Safety and effectiveness of edoxaban in Japanese patients with nonvalvular atrial fibrillation: Final report of a two‐year postmarketing surveillance study (ETNA‐AF‐Japan)

Abstract Background Direct oral anticoagulants (DOACs) are the recommended first‐line therapy for ischemic stroke prevention in patients with nonvalvular atrial fibrillation (NVAF). However, the safety and effectiveness of edoxaban for this indication requires monitoring over the long term in real‐world settings. Methods ETNA‐AF‐Japan (trial no. UMIN000017011) was a prospective, multicenter observational study (part of postmarketing surveillance in Japan). NVAF patients due to receive edoxaban to prevent ischemic stroke were enrolled between 13 April 2015 and 30 September 2017. Results A total of 11 569 patients were enrolled. For the 11 111 patients (female, 40.6%) whose data comprised the safety analysis set, age, body weight, creatinine clearance (CLcr), and CHADS2 score were 74.2 ± 10.0 years, 60.0 ± 12.7 kg, 63.9 ± 25.8 mL/min, and 2.2 ± 1.3, respectively (mean ± SD). The majority (86.3%) received edoxaban in accordance with package insert information. The mean duration of treatment was 561.9 ± 261.2 days. The annual incidence (95% confidence interval) of all bleeding events and major bleeding events was 5.60% (5.25%–5.98%) and 1.02% (0.88%–1.18%), respectively. The annual incidence of ischemic stroke (excluding transient ischemic attack, TIA) or systemic embolism was 1.08% (0.93%–1.25%). Multivariate analysis showed low body weight, low CLcr, history of gastrointestinal bleeding, anemia, and use of an antiplatelet agent to be associated with major bleeding, and history of ischemic stroke or TIA, vascular disease, and antiplatelet agent use to be associated with ischemic stroke (excluding TIA) or systemic embolism. Conclusions These results provide real‐world evidence for the long‐term good safety and effectiveness profile of edoxaban in Japanese NVAF patients under clinical practice.


| INTRODUC TI ON
The results of randomized clinical trials have shown the advantages of direct oral anticoagulants (DOACs) over warfarin for preventing stroke in patients with atrial fibrillation (AF). [1][2][3][4] Consequently, the Japanese Circulation Society guidelines, 5 the Japanese Circulation Society-Japanese Heart Rhythm Society joint guidelines, 6 and the European Heart Rhythm Association guidelines 7 now recommend DOACs in preference to warfarin as first-line anticoagulant therapy for ischemic stroke prevention in patients with nonvalvular atrial fibrillation (NVAF). However, the guidelines state that clinical experience of treatment with these newer drugs needs to be recorded and their performance monitored over the long term to ensure that they are acceptable in terms of the balance between risk and benefit; their safety profile is a particular concern, because AF patients commonly have risk factors such as concomitant diseases.
The DOAC edoxaban, taken once daily, directly and reversibly inhibits factor Xa. It is indicated for the prevention of ischemic stroke and systemic embolism in NVAF patients. [8][9][10][11] Two formulations are available: tablet and orally disintegrating. The latter is particularly useful for elderly patients who might otherwise have difficulty swallowing the tablets.
Confirmation of the efficacy and acceptable safety profile of edoxaban has been provided by the results of a pivotal confirmatory phase 3 study: ENGAGE AF-TIMI 48 (Effective aNticoaGulation with factor xA next GEneration in Atrial Fibrillation-Thrombolysis In Myocardial Infarction 48). 4 The inclusion and exclusion criteria for this randomized controlled trial were strict, therefore the study population did not include all potential beneficiaries of the treatment. Therefore, an investigation of the safety and effectiveness of long-term edoxaban use in a real-world clinical setting was necessary. ETNA-AF-Japan (Edoxaban Treatment in routiNe clinical prActice in patients with nonvalvular Atrial Fibrillation; trial no. UMIN000017011) was initiated to evaluate the occurrence of ischemic stroke/systemic embolism and bleeding episodes in ≥10 000 patients over 2 years. The findings of ETNA-AF-Japan will help determine how best to treat patients, particularly elderly patients, with AF. The findings are expected to be especially useful in Japan, where more than 700 000 people are estimated to have AF and over a quarter of the population is ≥65 years old (according to the 2015 census). 12,13 We reported the 3-month 14 and 1-year 15 interim analysis results from ETNA-AF-Japan in 2019 and 2020, respectively. These articles included data on patient demographics, clinical characteristics, dosing status, and edoxaban's safety and effectiveness. Here, we report the final results of ETNA-AF-Japan, including the results of analyses conducted to identify factors associated with the drug's safety and effectiveness.

| Study design
ETNA-AF-Japan is a real-world, prospective, multicenter observational study and part of postmarketing surveillance. Our aim in carrying out this study was to evaluate edoxaban's safety and effectiveness in Japanese NVAF patients by analyzing baseline and clinical data collected over 2 years. ETNA-AF-Japan was in accordance with the Good Post-marketing Study Practice of the Ministry of Health, Labor, and Welfare of Japan (MHLW). Methodological details are available in the report of the results of interim analyses based on 3-month data.14

| Patient population
The target number of study participants was 10 000. Adult NVAF patients for whom edoxaban had been prescribed for prevention of ischemic stroke and systemic embolism were eligible, on the condition that they had not received the drug before. Additional eligibility criteria were the patients' ability to start treatment during enrollment, their availability for long-term follow-up, and their agreement to provide, at the time of registration, written informed consent to participate.

| Study variables
The patients' baseline and clinical data, including information on medication status (for edoxaban and concomitant drugs), medical history and comorbidities, and details of nonpharmacological treatments for AF or surgical procedures, were recorded in case report forms by their physicians. The baseline data were used to calculate HAS-BLED, 16 CHADS 2 , 17 and CHA 2 DS 2 -VASc 18 scores. The HAS-BLED scores were based on a number of risk factors (namely hypertension, abnormal renal or liver function, stroke, bleeding, elderly status, and drug use, but not international normalized ratio or alcohol use).
The following events were recorded: death, stroke other than transient ischemic attack (TIA), systemic embolism, myocardial infarction, and adverse events (AEs), including bleeding events.

| Initial daily dose and administration
As specified on the Japanese package insert for edoxaban, the recommended oral daily dose is 60 and 30 mg for adult patients with body weight >60 kg and ≤60 kg, respectively. 11 The lower dose is also recommended for patients with either renal dysfunction (creatinine clearance, CLcr, ≤50 mL/min) or concomitant use of a P-glycoprotein inhibitor (erythromycin, cyclosporine, quinidine, or verapamil), or both.

| Study outcomes
The study outcomes were AEs, including bleeding events, for evaluation of safety, and clinical events, including death, ischemic stroke (excluding TIA), systemic embolism, and myocardial infarction, for evaluation of effectiveness. Physicians categorized bleeding as major bleeding, clinically relevant nonmajor bleeding (CRNMB), or minor bleeding, according to definitions in the report of the ENGAGE AF-TIMI 48 study, with minor modifications. 4,15

| Statistical analysis
To estimate the cumulative incidence of bleeding events from the start of treatment to the date of the last dose, the Kaplan-Meier method was used. To determine factors associated with bleeding events during this period, variables identified as significant (P < .05) in a univariate analysis (Cox proportional hazards model) were used as candidate variables in the forward-backward stepwise method, and selected factors were included in the subsequent multivariate analysis using the same model. Age was compulsorily included in the multivariate analysis. For hypothesis testing, the significance level was set at 5%, and the confidence intervals for both sides were set at 95%. Missing data were not computed. For all analyses, SAS System Release 9.2 (SAS Institute Inc, Cary, NC, USA) was used.

| Patient characteristics at baseline and patient disposition
A total of 11 569 patients were enrolled in 1367 Japanese centers in the period from 13 April 2015 to 30 September 2017. Of the patents whose case reports were collected, data from 358 were excluded for the reasons of serious deviation for the protocol, safety evaluation not carried out, withdrawal of consent, and edoxaban prescribed at an initial dose of 15 mg/day; consequently, data from 11 111 patients were included in the safety analysis. After the exclusion of 48 patients for the reason of off-label use, data from 11 063 patients were included in the effectiveness analysis ( Figure 1).
Patient characteristics at baseline are summarized in Table 1.
Of the 11 111 patients whose data comprised the safety analysis set, mean age, body weight, CLcr, and CHADS 2 score were 74.2 ± 10.0 years, 60.0 ± 12.7 kg, 63.9 ± 25.8 mL/min, and  Edoxaban treatment status is summarized in Table 2. At 2 years after treatment started, 7753 patients were continuing to receive edoxaban, and 3358 have completed or discontinued the drug; thus, the treatment continuation rate at 2 years was 69.8% (7753/11 111).

| Medication status
Regarding adherence, 94.6% of patients (10 513/11 111) took their medication every day. Mean ± SD and median treatment duration were 561.9 ± 261.2 days and 719.0 days, respectively. The most common reason for completing or discontinuing treatment was patients' failure to visit the hospital or their transfer to another hospital (11.7%), followed by the occurrence of AEs or clinical events (9.2%) and treatment completed as planned (4.7%).

| Safety analysis
Regarding bleeding events, their cumulative incidence increased over 2 years ( Figure 3A). Annual incidence of all bleeding events, major bleeding, CRNMB, and minor bleeding was 5.60%, 1.02%, 2.48%, and 2.27%, respectively ( Table 3; full details of the 183 major bleeding events recorded for 173 patients are available in Table S2).
The annual incidence of all bleeding was 6.00% and 5.76% in the recommended standard dose (60 mg/day) group and the recommended reduced dose (30 mg/day) group, respectively. The annual incidence of major bleeding was 0.75% and 1.21% in the recommended standard dose (60 mg/day) group and recommended reduced dose (30 mg/day) group, respectively.
Multivariate analysis identified bodyweight, CLcr, prior gastrointestinal bleeding, anemia, and antiplatelet agent use as risk factors for major bleeding (P < .05) ( Table 4). Table 5 shows edoxaban's safety and effectiveness when used at recommended doses in patients with specific backgrounds, for example, those who were not included in sufficient numbers in pivotal clinical trials. Compared with the total study population, the incidence of both all bleeding and major bleeding showed a tendency to be higher in each of the patient subgroups. Incidence of all bleeding showed a tendency to be higher in patients with prior bleeding (21.33%) and in older patients (≥80 years) with prior bleeding (24.32%) than in the total study population (8.60%). Incidence of major bleeding showed a tendency to be higher in older patients (≥80 years) with prior bleeding than in the total population (7.43% vs 1.64%).

| Effectiveness analysis
Regarding clinical events, their cumulative incidence increased over 2 years ( Figure 3B). The annual incidence of ischemic stroke (excluding TIA) or systemic embolism, stroke (excluding TIA) or systemic embolism, major cardiovascular events, death, and myocardial infarction was 1.08%, 1.44%, 1.83%, 1.28%, and 0.12%, respectively, and the annual incidence of all events was 2.72% ( Table 3). The annual incidence of ischemic stroke (excluding TIA) or systemic embolism recurrence was 1.03% in the group who received the recommended standard dose and 1.20% in those who received the recommended reduced dose.
Multivariate analysis identified history of ischemic stroke or TIA, vascular disease, and antiplatelet agent use as risk factors for ischemic stroke (excluding TIA) or systemic embolism (P < .05) ( Table 4).
In analyses by patient characteristics, incidence of ischemic stroke (excluding TIA) or systemic embolism showed a tendency to be higher in patients with low body weight (<40 kg) and low CLcr (<30 mL/ min) (4.55%), and in low-bodyweight patients and older patients (≥80 years) (3.47%), than in the total population (1.75%) ( Table 5).

| D ISCUSS I ON
Previously, we reported the results of an interim analysis of 1-year data from the ETNA-AF-Japan study, which were reassuring regarding the safety and effectiveness of edoxaban. 15 Here, we report the results of our final analyses at the completion of the 2-year study period, to show the safety and effectiveness of edoxaban under the real-world conditions of clinical practice.
In Japanese patients with factors associated with high risk for bleeding, such as low body weight, edoxaban is generally administered at the low dose of 30 mg/day, in accordance with the dose reduction criteria. Most patients in this study received edoxaban at 30 mg/day. Therefore, we consider information concerning the lower dose to be particularly important. In the previous 3-month and 1-year interim analyses, we evaluated data from patients classified into two groups according to edoxaban dose: 30 and 60 mg/day. 14,15 However, the former included under-dose patients (ie, those who received the lower dose although they did not meet the dose reduction criteria) and the latter included over-dose patients (ie those who received the higher dose although they did not meet the dose reduction criteria). In this study, we evaluated data from patients classified into over-dose, recommended standard dose (60 mg), recommended reduced dose (30 mg), and under-dose groups.
The present analysis included data from many patients with factors associated with high risk for bleeding, including older age, low body weight, and renal dysfunction. Because the dosing criteria for edoxaban specify body weight ≤60 kg and renal dysfunction as factors warranting dose reduction, there was a greater proportion of patients with the high risk of bleeding in the recommended reduced dose (30 mg) group than in the recommended standard dose (60 mg) group (Table 1)  A possible explanation is the higher bleeding risk in the reduced dose group as opposed to the standard dose group. Similar findings were also reported for apixaban, dabigatran, and rivaroxaban. 20 -23 Furthermore, the incidence of major bleeding in the recommended reduced dose was no higher than in ENGAGE AF-TIMI 48  17,18 Lower body weight has previously been reported as a factor associated with major bleeding. 23 However, no significant difference was found    Abbreviations: CI, confidence interval; CRNMB, clinically relevant non-major bleeding. a The safety analysis set minus data from the 230 patients whose dose adjustment factors were unknown. b The effectiveness analysis set minus data from the 227 patients whose dose adjustment factors were unknown.  Table S1 for the definition of liver dysfunction used in the present study. b Myocardial infarction, internal carotid artery stenosis, or arteriosclerosis obliterans.

TA B L E 4 (Continued)
between patients with body weight <40 kg and those with ≥60 kg.
Antiplatelet agent use may have been identified as one of the risk factors for ischemic stroke (excluding TIA) or systemic embolism because these drugs were likely to be prescribed to patients with complications and at high risk of thromboembolism.
In this study, we also evaluated edoxaban's safety and effectiveness in patients with high-risk factors that are considered particularly important (ie age, body weight, renal function, and history of bleeding). Although these patients may not have been sufficiently included in pivotal clinical trials, physicians often encounter such patients in clinical practice, therefore the information is important. We found that the incidence of bleeding events was particularly high in patients aged ≥80 years and with a history of bleeding, and the incidence of ischemic stroke (excluding TIA) or systemic embolism was high in those with body weight <40 kg and CLcr <30 mL/min (Table 5). Considering the results of the multivariate analysis and edoxaban's safety and effectiveness profile in patient subgroups, anticoagulation should prompt careful monitoring to detect bleeding in patients thought to be at particularly high risk of this AE.

| Limitations
This study has limitations. First, the nature of the study (an open-label observational study) meant that it had no comparative arm. Therefore, a comparison of the safety and effectiveness of edoxaban with those of other DOACs or a vitamin K anticoagulant (eg warfarin) is not possible. Second, the results are based on data collected from case report forms completed by physicians, and data were missing from the forms of patients who failed to attend follow-up visits. Consequently, event rates may be underestimated due to loss to follow-up. Third, the analyses by doses were conducted based on dosing status at the treatment start, and changes to the doses and dose adjustment factors during the 2-year follow-up period were not considered.

| CON CLUS IONS
The results obtained from this large-scale study (>11 000 patients) confirm the good safety and effectiveness profile of edoxaban used over the long term (2 years) by Japanese patients with NVAF. No major concerns arose regarding the safety and effectiveness of the recommended reduced dose (30 mg), which patients with high bleeding risk are most likely to be prescribed.

D I SCLOS U R E
ETNA-AF-Japan was carried out in accordance with the Good Post-

DATA S H A R I N G A N D DATA ACCE SS I B I LIT Y
Data from individual identified participants will not be shared. This includes information held in data dictionaries.