Early Clinical Benefit After Cardiac Resynchronization Therapy: Fortunately, Qrs Width and Ejection Fraction are Still the Best Predictors

This article refers to ‘Predictors of short-term clinical response to cardiac resynchronization therapy’ by C. Linde et al., published in this issue on pages 1056–1063. It's always comforting when results make sense and are compatible with pathophysiology and mechanistic understanding, and consistent with current clinical practice, especially when the information you need to make a wise clinical decision is routinely available. The paper ‘Predictors of short-term clinical response to cardiac resynchronization therapy’, by Linde et al.,1 contributes further to our ability to improve the identification of patients likely to have a favourable clinical response following the implantation of a cardiac resynchronization therapy (CRT) device. In symptomatic patients with systolic dysfunction and a prolonged QRS duration, randomized controlled trials (RCTs) of CRT have uniformly demonstrated improved long-term outcomes, defined as freedom from mortality or heart failure (HF) hospitalization. The most recent HF guidelines reflect this evidence.2 Meta-analyses consistently indicate that the greatest long-term outcome benefit is seen in patients with a left bundle branch morphology, wider QRS complex, and poorer ejection fractions (EFs).3 The paper by Linde et al.1 presents convincing results based on a substantial dataset sourced from three large RCTs of CRT devices from a single manufacturer. This permitted the entry of individual patient data into a meta-analysis performed with robust methodology. The paper reports the findings of an analysis of the clinical composite score (CCS) in 1591 patients at 6 months post-randomization and describes which patient characteristics predict short-term clinical response. The authors found that longer QRS duration and lower left ventricular EF (LVEF) independently predict short-term clinical response.1 The differences between the endpoint evaluated in this analysis1 and the extensive data reporting long-term benefit based on survival and hospitalization in published RCTs and meta-analyses should be emphasized. The CCS incorporates symptomatic improvement using both objective and subjective measures, and, importantly, requires that data for all patients contribute to the endpoint.4 It classifies patients in a hierarchical model as worsened, unchanged or improved based on mortality, HF hospitalization, crossover from assigned randomization, New York Heart Association (NYHA) class and the patient's own global assessment of his or her HF status. The addition of an important clinically meaningful endpoint at 6 months is novel and the paper by Linde et al.1 represents a welcome addition to the literature that enriches our understanding of the issue and our ability to select the patients most likely to respond favourably to CRT. To place these findings in context, the reader should pay attention to the nature of this cohort. The dataset is taken from three RCTS: MIRACLE (2002),5 MIRACLE-ICD (2004)6 and REVERSE (2008).7 Since these trials were conducted, practice has changed considerably with reference to both pharmacological and device-based therapy. The selection of patients has changed substantially in accordance with evolving guideline recommendations. Pre-implant imaging, implantation techniques and programming procedures have improved and follow-up is better organized. In addition, as the authors1 point out, these studies5-7 differ in their inclusion criteria, as well as in their policies on the inclusion of subjects with and without implantable cardioverter-defibrillators (ICDs). Furthermore, this analysis was not designed to detect differences between device types [CRT pacemaker (CRT-P) vs. CRT defibrillator (CRT-D)]. These issues should be kept in mind when interpreting the results. As expected, Linde et al.1 found that the best independent predictors of short-term outcome were QRS width and LVEF. However, after adjustment, the aetiology of HF, a left bundle branch block (LBBB) morphology, NYHA status, age and gender did not independently predict short-term benefit. These demographics have been shown to strongly predict improved long-term outcomes. Is this somewhat unexpected finding attributable to the specific characteristics of the particular cohort selected for this meta-analysis or do the predictors of short- and long-term responses differ? Making comparisons between this analysis and previous reports and meta-analyses is not straightforward. These analyses used different endpoints and follow-up times. It should be noted that in a previous meta-analysis, which referred to these three RCTs5-7 as well as to CARE-HF8 and RAFT,9 an LBBB morphology was a strong predictor of long-term outcomes on univariate analysis, but not after adjustment for factors including the powerful effect of QRS duration. The strong association of LBBB morphology with wider QRS duration in this report1 demonstrates the importance of this link. The complex relationships among LBBB morphology, aetiology, LVEF, degree of electrical dyssynchrony and extent of co-morbidity may explain the lack of power in the use of QRS morphology to predict short-term outcomes when EF and QRS duration are included as covariates. Importantly, previous meta-analyses evaluated benefit based on categories of QRS duration, the highest of which referred to QRS of >150 ms, whereas the current report1 included 1138 patients with a QRS of >150 ms and permitted more granular analysis based on subcategories of patients with very wide QRS complexes. The figures clearly demonstrate the additional information obtained by evaluating QRS durations that extend to 200 ms. The particularly favourable response in the subgroup of patients with QRS duration of 160–180 ms (n = 568) suggests a more dyssynchronous substrate that should potentially benefit from resynchronization. Regarding the predictive power of a reduced LVEF, it should be noted that the higher the EF, the more likely is a non-CRT patient to have an improved or unchanged CCS at 6 months; therefore, it is less probable that CRT will increase this patient's likelihood of an improved score at 6 months. Nonetheless, in a comparison of patients assigned to CRT-on vs. CRT-off, odds ratios for improved CCS at 6 months increased by 3.7% for every 1% decrease in EF. The three-dimensional plots display a large and complex dataset in an easily understood format. For example, Figure 1(A) elegantly quantifies and describes the continually changing interaction between QRS duration, LVEF and the odds ratio for 6-month benefit and identifies the sweetest spots for prediction of short-term clinical benefit.1 The study included adequate numbers of patients in the QRS >180 ms (n = 152) and QRS >190 ms (n = 220) groups. The data suggest that the short-term benefit was attenuated in the patients with the widest QRS complexes. This is an interesting finding, which the authors speculate may be attributable to the association between a very wide QRS complex and the extent of left ventricular dysfunction possibly requiring long-term CRT to demonstrate clinical benefit.1 As Table 2 shows, there were no differences regarding aetiology or gender although the wider complexes were associated with an increasing prevalence of LBBB, as well as with reduced LVEF.1 Meta-analyses of long-term benefit with CRT have not demonstrated this attenuation of effect with a QRS of >180 ms. The authors deserve credit for placing focus on a clinically relevant issue that has not been addressed adequately. Are the results convincing considering the methodological limitations? It's always easier to accept findings when they are biologically plausible and are compatible with the literature, as well as with our clinical impressions. This is the case here. It makes sense that QRS width and LVEF should be the strongest predictors of both short- and long-term outcomes. Of course, we would like to see systematic meta-analyses including a broader range of RCTs and there are surely datasets that could provide information on short-term clinical outcomes. As always, the reader should consider the differences between RCT cohorts for the evaluation of CRT that are selected according to rigid inclusion and exclusion criteria and the potential CRT candidates we encounter in everyday practice. In clinical practice, decision making must include considerations that are not necessarily pertinent when selecting subjects for RCTs. Issues such as advanced age, renal dysfunction and co-morbidities frequently complicate treatment strategy in practice. Many patients who are not within the wide QRS and low EF ‘sweet spot’ will respond favourably. The extent of local resources available for CRT dictates the number of devices that can be implanted. The challenge is to offer CRT to those patients most likely to benefit. Linde et al.1 have provided data that further inform decision making. Fortunately, their data confirm what most of us do today. Ultimately, in clinical practice, the decision to implant a device, as well as the choice of device type (CRT-P or CRT-D), should also be based on informed patient expectations, risk stratification that accounts for any co-morbidity, physician experience including adequate follow-up routines, and cost-effectiveness.10 An experienced clinician should be able to consider all of these factors in synchrony. Conflict of interest: K.D. has received research support from five companies manufacturing cardiac resynchronization therapy devices (Medtronic, Boston Scientific, St Jude, Biotronik and Sorin).