A novel risk prediction score in atrial fibrillation for a net clinical outcome from the ENGAGE AF-TIMI 48 randomized clinical trial

Abstract

Aims The choice between initiating a non-vitamin K antagonist oral anticoagulant (NOAC) and a vitamin K antagonist (VKA) in patients with atrial fibrillation (AF) may be challenging. To assist in this decision, we developed a risk score to identify patients for whom a therapeutic benefit of NOACs over VKA is predicted. Methods and results ENGAGE AF-TIMI 48 was a randomized clinical trial of edoxaban vs. warfarin in 21 105 patients with AF. Cox proportional hazard models identified factors associated with a serious net clinical outcome (NCO) of disabling stroke, life-threatening bleeding, and all-cause mortality in VKA naïve patients from the warfarin arm. These were used to develop an integer risk score. Performance was assessed by C-indices and validation by bootstrapping. Kaplan-Meier analyses were stratified by three score categories and treatment arm. Over a median of 2.7 years, 457 NCO events occurred in 2898 patients with a total person-time of 7549.5 years (6.05%/year). The risk prediction model (C = 0.693) for the NCO was translated into a 17-point integer score, with annualized event rates for the low, intermediate, and high-risk categories in the warfarin arm of 3.5%, 9.9%, and 20.8%, respectively. Therapeutic benefit of higher- and lower-dose edoxaban over warfarin was demonstrated in the high- and intermediate-risk, with equal benefit in the low-risk categories (P-interaction 0.008 and 0.014, respectively). Conclusion In VKA naive patients with AF, the TIMI-AF score can assist in the prediction of a poor composite outcome and guide selection of anticoagulant therapy by identifying a differential clinical benefit with a NOAC or VKA.