Do DOACs and specific blood biomarkers improve risk stratification for ischemic and haemorrhagic events in patients with atrial fibrillation and prior intracerebral haemorrhage?
In patients with AF and prior ICH, DOACs reduced ischemic events compared to no anticoagulation, and high-sensitivity troponin I emerged as a promising biomarker for predicting ischemic stroke risk.
Abstract Background and aims Patients with spontaneous intracerebral haemorrhage (ICH) and atrial fibrillation (AF) have a higher risk of both ischemic and haemorrhagic events, making anticoagulation decisions challenging. Blood biomarkers could serve as objective tools to balance these risks and guide post-ICH antithrombotic strategies. Methods The PRESTIGE-AF study evaluated the recurrence of IS and bleeding events in AF patients taking antiplatelets versus direct oral anticoagulants (DOACs) after ICH. Ischemic and bleeding-related biomarkers including NT-proBNP, high-sensitivity troponin I (TnIH), high-sensitivity C-reactive protein (hsCRP), and GDF-15 were measured and correlated with stroke and bleeding recurrence. Results Prospective study of 124 patients with AF and ICH (65% men). After ICH, 62 patients were not anticoagulated (control group) and 62 received DOACs. There were no differences in baseline clinical characteristics or biomarker concentrations between groups. During follow-up, ischaemic events (TIA/stroke) were more frequent in the control group than in DOACs (18% vs. 13%; p=0.023), with no significant differences in new haemorrhagic events (control 3.2% vs. DOACS 9.7%; p=0.2). Among biomarkers, TnIH showed the highest predictive accuracy for IS (AUC 0.74; p=0.017). GDF-15 showed a trend toward moderate prediction of IS (AUC 0.69; p=0.074). Regarding ICH events, hsCRP showed moderate discrimination for ICH (AUC 0.70; p=0.118) and overall haemorrhagic events (AUC 0.68; p=0.086). Conclusions TnIH emerged as the most promising biomarker for ischemic stroke prediction, while hsCRP may provide information on haemorrhagic risk. The limited number of events reduces precision, and further analyses are required to confirm these findings that might select the best ICH-AF patients to be anticoagulated. Conflict of interest Nothing to disclose
García et al. (2026) studied this question.