Drug interactions with apixaban: A systematic review of the literature and an analysis of VigiBase, the World Health Organization database of spontaneous safety reports

Abstract

Apixaban, a direct oral anticoagulant, has emerged over the past few years because it is considered to have a low risk of drug‐drug interactions compared to vitamin K antagonists. To better characterize these interactions, we systematically reviewed studies evaluating the drug‐drug interactions involving apixaban and analyzed the drug‐drug interactions resulting in an adverse drug reaction reported in case reports and VigiBase. We systematically searched Medline, Embase, and Google Scholar up to 20 August 2018 for articles that investigated the occurrence of an adverse drug reaction due to a potential drug interacting with apixaban. Data from VigiBase came from case reports retrieved up to the 2 January 2018, where identification of potential interactions is performed in terms of two drugs, one adverse drug reaction triplet and potential signal detection using Omega, a three‐way measure of disproportionality. We identified 15 studies and 10 case reports. Studies showed significant variations in the area under the curve for apixaban and case reports highlighted an increased risk of hemorrhage or thromboembolic events due to a drug‐drug interaction. From VigiBase, a total of 1617 two drugs and one adverse drug reaction triplet were analyzed. The most reported triplet were apixaban—aspirin—gastrointestinal hemorrhage. Sixty‐seven percent of the drug‐drug interactions reported in VigiBase were not described or understood. In the remaining 34%, the majority were pharmacodynamic drug‐drug interactions. These data suggest that apixaban has significant potential for drug‐drug interactions, either with CYP3A/P‐gp modulators or with drugs that may impair hemostasis. The most described adverse drug reactions were adverse drug reactions related to hemorrhage or thrombosis, mostly through pharmacodynamic interactions. Pharmacokinetic drug‐drug interactions seem to be poorly detected.