2015 Volume 79 Issue 12 Pages 2560-2567
Background: In 21,105 patients with atrial fibrillation in the ENGAGE AF-TIMI 48 trial, edoxaban was non-inferior to warfarin in preventing thromboembolic events while reducing bleeding. We compared results in Japan with the rest of East Asia (EA), including China, Korea, and Taiwan.
Methods and Results: We compared baseline characteristics, time-in-therapeutic range (TTR) for warfarin, and outcomes (efficacy: stroke or systemic embolic events [SEE], safety: major bleeding). Interaction P values were used to assess for effect modification of treatment (higher-dose edoxaban [HDE, 60 mg/30 mg] vs. warfarin; lower-dose edoxaban [LDE, 30 mg/15 mg] vs. warfarin) by region with adjustments for baseline characteristics. Fewer patients in Japan (n=1,010) were female, taking aspirin or amiodarone, naïve to warfarin (P<0.001 for each), had a history of stroke or transient ischemic attack (P=0.02), and more patients needed dose reduction (P<0.001) compared with EA (n=933). The mean TTR was higher in Japan (70% vs. 56%, P<0.001). Evidence for statistical interactions was observed for HDE vs. warfarin by region for stroke/SEE (adjusted P-int=0.052) and major bleeding (adjusted P-int=0.048) with greater relative efficacy and safety with HDE in EA compared with Japan. No interactions were observed for LDE vs. warfarin after adjustment.
Conclusions: HDE had a greater relative efficacy and safety in EA compared with Japan that was only partially explained by differences in baseline characteristics and TTR. (Circ J 2015; 79: 2560–2567)
Atrial fibrillation (AF) is a common arrhythmia that can lead to devastating complications such as stroke.1 Warfarin has been the mainstay in stroke prevention for the past 5 decades; however, several recent phase 3 clinical trials have been conducted with non-vitamin K antagonist oral anticoagulants (NOACs), with promising results.2–5 A recent meta-analysis combining 4 such trials displayed significant reduction in stroke or systemic embolic events (SEE), all-cause mortality and intracranial hemorrhage.6 The largest of these was the ENGAGE AF-TIMI 48 trial, which enrolled 21,105 patients with moderate-high risk AF and demonstrated that 2 once-daily edoxaban regimens (higher- and lower-dose edoxaban) were at least as efficacious as warfarin in reducing stroke or SEE while reducing major bleeding and cardiovascular mortality.3 Prior studies reported regional differences in terms of the safety and efficacy of NOACs, but similar data from ENGAGE AF-TIMI 48 have not been published.7,8 Regional variations in the genetic background of the population, socioeconomic and medical infrastructure, and unique regional patient characteristics may affect the relative efficacy and safety of the NOACs. However, it has been challenging to discern which factors explain regional differences, particularly when genetic factors are not controlled for. In contrast, it has been reported that East Asian (EA) populations are not genetically different in terms of drug disposition, metabolism, and elimination.9 In this context, the ENGAGE AF-TIMI 48 trial provides a unique opportunity to minimize the effects of a variable genetic background in EA because it enrolled a large and balanced number of patients in Japan and other EA countries.10 We investigated if there were regional differences in Japan compared with EA in the relative efficacy and safety of edoxaban compared with warfarin in the ENGAGE AF-TIMI 48 trial.
Editorial p 2537
ENGAGE AF-TIMI 48 was a multinational, randomized, double-blind, double-dummy trial that compared higher-dose once-daily edoxaban (HDE: 60/30 mg daily), lower-dose once-daily edoxaban (LDE: 30/15 mg daily) and warfarin with continual dose adjustment. Although Japanese guidelines recommend a target international normalized ratio (INR) between 1.6 and 2.6 for patients aged ≥70 years,11,12 the ENGAGE AF-TIMI 48 trial titrated warfarin to an INR of 2.0–3.0 in all of the participating countries, including Japan. Randomization was stratified by CHADS2 score (2–3 vs. 4–6) and factors affecting drug exposure that required edoxaban dose adjustment. The median follow-up period was 2.8 years. The doses of edoxaban were reduced by 50% in both edoxaban treatment arms in patients who had anticipated increased drug exposure caused by renal dysfunction (estimated creatinine clearance between 30 and 50 ml/min), low body weight (≤60 kg), or concomitant therapy with a strong P-glycoprotein inhibitor. Details of the trial design have been published previously.3,13 The time-in-therapeutic range (TTR) was calculated for each subject randomized to warfarin using the linear interpolation method with INR values rounded to the nearest 0.1, as described previously.14,15 Because there was a substantial difference in TTR between Japan and the other EA countries (China, Korea, and Taiwan), we compared results in Japan with those of the remaining EA countries. CHADS2 and HAS-BLED scores were calculated according to previously published methods.16,17
Study PopulationThe ENGAGE trial enrolled patients who were 21 years or older, had documented AF and a CHADS2 score of 2 or higher, and were scheduled to have anticoagulation therapy during the study period. The trial excluded patients with AF caused by a reversible disorder, an estimated creatinine clearance <30 ml/min, a high risk of bleeding, use of dual-antiplatelet therapy, moderate-to-severe mitral stenosis, other indications for anticoagulation, acute coronary syndromes/coronary revascularization/stroke within 30 days before randomization, and an inability to adhere to study procedures.
Efficacy and Safety OutcomesThe primary efficacy endpoint was a composite of stroke or SEE. The principal safety endpoint was major bleeding as defined by the International Society on Thrombosis and Haemostasis.18 An independent clinical endpoint committee, blinded to treatment assignment, adjudicated all deaths, cerebrovascular events, myocardial infarctions, SEE, and bleeding events. Details of the definitions used by the clinical endpoint committee were previously published in the protocol.3
Statistical AnalysisWe compared the main outcomes by randomized treatment in EA vs. Japan in the modified intention-to-treat population during the on-treatment period. The on-treatment period was defined as terminating either 3 days after the receipt of the last dose or at the end of the double-blind therapy treatment period, whichever came first. Interval censoring of events was performed if the study drug administration was interrupted for more than 3 days. Survival free of the endpoints was analyzed by the Kaplan-Meier method using the log-rank test. Multivariate models were constructed using backward selection with a threshold P-value of 0.10, forcing the randomized treatment, region, and an interaction term between treatment and region into the model. All the covariate candidate variables shown in Table 1 were included in the multivariate analysis. Results are presented as hazard ratios (HRs) with 2-sided 95% confidence intervals (CIs) based on a Cox proportional hazard model. All statistical analyses were performed with SAS software (version 9.2, SAS Institute Inc, Cary, NC, USA).
Characteristic | Japan | Other East Asia (China, Korea, Taiwan) | P value* | ||||||
---|---|---|---|---|---|---|---|---|---|
Warfarin (n=337) |
Higher- dose edoxaban (n=336) |
Lower- dose edoxaban (n=337) |
All (n=1,010) |
Warfarin (n=307) |
Higher- dose edoxaban (n=310) |
Lower- dose edoxaban (n=316) |
All (n=933) |
||
Age (years, median [interquartile range]) |
71 [65–77] | 72 [66–77] | 72 [66–77] | 72 [66–77] | 70 [64–76] | 72 [64–76] | 70 [64–76] | 71 [64–76] | 0.001 |
Female sex, n (%) | 74 (22) | 67 (20) | 79 (23) | 220 (22) | 103 (34) | 109 (35) | 113 (36) | 325 (35) | <0.0001 |
Paroxysmal AF, n (%) | 66 (20) | 56 (17) | 74 (22) | 196 (19) | 60 (20) | 66 (21) | 51 (16) | 177 (19) | 0.81 |
Qualifying risk factor, n (%) | |||||||||
Age ≥75 years | 125 (37) | 140 (42) | 134 (40) | 399 (40) | 102 (33) | 117 (38) | 111 (35) | 330 (35) | 0.06 |
Prior stroke or transient ischemic attack |
126 (37) | 135 (40) | 142 (42) | 403 (40) | 146 (48) | 130 (42) | 145 (46) | 421 (45) | 0.02 |
Congestive heart failure | 157 (47) | 154 (46) | 150 (45) | 461 (46) | 157 (51) | 148 (48) | 154 (49) | 459 (49) | 0.12 |
Diabetes mellitus | 128 (38) | 120 (36) | 113 (34) | 361 (36) | 99 (32) | 113 (37) | 107 (34) | 319 (34) | 0.47 |
Hypertension requiring treatment |
287 (85) | 275 (82) | 279 (83) | 841 (83) | 246 (80) | 256 (83) | 253 (80) | 755 (81) | 0.18 |
CHADS2 score (mean±standard deviation) |
2.8±0.9 | 2.9±1.0 | 2.8±1.0 | 2.8±1.0 | 2.9±1.0 | 2.9±1.0 | 2.9±1.0 | 2.9±1.0 | 0.18 |
≤3, n (%) | 265 (79) | 256 (76) | 259 (77) | 780 (77) | 227 (73) | 239 (77) | 241 (76) | 707 (76) | 0.45 |
HAS-BLED score (mean±standard deviation) |
2.3±1.1 | 2.2±1.1 | 2.3±1.2 | 2.3±1.2 | 1.7±0.9 | 1.8±1.0 | 1.8±1.0 | 1.8±1.0 | <0.0001 |
≥3, n (%) | 134 (39.8) | 123 (36.6) | 129 (38.3) | 386 (38.2) | 58 (18.9) | 65 (21.0) | 71 (22.5) | 194 (20.8) | <0.0001 |
Dose reduction at randomization, n (%) |
172 (51) | 171 (51) | 172 (51) | 515 (51) | 129 (42) | 125 (40) | 143 (45) | 397 (43) | 0.0002 |
Creatinine clearance ≤50 ml/min |
104 (31) | 111 (33) | 112 (33) | 327 (32) | 86 (28) | 86 (28) | 84 (27) | 256 (27) | 0.02 |
Weight ≤60 kg | 111 (33) | 113 (34) | 113 (34) | 337 (33) | 84 (27) | 79 (26) | 94 (30) | 257 (28) | 0.005 |
Use of verapamil or quinidine | 39 (12) | 38 (11) | 35 (10) | 112 (11) | 5 (1.6) | 7 (2.3) | 4 (1.3) | 16 (1.7) | <0.0001 |
Previous use of vitamin K antagonist for ≥60 days, n (%) |
292 (89) | 268 (80) | 269 (80) | 829 (82) | 101 (33) | 107 (35) | 116 (37) | 324 (35) | <0.0001 |
Medications at time of randomization, n (%) | |||||||||
Aspirin | 77 (23) | 72 (21) | 81 (24) | 230 (23) | 95 (31) | 112 (36) | 106 (34) | 313 (34) | <0.0001 |
Thienopyridine | 13 (3.9) | 13 (3.9) | 12 (3.6) | 38 (3.8) | 8 (2.6) | 10 (3.2) | 4 (1.3) | 22 (2.4) | 0.07 |
Amiodarone | 5 (1.5) | 4 (1.2) | 4 (1.2) | 13 (1.3) | 27 (8.8) | 22 (7.1) | 23 (7.3) | 72 (7.7) | <0.0001 |
Digoxin or digitalis preparation | 101 (30) | 110 (33) | 84 (25) | 295 (29) | 96 (31) | 103 (33) | 82 (26) | 281 (30) | 0.66 |
Student’s t-test used for continuous variables and Chi-square test for discrete variables. *Compares all Japan and all Other East Asia. AF, atrial fibrillation.
Among the 21,105 patients who were randomized into the ENGAGE AF-TIMI 48 trial, there were 1,010 patients enrolled in Japan and 933 in EA (China 469, Korea 230, Taiwan 234; Table S1). Patients in Japan were older and fewer were female, taking aspirin or amiodarone at baseline, naïve to warfarin (P<0.001 for each), and had a prior history of stroke or transient ischemic attack (P=0.02). Mean HAS-BLED score was higher in Japan than in EA (P<0.001). A larger number of patients enrolled in Japan needed dose reduction (P<0.001) compared with patients in EA (Table 1).
Data on TTR are shown in Table 2. The mean TTR (INR 2.0–3.0) was significantly higher in Japan (70±15%) compared with EA (56±21%, P<0.001). The results were similar when the therapeutic range was extended to INR 1.8–3.2 (84±13% in Japan vs. 70±22% in EA, P<0.001). Analysis of temporal changes in TTR in each country showed that TTR was higher in Japan than in any other country from the beginning of the study, and reached a plateau at approximately 6 months. The TTR in Japan remained higher than in EA during the rest of the study period (Figure S1).
INR | <1.5 | <2 | 2–3 | >3 | >4 | >5 | 1.8–3.2 | ||
---|---|---|---|---|---|---|---|---|---|
Japan | n=336 | Mean TTR | 3.2±7.1 | 23.6±15.4 | 70.3±15.3 | 6.1±1.9 | 0.5±1.9 | 0.1±0.3 | 84.3±13.0 |
China, Korea, Taiwan | n=293 | Mean TTR | 11.4±19.5 | 33.8±22.1 | 56.0±20.9 | 10.2±10.1 | 1.6±4.2 | 0.4±1.9 | 70.4±22.0 |
China | n=154 | Mean TTR | 11.1±19.3 | 35.3±22.5 | 55.7±21.0 | 9.0±9.3 | 1.2±3.0 | 0.2±0.9 | 70.4±22.1 |
Korea | n=63 | Mean TTR | 10.8±17.8 | 32.7±20.7 | 55.9±20.3 | 11.4±10.9 | 2.5±6.1 | 0.6±3.2 | 70.0±21.4 |
Taiwan | n=76 | Mean TTR | 12.5±21.5 | 31.8±22.6 | 56.7±21.4 | 11.6±10.7 | 1.8±4.2 | 0.4±2.0 | 70.7±22.4 |
P value* | <0.0001 | <0.0001 | <0.0001 | <0.0001 | <0.0001 | 0.004 | <0.0001 |
*Compares Japan with the rest of East Asia (China, Korea and Taiwan). Student’s t-test used for comparisons. INR, international normalized ratio; TTR, time-in-therapeutic range.
HDE vs. Warfarin Patients in EA randomized to HDE had a significantly lower rate of stroke/SEE (1.2%/year) as compared with warfarin (4.1%/year; HR 0.31; 95% CI 0.14–0.68; P=0.004). In contrast, patients in Japan had similar event rates regardless of the randomized treatment (1.5%/year with HDE vs. 1.6%/year with warfarin; HR 0.95; 95% CI 0.44–2.09; P=0.91) (Table 3A, Figure 1A). In the univariate analysis (including only treatment, region and interaction term as covariates), there was no significant effect modification of geographic region on treatment effect (P for interaction (P-int)=0.13, Table 3A). However, on multivariate analysis there was a borderline significant interaction, suggesting an enhanced relative benefit of HDE among patients in EA compared with Japan (adjusted P-int=0.052).
Comparison of stroke/systemic embolic event rates between (A) higher-dose edoxaban and warfarin and (B) lower-dose edoxaban and warfarin, stratified by region. P-int, P for interaction.
(A) Endpoint | Warfarin | Higher-dose edoxaban | Higher-dose edoxaban vs. warfarin |
Interaction P value |
Adjusted interaction P value |
|||
---|---|---|---|---|---|---|---|---|
Event rate | % of patients/ year |
Event rate | % of patients/ year |
Hazard ratio (95% CI) |
P value | |||
Stroke/SEE | ||||||||
Other East Asia | 25/304 | 4.07 | 8/306 | 1.17 | 0.31 (0.14–0.68) | 0.004 | 0.13 | 0.052 |
Japan | 13/337 | 1.56 | 12/336 | 1.47 | 0.95 (0.44–2.09) | 0.91 | ||
Major bleeding | ||||||||
Other East Asia | 35/304 | 5.85 | 15/306 | 2.23 | 0.39 (0.21–0.71) | 0.002 | 0.25 | 0.048 |
Japan | 33/337 | 4.03 | 27/336 | 3.38 | 0.84 (0.51–1.40) | 0.51 | ||
CRNM bleeding | ||||||||
Other East Asia | 91/304 | 17.88 | 63/306 | 10.39 | 0.60 (0.44–0.83) | 0.002 | 0.021 | 0.015 |
Japan | 112/337 | 16.46 | 107/336 | 15.99 | 0.98 (0.75–1.28) | 0.86 | ||
Major or CRNM bleeding | ||||||||
Other East Asia | 117/304 | 23.69 | 73/306 | 12.17 | 0.53 (0.40–0.72) | <0.001 | <0.01 | <0.01 |
Japan | 131/337 | 19.67 | 129/336 | 19.83 | 1.01 (0.79–1.30) | 0.91 | ||
Any overt bleeding | ||||||||
Other East Asia | 126/304 | 26.63 | 86/306 | 15.06 | 0.59 (0.45–0.78) | <0.001 | <0.01 | <0.01 |
Japan | 155/337 | 24.63 | 145/336 | 23.29 | 0.95 (0.76–1.19) | 0.66 | ||
Intracranial bleeding | ||||||||
Other East Asia | 15/304 | 2.42 | 4/306 | 0.58 | 0.24 (0.08–0.73) | 0.01 | 0.54 | 0.47 |
Japan | 13/337 | 1.55 | 5/336 | 0.61 | 0.39 (0.14–1.09) | 0.07 | ||
(B) Endpoint | Warfarin | Lower-dose edoxaban | Lower-dose edoxaban vs. warfarin |
Interaction P value |
Adjusted interaction P value |
|||
No. of patients | % of patients/ year |
No. of patients | % of patients/ year |
Hazard ratio (95% CI) |
P value | |||
Stroke/SEE | ||||||||
Other East Asia | 25/304 | 4.07 | 20/315 | 2.84 | 0.71 (0.39–1.27) | 0.25 | 0.044 | 0.19 |
Japan | 13/337 | 1.56 | 18/337 | 2.24 | 1.46 (0.72–2.99) | 0.30 | ||
Major bleeding | ||||||||
Other East Asia | 35/304 | 5.85 | 10/315 | 1.42 | 0.25 (0.12–0.51) | <0.001 | 0.061 | 0.24 |
Japan | 33/337 | 4.03 | 14/337 | 1.74 | 0.44 (0.24–0.82) | 0.010 | ||
CRNM bleeding | ||||||||
Other East Asia | 91/304 | 17.88 | 51/315 | 7.76 | 0.45 (0.32–0.63) | <0.001 | 0.022 | 0.020 |
Japan | 112/337 | 16.46 | 87/337 | 12.26 | 0.75 (0.57–0.995) | 0.046 | ||
Major or CRNM bleding | ||||||||
Other East Asia | 117/304 | 23.69 | 58/315 | 8.91 | 0.39 (0.28–0.53) | <0.001 | <0.01 | <0.01 |
Japan | 131/337 | 19.67 | 97/337 | 13.92 | 0.72 (0.56–0.94) | 0.014 | ||
Any overt bleeding | ||||||||
Other East Asia | 126/304 | 26.63 | 70/315 | 11.00 | 0.43 (0.32–0.58) | <0.001 | <0.01 | <0.01 |
Japan | 155/337 | 24.63 | 117/337 | 17.56 | 0.72 (0.57–0.92) | 0.008 | ||
Intracranial bleeding | ||||||||
Other East Asia | 15/304 | 2.42 | 5/315 | 0.70 | 0.29 (0.11–0.82) | 0.02 | 0.49 | 0.51 |
Japan | 13/337 | 1.55 | 2/337 | 0.24 | 0.16 (0.04–0.71) | 0.02 |
CI, confidence interval; CRNM, clinically relevant non-major; SEE, systemic embolic event.
LDE vs. Warfarin There were no statistical differences between LDE and warfarin in either EA (HR 0.71; 95% CI 0.39–1.27; P=0.25) or Japan (HR 1.46; 95% CI 0.72–2.99; P=0.30) (Table 3B, Figure 1B). Because the effects were directionally opposite between regions, however, there was a significant statistical interaction in the univariate analysis (P-int=0.04, Table 3B). After multivariate adjustment, the interaction was no longer significant (adjusted P-int=0.19). Figure S2 shows the Kaplan-Meier curves for the primary efficacy endpoint for HDE, LDE and warfarin stratified by region.
Safety EndpointsHDE vs. Warfarin Patients in EA who were assigned to HDE had significantly less major bleeding than patients assigned to warfarin (HR 0.39; 95% CI 0.21–0.71; P=0.002), whereas there was no significant difference between these treatment arms in Japan (HR 0.84; 95% CI 0.51–1.40; P=0.51) (Table 3A, Figure 2A). With univariate analysis, there was no significant effect modification by region in terms of major bleeding (P-int=0.25). In the multivariate analysis, the interaction was significant (adjusted P-int=0.048). Similar results were observed for clinically relevant non-major (CRNM) bleeding, major/CRNM bleeding and overt bleeding in that patients in EA had a higher event rate with warfarin than with HDE (P<0.01 in each comparison). The interaction was statistically significant with regard to CRNM bleeding (P-int=0.02), major/CRNM bleeding (P-int<0.01) and overt bleeding (P-int<0.01), showing that patients in EA had enhanced protection from bleeding by HDE vs. warfarin as compared with Japan. The interaction remained statistically significant in the multivariate analysis in terms of CRNM bleeding (adjusted P-int=0.015), major/CRNM bleeding (adjusted P-int=0.0004) and overt bleeding (adjusted P-int=0.006). In terms of intracranial hemorrhage, the event rate was higher with warfarin than HDE in both Japan and EA, with no evidence of effect modification by region (Table 3A).
Comparison of major bleeding rates between (A) higher-dose edoxaban and warfarin and (B) lower-dose edoxaban and warfarin, stratified by region. P-int, P for interaction.
LDE vs. Warfarin There were significant differences between LDE and warfarin in the safety endpoints in both Japan and EA (Table 3B, Figure 2B). Furthermore, the interaction analyses suggested that there was effect modification by region, with patients in EA experiencing significantly increased relative safety with LDE as compared with patients in Japan (P-int=0.06 for major bleeding, and <0.02 for the other safety endpoints; Table 3B). The P-int in the multivariate analysis did not reach statistical significance for major bleeding (adjusted P-int=0.24), whereas it was highly significant for other safety endpoints such as CRNM bleeding (adjusted P-int=0.020), major/CRNM bleeding (adjusted P-int=0.003) and overt bleeding (P-int=0.004). The risk of intracranial hemorrhage was similarly reduced by LDE compared with warfarin in both regions, with non-significant P values for interaction (Table 3B). Figure S3 shows the Kaplan-Meier curves for the primary safety endpoint for HDE, LDE and warfarin stratified by region.
This regional analysis performed among patients enrolled within EA demonstrated significant differences in baseline characteristics and TTR between patients enrolled in Japan vs. the rest of EA (China, Korea, Taiwan). Furthermore, there were several important regional differences in terms of the relative efficacy and safety of edoxaban that persisted after multivariate adjustment for several outcomes. For the efficacy endpoint, there was a marginally significant interaction between treatment with HDE and region with the multivariate analysis, but no significant interaction in a similar multivariate analysis with LDE. Similarly, for major bleeding, the interaction between HDE and region was significant in the multivariate analysis, but not with LDE. For the other safety endpoints, however, the interactions were significant for both HDE and LDE. Given the similarity in genetic background between these regions in terms of medication disposition, metabolism and elimination,9 this study provides a unique insight with regard to the possible contribution of environmental regional differences on the relative risk and benefit of edoxaban compared with warfarin. Our findings suggest that non-genetic factors resulted in slightly better efficacy and safety results with HDE in China, Korea, and Taiwan as compared with Japan.
Differences in Baseline CharacteristicsThere were 2 important findings that may explain some of the regional differences in the relative efficacy and safety of HDE compared with warfarin. The first finding was that there were fewer patients who were naïve to warfarin in Japan compared with EA. Patients who were warfarin-naïve experienced greater relative efficacy with edoxaban in the main trial.19 The second was that the TTR was significantly higher in Japan. These findings suggest that patients with AF in Japan and the treating physicians were either more accustomed to using warfarin and the fine dose adjustments required to keep the INR within the therapeutic range20 or the patients had already settled into a stable warfarin dose at enrollment into the study. Supporting the former hypothesis, the TTR in Japan was higher than that in EA at the beginning of the study, and both reached a plateau approximately 6 months after enrollment (Figure S1). In addition, the finding that the TTR in Japan remained higher than in EA during the rest of the study period suggests that the latter hypothesis also holds true (Figure S1). The rates of patients above age 65 and those with chronic renal insufficiency were higher in Japan, which may explain the higher HAS-BLED score in Japan compared with EA.
Efficacy EndpointsA higher incidence of stroke/SEE was observed in patients in EA who were treated with warfarin, whereas the rates in EA and Japan were similar in patients treated with once-daily edoxaban (HDE or LDE). These observations are consistent with the hypothesis that worse outcomes with warfarin in EA are driven by a lower TTR, specifically a longer time with a subtherapeutic INR. Regarding the comparison between EA and Japan for once-daily edoxaban, the rates of efficacy endpoints did not differ in the 2 regions (1.17 vs. 1.47 %/year with HDE; 2.84 vs. 2.24 %/year with LDE). This may be attributed to the similar risk of thromboembolism as reflected by the comparable CHADS2 scores in EA and Japan. A clinical implication of these observations is that in regions where the majority of patients are naïve to warfarin and/or the achieved TTR is relatively low, once-daily edoxaban may have enhanced relative efficacy compared with warfarin. On the other hand, in countries such as Japan where the majority of patients had already been on warfarin for a long time at the start of the trial and could expect a high TTR level, our findings suggest that once-daily edoxaban, especially HDE, may have a similar efficacy profile compared with warfarin.
Safety EndpointsIn terms of the comparisons between once-daily edoxaban (either with HDE or LDE) and warfarin, patients in EA appear to have enhanced safety with once-daily edoxaban as compared with patients in Japan with regard to bleeding outcomes. These findings may be explained by (1) an increase in the bleeding rate in EA compared with Japan in the warfarin group, and (2) an approximately 30% decrease in bleeding complications in EA in comparison with Japan in the once-daily edoxaban groups. These findings were consistent across all the safety endpoints evaluated. The former finding can be explained by the observation that patients who were randomized to warfarin in EA spent significantly longer time with a supratherapeutic INR than did patients in Japan. This held true whether a supratherapeutic INR was defined as INR >3, >4, or >5. In contrast, the latter findings could be explained well by a significantly higher bleeding risk in Japan as displayed by HAS-BLED scores, given that edoxaban theoretically achieves similar anticoagulation activity in both regions as there is little difference in genetic background. These data provide important clinical insights. In countries such as Japan where one can expect a high TTR with less time spent in a supratherapeutic INR with warfarin, the difference in safety between HDE and warfarin may be less pronounced. In addition, HDE would offer an enhanced safety profile in regions such as EA where the expected TTR is relatively low. LDE would provide a better safety profile than warfarin in both regions, and the difference in the rate of bleeding complications between LDE and warfarin would be more pronounced in EA compared with Japan based on the significant interaction P values.
Safety and Efficacy of Edoxaban in Japan Compared With the Rest of the WorldAn important finding regarding the safety of HDE compared with warfarin is that there were no significant differences in any of the safety endpoints in Japan. This is in contrast with the results in the main trial (n=21,105), which showed a significant reduction in major bleeding between HDE and warfarin.3 Although this may represent an important finding in Japan, it deserves caution because this is a small subpopulation (n=1,010) that had low statistical power to detect a significant difference. Supporting this explanation, the point estimate of the HR for major bleeding in Japan was 0.84, which was similar to that in the main trial (0.80), but with a wider CI (0.51–1.40 in Japan vs. 0.71–0.91 in the main trial).3
With regard to the comparison between LDE and warfarin in Japan, the results of this substudy indicated there was no significant difference in the primary efficacy endpoint, whereas safety was enhanced with LDE. These observations were in agreement with the findings in the main trial.3
Possible Source of Effect ModificationThere are several factors other than the significant difference in TTR between EA and Japan that could have affected the relative efficacy and safety of edoxaban vs. warfarin. The increased risk of bleeding among patients enrolled in EA may be explained by greater prevalence of aspirin use in the EA countries compared with Japan. Dose reduction status and concomitant use of P-glycoprotein inhibitors can also affect the relative efficacy and safety of edoxaban compared with warfarin. However, the interaction P values for all the safety endpoints except one remained significant even after adjusting for variables such as aspirin use, dose reduction status and concomitant use of P-glycoprotein inhibitors.
In summary, patients in EA were more likely to be naïve to warfarin, had lower TTRs and spent a longer time with either a supratherapeutic or subtherapeutic INR than patients enrolled in Japan. Consequently, patients in EA treated with warfarin had higher rates of stroke/SEE and major bleeding compared with Japan. HDE had an enhanced relative efficacy and safety vs. warfarin in EA compared with Japan, and the effect modification was only partially explained by differences in background characteristics, TTR and prior warfarin use.
Y.J.S. was reimbursed by Daiichi Sankyo, Inc for attending a conference. T.Y. has received honoraria from Daiichi Sankyo, Boehringer Ingelheim, Bayer Healthcare, Pfizer, Bristol-Myers Squibb, Eisai, Tanabe-Mitsubishi, and Ono Pharmaceutical and grants from Daiichi Sankyo, Boehringer Ingelheim, and Tanabe-Mitsubishi. Y.K. has received honoraria from Daiichi Sankyo, Boehringer Ingelheim, Bayer, Bristol-Meyers Squibb, and Pfizer. T.K. and K.A. are employees of Daiichi Sankyo Co, Ltd, Tokyo, Japan. M.M. and S.S. report being employees of Daiichi Sankyo. M.M. also reports holding a pending patent related to the clinical properties of edoxaban. C.T.R. served as a consultant and received honoraria from Daiichi Sankyo, Boehringer Ingelheim, and Bayer, and reports grants (through the Brigham and Women’s Hospital) from Daiichi Sankyo. R.P.G. has received honoraria for consulting and/or CME lectures from Bristol-Myers Squibb, Janssen, Portola, Pfizer, Daiichi Sankyo, Merck & Co, and Sanofi-Aventis; and grant support through his institution from Daiichi Sankyo, Merck & Co, Johnson & Johnson, Sanofi-Aventis and AstraZeneca.
Supplementary File 1
Table S1. Patients’ demographic and clinical characteristics by region
Figure S1. Trend of time-in-therapeutic range in each country.
Figure S2. Kaplan-Meier curves of stroke/systemic embolic event for higher-dose edoxaban, lower-dose edoxaban and warfarin in (A) Japan and (B) the rest of East Asia.
Figure S3. Kaplan-Meier curves of major bleeding for higher-dose edoxaban, lower-dose edoxaban and warfarin in (A) Japan and (B) the rest of East Asia.
Please find supplementary file(s);
http://dx.doi.org/10.1253/circj.CJ-15-0574