Author + information
- Received November 23, 2004
- Revision received January 25, 2005
- Accepted March 1, 2005
- Published online June 21, 2005.
- Brian Olshansky, MD⁎,⁎ (, )
- Eliot N. Heller, MD†,
- L. Brent Mitchell, MD‡,
- Mary Chandler, MD§,
- William Slater, MD∥,
- Martin Green, MD¶,
- Michael Brodsky, MD#,
- Patrick Barrell, BS§,
- H. Leon Greene, MD§,
- AFFIRM Investigators
- ↵⁎Reprint requests and correspondence:
Dr. Brian Olshansky, University of Iowa Hospitals, 200 Hawkins Drive, Iowa City, Iowa 52242.
Objectives The purpose of this study was to evaluate the associations of transthoracic echocardiographic parameters with recurrent atrial fibrillation (AF) and/or stroke.
Background The Atrial Fibrillation Follow-up Investigation of Rhythm Management (AFFIRM) study, an evaluation of elderly patients with AF at risk for stroke, provided an opportunity to evaluate the implications of echocardiographic parameters in patients with AF.
Methods Transthoracic echocardiographic measures of mitral regurgitation (MR), left atrial (LA) diameter, and left ventricular (LV) function were evaluated in the AFFIRM rate- and rhythm-control patients who had sinus rhythm resume and had these data available. Risk for recurrent AF or stroke was evaluated with respect to transthoracic echocardiographic measures.
Results Of 2,474 patients studied, 457 had ≥2+/4+MR, and 726 had a LA diameter >4.5 cm. The LV ejection fraction was abnormal in 543 patients. The cumulative probabilities of at least one AF recurrence/stroke were 46%/1% after 1 year and 84%/5% by the end of the trial (> 5 years), respectively. Multivariate analysis showed that randomization to the rhythm-control arm (hazard ratio [HR] = 0.64; p < 0.0001) and a qualifying episode of AF being the first known episode (HR = 0.70; p < 0.0001) were associated with decreased risk. Duration of qualifying AF episode >48 h (HR = 1.55; p < 0.0001) and LA diameter (p = 0.008) were associated with an increased risk of recurrent AF. Recurrent AF was more likely with larger LA diameters (HR = 1.21, 1.16, and 1.32 for mild, moderate, and severe enlargement, respectively). No transthoracic echocardiographic measures were associated with risk of stroke.
Conclusions In the AFFIRM study, large transthoracic echocardiographic LA diameters were associated with recurrent AF, but no measured echocardiographic parameter was associated with stroke.
Atrial fibrillation (AF) occurs most frequently in elderly patients (1), who are at risk for recurrent AF if an episode resolves or is cardioverted. Transthoracic echocardiographic parameters, including left atrial (LA) diameter, left ventricular (LV) ejection fraction, and mitral regurgitation (MR), may be associated with recurrent AF, stroke, or death (2–7).
The management of AF includes the prevention of thromboembolic events, slowing the ventricular response rate, and the establishment, followed by maintenance, of sinus rhythm. The Atrial Fibrillation Follow-Up Investigation of Rhythm Management (AFFIRM) study, a large, multicenter, randomized trial sponsored by the National Heart, Lung, and Blood Institute, compared rate- and rhythm-control approaches for AF patients at high risk for stroke or death. The primary end point was all-cause mortality (8–10).
The rate-control strategy included treatment designed to prevent thromboembolic events and to control the ventricular response rate, but not to establish or maintain sinus rhythm. The rhythm-control strategy included therapy designed to maintain sinus rhythm and to prevent thromboembolic events. The AFFIRM study offered the opportunity to evaluate echocardiographic LA diameter, LV ejection fraction, and/or MR measures as predictors of AF recurrence or stroke in patients with AF at high risk for stroke or death.
The AFFIRM study design, detailed elsewhere (8), was approved by the AFFIRM Steering Committee, by the Institutional Review Board of the University of Washington (for the Clinical Trial Center), by the Data and Safety Monitoring Board, and by each participating site’s local institutional review board. All patients gave informed, written consent to participate in the trial. The AFFIRM study indicated that the rhythm-control strategy offered no survival benefit (10).
This substudy considered all AFFIRM study patients in the rate- and rhythm-control arms in whom: 1) restoration of sinus rhythm was documented within two months of randomization, either spontaneously or with electrical and/or chemical cardioversion, and 2) a transthoracic echocardiogram was reported. We evaluated the association of specific echocardiographic parameters with AF recurrence and stroke (7,9,10).
Recurrent AF was identified on an electrocardiogram (ECG) or rhythm strip after the qualifying episode of AF had stopped spontaneously or was converted electrically and/or pharmacologically to sinus rhythm. Twelve-lead ECGs were obtained at 4 and 12 months after randomization, and a rhythm strip was obtained at each 4-month visit. The ECGs or rhythm strips were also recorded at any time that symptoms suggested AF recurrence. The time between restoration of sinus rhythm and AF recurrence was analyzed.
Stroke was defined as a focal neurologic deficit of any severity lasting >24 h. All strokes were adjudicated by a Stroke Events Committee composed of neurologists blinded to the treatment assignment.
Treatment during follow-up
Patients assigned to the rhythm-control strategy in the AFFIRM study were prescribed antiarrhythmic drugs commonly used to establish and/or maintain sinus rhythm. Patients assigned to the rate-control strategy received digoxin, non-dihydropyridine calcium antagonists, and/or beta-blockers (10). The treating physician prescribed rhythm- and rate-control drugs based on prespecified guidelines (7,9–11). Adequacy of rhythm- and rate-control was determined by history, physical examination, and an ECG rhythm strip performed at least at each four-month visit. Rate- and rhythm-control drugs were adjusted as necessary. Cardioversion was performed as needed in the rhythm-control group. If adequate rhythm or rate control could not be achieved with at least two medications given separately, the treating physician could proceed with an innovative therapy (7).
Patients were included in this study if they had a transthoracic two-dimensional and M-mode echocardiogram (with Doppler measurements to assess MR). The test was performed with various types and models of equipment and by various cardiac sonographers at 213 clinical sites, according to local practice. Echocardiograms performed within 12 months of qualifying AF episodes were reported and interpreted locally. The echocardiogram was performed within 30 days in 1,557 (63%) patients, between 31 and 59 days in 267 (11%) patients, between 60 and 89 days in 170 (7%) patients, and at >90 days in 479 (19%) patients. Measurements collected were LA diameter, LV ejection fraction, and qualitative assessment of MR. Transesophageal echocardiography data were not recorded for purposes of this trial.
Baseline clinical data were tabulated. Follow-up data related to AF recurrence and stroke was collected at two months and at four-month intervals thereafter.
The LA size was analyzed both as a dichotomous variable (≤4.5 cm vs. >4.5 cm) and as a categorical variable (<4.1 cm, 4.1 to 4.5 cm, 4.6 to 5.5 cm, or >5.5 cm). Mitral regurgitation was analyzed as a dichotomous variable (≥2+/4+vs. <2+). The LV ejection fraction, analyzed as a categorical variable, was divided into normal (≥50%), mild dysfunction (40% to 49%), moderate dysfunction (30% to 39%) and severe dysfunction (<30%).
Means and standard deviations were calculated for continuous variables. Frequencies and percentages were calculated for categorical variables. Continuous variables were analyzed using ttests. Categorical variables were analyzed using chi-square tests for homogeneity.
Cumulative survival probabilities were calculated by the Kaplan-Meier method. Survival distributions were compared using log-rank statistics.
Cox proportional hazards models were used for multivariate analyses of time to recurrent AF and time to stroke. An initial stepwise analysis was used to select significant demographic variables (age, gender, qualifying episode as first AF episode, duration of qualifying episode >48 h, history of: coronary artery disease [CAD], congestive heart failure [CHF], hypertension, diabetes, smoking, stroke, and myocardial infarction, and New York Heart Association functional class at baseline). A full model containing the demographic variables that were selected in the stepwise analysis, plus treatment arm and the echocardiographic variables (LA size, MR ≥2+and LV ejection fraction) was run. A second stepwise model was used to select the significant echocardiographic variables after controlling for significant demographic variables and treatment arm. Those variables found significant in this stepwise model were included in the final, reduced model. Significance was indicated by a p value <0.05.
The AFFIRM study enrolled 4,060 patients; 2,033 were randomized to the rhythm-control arm and 2,027 to the rate-control arm. A transthoracic echocardiogram was performed and reported for 3,311 patients (82%). Sinus rhythm resumed in 3,043 patients by the two-month visit; 2,474 (81%) of these patients had an echocardiogram reported and are the subjects of this analysis (Table 1).The mean time between the echocardiogram and the date of qualifying episode of AF was 49 ± 110 days.
Among patients who did not have sinus rhythm by the 2-month visit, a high proportion (70%) had been randomized to the rate-control arm of AFFIRM study and/or had experienced a qualifying episode of AF lasting >48 h. Otherwise, baseline characteristics of patients included in this analysis were similar to those who had echocardiograms but remained in AF and those who did not have an echocardiogram (Table 1).
Of the 2,474 patients in this analysis, 457 of 2,426 (19%) had ≥2+/4+MR, 264 of 1,918 (14%) had LV diastolic size >5.7 cm, 306 of 1,696 (18%) had LV systolic size >4.0 cm, 543 of 2,255 (24%) had an LV ejection fraction < 0.50, and 726 of 2,334 (31%) had an LA diameter >4.5 cm.
In the rhythm-control arm, amiodarone was the first antiarrhythmic drug prescribed in 448 patients (36%). Other antiarrhythmic drugs used initially were sotalol (438 patients, 35%), propafenone (127 patients, 10%), quinidine (69 patients, 6%), flecainide (65 patients, 5%), procainamide (68 patients, 5%), disopyramide (29 patients, 2%), and moricizine (10 patients, 0.8%). Of those taking an antiarrhythmic drug other than amiodarone as the first drug, 312 of 837 (37%) were prescribed amiodarone later.
In the rate-control arm, 504 patients (52%) were prescribed a beta-blocker and 491 patients (61%) were prescribed a calcium-channel blocker as the first rate-controlling drug (some received both). Overall, 361 (38%) were prescribed an angiotensin-converting enzyme (ACE) inhibitor.
Associations with LA diameter
Among the 2,334 patients who had their LA diameter documented, the mean follow-up time was 1,298 ± 460 days (range, 0 to 2,167 days). Of these, 999 were in the rate-control arm and 1,335 were in the rhythm-control arm (Table 2).
Patients with an enlarged LA (>4.5 cm) had significantly larger LV diameters, were more likely to be male, and were more likely to have CHF, angina, myocardial infarction, and an abnormal LV ejection fraction than patients with a smaller LA. In patients with an enlarged LA, the AF episode serving as the qualifying episode for the AFFIRM study was more likely to persist >48 h. Primary diagnoses differed in patients with and without an enlarged LA. A larger proportion of patients with an enlarged LA had a primary diagnosis of CAD or cardiomyopathy. Thirty-three percent with LA ≤ 4.5 cm were prescribed an ACE inhibitor, versus 49% with LA > 4.5 cm (p < 0.0001).
Patients with large LA diameters were more likely to be prescribed amiodarone as the first antiarrhythmic drug (34% with LA <4.5 cm vs. 40% with LA ≥4.5 cm; p = 0.03). Patients with MR were more likely to be prescribed amiodarone and ACE inhibitors as first drugs.
The AF recurred in 1,128 patients in the first year and in 1,686 patients by the end of the trial. The Kaplan-Meier estimates of the probabilities of AF recurrence were 46% after one year and 84% at the end of the trial. The risk of recurrent AF did not differ between those who had an echocardiogram performed and those who did not.
In a multivariate model of time to recurrent AF after establishment of sinus rhythm, having a first episode of AF as the qualifying episode and being randomized to the rhythm-control arm of the AFFIRM study were associated with a decreased risk of recurrent AF, whereas a qualifying episode lasting ≥48 h was associated with an increased risk of recurrent AF. After controlling for these variables, LA diameter was associated with recurrent AF (Table 3),with larger LA diameters associated with increased risk of recurrent AF compared with patients with normal LA diameters (Fig. 1).
Occurrence of stroke
Stroke occurred in 29 patients in the first year and in 86 patients by the end of the trial. The Kaplan-Meier estimates of the probability of stroke were 1% after one year and 5% at the end of the trial. The risk of stroke in patients who had an echocardiogram performed was not significantly different from that in patients who did not have an echocardiogram performed.
In a multivariate model of time to occurrence of stroke, female gender, history of stroke, history of myocardial infarction, and a qualifying episode lasting ≥48 h were associated with increased follow-up risk of stroke. After controlling for these variables, no transthoracic echocardiographic variables studied were associated with the risk of stroke (Table 4).Although MR ≥2+was of borderline significance in a model that included significant demographic variables plus all echocardiographic variables, when a stepwise selection model was run, neither MR nor any of the other echocardiographic parameters met the selection criterion of p < 0.05.
This AFFIRM substudy evaluated patients with AF treated with a rate- or a rhythm-control approach. It evaluated the association of frequently measured transthoracic echocardiographic parameters with AF recurrence and stroke. The follow-up probability of at least one episode of AF recurrence was high, even in the patients randomized to the rhythm-control arm (10,11). The LA diameter was associated with recurrent AF in the rate- and rhythm-control groups, but MR and depressed LV ejection fraction were not. No transthoracic echocardiographic parameter predicted stroke.
Echocardiographic predictors of AF recurrence
The initial occurrence of AF has been associated with LA enlargement, as determined by roentgenography (12), autopsy (13), or angiography (14). Small patient cohorts suggest that echocardiographic measurements of LA diameter correlate with the probability of maintaining sinus rhythm after spontaneous, chemical, or direct current cardioversion of AF. This association has been supported by some (4,15–21) but not all (22–25) studies.
The LA volume measured by transthoracic echocardiography may be superior to LA diameter for predicting adverse outcomes, including recurrent AF. Perhaps LA volume would be a more robust predictor of AF recurrence than LA diameter used here.
Previous reports showed an association between LA size and AF recurrence rates. Flaker et al. (26) performed a multivariate analysis on potential clinical and transthoracic echocardiographic predictors of AF recurrence in 341 patients in the Stroke Prevention in Atrial Fibrillation (SPAF)-I and -II trials. Recurrent AF was predicted by advancing age, presence of CHF, prior myocardial infarction, and LA diameter. Similarly, Alt et al. (27) performed a multivariate analysis on potential clinical and echocardiographic predictors of AF recurrence in 187 patients treated with sotalol after direct-current cardioversion of AF. Recurrent AF was predicted by advancing age, presence of coronary artery disease, prolonged (>2 months) index episode of AF, and LA diameter >6.0 cm. Notably, these analyses were performed on populations for which a limited number of co-variables could be assessed.
The primary analysis of this study involved 2,474 patients, a large population permitting a more inclusive multivariate analysis that confirmed an independent association of LA diameter with AF recurrence. The present study is more contemporary and includes a large patient population receiving antiarrhythmic drug therapy with frequent use of amiodarone. In contrast, it is noteworthy that two univariate analyses not reporting an association between LA diameter and AF recurrence were of patients receiving amiodarone therapy. However, these prior studies may have used amiodarone for refractory cases, unlike the treatment approach in the AFFIRM study (22,23).
Other potential transthoracic echocardiographic predictors of AF recurrence have been studied less extensively. The LV end-diastolic dimension has been suggested as a predictor of recurrent AF in one study (17), but not another (25). Fractional shortening did not predict recurrent AF in two studies (4,25), nor did LV wall thickness (25). We found no association between LV systolic function as measured by the ejection fraction and the follow-up probability of recurrent AF.
Echocardiographic predictors of thromboembolic events
Echocardiographic parameters have been evaluated as predictors of thromboembolic events, particularly thromboembolic stroke in follow-up. The most frequently studied echocardiographic parameter has been the LA diameter. Some studies have reported a statistically significant association between LA diameter and thromboembolic events (28,29); others have not (30–32). Multivariate analyses likewise yielded conflicting results (33–37).
Other transthoracic echocardiographic parameters have been evaluated in patients with AF in relation to thromboembolic events. Some univariate associations have been observed between thromboembolic events and LV end-diastolic diameter (38), LV end-systolic diameter (38), fractional shortening (38), and LV wall thickness or ventricular mass (29,38,39). Other univariate analyses have found no associations between thromboembolic events and LV end-diastolic diameter (30,35), LV end-systolic diameter (35), fractional shortening (29,37), or LV wall thickness or ventricular mass (30,35,37). None of these associations have been supported in multivariate analyses (33–37), with the exception of LV mass in one study (39) and qualitative assessment of global LV systolic dysfunction in another (38).
In the present study, neither LA diameter nor any other studied transthoracic echocardiographic parameter was associated with thromboembolic events in univariate or multivariate analyses. Transesophageal echocardiographic data (e.g., appendage velocities, appendage thrombus, and smoke), not recorded or assessed in this study, may have provided better predictive measures of stroke risk.
Mitral regurgitation and AF often coexist. In fact, 457 of 2,426 (19%) of the AFFIRM study patients had echocardiographic evidence of ≥2+/4+MR. Mitral regurgitation can cause atrial stretch and LA enlargement, either of which may change atrial electrophysiologic properties and alter atrial anatomy. Accordingly, MR may facilitate the occurrence of AF and make it more difficult to control with antiarrhythmic drug therapy.
The effect of the MR jet to lessen LA stasis may protect the LA from developing thrombi, whether or not the patient is taking an anticoagulant (40,41). Previous studies evaluating the relationship between MR and thromboembolic events (or their surrogates of LA thrombus or LA spontaneous echo contrast) provide conflicting results, but many of these studies involve patients with rheumatic mitral valve disease, and these data do not pertain to this patient population. Studies associating MR with a lower risk of thromboembolic events or stroke in patients with nonrheumatic AF reported disparate results. Some support a relationship (40,42), and some dispute it (31,32,38). In the present study not restricted to subjects with rheumatic heart disease, no statistically significant association was found between MR determined by the transthoracic echocardiogram and the follow-up probability of recurrent AF or thromboembolic stroke.
Patients randomized into the AFFIRM study had to be candidates for a therapeutic approach that attempted to establish and maintain sinus rhythm. Possibly, patients with exceedingly large LA diameters and those with severe MR were not included in the AFFIRM study because of the bias that they may not maintain sinus rhythm. The LA diameter was not corrected for body surface area. Those with larger LA diameters were more likely to be men, were sicker and, therefore, were less likely to maintain sinus rhythm. Patients with severe MR who were going to undergo valvular replacement or repair would not have been included in this study. Accordingly, this study could have been biased toward a less strong association between LA diameter and/or MR and the probabilities of recurrent AF and stroke. Also, AF may have recurred, been transient, or gone unrecognized. Given the large numbers of patients observed, it is likely that even if some recurrences were missed, the comparison between subgroups provided accurate differences in recurrence rates.
Entry criteria into the AFFIRM study did not require a transth oracic echocardiogram. Accordingly, not all patients had one performed, and those that did may have had an echocardiogram performed at a time other than at randomization. Those patients who had an echocardiogram performed differed slightly from patients who did not. Inclusion of only patients who had a transthoracic echocardiogram performed may incur study bias and impact the results.
Temporal changes in echocardiographic parameters for individual patients cannot be identified by a single echocardiogram. This analysis was limited to evaluation of widely measured and reported but locally performed and locally interpreted transthoracic echocardiographic parameters. Although this assessment represents the mainstream of echocardiographic practice, it is possible that if all transthoracic echocardiograms had been performed and/or interpreted at one site, results may have differed. It is also possible that similar parameters measured by transesophageal echocardiography may have provided different results or that other transthoracic echocardiographic or transesophageal echocardiographic parameters are associated with AF recurrence or stroke.
The incomplete nature of the AFFIRM study drug database complicates analysis. Patients could be and were switched from one antiarrhythmic drug to another and placed on and off warfarin. Although data regarding drug use were tabulated at each follow-up visit, the frequency and the dosage were not recorded. There was no way to distinguish whether the drug was taken for a few dosages or over the entire interval between follow-up visits.
Finally, the low stroke rate observed in this study may limit the power of an analysis to associate echocardiographic parameters with stroke.
For patients in the AFFIRM study who resumed sinus rhythm after an episode of AF, recurrent AF was associated with echocardiographic measurement of the LA diameter. The MR and LV ejection fraction, as measured by a standard transthoracic echocardiogram, were not associated with recurrent AF. No transthoracic echocardiographic parameter evaluated was associated with stroke occurrence.
Supported by the National Heart, Lung, and Blood Institute, Bethesda, Maryland, under contract N01-HC-55139. The AFFIRM Investigators and their affiliations are listed in reference 17.
- Abbreviations and Acronyms
- angiotensin-converting enzyme
- atrial fibrillation
- Atrial Fibrillation Follow-Up Investigation of Rhythm Management
- coronary artery disease
- congestive heart failure
- left atrial/atrium
- left ventricular/ventricle
- mitral regurgitation
- Received November 23, 2004.
- Revision received January 25, 2005.
- Accepted March 1, 2005.
- American College of Cardiology Foundation
- Psaty B.M.,
- Labovitz A.J.
- Henry W.L.,
- Morganroth J.,
- Pearlman A.S.,
- et al.
- AFFIRM First Antiarrhythmic Drug Substudy Investigators
- Davies M.J.,
- Pomerance A.
- Probst P.,
- Goldschlager N.,
- Selzer A.
- Vaziri S.M.,
- Larson M.G.,
- Benjamin E.J.,
- Levy D.
- Flaker G.C.,
- Fletcher K.A.,
- Rothbart R.M.,
- Halperin J.L.,
- Hart R.G.