Safety and Efficacy of Antithrombotic Strategies in Patients With Atrial Fibrillation Undergoing Percutaneous Coronary Intervention: A Network Meta-analysis of Randomized Controlled Trials

Key Points

Question  What is the most appropriate antithrombotic regimen to manage atrial fibrillation and coronary artery disease, in particular with acute coronary syndrome and/or percutaneous coronary intervention, while balancing ischemic and bleeding risk in an understudied high-risk patient population?

Findings  In this network meta-analysis, simultaneous comparisons of multiple antithrombotic strategies were performed for safety and efficacy outcomes in a study involving more than 10 000 participants. The study demonstrated that vitamin K antagonist plus dual antiplatelet therapy should be avoided, whereas the use of a non–vitamin K antagonist oral anticoagulant plus P2Y₁₂ inhibitor, without aspirin, should be the preferred treatment.

Meaning  Meaningful information about antithrombotic regimens may help physicians in their decision making when treating this high-risk group of patients.

Importance  The antithrombotic treatment of patients with atrial fibrillation (AF) and coronary artery disease, in particular with acute coronary syndrome (ACS) and/or percutaneous coronary intervention (PCI), poses a significant treatment dilemma in clinical practice.

Objective  To study the safety and efficacy of different antithrombotic regimens using a network meta-analysis of randomized controlled trials in this population.

Data Sources  PubMed, EMBASE, EBSCO, and Cochrane databases were searched to identify randomized controlled trials comparing antithrombotic regimens.

Study Selection  Four randomized studies were included (n = 10 026; WOEST, PIONEER AF-PCI, RE-DUAL PCI, and AUGUSTUS).

Data Extraction and Synthesis  The Preferred Reporting Items for Systematic Reviews and Meta-Analyses guidelines were used in this systematic review and network meta-analysis between 4 regimens using a Bayesian random-effects model. A pre hoc statistical analysis plan was written, and the review protocol was registered at PROSPERO. Data were analyzed between November 2018 and February 2019.

Main Outcomes and Measures  The primary safety outcome was Thrombolysis in Myocardial Infarction (TIMI) major bleeding; secondary safety outcomes were combined TIMI major and minor bleeding, trial-defined primary bleeding events, intracranial hemorrhage, and hospitalization. The primary efficacy outcome was trial-defined major adverse cardiovascular events (MACE); secondary efficacy outcomes were individual components of MACE.

Results  The overall prevalence of ACS varied from 28% to 61%. The mean age ranged from 70 to 72 years; 20% to 29% of the trial population were women; and most patients were at high risk for thromboembolic and bleeding events. Compared with a regimen of vitamin K antagonist (VKA) plus dual antiplatelet therapy (DAPT; P2Y₁₂ inhibitor plus aspirin), the odds ratios (ORs) for TIMI major bleeding were 0.58 (95% CI, 0.31-1.08) for VKA plus P2Y₁₂ inhibitor, 0.49 (95% CI, 0.30-0.82) for non-VKA oral anticoagulant (NOAC) plus P2Y₁₂ inhibitor, and 0.70 (95% CI, 0.38-1.23) for NOAC plus DAPT. Compared with VKA plus DAPT, the ORs for MACE were 0.96 (95% CI, 0.60-1.46) for VKA plus P2Y₁₂ inhibitor, 1.02 (95% CI, 0.71-1.47) for NOAC plus P2Y₁₂ inhibitor, and 0.94 (95% CI, 0.60-1.45) for NOAC plus DAPT.

Conclusions and Relevance  A regimen of NOACs plus P2Y₁₂ inhibitor was associated with less bleeding compared with VKAs plus DAPT. Strategies omitting aspirin caused less bleeding, including intracranial bleeding, without significant difference in MACE, compared with strategies including aspirin. Our results support the use of NOAC plus P2Y₁₂ inhibitor as the preferred regimen post–percutaneous coronary intervention for these high-risk patients with AF. A regimen of VKA plus DAPT should generally be avoided.

The combination of atrial fibrillation (AF) and coronary artery disease (CAD), and in particular acute coronary syndrome (ACS) or percutaneous coronary intervention (PCI), poses a significant dilemma to physicians who have to decide which antithrombotic regimen provides the best balance of safety and efficacy.^1,2 Vitamin K antagonists (VKAs) and non–vitamin K antagonist oral anticoagulants (NOACs) are indicated for stroke prevention, whereas prevention from recurrent atherothrombotic events and stent thrombosis usually requires dual antiplatelet therapy (DAPT) with a P2Y₁₂ inhibitor and aspirin.³

A number of randomized controlled trials (RCTs) have compared the safety and efficacy of antithrombotic regimens in patients with AF undergoing PCI or with concomitant ACS.^4-7 In 2018, the outcomes of these RCTs were combined in a meta-analysis⁸ that concluded that omitting aspirin resulted in fewer major bleeds while preserving efficacy compared with VKA or NOACs with DAPT. While the findings of this analysis are important, the sample size of approximately 5000 patients did not permit assessment of differences in less frequent but important complications such as stent thrombosis. Moreover, it also remains unclear whether different classes of anticoagulants (VKAs or NOACs) affect safety and efficacy in this high-risk population. To address these shortcomings, we performed a Bayesian network meta-analysis (NMA) that allowed for simultaneous comparisons of multiple antithrombotic strategies. This approach, with the inclusion of the 4614-patient 2019 Open-label, 2 × 2 Factorial, Randomized Controlled, Clinical Trial to Evaluate the Safety of Apixaban vs Vitamin K Antagonist and Aspirin vs Aspirin Placebo in Patients With Atrial Fibrillation and Acute Coronary Syndrome or Percutaneous Coronary Intervention (AUGUSTUS) trial,⁹ presents the largest evidence base to date to inform antithrombotic decisions in this high-risk group of patients.

The Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines (eTable 1 in the Supplement) were used in this systematic review and NMA.¹⁰ A prespecified statistical analysis plan was written, and the review protocol was registered at PROSPERO (CRD42019096584).

Two reviewers (R.D.L. and R.E.H.) performed a systematic review, and disagreements were resolved in a panel discussion of 3 reviewers (R.D.L., H.H., and R.E.H.). Study selection involved screening of titles and abstracts followed by full-text evaluation of possible eligible studies. The inclusion criteria for our study included (1) RCTs with at least 2 comparator arms; (2) study population of patients with ACS or undergoing PCI for stable CAD or ACS; (3) treatment with a combination of anticoagulation and antiplatelet therapy; (4) reported major bleeding and major adverse cardiovascular events (MACE); and (5) follow-up of at least 6 months. We excluded observational studies, crossover trials, studies where the method of allocation was not truly random, duplicate studies, and nonoriginal data studies. No language, publication date, or publication status restrictions were applied. References of prior systematic reviews and meta-analyses were also screened for related studies.

We searched PubMed/MEDLINE, Ovid/Embase, EBSCO/CINAHL, and Cochrane databases from database inception through the final search date of April 19, 2019. We used keywords related to PCI, ACS, NOACs or VKA, antiplatelets, and atrial fibrillation. The full search strategies are provided in eTable 2 in the Supplement. Study authors were contacted to identify other relevant studies or to provide unpublished data.

The primary safety outcome was major bleeding according to the Thrombolysis In Myocardial Infarction (TIMI) criteria. The secondary safety outcomes were combined TIMI major and minor bleeding, trial-defined primary bleeding events (eTable 3 in the Supplement for trial-specific definitions), intracranial hemorrhage, and hospitalization. The primary efficacy outcome was trial-defined MACE (eTable 3 in the Supplement), which was usually defined as a combination of either all-cause or cardiovascular mortality, myocardial infarction (MI), stroke, and stent thrombosis. Secondary efficacy outcomes were individual components of this composite MACE outcome.

Two reviewers (H.H. and J.L.) independently extracted data on the study design, baseline characteristics, interventions, and outcomes. Any disagreements of collected information between the 2 reviewers were reconciled through discussion. Unpublished outcome data were provided by the principal investigators of eligible RCTs. For the AUGUSTUS trial, the lead investigator (R.D.L.) provided the necessary information shortly after database lock in December 2018.

Two independent reviewers (R.D.L. and R.E.H.) assessed the risk of bias (low, intermediate, or high) of the included studies using the Cochrane Collaboration tool (eTable 4 in the Supplement). Given the limited number of publications, we did not assess the risk of publication bias.

We fitted a Bayesian random-effects NMA model to simultaneously compare multiple regimens. We extracted the sample size and total number of events for each of the prespecified outcomes in each treatment group from eligible RCTs. Data were analyzed between November 2018 and February 2019. The NMA model combines evidence about direct and indirect comparisons of regimens by accounting for the correlation among multiarm trials.^11,12 We assumed that the direct and indirect evidence for a treatment comparison had no discrepancy, called evidence consistency. To account for effect heterogeneity across trials, we allowed random effects to NMA and measured the magnitude of heterogeneity. In this model, we estimated odds ratios (ORs) of the effects of the 2 regimens and the associated 95% credible intervals using Markov chain Monte Carlo algorithms. All analyses were conducted using the gemtc package (version 0.8-2)¹³ in R, version 3.5.1 (The R Foundation).¹⁴ We used the package’s default setting including noninformative prior distributions with 4 parallel chains, where each chain consists of 50 000 samples after a 20 000-sample burn-in. The gemtc package uses a normal prior for log odds of baseline or log OR parameters and a uniform prior for the standard deviation of between-study heterogeneity and then determines the specifications of the prior distributions (eg, upper bound of the uniform prior) from the data so that the prior distributions are sufficiently vague but limit bias.¹⁵ We checked convergence of Markov chain Monte Carlo chains for all model parameters using trace plots and Gelman-Rubin diagnostic statistics.¹⁶

To evaluate and rank regimens, we calculated rank probabilities (ie, probability of a regimen being the best, second best, or worst for an outcome) and the Surface Under the Cumulative Ranking (SUCRA). The SUCRA is a numerical summary that accounts for both magnitude and uncertainty of the estimated effect for each regimen.¹⁷ A larger SUCRA value indicates better performance for the outcome. We ranked regimens based on SUCRA with respect to each safety and efficacy outcome.

We assessed statistical evidence for inconsistency defined as discrepancy between direct and indirect comparisons of treatment effects using 3 methods including loop-specific approach, node-splitting approach, and comparing Bayesian model fit with and without the assumption of evidence inconsistency.^18,19 In addition, we investigated the effect of different prior distributions for the heterogeneity parameter.²⁰ We conducted sensitivity analyses with 4 different networks of regimens. In addition, we added the landmark study of Triple Therapy in Patients on Oral Anticoagulation After Drug Eluting Stent Implantation (ISAR-TRIPLE)⁷ into the NMA.

All outcome data in each regimen used in the analyses (eAppendix 1 in the Supplement) and network structures for sensitivity analyses (eAppendix 3 in the Supplement) are available online.

Our database search resulted in 1448 unique studies. Of those, 1424 studies were deemed irrelevant based on title and abstract screening. Twenty-four studies were assessed in full-text for eligibility (eFigure 1 in the Supplement). Of those, 5 studies met the inclusion criteria.^4-7,9 However, 1 of these studies (ISAR-TRIPLE⁷) had a markedly different trial design, involving a randomized comparison of 2 durations of triple therapy with omitting clopidogrel after 6 weeks vs 6 months. While a landmark analysis was performed at 6 weeks, the study was not designed to compare dual-therapy vs triple-therapy regimens from time of index hospitalization (an important time window both from thromboembolic and bleeding perspective) and was included therefore only in a sensitivity analysis. As a result, the main analysis included 4 RCTs.

Trial design, population, treatment regimens, main results, and other characteristics of the 4 included RCTs are available online (eTables 5-7 in the Supplement). The RCTs were predominantly conducted at sites in Europe and North America and were published between 2013 and 2019. Follow-up duration was 6 to 14 months. All RCTs were judged to be at low risk of bias (eTable 4 in the Supplement). eTable 8 in the Supplement shows baseline characteristics of the patients in each RCT. A total of 10 026 patients were included in the NMA, and the sample size of the individual RCTs ranged from 563 to 4614. The overall prevalence of ACS varied from 28% to 61%. Except for AUGUSTUS, all patients underwent PCI, and the use of drug-eluting stents ranged from 66% to 83%. The mean age of patients ranged from 70 to 72 years, and 20% to 29% of the trial populations were women. Patients were predominantly white (>90%). Comorbidities, such as hypertension, dyslipidemia, and diabetes, were common. Most patients had annual thromboembolic and bleeding risk exceeding 3% (CHA₂DS₂-VASc ≥3 and HAS-BLED ≥3).

Figure 1 displays the network of treatment regimens used in the main analysis. We compared 4 treatment regimens: VKA plus DAPT; VKA plus P2Y₁₂ inhibitor; NOAC plus DAPT; and NOAC plus P2Y₁₂ inhibitor. In this network, we assumed that the 3 NOAC agents (apixaban, dabigatran, and rivaroxaban) and doses (apixaban, 5 mg, twice daily; dabigatran, 110 mg or 150 mg, twice daily; rivaroxaban, 2.5 mg, twice daily or 15 mg, once daily) included in this analysis have comparable safety and efficacy. We set VKA plus DAPT to be reference because all 4 RCTs investigated this regimen.

Compared with VKA plus DAPT, both the NOAC plus P2Y₁₂ inhibitor and VKA plus P2Y₁₂ inhibitor regimens were associated with less bleeding including less TIMI major, major or minor, and trial-defined primary bleeding (Figure 2). For intracranial hemorrhage, the use of NOAC plus P2Y₁₂ inhibitor had the lowest odds among the 4 regimens. Pairwise comparisons among regimens are displayed in eFigure 2 in the Supplement. The odds for any reported bleeding outcome were lower for NOAC plus P2Y₁₂ inhibitor when compared with VKA plus DAPT including intracranial hemorrhage. The VKA plus P2Y₁₂ inhibitor regimen was also more favorable for trial-defined bleeding outcomes and any TIMI bleed compared with VKA plus DAPT. When comparing these 2 regimens, NOAC plus P2Y₁₂ inhibitor had more favorable odds for intracranial hemorrhage. than VKA P2Y₁₂ inhibitor. In addition, the combination of NOAC plus DAPT did result in numerically lower rates of all reported bleeding outcomes when compared with VKA plus DAPT.

Overall, the 4 treatment regimens had comparable outcomes with respect to MACE and all-cause and cardiovascular death (Figure 3; eFigure 3 in the Supplement). Moreover, there were no significant differences for stroke, MI, and stent thrombosis between the antithrombotic regimens. Finally, the need for hospitalization was the same irrespective of the antithrombotic regimen used (Figure 3G; eFigure 3G in the Supplement).

The Table displays SUCRA values for safety and efficacy outcomes. The performance of the tested regimens was visualized in a 2-dimensional forest plot of ORs (Figure 4; eTables 9 and 10; eFigures 4 and 5 in the Supplement). The regimen with the highest SUCRA value (ie, best performance) for all bleeding outcomes was NOAC plus P2Y₁₂ inhibitor (SUCRA of 77-94). Most notably, this regimen was the preferred strategy for intracranial hemorrhage (94.2), followed by NOAC plus DAPT (63.5). A regimen of VKA plus P2Y₁₂ inhibitor was the runner-up for the other bleeding outcomes (67-79). For efficacy outcomes (MACE and its individual components), there was no treatment regimen clearly favored overall. For hospitalization, the combination of NOAC plus P2Y₁₂ inhibitor was equally favorable (69.2) to NOAC plus DAPT (70.0) and followed by VKA plus P2Y₁₂ inhibitor (50.2). All fitted models converged well (eFigure 6 in the Supplement) and we did not find evidence that indicated statistical inconsistency in our NMA (eAppendix 2 in the Supplement).

As shown in eAppendix 3 in the Supplement, we performed a number of additional analyses. We found similar outcomes when using an NMA in which rivaroxaban, apixaban, and dabigatran were compared with VKA plus DAPT (eAppendix 3A in the Supplement). Regimens with factor Xa inhibitors (rivaroxaban and apixaban) had similar safety and efficacy as those with a thrombin inhibitor (either as higher or lower dose of dabigatran or combined) when compared with VKA plus DAPT (eAppendix 3B and 3C in the Supplement). We found that including the results of the 6-week landmark analysis from ISAR-TRIPLE as part of a sensitivity analysis resulted in similar results compared with the main analysis (eAppendix 3E in the Supplement).

Navigating the delicate balance between bleeding and ischemic risk is a challenge to clinicians and patients who have AF and need for antiplatelet therapy. The general consensus is to continue oral anticoagulation and to modify antiplatelet intensity and/or duration, but the optimal regimen has yet to be identified.^21,22 In our NMA of RCTs investigating the safety and efficacy of antithrombotic regimens, we found that an antithrombotic regimen of a NOAC plus P2Y₁₂ inhibitor results in less bleeding, including intracranial hemorrhage, compared with VKA plus DAPT. Among antiplatelet strategies, omitting aspirin caused less bleeding, without significant difference in MACE, compared with strategies including aspirin. These results support the use of a NOAC plus P2Y₁₂ inhibitor regimen as the preferred treatment strategy in this high-risk patient population. Routine use of regimens that include a VKA plus a P2Y₁₂ inhibitor plus aspirin should generally be avoided.

A prior NMA²³ investigated antithrombotic strategies in patients with AF and ACS and/or PCI. The authors found that the combination of VKA and a single antiplatelet agent provided the best choice in terms of safety and efficacy, with a very low dose of rivaroxaban, 2.5 mg, twice daily plus DAPT as an acceptable alternative. These findings are of interest because they confirm that conventional triple therapy with VKA, P2Y₁₂ inhibitor, and aspirin should be avoided. However, the study also found that triple therapy with a low dose of NOAC in combination with P2Y₁₂ inhibitor and aspirin could be considered and was preferable vs a standard dose of NOAC plus P2Y₁₂ inhibitor. A 2018 study²⁴ using a bivariate analysis based on an Open-label, Randomized, Controlled, Multicenter Study Exploring Two Treatment Strategies of Rivaroxaban and a Dose-adjusted Oral Vitamin K Antagonist Treatment Strategy in Subjects With Atrial Fibrillation Who Undergo Percutaneous Coronary Intervention (PIONEER AF-PCI) and Randomized Evaluation of Dual Antithrombotic Therapy With Dabigatran vs Triple Therapy With Warfarin in Patients With Nonvalvular Atrial Fibrillation Undergoing Percutaneous Coronary Intervention (RE-DUAL PCI) found that all tested NOAC combinations were superior in terms of bleeding and noninferior in efficacy when compared with VKA plus DAPT.

During the last decade, the role of aspirin in combination antithrombotic regimens had been questioned. A 2018 meta-analysis,⁸ restricted to data from high-quality randomized clinical trials (n = 5317), concluded that dual antithrombotic therapy with an oral anticoagulant and P2Y₁₂ inhibitor may be a better option than triple therapy in many patients with AF following PCI, based on a 47% reduction in major or minor bleeding with comparable outcomes of MACE. Our meta-analysis, which adds almost 5000 patients, confirms these findings. This is particularly relevant for more rare events, such as stent thrombosis, which is an important reason to use DAPT in the month(s) following PCI prior to re-endothelialization of the stent struts and/or polymer material and is particularly of concern when PCI occurs in the (prothrombotic) setting of ACS.²⁵ A systematic review²⁶ that included observational data suggested comparable outcomes of P2Y₁₂ inhibitor with and without aspirin in terms of stent thrombosis and other MACE outcomes. In our NMA, we also found no clear signal of increased stent thrombosis in the groups taking antithrombotic regimens without aspirin. In patients who underwent PCI, a loading dose of aspirin followed by a few days of aspirin was typically administered, irrespective of subsequent regimen. It is therefore possible that, following PCI, dual antithrombotic therapy with an oral anticoagulant and a P2Y₁₂ inhibitor might be comparable with triple therapy for prevention of thrombotic events, including stent thrombosis, and should be preferred vs triple therapy for most patients owing to its better safety profile. Other studies in patients with ACS without AF have demonstrated the bleeding risk associated with full-dose triple therapy with a NOAC, P2Y12 inhibitor, and aspirin.^27,28

The scientific basis for VKA therapy to prevent arterial thrombosis in the setting of AF is not platelet dependent but includes interference with effects of tissue-factor expressed by activated endothelial cells and monocytes.²⁹ Prior research has demonstrated that oral anticoagulation is more effective in preventing reinfarction and stroke than aspirin plus clopidogrel in patients with AF.³⁰ Moreover, temporary discontinuation of oral anticoagulants or omission of P2Y₁₂ inhibitors in patients with AF undergoing PCI is associated with increased risk for ischemic events without reducing the risk of bleeding complications.^31,32 In addition, studies that have added aspirin to VKA vs VKA monotherapy did not clearly demonstrate an additional benefit in patients following MI while increasing the risk of bleeding.^33,34 While the exact underlying mechanisms may not be fully elucidated, it is known that oral anticoagulants, including NOACs, exert antiplatelet effects that result in prevention of both stroke and recurrent MI and that are further synergized by P2Y₁₂ inhibition.^35-38 Thus, a dual pathway of platelet inhibition with oral anticoagulants plus P2Y₁₂ inhibition might be enough for most patients with AF and ACS and/or PCI and adding aspirin might only increase the risk of bleeding, including intracranial bleeding, without a significant additional benefit in efficacy outcomes.

So how can we explain the favorable safety profile of NOACs plus P2Y₁₂ inhibition vs VKA-based regimens? Unlike VKAs, the NOACs exert their anticoagulant directly by either targeting factor Xa (apixaban, rivaroxaban, and edoxaban) or as a direct thrombin inhibitor (dabigatran). Thus, NOACs offer simplification of long-term anticoagulation therapy by promoting optimal thromboembolic protection with less bleeding complications. Finally, our main conclusions are based on the significance of the ORs and 95% credible intervals as well as ranking of treatment strategies using SUCRA values for best performance. Both VKA plus P2Y₁₂ inhibitor and NOAC plus P2Y₁₂ inhibitor have less TIMI major bleeding, TIMI major or minor bleeding, and trial-defined primary safety than VKA plus DAPT; however, NOAC plus P2Y₁₂ inhibitor had lower odds of intracranial hemorrhage compared with VKA plus P2Y₁₂ inhibitor and lower odds of TIMI major bleeding than VKA plus P2Y₁₂ inhibitor compared with VKA plus DAPT. The AUGUSTUS trial had an arm that included 1126 patients treated with VKA plus P2Y₁₂ inhibitor, which was much larger than the 279 patients receiving VKA plus P2Y₁₂ inhibitor in WOEST. Given these results and the other well-known advantages of NOACs over VKA, we concluded that NOAC plus P2Y12 inhibitor is the preferred treatment option for this patient population.

Although an NMA overcomes some of the limitations of routine pairwise meta-analyses and allows for evidence-based grading, our NMA also has limitations. First, the estimated between-study heterogeneity as a standard deviation on the log OR scale is close to or smaller than the standard deviation of the estimated log ORs, which could not be ignored. Although there is no definite cut point to determine small or large heterogeneity; in our analysis, between-study heterogeneity exists and the potential sources include design, indication, type, dose and duration of antithrombotic therapy, and duration of follow-up, which may potentially affect the interpretation of our results. Although the sample size was sufficient to draw firm conclusions on the overall populations for the primary safety and efficacy outcomes, it may still be insufficient to ascertain more modest effects on rare but serious events, such as stent thrombosis. A few other ongoing studies, including the Edoxaban Treatment Versus Vitamin K Antagonist in Patients With Atrial Fibrillation Undergoing Percutaneous Coronary Intervention (ENTRUST-AF-PCI) trial with 1508 participants, may add some information on the safety (and efficacy) of (standard dose) edoxaban vs VKA both in combination with a P2Y₁₂ inhibitor.³⁹ However, the relatively small sample size of these studies is unlikely to change the results of our NMA. In addition, to further explore which patients would benefit most from a given treatment combination, an individual patient-level data analysis is warranted. This analysis could shed further light on whether adjustments need to be made by sex, race/ethnicity, age, clinical presentation (ACS and stable CAD), thromboembolic risk, bleeding risk, stent placement, stent type, and other procedural characteristics. Most patients used clopidogrel as a P2Y₁₂ inhibitor; as such, we do not know whether the use of other more potent P2Y₁₂ inhibitors (ticagrelor and prasugrel) have the same effects on safety and efficacy in combination with oral anticoagulant therapyDirect comparisons between NOACs may also be warranted because it is possible that not all NOACs or doses of NOACs perform the same in terms of the balance of safety and efficacy when combined with platelet inhibitors in the setting of ACS and/or PCI. Mechanistic studies into the role of platelet function and genetic testing in patients treated with VKA or NOAC and antiplatelet therapy after PCI to develop personalized antithrombotic regimens is another area that remains poorly explored and warrants investigation. Genome-wide pharmacogenomics data suggest that several polymorphisms affect the response of VKA or NOACs.⁴⁰ In the future, a personalized medicine approach based on pharmacogenetics could be helpful in this complex heterogeneous patient population.

In patients with AF undergoing PCI, a regimen of NOAC plus P2Y₁₂ inhibitor was associated with fewer bleeding complications, including intracranial bleeding, without a significant difference in ischemic events compared with VKA plus DAPT. Although further data on outcomes of patients receiving VKA plus P2Y₁₂ inhibitor are needed for definitive conclusions, our results support the use of an NOAC and P2Y₁₂ inhibitor as the preferred treatment option for this high-risk patient population. In the absence of future trials showing incremental efficacy with acceptable bleeding resulting in a favorable net clinical benefit of triple therapy, regimens that include a VKA plus a P2Y₁₂ inhibitor plus aspirin should generally be avoided.

Corresponding Author: Renato D. Lopes, MD, PhD, Division of Cardiology, Duke Clinical Research Institute, Duke Health, 200 Morris St, Durham, NC 27701 (renato.lopes@duke.edu).

Accepted for Publication: May 2, 2019.

Published Online: June 19, 2019. doi:10.1001/jamacardio.2019.1880

Author Contributions: Dr Lopes had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Concept and design: Lopes, Hong, Harskamp, Bhatt, Cannon, Granger, Verheugt, Alexander.

Acquisition, analysis, or interpretation of data: Lopes, Hong, Harskamp, Bhatt, Mehran, Cannon, Li, ten Berg, Sarafoff, Gibson, Alexander.

Drafting of the manuscript: Lopes, Hong, Harskamp, Mehran.

Critical revision of the manuscript for important intellectual content: All authors.

Statistical analysis: Hong, Mehran, Li, Gibson.

Obtained funding: Lopes.

Administrative, technical, or material support: Lopes, Harskamp, Cannon, Verheugt, Sarafoff, Alexander.

Supervision: Lopes, Hong, Mehran, Cannon, Alexander.

Conflict of Interest Disclosures: Dr Lopes reported grants and personal fees from Bristol-Myers Squibb and Pfizer and personal fees from Boehringer Ingelheim and Bayer AG during the conduct of the study and grants from Amgen Inc, GlaxoSmithKline, Medtronic PLC, and Sanofi Aventis outside the submitted work. Dr Harskamp reported grants from Netherlands Organization for Scientific Research (NWO Rubicon) during the conduct of the study. Dr Bhatt reported personal fees from Harvard Clinical Research Institute (now Baim Institute for Clinical Research) and personal fees and other support from Boehringer Ingelheim during the conduct of the study; grants from Idorsia, Synaptic, Abbott, Regeneron, Amgen, Lilly, Chiesi, Ironwood, Pfizer, Forest Laboratories/AstraZeneca, Ischemix, Amarin, AstraZeneca, Bristol-Myers Squibb, Eisai, Ethicon, Medtronic, Sanofi Aventis, Roche, and The Medicines Company; other support from Novo Nordisk, Fractyl, Svelte, Merck, St Jude Medical (now Abbott), Biotronik, Cardax, Boston Scientific, Veterans Affairs, Clinical Cardiology, FlowCo, PLx Pharma, Medscape Cardiology, Regado Biosciences, Boston VA Research Institute, and Takeda; personal fees from Medtelligence/ReachMD, Bayer, TobeSoft, Cleveland Clinic, Mount Sinai School of Medicine, Journal of the American College of Cardiology, Duke Clinical Research Institute, Mayo Clinic, Population Health Research Institute, Belvoir Publications, Slack Publications, WebMD, HMP Global, Harvard Clinical Research Institute (now Baim Institute for Clinical Research), and Elsevier; personal fees, nonfinancial support, and other support from American College of Cardiology; personal fees and nonfinancial support from Society of Cardiovascular Patient Care; nonfinancial support from American Heart Association; and grants and other support from PhaseBio outside the submitted work. Dr Mehran reported grants from AstraZeneca, Bayer, Beth Israel Deaconess, Bristol Myers Squibb/Sanofi, CSL Behring, DSI, Medtronic, Novartis Pharmaceuticals, and OrbusNeich; personal fees from Boston Scientific, Medscape, Siemens Medical Solutions, Regeneron Pharmaceuticals, Roivant Sciences, and Sanofi; grants and personal fees from Abbott Vascular; other support from Abbott Laboratories, Spectranetics/Philips/Volcano, Janssen, BMS, Watermark Research, Medtelligence/Janssen, Claret Medical, and Elixir Medical; and personal fees and other support from PLx Opco/PLx Pharma outside the submitted work. Dr Cannon reported grants from Boehringer-Ingelheim, Daiichi Sankyo, and Janssen and personal fees from Boehringer-Ingelheim and Janssen during the conduct of the study and personal fees from Aegerion, Alnylam, Amarin, Amgen, BMS, Corvidia, Eisai, Innovent, Kowa, Merck, Pfizer, Regeneron, Sanofi and grants from Amgen, Bristol-Myers Squibb, and Merck outside the submitted work. Dr Granger reported grants and personal fees from Pfizer, Bristol-Myers Squibb, Daiichi Sankyo, Boehringer-Ingelheim, Janssen, and Bayer during the conduct of the study. Dr Verhuegt reported personal fees from Bayer HealthCare, Bristol-Myers Squibb/Pfizer, Boehringer-Ingelheim, and Daiichi-Sankyo during the conduct of the study. Dr Sarafoff reported travel grants from Bayer Healthcare, and Bristol-Myers Squibb/Pfizer. Dr Gibson reported grants and personal fees from Portola, Bayer, Janssen, and Johnson and Johnson; and grants from Bristol-Myers Squibb during the conduct of the study; grants and personal fees from Portola, Bayer, Janssen, and Johnson and Johnson, and grants from Bristol-Myers Squibb outside the submitted work. Dr Alexander reported grants and personal fees from Bristol-Myers Squibb, CSL Behring; personal fees from Pfizer, AbbVie Pharmaceuticals, Novo Nordisk, Portola Pharmaceuticals, Quantum Genomics, Teikoku Pharmaceuticals, VA Cooperative Studies Program, Zafgen, and Boehringer-Ingelheim; and grants from AstraZeneca, CryoLife, the US Food and Drug Administration, the National Institutes of Health, Sanofi, and VoluMetrix. No other disclosures were reported.

Funding/Support: Support for this study was provided by Duke Clinical Research Institute. Salary support for Dr Harskamp was provided by a Rubicon fellowship provided by the Netherlands Organisation for Scientific Research.

Role of the Funder/Sponsor: The funding sources had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.