Cardiac resynchronization therapy and electrical storm: results of the OBSERVational registry on long-term outcome of ICD patients (OBSERVO-ICD)

Abstract

Aims

Electrical storm (ES) is a condition defined as three or more episodes of ventricular fibrillation (VF) or ventricular tachycardia (VT) within 24 h, and usually coexist with advanced heart failure in patients with structural heart disease. The aim of the present study is to test whether cardiac resynchronization therapy (CRT) can be associated with a lower incidence of ES.

Methods and results

The OBSERVO-ICD (NCT02735811) is a multicentre, retrospective registry, enrolling all consecutive patients undergoing ICD or CRT-D implantation from 2010 to 2012 in five Italian high-volume arrhythmia centres. Propensity score matching was used to compare two equally sized cohorts of ICD and CRT-D patients with similar characteristics. The primary endpoint was the time free from ES. Secondary endpoints were time free from unclustered VT/VF episodes and time free from ES in CRT-D patients according to clinical or echographic response. CRT-D was associated with a 45% relative risk reduction in ES when compared with ICD (5.6% vs. 12.3%; log rank P = 0.014). CRT-responders presented lower rates of ES when compared with non-responders and negative responders according to both clinical and echographic criteria (log-rank P = 0.017 and 0.023, respectively). No ES was detected in any of the 133 full responders to CRT-D. Clinical and echographic positive responses, but not CRT-implant per se, were associated with lower estimate rates of unclustered VTs/VFs.

Conclusion

Patients with CRT had a lower incidence of ES when compared with propensity-matched ICD patients. The long-term benefit of CRT seems to be due to the improved haemodynamics, as CRT-responders performed markedly better over a long-term follow-up.

Introduction

Electrical storm (ES) is a clustering of at least three episodes of ventricular fibrillation (VF) or ventricular tachycardia (VT) within a short period of time, usually requiring immediate electrical treatment through an external defibrillator or, more frequently, an implantable cardiac defibrillator (ICD).^1,2 The incidence of ES is steadily rising worldwide along with the number of patients with ICD.³ Recent estimates show that 4–7% of primary prevention patients and 10–58% of secondary prevention patients will experience an ES at some time after the implant.^4,5 ES is associated with an increased cardiac and all-cause mortality,⁶ and is frequently related to heart failure (HF).⁷ The relationship between ES and HF is still unclear, but these two entities seem to share very similar rates of cardiac death, all-cause mortality, and rehospitalization, with HF representing the first cause of death in patients with ES.⁸ It has also been suggested that the occurrence of ES could represent a clinical manifestation of HF worsening rather than an independent event in patients with HF secondary to structural heart disease and ICD.^7,8 In this setting, cardiac resynchronization therapy (CRT) could potentially prevent ES episodes through the improvements in atrioventricular synchrony and left ventricular function and the reduction in reverse remodelling, functional mitral regurgitation, and septal dyskinesia.

On the other hand, recent reports have questioned the safety of biventricular pacing, showing how drug-refractory ES could start right after CRT implant, thus underlying a potential proarrhythmic effect of the epicardial pacing.^9,10

The aim of the present study was to determine whether CRT can be associated with a reduction in the incidence of ES and whether this association could vary according to the clinical and echocardiographic response to resynchronization.

Methods

The OBSERVational registry On long-term outcome of ICD patients (OBSERVO-ICD) is a multicentre, retrospective registry whose methods have been previously described.¹¹ The registry was endorsed by the Italian Association of Arrhythmology and Cardiac Pacing (AIAC) and enrolled all consecutive patients aged ≥18 years who underwent an ICD or CRT-D implant from 1 January 2010, to 31 December 2012, in one of the five participating high-volume arrhythmia centres.

The study was approved by the Ethics Committee of the proposing centre and was registered into www.clinicaltrials.gov (NCT02735811). All patients gave their written informed consent at the time of enrolment. The study was reported according to the STROBE statement (see Supplementary material online, Table S1).

Data collection

Demographics, clinical history, cardiovascular risk factors, and implant-related data were collected for all patients. A transthoracic echocardiogram was performed in all patients prior to device implant, and after 6 months in patients with CRT-D. Clinical response to CRT was defined as an improvement of at least one NYHA functional class at 6-month follow-up. The echocardiographic response was defined as an increase in left ventricular end-systolic volume ≥10% at 6-month.¹² Negative clinical response to CRT was defined as a worsening of at least one NYHA functional class at 6-monts follow-up. The negative echocardiographic response was defined as a lowering in left ventricular end-systolic volume after 6 months. All patients who did not meet both the definitions for positive and negative response were defined as non-responders.

Data regarding arrhythmic episodes were collected through ambulatory reports and remote monitoring transmissions, as previously reported.¹¹ ES was defined according to the currently accepted criteria as three or more VF or VT episodes within 24 h, each of those separated by at least 5 min and successfully terminated by the ICD, or by an incessant VT lasting more than 12 h.² Pseudo-storm, defined as ES due to inappropriate ICD shocks, was considered as device malfunctioning and not included in the present analysis.

Endpoints

The primary endpoint was the time free from ES in patients with ICD or CRT-D. Secondary endpoints were: (i) time free from unclustered VT/VF episodes; (ii) time free from ES in CRT-D patients according to the clinical and echocardiographic response after 6 months.

Statistical analysis

Quantitative variables were checked for normality by the Kolmogorov–Smirnov test. Normally distributed variables were described as mean ± standard deviation. Not-normally distributed variables were described as median and first to third quartile range (1st–3rd quartile). Categorical variables were assessed by using χ² analysis and described as absolute or relative prevalence. ANOVA was used to compare normally distributed quantitative variables. Kruskal–Wallis ANOVA was used to compare non-normally distributed quantitative variables.

A propensity score for the likelihood of CRT implant was obtained by means of multiple logistic regression. The variables included in the score were: age, sex, NYHA class, hypertension, diabetes, dyslipidaemia, chronic renal disease, chronic obstructive pulmonary disease, previous stroke or TIA, atrial fibrillation, ischaemic cardiomyopathy, idiopathic dilated cardiomyopathy, hypertrophic cardiomyopathy, ion channel disease, ARVD, medical therapy with ACE-Is, ARBs, β-blockers, MRAs, ASA, amiodarone, digitalis, other AADs, secondary prevention, and LVEF. Matching was then performed on log-transformed propensity score in a 1:1 fashion with a calliper of 0.1 in order to account for the different baseline characteristics between ICD and CRT-D patients. Kaplan–Meier analysis with a log-rank P test was used in order to compare the endpoints between ICD and CRT-D groups and between responders and non-responders to CRT. Given the enrolled population of 509 CRT patients and 810 non-CRT patients, we calculated that the present analysis had a 92% power to detect a significant difference in ES incidence between ICD and CRT-D groups, considering the incidence of ES already available in current literature.¹³ SPSS 21.0 for Windows (SPSS Inc., Chicago, IL, USA) was used for the statistical analysis. Values of P < 0.05 (two-tailed) were taken as statistically significant.

Results

General population

One thousand three hundred and nineteen consecutive patients underwent ICD (n = 810) or CRT-D (n = 509) implantation and were enrolled in the OBSERVO-ICD. During follow-up (median 39 months, 1st–3rd quartile 30–51 months), 195 (14.8%) patients experienced at least one unclustered episode of VT/VF and 62 (4.7%) patients experienced at least one ES.

Out of a total of 62 ES, 30 were associated with ischaemic heart disease, 24 with idiopathic dilated cardiomyopathy, 3 with hypertrophic cardiomyopathy, 3 with arrhythmogenic right ventricular dysplasia, and 2 with ion channel diseases. Moreover, among all patients experiencing ES during follow-up, 9 patients presented with NYHA I, 22 with NYHA II, 27 with NYHA III class, and 4 with NYHA IV class at the time of enrolment. According to the survival analysis, a class IV NYHA at the time of implant was associated with a significantly higher incidence of ES (19.8%) when compared with the other groups (7.3, 4.5, and 8.9% for NYHA classes I to III, respectively; log-rank P = 0.001).

General characteristics of the population and of the propensity score matched groups are shown in Table 1. CRT patients were on average 7 years older and more often males. They had a worse NYHA class and presented a lower LVEF at enrolment. Likewise, they had more concomitant diseases such as hypertension, diabetes, and dyslipidaemia; and they were treated more often with renin-angiotensin-aldosterone system inhibitors (Table 1). After propensity score matching for the likelihood of having a CRT-D implanted, 364 ICD and 364 CRT-D patients with similar baseline characteristics on 25 variables were selected.

Propensity score matched groups

During the follow-up, CRT-D was associated with a 45% relative risk reduction in ES when compared with patients with ICD (5.6% vs. 12.3%; log rank P = 0.014) with an annualized ES rate of 1.7% and 3.7%, respectively (Figure 1).

The first ES in the CRT-D and ICD groups occurred respectively 14 and 6 days after implantation. No significant differences in ES incidence were noted between CRT-D and ICD patients at 30-day (0.3% vs. 0.3%), 3-month (0.6% vs. 0.3%), and 1-year follow-up (2.6% vs. 1.4%; all P = ns).CRT-D and ICD showed similar incidence of unclustered VTs/VFs over time (23.2% vs. 22.7%; Log-rank P = ns; Figure 2).

Response to cardiac resynchronization therapy

Among all patients with a CRT-D device, 335 (65.8%) met the criteria for a positive clinical response and 370 (72.7%) met the criteria for a positive echographic response. Clinical responders experienced a 10- and 27-fold lower rates of ES when compared with clinical non-responders and negative responders, respectively (0.9% vs. 9.3% vs. 25.0%; Log-rank P = 0.017; Figure 3A). Similarly, echographic responders showed a 9- and 13-fold lower rates of ES when compared with echographic non-responders and negative responders (1.0% vs. 9.1% vs. 13.1%; log-rank P = 0.023; Figure 3B). No single episode of ES was detected in any of the 133 enrolled CRT-D patients who met the criteria for both clinical and echographic response during follow-up.

Clinical and echographic responses were also associated with lower estimate rates of unclustered VTs/VFs (log-rank P = 0.010 for clinical response and log-rank P = 0.005 for echographic response; Figure 4A and B). Eight (6%) out of 133 full-responders (clinical and echographic) experienced appropriate ICD therapies during follow-up.

Discussion

The main finding of the present study is that CRT-D is associated with a 45% relative risk reduction in ES incidence over a median of 39 months when compared with ICD in patients with left ventricular dysfunction. According to our data, CRT could have a protective effect on ES, which is even larger if the CRT implant is followed by a clinical and echographic response.

The potential antiarrhythmic or proarrhythmic effects of CRT are still a matter of debate. On one side, we have case series⁹ and small observational studies¹⁰ describing as ES can occur right after CRT implant up to the following few days. All of these reports include patients with structural heart disease and reduced LVEF, and most of them presented a previous history of monomorphic VT. In the described cases, pharmacological management was usually ineffective and ES termination was obtained by either turning off CRT or radiofrequency ablation, as the left ventricular lead was often located within an epicardial scar.¹⁰ The hypotheses raised to explain this serious complication are based on two main pathophysiological principles. First, the activation of the ventricular wall from the epicardium to the endocardium due to the epicardial pacing increases the transmural dispersion of repolarization similarly to what already seen in other arrhythmogenic substrates, such as long QT syndrome.¹⁴ Second, biventricular pacing alters the activation pattern of the LV, and the resulting wavefront may arrive at a different time on a scar-related circuit, potentially encountering an unidirectional block and thus enabling a re-entrant VT. According to that, biventricular pacing-induced ES can be considered as an acute complication.

On the other hand, no significant proarrhythmic effects have been associated to CRT in long-term randomized clinical trials,¹⁵ and previous evidence suggests that biventricular pacing, while not completely eliminating the need for an ICD, could reduce appropriate therapies.¹⁶

Regarding ES, the first evidence that CRT could have a beneficial effect came from Nordbeck et al.,¹³ who found an unadjusted ES rate of 7% and 0.6% in ICD and CRT-D patients, respectively over a 3.3 years follow-up. Unfortunately, the small sample size, the overall low event rates, and the many differences between the two populations prevented the authors from drawing any definitive conclusions.

Our findings strengthen the evidence that CRT could indeed have a protective effect on ES over a long-term follow-up, and this anti-arrhythmic effect is indeed more pronounced in CRT-responders. The main hypothesis behind this association relates to a potential deep connection between ES and HF. In fact, many recent studies support the hypothesis that ES could be reflective of end-stage systolic failure.^7,8 According to that, the occurrence of ES in patients with HF is not to be considered as an incidental bystander but rather as a warning sign of impending pump failure, deserving its own spot among the list of signs and symptoms of acute HF. The present data support this hypothesis in many ways.

First, as reported before,¹¹ ES was associated with a higher risk of HF-related mortality in the OBSERVO-ICD, while no arrhythmic deaths were recorded. Second, clinical and echographic responders had significantly lower rates of ES when compared with their respective non-responder counterparts. Third, patients who met both criteria for CRT response did not experience a single episode of ES during follow-up. Fourth, negative responders (i.e. patients with a deterioration of functional class or LVESV after CRT implant) seems to perform even worse than non-responders and ICD patients in term of ES incidence (Figure 3A and B). These findings, while strengthening the evidence linking ES and HF, also suggest that tackling the latter through CRT could deny the former from manifesting, at least in patients with left ventricular systolic dysfunction.

The relationship between unclustered VTs/VFs and LVEF has already been assessed in the MADIT-CRT study: a sub-analysis showed that each 5% increment in baseline LVEF was associated with a 30% reduction in the risk for fast VTs/VFs, with a corresponding reduced benefit from CRT-D in terms of arrhythmia reduction.¹⁷ Another sub-analysis from the same trial demonstrated that CRT-D patients achieving a LVEF >50% experienced very few ventricular tachyarrhythmias and received no shocks from the device.¹⁸ The same authors have also upped the ante saying that, in these patients, downgrading from CRT-D to CRT-P could be safe and cost-effective. Unfortunately, our data only partially support this statement. In fact, while clinical and echographic responses to CRT were associated with a much lower incidence of VTs/VFs, still a non-trivial part of CRT-responders, including eight full-responders, experienced appropriate ICD therapies during follow-up. Therefore, it can be hypothesized that, while CRT responders have a significant reduction in ES and VTs/VFs, patients who do not meet the definition of response could still have a reduction, albeit smaller, in total arrhythmic burden. The experienced reduction could not be enough in order to prevent VTs/VFs altogether but could be enough to reduce arrhythmic episodes and scatter them through time, thus avoiding the clusters commonly known as ES. This is also suggested by comparing Figures 3 and 4, where non-responders had a lower ES incidence but performed worse in term of unclustered VTs/VFs when compared with ICD patients, as if CRT was somewhat able to `downgrade' potential ES into isolated arrhythmic episodes in those patients. Of course, being an observational study, we are unable to draw any definitive conclusions. Moreover, the same definition of response, while surely useful from a clinical point of view, does not take into account many variables related to the device, the substrate, and the pharmacological treatment, which can potentially interfere with the arrhythmic burden.

Limitations

The OBSERVO-ICD is a retrospective, multicentre registry, and thus presents all the limitations related to observational studies. The propensity score-matching system, while now commonly regarded as one of the best methods to balance out potential differences in pre-treatment covariates, surely does not cancel out all the potential hidden biases due to the lack of randomization. For example, the propensity score-matching system had also excluded from the analysis many ICD patients with ion channel diseases, hypertrophic cardiomyopathy, arrhythmogenic right ventricular dysplasia, and implantation for secondary prevention, thus greatly reducing the weight of other arrhythmic triggers rather than HF in the ICD group. Moreover, some clinical features of our ICD and CRT-D samples could differ widely from other populations, and caution should be used in generalizing our results to other settings. Nonetheless, it is feasible to hypothesize that the present data could also be applicable in those sub-settings at higher risk of ventricular arrhythmias (such as old age or mild renal impairment) but in which CRT-related improvements are of similar magnitude to the ones reported in healthier patients.^19,20

Conclusions

Patients with CRT-D had a 45% lower incidence of ES when compared with propensity-matched ICD patients over a 3-year follow-up. The long-term benefit of CRT seems to be due to the improved haemodynamics, as CRT-responders experienced very low rates of ES. The present data strengthen the evidence that, in patients with HF, ES should be considered as a warning sign of impending pump failure rather than an incidental bystander.

Supplementary material

Supplementary material is available at Europace online.

Conflict of interest: none declared.

References