Abstract

Key Points

• Patients with paroxysmal and non-paroxysmal AF accounted for 21% and 16%, respectively, of patients undergoing TAVR in this cohort.

• Nonparoxysmal AF was associated with a higher risk of periprocedural and all-cause mortality compared with not having AF. In contrast, paroxysmal AF showed no such association.

• Risk stratification has traditionally been based on the presence or absence of AF; however, this study suggests the importance of stratification by AF type when assessing risk and prognosis for TAVR.

Abbreviations

AF

atrial fibrillation

AS

aortic stenosis

HF

heart failure

IPW

inverse probability weighting

STS

Society of Thoracic Surgeons

TAVR

transcatheter aortic valve replacement

Introduction

Atrial fibrillation (AF) is 1 of the most common cardiac arrhythmias, particularly in older patients and patients with heart failure (HF), and it is associated with an increased risk of stroke, HF, and mortality.^1,2 Transcatheter aortic valve replacement (TAVR) is a well-established treatment option for patients with symptomatic severe aortic stenosis (AS) that is considered surgically inoperable or who are considered intermediate to high risk.^3-5 Patients with AS who are candidates for TAVR tend to be older and also have HF, and AF is a common condition in these cohorts. According to the Society of Thoracic Surgeons (STS)/American College of Cardiology Transcatheter Valve Therapies Registry, approximately 40% of patients undergoing TAVR for AS had a history of AF.⁶ Atrial fibrillation is of concern in patients with AS because AS causes increased left ventricular end-diastolic pressure, resulting in poor left ventricular filling, which is exacerbated by AF because of the lack of atrial contraction. In fact, AF is a well-established risk factor for mortality, even after successful TAVR.^7-9 Little is known, however, about whether clinical outcomes differ between patients with different types of AF.^10,11 The sharing of important cardiovascular risk factors between AS and AF further complicates any direct comparison of their effects. To address these complexities, the current study aims to evaluate both early and late outcomes after TAVR by AF type using an inverse probability weighting (IPW) method to adjust for confounding.

Patients and Methods

This retrospective observational study reviewed the data of patients who underwent TAVR for severe AS at a participating institution. From January 2018 to September 2022, a total of 1,099 TAVR procedures was performed. Of the patients who underwent surgery, 982 qualified for inclusion in the current study. Patients who underwent valve-in-valve TAVR for a failed bioprosthesis (n = 114) were excluded, as were patients in whose cases the procedure was aborted because of the unsuccessful delivery of a transcatheter heart valve or in which TAVR was converted to surgical AVR (n = 3). The observational period for this study extended through September 30, 2023.

Patients were classified into 3 groups based on their diagnosed type of AF: (1) no AF, (2) paroxysmal AF, and (3) nonparoxysmal AF (persistent or permanent). The primary end point was all-cause mortality, and other outcomes of interest included periprocedural outcomes, all incidents of stroke, a composite of all-cause mortality or stroke, rehospitalization for HF, and overt bleeding. Definitions, terminology, and reported outcomes were consistent with the STS/American College of Cardiology Transcatheter Valve Therapies Registry and the Valve Academic Research Consortium 3 criteria.¹² The decision to perform a TAVR procedure was made by a dedicated heart team, primarily based on the patient's age and surgical risk according to the STS Predicted Risk of Mortality as well as on patient anatomy and patient-specific factors such as frailty. The choice of balloon-expanding or self-expanding valve for TAVR was generally not based on the presence or absence of AF but on several other factors, such as the presence of coronary artery disease and the anatomy of the aortic root complex, including annular size and valve/root calcification. For example, self-expanding valves are preferred for patients with small aortic annuli, while balloon-expandable valves are preferred in patients with coronary artery disease because they allow better access to the coronary ostium after TAVR. The general strategy for antithrombotic therapy after TAVR at the corresponding institution is as follows: double antiplatelet therapy for 4 weeks and single antiplatelet therapy thereafter in patients without AF; oral anticoagulation therapy and single antiplatelet therapy in patients with AF. This strategy, however, is individualized according to factors such as the risk of thrombosis or bleeding, the presence of coronary artery disease, the patient's history of deep vein thrombosis, left atrial appendage closure, and recent percutaneous coronary intervention. For follow-up, beginning from post-TAVR day 0 (time = 0), patients were censored at their last recorded contact, whether at the corresponding hospital, their primary clinic, or by telephone. Follow-up event information was collected accordingly, with the observation period for each patient concluding upon their reaching the primary end point. The study protocol was approved by the Main Line Hospitals Institutional Review Board (45CFR164.512). Individual patient consent was waived because of the retrospective nature of the study.

Statistical Analysis

Continuous values are presented as mean (SD) unless otherwise noted. The distribution of these variables was assessed using the Shapiro-Wilk test. For variables with a normal distribution, the t test was used for 2-group comparisons, and analysis of variance was used for 3-group comparisons. For non-normal distributions, the Wilcoxon rank sum test was used for 2-group comparisons, and the Kruskal-Wallis test was used for analyses involving 3 groups. Categorical values are reported as numbers (percentages) or percentages, and the χ² test or Fisher exact test was used to compare groups, as appropriate. When statistically significant differences were observed among the 3 groups, post hoc tests with the Bonferroni correction were used to determine specific group differences. Inverse probability weighting was employed to estimate the average treatment effect while controlling for confounding variables. Propensity scores were calculated using a generalized boosted model with 10,000 trees, a shrinkage parameter of 0.01, and a minimum of 10 observations per node. The model included the covariates listed in Table I. Weights were derived from the propensity scores using the twang package in R (R Foundation for Statistical Computing). These weights were then applied to the data to balance the distribution of the covariates between groups. The goodness of fit of the variables was assessed using the absolute standardized mean difference, with values less than 0.1 indicating an ideal fit and values of 0.1 to 0.2 indicating an acceptable fit. Logistic regression analysis was used to calculate odds ratios and 95% CIs for periprocedural outcomes by AF type. New permanent pacemaker implantation was not included in the logistic regression analysis after IPW because of the challenges associated with excluding patients with prior pacemaker or defibrillator implantation from the cohort after IPW. Kaplan-Meier curves were constructed to estimate event-free rates for the outcomes of interest, and Cox proportional hazards models were used to assess the effect of AF type on all-cause mortality, stroke, and a composite of all-cause mortality or stroke. In addition, forward-selection multivariate models were used to assess predictors of all-cause mortality and the composite outcome. Predictors of periprocedural outcomes and stroke were not assessed by multivariate analysis because of the expected small number of events. In addition to the following prespecified variables—AF type, age, sex, New York Heart Association classification, and STS Predicted Risk of Mortality scores—variables with a statistically significant association (P < .05) with each outcome in the univariate analysis were included in the multivariate analysis. For patients with missing data for certain variables (albumin, n = 7; brain-type natriuretic peptide, n = 23; aortic valve area, n = 2), the median value or the most frequent value was imputed for multivariate analyses and propensity-score estimation. All P values were 2 sided, and a 5% level was considered statistically significant. All analyses were conducted using R, version 4.2.3, software.

TABLE IBaseline and Procedural Characteristics Before and After IPW

View Fullscreen

Results

Periprocedural Outcomes

The overall periprocedural mortality rate was 2.0%, and periprocedural outcomes are summarized in Supplemental Table II. Results of logistic regression analyses for the periprocedural outcomes by AF type are shown in Table II. After IPW, nonparoxysmal AF was significantly associated with a high periprocedural mortality rate compared with not having AF, with an odds ratio of 4.71 (95% CI, 1.24-17.9), whereas paroxysmal AF did not show a statistically significant difference. Other major complications were comparable among groups.

TABLE IIPeriprocedural Outcomes Before and After IPW

View Fullscreen

A lower postoperative mean pressure gradient was observed in patients with nonparoxysmal AF compared with patients without AF.

Late Outcomes

The median (IQR) follow-up period was 22 (0-69) months, and the Kaplan-Meier estimated 1-year and 5-year overall mortality rates were 12% and 50%, respectively. In addition, the incidence of stroke in the entire cohort was 2.2% at 1 year and 6.5% at 5 years. Antithrombotic therapy at discharge in each group is shown in Supplemental Table III. In summary, 61% of patients with paroxysmal AF and 86% of patients with nonparoxysmal AF were receiving oral anticoagulants at discharge. During follow-up after TAVR, 8 patients with paroxysmal AF and 5 patients with nonparoxysmal AF underwent left atrial appendage closure. Figure 1 shows Kaplan-Meier curves for the late outcomes of interest comparing the 3 groups, and Table III and Supplemental Table IV show the results of Cox proportional hazards regression analyses for the late outcomes. Nonparoxysmal AF was associated with an increased risk of all-cause mortality and a composite of all-cause mortality or stroke compared with not having AF, with hazard ratios of 1.56 (95% CI, 1.14-2.14) and 1.54 (95% CI, 1.14-2.09), respectively, after IPW. Although a crude analysis showed an increased risk of all-cause mortality and the composite outcome in the patients with paroxysmal AF compared with patients without AF, the differences after IPW were not statistically significant. Consistent findings were observed in the multivariate analyses; nonparoxysmal AF was an independent predictor of all-cause mortality and the composite outcome compared with not having AF, but paroxysmal AF was not. Atrial fibrillation type was not significantly associated with stroke. Both paroxysmal and nonparoxysmal AF, however, were associated with a higher risk of rehospitalization for HF compared with not having AF both before and after IPW and in multivariate analysis. The incidence of overt bleeding was comparable between patients without AF and patients with paroxysmal AF. In contrast, the incidence of overt bleeding was significantly higher in patients with nonparoxysmal AF compared with patients without AF both before and after IPW. Multivariate analysis, however, showed no statistically significant difference in overt bleeding between the 2 groups (Supplemental Table IV).

View Figure

• Download as Powerpoint
• Download Figure

TABLE IIIRisk Analysis for All-Cause Mortality, Composite of All-Cause Mortality and Stroke, and Stroke

View Fullscreen

Discussion

This study found that nonparoxysmal AF was significantly associated with higher rates of periprocedural mortality, all-cause mortality, a composite of all-cause mortality or stroke, and rehospitalization for HF during follow-up compared with not having AF. Paroxysmal AF was not associated with increased rates of all-cause mortality, stroke, or overt bleeding, but the rate of rehospitalization for HF was significantly higher in patients with paroxysmal AF than in patients without AF. The prevalence of a history of AF in patients undergoing TAVR (37%), patients with periprocedural mortality (2.0%), and with 1-year and 5-year mortality rates (12% and 50%) were consistent with prevalence rates found in previous studies.^6,13 The current study is unique in that it compares TAVR outcomes by AF type while adjusting for confounders using the IPW method.

Atrial fibrillation is a well-known risk factor for allcause mortality after TAVR for severe AS,^7-9 but studies of the effect of the AF type have been limited. The higher rate of late mortality after TAVR in patients with nonparoxysmal AF compared with patients without AF has been reported by Shaul et al¹⁰ and Jaakkola et al.¹¹ The results of the current study are consistent with the results of these studies, even after adjustment of confounders using both IPW and multivariate analysis. The discrepancy in the results between patients with paroxysmal and nonparoxysmal AF compared with patients without AF was reasonable; a previous metaanalysis showed that nonparoxysmal AF was associated with a significantly higher risk of mortality than paroxysmal AF in the general population.¹⁴ The authors of the meta-analysis speculated that worsening HF caused by long-standing AF, more severe stroke events, or a higher burden of noncardiovascular disease were potential causes of the poorer outcomes. In patients with AS, AF presents a particular challenge because it leads to increased left ventricular end-diastolic pressure. This increase in pressure can compromise left ventricular filling, a situation further exacerbated by AF because of the lack of atrial contraction. The current study even showed a significantly higher periprocedural mortality despite a comparable incidence of other major complications, including acute stroke. The cardiac damage and degeneration caused by long-standing AF may contribute to these results. Further research with a larger cohort is needed to validate the results of the current study and elucidate the mechanism of the potential periprocedural risks of nonparoxysmal AF in TAVR.

Atrial fibrillation is an important risk factor for thromboembolism and stroke, but no association between any type of AF and the incidence of stroke was observed in this study. These findings are in line with a previous meta-analysis¹⁵ that showed no difference between patients who did not have AF and patients who did have AF after TAVR. Although not fully evaluated in the current study, optimal anticoagulation therapy could have played an important role in the lack of association of any subtype of AF with early or late stroke. Indeed, the observed low incidence of stroke over the 5-year follow-up period is consistent with findings from previous studies¹⁶ and suggests that the patients likely benefited from optimal medical management. The current study conversely suggests that the incidence of overt bleeding after TAVR may be higher in patients with nonparoxysmal AF than in patients without AF. In addition to individualized optimal management for antithrombotic medications, the potential benefits of left atrial appendage closure and discontinuation of oral anticoagulants in patients with AF at high risk of bleeding who undergo TAVR warrant further investigation. The observed association between nonparoxysmal AF and increased mortality after TAVR underscores the need for further research to explore whether early screening and prevention of progression to nonparoxysmal AF in patients with early-stage AS could improve the prognosis for patients with AS and patients with AF.

Risk stratification has traditionally been based on the presence or absence of AF, but the current study shows that patients with nonparoxysmal AF have a poorer prognosis, while patients with paroxysmal AF have an outcome comparable to that of patients without AF in terms of survival and risk of stroke or bleeding. These findings underscore the critical need for stratification by AF type when assessing risk and prognosis for TAVR. This approach could lead to more personalized care and better health outcomes for patients with different types of AF.

Conclusion

In patients undergoing TAVR for AS, nonparoxysmal AF was associated with worse clinical outcomes in terms of higher risk of periprocedural mortality, all-cause mortality, and rehospitalization for HF compared with not having AF. The risk of overt bleeding may also be higher. Patients with paroxysmal AF had comparable rates of mortality, stroke, and overt bleeding compared with patients without AF, though the rate of rehospitalization for HF was higher.