One‐year clinical outcomes of anticoagulation therapy among Japanese patients with atrial fibrillation: The Hyogo AF Network (HAF‐NET) Registry

Abstract Background Although anticoagulation therapy could reduce the risk of strokes in patients with atrial fibrillation (AF), large‐scale investigations in the direct oral anticoagulant (DOAC) and AF catheter ablation (CA) era are lacking. Methods This study was designed as a prospective, multicenter, observational study and a total of 2113 patients from 22 institutions were enrolled in the Hyogo area. Results The mean age and CHADS2 score were 70.1 ± 10.8 years old and 1.5 ± 1.1, respectively. The follow‐up period was 355 ± 43 days. CA was performed in 614 (29%) and DOACs were prescribed in 1118 (53%) patients. Ischemic strokes/systemic embolisms (SEs) and major bleeding occurred in 13 (0.6%) and 17 (0.8%) patients, respectively. New onset dementia, hospitalizations for cardiac events, and all‐cause death occurred in eight (0.4%), 60 (2.8%), and 29 (1.4%) patients, respectively. A multivariate analysis demonstrated that persistent AF and the body weight (BW) were associated with ischemic strokes/SEs and major bleeding, respectively (persistent AF: hazard ratio, 9.57; 95%CI, 1.2‐74.0; P = .03; BW: hazard ratio, 0.94; 95%CI, 0.90‐0.99; P = .02). AFCA history was associated with the cardiac events (hazard ratio, 0.44; 95%CI, 0.20‐0.99; P = .04). Age was associated with new onset dementia (hazard ratio, 1.1; 95%CI, 1.0‐1.2; P = .03). Conclusions In the DOAC and CA era, the incidence of ischemic strokes/SEs, major bleeding and cardiac events could be dramatically reduced in patients with AF. However, some unsolved issues of AF management still remain especially in elderly patients with persistent AF and a low BW.


| INTRODUC TI ON
The number of patients with atrial fibrillation (AF) is increasing at a rapid rate and is expected to reach beyond one million patients in Japan as the population ages. 1 AF has a major risk of thromboembolisms and heart failure. Several studies reported that AF is also related to new onset dementia. 2 The annual incidence of cerebral thromboembolisms in AF patients is almost 2%-4% in Japan and increases with the CHADS 2 score/CHA 2 DS 2 -VASc score number. 3,4 Direct oral anticoagulants (DOACs) have been widely used to prevent cerebral infarctions in patients with AF. The advantages of DOACs over warfarin in reducing cerebral infarctions and bleeding complications have been demonstrated in several randomized clinical trials (RCT). [5][6][7][8] However, the long-term outcomes of DOAC use remain unclear in the catheter ablation (CA) era.
AF catheter ablation (AFCA) is widely performed, and some investigations have reported that it is more effective for preventing AF recurrences than medical therapy. 9,10 CA in patients with heart failure has been reported to be associated with a significantly lower rate of a composite end point of death from any cause or hospitalization for worsening heart failure than medical therapy, while the impact is less in patients without heart failure. 11,12 Evidence that the mortality improves in patients who undergo CA is still limited. Therefore, it is important to reveal how to select the best treatment of AF based on each patient's background.
AF patients without strokes have been followed by cardiologists in Japan. However, once cerebral vascular events occur, those patients are followed by brain surgeons or neurologists. Therefore, it is difficult for primary care doctors to share the events. To share those events, we established the HAF-NET (HYOGO ATRIAL all-cause mortality and to clarify the reality of AF management in Japan by using the data form the HAF-NET Registry.

| Study cohort
The HAF-NET Registry is multicenter, prospective, observational study of Japanese patients with AF. The patients were enrolled from April 2015 to August 2016. Inclusion criteria were those aged 20 or older in whom AF was diagnosed by a 12-lead or Holter electrocardiogram. There were no exclusion criteria. A total of 22 institutions, all of which were located in Hyogo Prefecture, participated in this registry. They consisted of eight cardiovascular centers, two affiliated or community hospitals, and 12 primary care clinics. All patients were followed through a review of the inpatient and outpatient medical records, and additional information was obtained through contact with the patients, relatives, and/or referring physicians by mail or telephone. The data were checked by clinical the posted information will be updated as needed to reflect the protocol amendments and study progress.

| Registration card and data collection
All patients had a certification of attendance, which contained information including the anticoagulation therapy regimen and contact information of the primary care doctor ( Figure 1). Even though clinical adverse events occurred while being seen by secondary care doctors, they could inform the primary care doctor of the events by using this card. The primary care doctor was able to log into the website and register information about any adverse clinical events. The clinical patient data were registered on the online database system by the doctor in charge at each institution. The data were automatically checked for any missing or contradictory entries and values out of the normal range. Additional editing checks were performed by the clinical research coordinators at the general office of the registry.
The baseline clinical background data were as follows: patient clinical characteristics including the date of birth, age, gender, body weight, serum creatinine level, date when AF was diagnosed, history of treatment including CA, cardiac surgery, percutaneous coronary intervention, or coronary arterial bypass grafting, type of AF, comorbidities, and risk factors including heart failure, hypertension, diabetes mellitus, strokes/TIAs, vascular disease, valvular disease, ischemic heart disease, cardiomyopathy, dementia, whether patients smoked or consumed alcohol at the time of enrollment, a reduced left ventricular function (%FS < 25% or EF < 35%), current medications including anticoagulant drugs (DOACs or warfarin) and antiplatelet drugs, and subjective symptoms including palpitations, dyspnea and dizziness. Paroxysmal AF was defined as AF that terminated spontaneously within 7 days, while persistent AF was defined as AF that lasted for > 7 days but could be terminated with medication or electrical cardioversion. Long-lasting persistent AF was defined as AF that lasted > 1 year.
The risk of a stroke was evaluated by the CHADS 2 score and CHA 2 DS 2 -VASc score. 13 The risk of bleeding was evaluated by the HAS-BLED score. 14 A PT-INR of 1.6-2.6 was the optimal therapeutic range for patients aged 70 or older and a PT-INR of 2.0-3.0 was appropriate for patients aged 69 or younger.

| Primary and secondary endpoints
The primary endpoints of this registry were symptomatic cerebral infarctions including TIAs, SEs, and fatal bleeding complications requiring hospitalization including an intracranial hemorrhage. A TIA was defined as a sudden onset of focal neurologic symptoms and/or a sign lasting less than 24 hours, brought on by a transient decrease in the blood flow, which rendered the brain ischemic in the area pro-

| Statistical analysis
Continuous data were presented as the mean ± SD for normally distributed variables. Medians and quartiles were given for nonnormally distributed variables. If these data followed a normal distribution, they were tested with an unpaired t-test or Welch test. If not, they were tested with a Mann-Whitney test. Categorical variables were analyzed with the Fisher's exact test. Cox proportional hazards regression models were used to estimate the hazard ratios and 95% confidence intervals for each event. Previously reported variables including age, gender, BW, AF type, AFCA history, valvular disease, ischemic heart disease, cardiomyopathy, EF less than 35%, heart failure, hypertension, age > 75 years, diabetes mellitus, stroke/TIA, vascular disease, antiplatelet drug, DOAC use, and HAS-BLED were also selected as cofounders.
The multivariable Cox proportional hazards regression model included variables with a P < .05 using an unadjusted Cox proportional hazard regression analysis. To compare the clinical events between the warfarin and DOAC users, 667 age, BW, and CHADS 2 score-matched DOAC and warfarin users were tested. The cumulative incidence of a stroke or SE was determined by the Kaplan-Meier method. The survival analysis between warfarin and DOACs was performed using a log-rank test. A value of P < .05 was considered statistically significant. All statistical analyses were performed using SPSS, Release 25 software (SPSS).

| RE SULTS
A total of 2113 patients were enrolled from 38 institutions in Hyogo prefecture between April 2015 and August 2016. Of those, 1343 F I G U R E 1 Registration card. The left panel shows the front side of the registration card where the actual anticoagulation therapy could be checked. The following text was described in the registration card: "Please inform your primary care doctor of the following events: Ischemic stroke, SE, Hemorrhagic stroke, new onset dementia, hospitalization for major bleeding, hospitalization for cardiac event, all-cause mortality" was written by Japanese. The right panel shows the opposite side of the registration card where the patient name and birthday, primary care doctor's name and telephone number were written. The following text was described in the registration card: "Please always carry this card to inform your doctors of anticoagulation therapy. When you see a doctor, please show this card your doctors, dentists and pharmacists. According to the doctor's suggestion, please do not change the dosage of anticoagulants by self-determination. Please inform your primary care doctor when the anticoagulation therapy reluctantly stopped." (64%) were enrolled from cardiovascular centers, 66 (3%) from affiliated or community hospitals, and 704 (33%) from private clinics.
Two thousand and seventy (98%) of the 2113 patients were followed for 1 year after the enrollment and the mean follow-up period was 355 ± 43 days.

| Baseline characteristics of the registered patients
The baseline characteristics of the registered patients are summarized in Table 1. Almost 70% of the patients were male. The mean age was 70.1 years and 36% of them were aged 75 or over. Half of the patients had paroxysmal AF. Almost half of the patients were symptomatic and the most common subjective symptom was palpitations. The mean CHADS 2 and CHA 2 DS 2 -VASc scores were 1.5 ± 1.1 and 2.6 ± 1.6, respectively. Figure 2 shows the patients distribution according to the CHADS 2 score and CHA 2 DS 2 -VASc score. A CHADS 2 score = 1 and CHA 2 DS 2 -VASc score = 3 were the most common subpopulations. Table 2 shows the comorbidities of the patients. Hypertension was by far the most prevalent underlying disease, and 21.3% of the patients suffered from heart failure, of which 5.2% had an ejection fraction of <35%. Ischemic heart disease and valvular disease were present in 7.2% and 13.1%, respectively. Of those, mitral regurgitation was remarkably frequent. Of note, dementia was present in 2.7% of the patients. Almost 30% of the patients had a history of CA.

| Medications in HAF-NET patients
Low-dose users were common in both the dabigatran and edoxaban users, but not in the rivaroxaban and apixaban users.

| Main findings of the study
The data from the HAF-NET registry demonstrated a higher DOAC use and AFCA history as compared to the previous studies, which resulted in excellent outcomes after 1 year of follow-up among Japanese patients with AF in the DOAC and AF ablation era.
Persistent AF and a lower BW were strongly associated with stokes/ SEs and major bleeding, respectively. AFCA history as well as age

| Patient characteristics
One-third (800 patients) of the patients in this registry were from Chuo-Ku, which is located in the southern region of Kobe city. The population of Chuo-Ku is approximately 135 000 people. Based on the epidemiological prevalence of AF in the Japanese population of 0.6%, the number of AF patients in Chuo-Ku was estimated to be approximately 810. As the number in our registry was almost equal to the estimated AF patients in Chuo-Ku, the AF patents in this registry were assumed to fully reflect a typical ward in Kobe city.
In Japan, real-world data of the anticoagulation therapy in patients with AF have been published from two major AF registry such as FUSHIMI and SAKURA registry. 15,16 The enrollment of the

| Medications in HAF-NET patients
Warfarin was prescribed in only around 30% of the patients in the HAF-NET registry. As compared to the FUSHIMI and SAKURA reg- The prevalence of a low-dose usage of DOACs was significantly less in rivaroxaban/apixaban users than in dabigatran/edoxaban users.
Of importance, the SAKURA registry identified inappropriately low dosing in 19.7 to 27.6% of the DOAC users. The proportion of an adjusted low dosing was estimated to be almost 20% in the rivaroxaban or apixaban users and almost 50% in the dabigatran or edoxaban users. Furthermore, postmarketing studies for each DOAC also estimated that the proportion of an adjusted low dosing was almost 30% in the rivaroxaban or apixaban users and almost 60% in the dabigatran or edoxaban users. [19][20][21][22] The proportion of this estimated adequately low-dosing usage in the SAKURA registry was similar to our results. This indicated that the inadequate low dosing in the HAF-NET registry was extremely less than that in the previous AF registries. Almost 8 years have passed since dabigatran was released as the first DOAC in 2011. Over the past decade, we have experienced the importance of adequate dosing of DOACs, which has increased major bleeding as well as strokes or TIAs. Therefore, adequate dosing might be challenged in the recent real world of AF anticoagulation therapy in Japan. Actually, our data clearly supported this challenge and demonstrated excellent outcomes.

| Primary and secondary endpoints and clinical predictors
The two major postmarket surveillance (PMS) studies (J-Dabigatran surveillance, XAPASS) showed the incidence rates of major bleeding and thromboembolic events, suggesting that dabigatran and rivaroxaban were safe and effective in the Japanese clinical prac-

| Catheter ablation and anticoagulation therapy
Recently, several studies have reported the impact of CA on the mortality and cardiovascular hospitalization in patients with AF.
Especially in patients with heart failure, the impact has been greater.
The CATSLE-AF study clearly demonstrated that CA was associated with a significantly lower rate of the composite end point of death from any cause or hospitalization for worsening heart failure as compared to medical therapy. Although no statistical significance could be found, cerebrovascular accidents were dramatically reduced by CA as compared to medical therapy. 11 Furthermore, the impact of CA has been greater in patients with an age of <65 years old, heart failure of <NYHA functional class II, and EF of ≧ 25%. The CABANA study also reported that the impact of CA was greater in patients with an age of <65 years old. 12

| Impact of DOACs on preventing clinical events
Previous RCTs have revealed better clinical outcomes especially for fatal bleeding under DOAC therapy than under warfarin therapy. However, fewer Japanese patients could be enrolled in those RCTs. [5][6][7][8] The SAKURA registry showed no significant differences in the rates of strokes or SEs, major bleeding, and all-cause mortality for DOAC vs. warfarin users. Under propensity score matching, the incidence of strokes or SEs and all-cause death remained equivalent, but the incidence of major bleeding was significantly lower among DOAC than warfarin users. 25 In the HAF-NET registry, the incidence of strokes or SEs was significantly lesser in the DOAC users, but not that for major bleeding. This discrepancy might be caused by the frequency of the CA history. Progression of AF was reported to be associated with an increased risk of clinical adverse events during the arrhythmia progression period from paroxysmal to persistent AF among Japanese patients with AF. The risk of adverse events was also transiently elevated during the progression period from paroxysmal to persistent AF and declined to a level equivalent to persistent AF after the progression. 26 A CA history was found in almost 10% and 25% in the SAKURA an HAF-NET registries, respectively.
As compared to medical therapy, CA could strongly reduce the AF burden and no progression toward persistent AF was observed during a median follow-up of 6 years especially in patients with paroxysmal AF. 27 In such patients without AF recurrence after a successful CA, DOACs might be continued without a dose reduction, while the PT-INR level might be controlled at a lower level to avoid the fatal bleeding. This suggested the importance of CA and DOACs for preventing strokes or SEs and the awareness of an adequate DOAC lower dosing after a successful CA.

| Dementia and AF
A meta-analysis reported that AF was independently associated with an increased risk of all forms of dementia. 1 The incidence of dementia in the patients without AF was almost 3.0% during a followup period of over 5 years. After a dementia diagnosis, the presence of AF was associated with a marked increased risk of mortality. 2 Recently, individuals with AF have been reported to have an almost threefold increased risk of dementia during a 12 year follow-up (HR 2.8; 95% CI 1.3-5.7; P = .004). The population attributable risk for dementia resulting from AF was 13%. They concluded that patients with AF should be screened for cognitive symptoms. 28 In the HAF-NET registry, dementia diagnosed before enrollment was found in 56 (2.7%) of the patients and the incidence of new onset dementia was 8 (0.4%) patients. This annulus incidence of dementia was similar to that in patients without AF. This might be the impact from anticoagulation therapy with DOACs and a strong rhythm control therapy with CA. We hope that this impact would continue during the follow-up of over 3 years because the average time to the development of dementia has been reported to be almost 3 years.

| Study limitations
This study had several limitations. First, this study was designed as a prospective observational study, therefore, only associations were shown, not causality. The possibility of unmeasured or residual confounding factors was not ruled out. Second, anticoagulant therapy was assessed at the time of the enrollment, but the changes in the medical therapy could not be assessed. Third, to assess the impact of the DOAC therapy, age, BW, and CHADS 2 score-matched DOAC, and warfarin users were compared because of the small number in each medical therapy group. Fourth, this study involved AF patients recruited from a small region of Japan, and therefore, the results might not be generalizable to the overall population.

| CON CLUS ION
The HAF-NET registry was characterized as (a) having a high incidence of DOAC prescriptions and a CA history and (b) including relatively younger patients with lower CHADS 2 scores. In the DOAC and CA era, the incidence of ischemic strokes/SEs, major bleeding and hospitalization for the cardiac events could be strongly reduced in patients with AF. However, some unsolved issues of AF management still remain especially in elderly patients with persistent AF and a low BW.

ACK N OWLED G EM ENTS
We would like to thank John Martin for his linguistic assistance and Hiromasa Suzuki for his assistance with the clinical research coordination. The key personnel and institutions participating in the registry are as follows: Chief investigator: Yoshida A (Kita-harima Medical Center)