Prognostic Significance of AtrialFibrillation in Patients at a TertiaryMedical Center Referred for HeartTransplantation Because of Severe

Heart FailurePaul Mahoney, MD, Stephen Kimmel, MD, MS, David DeNofrio, MD, Peter Wahl, BA,

and Evan Loh, MD

Atrial fibrillation (AF) occurs frequently in advancedheart failure. The prognostic significance of AF remainscontroversial. To determine the relation of AF to survivalin patients with advanced heart failure, 234 consecutivepatients referred for heart transplantation evaluationfrom January 1993 to June 1996 were studied to deter-mine the effect of AF on event-free survival (freedomfrom death, heart transplantation, or placement of a leftventricular assist device). Clinical characteristics of thestudy population included: age, 51 6 17 years; maxi-mum exercise oxygen consumption, 14.2 6 5.3 ml/kg/min; left ventricular ejection fraction, 24 6 11%; pulmo-nary capillary wedge pressure, 23 6 9 mm Hg; andischemic etiology, 52%. Medical therapy included: di-uretics (86%), angiotensin-converting enzyme inhibitors(80%), digoxin (80%), and anticoagulation therapy(72%). Mean duration of follow-up was 1.1 6 1.0 years.

Sixty-two patients (27.4%) had AF. One-year event-freesurvival of the study population was 48%. No differencein event-free survival between patients with and withoutAF was observed. Univariate predictors of decreasedevent-free survival included: (1) advanced New YorkHeart Association class; (2) higher pulmonary capillarywedge pressure; (3) lower cardiac index; (4) lower max-imum exercise oxygen consumption; (5) use of inotropictherapy; and (6) greater pulmonary artery systolic pres-sure. By multivariate analysis, independent predictors ofdecreased event-free survival included advanced NewYork Heart Association class (p <0.002) and higherpulmonary capillary wedge pressure (p 5 0.02). Thus,AF in patients with advanced heart failure is not asso-ciated with decreased event-free survival. Q1999 byExcerpta Medica, Inc.

(Am J Cardiol 1999;83:1544–1547)

The prognostic significance of atrial fibrillation(AF) in patients with advanced left ventricular

dysfunction remains controversial.1–11 Carson et al1

reported no increase in either morbidity or mortalityattributable to the presence of AF on survival in 1,427patients with mild to moderate (New York Heart As-sociation [NYHA] class I and II) heart failure. How-ever, in patients with advanced heart failure (NYHAclass III and IV), AF has been associated with noeffect on survival,3,8 an adverse effect on survival insome studies,4,9,10and even an improved survival in 1study.11 Given the current lack of consensus surround-ing AF and its relation to survival in patients withadvanced heart failure, we sought to determine theprognostic impact of AF on event-free survival inpatients with advanced heart failure symptoms re-ferred to our institution for heart transplantation eval-uation between 1993 to 1996.

METHODSPatient population: Two hundred thirty-four con-

secutive patients with severe left ventricular dysfunc-tion referred for evaluation for heart transplantation tothe Hospital of the University of Pennsylvania fromJanuary 1, 1993, to June 1, 1996 were included in thisanalysis. All patients underwent an evaluation thatincluded a history and physical examination, right andleft heart catheterization, radionuclide ventriculogra-phy, laboratory evaluation, treadmill exercise test,transthoracic echocardiogram, electrocardiogram, andpulmonary function tests. Clinical variables from thisevaluation were prospectively entered into a comput-erized database designed by the authors.

Medical therapy: The goal of medical therapy wasfor all patients to be treated with digoxin, loop diuret-ics, and angiotensin-converting enzyme inhibitors (tothe maximally tolerated dosage). The approach to AFgenerally consisted of a conservative strategy of ven-tricular rate control and anticoagulation, unless con-traindicated. Final decisions on medical therapy wereat the discretion of the attending heart failure cardiol-ogist.

Definitions: AF was defined as irregular and/or dis-organized atrial activity with an irregular ventricularresponse (atrial flutter was considered a variant ofAF). Patients were assigned to the AF group if theyhad evidence of AF on the electrocardiogram, Holter,

From the Cardiovascular Division, Department of Medicine, and Cen-ter for Clinical Epidemiology and Biostatistics, Hospital of the Univer-sity of Pennsylvania, Philadelphia, Pennsylvania. This study was sup-ported in part by Grant MO1–RR00040 from the National Institutesof Health, Bethesda, Maryland. Manuscript received September 28,1998; revised manuscript received and accepted January 27, 1999.

Address for reprints: Evan Loh, MD, Cardiovascular Division,Hospital of the University of Pennsylvania, 3400 Spruce Street, Phila-delphia, Pennsylvania 19104. E-mail: lohe@mail.med.upenn.edu.

1544 ©1999 by Excerpta Medica, Inc. 0002-9149/99/$–see front matterAll rights reserved. PII S0002-9149(99)00144-7

or a history of clearly documented episodes of AF,even if in normal sinus rhythm (NSR) at the time ofevaluation (“paroxysmal AF”).

Event-free survival analysis: The primary end pointwas event-free survival, defined as freedom from (1)death, (2) placement of a left ventricular assist deviceas a bridge to cardiac transplantation, or (3) cardiactransplantation (as United Network of Organ SharingStatus I classification).

Follow-up: Follow-up data were obtained by reviewof active clinic charts, by phone contact with referringphysicians, or by phone contact with patients them-selves. Data on clinical end points (death, transplant,placement of left ventricular assist device, or alive atend of study period) was available on 226 of 234patients (96.6%); 8 patients (3.4%) were lost to fol-low-up, 6 with NSR and 2 with AF).

Statistical analysis: Statistical analysis was per-formed using SPSS software package on a WindowsNT platform (SPSS Inc., Chicago, Illinois). Compar-ative analysis of discrete variables between AF andNSR groups at baseline was done using the chi-squaretest; continuous baseline variables were analyzed us-ing the independentt test. The combined end point ofdeath, left ventricular assist device placement as abridge to transplant, or heart transplantation was ex-amined, and cumulative event-free survival rates be-tween the 2 groups were compared by log-rank anal-ysis. Event-free survival curves for each group weregenerated by the Kaplan-Meier method. Cumulativesurvival analysis was also performed between the 2groups for the end point of death alone (with rightcensoring of cardiac transplantation and placement ofleft ventricular assist device), as well as between the 2groups with those in paroxysmal AF included in theNSR group.

Univariate predictors for both event-free survivalas well as death alone were identified via Cox propor-tional-hazards modeling of 45 variables consideredpotentially significant out of 353 variables in the com-puterized database. Multivariable Cox models weredeveloped using variables associated with a p value,0.10 in univariate analysis; statistical significancewas defined at a 2-sideda level of ,0.05.

RESULTSStudy population: Demographic and clinical data on

the overall study population are listed in Table I.Medical therapy for the baseline population includedangiotensin-converting enzyme inhibitors (80.4%),digoxin (80.1%), and loop diuretics (85.7%). Seventy-two percent of patients, regardless of atrial rhythm,were taking anticoagulation therapy. Seventeen per-cent of patients were receiving intravenous inotropictherapy at the time of initial evaluation. Antiarrhyth-mic therapy included: amiodarone (15.8%), class IAantiarrhythmics (7.5%), procainamide (2.5%), quini-dine (5%), andb blockers (9.0%). Functional andhemodynamic parameters obtained at the time of eval-uation for heart transplantation are listed in Table II.

Comparison of clinical and hemodynamic parame-ters of heart failure patients—atrial fibrillation versusnormal sinus rhythm: A greater percentage of patientswith AF were Caucasian; other clinical and hemody-namic parameters between the groups were not differ-ent (Table III). Of the 63 patients classified as AF, 57(90%) were in chronic AF and 6 were consideredparoxysmal. Patients with AF had a higher rate ofdigoxin, amiodarone, and class IA antiarrhythmicdrugs usage (Table IV). No significant difference wasobserved in use of implantable cardiac defibrillators (4of 63 [6%] in the AF group vs 8 of 171 [5%] in NSRgroup).

Event-free survival analysis: Mean follow-up of thestudy cohort was 1.16 1.0 years. A total of 133 endpoints were reached in 226 patients for whom fol-low-up data were available and included: death (n540), placement of left ventricular assist device (n513), and heart transplantation (n5 80). All recipientswere United Network of Organ Sharing Status I at thetime of transplantation. One- and 2-year cumulativeevent-free survival of the entire population was 48%and 42%, respectively (Figure 1).

Ninety-seven patients (58%) in the NSR groupreached the predefined clinical end point during theperiod of the study (26 deaths [16%], 9 left ventricularassist devices [5%], and 62 heart transplantations[37%]). Thirty-six patients in the AF group (59%)reached a clinical end point (14 deaths [23%], 4 left

TABLE I Clinical Characteristics of Heart Failure Population(n 5 234)

Age (yr) (mean 6 SD) 51 6 17Men 185 (79%)Etiology of heart failure

Ischemic 122 (52%)Nonischemic 112 (48%)

NYHA classII 33 (14%)III 126 (54%)IV 75 (32%)

Atrial fibrillation 63 (27%)Diabetes mellitus 63 (27%)Race

White 178 (76%)Non-white 56 (24%)

History of tobacco use 157 (67%)

TABLE II Hemodynamics of Heart Failure Population

Left ventricular ejection fraction (%) 24 6 11Right atrial mean (mm Hg) 10 6 6Pulmonary artery systolic (mm Hg) 52 6 19Pulmonary artery diastolic (mm Hg) 25 6 10Pulmonary artery mean (mm Hg) 34 6 12Pulmonary capillary wedge pressure (mm Hg) 23 6 9Cardiac output (L/min) 4.3 6 1.3Cardiac index (L/min/m2) 2.1 6 0.5Pulmonary vascular resistance (Wood U) 2.9 6 1.6Systemic vascular resistance (Wood U) 18.3 6 7.3Maximum exercise O2 consumption (ml/kg/min) 14.2 6 5.3Peak heart rate (beats/min) 130 6 25

Values are expressed as mean 6 SD.O2 5 oxygen.

CONGESTIVE HEART FAILURE/ATRIAL FIBRILLATION, HEART FAILURE, AND PROGNOSIS 1545

ventricular assist devices [7%], and 18 heart trans-plantations [29%]). No significant difference in event-free survival between the 2 groups was observed (Fig-ure 2).

Predictors of event-free survival: Univariate compar-isons by Cox proportional-hazards regression analysisidentified 7 statistically significant variables associ-ated with decreased event-free survival (Table V).Multivariable regression analysis identified onlyNYHA class and pulmonary capillary wedge pressureas independently associated with decreased event-freesurvival (Table V). The unadjusted hazard ratio byCox proportional-hazards analysis of AF to event-freesurvival was 1.21. The multivariate hazard ratio forthe association between AF and event-free survivalafter adjusting for confounding variables was also notsignificant (hazard ratio 0.81; 95% confidence interval0.53 to 1.40). There were no statistically significantinteractions between AF and any of the followingvariables: anticoagulation, pulmonary capillary wedge

pressure, NYHA class, cardiac index, use of amioda-rone, pulmonary artery systolic pressure, right atrialpressure, and maximum exercise oxygen consump-tion.

A separate survival analysis was performed forfreedom from death, with right censoring of patientswho underwent transplantation or placement of leftventricular assist device. One- and 2-year freedomfrom death was 81% and 71%, respectively. No dif-ference in freedom from death between the AF andNSR groups was observed. A separate analysis includ-ing the 6 patients in paroxysmal AF in the NSR groupdid not affect results of the survival analysis.

DISCUSSIONIn this study, the presence of AF in patients with

advanced heart failure referred for heart transplanta-tion evaluation was not associated with decreasedevent-free survival.

Previous studies in smaller patient cohorts havedemonstrated that AF is a univariate, but not multi-variate, predictor of decreased survival in patientswith decreased left ventricular systolic function.9,10

Dries et al12 reported an independent association be-tween AF and mortality in a recent abstract from theStudy of Left Ventricular Dysfunction trial of 6,797

TABLE III Comparison of Clinical Variables: NSR Versus AF

VariablesNSR

(n 5 171)AF

(n 5 63)

Age (yr) 50 6 13 52 6 17Men 132 (77%) 54 (85%)Ischemic 84 (49%) 33 (52%)NYHA class (mean) 3.3 6 0.3 3.4 6 0.3Prior cerebrovascular accident 9 (5%) 6 (10%)Diabetes 47 (26%) (27%)Caucasian 123 (72%) 55 (87%)*Listed for transplant 99 (58%) 38 (60%)Left ventricular ejection fraction (%) 23 6 10 24 6 11Right atrial mean (mm Hg) 10 6 6 10 6 5Pulmonary artery mean (mm Hg) 34 6 13 35 6 10Pulmonary capillary wedge

pressure (mm Hg)23 6 9 24 6 9

Cardiac index (L/min/m2) 2.1 6 0.6 2.0 6 0.6Pulmonary vascular resistance

(Wood U)2.7 6 1.8 3.1 6 1.6

Systemic vascular resistance(Wood U)

18.0 6 6.0 19.2 6 9.7

Maximum exercise O2 consumption(ml/kg/min)

14.1 6 4.9 14.6 6 6.4

*p 5 0.02.All results are expressed as mean 6 SD, or as number of patients (%).O2 5 oxygen.

TABLE IV Medical Therapy: NSR Versus AF

MedicationsNSR

(n 5 171)AF

(n 5 63)

ACE inhibitors 135 (79%) 53 (84%)Digoxin 132 (77%) 56 (89%)*Diuretics 145 (84%) 57 (90%)Amiodarone 17 (10%) 19 (30%)*Anticoagulation 118 (69%) 50 (79%)b blockers 19 (11%) 2 (3%)Calcium antagonists 15 (9%) 6 (10%)Intravenous inotropes 31 (18%) 9 (15%)Nitrates 56 (33%) 16 (25%)Class 1A antiarrhythmics 5 (3%) 11 (17%)*

*p , 0.05.All results are expressed as number of patients (%).

FIGURE 1. Kaplan-Meier survival curve for event-free survival(n 5 226). Events: death (n 5 40), left ventricular assist device(n 5 13), and transplantation (n 5 80).

FIGURE 2. Kaplan-Meier survival curves for AF and NSR pa-tients. No statistically significant difference between the 2 curveswas observed (p 5 0.21).

1546 THE AMERICAN JOURNAL OF CARDIOLOGYT VOL. 83 JUNE 1, 1999

patients with predominantly class I to II NYHA heartfailure symptoms, while noting important baselinedifferences between the AF and NSR groups. In theVeteran’s Affairs Cooperative Study Group1 of 1,426patients with NYHA class II to III heart failure symp-toms in whom the groups were not greatly different inbaseline characteristics, AF was not associated withdecreased survival. In advanced heart failure, Middle-kauff et al4 first reported a significant associationbetween AF and decreased survival in patients re-ferred for transplantation evaluation. In contrast, in alater series of patients evaluated at the same institutionwho were treated with more contemporary medicaltherapy, AF was no longer associated with decreasedsurvival.3

The data presented in this study are consistent withand extend the most recent findings reported byStevenson et al.3 An important difference between ourdata and those of Stevenson et al3 is the definition ofevent-free survival used in this study. The need topredict freedom not only from death, but also fromneed for transplantation or left ventricular assist de-vice placement is an important clinical question in theevaluation of patients with advanced heart failure be-ing considered for cardiac transplantation.

These findings are consistent with the observationthat AF may not incrementally worsen the advancedhemodynamic disturbances seen in advanced heartfailure.13–16 Further, as others have proposed, thegreater heart rate variability manifest at rest and dur-

ing exercise by patients in advanced heart failure inAF compared with NSR13–16may compensate for thedecrease in ventricular filling secondary to loss ofatrial systole.17,18

Limitations to this study include the cross-sectionaldesign, the sample size, and the specific referral pop-ulation studied. We may not have been able to detectsmall differences in survival, and these results maynot be generalizable to patients with less severe heartfailure symptoms. We also did not address the issue ofpatients with advanced heart failure who subsequentlydeveloped AF after initial evaluation.

1. Carson PE, Johnson GR, Dunkman WB, Fletcher RD, Farrell L, Cohn JN : Theinfluence of AF on prognosis in mild to moderate heart failure.Circulation1993;87(suppl VI):VI-102–VI-110.2. Stevenson WG, Ganz P. Atrial fibrillation in heart failure.Heart Failure1997;13:22–29.3. Stevenson WG, Stevenson LW, Middlekauff HR, Fonarow GC, Hamilton MA,Woo MA, Saxon LA, Natterson PD, Steimle A, Walden JA, Tillisch JH. Im-proving survival for patients with atrial fibrillation and advanced heart failure.J Am Coll Cardiol1996;28:1458–1463.4. Middlekauff H, Stevenson WG, Stevenson LW. Prognostic significance ofatrial fibrillation in advanced heart failure.Circulation 1991;84:40–48.5. Aaronson KD, Schwartz JS, Chen TM, Wong KL, Goin JE, Mancini DM.Development and prospective validation of a clinical index to predict survival inambulatory patients referred for cardiac transplant evaluation.Circulation 1997;95:2660–2667.6. Keogh A, Baron DW, Hickie JB. Prognostic guides in patients with idiopathicor ischemic dilated cardiomyopathy assessed for cardiac transplantation.Am JCardiol 1990;65:903–908.7. Roberts WC, Siegel RJ, McManus BM. Idiopathic dilated cardiomyopathy:analysis of 152 necropsy patients.Am J Cardiol1987;60:1340–1355.8. Hofmann T, Meinertz T, Kaspar W Geibel A, Zehender M, Hohnloser S,Stienen U, Treese N, Just H. Mode of death in idiopathic dilated cardiomyopathy:a multivariate analysis of prognostic determinants.Am Heart J1988;116:1455–1483.9. Unverferth DM, Magorien RD, Moeschberger ML, Baker PB, Fetters JK, LeierCV. Factors influencing one year mortality of dilated cardiomyopathy.Am JCardiol 1984;54:147–152.10. Romeo F, Pellicia F, Cianfrocca C, Cristofani R. Predictors of sudden deathin idiopathic dilated cardiomyopathy.Am J Cardiol1989;63:138–140.11. Convert G, Delaye J, Biron A, Gonin A. Etude pronostique des myocardio-pathies primitives non obstructives.Arch Mal Coeur1980;73:227–237.12. Dries DL, Gersh BJ, Domanski MJ, Stevenson LW. Atrial fibrillation in mildto moderate heart failure is independently associated with progressive pumpfailure and death (abstr). J Am Coll Cardiol1998;31:219A.13. Ueshima K, Myers J, Ribisl PM, Atwood JE, Morris CK, Kawaguchi T, LiuJ, Froelicher VF. Hemodynamic determinants of exercise capacity in chronicatrial fibrillation. Am Heart J1993;125:1301–1305.14. Smith RF, Johnson G, Ziesche S, Bhat G, Blankenship K, Cohn JN. Func-tional capacity in heart failure: comparison of methods for assessment and theirrelation to other indices of heart failure.Circulation 1993;87(suppl VI):VI-88–VI-93.15. Atwood JE, Myers J, Sullivan M, Forbes S, Sandhu S, Callaham P, FroelicherV. The effect of cardioversion on maximal exercise capacity in patients withchronic atrial fibrillation.Am Heart J1989;118:913–918.16. Lipkin DP, Frenneaux M, Stewart R, Joshi J, Lowe T, McKenna WJ. Delayedimprovement in exercise capacity after cardioversion of atrial fibrillation to sinusrhythm.Br Heart J 1988;59:572–577.17. Kono T, Sabbah HN, Rosman H, Alam M, Stein PD, Goldstein S. Left atrialcontribution to ventricular filling during the course of evolving heart failure.Circulation 1992;86:1317–1322.18. Gosselink AT, Blanksma PK, Crijns HJ, Van Gelder IC, de Kam PJ, HillegeHL, Niemeijer MG, Lie KI, Meijler FL. Left ventricular beat to beat performancein atrial fibrillation: contribution of Frank Starling mechanism after short ratherthan long RR intervals.J Am Coll Cardiol1995;626:1516–1521.

TABLE V Predictors of Decreased Survival

Univariate

HazardRatio 95% CI p Value

NYHA class 2.87 1.93–4.25 ,0.001Pulmonary capillary wedge

pressure*1.32 1.21–1.52 ,0.001

Cardiac index 0.43 0.28–0.68 ,0.001Right atrial mean* 1.32 1.21–1.52 ,0.01Pulmonary artery systolic* 1.08 1.03–1.10 ,0.01Maximum exercise O2

consumption0.72 0.58–0.88 ,0.01

Amiodarone 1.78 1.11–2.80 ,0.02Atrial fibrillation 1.21 0.80–1.62 0.11

Multivariate

NYHA class† 2.64 1.69–4.11 ,0.001Pulmonary capillary wedge

pressure*1.37 1.24–1.47 ,0.001

*Analyzed as a continuous variable; hazard ratio expressed per 5 mm Hgincrease.

†Class IV versus class II/III.O2 5 oxygen.

CONGESTIVE HEART FAILURE/ATRIAL FIBRILLATION, HEART FAILURE, AND PROGNOSIS 1547