We hypothesized that the response to cardiac resynchronization therapy with a defibrillator (CRT-D) in patients with mildly symptomatic heart failure (HF) is more favorable than the commonly referenced figure of 70%. This study involves the Multicenter Automatic Defibrillator Implantation Trial with Cardiac Resynchronization Therapy (MADIT-CRT) study population in which paired echocardiograms from baseline and 1-year follow-up were available in 621 implantable cardioverter-defibrillator–treated patients and 749 patients treated with CRT-D. We prespecified CRT-D responders as the patients who at 1-year follow-up had a reduction in left ventricular end-systolic volume (LVESV) that corresponded to the top (best) quintile of LVESV reduction in the implantable cardioverter-defibrillator–treated patients, that is, a ≥17% reduction in LVESV. Using this metric, 88% of patients treated with CRT-D and 91% of the patients treated with CRT-D with left bundle branch block (LBBB) were identified as cardiac resynchronization therapy responders. Landmark multivariate Cox model analyses revealed a significant interaction (p = 0.038) involving LVESV (responders vs nonresponders) and LBBB (present vs not present) in risk reduction for HF or death. The interaction finding indicates that cardiac resynchronization therapy responders with LBBB have a significantly lower risk for HF or death (hazard ratio [HR] 0.24) than patients without LBBB (HR 0.62). In the patients treated with CRT-D, LVESV response was associated with reduction in the risk of death (HR 0.20, p <0.001). An increasing percent reduction in LVESV was associated with progressively lower rates of HF or death, a finding consistent with a dose-response relation. In conclusion, approximately 90% of CRT-D–treated patients in MADIT-CRT had a significant and meaningful reduction in LVESV, and these LVESV responders had reduced rates of cardiac events during long-term follow-up.
Several randomized clinical trials have shown that cardiac resynchronization therapy (CRT) is associated with reduction in the risk of death and heart failure (HF) in patients with New York Heart Association (NYHA) class III or IV, reduced ejection fraction, and a wide QRS complex. More recently, the Multicenter Automatic Defibrillator Implantation Trial with Cardiac Resynchronization Therapy (MADIT-CRT), the Resynchronization Reverses Remodeling in Systolic Left Ventricular Dysfunction (REVERSE) trial, and the Resynchronization-Defibrillation for Ambulatory Heart Failure trial have demonstrated that the benefits of CRT also apply to patients with mild HF symptoms (NYHA class I or II), with particularly good results in patients with left bundle branch block (LBBB) and those with wider QRS complexes. It is generally estimated that about 70% of patients treated with cardiac resynchronization benefit from this therapy. However, this percent benefit has not yet been thoroughly investigated in patients with mild HF using a control group. The present study uses the implantable cardioverter-defibrillator (ICD)–only patients in the MADIT-CRT trial as a comparison group to define response versus nonresponse to CRT based on a prespecified reduction in left ventricular end-systolic volume (LVESV) between baseline and the 1-year follow-up echocardiogram. We hypothesized that the response to cardiac resynchronization therapy with a defibrillator (CRT-D) in patients without advanced HF is more favorable than the commonly stated 70% figure, especially in patients with LBBB, and that the greater the reduction in LVESV, the greater the subsequent clinical benefit from CRT.
Methods
The MADIT-CRT trial included 1,820 patients with ischemic or nonischemic cardiomyopathy, QRS ≥130 ms, and left ventricular ejection fraction (LVEF) ≤30% who were in NYHA class I or II. Patients were randomized to CRT-D or ICD implantation in a 3:2 ratio. The primary aim of the MADIT-CRT trial was to investigate whether the implantation of a CRT-D would reduce the risk of HF or death, whichever occurred first, compared with implantation of an ICD device. Patients were excluded from enrollment for a variety of previously reported reasons. The study enrolled patients from 110 hospital centers in Europe and North America. Previous publications have presented the design and primary results of the MADIT-CRT study. The present study included 1,373 patients (75% of the total MADIT-CRT cohort) for whom paired baseline and 1-year echocardiographic measurements were available as previously published.
From December 22, 2004 through June 24, 2009, a total of 1,820 patients were enrolled. After device implantation, patients were seen at scheduled visits at 1 month and then every 3 months afterward until the termination of the trial. Extended follow-up was conducted until September 10, 2010. The mean follow-up of the enrolled patients was 29.4 months.
Echocardiographic evaluations were performed according to study-specific protocol at baseline before device implantation (n = 1,809) and at 1-year follow-up. Paired echocardiograms from baseline and 1-year follow-up were available in 621 ICD-treated patients and 749 patients treated with CRT-D. The echocardiograms for this study were read at the echocardiographic core laboratory at Brigham and Women’s Hospital based on the protocols established by the American Society of Echocardiography. The left ventricular and atrial volumes were measured by the Simpson method in the apical 4- and 2-chamber views and averaged, and the LVEFs were calculated according to standard methods. The coefficients of variation for end-diastolic volume, end-systolic volume, and LVEF were 5.2%, 6.2%, and 5.5%, respectively, as reported previously.
To differentiate between LVESV responders and nonresponders, we prespecified CRT-D responders as the patients who at 1-year follow-up had a reduction in LVESV that corresponded to the top (best) quintile of LVESV reduction in the ICD-treated patients. The reduction was measured as percent change from baseline to 1-year follow-up and the cut point of the top quintile was a 17% reduction in LVESV. This metric was then applied to the CRT-D group and the CRT-D LBBB subgroup to determine the percentage of patients who were responders versus nonresponders. The end points during follow-up after the 1-year echocardiogram included HF or death, whichever occurred first, and all-cause mortality.
Baseline characteristics were compared in LVESV responders versus nonresponders in the CRT-D group. We used the Kruskal-Wallis test for continuous variables and the chi-square test or Fisher’s exact test for dichotomous variables, as appropriate. Continuous variables are expressed as mean ± SD and categorical data are summarized as crude numbers and percentages. The baseline clinical factors associated with patients who were categorized as LVESV responders, that is, ≥17% reduction in LVESV, were determined by a standard logistic regression analysis reporting the parameter estimates and 2-sided p values.
The cumulative probability of the end points of HF and death after the 1-year echocardiogram (landmark analysis) for responders and nonresponders was displayed by the Kaplan-Meier method, with comparisons of cumulative event rates by the log-rank test. A multivariate Cox proportional hazards regression model was used in a landmark analysis to identify the risk associated with LVESV responders and nonresponders on the end points of HF and death. The Cox model was adjusted for relevant clinical covariates using the best subset analysis approach, setting the limit for entry into the model at p <0.05. Exploration was carried out for risk factor interactions. HRs for the risk of HF or death and for death with their 95% confidence intervals and 2-sided p values are reported. A 2-tailed p value <0.05 was considered statistically significant. Analyses were performed using SAS statistical system, 9.3 version (SAS Institute, Cary, North Carolina).
Results
The frequency distribution of the LVESV percent change from baseline to 1 year in the ICD-treated (n = 621) patients is presented in Figure 1 . We prespecified a favorable improvement in LVESV (responders) at the top quintile of the percent change in the LVESV distribution of the ICD-treated patients, and this metric was a 17% reduction in LVESV. That is, 20% of the ICD-treated patients had spontaneous reductions of ≥17% in LVESV. Using this metric as the definition for responders versus nonresponders, 88% (660 of 752) of the patients treated with CRT-D were identified as LVESV responders ( Figure 1 ) and 91% (486 of 534) of the LBBB subgroup of patients treated with CRT-D were identified as responders ( Figure 1 ).
The baseline clinical characteristics of patients treated with CRT-D who were classified as responders and nonresponders are listed in Table 1 . Several significant differences in the baseline characteristics between the responders and nonresponders are evident.
| Characteristic | Responders (n = 660) ∗ | Nonresponders (n = 92) ∗ |
|---|---|---|
| Age (yrs) | 64.6 ± 10.4 | 62.7 ± 12.1 |
| Men | 489 (74) | 78 (85) |
| White | 603 (92) | 82 (89) |
| African-American | 44 (7) | 8 (9) |
| Coronary heart disease NYHA class I † | 91 (14) | 18 (20) |
| Coronary heart disease NYHA class II † | 257 (39) | 46 (50) |
| Noncoronary heart disease NYHA class II † | 312 (47) | 28 (30) † |
| NYHA class III or IV >3 months before enrollment | 56 (9) | 10 (11) |
| Treatment for hypertension | 422 (64) | 54 (59) |
| Diabetes mellitus | 193 (29) | 24 (26) |
| Myocardial infarction | 263 (41) | 56 (61) † |
| Cigarette smokers | 71 (11) | 18 (20) |
| Body mass index (kg/m 2 ) | 28.6 ± 5.3 | 27.9 ± 4.8 |
| Heart rate (beats/min) | 67.0 ± 10.8 | 66.1 ± 10.0 |
| Systolic blood pressure (mm Hg) | 124 ± 17 | 120 ± 17 |
| Diastolic blood pressure (mm Hg) | 72 ± 10 | 72 ± 10 |
| Blood urea nitrogen >25 mg/dl | 167 (25) | 25 (27) |
| Creatinine ≥1.4 mg/dl | 146 (22) | 22 (24) |
| LBBB | 486 (74) | 48 (52) † |
| Non-LBBB | 174 (26) | 44 (48) † |
| QRS duration | 159.6 ± 9.6 | 151.0 ± 17.1 † |
| Echocardiogram/Doppler (baseline) | ||
| Left ventricular end-diastolic volume (ml) | 251 ± 54 | 245 ± 62 |
| LVESV (ml) | 178 ± 44 | 172 ± 49 |
| LVEF | 29 ± 3 | 30.1 ± 4 |
| Medications at baseline | ||
| Amiodarone | 42 (6) | 8 (9) |
| Angiotensin-converting enzyme inhibitors | 509 (77) | 69 (75) |
| Angiotensin receptor blockers | 139 (21) | 19 (21) |
| β Blockers | 620 (94) | 86 (93) |
| Digitalis | 173 (26) | 27 (29) |
| Diuretics | 451 (68) | 67 (73) |
| Lipid-lowering statin drugs | 445 (67) | 65 (71) |
∗ LVESV responders were defined by the top twentieth percentile change (reduction) in LVESV between the baseline and 1-year echocardiogram in ICD-treated patients, that is, a ≥17% reduction in LVESV; LVESV nonresponders were patients with <17% reduction in LVESV.
† Indicates p value <0.01 between responders and nonresponders.
For instance, responders tended to share the following characteristics: female gender, nonischemic NYHA class II, LBBB, no previous myocardial infarction, not currently smoking, and QRS >150 ms.
The factors associated with a favorable LVESV response to CRT therapy in the overall CRT-D population are presented in Table 2 . The important factors associated with a favorable response to CRT therapy include LBBB, age ≥65 years, body mass index ≥30 kg/m 2 , QRS ≥150 ms, and patients without a history of myocardial infarction.
| Parameter | Odds Ratio | 95% CI | p Value |
|---|---|---|---|
| LBBB | 1.74 | 1.05–2.90 | 0.03 |
| Age ≥65 yrs | 2.11 | 1.32–3.39 | <0.01 |
| Body mass index ≥30 kg/m 2 | 1.96 | 1.17–3.29 | 0.01 |
| QRS duration ≥150 ms | 1.82 | 1.12–2.96 | 0.02 |
| No history of myocardial infarction | 1.92 | 1.16–3.13 | 0.01 |
Univariate Kaplan-Meier analysis showed that among the patients treated with CRT-D, LVESV responders had a significant reduction in the cumulative probability for HF or death, whichever occurred first ( Figure 2 ), and for death ( Figure 2 ) compared with nonresponders, with similar but more marked findings for the LBBB subgroup (data not shown).
The findings from the landmark multivariate Cox model with adjustment for relevant clinical covariates are presented in Table 3 . In the overall population of patients treated with CRT-D (n = 752), the risk for HF or death, whichever occurred first, was significantly reduced in LVESV responders in patients with LBBB, but only a favorable trend was seen in patients without LBBB ( Table 3 ). However, there was a significant interaction (p = 0.038) involving LVESV (responders vs nonresponders) and LBBB (present vs not present) in risk reduction for HF or death. The interaction finding indicates that CRT responders with LBBB have a significantly lower risk for HF or death (HR 0.24) than patients without LBBB (HR 0.62). In the patients with LBBB, LVESV responders were associated with reduction in the risk of death (HR 0.20, p <0.001), a favorable trend in those without LBBB (HR 0.46), but no significant interactions of LVESV with LBBB or other covariates.
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
