Relationship of Electro-mechanical Remodeling to Survival Rates after Cardiac Resynchronization Therapy

Abstract

Cardiac resynchronization therapy, when added to optimal medical therapy, increases longevity in symptomatic congestive heart failure patients with left ventricular ejection fractions (LVEF) ≤0.35 and QRS durations >120 ms. Cardiac resynchronization therapy is also associated with electrical and mechanical reverse remodeling. We examined whether reverse remodeling predicts increased survival rates in non-trial settings.

Recipients of cardiac resynchronization therapy and defibrillators (n=112; 78 men; mean age, 69 ± 11 yr) underwent repeat echocardiography and electrocardiography at least 90 days after device implantation. Forty patients had mechanical responses of at least 0.05 improvement in absolute LVEF; 56 had electrical responses (any narrowing of biventricular-paced QRS duration compared with the electrocardiogram immediately after therapy). During a mean follow-up period of 3.1 ± 1.7 years, 55 patients died. The average death rate per 100 person-years was lower among mechanical responders than nonresponders (9.2% vs 23.9%; P=0.009); the unadjusted hazard ratio was 0.39 (95% confidence interval [CI], 0.19–0.79).

In a multivariate model adjusted for age, sex, baseline LVEF, and QRS duration, mechanical responders had 60% better survival than nonresponders (hazard ratio=0.40; 95% CI, 0.21–0.79; P=0.008). No difference in survival was observed in electrical response. In our association of absolute change in LVEF over the observed range with death (using restricted cubic splines), we observed a linear relationship with survival.

In patients given cardiac resynchronization therapy, mechanical but not electrical remodeling was associated with better survival rates, suggesting that mechanical remodeling underlies this therapy's mechanism of conferring a survival benefit.

In selected heart-failure patients, cardiac resynchronization therapy (CRT) improves exercise tolerance, maximal oxygen consumption, and quality of life, and reduces the risks of repeat hospitalization for heart failure or death.^1,2

Lower left ventricular ejection fraction (LVEF) is a predictor of cardiac events independently of QRS duration or electrical evidence of dyssynchrony.^3,4 Secondary data analyses from clinical trials yielded better clinical outcomes in the context of reverse mechanical remodeling.^5,6 In addition, electrical dyssynchrony—commonly observed in patients with left ventricular (LV) dysfunction⁷—is a predictor of LV systolic dysfunction.^8,9 Data from clinical practice are sparse in regard to associations of reverse mechanical and electrical remodeling with improved survival rates. In this study, we examined the association between electromechanical reverse remodeling and survival rates in a tertiary-care hospital.

Patients and Methods

This retrospective study was approved by the Institutional Review Board of the University of Pittsburgh. We included all patients who were implanted with a CRT device from 2002 through 2006 at the University of Pittsburgh Medical Center (a tertiary-care hospital) and who underwent electrocardiographic (ECG) and echocardiographic studies at baseline (upon implantation) and at follow-up at least 90 days later. Clinical and demographic data were obtained from each patient's electronic medical records. Follow-up data were ascertained from the medical records and the Social Security Death Index. Patients experiencing lead dislodgment or noncapture were excluded from the analysis.

Table I shows the baseline characteristics of the 112 patients, 78 of whom were white men with ischemic cardiomyopathy and severe LV dysfunction despite optimal medical therapy. The mean follow-up duration for this cohort was 407 ± 290 days (range, 92–1,439 d). Approximately 75% of patients were taking β-adrenergic-blocking agents, and 65% were taking angiotensin-converting enzyme inhibitors or angiotensin receptor blockers.

TABLE I. Characteristics of 112 Patients by Survival Status at the End of Follow-Up

Device Implantation. All CRT devices were implanted at the University of Pittsburgh Medical Center by staff electrophysiologists. Right atrial and right ventricular leads were placed in standard positions at the right atrial appendage and the right ventricular apex. Lateral and posterolateral coronary venous branches were preferentially targeted for placement of the LV lead, with alternative locations in the event of high pacing thresholds, diaphragmatic stimulation, or lack of suitable venous branches. No device reprogramming or optimization was performed in any patient during the study period.

Echocardiography. Echocardiography was performed as clinically indicated, in standard parasternal, apical, and subcostal transthoracic views. These data were abstracted for study upon CRT device implantation and at least 90 days later. A core group of university cardiologists interpreted the studies.¹⁰ Measurements of left atrial dimension were made in parasternal long-axis view at end-diastole. Left ventricular end-diastolic diameter was measured in parasternal long-axis view just before mitral valve closure, and LV end-systolic diameter was measured just before mitral valve opening. The LVEF was visually obtained. Mitral regurgitation severity was graded chiefly with use of color-Doppler mapping. Patients who had an absolute increase of 0.05 or more in LVEF after CRT were considered to be mechanical responders. The mean time between baseline and repeat echocardiography was 456 ± 266 days.

Electrocardiography. A 12-lead ECG was recorded in all patients after same-day CRT implantation and upon follow-up. All ECGs were recorded with use of the GE Marquette Mac® 5000 Resting ECG Analysis System (GE Medical Systems; Chicago, Ill). The QRS width was determined from the computer reading and was verified manually with use of calipers at a paper speed of 25 mm/s. The median time between the baseline, day-of-implantation, and follow-up ECG recordings was 418 days (range, 92–1,439 d). All patients were monitored in the device clinic at our institution. No patient underwent device reprogramming of the atrioventricular or interventricular timing during follow-up. Patients with an absolute decrease in the paced QRS width of 1 ms or more after CRT were considered to be electrical responders. This definition of electrical response was based on a mean change in QRS duration of 0.71 ± 34 ms in our overall study cohort, and on the reality that the course of heart failure would result in an increase in this duration with time. In addition, for comparison and for consistency with published articles, we defined electrical response by using a cutpoint of decrease of more than 10 ms.

Statistical Analysis

We used mean ± SD and percentage to describe continuous and categorical variables, respectively. Paired t tests were performed to determine any difference in the means of a variable at baseline. Cox proportional-hazard regression models were used to examine the unadjusted and adjusted (for potential confounders) estimate of association between mechanical and electrical response. The proportional-hazard assumption was evaluated by means of graphical methods (ln-ln S [t] graphs) and statistical tests involving continuous time-interaction terms (Cox tests), and the assumption was met for the relationship between mechanical response and death. Multivariable adjusted cumulative mortality curves were plotted by mechanical response status with use of inverse-probability weights.¹¹ Finally, we used restricted cubic splines¹² to examine the association of absolute LVEF change and QRS duration with death over the observed range. We used SAS software version 9.1.3 (SAS Institute; Cary, NC) for all statistical analyses; P values less than 0.05 were considered statistically significant.

Results

Forty of the 112 patients (36%) had mechanical responses, defined as an absolute improvement in LVEF of at least 0.05 in a repeat echocardiogram after at least 90 days. Fifty-six patients (50%) had electrical responses, defined as any narrowing (by 1 ms or more) of the biventricular-paced QRS duration seen on the ECG immediately after CRT implementation.

During follow-up, 55 patients died (49%). The average death rate per 100 person-years was lower among mechanical responders than nonresponders (9.2% vs 23.9%; P=0.009) with an unadjusted hazard ratio of 0.39 (95% confidence interval [CI], 0.19–0.79) (Fig. 1).

Fig. 1 Average mortality rate (per 100 person-years) by A) mechanical and B) electrical response. Response was determined at least 6 months after the initial evaluation of 112 study participants. Participants were monitored for a mean period of 407 ± 290 days. Mechanical response was defined as an absolute increase in left ventricular ejection fraction of at least 0.05; electrical response was defined as any narrowing of the paced QRS duration.

In a multivariate model adjusted for age, sex, baseline LVEF, and QRS duration, mechanical responders had 60% better survival rates than nonresponders (hazard ratio=0.40, 95% CI, 0.21–0.79; P=0.008) (Fig. 2). No difference in survival rate was observed in terms of electrical response. Of note, the type of cardiomyopathy (ischemic vs nonischemic) did not affect the degree of mechanical remodeling (change in LVEF of 0.05 ± 0.13 in ischemic vs 0.05 ± 0.12 in nonischemic cardiomyopathy patients; P=0.79) or electrical remodeling (change in QRS duration of −0.75 ± 34 ms in ischemic vs 1.72 ± 33.7 ms in nonischemic patients; P=0.736).

Fig. 2 Kaplan-Meier curve shows survival by mechanical-response status.

Hazard ratios for death were plotted, with use of restricted cubic splines, for absolute change in LVEF and QRS intervals over 6 months or longer (Figs. 3 and 4). We observed a linear relationship between absolute LVEF change and death. As Figure 4 shows, there was an increase in overall deaths in patients who had widening of 20 ms or more in the paced QRS complex after CRT. At 6 months or later, an increase in QRS duration despite CRT trended toward a worse prognosis, but this relationship was not statistically significant. Further, when we used a 10-ms decrease in paced QRS duration as a cutpoint at follow-up, the mortality rate among electrical responders trended toward benefit but was statistically insignificant (hazard ratio=0.67; 95% CI, 0.39–1.16; P=0.16). In the overall cohort, electrical remodeling did not predict mortality rates.

Fig. 3 Hazard ratio (curve) with 95% confidence-interval bands (gray-shaded area) of death by absolute change in left ventricular ejection fraction (LVEF), 6 months or longer after follow-up. A restricted cubic spline with 3 knots was used to allow for possible nonlinear relationships. The horizontal dots represent the decile cutpoint for QRS duration (change in ms) for the 112 study participants, adjusted for age, sex, baseline LVEF, and QRS duration.

Fig. 4 Hazard ratio (curve) with 95% confidence-interval bands (gray-shaded area) of death by absolute change in paced QRS duration, 6 months or longer after follow-up. A restricted cubic spline with 3 knots was used to allow for possible nonlinear relationships. The horizontal dots represent the decile cutpoint for QRS duration (change in ms) for the 112 study participants, adjusted for age, sex, baseline left ventricular ejection fraction, and QRS duration.

Electrical and mechanical remodeling were negatively correlated in the overall cohort (r = −0.304, P=0.001); that is, an increase in LVEF was associated with a decrease in QRS width. However, this relationship was weak.

Discussion

We found that an absolute increase of 0.05 or more in LVEF, at 90 days or longer after the implantation of a CRT device, is associated with a survival benefit. A similar association could not be established in the case of shortening of the paced QRS interval.

Cardiac resynchronization therapy has been associated with a survival benefit,1,2,5,13–17 but little is known about the mechanisms that underlie this benefit. In our study, even a modest improvement in the mechanical-pump function of the LV was associated with a significant decrease in mortality rates, suggesting—albeit not proving—a cause-and-effect relationship. This relationship persisted after adjustment for important covariates known to affect survival in heart-failure patients.

Mechanical response to CRT has been shown to improve survival rates. In one study, a reduction of LV end-systolic volume (LVESV) of at least 10% was associated with a very low event rate (6.9% all-cause mortality rate) during a mean follow-up period of approximately 2 years.¹⁸ Another study showed that either an absolute increase in LVEF of more than 0.06 or a reduction in LVESV of at least 10% was associated with an event-free survival rate of 66% at a follow-up duration of 5 years.^19,20 Furthermore, super-responders to CRT—patients who had a 30% or greater reduction in LVESV after CRT—had much better long-term survival rates than did other CRT recipients who had less pronounced responses.²¹

In contrast, reports associating electrical reverse remodeling with better survival rates have been inconsistent. Some of these studies reported no improvement in clinical and echocardiographic outcomes on the basis of baseline QRS duration.²⁰ Others, such as the Comparison of Medical Therapy, Pacing, and Defibrillation in Congestive Heart Failure (COMPANION) trial, reported improvement in the primary endpoint if the baseline QRS duration was wider than 168 ms.³ Similar results were noted in the Pacing Therapies for Congestive Heart Failure (PATH-CHF) II trial, in which patients with baseline QRS durations wider than 150 ms experienced significant improvement in quality of life and exercise tolerance.²²

In our review of the literature, only one study²³ examined the relationship of reverse electrical remodeling to survival rates. In that study, reverse electrical remodeling after CRT was associated with a 4-fold decrease in the risk of death or in the risk of sustained ventricular tachycardia that necessitated appropriate implantable cardioverter-defibrillator therapies. Our study—with more patients—did not show that association. In the other study,²³ notably unlike ours, reverse electrical remodeling was defined as a decrease in native QRS duration of 10 ms or more.

The mechanisms associating reverse electrical remodeling with improved survival are not fully elucidated. Improved cardiac hemodynamics after CRT might result in a less passive stretch of the myocardium, possibly leading to fewer arrhythmic events and a lower incidence of pump failure. In fact, changes in the trafficking, synthesis, and degradation of myocyte gap junctions have been affected by mechanical stretch through changes in cell-to-cell coupling and in the composition of the extracellular matrix.^24–26 In addition, the risk of deadly arrhythmias decreases upon improved cell-to-cell coupling.²⁷ In the InSync III Marquis study,²⁸ LVESV reduction by 15% or more was indeed associated with a lower incidence of premature ventricular contractions and episodes of ventricular tachycardia and fibrillation, in comparison with no LV reverse remodeling. Similarly, the InSync ICD Italian Registry²⁹ showed that LVESV reduction by at least 10% led to a decreased incidence of ventricular arrhythmias. Whether these mechanisms underlie the survival benefits of CRT in the presence of a favorable mechanical response seems likely but is unproven.

Strengths and Limitations of the Study

Other studies have evaluated the importance of reverse remodeling after CRT on the basis of clinical endpoints, such as changes in functional class of heart failure, distance walked in 6 minutes, or quality-of-life score. Conversely, our study correlates reverse remodeling with the harder endpoint of total, all-cause death, which is a strength of our study. However, our study has some limitations. First, it is retrospective and observational, so its results might be affected by biases that could not be accounted for and might not be readily applicable to other patient populations (even though observational studies arguably parallel real-world clinical practice). In addition, although we collected no data about causes of death, these are most likely to be of a cardiovascular nature in CRT recipients. Third, although lateral and posterolateral LV locations were preferentially targeted for LV lead placement, we did not analyze the actual position of the LV leads in our cohort. Fourth, we visually estimated rather than measured LVEF. Although estimates of LVEF can be subjective, they constitute the methods used in everyday practice that drive clinical decisions, such as those pertaining to device implantation. However, it is worth noting that the baseline and follow-up echocardiograms were analyzed by the same method, thus minimizing biases in the interpretation of the studies. Last, all the analyzed ECGs were ventricular-paced. Therefore, we did not analyze the morphology of the intrinsic QRS complex, which could influence the response to CRT.