Biomarker-Based Risk Prediction With the ABC-AF Scores in Patients With Atrial Fibrillation Not Receiving Oral Anticoagulation

BACKGROUND: The novel ABC (Age, Biomarkers, Clinical History) scores outperform traditional risk scores for stroke, major bleeding, and death in patients with atrial fibrillation (AF) receiving oral anticoagulation. To refine their utility, the ABC-AF scores needed to be validated in patients not receiving oral anticoagulation. METHODS: We measured plasma levels of the ABC biomarkers (N-terminal pro-B-type natriuretic peptide, cardiac troponin-T, and growth-differentiation factor 15) to apply the previously developed ABC-AF scores in patients with AF receiving aspirin (n=3195) or aspirin and clopidogrel (n=1110) in 2 large clinical trials. Calibration was assessed by comparing estimated with observed 1-year risks. Cox regression models were used for recalibration. Discrimination was evaluated separately for the aspirin-only and the overall cohort (n=4305). RESULTS: The ABC-AF-stroke score yielded a c-index of 0.70 (95% CI, 0.67-0.73) in both cohorts. The ABC-AF-bleeding score had a c-index of 0.76 (95% CI, 0.71-0.81) in the aspirin-only cohort and 0.73 (95% CI, 0.69-0.77) overall. Both scores were superior to risk scores recommended by current guidelines. The ABC-AF-death score yielded a c-index of 0.78 (95% CI, 0.76-0.80) overall. Calibrated in patients receiving oral anticoagulation, the ABC-AF-stroke score underestimated and the ABC-AF-bleeding score overestimated the risk of events in both cohorts. These scores were recalibrated for prediction of absolute event rates in the absence of oral anticoagulation. CONCLUSIONS: The biomarker-based ABC-AF scores showed better discrimination than traditional risk scores and were recalibrated for precise risk estimation in patients not receiving oral anticoagulation. They can now provide improved decision support on treatment of an individual patient with AF.