Dual Versus Triple Therapy for Atrial Fibrillation After Percutaneous Coronary Intervention: A Systematic Review and Meta-analysis

Abstract

Background

The safety and effectiveness of dual therapy (direct oral anticoagulant [DOAC] plus P2Y12 inhibitor) versus triple therapy (vitamin K antagonist plus aspirin and P2Y12 inhibitor) in patients with nonvalvular atrial fibrillation (AF) after percutaneous coronary intervention (PCI) is unclear.

Purpose

To examine the effects of dual versus triple therapy on bleeding and ischemic outcomes in adults with AF after PCI.

Data Sources

Searches of PubMed, EMBASE, and the Cochrane Library (inception to 31 December 2019) and ClinicalTrials.gov (7 January 2020) without language restrictions; journal Web sites; and reference lists.

Study Selection

Randomized controlled trials that compared the effects of dual versus triple therapy on bleeding, mortality, and ischemic events in adults with AF after PCI.

Data Extraction

Two independent investigators abstracted data, assessed the quality of evidence, and rated the certainty of evidence.

Data Synthesis

Four trials encompassing 7953 patients were selected. At the median follow-up of 1 year, high-certainty evidence showed that dual therapy was associated with reduced risk for major bleeding compared with triple therapy (risk difference [RD], —0.013 [95% CI, —0.025 to —0.002]). Low-certainty evidence showed inconclusive effects of dual versus triple therapy on risks for all-cause mortality (RD, 0.004 [CI, —0.010 to 0. 017]), cardiovascular mortality (RD, 0.001 [CI, —0.011 to 0.013]), myocardial infarction (RD, 0.003 [CI, —0.010 to 0.017]), stent thrombosis (RD, 0.003 [CI, —0.005 to 0.010]), and stroke (RD, —0.003 [CI, —0.010 to 0.005]). The upper bounds of the Cis for these effects were compatible with possible increased risks with dual therapy.

Limitation

Heterogeneity of study designs, dosages of DOACs, and types of P2Y12 inhibitors.

Conclusion

In adults with AF after PCI, dual therapy reduces risk for bleeding compared with triple therapy, whereas its effects on risks for death and ischemic end points are still unclear.

Primary Funding Source

None.

Revascularization by percutaneous coronary intervention (PCI) is considered the standard of care for patients with acute coronary syndrome (ACS) (1, 2). Dual antiplatelet therapy (DAPT; aspirin plus P2Y12 inhibitor) prevents major adverse cardiovascular events (MACE) after PCI for ACS or stable coronary artery disease (1–3). However, approximately 5% to 10% of patients undergoing PCI have atrial fibrillation (AF), which complicates the choice of optimal antithrombotic therapy (4–6). Evidence has favored direct oral anticoagulant (DOAC) agents over vitamin K antagonists (VKAs) in patients with nonvalvular AF for better safety (interms of bleeding) and effectiveness (in terms of MACE) (7, 8). More recently, randomized controlled trials (RCTs) compared an alternative approach-dual therapy consisting of a DOAC and a P2Y12 inhibitorm versus triple therapy comprising a VKA and DAPT-to identify an ideal antithrombotic strategy in patients with AF after PCI (6, 9–11). Theoretically, the cardiovascular benefits gained by using triple therapy could be offset by higher risk for bleeding, whereas withdrawal of aspirin might lead to higher rates of stent thrombosis and ischemic events with dual therapy (2, 12). We did a meta-analysis of contemporary RCTs to address the clinical conundrum of which antithrombotic regimen- dual (DOAC plus P2Y12 inhibitor) or triple (VKA plus DAPT) therapy-is most appropriate for the management of patients with AF and coronary artery disease who had PCI.

Methods

This meta-analysis was conducted following the Cochrane Collaboration guidelines and reported according to PRISMA (Preferred Reporting Items for Systematic reviews and Meta-Analyses) (13, 14). The protocol, although not registered, was submitted to PROSPERO on 21 November 2019.

Data Sources and Searches

The literature search was done without language restrictions using the electronic databases of PubMed, EMBASE, and the Cochrane Library from inception to 31 December 2019 and ClinicalTrials.gov (7 January 2020). Additional sources were Web sites of major cardiovascular and medical journals (www.onlinejacc.org, http://annals.org/aim, https://academic.oup.com/eurheartj, www.nejm.org, www.ahajournals.org/journal/circ, and https://jamanetwork.com ) and bibliographies of relevant studies. The search strategy included broad search terms, such as new oral anticoagulant, direct oral anticoagulant, DOAC, vitamin K antagonist, VKA, anti-platelets, percutaneous coronary intervention, PCI, acute coronary syndrome, ACS, and atrial fibrillation (Appendix Tables 1 to 4, available at Annals.org).

Study Selection

The prespecified inclusion criteria were RCTs that reported bleeding, mortality, and ischemic outcomes of interest and compared dual therapy using DOACs plus P2Y12 inhibitors versus triple therapy using VKAs plus DAPT in adult patients (aged >18 years) with AF receiving PCI. We placed no limitations on language, sample size, or follow-up duration. We excluded observational studies, registries, and post hoc analyses of RCTs. We also excluded RCTs that assessed dual therapy comprising a VKA plus a P2Y12 inhibitor or any oral anticoagulant plus aspirin and trials where a minority of patients (<50%) received PCI.

After removing duplicates, we screened articles first at the title and abstract level and then at the full- text level. Two investigators (M.U.K. and M.S.K.) independently screened and selected articles, and conflicts were resolved by discussion or opinion of an additional investigator (S.U.K.).

Data Extraction, Outcomes, and Quality Assessment

Two investigators (M.U.K. and M.S.K.) who were not involved in any of the selected trials independently abstracted data using prespecified collection forms, appraised the accuracy of the abstractions, and resolved any discrepancies by consensus after discussion with a third investigator (S.U.K.). The following information was abstracted: first author, journal, characteristics of the trial and participants, crude point estimates, number of events, sample size, and follow-up duration. The dual therapy group of RE-DUAL PCI (Randomized Evaluation of Dual Antithrombotic Therapy With Dabigatran vs. Triple Therapy With Warfarin in Patients with Nonvalvular AF Undergoing PCI) (10) used 2 dosages of dabigatran, 110 mg and 150 mg twice daily. As per our prespecified plan to analyze the 2 dosages separately, we abstracted outcomes from dual therapy with dabigatran, 150 mg, for our main analysis because this dose is approved in the United States for stroke prevention in patients with AF without significant renal impairment. We used data regarding dual therapy with dabigatran, 110 mg, for sensitivity analyses.

We also reviewed other meta-analyses of the selected trials for any key information not available in the original trial report (15–17). Specifically, data on all of the outcomes of interest were unavailable in the main article for AUGUSTUS (Open-Label, 2×2 Factorial, Randomized Controlled, Clinical Trial to Evaluate the Safety of Apixaban vs. VKA and Aspirin vs. Aspirin Placebo in Patients With AF and ACS or PCI) (6) but were subsequently provided in the meta-analysis by Lopes (the lead investigator of the trial) and colleagues (15). We abstracted the exclusive data for DOAC-based dual therapy and VKA-based triple therapy from this meta-analysis.

Two unblinded investigators (M.U.K. and M.S.K.) independently appraised the potential risk of bias of the RCTs using the Cochrane Risk of Bias Tool at the study and outcome levels (Appendix Table 5, available at Annals.org) (18, 19). The data abstraction of outcomes was based on the intention-to-treat principle. The main outcomes of interest were major bleeding events that met TIMI (Thrombolysis in Myocardial Infarction) criteria and all-cause mortality. The other end points were TIMI major and minor bleeding, intracerebral hemorrhage, trial-defined bleeding events, cardiovascular mortality, myocardial infarction (MI), stent thrombosis, stroke, and MACE.

Data Synthesis and Analysis

Two reviewers (M.U.K. and M.O.) assessed the certainty of the evidence under the supervision of a third reviewer (S.U.K.) using the GRADE (Grading of Recommendations Assessment, Development and Evaluation) approach (GRADEpro GDT) (https://gdt.gradepro.org/app ) (Appendix Table 6, available at Annals.org) (20). Certainty of evidence was classified as high, moderate, low, or very low.

Estimates were pooled using an inverse-variance random-effects model. The Paule-Mandel method was used for estimation of τ ². We applied standard or modified (in case τ ² = 0) Hartung-Knapp-Sidik-Jonkman small-sample adjustments because we had fewer than 10 studies (21). We reported effect sizes as the risk difference (RD) with 95% CI. We used I² statistics to measure the extent of unexplained statistical heterogeneity: I² greater than 50% was considered a high degree of between-study heterogeneity (22). We did not examine publication bias because the small number of studies (<10) meant that our meta-analysis was underpowered to detect it. We used “meta” commands from Stata, version 16 (StataCorp), for all analyses.

Role of the Funding Source

The study received no funding.

Results

Design of Included Trials

Of 265 articles reviewed for eligibility, we selected 4 RCTs encompassing 7953 patients (Appendix Figure 1, available at Annals.org). The RCTs varied substantially in design and participant characteristics (Table and Appendix Table 7 [available at Annals.org]). PIONEER AF-PCI (Open-Label, Randomized, Controlled, Multicenter Study Exploring Two Treatment Strategies of Rivaroxaban and a Dose-Adjusted Oral VKA Treatment Strategy in Subjects With AF Who Undergo PCI) (9) randomly assigned 2124 patients with AF receiving PCI into the following 3 antithrombotic groups: rivaroxaban, 15 mg daily, plus a P2Y12 inhibitor (dual therapy); rivaroxaban, 2.5 mg twice daily, plus DAPT; and a VKA plus DAPT (triple therapy). RE-DUAL PCI (10) randomly assigned 2725 patients with AF receiving PCI (51% with ACS) to triple therapy or 1 of 2 dosages of dabigatran-based dual therapy (150 or 110 mg twice daily). AUGUSTUS (6) randomly assigned 4614 patients with AF presenting with ACS or having PCI in a 2 × 2 factorial design to either 5 mg of apixaban twice daily versus VKA or aspirin versus placebo, both on a background of P2Y12 inhibitor therapy. The ENTRUST-AF PCI (Edoxaban Treatment Versus VKA in Patients With AF Undergoing PCI) trial (11) randomly assigned 1506 patients with AF and recent PCI to full- dose edoxaban-based dual therapy versus VKA-based triple therapy. Follow-up was 12 months in PIONEER AF-PCI and ENTRUST-AF PCI, 14 months in RE-DUAL PCI, and 6 months in AUGUSTUS. The definition of the primary bleeding end point varied across trials: RE-DUAL PCI, AUGUSTUS, and ENTRUST-AF PCI defined it as a major or clinically relevant nonmajor bleeding event according to criteria from the International Society on Thrombosis and Haemostasis, whereas PIONEER AF-PCI defined it as a composite of TIMI major or minor bleeding or bleeding requiring medical attention. All patients received PCI except in AUGUSTUS, where 23.9% were treated with medical therapy (6).

Table.

Characteristics of Included Trials

Characteristic	PIONEER AF-PCI (9)	RE-DUAL PCI (10)	AUGUSTUS (6)	ENTRUST-AF PCI (11)
Trial name	Open-Label, Randomized, Controlled, Multicenter Study Exploring Two Treatment Strategies of Rivaroxaban and a Dose-Adjusted Oral VKA Treatment Strategy in Subjects With AF Who Undergo PCI	Randomized Evaluation of Dual Antithrombotic Therapy With Dabigatran vs. Triple Therapy With Warfarin in Patients With Nonvalvular AF Undergoing PCI	Open-Label, 2 × 2 Factorial, Randomized Controlled, Clinical Trial to Evaluate the Safety of Apixaban vs. VKA and Aspirin vs. Aspirin Placebo in Patients With AF and ACS or PCI	Edoxaban Treatment Versus VKA in Patients With AF Undergoing PCI
Patients, n	2124	2725	4614	1506
Enrollment initiation	May 2013	July 2014	June 2015	February 2017
Enrollment completion	July 2015	October 2016	April 2018	June 2019
Year of publication	2016	2017	2019	2019
Population	Patients who had paroxysmal, persistent, or permanent nonvalvular AF and who had received PCI with stent placement within 72 h	Patients who had nonvalvular AF and had successfully undergone PCI with a bare-metal or drug-eluting stent within the previous 6–120 h	Patients with AF with recent ACS (within 14 d) and/or PCI	Patients with AF who had a successful PCI for stable CAD or ACS (within 4–5 d)
Trial type	Open-label, multicenter RCT	Open-label, multicenter RCT	Open-label, multicenter RCT	Open-label, multicenter RCT
Inclusion criteria	Age ≥18 y AF that occurred within past 1 y, or AF that occurred >1 y ago and the participants had been receiving an OAC for AF for the past 3 mo	Age ≥18 y Patients with nonvalvular AF who just had PCI with a bare-metal or drug-eluting stent for ACS or unstable angina Patients who have been receiving an OAC or who were treatment-naive before PCI	Age ≥18 y Patients with either active or a history of AF or flutter with planned or existing use of an OAC for prophylaxis of thromboembolism Patients who have had ACS and/or PCI within the prior 14 d Planned use of an approved P2Y12 inhibitor for ≥6 mo	Age ≥18 y Nonvalvular AF and PCI for stable CAD or ACS OAC for >12 mo
Exclusion criteria	History of stroke or transient ischemic attack Significant gastrointestinal bleeding within 12 mo Calculated creatinine clearance <30 mL/min/1.73 m2 Anemia with a hemoglobin level <100 g/L	Presence of bioprosthetic or mechanical heart valves Creatinine clearance <30 mL/min/1.73 m2	Patients with other conditions that require anticoagulation (such as prosthetic valves or moderate or severe mitral stenosis) Severe renal insufficiency History of intracranial hemorrhage	Other conditions that require anticoagulation, such as moderate to severe mitral stenosis, mitral valve rheumatic disease, mechanical heart valve, or pulmonary embolism History of LAA closure, left ventricular or LAA thrombus High risk for bleeding Contraindication to anticoagulation Anemia with a hemoglobin level <80 g/L and platelet count <50 × 109 cells/L Renal failure, liver disease
Randomization sequence	Central randomization, computer-generated	Interactive response technology	Interactive voice response system	Web-based, central randomization, computer-generated
Treatments	DOAC + P2Y12 inhibitor (rivaroxaban, 15 mg/d; clopidogrel, 75 mg/d) DOAC + DAPT (rivaroxaban, 2.5 mg twice daily; aspirin, 75–100 mg/d; clopidogrel, 75 mg/d) VKA + DAPT (warfarin*; aspirin, 75–100 mg/d; clopidogrel, 75 mg/d)	DOAC (low dose) + P2Y12 inhibitor (dabigatran etexilate, 110 mg twice daily; either clopidogrel or ticagrelor) DOAC (high dose) + P2Y12 inhibitor (dabigatran etexilate, 150 mg twice daily; either clopidogrel or ticagrelor) VKA + DAPT (warfarin; aspirin, ≤100 mg/d; either clopidogrel or ticagrelor)	DOAC + DAPT (apixaban, 5 mg twice daily; aspirin, 81 mg/d; clopidogrel or ticagrelor or prasugrel) DOAC + P2Y12 inhibitor (apixaban, 5 mg twice daily; clopidogrel or ticagrelor or prasugrel) VKA + DAPT (warfarin; aspirin, 81 mg/d; clopidogrel or ticagrelor or prasugrel) VKA + P2Y12 inhibitor (warfarin; clopidogrel or ticagrelor or prasugrel)	DOAC + P2Y12 inhibitor (edoxaban, 60 mg/d or 30 mg/d; clopidogrel, 75 mg/d, or prasugrel, 5–10 mg/d, or ticagrelor, 90 mg twice daily) VKA + DAPT (warfarin; aspirin, 100 mg/d; clopidogrel, 75 mg/d, or prasugrel, 5–10 mg/d, or ticagrelor, 90 mg twice daily)
Definition of primary bleeding outcome	Composite of major bleeding or minor bleeding according to TIMI or bleeding requiring medical attention	Major or CRNM bleeding event according to ISTH	Composite of major or CRNM bleeding event according to ISTH	Composite of major or CRNM bleeding event according to ISTH
Definition of MACE	Composite of death from cardiovascular causes, MI, or stroke	Thromboembolic events, death, or unplanned revascularization	Composite of all-cause death or ischemic event	Composite of cardiovascular death, stroke, systemic embolic events, MI, or definite stentthrombosis
Definition of stent thrombosis	Definite stent thrombosis (not clearly reported)	Definite stent thrombosis	Definite/probable stent thrombosis	Definite stent thrombosis
Type of index event, %†
Non-STEMI	18.1	21.7	–	52
STEMI	11.5	14.7	–	52
Unstable angina	22.2	17.7	–	52
Type of stent, %
Bare-metal	32.2	14.9	–	12.2
Drug-eluting	65.9	82.7	–	84.7
Bare-metal and drug-eluting	1.9	1.5	–	1.5
Type of P2Y12 inhibitor, %
Clopidogrel	94.7	87.8	–	92.5
Prasugrel	1.2	0	–	<1
Ticagrelor	4.1	12.2	–	7.5
Analysis	Modified ITT	ITT	Modified ITT and ITT	ITT
Follow-up	12 mo	14 mo	6 mo	12 mo
Funding	Janssen Scientific Affairs and Bayer Pharmaceuticals	Boehringer Ingelheim	Bristol-Myers Squibb and Pfizer	Daiichi Sankyo

ACS = acute coronary syndrome; AF = atrial fibrillation; AUGUSTUS = 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; CAD = coronary artery disease; CRNM = clinically relevant nonmajor; DAPT = dual antiplatelet therapy; DOAC = direct OAC (non-VKA); ENTRUST-AF PCI = Edoxaban Treatment Versus Vitamin K Antagonist in Patients With Atrial Fibrillation Undergoing Percutaneous Coronary Intervention; ISTH = International Society on Thrombosis and Haemostasis; ITT = intention-to-treat; LAA = left atrial appendage; MACE = major adverse cardiovascular events; MI = myocardial infarction; OAC = oral anticoagulant; PCI = percutaneous coronary intervention; PIONEER AF-PCI = 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; RCT = randomized controlled trial; RE-DUAL PCI = Randomized Evaluation of Dual Antithrombotic Therapy With Dabigatran vs. Triple Therapy With Warfarin in Patients With Nonvalvular Atrial Fibrillation Undergoing Percutaneous Coronary Intervention; STEMI = ST-segment elevation MI; TIMI = Thrombolysis in Myocardial Infarction; VKA = vitamin K antagonist.

^*Target international normalized ratio, 2–3.

^†100% of patients had ACS in AUGUSTUS.

Overall, mean patient age ranged from 68.6 to 71.7 years, and the proportion of enrolled women varied from 23.6% to 30.7%. Disease prevalence ranged from 73.3% to 91.0% for hypertension, 28.8% to 37.9% for diabetes, and 17.1% to 52.0% for ACS. Between 65.4% and 86.2% of patients received a drug-eluting stent. Most patients had annual thromboembolic and bleeding risks greater than 3%. The median follow-up duration across the trials was 1 year (interquartile range, 0.87 to 1.04 years).

All trials were industry-funded, and all used an open-label design that could affect treatment or reporting bias. In PIONEER AF-PCI (9), randomization was adequate across main comparative groups but patients were not randomly assigned to a stratum of DAPT duration (1,6, or 12 months); this stratum was allocated at the clinician’s discretion, generating risk for selection bias. All trials reported minimal loss to follow-up and had minimal missing outcome data (the figures show full risk-of-bias assessments).

Outcomes

High-certainty evidence showed that dual therapy was associated with lower risks for TIMI major bleeding (RD, −0.013 [95% CI, −0.025 to −0.002]) (Figure 1, top), TIMI major and minor bleeding (RD, −0.031 [−0.049 to −0.012]) (Figure 1, bottom), and trial- defined bleeding (RD, −0.072 [CI, −0.129 to −0.015]) (Appendix Figure 2 [top], available at Annals.org) compared with triple therapy. We found no statistically significant difference between dual and triple therapy in terms of intracerebral hemorrhage (RD, −0.004 [CI, −0.009 to 0.000]; moderate-certainty evidence) (Appendix Figure 2, bottom).

Figure 1.

Effect of dual versus triple therapy on TIMI major bleeding (top) and TIMI major and minor bleeding (bottom).

AUGUSTUS = 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; ENTRUST-AF PCI = Edoxaban Treatment Versus Vitamin K Antagonist in Patients With Atrial Fibrillation Undergoing Percutaneous Coronary Intervention; GRADE = Grading of Recommendations Assessment, Development and Evaluation; PIONEER AF-PCI = 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; RD = risk difference; RE-DUAL PCI = Randomized Evaluation of Dual Antithrombotic Therapy With Dabigatran vs. Triple Therapy With Warfarin in Patients With Nonvalvular Atrial Fibrillation Undergoing Percutaneous Coronary Intervention; TIMI = Thrombolysis in Myocardial Infarction.

Low-certainty evidence showed that dual therapy had an inconclusive effect compared with triple therapy on risks for all-cause mortality (RD, 0.004 [CI, −0.010 to 0.017]) (Figure 2, top), cardiovascular mortality (RD, 0.001 [CI, −0.011 to 0.013]) (Figure 2, bottom), MI (RD, 0.003 [CI, −0.010 to 0.017]) (Figure 3, top), stent thrombosis (RD, 0.003 [CI, −0.005 to 0.010]) (Figure 3, bottom), and MACE (RD, 0.003 [CI, −0.016 to 0.023]) (Figure 4, top). The upper bounds of the CIs of these estimates were compatible with a possible increased risk for ischemic outcomes with dual versus triple therapy. Low-certainty evidence showed no statistically significant difference in effect on stroke risk between dual and triple therapy (RD, −0.003 [CI, −0.010 to 0.005]) (Figure 4, bottom).

Figure 2.

Effect of dual versus triple therapy on all-cause mortality (top) and cardiovascular mortality (bottom).

AUGUSTUS = 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; ENTRUST-AF PCI = Edoxaban Treatment Versus Vitamin K Antagonist in Patients With Atrial Fibrillation Undergoing Percutaneous Coronary Intervention; GRADE = Grading of Recommendations Assessment, Development and Evaluation; PIONEER AF-PCI = 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; RD = risk difference; RE-DUAL PCI = Randomized Evaluation of Dual Antithrombotic Therapy With Dabigatran vs. Triple Therapy With Warfarin in Patients With Nonvalvular Atrial Fibrillation Undergoing Percutaneous Coronary Intervention.

Figure 3.

Effect of dual versus triple therapy on myocardial infarction (top) and stent thrombosis (bottom).

AUGUSTUS = 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; ENTRUST-AF PCI = Edoxaban Treatment Versus Vitamin K Antagonist in Patients With Atrial Fibrillation Undergoing Percutaneous Coronary Intervention; GRADE = Grading of Recommendations Assessment, Development and Evaluation; PIONEER AF-PCI = 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; RD = risk difference; RE-DUAL PCI = Randomized Evaluation of Dual Antithrombotic Therapy With Dabigatran vs. Triple Therapy With Warfarin in Patients With Nonvalvular Atrial Fibrillation Undergoing Percutaneous Coronary Intervention.

Figure 4.

Effect of dual versus triple therapy on major adverse cardiovascular events (top) and stroke (bottom).

AUGUSTUS = 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; ENTRUST-AF PCI = Edoxaban Treatment Versus Vitamin K Antagonist in Patients With Atrial Fibrillation Undergoing Percutaneous Coronary Intervention; GRADE = Grading of Recommendations Assessment, Development and Evaluation; PIONEER AF-PCI = 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; RD = risk difference; RE-DUAL PCI = Randomized Evaluation of Dual Antithrombotic Therapy With Dabigatran vs. Triple Therapy With Warfarin in Patients With Nonvalvular Atrial Fibrillation Undergoing Percutaneous Coronary Intervention.

Sensitivity analyses after pooling dual therapy using dabigatran, 110 mg, showed consistent bleeding and ischemic outcomes compared with triple therapy (Appendix Figures 3 to 7, available at Annals.org).

Discussion

The following are the 2 principal findings of our meta-analysis: high-certainty evidence shows that DOAC-based dual therapy reduces risk for bleeding events compared with VKA-based triple therapy, and low-certainty evidence shows inconclusive effects of dual versus triple therapy on risks for death and ischemic end points, such as MI, stent thrombosis, stroke, and MACE. Results including sensitivity analyses, however, were compatible with a possible increased risk for ischemic end points with use of dual versus triple therapy.

Our results reflect the outcome trends in all 4 included RCTs. In PIONEER AF-PCI (9), both rivaroxaban- based antithrombotic regimens showed lower risk for clinically significant bleeding than triple therapy (16.8% vs. 18.0% vs. 26.7%) at 12 months. In RE-DUAL PCI (10) at 14 months, triple therapy was associated with a 5.5- percentage point increase in major or clinically relevant nonmajor bleeding events compared with dabigatran, 150 mg; the corresponding increase compared with dabigatran, 110 mg, was 11.5 percentage points.

These findings were consistent regardless of index PCI subgroup (such as ACS or stable coronary artery disease), stent type, or type of P2Y12 inhibitor therapy. What remained unclear from these 2 trials was whether the decreased bleeding risk seen with dual therapy was exclusively driven by using a DOAC instead of a VKA or was due to an aspirin-free strategy (12). AUGUSTUS (6) explored this issue: At 6 months, apixaban was associated with a 4.2-percentage point reduction in major or clinically relevant nonmajor bleeding compared with VKA; addition of aspirin was associated with a 7.1- percentage point increase in bleeding compared with placebo. Contrary to all previous trials, ENTRUST-AF PCI showed no significant differences for major or clinically relevant nonmajor bleeding events between dual therapy (17%) and triple therapy (20%) (11). These results could have been driven by lower bleeding rates in the VKA group during the first 2 weeks of treatment, when a high proportion of patients did not achieve an international normalized ratio of 2 (69% in the first week and 42% in the second week).

All RCTs showed a numerical increase in ischemic end points with dual therapy. These findings should be cautiously interpreted in light of the limitations of the individual RCTs. In PIONEER AF-PCI, statistical power to detect important differences between groups for all efficacy end points was low, varying between 5.4% and 13% (9). The stent thrombosis end point was not centrally adjudicated, and the directionality of efficacy end points was inconsistent among antithrombotic groups. All dosing regimens of rivaroxaban were less than the 20-mg daily dose approved for stroke prevention in AF for patients without significant renal impairment. In RE-DUAL PCI, the composite efficacy end point was underpowered to detect clinical differences (10). Of note, a pooled analysis of both doses of dabigatran-based dual therapy in RE-DUAL PCI showed a 1.1-percentage point increase in the efficacy end point (10). However, the rate of the efficacy end point was numerically lower in the comparison of dual therapy using 150 mg of dabigatran (11.8%) versus triple therapy (12.8%) than in the comparison of dual therapy using 110 mg of dabigatran (13.7%) versus triple therapy (13.4%); thus, it can be inferred that higher thromboembolic tendency might be solely limited to the 110-mg- based regimen. In AUGUSTUS, rates of MI and stent thrombosis increased by 0.5% and 0.4%, respectively, among patients who were not receiving aspirin (6). Nevertheless, low event rates, inadequate power to assess cardiovascular outcomes, and limited follow-up duration were major limitations of this trial.

The management of AF after PCI is a common clinical conundrum. Fundamental questions include the timing of aspirin withdrawal, whether discontinuation of aspirin therapy can compromise any potential cardiovascular benefits, and the choice of DOAC versus VKA for oral anticoagulation. All included RCTs tested the safety of early withdrawal of aspirin-that is, before hospital discharge-on the basis of the rationale that risk for bleeding is generally higher during the first month after PCI because of periprocedural use of antithrombotic therapy (6, 9–12, 23). A guideline from the American College of Cardiology, American Heart Association, and Heart Rhythm Society recommends limiting aspirin use in the preprocedural period and during hospitalization (24). However, 2018 guidelines from the European Society of Cardiology restrict use of dual therapy only in patients with high bleeding risk at baseline and recommend 1 to 6 months of triple therapy for all other patients, based on thrombotic and bleeding risk assessment (class of recommendation, Ila for both guidelines). These recommendations were generated before the AUGUSTUS and ENTRUST-AF PCI trials (6, 11). Our meta-analysis might promote more consistency between American and European society guidelines regarding the safety of early withdrawal of aspirin from triple therapy.

Historically, aspirin has remained the cornerstone of the secondary prevention strategies after PCI (23). However, an emerging paradigm shift favors P2Y12 inhibitor monotherapy over aspirin monotherapy after discontinuation of DAPT. Four recent RCTs (25–28) compared an early deescalation of DAPT to P2Y12 inhibitor monotherapy (1 to 3 months) versus 12 months of DAPT; these trials unanimously showed better safety in terms of bleeding events without worsening MACE. Aspirin has very limited efficacy in preventing cardio- embolic stroke particular to AF given the specific nature of thrombi (that is, less platelet mass than arterial thrombi) (23). The post hoc analysis of the SPAF (Stroke Prevention in Atrial Fibrillation) trial confirmed that aspirin therapy resulted in greater reductions in the incidence of noncardioembolic than cardioembolic stroke (risk reduction, 100% vs. 31%; P = 0.01) (23, 29).

Regarding oral anticoagulation, DOACs have shown a more favorable risk-benefit profile than VKAs in AF (7). AUGUSTUS (6) showed that apixaban was associated with lower rates of death and hospitalization, driven by the incident hospitalization rate and 50% relative risk (RR) reduction in stroke compared with VKA. Current American and European guidelines also favor DOACs over VKAs for AF management in the absence of contraindications (1, 24). Finally, the factorial design of AUGUSTUS signaled that perhaps both aspects of antithrombotic strategies-early withdrawal of aspirin and preference for DOACs over VKAs-offer the optimal balance of safety and effectiveness in the management of AF after PCI.

Our results are generally consistent with those of other meta-analyses using the same RCT data (11, 15). However, a recent meta-analysis by Gargiulo and colleagues (30) showed a statistically significant 59% higher RR for stent thrombosis with dual therapy than triple therapy. Their study varied from the current meta-analysis in several ways. Given the unique design of AUGUSTUS, it was critical to pool data specific to the DOAC-based dual therapy and VKA-based triple therapy groups (6). Pooling this data set using a DerSimo- nian-Laird random-effects model (a method used by Gargiulo and colleagues) will generate an RR of 1.55 (CI, 0.99 to 2.41) with dual versus triple therapy, as shown in the secondary analysis by Gargiulo and colleagues (30). However, they pooled the aspirin and placebo groups for their primary analyses without accounting for the fact that participants in both oral anticoagulant groups were randomly assigned to aspirin or placebo. The DerSimonian-Laird random-effects model is known to substantially underestimate the 95% CI when only a few studies report effect estimates (31). In the study by Gargiulo and colleagues (30), the RR for stent thrombosis was 1.59 (CI, 1.01 to 2.50), which barely excluded 1. This marginal effect could easily be due to the use of the DerSimonian-Laird estimator. We used a more robust statistical method that is consistent with general recommendations from Hartung and Knapp (32) and Veroniki and colleagues (21). Another meta-analysis by Vranckx and colleagues (11) had findings consistent with our own.

The included trials in our meta-analysis had methodological heterogeneities. In AUGUSTUS, patients were enrolled and randomly assigned to a group within 14 days (median time to randomization, 6.6 days) after an ACS episode or PCI. Time to randomization was shorter in other trials. In RE-DUAL PCI, aspirin therapy was discontinued after 1 month in patients with a bare- metal stent or after 3 months in patients with a drugeluting stent. In PIONEER AF-PCI, patients received DAPT for 1, 6, or 12 months. Because we relied on trial-level information, we could not analyze results according to type of DOAC or P2Y12 inhibitor, age, or comorbid conditions (for example, diabetes; renal failure; or history of bleeding or procedure-related factors, such as coronary anatomy complexity, stent length, or left main stenting). Clopidogrel was used in more than 90% of participants, and a very small proportion received ticagrelor and prasugrel. Thus, we could not examine the influence of different P2Y12 inhibitors on outcomes. However, the component RCTs of this meta-analysis did not find statistically significant subgroup interactions across different P2Y12 inhibitors with regard to bleeding or cardiovascular end points (6, 10, 11). Such differences can be comprehensively addressed by pooling individual patient data across trials. In all RCTs, lower rates of ischemic events were observed than anticipated, resulting in limited statistical power to detect differences between groups in ischemic and mortality outcomes. All RCTs excluded patients with renal dysfunction and enrolled patients with relatively low bleeding risk, limiting the generalizability of the findings. The open-label design of the trials might bias the results in favor of dual therapy.

In conclusion, our meta-analysis shows high- certainty evidence that supports the use of DOAC- based dual therapy over VKA-based triple therapy for reducing bleeding risk in patients with AF who received PCI. The use of dual therapy versus triple therapy had an inconclusive effect on risks for death and ischemic outcomes. The certainty of this evidence was low, and upper bounds of CIs signaled a possible increased risk for the ischemic end points. Future RCTs, such as COACH-AF-PCI (Dabigatran vs. Warfarin With Nonvalvular AF Who Undergo PCI) (NCT03536611), APPROACH-ACS-AF (APixaban vs. PhenpRocoumon in Patients With ACS and AF) (NCT02789917), OPTIMAL (Optimal Antithrombotic Therapy for ACS Patients Concomitant AF Undergoing New Generation Drug-Eluting Stent Implantation) (NCT03234114), and the Japanese SAFE-A (SAFety and Effectiveness Trial of Apixaban Use in Association With DAPT in Patients With AF Undergoing PCI) study (33), will be valuable to gain further insight regarding ischemic end points.