Antithrombotic Treatment in Cryptogenic Stroke Patients With Patent Foramen Ovale

Background and Purpose— It is unclear whether treatment with anticoagulants or antiplatelets is the optimal strategy in patients with stroke or transient ischemic attack of undetermined cause and patent foramen ovale that is not percutaneously closed. We aimed to perform a systematic review and meta-analysis of randomized controlled trials to compare anticoagulant or antiplatelet treatment in this population. Methods— We searched PubMed until July 16, 2019 for trials comparing anticoagulants and antiplatelet treatment in patients with stroke/transient ischemic attack and medically treated patent foramen ovale using the terms: “cryptogenic or embolic stroke of undetermined source” and “stroke or cerebrovascular accident or transient ischemic attack” and “patent foramen ovale or patent foramen ovale or paradoxical embolism” and “trial or study” and “antithrombotic or anticoagulant or antiplatelet.” The outcomes assessed were stroke recurrence, major bleeding, and the composite end point of stroke recurrence or major bleeding. We used 3 random-effects models: (1) a reference model based on the inverse variance method with the Sidik and Jonkman heterogeneity estimator; (2) a strict model, implementing the Hartung and Knapp method; and (3) a commonly used Bayesian model with a prior that assumes moderate to large between-study variance. Results— Among 112 articles identified in the literature search, 5 randomized controlled trials were included in the meta-analysis (1720 patients, mean follow-up 2.3±0.5 years). Stroke recurrence occurred at a rate of 1.73 per 100 patient-years in anticoagulant-assigned patients and 2.39 in antiplatelet-assigned patients (hazard ratio, 0.68; 95% CI, 0.32–1.48 for the Sidik and Jonkman estimator). Major bleeding occurred at a rate of 1.16 per 100 patient-years in anticoagulant-assigned patients and 0.68 in antiplatelet-assigned patients (hazard ratio, 1.61; 95% CI, 0.72–3.59 for the Sidik and Jonkman estimator). The composite outcome occurred in 52 anticoagulant-assigned and 54 antiplatelet-assigned patients (odds ratio, 1.05; 95% CI, 0.65–1.70 for the Sidik and Jonkman estimator). Conclusions— We cannot exclude a large reduction of stroke recurrence in anticoagulant-assigned patients compared with antiplatelet-assigned, without significant differences in major bleeding. An adequately powered randomized controlled trial of a non–vitamin K antagonist versus aspirin is warranted.