ABSTRACT NUMBER: ESOC2026A674 NET CLINICAL BENEFIT OF STARTING ANTICOAGULATION EARLY AFTER ISCHEMIC STROKE IN ATRIAL FIBRILLATION: A META-ANALYSIS OF RANDOMIZED TRIALS FROM THE CATALYST COLLABORATION

Abstract Background and aims Starting direct oral anticoagulants (DOAC) early after atrial fibrillation(AF)-associated ischaemic stroke reduces the composite of subsequent ischaemic or haemorrhagic stroke. However, whether early treatment confers net clinical benefit (NCB) after accounting for the different clinical importance of ischaemic and haemorrhagic events is unclear. Methods Prespecified analysis of CATALYST pooling individual participant data from four trials (ELAN, OPTIMAS, START, TIMING). We included 5,429 participants randomized to start DOAC early (≤4 days; n = 2,683) or later (≥5 days; n = 2,746) after onset of AF-associated ischaemic stroke. We calculated the cerebrovascular NCB (cNCB) by subtracting the weighted rate of excess intracerebral hemorrhage (ICH) attributable to early treatment from the rate of excess ischaemic stroke prevented by early treatment within 30 days using an established event weighting scheme. We estimated event rates from generalized linear mixed-effects models, and quantified uncertainty using non-parametric bootstrapping. Results We observed 115 recurrent ischaemic strokes (45 [1.7%] with early; 70 [2.5%] with later treatment) and 20 ICH (10 [0.4%] in each arm). The cNCB of early over later treatment ranged from +0.94 (95%-CI 0.00 to +2.08) to +0.92 (95%-CI -0.49 to +2.16) weighted cerebrovascular events prevented per 100 participants across ICH weights from 1.5 to 3.3 (number-needed-to-treat≈107). Subgroup analyses and global NCB analyses incorporating additional vascular events yielded consistent results. Conclusions We estimated a modest NCB from starting DOAC early after AF-associated ischaemic stroke. Our findings add support to the early treatment approach, although estimates cannot exclude the possibility of no benefit or small net harm. Conflict of interest The authors have nothing to disclose. The CATALYST collaboration was facilitated by a British Heart Foundation grant for OPTIMAS (grant reference number CS/17/6/33361), with support from researchers at the National Institute for Health and Care Research University College London Hospitals Biomedical Research Centre, and a Swiss National Science Foundation grant for ELAN (32003B_197009; 32003B_169975). Alexandros Polymeris and Konstantina Chalkou contributed equally as first authors. Steven J Warach, Signild Åsberg, David Werring, and Urs Fischer contributed equally as senior authors. Figure 1 - belongs to Results