Original Article. End of Study Transition From Study Drug to Open-Label Vitamin K Antagonist Therapy The ROCKET AF Experience

Original Article End of Study Transition From Study to Open-Label Vitamin K Antagonist Therapy The ROCKET AF Experience Kenneth W. Mahaffey, MD; Anne S. Hellkamp, MS; Manesh R. Patel, MD; Karen L. Hannan, BA; Kimberly Schwabe; Christopher C. Nessel, MD; Scott D. Berkowitz, MD; Jonathan L. Halperin, MD; Graeme J. Hankey, MD; Richard C. Becker, MD; Jonathan P. Piccini, MD; Günter Breithardt, MD; Werner Hacke, MD, PhD; Daniel E. Singer, MD; Robert M. Califf, MD; Keith A.A. Fox, MB, ChB Background To evaluate the previously reported excess of thromboembolic events during the 30 days after the end of study (EOS) visit when participants transitioned from blinded therapy to open-label vitamin K antagonist. Methods and Results At the EOS visit, open-label vitamin K antagonist was recommended, and the international normalized ratio (INR) was not to be measured until 3 days later to preserve blinding. We analyzed transition strategies, clinical outcomes, and INR values. Event rates are per 100 patient-years. A total of 9248 (65%) participants were taking study drug at EOS, and, between days 3 and 30, an excess of stroke and systemic embolic events were observed in participants assigned to rivaroxaban (rivaroxaban 22 events, event rate 6.42; warfarin 6 events, event rate 1.73; hazard ratio, 3.72; 95% confidence interval, 1.51 9.16; P=0.0044). No INR values were reported for 5% of participants transitioned to warfarin. By 30 days after EOS, 83% of the warfarin group and 52% of the rivaroxaban group had 1 therapeutic INR value. Median time to first therapeutic INR was 3 days in the warfarin group and 13 days in the rivaroxaban group. Conclusions The excess of events at EOS was likely because of a period of inadequate anticoagulation in rivaroxaban participants switched to vitamin K antagonist therapy. If transition from rivaroxaban to vitamin K antagonist is needed, timely monitoring and careful dosing should be used to ensure consistent and adequate anticoagulation. (Circ Cardiovasc Qual Outcomes. 2013;6:00-00.) Key Words: atrial fibrillation bridging discontinuations factor Xa randomized controlled trial rivaroxaban stroke Atrial fibrillation (AF) is common, and the prevalence will increase as the population ages. 1 Patients with AF often have other comorbidities and undergo invasive cardiac and noncardiac procedures that require interruption of anticoagulant therapy. Routine clinical care procedures can be either elective or urgent, requiring different approaches to management of anticoagulant therapy with rapid reversal or careful planning with potential bridging therapy. In clinical trials, interruption of study drug therapy for similar procedures is anticipated, and guidelines or recommendations are provided for site investigators. In contrast to clinically mandated interruptions during the trial, at the end of the blinded Rivaroxaban Once-Daily, Oral, Direct Factor Xa Inhibition Compared With Vitamin K Antagonism for Prevention of Stroke and Embolism Trial In Atrial Fibrillation (ROCKET AF) study, there was a necessary but artificial discontinuation of study therapy and a necessary transition to open-label vitamin K antagonist (VKA) therapy. ROCKET AF was a double-blind, event-driven trial comparing rivaroxaban with warfarin in patients with nonvalvular AF at high risk of stroke and noncentral nervous system embolism. 2,3 When the requisite number of events was accrued, site investigators and coordinators were notified to perform an end of study (EOS) visit, during which information was collected, and participants were to be transitioned to open-label VKA therapy per local physician decision. An excess of stroke and noncentral nervous system embolic events were observed in participants randomly assigned rivaroxaban compared with those who were randomly assigned warfarin during the 30 days after discontinuation of study drug in participants still Received October 16, 2012; accepted May 13, 2013. From the Duke Clinical Research Institute (K.M.W., A.S.H., M.R.P., K.L.H., R.C.B., J.P.P.) and Duke Translational Medicine Institute (R.M.C.), Duke University Medical Center, Durham, NC; Johnson & Johnson Pharmaceutical Research and Development, Raritan, NJ (K.S., C.C.N.); Bayer HealthCare Pharmaceuticals, Montville, NJ (S.D.B.); The Cardiovascular Institute, Mount Sinai Medical Center, New York, NY (J.L.H.); Department of Neurology, Royal Perth Hospital, Perth, Australia (G.J.H.); Department of Cardiology and Angiology, Hospital of the University of Münster, Münster, Germany (G.B.); Department of Neurology, Ruprecht-Karls-University, Heidelberg, Germany (W.H.); Clinical Epidemiology Unit, General Medicine Division, Massachusetts General Hospital, and Harvard Medical School, Boston, MA (D.E.S.); and University of Edinburgh, Royal Infirmary of Edinburgh, Edinburgh, UK (K.A.A.F.). Correspondence to Kenneth W. Mahaffey, MD, Duke Clinical Research Institute, Duke University Medical Center, Room 0311 Terrace Level, 2400 Pratt St, Durham, NC 27705. E-mail kenneth.mahaffey@duke.edu 2013 American Heart Association, Inc. Circ Cardiovasc Qual Outcomes is available at http://circoutcomes.ahajournals.org DOI: 10.1161/CIRCOUTCOMES.113.000132 1

2 Circ Cardiovasc Qual Outcomes July 2013 WHAT IS KNOWN An excess of thromboembolic events occurred in the ROCKET AF trial during the 30 days after the end of study visit when participants transitioned from blinded therapy to open-label vitamin K antagonist. The excess in events during the transition phase resulted in a US Food and Administration black box warning in the prescribing information for rivaroxaban cautioning providers that discontinuation could increase the risk of thrombotic events. WHAT THE STUDY ADDS These analyses provide clinicians with a possible explanation for the excess in events at the end of the study. Based on these data, clinicians transitioning a patient with atrial fibrillation from rivaroxaban to a vitamin K antagonist should do so with careful dosing and monitoring to ensure consistent and adequate anticoagulation. on their assigned therapy at the EOS. 3 The US Food and Administration issued a black box warning in the prescribing information for rivaroxaban, cautioning providers that discontinuing rivaroxaban could increase the risk of thrombotic events. 4 The analyses presented here further evaluate the ROCKET AF findings and provide potential explanations. Methods The ROCKET AF study has been described previously. 2,3 Briefly, the study was a multicenter, randomized, double-blind, double-dummy, event-driven trial comparing rivaroxaban (20 mg daily; 15 mg daily in participants with creatinine clearance 30 49 ml/min) with adjusted-dose warfarin (target international normalized ratio [INR], 2.0 3.0) for prevention of stroke or systemic embolism. 2,3 The study was supported by Johnson & Johnson Pharmaceutical Research & Development (Raritan, NJ) and Bayer HealthCare AG (Leverkusen, Germany), and an international executive committee designed the study and oversaw the trial conduct. The protocol was approved by national regulatory authorities and ethics committees at participating centers. All participants provided written consent. Study Participants and Follow-up Participants with nonvalvular AF at moderate to high risk of stroke were recruited at 1178 participating sites in 45 countries. Participants who were included had a history of stroke, transient ischemic attack, or systemic embolism or 2 (capped at 10%) or 3 of the following: heart failure or left ventricular ejection fraction 35%, hypertension, 75 years of age, or diabetes mellitus (CHADS 2 score 2). Participants were followed up monthly for assessment of clinical outcomes and adverse events and monitoring of INR values while on therapy. If study drug was discontinued before the EOS, participants were to be contacted by telephone every 3 months for ascertainment of events. An independent clinical end point committee applied protocol definitions to adjudicate all suspected stroke, systemic embolism, myocardial infarction, death, and bleeding events, including those after the EOS visit, although blinded to treatment assignment. EOS Recommendations Recommendations for EOS transition to open-label anticoagulation were provided to site investigators and coordinators through Webinars, frequently asked question documents, and a clinical helpline. If openlabel VKA was indicated, investigators were instructed to start that therapy on the day of the EOS visit and to simultaneously discontinue study medication. To maintain the study blind, an unblinded INR was to be collected on the third day after the EOS visit. If the INR was subtherapeutic, investigators were encouraged to consider bridging therapy with enoxaparin or unfractionated heparin, if the participant was considered high risk for thromboembolism according to local standards and practices based on the unblinded INR. Investigators were advised to carefully select the appropriate dose of VKA that would obtain an INR of 2.0 to 3.0 or to consider using a dose similar to that used before entry in the trial. Investigators could also call a clinical helpline for assistance if a nonwarfarin VKA was going to be used. Investigators were instructed to submit all INR values gathered during the transition phase and report all adverse events and safety and efficacy events for adjudication. All participants still on study drug at the end of the trial were to be seen 30 days after the EOS visit for a final study visit. Statistical Analysis By design, ROCKET AF was a noninferiority trial, and the primary analysis was performed in the per-protocol, on-treatment population. 2,3 These analyses were also performed on this population to explore the consequences of transition from study drug to standard clinical care at the end of the trial in participants still on study drug. The end of the on-treatment period was defined as the last study drug administration plus 2 days to account for continued drug effect. Therefore, the focus of these analyses is on days 3 to 30 after study drug discontinuation. Descriptive statistics of baseline characteristics are given as numbers and percentages or medians with 25th and 75th percentiles, as appropriate. Event rates are presented as per 100 patient-years. The primary end point for this analysis was the primary efficacy end point in the trial, the composite of any stroke or noncentral nervous system systemic embolism. Secondary end points included both efficacy and safety end points from the trial. With respect to the time to each end point, an unadjusted Cox proportional hazards model was used to evaluate the association between rivaroxaban versus warfarin and risk of event. Risk relationships are expressed as hazard ratios with 95% confidence intervals derived from the Cox models. The time to attainment of therapeutic INR ( 2) within 30 days after study drug discontinuation was compared between treatment arms with a logrank test. Analyses of INR values and comparison of event rates during the trial and after the EOS visit are considered exploratory. All statistical analyses were performed with SAS software (version 9.2, SAS Institute, Inc, Cary, NC). All subjects enrolled in the study gave informed consent. All appropriate national regulatory authorities and ethics committees at participating centers approved the study. Results Overall, 9248 (65%) of the 14 143 per-protocol, on-treatment participants in ROCKET AF were still taking study drug at the EOS visit. Baseline characteristics in participants on study drug at the EOS visit are shown in Table 1. Participants completing the study still on study drug tended to be younger, had better renal function, and had generally less comorbidity but had more previous stroke or transient ischemic attack, although the differences are minor. These characteristics were balanced between the treatment groups in participants on study drug at the EOS. At the EOS, 92.2% of participants still on study drug were transitioned to open-label VKA therapy within 30 days (rivaroxaban: 4232/4591 [92.2%]; warfarin: 4292/4657 [92.2%]). Bridging therapy with low molecular weight or unfractionated heparin was used infrequently (rivaroxaban: 83/4591 [1.8%]; warfarin: 56/4657 [1.2%]).

Mahaffey et al ROCKET AF: End of Study Transition to VKA Therapy 3 Table 1. Baseline Characteristics All Participants Enrolled (n=14 143) Participants on Study at EOS (n=9248) Rivaroxaban (n=4591) Warfarin (n=4657) Baseline characteristics Age, median, y 73 (65 78) 72 (64 77) 72 (64 77) Female, n (%) 5590 (39.5) 1815 (39.5) 1780 (38.2) Atrial fibrillation, n (%) Persistent 11 462 (81.0) 3763 (82.0) 3763 (80.8) Paroxysmal 2487 (17.6) 765 (16.7) 836 (18.0) Newly diagnosed 194 (1.4) 63 (1.4) 58 (1.2) Presenting characteristics BMI, median, kg/m 2 28.2 (25.1 32.0) 28.4 (25.3 32.2) 28.2 (25.2 32.0) Systolic BP, median, mm Hg 130 (120 140) 130 (120 140) 130 (120 140) Diastolic BP, median, mm Hg 80 (70 85) 80 (70 86) 80 (71 86) Previous aspirin use, n (%) 5461 (38.6) 1665 (36.3) 1712 (36.8) Previous VKA use, n (%) 8838 (62.5) 2900 (63.2) 3015 (64.7) CHADS 2 score, median 3 (3 4) 3 (3 4) 3 (3 4) Creatinine clearance, median, ml/min 67 (52 87) 70 (55 90) 70 (55 89) History, n (%) Stroke, TIA, or non-cns embolism 7750 (54.8) 2613 (56.9) 2639 (56.7) Hypertension 12 801 (90.5) 4137 (90.1) 4194 (90.1) Congestive heart failure 8837 (62.5) 2836 (61.8) 2845 (61.1) Diabetes mellitus 5635 (39.8) 1829 (39.8) 1801 (38.7) COPD 1477 (10.4) 422 (9.2) 390 (8.4) MI 2438 (17.2) 668 (14.6) 747 (16.0) PAD 829 (5.9) 226 (4.9) 263 (5.6) BMI indicates body mass index; BP, blood pressure; CNS, central nervous system; COPD, chronic obstructive pulmonary disease; EOS, end of study; MI, myocardial infarction; PAD, peripheral arterial disease; TIA, transient ischemic attack; and VKA, vitamin K antagonist. Figure 1 shows the stroke and systemic embolic events that occurred in participants still on study drug at the EOS visit. For days 3 to 30, which were defined as off treatment, there was an excess of events in the participants assigned to rivaroxaban compared with those assigned to warfarin (rivaroxaban 22 events, event rate 6.42; warfarin 6 events, event rate 1.73; hazard ratio, 3.72; 95% confidence interval [1.51 9.16]; P=0.0044). Three events occurred during the first 2 days after EOS and were not included in the analyses for 3 to 30 days (rivaroxaban 0 events; warfarin 3 events). The rates of stroke and systemic embolism within 30 days after the EOS visit were similar in participants who completed the study and did not transition to open-label VKA (n=724; 0.31%) compared with participants who did transition to open-label VKA Figure 1. Occurrence of stroke or systemic embolic events after end of study visit in those on study drug at end of study.

4 Circ Cardiovasc Qual Outcomes July 2013 Table 2. Characteristics of 28 Participants With Stroke or Non-CNS Embolism After EOS* Patient Age Sex Regions Baseline CHADS 2 History of Stroke, TIA, Randomized Embolism Treatment Transitioned to Open-Label VKA No. INR Tests in 30 Days No. IVRS INR Tests in 30 Days No. Local Laboratory INR Tests in 30 Days No. INR Tests 2 Within 30 Days Days From Stop to Therapeutic INR Bridging Therapy (UFH/LMWH) Within 30 Days 1 75 F North America 4 Y Rivaroxaban Y 2 1 1 0 na N 22 2 61 F Eastern 4 Y Warfarin Y 0 0 0 0 na N 20 3 79 F Eastern 5 Y Warfarin Y 3 1 2 2 1 N 16 4 81 M Western 3 N Rivaroxaban Y 4 1 3 0 na N 11 5 78 F Eastern 5 Y Rivaroxaban Y 17 1 16 8 8 Y 1 6 75 F Eastern 6 Y Rivaroxaban Y 1 1 0 1 1 N 22 7 72 F Eastern 3 N Warfarin Y 3 1 2 2 1 N 12 8 75 M Eastern 5 Y Rivaroxaban Y 7 1 6 6 6 N 11 9 85 F Eastern 5 Y Rivaroxaban Y 5 1 4 2 15 N 9 10 75 M Western 4 Y Rivaroxaban Y 7 1 6 3 6 N 3 11 81 F Eastern 5 Y Warfarin Y 2 1 1 1 1 N 24 12 84 F Latin America 4 Y Rivaroxaban Y 1 0 1 1 8 N 20 13 57 F Latin America 4 Y Rivaroxaban Y 2 1 1 0 na N 4 14 80 F Latin America 5 Y Rivaroxaban N 2 1 1 0 na N 1 15 57 M Latin America 5 Y Rivaroxaban Y 2 0 2 2 21 N 18 16 68 M Latin America 3 Y Rivaroxaban Y 2 0 2 1 20 N 18 17 70 F Latin America 2 Y Rivaroxaban Y 2 1 1 0 na N 11 18 65 M Latin America 3 Y Warfarin Y 4 1 3 0 na N 2 19 66 F Latin America 4 Y Rivaroxaban Y 3 1 2 1 9 N 27 20 75 F Latin America 5 Y Rivaroxaban Y 4 1 3 1 29 N 4 21 86 F Latin America 5 Y Warfarin Y 1 0 1 1 6 N 18 22 75 M Latin America 5 Y Rivaroxaban Y 2 0 2 2 12 N 3 23 71 M Asia/Pacific 4 Y Rivaroxaban Y 2 1 1 0 na N 27 24 47 M Asia/Pacific 3 N Rivaroxaban Y 4 1 3 0 na Y 10 25 71 F Asia/Pacific 2 Y Rivaroxaban Y 2 1 1 0 na N 9 26 75 F Asia/Pacific 3 N Rivaroxaban N 1 1 0 0 42 Y 10 27 50 M Eastern 5 Y Rivaroxaban Y 1 1 0 0 na N 23 28 67 M Eastern 4 Y Rivaroxaban Y 2 1 1 1 6 N 27 CNS indicates central nervous system; EOS, end of study; INR, international normalized ratio; IVRS, interactive voice response system; LMWH, low molecular weight heparin; TIA, transient ischemic attack; UFH, unfractionated heparin; and VKA, vitamin K antagonist. *For each patient, the following information needs to be provided: age; sex; region; baseline CHADS 2 ; history of stroke, TIA, or non-cns embolism; randomized treatment; open-label VKA treatment post-eos (yes/no); number of INR values collected from EOS to 30 days post-eos total and separately provided via IVRS and via local labs; number of INR values collected from EOS to 30 days post-eos that were 2.0; time to first INR 2.0; use of bridging therapy post-eos (yes/no). Days to Event (n=8524; 0.28%), although the number of participants and events is small. Detailed information for each of the 28 participants with stroke or systemic embolic events that occurred between 3 and 30 days after the EOS visit is shown in Table 2. The mean CHADS 2 score in these participants was higher than the overall trial (4.12 versus 3.48), and a higher proportion of participants had previous stroke, transient ischemic attack,

Mahaffey et al ROCKET AF: End of Study Transition to VKA Therapy 5 Table 3. HR and 95% CIs for Time to First Occurrence of All Efficacy End Points, and Safety End Points of Major and Minor Bleeding (All Adjudicated by CEC) From Day 3 to Day 30 After Last Dose for Participants Who Completed Study Medication Rivaroxaban Warfarin Rivaroxaban vs Warfarin End Points, n (%) n=4587 Event Rate (100 Pt-Y) n=4652 Event Rate (100 Pt-Y) HR (95% CI) P Value Primary efficacy* 22 (0.48) 6.42 6 (0.13) 1.73 3.72 (1.51 9.16) 0.004 Stroke type 22 (0.48) 6.42 6 (0.13) 1.73 3.72 (1.51 9.16) 0.004 Primary hemorrhagic 4 (0.09) 1.17 0 (0.00) 0.00 Primary ischemic 18 (0.39) 5.25 4 (0.09) 1.15 4.56 (1.54 13.5) 0.006 Unknown 0 (0.00) 0.00 2 (0.04) 0.58 Nondisabling stroke 5 (0.11) 1.46 4 (0.09) 1.15 1.26 (0.34 4.71) 0.726 Non-CNS systemic embolism 0 (0.00) 0.00 0 (0.00) 0.00 Myocardial infarction 1 (0.02) 0.29 2 (0.04) 0.58 0.50 (0.05 5.55) 0.575 All-cause mortality 14 (0.31) 4.08 8 (0.17) 2.30 1.77 (0.74 4.22) 0.197 Vascular death 12 (0.26) 3.49 7 (0.15) 2.01 1.73 (0.68 4.41) 0.247 Nonvascular death 2 (0.04) 0.58 0 (0.00) 0.00 Unknown death 0 (0.00) 0.00 9 (0.02) 0.29 Major bleeding 25 (0.55) 7.29 7 (0.15) 2.01 3.62 (1.56 8.36) 0.0026 Minor bleeding 3 (0.07) 0.87 6 (0.13) 1.73 0.51 (0.13 2.02) 0.33 CI indicates confidence interval; CNS, central nervous system; and HR, hazard ratio. *Primary efficacy end point is the composite of stroke and non-cns systemic embolism. Statistically significant at 0.05 (2 sided, not adjusted for multiplicity). or systemic embolism compared with the overall trial (86% versus 55%). Table 3 shows all efficacy and cardiovascular outcomes occurring between 3 and 30 days after the EOS visit in participants still on study drug at the EOS. Few myocardial infarctions occurred (rivaroxaban 1; warfarin 2). A total of 22 deaths were reported (rivaroxaban 14; warfarin 8), but the rates per 100 patient-years were low. None of the 4 hemorrhagic strokes in the rivaroxaban group occurred in participants treated with bridging therapy. Table 4 details the INR measurements reported in participants on study drug at the EOS and transitioned to open-label VKA therapy. Approximately 5% of participants in both treatment groups Table 4. Number of Local or IVRS INR Measurements for Participants Completing Treatment and Transitioning to VKA Post-Treatment Days 1 to 30 End Point Rivaroxaban Warfarin Stroke/SE Event Rate/100 Pt-Y (n) Major Bleed Event Rate/100 Pt-Y (n) Total participants 4242 4309 transitioned to VKA INR measurements 0 210 (5.0%) 236 (5.5%) 5.71 (2) 0.00 (0) 1 1218 (28.7%) 1298 (30.1%) 1.98 (4) 2.97 (6) 2 1273 (30.0%) 1406 (32.6%) 5.57 (12) 4.17 (9) 3 734 (17.3%) 789 (18.3%) 2.46 (3) 6.57 (8) >3 807 (19.0%) 580 (13.5%) 7.22 (8) 9.91 (11) INR indicates international normalized ratio; IVRS, interactive voice response system; SE, systemic embolism; and VKA, vitamin K antagonist. had no INR values reported. No relationship between the number of events or event rates and number of INR values is apparent. In participants with INR values reported, the cumulative proportion of participants with 1 INR value 2.0 during the first 30 days after the last dose of study drug was 83% for the warfarin group and 52% for the rivaroxaban group (P<0.0001; Figure 2A). Of the 5589 participants who transitioned to open-label VKA at the EOS visit having 2 INR measurements within 30 days of study drug discontinuation, 63.0% (1748/2775) of warfarin participants and 33.0% (928/2814) of rivaroxaban participants had 2 INRs 2 (P<0.0001; Figure 2B). Table 5 shows the distribution of the 28 stroke and systemic embolic events and event rates between 3 and 30 days after the EOS visit and the number of INR measurements available by geographic region. Regions with more INR assessments had numerically fewer stroke or systemic embolic events. In participants on study drug at the EOS, the median (25th 75th) time from last dose of study drug and start of open-label VKA therapy was 1 (1 2). A total of 92% of participants who were transitioned did so on the same day as the EOS visit or the day after. The transition time and proportions were balanced between the assigned study drug groups. Another period of transition of anticoagulation therapy was at the beginning of the trial. Of the 8834 (62%) participants on VKA therapy at the time of randomization, there were 20 strokes or systemic emboli during the first 30 days of the study after the transition from VKA therapy to blinded study drug (8 rivaroxaban, 12 warfarin). In the same population, major bleeding within 30 days was uncommon, and the rates were similar (19 rivaroxaban, 17 warfarin). The median screening INR in participants with major bleeding was 2.4 (1.7 2.7) compared with 1.9 (1.5 2.3) for participants with no bleeding.

6 Circ Cardiovasc Qual Outcomes July 2013 Figure 2. A, Cumulative proportion of participants on study drug at end of study visit transitioned to vitamin K antagonist and had 1 international normalized ratio measurement 2 within 30 days of end of study visit. B, Cumulative proportion of participants on study drug at end of study visit transitioned to vitamin K antagonist and had 2 international normalized ratio measurements 2 within 30 days of end of study visit. Figure 3 shows stroke and systemic embolic events in participants assigned to receive warfarin and not on VKA therapy at the screening visit. A total of 8 events occurred during the initiation of warfarin in this participant subgroup without a clustering of early events, suggesting EOS was a unique and artificial period of transition. Discussion In the ROCKET AF trial, 9248 (65%) participants remained on study drug until the end of the trial. After participants were seen at the EOS visit and transitioned to open-label VKA therapy at the discretion of their physician, an excess of stroke and systemic embolic events were observed in the subsequent 28 days in participants who had been assigned to receive rivaroxaban compared with those who had been assigned to receive warfarin. These analyses support the hypothesis that the excess of events (rivaroxaban, 22 versus warfarin, 6) was likely because of a period of inadequate anticoagulation in rivaroxaban participants transitioned to VKA therapy. The reason for the period of inadequate anticoagulation is not clear. The protocol intentionally did not stipulate a prescribed algorithm for transition; therefore, physicians could manage participants using local standards, including anticoagulation clinics or clinical practice algorithms for management of VKA therapy. However, site investigators were advised through the use of study memos, flyers, and live Webcasts about the appropriate transition to open-label VKA therapy. Lack of reporting of INR values rather than lack of measuring the values may have contributed because 5% of participants did not have an INR checked between 3 and 30 days after rivaroxaban was stopped at EOS. The regions with more frequent INR monitoring reported the fewest events, and, overall, there was no apparent relationship between number of INR values reported and event rates after the EOS visit. However, appropriate adjustment of therapy is a factor as well, and we were unable to explore this with the data collected. Importantly, although no evidence of a rebound phenomenon was observed, it is not entirely excluded. However, these data from ROCKET AF and related clinical trial experiences with rivaroxaban make rebound unlikely, although the populations and stroke risk are different in these other cohorts. Key observations, including the pattern of events

Mahaffey et al ROCKET AF: End of Study Transition to VKA Therapy 7 Table 5. Time to First Occurrence of Post-Treatment Events: Primary Efficacy End Point by Region Safety/Days 3 to 30 After Last Dose in Participants Who Completed Regions Rivaroxaban (n=4587) Warfarin (n=4652) HR (95% CI) North America Event, n (rate) 1 (1.75) 0 Participants, n 751 781 Any INR measured, n 714 728 Number of INR tests, median 2 (1 4) 2 (1 3) Latin America Event, n (rate) 9 (19.24) 2 (4.07) 4.71 (1.02 21.8) Participants, n 609 631 Any INR measured, n 554 560 Number of INR tests, median 2 (1 3) 2 (1 2) Western Event, n (rate) 2 (3.86) 0 Participants, n 665 658 Any INR measured, n 644 638 Number of INR tests, median 3 (1 4) 2 (1 4) Eastern Event, n (rate) 6 (4.36) 4 (2.87) 1.52 (0.43 5.38) Participants, n 1629 1665 Any INR measured, n 1549 1593 Number of INR tests, median 2 (1 3) 2 (1 3) Asia/Pacific Event, n (rate) 4 (8.11) 0 Participants, n 578 557 Any INR measured, n 562 537 Number of INR tests, median 2 (1 3) 2 (1 3) CI indicates confidence interval; HR, hazard ratio; and INR, international normalized ratio. with a lack of clustering in the days shortly after rivaroxaban cessation, the absence of excess events during other periods of the trial when participants either temporarily or permanently stopped rivaroxaban, 5 and the lack of a similar pattern of events in trials of rivaroxaban in deep vein thrombosis, pulmonary embolism, and after orthopedic surgery provide clinical evidence to support a lack of a rebound phenomenon. 6 Several trial methodologies have been used to censor events during long-term, event-driven trials, and no consensus exists on the best approach. Some clinical trialists have censored on the date the sites were notified (site notification) that the requisite number of events had occurred. Others have censored on the date the participant was seen after site notification and study therapy was discontinued (EOS visit), and others have censored at various intervals after the EOS visit, including 30 days, when the participant may be contacted a final time for ascertainment of safety or efficacy events poststudy drug cessation. For ROCKET AF, it was prespecified that primary analyses include events that occurred up to site notification, but we reported the events after site notification for full transparency in the primary publication. Investigators from 2 other large trials of AF have not reported information about events after EOS in publications to date. 7,8 Consistent approaches should be considered so that trial data are comparable and transparency is achieved. During ROCKET AF and before unblinding of the data, the trial leadership was informed about the preliminary results of the Japanese ROCKET trial and a modest excess in events after discontinuation of rivaroxaban compared with warfarin. 9 The trial leadership notified the ROCKET AF data safety monitoring board (DSMB) about these data from the Japanese- ROCKET trial to seek any recommendation about EOS transition. The DSMB informed the trial leadership that participants should have an unblinded INR within a very short time period after discontinuation of study medication, for example, 3 to 5 days (mean 4 days), to determine the adequacy of anticoagulation. The ROCKET AF executive committee considered several options for transition from rivaroxaban to open-label VKA therapy at the EOS visit. The approach implemented represented an attempt to maintain participant safety, provide continued effective anticoagulation, protect the study blind, and limit bias in reporting of events at the end of the trial. Discontinuation of anticoagulant therapy in patients with AF would be expected to result in an increase in the rate of thromboembolic events attributable to removal of effective therapy. The observed stroke and systemic embolic event rate Figure 3. Number of events in those with no vitamin K antagonist (VKA) at screening.

8 Circ Cardiovasc Qual Outcomes July 2013 after EOS in ROCKET AF was 6.42 and consistent with rates (5.4 7.5) reported in similar AF populations. 10,11 No clinical trial data are available to guide physicians when converting patients from rivaroxaban to warfarin. Rivaroxaban affects the INR, and, therefore, INR measurements may not accurately reflect the VKA effect when both VKA and rivaroxaban are present. The rivaroxaban package insert suggests that when a clinical situation requires discontinuation of rivaroxaban, the physician could discontinue XARELTO and begin both a parenteral anticoagulant and warfarin at the same time the next dose of XARELTO would have been taken. Alternatively, we suggest that rivaroxaban could be continued while warfarin is started, and rivaroxaban could be stopped when the INR measured at rivaroxaban trough is 2.0. The experience of overlap of rivaroxaban with warfarin at the beginning of the trial when VKA-experienced participants were randomly assigned to rivaroxaban with no excess of stroke, systemic embolic, or major bleeding events in the first 30 days of the trial provides circumstantial evidence of the safety of overlapping rivaroxaban and warfarin. However, in this setting, participants had rivaroxaban added to warfarin temporarily rather than warfarin added to rivaroxaban as occurred at the EOS visit. Several limitations should be acknowledged. Site investigators were instructed to report INR values after the EOS visit on the electronic case report form; however, we suspect but cannot confirm that some values may not have been reported. Not all participants were transitioned to open-label VKA therapy, and the decision to do so was made by the local physician; hence, it was not random. Therefore, potential biases may have been introduced that could neither be identified nor evaluated. No prospective clinical outcome trials have evaluated either the use of bridging therapy 12 or the strategy of overlapping of rivaroxaban and warfarin; therefore, the risks and benefits of these approaches are not known. Conclusions An excess in stroke and systemic embolic events during the EOS transition from rivaroxaban to open-label VKA occurred in participants assigned to receive rivaroxaban compared with warfarin in ROCKET AF. The pattern of events was not consistent with a rebound phenomenon. The excess events were likely associated with a period of inadequate therapeutic levels of anticoagulation during transition to VKA therapy, which was attributable to study-specific procedures implemented in part to protect the study blind. If a clinical situation requires rivaroxaban therapy to be stopped and a participant transitioned to VKA therapy, then suggested transition strategies include overlapping rivaroxaban and warfarin until the INR is 2.0 taken at the rivaroxaban trough. An alternative strategy is to stop rivaroxaban and simultaneously start parenteral anticoagulation and VKA therapy. Acknowledgments The authors thank Morgan deblecourt and Elizabeth Cook for editorial support with the article. We also thank all the site investigators, study coordinators, and patients who contributed to and participated in the trial. Sources of Funding This work was sponsored by Johnson & Johnson Pharmaceutical Research & Development, Raritan, NJ; and Bayer HealthCare AG, Leverkusen, Germany. Disclosures K. Mahaffey has disclosed grant support (significant) from AstraZeneca, Bayer, Boehringer Ingelheim, Bristol-Myers Squibb, Daiichi Sankyo, Eli Lilly, GlaxoSmithKline, Johnson & Johnson, Merck, Momenta Pharmaceuticals, Novartis, Portola, Pozen, Regado Biotechnologies, Sanofi-Aventis, Schering-Plough (now Merck), and The Medicines Company; significant consulting fees from AstraZeneca and Johnson & Johnson; and modest consulting fees from Bayer, Boehringer Ingelheim, Bristol-Myers Squibb, Daiichi Sankyo, Eli Lilly, GlaxoSmithKline, Merck, Novartis, Ortho/ McNeill, Pfizer, Polymedix, Sanofi-Aventis, and Schering-Plough (now Merck). M.R. Patel has disclosed honoraria from Johnson & Johnson and Bayer HealthCare for serving on the executive committee of the ROCKET AF; consulting fees from Ortho McNeil Janssen and Bayer HealthCare; and advisory board fees from Genzyme. K. Schwabe and C. Nessel are employees of Johnson & Johnson Pharmaceutical Research & Development. S. Berkowitz is an employee of Bayer HealthCare Pharmaceuticals. J.L. Halperin has disclosed honoraria from Johnson & Johnson and Bayer; and advisory board fees from Boehringer Ingelheim, Bristol Myers-Squibb, and Pfizer. G.J. Hankey has disclosed honoraria from Johnson & Johnson, Bayer, and Sanofi-Aventis; and has received fees for serving on trial adjudication committees and an advisory board for Boehringer Ingelheim. R.C. Becker has disclosed research support from Bayer and Johnson & Johnson. J. Piccini has disclosed grants for clinical research from Johnson & Johnson and Boston Scientific; and consulting and advisory board fees from Medtronic, Forest Laboratories, Sanofi-Aventis, and Johnson & Johnson. G. Breithardt has disclosed honoraria from Johnson & Johnson and Bayer; and advisory board fees from Boehringer Ingelheim, Bristol-Myers Squibb, Pfizer, and Sanofi-Aventis. W. Hacke has disclosed honoraria from Johnson & Johnson and Bayer; and advisory board fees from Boehringer Ingelheim. D.E. Singer has disclosed support, in part, by the Eliot B. and Edith C. Shoolman Fund of the Massachusetts General Hospital (Boston, MA); consulting fees from Bayer HealthCare, Boehringer Ingelheim, Bristol-Myers Squibb, Daiichi Sankyo, Johnson & Johnson, Pfizer, and Sanofi; and serves as a member of the executive committee of the ROCKET AF of rivaroxaban versus warfarin in patients with AF sponsored by Johnson & Johnson and Bayer HealthCare. R.M. Califf has disclosed consulting fees and research funding from Johnson & Johnson; and all other industry interactions are listed at www.dcri.org. K. Fox has disclosed grants and honoraria from Bayer, Lilly, Boehringer Ingelheim, Sanofi-Aventis, and GlaxoSmithKline. The other authors report no conflicts. References 1. Go AS, Hylek EM, Phillips KA, Chang Y, Henault LE, Selby JV, Singer DE. Prevalence of diagnosed atrial fibrillation in adults: national implications for rhythm management and stroke prevention: the Anticoagulation and Risk Factors in Atrial Fibrillation (ATRIA) study. JAMA. 2001;285:2370 2375. 2. ROCKET AF Study Investigators. Rivaroxaban-once daily, oral, direct factor Xa inhibition compared with vitamin K antagonism for prevention of stroke and Embolism Trial in Atrial Fibrillation: rationale and design of the ROCKET AF study. Am Heart J. 2010;159:340 347. 3. Patel MR, Mahaffey KW, Garg J, Pan G, Singer DE, Hacke W, Breithardt G, Halperin JL, Hankey GJ, Piccini JP, Becker RC, Nessel CC, Paolini JF, Berkowitz SD, Fox KA, Califf RM; ROCKET AF Investigators. Rivaroxaban versus warfarin in nonvalvular atrial fibrillation. N Engl J Med. 2011;365:883 891. 4. U.S. Food and Administration. XARELTO (Rivaroxaban) tablets. Risk Evaluation and Mitigation Strategy (REMS). http://www.fda.gov/ downloads/s/safety/postmarketsafetyinformationfor PatientsandProviders/UCM295405.pdf?utm_source=fdaSearch&utm_

Mahaffey et al ROCKET AF: End of Study Transition to VKA Therapy 9 medium=website&utm_term=rivaroxaban box warning&utm_content=1. Accessed April 4, 2012. 5. Patel MR, Hellkamp AS, Lokhnygina Y, Piccini JP, Zhang Z, Mohanty S, Singer DE, Hacke W, Breithardt G, Halperin JL, Hankey GJ, Becker RC, Nessel CC, Berkowitz SD, Califf RM, Fox KAA, Mahaffey KW. Outcomes of discontinuing rivaroxaban compared with warfarin in patients with nonvalvular atrial fibrillation: analysis from the ROCKET AF trial. J Am Coll Cardiol. 2013;61:651-658. 6. FDA Approves Xarelto to Reduce Risk of Blood Clots After Hip, Knee Replacements [News Release]. Silver Spring, MD: US Food & Administration; 2011. http://www.fda.gov/newsevents/newsroom/ PressAnnouncements/ucm261839.htm. Accessed July 10, 2012. 7. Connolly SJ, Ezekowitz MD, Yusuf S, Eikelboom J, Oldgren J, Parekh A, Pogue J, Reilly PA, Themeles E, Varrone J, Wang S, Alings M, Xavier D, Zhu J, Diaz R, Lewis BS, Darius H, Diener HC, Joyner CD, Wallentin L. Dabigatran versus warfarin in patients with atrial fibrillation. N Engl J Med. 2009;361:1139 1151. 8. Granger CB, Alexander JH, McMurray JJ, Lopes RD, Hylek EM, Hanna M, Al-Khalidi HR, Ansell J, Atar D, Avezum A, Bahit MC, Diaz R, Easton JD, Ezekowitz JA, Flaker G, Garcia D, Geraldes M, Gersh BJ, Golitsyn S, Goto S, Hermosillo AG, Hohnloser SH, Horowitz J, Mohan P, Jansky P, Lewis BS, Lopez-Sendon JL, Pais P, Parkhomenko A, Verheugt FW, Zhu J, Wallentin L. Apixaban versus warfarin in patients with atrial fibrillation. N Engl J Med. 2011;365:981 992. 9. Hori M, Matsumoto M, Tanahashi N, Momomura S, Uchiyama S, Goto S, Izumi T, Koretsune Y, Kajikawa M, Kato M, Ueda H, Iwamoto K, Tajiri M; J-ROCKET AF study investigators. Rivaroxaban vs. warfarin in Japanese patients with atrial fibrillation the J-ROCKET AF study. Circ J. 2012;76:2104 2111. 10. Gage BF, Waterman AD, Shannon W, Boechler M, Rich MW, Radford MJ. Validation of clinical classification schemes for predicting stroke: results from the National Registry of Atrial Fibrillation. JAMA. 2001;285:2864 2870. 11. Go AS, Hylek EM, Chang Y, Phillips KA, Henault LE, Capra AM, Jensvold NG, Selby JV, Singer DE. Anticoagulation therapy for stroke prevention in atrial fibrillation: how well do randomized trials translate into clinical practice? JAMA. 2003;290:2685 2692. 12. Harrison RW, Ortel TL, Becker RC. To bridge or not to bridge: these are the questions. J Thromb Thrombolysis. 2012;34:31 35.