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

Analysis

The primary objective of this analysis was to examine the risk of cardiovascular (CV) events related to baseline AF according to baseline EF. We also assessed the frequency of CV events in patients in whom new AF developed during follow-up. These were prespecified analyses, but the data on new development of AF were collected without relationship to the timing of new-onset AF. In secondary analyses, we examined the influence of baseline rhythm (AF or other) on the effect of candesartan on outcomes and on the need for permanent withdrawal from study drug because of serious adverse effects such as hypotension, hyperkalemia, or increased serum creatinine level. Patients were classified as having AF or no AF according to the investigator interpretation of their baseline ECG. Patients in sinus rhythm at baseline but with a history of AF were categorized as no AF. Analyses were carried out for all patients or with patients divided by EF. The CHARM-Added and CHARM-Alternative participants were considered the low EF group and CHARM-Preserved participants were considered the PEF group. The primary outcome of the component trials in the CHARM program was the composite of CV death or unplanned admission to the hospital for the management of worsening CHF, and these events were adjudicated by a blinded committee. Prespecified secondary outcomes included all-cause mortality, CV death, admission to the hospital for CHF, and fatal or nonfatal stroke. We classified all deaths as CV unless an unequivocal non-CV cause was established.

Statistical analysis

All outcome variables were defined as the time to an event or censoring and were analyzed with the proportional hazards model. Both simple Cox regression models and multiple Cox regression models were fitted to data. The explanatory variables included in the multiple regression models were the same set of 33 variables that were adjusted for in the CHARM program (20), except for the variable ACE-I at baseline. All subgroups analyzed for the low EF group were stratified by component trial (CHARM-Added or CHARM-Alternative). When an analysis included new-onset AF as an explanatory variable, a logistic regression model was fitted to data because information on timing of occurrence of AF was lacking, and these analyses are therefore presented as odds ratios rather than hazard ratios (HR). In this analysis, the response was considered to be a binary variable, indicating whether or not a patient experienced a CV event. All p values were generated from the Wald test statistics.

Baseline characteristics

A total of 7,601 patients were randomized to candesartan or placebo, 7,599 with available data. Baseline characteristics are summarized in Table 1.In general, patients with AF were older, had a higher baseline heart rate, and more often had a cardiothoracic ratio of >0.5 and a history of hospitalization for heart failure. Patients with AF less frequently had a history of prior myocardial infarction, angina pectoris, or diabetes mellitus. Ischemic heart disease was the most common etiology of heart failure regardless of presence or absence of AF, but it was a less common cause in patients with AF (43.5%) than without AF (64.8%). Hypertension was reported as the etiology of HF more commonly in patients with AF, much more commonly in patients with PEF, and most commonly in those with AF and PEF. Patients with AF had similar EF to those in sinus rhythm but had a worse New York Heart Association functional classification and were more likely to have had a pacemaker implanted. Patients with AF were more often treated with digitalis, diuretic agents, and oral anticoagulants and less often with beta-blocker and acetylsalicylic acid. A small number of patients were treated with an antiarrhythmic drug in either group, regardless of AF, with more frequent use in patients with a low EF.

Outcomes by baseline AF

Atrial fibrillation recorded on baseline ECG was associated with an increased risk of morbidity and mortality (Figs. 1 to 3).⇓⇓⇓Patients with AF and low EF had the highest absolute risk for adverse CV outcomes (e.g., 45% with CV death or CHF hospitalization) relative to those with low EF and sinus rhythm (37% with an event), PEF and AF (34% with an event), or PEF and sinus rhythm (21% with an event) (Figs. 1 to 3). However, AF was associated with a greater increase in the risk of CV death or hospitalization for worsening heart failure in patients with PEF (HR 1.72, 95% confidence interval [CI] 1.45 to 2.06) than in patients with low EF, (HR 1.29, 95% CI 1.14 to 1.46, p for interaction 0.008). The same was true for all-cause mortality: PEF HR 1.80 (95% CI 1.46 to 2.21) and low EF HR 1.38 (95% CI 1.21 to 1.59, p for interaction 0.041). Similarly, for each of these adverse CV outcomes, with the exception of stroke, patients with low EF and AF at baseline had the highest absolute risk, and patients with PEF and AF at baseline had a greater increase in risk than those with low EF and AF. When adjusted for 32 covariates in multiple regression analysis, baseline AF remained an independent risk factor for CV death or hospitalization for heart failure in patients with PEF (HR 1.32, 95% CI 1.06 to 1.65, p = 0.015) but not in those with low EF (HR 1.12, 95% CI 0.97 to 1.29, p = 0.12). After covariate adjustment, AF at baseline remained an independent predictor of all-cause mortality regardless of baseline EF: PEF HR 1.37 (95% CI 1.06 to 1.79) and low EF HR 1.22 (95% CI 1.04 to 1.43).

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

Risk of baseline atrial fibrillation (AF) for cardiovascular (CV) events depending on ejection fraction (EF). CHF = chronic heart failure.

• Download figure
• Open in new tab
• Download powerpoint

Figure 2

Time to cardiovascular (CV) death or hospitalization because of heart failure. AF = atrial fibrillation; EF = ejection fraction; PEF = preserved left ventricular ejection fraction.

• Download figure
• Open in new tab
• Download powerpoint

Figure 3

Time to all-cause mortality. AF = atrial fibrillation; CI = confidence interval; EF = ejection fraction; PEF = preserved left ventricular ejection fraction.

Outcomes in patients with new-onset AF

In 392 patients, AF developed during follow-up, 263 (7.8%) in the low EF group and 129 (4.9%) in the PEF group. Patients with new-onset AF experienced a higher risk of morbidity and mortality regardless of baseline EF. The odds ratio for CV death or hospitalization for worsening heart failure was 4.22 (95% CI 2.90 to 6.13) in the PEF group and 3.17 (95% CI 2.45 to 4.09) in the low EF group; p = 0.19 for interaction. For all-cause mortality the odds ratio was 2.57 (95% CI 1.70 to 3.90) in the PEF group and 1.85 (95% CI 1.44 to 2.37) in the low EF group; p = 0.18 for interaction. Again, the absolute risk of an adverse CV outcome was highest in the low EF-new AF patient group, but the patients with PEF and new AF had a greater relative increase in risk than those with low EF and new AF (Fig. 4).For example, the absolute risk of CV death or CHF hospitalization was 66% in the low EF-new AF group and was increased from 20% to 47% by the new development of AF in the PEF group (Fig. 4).

• Download figure
• Open in new tab
• Download powerpoint

Figure 4

Risk of new-onset atrial fibrillation (AF) for cardiovascular (CV) events depending on ejection fraction (EF). CHF = chronic heart failure.

Treatment effects

In the CHARM-Overall program, candesartan reduced the risk of the composite primary end point in both AF (HR 0.83, 95% CI 0.69 to 0.99) and non-AF groups (HR 0.84, 95% CI 0.77 to 0.92; p for interaction 0.80). There were also trends toward a reduced risk of all-cause mortality in both groups that did not reach statistical significance, HR 0.82 (95% CI 0.67 to 1.01) for AF and HR 0.94 (95% CI 0.85 to 1.04) for no AF, p for interaction 0.22 (Fig. 5).In the PEF group there was no significant effect of candesartan on the primary composite end point in the overall population and no difference in effect according to presence or absence of AF (p for interaction 0.82). In patients with low EF, candesartan reduced CV morbidity and mortality to a similar degree regardless of the presence of AF at baseline (Table 2).

• Download figure
• Open in new tab
• Download powerpoint

Figure 5

Treatment effects depending on baseline heart rhythm in CHARM-Overall. AF = atrial fibrillation; CHF = chronic heart failure; CV = cardiovascular.

Adverse effects

Patients with baseline AF had higher discontinuation rates than those without because of hypotension (3.4% vs. 2.5%), increased creatinine (6.3% vs. 4.3%), and hyperkalemia (1.8% vs. 1.3%), respectively.

Baseline characteristics

Patients with AF were in general older, had a worse functional class, and had more enlarged hearts despite EF similar to patients with no AF. Several clinical trials, outpatient cohorts, and epidemiologic studies have shown a mutual relationship between the two conditions in which heart failure begets AF (13,22) and AF begets CHF (13,23). In 9,193 patients with hypertension and left ventricular hypertrophy, new-onset AF was associated with a five-fold increase in hospitalizations for CHF (24). Beta-blockers were less often used in patients with AF, which could reflect the higher age and more co-morbidity. Treatment with digitalis was more frequent among AF patients, but higher baseline heart rates suggest that single-drug therapy may be insufficient for adequate rate control (25). Around 75% of patients with AF were treated with oral anticoagulants regardless of baseline EF. This is higher than reported from the EuroHeart Failure Survey (26).

Outcomes

Our analysis extends previous findings by showing that AF, whether present at baseline or of new onset, is an independent risk factor for adverse CV outcomes in patients with heart failure regardless of baseline EF. After adjustment for baseline factors, there are still differences in risk increase associated with AF between patients with low EF and PEF. The reason for the difference is unclear, but the presence of AF may reflect different stages of CHF between the two patient groups. Although patients with a low EF are at a higher absolute risk for CV events than those with PEF (Figs. 1 to 4) (27), the added risk of AF is particularly important for patients with PEF. Patients with PEF might have a lower tolerability for uncontrolled ventricular rate, thus being more prone to deteriorate (3,4).

Earlier studies

To this date, we are aware of only one trial published in full evaluating the prognostic impact of supraventricular tachycardia (including AF) in heart failure patients with evidence of preserved systolic function and clinical evidence of heart failure. The DIG trial included 988 patients with LVEF >0.45 and clinical signs of heart failure in an ancillary study, and these patients were analyzed for incidence of supraventricular tachycardia together with patients in the main trial. The development of supraventricular tachycardia during the study was independently associated with a higher risk of mortality and stroke regardless of baseline EF (28).

In the previous literature there is conflicting evidence about the prognostic importance of AF. Four of five trials with follow-up started before 1990 showed an independent relationship between AF and mortality: one was an epidemiologic study (13); one included patients in the SOLVD trial, in which AF also predicted CV morbidity (6); and two were in severe heart failure (9,10). One (12) of five (7,8,10,11) trials with follow-up starting in 1990 or later showed an independent relationship between AF and mortality. That was an unselected cohort of 944 elderly patients hospitalized for heart failure. AF did not predict 1-month rehospitalization. The other substantial recent analysis was of the 3,029 patients randomized to either carvedilol or metoprolol tartrate and followed up for 5 years within the COMET study (8). The AF did not independently predict all-cause mortality, but did predict all-cause mortality or hospitalization and CV death or hospitalization for heart failure. Of note, in trials with follow-up started in 1990 or later, patients were almost universally treated with ACE-I and the use of class I antiarrhythmic drugs had decreased considerably.

Four of five published analyses of new-onset AF showed a prognostic relationship with mortality, one an analysis from the Framingham study (13). and three analyses from the DIG, COMET, and Val-HeFT clinical studies (8,28,29). The exception was a small group of patients with severe CHF enrolled in the PRIME II study (7).

Angiotensin II receptor blockers in patients with heart failure and AF

This is the first trial to evaluate the effect of an angiotensin II receptor blocker on CV outcomes in patients with CHF and concomitant AF, and moreover, over a broad range of EF. In Val-HeFT study, valsartan reduced the incidence of AF in patients with reduced EF (29). In a recent analysis of the CHARM program, we have shown that candesartan reduces the risk of developing AF in patients with both reduced and preserved EF (30). The patients enrolled in the CHARM-Preserved component trial, which had symptomatic CHF and PEF, had some similarities to those in the LIFE trial, in which losartan was superior to atenolol in reducing a composite end point of CV death, stroke, or myocardial infarction in 342 patients with hypertension, left ventricular hypertrophy, and AF despite similar decreases in blood pressure (31).

Study limitations

The primary analysis of this study was based on the diagnosis of AF using the baseline ECG. New-onset AF was documented at the end of the study from investigator reports and not by systematic recordings, so episodes of AF may have been missed, particularly paroxysmal AF. Also, patients who died early did not have the opportunity to develop AF, so collectively the relationship reported here probably underestimates the true relationship of new-onset AF and mortality. The relationship in time between an adverse CV outcome and new-onset AF was not possible to analyze because of lack of information when onset of AF occurred. However, our findings are supported by the fact that three previous randomized trials including >15,000 patients have shown that new-onset supraventricular tachycardia or AF are strong predictors of mortality in time-dependent analyses.

Conclusions

Atrial fibrillation is associated with an increased risk of adverse CV outcomes in patients with CHF and either reduced or preserved left ventricular systolic function. Patients with AF and reduced EF have the highest absolute risk, and those with AF and PEF have a greater increase in risk. New-onset AF is also associated with an increased risk with or without left ventricular systolic dysfunction. Candesartan improved clinical outcomes regardless of baseline rhythm.