Author + information
- Received April 29, 2014
- Revision received June 26, 2014
- Accepted July 23, 2014
- Published online November 4, 2014.
- Alan D. Enriquez, MD∗∗ (, )
- Brandon Calenda, MD†,
- Parul U. Gandhi, MD†,
- Ajith P. Nair, MD†,
- Anelechi C. Anyanwu, MD† and
- Sean P. Pinney, MD†
- ∗Cardiovascular Division, Brigham and Women’s Hospital, Boston, Massachusetts
- †Zena and Michael A. Wiener Cardiovascular Institute, Icahn School of Medicine at Mount Sinai, New York, New York
- ↵∗Reprint request and correspondence:
Dr. Alan D. Enriquez, Brigham and Women’s Hospital, Cardiovascular Division, 75 Francis Street, Boston, Massachusetts 02115.
Background Atrial fibrillation (AF) is common in patients with the HeartMate II (HMII) left ventricular assist device (LVAD), but the impact of AF on clinical outcomes is uncertain.
Objectives This study sought to determine the effect of AF on outcomes in patients with the HMII LVAD.
Methods Records of 106 patients who underwent HMII implantation at a single center were reviewed. The associations of paroxysmal atrial fibrillation (PAF) and persistent atrial fibrillation (PeAF) with survival, heart failure (HF) hospitalization, bleeding, and thromboembolism were examined using Kaplan-Meier survival analysis and Cox proportional hazards regression.
Results Mean age was 56.6 ± 11.4 years, 87.7% of the implants were intended as a bridge to transplantation, and median length of support was 217 days (range: 1 to 952 days). AF was present in 55 patients (51.9%); 36 patients (34.0%) had PAF and 19 (17.9%) had PeAF. Twenty-one patients (19.8%) died, and 18 (17.0%) were hospitalized for HF. There were 0.75 major bleeding events and 0.28 thromboembolic events per patient year of follow-up. PAF was not associated with increased mortality, HF hospitalization, bleeding, or thromboembolism. PeAF, however, was an independent predictor of the composite endpoint of death or HF hospitalization (hazard ratio: 3.54; 95% confidence interval: 1.52 to 8.25; p < 0.01). Although there was no increase in bleeding or thromboembolism, patients with AF had thromboembolic events at higher international normalized ratios (INRs).
Conclusions Although PAF is not associated with worse outcomes in patients with the HMII LVAD, PeAF may be associated with increased mortality and HF hospitalization. Patients with AF also may have thromboembolic events at higher INR levels.
Atrial fibrillation (AF) is common in patients with end-stage heart failure (HF) and is present in up to 50% of patients (1). AF prevalence in patients with continuous flow left ventricular assist devices (CF-LVADs) is similarly high (2). In conventional HF patients, AF has been associated with worse outcomes and increased mortality (1,3). In addition to an increased risk of thromboembolism, AF may lead to HF exacerbations because of a reduction in ventricular filling through the loss of atrial systole and rapid heart rates. However, the effect of AF on outcomes and mortality has not been well studied in LVAD patients.
AF may influence outcomes in patients with a CF-LVAD in several ways. Although LV filling is likely unaffected by AF due to support from the LVAD, right ventricular (RV) filling, and consequently cardiac output, may still be compromised by loss of atrial contraction, especially in patients with poor RV function and pulmonary hypertension. In addition, AF may confer an increased risk of thromboembolism in patients with a CF-LVAD (4), and some recommend targeting an international normalized ratio (INR) of 1.5 to 2.0 for patients without AF and 2.0 to 2.5 for those with AF (5). Accordingly, AF may affect outcomes through an increased risk of bleeding due to the higher level of anticoagulation therapy. Therefore, we sought to determine the effect of AF on mortality, HF hospitalization, bleeding, and thromboembolism in patients with CF-LVADs.
We reviewed the records of consecutive adult patients receiving the HeartMate II (HMII; Thoratec Corp., Pleasanton, California) LVAD at the Mount Sinai Medical Center in New York between June 2008 and April 2012. Patients were followed until 1 of the following endpoints was reached: death, transplantation, HMII explantation, end of follow-up period, or loss to follow-up. The Institutional Review Board of the Icahn School of Medicine at Mount Sinai approved this study.
Atrial fibrillation and anticoagulation
Retrospective chart reviews of electrocardiograms, device interrogations, and progress notes were performed to evaluate for AF occurrence. AF was defined as the presence of preoperative AF or the development of AF post-LVAD past the perioperative period (>30 days). AF was further subdivided into paroxysmal atrial fibrillation (PAF) and persistent atrial fibrillation (PeAF), using standard definitions (1). Because outcomes may differ between these groups, patients were analyzed in 3 groups: 1) those who did not have AF; 2) those who had PAF; and 3) those who had PeAF. The management of AF post-LVAD was left to the discretion of the HF specialist. In terms of anticoagulation, all patients received aspirin, 81 mg daily, and warfarin. For patients without AF, the INR goal was 1.5 to 2.0. For all patients with AF, the INR goal was 2.0 to 2.5. If a patient had multiple bleeding events, the INR goal was decreased to 1.5 to 2.0 in patients with AF.
Atrial fibrillation and outcomes
The focus of the study was the impact of PAF and PeAF on the following 3 outcomes: survival or HF hospitalization, thromboembolism, and bleeding. All deaths were confirmed through examination of the medical record, and the cause of death was noted. HF hospitalization was defined as hospitalization for signs of right HF (e.g., jugular venous distension, lower extremity edema) requiring escalation of diuretic therapy and/or initiation of inotrope therapy. Thromboembolism was defined as cerebrovascular accident (CVA), transient ischemic attack (TIA), arterial thromboembolism, or confirmed LVAD pump thrombosis. Major bleeding was defined using the Interagency Registry of Mechanically Assisted Circulatory Support (INTERMACS) definition (6). However, intracranial hemorrhage (ICH) also was included in major bleeding. The INR at the time of each bleeding and thromboembolic event was recorded. For thromboembolic events, a 4-week mean INR prior to the event also was calculated.
Atrial fibrillation and functional status
Patients deemed suitable by the HF specialist underwent a cardiopulmonary exercise test (CPET) ≥3 months after LVAD implantation. The CPET was performed on a treadmill, using a modified Naughton protocol. Oxygen consumption (Vo2) was continuously measured, and the test was symptom limited. Peak Vo2 levels of patients without AF were compared with those of patients with PAF and PeAF.
Categorical variables were evaluated using the chi-square or Fisher exact test. Continuous variables were analyzed with the t-test or Mann-Whitney U test. Normally distributed continuous variables were expressed as mean ± SD; non-normal variables were expressed as medians with interquartile ranges (IQR). For survival analysis, Kaplan-Meier time-to-event curves stratified by AF status were generated for death or HF hospitalization, thromboembolism, and bleeding. Statistical significance between the curves was analyzed using the log-rank test. The effect of AF and other variables on each outcome was analyzed using Cox proportional hazards regression. Because of the relatively small number of events for each outcome, multivariable regression was performed only for the composite outcome. Variables with a p value of <0.10 in univariable analysis were included in the multivariable model. All p values were 2-tailed, and the level of significance for all p values was <0.05. No corrections were used for multiple comparisons. Confidence intervals (CIs) were computed at the 95% confidence level. All statistics were computed using Stata version 12.0 (StataCorp, College Station, Texas).
During the study period, 106 patients received the HMII device, and 9 patients also received a temporary RV assist device at the time of LVAD implantation. Only 1 patient underwent an (unsuccessful) Cryo-Maze (Medtronic, Minneapolis, Minnesota) procedure during LVAD implantation, and 2 patients underwent left atrial appendage ligation. Overall, the majority of patients were male (82.1%), had a non-ischemic cardiomyopathy (57.5%), and received the HMII as a bridge to transplantation (87.7%). Mean age was 56.6 ± 11.4 years, and mean left ventricular ejection fraction (LVEF) was 18.4 ± 7.2%. The median support time was 217 days (range: 1 to 952 days), and there were 73.7 patient years of total support time. All patients were classified as New York Heart Association functional class III to IV and had INTERMACS profiles 1 to 4 at the time of HMII implantation. Twenty-one patients (19.8%) died, 60 survived to transplantation (56.6%), 3 (2.8%) had the HMII explanted because of myocardial recovery, 19 (17.9%) reached the end of follow-up, and 3 patients (2.8%) were followed at other centers after implantation.
AF was present in 55 patients (51.9%). Fifty patients had pre-LVAD AF, and 5 patients developed AF past the postoperative period (>30 days after implantation). Thirty-six patients (34.0%) had PAF, and 19 (17.9%) had PeAF. The baseline characteristics of patients with PAF, PeAF, and no AF are shown in Table 1. Compared to patients without AF, those with PAF and PeAF were more likely to be older, Caucasian, have a diagnosis of HF for >1 year, have an implantable cardioverter-defibrillator, and be receiving a beta-blocker and/or amiodarone prior to LVAD implantation. Patients with PAF were more likely to have had CVA, and those with PeAF were more likely to have chronic kidney disease. There were no statistically significant differences between the PAF and PeAF groups.
Atrial fibrillation, survival, and hospitalization for heart failure
Thirty-seven patients reached the composite endpoint of death or HF hospitalization: 13 (25.5%) in the no AF group, 11 (30.6%) in the PAF group, and 13 (68.4%) in the PeAF group. Patients with PeAF were at increased risk for death or HF hospitalization compared to the other groups (log-rank: p < 0.001) (Figure 1). In multivariable regression analysis with PAF, PeAF, destination therapy LVAD, and LVAD implant creatinine level as covariates, PAF was not associated with an increased risk of reaching the composite endpoint (hazard ratio [HR]: 0.91; 95% CI: 0.38 to 2.20; p = 0.83), but PeAF remained an independent predictor of the composite endpoint (HR: 3.54; 95% CI: 1.52 to 8.25; p = 0.003). Implant creatinine level was also a significant predictor of the composite endpoint (HR: 1.48; 95% CI: 1.21 to 1.81; p < 0.01).
The associations between PAF and PeAF and the individual components of the composite endpoint are shown in Table 2. Twenty-one patients died during the study period, with 9 (17.7%) in the no AF group, 5 (13.9%) in the PAF group, and 7 (36.8%) in the PeAF group. The causes of death in each group are shown in Table 3. In regression analysis, PAF was not associated with an increased risk of death (HR: 0.77; 95% CI: 0.25 to 2.40; p = 0.65). There was a strong trend toward increased mortality in the PeAF group, but this did not reach statistical significance (HR: 2.65; 95% CI: 0.96 to 7.35; p = 0.06). Having the LVAD implanted as destination therapy was the only significant predictor of death in regression analysis (HR: 3.69; 95% CI: 1.40 to 9.72; p < 0.01). Eighteen patients (17.0%) were hospitalized for HF, with 4 (7.8%) in the no AF group, 7 (19.4%) in the PAF group, and 7 (36.8%) in the PeAF group. There were a total of 28 hospitalizations, with 0.38 hospitalizations per patient year. During regression analysis, PAF was not associated with an increased risk of HF hospitalization (HR: 2.18; 95% CI: 0.64 to 7.47; p = 0.21), but PeAF was a significant predictor of HF hospitalization (HR: 7.37; 95% CI: 2.12 to 25.64; p = 0.002). LVAD implant creatinine level (HR: 1.69; 95% CI: 1.26 to 2.26; p < 0.001) and diabetes mellitus (HR: 3.36; 95% CI: 1.25 to 9.06; p = 0.02) also were predictors of HF hospitalization.
Atrial fibrillation and functional status
Thirty-five patients (33.0%) had a CPET performed at a median of 134 days (IQR: 109 to 231 days) post-LVAD implantation. The median peak Vo2 for this group was 13.6 (range: 4.0 to 21.4) ml/kg/min. Nineteen patients (54.3%) without AF, 9 patients (25.7%) with PAF, and 7 patients (20.0%) with PeAF underwent CPET. Only patients with PeAF were in AF at the time of the study. Median peak Vo2 in the PeAF group (11.2 [IQR: 8.7 to 14] ml/kg/min) was lower than in the no AF and PAF groups (13.8 [IQR: 12.7 to 16] ml/kg/min and 14.2 [IQR: 10.8 to 15] ml/kg/min, respectively) (Figure 2). However, there were no statistically significant differences in peak Vo2 between the groups (p = 0.13 for PeAF versus no AF; p = 0.46 for PeAF versus PAF; and p = 0.42 for PAF versus no AF) or in peak Vo2 between those in AF and those not in AF during the CPET (median peak Vo2: 11.2 versus 14.0 ml/kg/min; p = 0.17). There were also no significant differences between the median respiratory exchange ratios of the 3 groups (1.00 [IQR: 0.93 to 1.09] for no AF; 0.92 [IQR: 0.9 to 1.0] for PAF; and 0.94 [IQR: 0.9 to 1.03] for PeAF).
Atrial fibrillation, bleeding, and thromboembolism
Thirty-one patients (29.2%) experienced 55 bleeding events during follow-up, with 0.75 events per patient year. Sixteen patients (15.1%) had 21 thromboembolic events during follow-up, with 0.28 events per patient year. Characteristics of bleeding and thromboembolic events stratified by AF status are shown in Table 4. Thirteen patients (25.5%) without AF, 13 (36.1%) with PAF, and 5 (26.3%) with PeAF experienced bleeding events. The majority of episodes (70.9%) were gastrointestinal, with many requiring endoscopic intervention. ICH was the cause in 9.1% of the episodes, and 18.1% were from other causes, primarily significant epistaxis. There were no significant differences in the type of bleeding event or in the INR at the time of the bleeding event between patients with no AF, PAF, and PeAF (Table 4). There were also no significant differences in survival free from major bleeding among the 3 groups (Figure 3A) (log-rank: p = 0.70). In regression analysis, the development of new AF post-LVAD (HR: 3.63; 95% CI: 1.26 to 10.45; p = 0.02) and implant creatinine level (HR: 1.36; 95% CI: 1.08 to 1.61; p < 0.01) were significant predictors of bleeding. There was also a strong trend toward increased bleeding in patients with post-LVAD thromboembolism (HR: 2.23; 95% CI: 1.0 to 5.0; p = 0.05).
In terms of thromboembolism, 9 patients (17.7%) without AF, 6 (16.7%) with PAF, and 1 (5.3%) with PeAF had thromboembolic events. The majority of episodes were due to CVA (61.9%) followed by LVAD thrombosis (28.6%) (Table 4). One episode (4.8%) was due to a popliteal arterial embolus, and another (4.8%) was due to LV apical thrombus. There were no significant differences in the type of thromboembolism or in survival free from thromboembolism among the 3 groups (Figure 3B) (log-rank: p = 0.48). Because there was only 1 event in the PeAF group, INRs were compared between patients with PAF or PeAF and those without AF. Patients with any AF had a higher INR both at the time of the thromboembolic event (2.70 ± 0.94 vs. 1.54 ± 0.34; p = 0.003) and for the 4 weeks leading up to the event (2.33 ± 0.65 vs. 1.57 ± 0.31; p = 0.006) (Figure 4). In regression analysis, there was a trend toward increased risk for thromboembolism in patients with post-LVAD bleeding, but this did not quite reach statistical significance (HR: 2.52; 95% CI: 0.94 to 6.74; p = 0.06).
Our study has 3 primary findings. First, PAF was not associated with an increased risk of death, HF hospitalization, bleeding, or thromboembolism. Second, PeAF was an independent predictor of death or HF hospitalization. Third, despite a higher level of anticoagulation therapy, patients with any AF had nearly the same number of thromboembolic events as those without AF (Central Illustration). To our knowledge, this is the first study to show an increased risk of mortality or HF hospitalization with AF in this patient population and the first to support a higher target INR in patients with AF.
It has long been recognized that AF and chronic HF are interrelated. AF can worsen symptoms of HF by reducing ventricular filling through loss of atrial systole, decreasing diastolic filling time, and impairing systolic function. In turn, decompensated HF also can aggravate AF, leading to rapid ventricular response. With this interplay, it is not surprising that AF has been associated with increased mortality and HF hospitalization in patients with chronic HF (3,7). However, this finding has not been consistent (8,9). One possible reason for this discrepancy is that the type of AF was not differentiated in many studies. In the SOLVD (Studies of Left Ventricular Dysfunction) trial, AF was defined as AF present on the pre-randomization electrocardiogram (3). This definition likely selected for PeAF patients, and the study found that AF was associated with an increased risk of the composite endpoint of death or HF hospitalization.
Results from our study are similar to those from SOLVD. Although PAF was not associated with an increased risk of reaching the composite endpoint of death or HF hospitalization, PeAF was an independent predictor of the composite endpoint (HR: 3.54; p < 0.01). This finding was driven in large part by an increased risk of HF hospitalization in patients with PeAF (HR: 7.37; p < 0.01), but there was also a strong trend toward increased mortality in patients with PeAF (HR: 2.65; p = 0.06). These results suggest that the AF burden is of great importance in this patient population. An analysis using a quantitation of AF burden would have been ideal, but this was not possible because these data were unavailable for many patients.
Only 1 other study has examined the effect of AF on mortality in CF-LVAD patients. In a study of 389 patients with a CF-LVAD, Stulak et al. (4) reported that preoperative AF was not associated with increased mortality. However, AF was not subclassified into PAF and PeAF in the study, which potentially explains the discrepant result. No other studies have examined the association of AF and HF hospitalization in CF-LVAD patients.
The cause of the trend toward increased mortality in patients with PeAF is unclear. Patients with PeAF were more likely to be older, have chronic kidney disease, and have had a diagnosis of HF for more than 1 year. It is possible that PeAF is simply a marker for sicker patients with worse outcomes after LVAD implantation, as sepsis was the primary cause of death in the PeAF group (Table 3). The increased risk of HF hospitalization in patients with PeAF in our study suggests that the RV may remain sensitive to the hemodynamic effects of AF despite CF-LVAD support. Hemodynamic compromise from AF with improvement after catheter ablation in a patient with a CF-LVAD has been reported (10). The trend toward lower peak VO2 in PeAF patients also suggests a possible hemodynamic effect of AF. However, the number of patients undergoing CPET in our study was very small, and the effect of AF on functional status warrants further study.
Although patients with AF had a higher target INR level in our study, there were no differences in bleeding events. This result underlies the observation that acquired von Willebrand disease plays a major role in bleeding in patients with CF-LVADs, and a relatively small difference in target INR may be less contributory (11). It is also important to note that the rate of major bleeding was lower in our cohort than in other studies (12). In our study, the rate of thromboembolic events was 0.28 events per patient year, which is slightly higher than that in other studies (12,13). An important finding in our study is that patients with AF experienced thromboembolic events at significantly higher INR levels than those without AF. Despite this higher level of anticoagulation, the number of events between the 2 groups was nearly identical, suggesting an increased risk of thromboembolism with AF. This is consistent with the only other study examining thromboembolism and AF in patients with CF-LVADs, as Stulak et al. (4) reported an increased risk of thromboembolism with preoperative AF (HR: 1.89; p < 0.01). It is also interesting to note that in our study, thromboembolism was a predictor of bleeding and vice versa. This relationship has been shown before in LVAD patients and underscores the need for finding the optimal level of anticoagulation (14).
Our study may have implications for the management of patients with CF-LVADs. If PeAF is associated with increased mortality and worsening HF, patients may benefit from a more aggressive rhythm control strategy to reduce the burden of AF. Consideration should be given to performing surgical Cryo-Maze at the time of LVAD implantation, more liberal use of antiarrhythmic drugs, and even catheter ablation in select patients. Although our study was not designed to determine the optimal level of anticoagulation, our results support targeting a higher INR in patients with AF.
Our study has several important limitations. First, this study is retrospective and non-randomized and therefore subject to selection bias and confounding with the data limited by documentation in the chart. There were a relatively small number of patients and outcomes, and the findings should be considered only exploratory and hypothesis-generating. Second, the major positive finding in our study was found with subgroup analysis. Thus, the findings are less significant, although the p values for the composite endpoint were robust. Third, the generalizability of our results is somewhat limited. More than 85% of the patients in our study had the HMII implanted as a bridge to transplantation, and all of the patients had a HMII implanted with INR goals based on our institutional policy. Therefore, our bleeding and thromboembolism results may not be applicable with different target INR levels or in patients with the other commercially available CF-LVAD (HeartWare HVAD, Framingham, Massachusetts).
AF is common in patients with CF-LVADs. Although PAF is not associated with worse clinical outcomes, PeAF may be associated with increased mortality and hospitalization for HF. Although PAF and PeAF are not associated with a significantly increased risk of bleeding or thromboembolic events, patients with any AF may have thromboembolic events at higher INR levels. Prospective studies with monitoring of AF burden are needed to confirm our results.
COMPETENCY IN MEDICAL KNOWLEDGE: AF develops in up to 50% of patients with CF-LVADs. Although paroxysmal AF is not associated with appreciable worsening of clinical outcomes, persistent AF may be associated with more frequent hospitalization for decompensated heart failure and higher rates of mortality. Patients with AF may also have thromboembolic events at significantly higher INR levels.
TRANSLATIONAL OUTLOOK: Prospective studies are needed to evaluate the relative safety and efficacy of more intensive anticoagulation and/or more aggressive rhythm control strategies in patients with CF-LVADs who develop persistent AF.
Dr. Pinney is on the advisory board of and has received consulting and speaker’s fees from CareDX; is a paid consultant for Acorda Therapeutics; and has received clinical trial support and consulting and educational fees from Thoratec. All other authors have reported that they have no relationships relevant to the contents of this paper to disclose.
- Abbreviations and Acronyms
- atrial fibrillation
- continuous flow left ventricular assist device
- cardiopulmonary exercise test
- cerebrovascular accident
- heart failure
- HM II
- HeartMate II
- intracranial hemorrhage
- paroxysmal atrial fibrillation
- persistent atrial fibrillation
- right ventricular/ventricle
- transient ischemic attack
- Received April 29, 2014.
- Revision received June 26, 2014.
- Accepted July 23, 2014.
- American College of Cardiology Foundation
- Fuster V.,
- Ryden L.E.,
- Cannom D.S.,
- et al.
- Dries D.L.,
- Exner D.V.,
- Gersh B.J.,
- et al.
- Olsson L.G.,
- Swedberg K.,
- Ducharme A.,
- et al.
- Suarez J.,
- Patel C.B.,
- Felker G.M.,
- et al.
- Pagani F.D.,
- Miller L.W.,
- Russell S.D.,
- et al.