OUP user menu

Predictors of VT/VF-occurrence in ICD patients: results from the PROFIT-Study

Gunnar Klein, Christoph Lissel, Anne-Catherine Fuchs, Ajmal Gardiwal, Hanno Oswald, Marcos deSousa, A. Maximilian Pichlmaier, Ralf Lichtinghagen, Heinz Geerlings, Peter Lippolt, Michael Niehaus, Helmut Drexler, Thomas Korte
DOI: http://dx.doi.org/10.1093/europace/eul082 618-624 First published online: 24 July 2006

Abstract

Aims Identification of risk factors for ventricular tachycardia/ventricular fibrillation (VT/VF) occurrence in patients with implantable cardioverter-defibrillators (ICD) is reasonable, because ICD patients with multiple risk factors might benefit from more aggressive anti-arrhythmic therapy for the prevention of arrhythmic events. Furthermore, in the era of prophylactic ICD therapy and limited healthcare resources, additional markers are needed for improved patient selection.

Methods and results Thus, in Prospective Analysis of Risk Factor for Appropriate ICD Therapy (PROFIT), we prospectively analyzed the role of ejection fraction (EF), N-terminal probrain natriuretic peptide (NT-proBNP), New York Heart Association (NYHA) class, atrial fibrillation, and QRS-duration as independent predictors for VT/VF occurrence in 250 ICD patients. Kaplan–Meier analysis showed that EF<40% (log-rank P=0.001), NT-proBNP levels higher than median (≥405 ng/L; log-rank P=0.04), QRS-duration ≥150 ms (log-rank P=0.016), permanent atrial fibrillation (log-rank P=0.008), and higher NYHA class (log-rank P=0.029) were associated with VT/VF occurrence. By multivariate Cox regression analysis EF, QRS-duration and atrial fibrillation remained significantly associated with appropriate VT/VF therapy, whereas there was no relationship among NT-proBNP, NYHA class, and VT/VF occurrence. Stratifying patients according to the number of their independent risk factors (EF<40%, AF, QRS-width≥150 ms) showed that patients with greater than or equal to two risk factors had a 100% 2-year risk of VT/VF occurrence, whereas patients with no or one risk factor had a 19.3 and 25% 2-year risk, respectively.

Conclusions EF<40%, permanent atrial fibrillation, and QRS≥150 ms are independent predictors for VT/VF occurrence in predominantly secondary prophylactic ICD patients. Combining all independent predictors, we developed a risk score for VT/VF occurrence identifying a subgroup of patients with two or more risk factors who had a 100% 2-year risk. Future studies will reveal if this risk score helps to identify ICD patients suitable for empirical anti-arrhythmic therapy and to improve patient selection for prophylactic ICD therapy.

  • Implantable cardioverter-defibrillator
  • Risk factors for VT/VF occurrence
  • Ventricular tachyarrhythmias

Introduction

Recent trials have confirmed the superiority of the implantable cardioverter-defibrillator (ICD) compared with anti-arrhythmic therapy for primary or secondary prevention of sudden cardiac death and mortality.17 MUSTT7 and MADIT-I2 revealed that patients with reduced ejection fraction (EF) due to ischaemic cardiomyopathy, non-sustained ventricular tachycardias, and inducibility on electrophysiological (EP) study are suitable candidates for primary prophylactic ICD therapy.

The aim of the present study was prospectively to assess predictors of ventricular tachycardia/ventricular fibrillation (VT/VF) occurrence in ICD patients. Identification of risk factors for VT/VF occurrence in ICD patients might be of benefit: First, patients with multiple risk factors might benefit from more aggressive anti-arrhythmic therapy for the prevention of arrhythmic events. Secondly, as in AVID,1 32% of patients had not experienced spontaneous sustained VT or VF at 2 years, patient selection for ICD therapy still has to be improved. Recent primary prophylactic device studies (MADIT-II3 and SCD-HeFT5) emphasize this need, because ICD implant rates are expected to rise dramatically and healthcare resources are limited. Identification of VT/VF risk will facilitate allocation of this life-saving therapy to those patients who will benefit the most.

Thus, PROFIT (Prospective Analysis of Risk Factors for appropriate ICD Therapy) is a prospective single-centre study, investigating the role of EF, N-terminal probrain natriuretic peptide (NT-proBNP), New York Heart Association (NYHA) class, atrial fibrillation, and QRS-duration as independent predictors for VT/VF occurrence in 250 ICD patients.

Materials and methods

Inclusion criteria

The PROFIT study is a single-centre prospective cohort study of consecutive ICD patients designed to identify pre-specified risk factors, i.e. EF, NT-proBNP, NYHA class, atrial fibrillation, and QRS-duration as predictors for VT/VF events. For inclusion in the study, patients had to meet the 2002 American College of Cardiology/American Heart Association guidelines for ICD implantation (Class I–IIb).8 Enrolment began in March 2003 and ended in March 2004. At entry into the study, clinical characteristics were collected through medical record review and interrogation of the patients. Congestive heart failure (CHF) severity was categorized according to NYHA class at baseline interview. EF was determined either by echocardiography or LV-angiography. For analysis of NT-proBNP blood was collected at baseline interview. NT-pro-BNP was measured using an electrochemiluminescence immunoassay ‘ECLIA’ (Roche Diagnostics, Mannheim, Germany)9.

Follow-up

Patients were monitored for ICD therapies (only sustained VT/VF episodes either treated by anti-tachycardia pacing or shock were included) and clinical endpoints (death, hospitalization due to heart failure, or myocardial infarction) during routine follow-up visits every 6 months. Hospitalization due to heart failure was defined as NYHA class IV (pulmonary oedema) or ankle oedema, making intravenous diuretic treatment necessary. Myocardial infarction (STEMI and NSTEMI) were defined as previously described.10,11 Appropriate device therapy for sustained VT/VF was confirmed by two electrophysiologists (GK/TK) by analysis of stored electrograms as previously described.12,13 The electrophysiologists were unaware of the patients' EF, NT-proBNP levels, NYHA-class, and QRS-duration.

Statistical analysis

Sample size was calculated by nQuery Advisor Release 6.0 based on previous results showing a 2.4-fold increased risk for appropriate shocks in patients with heart failure NYHA≥III when compared with NYHA≤II.14 Assuming an equal distribution of patients NYHA≥III and NYHA≤II, we estimated a sample size of 125 patients per group to reach a statistical power >90%. Fisher's exact test was used to compare categorical variables, and Student's t-test was used to compare continuous variables. Cumulative event proportions were calculated by the Kaplan–Meier method, and outcome differences of patients were assessed by log-rank test. Cox regression analysis was performed to evaluate associations between risk factors at enrolment and appropriate ICD therapy. As prospectively defined, patients were stratified for EF<or ≥40% (median), NT-proBNP<or ≥405 ng/L (median), no CHF/NYHA I or NYHA II/III, and QRS-duration<or≥150 ms. QRS-duration was only measured in spontaneous beats in sinus rhythm or atrial fibrillation. P<0.05 was considered statistically significant. Values are means±SD. All statistical analysis was performed using SPSS for Windows 12.0 (SPSS Inc., Chicago, USA).

Results

Baseline characteristics

PROFIT study enrolled 250 ICD patients. At inclusion, patients were predominantly male, on average 63±13 years old (Table 1). Underlying heart disease was coronary artery disease in 62%, non-ischaemic dilated cardiomyopathy in 21%, arrhythmogenic right ventricular cardiomyopathy (ARVC), or suspected ARVC in 7% and valvular heart disease, hypertensive cardiomyopathy, long-QT-syndrome, and hypertrophic cardiomyopathy in the remaining patients (10%). Patients had a mean EF of 40±16% and a mean NYHA class of 2.1±0.8. A total of 17% had permanent atrial fibrillation, baseline QRS-duration was 144±41 ms. Indication for ICD therapy was primary prophylactic in 7.6% (non-sustained VT, EF ≤35% due to ischaemic cardiomyopathy, and inducibility in EP study) and secondary prophylactic in 92.4% (successful resuscitation of VT/VF or recurrent sustained VT with or without syncope). Single-chamber devices were seen in 69%; 27% had dual-chamber devices; and 4% had biventricular ICDs for primary prophylactic reasons. Lowest VT-detection zone was 174±18 bpm. In the VT-zone, three bursts and three ramps were programmed for anti-tachycardia pacing, whereas in VF-zone shocks of 30–40 J were programmed. Patients had a mean NT-proBNP serum level of 1663 ng/L at inclusion. Patients had a mean follow-up of 549±118 days (18.3 months).

View this table:
Table 1

Baseline characteristics

Baseline characteristicsAll patients (n = 250) n(%)/mean ± SD
Male192 (76.8)
Age63 ± 13
Heart disease
 CAD155 (62.0)
 DCM52 (21.0)
 Hypertensive CMP6 (2.4)
 ARVC or suspected ARVC18 (7.2)
 Valvular heart disease8 (3.2)
 Long-QT-syndrome4 (1.6)
 Status post-inflammatory CMP3 (1.2)
 Status post-heart transplantation2 (0.8)
 HOCM1 (0.4)
 Idiopathic VF1 (0.4)
Ejection fraction (%)40±16
NYHA2.1±0.8
No CHF11 (4.4)
NYHA I28 (11.2)
NYHA II147 (58.8)
NYHA III63 (25.2)
NYHA IV1 (0.4)
Permanent AF42 (16.8)
QRS-width (ms)144 ± 41
Medication
 ACE-inhibitor179 (72.0)
 Beta-blocker175 (70.0)
 Diuretics106 (42.4)
 Spironolactone58 (23.2)
 Sotalol52 (20.8)
 Amiodarone50 (20.0)
ICD indication
 Primary prophylaxis19 (7.6)
 Secondary prophylaxis231 (92.4)
Lowest VT-detection zone (/min.)174 ± 18
Follow-up (days)549 ± 118
Serum-NT-proBNP (ng/L)1663 ± 5682
  • CAD, coronary artery disease; DCM, dilated cardiomyopathy; CMP, cardiomyopathy; HOCM, hypertrophic obstructive cardiomyopathy; ACE, angiotensin-conversting enzyme.

NT-proBNP and EF are independent predictors for death, myocardial infarction, and hospitalization for heart failure in ICD patients

Kaplan–Meier analysis showed that patients with EF<40% or NT-proBNP levels higher than the median (≥405 ng/L) had a significantly higher probability of experiencing any of the clinical endpoints (Figure 1A and B). By univariate analysis, NT-proBNP and EF, but not atrial fibrillation or QRS-duration ≥150 ms were significant predictors for the combined clinical endpoint (Table 1). Multivariate Cox regression analysis including EF, atrial fibrillation, QRS-duration, and NT-proBNP revealed that NT-proBNP was an independent predictive factor for the occurrence of any clinical event (Table 2).

Figure 1

Kaplan–Meier curves of freedom from any clinical event (death, myocardial infarction, and hospitalization due to heart failure) for patients (A) with EF<median (closed circles) and ≥median (closed squares) (EF=40%) and (B) with NT-proBNP lower (closed squares) and patients with NT-proBNP higher than median (closed circles).

View this table:
Table 2

Cox regression analysis for death, myocardial infarction, and hospitalization due to heart failure

ParameterUnivariate analysisMultivariate analysis
[Exp(B)]95% CIP[Exp(B)]95% CIP
NT-proBNP0.2520.223–0.7680.0010.3440.149–0.7950.013
EF0.3160.198–0.6860.0020.4460.199–1.0000.050
Atrial fibrillation0.7730.337–1.7440.5441.0070.433–2.3420.987
QRS1.1800.245–0.8770.6351.4920.741–3.0040.262

EF, permanent atrial fibrillation, and QRS-duration are independent predictors for occurrence of VT/VF

Among the 250 patients enrolled in PROFIT, 427 sustained VT/VF episodes treated by the ICD were recorded. A total of 46 patients suffered from sustained VT/VF. Mean time to first episode was 529±128 days and mean cycle length of VT/VF was 289±55 ms.

Kaplan–Meier analysis showed that NT-proBNP levels higher than median (≥405 ng/L) were associated with a significantly higher probability of VT/VF occurrence (Figure 2A, log-rank P=0.004). Furthermore, Kaplan–Meier analysis showed that EF<40% (log-rank P=0.001), QRS-duration ≥150 ms (log-rank P=0.016), permanent atrial fibrillation (log-rank P=0.008), and higher NYHA class (log-rank P=0.029) were associated with VT/VF occurrence (Figure 2BE). Of those patients with wide QRS (101 patients), 81% had left bundle branch block pattern, 8% had right bundle branch block pattern, and another 8% had intraventricular conduction delay. Kaplan–Meier analysis showed a higher incidence of VT/VF occurrence in patients with left than with right bundle branch block pattern (log-rank P=0.09, non-significant). Even after multivariate analysis adjusted for underlying heart disease, baseline medication (beta-blocker, sotalol, amiodarone), ICD indication, and CRT-treatment, QRS-duration ≥150 ms remained an independent predictor of VT/VF occurrence [P=0.042; Exp(B) 0.499; SE 0.341].

Figure 2

Kaplan–Meier curves of freedom from sustained VT/VF stratified according to (A) NT-proBNP≤median or >median, (B) EF<or ≥median, (C) QRS-duration <150 ms (n=149) or ≥150 ms (n=101), (D) sinus rhythm (n=208) and permanent atrial fibrillation (n=42), and (E) NYHA class (n=39 with no CHF or NYHA I and n=239 with NYHA II or III).

By multivariate Cox regression analysis, EF, QRS-duration, and atrial fibrillation remained significantly associated with appropriate VT/VF therapy, whereas there was no relationship among NT-proBNP, NYHA-class, and VT/VF occurrence (Table 3).

View this table:
Table 3

Cox regression analysis for VT/VF occurrence

ParameterUnivariate AnalysisMultivariate Analysis
[Exp(B)]95% CIP[Exp(B)]95% CIP
NT-proBNP0.4140.223–0.7680.0051.0001.000–1.0000.490
EF2.7180.198–0.6860.0020.9710.947–0.9950.019
QRS2.1570.245–0.8770.0180.4580.237–0.8850.020
Atrial fibrillation2.3350.223–0.8220.0110.4980.254–0.9760.042
NYHA0.1290.018–0.9430.0440.5040.151–1.6820.265

Risk score for prediction of VT/VF occurrence in ICD patients

According to the number of independent risk factors for VT/VF occurrence (EF<40%, QRS-duration ≥150 ms, and atrial fibrillation), we stratified patients in two groups: patients with no or one risk factor and patients with two or more risk factors. Patients with greater than or equal to two risk factors had a 100% 2-year risk of VT/VF occurrence, whereas patients with no or one risk factor had a 19.3% and 25% 2-year risk of experiencing VT/VF, respectively (Figure 3).

Figure 3

Kaplan–Meier curves of freedom from sustained VT/VF for patients (A) with no or one (closed circles, n=215) or greater than or equal to (closed triangles, n=35) independent risk factors (EF <40%, permanent atrial fibrillation, QRS-duration ≥150 ms). Thirteen patients had three risk factors, only half of them were treated with class III anti-arrhythmic medication, six received sotalol, and one received amiodarone.

Discussion

Major findings

The present study shows that EF, permanent atrial fibrillation, and QRS-duration are independent predictors for VT/VF therapy in predominantly secondary prophylactic ICD patients. Furthermore, NT-proBNP and EF were both independent predictors for the combined clinical endpoint of death, myocardial infarction, and hospitalization for heart failure. By stratifying patients according to their independent risk factors for appropriate VT/VF therapy, we developed a risk score identifying patients with a 100% 2-year risk for VT/VF occurrence.

NT-proBNP for risk stratification

BNP is mainly synthesized and released from ventricular myocytes as a 76 aminoacid N-terminal fragment (N-terminal BNP) and a 32-aminoacid active hormone (BNP) in response to myocyte stretch and transmural pressure load.15 Our data show that NT-proBNP is an independent predictor for clinical events such as death, myocardial infarction, or hospitalization due to heart failure in ICD patients. This is in accordance with various previous studies in heart failure and acute coronary syndrome.1618

Regarding ventricular arrhythmia burden, our study shows that patients with elevated NT-proBNP are at higher risk for VT/VF. This appears to be reasonable, as BNP is released as a response of the heart to intraventricular pressure or stretch, which in turn causes EP abnormalities like slowing of intraventricular conduction, triggering after depolarizations, and ventricular ectopic beats.14,15,1921 Moreover, recent studies in mild-to-moderate and severe heart failure identified BNP as an independent risk factor for sudden death.22,23 However, neither study specified arrhythmic death as an outcome. Non-arrhythmic causes of sudden death such as pulmonary emboli, ruptured aneurysms, myocardial infarctions, hyperkalaemia, and electromechanical dissociation could have been underestimated in these studies. In our study, NT-proBNP was not an independent predictor for VT/VF occurrence, possibly because EF was the most reliable predictor for appropriate ICD therapy. The higher incidence of VT/VF episodes in the group with higher NT-proBNP-levels might, therefore, simply reflect the higher risk of ventricular arrhythmias in patients with LV-dysfunction.

EF, atrial fibrillation, and QRS-duration for risk stratification

In accordance with our data, reduced EF has been shown to be a strong risk factor for VT/VF occurrence in various published studies including ICD patients.14,2427 In our study, patients with EF<40% had a 2.1-fold increase in 2-year risk for experiencing appropriate ICD therapy. The TOVA study has recently shown that patients with EF<20% and NYHA class III have a 3.6-fold 1-year risk for appropriate ICD shock when compared with patients with EF≥20% and NYHA class I and II.14 However, in TOVA only ICD shocks but not ATP-therapies were included and neither NT-proBNP nor other risk factors, i.e. atrial fibrillation and QRS-duration were studied as predictors for VT/VF occurrence. In our study, multivariate analysis also revealed QRS-duration as an independent predictor for arrhythmia burden. Notably, Kaplan–Meier analysis showed a higher incidence of VT/VF occurrence in patients with left than with right bundle branch block pattern (log-rank P=0.09), although this did not reach significance because of the low number of patients with right bundle branch block pattern. This is in accordance with data showing that heart failure patients with prolonged QRS-duration have higher mortality, show a significant benefit from ICD therapy, and have more recurrences of fast VTs.4,28,29 Our study does not permit the drawing of any conclusions on the role of CRT–ICDs on the incidence of VT/VF occurrence in patients with wide QRS-complexes, but our data suggest that a wide QRS-complex is an independent risk factor for VT/VF occurrence. Controlled, randomized, prospective, and adequately powered studies to assess any beneficial effect of CRT on episode occurrence in ICD patients are still lacking. However, a retrospective ad hoc analysis of the MIRACLE ICD population did not find a beneficial effect of CRT on VT/VF occurrence, neither in patients with primary nor secondary prophylactic ICD indication.30

Another independent predictor of VT/VF by multivariate analysis in our study was permanent atrial fibrillation. Patients with atrial fibrillation had a 1.8-fold increase in 2-year risk of VT/VF. Many hypotheses have been put forward to explain this pathophysiological phenomenon such as tachycardia-induced changes in ventricular refractoriness or indirectly haemodynamic alteration, ischaemia, or neuroendocrine activation because of atrial fibrillation.31,32 In accordance with our data, Gronefeld et al.32 have recently shown that patients with persistent atrial fibrillation at ICD implant had a higher incidence of appropriate ICD therapy than patients in sinus rhythm. Furthermore, the ‘Jewel-AF-study’ showed that atrial tachyarrhythmias increase the risk of ventricular tachyarrhythmias in ICD patients.31

Risk score for VT/VF occurrence and clinical impact

Combining all three independent risk factors (EF<40%, atrial fibrillation, and QRS-duration ≥150 ms) and stratifying patients according to the number of their individual risk factors showed that patients with two or more risk factors had a 100% 2-year risk of suffering from VT/VF. In the era of primary prophylactic ICD implantation more patients receive ICD therapy. However, after SCD-HeFT, a further risk stratification of heart failure patients beyond EF will be necessary especially on economic grounds. A risk score as described in our study identifies patients at high risk for VT/VF occurrence, who might profit most from ICD therapy. Since in our study most of the patients had a secondary prophylactic ICD indication, further prospective and randomized studies are needed to confirm the value of this easily obtainable risk factor score for patients with primary prophylactic ICD indications.

Apart from patient selection in prophylactic ICD therapy a risk score for appropriate ICD interventions might also be helpful in allocating patients to an empirical anti-arrhythmic medication to prevent ICD shocks and thus improve quality of life. It has recently been shown, that sotalol, azimilide, and amiodarone are effective in reducing appropriate and inappropriate ICD interventions.3336 Patients with a higher risk score could be good candidates for such therapy.

A potential limitation of our study is the lack of standardized device programming. However, rate-detection settings in patients with and without ICD therapy were identical, arguing against programming differences accounting for the results in the present study.

Another limitation of our study refers to the patient population, which is a predominantly secondary prophylactic ICD population. Thus, our results do not reflect the situation in primary prophylactic ICD patients. Further prospective studies are needed to show if the risk score holds true also for primary prophylactic ICD patients.

Conclusions

EF <40%, permanent atrial fibrillation, and QRS ≥150 ms are independent predictors for VT/VF occurrence in predominantly secondary prophylactic ICD patients. Combining all independent predictors, we developed a risk score for VT/VF occurrence identifying a subgroup of patients with two or more risk factors who had a 100% 2-year risk. Future studies will reveal if this risk score helps to identify ICD patients suitable for empirical anti-arrhythmic therapy and improve patient selection in prophylactic ICD therapy.

References

View Abstract