Is there a role for half-dose anticoagulation in atrial fibrillation? Randomized comparison of 2 doses of edoxaban in 14,014 patients in ENGAGE AF-TIMI 48

Abstract

Background: In the phase 3 RCT ENGAGE AF‐TIMI 48, both the higher‐dose (HDER/edoxaban 60/30mg) and lower‐dose edoxaban regimen (LDER/edoxaban 30/15mg) were non‐inferior to well‐managed warfarin for the prevention of stroke or systemic embolic events in patients with atrial fibrillation (AF). While LDER was associated with a 41% relative increase in ischemic stroke as compared to warfarin, LDER showed a larger relative reduction in bleeding events than HDER. An interaction tree analysis was performed to identify subgroups of patients likely to derive greater benefit from LDER vs. HDER based on a serious net outcome (SNO) that combines the most serious stroke and bleeding events: disabling/fatal ischemic stroke or life‐threatening/fatal bleeding. Methods and results: LDER (N=7002) had a higher risk for disabling/fatal ischemic stroke (HR 1.37, 95% CI 1.06‐1.77; p=0.017), but fewer lifethreatening/ fatal bleeds (0.61 (0.44, 0.84), p=0.002) than HDER (N=7012). The SNO was similar with LDER vs HDER in the total population (0.95 (0.78‐1.16), p=0.61). In the 30% of patients who had either malignancy or paroxysmal AF (Figure) the SNO was significantly lower with LDER vs HDER (HR 0.59 (0.40‐0.86)), driven by large reductions in bleeding without an increase in ischemic stroke in this subgroup (figure). Meanwhile the remaining 70% of patients with neither malignancy nor paroxysmal AF tended to have a higher risk with LDER vs. HDER (HR 1.15 (0.91‐1.46), P‐interaction 0.003). Conclusions: In our current post‐hoc analysis we were able to demonstrate that the combined rate of serious stroke and bleeding events was similar in patients treated with LDER and HDER. Moreover, patients with a malignancy as well as those with paroxysmal AF derived a greater net benefit from LDER vs. HDER regarding serious AF‐/anticoagulationrelated outcome events. These results need to be viewed as hypothesisgenerating which may, if confirmed, aid physicians in individualizing edoxaban dosing. (Figure Presented).