In the current study we showed that ischaemic cardiomyopathy and a BNP level > 163 pmol/l at baseline were associated with an increased risk of ventricular arrhythmias after CRT implantation. Other studies evaluating baseline predictors of ventricular arrhythmias in a CRT population found variable results but identified gender, no beta-blocker or ACE inhibitor use, severely decreased LVEF and NYHA class IV as predictors of ventricular arrhythmias [
18‐
20]. Friedman et al. found that LV end-diastolic diameter was a strong predictor of ventricular arrhythmias in a CRT population even when corrected for reverse remodelling after CRT [
20]. Moreover, Van der Heijden et al. found LV end-systolic volume > 130 ml predictive of appropriate ICD therapy [
21]. Although we could not confirm these results we did see a trend for larger cardiac volumes at baseline to be associated with a higher risk of appropriate ICD therapy. Furthermore, most of these baseline predictors of ventricular arrhythmias after CRT implantation seem to relate to the degree of heart failure prior to implant, as does BNP level, which was a predictor in our population. The variability among studies indicates that prior to implant it is hard to identify valid predictors of ventricular arrhythmias, probably caused by the dynamic nature of CRT which induces changes in baseline variables associated with processes important in arrhythmogenesis, such as wall stress [
4], oxygen consumption and supply [
22], and neurohormonal activation [
23].
CRT and arrhythmogenesis
Regarding our second aim, we found that patients with a favourable echocardiographic outcome at 6 months had a significantly lower risk of ventricular arrhythmias compared with those showing a less favourable outcome. Although the majority of previous single-centre studies analysing the effect of reverse remodelling after CRT implantation on the risk of ventricular arrhythmias found comparable results, there was some controversy. The observed discrepancy might be due to different definitions of volume response, inclusion of both primary and secondary prophylactic ICD patients and different follow-up durations [
2‐
6,
19]. We showed absolute LVEF at 6 months to be a stronger predictor of ventricular arrhythmias than the occurrence of volume response. An explanation might be that volume response only reflects a relative improvement while absolute LVEF after CRT implantation is more strongly associated with pre- and post-CRT cardiac volumes. Sub-analysis of heart failure aetiology showed that volume response in patients with ischaemic cardiomyopathy was not significantly associated with the risk of ventricular arrhythmias while there was a trend towards a significant correlation with LVEF at 6 months and appropriate ICD therapy. These findings might have multiple explanations. First of all, it is known that patients with ischaemic cardiomyopathy respond less to CRT than those with non-ischaemic cardiomyopathy [
24]. Moreover, the mechanism of ventricular arrhythmias in ischaemic cardiomyopathy patients is most often a re-entry tachyarrhythmia involving the scar tissue [
25]. It is unlikely that response to CRT is of influence on these ventricular arrhythmias. Finally, response to CRT enables the patient to perform at a higher level of activity; in patients with ischaemic cardiomyopathy this could increase the ischaemic events and therefore trigger ischaemia-induced ventricular arrhythmias.
Although the majority of the evidence shows that favourable echocardiographic outcome after CRT is associated with a lower risk of ventricular arrhythmias, this is compared with patients who show a less favourable response to CRT. It is unclear whether CRT itself is antiarrhythmic as most studies did not have an ICD-only group for comparison. Moreover, there are also some suggestions for a proarrhythmic effect based on small case cohorts and reports [
26]. Most of these proarrhythmic events after CRT are linked to pacing in scar tissue but it is also possible that reversal of transmural activation pattern, due to LV epicardial pacing, increases repolarisation dispersion and thereby increases the risk of ventricular arrhythmias. Although we did not have an ICD-only group either, previous occurrence of first appropriate ICD therapy among primary prophylactic one- or two-chamber ICD patients implanted in our hospital was analysed by Wijers et al. [
27]. Eighteen percent (
N = 55) (mean LVEF 24 ± 6 %) received appropriate ICD therapy during a follow-up of 31 ± 17 months. As 25 % of our CRT-D patients received appropriate ICD therapy during a median follow-up of 37 months, it seems the overall effect of CRT on arrhythmogenesis is neutral. However, when looking at the occurrence of ventricular arrhythmias in non-responders (32 %) and responders (14 %) it suggests competing proarrhythmic effects in non-responders and antiarrhythmic effects in responders. Although this is in line with the results of the REVERSE study and the MADIT CRT trial, any conclusion based on comparison of these numbers is limited since an adequate baseline comparison was not performed and therefore other patient characteristics could have influenced the results [
2,
3].
CRT-D or CRT-P
The majority of CRT devices implanted in the Netherlands are CRT-D devices [
28]. Even though all patients eligible for CRT have an indication for primary prophylactic ICD therapy based on their depressed LVEF, the survival benefit of CRT-D over CRT-P is still matter of debate. In the choice between CRT-D and CRT-P, the 2013 ESC guideline, endorsed by our national society, the NVVC, states that no strict recommendations on device choice can be made due to insufficient evidence from randomised controlled trials. The guideline merely aims to offer guidance regarding the selection of patients for CRT-D versus CRT-P based on clinical condition, device-related complications as inappropriate ICD therapy, and costs [
1]. We showed that 10 % of our CRT-D population received inappropriate ICD therapy, which is associated with healthcare costs, and CRT-D device costs are much higher than CRT-P costs. Therefore, careful consideration in the choice of CRT device is warranted. Moreover, basing CRT device choice on pre-CRT clinical condition is questionable as we, and the majority of studies, have shown response to CRT to influence the risk of ventricular arrhythmias.
Only first appropriate ICD therapy was analysed, as we opted that the risk of first appropriate ICD therapy was of most interest in the decision to implant a CRT-D or a CRT-P, because one episode of appropriate therapy can in theory have prevented a sudden cardiac death. Whether the patient receives more appropriate ICD therapy after this episode is of limited additional value in CRT device choice.
Patients with an LVEF > 35 % after CRT implantation were at a low risk (6 %) of appropriate ICD therapy during 37 months of follow-up. Previously, Van Boven et al. [
8] actually showed that none of the patients who reached an LVEF ≥ 35 % after 4 months of CRT needed appropriate shock therapy during a follow-up of 3 years. Manfredi et al. [
9] and Schaer et al. [
10], who used a higher cut-off in LVEF, found similar results and Ruwald et al. [
11] showed with 752 patients from the MADIT CRT trial that patients who achieve LVEF normalisation (> 50 %) have a very low absolute and relative risk of ventricular tachyarrhythmias within 2.2 years of follow-up.
Therefore it would be of great value to correctly predict which patients are most likely to reach an LVEF > 35 % as these patients could be candidates for CRT-P implantation. We propose that the characteristics associated with reaching this LVEF in our cohort, namely non-ischaemic cardiomyopathy, higher IVMD, and higher LVEF, could be used as pre-implantation predictors of patients suitable for CRT-P.
Although these characteristics have also been previously linked to response to CRT in other cohorts [
29,
30], prediction of echocardiographic outcome has proven to be difficult. As non-responders might actually be at increased risk of ventricular arrhythmias and responders could still be at risk during the initial 6 months the association found might be of more value for the decision to downgrade a CRT-D in case of generator replacement. Meaning that if a patient has an LVEF > 35 % at the time of generator replacement, CRT-D downgrading to a CRT-P could be considered. This consideration should be restricted to patients with non-ischaemic cardiomyopathy as the association between echocardiographic outcome and appropriate ICD therapy was much weaker in ischaemic cardiomyopathy patients. Moreover, it is unclear whether the relationship between LVEF and risk of ventricular arrhythmias at the time of generator replacement, approximately 5 years after implant, is comparable with the relationship found during the currently evaluated period. We recommend further research into this correlation.