Eva M. Lonn Sherryn Rambihar Peggy Gao Florian F. Custodis Karen Sliwa Koon K. Teo Salim Yusuf Michael Böhm

Transcription

1 Clin Res Cardiol (14) 3: DOI.7/s ORIGINAL PAPER Heart rate is associated with increased risk of major cardiovascular events, cardiovascular and all-cause death in patients with stable chronic cardiovascular disease: an analysis of ONTARGET/TRANSCEND Eva M. Lonn Sherryn Rambihar Peggy Gao Florian F. Custodis Karen Sliwa Koon K. Teo Salim Yusuf Michael Böhm Received: 8 October 13 / Accepted: 21 November 13 / Published online: December 13 Ó Springer-Verlag Berlin Heidelberg 13 Aims Heart rate was proposed as an emergent cardiovascular (CV) risk factor. Previous studies have shown associations between increased heart rate and CV risk in various populations. We aimed to evaluate the prognostic relevance of heart rate in a large contemporaneous medically optimized cohort of patients with stable chronic CV disease. Methods and results In a post hoc analysis of the ON- TARGET/TRANSCEND trials, we evaluated associations between baseline and average heart rate in trial with CV risk in 31, 531 patients followed for a median of 5 years. The primary outcome, major vascular events (MVE), was a composite of CV death, myocardial infarction (MI), stroke, and congestive heart failure (CHF). Pre-specified secondary outcomes included all-cause death and the individual components of the primary outcome. Associations between heart rate and outcomes were computed with heart rate as a continuous variable, baseline heart rate [7 vs B7 bpm, For the ONTARGET/TRANSCEND Investigators. Clinical Trial Registration: URL: Unique Identifier NCT E. M. Lonn S. Rambihar K. K. Teo S. Yusuf Division of Cardiology, Department of Medicine, McMaster University, Hamilton, ON, Canada E. M. Lonn P. Gao K. K. Teo S. Yusuf Population Health Research Institute, Hamilton, ON, Canada F. F. Custodis M. Böhm (&) Klinik für Innere Medizin III, Kardiologie, Angiologie und Internistische Intensivmedizin, Universitätsklinikum des Saarlandes, Kirrberger Str. 1, Homburg/Saar, Germany michael.boehm@uks.eu K. Sliwa Hatter Institute for Cardiovascular Research in Africa, Cape Town, South Africa and across heart rate quintiles, adjusting for other markers of risk, beta-blocker and non-dihydropyridine calcium channel blocker use. For each bpm increase in baseline and average heart rate, we observed a significant increase in risk of MVE, CV death, CHF and all-cause death. There was a continuous relationship between MVE and baseline and, more importantly, average in-trial heart rate, with no observed threshold. MVE, CV death, stroke, CHF, and allcause death increased across heart rate quintiles. There was no association between MI and HR. Results were consistent in clinically relevant subgroups. There were modest but significant improvements in C-statistic and in statistical measures of model calibration for models that included heart rate for MVE, CV death, CHF and all-cause death. Conclusions This large study examined and quantitated associations between heart rate and CV events in a contemporary medically optimized population with stable CV disease. Resting and, in particular, in-trial average heart rate are independently associated with significant increases in CV events and all-cause death. Keywords Heart rate Coronary artery disease Population health Epidemiology Introduction A considerable number of epidemiological studies have reported strong associations between elevated heart rate and cardiovascular (CV) risk, independent of other major risk factors [1]. This association was observed in healthy populations among men and women [2 4], in hypertensive subjects [5 7], patients with coronary artery disease [8 11], patients after myocardial infarction (MI) [12 14] and in those with left ventricular dysfunction [14] after MI and/or

2 1 Clin Res Cardiol (14) 3: heart failure (HF) [15] with reduced ejection fraction. Heart rate is associated with atherosclerosis in experimental models [16, 17] and in studies on human subjects [18]. In spite of the substantial existent body of evidence linking elevated heart rate to increased CV risk, knowledge gaps remain. Thus, many studies were conducted prior to the wide spread use of various cardiac and vascular protective therapies [2 6] and their findings require confirmation in contemporary settings. Almost all current data are based on studies of associations between heart rate measured at a single point in time and subsequent CV events. Heart rate may vary, however, substantially over time. To date, there are limited data on associations between multiple heart rate measurements over prolonged periods of time and CV outcomes. Finally, the value of heart as a biomarker of risk beyond measures of associations is not well studied. Therefore, we evaluated the associations with major clinical events and the prognostic relevance of baseline and in-trial resting heart rate in the Ongoing Telmisartan Alone and in Combination with Ramipril Global Endpoint Trial (ONTARGET) and Telmisartan Randomized Assessment Study in ACE Intolerant Subjects with Cardiovascular Disease (TRANSCEND) trials [19, ], in a large, contemporary population with chronic stable high-risk CVD followed for a long period of time. Methods Study design and study population This is a post hoc analysis of the role of heart rate in the prediction of major vascular events (MVE) in the ON- TARGET and TRANSCENDS trials. The design of the ONTARGET and TRANSCEND trials has been described previously [19, ]. In ONTARGET, men and women 55 years or older with coronary, peripheral, or cerebrovascular disease, or diabetes with end organ damage, were enrolled in a multicentre, double-blind randomized trial with three study arms: mg of ramipril daily, 8 mg of telmisartan daily, or their combination. Patients with symptoms of HF were not eligible. Patients who could not tolerate ACE inhibitors were randomly assigned to receive either 8 mg of telmisartan daily or placebo in TRAN- SCEND. The primary outcome was a composite of MVE, including death from CV causes, MI, stroke, or hospitalization for HF. Secondary outcomes included all-cause death and the individual components of the primary outcome. This analysis is based on the primary outcome measures. All endpoints were evaluated by an independent adjudication committee [19]. Between November 1 and June 4, 25,6 patients were enrolled in ONTARGET and 5,927 in TRANSCEND. Follow-up visits occurred at 6 weeks, at 6 months and thereafter every 6 months until the last scheduled visit, with a median follow-up of 56 months. At each of the visits, resting heart rate was measured in duplicate in sitting position after 3 min of rest using an automated validated device (OMRON: Model HEM 757; Omron Corp., Japan). In-trial heart rate (at baseline and at follow-up visits) was averaged using all per protocol heart rate measurements in the trial. Statistical analysis Baseline characteristics of the study population were compared by quintiles of baseline resting heart rate. Continuous data are presented as means and standard deviations and categorical data as frequencies and percentages. Differences in clinical characteristics were analyzed by ANOVA for continuous variables and v 2 tests for categorical variables. Associations between baseline and intrial HR with the first occurrence of primary and secondary outcomes were computed with heart rate as a continuous variable per bpm heart rate increments, by heart rate quintiles, and in pre-specified heart rate ranges\, 59, 6 69, 7 79, 8 89, 9 99 and C beats per minute (bpm). In addition, analyses were performed with heart rate as a dichotomous variable, above and below median heart rate and above and below 7 bpm as used in previous studies [12]. All time-to-event regression analyses were conducted based on Cox proportional hazards models. We checked the assumption of proportionality of hazard with the log[-log(survival)] plot and by the time-dependent covariate test. The proportional hazard assumption was satisfied for all outcomes evaluated. Unadjusted, age- and sex-adjusted and fully multivariable adjusted hazard ratios (HRs) and 95 % confidence intervals (CIs) were estimated. Full adjustment was performed using additional markers of risk (diabetes, hypertension, hypercholesterolemia, current smoking, baseline creatinine), the HOPE risk score (HOPE risk score = age? [if patient is male]?.458 [if current smoker]? [if history of hypertension]? [if left ventricular hypertrophy]? [if history of diabetes]?.343 [if history of stroke]? [if history of peripheral arterial disease]? 1836 [if history coronary artery disease]?.4356 [if microalbuminuria) [19], shown to reliably discriminate risk in a similar population, baseline use of beta-blocker and non-dihydropyridine calcium channel blocker use and stratified for treatment allocation to telmisartan or placebo in TRANSCEND. In a sensitivity analysis, patients with baseline atrial fibrillation were excluded. Kaplan Meier curves were used to estimate cumulative event rates by baseline and in-trial heart rate quintiles and were compared by log rank models.

3 Clin Res Cardiol (14) 3: We further compared test characteristics of the multivariate regression models with and without addition of baseline and in-trial heart rate. Discrimination was assessed with the Harrell C index. The C index for the extended multivariable model which included heart rate was compared with the C index for the multivariable model without heart rate in Cox models using the bootstrapping method to obtain CIs for the adjusted C index. The models with and without heart rate were also compared through the Akaike information criteria (AIC) and the Bayes information criteria (BIC). The goodness of fit for each model was compared by the use of the O Quigley likelihood ratio R 2 measure. Model calibration was compared with the likelihood ratio v 2 statistic. A two-tailed p \.5 was considered statistically significant. Pre-specified subgroup analyses of clinically relevant patient subsets defined by age, sex, history of diabetes, hypertension, coronary artery disease, baseline use of beta-blocker and non-hydropyridine calcium channel blocker use, HOPE risk score, baseline activity level, sinus rhythm vs atrial fibrillation and enrollment in ONTARGET vs. TRANSCEND were performed, and statistical tests for interaction were done in the Cox regression analysis. Analysis was performed with SAS version 8.2 (SAS Institute, Cary, NC, USA). Results Study population Baseline and in-trial heart rate were available for 31,531 patients enrolled in ONTARGET/TRANSCEND (99.9 %). Baseline characteristics by baseline heart rate quintiles are shown in Table 1. Overall, % of patients were women, 75 % had a history of CAD, 7 % of hypertension, 37 % of diabetes, 38 % of hypercholesterolemia and 12 % reported current smoking. Patients mean age was 66.5 years, mean baseline heart rate was 68.1 ± 12.2 bpm and mean blood pressure was 142/82 ± 17/ mmhg. In-trial heart rate was 69.1 ± 8.8 bpm. At baseline, patients with lower heart rates were more likely to have known CAD, whereas patients with higher heart rates were more likely to be women, and had higher reported rates of diabetes, hypertension, hypercholesterolemia, and current smoking. At baseline, 76 % of patients were taking aspirin, 9 % clopidogrel, 57 % a betablocker, % a non-dihydropyridine calcium channel blocker and 6 % a statin. Patients with lower heart rates were more likely to be treated with aspirin, clopidogrel, betablockers and statins and less likely to be on diuretics and nondihydropyridine calcium channel blockers. Table 1 Baseline characteristics by quintiles of baseline resting heart rate Heart rate Quintile 1 B58 Quintile Quintile Quintile Quintile 5 C79 p value Number of patients (%) 6,762 (21) 6,396 () 6,146 (19) 6,6 (21) 5,587 (19) Female (%) 1,425 (21.1) 1,733 (27.1) 1,885 (.7) 2,283 (34.4) 2,47 (36.6) \.1 Age ± SD ± ± ± ± ± 7.33 \.1 Heart rate ± SD ± 5 66 ± ± ± ± 9.72 \.1 Systolic BP (mmhg) ± SD 142 ± ± ± ± ± Diastolic blood pressure (mmhg) ± SD 8 ± ± 82 ± 83 ± 84 ± \.1 Medical history and risk factors CAD (%) 5,8 (86.5) 5,129 (8.2) 4,621 (75.2) 4,539 (68.4) 3,369 (6.3) \.1 Stroke/TIA (%) 1,142 (16.9) 1,218 (19.) 1,299 (21.1) 1,5 (23.2) 1,445 (25.9) \.1 PAD (%) 684 (.1) 773 (12.1) 866 (14.1) 952 (14.3) 861 (15.4) \.1 Hypertension (%) 4,396 (6) 4,334 (67.8) 4,2 (7.) 4,849 (7) 4,245 (76.) \.1 Diabetes (%) 1,646 (24.3) 2,2 (32.9) 2,327 (37.9) 2,8 (4) 2,831 (.7) \.1 Current smoker (%) 644 (9.5) 7 (.9) 756 (12.3) 89 (13.4) 816(14.6) \.1 Former smoker (%) 3,897 (57.6) 3,353 (52.4) 3,88 (.2) 3,29 (45.6) 2,456 (4) \.1 Hypercholesterolemia (%) 1,948 (28.8) 2,215 (34.6) 2,354 (38.3) 2,886 (43.5) 2,564 (45.9) \.1 Medications at randomization Aspirin (%) 5,583 (82.6) 5,59 (79.1) 4,685 (76.2) 4,779 (7) 3,711 (66.4) \.1 Clopidogrel (%) 673 (.) 544 (8.5) 519 (8.4) 538 (8.1) 397 (7.1) \.1 Beta-blocker (%) 5,7 (78.5) 4,187 (65.5) 3,38 (5) 3,95 (46.6) 2,59 (36.9) \.1 Non-dihydropyridine CCB (%) 544 (8.) 5 (8.6) 656 (.7) 679 (.2) 655 (11.7) \.1 Statin (%) 4,765 (7) 4,128 (64.5) 3,743 (6.9) 3,686 (55.5) 2,724 (48.8) \.1 Diuretics (%) 1,646 (24.3) 1,727 (27.) 1,763 (28.7) 2,59 (3) 1,9 (34.4) \.1

4 152 Clin Res Cardiol (14) 3: Outcomes The primary outcome, MVE, occurred in 5,188 patients (3.82 %/year), CV death in 2,269 (8 %/year), MI in 1,558 (1.1 %/year), HF hospitalizations in 1,343 (.95 %/ year), stroke in 1,395(.99 %/year), and all-cause death in 3,779 (2.63 %/year) patients at a median follow-up of 56 months. The association of baseline heart rate as a continuous variable with outcomes is shown in Table 2. For each bpm increment in baseline resting heart rate, the risk of subsequent MVE increased by 12 % in unadjusted analyses, and by 13 % in multivariate analyses; the risk of CV death increased by 19 % in unadjusted and % in multivariate analyses; the risk of HF increased by 18 % in unadjusted and by 18 % in multivariate analyses, and the risk of all-cause death increased by 16 % in unadjusted and by % in multivariate analyses (p \.1 for all). Baseline heart rate was modestly associated with risk for subsequent stroke in unadjusted and in age- and sex-adjusted analyses, but not in multivariable models, while no significant associations with MI were seen. The associations of heart rate and CV outcomes were substantially strengthened when in-trial heart rate was considered (Table 2). For each bpm increase in in-trial heart rate, there was a % increase in risk of MVE, % increase in risk of CV death, % increase in risk of hospitalized HF, and % increase in risk of all-cause death in unadjusted, age and sex and multivariate adjusted models (Table 2). Associations with stroke were also strengthened, % increased risk in the unadjusted analysis, 23 % increased risk after adjustment for age and sex and % increase in risk in multivariable models. There was no significant association between in-trial heart rate and subsequent MI. The relationship between baseline and in-trial resting heart rate and MVE, CV death, HF and all-cause death was linear with no observed threshold within the studied HR ranges (Fig. 1A, B). Findings were similar for associations between baseline and in-trial standard deviation change in heart rate and CV events. Table 2 Hazards ratios for major vascular outcomes by bpm increment in baseline and in-trial heart rate Heart rate MVE (95 % CI) CV death (95 % CI) MI (95 % CI) Stroke (95 % CI) Hospitalized HF (95 % CI) All-cause death (95 % CI) Model 1 Baseline 1.12 (9 1.14) 1.19 ( ) 1 (.97 6) 7 (3 1.12) 1.18 ( ) 1.16 ( ) p \.1 p \.1 p =.4932 p =.11 p \.1 p \.1 In trial 1.23 ( ) 1.36 ( ).96 (.91 2) 1. ( ) 1.44 (1.36 3) 1.34 ( ) p \.1 p \.1 p =.95 p \.1 p \.1 p \.1 Model 2 Baseline 1.13 ( ) 1.19 ( ) 3 (.99 7) 8 (3 1.12) 1.18 ( ) 1.17 ( ) p \.1 p \.1 p =.1392 p =.4 p \.1 p \.1 In trial 1.26 ( ) 1.41 ( ) (.94 5) 1.23 ( ) 1.47 (1.39 6) 1.39 ( ) p \.1 p \.1 p =.872 p \.1 p \.1 p \.1 Model 3 Baseline 1. (8 1.12) 1.16 ( ) 4 (.99 8) 1 (.97 6) 1.18 ( ) 1.14 ( ) p \.1 p \.1 p =.94 p = 726 p \.1 p \.1 In trial 1.22 ( ) 1.33 ( ) (.94 6) 1.12 (5 1.) 1.48 (1.39 8) 1.33 ( ) p \.1 p \.1 p =.9533 p =.6 p \.1 p \.1 Model 4 Baseline 1.12 ( ) 1.18 ( ) 5 (1 9) 4 ( 9) 1. ( ) 1.15 ( ) p \.1 p \.1 p =.248 p =.752 p \.1 p \.1 In trial 1.26 ( ) 1.38 ( ) 3 (.97 9) 1.17 ( ) 5 ( ) 1.35 (1. 1.) p \.1 p \.1 p =.3762 p \.1 p \.1 p \.1 Model 1: unadjusted Model 2: adjusted for age, sex Model 3: adjusted for the following baseline variables: age, sex, diabetes, hypertension, dyslipidemia, current smoking, creatinine, use of betablocker, use of diltiazem/verapamil and stratified by treatment allocation in the trial Model 4: adjusted for the baseline HOPE risk score, baseline use of beta-blocker, baseline use of diltiazem/verapamil and stratified by treatment allocation in the trial MVE major vascular events, CV cardiovascular, MI myocardial infarction, HF heart failure

5 Clin Res Cardiol (14) 3: Similar results were seen in the analysis by quintiles, with increasing risk of MVE, CV death, hospitalized HF, all-cause death and stroke across quintiles of baseline and in-trial resting HR, but lack of increase in risk for MI with increasing baseline and in-trial heart rate (Fig. 2A, B). Patients with baseline and in-trial heart rate above median and those with heart rate C7 bpm had a higher risk of MVE, CV death, hospitalized HF and all-cause death compared to those with HR below median and below 7 bpm, respectively (results available on request). The addition of heart rate resulted in modest, but statistically significant improvements in measures of model discrimination and calibration, for MVE, CV death, hospitalization for HF, all-cause death (Table 3). In-trial, but not baseline heart rate, improved these model characteristics for stroke. There were no improvements in measures of model discrimination and calibration for MI with the addition of either baseline or in-trial heart rate. Comparisons of key subgroups showed similar effects of baseline and in-trial heart rate on MVE in most clinically relevant subgroups (Fig. 3a, b). Significant interactions were noted in subgroups defined by age, history of diabetes, coronary artery disease and use of beta-blockers suggesting a stronger association in younger patients, those without prior history of diabetes, those already on betablocker therapy at baseline and particularly those with known coronary artery disease at baseline. There were 1,52 patients with baseline atrial fibrillation. The associations between baseline and average heart rate and CV risk did not differ after exclusion of patients with baseline atrial fibrillation. Discussion Heart rate and outcomes The major finding of our study is that heart rate is an independent predictor of MVE, CV death, hospitalization for HF and all-cause death in a contemporaneous population of patients with chronic stable CVD or diabetes with end organ damage. These associations were robust, with no apparent threshold and persisted in analyses adjusted for multiple CV risk factors, as well as baseline beta-blocker and non-dihydropyridine CCB use. Associations between baseline heart rate and stroke were significant in crude analyses and after adjusting for age and sex alone, but not after adjusting for other prognostic factors, while in-trial heart rate was significantly associated with stroke even after multivariate adjustment. All associations were substantially strengthened when considering in-trial heart rate, derived from multiple measurements over time. Of note, we found no significant associations between baseline or in-trial heart rate and risk for MI. Our study shows also for the first time in a large dataset that the addition of heart rate, measured at one point in time (baseline heart rate) and even more so when measured repeatedly over time (in-trial heart rate), improved the discrimination and precision of prognostic models for major CV events, with the exception of MI. Associations between baseline resting heart rate and mortality were described in previous older studies [2 8] and were confirmed more recently in contemporary populations, including patients with chronic coronary disease [11], hypertension [7], left ventricular dysfunction [14] and HF [15] and after stroke [21]. In spite of the existent large body of knowledge in this field, our study provides important additional information. First, the large number of CV events in the ONTARGET/TRANSCEND [19, ] allows the quantitation of associations between heart rate and individual CV outcomes and in subgroups. Second, we provide novel information on associations of serial heart rate measurements (in-trial heart rate) and CV outcomes, for which there are very limited previous data. Third, we report not only on associations between heart rate and outcomes, but also on additional measures of the prognostic relevance of heart rate in the prediction of CV disease. Our findings raise the hypothesis of possible benefit for therapies directed specifically towards heart rate lowering, what, however, must be scrutinized in prospective randomized clinical trials. There are many hypotheses of how heart rate may affect mortality and CV risk [22]. It is possible that beyond sympathetic activity heart rate itself is a contributing factor to atherogenesis, CV death and all-cause death, as well as reflecting other pathophysiologic mechanisms which increase risk of CV events and death. High entry heart rates increase myocardial oxygen demand [23], increase shear stress on the endothelium [22], and cause accelerated coronary and carotid atherosclerosis [17, 18] and plaque rupture [24]. Increased sympathetic tone may promote lifethreatening ventricular arrhythmias and sudden cardiac death [, 25]. High heart rates are also independently associated with development of cardiac risk factors such as HTN [26], diabetes, high waist circumference and the metabolic syndrome [27] and strongly influenced by smoking and lack of physical activity [27]. Irrespective of the mechanism, heart rate has long been a therapeutic target in coronary disease and HF. In CAD, beta blockade decreases myocardial oxygen demand, lengthens diastole and promotes myocardial perfusion, and is associated with decreased infarct size [8]. In HF, betablockers protect the myocardium from excessive catecholamine stimulation [28]. More recently ivabradine, a specific inhibitor of the f-currents in the sinus node cells provides pure heart rate reduction and was shown to

6 154 Clin Res Cardiol (14) 3: A a Major vascular Events b Cardiovascular Death c Myocardial Infarction % of patientswith event < < < d Stroke e Hospitalized Heart Failure f All-Cause Death < < < B a Major vascular Events b Cardiovascular Death c Myocardial Infarction < < < d Stroke e Hospitalized Heart Failure f All-Cause Death 6 < < >= < Fig. 1 Association of hazard ratios to baseline heart rate (A), in-trial heart rate (B) for major vascular events (a), cardiovascular death (b), myocardial infarction (c), stroke (d), hospitalized heart failure (e) and all-cause death (f)

7 Clin Res Cardiol (14) 3: Major Vascular Events : Heart Rate 58 : Heart Rate >58 & 64 : Heart Rate >64 & 7 : Heart Rate >7 & 78 : Heart Rate > A. 5 5 a : Heart Rate 58 : Heart Rate >58 & 64 : Heart Rate >64 & 7 : Heart Rate >7 & 78 : Heart Rate >78 p< p= b Cardiovascular Death d Stroke e HospitalizedHeartFailure f : Heart Rate 58 : Heart Rate >58 & 64 : Heart Rate >64 & 7 : Heart Rate >7 & 78 : Heart Rate > : Heart Rate 58 : Heart Rate >58 & 64 : Heart Rate >64 & 7 : Heart Rate >7 & 78 : Heart Rate >78 p< p< c Myocardial Infarction : Heart Rate 58 : Heart Rate >58 & 64 : Heart Rate >64 & 7 : Heart Rate >7 & 78 : Heart Rate > All Cause Death : Heart Rate 58 : Heart Rate >58 & 64 : Heart Rate >64 & 7 : Heart Rate >7 & 78 : Heart Rate >78 p=.6718 p< a Major Vascular Events b Cardiovascular Death c Myocardial Infarction : Heart Rate 58 : Heart Rate >58 & 64 : Heart Rate >64 & 7 : Heart Rate >7 & 78 : Heart Rate > B p< : Heart Rate 58 : Heart Rate >58 & 64 : Heart Rate >64 & 7 : Heart Rate >7 & 78 : Heart Rate > p< : Heart Rate 58 : Heart Rate >58 & 64 : Heart Rate >64 & 7 : Heart Rate >7 & 78 : Heart Rate > P= d Stroke e Hospitalized Heart Failure f All-Cause Death : Heart Rate 58 : Heart Rate >58 & 64 : Heart Rate >64 & 7 : Heart Rate >7 & 78 : Heart Rate > p< : Heart Rate 58 : Heart Rate >58 & 64 : Heart Rate >64 & 7 : Heart Rate >7 & 78 : Heart Rate > p< : Heart Rate <=58 : Heart Rate >58 & <=64 : Heart Rate >64 & <=7 : Heart Rate >7 & <=78 : Heart Rate > p< Fig. 2 Kaplan Meier estimates of baseline heart rate (A), in-trial heart rate (B) for major vascular events (a), cardiovascular death (b), myocardial infarction (c), stroke (d), hospitalized heart failure (e) and all-cause death (f)

8 156 Clin Res Cardiol (14) 3: Table 3 C-statistic, AIC, BIC, Quigley R 2 and v 2 for major vascular events with and without baseline and in-trial heart rate in multivariable model C index (95 % CI), p AIC BIC O Quigley R 2 LR v 2, p Major vascular events Model ( ) 87,824 87, Model 3? baseline HR.631 ( ), \.1 87,77 87, , \.1 Model 3? in-trial HR.633 ( ), \.1 87,721 87, , \.1 Cardiovascular death Model ( ) 38,786 38, Model 3? baseline HR.673 ( ), \.1 38,7 38, , \.1 Model 3? in-trial HR.675 ( ), \.1 38,71 38, , \.1 Myocardial infarction Model 3.69 (94.623) 27,59 27,113.1 Model 3? baseline HR.6 (96.625), ,63 27, ,.393 Model 3? in-trial HR.6 (95.624),.7 27,65 27, ,.699 Stroke Model ( ) 24,266 24, Model 3? baseline HR.629 ( ), ,273 24, ,.982 Model 3? in-trial HR.633 ( ),.29 24,256 24, ,.1 Hospitalized heart failure Model ( ) 23,87 23, , \.1 Model 3? baseline HR.691 ( ), \.1 23,24 23,97 Model 3? in-trial HR.697 ( ), \.1 22,962 23, , \.1 All-cause death Model ( ) 64,35 64,97.7 Model 3? baseline HR.669 ( ), \.1 63,948 64, , \.1 Model 3? in-trial HR.672 ( ), \ , , \.1 Model 3: adjusted for the following baseline variables: age, sex, diabetes, hypertension, dyslipidemia, current smoking, creatinine, use of betablocker, use of diltiazem/verapamil and stratified by treatment allocation in the trial p values refer to the comparisons of the model characteristics without heart rate and with baseline heart rate and in-trial heart, respectively AIC Akaike information criteria, BIC Bayes information criteria, O Quigley R 2 O Quigley likelihood ratio R 2, LR likelihood ratio v 2, HR heart rate improve outcomes in HF patients on a background of betablockers and other therapies [29] and this drug is currently under investigation in patients with chronic stable CAD and preserved left ventricular systolic function with baseline resting heart rate [7 bpm (SIGNIfY, ISRCTN ). We observed a continuous relationship between heart rate and MVE, CV death, all-cause death and HF, with no apparent threshold. This finding differs from some previous studies, which suggested a threshold effect [2] and studies which reported a J-shaped curve [7] for the relationship between heart rate and all-cause mortality, with risk at the higher and lower ends of the heart rate spectrum. However, our data similar to reports from other studies, such as BEAUTIFUL [14], SHIFT [15] and TNT [12], identified a linear relationship with lower CV risk extending to lower heart rates and with no apparent threshold. Biologic systems are in flux, and variables such as heart rate and blood pressure vary over time. The true impact of a biological variable on the risk of a clinical endpoint is likely to be complex and dependent on multiple values and changes in the variable over time. However, many clinical studies report only associations between a single value at a particular point in time and risk. New time-varying statistical analyses take into account changing values for a given variable, for example the extended Kaplan Meier Estimator permits patients to shift from one quantile to another []. This has been used to evaluate the risk of CV disease with changes in left ventricular hypertrophy [31], and to assess the prognostic value of changes in blood pressure with CV outcomes [32]. Limitations of this approach relate primarily to the possibility of including the same individual in a higher or lower risk category at different points in time. In our study, we employed a simpler approach to account for heart rate over time, using multiple readings of

9 Clin Res Cardiol (14) 3: Primary(unadj) Men Women >65 <=65 DM No DM HTN No HTN CAD No CAD BB No BB ndhp No ndhp BB or ndhp No BB or ndhp Hope score >3.94 Hope score <=3.94 # Pts % Incidence HR <=7 HR >7 p for interaction HR <= % Incidence p for HR >7 interaction Activity No Activity ONTARGET TRANSCEND AF Sinus beats/minute HR (95% CI) HR (95% CI) Fig. 3 Subgroup analyses: risk of major vascular events by baseline (left) and in-trial heart rate (right) B7 beats/minute vs [7 beats/ minute. The plot shows hazard ratios in 95 % CIs for the cardiovascular events in pre-specified subgroups. Significant interaction was observed for age, coronary artery disease (CAD) and betablocker treatment (BB) heart rate for a given patient throughout the study, and computing the average heart rate over time. The average heart rate analysis strengthened associations between heart rate and MVE, CV death, HF, and all-cause death. This study demonstrated associations of CV events with dynamic heart rate measurements over time. Interestingly, the effects of high resting heart rates on MI in this population after a stroke, MI, diabetics with two risk factors or proven peripheral artery disease were absent or less robust. It might be premature to conclude that resting heart rate reduction in SIGNIFY is not useful, just because heart rate does not impose an added risk for MI. Although 74.5 % of the ONTARGET population [19] had CAD, in this population the association of resting heart rate to MI was different to non-cad patients (Forrest plot in Fig. 3). Herein, the hypothetical effects of heart rate on MI in CAD patients might have been diluted because poststroke patients and diabetics could be prone to develop another second event like a stroke or HF hospitalization. Furthermore, the Forrest plot presents non-adjusted data but further emphasizes the importance of the SIGNIFY data to show whether heart rate is a treatment target to reduce MIs and events in a CAD population. We describe also novel observations of improvement in statistical measures of discrimination and model calibration when heart rate is added to prognostic models. While it is compelling to contemplate potential use of heart rate as a bedside prediction tool adding to traditional risk models, we should consider the incremental benefit [33]. The C-statistic, used in our analysis, is a widely reported measure for formally assessing new CV risk prediction models [34] and confirms the overall results of this analysis.

10 158 Clin Res Cardiol (14) 3: Conclusion In conclusion, this is a large study examining and quantitating associations between heart rate and CV events in a contemporaneous well-treated population with chronic stable CV disease. Resting and average heart rate are independently associated with significant increases in MVE, CV death, CHF and all-cause death but less robust for MI. Importantly, these associations were observed even in patients treated with beta-blockers. Potential therapeutic implications of heart rate lowering in this population require further exploration in prospective clinical trials on each individual endpoint, because risk associations are of different strength. The recent availability of drugs with pure heart rate lowering effects, shown to be very safe and well tolerated, such as ivabradine, makes such research highly clinically relevant. Conflict of interest All authors obtained scientific support by Boehringer Ingelheim. The ONTARGET/TRANSCEND trial was sponsored by Boehringer Ingelheim. Ethical standard statement The study was performed according to the Declaration of Helsinki. The approval was given by the ethical committees of the study centers participating in the ONTARGET/ TRANSCEND trial. References 1. Palatini P, Julius S (4) Elevated heart rate: a major risk factor for cardiovascular disease. Clin Exp Hyperten 26: Dyer AR, Persky V, Stamler J, Paul O, Shekelle RB, Berkson DM, Lepper M, Schoenberger JA, Lindberg HA (198) Heart rate as a prognostic factor for coronary heart disease and mortality: findings in three Chicago epidemiologic studies. Am J Epidemiol 112: Kannel WB, Kannel C, Paffenbarger RS Jr, Cupples LA (1987) Heart rate and cardiovascular mortality: the Framingham study. Am Heart J 113: Gillum RF, Makuc DM, Feldman JJ (1991) Pulse rate, coronary heart disease, and death: the NHANES I epidemiologic follow-up study. Am Heart J 121: Gillman MW, Kannel WB, Belanger A, D Agostino RB (1993) Influence of heart rate on mortality among persons with hypertension: the Framingham study. Am Heart J 125: Palatini P, Thijs L, Staessen JA, Fagard RH, Bulpitt CJ, Clement DL, de Leeuw PW, Jaaskivi M, Leonetti G, Nachev C, O Brien ET, Parati G, Rodicio JL, Roman E, Sarti C, Tuomilehto J (2) Systolic Hypertension in Europe (Syst-Eur) Trial Investigators. Predictive value of clinic and ambulatory heart rate for mortality in elderly subjects with systolic hypertension. Arch Intern Med 162: Kolloch R, Legler UF, Champion A, Cooper-Dehoff RM, Handberg E, Zhou Q, Pepine CJ (8) Impact of resting heart rate on outcomes in hypertensive patients with coronary artery disease: findings from the International VErapamil-SR/trandolapril Study (INVEST). Eur Heart J 29: Shaper AG, Wannamethee G, Macfarlane PW, Walker M (1993) Heart rate, ischaemic heart disease, and sudden cardiac death in middle-aged British men. Br Heart J 7: Lee KL, Woodlief LH, Topol EJ, Weaver WD, Betriu A, Col J, Simoons M, Aylward P, Van de Werf F, Califf RM, For the GUSTO-I Investigators (1995) Predictors of -day mortality in the era of reperfusion for acute myocardial infarction: results from an international trial of 41,21 patients. Circulation 91: Zuanetti G, Mantini L, Hernández-Bernal F, Barlera S, di Gregorio D, Latini R, Maggioni AP (1999) Relevance of heart rate as a prognostic factor in myocardial infarction: the GISSI experience. Eur Heart J 1(Suppl J):H52 H Diaz A, Bourassa MG, Guertin MC, Tardif JC (5) Long-term prognostic value of resting heart rate in patients with suspected or proven coronary artery disease. Eur Heart J 26: Ho JE, Bittner V, Demicco DA, Breazna A, Deedwania PC, Waters DD () Usefulness of heart rate at rest as a predictor of mortality, hospitalization for heart failure, myocardial infarction, and stroke in patients with stable coronary heart disease (data from the Treating to New Targets [TNT] Trial). Am J Cardiol 5: Bangalore S, Messerli FH, Ou FS, Tamis-Holland J, Palazzo A, Roe MT, Hong MK () Peterson ED; CRUSADE Investigators. The association of admission heart rate and in-hospital cardiovascular events in patients with non-st-segment elevation acute coronary syndromes: results from patients in the CRUSADE quality improvement initiative. Eur Heart J 5: Fox K, Ford I, Steg PG, Tendera M, Robertson M, Ferrari R, BEAUTIFUL investigators (8) Heart rate as a prognostic risk factor in patients with coronary artery disease and left-ventricular systolic dysfunction (BEAUTIFUL): a subgroup analysis of a randomised controlled trial. Lancet 372: Böhm M, Swedberg K, Komajda M, Borer J, Ford I, Dubost- Brama A, Lerebours G, Tavazzi L, On behalf of the SHIFT investigators () Heart rate as a risk factor in chronic heart failure (SHIFT): the association between heart rate and outcomes in a randomised placebo-controlled trial. Lancet 376: Custodis F, Baumhäkel M, Schlimmer N, List F, Gensch C, Böhm M, Laufs U (8) Heart rate reduction by ivabradine reduces oxidative stress, improves endothelial function, and prevents atherosclerosis in apolipoprotein E-deficient mice. Circulation 117: Beere PA, Glagov S, Zarins CK (1984) Retarding effect of lowered heart rate on coronary atherosclerosis. Science 226: Perski A, Hamsten A, Lindvall K, Theorell T (1988) Heart rate correlates with severity of coronary atherosclerosis in young post infarction patients. Am Heart J 116: ONTARGET Investigators, Yusuf S, Teo K, Pogue J, Dyal L, Copland I, Schumacher H, Dagenais G, Sleight P, Anderson C (8) Telmisartan, Ramipril, or both in patients at high risk for vascular events. N Engl J Med 358: Telmisartan Randomised AssessmeNt Study in ACE intolerant subjects with cardiovascular Disease (TRANSCEND) Investigators, Yusuf S, Teo K, Anderson C, Pogue J, Dyal L, Copland I, Schumacher H, Dagenais G, Sleight P (8) Effects of the angiotensin-receptor blocker telmisartan on cardiovascular events in high-risk patients intolerant to angiotensin-converting enzyme inhibitors: a randomised controlled trial. Lancet 372: Böhm M, Cotton D, Foster L, Custodis F, Laufs U, Sacco R, Bath PM, Yusuf S, Diener HC (12) Impact of resting heart rate on mortality, disability and cognitive decline in patients after ischaemic stroke. Eur Heart J 33: Custodis F, Schirmer SH, Baumhäkel M, Heusch G, Böhm M, Laufs U () Vascular pathophysiology in response to increased heart rate. J Am Coll Cardiol 56:

11 Clin Res Cardiol (14) 3: Reil JC, Custodis F, Swedberg K, Komajda M, Borer JS, Ford I, Tavazzi L, Laufs U, Böhm M (11) Heart rate reduction in cardiovascular disease and therapy. Clin Res Cardiol : Heidland UE, Strauer BE (1) Left ventricular muscle mass and elevated heart rate are associated with coronary plaque disruption. Circulation 4: Palatini P, Casiglia E, Julius S, Pessina AC (1999) High heart rate: a risk factor for cardiovascular death in elderly men. Arch Intern Med 159: Palatini P, Dorigatti F, Zaetta V, Mormino P, Mazzer A, Bortolazzi A, D Este D, Pegoraro F, Milani L, Mos L, HARVEST Study Group (6) Heart rate as a predictor of development of sustained hypertension in subjects screened for stage 1 hypertension: the HARVEST Study. J Hypertens 24: Vergnaud AC, Protogerou AD, Li Y, Czernichow S, Vesin C, Blacher J, Safar ME (8) Pulse pressure amplification, adiposity and metabolic syndrome in subjects under chronic antihypertensive therapy: the role of heart rate. Atherosclerosis 199: Lohse MJ, Engelhardt S, Danner S, Böhm M (1996) Mechanisms of beta-adrenergic receptor desensitization: from molecular biology to heart failure. Basic Res Cardiol 91(suppl 2): Swedberg K, Komajda M, Böhm M, Borer JS, Ford I, Dubost- Brama A, Lerebours G, Tavazzi L, SHIFT Investigators () Ivabradine and outcomes in chronic heart failure (SHIFT): a randomised placebo-controlled study. Lancet 376: Snappin S, Iglewicz B (5) Illustrating the impact of a timevarying covariate with an extended Kaplan Meier indicator. Am Stat 59: Mathew J, Sleight P, Lonn E, Johnstone D, Pogue J, Yi Q, Bosch J, Sussex B, Yusuf S, Heart Outcomes Prevention Evaluation (HOPE) Investigators (1) Reduction of cardiovascular risk by regression of electrocardiographic markers of left ventricular hypertrophy by the angiotensin-converting enzyme inhibitor ramipril. Circulation 4: Sleight P, Redon J, Verdecchia P, Mancia G, Gao P, Fagard R, Schumacher H, Weber M, Böhm M, Williams B, Pogue J, Koon T, Yusuf S, ONTARGET investigators (9) Prognostic value of blood pressure in patients with high vascular risk in the Ongoing Telmisartan Alone and in combination with Ramipril Global Endpoint Trial study. J Hypertens 27(7): Hlatky MA, Greenland P, Arnett DK, Ballantyne CM, Criqui MH, Elkind MS, Go AS, Harrell FE Jr, Hong Y, Howard BV, Howard VJ, Hsue PY, Kramer CM, McConnell JP, Normand SL, O Donnell CJ, Smith SC Jr, Wilson PW, American Heart Association Expert Panel on Subclinical Atherosclerotic Diseases and Emerging Risk Factors and the Stroke Council (9) Criteria for evaluation of novel markers of cardiovascular risk: a scientific statement from the American Heart Association. Circulation 119: Cook N (8) Statistical evaluation of prognostic versus diagnostic models. Clin Chem 54:17 23