Reduced stroke risk without increased bleeding risk in patients with atrial fibrillation and complicated liver cirrhosis treated with oral anticoagulation

Abstract Introduction Patients with atrial fibrillation (AF) have an increased risk of thromboembolic events (TE), while patients with complicated liver cirrhosis have an increased risk of both TE and bleeding. Oral anticoagulation reduces the risk of TE in the general group of patients with AF but its use in patients with liver cirrhosis is obscured by their imbalance between endogenous procoagulants and anticoagulants, as well as the lack of data from randomized controlled trials. Purpose To examine the risks of TE and bleeding in patients with AF and complicated liver cirrhosis according to whether oral anticoagulation is initiated. Methods We conducted a nationwide registry-based study of anticoagulant-naive patients with complicated liver cirrhosis and first-time AF diagnosed between 2010–2017. Complicated liver cirrhosis was defined as liver cirrhosis plus one of the following: alcoholism, esophageal varices, ascites or hepatorenal syndrome. Patients were followed for a maximum of 5 years. TE was defined as a composite of ischemic stroke, transient ischemic attack or systemic thromboembolism; and the bleeding endpoint was defined as gastrointestinal, cerebral or urogenital bleeding requiring hospitalization, or any hospital contact with epistaxis. TE risk was estimated by use of the CHA2DS2-VASc score, while bleeding risk was estimated by use of the HAS-BLED score. Outcomes were stratified according to whether an oral anticoagulant (vitamin K antagonists [VKA] or direct oral anticoagulants [DOAC]) was initiated. Results We identified 770 patients with complicated liver cirrhosis and first-time AF. TE events occurred in 7.0% (n=25/359) of patients with a CHA2DS2-VASc score ≤2 versus 20.7% (n=85/411) of patients with a CHA2DS2-VASc score >2. Among 411 patients with a high CHA2DS2-VASc score, 111 (27.0%) were prescribed an oral anticoagulant (OAC+; VKA, n=53 [47.7%], DOAC, n=58 [52.3%]), while 300 (73.0%) were not treated with oral anticoagulation (OAC−). These two groups had comparable baseline data, including HAS-BLED (OAC+ 3.0 [2.5–4.0] versus OAC− 3.0 [2.0–4.0]) and CHA2DS2-VASc (OAC+ 4.0 [3.0–5.0] versus OAC− 4.0 [3.0–5.0]) scores. The 5-year TE risk among patients receiving anticoagulant therapy was 14.4% (n=16/111) versus 23.0% (n=69/300) in patients not treated with anticoagulant therapy (hazard ratio (HR) 0.55 [0.32–0.95]). The difference in bleeding risk was insignificant between the two groups (HR 0.67 [0.35–1.30]). Adjusting for CHA2DS2-VASc