Efficacy of beta-blockers in heart failure patients with atrial fibrillation: An individual patient data meta-analysis Dipak Kotecha, MD PhD on behalf of the
Selection of slides presented at the European Society of Cardiology Congress 2014 (Barcelona) Clinical Trial Update Hot Line: Stable CAD and atrial fibrillation 2 nd September 2014
Disclosures/Conflicts of interest Beta-blockers in Heart Failure Collaborative Group: The majority of the group have received speaker fees, honoraria or grant support from pharmaceutical companies involved in beta-blocker therapies. Personal: Honoraria/research grants; Menarini Farmaceutica. Steering committee lead for BB-meta-HF and the RATE-AF trial. Funding: Investigator-driven. Administrative financial support from Menarini Farmaceutica and data extraction support from GlaxoSmithKline. DK is funded by the National Institute for Health Research (NIHR); the opinions herein do not represent the views of the NIHR or the UK Department of Health.
Beta-blockers in heart failure Beta-blocker therapy has a class 1A recommendation for symptomatic heart failure (HF) due to reduced left-ventricular ejection fraction (LVEF). Uptake in clinical practice remains sub-optimal, with those at the greatest risk of death less likely to receive therapy. There have also been concerns over treatment efficacy in certain under-represented groups, notably women, the elderly and those with atrial fibrillation (AF). * Lee et al. JAMA. 2005;294:1240
Heart failure and atrial fibrillation HF and AF are two emerging epidemics of the 21 st century. AF is present in 14-50% of heart failure patients with symptoms and is closely related to NYHA class. HF patients with AF have even higher rates of death and hospitalisation, regardless of which arises first. Beta-blocker therapy in AF is advocated in current heart failure guidelines due to the benefit seen predominantly in patients with sinus rhythm. * Maisel & Stevenson. Am J Cardiol. 2003;91:2D
Methods Randomised controlled trials Reporting mortality as a major trial endpoint Unconfounded head-to-head Planned >6m follow-up >300 patients (accounts for >95% of eligible RCT participants) Pooling of individual data from 18,254 heart failure patients randomised to beta-blockers or placebo, according to a published extraction and analysis plan. * Excluded from sinus rhythm versus AF analysis due to study exclusion criteria Kotecha et al. Syst Rev. 2013;2:7
Individual patient data meta-analysis Considered the gold-standard of meta-analysis* Appropriately combine original data, thereby improving data quality. Inclusion of outcomes not originally reported. Robust examination of sub-groups with enhanced sample size. Full time-to-event analyses and generation of hazard ratios adjusted for individual baseline covariates. Stratified one-stage Cox proportional hazards model adjusted for age, gender, LVEF, heart rate and use of ACEi/ARB, presented as hazard ratios (HR) and 95% CI, censored at 40 months (3.3 years). Intention-to-treat; range of sensitivity and exploratory analyses. * Stewart & Tierney. Eval Health Prof. 2002;25:76; Simmonds et al. Clin Trials. 2005;2:209
Pooled baseline characteristics 18,254 individual RCT participants from 10 trials Sinus rhythm 13,946 (76.4%) AF/Flutter 3,066 (16.8%) Other rhythms 1,242 (6.8%) Heart block/paced 1,124 (6.2%) Missing ECG 118 (0.6%) No differences in any group between those allocated to beta-blockers or placebo Heart rhythm according to baseline ECG
Pooled baseline characteristics Characteristic Sinus rhythm (n=13,946) Atrial fibrillation (n=3,066) Age, median years (IQR) 64 (54-71) 69 (60-74) Women 25% 19% Diabetes mellitus 25% 23% Years with HF diagnosis, median (IQR) 3.0 (1.0-6.0) 3.0 (1.0-7.0) LVEF, median (IQR) 0.27 (0.21-0.33) 0.27 (0.22-0.33) NYHA class III/IV 63% 72% Systolic BP, median mmhg (IQR) 123 (110-140) 127 (113-140) Heart rate, median bpm (IQR) 80 (72-88) 81 (72-92) ACEi or ARB 95% 95% Any diuretic therapy 85% 94% Digoxin 53% 84% Oral anticoagulation 26% 58%
Mortality according to baseline rhythm Number of deaths (%) Sinus rhythm Atrial fibrillation All reported deaths 1 2,237 / 13,946 (16%) 633 / 3,066 (21%) Cause of death (% of group): Sudden death 927 (41%) 231 (37%) Heart failure 539 (24%) 184 (29%) Acute myocardial infarction 126 (6%) 13 (2%) Stroke 43 (2%) 27 (4%) Other cardiac/vascular 158 (7%) 49 (8%) Non-cardiovascular/unknown 444 (20%) 129 (20%) Deaths during study period 2 2,021 / 13,946 (14%) 556 / 3,066 (18%) 1. Mean 1.5 years until death or censoring (SD 1.1). 2. Mean 1.4 years until death or censoring (SD 1.1).
Hospitalisation according to baseline rhythm Hospitalisation type Sinus rhythm Atrial fibrillation CV-hospitalisation: Percentage with 1 or more admission 26% 29% Average number of admissions per patient 0.45 (range 0-16) 0.49 (range 0-14) Annualised hospitalisation rate per patient 0.52/year 0.60/year Average length of first five admissions Mean 9.7, median 6 days Mean 11.9, median 8 days HF-related hospitalisation: Percentage with 1 or more admission 16% 21% Average number of admissions per patient 0.30 (range 0-16) 0.36 (range 0-14) Annualised hospitalisation rate per patient 0.36/year 0.41/year Average length of first five admissions Mean 9.8, median 6.5 days Mean 12.0, median 8 days
Efficacy of beta-blockers for preventing death Unadjusted Kaplan-Meier survival (includes all reported deaths). Hazard ratios (HR) derived from the adjusted one-stage Cox model.
Efficacy of beta-blockers for preventing death Unadjusted Kaplan-Meier survival (includes all reported deaths). Hazard ratios (HR) derived from the adjusted one-stage Cox model.
Mortality outcomes Outcome Events/ sample size Sinus rhythm Beta-blockers versus placebo Atrial fibrillation Beta-blockers versus placebo Interaction AF versus sinus rhythm HR (95% CI) p-value HR (95% CI) p-value p-value All-cause mortality (all reported deaths) 2870/17009 0.73 (0.67, 0.80) <0.001 0.97 (0.83, 1.14) 0.73 0.002 All-cause mortality (study period only) 2577/17009 0.73 (0.67, 0.80) <0.001 0.93 (0.79, 1.10) 0.43 0.01 Cardiovascular death (all reported deaths) 2297/17009 0.72 (0.65, 0.79) <0.001 0.92 (0.77, 1.10) 0.35 0.02 Hazard ratios derived from the one-stage Cox regression model, adjusted for age, gender, baseline LVEF, heart rate and use of ACEi/ARB.
Sensitivity/exploratory analyses Outcome Sinus rhythm Beta-blockers versus placebo Atrial fibrillation Beta-blockers versus placebo Interaction AF versus sinus rhythm HR (95% CI) p-value HR (95% CI) p-value p-value Additional adjustment for digoxin and oral anticoagulation 0.73 (0.67, 0.80) <0.001 0.97 (0.83, 1.14) 0.75 0.002 Exclusion of BEST 0.66 (0.60, 0.74) <0.001 1.03 (0.86, 1.24) 0.74 <0.001 Exclusion of CAPRICORN 0.72 (0.66, 0.79) <0.001 0.98 (0.83, 1.15) 0.81 0.002 Censor at 770 days 0.72 (0.66, 0.79) <0.001 1.00 (0.84, 1.18) 0.98 <0.001 Censor at 365 days 0.69 (0.61, 0.77) <0.001 0.97 (0.79, 1.19) 0.75 0.005
Sensitivity/exploratory analyses Heterogeneity: I 2 =56%, p=0.016 Heterogeneity: I 2 =0%, p=0.65 Two stage Cox regression model, adjusted for age, gender, baseline LVEF, heart rate and use of ACEi/ARB. Includes all reported deaths.
Sensitivity/exploratory analyses Sub-group analysis of all reported deaths for patients in AF at baseline:
Sensitivity/exploratory analyses Sub-group analysis of all reported deaths for patients in AF at baseline:
Hospitalisation, composite outcomes & stroke Outcome Events/ sample size Sinus rhythm Beta-blockers versus placebo Atrial fibrillation Beta-blockers versus placebo Interaction AF versus sinus rhythm HR (95% CI) p-value HR (95% CI) p-value p-value First CV hospitalisation 4374/16644 0.78 (0.73, 0.83) <0.001 0.91 (0.79, 1.04) 0.15 0.05 First HF-related hospitalisation 2872/16644 0.71 (0.65, 0.77) <0.001 0.91 (0.78, 1.07) 0.26 0.005 Death or CV hospitalisation 5670/16644 0.76 (0.72, 0.81) <0.001 0.89 (0.80, 1.01) 0.06 0.01 CV-death or HFhospitalisation * 4151/16644 0.70 (0.65, 0.75) <0.001 0.90 (0.79, 1.03) 0.13 0.001 Non-fatal stroke 296/16644 1.02 (0.78, 1.32) 0.91 1.04 (0.66, 1.63) 0.87 0.94 MDC does not contribute to hospitalisation or incident stroke. All deaths within study period only. * Outcome was not pre-specified.
Strengths & limitations Individual patient data from large, high-quality RCTs Near totality of randomised data (>95% eligible) Improved data quality, harmonised across trials Additional mortality outcomes Ability to perform adjusted time-to-event analyses Largest analysis of treatment efficacy in HF with AF Limited to data collected in the individual trials; retrospective AF group included atrial flutter (approximately 4%) Grouped according to baseline ECG; no data for AF type Missing data (minimal impact on this analysis) Insufficient patients with preserved LVEF for separate analysis of treatment efficacy (1.8% with LVEF 0.50)
Summary HF patients with AF have more hospital admissions, longer length of stay and higher death rates compared to sinus rhythm. For HF patients with reduced LVEF in sinus rhythm, there were clear benefits associated with beta-blocker therapy (27% reduction for incidence of death over 3.3 years). Beta-blockers in patients with concomitant AF had no significant impact on all-cause mortality, CV mortality, CV hospitalisation or HF-related hospitalisation. Our results dispute the preferential use of beta-blockers compared to other rate-control medications and suggest that current guideline recommendations should be revised.
Thank you for your attention The full article is now available in the Lancet http://www.thelancet.com/ journals/lancet/onlinefirst