progression to clinical AF. The importance of the cumulative duration of AF has also been supported by several recent studies showing that persistent/permanent AF was more closely associated with increased risks of adverse outcomes than paroxysmal AF.²⁴ In our 6 patients with non-AF progression composite adverse events, the presence of high-burden SCAF was always detected before the occurrence of such adverse events (Table in the Data Supplement).

LA size, one of the well-known markers of atrial cardio(myo)pathy, was also closely associated with composite outcomes including ischemic stroke even after adjusting for SCAF in our data. Therefore, atrial cardio(myo)pathy and SCAF, independently or interacting with each other, might elevate the risk of ischemic stroke in PPM patients.²⁵ In addition, chronic kidney disease (CKD) was also independently associated with ischemic stroke even if adjusted with variables in Table 3 (odds ratio, 14.1 [95% CI, 1.60−121], P=0.016). This finding may be the function over-fitting. However, CKD, sharing many common risk factors with ischemic stroke, was reportedly prone to prothrombotic/procoagulant state.^26,27 Indeed, in our data, patients with CKD had a significantly higher CHA₂DS₂-VASc score than those without CKD (3.5±1.3 versus 2.3±1.3, P<0.001). Besides, CKD was identified as an independent risk factor of ischemic stroke in previous studies.^28–30 Therefore, CKD could serve as a real risk factor for ischemic stroke in PPM patients with SCAF. However, more data are needed to confirm this result.

However, overall event rates in our study were lower than those in other studies, which may have been at least partly a function of the less severe risk profiles of our patients, as suggested by their lower CHA₂DS₂-VASc scores and much younger ages compared with those in previous studies.^5–7

The causal relationship between device-detected SCAF with the risk of stroke or systemic embolism remains controversial, as does the issue of optimal anticoagulation strategies for SCAF patients, particularly when the burden or duration of SCAF is significant. In a recent study, anticoagulation therapy for SCAF patients, mostly using non-vitamin K antagonist oral anticoagulants, was not associated with a significant difference in the risk of thromboembolism or major bleeding.³¹ However, this study was not randomized and had a small sample size, comparing only 69 patients with anticoagulation therapy and 17 without. Additionally, their SCAF duration (≥6 minutes) may not have been sufficiently long enough to justify anticoagulation therapy. In contrast, in a retrospective analysis of 1712 PPM or ICD patients with CHA₂DS₂-VASc scores ≥2, Perino et al³² found that anticoagulation therapy in patients with device-detected AF >24 hours was associated with a considerable reduction in stroke risk (hazard ratio, 0.28 [95% CI, 0.10–0.81]; P=0.02).³² Similarly, our 4 stroke patients in the high-burden SCAF group had also high CHA₂DS₂-VASc scores (5.8±2.2). Therefore, anticoagulation therapy using non-vitamin K antagonist oral anticoagulants with a lower bleeding risk may be a reasonable option for CIED patients with both high SCAF burdens (eg, >24 hours) and high CHA₂DS₂-VASc scores. Two large randomized trials currently ongoing may provide further clarification of the role of non-vitamin K antagonist oral anticoagulants for SCAF in CIED patients: the ARTESiA (Apixaban for the Reduction of Thromboembolism in Patients With Device-Detected Subclinical Atrial Fibrillation; URL: https://www.clinicaltrials.gov. Unique identifier: NCT01938248) and NOAH studies (Non-Vitamin K Antagonist Oral Anticoagulants in Patients With Atrial High Rate Episodes; URL: https://www.clinicaltrials.gov. Unique identifier: NCT02618577) using apixaban and edoxaban, respectively. Regarding nonstroke outcomes, additional research is warranted to determine whether more aggressive rhythm control might improve the prognosis of patients with high SCAF burdens.

We acknowledge that this study has several limitations inherent to a single-center retrospective observational study. Only patients with PPM were evaluated in this study; accordingly, our results might not be applied to those with ICDs or biventricular pacemakers. However, patients with ICDs or biventricular pacemakers are more likely to have reduced ejection fraction and higher risks of AF and stroke. Therefore, more attention needs to be paid to SCAF patients with such devices. Unfortunately, not all device interrogation data were stored in our registry, so a reliable assessment of the association between clinical outcomes and the longest SCAF episode with many missing values was not feasible. The study sample size and the number of clinical outcomes were small to clearly demonstrate the effects of SCAF burden on more serious end points (ischemic stroke, MI, HF events, and cardiac death) or to address the issue of anticoagulation therapy in this patient group. Additionally, our definition of high-burden SCAF was arbitrary. Therefore, additional prospective studies with larger sample sizes should be conducted to validate our findings and resolve these limitations.

SCAF episodes were recorded in a significant proportion of PPM patients without a previous history of AF. Approximately 10% of patients were classified into the high-burden SCAF group, which was found to have increased risks of composite adverse outcomes, including progression to clinical AF, ischemic stroke, MI, HF hospitalization, and cardiac death, than the low-burden or no SCAF groups. The presence of high-burden SCAF, which always preceded ischemic stroke events, was identified as an independent predictor for composite adverse outcomes and progression to clinical AF. Therefore, anticoagulation therapy using non-vitamin K antagonist oral anticoagulants with a lower bleeding risk may be a reasonable option for PPM patients with both high-burden SCAF and high CHA₂DS₂-VASc scores.