OUP user menu

Cardiac resynchronization therapy in combination with implantable cardioverter–defibrillator

J. Claude Daubert , Christophe Leclercq , Philippe Mabo
DOI: http://dx.doi.org/10.1093/europace/eup305 v87-v92 First published online: 27 October 2009


Device therapy for heart failure has a spectacular development during the last 10 years. Patients with chronic heart failure might benefit from electrical therapy with a view to: (i) resynchronize the failing and dyssynchronized heart and improve its mechanical performance or (ii) prevent the risk of sudden death by automatic defibrillation. These two therapies can be applied together with a combined device, the biventricular implantable cardioverter-defibrillator (CRT-D). Today, the proportion of CRT-D devices among all CRT devices implanted around the world is more than 75% and more than 85% in the USA. This review article will try to answer some important questions for clinical practice: is the growing use of CRT-D devices supported by clinical evidence? Is the risk-benefit profile of CRT-D favourable in particular in mildly symptomatic patients?

  • Cardiac resynchronisation therapy
  • Implantable cardioverter defibrillator (ICD)
  • Biventricular ICD
  • Heart failure


Device therapy for heart failure (HF) has known as a spectacular development during the past 10 years. Patients with chronic HF might benefit from electrical therapy with a view to: (i) resynchronize the failing and dyssynchronized heart and improve its mechanical performance, or (ii) prevent the risk of sudden death by automatic defibrillation. These two therapies can be applied separately with dedicated devices, using specific pacemakers for cardiac resynchronization therapy (CRT-P), or an implantable cardioverter–defibrillator (ICD). Or they can be applied together with a combined device, the biventricular ICD (CRT-D). Today, the proportion of CRT-D devices among all CRT devices implanted is more than 75% worldwide and more than 85% in the USA. Is the growing use of CRT-D devices supported by clinical evidence? Is the risk-benefit profile of CRT-D favourable in HF patients? This review article will try to answer these important questions for clinical practice.

Rationale for the use of implantable cardioverter–defibrillator for primary prevention in heart failure patients

The use of ICDs for primary prevention is based on the high proportion of sudden cardiac deaths in HF. This percentage, observed in the control groups of large drug trials in New York Heart Association (NYHA) functional class II to IV patients ranges between 35 and 50%, reaching a maximum of 58% in the Metoprolol Randomized Intervention Trial in Congestive Heart Failure (MERIT-HF),1 which is probably explained by the inclusion of 41% of patients in NYHA functional class II. As the severity of HF increases, the proportion of sudden deaths relative to overall mortality decreases, and conversely, the percentage of deaths attributed to worsening HF increases.1 Therefore, the potentially greatest benefit of ICDs is likely to be conferred to patients with mild-to-moderate HF. The Sudden Cardiac Death in Heart Failure (SCD-HeFT) trial is thus far the only trial showing a lowering of all-cause mortality by ICD therapy in patients with mild-to-moderate HF and left ventricular ejection fraction (LVEF) <35%.2 In that study, which enrolled patients optimally treated for HF, ICD therapy was compared with amiodarone or placebo. Although amiodarone conferred no benefit compared with placebo, a significant 23% relative reduction in the risk of overall mortality (RRR) was observed among the ICD recipients. These apparently spectacular results must be interpreted cautiously. The demographic characteristics of the SCD-HeFT population were quite peculiar (see what follows). In addition, the trial had to include over 2500 patients followed up for a median of 45.5 months and had to be extended for 1 year to show a statistically significant albeit modest absolute risk reduction (ARR) of 7.2% at 5 years. Finally, in agreement with the epidemiologic data mentioned earlier, a subgroup analysis showed that the risk reduction was limited to functional class II patients (RRR, 46%; ARR, 11.9% at 5 years). No treatment benefit was observed among the 30% of patients who were in NYHA functional class III (hazard ratio, 1.16). These latter observations, which place the impact of ICD therapy in perspective, must also have been considered when choosing a device for a cardiac resynchronization therapy (CRT) candidate, because at that time CRT has only been validated in NYHA functional class III to IV patients.

Evidence-based clinical efficacy of cardiac resynchronization therapy in patients with moderate-to-severe heart failure: is there a superiority of CRT-D over CRT-P?

During the last decade, several randomized trials have been conducted to ascertain the clinical impact of CRT in patients with advanced HF and in sinus rhythm, with or without indications for an ICD.310 Meta-analyses have also been published.1113 The usual enrolment criteria have been: (i) NYHA functional class III or IV function despite optimal drug treatment, (ii) LVEF <35%, (iii) QRS duration >120–150 ms, and (iv) normal sinus rhythm (Table 1).

View this table:
Table 1

Main cardiac resynchronization therapy randomised studies: inclusion criteria

StudyAuthors, countryNo. of patientsAgeNYHA classLVEF (%)LVEDD (mm)QRS (ms)Follow-upMain result
MUSTIC-SR 2001Cazeau et al., Europe5863III≤35%≥60≥1503 monthsPositive effect of CRT
MUSTIC-AF 2003Leclercq et al., Europe4363III≤35%≥60>200 (paced)3 monthsNo significant effect
MIRACLE 2002Abraham et al., USA45364III, IV≤35%≥55≥1306 months40% reduction for ‘CE’
PATH CHF 2002Auricchio et al., Germany4160III, IV≤35%NA≥1206 monthsPositive effect of CRT
MIRACLE ICD 2003Young et al., USA36963III, IV≤35%≥55≥1306 monthsPositive effect of CRT
CONTAK CD 2003Higgins et al., USA227NAII-IV≤35%NA≥1206 monthsPositive effect of CRT
MIRACLE ICD II 2004Abraham et al., USA18663II≤35%≥55≥1306 monthsPositive effect of CRT
PATH CHF II 2004Auricchio et al., Germany8960III, IV≤35%NA≥1203 monthsPositive effect of CRT
COMPANION 2004Bristow et al., USA152067III, IV≤35%NA≥12012 months12% reduction for ‘CE’
CARE HF 2005Cleland et al., Europe81467III, IV≤35%≥30 mm/height≥12029.4 months37% reduction for ‘CE’
CARE HF extension 2006Cleland et al., Europe81367III, IV≤35%≥30 mm/height≥12036.4 months55% reduction in sudden death
REVERSE 2008Linde et al. Europe-USA61062I, II≤40%≥55≥12012 monthsClinical composite response: NS
REVERSE Europe 2009Daubert et al., Europe–USA26261I, II≤40%≥55≥12024 months44% reduction in risk of worsening
MADIT-CRT 2009Moss et al., US and Europe182064I, II≤30%NA≥13026 months34% reduction in risk of HF events and death
  • LVEF, left ventricular ejection fraction; LVEDD, left ventricular end-diastolic diameter; NA, not availabable; CE, composite endpoint = death or hospitalization for major cardio-vascular event.

The concordant results of these studies have confirmed that in this patient population, CRT combined with optimal pharmacological therapy has a highly favourable and sustained impact12,13 on all of the treatment objectives: (i) improvement of symptoms and exercise capacity with a mean decrease in NYHA functional class by 0.5 to 0.8 points, a mean increase in exercise capacity by 10% (Vo2 peak) to 20% (6 min walking distance), and a highly significant improvement in quality-of-life score; (ii) major reduction in HF-related morbidity with, in particular, a 30 to 52% decrease in the number of hospitalizations for worsening HF;912 (iii) reverse ventricular remodelling: a consistent finding in the randomized trials designed with ≥6 months of follow-up has been an up to 15% absolute reduction in LV end-diastolic diameter, and an up to 6% increase in LVEF, conferred by CRT10,13,14—these effects were significantly greater in patients with non-ischaemic than in patients with ischaemic heart disease;14,15 and (iv) impact of CRT on mortality: the CARE-HF and Comparison of Medical Therapy, Pacing, and Defibrillation in Heart Failure (COMPANION) trials examined the effects of CRT on morbidity and mortality.9,10 In the COMPANION trial,9 CRT-P and CRT-D were both associated with a 20% reduction in the primary combined endpoint of all-cause mortality and all-cause hospitalization (P < 0.01). However, only CRT-D, compared with controls, was associated with a significant decrease in total mortality at 1 year (RRR, 36%; ARR, 7%; P = 0.003), whereas the 24% relative (absolute, 4%) reduction in mortality associated with CRT-P was nearly statistically significant (P = 0.059), but there was no head-to-head comparison to be planned in the study protocol between the two treatment modalities. The comparison was actually performed as a post hoc analysis16 and failed to show any statistically significant difference (HR: 0.92; P = 0.33) in favour of CRT-D when compared with CRT-P in terms of incremental survival benefit.

The CARE-HF trial enrolled 813 patients.10 The CRT-P plus standard HF pharmacological treatment was compared with pharmacologic treatment alone. At the end of a mean follow-up of 29 months, a 37% relative risk reduction in the composite endpoint of death and hospitalization for major cardiovascular events (P < 0.001) and 36% in the risk of death (ARR, 10%; P < 0.002) were observed. The effect on mortality was mainly attributable to a marked reduction in HF-related deaths. In the main study, 35% of the deaths in the CRT arm were sudden, very similar to what was observed in the CRT-P arm of the COMPANION trial,9 but longer follow-up (36 months) in the CARE-HF extension study showed a delayed but highly significant (HR: 0.54; P = 0.006) reduction in the risk of sudden cardiac death in the CRT group.17

Based on COMPANION data, it thus appears that CRT-D reduces the risk of sudden death when compared with CRT-P, at least in a mid-term perspective. At 1 year follow-up, the sudden cardiac death incidence was 16% in the CRT-D arm, a 55% relative risk reduction when compared with CRT-P. But there is no current evidence that this early benefit from adding an ICD is long-lived and that it may positively impact total mortality. The results of the post hoc analysis of COMPANION do no support this hypothesis.16

In practice, only a new randomized study comparing the two treatment modalities might resolve this issue. Based on the results of the CARE-HF trial,10 and assuming that the combination of CRT and defibrillation back-up could prevent two-thirds of sudden deaths, a study would require 1300 patients per group and a follow-up period equivalent to that of the CARE-HF trial to have a statistical power of 90% to detect a 5% absolute relative risk reduction of death from any cause with the use of combined therapy compared with CRT alone.

Based on current scientific evidence, a class I-a recommendation for CRT was issued by both electrophysiology18,19 and HF20,21 communities. Recent ESC guidelines on the management of acute and chronic HF20 gave the same level of recommendation for CRT-P and CRT-D use. The two treatment modalities are both recommended to reduce morbidity and mortality in patients in NYHA class III-IV who are symptomatic despite optimal medical therapy, and who have a reduced LVEF (<35%) and QRS prolongation (QRS width >120 ms). No clear preference is given to one treatment modality over the other.

Influence of heart failure stage

Severely diseased patients

Subgroup analyses in COMPANION9,22 suggest that the more diseased patients, notably non-ambulatory NYHA class IV patients or patients with renal dysfunction, do not benefit from adding an ICD to CRT even if they are at higher risk of sudden cardiac death than less diseased patients.

Less severe patients

The clinical efficacy of CRT in patients with NYHA class I to II symptoms and wide QRS has been a matter of extensive debate during the last 5 years. Demonstrating significant benefit and favourable risk-benefit ratio in these relatively low-risk patients is likely to be lengthy and arduous because, a priori, patients have few or no symptoms, are rarely hospitalized for decompensated HF, and have a low death rate when receiving optimized medical therapy. As a consequence, the main goal of treatment for NYHA class I to II patients is largely different from those set for patients in class III to IV, consisting primarily of: (i) preventing HF and disease progression and (ii) lowering cardiac mortality, mainly by reducing the risk of sudden death. The role of combined CRT–ICD could be of particular interest in this population as suggested by the results of SCD-HeFT.2

Assessing these new objectives requires the use of specific endpoints, the most relevant being: (i) a composite of HF events and mortality as used in MADIT-CRT or of symptoms, HF morbidity and mortality like the Packer's clinical composite response used as primary endpoint in REVERSE23 and (ii) change in cardiac anatomy and function indicating reverse remodelling. The link between reverse remodelling and clinical outcome remains uncertain. There is still no evidence from randomized trials that reverse remodelling is an independent predictor of outcome.

In the study by Higgins et al.,6 a significant degree of reverse remodelling was observed after 6 months of CRT in a subgroup of patients in NYHA functional class I to II, although the benefits were less prominent than in the much larger group of patients in NYHA functional class III to IV. Similar observations were made in the Multicenter Insync Randomized Clinical Evaluation (MIRACLE ICD II) study.8 This small trial randomly assigned NYHA class II patients to CRT vs. no CRT, who all received a CRT-D for an accepted ICD indication. At the end of the 6 month blinded period, there was no significant difference in the primary study objective (peak Vo2), although a significant improvement in the clinical composite endpoint was observed in the group assigned to CRT compared with the controls. These observations suggest that CRT might have a favourable impact on the outcome of patients with less-advanced HF and ventricular dyssynchrony.

This issue was recently examined in two large randomized trials. The purpose of the REVERSE study was to assess the safety and efficacy of CRT in addition to optimal medical therapy, in patients with asymptomatic or mild (ACC/AHA stage C NYHA class I or II) HF in a large parallel randomized trial with a 12 month double-blind duration in the USA24 and 24 months in Europe.25 Since only mildly or asymptomatic patients were included, the primary endpoint was the percentage worsened by HF clinical composite response and not the traditional analysis of this endpoint that is the distribution of worsened, unchanged, or improved. The prospectively powered secondary endpoint was LV end-systolic volume index (LVESVi). Among other secondary endpoints were hospitalizations for worsening of HF and mortality. A total of 610 patients with NYHA Class I or II HF with a QRS duration ≥120 ms and a LVEF ≤40% received a CRT device (±defibrillator) and were randomly assigned with a 2:1 ratio to active CRT (CRT-ON; n = 419) or control (CRT-OFF; n = 191). The baseline characteristics showed that REVERSE patients were younger, had more narrow QRS complex, better quality of life score, and larger 6 min walk distance than patients in previous trials of both NYHA III-IV patients or NYHA II patients. In spite of better compliance to medication according to HF guidelines than in previous CRT studies, reduction of LV function in REVERSE was comparable to the patients in the previous studies.

At 12 months in the main study, 16% of patients in CRT-ON worsened by the clinical composite HF endpoint compared to 21% in CRT-OFF (P = 0.10). The change in the 6 min walk distance at 12 months, quality of life by the Minnesota score and the Kansas City score and NYHA classification did not improve with CRT. Importantly, patients assigned to CRT-ON experienced a greater improvement in LVESVi (−18.4 ± 29.5 mL/m2 vs. −1.3 ± 23.4 mL/m2, P < 0.0001) and other measures of LV remodelling. Like previous studies, the extent of reverse remodelling was two to three times greater in non-ischaemic compared with ischaemic HF patients. The effect of CRT on the HF clinical composite response in various subgroups was evaluated using odds ratios. The overall odds ratio for the study was 0.70 in favour of CRT-ON and the point estimates for nearly all subgroups favoured CRT-ON, demonstrating consistent results across the study population. Time to first HF hospitalization was significantly delayed in CRT-ON (hazard ratio 0.47, P = 0.03).

The 24 month results of the European patients who remained in their double-blind assignment for 24 months were recently reported.25 In this long term follow-up of 262 European patients, 180 were assigned to CRT-ON and 82 to CRT-OFF. Unlike in the main study, 19% patients worsened in the CRT-ON group vs. 34% in the CRT-OFF group (P = 0.01). In spite of this, no significant difference in 6 min walk distance, quality of life, or NYHA classification was observed between groups. LV end-systolic volume index decreased by a mean of 27.5 ± 31.8 mL/m2 in the CRT-ON, vs. 2.7 ± 25.8 mL/m2 in the CRT-OFF group (P < 0.0001). Reverse remodelling by CRT was thus progressive with the greatest effect during the first 6 months and with further improvement developing over the following 12 months. This progressive reverse remodelling was accompanied by a significant delay in time to first HF hospitalization or death (hazard ratio 0.38; P = 0.003) by CRT suggesting that CRT prevents the progression of disease in patients with asymptomatic or mildly symptomatic LV dysfunction when applied over 1–2 years. Whether this reverse remodelling effect translates in a better clinical outcome remains to be demonstrated.

The MADIT-CRT trial26,27 has been recently completed The primary objective of MADIT CRT is to determine if CRT-D in high risk relatively asymptomatic patients with ischaemic and non-ischaemic cardiomyopathy will significantly reduce the combined endpoint of all-cause mortality or HF events, whichever appears first when compared with ICD therapy alone. The superiority of CRT-D was solely driven by a 41% reduction in the risk of a first heart-failure event, since the mortality was not influenced by the allocated type of device. The secondary objectives are measures of reverse remodelling after 12 months with all cause mortality as one of the tertiary endpoints. To be included, patients with ischaemic aetiology could be in NYHA I or II and those with non-ischaemic aetiology had to be in NYHA II. All needed to have LV dysfunction with LVEF <30% and sinus rhythm with QRS duration >130 ms. Patients were randomly assigned in a 3:2 ratio to receive CRT-D or a stand-alone single chamber ICD. Randomization was stratified by aetiology and centre. Length of follow-up for the individual patient was dependent on date of study entry, since all patients will be followed to a common study termination date. The assumption was that CRT-D will result in a 25% reduction in the 2 year cumulative endpoint when compared with ICD alone i.e. from 30 to 22.6%. A total of 1820 patients were included, with a mean age of 64.5 years. The proportion of women was 25%; HF aetiology was classified as ischaemic in 54.9% patients with a small minority of NYHA class I (15%). All other patients had NYHA class II symptoms. Patients were well treated with regard to guidelines with notably a prescription rate of 97% for ACE-inhibitor and/or Angiotensin receptor blocker, 93.2% for beta-blockers, 31% for aldosterone antagonist, and 67.3% for lipid lowering drugs. As expected by the inclusion criteria, the mean LVEF was lower (24 ± 5%) than in REVERSE.24 A total of 1089 patients received a CRT-D device and 731 an ICD device. The implantation success rate was 92.5% in the CRT-D group. The trial was stopped on 22 June 2009 after the superiority boundary was crossed. The mean follow-up time was 2.4 years.

A primary endpoint event occurred in 17.2% patients in the CRT-D group and in 25.3% in the ICD group [HR: 0.66 (0.52–0.84); P < 0.001], indicating a 34% relative risk reduction. The primary endpoint was driven by HF events (13.9 vs. 22.8%). The rate of deaths was similar in the two groups: 6.8% in the CRT-D group and 7.3% in the ICD group (HR: 1.00; P = 0.99). The mean annual mortality rate was ∼3% in the two groups. Stratification by aetiology did not show any significant difference between ischaemic and non-ischaemic HF patients (HR: 0.67 and 0.62, respectively), but the total mortality rate was twice greater in the ischaemic group when compared with the non-ischaemic group.

Subgroup analysis identified two main interactions with regard to gender and QRS duration. The clinical benefit was significantly greater in women [HR: 0.37 (0.22–0.61)] when compared with men [HR: 0.76 (0.59–0.97), P = 0.01 for interaction] and in patients with QRS duration >150 ms [HR: 0.48 (0.37–0.64)] when compared with QRS duration <150 ms [HR: 1.06 (0.74–1.52), P = 0.001 for interaction]. Due to insufficient power, no conclusion could be drawn for NYHA class I.

At 1 year follow-up, the LVESV was reduced by, on average, 52 mL in the CRT-D group vs. 18 mL in the ICD group (P < 0.001). The LVEF increased by a mean of 11% in the CRT-D group vs. 3% in the ICD group (P < 0.001). These data were remarkably consistent with those reported previously in REVERSE.24 Reported safety data do not allow a precise evaluation of the risk-benefit ratio.

Based on the consistent quite similar results of REVERSE and MADIT CRT, it can be concluded that CRT prevents HF progression and delays disease progression in mildly symptomatic HF (NYHA class II) patients with wide QRS complex. The results of an additional trial in NYHA class II patients are expected in 2010.28

Cost-effectiveness and risk-benefit ratio of CRT-D vs. CRT-P

Cost-effectiveness analyses based on the COMPANION and CARE-HF data29,30 indicate that both CRT-P and CRT-D are cost-effective with regards to accepted benchmarks for therapeutic interventions. The two therapies are cost-effective at the notional willingness to pay thresholds of approximately €15 000 for CRT-P and €35 000–50 000 for CRT-D per Qaly (quality adjusted life year) gained. The incremental cost to pay per Qaly gained is thus two to three times higher with CRT-D when compared with CRT-P. Further analyses30 also showed that the overcost of using a CRT-D strategy compared with a CRT-P strategy is age-sensitive. Overcost increases progressively with age at the time of implantation. For example, Yao et al.30 calculated an incremental cost of €48 800 for a patient implanted at 65 years and of €73 300 when implanted at 75 years. CRT-D thus appears less attractive in older patients.

Another important issue to discuss is safety. The main safety issue for CRT devices is linked to the ventricular defibrillation lead and thus only concerns the CRT-D systems. Poor ICD lead reliability contrasts with the excellent long-term performances of pacing leads, including LV leads. Recent reports indicate that the overall risk of ICD lead failure is much higher than expected, up to 15% at a median follow-up of 2.6 years.31 It increases progressively over time with an estimated survival rate without lead defect of 85% at 3 years and 60% at 8 years follow-up. Newer lead models are as much affected as older models. That was recently evidenced with the Sprint Fidelis alert.32 Implantable cardioverter–defibrillator lead failure may have serious clinical consequences, including life-threatening complications. The most frequent are inappropriate shocks (33% as reported by Kleeman et al.31) that may severely impair the patient quality of life and compromise psychological equilibrium.

Risk-benefit ratio becomes a major issue when considering low-risk patients, typically asymptomatic or mildly symptomatic HF patients. Currently, solid data on risk-benefit ratio of CRT-D in NYHA class I-II patients are lacking.

The real world and the individual patient

Randomized trials are often criticized for their enrolment of highly selected patients, unlike those encountered in real life. This criticism is particularly applicable in the treatment of HF. Considering the large randomized studies of drugs or devices in HF conducted in the last years, the mean age of the populations was relatively young, between 60 and 67 years, and the proportion of women was small, between 20 and 32%. The most striking example is the SCD-HeFT trial, with a population with a mean age of 60.1 years and 23% women included.2 These demographic characteristics are largely different from those of registries in which the mean ages are more than 70 years and the proportion of women approaches 50%.33 It is therefore problematic to apply recommendations issued from randomized trials to the general population of patients with HF. Common sense dictates that these recommendations would have to be applied to similar or identical patient populations as the studies used to support the clinical indication. In the case of the SCD-HeFT trial example, the treatment should be logically offered to ‘young’ or relatively young patients without serious co-morbidity. However, the majority of patients with advanced HF are older and have various concomitant disorders.34

It is not morally wrong to address the issue of priority of therapeutic objectives in this majority of patients with HF, whose longevity remains limited despite all therapeutic efforts. Improving quality of life, lowering the rates of hospitalizations for management of HF, and preserving patient autonomy are probably priorities for elderly patients. A treatment that moreover lowers mortality and prolongs life under comfortable conditions, which the CRT-P achieves at a modest cost, may be viewed as successful.


Both CRT-P and CRT-D are electrical treatment modalities that have been validated for the management of HF. There is currently no strong scientific evidence indicating that a CRT-D must be offered to all candidates for CRT. Plain common sense should limit the prescription of these costly and complicated devices for patients in need of secondary prevention, or for the purpose of primary prevention in younger patients without major co-morbidity. NYHA class II patients who are younger and less diseased than NYHA class III-IV patients could be a target population. The preferential choice of CRT-P in older and more diseased patients can be seen as an acceptable one.


View Abstract