Heterogeneity of Treatment Effect of DOACs Vs Warfarin in Atrial Fibrillation

BACKGROUND:Direct oral anticoagulants (DOACs) are recommended over vitamin K antagonists (VKAs) for stroke prevention in atrial fibrillation (AF). However, limited accessibility in resource-constrained settings makes it important to identify patient subgroups in whom VKAs may achieve comparable net clinical outcomes (NCOs). OBJECTIVES:The purpose of this study was to identify AF patient subgroups and explore potential predictors of differential treatment benefit to inform anticoagulant selection where DOAC access is limited. METHODS:Individual patient data from the COMBINE-AF data set pooling 4 pivotal randomized trials of DOACs vs VKAs in AF were analyzed. Model-based clustering identified patient subgroups. Heterogeneity of treatment effects for the NCO-a composite of all-cause death, disabling or fatal stroke, and intracranial or fatal bleeding- was explored using Cox regression and gradient boosting analyses adapted for time-to-event data. RESULTS:A total of 58,634 patients were included (29,272 warfarin; 29,362 standard dose DOACs). Two subgroups were identified: a high-risk cluster (mean creatinine clearance 55.1 mL/min, age 75.6 years, body mass index [BMI] 26.6 kg/m2) and a low-risk cluster (mean creatinine clearance 101 mL/min, age 64.2 years, BMI 32.3 kg/m2). DOACs significantly reduced NCOs in the high-risk cluster (HR: 0.86; 95% CI: 0.81-0.92; P < 0.001), but not in the low-risk cluster (HR: 0.93; 95% CI: 0.85-1.03; P = 0.154, p-interaction = 0.115). Primary predictors of treatment heterogeneity were renal function, age, and BMI; no statistically significant treatment-by-subgroup interactions were identified. CONCLUSIONS:Renal function, age, and BMI are key determinants of anticoagulant treatment benefit in AF. In regions with limited access to DOACs, these findings suggest that some lower-risk patient profiles may derive similar outcomes with VKAs, although differences were not statistically significant.