Right ventricular function and survival following cardiac resynchronisation therapy
- Darryl P Leong1,2,
- Ulas Höke1,
- Victoria Delgado1,
- Dominique Auger1,
- Tomasz Witkowski1,
- Joep Thijssen1,
- Lieselot van Erven1,
- Jeroen J Bax1,
- Martin J Schalij1,
- Nina Ajmone Marsan1
- 1Department of Cardiology, Leiden University Medical Centre, Leiden, The Netherlands
- 2The Discipline of Medicine, University of Adelaide and Flinders University, Adelaide, Australia
- Correspondence to Dr Nina Ajmone Marsan, Department of Cardiology, Leiden University Medical Centre, Albinusdreef 2, Leiden 2333 ZA, The Netherlands;
- Received 17 September 2012
- Revised 23 November 2012
- Accepted 27 November 2012
- Published Online First 12 January 2013
Objectives Right ventricular (RV) function is an important prognostic marker in heart failure. However, its impact on all-cause mortality following cardiac resynchronisation therapy (CRT) independent of confounding factors has not been evaluated. Furthermore, evidence concerning the effect of CRT on RV function is limited. The study's aims were to: (1) assess the prognostic importance of RV function among CRT recipients, and (2) characterise RV functional change following CRT and its determinants.
Design Retrospective observational study.
Setting Single tertiary centre.
Patients A total of 848 CRT recipients (median age 65 years, 78% male, 60% ischaemic) underwent echocardiography before and 6 months after CRT. RV function was evaluated using tricuspid annular plane systolic excursion (TAPSE), with a ≤14 mm threshold indicating severe RV impairment. The primary endpoint was long-term all-cause mortality.
Results Significant baseline RV dysfunction was observed in 286 (34%) individuals. After a median 44 months, 288 deaths occurred. RV impairment was associated with a greater incidence of all-cause mortality (log-rank p<0.001). Independent predictors of this endpoint were functional class, ischaemic aetiology, diabetes, atrial fibrillation, renal dysfunction, bigger left ventricular (LV) end-systolic volume, less LV dyssynchrony and reduced TAPSE. Importantly, TAPSE added prognostic value to these recognised prognostic parameters (likelihood-ratio test p<0.001). Furthermore, improvement in RV function after CRT was independent of the improvement in LV systolic function but significantly associated with the improvement in LV diastolic function. Importantly, a favourable RV functional response to CRT was associated with superior survival.
Conclusions RV function is an independent predictor of long-term outcome following CRT.
Cardiac resynchronisation therapy (CRT) reduces mortality among patients with advanced heart failure, impaired left ventricular (LV) systolic function and QRS duration ≥120 ms.1 However, long-term death rates remain high in CRT recipients.2 Identification of potential CRT candidates who are unlikely to derive survival benefit from this therapy, therefore, remains an important goal. Characterisation of patients who will have a poor outcome despite CRT may in fact allow the development of more targeted treatment approaches in this population.
Right ventricular (RV) function is recognised as a cardinal prognostic marker in patients with heart failure and reduced LV systolic function.3 Its impact on all-cause mortality following CRT independent of confounding factors has not been evaluated, however. Furthermore, evidence on the effect of CRT on RV contractile function is limited.4 The aims of this evaluation were therefore to: (1) assess the prognostic importance of RV function among CRT recipients, and (2) characterise RV function changes following CRT and its determinants.
The study population comprised of 905 consecutive patients from the ongoing registry of CRT recipients at the Leiden University Medical Centre.5 Patient data were collected prospectively in the departmental Cardiology Information System (EPD-Vision, Leiden University Medical Centre, Leiden, the Netherlands) and retrospectively analysed. Indication for CRT was heart failure with LV ejection fraction (LVEF) ≤35%, and QRS duration ≥120 ms.6 Heart failure aetiology was considered ischaemic in the presence of significant coronary artery disease (>50% stenosis in ≥1 major epicardial coronary artery on coronary angiography) and/or history of prior myocardial infarction or revascularisation.
According to the institutional protocol, prior to CRT implantation, all patients underwent extensive clinical and echocardiographic evaluation.7 Clinical evaluation was performed in particular for identification of comorbid conditions, medications and functional status, which was quantified by the New York Heart Association (NYHA) functional class. In addition, 12-lead electrocardiography and 6-min walk test were performed. Renal function was determined with estimated glomerular filtration rate calculated using the Cockcroft-Gault formula. Transthoracic echocardiography was undertaken according to a standardised protocol (see below), which included the assessment of LV and RV function. The same protocol was repeated at 6 months' follow-up to evaluate the changes in LV and RV function after CRT. Clinical long-term follow-up was performed to determine the incidence of all-cause mortality.
Patients were imaged in the left lateral decubitus position using a commercially available system equipped with a 3.5 Mhz transducer (Vingmed Vivid7 and E9, General Electric Healthcare, Horten, Norway). Two-dimensional grey-scale, pulsed, continuous and colour Doppler data were acquired in the parasternal and apical views, the latter including dedicated imaging of the RV. Images were recorded digitally in cine-loop format and analysed offline with commercial software EchoPAC (110.0.0, GE-Vingmed, Horten, Norway).
LV end-systolic and end-diastolic volumes were measured from the apical 2-chamber and 4-chamber views and LVEF calculated using the biplane Simpson's technique.8 RV function was evaluated by tricuspid annular plane systolic excursion (TAPSE), which was displayed by applying an M-mode algorithm to the lateral tricuspid annulus in the RV apical view. This approach to the assessment of RV function was chosen as a widely available and highly feasible one, which is recommended as a routine part of RV evaluation.9 RV impairment was defined as TAPSE≤14 mm, as this cut-off has demonstrated important prognostic value in heart failure patients.10 RV systolic pressure and mitral regurgitation (MR) severity were estimated according to current guidelines.11 ,12 LV dyssynchrony was assessed by septal-to-lateral delay on colour tissue Doppler imaging as previously described.13 In addition, LV diastolic function was assessed by measuring transmitral E and A wave velocities using pulsed wave Doppler at the level of the mitral leaflet tips, and E′ by offline analysis using colour tissue Doppler imaging at the septal mitral annulus. The E/E′ ratio was then calculated. Echocardiographic measurements were performed blind to patient outcome.
The LV lead was inserted transvenously into the coronary sinus and advanced where possible to the posterolateral cardiac vein under the guidance of venography. The RV and right atrial leads were positioned at the RV apex and right atrial appendage respectively. Implanted devices included Contak Renewal, Contak TR or Contak CD (Guidant USA), Insync Marquis, Insync III, Insync Sentry or Protecta (Medtronic Inc. USA), Atlas HF (St Jude Medical USA) and Lumax (Biotronik, Germany), while LV leads included Attain (Medtronic Inc. USA), Corox (Biotronik, Germany) and Easytrak (Guidant USA).
Long-term follow-up was performed by case record review, telephone contact with patient primary healthcare providers, and through the national death registry. The primary endpoint was all-cause mortality.
Continuous variables are presented as mean±SD (where normally distributed) or median (IQR) and compared between groups at baseline using analysis of variance or the Kruskal-Wallis test as appropriate. Categorical data are summarised as frequencies and percentages, and compared using the χ2 test.
For survival analysis, rates of survival were estimated using the Kaplan-Meier technique and tested between groups using the log-rank test. To reduce type I error, pair-wise comparison among multiple groups was undertaken with Bonferroni's adjustment to the threshold significant p value.14 The proportional hazards assumption was checked for continuous variables by visual inspection of scaled Schoenfeld residuals15 and confirmed by the method of Grambsch and Therneau,16 and for categorical variables by visual inspection of log-log plots. Cox proportional hazards models were fitted to evaluate the predictive value of patient characteristics for the primary endpoint. Predictors associated with a p value <0.2 on univariate analysis were entered into a multivariate analysis using backward elimination to identify independent predictors of the primary endpoint. Results are expressed using HRs. Model fit was evaluated by visual inspection of the cumulative hazard of Cox-Snell residuals and by the Harrell's C-statistic.17 The incremental value of successive predictors of the primary endpoint was assessed by the likelihood ratio test for the nesting model containing the covariate of interest. Covariates for this analysis were those demonstrating independent predictive value in the Cox model.
For repeated-measures analysis of TAPSE, mixed effects modelling was employed. The significance of predictor variables of interest was evaluated by including them in a mixed model as part of an interaction term with subjects’ visit time (ie, baseline vs follow-up). If this predictor-time interaction term was significant, it implied that the candidate predictor variable's influence on TAPSE was time-dependent. Post hoc testing was then performed to determine whether the candidate predictor variable's influence on TAPSE was significant at baseline and follow-up visit. All statistical tests were two-sided, and a p value <0.05 was considered statistically significant. All analyses were undertaken using STATA software, V.12 (Stata Corp, College Station, Texas, USA).
Nine hundred and five consecutive patients who successfully underwent CRT were evaluated. In a total of 57 (6%) patients, TAPSE was deemed non-evaluable owing to image quality. Baseline clinical and echocardiographic characteristics of the remaining 848 patients are displayed in table 1. One hundred and thirty-seven (16%) patients had atrial fibrillation (AF) at the time of implantation and 42 (5%) were upgrade procedures from RV pacing. Six hundred and fifty-two (77%) patients were in NYHA class III or IV at implantation; the remaining were in NYHA class II.
Significant RV dysfunction (as defined by a TAPSE≤14 mm10) was observed in 286 (34%) patients. Comparisons between patients with TAPSE≤14 mm and >14 mm are summarised in table 1. Patients with significant RV dysfunction exhibited a greater prevalence of ischaemic cardiomyopathy, diabetes mellitus and AF. Furthermore, functional status, exercise capacity, renal function and LV systolic and diastolic function were poorer, while RV systolic pressure was higher among patients with RV impairment. Individuals with RV impairment also exhibited less LV dyssynchrony than those with more preserved RV function.
Over the course of a median follow-up time of 44 months (IQR 29–65 months), 288 deaths occurred. Death rate was observed to be significantly higher among CRT recipients with impaired baseline RV function (figure 1). The Kaplan-Meier curves of those with impaired versus preserved baseline RV function diverged early during follow-up, and continued to diverge over time. One-year death rates among those with impaired and preserved baseline RV function were 18% (95% CI 14% to 23%) and 3% (95% CI 2% to 5%), respectively. At 2 years, these death rates were 27% (95% CI 22% to 33%) and 8% (95% CI 6% to 10%), while at 5 years’ follow-up, death rates were 50% (95% CI 44% to 58%) and 21% (95% CI 18% to 25%), respectively.
TAPSE was tested as a predictor of the incidence of mortality in univariate Cox analysis together with age, gender, ischaemic aetiology, diabetes mellitus, AF, NYHA functional class, 6 min walk distance, estimated glomerular filtration rate, QRS duration, LVEF, LV end-systolic volume, MR grade, LV dyssynchrony, E/E′, and RV systolic pressure (table 2). For the multivariate model, 6-min walk distance was omitted in favour of NYHA class as a measure of functional status in the interest of feasibility, as NYHA class was evaluable in all patients, whereas 179 (21%) patients were unable to perform the 6 min walk test. Similarly, RV systolic pressure was not included in the multivariate model in favour of TAPSE for the purposes of model parsimony (to avoid multicollinearity and model overfit) and technical feasibility (RV systolic pressure could not be measured echocardiographically in 419 (49%) patients due to the absence of clear tricuspid regurgitation Doppler signal). By this model, independent predictors of the primary endpoint were more advanced NYHA class, ischaemic aetiology, the presence of diabetes mellitus, AF, poorer renal function, bigger LV end-systolic volume, less LV dyssynchrony and reduced TAPSE. Model fit was found to be acceptable graphically by inspection of the cumulative hazard of Cox-Snell residuals (see online supplementary appendix) with Harrell's C statistic 0.76. The incremental value of evaluation of TAPSE for the prediction of death over and above recognised prognostic indices is displayed in figure 2. The unadjusted relative risk of the primary endpoint as a function of baseline TAPSE is presented in figure 3. This demonstrates progressive increase in the risk of death with poorer baseline TAPSE, in spite of CRT.
RV functional changes following CRT
During the first 6 months following CRT, 54 (6.4%) individuals died. The following analysis was based on the remaining 794 patients, of whom 738 (93%) had echocardiography at 6 months. Across the cohort as a whole, a significant improvement of TAPSE and RV systolic pressure was demonstrated. In particular, TAPSE increased from 17±5 mm at baseline to 19±6 mm (p<0.001), while estimated RV systolic pressure declined from 38±14 mmHg to 35±12 mmHg (p<0.001). In addition, the mean left ventricular end-systolic volume (LVESV) decreased from 165±70 ml at baseline to 136±64 ml (p<0.001), LVEF increased from 26±8% at baseline to 32±10% (p<0.001) and E/E′ ratio improved from 21±14 at baseline to 18±11 (p<0.001) at 6 months.
Potential determinants of the improvement in RV function following 6 months’ CRT, and in particular the association with improvement in LV function, were explored using mixed effects modelling. While at baseline and follow-up there was a significant positive association between LVEF and TAPSE (coefficient 0.11, 95% CI 0.075 to 0.15, p<0.001), the change in TAPSE from baseline to follow-up was independent of the change in LVEF (p value of the LVEF-visit time interaction term=0.4). A trend towards a time-dependent relationship between TAPSE and LVESV was observed (p value of the LV end systolic volume-visit time interaction term=0.06) but did not reach statistical significance. However, the change in TAPSE was significantly and inversely related to the change in E/E′ (p value of the E/E′-visit interaction term <0.001), such that reduction in E/E′ following 6 months’ CRT was associated with improvement in TAPSE. Change in TAPSE was not directly related to the improvement in MR (MR-visit interaction term p value 1.0), although there was a significant negative relationship between MR grade and TAPSE at baseline and follow-up (coefficient −0.50, 95% CI −0.79 to −0.21, p=0.001). Finally, at baseline there was no significant relationship between TAPSE and estimated RV systolic pressure; however, improvement in TAPSE over 6 months was associated with the reduction in RV systolic pressure (RV systolic pressure-visit interaction term p=0.003).
Of 236 individuals with impaired baseline RV function (TAPSE≤14 mm) who were evaluated at 6 months, 126 (53%) exhibited improvement in RV function with a follow-up TAPSE>14 mm. Among the 502 patients with preserved baseline RV function, only 36 (7%) showed a worsening of TAPSE to 14 mm or below, despite CRT. The relationship between change in TAPSE over the 6 months and subsequent long-term mortality was examined. CRT recipients whose TAPSE was >14 mm at 6 months follow-up displayed significantly superior survival to those whose 6-month TAPSE was ≤14 mm, irrespective of the degree of baseline RV function (figure 4).
The key findings of this study are: (1) baseline RV function, as quantified by TAPSE as a highly feasible echocardiographic measure, is an independent predictor of all-cause mortality following CRT and confers incremental prognostic value over a broad range of clinical and echocardiographic parameters; (2) CRT exerts a beneficial effect over RV function which is independent of the improvement in LV systolic function or of the reduction in MR, but is associated with improvement in LV diastolic function.
RV dysfunction in heart failure and outcome following CRT
In the present study of a large cohort of heart failure patients undergoing CRT, significant RV dysfunction (using a 14 mm TAPSE threshold) was observed in 34% of individuals. In these patients, several causes for the development of RV dysfunction have been proposed, including the same cause (ischaemic or idiopathic) as the LV dysfunction (direct myocardial involvement), pulmonary venous hypertension secondary to LV impairment, ventricular interdependence and neurohormonal interactions.18 Accordingly in the current study, patients with impaired RV function shared features of lower LVEF, less LV dyssynchrony and higher pulmonary pressures compared with patients with preserved RV function. They also exhibited a greater prevalence of ischaemic heart disease and AF, a poorer functional status and target organ compromise, such as renal dysfunction. These findings therefore represent one of the most comprehensive characterisations of patients with RV impairment in the context of LV failure, and suggest RV dysfunction as an important marker for more advanced heart failure.
In line with this hypothesis, the presence of RV dysfunction is also a well-known independent prognostic factor in heart failure patients.3 However, the role of RV dysfunction in predicting outcome following CRT is only of emerging interest. Scuteri et al19 have shown in a small group of patients that TAPSE is predictive of reverse LV remodelling following CRT. Among 130 CRT recipients, Tabereaux et al20 recently demonstrated RV dysfunction to be an independent predictor of a heterogeneous composite adverse outcome including death, transplantation, need for LV assist device, lack of improvement in NYHA class and hospital care. Field et al21 reported similar findings using RV myocardial performance as measurement of RV function. Finally, Kjaergaard et al22 showed in 450 minimally symptomatic patients (NYHA class I or II) receiving CRT, that baseline RV dysfunction is an independent predictor of symptomatic deterioration and reduced reverse LV remodelling. The same authors acknowledged the need to study the relationship between CRT and RV function in CRT recipients with more advanced symptoms, and call for research on the role of RV function in survival following CRT.22
Thus, existing research has yet to demonstrate prognostic value for evaluation of RV function for all-cause mortality, and importantly has examined only to a limited extent whether the prognostic capacity of RV function assessment is confounded by other important determinants of outcome, such as LV function. The present study has applied TAPSE as a sensitive and easily acquired echocardiographic index of RV function9 to address these remaining uncertainties. It has shown in a large series of CRT patients that RV dysfunction is associated with overall mortality independent of a broad range of potential confounding factors, conferring additive prognostic value to routinely evaluated clinical and echocardiographic parameters. Therefore, the evaluation of RV function among other recognised prognostic factors may help clinical decision-making prior to CRT prescription. Such a priori risk stratification may be particularly relevant for individuals in whom the risk-benefit relationship for CRT is marginal.
RV function after CRT
To date, most evidence on the effects of CRT has focussed on its ability to promote LV reverse remodelling. There is a relative paucity of literature on its influence on the RV. Bleeker et al23 have demonstrated that CRT results overall in significant RV reverse remodelling and that this beneficial effect was most marked in patients with more severe RV dilation at baseline. Beyond RV remodelling, Rajagopalan et al24 and Donal et al25 showed using tissue Doppler imaging that CRT might improve RV contractile function. Similarly, the present study showed in a large series of patients a significant improvement in RV function after CRT, using TAPSE as a simple and widely available echocardiographic measure.
A mechanistically important aim of this study was to identify potential determinants of improvement in RV function following CRT. Previous cross-sectional evidence in heart failure patients suggested that LV and RV function are closely associated.26 Less clear is the temporal and causal relationship between the two, particularly in response to heart failure therapy. The present study has shown that improvement in RV function following CRT is independent of rather than secondary to the increase in LV systolic function, despite the association between LVEF and RV function observed before CRT implantation. However, a trend towards a significant association between change in RV function and change in LVESV was noted, and importantly the improvement in RV function was significantly associated with the reduction in LV filling pressures as estimated by the E/E′ ratio. However, the current study could not demonstrate a significant association between improvement in RV function and reduction in MR severity, despite a significant negative relationship between MR grade and TAPSE at baseline and follow-up. Taken together, these findings prompt speculation that CRT, in addition to its effect on LV systolic function, exerts a beneficial effect on RV function that is probably partially direct to the RV myocardium and partially secondary to the improvement in LV diastolic function and therefore related to ventricular interdependency.
Novel to the current study is also the observation that improvement in RV function is not uniform following CRT. While it was observed that baseline RV function is poorer among CRT recipients who experienced an adverse outcome during follow-up, this study importantly also demonstrated that amelioration or preservation of good RV function after CRT was associated with improved survival. In contrast, progressive decline in RV function or lack of significant RV function improvement after CRT were associated with a poor outcome.
Owing to its observational nature, this study is unable to conclude whether patients with severe RV dysfunction should not receive otherwise indicated CRT. This postulate should be addressed in a prospective study of CRT on the basis of RV function. However, the current study emphasises the importance of assessing RV function before and after CRT for optimal patient management.
This study indicates that baseline evaluation of RV function confers additive predictive value over and above a wide range of recognised prognostic factors. Improvement in RV function after CRT was independent of improvement in LV systolic function and associated with improvement in LV diastolic function. Furthermore, favourable RV functional response to CRT was associated with superior survival.
Contributors DPL: conception and design of the study; collection, analysis and interpretation of data; drafting of the manuscript; final approval of the manuscript. UH: conception and design of the study; collection, analysis and interpretation of data; final approval of the manuscript. VD: conception and design of the study; collection, analysis and interpretation of data; drafting of the manuscript; final approval of the manuscript. DA: conception and design of the study; collection, analysis and interpretation of data; final approval of the manuscript. JT: conception and design of the study; collection, analysis and interpretation of data; final approval of the manuscript. TW: collection, analysis and interpretation of data; final approval of the manuscript. LvE: analysis and interpretation of data; drafting of the manuscript; final approval of the manuscript. JJB: conception and design of the study; revising of the manuscript; final approval of the manuscript. MJS: conception and design of the study; revising of the manuscript; final approval of the manuscript. NAM: conception and design of the study; collection, analysis and interpretation of data; drafting of the manuscript; final approval of the manuscript.
Funding DPL is supported by the National Health and Medical Research Council of Australia (Grant 1016627), the National Heart Foundation of Australia (Grant O10A 5372), and the Royal Australasian College of Physicians, and is the recipient of the Earl Bakken Electrophysiology Fellowship. DA is 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. TW is supported by a Research Fellowship of the European Society of Cardiology.
Competing interests The department of Cardiology received research grants from Biotronik, Medtronic, Boston Scientific Corporation, St Jude Medical, Lantheus Medical Imaging and GE Healthcare. VD received consultant fees from St Jude Medical. The remaining authors have no competing interest to disclose.
Provenance and peer review Not commissioned; internally peer reviewed.