CLINICAL RESEARCH

Europace (2014) 16, 1069–1077 doi:10.1093/europace/eut303

Electrocardiology and risk stratification

Heart rate turbulence predicts ICD-resistant mortality in ischaemic heart disease Thomas Marynissen 1,2, Vincent Flore´ 1,2, Hein Heidbuchel 1,2, Dieter Nuyens 1,2, Joris Ector 1,2, and Rik Willems 1,2* 1

Department of Cardiology, University Hospitals Leuven, Herestraat 49, 3000 Leuven, Belgium; and 2Department of Cardiovascular Sciences, University of Leuven, Leuven, Belgium

Received 5 June 2013; accepted after revision 22 October 2013; online publish-ahead-of-print 6 November 2013

Aims

----------------------------------------------------------------------------------------------------------------------------------------------------------Keywords

ICD † Non-invasive risk stratification † Sudden cardiac death † Ischaemic heart disease † Heart rate turbulence

Introduction Sudden cardiac death (SCD) caused by ventricular arrhythmia is an important cause of death in patients with ischaemic heart disease (IHD).1 The implantation of an implantable cardioverter-defibrillator (ICD) effectively decreases mortality due to SCD.2 – 4 To keep the cost-effectiveness and complication burden of ICD therapy acceptable, it is crucial to identify patients at high risk of arrhythmic death but at low risk of dying from other causes, e.g. heart failure or noncardiac death. Currently, the selection of ICD candidates is based mainly on the assessment of heart failure symptoms and left ventricular ejection fraction (LVEF).2 – 6 However, using current guidelines, only 20% of patients implanted with a prophylactic ICD receive appropriate therapy during long-term follow-up.2,5,7 Moreover, less than one-third of SCD victims have an impaired LVEF.8 The

Belgian Health Care Knowledge Center calculated that the poor sensitivity and specificity of LVEF to predict arrhythmia results in questionable cost-effectiveness of ICD therapy in our healthcare system.9 Given the complex and multifactorial pathophysiology of SCD, multivariate risk models combining LVEF with other parameters will likely be required to optimize risk stratification.8 Several electrocardiogram (ECG)-based techniques have been investigated, including QRS duration, QT length/dispersion, signal-averaged ECG, T-wave alternans, presence of ventricular ectopy on Holter, heart rate variability (HRV), and heart rate turbulence (HRT).10 Heart rate variability is the temporal beat-to-beat variation in successive RR intervals.11 Heart rate turbulence describes the physiological short-term oscillation of beat-to-beat intervals after spontaneous ventricular premature beats (VPBs).12 Both measure autonomic modulation of heart rate and can be

* Corresponding author. Tel: +3216/344235; fax: +3216 344240, E-mail: [email protected] Published on behalf of the European Society of Cardiology. All rights reserved. & The Author 2013. For permissions please email: [email protected].

Downloaded from by guest on November 13, 2015

In high-risk patients, implantable cardioverter-defibrillators (ICDs) can convert the mode of death from arrhythmic to pump failure death. Therefore, we introduced the concept of ‘ICD-resistant mortality’ (IRM), defined as death (a) without previous appropriate ICD intervention (AI), (b) within 1 month after the first AI, or (c) within 1 year after the initial ICD implantation. Implantable cardioverter-defibrillator implantation in patients with a high risk of IRM should be avoided. ..................................................................................................................................................................................... Methods Implantable cardioverter-defibrillator patients with ischaemic heart disease were included if a digitized 24 h Holter was and results available pre-implantation. Demographic, electrocardiographic, echocardiographic, and 24 h Holter risk factors were collected at device implantation. The primary endpoint was IRM. Cox regression analyses were used to test the association between predictors and outcome. We included 130 patients, with a mean left ventricular ejection fraction (LVEF) of 33.6 + 10.3%. During a follow-up of 52 + 31 months, 33 patients died. There were 21 cases of IRM. Heart rate turbulence (HRT) was the only Holter parameter associated with IRM and total mortality. A higher New York Heart Association (NYHA) class and a lower body mass index were the strongest predictors of IRM. Left ventricular ejection fraction predicted IRM on univariate analysis, and was the strongest predictor of total mortality. The only parameter that predicted AI was non-sustained ventricular tachycardia. ..................................................................................................................................................................................... Conclusion Implantable cardioverter-defibrillator implantation based on NYHA class and LVEF leads to selection of patients with a higher risk of IRM and death. Heart rate turbulence may have added value for the identification of poor candidates for ICD therapy. Available Holter parameters seem limited in their ability to predict AI.

1070

Methods Patient population Patients were selected from the ICD registry of the Cardiology Department of the University Hospitals Leuven, Belgium. This registry contains data on all 1961 patients who underwent ICD implantation from 1989 to 2012. We selected patients with IHD (presence of a stenosis of at least 50% in at least one coronary artery15), who received an ICD for primary or secondary prevention of SCD. Patients were included if a digitized 24 h Holter ECG was available at the time of ICD implantation. Devices were implanted and programmed according to the discretion of the treating cardiologist. The study complies with the Declaration of Helsinki, and the research protocol was approved by the locally appointed ethics committee.

1160

Turbulence slope

1000

840 –32

Turbulence onset –24

–16

–8 0 8 Interval number

16

24

32

–24

–16

–8 0 8 Interval number

16

24

32

B 880

720

560 –32

Figure 1 Heart rate turbulence describes the physiological short-term oscillation of beat-to-beat intervals after spontaneous VPBs. The physiological pattern of HRT consists of brief heart rate acceleration (RR shortening) [quantified by turbulence onset (TO)] followed by more gradual heart rate deceleration (RR prolongation) [quantified by turbulence slope (TS)] before the heart rate returns to baseline.18 The local averaged VPB tachogram is constructed by aligning and averaging RR interval sequences surrounding isolated VPBs and is shown for (A) a 70-year-old male who received an ICD for primary prevention of SCD. An appropriate intervention occurred 71 days after implantation. Both TO and TS are within the normal range. (B) an 80-year-old male who received an ICD for secondary prevention of SCD and died due to pump failure 1.5 years after ICD implantation without ever receiving an appropriate intervention. TO and TS are both abnormal.

Retrospective data collection All tested variables were defined and categorized according to literature or common practice. At device implantation, we collected demographic data [age, gender, body mass index (BMI)], glomerular filtration rate (calculated using the Cockroft– Gault equation), and LVEF (based on echocardiography or radionuclide ventriculography). Heart failure symptoms were classified as mild or severe, according to New York Heart Association (NYHA) functional class (I– II vs. III – IV). A patient was considered to suffer from diabetes mellitus if he was receiving any anti-diabetic medication. The last ECG prior to ICD implantation was used for analysis of QRS duration and heart rate corrected QT time (QTc), using the Bazett formula. Twenty-four hour ambulatory two-channel ECG recordings were analysed by means of Synescope Holter software (ELA medical, Sorin group). The presence of VPB (≥10 per hour10) and non-sustained ventricular tachycardia (nsVT) (≥3 consecutive beats at .120 b.p.m., terminating spontaneously in ,30 s4) were assessed. Time- and frequency-domain parameters of HRV were automatically calculated, including standard deviation of the NN intervals (SDNN), square root

Downloaded from by guest on November 13, 2015

easily calculated from 24 h Holter ECG. An increasing amount of evidence supports the association of these parameters with adverse outcome.10 Unfortunately, most of the contemporary risk stratification techniques predict total mortality and do not discriminate arrhythmic from non-arrhythmic death. Ideally, a parameter would only select patients at high arrhythmic risk and good candidates for ICD therapy, and not patients likely to die from other causes, thus without using their device. The balance between arrhythmic and non-arrhythmic death prediction complicates interpretation of literature and implementation of many available risk parameters into clinical practice.10 An illustration of this conundrum is the so-called ‘conversion of mode of death’ from arrhythmic SCD to heart failure death by ICDs, as was described in high-risk subgroups early after myocardial infarction.13,14 To identify patients at high risk of dying without benefiting from an ICD, we introduced the concept of ICD-resistant mortality (IRM), defined as death, regardless the cause, (a) without previous appropriate ICD intervention (AI), (b) within 1 month after the first AI, or (c) within 1 year after the initial ICD implantation. The 1-month period was chosen referring to the classical 30-day perioperative mortality from surgical literature. The 1-year period after implantation is integrated in the current guidelines for ICD implantation, demanding an expected life expectancy of .1 year after implantation.6 While other authors have only used criterion (a),15 we believe criteria (b) and (c) provide a more correct reflection of everyday practice by also taking into account patients that die shortly after AI or ICD implantation. In this retrospective cohort study, we investigated the predictive value of demographic, electrocardiographic, echocardiographic, and 24 h Holter-based risk factors for IRM, because ICD implantation in patients at high risk of IRM should be avoided.

A RR interval (ms)

† Implantable cardioverter-defibrillator (ICD) implantation based on New York Heart Association class and left ventricular ejection fraction leads to selection of patients with a higher risk of ICD-resistant mortality and death. † Heart rate turbulence may have added value for the identification of poor candidates for ICD therapy. † Available Holter parameters seem limited in their ability to predict appropriate ICD interventions.

RR interval (ms)

What’s new?

T. Marynissen et al.

1071

HRT predicts ICD-resistant mortality in IHD

of the mean squared differences of successive NN intervals (RMSSD), low- and high-frequency power. Heart rate turbulence and deceleration capacity (an integral measure of all deceleration-related oscillation observed over 24 h) were calculated using previously published methods.12,16 Turbulence onset (TO) and turbulence slope (TS) were dichotomized at predefined cut points [TO , 0 and ≥0% and TS . 2.5 and ≤2.5 ms per normal-to-normal interval (RRI)].12 Heart rate turbulence was categorized as normal (TO , 0% and TS . 2.5 ms/RRI) or abnormal (Figure 1). Patients for whom HRT parameters could not be calculated due to insufficient VPB were classified as having normal HRT.12 In 10 patients (7.7%), HRT and HRV parameters could not be calculated due to atrial fibrillation (AF). Finally, QT analysis software (ELA medical, Sorin group) was used to calculate 24 h QT apex (QTa), QT end (QTe), and the QTa/RR and QTe/RR slope, according to previously published methods.17 Data on survival and ICD interventions were gathered, based on patient records and ICD interrogation printouts. The database was locked on 1 September 2012.

1961 patients were screened (ICD database from University Hospitals Leuven)

130 patients were included in the study (IHD and available 24h Holter ECG pre-implantation) Follow-up 52 + 31 months IRM: n = 21 (16.2%) Death (any cause): n = 33 (25.3%) Pump failure death: n = 18 (13.8%)* Heart transplantation: n = 15 (11.5%)* Al: n = 56 (43.8%) IAI: n = 27 (20.8%)

Endpoints * Heart failure death/heart transplantation: n = 33 (25.4%)

Figure 2 Patient selection and outcome.

Table 1 Patient characteristics and their association with IRM Total population (n 5 130)

IRM (n 5 21)

Non-IRM (n 5 94)

HR (95% CI) (univariate)

P

HR (95% CI) (multivariate)

P

–

–

............................................................................................................................................................................... Male gender

115 (88.5%)

17 (81%)

85 (90.4%)

0.59 (0.2– 1.77)

0.347

Age (years) BMI (kg/m2)

64 + 9.6 25.6 + 3.8

70.7 + 7.7 23.4 + 2.8

64.2 + 8.8 26.2 + 3.7

2.44 (1.04–5.78)a 3.64 (1.40–9.48)b

0.041 1.92 (0.66– 5.59)a 0.231 0.008 3.89 (1.43– 10.56)b 0.008

47.3 + 16.7

69.6 + 23.2

3.92 (1.43–10.75)c 0.008 1.59 (0.47– 5.39)c

0.454

31.4 + 9.5

35.1 + 10.6

4.15 (1.34– 12.82)d 0.013 2.39 (0.72– 7.92)d 3.75 (1.55–9.08) 0.003 4.17 (1.62– 10.75)

0.155 0.003

GFR (mL/min) (Cockroft –Gault) 66.5 + 25.1 LVEF (%) NYHA class III/IV vs.

33.6 + 10.3 47 (36.2%)

12 (57.1%)

25 (26.6%)

I/II Medication

83 (63.8%)

9 (42.9%)

69 (73.4%)

Diuretics

82 (63.1%)

14 (66.7%)

54 (57.4%)

1.49 (0.6– 3.71)

0.388

–

–

Beta-blockers ACE-inhibitors/ARB

122 (93.8%) 119 (91.5%)

21 (100%) 20 (95.2%)

86 (91.5%) 87 (92.6%)

23.8 (0.03–1) 1.65 (0.22–12.35)

0.348 0.626

– –

– –

Statins

110 (84.6%)

15 (71.4%)

82 (87.2%)

0.47 (0.18–1.23)

0.123

–

–

26 (20%) 56 (43.1%)

2 (9.5%) 13 (61.9%)

24 (25.5%) 38 (40.4%)

0.36 (0.03–1.55) 1.76 (0.72–4.26)

0.170 0.211

– –

– –

0.90 (0.37–2.16)

0.807

–

–

88 (67.7%) 42 (32.3%)

13 (61.9%) 8 (38.1%)

62 (66%) 32 (34%)

QRS duration (ms)

120 (102– 148)

128 (101– 178) 118 (100.5– 147.5) 1.01 (0.99–1.02)

0.088

–

–

QTc (ms) (Bazett)

447 (423– 484)

442 (411– 494) 447.5 (424.5– 484) 1.00 (0.99–1.01)

0.612

–

–

Diabetes Atrial fibrillation ICD indication Primary vs. Secondary prevention

IRM, ICD-resistant mortality; HR, hazard ratio; CI, confidence interval; BMI, body mass index; GFR, glomerular filtration rate; LVEF, left ventricular ejection fraction; NYHA, New York Heart Association. Bold values are considered statistically significant (P , 0.05). a age .70 vs. ≤70 years. b BMI , 25 vs. ≥25 kg/m2. c GFR , 60 vs. ≥60 mL/min. d LVEF ≤ 35 vs. .35%.

Downloaded from by guest on November 13, 2015

The primary endpoint was IRM, defined as death, regardless the cause, (a) without previous AI, (b) within 1 month after the first AI, or (c) within 1 year after the initial ICD implantation. We also evaluated total mortality, as well as heart failure progression, a combined endpoint consisting of heart failure death and heart transplantation. Appropriate ICD intervention was defined as anti-tachycardia pacing or shock delivered for an

1072

T. Marynissen et al.

on the availability of a digitized 24 h Holter ECG pre-implantation (Figure 2). The Holter ECG was recorded 102 + 167 days before ICD implantation. Baseline characteristics are summarized in Table 1. Most patients were male (88.5%), with a mean age of 64 + 9.6 years. The average LVEF was 33.6 + 10.3%. Nine three percent of patients had prior myocardial infarction. In 67.7% of patients, an ICD was implanted in primary prevention of SCD. Twenty percent received a dual-chamber ICD, 30.8% a cardiac resynchronization therapy defibrillator.

Percent event-free

A 100 80 60 40 Mortality IRM

20 0 0

24

48

96

120

72

96

120

Time

Percent event-free

B 100 80 60 40 AI IAI

20 0 0

24

48 Time

Figure 3 Kaplan– Meier curves for (A) IRM and total mortality, (B) appropriate and inappropriate ICD intervention (AI/IAI).

episode of ventricular tachycardia or ventricular fibrillation. Other ICD interventions were classified as inappropriate (IAI).

Statistical analysis Continuous variables are expressed as mean with standard deviation or median with interquartile range when appropriate. Dichotomous variables are presented as numbers and percentages. For analysis of the association between predictors and outcome, uni- and multivariate Cox regression analyses were performed. Univariate Cox regression was used to identify factors with potential prognostic significance. Using predictors with P , 0.05 in univariate Cox regression, a multivariate model was created. Finally, the additional value of Holter parameters to this model was tested. The effects of the factors investigated are given as hazard ratios (HRs) with 95% confidence intervals (CIs). A difference in event-free rates was shown using Kaplan – Meier survival curves and the log-rank test. All comparisons were done using SPSS for windows, version 20 (IBM). A value for P , 0.05 was considered statistically significant. All P values were two-sided.

Results Patient characteristics Eight hundred and ninety-four patients with IHD underwent ICD implantation at the University Hospitals Leuven from 1995 to 2012. One hundred and thirty patients were included in this study based

Outcome During a mean follow-up of 52 + 31 months, 33 patients died (25.3%) and 15 patients underwent heart transplantation (11.5%). The cause of death was terminal pump failure in 18 patients (54.5%) and SCD (as defined elsewhere10) in 4 patients (12.1%). The remaining 11 patients died due to non-cardiac causes [infection (n ¼ 5), malignancy (n ¼ 2), haemorrhagic stroke (n ¼ 1), lower limb ischaemia (n ¼ 1), intestinal ischaemia (n ¼ 1), and acute pancreatitis (n ¼ 1)]. There were 21 cases of IRM (16.2%) [criterium (a): n ¼ 15; criterium (b): n ¼ 1; criterium (c): n ¼ 5]. Fifty-six patients received at least one AI (43%), whereas IAI were documented in 27 patients (20.8%). Kaplan –Meier curves for IRM, total mortality, AI, and IAI are shown in Figure 3.

Implantable cardioverter-defibrillator-resistant mortality The results of Cox uni- and multivariate regression analyses are presented in Tables 1 and 2. Implantable cardioverter-defibrillatorresistant mortality risk was significantly elevated in patients with lower BMI (HR: 3.64; CI: 1.4 –7.48), lower LVEF (HR: 4.15; CI: 1.34–12.82), worse renal function (HR: 3.92; CI: 1.43–10.75), higher NYHA functional class (HR: 3.75; CI: 1.55 –9.08), and advanced age (HR: 2.44; CI: 1.04 –5.78). In multivariate analysis, only BMI (HR: 3.89; CI: 1.43–10.56) and NYHA class (HR: 4.17; CI: 1.62–10.75) remained significant predictors of adverse outcome. Among Holter parameters, only HRT was associated with IRM, with a HR of 3.17 (CI: 1.14–8.85). This association remained significant when HRT was entered in a Cox model consisting of BMI and NYHA class (HR: 3.38; CI: 1.17–9.71). Kaplan–Meier curves for association between IRM and BMI, NYHA class, and HRT are presented in Figures 4A–C.

Other endpoints This analysis was repeated for total mortality, heart failure death/ heart transplantation, AI, and IAI. Results are presented in Tables 3–5 and Figure 5A–C.

Discussion In 130 ICD patients with IHD, there were 21 cases of IRM on a total of 33 deaths (63.6%), during an average follow-up of over 4 years. Heart rate turbulence was the only Holter parameter associated with IRM and total mortality. Body mass index and NYHA functional class were the strongest demographic predictors of IRM. Left ventricular

Downloaded from by guest on November 13, 2015

72

1073

HRT predicts ICD-resistant mortality in IHD

Table 2 Holter parameters and their association with IRM Total population IRM (n 5 130) (n 5 21)

Non-IRM (n 5 94)

HR (95% CI) (univariate)

P HR (95% CI) P value (multivariatea) value

............................................................................................................................................................................... Presence of ≥10 VPB per hour 72 (55.4%)

11 (52.4%)

58 (61.7%)

0.71 (0.3–1.67)

0.432

–

–

Presence of nsVTb

64 (49.2%)

10 (47.6%)

51 (54.3%)

0.82 (0.35–1.94)

0.656

–

–

Minimum heart rate (b.p.m.) Maximum heart rate (b.p.m.)

43.9 + 9.7 110.8 + 28.1

42.1 8.5 114.4 31.8

42.9 9.2 111.2 28.7

0.99 (0.94–1.04) 1 (0.99– 1.02)

0.699 0.507

– –

– –

Mean heart rate (b.p.m.)

65.5 + 11.3

67.6 12.6

64.1 10.7

1.02 (.99–1.06)

0.267

–

–

SDNN (ms) RMSSD (ms)

110.5 + 41.9 30.7 (20.4–50.2)

100.3 40.2 29.8 (18.5– 45.2)

116.2 40.4 32.3 (22.1– 56)

2.37 (0.79–7.18)c 0.126 0.99 (0.97–1.01) 0.178

– –

– –

376 (148–773)

1 (0.99– 1.01)

LF

304 (115– 649)

226 (75– 425)

0.176

–

–

HF TO (%)

131 (59–298) 20.5 (21.5–0.4)

104 (49– 228) 142 (81– 396) 1 (0.99– 1.01) 0.227 .08 (21.05– 0.38) 20.78 (21.62–0.09) 2.17 (0.88–5.35)d 0.093

– –

– –

TS (ms/RRI)

2.9 (1.5– 5.5)

2.03 (1.06– 3.1)

2.24 (0.88–5.71)e 0.090

–

HRT category I/II vs.

3.08 (1.74– 5.5)

–

3.17 (1.14–8.85)

0.028 3.38 (1.17– 9.71)

0.024

59 (49.2%)

14 (73.7%)

38 (43.7%)

61 (50.8%)

5 (26.3%)

49 (56.3%)

DC (ms) QTa/RR (ms)f

3.5 (1.0– 5.2) 0.17 + 0.19

2.5 (22.8–5.18) 0.12 (0.09– 0.24)

3.6 (1.4–5.1) 0.17(0.12–0.22)

1 (0.96– 1.05) 0.37 (0.03–5.26)

0.994 0.461

– –

– –

QTe/RR (ms)g

0.18 + 0.24

0.14 (0.07– 0.23)

0.18 (0.11– 0.23)

0.55 (0.07–4.62)

0.584

–

–

0

ejection fraction was associated with IRM on univariate analysis, and was the strongest independent predictor of total mortality and heart failure death/heart transplantation. Only the occurrence of nsVT on Holter predicted AI.

Additional information provided by 24 h Holter electrocardiogram In this trial we specifically studied the association between 24 h Holter-based parameters and IRM. Heart rate turbulence and heart rate variability Heart rate turbulence is a strong and independent predictor of SCD and total mortality in IHD, as was demonstrated in six retrospective and four prospective studies.18 The predictive power of HRT increases when it is combined with other parameters, such as repolarization alternans and LVEF.19 Heart rate turbulence has proven to be related to adverse outcome both in patients with reduced and preserved LVEF.19,20 The usefulness of HRV in the prediction of arrhythmia and death has also been extensively studied in patients with IHD.21,22 In 1284 patients with recent myocardial infarction included in the ATRAMI trial, SDNN , 70 ms was an independent predictor of cardiac mortality during a follow-up period of 21 months. In a subgroup of

patients with reduced LVEF, SDNN identified patients at particularly high risk.23 Despite the evidence linking HRT and HRV with arrhythmia, their usefulness to identify patients likely to benefit from ICD implantation might be limited, because they are both also associated with nonarrhythmic death.18,21 In our study population, HRT was a strong and independent predictor of IRM and total mortality, but not of AI. This lack of a relationship between HRT and AI was also reported in a study by Berkowitsch et al.,24 based on 10 min Holter recordings from 884 MADIT (Multicenter Automatic Defibrillator Implantation Trial)-II patients. Other trials have shown that HRT is a marker of heart failure severity, further providing a link between HRT and adverse outcome in general.18 We did not find an association between HRV parameters and IRM or any other endpoint, which might be due to power issues. SDNN showed a trend towards significance for heart failure death/heart transplantation (P ¼ 0.085). Only 15 patients (12.9%) had severely depressed SDNN (,70 ms). Five of them underwent heart transplantation, two died from terminal heart failure.

Ventricular ectopy There is extensive literature linking nsVT and frequent VPB (usually ≥10 per hour) with adverse outcome.10 However, most studies

Downloaded from by guest on November 13, 2015

IRM, ICD-resistant mortality; HR, hazard ratio; CI, confidence interval; VPB, ventricular premature beats; nsVT, non-sustained ventricular tachycardia; SDNN, standard deviation of the NN intervals; RMSSD, square root of the mean squared differences of successive NN intervals; LF, low-frequency power; HF, high-frequency power; TO, turbulence onset; TS, turbulence slope; RRI, R –R interval; HRT, heart rate turbulence; DC, deceleration capacity. Bold values are considered statistically significant (P , 0.05) a Cox model HR adjusted for BMI and NYHA class. b Defined as three or more consecutive beats at .120 b.p.m., terminating spontaneously in ,30 s. c SDNN , 70 vs. ≥70 ms. d TO ≥ 0 vs. ,0%. e TS ≤ 2.5 vs. .2.5 ms/RRI. f Slope of QTa plotted against RR interval. g Slope of QTe plotted against RR interval.

1074

T. Marynissen et al.

Role of left ventricular ejection fraction

Percent IRM-free

A 100 80 60 P = 0 . 008 40 BMI < 25 kg/m2

20

BMI < 25 kg/m2 0 0

24

48

96

120

72

96

120

Percent IRM-free

B 100 80 60 P = 0 . 003 40 20

NYHA III–IV NYHA I–II

0 0

24

48 Time

Percent IRM-free

C 100

Other clinical parameters

80 60 P = 0 . 028 40 HRTI–II HRT 0

20 0 0

24

48

72

96

120

Time

Figure 4 Cumulative IRM-free survival for patients stratified to (A) BMI (,25 vs. ≥25 kg/m2), (B) NYHA class (I – II vs. III – IV), (C) HRT category (I– II vs. 0).

showed only limited predictive power when other factors such as LVEF were taken into account.25 We found that the presence of nsVT on 24 h Holter was associated with AI, but not with total mortality in this population treated with an ICD. Frequent VPB showed a trend towards significance for prediction of AI (P ¼ 0.056). Similar results were found in a cohort of MADIT-II patients.24 In our study population, the occurrence of frequent VPB or nsVT was associated with a reduced risk of heart failure death/heart transplantation, a surprising finding that warrants further study. Ventricular ectopy was not associated with IRM.

Body mass index and NYHA functional class were the strongest independent predictors of IRM. A higher NYHA functional class is a marker of decompensated heart failure. Although mortality increases with the severity of heart failure, the proportion of deaths due to SCD decreases as deaths due to progressive pump failure increase.10 This is reflected in the guidelines, which limit ICD implantation to patients in NYHA I–III. The benefit of ICD treatment in patients with NYHA III remains a matter of controversy. In the SCD-HEFT trial ICD implantation in patients in NYHA III did not affect mortality.2 This was not the case in MADIT-II and DEFINITE (Defibrillators In Nonischemic Cardiomyopathy Treatment Evaluation).4,5 We found that patients in NYHA III –IV had a four-fold risk of dying without benefit from their ICD. This patient group included only two individuals in NYHA IV. Combining NYHA class with other risk stratification techniques might help selecting patients in NYHA III, who are unlikely to benefit from ICD therapy. The finding that an increased BMI is associated with a decreased risk of IRM (with a trend for total mortality) seemed to be paradoxical, taking into account the association between obesity and cardiovascular disease. On the other hand, patients with end-stage cardiac or non-cardiac disease (e.g. terminal heart failure, cancer) can be expected to have a lower body weight and BMI, due to cachexia. These patients are more likely to die from non-arrhythmic causes. An inverse association between BMI and outcome was also demonstrated by Choy et al.27 in 1231 patients with IHD and left ventricular dysfunction enroled in MADIT-II. Large population trials have also shown a U-shaped relationship between BMI and total mortality,

Downloaded from by guest on November 13, 2015

72 Time

Left ventricular ejection fraction has been identified as a predictor of SCD and total mortality in a large number of observational trials.10 In patients with IHD and LVEF ≤ 30 –35%, ICD implantation was shown to relatively reduce total mortality by 21– 54% over a followup period of 20–45.5 months in MADIT-I, MADIT-II, and SCD-HEFT (Sudden Cardiac Death in Heart Failure Trial).2 – 4 In the ACC/AHA/ HRS Guidelines for Device-Based Therapy of Cardiac Rhythm Abnormalities (2008), ICD implantation is therefore considered a class IA indication in primary prevention patients with a LVEF ≤ 30% in NYHA class I and patients with a LVEF ≤ 35% in NYHA class II –III.6 On the other hand, LVEF is a marker of total mortality not specifically related to arrhythmic death and the majority of SCD occur in patients with normal LVEF.10 In the absence of other risk factors, poor LVEF is associated with a low arrhythmic mortality.26 The high incidence of IRM in our patient group [21 cases of IRM on a total of 33 deaths (63.6%)] illustrates the limitations of the current guidelines. We showed that a LVEF ≤ 35% was a predictor of IRM, total mortality, and heart failure death/heart transplantation, but not of AI. Patients with a severely reduced left ventricular function were four times more likely to die without benefit from their ICD. Our results confirm the general consensus that there is a need for other risk predictors to identify patients who are less likely to benefit from an ICD, despite a reduced LVEF.

1075

HRT predicts ICD-resistant mortality in IHD

Table 3 Predictors of total mortality (only parameters with univariate P < 0.10 are shown) Fatal outcome (n 5 33)

Survivors (n 5 82)

HR (95% CI) (univariate)

69.6 + 7.9

63.7 + 8.9

2.39 (1.20 –4.76)a

P

HR (95% CI) (multivariate)

P

0.013

1.57 (0.67 – 3.69)a

0.297

0.050

–

–

............................................................................................................................................................................... Clinical parameters Age (years) 2

b

BMI (kg/m )

24.7 + 3.7

26.1 + 3.6

2.01 (1– 4.05)

GFR (mL/min) (Cockroft – Gault)

51.7 + 20.8

71.1 + 22.7

2.38 (1.15 –4.92)c

0.020

1.59 (0.66 – 3.84)c

0.305

LVEF (%)

31.4 + 8.8

35.6 + 10.9

5.31 (2.1 –13.41)d

,0.001

NYHA class

3.78 (1.45 – 9.86)d

0.007

2.93 (1.45 –5.92)

0.003

2.16 (1.05 – 4.46)

0.037

–

III/IV vs.

16 (48.5%)

21 (25.6%)

I/II

17 (51.5%)

61 (74.4%)

LF

238 (135 – 427)

396 (131 –822)

0.99 (0.98 –1)

0.089

–

TO (%)

0.1 + 1.9

20.8 + 1.6

2.03 (0.99 –4.16)e

0.054

–

–

TS (ms/RRI)

2.15 (1.27 –3.20)

3.28 (1.74 – 5.53)

2.63 (1.23 –5.62)f

0.013

1.9 (0.87 –4.13)f,g

0.107

3.77 (1.61 –8.85)

0.002

2.72 (1.13 – 6.54)g

0.025

I/II vs.

23 (76.7%)

29 (38.2%)

0

7 (23.3%)

47 (61.8%)

Holter parameters

HRT category

Table 4 Predictors of heart failure death/heart transplantation (only parameters with univariate P < 0.10 are shown) Event (n 5 33)

Event-free (n 5 97)

HR (95% CI) (univariate)

P

25.1 + 4.2

25.8 + 3.7

2.21 (1.09–4.48)a b

HR (95% CI) (multivariate)

P

0.028

2.72 (1.32–5.59)a

0.007

.............................................................................................................................................................................. Clinical parameters BMI (kg/m2) GFR (mL/min) (Cockroft– Gault)

59.4 + 30.3

68.9 + 22.7

1.81 (0.91– 3.62)

0.091

–

–

LVEF (%) NYHA class

28.5 + 5.8 19 (57.6%)

35.3 + 11.0 28 (28.9%)

8 (2.42– 26.43)c 3.38 (1.68–6.81)

0.001 0.001

5.25 (1.52–18.09)c 2.58 (1.21–5.49)

0.009 0.014

27 (81.8%)

55 (56.7%)

3.04 (1.25–7.37)

0.014

1.71 (0.62–4.76)

0.303

10 (30.3%)

62 (63.9%)

0.30 (0.14–0.62)

0.001

0.33 (0.15–0.75)d

0.008

Presence of nsVT Minimum heart rate (b.p.m.)

9 (27.3%) 46.5 + 9.9

55 (56.7%) 43 + 9.5

0.35 (0.16–0.76) 1.04 (1– 1.08)

0.008 0.028

0.39 (0.33–0.89)d 1.03 (0.99–1.06)d

0.026 0.166

SDNN (ms)

103.5 + 41.8

112.8 + 42

2.21 (0.90–5.43)f

0.085

–

–

1.87 (0.89–3.94)

0.098

–

–

III/IV vs. I/II Use of diuretics Holter parameters Presence of ≥10 VPB per hour e

HRT category I/II vs. 0

19 (63.3%)

40 (44.4%)

11 (36.7%)

50 (55.6%)

HR, hazard ratio; CI, confidence interval; BMI, body mass index; GFR, glomerular filtration rate; LVEF, left ventricular ejection fraction; NYHA, New York Heart Association; VPB, ventricular premature beats; nsVT, non-sustained ventricular tachycardia; SDNN, standard deviation of the NN intervals; HRT, heart rate turbulence. Bold values are considered statistically significant (P , 0.05). a BMI , 25 vs. ≥25 kg/m2. b GFR , 60 vs. ≥60 mL/min. c LVEF ≤ 35 vs. .35%. d Cox model HR adjusted for BMI, LVEF, and NYHA class. e Defined as three or more consecutive beats at .120 b.p.m., terminating spontaneously in ,30 s. f SDNN , 70 vs. ≥70 ms.

Downloaded from by guest on November 13, 2015

HR, hazard ratio; CI, confidence interval; BMI, body mass index; GFR, glomerular filtration rate; LVEF, left ventricular ejection fraction; NYHA, New York Heart Association; LF, low-frequency power; TO, turbulence onset; TS, turbulence slope; RRI, R –R interval; HRT, heart rate turbulence. Bold values are considered statistically significant (P , 0.05) a age .70 vs. ≤70 years. b BMI , 25 vs. ≥25 kg/m2. c GFR , 60 vs. ≥60 mL/min. d LVEF ≤ 35 vs. .35%. e TO ≥ 0 vs. ,0%. f TS ≤ 2.5 vs. .2.5 ms/RRI. g Cox model HR adjusted for LVEF and NYHA class.

1076

T. Marynissen et al.

Table 5 Predictors of appropriate and inappropriate ICD interventions (only parameters with P < 0.10 are shown) Appropriate ICD intervention (n 5 56)

Others (n 5 74)

HR (95% CI) (univariate)

P

............................................................................................................................................................................... Presence of ≥10 VPB per hour

39 (69.6%)

33 (44.6%)

1.75 (0.99–3.10)

0.056

Presence of nsVTa

36 (64.3%)

28 (37.8%)

1.95 (1.13–3.38)

0.017

Atrial fibrillation

Inappropriate ICD intervention (n ¼ 27) 18 (66.7%)

Others (n ¼ 103) 38 (36.9%)

2.69 (1.20–5.99)

0.016

HR, hazard ratio; CI, confidence interval; VPB, ventricular premature beats; nsVT, non-sustained ventricular tachycardia. a Defined as three or more consecutive beats at .120 b.p.m., terminating spontaneously in ,30 s.

with the lowest mortality rate in patients with a BMI of 22.5–25 kg/m2.28

A 100

Future directions

60 P = 0 .007 40 LVEF < 35% LVEF > 35%

20 0 0

24

48

72

96

120

72

96

120

Time

B 100 Percent survival

80 60

P = 0 .037

40 NYHA III–IV NYHA I–II

20 0 0

24

48 Time

C 100 Percent survival

80 60 P = 0 .025 40

In an era of increasing healthcare cost, ICD implantation based on risk stratification techniques for SCD leads to mortality reduction. We expect it to reach this goal in a cost-effective way. From this perspective, it is of particular interest to know which patients, currently receiving an ICD, die without benefiting from it. The endpoint of IRM (defined as death without AI) was used recently by Van Rees et al.15 In a cohort of 900 ICD patients with IHD, they identified five parameters (NYHA class, age, presence of diabetes, LVEF, history of smoking) that could classify patients into low, intermediate, or high risk for IRM. Five-year cumulative incidence of IRM ranged from 10% in low-risk patients to 41% in high-risk patients. Interesting information on the prediction of ICD effectiveness also comes from risk models derived from major clinical ICD trials. Based on data from MADIT-II, Goldenberg et al. developed a risk score consisting of five clinical parameters (NYHA class, age, blood urea nitrogen, QRS duration, presence of AF), which could predict all-cause mortality during long-term follow-up.29,30 A group of high-risk patients was identified, showing no reduction in 8-year all-cause mortality when implanted with an ICD. In SCD-HEFT, similar results were found for patients with a predicted annual mortality .20%, based on a prognostic model validated primarily in heart failure patients (the Seattle Heart Failure Model).31 – 33 In conclusion, extensive evidence supports the existence of a patient group at particularly high risk of dying without benefiting from ICD therapy. Further research is needed to optimize and validate risk models for the identification of these patients. Based on the results of our trial, we believe Holter parameters (especially HRT) might also be of value.

Limitations and strengths HRTI–II HRT 0

20 0 0

24

48

72

96

120

Time

Figure 5 Cumulative survival for patients stratified to (A) LVEF (≤35 vs. .35%), (B) NYHA class (I– II vs. III– IV), (C) HRT category (I– II vs. 0).

Main limitations are the relatively small number of patients and the retrospective methodology. About one-third of our study population consisted of secondary prevention ICD patients, a patient group for which there is general consensus that ICD implantation is useful. However, we found no significant difference between primary and secondary prevention patients with regard to total mortality, AI, and IRM. When uni- and multivariate analyses were repeated for primary prevention patients only, similar results were found (data not shown).

Downloaded from by guest on November 13, 2015

Percent survival

80

HRT predicts ICD-resistant mortality in IHD

Conclusion Implantable cardioverter-defibrillator implantation in IHD based on NYHA class and poor LVEF leads to selection of patients with a high risk of dying despite their ICD. Heart rate turbulence may have added value for the identification of poor candidates for ICD therapy, but in our experience Holter parameters are limited in their ability to predict AI. Research for new non-invasive risk stratification tools should focus more on the discrimination between arrhythmic and non-arrhytmic mortality. Conflict of interest: none declared.

Funding The research leading to these results has received funding from the European Community’s Seventh Framework Program FP7/2007– 2013 (under grant agreement no. HEALTH-F2-2009-241526, EUTrigTreat) and from an unconditional grants from Biotronik, Boston Scientific, and Medtronic Belgium. R.W. is supported as a clinical researcher by the Fund for Scientific Research Flanders.

References

12. Bauer A, Malik M, Schmidt G, Barthel P, Bonnemeier H, Cygankiewicz I et al. Heart rate turbulence: standards of measurement, physiological interpretation, and clinical use: International Society for Holter and Noninvasive Electrophysiology Consensus. J Am Coll Cardiol 2008;52:1353 –65. 13. Hohnloser SH, Kuck KH, Dorian P, Roberts RS, Hampton JR, Hatala R et al. Prophylactic use of an implantable cardioverter-defibrillator after acute myocardial infarction. N Engl J Med 2004;351:2481 – 8. 14. Steinbeck G, Andresen D, Seidl K, Brachmann J, Hoffmann E, Wojciechowski D et al. Defibrillator implantation early after myocardial infarction. N Engl J Med 2009;361: 1427 –36. 15. Van Rees JB, Borleffs CJ, van Welsenes GH, van der Velde ET, Bax JJ, van Erven L et al. Clinical prediction model for death prior to appropriate therapy in primary prevention implantable cardioverter defibrillator patients with ischaemic heart disease: the FADES risk score. Heart 2012;98:872 –7. 16. Bauer A, Barthel P, Schneider R, Ulm K, Mu¨ller A, Joeinig A et al. Improved Stratification of Autonomic Regulation for risk prediction in post-infarction patients with preserved left ventricular function (ISAR-Risk). Eur Heart J 2009;30:576 –83. 17. Pathak A, Curnier D, Fourcade J, Roncalli J, Stein PK, Hermant P et al. QT dynamicity: a prognostic factor for sudden cardiac death in chronic heart failure. Eur J Heart Fail 2005;7:269 –75. 18. Bauer A, Zu¨rn CS, Schmidt G. Heart rate turbulence to guide treatment for prevention of sudden death. J Cardiovasc Pharmacol 2010;55:531 –8. 19. Exner DV, Kavanagh KM, Slawnych MP, Mitchell LB, Ramadan D, Aggarwal SG et al. Noninvasive risk assessment early after a myocardial infarction the REFINE study. J Am Coll Cardiol 2007;50:2275 –84. 20. La Rovere MT, Pinna GD, Maestri R, Barlera S, Bernardinangeli M, Veniani M et al. Autonomic markers and cardiovascular and arrhythmic events in heart failure patients: still a place in prognostication? Data from the GISSI-HF trial. Eur J Heart Fail 2012;14:1410 –9. 21. Huikuri HV, Stein PK. Clinical application of heart rate variability after acute myocardial infarction. Front Physiol 2012;3:41. 22. Heart rate variability. Standards of measurement, physiological interpretation, and clinical use. Task Force of the European Society of Cardiology and the North American Society of Pacing and Electrophysiology. Eur Heart J 1996;17:354 – 81. 23. La Rovere MT, Bigger JT Jr, Marcus FI, Mortara A, Schwartz PJ. Baroreflex sensitivity and heart-rate variability in prediction of total cardiac mortality after myocardial infarction. Lancet 1998;351:478 –84. 24. Berkowitsch A, Zareba W, Neumann T, Erdogan A, Nitt SM, Moss AJ et al. Risk stratification using heart rate turbulence and ventricular arrhythmia in MADIT II: usefulness and limitations of a 10-minute Holter recording. Ann Noninvasive Electrocardiol 2004;9:270 –9. 25. Katritsis DG, Zareba W, Camm AJ. Nonsustained ventricular tachycardia. J Am Coll Cardiol 2012;60:1993 –2004. 26. Buxton AE, Lee KL, Hafley GE, Pires LA, Fisher JD, Gold MR et al. Limitations of ejection fraction for prediction of sudden death risk in patients with coronary artery disease: lessons from the MUSTT study. J Am Coll Cardiol 2007;50:1150 –7. 27. Choy B, Hansen E, Moss AJ, McNitt S, Zareba W, Goldenberg I. Relation of body mass index to sudden cardiac death and the benefit of implantable cardioverterdefibrillator in patients with left ventricular dysfunction after healing of myocardial infarction. Am J Cardiol 2010;105:581 –6. 28. Whitlock G, Lewington S, Sherliker P, Clarke R, Emberson J, Halsey J et al. Body-mass index and cause-specific mortality in 900 000 adults: collaborative analyses of 57 prospective studies. Lancet 2009;373:1083 –96. 29. Goldenberg I, Vyas AK, Hall WJ, Moss AJ, Wang H, He H et al. Risk stratification for primary implantation of a cardioverter-defibrillator in patients with ischemic left ventricular dysfunction. J Am Coll Cardiol 2008;51:288–96. 30. Barsheshet A, Moss AJ, Huang DT, McNitt S, Zareba W, Goldenberg I. Applicability of a risk score for prediction of the long-term (8-year) benefit of the implantable cardioverter-defibrillator. J Am Coll Cardiol 2012;59:2075 –9. 31. Levy WC, Lee KL, Hellkamp AS, Poole JE, Mozaffarian D, Linker DT et al. Maximizing survival benefit with primary prevention implantable cardioverter-defibrillator therapy in a heart failure population. Circulation 2009;120:835 –42. 32. Levy WC, Mozaffarian D, Linker DT, Sutradhar SC, Anker SD, Cropp AB et al. The Seattle Heart Failure Model: prediction of survival in heart failure. Circulation 2006; 113:1424 –33. 33. Mozaffarian D, Anker SD, Anand I, Linker DT, Sullivan MD, Cleland JG et al. Prediction of mode of death in heart failure: the Seattle Heart Failure Model. Circulation 2007;116:392 –8.

Downloaded from by guest on November 13, 2015

1. Chugh SS, Reinier K, Teodorescu C, Evanado A, Kehr E, Al Samara M et al. Epidemiology of sudden cardiac death: clinical and research implications. Prog Cardiovasc Dis 2008;51:213 –28. 2. Bardy GH, Lee KL, Mark DB, Poole JE, Packer DL, Boineau R et al. Amiodarone or an implantable cardioverter-defibrillator for congestive heart failure. N Engl J Med 2005; 352:228 –37. 3. Moss AJ, Hall WJ, Cannom DS, Daubert JP, Higgins SL, Klein H et al. Improved survival with an implanted defibrillator in patients with coronary disease at high risk for ventricular arrhythmia. N Engl J Med 1996;335:1933 –40. 4. Moss AJ, Zareba W, Hall WJ, Klein H, Wilber DJ, Cannom DS et al. Prophylactic implantation of a defibrillator in patients with myocardial infarction and reduced ejection fraction. N Engl J Med 2002;346:877 –83. 5. Kadish A, Dyer A, Daubert JP, Quigg R, Estes NA, Anderson KP et al. Prophylactic defibrillator implantation in patients with nonischemic dilated cardiomyopathy. N Engl J Med 2004;350:2151 –8. 6. Epstein AE, DiMarco JP, Ellenbogen KA, Estes NA III, Freedman RA, Gettes LS et al. ACC/AHA/HRS 2008 Guidelines for Device-Based Therapy of Cardiac Rhythm Abnormalities: a report of the American College of Cardiology/American Heart Association Task Force on Practice Guidelines (Writing Committee to Revise the ACC/AHA/ NASPE 2002 Guideline Update for Implantation of Cardiac Pacemakers and Antiarrhythmia Devices): developed in collaboration with the American Association for Thoracic Surgery and Society of Thoracic Surgeons. Circulation 2008;117:e350–408. 7. Moss AJ, Greenberg H, Case RB, Zareba W, Hall WJ, Brown MW et al. Long-term clinical course of patients after termination of ventricular tachyarrhythmia by an implanted defibrillator. Circulation 2004;110:3760 –5. 8. Liew R. Electrocardiogram-based predictors of sudden cardiac death in patients with coronary artery disease. Clin Cardiol 2011;43:466 –73. 9. Van Brabandt H, Neyt M. Pitfalls in health-economic evaluations: the case of cost effectiveness of prophylactic implantable cardioverter-defibrillator therapy in Belgium. Europace 2009;11:1571 –3. 10. Goldberger JJ, Cain ME, Hohnloser SH, Kadish AH, Knight BP, Lauer MS et al. American Heart Association/American College of Cardiology Foundation/Heart Rhythm Society Scientific Statement on Noninvasive Risk Stratification Techniques for Identifying Patients at Risk for Sudden Cardiac Death. A scientific statement from the American Heart Association Council on Clinical Cardiology Committee on Electrocardiography and Arrhythmias and Council on Epidemiology and Prevention. J Am Coll Cardiol 2008;52:1179 –99. 11. Nicolini P, Ciulla MM, De Asmundis C, Magrini F, Brugada P. The prognostic value of heart rate variability in the elderly, changing the perspective: from sympathovagal balance to chaos theory. Pacing Clin Electrophysiol 2012;35:622 –38.

1077