Gastrointestinal Safety of Antithrombotic Agents after Left Atrial Appendage Occlusion in Atrial Fibrillation Patients

ABSTRACT In the United States, 12 million atrial fibrillation patients (AF) are at a 5-fold increased risk of stroke, and a 2-fold increased risk of mortality. Chronic anticoagulant prescription (warfarin or a direct oral anticoagulant [DOAC]) is the foundation of stroke prevention in this high-risk population. In 90% of AF patients thrombi in theleft atrial appendage (LAA) serves as the nidus of stroke. Mechanical occlusion of the LAA is an alternative forthe 10% of AF patients intolerant of lifelong anticoagulation (i.e., greater than 1 million Americans). After leftatrial appendage occlusion (LAAO) device implantation, patients transition from anticoagulation to antiplateletmonotherapy, over 180 days, while the device undergoes endothelialization to prevent clot formation whenexposed to circulating blood. Following LAAO, gastrointestinal bleeding (GIB) is a frequent adverse event (4%to 6%) with risk closely tied to the prescribed drug regimen. Many post-LAAO drug regimens have beenproposed including dual antithrombotic therapy with aspirin (ASA) and warfarin, dual antiplatelet therapy (ASAand a thienopyridine antiplatelet agent), and monotherapy with DOAC or thienopyridine agent. Some strategies are considered off-label and none have been compared to the post-LAAO drug strategy used in randomized controlled trials (warfarin and ASA) for GIB and stroke risk. So which post-LAAO strategy is safest? LAAO use has increased rapidly over the last few years, yet, little is known about the optimal management of these patients after the procedure. In this study, we will address the comparative safety of post-LAAO antithromboticstrategies among AF patients providing much needed real-world evidence where it is lacking or insufficient tomake safer drug choices. In Aim #1, we will examine GIB risk post-LAAO in a large, geographically diverse population of elderly and non-elderly AF patients who are privately insured, enrolled in Medicare Advantage orMedicare FFS. We will quantify GIB stratifying the cohort by the prescription drug strategy. Incidence rates andpropensity-matched Cox proportional models will be compared across treatments. For Aim #2, we will use thesame sample as Aim #1 to examine the heterogeneity of treatment effects related to age, multiple chronic co-morbidity, renal and hepatic dysfunction, and underlying gastrointestinal lesions. In Aim #3, we use machinel earning techniques to derive and validate an algorithm to identify the safest post-LAAO regimen consideringthe underlying gastrointestinal lesion, the prescribed post-LAAO regimen, and the competing outcomes of GIBand stroke.