OUP user menu

Impact of clinical and echocardiographic response to cardiac resynchronization therapy on long-term survival

Matteo Bertini, Ulas Höke, Rutger J. van Bommel, Arnold C.T. Ng, Miriam Shanks, Gaetano Nucifora, Dominique Auger, C. Jan Willem Borleffs, Eva P.M. van Rijnsoever, Lieselot van Erven, Martin J. Schalij, Nina Ajmone Marsan, Jeroen J. Bax, Victoria Delgado
DOI: http://dx.doi.org/10.1093/ehjci/jes290 774-781 First published online: 7 December 2012

Abstract

Background Clinical or echocardiographic mid-term responses to cardiac resynchronization therapy (CRT) may have a different influence on a long-term prognosis of heart failure patients treated with CRT. The aim of the evaluation was to establish which definition of response to CRT, clinical or echocardiographic, best predicts long-term prognosis.

Methods and results A total of 679 heart failure patients treated with CRT were included. All the patients underwent a complete history and physical examination and transthoracic echocardiogram prior to CRT implantation and at 6-month follow-up. The clinical and echocardiographic responses to CRT were defined based on clinical improvement (≥1 NYHA class) and LV reverse remodelling (reduction in LV end-systolic volume ≥15%) at 6-month follow-up, respectively. All the patients were prospectively followed up for the occurrence of death. The mean age was 65 ± 11 years and 79% of the patients were male. At 6-month follow-up, 510 (77%) patients showed clinical response to CRT and 412 (62%) patients showed echocardiographic response to CRT. During a mean follow-up of 37 ± 22 months, 140 (21%) patients died. Clinical and echocardiographic responses to CRT were both significantly related to all-cause mortality on univariable analysis. However, on multivariable Cox-regression analysis only echocardiographic response to CRT was independently associated with superior survival (hazard ratio: 0.38; 95% CI: 0.27–0.50; P < 0.001).

Conclusion In a large population of heart failure patients treated with CRT, the reduction in LV end-systolic volume at the mid-term follow-up demonstrated to be a better predictor of long-term survival than improvement in the clinical status.

  • Cardiac resynchronization therapy
  • Echocardiographic response
  • Clinical response
  • Prognosis

Introduction

The long-term survival benefits of cardiac resynchronization therapy (CRT) demonstrated in various trials have changed the clinical management of heart failure patients.13 CRT improves left ventricular (LV) performance by partially restoring more physiological and synchronous contraction. As a consequence, significant improvements in the clinical status and reduction in LV volumes and mitral regurgitation have been reported. Ultimately, these favourable changes result in improved long-term morbidity and mortality rates.4,5 Most of the landmark trials have evaluated the efficacy of CRT by means of improvement in the clinical status and/or a reduction in LV end-systolic volume (LVESV) at the mid-term follow-up.6,7 The value of these surrogate endpoints relies on their ability to predict long-term survival.8,9 Thus far, only few studies have attempted to evaluate the long-term survival implications of mid-term clinical and echocardiographic responses to CRT (commonly assessed at 3–6-month follow-up).4,5,10 In a series of 141 heart failure patients treated with CRT, mid-term LV reverse remodelling had superior accuracy to predict long-term survival than improvement in clinical parameters.5 A recent subanalysis of the CARE-HF trial showed that LV volumes measured 3 months after CRT implantation predicted long-term survival on univariable analysis with a hazard ratio higher than LV volumes at baseline; however, LV volumes (baseline and after 3 months) were not independent determinants of long-term survival.10 In view of these results, it remains unclear whether clinical or echocardiographic mid-term responses to CRT have an influence on a long-term prognosis of heart failure patients treated with CRT. Accordingly, the present evaluation aimed to establish which definition of CRT response at the mid-term follow-up (clinical improvement or LV reverse remodelling) best predicts long-term prognosis.

Methods

Patient population and protocol

A total of 679 heart failure patients undergoing CRT implantation were included. According to current guidelines, the inclusion criteria were: New York Heart Association (NYHA) functional class III–IV, sinus rhythm, LV ejection fraction ≤35%, and QRS duration ≥120 ms. Aetiology of heart failure was considered ischaemic in the presence of significant coronary artery disease (>50% stenosis in ≥1 major epicardial coronary artery) on coronary angiography and/or a history of myocardial infarction or revascularization.

All the patients underwent a complete history and physical examination, 12-lead surface ECG and transthoracic echocardiogram prior to CRT implantation, and at 6-month follow-up. Clinical and ECG variables recorded included NYHA functional class, medication, haemoglobin level, renal dysfunction (estimated glomerular filtration rate ≤60 mL/min/1.73m2),11,12 and QRS duration.

In addition, the number of hospitalizations for heart failure within 6-month follow-up was recorded. All clinical data were prospectively entered into the departmental Cardiology Information System (EPD-Vision®, Leiden University Medical Center, Leiden, the Netherlands) and retrospectively analysed.

The echocardiographic examination consisted of a comprehensive evaluation of LV volumes and function, and severity of mitral regurgitation, if present.

Finally, the clinical and echocardiographic responses to CRT were defined based on clinical improvement and LV reverse remodelling at 6-month follow-up, respectively, as previously described.6

All the patients were prospectively followed up for the occurrence of death. To test whether the clinical and echocardiographic mid-term responses to CRT could independently predict mortality, the long-term follow-up started at 6 months after CRT implantation.

Echocardiography

Transthoracic echocardiography was performed with the patients in the left lateral decubitus position using a commercially available ultrasound transducer and equipment (M4S probe, Vivid 7, GE-Vingmed, Horten, Norway). All transthoracic echocardiographic examinations were performed prior to CRT implantation and at 6-month follow-up. All images were digitally stored on hard disks for offline analysis (EchoPAC version 7.0.0 and 108.1.5 GE-Vingmed, Horten, Norway).

A complete two-dimensional and colour Doppler echocardiographic examination was performed. LV end-diastolic (LVEDV) and LVESV were calculated using Simpson's biplane method of discs. The LV ejection fraction was calculated and expressed as a percentage.13 Intra- and inter-observer variability for the assessment of LV volumes and LVEF were previously reported.14 Severity of mitral regurgitation was graded semi-quantitatively from colour-flow Doppler data using the four-chamber apical views according to the ACC/AHA guidelines. Mitral regurgitation was classified as mild (jet area/left atrial area <20%), moderate (jet area/left atrial area 20–40%), and severe (jet area/left atrial area >40%).15

CRT implantation

All the patients received a biventricular pacemaker with cardioverter-defibrillator function (Contak Renewal 4RF, Boston Scientific St Paul, MN, USA; or InSync Sentry, Medtronic, Inc., Minneapolis, MN, USA; Lumax 340 HF-T, Biotronik, Berlin). The right atrial and ventricular leads were positioned conventionally. All the LV leads were implanted transvenously, and positioned preferably in a (postero-)lateral vein. A coronary sinus venogram was obtained using a balloon catheter, followed by the insertion of the LV pacing lead. An 8-F guiding catheter was used to place the LV lead (Easytrak, Boston Scientific, or Attain-SD, Medtronic, or Corox OTW Biotronik) in the coronary sinus.

Definition of CRT response

As previously reported, clinical response to CRT was defined as improvement ≥1 in NYHA functional class at 6-month follow-up and echocardiographic response to CRT was defined by the occurrence of LV reverse remodelling (reduction ≥15% in LVESV at 6-month follow-up).6

All clinical and echocardiographic analyses were performed by independent blinded physicians.

Study endpoints

All the patients were followed up regularly (every 3–6 monthly intervals) and all deaths occurring after 6-month follow-up were recorded as events. All-cause mortality was adjudicated by physicians blinded to the clinical and echocardiographic data.

Statistical analysis

All continuous variables are presented as mean and standard deviation. Categorical variables are presented as frequencies and percentages, and were compared using the χ2 test. Student's t-test was used to compare unpaired continuous variables. First, the cumulative event rates 6 months after CRT implantation were calculated using the Kaplan–Meier method and dichotomizing the population according to the clinical and echocardiographic response to CRT. The log-rank tests for time-to-event data with respect to all-cause mortality were used for statistical comparison between two patient groups. To identify independent predictors of all-cause mortality, multivariable Cox proportional hazards models were constructed with backward selection model. All significant univariable clinical and echocardiographic predictors at baseline, clinical, and echocardiographic response to CRT at 6-month follow-up, and hospitalizations for heart failure within 6-month follow-up were entered in the multivariable model as covariates. The Cox proportional hazards models were then used to estimate hazard ratios and 95% confidence intervals (CIs) for those independent variables. To avoid multicollinearity between the univariable predictors, a correlation coefficient of <0.7 (corresponding to a tolerance level of >0.5) was set. Secondly, the cumulative event rates 6 months after CRT implantation were calculated using the Kaplan–Meier method after dividing the populations into four different subgroups based on combined clinical and/or echocardiographic response: patients with both clinical and echocardiographic response, patients with clinical but not echocardiographic response, patients with echocardiographic but not clinical response, and patients with neither clinical or echocardiographic response. The log-rank tests for time-to-event data with respect to all-cause mortality were used for statistical comparison between four patient subgroups All the statistical tests were two-sided, and a P-value <0.05 was considered significant. A statistical software program SPSS 16.0 (SPSS, Inc., Chicago, IL, USA) was used for all statistical analyses.

Results

Overall patient population

Of the 679 patients included, 16 (2.4%) patients died before 6-month follow-up and these patients were excluded from further analysis. Therefore, the patient population consisted of 663 heart failure patients.

Baseline clinical, ECG, and echocardiographic characteristics are reported in Table 1. The mean age was 65 ± 11 years and 79% of the patients were male. All the patients had dilated with depressed LV systolic function.

View this table:
Table 1

Baseline characteristics of overall patient population, and survivors vs. non-survivors

Overall population (n = 663)Survivors (n = 523)Non-survivors (n = 140)P-value
Demographic characteristics
 Age (years)65 ± 1165 ± 1168 ± 100.003
 Male [n (%)]523 (79)405 (77)118 (84)0.078
 Body surface area (m2)1.97 ± 0.221.98 ± 0.221.94 ± 0.220.041
 Body mass index (kg/m2)26.4 ± 4.326.6 ± 4.225.6 ± 4.30.013
Medical history
 Diabetes [n (%)]131 (20)90 (17)41 (29)0.001
 Ischaemic aetiology [n (%)]398 (60)300 (57)98 (70)0.007
 Systolic blood pressure (mmHg)121 ± 20123 ± 20114 ± 19<0.001
 Diastolic blood pressure (mmHg)72 ± 1273 ± 1269 ± 11<0.001
Medications [n (%)]
 ACE inhibitor/angiotensin receptor blockers597 (90)477 (91)120 (86)0.054
 Beta-blockers466 (70)381 (73)85 (61)0.005
 Diuretics551 (83)422 (81)129 (92)0.001
 Nitrates161 (24)120 (23)41 (25)0.12
 Statins386 (58)314 (60)72 (51)0.067
 Oral anticoagulants/aspirin619 (93)490 (94)129 (92)0.51
Clinical characteristics
 NYHA functional class IV [n (%)]40 (6.0)26 (5.0)14 (10.0)0.026
 Heart rate (b.p.m.)72 ± 1772 ± 1774 ± 160.20
 QRS duration (ms)155 ± 33155 ± 33156 ± 310.76
 Haemoglobin level (mmol/L)8.3 ± 0.98.4 ± 0.98.1 ± 1.00.017
 Renal dysfunction [(n (%)]269 (40.6)176 (33.7)93 (66.4)<0.001
Echocardiographic characteristics
 LV end-diastolic volume (mL)218 ± 80215 ± 74231 ± 960.22
 LV end-systolic volume (mL)164 ± 69160 ± 64180 ± 840.029
 LV ejection fraction (%)25 ± 826 ± 823 ± 8<0.001
 Severe mitral regurgitation [n (%)]113 (17)72 (14)41 (29)<0.001
  • ACE, angiotensin-converting enzyme; LV, left ventricular; NYHA, New York Heart Association.

  • Renal dysfunction: glomerular filtration rate ≤60 mL/min/1.73 m2.

At 6-month follow-up, 510 (77%) patients showed an improvement ≥1 in NYHA functional class (clinical response to CRT) and 412 (62%) patients showed ≥15% reduction in LVESV (echocardiographic response to CRT). Furthermore, the patient population (n = 663) was divided in four subgroups based on combined clinical and/or echocardiographic response to CRT at 6-month follow-up: 348 (52.4%) patients showed both clinical and echocardiographic response, 159 (24%) patients showed clinical but not echocardiographic response, 64 (9.7%) were echocardiographic but not clinical responders, and 92 (13.9%) patients did not show either clinical or echocardiographic response.

Survivors vs. non-survivors

During a mean follow-up of 37 ± 22 months, 140 (21%) patients died. Baseline parameters of the survivors vs. the non-survivors are reported in Table 1. Of note, the survivors were younger, had less frequently diabetes, ischaemic aetiology of heart failure, and NYHA functional class IV. Interestingly, the survivors had higher haemoglobin levels and also more preserved renal function. Regarding the echocardiographic data at baseline, the survivors had smaller LVESV, higher LV ejection fraction, and less frequently severe mitral regurgitation. There were seven (1.3%) hospitalizations for heart failure in the survivors and eight (5.7%) in the non-survivors (P = 0.002) within 6-month follow-up.

The survivors were more frequently clinical responders when compared with the non-survivors (78 vs. 70%, P = 0.037). Moreover, the survivors showed a higher echocardiographic response rate when compared with the non-survivors (69 vs. 37%, P < 0.001). The clinical (change in NYHA functional class at 6-month follow-up) and the echocardiographic (change in LVEDV, LVESV, and LV ejection fraction) improvements are illustrated in Figure 1.

Figure 1

(A) The absolute change between 6-month follow-up and baseline in New York Heart Association (NYHA) functional class. (B and C) The relative changes in left ventricular end-diastolic (LVEDV) and end-systolic volumes (LVESV). (D) The absolute change in the left ventricular ejection fraction (LVEF).

Clinical and echocardiographic response to CRT vs. long-term mortality

Clinical and echocardiographic responses to CRT were both significantly related to all-cause mortality. When the patient population was dichotomized based on clinical response to CRT, a cumulative 2, 7, and 12% of the patients with improvement in ≥1 NYHA functional class died by 12-, 24-, and 36-month follow-up, respectively. In contrast, a respective 7, 18, and 23% of the patients without improvement in ≥1 NYHA functional class died during the same time period (log-rank P = 0.004: Figure 2A).

Figure 2

Kaplan–Meier curves estimate of all-cause mortality. (A) The probability of all-cause mortality which differed significantly between the clinical responders and clinical non-responders. (B) The probability of all-cause mortality which differed significantly between the echocardiographic responders and echocardiographic non-responders. The orange shadowed bar indicates the 6-month follow-up period after CRT implantation.

When the patient population was dichotomized based on an echocardiographic response to CRT, a cumulative 1, 4, and 8% of the patients with LV reverse remodelling died by 12-, 24-, and 36-month follow-up, respectively. In contrast, a respective 8, 19, and 27% of the patients without LV reverse remodelling died during the same time period (log-rank P < 0.001) (Figure 2B).

To identify whether clinical and echocardiographic responses were independent predictors of all-cause mortality during the follow-up, significant univariable predictors with a P-value <0.05 were entered into the Cox proportional-hazard model as covariates (Table 2). On multivariable analysis, echocardiographic response (hazard ratio: 0.35; 95% CI: 0.25–0.50; P < 0.001) but not clinical response was independently associated with superior survival.

View this table:
Table 2

Cox uni- and multivariable regression analyses for all-cause mortality

Dependent variable: All-cause mortality Univariable analysisMultivariable analysis
HR (95% CI)P-valueHR (95% CI)P-value
Independent variables
Age (years)1.03 (1.01–1.05)0.0011.03 (1.01–1.05)0.012
Body mass index (kg/m2)0.95 (0.91–0.99)0.012
Diabetes2.06 (1.42–2.97)<0.0012.05 (1.40–3.01)<0.001
Ischaemic aetiology1.85 (1.29–2.67)0.001
Systolic blood pressure (mmHg)0.98 (0.97–0.99)<0.0010.99 (0.98–0.99)0.023
Diuretics2.24 (1.21–4.15)0.010
Haemoglobin level (mmol/L)0.75 (0.63–0.90)0.002
Renal dysfunction3.05 (2.15–4.33)<0.0011.93 (1.29–2.89)0.001
LV ejection fraction (%)0.97 (0.95–0.99)0.0030.98 (0.96–1.00)0.088
Severe mitral regurgitation1.91 (1.32–2.75)0.0011.47 (1.00–2.14)0.048
Hospitalizations for heart failure within 6-month follow-up3.93 (1.91–8.06)<0.0014.03 (1.91–8.50)<0.001
Clinical response0.59 (0.41–0.85)0.004
LV reverse remodelling0.31 (0.22–0.43)<0.0010.35 (0.25–0.50)<0.001
  • CI, confidence intervals; HR, hazards ratio; LV, left ventricular.

  • Renal dysfunction: glomerular filtration rate ≤60 mL/min/1.73 m2.

As expected, the main cause of death was cardiovascular [91 (65%) of 140 patients] and specifically related to the progression of heart failure [81 (89%) of 91 patients]. Interestingly, echocardiographic non-responder patients died more frequently for cardiovascular reasons when compared with responder patients [65 (74%) vs. 26 (50%) patients, P = 0.004].

Finally, dividing the patient population in four subgroups based on combined clinical and/or echocardiographic response, patients with both clinical and echocardiographic response to CRT and patients with echocardiographic but not clinical response had significantly lower mortality (12.9 and 10.9%, respectively) when compared with patients with clinical but not echocardiographic response or patients without any response to CRT (nor clinical or echocardiographic), in whom mortality rates were 32.7 and 39.1%, respectively (χ2 = 48.72, P < 0.001). Figure 3 represents the Kaplan–Meier survival curves of these four subgroups of patients showing superior survival for patients with both clinical and echocardiographic response to CRT and for patients with echocardiographic but not clinical response to CRT.

Figure 3

Kaplan–Meier curves estimate of all-cause mortality in four subgroups of patients: clinical + echocardiographic responders, only echocardiographic responders, only clinical responders, and non-responders (nor clinical or echocardiographic). The orange shadowed bar indicates the 6-month follow-up period after CRT implantation.

Discussion

The present evaluation demonstrated that reduction in ≥15% LVESV, regardless changes in the clinical status, was a predictor of all-cause mortality in heart failure patients treated with CRT. Specifically, LV reverse remodelling was an important and independent predictor of the outcome in CRT recipients over other well-established predictors of mortality in the general heart failure population.

Clinical and echocardiographic response to CRT

Along the various single- and multi-centre trials on CRT, definition of response to CRT has widely varied.16 Different clinical and echocardiographic parameters have been used to evaluate the efficacy of CRT. In addition, the response to CRT has been evaluated at different time points, most commonly at 3- or 6-month follow-up. However, it has been repeatedly shown that clinical and echocardiographic CRT response may not coincide and, indeed, a significant percentage of patients who showed an improvement in clinical endpoints did not show any improvement in echocardiographic parameters (≥15% LV reverse remodelling or significant improvement in LV ejection fraction).6,16,17 In the present evaluation, 77% of the patients showed an improvement in NYHA function class of at least 1-point, whereas 62% showed ≥15% reduction in the LVESV. This confirms previous findings and suggests the presence of a placebo effect that may overestimate the benefits of CRT when the response is defined according to clinical criteria. However, it may be more interesting to evaluate whether this improvement in the clinical status or in echocardiographic parameters at the mid-term follow-up conveys or not superior long-term survival. Indeed, the appropriateness of surrogate endpoints, such as NYHA functional class or LV reverse remodelling, to evaluate the efficacy of heart failure therapies depends on the strength of the statistical relationship between the change in the surrogate endpoints over time and the clinical outcome.8

Response to CRT vs. long-term survival

The present evaluation demonstrated that the occurrence of significant LV reverse remodelling at 6-month follow-up was related with superior long-term survival. In contrast, improvement in NYHA functional class was not associated with improved long-term survival. This was also confirmed dividing the patient population in four subgroups based on combined clinical and/or echocardiographic response CRT response. The patients with echocardiographic response to CRT (clinical and echocardiographic or only echocardiographic CRT response) had better survival when compared with the patients without echocardiographic response to CRT (Figure 3). Previously, Yu et al.5 have shown in 141 heart failure patients that LVESV reduction was independently associated to the mid-term outcome after CRT. The authors found a relationship between reduction in the LVESV and the mid-term clinical outcome. However, the association between improvement in the clinical status (defined by ≥1 point in NYHA functional class) and all-cause and cardiovascular mortality was not evaluated. More recently, the results of the REVERSE, MADIT-CRT, and RAFT trials have shown that in mildly symptomatic heart failure patients (NYHA functional class I–II), CRT reduced the number of heart failure re-hospitalizations, induced a significant LV reverse remodelling and improved the long-term survival.1820 This improvement in long-term survival might be related to improvement in LV performance and LV reverse remodelling rather than improvements in NYHA functional class since the included patients were asymptomatic or mildly symptomatic.

The present evaluation confirms and extends these preliminary results in over 650 heart failure patients treated with CRT. LV reverse remodelling was independently related to all-cause mortality (hazard ratio: 0.34; 95% CI: 0.25–0.50; P < 0.001) together with age, presence of diabetes, renal dysfunction, and heart failure hospitalizations. These clinical parameters (age, diabetes, systolic blood pressure, renal dysfunction, and heart failure hospitalizations) are well-known strong prognosticators in patients with advanced heart failure.2123 However, thus far, the effects of CRT on LV dimensions and its implications on long-term survival have not been evaluated together with these clinical variables.4,5 The fact that LV reverse remodelling at the mid-term follow-up was still an independent predictor of the long-term outcome after adjusting for these powerful risk factors underlines the relevance of the assessment of CRT response at the mid-term follow-up based on LV reverse remodelling.

Study limitations

The study was retrospective and reported the experience of a single centre. Data on LV lead position were not included in the present evaluation. The influence of this parameter on CRT response and outcome has been well described in observational and randomized trials. However, the impact of LV lead position on CRT response and outcome was beyond the scope of the present study that focused on the prognostic implications of clinical and/or echocardiographic response. The percentage of beta-blocker use was relatively low but still higher when compared with previous randomized trials (MIRACLE and COMPANION) and similar to the CARE-HF trial but lower than the recent REVERSE, MADIT-CRT, and RAFT trials.1,3,1820,24 However, all the patients received the maximum tolerated dose of beta-blocker.

Conclusions

In a large population of heart failure patients treated with CRT, the reduction in LVESV at the mid-term follow-up demonstrated to be a better predictor of long-term survival than improvement in the clinical status. In addition, LV reverse remodelling was an important and independent predictor of the outcome in these patients together with other well-established predictors of long-term survival.

Funding

D.A. is financially supported by the ‘Programme de bourse de perfectionnement et de fellowship du Centre Hospitalier de l'Université de Montréal (CHUM) et de la Fondation du CHUM'.

Conflict of interest: The department of cardiology received grants from Medtronic, Biotronik, Boston Scientific, St. Jude Medical & GE Healthcare. Dr Delgado received consulting fees from Medtronic and St. Jude Medical. The remaining authors have nothing to disclose.

References

View Abstract