Characterization of super-response to cardiac resynchronization therapy

Characterization of super-response to cardiac resynchronization therapy John Rickard, MD, Dharam J. Kumbhani, MD, SM, Zoran Popovic, MD, David Verhaert, MD, Mahesh Manne, MD, MPH, Daniel Sraow, MD, Bryan Baranowski, MD, David O. Martin, MD, MPH, Bruce D. Lindsay, MD, FHRS, Richard A. Grimm, DO, Bruce L. Wilkoff, MD, FHRS, Patrick Tchou, MD From the Heart and Vascular Institute, Cleveland Clinic, Cleveland, Ohio. BACKGROUND In patients with chronic systolic heart failure who undergo cardiac resynchronization therapy (CRT), improvements in left ventricular ejection fraction (LVEF) and reductions in left ventricular volume are generally modest. A minority of patients experience a dramatic response to CRT (super-responders), but the attributes associated with these patients have not been fully characterized. OBJECTIVE The purpose of this study was to identify baseline clinical attributes of super-responders and to assess the survival benefit associated with this response. METHODS We reviewed clinical, echocardiographic, and ECG data from a cohort of 233 patients undergoing new implantation of a CRT device between December 2001 and November 2006. All patients had a baseline LVEF ⱕ40% and New York Heart Association class II to IV symptoms on standard medical therapy. Patients whose absolute LVEF improved by ⱖ20% were termed super-responders. A multivariate model was constructed to determine factors predictive of superresponse, and an assessment of mortality was made.

more likely to be female and have a native left bundle branch block, lower preimplant brain natriuretic peptide and red cell distribution width levels, and smaller baseline left ventricular volumes with trends toward having more nonischemic cardiomyopathy and midventricular lead positions. In multivariate analysis, only left bundle branch block remained significantly associated with super-response. Super-responders had a considerably lower incidence of mortality compared to non–super-responders (9.4% vs 43.2%, P ⫽ .006) at mean follow-up of 5.5 ⫾ 1.2 years. CONCLUSION Baseline left bundle branch block is strongly associated with super-response to CRT. Super-responders derive better long-term outcomes with CRT than do non–super-responders. KEYWORDS Cardiac resynchronization therapy; Left bundle branch block; Resynchronization; Super-response ABBREVIATIONS CRT ⫽ cardiac resynchronization therapy; LBBB ⫽ left bundle branch block; LVEDV ⫽ left ventricular enddiastolic volume; LVEF ⫽ left ventricular ejection fraction; LVESV ⫽ left ventricular end-systolic volume

RESULTS In this cohort of 233 patients, 32 (13.7%) met criteria for super-response. In univariate analysis, super-responders were

(Heart Rhythm 2010;7:885– 889) © 2010 Heart Rhythm Society. All rights reserved.

Introduction

LVEF with CRT have been termed hyperresponders, whereas those who realize a dramatic improvement in LVEF but not necessarily normalization have been called “super-responders.” Limited data exist regarding the baseline characteristics associated with such a salutary response. The purpose of this study was to identify baseline clinical attributes of super-responders and to assess the survival benefit associated with this response.

In multiple large studies of patients with chronic systolic heart failure, cardiac resynchronization therapy (CRT) has been shown to improve quality of life and exercise capacity, induce reverse cardiac remodeling, reduce congestive heart failure hospitalizations, and improve overall survival.1–3 Multiple studies have demonstrated improvements in left ventricular ejection fraction (LVEF) following CRT often in the range from 3% to 5%.4 However, a few small studies recently have reported patients who realized more dramatic improvements in LVEF with CRT. Patients who normalize Dr. Martin has received support from Boston Scientific. Dr. Lindsay has received speaking fees from Medtronic and Boston Scientific. Dr. Wilkoff has received consulting and research support from Medtronic, St. Jude Medical, and Boston Scientific. Dr. Tchou has received research support from Boston Scientific and St. Jude Medical. Address reprint requests and correspondence: Dr. John Rickard, Cleveland Clinic, Department of Cardiovascular Medicine, 9500 Euclid Avenue, Cleveland, Ohio 44195. E-mail address: [email protected]. (Received February 9, 2010; accepted April 2, 2010.)

Methods This retrospective study involved the analysis of a cohort of 233 patients who underwent new implantation of a CRT device at the Cleveland Clinic, Cleveland, Ohio, between December 27, 2001 and November 3, 2006. The study was approved by the Institutional Review Board of the Cleveland Clinic for retrospective medical records review and performed according to institutional guidelines. From a total consecutive cohort of 1,179 patients, high-quality preimplant and postimplant ECGs and echocardiograms were available for 233 patients, who formed the study population.

1547-5271/$ -see front matter © 2010 Heart Rhythm Society. All rights reserved.

doi:10.1016/j.hrthm.2010.04.005

886

Heart Rhythm, Vol 7, No 7, July 2010

The follow-up echocardiogram had to have been obtained at least 2 months after CRT implantation. All patients had New York Heart Association (NYHA) class II to IV heart failure symptoms and baseline LVEF ⱕ40%. Clinical data were gathered via chart review. Super-response was defined as an absolute improvement in LVEF ⱖ20% between baseline and follow-up echocardiograms. Left bundle branch block (LBBB) was defined as QRS duration ⱖ120 ms, monophasic QS or rS complex in lead V1, and monophasic R wave in lead V6. Right bundle branch block was defined as QRS duration ⱖ120 ms, deep terminal S wave in leads I and V6, and RSR=, wide R, or qR pattern in lead V1. Nonspecific intraventricular conduction delay was defined as QRS ⱖ120 not meeting criteria for either right or left Table 1

bundle branch block. Echocardiograms were reanalyzed by two board-certified cardiologists blinded to the current study. Left ventricular end-diastolic volume (LVEDV), left ventricular end-systolic volume (LVESV), left ventricular end-diastolic diameter, and left ventricular end-systolic diameter were measured manually, and LVEFs were calculated from the volumetric data. Lead position was assessed by two cardiologists blinded to the current study via postCRT posteroanterior and lateral chest X-ray film using a standardized technique that has been described previously.5 CRT device implantations were performed transvenously in the vast majority of patients by electrophysiologists targeting a posterolateral vein for the left ventricular lead position. In instances when a transvenous lead could not be

Baseline patient characteristics Total (n ⫽ 233)

Super-responders (n ⫽ 32 [13.7%])

Non–super-responders QRS (n ⫽ 201 [86.3%]) P value

Age (years) 65.0 (57.3–73.3) 65.6 (59.6–74.8) 65.0 (57.0–73.1) Female gender 62 (26.6%) 16 (50.0%) 46 (22.9%) Caucasian race 195 (83.7%) 27 (84.4%) 168 (83.6%) Nonischemic cardiomyopathy 110 (47.2%) 20 (62.5%) 90 (44.8%) ICD* 218 (93.6%) 29 (90.6%) 189 (94.0%) Epicardial left ventricular lead 25 (10.7%) 5 (15.6%) 20 (10.0%) Diabetes mellitus 83 (35.6%) 6 (18.8%) 77 (38.3%) BMI† 27.4 (24.5–31.7) 26.7 (23.3–29.9) 27.5 (24.6–32.2) Chronic obstructive pulmonary disease 36 (15.5%) 4 (12.5%) 32 (15.9%) Hypertension 134 (57.5%) 20 (62.5%) 114 (56.7%) Hyperlipidemia 127 (54.5%) 20 (62.5%) 107 (53.2%) Serum creatinine (mg/dL) 1.1 (0.85–1.2) 1.1 (0.85–1.2) 1.2 (1.0–1.5) Serum hemoglobin (g/dL) 12.8 (11.9–14.3) 13.0 (11.9–14.6) 12.8 (11.9–14.3) Serum RDW‡ 14.6 (13.6–16.0) 14.2 (13.3–15.0) 14.7 (13.7–16.1) Serum WBC (k/␮L)§ 7.2 (5.6–8.5) 7.2 (6.1–8.2) 7.2 (5.6–8.7) Serum brain natriuretic peptide (pg/mL)¶ 376.0 (134.5–747.5) 159.5 (76.5–310.3) 402.0 (143.0–802.0) QRS duration pre-CRT (ms) 160 (142–182) 161 (149–172) 160 (141–184) QRS duration post-CRT (ms) 158 (144–170) 153 (138–163) 160 (144–172) Change in QRS (ms) ⫺6 (⫺22–16) ⫺12 (⫺22–1) ⫺4 (⫺23–18) Left bundle branch block 101 (43.3%) 22 (68.8%) 79 (39.3%) Right bundle branch block 12 (5.2%) 1 (3.1%) 11 (5.5%) Nonspecific intraventricular conduction delay 56 (24.0%) 2 (6.3%) 54 (26.9%) Paced rhythm 50 (21.5%) 6 (18.8%) 44 (21.9% Narrow (⬍120 ms) 14 (10.5%) 1 (3.1%) 13 (6.5%) History of AV nodal ablation 11 (4.7%) 1 (3.1%) 10 (5.0%) History of transient ischemic attack or cerebrovascular 27 (11.6%) 2 (6.3%) 25 (12.4%) accident Beta-blocker†† 169 (76.1%) 26 (81.3%) 143 (75.3%) Angiotensin-converting enzyme inhibitor or 178 (80.2%) 26 (81.3%) 152 (80.0%) angiotensin receptor blocker†† Aspirin†† 115 (51.8%) 16 (50.0%) 99 (52.1%) Diuretic†† 182 (82.0%) 22 (69.0%) 160 (84.2%) Nitrates†† 76 (34.2%) 6 (18.8%) 70 (36.8%) Hydralazine†† 31 (14.0%) 2 (6.3%) 29 (15.3%) Statin†† 107 (48.2%) 11 (34.3%) 96 (50.5%) Antiarrhythmic medication†† 58 (26.1%) 4 (12.9%) 54 (28.4%) Mean follow-up (years) 5.5 ⫾ 1.2 5.7 ⫾ 1.3 5.5 ⫾ 1.2 CRT ⫽ cardiac resynchronization therapy. *History of implantable cardioverter-defibrillator (ICD) placement. †Body mass index (BMI) prior to CRT implant. ‡Serum red cell distribution width (RDW) prior to CRT. §Serum white blood cell (WBC) count prior to CRT. ¶Serum brain natriuretic peptide prior to CRT. ††Medications available for 32/33 super-responders and 190/201 non–super-responders.

.57 .0013 .91 .06 .44 .36 .03 .21 .79 .54 .33 .19 .64 .04 .89 .004 .97 .06 .22 .002 1 .01 .82 .70 1 .55 .29 .61 .85 .07 .07 .27 .13 .07 .63

Rickard et al

Super-response to Cardiac Resynchronization Therapy

placed due to procedural difficulty, a minimally invasive epicardial lead was placed by a cardiothoracic surgeon. CRT devices were commonly programmed with an atrioventricular sensed delay of 100 ms and paced delay of 130 ms, with optimization performed according to the standard protocols of the Cleveland Clinic. Medications were recorded immediately prior to implantation of the CRT device, with titration of medications made at the discretion of the patient’s physician. Mortality was assessed using the U.S. Social Security Death Index. Patients who lacked a U.S. social security number were excluded from the analysis of mortality.

Statistical analysis For continuous variables, comparisons were made using Student’s t-test for parametric variables and the MannWhitney test for nonparametric variables. Fisher exact test was used for discrete variables. Continuous variables are presented as median with 25% to 75% interquartile range and dichotomous variables as absolute number with percentages. Multivariate logistic regression models were constructed with super-response as the dependent variable. Based on Hosmer-Lemeshow guidelines, only variables with P ⬍.2 in the univariate analysis were entered into a forward stepwise procedure.6 Odds ratios with corresponding 95% confidence intervals were estimated. The c-statistic, which is analogous to the area under the receiver operating characteristic curve, was calculated to estimate the discriminative value of the predictive models.7 KaplanMeier curves, with log-rank P values, compared long-term mortality between super-responders and non–super-responders.8 Cox proportional hazards models were constructed to identify independent predictors of long-term mortality, and hazard ratios with corresponding 95% confidence interval were estimated.9 The same variable selection techniques as described earlier were used to develop these models. Two-tailed P ⱕ.05 was considered significant. Analyses were performed using SAS version 9.1 (SAS Table 2

887

Institute, Cary, NC, USA) and Graphpad Prism version 5.0 (Graphpad Software Inc., La Jolla, CA, USA).

Results Of 233 patients, 32 (13.7%) met criteria for super-response. Preimplant echocardiograms were obtained a mean of 2.9 ⫾ 5.0 months prior to device implantation, and follow-up transthoracic echocardiograms were obtained a mean of 11.6 ⫾ 9.0 months following device implantation. Baseline characteristics of the study population are listed in Table 1. Super-responders were significantly more likely to be female, have a native LBBB, have lower preimplant brain natriuretic peptide and red cell distribution width levels, smaller LVEDV and LVESV (Table 2), and absence of diabetes mellitus. There were borderline significant trends toward a correlation with super-response and nonischemic cardiomyopathy (Tables 1 and 3), lower post-CRT paced QRS duration, midventricular lead position, and lack of lateral lead position (Table 4). A multivariate regression model was constructed adjusting for gender, native LBBB, baseline LVESV and LVEDV, lead position, and preimplant brain natriuretic peptide level. Only LBBB remained significantly associated with super-response (odds ratio ⫽ 5.72, 95% confidence interval 1.66 –19.71, P ⫽ .006) with a c-index of 0.86 for the final model. For the mortality analysis, two patients (both non–superresponders) lacked a U.S. Social Security number and were excluded. There were 3 (9.4%) deaths in the super-responder group and 86 (43.2%) in the non–super-responder group over a mean follow-up of 5.5 ⫾ 1.2 years. Figure 1 shows Kaplan-Meier curves comparing overall survival between super-responders and non–super-responders (log rank P ⫽ .006). After adjusting for baseline characteristics, including gender, type of cardiomyopathy, LVESV, LVEDV, and brain natriuretic peptide levels, super-responders had a trend toward a significant mortality benefit compared with

Echocardiographic statistics

LVEF pre (%) LVEF post (%) LVEF change (%) LVEDV pre (mL) LVEDV post (mL) LVEDV change (mL) LVESV pre (mL) LVESV post (mL) LVESV change (mL) LVEDD pre (cm) LVEDD post (cm) LVEDD change (cm) LVESD pre (cm) LVESD post (cm) LVESD change (cm)

Total (n ⫽ 233)

Super-responders (n ⫽ 32 [13.7%])

Non–super-responders (n ⫽ 201 [86.3%])

P value

23.0 29.0 5.0 234.0 215.0 ⫺9.0 178.0 157.0 ⫺15.0 6.5 6.2 ⫺0.2 5.7 5.5 ⫺0.2

21.0 52.0 29.0 196.0 125.5 ⫺66.5 163.0 66.0 ⫺85.0 6.3 5.3 ⫺0.9 5.5 4.0 ⫺1.6

23.0 26.0 3.0 238.0 229.0 ⫺5.0 181.0 165.0 ⫺9.3 6.5 6.4 ⫺0.1 5.7 5.6 ⫺0.1

.48 ⬍.0001 ⬍.0001 .0035 ⬍.0001 ⬍.0001 .02 ⬍.0001 ⬍.0001 .06 ⬍.0001 ⬍.0001 .16 ⬍.0001 ⬍.0001

(18.0–29.0) (21.0–39.0) (⫺1.0–11.5) (189.3–302.0) (165.0–295.0) (⫺39.5–11.0) (139.0–237.4) (99.9–214.0) (⫺56.5–8.3) (6.0–7.2) (5.6–7.3) (⫺0.7–0.1) (5.1–6.6) (4.5–6.4) (⫺0.9–0.1)

(18.0–27.8) (47.3–60.3) (23.3–34.8) (158.3–268.3) (108.3–170.0) (⫺105.3–⫺32.8) (116.0–207.2) (42.8–79.5) (⫺142.0–⫺61.2) (5.8–6.5) (5.0–5.6) (⫺1.2–⫺0.5) (5.1–5.9) (3.6–4.6) (⫺1.8–⫺1.2)

(18.0–30.0) (20.0–33.0) (⫺2.0–8.0) (193.0–310.5) (180.0–306.1) (⫺26.5– 12.0) (144.0–245.5) (123.5–233.5) (⫺34.5⫺11.0) (6.0–7.3) (5.9–7.4) (⫺0.5–0.2) (5.1–6.7) (5.0–6.5) (⫺0.6–0.2)

LVEDD ⫽ left ventricular end-diastolic diameter; LVEDV ⫽ left ventricular end-diastolic volume; LVEF ⫽ left ventricular ejection fraction; LVESD ⫽ left ventricular end-systolic diameter; LVESV ⫽ left ventricular end-systolic volume.

888 Table 3

Heart Rhythm, Vol 7, No 7, July 2010 Nonischemic cardiomyopathy subtypes

Idiopathic dilated Valvular Viral Chemotherapy induced Hypertensive Sarcoid Peripartum Genetic Alcohol induced

Total (n ⫽ 110)

Super-responders (n ⫽ 20 [18.2%])

Non–super-responders QRS (n ⫽ 90 [81.8%])

69 18 7 6 5 1 2 1 1

16 (80.0%) 3 (15.0%) 0 0 1 (5.0%) 0 0 0 0

53 15 7 6 4 1 2 1 1

(62.7%) (16.4%) (6.4%) (5.5%) (4.5%) (0.9%) (1.8%) (0.9%) (0.9%)

non–super-responders (hazard ratio ⫽ 0.25, 95% confidence interval 0.06 –1.03, P ⫽ .055).

Discussion The current study sought to identify baseline characteristics in a “real-world” cohort of heart failure patients who met criteria for super-response following CRT. Although multiple variables were associated with super-response in univariate analysis, only native LBBB was associated with such a response in a multivariate model. Super-responders were found to have better long-term outcomes than non– super-responders. Improvement in LVEF with CRT has been well documented in multiple randomized studies and has been correlated with improved survival.2,10 In a systematic review of many of the large trials involving CRT, a weighted mean improvement in LVEF of 3.0% was observed.4 It has been noted recently that some patients undergoing CRT realize a dramatic improvement in LVEF. However, the literature describing the clinical characteristics of these patients is limited to case reports and smaller case series. In addition, the longterm outcomes in patients realizing marked response to CRT compared to all others have been lacking. The current study contains one of the largest cohorts of patients experiencing “super-response” to CRT, defined as a dramatic improvement in LVEF not necessarily to normalization. Bulava et al11 were among the first to describe this phenomenon, reporting the case of a 72-year-old woman with ischemic cardiomyopathy and intermittent LBBB whose LVEF increased from 15% to normalization at 12Table 4

Anterior* Posterior* Lateral* Basal† Apical† Mid†

(58.9%) (16.7%) (7.8%) (6.7%) (4.4%) (1.1%) (2.2%0 (1.1%) (1.1%)

P value .12 1 .35 .59 1 1 1 1 1

month follow-up following CRT. Blanc et al12 reported an early case series of five patients with nonischemic cardiomyopathy whose LVEF normalized and experienced clinical improvement following CRT. Castellant et al13 were among the first to coin the term hyperresponder in two separate small cohorts, reporting that up to 20% of patients presenting for CRT may realize this response. In the first cohort, the investigators found no baseline variables predictive of hyperresponse. In the second cohort, nonischemic etiology was reported to have an association; however, this cohort had very few patients with ischemic cardiomyopathy.14 Both cohorts had very few patients; the larger contained only 11 hyperresponders. In our cohort we report an incidence of 13.7% based on our definition of super-response. Antonio et al15 were the first to use the term superresponder in a cohort of 87 patients undergoing CRT. In this study, super-response was defined as an improvement in New York Heart Association functional class, a twofold or greater improvement in LVEF or a final LVEF⬎45%, and a decrease in LVESV ⬎15%; 10 patients met these criteria.15 Duration of heart failure symptoms, left ventricular enddiastolic diameter, and mitral regurgitation were significantly associated with super-response in univariate analysis, with only duration of heart failure symptoms remaining significant in a multivariate model.15 These characteristics were not found to be associated with hyperresponse as reported by Castellant et al. The only large series investigating super-response was reported by van Bommel et al16 in a subanalysis from the PROSPECT trial.17 In this study,

Lead position Total

Superresponders

Non–superresponders

P value

8 140 52 10 52 141

2 22 3 0 4 23

6 118 49 10 48 118

.29 .18 .06 .36 .24 .07

(4.0%) (70.0%) (26.0%) (4.9%) (25.6%) (69.5%)

(6.3%) (81.5%) (11.1%) (14.8%) (85.2%)

(3.5%) (68.2%) (28.3%) (5.7%) (27.3%) (67.0%)

*Available in 27/32 super-responders and 173/201 non–superresponders. †Available in 27/32 super-responders and 176/201 non–super-responders.

Figure 1 Kaplan-Meier survival curves based on super-response. Log rank P ⫽ .0006.

Rickard et al

Super-response to Cardiac Resynchronization Therapy

super-response, which was defined as a reduction in LVESV ⱖ30%, was associated with female gender, nonischemic etiology, lack of history of ventricular tachycardia, wider baseline QRS duration, and greater baseline dyssynchrony.16 No multivariate analysis was performed to determine whether any of these factors were predictive of such response.16 In all of these studies, either baseline LBBB was an inclusion criterion or QRS morphology was not specified. Although LBBB has commonly been thought to be the result of an underlying cardiomyopathic process, it has been postulated that LBBB, in and of itself, may be solely or at least a contributor to left ventricular dysfunction in some patients.18 Grines et al18 reported on 18 patients with LBBB in the absence of other cardiovascular disease. Using radionuclide studies, they showed that LVEF in patients with LBBB was 54.7% ⫾ 7% compared to 62% ⫾ 5% in controls. One postulated mechanism is that LBBB leads to delayed left ventricular contraction producing both left and right ventricular asynchrony. This may then cause a decrease in left ventricular diastole leading to abnormal septal motion and, finally, to reduction in LVEF.18 This chain of events may be mitigated by CRT, thus explaining our results. Why LBBB results in more dramatic left ventricular dysfunction in some patients than others is unknown. Lead position is thought to be a major determinant of response to CRT.19 To our knowledge, this is the first study to evaluate the correlation between lead position and a dramatic response to CRT. Although there were trends toward superresponders having more midventricular and less laterally placed leads, lead position did not prove to be a significant predictor of super-response in multivariate analysis. This lack of predictive value may be due in part to the large majority of patients having their left ventricular lead implanted in the posterior, midventricular position. Female gender was also associated with super-response in univariate analysis similar to that reported by van Bommel et al.16 In the recently published MADIT-CRT study, females tended to have better outcomes than males.20 Whether the observation made in the current study and others is truly due to a gender difference in response to CRT or confounding with nonischemic cardiomyopathy, a factor commonly thought to have improved outcomes following CRT, remains in question.

Study limitations The retrospective nature cannot account for all confounders despite our best efforts to identify important baseline differences. Patients in our cohort come from a single tertiary care center and therefore may not be representative of patients presenting to other centers. There is a relatively wide disparity in the timing of follow-up echocardiograms. This reflects different follow-up practices among cardiologists and the wide referral area of the Cleveland Clinic. The number of super-responders is relatively small but represents one of the largest such cohorts yet reported. Given this

889

is a retrospective study, accurately identifying duration of heart failure symptoms and medication use prior to CRT placement was not possible.

Conclusion To our knowledge, the current study is the first to demonstrate that LBBB is a characteristic of super-responders to CRT. Not only was LBBB the only variable associated with super-response in multivariate analysis, but super-responders were six times more likely than non–super-responders to have LBBB at baseline. Moreover, the results of this study show that super-responders have a lower mortality rate than patients who do not meet these criteria.

References 1. Bristow MR, Saxon LA, Boehmer J, et al. Cardiac resynchronization therapy with or without an implantable defibrillator in advanced chronic heart failure. N Engl J Med 2004;350:2140 –2150. 2. Abraham WT, Fisher WG, Smith AL, et al. Cardiac-resynchronization in chronic heart failure. N Engl J Med 2002;346:1845–1853. 3. Cleland JG, Daubert JC, Erdmann E, et al. The effect of cardiac resynchronization on morbidity and mortality in heart failure. N Engl J Med 2005;352:1539 –1549. 4. McAlister FA, Ezekowitz J, Hooton N, et al. Cardiac resynchronization therapy for patients with left ventricular systolic dysfunction: a systematic review. JAMA 2007;297:2502–2514. 5. Wilton SB, Shibata MA, Sondergaard R, Cowan K, Semeniuk L, Exner DV. Relationship between left ventricular lead position using a simple radiographic classification scheme and long-term outcome with resynchronization therapy. J Interv Card Electrophysiol 2008;23:219 –227. 6. Hosmer DW. Model building. In: Hosmer DW, Lemeshow S, editors. Applied Logistic Regression. Second Edition. New York: Wiley, 2000:91–114. 7. Hanley JA, McNeil BJ. The meaning and use of the area under a receiver operating characteristic (ROC) curve. Radiology 1982;143:29 –36. 8. Kaplan EL, Meier P. Nonparametric estimation from incomplete observations. J Am Stat Assoc 1958;53:457– 481. 9. Cox DR: Regression models and life tables. J R Stat Soc 1972;B34:187–202. 10. Cha YM, Rea RF, Wang M, et al. Response to cardiac resynchronization therapy predicts survival in heart failure: A single-center experience. J Cardiovasc Electrophysiol 2007;18:1015–1019. 11. Bulava A, Lukl J, Skvarilova M, Marek D. Dramatically improved left ventricular function after biventricular pacemaker implantation—a case report. Eur J Heart Fail 2005;7:231–233. 12. Blanc JJ, Fatemi M, Bertault V, Baraket F, Etienne Y. Evaluation of left bundle branch block as a reversible cause of non-ischemic dilated cardiomyopathy with severe heart failure. A new concept of left ventricular dyssynchrony induced cardiomyopathy. Europace 2005;7:604 – 610. 13. Castellant P, Fatemi M, Orhan E, Etienne Y, Blanc JJ. Patients with non-ischaemic dilated cardiomyopathy and hyper-responders to cardiac resynchronization therapy: characteristics and long-term evolution. Europace 2009;11:350 –355. 14. Castellant P, Fatemi M, Bertault-Valls V, Etienne Y, Blanc JJ. Cardiac resynchronization therapy: “nonresponders” and “hyperresponders.” Heart Rhythm 2008;5:193–197. 15. Antonio N, Teixeira R, Coelho L, et al. Identification of ‘super-responders’ to cardiac resynchronization therapy: the importance of symptom duration and left ventricular geometry. Europace 2009;11:343–349. 16. van Bommel RJ, Bax JJ, Abraham WT, et al. Characteristics of heart failure patients associated with good and poor response to cardiac resynchronization therapy: a PROPSECT sub-analysis. Eur Heart J 2009;30:2470 –2477. 17. Chung ES, Leon AR, Tavazzi L et al. Results of the predictors of response (PROSPECT) trial. Circulation 2008;117:2567–2569 18. Grines CL, Bashore TM, Boudoulas H, Olson S, Shafer P, Wooley CF. Functional abnormalities in isolated left bundle branch block. Circulation 1989;79:845– 853. 19. Wilfied M, Grimm RA, Verga T, et al. Insights from a cardiac resynchronization optimization clinic as part of a heart failure disease management program. J Am Coll Cardiol 2009;53:765–773. 20. Moss AJ, Hall WJ, Cannom DS, et al. Cardiac resynchronization therapy for the prevention of heart-failure events. N Engl J Med 2009; 361:1329 –1338.