Erratum: Anticoagulation After Transcatheter Aortic Valve Replacement: Evolving Answers and Still Unaddressed Questions (JACC: Cardiovascular Interventions (2021) 14(15) (1714–1716), (S1936879821010463), (10.1016/j.jcin.2021.06.004))

Transcatheter aortic valve replacement (TAVR) is an established treatment for aortic valve stenosis (1). TAVR effectiveness and safety have considerably improved during the past decade, and current indications include most patients with severe aortic stenosis (1). Stroke and thromboembolic events are known to occur after initially successful TAVR, and the proportion of patients undergoing TAVR who have or subsequently develop indications for oral anticoagulation (OAC) is not negligible (1,2). In addition, in recent years, the relatively common observation of subclinical bioprosthesis thrombosis causing hypoattenuated leaflet thickening and reduced leaflet motion by computed tomography and transesophageal echocardiography has raised questions related to accelerated bioprosthesis degeneration (3). Treatment with OAC has been shown to reduce subclinical bioprosthesis thrombosis, but the net clinical benefit in the absence of another indication is uncertain (3). Recommendations on antithrombotic treatment after TAVR are based on a limited amount of high-quality data, and in patients who have indications for OAC, vitamin K antagonists (VKAs) have been preferentially recommended (4,5). In these patients, it has also been common practice to add single or dual antiplatelet therapy (DAPT) for 3 to 6 months after TAVR to reduce the risk for thromboembolic complications (4,5). The net benefit between thrombotic prevention and bleeding propensity associated with available antithrombotic strategies across the main subsets of patients undergoing TAVR still requires elucidation. In a previous issue of JACC: Cardiovascular Interventions, Didier et al (6) present the results of a nationwide observational study of patients with indications for OAC who underwent TAVR for severe aortic stenosis in France from 2010 to 2017. After 1:1 nearest neighbor propensity score matching according to OAC type among 8,960 patients who underwent TAVR, 1,093 patients treated with VKAs were compared with 1,378 patients treated with direct oral anticoagulant agents (DOACs) (6). At 3-year follow-up, patients receiving VKAs died more frequently (35.7% vs. 31.2%; HR: 1.37, 95% CI: 1.12-1.67) and experienced a higher incidence of major bleeding (12.3% vs. 8.4%; HR: 1.64, 95% CI: 1.17-2.29) compared with patients receiving DOACs. Ischemic stroke (4.8% vs. 3.3%; HR: 1.32, 95% CI: 0.81-2.15) and acute coronary syndrome (4.0% vs. 4.3%; HR: 1.17, 95% CI: 0.68-1.19) incidences were not significantly different between groups (6). The detrimental association between major bleeding occurrence and survival have been reported in other settings (7,8). Although in the study by Didier et al (6) a causal effect between major bleeding and mortality was not assessed, it might be reasonable to assume that the reduced mortality observed in patients treated with DOACs occurred in part through the mediation of reduced major bleeding. A plausible explanation for the significant reduction in major bleeding associated with DOACs therapy observed in the present study might rely on the highly predictable pharmacokinetics and pharmacodynamics of these drugs that do not require routine monitoring and present limited food and drug interactions (9). In contrast, VKAs have a less favorable profile and narrow therapeutic window that not infrequently result in unstable and supra- or subtherapeutic values of international normalized ratio, leading to increased risks of bleeding or thromboembolic events, respectively (9). Elderly and frail patients with multiple significant comorbid conditions and concomitant medications including OAC undergoing TAVR may take advantage of DOACs safer therapeutic profile (10). However, previous observational investigations comparing outcomes in patients undergoing TAVR treated with VKAs or DOACs have provided mixed results, with some studies supporting a benefit from DOACs over VKAs in terms of both mortality and major bleeding, others showing nonsignificant differences between treatments with respect to the same outcomes, and others even revealing higher rates of ischemic outcomes associated with DOACs as compared with VKAs (11-13). Of note, superiority of DOACs over VKAs after TAVR was not observed in the recent ATLANTIS (Antithrombotic Strategy After Trans-Aortic Valve Implantation for Aortic Stenosis) trial (14). In the cohort of patients requiring OAC included in that trial, there was no significant difference between treatments with respect to the primary net composite endpoint (HR: 1.02; 95% CI: 0.68-1.51). Similarly, no significant differences between treatments across secondary composite and individual endpoints were detected (14). Regardless of the presence of confounding effects, the most important result of the study by Didier et al (6) is the absence of any signal of harm associated with DOACs therapy after TAVR in patients with indications for OAC. Indeed, the recent results of the GALILEO (Global Study Comparing a Rivaroxaban-Based Antithrombotic Strategy to an Antiplatelet-Based Strategy After Transcatheter Aortic Valve Replacement to Optimize Clinical Outcomes) trial, including patients with low to intermediate surgical risk without established indications for OAC, have generated concerns related to DOACs use in patients undergoing TAVR (15). In this trial, treatment with DOACs was associated with increased incidences of all-cause death (HR: 1.69; 95% CI: 1.13-2.53) and Valve Academic Research Consortium major bleeding (HR: 2.02; 95% CI: 1.09-3.76) compared with antiplatelet therapy (15). However, the results of the GALILEO trial cannot be translated to the setting of patients with indications for OAC, as outlined by the findings of Didier et al (6,15). The results of the study by Didier et al (6) need to be interpreted in light of the following considerations. First, as with other statistical techniques used to account for confounding effects and selection bias in observational studies, propensity score matching relies on the quality and extent of available information and unmeasured confounders can continue to influence outcomes despite the achievement of an acceptable balance between baseline covariates. However, although the favorable findings of the study by Didier et al (6) warrant confirmation in focused randomized clinical trials, matching patients according to the probability of treatment assignment conditional on baseline covariates (i.e., propensity score matching) can be a powerful method to balance treatment groups in observational studies. Moreover, in the present study outcomes between unmatched and matched cohorts were quite consistent. Second, in the study by Didier et al (6) working sample size was significantly reduced by propensity score matching likely as a result of the substantial heterogeneity between treatment groups in the original study cohort (6). In addition, results beyond 2-year follow-up are based on less than one-third of the original sample size due to limited follow-up completeness (6). These conditions have reduced statistical power and may underlie selection bias. Third, restricted numbers of variables and outcomes and absence of independent adjudication of events are known limitations of large-scale observational studies with administrative registry-linked follow-up. With respect to some outcomes, such as major bleeding and stroke, it is possible that heterogeneity in reporting, inconsistency of definitions, and variability of assessment methods across centers played a role. Finally, in the study no account was made for the influence of concomitant antithrombotic therapy duration and composition (i.e., single aspirin- or clopidogrel-based antiplatelet therapy or DAPT) between treatment groups. The results of the present study should be revisited in light of the POPular-TAVI (Antiplatelet Therapy for Patients Undergoing Transcatheter Aortic Valve Implantation) trial, which recently showed that supplementation of OAC with clopidogrel for 3 months after TAVR led to a significant excess of non-procedure-related bleeding (34.0% vs 21.7%; P = 0.02), without significant differences in cardiovascular death, ischemic stroke, or myocardial infarction and major individual endpoints (16). These results suggest that OAC alone may be the standard antithrombotic strategy after TAVR in patients with indications for OAC and, when percutaneous coronary intervention is required before or after TAVR, recent evidence supports minimalist antithrombotic strategies (17). In conclusion, although heterogeneity in accumulated evidence should continue to promote high-quality research, answers related to OAC alone or associated with single antiplatelet therapy or DAPT after TAVR are growing. According to available data, OAC alone seems to be the safest therapeutic approach in patients with clinical conditions requiring thromboembolic prevention undergoing TAVR, especially in relation to the frequently relevant inherent bleeding risk. However, unaddressed questions related to the comparison between VKAs and DOACs require further analysis. In this regard, the results of the ongoing ENVISAGE-TAVI AF (Edoxaban Compared to Standard Care After Heart Valve Replacement Using a Catheter in Patients With Atrial Fibrillation; NCT02943785) trial may provide important answers. According to available data, including the study by Didier et al (6), there is no strong evidence of unsafety of DOACs use in patients treated with TAVR with indications for OAC.