Author + information
- Received June 22, 2015
- Revision received September 4, 2015
- Accepted September 5, 2015
- Published online December 1, 2015.
- ∗United Heart & Vascular Clinic, Allina Health, St. Paul, Minnesota
- †Department of Biomedical Engineering, University of Minnesota, Minneapolis, Minnesota
- ↵∗Reprint requests and correspondence:
Dr. Alan J. Bank, United Heart & Vascular Clinic, 225 N. Smith Avenue, Suite #400, St. Paul, Minnesota 55102.
Objectives The aim of this study was to investigate the frequency and clinical implications of a delayed echocardiographic response to cardiac resynchronization therapy (CRT).
Background Long-term prognosis for CRT patients is routinely based on the assessment of echocardiograms after 6 to 12 months of therapy. Some patients, however, may require a longer period of therapy before echocardiographic improvements are detectable.
Methods This observational study included all patients with heart failure (HF) receiving a CRT device at a single center from 2003 to 2011. Eligible patients met current indications and had technically adequate echocardiograms from before implantation, approximately 1 year after implantation (mid-term), and ≥3 years after implantation (long-term). A positive echocardiographic response to CRT was defined as a reduction in left ventricular end-systolic volume ≥15%. All-cause mortality was compared for patients in 3 response groups: mid-term responders, long-term responders, and nonresponders.
Results During this study, 294 patients met the study criteria. Of the 120 patients who were nonresponders after 1 year, 52 (43%) experienced a delayed positive response. Delayed, long-term responders had mortality and hospitalization rates similar to mid-term responders and significantly lower than nonresponders.
Conclusions Among patients surviving at least 3 years after implantation of a CRT device and with echocardiographic follow-up, a significant portion of nonresponders after 1 year of CRT experience a delayed echocardiographic response after a longer period of time. Survival and hospitalization rates were similar for all echocardiographic responders, regardless of the time at which the response occurred.
Cardiac resynchronization therapy (CRT) is an established treatment for select patients with symptomatic heart failure (HF) with reduced ejection fraction (EF) and a prolonged QRS complex (1,2). In these patients, CRT alleviates symptoms, improves left ventricular (LV) size and systolic function, reduces hospitalizations, and improves survival. When the response to CRT is assessed according to echocardiography after 6 to 12 months of therapy, approximately 30% to 40% of patients do not seem to respond favorably, depending on the specific criteria used to define response (3). Patients who have increases in EF ≥5 units or decreases in LV volume ≥15% within this time frame reportedly also exhibit improved morbidity and mortality compared with patients whose response does not meet these thresholds (4–7).
A few studies have examined the time course of improvements in LV size and function after the initiation of CRT, and they found that most of the echocardiographic evidence regarding CRT response is present after 6 to 12 months of therapy (8–10). Other studies, however, have noted that a minority of patients exhibit a slower, delayed positive response (11,12). To the best of our knowledge, this subgroup of CRT patients has not been previously examined in detail. The prevalence of delayed CRT response is not well established; the impact of this response on clinical endpoints such as survival or hospitalizations has also not been well established.
The main purpose of the present study was to investigate mid-term and long-term changes in LV size and function in a large consecutive cohort of CRT patients to assess for delayed response to therapy. We were particularly interested in patients with a delayed therapeutic response. We also investigated the effects of a delayed response on clinical outcomes of mortality and hospitalization for HF.
All patients at the United Heart & Vascular Clinic (St. Paul, Minnesota) who received a new CRT device and met the study inclusion criteria between January 1, 2003, and December 31, 2011, were included in this study. Patients were required to have a QRS duration ≥120 ms, an EF ≤35%, symptoms classified as New York Heart Association functional class II through IV, and be receiving optimal medical therapy before implantation unless contraindicated. Patients receiving replacement CRT devices were excluded. In addition, only patients with technically adequate echocardiograms before receiving the CRT device and approximately 1 year after implantation, and who were alive 3 years after implantation, were identified. Of these patients, those with a technically adequate echocardiogram ≥3 years after implantation of their CRT device were included in the main study analysis. Institutional review board approval was obtained.
Echocardiographic Clinical Outcomes
Echocardiographic studies from the 3 time points were retrospectively analyzed (EchoPAC version 113; GE Medical Systems, Horten, Norway) by experienced readers from the United Heart & Vascular Clinic Echo Core Laboratory, blinded to the time point, baseline characteristics, and clinical outcomes of the subjects. Simpson’s biplane method was used to measure LV volumes from apical 4-chamber and 2-chamber views, and EF was calculated from these volumes. Echocardiographic response was defined as a decrease in end-systolic volume (ESV) ≥15% (13–15). Mid-term response was assessed by comparing the pre-implantation data and the 1-year follow-up data. Long-term response was determined by comparing the most recent echocardiogram with the preimplantation echocardiogram for each patient. Patients were divided into 3 echocardiographic response groups: 1) mid-term responders who showed a significant echocardiographic improvement at the 1-year follow-up; 2) long-term responders who showed no improvement at 1 year but who exhibited significant improvement at long-term follow-up; and 3) nonresponders who never displayed significant echocardiographic improvement.
Electronic medical records of 2 large health systems providing follow-up care for almost all study patients were searched for patient all-cause mortality, cardiac transplantation, LV assist device implantation, and HF hospitalization data. Online databases, including the Social Security Death Index and local newspaper obituary archives, were also searched for mortality data.
All continuous variables are expressed as mean ± SD and categorical variables are expressed as count (percentage), unless otherwise noted. Comparisons between the group of patients included in the study and those excluded because of missing echocardiograms were performed by using unpaired Student t tests or chi-square tests, as appropriate. Comparisons between the 3 CRT response groups were made by using 1-way analysis of variance. When a significant difference was detected among the groups, pair-wise Student t tests were performed with a Bonferroni correction to identify specific differences. All-cause mortality (including death, cardiac transplantation, or LV assist device implantation) was examined with a Kaplan-Meier analysis. To determine the association between echocardiographic response and mortality outcomes, a multivariate Cox proportional hazards model was performed controlling for the baseline characteristics of age, sex, presence of left bundle branch block (LBBB), etiology of HF (ischemic or nonischemic), angiotensin-converting enzyme inhibitor or angiotensin II receptor blocker usage, atrial fibrillation (AF), right ventricular dysfunction, serum creatinine, ESV, and EF.
Patients were considered to have significant coronary artery disease if they had a previous myocardial infarction, percutaneous coronary intervention, or coronary artery bypass grafting. AF was considered present if the patient had either paroxysmal or sustained AF. Right ventricular systolic dysfunction was qualitatively judged as present or absent. Beta-blocker, angiotensin-converting enzyme inhibitor, and angiotensin II receptor blocker medication dosages were converted to equivalent doses of carvedilol, lisinopril, and losartan, respectively. A significant increase in dosage was defined as a >25% increase from the dosage at the previous time point. Proportions of patients with either a new prescription for 1 of these drugs or a significant increase in dosage at either follow-up time were compared by using the Fisher exact test.
Differences between groups were reported by using hazard ratios (HRs) and 95% confidence intervals (CIs). Because no patients who died within 3 years of receiving CRT were enrolled in the study, the first 3 years of CRT were disregarded for mortality comparisons. However, HF hospitalizations occurring within the first 3 years of CRT were included in the analysis. STATA/SE software version 12.1 (StataCorp, College Station, Texas) was used for data analysis, and a 2-sided p value <0.05 was considered statistically significant. Because the primary outcome of the present study was the percentage of patients with a late positive response to CRT, and the remaining follow-up data were primarily exploratory, no further adjustments were made for multiple comparisons.
Figure 1 presents a flow diagram showing patient selection for this study. A first-time CRT device was implanted in 790 patients meeting standard indications during the study period. Of these, 185 (23%) died, and 117 (15%) were lost to follow-up before a 3-year follow-up visit was possible. An additional 48 (6%) patients were missing 1-year follow-up echocardiograms and were excluded. Of the remaining 441 patients, long-term echocardiograms were available for 294 (67%) patients, with a mean follow-up duration of 4.9 ± 1.4 years (range 3 to 9 years). The bottom portion of Figure 1 shows the classification of patients with long-term echocardiograms into groups of mid-term responders, long-term responders, and nonresponders. After approximately 1 year, 174 (59%) of these patients were classified as mid-term echocardiographic responders; of these, 149 (86%) maintained their echocardiographic improvements at long-term follow-up. However, of the 124 (42%) patients with long-term echocardiograms who were not classified as mid-term responders, 52 (43%) met the criteria for echocardiographic response at long-term follow-up. The remaining 68 (57%) patients were considered nonresponders. As a result, of the 298 patients surviving at least 3 years after CRT device implantation with echocardiograms at the 3 time points required in this study, 77% met the echocardiographic definition of response at ≥1 of the follow-up time points.
Table 1 compares the baseline demographic and echocardiographic characteristics of those patients with and without long-term echocardiograms. Compared with those with no long-term echocardiograms, patients included in the analysis were, on average, younger (mean age, 68 ± 11 years vs. 71 ± 12 years; p = 0.005) and less likely to have significant coronary artery disease (50% vs. 65%; p = 0.002). Changes in ESV (–27 ± 48 ml vs. –22 ± 42 ml; p = 0.329) and EF (6.6 ± 9.4% vs. 7.1 ± 10.1%; p = 0.610) after approximately 1 year of CRT were similar between patients with long-term follow-up and those without. Similarly, the proportion of echocardiographic responders was similar between those with and without long-term follow-up (59% vs. 54%, respectively; p = 0.276). However, patients with no long-term echocardiographic follow-up data were more likely to have died (adjusted HR: 3.7; p < 0.001) or been hospitalized for HF (adjusted HR: 1.82; p = 0.002) during the study.
Baseline demographic and echocardiographic analysis of the 3 patient groups is presented in Table 2. Compared with the mid-term responders, both long-term responders (43% vs. 60%; p = 0.036) and nonresponders (43% vs. 60%; p = 0.016) had a lower percentage with ischemic etiology. In addition, comparing long-term responders versus nonresponders, there was no significant difference in the proportion of patients with either a new prescription for a beta-blocker or a significant increase in dosage at mid-term (44% vs. 29%; p = 0.124) or long-term (17% vs. 13%; p = 0.999) follow-up. Similarly, there was no difference in the proportion of long-term responders and nonresponders who had a new prescription or significant dose increase for an angiotensin-converting enzyme inhibitor or an angiotensin II receptor blocker at either mid-term (21% vs. 19%; p = 0.821) or long-term (17% vs. 16%; p = 0.999) follow-up.
Long-Term Echocardiographic Response
Echocardiographic characteristics of each group at baseline, mid-term, and long-term follow-up are summarized in Figure 2. In the study group as a whole, both EF (Figure 2A) and ESV (Figure 2B) improved most from the pre-CRT baseline to the mid-term echocardiogram (ΔEF, 6.76 ± 9.4% [p < 0.001]; ΔESV, –26.9 ± 47.8 ml [p < 0.001], respectively), with relatively small improvements between the mid-term and long-term follow-ups (ΔEF, 1.4 ± 9.0% [p = 0.007]; ΔESV, –12.6 ± 38.0 ml [p < 0.001]). Mid-term responders followed this same trend, with rapid improvement at the mid-term echocardiogram (ΔEF, 10.6 ± 9.0% [p < 0.001]; ΔESV, –54.6 ± 39.1 ml [p < 0.001]) and slight improvement between the mid-term and long-term echocardiogram (ΔEF, 0.6 ± 9.6% [p = 0.434]; ΔESV, 5.3 ± 33.5 ml [p = 0.038]). Long-term responders exhibited only a slight improvement in EF and, by definition, no change in ESV at the mid-term echocardiogram (ΔEF, 2.1 ± 7.2% [p = 0.041]; ΔESV, 5.3 ± 22.7 ml [p = 0.101]) but large improvements between the mid-term and long-term echocardiogram (ΔEF, 5.4 ± 8.9% [p < 0.001]; ΔESV, –49.1 ± 41.9 ml [p < 0.001]). Finally, as expected, nonresponders exhibited no change or worsening EF or ESV at the mid-term echocardiogram (ΔEF, –0.1 ± 5.8% [p = 0.901]; ΔESV, 19.3 ± 26.4 ml [p < 0.001]) and no additional changes between mid-term and long-term echocardiograms (ΔEF, 0.5 ± 6.5% [p = 0.497]; ΔESV, –3.1 ± 28.2 ml [p = 0.376]).
Long-Term Clinical Response
Kaplan-Meier curves comparing all-cause mortality in the 3 study groups are presented in Figure 3A, and unadjusted and adjusted Cox proportional hazards model analyses are presented in Table 3. Beginning 3 years after CRT device implantation, 49 (17%) of the subjects died, at a rate of 6% per year. Mortality was similar in mid-term responders (22 deaths [13%], 5% per year) and long-term responders (8 deaths [15%], 5% per year; adjusted HR: 0.60; p = 0.312). Both of these groups had lower mortality than nonresponders (19 deaths [28%], 13% per year; adjusted HR vs. mid-term responders, 3.82 [p = 0.001]; adjusted HR vs. long-term responders, 6.40 [p = 0.002]).
Kaplan-Meier curves comparing the combined outcome of HF hospitalizations or death in the 3 study groups are presented in Figure 3B, and the results of unadjusted and adjusted Cox proportional hazards model analyses are presented in Table 3. After CRT device implantation, 95 (32%) of the subjects met the combined endpoint, at a rate of 6% per year. Survival free from HF hospitalization was similar in mid-term responders (42 events [24%], 5% per year) compared with long-term responders (19 events [37%], 7% per year; adjusted HR: 1.22 [p = 0.511]). Both of these groups had higher survival free from HF hospitalization than nonresponders (34 events [50%], 12% per year; adjusted HR vs. mid-term responders, 2.64 [p < 0.001]; adjusted HR vs. long-term responders, 2.24 [p = 0.023]).
The primary finding of the present study was that a significant portion of patients who did not exhibit an echocardiographic response to CRT at 1-year follow-up proceeded to demonstrate a positive response at longer term follow-up. Furthermore, those patients with a delayed response had mortality and hospitalization outcomes similar to earlier echocardiographic responders and were significantly better than patients who never responded (according to echocardiographic criteria).
Previous studies have shown that approximately 30% to 40% of patients receiving CRT do not respond favorably within 6 to 12 months of therapy, as assessed by echocardiography (3). Our study confirmed these results. We found that 41% of our subjects failed to show significant echocardiographic improvement with CRT by the 1-year follow-up. In addition, our finding that earlier responders tended to include more patients with nonischemic disease (9), LBBB (16–18), or wide QRS duration (17,19) also corresponds to previous results.
A few previous studies have reported the time course of echocardiographic reverse remodeling after CRT (9,10). In a study of 313 CRT patients who were followed up for 3.5 years, Verhaert et al. (10) found that the rate of change in left ventricular end-systolic volume (LVESV) was 10 times greater during the first 6 months of therapy than it was thereafter. Similarly, in a long-term echocardiographic follow-up of the CARE-HF (Cardiac Resynchronization–Heart Failure) trial, LVESV decreased and EF increased rapidly in the first 9 months, and more slowly for the remainder of the follow-up period (9). Our results are similar. We found that, on average, both LVESV and EF improved rapidly in the first year and more slowly thereafter. We also investigated individual differences in the time course of reverse remodeling after CRT. A significant subset of patients exhibited a delayed response, which was not evident when examining the mean response of the entire study group.
Isolated reports of delayed response to CRT have been described previously. Castellant et al. (11) found that 11 of the 84 CRT patients in their study experienced full normalization of EF at long-term follow-up; in 2 of the 11, however, no significant changes in EF were detected for the first 18 to 24 months of therapy. In addition, Gasparini et al. (12) reported that approximately 10% of the 520 CRT patients in their study had a normalization of EF that occurred very late in therapy (up to 3 years after implantation). However, the median follow-up in this study was only 28 months; therefore, the long-term response rate was not well established. Our study used a smaller threshold of response, which has been helpful in predicting long-term outcomes in previous large-scale studies (13–15), and we ensured that all patients were followed up for at least 3 years. To our knowledge, our attempt is the first to characterize this phenomenon in a large cohort of patients.
Long-Term Clinical Outcomes
In previous studies, reverse remodeling response has typically been measured 6 and 12 months after CRT device implantation, and long-term outcomes have been compared between response groups based on those 6- to 12-month echocardiographic assessments. Patients with no significant improvement in LV volume or EF have worse long-term outcomes than those who respond favorably to CRT according to echocardiography (5,7,20). In agreement with these studies, our patients with no echocardiographic response to CRT after 1 year had higher mortality than those with significant echocardiographic improvement. However, an important finding in our study is that the subset of patients who exhibited a delayed echocardiographic response to CRT had long-term clinical outcomes similar to those who exhibited a more rapid response.
Mechanisms of Delayed Response
The mechanisms that may explain the delayed echocardiographic response in some patients remain unclear. Several smaller studies have demonstrated that the correction of mechanical dyssynchrony is an important factor in the echocardiographic response to CRT (21–23). Echocardiographic assessment of response has generally occurred within the first year of CRT in these studies, suggesting that when LV mechanical dyssynchrony is acutely corrected by CRT, reverse remodeling occurs rapidly. Patients with less mechanical dyssynchrony before CRT may require a longer period of therapy before noticeable improvements in systolic function or reverse remodeling occur. Supporting this conjecture, we found that the percentage of mid-term responders exhibiting LBBB morphology on their pre-implantation electrocardiogram tended to be higher than the percentage of long-term responders with LBBB morphology.
Alternatively, patients with a delayed response to CRT may benefit through a mechanism other than improved mechanical dyssynchrony. We did not investigate diastolic function or neurohormonal factors, but improvements in these areas may have been responsible for a more gradual improvement in LV size and function.
As previously shown in other studies, we found that mid-term responders in the present study tended to have more prolonged QRS durations, a lower incidence of ischemic disease, and more frequent LBBB patterns on electrocardiogram than nonresponders (7,16,20). However, long-term responders and nonresponders did not differ in these or the other variables measured.
Our center is aggressive in optimizing the device settings of CRT nonresponders (24,25). The proportion of patients who underwent CRT optimization was the same in the group of patients with a delayed response (18 [35%] of 52 patients) as in those with no response (21 [30%] of 68 patients; p = 0.665). Furthermore, those who had been optimized had no better clinical outcomes than the initial nonresponders who were not optimized (survival: adjusted HR, 1.45 [p = 0.442]; hospitalization: adjusted HR, 0.92 [p = 0.804]). This finding suggests that the delayed response in some patients was unlikely to be caused by the optimization procedure.
A primary limitation is that long-term echocardiograms were not obtained from all patients who underwent implantation during the study period. Of the 790 CRT device implants at our institution during the study period, these results are on the basis of a sample of 294 (37%). It is possible that patients with continuing HF symptoms after CRT were more likely to have long-term echocardiograms ordered by their cardiologists, for example. However, demographic and echocardiographic data were similar between subjects with long-term follow-up echocardiograms and those without, including 1-year response rates. This finding suggests that our patients receiving long-term follow-up echocardiograms may be representative of all our CRT patients. In addition, we only included subjects who were alive 3 years after implantation and thus could potentially have undergone a long-term echocardiogram. Our results thus apply only to those patients who survive at least 3 years after they undergo CRT device implantation and have had the echocardiographic follow-up required for study inclusion.
In addition, as a retrospective study, we relied on hospital records and online obituaries to determine mortality. Although it is possible that our search process missed some hospitalizations or deaths, we expect that any underreporting of these events would be of small magnitude and would be similar between the study groups. Also, because this was a retrospective study, not all data of interest were available for many subjects (e.g., biomarker data, detailed measurement of right ventricular size and function, scar location, New York Heart Association functional classification at follow-up time points, quality of life scores, 6-min hall walk distances). Future prospective studies should include these measurements. Finally, due to the relatively small number of long-term responders in this study, it was impossible to identify the demographic or echocardiographic factors that may predict delayed response in CRT patients.
Among patients who survive at least 3 years after undergoing CRT device implantation with echocardiographic follow-up, 43% of nonresponders at 1-year follow-up experienced a delayed echocardiographic response after ≥3 years of therapy. Survival and hospitalizations were similar for all echocardiographic responders, regardless of the time at which the response occurred. As a result, the echocardiographic response rate among those surviving 3 years after CRT device implantation was 77%.
COMPETENCY IN MEDICAL KNOWLEDGE: In this study, we found that nearly one-half of patients who were not echocardiographic responders after 6 to 12 months of CRT became responders long term. Patients who are considered nonresponders should be followed up with serial echocardiograms because they may exhibit a beneficial delayed remodeling response many years after CRT device implantation.
TRANSLATIONAL OUTLOOK: Additional prospective clinical trials are needed not only to validate the rate of delayed response among CRT patients but also to investigate potential factors resulting in a delayed response. Future studies should consider both baseline predictors of delayed response as well as interventions after CRT device implantation, which could contribute to the conversion of initial nonresponders to long-term responders.
This study was supported in part by a grant from Boston Scientific. Dr. Burns, Mr. Gage, and Ms. Curtin have received research funding from Medtronic, Boston Scientific, and Biotronik. Dr. Bank has received consulting fees from Medtronic, St. Jude Medical, Boston Scientific, and Sorin; honorarium for education from Medtronic; and research funding from Medtronic, Boston Scientific, and Biotronik.
- Abbreviations and Acronyms
- confidence interval
- cardiac resynchronization therapy
- ejection fraction
- end-systolic volume
- heart failure
- hazard ratio
- left bundle branch block
- left ventricular
- left ventricular end-systolic volume
- Received June 22, 2015.
- Revision received September 4, 2015.
- Accepted September 5, 2015.
- American College of Cardiology Foundation
- Brignole M.,
- Auricchio A.,
- Baron-Esquivias G.,
- et al.
- Russo A.M.,
- Stainback R.F.,
- Bailey S.R.,
- et al.
- Ypenburg C.,
- van Bommel R.J.,
- Borleffs C.J.,
- et al.
- Daubert C.,
- Gold M.R.,
- Abraham W.T.,
- et al.
- Verhaert D.,
- Grimm R.A.,
- Puntawangkoon C.,
- et al.
- Bertini M.,
- Hoke U.,
- van Brommel R.J.,
- et al.
- Birnie D.H.,
- Ha A.,
- Higginson L.,
- et al.
- Cleland J.G.,
- Abraham W.T.,
- Linde C.,
- et al.
- Bleeker G.B.,
- Schalij M.J.,
- Nihoyannopoulos P.,
- et al.
- Risum N.,
- Williams E.S.,
- Khouri M.G.,
- et al.
- Bank A.J.,
- Burns K.V.,
- Kelly A.S.,
- Thelen A.M.,
- Kaufman C.L.,
- Adler S.W.