OBJECTIVES: To evaluate further evidence on the clinical profile of new oral anticoagulants (NOACs) for patients with atrial fibrillation (AF) and percutaneous coronary intervention (PCI).

METHODS: A systematic review with network meta-analysis (NMA) was performed (CRD42019146813). Searches were conducted in PubMed and Scopus. Randomized controlled trials (RCT) evaluating anticoagulant regimens compared head-to-head or against placebo were included. Networks were built for each outcome of interest (i.e. major or clinically relevant non-major (ISTH), major bleeding (TIMI), major or minor bleeding (TIMI), stent thrombosis, stroke, myocardial infarction, death). Results were reported as odds ratio with 95% credibility intervals. Surface under the cumulative ranking curve analyses (SUCRA) were calculated. A multicriteria decision analysis (stochastic multicriteria acceptability analysis - SMAA) was used to quantitatively estimate the benefit-risk of the therapies. Different scenarios were built considering the combination of one or more benefit (efficacy) or risk (safety) criteria.

RESULTS: Five RCTs were included (n=11532; WOEST, PIONEER AF-PCI, RE-DUAL PCI, AUGUSTUS and ENTRUST AF-PCI). No significant statistical differences were observed among the therapeutic alternatives (vitamin K antagonist + P2Y12 inhibitor, vitamin K antagonist + P2Y12 inhibitor + aspirin, apixaban 5mg + P2Y12 inhibitor, dabigatran 110mg + P2Y12 inhibitor, dabigatran 150mg + P2Y12 inhibitor, rivaroxaban 2.5mg + P2Y12 inhibitor + aspirin, rivaroxaban 15mg + P2Y12 inhibitor, edoxaban 60mg + P2Y12 inhibitor) for all outcomes, including bleeding, stroke and death. SUCRA and SMAA demonstrated that the association edoxaban 60mg + P2Y12 inhibitor was the worst option (28% chances), leading to more major bleeding, while apixaban 5mg + P2Y12 inhibitor seemed to be the safest alternative (63%).

CONCLUSIONS: There is a lack of evidence on the clinical superiority of different NOACs regimens for AF and PCI, although apixaban 5mg + P2Y12 inhibitor slightly stands out. Further well-designed RCTs with larger sample sizes are required to strengthen these conclusions.