Atrial fibrillation and implantable cardioverter-defibrillator in non-ischaemic heart failure with reduced ejection fraction: insights from the DANISH trial (VSports最新版本)

Study participants

Patients with a diagnosis of HF were eligible if they had a non-ischaemic cause of HF (preferably determined by coronary angiography, but a normal computed tomographic angiogram or nuclear myocardial perfusion imaging was accepted), had a left ventricular ejection fraction (LVEF) of ≤35%, were in New York Heart Association (NYHA) functional Class II or III [or Class IV if cardiac resynchronization therapy (CRT) was planned], had a N-terminal pro-B-type natriuretic peptide (NT-proBNP) concentration >200 pg/mL, and were optimally treated with medical therapy for HF. Patients with preexisting pacemakers or CRT-pacemakers could also be included in the study if they were willing to accept a potential upgrade. Patients fulfilling indications for a CRT device received a CRT-defibrillator (if randomized to the ICD arm) or a CRT-pacemaker (if randomized to the control arm). Key exclusion criteria were a resting heart rate >100 beats per minute in patients with permanent AF and renal failure that was being treated with dialysis. A full list of exclusion criteria is provided in the main paper.¹⁵

Definition of atrial fibrillation

Data on a history (and chronicity) of AF were collected from the trial case report forms (CRF). Investigators were first asked whether patients had a history of AF any time before enrollment and then to specify the type of AF according to the following options: paroxysmal, persistent, and permanent. Data on heart rhythm on the electrocardiogram (ECG) at enrollment were also collected on the CRF, and investigators were asked to specify the heart rhythm from the following options: sinus rhythm, AF, atrial flutter, paced rhythm, and other.

As in previous studies, the definition of AF in the present study included AF or atrial flutter.^17,18 The assessment of the effect of ICD implantation according to permanent AF status at baseline was a prespecified subgroup analysis. In the present analysis, we made the following comparisons based on AF status at baseline: (i) patients without any AF (no history of AF and no AF on enrollment ECG) vs. any AF (a history of AF or AF on enrollment ECG); (ii) patients without any AF (no history of AF and no AF on enrollment ECG) vs. non-permanent AF (a history of paroxysmal/persistent AF or AF on enrollment ECG without a history of AF) vs. permanent AF (a history of permanent AF). This classification was used in all analyses.

Follow-up and outcomes

In the main trial, patients were followed from randomization until death or 30 June 2016, whichever came first. In the present study with extended follow-up, patients were followed until death or 18 May 2020, whichever came first, and no patients were lost to follow-up.¹⁹ As in the main trial, the primary outcome was all-cause death, and secondary outcomes were cardiovascular death and sudden cardiovascular death. All outcomes were assessed by review of patients’ electronic and hard copy medical files and adjudicated by an event committee blinded to treatment allocation.

Device complications and shock therapy among patients randomized to an ICD were also examined, although these data were available only for the original, and not the extended, follow-up duration.

Statistical analyses

Baseline characteristics were summarized as frequencies with percentages, means with standard deviations, or medians with 25th–75th percentiles, and differences were tested using the χ² test or the Fisher’s exact test for categorical variables and unpaired t-test, the Wilcoxon test, ANOVA, or the Kruskal–Wallis test for continuous variables, as appropriate.

The association between AF status at baseline and outcomes was evaluated using Cox proportional hazards regression models, adjusted for age, sex, treatment assignment, centre, CRT (preexisting or planned), body mass index (BMI), duration of HF, NYHA functional class, log of NT-proBNP, LVEF, estimated glomerular filtration rate (eGFR), history of HF hospitalization, hypertension, and diabetes.

The effect of ICD implantation vs. usual care according to AF status at baseline was examined with the Kaplan–Meier estimator, Aalen–Johansen estimator, and Cox proportional hazards regression models, stratified according to centre and CRT implantation (preexisting or planned). Since there was a difference in sex between the ICD and control arm in patients with non-permanent AF, these analyses were also adjusted for sex.

The effect of ICD implantation was also examined according to continuous age in patients with and without AF at baseline, respectively, using fractional polynomial models, restricted cubic spline models, or linear regression models, whichever had the lowest Akaike information criterion score. These analyses were conducted, because the main DANISH trial demonstrated that ICDs reduced the rate of all-cause death in younger, but not in older, patients.

Data were analysed according to the intention-to-treat principle. All analyses were conducted using STATA version 17.0 and SAS version 9.4. A P-value of 0.05 was considered statistically significant.

Patient characteristics

Type of atrial fibrillation at baseline

Patient characteristics according to type of AF at baseline (i.e. no AF vs. non-permanent AF vs. permanent AF) are presented in Supplementary material online, Table S2. Patients with permanent AF were older, more often male, and had a lower eGFR and a higher NT-proBNP level. They were also more likely to have a prior stroke and a longer duration of HF. Regarding background HF therapy, there was no significant difference between the groups. The proportion of patients treated with amiodarone was highest in the non-permanent AF group, whereas the proportion of patients treated with anticoagulants was the highest in the permanent AF group. The proportion of patients treated with loop diuretics was highest in patients with AF, with no difference between the non-permanent and permanent AF groups. Patients with permanent AF less often had a preexisting or planned CRT implantation.

When baseline characteristics were examined according to treatment assignment and type of AF at baseline, there was no significant differences between the usual care arm and the ICD arm across AF types, with one exception: among participants with non-permanent AF, the proportion of males was significantly lower in the usual care arm than in the ICD arm (see Supplementary material online, Table S3).

Outcomes according to atrial fibrillation status at baseline

Effect of implantable cardioverter-defibrillator implantation on outcomes according to AF status at baseline

Outcomes according to AF status at baseline

It remains uncertain if AF is an independent predictor of adverse outcomes or simply a marker of more advanced HF. The conflicting findings in previous studies may, in part, reflect residual confounding and varying durations of follow-up. In the present analysis of the DANISH trial, during a median follow-up of more than 9 years, AF was associated with a higher mortality risk, including sudden cardiovascular death, even after comprehensive adjustment for potential confounders. However, these associations were no longer significant after adjustment for NT-proBNP, the single strongest predictor of adverse outcomes in HF. These findings are in line with those from a pooled analysis of the Prospective comparison of ARNI with ACEI to Determine Impact on Global Mortality and Morbidity in Heart Failure (PARADIGM-HF) and Aliskiren Trial of Minimizing Outcomes for Patients with Heart Failure (ATMOSPHERE) trials, but not the Dapagliflozin And Prevention of Adverse outcomes in Heart Failure (DAPA-HF) trial, where patients with AF did not have a higher risk of all-cause and cardiovascular death, even in unadjusted analyses.^13,17,20 Although these observations from the DAPA-HF trial do not support the view that AF is a marker of more advanced HF, the median follow-up of 18 months in the trial might have been too short to detect a mortality difference. Collectively, our observations, coupled with the findings from prior reports, do not fully support the view that AF is an independent prognostic factor of adverse outcomes in patients with HFrEF, but rather a marker of more advanced disease.

Effect of implantable cardioverter-defibrillator implantation on outcomes according to AF status at baseline

As alluded to in the Introduction, there is some biological plausibility that the effects of primary prevention ICD implantation may differ according to AF status. While the effects of ICD implantation in patients with HFrEF and concomitant AF have been examined in landmark primary prevention ICD trials, the findings have been inconsistent, data on the effects of ICD implantation on sudden cardiac death were not reported, and the proportion of patients with AF was low (≤25%). In the Multicenter Automatic Defibrillator Implantation Trial II (MADIT-II) trial, ICD implantation, compared with usual care, significantly reduced the risk of all-cause death in patients with ischaemic HFrEF, and this beneficial effect was evident in participants with concomitant AF, although a direct comparison of the treatment effect in patients with and without AF was not performed.¹⁰ Conversely, in the Sudden Cardiac Death in Heart Failure (SCD-HeFT) trial, which enrolled patients with both ischaemic and non-ischaemic HFrEF, the opposite appeared to be true since the beneficial effect of an ICD appeared to be attenuated (or absent) in individuals with AF, although, again, no formal test of interaction was performed.¹¹ In the Defibrillators in Non-ischemic Cardiomyopathy Treatment Evaluation (DEFINITE) trial, the association between ICD implantation and all-cause death did not seem to be different in patients with and without AF, although this trial did not demonstrate an overall survival benefit with a primary prevention ICD.¹²

In the DANISH trial, the largest and most contemporary primary prevention ICD trial in patients with non-ischaemic HFrEF, the proportion of patients with AF was substantially higher than in any of the landmark ICD trials (38% vs. ≤25%). In the present extended follow-up analysis of DANISH, AF status significantly modified the effect of ICD implantation on cardiovascular death, such that ICD implantation was associated with a lower rate of this outcome in patients with AF, but not in those without AF. Although the interaction tests for all-cause death and sudden cardiovascular death were not statistically significant, there appeared to be a trend towards a greater effect of ICD implantation on these outcomes in patients with vs. without AF. These effects were generally consistent, irrespective of AF type (i.e. permanent and non-permanent AF).

As with any subgroup analyses, prespecified or not, these cannot provide conclusive evidence of a treatment effect given the lack of statistical power and the risk of spurious associations (i.e. a chance finding). Therefore, the results from the present, and previous, analyses of the effect of ICD implantation according to AF status should be interpreted with caution and considered as hypothesis-generating at best. However, although it is possible that the observed interaction between AF status and the effect of ICD implantation in the DANISH trial may simply have resulted by chance due to multiple testing, there is some biological plausibility for a differential effect of ICD implantation in patients with and without AF. An ICD can prevent sudden cardiovascular death caused by ventricular tachyarrhythmia, severe bradycardia, or complete heart block, but cannot provide protection against other causes of sudden death including ‘pump failure’ or terminal HF. Patients with AF may derive greater benefit from ICD therapy because AF contributes to both ventricular tachyarrhythmias, through mechanisms such as irregular ventricular conduction, short–long–short sequences, structural remodeling, and medication-related pro-arrhythmia, and bradyarrhythmias related to conduction disease or rate-controlling medications. In contrast, in patients without AF, non-arrhythmic mortality may represent a larger competing risk, attenuating the relative benefit of ICD implantation. Indeed, in the DANISH trial, patients with AF had a higher risk of sudden cardiovascular death than those without AF, and the proportion of deaths attributed to sudden cardiovascular death was substantially higher in patients with vs. without AF. Given that the beneficial effect of ICD implantation was mainly observed in younger (and not older) individuals with AF, this hypothesis may be especially relevant in these individuals.

Although our findings support the use of ICD implantation for the prevention of sudden cardiovascular death in patients with non-ischaemic HFrEF and concomitant AF (and especially those of younger age), regardless of AF type, it is important to emphasize that the risk of inappropriate shock is not negligible in these patients. Indeed, despite improvements in ICD devices, the use of less aggressive ICD settings, and the exclusion criterion of dysregulated permanent AF, more than 8% of patients with AF in the DANISH trial experienced inappropriate shocks, corresponding to a two-fold higher risk compared to patients without AF. This highlights the importance of careful patient selection,²¹ optimization of device programming, and continued strategies to reduce the risk of inappropriate shocks in patients with AF.

Limitations

The main strength of the present study is the long follow-up in the setting of a randomized, controlled multicentre trial with no loss to follow-up. However, there are some important limitations. First, although the assessment of the effect of ICD implantation on death from any cause according to permanent AF status was prespecified, the assessment of the effect of ICD implantation according to any AF status (as well as type of AF), including on secondary outcomes, was not. Second, our findings may not be generalizable to all patients with HFrEF, since the DANISH population was predominantly White. Furthermore, the enrollment of a specific subgroup at higher risk of adverse outcomes such as all-cause mortality and new-onset HF, for example, those with unregulated permanent AF,²² was precluded by the prespecified inclusion and exclusion criteria of the DANISH trial. Third, since the DANISH trial only enrolled patients with non-ischaemic HFrEF, our findings cannot be extrapolated to those with HFrEF of ischaemic origin. Fourth, data on the type of AF were investigator-reported, and the possibility of some degree of misclassification cannot be excluded. In addition, data on AF duration at the time of enrollment, new-onset AF or progression of existing AF during follow-up (e.g. from paroxysmal to persistent or permanent AF), treatment of AF during follow-up (including pharmacological treatment and ablation), and echocardiographic data (including data on left atrial size and function) were not available. In this regard, catheter ablation of AF is of particular interest, since it may lead to recovery of left ventricular function in selected patients with non-ischaemic HFrEF, thereby mitigating the indication for ICD implantation and possibly the risk of sudden cardiovascular death.^23–27 Fifth, the DANISH trial was conducted before angiotensin receptor–neprilysin inhibitors and sodium-glucose cotransporter-2 inhibitors were recommended by HFrEF guidelines, and since both drug classes have been shown to reduce sudden cardiovascular death when added to conventional therapy, the absolute (and relative) risk of sudden cardiovascular death is expected to decline even further, potentially diminishing the benefit of ICD implantation on preventing this mode of death (and perhaps the effect modification of AF observed in this study).^28,29 Sixth, data on specific subtypes of cardiovascular and sudden cardiovascular death (e.g. ventricular tachyarrhythmia, bradyarrhythmia or pump failure) were not available. Seventh, 56% of the DANISH population received CRT shortly after randomization, with a higher rate in individuals without AF. However, in a previous analysis of the DANISH trial, we demonstrated that the effects of ICD implantation were not modified by CRT.³⁰ Finally, given the observational nature of the analyses on the association between AF status and outcomes, the possibility of residual confounding cannot be fully excluded despite adjustment for measured known confounders.