In early randomized trials of New York Heart Association (NYHA) Class III–IV patients (Multisite Stimulation in Cardiomyopathies, MUSTIC [13]; Multicenter InSync Randomized Clinical Evaluation, MIRACLE [14]; MIRACLE ICD [15]; Contak CD [16]); and NYHA Class II patients (MIRACLE ICD II [17]), CRT improved one or more measures of clinical function such as 6-minute walk (6 MW) distance, peak oxygen consumption (VO₂), quality of life (QOL) based on standardized questionnaires, and NYHA functional class.

In recent years, greater focus has been given to additional measures of “benefit,” especially those indicating a diminished cost of care. Thus, reduction in hospitalization is a measure of clinical response and reflects the economic benefit of therapy. Although a few early trials (MUSTIC [13], MIRACLE [14]) showed a decrease in hospitalization with CRT, others did not (MIRACLE ICD [15], Contak CD [16]). More recently, larger trials have employed a composite primary endpoint, using all-cause mortality plus a measure of HF or other cardiovascular morbidity (◘ Table 21.1).

In the Cardiac Resynchronization-Heart Failure (CARE-HF) trial [21], 813 patients (94 % NYHA Class III, 6 % NYHA Class IV) were followed for a mean of 29.4 months. Patients were randomized to standard pharmacologic therapy alone for treatment of HF or with drug therapy plus a CRT pacemaker (CRT-P). The addition of CRT therapy significantly decreased the primary endpoint of death from any cause or an “unplanned hospitalization for a major cardiovascular event” (39 % vs. 55 %, HR 0.63, 95 % CI 0.51–0.77, p < 0.001). Subgroup analysis showed that CRT reduced unplanned hospitalization for a major cardiovascular event (p < 0.001) and unplanned hospitalization for worsening HF (p < 0.001).

The Comparison of Medical Therapy, Pacing, and Defibrillation in Heart Failure (COMPANION) trial [25] assigned 1520 patients in a 1:2:2 randomization pattern to optimal pharmacological HF therapy (OPT) alone, OPT plus CRT-P, or OPT plus a CRT defibrillator (CRT-D). Inclusion criteria included left ventricular ejection fraction (LVEF) ≤ 35 %, NYHA Class III or IV, and a QRS duration (QRSd) ≥120 ms. The main observation of COMPANION was that compared to OPT alone, CRT-P decreased the risk of the primary composite endpoint of time to death or hospitalization for any cause (HR 0.81, p = 0.014), as did CRT-D (HR 0.80, p = 0.01).

The findings in the studies noted above were subsequently extended to patients with less severe HF. The Multicenter Automatic Defibrillator Implantation Trial-Cardiac Resynchronization Therapy (MADIT-CRT) trial [31] followed 1820 patients (85 % NYHA Class II, 15 % NYHA Class I) for a mean of 2.4 years. Patients were randomized to a conventional ICD or a CRT-D device. CRT significantly decreased the combined endpoint of death from any cause or nonfatal HF event (17.2 % vs. 25.3 %, HR 0.66, 95 % CI 0.52–0.84, p = 0.001). The superiority of CRT was driven by a 41 % reduction in the risk of HF events (p < 0.001).

The Resynchronization-Defibrillation for Ambulatory Heart Failure Trial (RAFT ) [42] followed 1798 patients (80 % NYHA Class II, 20 % NYHA Class III) for a mean of 40 months. Patients were randomized to implantation of either a CRT-D or an ICD without CRT. CRT decreased the primary combined outcome of death from any cause or hospitalization for HF (33.2 % vs. 40.3 %, HR 0.75, 95 % CI 0.64–0.87, p < 0.001) and a secondary endpoint of hospitalization for HF (p < 0.001). In the Resynchronization Reverses Remodeling in Systolic Left Ventricular Dysfunction (REVERSE) trial [45], 610 patients (82 % NYHA Class II, 18 % NYHA Class I) were followed for 12 months. Patients received either a CRT-P or CRT-D and were randomized to having CRT turned on or off for a mean follow-up period of 40 ± 20 months. The primary endpoint was a composite HF score that judged patients to be improved, unchanged, or worsened. In this scheme, a statistically nonsignificantly higher percentage of patients worsened with CRT (p = 0.1). On the other hand, the time to first hospitalization for worsening HF was a prospective secondary endpoint and was significantly delayed by CRT (HR 0.47, p = 0.03).

Some trials have demonstrated CRT to be associated with improvement in LV systolic function and favorable electrical remodeling. The MIRACLE [14], Contak CD [16], MIRACLE ICD II [17], and MADIT-CRT [31] trials showed statistically significant improvement in LVEF with CRT. The REVERSE trial [45] was prospectively powered to use the left ventricular end-systolic volume index (LVESVI ) as a secondary endpoint that assessed LV remodeling. Compared with controls, CRT was associated with significantly greater reduction in LVESVI (−18.4 ± 29.5 ml/m² vs. −1.3 ± 23.4 ml/m², p < 0.0001) and significantly greater increase in LVEF at 12 months.

CRT and systems for implantable defibrillation have an independent beneficial impact on mortality in appropriately selected patients. Trials examining the effect of CRT on mortality may compare medical therapy with CRT-P or CRT-D with ICD. Some trials have compared medical therapy with CRT-P or CRT-D. The vast majority of trials examined mortality as part of a composite endpoint (typically combined with hospitalization, HF exacerbation, or hospitalization for HF) or as a secondary endpoint.

As mentioned previously, the COMPANION trial [25] examined patients (LVEF ≤ 35 %, NYHA Class III or IV, QRS duration ≥120 ms) randomized to OPT alone, OPT plus CRT-P, or OPT plus CRT-D. Death from any cause was a secondary endpoint with a mean follow-up duration of about 15 months. Compared with medical therapy (i.e., OPT), CRT-P showed a trend toward lower total mortality of 24 % (p = 0.059), while CRT-D showed a significant, 36 % reduction in total mortality (p = 0.003). The latter observation was consistent with contemporary trials that were showing a benefit to implantation of an ICD as primary prevention in patients with reduced LVEF and HF [30, 51].

The CARE-HF trial [21] was the first to show a mortality benefit that could be solely attributed to CRT. As a principal secondary endpoint, the study demonstrated a significantly lower rate of death from any cause in the medical therapy plus CRT-P group compared to medical therapy alone (20 % vs. 30 %, p < 0.002). All-cause mortality was used as the primary endpoint in an extended follow-up assessment (mean 37.4 months) of patients in the CARE-HF trial and showed that CRT significantly decreased mortality compared to medical therapy (HR 0.60, 95 % CI 0.47–0.77, p < 0.0001) [22].

The RAFT trial [42] examined whether adding CRT to ICD provides an additional benefit, given that the mortality benefit of primary prevention ICD in patients with systolic HF was already well established. Patients were randomized to receive ICD or CRT-D. Adding CRT reduced the death rate from any cause by 25 % (HR 0.75, 95 % CI 0.62–0.91, p = 0.003).

Several smaller studies, and at least one large study (MADIT-CRT [31]), have not shown a significant decrease in total mortality. A recent meta-analysis of 25 trials (9082 patients), of which the COMPANION, CARE-HF, RAFT, and MADIT-CRT trials accounted for 59 % of the total population, concluded that there was a reduction in all-cause mortality with CRT for patients with Class II–IV congestive HF [58].

Most clinical trials of CRT used QRS duration ≥120 ms as an inclusion criterion, and initial guidelines used this measure in recommendations for patient selection for CRT. Since the degree of QRS prolongation should, to a large extent, reflect the severity of electrical dyssynchrony, it seems reasonable to expect that patients with the longest QRS duration would have the greatest potential for clinical benefit from CRT. Several clinical trials included subgroup analyses based on QRS duration. Many showed that the benefits of CRT were limited to those patients with QRS duration longer than approximately 150 ms. Two published meta-analyses using the same five studies (COMPANION, CARE-HF, REVERSE, MADIT-CRT, and RAFT) concluded that the benefit of CRT was demonstrated in patients with QRS duration ≥150 ms, but not in patients with QRS duration <150 ms [59, 60].

Although the COMPANION trial showed that CRT decreased the risk of the combined endpoint of death from any cause or first hospitalization for any cause in patients with QRS duration >120 ms (LVEF ≤ 35 %, NYHA Class III–IV), subgroup analysis showed the benefit was seen only in patients with the longest QRS duration and not in patients with QRS duration of 120–147 ms [25]. The REVERSE trial showed that CRT produced overall favorable structural remodeling at 1 year among patients with QRS duration ≥120 ms (LVEF ≤ 40 %, NYHA Class I–II). Subgroup analysis, however, showed no benefit in LVESVI for QRS duration <140 ms, but significant benefit for QRS duration of 140–160 ms, and greater benefit for QRS duration >160 ms [46].

In the MADIT-CRT trial, CRT significantly decreased the primary endpoint of death from any cause or a nonfatal heart failure event in patients with QRS duration ≥130 ms (LVEF ≤ 30 %, NYHA Class I or II). In a prespecified subgroup analysis, however, the benefit was seen in patients with QRS duration ≥150 ms (about 65 % of the enrolled patients), but not in patients with QRS duration < 150 ms [31]. In the RAFT trial, CRT significantly decreased the primary outcome of death from any cause or hospitalization for HF in patients with QRS duration ≥120 ms (LVEF ≤ 30 %, NYHA Class II or III), but a prespecified subgroup analysis showed the benefit to be limited to patients with QRS duration ≥150 ms [42].

Other trials have shown benefit among patients with QRS duration of 120–150 ms. In the extended follow-up of patients in the CARE-HF trial, a prespecified subgroup analysis showed that CRT significantly decreased the primary endpoint of all-cause mortality whether the QRS duration was 120–159 ms or ≥160 ms [22]. In the MIRACLE trial, QRS duration was treated as a continuous variable and did not influence the significant improvements in the primary endpoints: 6 MW distance, QOL score, and NYHA functional class by CRT [14]. In the REVERSE trial [47], the favorable and significant effect of CRT on LVESVI showed a fairly linear relationship with baseline QRS duration that began at about 120 ms. In the same study, clinical response measured with a “clinical composite score” (including mortality, hospitalization for HF, whether the patient crossed over or discontinued double-blind treatment due to worsening HF, change in NYHA class, and patient global assessment) showed a similar relationship.

In view of these analyses, the European Society of Cardiology (ESC), American College of Cardiology Foundation (ACCF)/American Heart Association (AHA)/Heart Rhythm Society (HRS), and the Heart Failure Society of American (HFSA) revised their guidelines for CRT to limit a Class I or “recommended” indication for those patients with QRS duration >150 ms [11, 61, 62]. Indications remain for QRS duration of 120–150 ms, but at a lower level of recommendation (Class IIa or “may consider”) and/or a lower level of evidence base. QRS duration is, of course, a continuous variable and may be viewed as an indicator of the likelihood and/or magnitude of clinical response to CRT. Guideline revisions reflect clinical findings and the level of evidence supporting these findings. Of concern to caregivers, however, is that these interpretations may be used by third-party payers to limit coverage for procedures and, most importantly, may limit patients’ accessibility to therapies. A thoughtful, invited debate has been published on the application of QRS duration data to clinical practice [63, 64].

Some single-center studies have used various imaging techniques to identify mechanical dyssynchrony in the face of QRS duration < 120 ms. In spite of this, at least three multicenter studies do not support the use of CRT in such patients. The Cardiac Resynchronization Therapy in Patients with Heart Failure and Narrow QRS (RETHINQ) study examined the effect of CRT in 172 patients with LVEF ≤ 35 %, NYHA Class III HF, QRS duration ≤130 ms, and evidence on echocardiography of mechanical dyssynchrony. CRT did not improve peak VO₂ (the primary endpoint) or quality of life [44]. The Echocardiography Guided Cardiac Resynchronization Therapy (EchoCRT) study examined patients with LVEF ≤ 35 %, NYHA Class III or IV, and echocardiographic evidence of dyssynchrony. All patients received a CRT-D with CRT randomized “on” or “off.” The study was stopped prematurely for futility after 809 patients had been enrolled. CRT did not reduce the composite endpoint of death or HF hospitalization and may have increased mortality [27].

The Evaluation of Resynchronization Therapy for Heart Failure (LESSER-EARTH) trial examined patients with LVEF ≤ 35 %, QRS duration < 120 ms, and symptoms of HF as indicated by 6 MW distance <400 m due to HF symptoms. There was no prerequisite for LV dyssynchrony . The trial was stopped prematurely as CRT was associated with a significant decrease in the 6 MW distance, and there was a nonsignificant trend toward increased HF hospitalization [29]. It may be that the electrical synchrony provided by the Purkinje system that produces a narrow QRS complex is undermined by biventricular pacing that does not employ the Purkinje fibers. Thus, CRT may actually lengthen QRS duration.

In the LESSER-EARTH trial , CRT increased QRS duration at 6 months by 41.5 ms [29]. Thus, CRT produces a degree of electrical dyssynchrony in those patients with QRS duration < 120 ms. In essence, current technology cannot duplicate the rapid, “synchronized” ventricular activation that the intact conduction system provides, but nevertheless may remain advantageous if the conduction system is sufficiently impaired.

Under normal conditions, the last part of the LV to be activated is the posterior base [65]. LBBB prolongs LV activation while isolated right bundle branch block (RBBB), i.e., without fascicular delay in the LV, lengthens QRS duration without delaying LV activation. As such, patients with LV systolic dysfunction might be predicted to be more likely to respond to CRT if they have an underlying LBBB rather than RBBB. Data from CRT studies are consistent with this expectation, an outcome that is helpful since LBBB is more common than RBBB among patients with LV systolic dysfunction [33].

Subgroup analysis of MADIT-CRT [34] and REVERSE [47] studies showed that the favorable response to CRT was limited to patients with baseline LBBB. It was not seen in patients with RBBB or nonspecific intraventricular conduction delay (IVCD). Subgroup analysis of RAFT [42] showed a weak interaction between treatment and QRS morphology. In fact, patients with LBBB appeared to derive greater benefit than patients with nonspecific IVCD.

A meta-analysis of 5356 patients in COMPANION, CARE-HF, MADIT-CRT, and RAFT showed a highly significant reduction in adverse clinical events (mortality from any cause or hospitalization either for HF or, in some studies, other causes) among patients with LBBB. No benefit was seen among all patients with non-LBBB conduction abnormalities or when subdivided to RBBB and nonspecific IVCD [66]. A study of patients with baseline RBBB pooled from the MIRACLE and Contak CD found no benefit from CRT in any subjective or objective measure at 3 or 6 months except for NYHA class. However, control patients also showed a significant improvement in NYHA class at 6 months, consistent with a placebo effect [67].

CRT is contemplated for patients with AF in two major scenarios: (1) patients with HF and a hemodynamic indication for CRT who have concurrent AF and (2) AF patients with a rapid ventricular rate who are being considered for AV junction ablation and permanent pacing. Although AF is a common arrhythmia in HF patients [33], the vast majority of patients enrolled in CRT trials have been in sinus rhythm. The limited data addressing use of CRT in patients with HF and permanent AF have been largely disappointing.

The MUSTIC trial was an early, single-blind crossover trial examining the clinical efficacy of CRT using 6 MW distance as the primary endpoint [13]. Secondary endpoints were peak VO₂, QOL, hospitalizations, patients’ preferred study period, and mortality. The original study examined patients in sinus rhythm.

Using the same protocol, MUSTIC subsequently recruited the first and, to date, only trial of CRT in HF patients specifically with AF [38]. Patients had NYHA Class II HF, persistent AF (defined as >3 months), and a slow ventricular rate that necessitated ventricular pacing. Fifty-nine patients began the trial but, due to a high drop-out rate, only 37 completed both crossover phases. In the intention-to-treat analysis, none of the clinical endpoints were met. Nevertheless, blind questioning at the end of the study showed 84.6 % of patients preferred the period during which they received CRT (p < 0.001). On-therapy analysis showed CRT to improve 6 MW distance (p = 0.05) and peak VO₂ uptake (p = 0.04).

The RAFT trial [43] included 229 patients with permanent AF randomized to ICD or CRT-D, the largest population of AF patients in whom CRT has been evaluated to date. It found no difference in the primary endpoint of a composite of death or HF hospitalization. Likewise, there was no difference in cardiovascular death or 6 MW distance. It did show a trend favoring CRT with respect to fewer HF hospitalizations and a greater improvement in the Minnesota Living with Heart Failure score. A recent meta-analysis of 23 observational studies of response to CRT in AF patients concluded that CRT benefits are attenuated in AF patients, with a lower response rate compared to patients in sinus rhythm (RR 1.32, 95 % CI 1.22–1.55, p = 0.001) [68].

AF may pose a number of hurdles that contribute to limited CRT response (◘ Table 21.2). One of the key factors is the need to maintain a very high percentage of biventricular pacing, typically targeting 92 % or higher [69, 70]. Indeed, even though eligible patients in the RAFT trial were required to have bradycardia that necessitated pacing (resting heart rate ≤60 bpm and ≤90 bpm after a 6 MW test), two-thirds had <95 % and half had <90 % CRT pacing [43]. These numbers, in fact, probably overestimate the amount of effective biventricular pacing as some of the so-called paced beats will be fusion or pseudofusion beats.

Better results have been reported when CRT is combined with AV junction ablation for patients with AF. This strategy may ameliorate a component of tachycardia-mediated cardiomyopathy and maximizes biventricular pacing by eliminating conducted beats that produce fusion. The Post AV Nodal Ablation Evaluation (PAVE) study [41] enrolled patients with chronic AF undergoing AV node ablation for medically refractory rapid ventricular rates and randomized them to receive a single-chamber pacemaker with its lead in the RV apex (n = 81) or a CRT device consisting of pacing leads in the right ventricle and coronary sinus (n = 103). This was not a heart failure study, as the groups had an initial LVEF of 45 ± 15 % and 47 ± 16 %, respectively. CRT patients showed greater improvement in the primary endpoint of change in 6 MW distance over 6 months following implant (p = 0.04). There was no difference in the secondary endpoint of QOL assessment. The LVEF in the CRT group was significantly higher (by 5 % absolute) than the RV pacing group after 6 months (p < 0.05) due to an LVEF decrease in the RV pacing group that did not occur in the CRT group.

The Ablate and Pace in Atrial Fibrillation (APAF) trial [19] enrolled patients with permanent AF who either (1) underwent a previously planned AV junction ablation to control high ventricular rates or (2) had permanent AF, drug refractory HF, and depressed LV function and in whom a clinical decision had been made to perform AV junction ablation and implantation of a CRT device. All received a CRT device (CRT-P or CRT-D at the enrolling physicians’ discretion) but were randomized to RV or CRT pacing. After a mean 20-month follow-up, the primary composite endpoint of HF death, HF hospitalization, or worsening HF was significantly lower in the CRT groups (p = 0.005).

Total mortality was similar in both groups. The patients were analyzed in two subgroups: 25 % met contemporary United States [71] and European [72] guidelines for CRT (LVEF ≤ 35 %, QRS width ≥120 ms, and NYHA Class ≥III) and 75 % did not meet these guidelines. The primary endpoint was significantly lower (i.e., improved outcome) in both subgroups when treated with CRT rather than RV pacing.

The favorable response to CRT seen in these “ablate-and-pace trials” [19, 41] vs. HF trials that included AF patients reflects several important factors. First, AV junction ablation maximizes the percentage of time with biventricular pacing (limited, with normal pacemaker function, only by the quantity of ventricular ectopy). Second, as noted previously, even when AF patients were selected for bradycardia, the time spent with biventricular pacing was limited [43]. Third, AF patients receiving AV junction ablation in addition to CRT therapy may benefit by alleviating any component of tachycardia-mediated LV dysfunction. Finally, ablate-and-pace trials compare the electrical remodeling effects of RV and CRT pacing that may produce structural remodeling. That is, electrical dyssynchrony caused by RV pacing (or LBBB, or high-burden ventricular ectopy) may cause mechanical dyssynchrony.

By their nature, ablate-and-pace trials assess the extent to which CRT may prevent such electrical dyssynchrony. On the other hand, HF patients enrolled in CRT trials who also have AF may primarily have mechanical dyssynchrony (e.g., from ischemic heart disease, hypertensive heart disease, myocarditis) with secondary electrical dyssynchrony. Such patients may have a hemodynamic improvement with CRT, but they may be less likely to have favorable mechanical remodeling.

The potential detrimental effects of RV-only pacing have been described in pacemaker and ICD studies. Dual-chamber pacing modes are designed to maintain AV synchrony over a wide range of sinus rates so that, in the setting of prolonged AV conduction or AV block, the ventricle is paced at the sinus rate.

The Mode Selection Trial (MOST) showed that in patients with sinus node dysfunction, dual-chamber pacing, and normal LV function, RV pacing >40 % of the time was associated with a 2.6-fold increased risk of HF hospitalization compared to patients with less RV pacing [8]. The Dual Chamber and VVI Implantable Defibrillator (DAVID) trial showed that in ICD patients with LVEF < 40 % but without an indication for anti-bradycardia pacing, patients with RV pacing due to a dual-chamber ICD system are more likely to develop HF than patients with an ICD that minimizes RV pacing by only doing so when the ventricular rate is less than 40 beats per minute without attempting to maintain AV synchrony [26]. The Biventricular Versus Right Ventricular Pacing in Heart Failure Patients with Atrioventricular Block (BLOCK-HF) trial showed that in patients with LVEF < 50 % who require ventricular pacing due to AV block, CRT was associated with a significantly lower risk of death or HF than RV-only pacing [20].