Introduction
Implantable cardioverter defibrillators (ICD) are a well-established and effective therapy for patients with heart failure with reduced ejection fraction (HFrEF) who survived sustained ventricular tachycardia (VT) or ventricular fibrillation (VF), or those who are exposed to an increased risk of these arrhythmias. (1) With regard to poor left ventricular systolic function, AF is often present in patients with ICD, with a prevalence from 17% to 36% at time of implantation in clinical practice. (2,3) Except for the negative effect on survival, (4,5) AF may interfere with ICD therapy and limits its clinical usefulness by inducing inappropriate and appropriate shocks. This issue is important because ICD shocks are painful, cause psychological disturbances, and may induce ventricular pro-arrhythmia and even death. (6-9) The aim of this review was to explore the possible interference mechanisms of AF with ICD therapy, examine the impact of this interference on survival, and discuss the treatment options available for its reduction.
Mechanisms of atrial fibrillation interference with implantable cardioverter-defibrillator therapy
ICDs are an important therapeutic option in reducing mortality due to ventricular arrhythmias in patients with HFrEF. (10) The primary goal of ICD therapy is to detect and stop the life-threatening ventricular arrhythmia by delivering electrical shock and thus prevent a sudden cardiac death. In that context, ICD shocks delivered for VT/VF are considered appropriate because they save lives. (11) In contrast, ICD shocks delivered for the non-ventricular arrhythmias, like AF, atrial flutter (AFL), supraventricular tachycardia (SVT), atrial tachycardia (AT), sinus tachycardia (ST), or non-arrhythmic events, like electromagnetic interference, myopotentials, and device malfunction or oversensing, are unnecessary and are considered inappropriate. (12)
AF may interfere with ICD therapy in two ways. The first way is more common, and occurs when the fast ventricular rate of AF with relatively regular R-R intervals reaches a device’s programmed detection zone for VT/VF, is misclassified as ventricular arrhythmia, and induces the delivery of inappropriate shock (Figure 1). According to data from randomized clinical trials, a history of AF was found in 9% to 24% of patients with ICD (Table 1), more often in patients with non-ischemic than in those with ischemic cardiomyopathy. (13-19) On the other hand, the incidence of new-onset AF lasting at least 24 hours per day was 6.3% and 7.1% in single chamber and doubled chamber patients with ICD, respectively. (20) Therefore, it is not surprising that AF/AFL was the most common reason for an inappropriate ICD shock, followed by ST/SVT and non-arrhythmic events. (21,22) In earlier randomized clinical trials, (15,21,23,24) the rate of inappropriate ICD shocks ranged from 15% to 21% over 20 to 45.5 months of follow-up (Table 1). Later, thanks to device programming adjustment and enhanced algorithm discrimination of arrhythmias, the rate of inappropriate shocks decreased, (18,19,25) but even with these improvements, they occur in 1.9% to 3.7% of patients with ICD over a period of two years. (26,27)
| RCT | No of ICD patients | Heart disease (%) | History of AF (%) | EF (median) | FU (m) (mean) | ICD shocks (%) Approp Inapprop |
|---|---|---|---|---|---|---|
| AVID, (13,23) | 507 | ICM 81 DICM 19 | 21 | 32% | 18 | 39 20 |
| MADIT II (14,21) | 719 | ICM 100 | 9 | 23% | 20 | 18 15 |
| DEFINITE (15) | 229 | DICM 100 | 23 | 29% | 29 | 18 21 |
| SCD-HeFT (16,24) | 811 | ICM 53 DICM 47 | 17 | 25% | 45 | 22 17 |
| MADIT-CRT (17,25) | 1790 | ICM 55 DICM 45 | 11 | 28% | 39 | 13 7 |
| MADIT-RIT (18) | 1500 | ICM 53 DICM 47 | 13 | 26% | 17 | 13 4 |
| DANISH (19) | 556 | DICM 100 | 24 | 25% | 67 | 12 6 |
| RCT = randomized controlled trial, No = number, AF = atrial fibrillation, EF = ejection fraction, FU = follow-up, m = months, ICD = implantable cardioverter-defibrillator, approp = appropriate, inapprop = inappropriate, ISM = ischemic cardiomyopathy, NICM = non-ischemic cardiomyopathy. | ||||||
The second way in which AF may interfere with ICD therapy is direct action of AF in causing an episode of VT/VF, with consequent appropriate shock therapy. An earlier analysis of stored ICD electrograms revealed that approximately 10% of VT episodes were preceded by atrial fibrillation typically associated with a rapid ventricular response. (28) Subsequently, Gronefeld et al. (29) reported that patients in AF experienced appropriate ICD therapy for recurrent ventricular arrhythmias more frequently than those in sinus rhythm (63% vs. 38%, p=0.01) because of a higher incidence of short-long-short cycles preceding ventricular arrhythmias in AF (50% vs. 16%, p=0.002). In the study by Stein et al. (30) which included 537 patients with dual-chamber ICD, 233 (8.6%) of 2602 VT/VF episodes were preceded by a paroxysm of AT or AF. The median duration of these arrhythmias preceding VT/VF was approximately one hour. Ventricular therapies were delivered to treat 205 of these VT/VF episodes, while the remaining 28 episodes terminated spontaneously before an ICD therapy could be delivered. On the other hand, Borleffs et al. reported that patients with permanent AF had higher risk of ventricular arrhythmias triggering ICD discharge compared with those without AF and those with paroxysmal or persistent AF (49% vs. 29%, 26%, and 26%, respectively; p<0.001). (4)
There are several mechanisms that can explain the association between AF and appropriate ICD shocks. First, it is possible that AF and ventricular arrhythmias share risk factors such as ischemia, renal failure, increased sympathetic tone, or increased left ventricular filling pressure, (31,32) which facilitate the initiation of both arrhythmias. Second, in structural heart disease with reduced EF, the rapid ventricular response caused by AF may directly affect ventricular refractoriness and therefore induce ventricular arrhythmia (Figure 2). (30,33) Third, the irregular rhythm of AF results in irregular ventricular activation, creating the short-long-short sequences, which lead to inhomogeneous depolarization and thus to a higher myocardium susceptibility for ventricular arrhythmias. (29)
Prognostic significance of appropriate and inappropriate implantable cardioverter defibrillator shocks
Two landmark primary prevention ICD studies, MADIT-II (Multicenter Automatic Defibrillation Trial) and SCD-HeFT (Sudden Cardiac Death in Heart Failure Trial), have found that both appropriate and inappropriate ICD shocks were significant predictors of death, increasing risk of death for appropriate shocks by 3 and 5 times, respectively, and with a 2-fold increase for inappropriate shocks. (21,24,34) In both studies, progressive HF was the leading cause of death among the patients who received at least one ICD shock. (24,34) A meta-analysis by Proietti et al. (35) summarized the existing evidence on the prognostic significance of ICD shocks in patients with HFrEF. In a pooled analysis of almost 200 000 patients with ICD, a significant association was found between ICD shocks and mortality, stronger for appropriate (HR 2.95, p<0.001) than inappropriate shocks (HR 1.71, p<0.001), while the combination of both appropriate and inappropriate shocks was associated with a higher risk of death (HR 4.18, p<0.001) than either type of shock alone.
How we can explain the association between ICD shocks and increased mortality? There are a number of potential mechanisms by which shocks may directly increase the risk of death, such as induced ventricular proarrhythmia, transient myocardial dysfunction, post-shock vasospasm, myocardial stunning, or post-shock pulseless electrical activity, (7,24,36-38) but these mechanisms can potentially be the cause of death within the first 24 hours of a shock. Another explanation arises from the fact that patients with HFrEF and more severe forms of heart disease have an increased tendency to develop ventricular arrhythmias and AF (34,39,40) and consequently a tendency towards a higher incidence of appropriate and inappropriate shocks, (4) which suggests the possibility that ICD shocks are indicator of a higher-risk patient. This explanation is illustrated in the study by Powel et al. (22) who analyzed the rhythms of the first shock delivery in 7 439 randomly-selected patients with an ICD or cardiac resynchronization therapy defibrillator (CRT-D) to determine their relationship with survival. Of the first shock episodes, 58.7% were appropriate shocks for sustained ventricular arrhythmias, and 41.3% were inappropriate shocks for non-ventricular arrhythmias or non-arrhythmic events. Compared with no-shock patients, those who received their first shock for VT/VF or AF/AFL had an increased risk of death (Table 2). In contrast, patients who received an inappropriate shock for ST/SVT or non-arrhythmic events (noise, artifact, and oversensing) had similar survival to those who did not receive a shock. Two important messages emerge from this study: 1) the risk associated with inappropriate shocks is limited to patients receiving shocks for AF/AFL, and 2) increased long-term mortality after shock is more related to the underlying arrhythmia substrate than to an adverse effect from the shock itself. Taking all this together, it is a plausible explanation that the association between ICD shocks and increased mortality is multifactorial, including more advanced heart disease, progression of the underlying arrhythmia substrate, co-morbid conditions, and, to a lesser extent, detrimental effects from the ICD shocks themselves. (22,24,25,41)
Strategies to minimize the interference of atrial fibrillation with implantable cardioverter therapy
According to recent guidelines, the optimization of ICD programming is the initial option to minimize the rate of ICD shocks induced by AF and to improve patient outcomes. (1) This recommendation is based on the results of two meta-analyses, (42,43) which examined the overall effect of ICD therapy reduction programming on several clinical outcomes, including mortality (Table 3). In both meta-analyses, therapy reduction programming was associated with a 50% reduction in inappropriate shocks and with a 30% and 23% reduction in mortality, respectively, compared with the conventional programming. No increase was seen in risk of appropriate shocks or syncope in therapy reduction programming. In line with this recommendation, ICD programming should include: 1) customized bradycardia pacing mode to prevent unnecessary right ventricular pacing in patients without an indication for permanent pacing, (44,45) 2) multi-zone tachycardia detection programming with incorporated prolonged detection time settings and high rate thresholds to avoid unnecessary shocks due to non-sustained VT or clinically stable VT, (46-48) 3) consistent use of algorithms for supraventricular versus ventricular arrhythmias discrimination even for tachycardias with rates up to 230 bpm, (17,26,48) 4) proper atrial sensing for activation of dual-chamber discriminators, (48) 5) systematic use of ATP before shock therapy, also for very fast VT, (46,49,50) and 6) specific algorithms for T-wave discrimination and right ventricular lead noise. (48)
Pharmacological therapy is the next option for the reduction of ICD shocks caused by AF, (1) but data on this issue are limited. The goal of pharmacological therapy is to slow down the heart rate so that it does not reach the detection zone for VT/VF, or to reduce AF recurrences. In a secondary analysis of the MADIT II study, (51) beta-blocker therapy (metoprolol 44%, carvedilol 42%, atenolol 13%) reduced ICD therapy for VT or VF, but failed to prevent inappropriate shocks induced by AF or SVT. On the other hand, a retrospective analysis of the MADIT-CRT (Multicenter Automatic Defibrillator Implantation with Cardiac Resynchronization Therapy) study found that the risk of inappropriate therapy due to AF was significantly reduced in patients receiving carvedilol compared with those who received metoprolol (HR: 0.50, p=0.004). (52) The superior clinical effect of carvedilol versus metoprolol was probably the result of its a larger electrophysiological and pharmacological profile. (53) In a randomized controlled trial on 412 patients with implanted ICD for secondary prevention, (54) a combination of amiodarone and beta-blocker was more effective in the prevention of inappropriate shocks (mostly for supraventricular tachyarrhythmias) than beta-blocker alone (3.3% vs. 15.4%, p=0.006), but with higher rates of adverse thyroid and pulmonary effects and of symptomatic bradycardia. On the other hand, sotalol was not more effective than beta-blockers in reducing risk of inappropriate shocks. Amiodarone can be used for rhythm control after cardioversion, but its success rate in the prevention of AF recurrences in patients with ICD or CRT-D with HFrEF and persistent AF was not higher than 34% at two-year follow-up. (55)
Catheter ablation of the AF or AV junction is recommended in patients with AF-related inappropriate shocks who are unresponsive to pharmacological therapy. (1) Several studies confirmed the feasibility and usefulness of these therapeutic options. (55-58) In 2014, Kosiak et al. (56) published the results of AF catheter ablation in 73 patients with implanted ICD due to ischemic or non-ischemic cardiomyopathy. Complete pulmonary vein isolation (PVI) as the procedural endpoint was achieved in all patients, and re-ablation was necessary in 20 of them. In comparison with the period prior to ablation, there was significant reduction of appropriate and inappropriate shocks after ablation (p=0.03 and p=0.001, respectively). In the study by Miyazaki et al., (57) 14 of 106 ICD patients experienced inappropriate ICD shocks due to AF. PVI was performed in 13 patients, one patient underwent an AVJ ablation because of the absence of an inferior vena, and 4 patients were re-ablated for recurrent AF. After the last ablation procedure, no patients experienced inappropriate shock during the median follow-up of 19 months. In a pooled analysis of 664 patients with CRT-D and permanent AF, (58) patients with AVJ ablation had a significantly lower rate of inappropriate shocks induced by AF and a lower rate of appropriate shocks than those treated with drugs (both p<0.001). These data are important for the treatment of patients with CRT-D, because approximately 26% of this population have AF. (59)
Conclusion
Atrial fibrillation may interfere with ICD therapy in two ways: 1) by inducing inappropriate ICD shock when its rapid ventricular rate reaches a device’s programmed detection zone for VT/VF, and 2) by directly inducing a VT or VF episode with consequent appropriate shock therapy. Patients with ICD who receive appropriate or inappropriate shocks related to AF have a substantially higher risk of death than those who do not receive them. The association between ICD shocks and increased mortality is multifactorial, including more advanced heart disease, progression of the underlying arrhythmia substrate, co-morbid conditions, and, to a lesser extent, detrimental effects from ICD shock itself.
The optimization of ICD programming is the initial option for reducing inappropriate and appropriate shocks induced by AF. Pharmacological therapy with beta-blockers for rate or amiodarone for rhythm control is the next option to this effect. The catheter ablation of AF or AV junction is recommended in patients unresponsive to pharmacological therapy depending on whether it is paroxysmal and persistent AF, or permanent AF.