Epistaxis Versus Nonepistaxis Bleeding in Anticoagulated Patients With Atrial Fibrillation: Results From the ENGAGE AF-TIMI 48 Trial

Abstract

Background: Epistaxis is common with antithrombotic therapy and is often troublesome to patients, yet its frequency, severity, and outcomes are poorly characterized.

Methods and results: Effective Anticoagulation with Factor Xa Next Generation in Atrial Fibrillation-Thrombolysis in Myocardial Infarction 48 (ENGAGE AF-TIMI 48) randomized 21 105 patients with atrial fibrillation and CHADS2 risk score ≥2 to higher-dose edoxaban regimen (60 mg daily, dose-reduced to 30 mg), lower-dose edoxaban regimen (30 mg, dose reduced to 15 mg, daily), or warfarin. Bleeds were adjudicated using International Society on Thrombosis and Haemostasis criteria. Patients with intracranial hemorrhage during follow-up were excluded; those with >1 bleeding event were categorized according to their most severe event. The safety cohort with interval censoring during drug interruption was analyzed. Proportions were compared using Pearson's chi-square test and treatment arms were compared using a Cox proportional hazards model. Among 5247 patients with a bleeding event, 1008 (19.2%) had epistaxis and 4239 (80.8%) had nonepistaxis bleeding. Epistaxis events were less severe than nonepistaxis bleeds (International Society on Thrombosis and Haemostasis major: 3.2% versus 20.7%; clinically relevant nonmajor: 64.7% versus 60.1%; minor: 32.1% versus 19.2%; P<0.001). Permanent drug discontinuation was similar following epistaxis versus nonepistaxis bleeding in patients with major (59.4% versus 53.6%; P=0.52) or clinically relevant nonmajor (32.5% versus 33.3%; P=0.70) bleeding but was significantly higher in patients with minor epistaxis versus other minor bleeds (33.3% versus 23.9%; P=0.001). Compared with warfarin, higher-dose edoxaban regimen had similar risk of epistaxis (hazard ratio [HR], 1.09 [95% CI, 0.95-1.26]), whereas lower-dose edoxaban regimen conferred reduced risk (HR, 0.73 [95% CI, 0.62-0.86]).

Conclusions: Epistaxis was frequent, and despite being overall less severe than nonepistaxis bleeding, was associated with similar rates of anticoagulant discontinuation. Compared with warfarin, lower-dose edoxaban regimen reduced the risk of epistaxis by 27% whereas higher-dose edoxaban regimen had no effect.

Registration: URL: https://clinicaltrials.gov; Unique Identifier: NCT00781391.

Keywords: anticoagulation; atrial fibrillation; bleeding; discontinuation; epistaxis.

Figure 1. Highest bleed severity for patients with epistaxis vs nonepistaxis bleeding events.

Patients with any epistaxis events were assigned to the epistaxis group, and patients with exclusively nonepistaxis bleeds were assigned to the nonepistaxis group. Patients were categorized according to their most severe event. All reported bleeding events were adjudicated by the independent Clinical Event Committee, and bleeding severity was determined according to modified ISTH criteria. Cochrane‐Armitage test for trend (P<0.001) indicates significantly less severe bleeding with epistaxis as compared with nonepistaxis bleeding events. CRNM, clinically relevant nonmajor; and ISTH, International Society on Thrombosis and Haemostasis.

Figure 2. Rates of permanent study drug discontinuation after epistaxis and nonepistaxis bleeding events.

The percent of patients who permanently discontinued use of the study drug following epistaxis or nonepistaxis bleeding events were compared within each bleeding severity category, and for all bleeds, using Pearson's chi‐square test. All reported bleeding events were adjudicated by the independent Clinical Events Committee, and bleeding severity was determined according to modified International Society on Thrombosis and Haemostasis criteria. Rates differed significantly only for patients with minor bleeds, with a higher discontinuation rate following epistaxis compared with nonepistaxis events. CRNM, clinically relevant nonmajor.

Figure 3. Effect of lower‐dose and higher‐dose edoxaban regimens on epistaxis events relative to warfarin.

A total of 20 792 patients with AF and CHADS₂ ≥2 were randomized to higher‐dose edoxaban regimen (HDER; 60 mg daily or dose‐reduced to 30 md daily), lower‐dose edoxaban regimen (LDER; 30 mg daily or dose‐reduced to 15 mg daily), or warfarin, received at least 1 dose of study drug, and did not experience an ICH during the trial. All reported bleeding events were CEC adjudicated and bleeding severity was determined according to modified ISTH criteria. Annualized rates of epistaxis were calculated for the 3 study arms and hazard ratios with 95% CIs were analyzed for LDER and HDER relative to warfarin. Epistaxis rates were similar between HDER and warfarin, whereas compared with warfarin, LDER reduced the rate of major or clinically relevant nonmajor (CRNM) bleeds combined, and all epistaxis bleeds overall. AF indicates atrial fibrillation; CEC, clinical end point committee; CHADS₂, 1 point each for history of congestive heart failure, hypertension, age ≥75, diabetes, and 2 points for prior stroke or transient ischemic attack; CRNM, clinically relevant non‐major; HDER, higher‐dose edoxaban regimen (60 mg, dose‐reduced to 30 mg, once daily); HR, hazard ratio; ISTH, International Society on Thrombosis and Haemostasis; and LDER, lower‐dose edoxaban regimen (30 mg, dose‐reduced to 15 mg, once daily).