A Systematic Review of Randomized Trials Comparing Double versus Triple Antithrombotic Therapy in Patients With Atrial Fibrillation Undergoing Percutaneous Coronary Intervention

Abstract

BACKGROUND:

Prior randomized controlled trials (RCT) evaluating the optimal antithrombotic therapies for patients with atrial fibrillation (AF) undergoing percutaneous intervention (PCI) have not been powered to evaluate ischemic outcomes. We compared double therapy with oral anticoagulation (OAC) and a P2Y12 inhibitor to triple therapy with an OAC + dual antiplatelet therapy in patients with AF requiring PCI.

METHODS:

Using PRISMA guidelines, we searched for RCTs including patients with AF as an indication for OAC and undergoing PCI or medical management of acute coronary syndrome. The results were pooled using fixed-effects and random-effects models to estimate the overall effect of double therapy vs. triple therapy on ischemic and bleeding outcomes.

RESULTS:

We identified 4 RCTs, comprising 10,238 patients (5498 double therapy, 4740 triple therapy). Trial-reported major adverse cardiovascular events were similar between double therapy and triple therapy (fixed effect model OR 1.09, 95% CI 0.94–1.26). However, stent thrombosis (61/5496 double therapy v 33/4738 triple therapy; fixed effect model OR 1.57, 95% CI 1.02–2.40; NNT with triple therapy=242) favored triple therapy. Bleeding outcomes were less frequent with double therapy (746/5470 vs 950/4710; fixed effect model OR 0.59, 95% CI 0.53–0.65; NNH with triple therapy=16), but with significant heterogeneity (Q=8.33, p=0.04; I²=64%), as were intra-cranial hemorrhages (19/5470 vs 30/4710; fixed effect model OR 0.54, 95% CI 0.31–0.96).

CONCLUSIONS:

Double therapy in patients with AF requiring oral anticoagulation following PCI or ACS has a significantly better safety profile than triple therapy but may be associated with a modest increased risk of stent thrombosis.

INTRODUCTION

Patients presenting with atrial fibrillation (AF) undergoing percutaneous coronary intervention (PCI) for obstructive coronary artery disease present a unique therapeutic challenge. The American College of Cardiology/American Heart Association/European Society of Cardiology guideline recommends triple therapy (oral anticoagulant, P2Y12 inhibitor, and aspirin) in patients with atrial fibrillation who require anticoagulation after undergoing PCI.^1,2 The triple therapy approach has been challenged recently by a number of randomized controlled trials that demonstrated lower bleeding rates without a significant difference in ischemic cardiac events.^3–6 As ischemic events are less frequent than bleeding outcomes in this population, these studies have not been adequately powered to detect a difference in ischemic endpoints. Prior meta-analyses have achieved adequate power but have considered OAC with aspirin and OAC with P2Y12 inhibitor as equivalent double therapy regimens^7–10. They have also included patients with other indications for oral anticoagulation (such as pulmonary embolism or mechanical valve), which limits clinical utility in patient with AF.^7–10 We set out to perform a systematic review and meta-analysis to evaluate the safety and efficacy of double therapy, specifically with an OAC and P2Y12 inhibitor compared to traditional triple therapy with OAC, P2Y12 inhibitor, and aspirin, in preventing ischemic events.

METHODS

Following the Preferred Reporting Items for Systematic Reviews and Meta-Analysis guidelines¹¹, we conducted a systematic review and meta-analysis. The review protocol was registered at PROSPERO (http://www.crd.york.ac.uk/PROSPERO/ID=CRD42018091486).

Study Selection and Inclusion Criteria

This systematic review was carried out by two investigators (AES, MGN). All disagreements regarding eligibility or data extraction were resolved through discussion between the reviewers. The investigators initially screened studies by titles and abstracts followed by full-text review. Double therapy was defined as any oral anticoagulant plus any P2Y12 inhibitor without aspirin. Triple therapy was defined as any oral anticoagulant plus any P2Y12 inhibitor with aspirin.

Study inclusion criteria were 1) randomized trial with both a double and triple therapy arm; 2) double therapy had to utilize a P2Y12 inhibitor as the antiplatelet agent with an OAC as the oral anticoagulation agent; 3) all patients included had to have AF as the indication for anticoagulation; 4) all patients had to have been managed for ACS and/or be undergoing PCI; 5) data had to be presented in the manuscript or any available supplemental material in a manner amenable to data abstraction; 6) follow-up had to be greater than or equal to 3 months.

Observational studies, crossover trials, studies without both therapeutic strategies, and non-original data studies were excluded. Abstracts were reviewed to assess for publication bias but were not included in this study. No language, publication date, or publication status restrictions were utilized. Other previously published meta-analyses and systematic reviews to ensure all related studies were screened.

Search Strategies

We searched MEDLINE (via PubMed), EMBASE (via Elsevier), Cochrane Central Register of Controlled Trials (CENTRAL) (via Wiley), Web of Science (via Clarivate), and ClinicalTrials.gov from inception through September 10th, 2019. Our search strategy utilized key words and corresponding subject headings for anticoagulation, antiplatelet drugs, atrial fibrillation, PCI, acute coronary syndrome and was designed by a medical librarian with input from practicing clinicians. Full search details can be found in the supplement.

Outcomes

The primary outcome was trial-defined major adverse cardiac events (MACE). The trial-defined MACE endpoint varied slightly across trials.^3,5,6,12 We assessed all outcomes at the trial-specified follow-up point. Secondary outcomes included: 1) all-cause mortality; 2) myocardial infarction; 3) ischemic stroke; 4) stent thrombosis; 5) the trial-defined primary bleeding endpoint, 6) intracranial hemorrhage. In the case of PIONEER AF-PCI, all-cause death was not reported in the primary manuscript but was able to be abstracted from the meta-analysis by Lopes et al.¹³ We abstracted “definite” stent thrombosis from RE-DUAL PCI³ and “definite or probable” stent thrombosis from AUGUSTUS⁶ and ENTRUST-AF PCI¹²; stent thrombosis was not further classified in PIONEER AF-PC.I⁵ We permitted the bleeding endpoint to be trial-defined as trials utilized different bleeding definitions, including the Thrombolysis in Myocardial Infarction (TIMI) major or bleeding requiring medical attention⁵, clinically relevant non-major bleeding and the International Society on Thrombosis and Haemostasis (ISTH) bleeding scale.^3,6,12 PIONEER AF-PCI also compared very low dose rivaroxaban 2.5 twice daily (plus DAPT); this was not included in the meta-analysis as it has never been evaluated for stroke prevention in patients with AF.

Risk of Bias

Risk of bias were assessed by two authors (MGN, AES) using the Cochrane risk-of-bias tool for randomized controlled trials (Supplementary Table S1).¹⁴ Given the number of studies included in the meta-analysis, we did not assess for risk of publication bias.

Statistical Analysis

Following PRISMA guidelines, all available randomized controlled trials (RCT) were pooled using fixed-effects and random-effects models to estimate the overall effect of double therapy vs. triple therapy on ischemic and bleeding outcomes. Effects were described as odds ratios (OR) and 95% confidence intervals (CI). Heterogeneity was assessed using Q and I² statistics. Number needed to treat (NNT) for triple therapy vs. double therapy was calculated for stent thrombosis and number needed to harm (NNH) for triple therapy vs. dual therapy was calculated for the combined trial-defined bleeding endpoint and intracranial hemorrhage. Summary analyses were performed using RevMan 5.3.5.¹⁵

RESULTS

Study Identification

After initial database search, we screened 2001 studies by title and abstract. We identified a total of 6 studies potentially eligible for data extraction (Figure 1). Full-text screening led to exclusion of 2 of these studies. The 4 remaining studies met all criteria for inclusion and underwent data extraction to create our study population. Study characteristics can be found in table 1. A total of 10,238 patients were included, of which 5498 received double therapy and 4740 received triple therapy.

Figure 1. PRISMA Selection Process.

An outline of the number of references screened, studies screened, full-text studies assessed for eligibility and studies included.

Table 1.

Study Characteristics.

Studies	Study Design	Double Therapy (n)	Triple Therapy (n)	Age (yr)	Type of P2Y12	Type of anticoagulation	Anticoagulation indication	CHA2DS2-VASc Score	Follow-up	Primary Endpoint	Primary Bleeding endpoint	Ischemic Endpoints
PIONEER AF-PCI	RCT	709	706	69.9 triple therapy 70.4 double therapy	Clopidogrel, ticagrelor or prasugrel for 1, 6 or 12 months*	Low dose rivaroxaban 15 daily in double therapy vs. warfarin in triple therapy**	Atrial fibrillation	92.8 % ≥2 for triple therapy vs. 89.1% ≥2 for double therapy	12 months	Clinically significant bleeding	TIMI bleeding or bleeding requiring medical attention at one year.	Major adverse CV event (a composite of death from cardiovascular causes, MI, or stroke), each of the components, and stent thrombosis
RE-DUAL PCI	RCT	1744	981	71.7 and 68.8 for triple therapy/71.5 and 68.6 for double therapy	Clopidogrel or ticagrelor***	Dabigatran 110 mg PO BID or 150 mg PO BID	Non-valvular atrial fibrillation	Mean of 3.8 and 3.6 for triple therapy/Mean of 3.7 and 3.3 for double therapy	Mean 14 months	ISTH major or clinically relevant nonmajor bleeding	ISTH major or clinically relevant nonmajor bleeding	Composite efficacy endpoint of thromboembolic events, death, or unplanned revascularization and definite stent thrombosis
AUGUSTUS	RCT	2307	2307	70.6 for double therapy, 70.8 for triple therapy	Clopidogrel, ticagrelor or prasugrel	Apixiban or Warfarin	Atrial fibrillation	Mean of 3.9 for triple therapy/Mean of 3.9 for double therapy	6 months	ISTH major or clinically relevant nonmajor bleeding	ISTH major or clinically relevant nonmajor bleeding	Composite of death or hospitalization and the composite of death or ischemic events (stroke, myocardial infarction, stent thrombosis [definite or probable], or urgent revascularization)
ENTRUST-AF PCI	RCT	751	755	69 for double therapy, 70 for triple therapy	Clopidogrel, ticagrelor or prasugrel	Edoxaban or warfarin	Atrial Fibrillation	4.0 for both groups	12 months	ISTH major or clinically relevant non-major	ISTH major or clinically relevant non-major	Composite of cardiovascular death, stroke, systemic embolic events, myocardial infarction, and definite stent thrombosis.

^*In the 1- and 6-months groups, after that time period, patients stopped the P2Y12 inhibitor and continued the OAC + ASA

^**PIONEER AF-PCI also compared very low dose rivaroxaban 2.5 twice daily (plus DAPT). This was not included in the meta-analysis.

^***Continued for 1 month (BMS) or 3 months (DES), after which ASA was discontinued in the Triple Therapy group

Notable studies that have been included in other meta-analyses but excluded in this analysis were WOEST and ISAR-TRIPLE. WOEST was excluded as only 69% of patients had AF or atrial flutter as an indication for anticoagulation. The remainder were on anticoagulation for mechanical valves, apical aneurysm, pulmonary embolism, peripheral artery disease, or poor left ventricular function, which confer variable thrombotic risk from atrial fibrillation. ISAR-TRIPLE was excluded as patients were not randomized to receive double therapy vs triple therapy from time of index hospitalization and patients on oral anticoagulation + aspirin (rather than a P2Y12 inhibitor) comprised the double therapy arm, which began after 6 weeks of triple therapy.

Ischemic Outcomes

A total of 859 patients experienced a MACE event (8.4%) with 94 episodes of stent thrombosis across the four trials (0.92%). A composite outcome of trial-defined major adverse cardiovascular events was similar between double therapy (497/5496=9.0%) and triple therapy (362/4738=7.6%) (Figure 2; MH OR 1.09, 95% CI 0.94–1.26) and did not exhibit significant statistical heterogeneity among studies (Q=0.42, p=0.94; I²=0%). Double therapy patients had a higher risk of stent thrombosis (61/5496=1.11%) compared to those on triple therapy (33/4738=0.70%) (Figure 3; MH OR 1.57, 95% CI 1.02–2.40). Stent thrombosis included “definite” stent thrombosis from RE-DUAL PCI³, “definite or probable” stent thrombosis from AUGUSTUS⁶ and ENTRUST-AF PCI¹², and was not further classified in PIONEER AF-PCI⁵. There was a non-significant numerical excess of increased risk of myocardial infarction (MH OR 1.23, 95% CI 0.99–1.54) with double therapy. Ischemic stroke (MH OR 1.04, 95% CI 0.70–1.54) and all-cause death (MH OR 1.11, 95% CI 0.90–1.36) were similar between the two groups. The number needed to treat with triple therapy vs. dual therapy to prevent one stent thrombosis was 242.

Figure 2. Rates of Major Adverse Cardiovascular Events.

Forest plot demonstrating a composite value of trial-defined major adverse cardiovascular events between double therapy and triple therapy.

Figure 3. Rates of Stent Thrombosis.

Forest plot demonstrating the rates of stent thrombosis between double therapy and triple therapy in PIONEER AF-PCI, RE-DUAL PCI, AUGUSTUS and ENTRUST-AF PCI. Stent thrombosis was defined as “definite” stent thrombosis from RE-DUAL PCI³ and “definite or probable” stent thrombosis from AUGUSTUS⁶ and ENTRUST-AF PCI¹²; stent thrombosis was not further classified in PIONEER AF-PCI⁵.

Bleeding Outcome

A total of 1696 (16.7%) patients had a major bleeding event across four trials, with 49 (0.48%) ICH episodes. The trial-defined primary bleeding outcome significantly favored double therapy (746/5470=13.6%) over triple therapy (950/4710=20.2%) (Figure 4; MH OR 0.59, 95% CI 0.53–0.65) but with significant heterogeneity (Q=8.33, p=0.04; I² =64%). Intra-cranial hemorrhage also significantly favored double therapy (MH OR 0.54, 95% CI 0.31–0.96) with moderate heterogeneity (Q=5.62, p=0.13; I²=47%) Results were similar when random effects modeling was used (Supplementary Table S2). The NNH for the combined bleeding endpoint for triple therapy vs. dual therapy was 16; NNH for intracranial hemorrhage was 346.

Figure 4. Rates of Bleeding.

Forest plot demonstrating combined bleeding outcomes in the double therapy group compared to triple therapy, using the trial-defined primary bleeding endpoints.

DISCUSSION

The conversation regarding double therapy vs triple therapy in patients with atrial fibrillation after PCI has been a topic of debate for almost 2 decades. While multiple RCTs have demonstrated improved bleeding outcomes, neither RCT data nor prior meta-analyses have adequately assessed the ischemic consequences of using double therapy with P2Y12 inhibitors vs. triple therapy in patients with AF undergoing PCI. We present data assessing the specific use of a double therapy with a P2Y12 inhibitor as the lone antiplatelet agent in patients with AF, finding that this commonly used dual therapy approach may place patients at a modest increased risk of stent thrombosis. As has already been well-established, we confirmed that this small increased risk of stent thrombosis is outweighed by a significantly higher bleeding risk, including intracranial bleeding, seen with a triple therapy approach.

Prior meta-analyses have evaluated ischemic outcomes in patients on OAC requiring PCI,^7,9,13,16 but this meta-analysis demonstrates that AF patients on double therapy with a P2Y12 inhibitor alone are at modest increased risk for stent thrombosis. One major limitation of prior meta-analyses has been considering aspirin and P2Y12 inhibitors as equivalent therapies.^7,9,10 The differences in platelet inhibition^17–19 and clinical outcomes²⁰ between aspirin and P2Y12 inhibitors are well-established, suggesting that they should not be considered together when assessing double therapy. One recent network meta-analysis by Lopes et al¹³ provides an analysis of ischemic risk stratified by OAC/antiplatelet regimen. However, this meta-analysis included the WOEST trial, in which >30% patients did not have AF or atrial flutter. Our meta-analysis strictly considered patients with atrial fibrillation as the indication for oral anticoagulation, in whom the large majority underwent PCI. This is an important distinction as patients with atrial fibrillation are at a 1%−10% risk of stroke based on CHA2DS2-VASc score and an increased risk of MI and cardiovascular death, respectively, compared to other patients requiring OAC.²¹ These are risks that are not shared amongst patients who are anticoagulated for indications such as pulmonary embolus.²²

Without question, double therapy with NOAC is the preferred regimen in most patients requiring anticoagulation following PCI, especially in those at high risk for bleeding, and this is been supported by RCTs and several meta-analyses.^{3–6,9,13,16} However, the four RCTs that enrolled only patients with AF - PIONEER AF-PCI, RE-DUAL PCI, AUGUSTUS and ENTRUST-AF PCI - all had signals forewarning increased ischemic events in the double therapy group. This data supports that signal and suggests that there could be a minority of very high-risk patients who may benefit from a triple therapy approach. In patients with AF, the CHA2DS2-VASc score provides risk stratification for ischemic stroke, but identifying patients at risk of recurrent MI and stent thrombosis remains challenging for clinicians prescribing double therapy vs. triple therapy. Another challenge for clinicians is the fact that the risk factors that confer risk for thrombotic events, such as those used in the CHA2DS2-VASc score, also confer higher risk for bleeding.²³ More data is needed to help identify patients at increased risk for ischemic events, such as stent thrombosis, who may benefit the most from longer treatment with aspirin. Potential avenues for future investigation include a closer examination of more granular angiographic factors such as PCI characteristics and coronary lesion characteristics that may lead to greater ischemic risk. The bleeding cost of employing a triple therapy approach, even in high-risk patients, would be significant: in our cohort, we found a NNT of 242 to prevent 1 stent thrombosis with triple therapy, at the cost of 16 additional bleeding events. Given this disproportionate risk, the harm associated with increased bleeding with a triple therapy approach would appear to outweigh the potential benefits of this approach to prevent stent thrombosis in the majority of patients. Ultimately, we must balance the increased risk of rare but potentially catastrophic stent thrombosis demonstrated here with the far more common risk of bleeding when making patient-centered decisions regarding double therapy vs. triple therapy in patients with AF undergoing PCI.

We acknowledge a few limitations to our study. We would have preferred that 100% of patients underwent PCI to narrow our population of interest but we did not have access to patient-level data in each of the individual trials. However, by rigorously selecting studies where patients had either ACS and/or underwent PCI, only 10.7% of patients (all from AUGUSTUS) included in this meta-analysis did not undergo PCI, which we feel maintains reasonable clinical applicability of our results. Furthermore, the inclusion of patients who did not undergo PCI would actually result in the underestimation of the risk of stent thrombosis. Second, our MACE and bleeding endpoints were trial-defined due to endpoint differences across the four studies and limited access to patient-level data. Despite this, only our composite bleeding endpoint exhibited heterogeneity, which was driven by disproportionately low bleeding rates in the ENTRUST-AF PCI triple therapy group due to delayed achievement of therapeutic INRs and early discontinuation of aspirin.¹² However, the increased bleeding risk associated with triple therapy vs. double therapy has been well-established and consistent across RCTs. It is noteworthy that, while most RCTs compared a NOAC-based double therapy to warfarin-based triple therapy, AUGUSTUS was the only study allowing determination of how much risk reduction was due to avoiding aspirin rather than use of apixiban vs VKA. As indications for VKA dwindle, future studies should consider comparisons of NOAC-only regiments. Third, some of the studies included variable NOAC doses, some of which have not been evaluated for stroke prevention in AF. For example, the dabigatran 110 mg twice daily dose from RE-DUAL PCI is approved for use in patients with DVT/PE rather than AF in the United States, though was non-inferior to warfarin for prevention of stroke and systemic embolism in patient with AF in the RE-LY trial.²⁴ We did specifically exclude patients from PIONEER AF-PCI who received the very low dose rivaroxaban 2.5 mg dose for that reason. Fourth, we were not able to account for differences in the type of P2Y12 inhibitor used in the individual studies. Whether the selection of certain P2Y12 therapies, such as clopidogrel or ticagrelor, alter the relative risk of bleeding or ischemic outcomes remains to seen in this population. Importantly, clopidogrel was used in more than 90% of patients included in this meta-analysis, but epidemiologic studies have suggested that >30% of patients are resistant to clopidogrel.²⁵ Despite this, we observe a small absolute number of stent thrombosis with double therapy. This is likely due to the protective effect of NOACs, which was the predominant double therapy anticoagulation class in our meta-analysis and have been shown to reduce stent thrombosis after ACS.²⁶ Finally, we did not have access to specific procedural aspects of the PCIs performed in these trials, nor do we have details on the timing of stent thrombosis, both of which may be important in understanding risks and timing of stent thrombosis beyond the double therapy vs. triple therapy approach.

CONCLUSION

This meta-analysis of 4 RCTs shows that in patients with AF on OAC undergoing PCI, double therapy with a P2Y12 inhibitor appears to modestly increase the risk of stent thrombosis compared with triple therapy, but significantly reduces bleeding risk, including the risk of intra-cranial hemorrhage. Improving the available objective measures for assessing ischemic and bleeding risk will be important to choose the optimal regimens for individual patients