Prognostic associations of major bleeding in anticoagulated patients with atrial fibrillation after TAVI: a post hoc analysis of the ENVISAGE-TAVI AF trial

Background Major bleeding (MB) is a serious complication in anticoagulated patients with atrial fibrillation (AF) after transcatheter aortic valve implantation (TAVI). The prognostic implications of bleeding occurrence and anatomical location in this high-risk population remain insufficiently characterised. Purpose To investigate clinical outcomes after MB in anticoagulated patients with AF after TAVI. Methods ENVISAGE-TAVI AF (NCT02943785) was a randomised trial comparing edoxaban with vitamin K antagonist therapy in patients with AF following successful TAVI. This post hoc analysis included 1377 patients from the trial safety population who received at least one dose of the study drug. Patients were stratified by the occurrence of on-treatment MB and bleeding location (gastrointestinal [GI], intracranial, or non-GI/non-intracranial), as defined by the International Society on Thrombosis and Haemostasis. Outcomes included all-cause death, net adverse clinical events, major adverse cardiovascular events, cardiovascular death, ischaemic stroke/systemic embolic events, intracranial haemorrhage (ICH), and recurrent MB. Results Of 1377 patients, 138 (10.0%) had at least one on-treatment MB event. Patients with MB more frequently presented with a history of recent percutaneous coronary intervention (9.4% vs 4.0%, P = 0.008), abnormal renal function (4.3% vs 1.5%, P = 0.03), higher mean HAS-BLED scores (1.9 vs 1.5, P <0.001), edoxaban assignment (60.9% vs 49.1%, P = 0.009), and hypercholesterolaemia (80.4% vs 68.4%, P = 0.003) at randomisation than those without MB. Fatal outcomes of MB events were most frequent with ICH and uncommon with GI or non-GI/non-ICH bleeding, with no meaningful differences between treatment arms (Figure 1A). Annualised all-cause mortality was markedly higher in patients with vs without on-treatment MB (40.29%/year vs 6.60%/year; hazard ratio [HR], 5.71; 95% confidence interval [CI], 4.00–8.13), with similar higher risk for cardiovascular death (HR, 6.68; 95% CI, 4.18–10.68). Similar patterns were observed with the other clinical outcomes (Figure 2). Within one year of a first on-treatment MB event, recurrent MB was most frequent after GI bleeding, intermediate after non-GI/non-ICH bleeding, and rare after non-fatal ICH, with similar patterns across treatment arms (Figure 1B). Conclusions In anticoagulated patients with AF after TAVI, MB was associated with a worse prognosis. GI bleeding identified patients at high risk for recurrence, while ICH conferred the greatest early mortality risk.