Impact of initial heart failure emergence on clinical outcomes of atrial fibrillation patients in the AFFIRM trial

Journal Pre-proof Impact Of Initial Heart Failure Emergence On Clinical Outcomes Of Atrial Fibrillation Patients In the AFFIRM Trial

April Slee, Sanjeev Saksena PII:

S0002-8703(19)30280-7

DOI:

https://doi.org/10.1016/j.ahj.2019.10.005

Reference:

YMHJ 5996

To appear in:

American Heart Journal

Received date:

1 June 2019

Accepted date:

6 October 2019

Please cite this article as: A. Slee and S. Saksena, Impact Of Initial Heart Failure Emergence On Clinical Outcomes Of Atrial Fibrillation Patients In the AFFIRM Trial, American Heart Journal(2019), https://doi.org/10.1016/j.ahj.2019.10.005

This is a PDF file of an article that has undergone enhancements after acceptance, such as the addition of a cover page and metadata, and formatting for readability, but it is not yet the definitive version of record. This version will undergo additional copyediting, typesetting and review before it is published in its final form, but we are providing this version to give early visibility of the article. Please note that, during the production process, errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

© 2019 Published by Elsevier.

Journal Pre-proof Impact Of Initial Heart Failure Emergence On Clinical Outcomes Of Atrial Fibrillation Patients In the AFFIRM Trial April Slee MS 1 Sanjeev Saksena MD1, 2 From the Electrophysiology Research Foundation, Warren, NJ1, and the Rutgers’-

of

Robert Wood Johnson Medical School, Piscataway, NJ2.

ro

Abbreviated Title: Heart Failure in Atrial Fibrillation

-p

Presented in part at the Annual Scientific Sessions of the European Society of

re

Cardiology, Munich, Germany, August 2018 and the Annual Scientific Sessions of the

lP

Heart Rhythm Society, San Francisco, May 2019.

na

No extramural funding was used to support this work Corresponding author: Sanjeev Saksena MD

Jo

ur

Address: 161 Washington Valley Road, Suite 201, Warren, NJ 07059 E-mail: [email protected] ; Tel: 732-302-9990; Fax: 732-302-9911 Declaration of Interest: Both the authors have no conflicts of interest with the subject of this manuscript The authors are solely responsible for the design and conduct of this study, all study analyses and drafting and editing of the paper. Word count: Abstract 250; Manuscript text including references:4801

1

Journal Pre-proof

2

ABSTRACT: Background: Heart failure (HF) emergence in atrial fibrillation (AF) patients undergoing different treatment strategies has not been studied. Methods: AFFIRM trial subjects with no history of HF, without clinical HF & normal left ventricular ejection fraction at enrollment were identified. The principal outcome was time to development of a composite of NYHA class >II HF &/or cardiovascular (CV)

of

death. It was compared for Rate and Rhythm strategies and correlated with ECG

ro

parameters on follow up (FU).

-p

Results: 1771 patients (880 Rate, 891 Rhythm) were evaluated. The principal outcome occurred in 21.4% of Rate and 16.8% of Rhythm subjects at 5 years (HR 1.32, 95%

re

confidence intervals {CI :} 1.04, 1.69, p = 0.024). HF inrement by 2 classes increased

lP

total mortality (HR 2.83, 95% CI 1.91, 4.18, p<.0001), cardiac mortality, (HR 4.27, 95%

na

CI 2.03, 9.04, p=.0001), & CV Hospitalizations (HR 3.04, 95% CI 2. 15, 4.29, p<.0001). HF emergence during FU was associated with AF (p=0.0004), ventricular rate > 80 bpm

ur

(p=0.0106) and higher frequency of recorded AF in the Rhythm arm (25-75% vs. <25%

p = <.001).

Jo

HR: 1.69, 95% CI: 1.09, 2.64, p = 0.020; >75% vs. <25% HR: 3.15, 95% CI: 1.87, 5.34,

Conclusions: 1. In AF pts without HF, symptomatic HF emergence was more frequent with Rate control than with Rhythm control, 2. HF appearance presages increased mortality risk. 3. Delaying HF emergence is associated with effective Rhythm control with documented sinus rhythm during >75% of FU visits as well as ventricular rate control.

Journal Pre-proof Key words: Heart Failure; Atrial Fibrillation; Antiarrhythmic therapy: Clinical Trials;

Jo

ur

na

lP

re

-p

ro

of

Outcomes Research, Heart Failure with Preserved Cardiac Function

3

Journal Pre-proof

4

INTRODUCTION Atrial fibrillation (AF) is a major event in the clinical course of heart failure (HF) and also adversely impacts its outcomes.(1,2) The converse, the development of new clinical HF in AF patients has not been well studied and impact of HF emergence on hard clinical outcomes in AF has not been quantified. It has often been debated which condition

of

leads to the other, but it is generally accepted that, when present, both promote a

ro

vicious cycle that leads to worsening and often rapid deterioration in the patient’s

-p

condition. (3) Rhythm and Rate control of AF in advanced HF has been the focus of

re

research efforts using both pharmacologic and non-pharmacologic approaches. (4-8) The impact of different treatment approaches on HF onset in AF patients has not been

na

lP

examined.

The AFFIRM trial (conducted 1996-2002) provided an evidence based foundation for

ur

two AF treatment strategies, namely Rate and Rhythm control.(9) It demonstrated

Jo

comparable efficacy with these two strategies with respect to patient survival, and many other secondary endpoints such as stroke, major bleeding, myocardial infarction and systemic embolism. The impact of the AF treatment strategy on symptomatic HF emergence was not reported. In this report, we studied the development of new onset HF in AF patients without prior HF, no HF symptoms or signs or left ventricular systolic dysfunction at enrolment in the AFFIRM trial. We examined this outcome in the two treatment strategies used in the AFFIRM study and its relationship, if any, to the measured cardiac rhythm and ventricular rate during the follow up period.

Journal Pre-proof

5

OBJECTIVES OF THE STUDY

The purpose of this study was to evaluate the likelihood of developing symptomatic HF in the Rate and Rhythm treatment strategy arms in the AFFIRM trial in AF patients without a prior history of HF or clinical symptoms or signs of HF and normal LV systolic

of

function at study entry. Furthermore, we examined impact of HF emergence on hard

ro

clinical outcomes, and stratified by outcomes by rhythm and frequency of AF detection

re

-p

on follow up electrocardiograms in each treatment arm

na

Patient selection in AFFIRM

lP

MATERIAL AND METHODS

AFFIRM recruited consenting patients who had prior AF that was likely to be recurrent,

ur

warranted therapy and had risk factor(s) for stroke or death. They were either ≥65 years

Jo

of age or <65 years of age with at least 1of the following additional risk factors for stroke: Hypertension; diabetes; congestive heart failure; prior stroke; prior transient ischemic attack; systemic embolism; left atrial size ≥50 mm by echocardiogram; left ventricular ejection fraction (LVEF) <45% by echocardiogram. Patients were candidates for at least two drugs within each treatment strategy for rate or rhythm control and for anticoagulation. (7) During follow up, visits and ECG recordings occurred at 2 and 4 months after randomization, and thereafter every 4 months.

Journal Pre-proof

6

Patients included in this analysis had no history of HF, classified as having no HF based on clinical history and physical examination at baseline enrollment, normal left ventricular systolic function by echocardiography as reported by the investigators and at least 1 or more heart failure ssessment during follow-up. Study cohorts were based on the initial randomized treatment strategy in the trial.

ro

AFFIRM patients that were excluded were as follows:

of

Patient Exclusions:

a. 749 pts (18.4%) had no baseline echocardiogram reported in the study time

-p

window (in the prior year) at enrollment.

re

b. Of the remainder, only 2244 of the remaining patients had a documented LVEF

lP

of 50% or greater.

c. 418 pts of these patients had either a history of HF or clinical HF.

ur

patients.

na

d. Lack of an adequate number of follow up HF assessments eliminated 55

Jo

The final study cohort consisted of 1771 patients.

Study Outcomes and Definitions HF was assessed at each follow-up visit, and HF emergence was defined as an increase in NYHA HF class by at least two or more classes compared to baseline. HF onset was defined as the midpoint between the follow-up visit at which the ≥ 2 NYHA class increase was observed and the previous follow-up visit. Deaths were identified during study follow-up and using a vital status sweep (telephone contact with all subjects and national death index scan) at the end of the study. Cause of death was adjudicated by a blinded clinical events committee.

Journal Pre-proof

7

Ventricular rate and rhythm were assessed by ECG at each follow-up visit. Patients were stratified into cohorts based on the proportion of follow-up ECGs recording AF from randomization through HF onset. Three strata were established, namely, AF was present in < 25%, 25 to 75% and > 75% of follow up visits with ECG documented

of

cardiac rhythm. Rate control was defined as a resting ventricular rate ≤ 80 bpm.

Principal and Secondary analyses

ro

The principal clinical outcome analysis was a composite of the emergence of HF,

re

Other secondary clinical outcomes were:

-p

(defined as development of NYHA class > 2 HF) and cardiac mortality.

lP

a. Development of NYHA class II or greater HF

na

b. Development of NYHA class I or greater HF c. All Cause Death (which was defined as death from any cause). Time from

ur

randomization to the first occurrence of the following: Death, lost to follow up or

Jo

end of study, and

d. Cardiac death, defined as death due to cardiac etiologies as determined by a clinical events adjudication committee e. CV hospitalization as defined previously.(10)

For cardiac and all-cause death, time zero was defined as the day of randomization through the first of death, loss to follow-up or study termination. Time zero was also the day of randomization for HF onset and CV hospitalization, but the event time was

Journal Pre-proof

8

defined as the midpoint between the visit at which the event was reported and the previous visit. Patients who did not experience these events were censored at the last follow-up visit. Statistical Methods The comparisons of event time using the log-rank test was similar to the primary

of

AFFIRM analysis. Proportional hazards models were used to obtain hazard ratios and

ro

95% confidence intervals. Comparisons were performed by treatment strategy for baseline covariates using the t-test, chi-square test and the Fisher exact test. Kaplan

-p

Meier survival analysis was used to produce survival curves and, when the proportional

re

hazards assumption was met, Cox proportional hazards models were used to adjust for

lP

imbalances in baseline characteristics and to calculate hazard ratios. The impact of HF emergence on subsequent events was quantified using a proportional hazards model

na

with a cumulative time-dependent indicator of whether ≥ NYHA class II HF had been

ur

reported at each follow-up interval.

Jo

The three AF subgroups based on strata of ECG recorded AF (< 25%, 25% to 75% and > 75%) were also compared within each treatment strategy. A sensitivity analysis to examine the relationships between time-dependent AF, lack of rate control and development of HF was performed using the actual ECG results at each follow-up visit through the emergence of HF or end of follow-up. This analysis adjusted for first episode of AF, which differed across the ECG strata for the Rate cohort. The purpose of this analysis was to determine whether the reduction of rhythm information into the strata described above was appropriate, and to determine whether ventricular rate was an independent predictor of HF emergence.

Journal Pre-proof

9

Predictors of HF emergence were identified based on a subset of covariates used in other AFFIRM analyses (age, sex, a first episode of atrial fibrillation, the presence or absence of coronary artery disease, hypertension), as well as minority race, baseline antiarrhythmic drug failure, cardiomyopathy, valvular disease, diabetes, stroke and maximum ventricular rate. (9) These variables were assessed in univariate proportional hazards models. Backward selection was used to reduce the set of variables with p-

of

values < 0.05 in univariate analyses. After the best set of baseline covariates were

ro

identified, randomization assignment was added to the final model.

-p

.Study support and reporting

re

No extramural funding was used to support this work. The authors are solely

lP

responsible for the design and conduct of this study, all study analyses, the drafting and

Patient population

ur

RESULTS:

na

editing of the paper and its final contents."

Jo

1771 patients (1022 male and 749 female), mean age 69.5 years with no baseline HF met the inclusion criteria for this analysis. One patient in the rate control arm had no follow-up ECG information, and was excluded from the analyses of ECG parameters. Patient clinical and arrhythmia characteristics are detailed in left column in Table 1. Table 1 also shows a comparison of this subgroup with the remaining patients in the AFFIRM trial in the adjoining column. Male and minority subjects were a lower proportion in our selected study population compared to the remaining patients (who were mostly excluded due to baseline

Journal Pre-proof

10

evidence of HF and ventricular dysfunction). Other clinically significant differences include a lower incidence of coronary disease, dilated nonischemic cardiomyopathy, valvular heart disease and diabetes. A minimally elevated mean ventricular rate during the index AF event (110 bpm) versus the rest of the population (106 bpm) was not considered clinically significant. ACE inhibitors were used at baseline in 258 patients in the Rate arm and 257 patients in the Rhythm arm, most probably for treatment of

of

hypertension (p=ns).

ro

The comparison of the two cohorts randomized by treatment strategy at baseline is

-p

shown in the two columns in the right in Table 1. The two cohorts were comparable for

re

demographic, disease states, and arrhythmia characteristics and previous

lP

antiarrhythmic drug therapy. Table 2 shows initial drug selection for AF therapy in the Rate and Rhythm arm at randomization. Table 3 shows clinical characteristics of the

na

two treatment subgroups in each recorded AF frequency strata. They are largely

ur

comparable except for mean ventricular rate in AF detected at baseline, which was higher in both treatment arms with <25% of recorded AF during follow-up visits. First

groups.

Jo

(versus recurrent) AF was higher in the <25% strata for Rate compared to the other

Principal Outcome analysis: The principal outcome showed a significantly increased risk of HF emergence and/or cardiac death in the Rate cohort compared to the Rhythm cohort (Hazard ratio (HR) =1.32, 95% confidence interval CI) =1.04-1.69, p=.024) in this population (Figure 1 Panel A). This risk was quantitatively estimated at 21.4% by year 5 of follow up in the Rate cohort and 16.8% in the Rhythm cohort.

Journal Pre-proof

11

Secondary outcomes analyses showed a higher likelihood of HF emergence alone in the Rate cohort as compared to the Rhythm cohort (HR= 1.30, 95% CI = 1.00, 1.69, p=.047, Figure 1, Panel B). By year 5, 20.0% of Rate and 15.4% of Rhythm subjects had developed NYHA class II or greater HF. There were no differences in cardiac mortality (HR= 1.20, 95% CI 0.67, 2.14, p=.539 Figure 2, panel A) or all-cause mortality (HR= 0.79, CI=0.59, 1.05, p= 0.108 Figure 2, panel B). At 5 years, the total mortality in

of

our study group patients was 13.3% in the Rate subgroup and 15.4. % in the Rhythm

ro

subgroup (p=0.108 by the Log-rank test). In the rest of the AFFIRM population with

-p

echo data, it was higher (Rate control 29.3%, Rhythm control 33.5%, p=0.275). Time to

re

death analyzed by use of digoxin therapy at study entry demonstrated no differences in

lP

outcomes (p= 0.483).

na

Table 4 shows a Cox proportional hazard analysis to identify baseline clinical predictors of HF emergence in all AF patients meeting entry criteria for this analysis. Multivariate

ur

analysis identified three baseline characteristics namely, patients entering the study with

Jo

their first episode of AF meeting the duration criteria of 6 hours, and the co-morbidities of coronary artery disease and valvular heart disease as being predictive of HF emergence. Multivariate analysis of baseline factors predicting all-cause mortality showed increasing age (HR 1.07 95% CI 1.05 – 1.09, p<.001), coronary artery disease ( HR 1.42 95% CI 1.06-1.91, p=.02), diabetes ( HR 1.87 95% CI 1.33-2.62 p<001) and stroke (1.60 95% CI 1.11-2.29, p=0.011)

Journal Pre-proof

12

An analysis of the development of class II or greater heart failure with time-dependent covariates (including clinical characteristics, preexisting or continuing therapies and electrocardiographic parameters recorded at each follow up) is shown in Table 5. This analysis identified a Rhythm control treatment strategy as reducing risk of HF emergence. Conversely, first episode of AF, resting ventricular rate > 80 bpm on follow up ECG and recorded AF on ECG were associated with increased risk of HF emergence.

of

In the same model, ventricular rate as a continuous time dependent variable was

ro

examined. Incremental ventricular rates in steps of 10 bpm on follow up increased the

-p

likelihood of HF emergence as defined above (HR 1.092 95% CI 1.054-1.131, p<.0001).

re

Digoxin use in this study at entry was confined to Rate control and when it was introduced later in the study this was for patients with inadequate ventricular rate

lP

control. Thus, the time dependent effect of digoxin therapy cannot be segregated from

na

the underlying baseline condition requiring it, which our analyses show has an independent association with adverse outcome related to HF emergence. This

Jo

ur

precludes analysis of the effect of digoxin Rs an independent variable.

The development of HF in the two strategy cohorts by AF recorded frequency is shown in Figure 3. Among rate control subjects, 9.1% of subjects with AF at < 25% of visits, 12.5% of subjects with AF at 25 – 75% of visits and 15.0% of subjects with AF at > 75% of visits developed NYHA Class ≥ II HF by year 3. Risk of HF development did not differ significantly in the three recorded AF frequency strata of the Rate cohort {25-75% vs. <25%, HR: 1.50, 95% CI: 0.93, 2.44, p = 0.099 ;> 75%vs. <25% HR = HR: 1.36, 95% CI: 0.85, 2.18, p = 0.203, Figure 3 Panel A}. Among rhythm control subjects, 7.5% of

Journal Pre-proof

13

subjects with AF at < 25% of visits, 12.0% of subjects with AF at 25 – 75% of visits and 21.8% of subjects with AF at > 75% of visits developed NYHA Class ≥ II HF by year 3. There was a significantly greater likelihood of developing progressive HF in two recorded AF frequency strata of the Rhythm cohort with a recorded AF frequency >25% {25-75% vs. <25%: HR = 1 HR: 1.69, 95% CI: 1.09, 2.64, p = 0.020; ≥ 75% vs. < 25%:

of

HR = HR: 3.15, 95% CI: 1.87, 5.34, p = <.001, Figure 3 Panel B)

ro

The impact of time-dependent NYHA class increases on clinical events was highly

-p

significant. HF increment by 2 classes increased risk of cardiac mortality, (HR 4.27, 95% CI 2.03, 9.04, p=.0001), the composite of CV hospitalization and death (HR3.00 2.19,

re

4.12, p<0.0001), total mortality (HR 2.83, 95% CI 1.91, 4.18, p<.0001),and CV

lP

hospitalizations (HR 3.04, 95% CI 2.15, 4.29, p<.0001). An increase of 1 NYHA class

ur

DISCUSSION:

na

was associated with slightly smaller but statistically significant increases in event risk.

Jo

HF progression is widely recognized to predict subsequent mortality in numerous populations with heart disease, and is specifically recognized as a predictor of adverse outcomes in patients with HF. Preventing progression of HF is an important objective of HF therapies and reversing HF class status is often used to suggest therapeutic value of the treatment strategy. Primary HF prevention with upstream therapies has not been widely employed in patients with AF at risk for HF. In the AFFIRM trial, progression of HF was identified as a predictor of CV hospitalization and death during the study but no detailed analysis has been available in symptomatic AF patients without HF. (8).

Journal Pre-proof

14

This analysis examined: a. The likelihood of developing HF in AF patients entering the AFFIRM study with no prior or presenting history of HF or evidence of systolic ventricular dysfunction. b. Furthermore, we examined HF emergence in each of the treatment arms of this trial and determined its impact on mortality

of

c. Finally, we studied these outcomes based on rhythm recorded prior to HF

ro

emergence and the frequency of electrocardiographically documented AF recorded

-p

during follow up in the two treatment arms of the study.

re

These are novel analyses that provide new information on the emergence of HF and

lP

impact on mortality and CV hospitalizations during the life cycle of AF patients with

na

these two treatment strategies in this foundational study.

ur

The major findings of this study are:

Jo

1. The study population selected by the entry criteria for this analysis encompassed 44% of the entire population of the AFFIRM trial 2. Rate and Rhythm subgroups were well matched for clinical and demographic criteria 3. The risk of development of HF was higher in the Rate cohort compared to the Rhythm cohort. 4. Emergence of HF was associated with adverse clinical outcomes including total and cardiac mortality and CV hospitalization risk. HF increase by 2 HF NYHA

Journal Pre-proof

15

classes had a more severe impact on these outcomes and trended to increase cardiac mortality risk to a greater extent 5. Recorded AF and its frequency on ECG follow-up was associated with the emergence of HF in the Rhythm cohort, with patients with <25% of visits with recorded AF having significantly less likelihood of HF emergence. 6. Ventricular rate control as defined also plays an important role in HF emergence,

of

and this effect appears to be independent of rhythm control.

ro

Heart failure Emergence in AF and Cardiovascular events or Death

-p

Our study data provide a quantitative estimate of HF emergence in AF patients who

re

initially present without a prior history or present evidence of HF and normal systolic left

lP

ventricular function. HF emergence presaged CV hospitalizations and death. HF emergence, initially even by just 1 NYHA class, significantly increased risk for total

na

mortality, cardiac mortality and CV hospitalization. This risk increased further with

ur

worsening by>2 NYHA classes.

Jo

We chose HF emergence rather than HF hospitalizations for several reasons. Hospitalizations in AFFIRM were classified by the investigator as cardiac and non cardiac; In our analysis of CV hospitalizations that we have used in all our prior publications commencing from reference 10, these hospitalizations could include causes other than heart failure such as PCI, MI etc.In other cases, more than one cardiac event occurred including HF. These considerations led us to conclude that we could not reliably identify HF alone.

Journal Pre-proof

16

HF emergence may reflect an inflection point in the longitudinal natural history of this AF population. Furthermore, it may help identify appropriate endpoints for AF population’s enrolled in future clinical trials. HF emergence may be an intermediate event in the longitudinal history of AF, occurring after emergence of AF symptoms and preceding, but predicting, entry of the patient into a higher mortality and CV hospitalization phase of the disease. Mortality trials in this condition may be more likely to achieve higher

of

event rates after the initial emergence of HF symptoms. The absence of this inflection

ro

point may explain the need for larger AF patient enrollment in trials such as CABANA

-p

versus CASTLE – AF. (5, 11)

re

Impact of Atrial Fibrillation treatment strategy on Heart failure emergence

lP

The benefits of effective rhythm control on HF emergence are suggested from our data and are hypothesis generating warranting further investigation. However, no overall all

na

cause mortality differences were elicited during the follow up period with Rate or

ur

Rhythm strategies and this finding warrants comment. In another study in early HF,

Jo

improvement in HF hospitalizations was observed by study termination, but mortality benefits accrued only with extended follow-up up to nine years after initiating cardiac resynchronization therapy. (12, 13) Similar considerations may apply here in AF. Atrial Fibrillation Recording Methodology and Impact on Outcomes As discussed in the introduction to this report, AF recorded in HF patients either prior to or at study enrollment predicted a poorer clinical outcome for these individuals. (1,2) AF recorded after study entry also had similar implications and support use of intermittent ECG recordings having prognostic value.(2) Our study used every 4 monthly ECG

Journal Pre-proof

17

recordings as required in AFFIRM to provide more detailed longitudinal data. Our time dependent analyses in Table 5 clearly indicates that recorded AF on follow up was associated with HF emergence and the frequency of recorded AF in the higher density strata in the rhythm arm were consistent with this finding.(Figure 3) Prior studies have also only reported the frequency of rhythm control at fixed follow up

of

points in the entire study population Thus, AF – CHF and CASTLE-AF reported frequency of detected AF in the entire randomized cohort for each treatment strategy.

ro

(4, 5) In contrast to these and other trials, we have examined the longitudinal behavior

-p

of individual patients with respect to recorded AF frequency and correlated this with

re

development of HF. We categorized patients into frequency strata that could be

lP

practical and widely available in a cardiology clinic follow up setting. While total AF burden is available in implanted devices, device insertion is not routinely employed in all

na

AF patients. However, ECG based AF detection may reliably reflect progression of AF

ur

as seen by implanted devices.

Jo

In a device based study of AF progression, our group reported that the first persistent AF event was often self- terminating, but it rapidly evolved into persistent AF which became established within months. (14, 15) Thus, serial ECG recordings at 2 or 4 month intervals would record conversion to persistent AF. In a different subgroup of AFFIRM patients, those with prior HF event or HF at enrollment, “continuous “ AF was more likely to be reported by the investigators in the Rate arm (36.1%) as compared to the Rhythm arm (20.2%). (16, 17) Role of Rhythm and Rate Control on HF Emergence

Journal Pre-proof

18

. In other studies, restoration of sinus rhythm with catheter ablation after HF has emerged, with catheter ablation has been shown to reduce HF progression and improve LV systolic function. (5, 18) Our data extend this observation by suggesting that a high degree of rhythm control on follow-up with avoidance of recorded AF could potentially delay emergence of the first HF event in AF. A high degree of recorded rhythm control (< 25% of follow up visits with recorded AF) in the Rhythm arm was noted in a majority

of

(n=522, .58.5%) of Rhythm arm subjects. Device based studies of the natural history of

ro

AF progression would suggest it is likely that most of these patients in the <25% AF

-p

frequency subgroup did not evolve to persistent AF. (14, 15) Our observations could

lP

persistent AF at study termination.

re

also suggest longer periods of sinus rhythm, in addition to a lower incidence of

Ventricular rate control on follow-up had an independent and significant impact on HF

na

emergence. Good resting ventricular rate control (resting rate <80 bpm) in a time

ur

dependent analysis predicted lower risk of this outcome. Thus, our analysis suggests

emergence.

Jo

that good rhythm and rate control have additive potential in lowering the risk of HF

Finally, in another retrospective time-dependent analysis, the AFFIRM investigators suggested that AF patients were less likely to have a mortal event during a period of sinus rhythm than during periods of AF during follow up in either treatment strategy. (19) While this finding has been widely quoted to suggest benefits with a rhythm control strategy, it does not provide evidence of definitive benefit. However, HF emergence has definitive adverse mortality and hospitalization outcomes as seen in this study and

Journal Pre-proof

19

effective rhythm control could confer benefits by delaying HF emergence and its consequences. Limitations of the study The AFFIRM trial excluded patients with AF both in whom a rate or a rhythm control strategy could not be implemented (such as permanent AF), patients declining rate or

of

rhythm control and highly symptomatic AF patients requiring immediate intervention.

ro

This was a retrospective analysis of the AFFIRM trial. Randomization was limited to the

-p

entire treatment strategy populations rather than the subgroups drawn from them in this report and this analysis was not prospectively specified. Furthermore, our data

re

demonstrate that comparisons of rate and rhythm control require good rate control to be

lP

achieved as this is an independent contributor to AF outcomes. Serial echocardiographic studies were not performed in the AFFIRM trial. Total AF burden in

na

individual patients and asymptomatic AF between follow up visits cannot be measured

Jo

Conclusions:

ur

based on recorded frequency of AF on follow up visit.

In AF patients with no HF event or clinical HF signs, an effective Rhythm control strategy with ECG documented sinus rhythm during >75% of follow up visits was associated with delayed emergence of symptomatic HF. HF emergence significantly increased CV morbidity and mortality. There is an implication from these data that effective AF rhythm control could be a potential goal for HF prevention in AF, and open a new avenue for future study. These data are hypothesis generating observations and warrant prospective well designed clinical studies in the future.

Journal Pre-proof

20

Acknowledgements: The authors gratefully acknowledge the important contributions of the AFFIRM investigators in recruiting patients and executing this trial and the National Heart , Lung

of

and Blood Institute for funding this trial. The AFFIRM database was made available at

ro

the conclusion of the trial to the Steering Committee members who included Dr.

Jo

ur

na

lP

re

-p

Saksena.

Journal Pre-proof

21

REFERENCES: 1. Olsson LG, Swedberg K, Ducharme A, Granger CB, Michelson EL, McMurray J, et al. Atrial fibrillation and risk of clinical events in chronic heart failure with and without left ventricular systolic dysfunction: results from the candesartan in heart failure assessment of reduction in mortality and morbidity (CHARM) program. J

of

Am Coll Cardiol. 2006; 47(10):1997–2004. 2. Cikes M, Claggett B, Shah AM, Desai AS, Lewis M, Shah SJ, et al. Atrial

ro

fibrillation in heart failure with preserved ejection fraction: the treatment of

-p

preserved cardiac function heart failure with an aldosterone antagonist

re

(TOPCAT) trial. JACC Heart Fail. 2018;6: 689–97

lP

3. Maisel WH, Stevenson LW Atrial fibrillation in heart failure: epidemiology, pathophysiology, and rationale for therapy. Am J Cardiol 2003;91:Suppl:2D-8D

na

4. Roy D, Talajic M, Nattel S et al: Rhythm Control versus Rate Control for Atrial

ur

Fibrillation and Heart Failure. New Eng. J. Med 2008: 358:2667-77 5. Marrouche N, Brachmann J, Andresen D et al: Catheter ablation for atrial

Jo

fibrillation and heart failure. New Eng. J. Med 2018: 378:417-27 6. Briceño DF, Markman T.M, Lupercio F et al: Catheter ablation versus conventional treatment of atrial fibrillation in patients with heart failure with reduced ejection fraction: a systematic review and meta-analysis of randomized controlled trials. J. Intervent. Card. Electrophysiol. 2018: 51(1): 19-30. 7. Mangla A, Kane J, Beaty E, Richardson D, Powell LH, Calvin JE Jr.: Comparison of predictors of heart failure-related hospitalization or death in patients with versus

Journal Pre-proof

22

without preserved left ventricular ejection fraction. Am J Cardiol. 2013 Dec 15;112(12):1907-12. 8. Bittner V1, Weiner DH, Yusuf S, Rogers WJ, McIntyre KM, Bangdiwala SI, Kronenberg MW, Kostis JB, Kohn RM, Guillotte M, et al. Prediction of mortality and morbidity with a 6-minute walk test in patients with left ventricular dysfunction. SOLVD Investigators. JAMA. 1993 Oct 13;270(14):1702-7.

of

9. The AFFIRM investigators: A comparison of rate control and rhythm control in

ro

patients with atrial fibrillation: New Eng. J. Med. 2002: 347:1825-33

-p

10. Saksena S, Slee A, Waldo AL, Freemantle N, Reynolds M, Rosenberg Y,

AFFIRM

Trial

re

Rathod S, Grant S, Thomas E, Wyse DG: Cardiovascular Outcomes in the (Atrial

Fibrillation

Follow-Up

Investigation

of

Rhythm

lP

Management): An Assessment of Individual Antiarrhythmic Drug Therapies

na

Compared With Rate Control With Propensity Score-Matched Analyses. J Am

ur

Coll Cardiol 2011;58:1975–85

11. Packer DL, Mark DM, Robb RA et al: Effect of catheter ablation vs.

Jo

antiarrhythmic drug therapy on mortality, stroke, bleeding and cardiac arrest in patients with atrial fibrillation. J. Am. Med. Assoc. 2019: doi:10.1001 /jama.2019.0693 12. Goldenberg I, Kutyifa V, Klein HU et al: Survival in cardiac resynchronization therapy with mild heart failure. N Engl J Med 2014; 370:1694-701. 13. Herscovici R, Kutyifa V, Barsheshet A, Solomon S, McNitt S, Polonsky B, Lee AY, Zareba W, Moss AJ, Goldenberg I : Early intervention and long-term

Journal Pre-proof

23

outcome with cardiac resynchronization therapy in patients without a history of advanced heart failure symptoms. Eur J Heart Fail. 2015: 9:964-70. 14. Saksena S, Hettrick DA, Koehler JL, Grammatico A, Padeletti L. Progression of paroxysmal atrial fibrillation to persistent atrial fibrillation in patients with bradyarrhythmias. Am Heart J. 2007;154(5):884–92 15. Nagarakanti R, Saksena S, Hettrick D, Koehler JL, Grammatico A, Padeletti L:

of

Progression of new onset to established persistent atrial fibrillation: an

ro

implantable device-based analysis with implications for clinical classification of

re

doi: 10.1007/s10840-011-9601-1

-p

persistent atrial fibrillation. J Intervent Card Electrophysiol. 2011 Oct; 32(1):7-15.

16. Slee A, Saad M, Saksena S:Progression of heart failure differs in patients with

lP

atrial fibrillation with rate and rhythm control strategies (abstract) Heart Rhythm,

na

2019: 16, No. 5, Supplement: S176.

17. Slee A, Saad M, Saksena S:Heart failure progression and mortality in atrial

ur

fibrillation patients with preserved or reduced left ventricular ejection fraction:

Jo

Journal of Interventional Cardiac Electrophysiology (2019) 55:325–331 18. Hsu LF, Jaïs P, Sanders P et al: Catheter ablation for atrial fibrillation in congestive heart failure. N. Engl. J. Med. 2004 Dec 02;351 (23):2373–83. 19. The AFFIRM Investigators: Relationships Between Sinus Rhythm, Treatment, and Survival in the Atrial Fibrillation Follow-Up Investigation of Rhythm Management (AFFIRM) Study Circulation 2004;109;1509-1513

Journal Pre-proof

24

TABLE 1: PATIENT CHARACTERISTICS OF STUDY POPULATION, TREATMENT STRATEGY

Jo

ur

na

lP

re

-p

ro

of

SUBGROUPS AND RESIDUAL AFFIRM POPULATION

Journal Pre-proof

25

25

TABLE 2: RATE AND RHYTHM CONTROL DRUG USE IN STUDY POPULATION

Diltiazem

Verapamil

366 ( 41.6%)

161 ( 18.1%)

527 ( 29.8%)

No

514 ( 58.4%)

730 ( 81.9%)

1244 ( 70.2%)

Yes

415 ( 47.2%)

120 ( 13.5%)

535 ( 30.2%)

No

465 ( 52.8%)

771 ( 86.5%)

1236 ( 69.8%)

Yes

255 ( 29.0%)

86 ( 9.7%)

341 ( 19.3%)

No

625 ( 71.0%)

805 ( 90.3%)

1430 ( 80.7%)

88 ( 10.0%)

27 ( 3.0%)

115 ( 6.5%)

792 ( 90.0%)

864 (97.0%)

1656 ( 93.5%)

0 ( 0.0%)

253 (28.4%)

253 ( 14.3%)

880 (100.0%)

638 (71.6%)

1518 ( 85.7%)

1 ( 0.1%)

305 ( 34.2%)

306 ( 17.3%)

879 ( 99.9%)

586 ( 65.8%)

1465 ( 82.7%)

1 ( 0.1%)

108 (12.1%)

109 ( 6.2%)

879 ( 99.9%)

783 ( 87.9%)

1662 ( 93.8%)

Yes

0 ( 0.0%)

46 ( 5.2%)

46 ( 2.6%)

No

880 (100.0%)

845 ( 94.8%)

1725 ( 97.4%)

Yes

2 ( 0.2%)

32 ( 3.6%)

34 ( 1.9%)

No

878 ( 99.8%)

859 ( 96.4%)

1737 ( 98.1%)

Yes

0 ( 0.0%)

69 ( 7.7%)

69 ( 3.9%)

No

880 (100.0%)

822 ( 92.3%)

1702 ( 96.1%)

Yes

0 ( 0.0%)

25 ( 2.8%)

25 ( 1.4%)

No

880 (100.0%)

866 ( 97.2%)

1746 ( 98.6%)

Yes

0 ( 0.0%)

6 ( 0.7%)

6 ( 0.3%)

No

880 (100.0%)

885 ( 99.3%)

1765 ( 99.7%)

of

Yes

ro

Beta Blocker

Rhythm (N=891)

-p

Digoxin

Rate (N=880)

Yes

Yes No

Yes No

Yes

ur

Propafenone

na

Sotalol

lP

Amiodarone

re

No

Procainamide

Quinidine

Flecainide

Disopyramide

Moricizine

Jo

No

Total (N=1771)

Journal Pre-proof

26

TABLE 3: COMPARISON OF PATIENT CHARACTERISTICS IN SUBGROUPS BASED ON FREQUENCY OF AF RECORDED DURING FOLLOW-UP IN THE RATE AND RHYTHM COHORTS

Jo

ur

na

lP

re

No Baseline HF - Rhythm

of

-p

ro

No Baseline HF – Rate

Journal Pre-proof TABLE 4:

Jo

ur

na

lP

re

-p

ro

of

PREDICTORS OF HEART FAILURE EMERGENCE IN STUDY POPULATION

27

Journal Pre-proof

28

TABLE 5:

RELATIONSHIP OF RATE AND RHYTHM CONTROL AND INDIVIDUAL STUDY PARAMETERS TO HF EMERGENCE: A TIME DEPENDENT ANALYSIS

Hazard Ratio

of

0.88 1.23 1.20 1.24

Jo

ur

na

lP

re

-p

Rhythm control (versus Rate control) First episode of AF (vs. recurrent) Follow-up heart rate > 80 bpm (vs. <= 80 bpm) Follow-up AF (versus SR)

95% CI 0.79 1.11 1.04 1.10

ro

Parameter

0.98 1.37 1.37 1.39

pvalue 0.0211 0.0001 0.0106 0.0004

Journal Pre-proof

LEGENDS TO TABLES: Table 1: Comparison of baseline demographic and clinical characteristics of entire study cohort and residual AFFIRM study population (middle panel) and of the two study cohorts by

of

treatment strategy (right panel).

ro

Abbreviations: As in text; age is shown in years; bpm = beats per minute;

-p

Cardiomyopathy = Dilated nonischemic cardiomyopathy

lP

re

Table 2:

Table 3:

na

Rate and Rhythm Control Drug therapy selected at Baseline in Study Population

ur

Comparison of baseline demographic and clinical characteristics of patients classified

Jo

by frequency of AF recording during follow-up in the Rate and Rhythm subgroups Abbreviations: As in text; age is shown in years; bpm = beats per minute. Table 4. Cox proportional hazards analysis to identify predictors of progressive HF in all AF patients meeting entry criteria for this analysis Abbreviations: As in text; age is shown in years; bpm = beats per minute; AAD = antiarrhythmic drug; CAD = coronary artery disease; VR = ventricular rate

29

Journal Pre-proof

30

able 5. Development of class II or greater heart failure with time-dependent covariates. This figure shows time dependent analysis of Rate and Rhythm control strategies and individual covariates (comprising clinical, baseline and concomitant therapy,

of

randomized strategy, electrocardiographic parameters on follow up that had significant

Jo

ur

na

lP

re

-p

ro

relationship to heart failure emergence ) in the study population.

Journal Pre-proof

31

LEGENDS TO FIGURES Figure 1 Caption: Principal and Secondary Outcomes with Rate and Rhythm Strategies in Study Population Panel A: Time to development of NYHA Class II or Greater HF or Death from

of

Cardiovascular Cause,

-p

ro

Panel B: Time to development of NYHA Class II or Greater HF

re

Abbreviations: as in text;

lP

Figure 2

na

Caption: Mortality Outcomes with Rate and Rhythm Strategies in Study Population Panel A: Time to Death from Cardiovascular Cause

Jo

ur

Panel B: Time to Death from Any Cause Abbreviations: as in text Figure 3 Caption: Secondary Outcome with Rate and Rhythm Strategies in Study Population Time to development of NYHA Class II or Greater HFin each treatment strategy stratified by Recorded AF Frequency during Follow-up Visits in study population.

Journal Pre-proof Panel A: Rate Control Cohort Panel B: Rhythm Control Cohort Abbreviations: as in text Figure 4

of

Caption: impact of Heart Failure Emergence on clinical outcomes

ro

Time dependent impact of Heart Failure Emergence on clinical outcomes in study

-p

population (Forrest plot) ;

re

X axis shows hazard ratio on log scale

lP

Y axis: Clinical outcome

Jo

ur

heart failure class

na

Abbreviations: CV=cardiovascular; HF= heart failure; HR= hazard ratio; class =NYHA

32

Journal Pre-proof

33

Declarations of interest: Ms. Slee is a statistician with no financial or other conflicts of interest in the atrial fibrillation space. She was a senior member of the original statistical and data management/coordinating center, AxioResearch Corporation in Seattle till 2015. Dr. Saksena is a member of the Planning and Steering Committee of the AFFIRM trial,

of

an investigator in the Trial, and the head of the Non-Pharmacologic therapy section of the Trial. The Foundation received funding as a clinical site during the period 1996-

ro

2001. The only other relevant analysis for the AFFIRM study that was funded from

-p

outside sources was the Cardiovascular Outcomes study in 2011 funded by an

re

educational grant from sanofi. His most recent funding in the atrial fibrillation space was

lP

the ENTICED-AF trial from Daiichi-Sankyo and this terminated in 2017. All AFFIRM analyses published by this group since 2012 (9 abstracts and two manuscripts) have

Jo

ur

na

been solely supported by the Electrophysiology Research Foundation.

Journal Pre-proof Highlights 1. Treatment strategy for atrial fibrillation impacts heart failure (HF) emergence 2. HF emergence & Cardiac Death occurred more frequently in Rate than Rhythm 3. Absent/infrequent AF recurrence and strict ventricular rate control delayed HF emergence

control

Jo

ur

na

lP

re

-p

ro

5. HF emergence presaged higher mortality risk .

of

4. HF emergence less likely with recorded AF < 25% of followup visits in Rhythm

34

Figure 1

Figure 2

Figure 3

Figure 4