Circulation Journal

Official Journal of the Japanese Circulation Society

ORIGINAL ARTICLE
http://www. j-circ.or.jp Arrhythmia/Electrophysiology

Rivaroxaban vs. Warfarin in Japanese

Patients With Atrial Fibrillation
– The J-ROCKET AF Study –
Masatsugu Hori, MD, PhD; Masayasu Matsumoto, MD, PhD; Norio Tanahashi, MD;
Shin-ichi Momomura, MD; Shinichiro Uchiyama, MD, PhD; Shinya Goto, MD, PhD;
Tohru Izumi, MD, PhD; Yukihiro Koretsune, MD, PhD; Mariko Kajikawa, MD, PhD;
Masaharu Kato; Hitoshi Ueda, PhD; Kazuya Iwamoto, MD, PhD; Masahiro Tajiri, BSc;
on behalf of the J-ROCKET AF study investigators

Background: The global ROCKET AF study evaluated once-daily rivaroxaban vs. warfarin for stroke and systemic
embolism prevention in patients with atrial fibrillation (AF). A separate trial, J-ROCKET AF, compared the safety of a Japan-
specific rivaroxaban dose with warfarin administered according to Japanese guidelines in Japanese patients with AF.

Methods and Results: J-ROCKET AF was a prospective, randomized, double-blind, phase III trial. Patients
(n=1,280) with non-valvular AF at increased risk for stroke were randomized to receive 15 mg once-daily rivaroxaban
or warfarin dose-adjusted according to Japanese guidelines. The primary objective was to determine non-inferiority
of rivaroxaban against warfarin for the principal safety outcome of major and non-major clinically relevant bleeding,
in the on-treatment safety population. The primary efficacy endpoint was the composite of stroke and systemic
embolism. Non-inferiority of rivaroxaban to warfarin was confirmed; the rate of the principal safety outcome was
18.04% per year in rivaroxaban-treated patients and 16.42% per year in warfarin-treated patients (hazard ratio [HR]
1.11; 95% confidence interval 0.87–1.42; P<0.001 [non-inferiority]). Intracranial hemorrhage rates were 0.8% with
rivaroxaban and 1.6% with warfarin. There was a strong trend for a reduction in the rate of stroke/systemic embolism
with rivaroxaban vs. warfarin (HR, 0.49; P=0.050).

Conclusions: J-ROCKET AF demonstrated the safety of a Japan-specific rivaroxaban dose and supports bridging
the global ROCKET AF results into Japanese clinical practice.   (Circ J 2012; 76: 2104 – 2111)

Key Words: Anticoagulants; Atrial fibrillation; Japanese; Prevention; Stroke

F
or more than 50 years, vitamin K antagonists (VKAs),
primarily warfarin, have been the most effective anti-
thrombotic therapy available for the prevention of isch-
emic stroke in patients with atrial fibrillation (AF).1 However,
warfarin therapy is associated with numerous issues that limit
its use, such as multiple food, drug, and pharmacogenomic in-
teractions, which contribute to its unpredictable pharmacoki-
netics and pharmacodynamics.1 Frequent coagulation moni-
toring and dose adjustment is necessary to maintain warfarin
anticoagulation within the therapeutic international normalized
ratio (INR) range.2
Editorial p 2086
Rivaroxaban is a novel oral, direct Factor Xa inhibitor in
advanced clinical development that may overcome the many
drawbacks of warfarin. ROCKET AF was a global phase III trial
that evaluated the safety and efficacy of rivaroxaban 20 mg once
daily (o.d.) for the prevention of stroke and systemic embolism
in patients with non-valvular AF; a reduced dose of 15 mg o.d.

Received April 4, 2012; accepted May 8, 2012; released online June 5, 2012   Time for primary review: 13 days
Osaka Medical Center for Cancer and Cardiovascular Diseases, Osaka (M.H.); Department of Clinical Neuroscience and Therapeutics,
Hiroshima University, Hiroshima (M.M.); Department of Neurology, Saitama Medical University International Medical Center, Hidaka
(N.T.); Division of Cardiovascular Medicine, Saitama Medical Center, Jichi Medical University, Saitama (S.M.); Department of Neurol-
ogy, Tokyo Women’s Medical University, Tokyo (S.U.); Department of Medicine (Cardiology), Tokai University School of Medicine,
Tokyo (S.G.); Department of Cardio-angiology, Kitasato University School of Medicine, Sagamihara (T.I.); Institute for Clinical Re-
search, Osaka National Hospital, Osaka (Y.K.); Bayer Yakuhin Ltd, Osaka (M. Kajikawa, M. Kato, H.U., K.I., M.T.), Japan
Clinical Trial Registration: http://www.clinicaltrials.gov; NCT00494871.
Mailing address: Masatsugu Hori, MD, PhD, Osaka Medical Center for Cancer and Cardiovascular Diseases, 1-3-3 Nakamichi, Higashinari-ku,
Osaka 357-8511, Japan.   E-mail: hori-ma@mc.pref.osaka.jp
ISSN-1346-9843   doi: 10.1253/circj.CJ-12-0454
All rights are reserved to the Japanese Circulation Society. For permissions, please e-mail: cj@j-circ.or.jp

Circulation Journal Vol.76, September 2012

Rivaroxaban vs. Warfarin in Japanese AF Patients
was evaluated in patients with moderate renal impairment, de-
fined as baseline creatinine clearance (CrCl) 30–49 ml/min.3
Patients in Japan were not enrolled into the global ROCK-
ET AF trial for 2 reasons. First, pharmacokinetic modeling data
indicated that, at steady state, the distribution of both the max-
imum concentration (Cmax: median: 259.48 μg/L) and area under
the curve from 0 to 24 h (AUC0–24: median: 3,193.89 μg*h/L)
of rivaroxaban in Japanese patients with AF who received a
15 mg o.d. dose of rivaroxaban would be comparable to the Cmax
(median: 289.05 μg/L), and AUC0–24 (median: 3,243.04 μg*h/L),
in Caucasian patients with AF who received a 20 mg o.d. dose
(Tanigawa et al unpublished data, 2012). Second, lower anti-
coagulation targets are used in Japanese clinical practice:
Japanese guidelines recommend a reduced INR target level of
1.6–2.6 in patients with AF aged ≥70 years,4 and in practice a
tendency for Japanese physicians to favor lower levels of an-
ticoagulation in patients aged <70 years has been reported.5
Accordingly, the 15 mg o.d. dose of rivaroxaban was selected
for the phase III J-ROCKET AF study to provide a margin
consistent with the lower target INRs recommended for VKA
therapy in clinical practice in Japan.4,5
The J-ROCKET AF study was conducted entirely in Japan
and was designed specifically to evaluate the safety of the
15 mg o.d. dose of rivaroxaban vs. warfarin comparator dosed
according to Japanese practice, with the lower target INR range
for patients ≥70 years.4
Methods
Study Design and Enrolment Criteria
J-ROCKET AF was a prospective, randomized, double-blind,
double-dummy, parallel-group, active-controlled, multicenter
clinical trial that evaluated the safety of rivaroxaban vs. dose-
adjusted warfarin. Safety was assessed with respect to on-treat-
ment bleeding events in the safety population (ie, all patients who
received ≥1 dose of the study drug). The trial was conducted in
accordance with Japanese Good Clinical Practice, the Declara-
tion of Helsinki, and the International Conference on Harmo-
nization guideline E6. The study was approved by the Institu-
tional Review Boards and all patients gave informed consent.
Study Participants
Japanese patients aged ≥20 years with non-valvular AF, docu-
mented electrocardiographically ≤30 days before randomiza-
tion, were recruited from 167 participating sites in Japan. Pa-
tients had a history of prior ischemic stroke, transient ischemic
attack (TIA), or non-central nervous system (CNS) systemic
embolism or had ≥2 of the following risk factors for thrombo-
embolism: congestive heart failure and/or left ventricular ejec-
tion fraction ≤35%, hypertension, age ≥75 years, or diabetes
mellitus. Recruitment of patients without prior stroke, TIA, or
non-CNS systemic embolism and with only 2 stroke risk fac-
tors was limited to 10% of the total number of patients. Detailed
inclusion and exclusion criteria are available in Data S1.
Study Treatment
A double-blind, double-dummy design was chosen to minimize
bias with respect to concomitant interventions and reporting
of clinical events. Patients with AF were randomized to receive
either oral rivaroxaban 15 mg o.d. (10 mg o.d. in patients with
CrCl 30–49 ml/min at randomization) or warfarin dose-adjusted
to a target INR of 2.0–3.0 in patients aged <70 years, or a re-
duced INR of 1.6–2.6 in patients aged ≥70 years. As part of
the double-dummy design, patients in each group also received
a tablet of either titrated warfarin placebo or rivaroxaban pla-
Circulation Journal Vol.76, September 2012
2105
cebo, respectively, to preserve the treatment blind. INR mon-
itoring procedures to maintain double-blinding are described
in Data S1. Treatment compliance was evaluated by express-
ing the proportion of days a patient took study medication as
a percentage of the total treatment duration.
Study Procedures
The study was divided into several periods: screening, double-
blind treatment closing with an end-of-study visit, and a 30-day
post-treatment observation period. Patients returned for visits
at weeks 2 and 4, and every 4 weeks thereafter for the duration
of the double-blind treatment period. The prespecified maxi-
mum exposure period, and the expected study duration, were
30 months. At the end-of-study visit, or at an early discontinu-
ation visit, patients were transitioned by the investigator from
study medication to open-label warfarin or other appropriate
therapy according to usual clinical practice, and a follow-up
visit was performed 30 days after the end-of-study or early
discontinuation visit.
Adjudication of Clinical Outcomes
An independent clinical endpoint committee adjudicated all
suspected strokes, systemic embolisms, myocardial infarctions
(MIs), deaths, and bleeding events contributing to the prespeci-
fied endpoints. Events taking place from the time of random-
ization through to the end of the study treatment period and until
the end of the 30-day follow-up period were adjudicated.
Safety Outcomes
The principal safety outcome was the composite of major and
non-major clinically relevant bleeding assessed by a blinded
clinical endpoint committee. Bleeding events adjudicated by the
clinical endpoint committee to have met the definition of hem-
orrhagic stroke were included in the safety outcome analysis
as well as the efficacy analysis. Major and non-major clinically
relevant bleeding events are defined in the Table S1. Other
overt bleeding episodes that did not meet the criteria for major
or non-major clinically relevant bleeding were classified as min-
imal. Treatment-emergent adverse events (AEs), hepatic liver
enzyme activity, and total bilirubin were also assessed. Safety
outcomes were assessed in the safety population, which included
all patients who received ≥1 dose of study drug, using an on-
treatment analysis, defined as the period from first dose of study
drug up to 2 days after last dose.
Efficacy Endpoints
The primary efficacy endpoint was the composite of adjudi-
cated all-cause stroke (ischemic or hemorrhagic) and non-CNS
systemic embolism. Secondary efficacy endpoints included a
composite of stroke, systemic embolism, and vascular death
and a composite of stroke, systemic embolism, vascular death,
and MI. Individual components of the composite secondary
endpoints were also included. Definitions of efficacy endpoints
and their components are reported in the Table S1).
Statistical Analysis
Safety   The primary objective was to test whether rivaroxa-
ban was non-inferior to warfarin with respect to the principal
safety outcome in the safety population, on-treatment, as eval-
uated by non-stratified Cox proportional hazards modeling.
Based on the expected incidence of adjudicated major bleeding
events and non-major clinically relevant bleeding events, a sam-
ple size of 1,200 patients with 600 per arm was considered suf-
ficient to test the non-inferiority of rivaroxaban with respect to
the principal safety outcome, with non-inferiority to be con-
2106 HORI M et al.

Table 1. Baseline Demographics

Rivaroxaban Warfarin Total
Patient characteristics
(n=639) (n=639) (n=1,278)
Sex, n (%)
   Male 530 (82.9) 500 (78.2) 1030 (80.6)
   Female 109 (17.1) 139 (21.8) 248 (19.4)
Age, years
   Mean (range) 71.0 (34–89) 71.2 (43–90) 71.1 (34–90)
Baseline creatinine clearance, n (%)
   30–49 ml/min* 141 (22.1) 143 (22.4) 284 (22.2)
   50–<80 ml/min 328 (51.3) 328 (51.3) 656 (51.3)
   ≥80 ml/min 170 (26.6) 168 (26.3) 338 (26.4)
Prior warfarin used,† n (%) 577 (90.3) 573 (89.7) 1,150 (90.0)
Prior acetylsalicylic acid used, n (%) 243 (38.0) 222 (34.7) 465 (36.4)
CHADS2 score
   Mean 3.27 3.22 3.25
   0–1 (%)‡ 0 0 0
   2 (%) 15.2 18.0 16.6
   ≥3 (%) 84.8 82.0 83.4
Congestive heart failure (%) 41.3 40.2 40.8
Hypertension (%) 79.5 79.5 79.5
Age ≥75 years (%) 39.4 38.5 39.0
Diabetes mellitus (%) 39.0 37.1 38.0
 aseline stroke/transient ischemic attack/systemic
B 63.8 63.4 63.6
embolism (%)
Prior MI (%) 7.0 8.3 7.7
*These patients received a reduced 10 mg once-daily dose. †Warfarin
use for ≥6 weeks at time of screening. ‡Patient
with CHADS2 score=2 incorrectly registered as CHADS2 score=0.
CHADS2, congestive heart failure, hypertension, age ≥75 years, diabetes mellitus, prior stroke, transient ischemic
attack, or thromboembolism (2 points); MI, myocardial infarction.

cluded if the upper boundary of the 95% confidence interval
(CI) for the hazard ratio (HR) of rivaroxaban to warfarin did
not exceed 2.0. This margin was chosen based on studies in
Asian patients with AF, which demonstrated at least a 2-fold
increase in bleeding risk with warfarin treatment at INRs ≥2.6
compared with <2.6 (Table S6).6,7 Therefore in this study, as-
suming good control of INR, if bleeding resulting from rivar-
oxaban treatment was less than 2-fold higher compared with
that seen with warfarin, rivaroxaban would be at least non-in-
ferior to warfarin treatment at an INR <2.6. Further details are
available in Data S1.
Efficacy   The study was not powered to test efficacy hypoth-
eses and efficacy endpoints were evaluated in both the per-pro-
tocol and intention-to-treat (ITT) populations. The per-proto-
col population was defined as all patients randomized without
any major protocol violations, using an on-treatment analysis
as described previously. ITT population analyses included all
randomized patients, and were carried out using both an on-
treatment analysis (including events occurring up to 2 days after
last dose), and an analysis including 30-day follow-up (includ-
ing events occurring until the 30-day follow-up visit). Events
occurring after the 30-day follow-up visit for discontinued pa-
tients were not collected in the analysis including 30-day fol-
low-up.
Time in Therapeutic Range (TTR)   For each patient receiv-
ing warfarin, individual TTR was based on regression analy-
ses, calculated from INR values using linear interpolation.8 The
cumulative total of patient treatment days was the denominator,
and the total number of days that INR was in the target range
was the numerator. Center TTR was calculated by dividing the
total number of INR values within the target range at a center
by the total number of INR values from all warfarin-treated
patients at the center.
Bayer Yakuhin Ltd funded the trial and was responsible for
trial design and study data collection. A Steering Committee,
chaired by Dr M. Hori, approved the design of the trial and was
responsible for oversight of the conducting of the study. All
authors take responsibility for the accuracy and completeness
of the data and all analyses presented here.
Results
Patients
A total of 1,439 patients were screened for eligibility and 1,280
were randomized beginning June 8, 2007. The last patient visit
was recorded on January 19, 2010. Of the 1,280 randomized
patients, 1,278 (639 in each group) received ≥1 dose of study
medication and were included in the safety population, and
1,274 patients without major protocol violations were included
in the per-protocol population (Figure S1). Baseline demograph-
ics were balanced across both treatment arms (Table 1). Mean
patient age was 71.1 years. At baseline, moderate renal impair-
ment, defined as CrCl 30–49 ml/min, was present in 22.1% of
patients randomized to rivaroxaban; these patients received a
reduced 10 mg o.d. rivaroxaban regimen. Mean treatment com-
pliance was >99% for both treatment groups.
Safety Outcomes
The principal safety outcome, the composite of adjudicated major
bleeding and non-major clinically relevant bleeding events, eval-

Circulation Journal Vol.76, September 2012

Rivaroxaban vs. Warfarin in Japanese AF Patients
uated in the safety population, on-treatment analysis, occurred
in 138 patients in the rivaroxaban group (18.04% per year) com-
pared with 124 patients in the warfarin group (16.42% per year;
HR 1.11; 95% CI 0.87–1.42). The upper boundary of the 95%
CI was 1.42, well below the prespecified non-inferiority bound-
ary of 2.0 (Figure 1A), indicating the non-inferiority of rivar-
oxaban to warfarin with respect to the principal safety outcome.
The observed rate of major bleeding events was 3.00% per
year in the rivaroxaban arm compared with 3.59% per year in
the warfarin arm (HR 0.85; 95% CI 0.50–1.43), and observed
rates also tended to be lower with rivaroxaban for all indi-
vidual components of the major bleeding outcome (Figure 1B),
although none of the differences was statistically significant.
Non-major clinically relevant bleeding event rates were 15.42%
per year in rivaroxaban-treated patients compared with 12.99%
per year in warfarin-treated patients (HR 1.20; 95% CI 0.92–
1.56), and this difference was also not statistically significant.
Major bleeding from the upper gastrointestinal tract occurred
in 6 patients (0.9%) who received rivaroxaban and in 12 patients
(1.9%) who received warfarin. Treatment-emergent fatal bleed-
Circulation Journal Vol.76, September 2012
2107
Figure 1. Principal safety outcomes
in the safety population on-treatment
analysis. (A) Kaplan-Meier curve of
time to first major bleeding or non-
major clinically relevant bleeding
event. (B) Components of the prin-
cipal safety outcomes: major or non-
major clinically relevant bleeding. CI,
confidence interval; HR, hazard ratio;
PRBC, packed red blood cells.
ing events occurred in 1 patient who received rivaroxaban and
in 3 patients who received warfarin. In terms of critical organ
bleeding, intracranial hemorrhages were observed in 5 patients
(0.8%) in the rivaroxaban group and in 10 patients (1.6%) in the
warfarin group. These results were not tested for statistical sig-
nificance. Sites of major bleeding are shown in Figure 2.
No significant differences in principal safety outcome rates
were observed between the rivaroxaban and warfarin treatment
groups, either in patients with moderate renal impairment (HR
1.22; 95% CI 0.78–1.91) or in patients with mild or no renal
impairment and baseline CrCl ≥50 ml/min (HR 1.07; 95% CI
0.80–1.43; interaction P-value=0.628) (Table S2).
Adverse Events
The observed incidence of treatment-emergent AEs in the safe-
ty population, including drug-related AEs, was similar between
the 2 treatment groups (Table 2). The incidence of AEs lead-
ing to permanent discontinuation of the study drug was 13.1%
in the rivaroxaban arm and 15.0% in the warfarin arm. The most
frequently reported treatment-emergent bleeding events (ie, those
2108 HORI M et al.

Figure 2. Most frequent major bleed-

ing by site. Safety population, on-treat-
ment analysis. GI, gastrointestinal.

Efficacy Endpoints
Table 2. Treatment-Emergent Adverse Event Summary and
Liver Enzyme Data In the primary efficacy analysis in the per-protocol population,
while on treatment, stroke or non-CNS systemic embolism oc-
Rivaroxaban, Warfarin,
Adverse event*
n=639 (%) n=639 (%) curred at a rate of 1.26% per year in patients receiving rivar-
Any serious AE 23.6 24.3
oxaban, compared with 2.61% per year in warfarin-receiving
patients (HR 0.49; 95% CI 0.24–1.00; P=0.050; Figure 3A,
 ny AE leading to study drug
A 13.1 15.0
discontinuation Table 3).
Treatment-emergent AEs
In terms of the secondary efficacy endpoints in the per-pro-
tocol population, on-treatment analysis, the observed incidence
   Nasopharyngitis 32.2 36.3
of the composite of adjudicated stroke, non-CNS systemic em-
   Epistaxis 16.3 9.4
bolism, and vascular death was 1.83% per year in the rivaroxa-
   Subcutaneous hemorrhage 10.5 12.5 ban group compared with 2.85% per year in the warfarin group
   Contusion 9.2 8.8 (HR 0.65; 95% CI 0.34–1.22), and the event rates of the com-
   Diarrhea 8.9 6.3 posite of adjudicated stroke, non-CNS systemic embolism, MI,
  Upper respiratory tract 8.8 11.6 and vascular death were 2.18% per year and 2.97% per year
inflammation for rivaroxaban and warfarin, respectively (HR 0.74; 95% CI
   Gingival bleeding 8.5 4.9 0.41–1.34; Table 3). All-cause stroke occurred at a lower rate
   Back pain 7.8 8.8 in patients treated with rivaroxaban than with warfarin (HR
   Diabetes mellitus 6.9 4.9 0.46; 95% CI 0.22–0.98), as did primary ischemic stroke (HR
   Cardiac failure 6.6 5.5 0.40; 95% CI 0.17–0.96). Primary hemorrhagic stroke occur-
ALT elevation† rence was similar in both treatment arms (HR 0.73; 95% CI
   >3×ULN 2.19 2.19 0.16–3.25). The numbers of MI, vascular death, and all-cause
  >3×ULN and total bilirubin 0.47 0.47
mortality were also low in both treatment groups (Table 3);
>2×ULN‡ however, the numbers of each event were too few to permit
meaningful statistical analysis.
*Safety population, on-treatment. †Safety population, post base-
line. ‡On the same calendar day. In the ITT population analysis including 30-day follow-up,
AE, adverse event; ALT, alanine aminotransferase; ULN, upper the primary efficacy endpoint occurred at a rate of 2.38% per
limit of normal. year and 2.91% per year in patients receiving rivaroxaban and
warfarin, respectively (HR 0.82; 95% CI 0.46–1.45; Figure S2).
In the on-treatment analysis of the ITT population, the primary
occurring at a rate ≥10% in either treatment group) were epi- efficacy endpoint occurred at a rate of 1.26% per year and 2.60%
staxis (16.3% in rivaroxaban patients vs. 9.4% in warfarin pa- per year in patients receiving rivaroxaban and warfarin, respec-
tients) and subcutaneous hemorrhage (10.5% vs. 12.5%, respec- tively (HR 0.48; 95% CI 0.23–1.00).
tively). The efficacy of rivaroxaban relative to warfarin in the ITT
Treatment-emergent serious AEs were reported in 23.6% and population analysis including 30-day follow-up was lower than
24.3% of patients receiving rivaroxaban or warfarin, respec- in the on-treatment analyses of the per-protocol and ITT pop-
tively. Elevations of hepatic enzyme activity and total biliru- ulations, partly because of a higher rate of primary efficacy events
bin during the study were similar in both treatment groups, and in the off-treatment period. These data are discussed in Data S1.
there was no indication of severe liver damage (Table 2).

Circulation Journal Vol.76, September 2012

Rivaroxaban vs. Warfarin in Japanese AF Patients 2109

Figure 3. Kaplan-Meier curve of time

to first primary efficacy endpoint event
in the per-protocol population on-treat-
ment analysis. CI, confidence interval;
HR, hazard ratio.

Table 3. Primary and Secondary Efficacy Endpoints

Rivaroxaban Warfarin HR
Composite efficacy endpoints*
(n=637) (n=637) (95% CI)
Primary efficacy endpoint (stroke plus non-CNS systemic 11 (1.26) 22 (2.61) 0.49 (0.24–1.00)
embolism), n (% per year)
   All-cause stroke, n 10 21 0.46 (0.22–0.98)
   Primary hemorrhagic stroke, n 3 4 0.73 (0.16–3.25)
   Primary ischemic stroke, n 7 17 0.40 (0.17–0.96)
 econdary efficacy endpoint 1 (stroke, non-CNS systemic
S 16 (1.83) 24 (2.85) 0.65 (0.34–1.22)
embolism and vascular death), n (% per year)
 econdary efficacy endpoint 2 (stroke, non-CNS systemic
S 19 (2.18) 25 (2.97) 0.74 (0.41–1.34)
embolism, MI and vascular death), n (% per year)
Other endpoints†
   Non-CNS systemic embolism, n 1 1 –
   MI, n 3 1 –
   Vascular death, n 6 2 –
   Stroke with serious residual disability, n 5 10 –
   All-cause mortality, n 7 5 –
*Per-protocol population on treatment. †Too
few events occurred to provide a robust statistical evaluation.
HR, hazard ratio; CI, confidence interval; CNS, central nervous system; MI, myocardial infarction.

TTR Safety Outcomes

Over the entire treatment period, 65.0% of the INR values of
warfarin-treated patients were within the prespecified, age-de-
pendent target range. For those aged ≥70 years, 74.0% of val-
ues were within the therapeutic INR range, but for those aged
<70 years (for whom guidelines direct that the target INR range
should be 2.0–3.0) only 51.8% of values were within this range
(Table S3). Nevertheless, HRs for the treatment effect of ri-
varoxaban vs. warfarin for both the principal safety outcome
and primary efficacy endpoints (on-treatment analysis) were
consistent across quartiles of center TTR (Table S4,S5).
Discussion
This study was the first double-blind clinical trial conducted
entirely in Japan to compare Japanese guideline-directed war-
farin therapy for stroke prevention in AF with the newer oral
anticoagulant, rivaroxaban, at a dose specifically adjusted for
Japanese patient characteristics and Japanese clinical practice.4
The primary objective of this trial was met, because non-infe-
riority of rivaroxaban vs. warfarin was observed with respect
to the principal safety outcome of major plus non-major clini-
cally relevant bleeding. Although no significant differences in
overall bleeding rates were observed, rivaroxaban use was as-
sociated with a non-significant lower major bleeding rate and
a slightly higher non-major clinically relevant bleeding rate than
with warfarin. In the global ROCKET AF study, there was no
significant difference in either the composite of major plus non-
major clinically relevant bleeding or in the individual compo-
nents of the composite endpoint.3 As with the global ROCKET
AF study, fewer intracranial hemorrhages were observed with
rivaroxaban therapy compared with warfarin therapy, with fewer
fatal bleeding events, although there were fewer total events
in J-ROCKET AF, limiting the robustness of the analyses of
these outcomes. Fewer intracranial hemorrhages with rivaroxa-
ban than warfarin might be attributable to the different mech-
anisms of inhibition of the coagulation cascade; warfarin sup-

Circulation Journal Vol.76, September 2012

2110
presses the tissue factor–FVIIa complex and other multiple
factors (ie, FIXa, FXa and thrombin), whereas rivaroxaban does
not inhibit the tissue factor–FVIIa complex and directly inhib-
its only FXa.3,9 In J-ROCKET AF, despite the lower PT-INR
target than in the global ROCKET AF, rivaroxaban also showed
a tendency of fewer intracranial hemorrhages compared with
warfarin. AE rates were similar across both treatment groups,
and there were no significant differences between treatment
arms in the elevation of liver enzyme levels.
Major Gatrointestinal Bleeding in J-ROCKET AF and Global
ROCKET AF
J-ROCKET AF demonstrated that the major GI bleeding rate
of the rivaroxaban group tended to be lower than that of the
warfarin group, whereas the global ROCKET AF showed major
GI bleeding more frequently in the rivaroxaban group than in
the warfarin group. This discrepancy in the rate of GI bleeding
between J-ROCKET AF and global ROCKET AF might be at-
tributable to ethnic difference in GI bleeding, or to healthcare
divergence by country in the endoscopic diagnosis/treatment
for GI tract diseases and different patient awareness of GI bleed-
ing, both of which may influence the entry of patients to the
study. The possibility of a lack of robustness of the outcome
analysis because of the relatively limited number of patients
in J-ROCKET AF can not be excluded.
Bleeding Rates With Warfarin in Japanese Patients With AF
The results of J-ROCKET AF provide additional insight into
the Japanese perception of bleeding rates with warfarin, within
the wider context of recently published clinical trials of the
newer anticoagulants for stroke prevention in AF. Patients re-
ceiving warfarin in the standalone J-ROCKET AF trial had their
INR levels managed according to Japanese clinical guideline
standards. The major bleeding rate in patients in the warfarin
group was 3.59% per year in J-ROCKET AF, which was rough-
ly comparable to that observed in warfarin-receiving patients in
the subanalysis of the Japanese population in the RE-LY phase
III clinical trial of dabigatran etexilate vs. warfarin for stroke
prevention in patients with AF (3.31% per year);10 however,
differences in study design (eg, double-blind vs. open-label),
patient background (eg, median CHADS2 score), and the proto-
col-directed target INR ranges for the warfarin group (2.0–3.0
for patients aged <70 years and 2.0–2.6 for patients aged ≥70
years in the RE-LY study) preclude cross-trial comparisons.
Efficacy Endpoints
J-ROCKET AF shared the same efficacy endpoints as the glob-
al ROCKET AF study. In the global trial, it was estimated that
approximately 14,000 patients would be needed to demonstrate
the non-inferiority of rivaroxaban vs. warfarin with regard to
the primary efficacy endpoint. However, a total of 1,280 patients
were enrolled into J-ROCKET AF. Although J-ROCKET AF
was not powered for efficacy, efficacy endpoint data collection
was prespecified. The rate at which patients receiving rivar-
oxaban experienced the primary efficacy endpoint of stroke
and non-CNS systemic embolism in the on-treatment analysis
was approximately half the rate for patients receiving warfa-
rin, narrowly missing nominal statistical significance. In par-
ticular, rivaroxaban was associated with a strong trend towards
a reduction in all-cause stroke compared with warfarin; this
reduction was driven primarily by a reduction in ischemic stroke,
and the hemorrhagic stroke rates were similar in both treatment
arms. These results are consistent with those of the global
ROCKET AF trial, in which rivaroxaban was non-inferior to
warfarin for the primary efficacy endpoint with no significant
Circulation Journal Vol.76, September 2012
HORI M et al.
differences in the rates of either the principal safety out-
come, major bleeding, or non-major clinically relevant bleed-
ing (Table S7).3 Evaluation of the primary efficacy endpoint
in the ITT population confirmed these observations, which are
discussed further in Data S1.
In terms of the secondary efficacy endpoints and subgroup
analyses, absolute event numbers were low, precluding mean-
ingful comparisons between treatment groups. Notably, in the
global ROCKET AF trial, which had a 10-fold larger sample
size, there was a trend towards fewer MIs in patients while they
were receiving treatment with rivaroxaban compared with war-
farin (0.9% vs. 1.1%; HR 0.81; 95% CI 0.63–1.06), whereas
in J-ROCKET AF there were 3 and 1 MIs in the rivaroxaban
and warfarin treatment arms, respectively.
TTR Analyses
Patients assigned to warfarin therapy in this study had an over-
all mean TTR of 65.0%, which is similar to other clinical trials
of the newer oral anticoagulants.11,12 Japanese guidelines rec-
ommend that patients aged ≥70 years should receive warfarin
with a target INR range of 1.6–2.6, or 2.0–3.0 for those aged
<70 years.4 However, the Japanese physicians in this trial tend-
ed to anticoagulate to an INR range of 1.6–2.6 regardless of
the patient’s age. Patients aged <70 years had a TTR (INR
2.0–3.0) of only 51.8%. Analysis of time spent within the INR
range of 1.6–2.6 in patients aged <70 years revealed that 72.7%
of the INR values were within the range of 1.6–2.6 (Table S3),
a result that was similar to the 74.0% TTR in patients aged
≥70 years. This observation is consistent with Japanese clini-
cal practice, as demonstrated by TTR data from the J-RHYTHM
registry, which reported that in patients aged <70 years, most
INR values outside the therapeutic range were below the ther-
apeutic range that applied to their age group (ie, below 2.0),
reflecting a tendency of Japanese physicians to favor lower
levels of anticoagulation.5 Compared with warfarin, the safety
and efficacy of rivaroxaban were consistent across all quartiles
of TTR by clinical site, despite the relatively high risk profile
(CHADS2 score=3.25) of the enrolled patients (P-values for
interaction 0.591 and 0.963, respectively; Table S4,S5).
Renal Impairment Subgroup Analysis
In a similar manner to the global ROCKET AF,13 a reduced
rivaroxaban dose for patients with moderate renal impairment
was proactively assessed in J-ROCKET AF. Patients with a
baseline CrCl of 30–49 ml/min received a reduced rivaroxaban
dosing regimen of 10 mg o.d.. Notably, there was no significant
difference in the relative risk of either the principal safety out-
come or the primary efficacy endpoint for rivaroxaban vs. war-
farin with either dose of rivaroxaban (Table S2).
Overall, the trial results support the use of a reduced dose
of rivaroxaban (15 mg o.d.) for evaluation in Japanese patients
with non-valvular AF. The results of J-ROCKET AF also sup-
plement and are supported by those of the global ROCKET
AF, which demonstrated that, while on treatment, patients re-
ceiving rivaroxaban experienced significantly fewer strokes or
systemic emboli than patients receiving warfarin, with similar
rates of major plus non-major clinically relevant bleeding.
Summary
J-ROCKET AF demonstrated the safety of a Japan-specific dose
of rivaroxaban for the prevention of stroke and systemic embo-
lism in Japanese patients with AF and a moderate to high stroke
risk; that is, rivaroxaban was non-inferior to warfarin dose-ad-
justed according to Japanese guideline recommendations with
respect to the principal safety outcome. Furthermore, the trial
Rivaroxaban vs. Warfarin in Japanese AF Patients
design had a particular focus on the unique aspects of Japanese
clinical practice. Compared with warfarin therapy, there was
also a strong trend towards reduction in stroke and non-CNS
systemic embolism with rivaroxaban compared with warfarin.
The findings of J-ROCKET AF support bridging of the
broader safety and efficacy data from the larger, pivotal, glob-
al ROCKET AF study to Japanese patients with AF.
Acknowledgments
The authors thank Mark Hillen who provided medical writing services
with funding from Bayer HealthCare Pharmaceuticals AG and Janssen
Pharmaceuticals, Inc.
Sources of Funding
The rivaroxaban clinical development program is co-sponsored by
Janssen Pharmaceuticals, Inc (Raritan, NJ, USA) and Bayer HealthCare
Pharmaceuticals AG (Leverkusen, Germany). The trial was funded by Bayer
Healthcare Pharmaceuticals AG’s Japanese subsidiary, Bayer Yakuhin Ltd.
Disclosures
Drs Iwamoto, Kajikawa and Ueda, and Mr Tajiri and Mr Kato report
employment by Bayer Yakuhin. Mr Tajiri reports an equity interest in
Bayer HealthCare Pharmaceuticals. Dr Hori has received consultancy fees
from Bayer, Boehringer Ingelheim, Bristol Myers-Squibb and Pfizer. Dr
Matsumoto and Dr Momomura have received consultancy fees from
Bayer. Dr Tanahashi has received consultancy fees from Bayer and Mitsubishi
Tanabe, and honoraria from Mitsubishi Tanabe and Sanofi-Aventis. Dr
Goto has received research grants from Astellas, AstraZeneca, Daiichi,
Eisai, Kowa, Ono, Otsuka, Pfizer, Sanofi-Aventis, and Takeda, and hono-
raria from Daiichi-Sankyo, Eisai, Otsuka, Sanofi-Aventis and Schering-
Plough. Dr Izumi has received consultancy fees from Bayer and Pfizer. Dr
Koretsune has received honoraria from Bayer, Boehringer Ingelheim,
Daiichi-Sankyo and Bristol Myers-Squibb. Dr Uchiyama has received
consultancy fees from Bayer and Boehringer Ingelheim and research
grants from Bayer, Boehringer Ingelheim and Daiichi-Sankyo. No other
conflicts of interest are reported.
References
1. Ansell J, Hirsh J, Hylek E, Jacobson A, Crowther M, Palareti G.
Pharmacology and management of the vitamin K antagonists: Amer-
ican College of Chest Physicians evidence-based clinical practice
guidelines (8th edition). Chest 2008; 133: 160S – 198S.
2. Fuster V, Rydén LE, Cannom DS, Crijns HJ, Curtis AB, Ellenbogen
KA, et al. ACC/AHA/ESC 2006 guidelines for the management of
patients with atrial fibrillation: Executive summary: A report of the
American College of Cardiology/American Heart Association Task
Force on Practice Guidelines and the European Society of Cardiol-
ogy Committee for Practice Guidelines (Writing Committee to Re-
vise the 2001 Guidelines for the Management of Patients With Atrial
Fibrillation): Developed in collaboration with the European Heart
Rhythm Association and the Heart Rhythm Society. Circulation 2006;
114: 700 – 752.
3. Patel MR, Mahaffey KW, Garg J, Pan G, Singer DE, Hacke W, et al.
Rivaroxaban versus warfarin in nonvalvular atrial fibrillation. N Engl
J Med 2011; 365: 883 – 891.
4. Japanese Circulation Society Joint Working Group. Guidelines for
pharmacotherapy of atrial fibrillation (JCS 2008): Digest version.
Circ J 2010; 74: 2479 – 2500.
Circulation Journal Vol.76, September 2012
2111
5. Atarashi H, Inoue H, Okumura K, Yamashita T, Kumagai N, Origasa
H. Present status of anticoagulation treatment in Japanese patients
with atrial fibrillation. Circ J 2011; 75: 1328 – 1333.
6. Cheung CM, Tsoi TH, Huang CY. The lowest effective intensity of
prophylactic anticoagulation for patients with atrial fibrillation.
Cerebrovasc Dis 2005; 20: 114 – 119.
7. Yasaka M, Minematsu K, Yamaguchi T. Optimal intensity of inter-
national normalized ratio in warfarin therapy for secondary preven-
tion of stroke in patients with non-valvular atrial fibrillation. Intern
Med 2001; 40: 1183 – 1188.
8. Rosendaal FR, Cannegieter SC, van der Meer FJ, Briet E. A method
to determine the optimal intensity of oral anticoagulant therapy.
Thromb Haemost 1993; 69: 236 – 239.
9. Hirsh J. Oral anticoagulant drugs. N Engl J Med 1991; 324: 1865 –
1875.
10. Hori M, Connolly SJ, Ezekowitz MD, Reilly PA, Yusuf S, Wallentin
L. Efficacy and safety of dabigatran vs. warfarin in patients with atrial
fibrillation: Sub-analysis in Japanese population in RE-LY trial. Circ J
2011; 75: 800 – 805.
11. Wallentin L, Yusuf S, Ezekowitz MD, Alings M, Flather M, Franzosi
MG, et al. Efficacy and safety of dabigatran compared with warfarin
at different levels of international normalised ratio control for stroke
prevention in atrial fibrillation: An analysis of the RE-LY trial. Lan-
cet 2010; 376: 975 – 983.
12. Amouyel P, Mismetti P, Langkilde LK, Jasso-Mosqueda G, Nelander
K, Lamarque H. INR variability in atrial fibrillation: A risk model for
cerebrovascular events. Eur J Intern Med 2009; 20: 63 – 69.
13. Fox KAA, Piccini JP, Wojdyla D, Becker RC, Halperin JL, Nessel
CC, et al. Prevention of stroke and systemic embolism with rivar-
oxaban compared with warfarin in patients with non-valvular atrial
fibrillation and moderate renal impairment. Eur Heart J 2011; 32:
2387 – 2394.
Supplementary Files
Supplementary File 1
Data S1.   Online Supplement
Table S1.   Definitions of Outcome and Endpoint Components
Table S2.   Renal Impairment Subgroup Analysis for the Principal
Safety Outcome and Primary Efficacy Endpoint
Table S3.   Proportion of INR Values Spent Within Therapeutic (Tar-
get) Range in Warfarin-Receiving Patients
Table S4.   Treatment Comparisons for the Principal Safety Outcome
According to Center Time in Therapeutic Range in the Safety Popu-
lation
Table S5.   Treatment Comparisons for the Primary Efficacy Endpoint
According to Center Time in Therapeutic Range in the Safety Popu-
lation (on Treatment Analysis)
Table S6.   Bleeding Event Rates at Different Levels of INR Control
in Japanese and Chinese Patients With AF
Table S7.   Key Trial Data From the Global ROCKET AF and J­ROCKET
AF Studies
Figure S1.   Patient disposition.
Figure S2.   Primary efficacy endpoint observed in different popula-
tions and analysis periods.
Please find supplementary file(s);
http://dx.doi.org/10.1253/circj.CJ-12-0454