Benefits and risks of extended treatment of venous thromboembolism with rivaroxaban or with aspirin

Thrombosis Research 168 (2018) 121–129

Contents lists available at ScienceDirect

Thrombosis Research journal homepage: www.elsevier.com/locate/thromres

Full Length Article

Benefits and risks of extended treatment of venous thromboembolism with rivaroxaban or with aspirin

T

Paolo Prandonia, Anthonie W.A. Lensingb, Martin H. Prinsc, Martin Gebelb, Akos F. Papb, Martin Homeringb, Rupert Bauersachsd, Jan Beyer-Westendorfe,f, Henri Bounameauxg, Alexander T. Cohenh, Bruce L. Davidsoni, Bonno van Bellenj, Peter Verhammek, Philip S. Wellsl, ⁎ Zhong Yuanm, Bennett Levitanm, Jeffrey I. Weitzn, a

Department of Cardiothoracic & Vascular Sciences, Vascular Medicine Unit, University of Padua, Italy Bayer AG, Leverkusen, Germany Department of Epidemiology and Technology Assessment, University of Maastricht, Maastricht, The Netherlands d Department of Vascular Medicine, Klinikum Darmstadt GmbH, Germany e Department of Hematology, Medical Clinic I, University Hospital “Carl Gustav Carus” Dresden, Germany f Department of Haematology, Oncology Kings College London, UK g Division of Angiology and Hemostasis, University Hospitals of Geneva, Faculty of Medicine, Geneva, Switzerland h Department of Haematological Medicine, Guys and St Thomas' Hospitals, King's College Hospital, London, UK i University of Washington School of Medicine, Seattle, WA, USA j Hospital Beneficência Portuguesa, São Paulo, Brazil k Vascular Medicine and Hemostasis, University of Leuven, Belgium l Department of Medicine, University of Ottawa, Ottawa Hospital Research Institute, Ontario, Canada m Janssen Research & Development, Raritan, NJ, USA n Thrombosis and Atherosclerosis Research Institute, McMaster University, Hamilton, Ontario, Canada b c

A B S T R A C T

Background: Full- or lower-dose anticoagulant therapy or aspirin can be used for extended therapy in patients with venous thromboembolism (VTE), but information on their relative benefit-risk profiles is limited. Methods: Data from the EINSTEIN-CHOICE trial were used to compare the benefit-risk profiles of extended treatment with rivaroxaban (20 or 10 mg once daily) and aspirin (100 mg once daily) in VTE patients who had completed 6 to 12 months of anticoagulation therapy. One-year cumulative incidences of recurrent VTE and major bleeding were estimated and benefits and risks were calculated by determining the between group differences in a hypothetical population of 10,000 VTE patients treated for 1 year. Findings: A total of 1107 patients were treated with 20 mg of rivaroxaban, 1127 with 10 mg of rivaroxaban, and 1131 with aspirin. The cumulative incidences of recurrent VTE in the rivaroxaban 20-mg, rivaroxaban 10-mg and aspirin groups were 1.9%, 1.6%, and 5.0%, respectively, whereas the cumulative incidences of major bleeding were 0.7%, 0.4% and 0.5%, respectively. The incidences of the combined outcome of recurrent VTE and major bleeding were 2.8% and 3.4% lower in the rivaroxaban 20-mg and 10-mg groups than in the aspirin group. For 10,000 patients treated for 1 year, there would be 284 (95% confidence interval [CI] 106 to 462) and 339 (95% CI 165 to 512) fewer events with rivaroxaban 20 mg or 10 mg than with aspirin. Interpretation: Compared with aspirin, extended anticoagulation with once daily rivaroxaban reduces recurrent VTE with a favourable benefit-risk profile. Funding: Bayer AG.

1. Introduction Venous thromboembolism (VTE), which includes deep-vein thrombosis and pulmonary embolism, is the third most common cause of

⁎

cardiovascular death [1–3]. Anticoagulation is the mainstay of treatment of VTE and many patients require extended therapy [4]. Extended anticoagulation with vitamin K antagonists such as warfarin is unwieldy because of the need for frequent coagulation monitoring and

Corresponding author at: Thrombosis and Atherosclerosis Research Institute, 237 Barton St. E, Hamilton L8L 2X2, Ontario, Canada. E-mail address: [email protected] (J.I. Weitz).

https://doi.org/10.1016/j.thromres.2018.06.009 Received 4 April 2018; Received in revised form 25 May 2018; Accepted 13 June 2018 0049-3848/ © 2018 Elsevier Ltd. All rights reserved.

Thrombosis Research 168 (2018) 121–129

P. Prandoni et al.

appearing, immediate release film-coated tablets, whereas aspirin and matching placebo were provided as enteric-coated tablets. The intended duration of administration of study drug was 12 months, but patients randomized after the requisite number of primary efficacy outcomes was reached were treated for at least 6 months. The risk profiles for recurrent VTE at the time of the index event were systematically assessed by an independent adjudication committee whose members were blinded to treatment assignment and outcome status. Index VTE events were classified as provoked by (a) major persistent risk factors (i.e., active cancer excluding basal-cell or squamous-cell skin cancer), (b) minor persistent risk factors (i.e., inflammatory bowel disease, lower extremity paralysis or paresis, congestive heart failure, body mass index over 30 kg/m2, calculated creatinine clearance below 50 mL/min, family history of VTE, or known thrombophilia, including deficiency of antithrombin, protein C or protein S, factor V Leiden or prothrombin gene mutation, and antiphospholipid syndrome), (c) minor transient risk factors (i.e., immobilization, travel over 8 h, pregnancy, puerperium or use of oestrogen, or lower limb trauma with transient impairment of mobility), or (d) major transient risk factors (i.e., major surgery or trauma, or caesarean section) using pre-specified criteria. Patients without any of these risk factors were classified as having unprovoked VTE. The EINSTEIN-CHOICE study was performed in a manner consistent with the principles of the Declaration of Helsinki. Institutional review board approval was obtained in all participating centres and all patients provided written informed consent.

dose adjustment to ensure that the international normalized ratio (INR) is in the therapeutic range [5–7]. Furthermore, major bleeding can occur with vitamin K antagonists even when the INR is therapeutic [8]. Therefore, more convenient and safer anticoagulants are needed. Direct oral anticoagulants such as rivaroxaban simplify VTE treatment because they can be given in fixed doses without routine coagulation monitoring [9–12]. Although direct oral anticoagulants are safer than vitamin K antagonists [10, 13–15], major bleeding can still occur. Consequently, patients and physicians are often reluctant to continue anticoagulation treatment except in patients at highest risk for recurrence and aspirin may be used instead of an anticoagulant because of the perception that it is safer. Traditionally, patients with unprovoked VTE were considered to be at high risk for recurrence, whereas those with provoked VTE were considered to be at lower risk. However, there is emerging evidence that patients with VTE provoked by minor risk factors also have an appreciable risk of recurrence. Therefore, more information is needed on the benefit-risk profile of rivaroxaban relative to aspirin in patients with unprovoked VTE and VTE provoked by minor risk factors. In this analysis, data from the Reduced-dosed Rivaroxaban in the Long-term Prevention of Recurrent Symptomatic Venous Thromboembolism trial (EINSTEIN-CHOICE, ClinicalTrials.gov number NCT02064439) were used to compare the benefit-risk profiles of extended treatment with rivaroxaban (20 mg or 10 mg once daily) and aspirin (100 mg once daily) in patients with symptomatic VTE who had completed 6 to 12 months of anticoagulation therapy and for whom there was equipoise regarding the need for extended anticoagulation treatment [19, 20]. Results in patients with unprovoked VTE were compared with those in patients with VTE provoked by minor risk factors.

2.2. Outcome assessment The primary efficacy outcome was symptomatic recurrent VTE, defined as the composite of deep-vein thrombosis or nonfatal or fatal pulmonary embolism, using the diagnostic criteria described previously [19]. Death was classified as due to pulmonary embolism, bleeding, or other established causes. An overt bleeding event was defined as major if it was fatal, occurred in a critical site, was associated with a decrease in the haemoglobin level of ≥2.0 g/dl or required a transfusion of ≥2 units of whole blood or packed red blood cells [21]. Clinically relevant bleeding was defined as the composite of major bleeding and clinically relevant nonmajor bleeding (i.e., any overt bleeding that required a medical intervention and did not meet the criteria for major bleeding) [21, 22]. The predefined outcome of net clinical benefit was the composite of recurrent VTE and major bleeding. An independent committee blinded to treatment allocation adjudicated all suspected study outcomes. Patients were followed for the intended treatment duration and seen at fixed intervals, at which time a checklist was used to assess symptoms and signs of recurrent VTE, bleeding, and adverse events. Patients were instructed to report to the study centre if any of these events occurred. In cases of suspected VTE, the study protocol required objective testing for confirmation of the diagnosis.

2. Patients and methods 2.1. Study design The EINSTEIN-CHOICE study was a double-blind, randomized, double-dummy, event-driven, superiority study that compared two once daily doses of rivaroxaban (20 mg and 10 mg) with aspirin (100 mg once-daily) for extended treatment of VTE [19, 20]. Study drugs were administered for up to 12 months. Patients were eligible for inclusion in the study if they were 18 years of age or older; if they had objectively confirmed, symptomatic proximal deep-vein thrombosis or pulmonary embolism; if they had been treated for 6 to 12 months with an anticoagulant including vitamin K antagonists or direct oral anticoagulants; and if they had not interrupted therapy for > 7 days prior to randomization. Patients were ineligible if they had a contraindication to extended anticoagulant therapy or if they required extended anticoagulant therapy at therapeutic dosages or antiplatelet therapy. Additional ineligibility criteria included a calculated creatinine clearance < 30 mL/ min; hepatic disease associated with a coagulopathy; active bleeding or a high risk of bleeding that contraindicated anticoagulant treatment; systolic blood pressure > 180 mm Hg or diastolic blood pressure > 110 mm Hg; childbearing potential without proper contraceptive measures, pregnancy, or breast-feeding; concomitant use of potent cytochrome P450 3A4 inhibitors or inducers; or a life expectancy of < 3 months. Randomization was performed in blocks of 6 using an interactive voice-response system and was stratified according to the index diagnosis (deep-vein thrombosis or pulmonary embolism) and country. Patients were enrolled at least 24 h after receiving their last dose of direct oral anticoagulant or, if they were receiving a vitamin K antagonist, when the INR was 2.5 or lower. Patients were assigned in a 1:1:1 ratio to receive 20 mg of rivaroxaban, 10 mg of rivaroxaban, or 100 mg of aspirin, all given once daily with food. Rivaroxaban (20 mg and 10 mg) and matching placebo were provided as identical

2.3. Statistical analysis To compare benefits and risks, efficacy and safety outcomes were analysed in the full analysis set, consisting of all randomized patients who received at least one dose of study medication. Results were computed using 1-year Kaplan-Meier cumulative incidences and events that occurred in the interval between the start of study drug and the end of the individual intended treatment duration were included. Additional analyses were performed for the defined risk factor groups and for the following pre-specified subgroups: age (< 65, 65–75, and > 75 years), sex, body weight (< 50 and ≥50 kg), index event (deep-vein thrombosis and pulmonary embolism with or without deepvein thrombosis), known thrombophilia, creatinine clearance (< 50, 50– < 80, and > 80 mL/min), fragility (defined as age > 75 years, creatinine clearance < 50 mL/min, or body weight < 50 kg), active 122

Thrombosis Research 168 (2018) 121–129

P. Prandoni et al.

Table 1 A. Rivaroxaban 20 mg vs aspirin: one-year Kaplan-Meier cumulative incidences, differences, and NNT/NNH for key efficacy and safety outcomes per 10,000 patients. B. Rivaroxaban 10 mg vs aspirin: one-year Kaplan-Meier cumulative incidences, differences, and NNT/NNH for key efficacy and safety outcomes per 10,000 patients. Outcome

Number of events/10,000 patients

Rate difference/10,000 patients (95% CI)

NNT or NNHa

Rivaroxaban 20 mg

Aspirin

Efficacy Recurrent venous thromboembolism Pulmonary embolism Deep-vein thrombosis All-cause mortality

185 80 106 79

497 203 304 69

−312 (−479 to 144) −123 (−226 to −21) −198 (−333 to −62) 10 (−66 to 85)

−33 −82 −51 1018

Safety Major bleeding Fatal or intracranial bleeding Clinically relevant nonmajor bleeding

73 39 287

45 18 258

28 (−43 to 99) 20 (−25 to 66) 29 (−151 to 209)

356 488 345

Other Net clinical benefit

247

530

−284 (−462 to −106)

−36

Outcome

Number of events/10,000 patients

Rate difference/10,000 patients (95% CI)

NNT or NNHa

Rivaroxaban 10 mg

Aspirin

Efficacy Recurrent venous thromboembolism Pulmonary embolism Deep-vein thrombosis All-cause mortality

156 82 87 18

497 203 304 69

−341 (−507 to −175) −121 (−238 to −4) −217 (−342 to −92) −51 (−108 to 6)

−30 −83 −47 −196

Safety Major bleeding Fatal or intracranial bleeding Clinically relevant nonmajor bleeding

45 9 224

45 18 258

0 (−60 to 59) −9 (−40 to 21) −34 (−213 to 145)

−58,376 −1099 −295

Other Net clinical benefit

192

530

−339 (−512 to −165)

−30

CI, confidence interval; NNT, number needed to treat; NNH, number needed to harm. a A negative number denotes the number of patients needed to be followed for 1 year with rivaroxaban instead of aspirin to prevent 1 additional event (NNT), while a positive number denotes the number of patients needed to be followed for 1 year with rivaroxaban instead of aspirin to induce 1 additional event (NNH).

access to the raw data. The corresponding author had full access to all of the data, and the final responsibility for submission of the paper for publication.

cancer, cardiac disease, and RIETE score for risk of bleeding (low, intermediate, or high) [23]. Benefits and risks were assessed using excess numbers of events, defined as the difference in cumulative incidences scaled to a hypothetical population of 10,000 VTE patients treated for 1 year to best reflect benefits and risks at a population level [24, 25]. Excess numbers of events can be interpreted as the number of patients in this hypothetical population who would experience a particular event when treated with rivaroxaban (20 mg or 10 mg once daily) minus that in the same population treated with aspirin (100 mg once daily), such that negative values would favour rivaroxaban. Results are shown in forest plot and tabular form, both for the total population and for the prespecified subgroups. The number needed to treat (NNT) and the number needed to harm (NNH) were calculated as the reciprocals of the differences in the cumulative incidences between rivaroxaban (20 mg or 10 mg) and aspirin. To assess benefits and risks over time, between treatment group differences in the Kaplan–Meier cumulative incidence estimates of recurrent VTE and major bleeding were computed and plotted over the 1year period. The curves reflect the difference between the Kaplan–Meier curves for the rivaroxaban and aspirin groups, with negative values indicating fewer events with rivaroxaban. Cumulative incidences were converted to excess numbers of events, as described above.

3. Results 3.1. Study patients From March 2014 through March 2016, a total of 3396 patients from 244 sites in 31 countries were enrolled. Of these, 31 patients (0.9%) were excluded because they did not take any study drug. Therefore, 3365 patients were included in the analyses. Baseline characteristics of the patients were similar in the 3 treatment arms [20]. 3.2. Efficacy outcomes In the rivaroxaban 20-mg group, a primary efficacy outcome event occurred in 17 of the 1107 patients (8 had pulmonary embolism and 9 had deep-vein thrombosis) during a median treatment duration of 349 days (interquartile range [IQR], 189–362 days), with a 1-year cumulative incidence of 1.9%; in the rivaroxaban 10-mg group, a primary efficacy outcome occurred in 13 of the 1127 patients (5 had pulmonary embolism and 8 had deep-vein thrombosis) during a median treatment duration of 353 days (IQR, 190–362 days) with a 1-year cumulative incidence of 1.6%, and in the aspirin group, a primary efficacy outcome occurred in 50 patients of the 1131 patients (21 had pulmonary embolism and 29 had deep-vein thrombosis) during a median treatment duration of 350 days (IQR, 186–362 days) with a 1-year cumulative incidence of 5.0%. Fatal VTE occurred in 0.2%, 0%, and 0.2% of patients in the rivaroxaban 20-mg, rivaroxaban 10-mg and aspirin groups,

2.4. Role of the funding source Bayer AG, which funded the EINSTEIN-CHOICE study, gathered, maintained, and extracted data. The authors had responsibility for interpreting the data and writing the article. PP, MHP, and JIW had 123

Thrombosis Research 168 (2018) 121–129

P. Prandoni et al.

Fig. 1. One-year Kaplan-Meier cumulative incidence differences per 10,000 patients for key efficacy and safety outcomes. Panel A, rivaroxaban 20 mg vs aspirin; Panel B, rivaroxaban 10 mg vs aspirin. Negative values favour rivaroxaban; positive values favour aspirin. Vertical lines reflect 95% confidence intervals. Footnote in this figure: Confidence limits based on asymptomatic method (Greenwood's formula) for standard error estimation.

pulmonary embolism and 198 (95% CI 62 to 333) fewer episodes of deep-vein thrombosis (Table 1A, Fig. 1A), whereas in those given 10 mg of rivaroxaban instead of aspirin, there would be 121 (95% CI 4 to 238) fewer episodes of pulmonary embolism and 217 (95% CI 92 to 342) fewer episodes of deep-vein thrombosis (Table 1B; Fig. 1B). The NNT to prevent one episode of deep-vein thrombosis or pulmonary embolism with rivaroxaban rather than with aspirin is 33 with the 20-mg rivaroxaban dose and 30 with the 10-mg dose (Table 1A and B). Cumulative incidences of death from any cause were low. There were 8 (0.8%) deaths in the rivaroxaban-20 mg group, 2 (0.2%) deaths

respectively. The 1-year cumulative incidence difference between rivaroxaban 20 mg and aspirin is 3.1% (95% confidence interval [CI] 1.4 to 4.8) in favour of rivaroxaban and the 1-year cumulative incidence difference between rivaroxaban 10 mg and aspirin is 3.4% (95% CI 1.7 to 5.1) in favour of rivaroxaban. For 10,000 VTE patients treated for 1 year, treatment with rivaroxaban 20 mg or 10 mg instead of aspirin would have resulted in 312 (95% CI 144 to 479) and 341 (95% CI 175 to 507) fewer recurrent VTE events, respectively (Table 1A and B). Thus, in patients treated with 20 mg of rivaroxaban instead of with aspirin there would be 123 (95% CI 21 to 226) fewer episodes of 124

Thrombosis Research 168 (2018) 121–129

P. Prandoni et al.

Table 2 A. Rivaroxaban 20 mg vs aspirin: recurrent VTE, major bleeding, and net clinical benefit 1-year Kaplan-Meier cumulative incidences, NNT and NNH per 10,000 patients for patients with unprovoked or provoked VTE. B. Rivaroxaban 10 mg vs aspirin: recurrent VTE, major bleeding, and net clinical benefit 1-year Kaplan-Meier cumulative incidences, NNT and NNH per 10,000 patients for patients with unprovoked or provoked VTE. Risk factor profile

A All Unprovoked Provoked by major persistent risk factors Provoked by minor persistent risk factors Provoked by minor transient risk factors Provoked by major transient risk factors B All Unprovoked Provoked by major persistent risk factors Provoked by minor persistent risk factors Provoked by minor transient risk factors Provoked by major transient risk factors

Primary efficacy outcome

Major bleeding

Net clinical benefit a

Number of patients

Rate difference/10,000 patients (95% CI)

NNT or NNHa

Rate difference/10,000 patients (95% CI)

NNT or NNH

Rate difference/10,000 patients (95% CI)

NNT or NNHa

Rivaroxaban

Aspirin

−312 (−479 to −144) −386 (−646 to −125) −714 (−1668 to 240)

−33 −26 −14

28 (−43 to 99) 75 (−29 to 179) −764 (−1823 to 296)

356 134 −14

−284 (−462 to −106) −311 (−590 to −32) −1106 (−2320 to 108)

−36 −33 −10

1107

1131

−226 (−503 to 51)

−45

21 (−20 to 63)

471

−205 (−485 to 75)

−49

−323 (−775 to 130)

−31

150 (−308 to 609)

67

−293 (−892 to 306)

−35

0

–

0

–

0

–

−341 (−507 to −175) −419 (−672 to −167) −714 (−1668 to 240)

−30 −24 −14

0 (−60 to 59) 21 (−52 to 93) −764 (−1823 to 296)

−58,376 483 −14

−339 (−512 to −165) −419 (−678 to −160) −1106 (−2320 to 108)

−30 −24 −10

1127

1131

−243 (−533 to 46)

−42

0

–

−243 (−533 to 46)

−42

−419 (−831 to −7)

−24

72 (−299 to 443)

140

−349 (−900 to 203)

−29

0

–

179 (−168 to 525)

56

179 (−168 to 525)

56

CI, confidence interval; NNT, number needed to treat; NNH, number needed to harm; VTE, venous thromboembolism. a A negative number denotes the number of patients needed to be followed for 1 year with rivaroxaban instead of aspirin to prevent 1 additional event (NNT), while a positive number denotes the number of patients needed to be followed for 1 year with rivaroxaban instead of aspirin to induce 1 additional event (NNH).

rivaroxaban, whereas there would be one less clinically relevant nonmajor bleed for every 295 patients treated with 10 mg of rivaroxaban (Table 1A and B).

in the rivaroxaban 10-mg group; and 7 (0.7%) deaths in the aspirin group. For 10,000 VTE patients treated with rivaroxaban 20 mg for 1 year instead of with aspirin there would be 10 (95% CI −66 to 85) more deaths with a corresponding NNH of 1018; for 10,000 VTE patients treated with rivaroxaban 10 mg instead of with aspirin there would be 51 (95% CI −6 to 108) less deaths with a corresponding NNT of 196 (Table 1A and B).

3.4. Net clinical benefit Net clinical benefit, defined as the composite of symptomatic recurrent VTE and major bleeding events, occurred in 23 patients in the rivaroxaban 20-mg group, in 17 patients in the rivaroxaban 10-mg group, and in 53 patients in the aspirin group, yielding differences in the 1-year cumulative incidences of 2.8% (95% CI 1.1% to 4.6) and 3.4% (95% CI 1.6% to 5.1) in favour of the 20 mg and 10 mg doses of rivaroxaban, respectively. For 10,000 VTE patients treated for 1 year with rivaroxaban 20 mg or 10 mg instead of aspirin, there would be 284 (95% CI 106 to 462) and 339 (95% CI 165 to 512) fewer net clinical benefit outcomes, respectively. Thus, compared with aspirin, one additional symptomatic recurrent VTE or major bleed would be avoided for every 36 or 30 patients treated with rivaroxaban 20 mg or 10 mg, respectively (Table 1A and B).

3.3. Safety outcomes Major bleeding occurred in 7 patients (4 intracranial, 1 pericardial, 1 pulmonary and 1 gastrointestinal) in the 20-mg rivaroxaban group for a 1-year cumulative incidence of 0.7%. There was 1 fatal bleed and 4 bleeds in a critical site. Major bleeding occurred in 5 patients (1 intracranial, 1 intramuscular, 2 gastrointestinal, and 1 abdominal) in the 10-mg rivaroxaban group for a 1-year cumulative incidence of 0.4%. There were no fatal bleeds, but 2 bleeds were in a critical site. Major bleeding occurred in 4 patients (2 intracranial, 2 gastrointestinal) in the aspirin group for a 1-year cumulative incidence of 0.5%. There was one fatal bleed and one bleed in a critical site. For 10,000 VTE patients treated with rivaroxaban 20 mg or 10 mg for 1 year instead of with aspirin there would be 28 (95% CI −43 to 99) and 0 (95% CI −60 to 59) more major bleeding events, respectively, and the corresponding NNHs would be 356 and > 10,000, respectively (Table 1A and B). The cumulative incidences of clinically relevant nonmajor bleeding were 2.9%, 2.2%, and 2.6% in the rivaroxaban 20-mg, rivaroxaban 10mg and aspirin groups, respectively. Therefore, for 10,000 VTE patients treated with rivaroxaban 20 mg for 1 year instead of with aspirin there would be 29 (95% CI −151 to 209) more clinically relevant non-major bleeding events, whereas for those treated with rivaroxaban 10 mg instead of with aspirin there would be 34 (95% CI −145 to 213) fewer clinically relevant non-major bleeding events. Thus, compared with aspirin treatment, there would be one additional clinically relevant nonmajor bleed for every 345 patients treated with 20 mg of

3.5. Subgroups Cumulative rates of recurrent VTE, major bleeds, and clinically relevant nonmajor bleeds are provided for patients with unprovoked or provoked VTE in Table 2A and B. In general, the number of episodes prevented or induced and the NNT and NNH values in patients with unprovoked VTE were similar to those in patients with VTE provoked by minor persistent or minor transient risk factors, although the 95% CIs are wide. Results for all prespecified subgroups are shown in Table 3A and B. The cumulative incidence of recurrent VTE was lower with both doses of rivaroxaban than with aspirin in all subgroups. Due to the small numbers, the cumulative incidence of major bleeding varied between subgroups. Nonetheless, net clinical benefit as well as the ratio of NNT 125

Thrombosis Research 168 (2018) 121–129

P. Prandoni et al.

Table 3 A. Rivaroxaban 20 mg vs aspirin: recurrent VTE, major bleeding, and net clinical benefit 1-year Kaplan-Meier cumulative incidences, NNT and NNH per 10,000 patients for all subgroups. B. Rivaroxaban 10 mg vs aspirin: recurrent VTE, major bleeding, and net clinical benefit 1-year Kaplan-Meier cumulative incidences, NNT and NNH per 10,000 patients for all subgroups. Subgroup

Primary efficacy outcome Rate difference/10,000 patients (95% CI)

A All −312 (−479 to −144) Age (years) < 65 −309 (−529 to −90) 65–75 −303 (−627 to 21) > 75 −348 (−734 to 38) Sex Female −310 (−532 to −88) Male −311 (−553 to −69) Weight (kg) < 50 −769 (−2218 to 679) ≥50 −306 (−474 to −138) Type of index event DVT −460 (−682 to −239) PE ± DVT −165 (−414 to 84) Fragility≠ No −309 (−496, −122) Yes −348 (−680 to −16) Creatinine clearance (mL/min) < 50 −500 (−1054 to 53) 50– < 80 −311 (−568 to −55) ≥80 −301 (−514 to −87) Known thrombophilia No −277 (−445 to −108) Yes −790 (−1661 to 80) Active cancer No −300 (−470 to −131) Yes −714 (−1668 to 240) Cardiac disease No −327 (−503 to −150) Yes −168 (−642 to 305) RIETE risk score Low −366 (−677 to −56) Intermediate −268 (−456 to −81) High −1667 (−4649 to 1315) B All −341 (−507 to −175) Age (years) < 65 −359 (−576 to −141) 65–75 −415 (−712 to −117) > 75 −89 (−572 to 394) Sex Female −236 (−473 to 1) Male −430 (−656 to −204) Weight (kg) < 50 −769 (−2218 to 679) ≥50 −335 (−503 to −168) Type of index event DVT −394 (−658 to −130) PE ± DVT −283 (−493 to −72) Fragility≠ No −378 (−560 to −196) Yes −138 (−546 to 269) Creatinine clearance (mL/min) < 50 −500 (−1054 to 53) 50– < 80 −221 (−506 to 64) ≥80 −372 (−581 to −164) Known thrombophilia No −314 (−480 to −148) Yes −729 (−1639 to 181) Active cancer No −330 (−499 to −162) Yes −714 (−1668 to 240) Cardiac disease No −334 (−514 to −154) Yes −385 (−754 to −15) RIETE risk score Low −435 (−754 to −116)

Major bleeding

Net clinical benefit

NNT or NNHa

Rate difference/10,000 patients (95% CI)

NNT or NNH

−33

28 (−43 to 99)

−33 −33 −29

a

Number of patients

Rate difference/10,000 patients (95% CI)

NNT or NNHa

Rivaroxaban

Aspirin

356

−284 (−462 to −106)

−36

1107

1131

72 (−10 to 154) −36 (−205 to 133) −71 (−209 to 68)

139 −282 −141

−255 (−486 to −24) −339 (−701 to 23) −338 (−717 to 41)

−40 −30 −30

691 301 115

684 301 146

−33 −33

33 (−63 to 129) 25 (−76 to 127)

305 396

−277 (−518 to −36) −288 (−543 to −33)

−37 −35

505 602

488 643

−13 −33

0 (0 to 0) 28 (−43 to 100)

. 352

−769 (−2218 to 679) −278 (−457 to −99)

−13 −36

14 1093

13 1118

−22 −61

43 (−64 to 150) 14 (−79 to 107)

233 715

−439 (−680 to −198) −129 (−390 to 132)

−23 −78

556 551

567 564

−33 −29

42 (−38 to 122) −58 (−173 to 56)

241 −171

−280 (−481 to −80) −340 (−667 to −14)

−36 −30

958 149

953 178

−20 −33 −34

0 (0 to 0) 39 (−140 to 218) 25 (−55 to 105)

– 258 402

−500 (−1054 to 53) −273 (−585 to 38) −275 (−498 to −53)

−20 −37 −37

41 279 787

64 277 790

−37 −13

31 (−45 to 107) 0 (0 to 0)

323 –

−246 (−427 to −65) −790 (−1661 to 80)

−41 −13

1028 79

1061 70

−34 −14

52 (−13 to 117) −748 (−1777 to 281)

193 −14

−260 (−439 to −81) −1091 (−2281 to 99)

−39 −10

1082 25

1094 37

−31 −60

8 (−63 to 79) 238 (−89 to 566)

1220 42

−318 (−505 to −132) 68 (−503 to 638)

−32 148

1012 95

1027 104

−28 −38 −6

52 (−46 to 149) 29 (−64 to 121) −1429 (−4021 to 1164)

193 349 −7

−343 (−663 to −24) −240 (−448 to −31) −1429 (−4021 to 1164)

−30 −42 −7

428 677 2

425 699 7

−30

0 (−60 to 59)

−58,376

−339 (−512 to −165)

−30

1127

1131

−28 −25 −112

45 (−20 to 110) −71 (−225 to 84) −71 (−209 to 68)

223 −142 −141

−328 (−553 to −103) −485 (−817 to −152) −80 (−557 to 398)

−31 −21 −126

678 316 133

684 301 146

−43 −24

8 (−75 to 92) −6 (−90 to 77)

1180 −1594

−247 (−495 to 0) −416 (−653 to −179)

−41 −25

507 620

488 643

−13 −30

0 (0 to 0) 0 (−60 to 61)

– 136,109

−769 (−2218 to 679) −333 (−508 to −157)

−13 −31

20 1107

13 1118

−26 −36

9 (−78 to 96) −9 (−90 to 72)

1093 −1085

−384 (−661 to −107) −287 (−505 to −69)

−27 −35

567 560

567 564

−27 −73

10 (−58 to 77) −58 (−173 to 56)

1038 −171

−378 (−570 to −186) −130 (−533 to 273)

−27 −77

963 164

953 178

−20 −46 −27

0 (0 to 0) −25 (−157 to 107) 8 (−65 to 80)

– −402 1323

−500 (−1054 to 53) −246 (−559 to 67) −361 (−577 to −145)

−20 −41 −28

51 302 774

64 277 790

−32 −14

0 (−64 to 64) 0 (0 to 0)

−61,902 –

−311 (−486 to −136) −729 (−1639 to 181)

−33 −14

1053 74

1061 70

−31 −14

23 (−28 to 75) −748 (−1777 to 281)

434 −14

−316 (−491 to −141) −1091 (−2281 to 99)

−32 −10

1100 27

1094 37

−30 −26

1 (−65 to 67) 0 (0 to 0)

10,631 –

−330 (−518 to −142) −385 (−754 to −15)

−31 −26

1005 122

1027 104

−24

50 (−45 to 146)

199

−383 (−716 to −51)

−27

411

425

(continued on next page) 126

Thrombosis Research 168 (2018) 121–129

P. Prandoni et al.

Table 3 (continued) Subgroup

Intermediate High

Primary efficacy outcome

Major bleeding

Net clinical benefit

Number of patients

Rate difference/10,000 patients (95% CI)

NNT or NNHa

Rate difference/10,000 patients (95% CI)

NNT or NNHa

Rate difference/10,000 patients (95% CI)

NNT or NNHa

Rivaroxaban

Aspirin

−270 (−457 to −83) −1667 (−4649 to 1315)

−37 −6

−15 (−86 to 57) −1429 (−4021 to 1164)

−685 −7

−299 (−497 to −102) −1429 (−4021 to 1164)

−34 −7

712 4

699 7

CI, confidence interval; DVT, deep vein thrombosis; NNT, number needed to treat; NNH, number needed to harm; PE, pulmonary embolism. a A negative number denotes the number of patients needed to be followed for 1 year with rivaroxaban instead of aspirin to prevent 1 additional event (NNT), while a positive number denotes the number of patients needed to be followed for 1 year with rivaroxaban instead of aspirin to induce 1 additional event (NNH). ≠Fragility is defined as creatinine clearance < 50 mL/min, age ≥ 75 years, or body weight < 50 kg.

aspirin, rivaroxaban is associated with a reduction in recurrent VTE and a minimal increase in bleeding. Therefore, there is no longer a place for aspirin for extended VTE treatment. Like aspirin, rivaroxaban is taken once daily and requires no coagulation monitoring [20]. Furthermore, rates of drug discontinuation due to adverse events with rivaroxaban are similar to those with aspirin [20]. Consequently, except for its higher cost, rivaroxaban is a better choice for extended VTE treatment than aspirin. Selection of the 20 mg or 10 mg dose of rivaroxaban for extended VTE needs to be individualized. The 10 mg dose can be used for most patients. However, patients at high risk for recurrent VTE, such as those with antiphospholipid syndrome, metastatic cancer, or with recurrence while on therapy may do better with the 20 mg rivaroxaban dose because few such patients were enrolled in the EINSTEIN-CHOICE study. Regardless of which dose is chosen, however, the results of this analysis indicate that rivaroxaban has a favourable benefit-risk profile relative to aspirin. Therefore, the findings facilitate the translation of the results into practice. The estimates of recurrent VTE and bleeding with rivaroxaban and aspirin are likely to be accurate because the EINSTEIN-CHOICE study was double-blind and outcome events were centrally adjudicated by a committee unaware of treatment allocation. Although not all patients were treated with study drug for 12 months, the median duration was 350 days and over 80% of patients received study drug treatment for 9 months or longer. Therefore, the 1-year Kaplan-Meier estimates of cumulative incidences of recurrent VTE and major or clinically-relevant nonmajor bleeds are likely to be robust. In summary, this study shows that compared with aspirin, extended anticoagulation with once daily rivaroxaban at either the 20 mg or 10 mg dose provides a clinically important benefit in terms of reduction in recurrent VTE, and a favourable benefit–risk profile. This observation holds among important subgroups of patients including those with unprovoked VTE and those with VTE provoked by minor persistent or transient risk factors. Therefore, rivaroxaban is convenient and safe for extended VTE treatment.

to NNH favoured rivaroxaban in all subgroups, except for the comparison of rivaroxaban 20 mg with aspirin in patients with cardiac disease. 3.6. Benefits and risks over time Fig. 2A and B illustrate the time-to-event curves for recurrent VTE and major bleeding in a hypothetical population of 10,000 VTE patients treated for 1 year. The curves for recurrent VTE with both doses of rivaroxaban relative to aspirin diverge soon after starting treatment and continue to separate thereafter. The increase in major bleeds occurs gradually over time, suggesting that throughout the course of treatment, benefit would exceed risk. 4. Discussion Decisions about extended anticoagulation therapy for patients with VTE are complicated. Guidelines recommend extended therapy if the risk of recurrent VTE off treatment exceeds the risk of bleeding on treatment [4]. Patient preference also needs to be considered and some patients elect to stop anticoagulation therapy mainly because of the fear of bleeding [27]. Guidelines suggest aspirin over no treatment in such patients [4] because compared with placebo, aspirin reduced the risk of recurrent VTE by 32% without significantly increasing the risk of major bleeding [28]. However, the results of the EINSTEIN-CHOICE trial revealed that full-dose and half-dose once daily rivaroxaban (20 and 10 mg, respectively) are superior to aspirin for prevention of recurrent VTE (hazard ratio for 20 mg of rivaroxaban vs aspirin, 0.34; 95% CI 0.20 to 0.59; hazard ratio for 10 mg of rivaroxaban vs aspirin, 0.26; 95% CI, 0.14 to 0.47; P < 0.001 for both comparisons) and are associated with similar low rates of major bleeding (0.5% in the 20-mg rivaroxaban group and 0.4% in the 10-mg rivaroxaban group, as compared with 0.3% in the aspirin group) [20]. This observation appears to be consistent in patients with unprovoked VTE and VTE provoked by minor persistent or transient risk factors. To better compare the benefit-risk profiles of rivaroxaban and aspirin, we used the Kaplan-Meier method to estimate the 1-year cumulative incidences of recurrent VTE and major bleeding and we then compared the between-group differences in a hypothetical population of 10,000 VTE patients followed for 1 year. Compared with aspirin, treatment with 20 mg or 10 mg of rivaroxaban would result in 312 (95% CI, 144 to 479) and 341 (95% CI, 175 to 507) fewer recurrent VTE events (NNTs of 33 and 30, respectively) and in 28 (95% CI −43 to 99) and 0 (95% CI −60 to 59) more major bleeds (NNHs of 356 and > 10,000, respectively). Therefore, both doses of rivaroxaban have favourable benefit-risk profiles compared with aspirin. It should be realized that all outcome rates are reported for the intention-to-treat population to enable optimal comparison of benefits and risks. Therefore, the rates of bleeding reported in this analysis differ from those in the original publication where crude incidences during study treatment were reported. The clinical implications of these findings are clear. Compared with

Contributors JIW, MHP, AWAL, and PP prepared the initial draft of the manuscript. ÁFP, MG, ZY and LB performed the statistical analyses. All authors participated in the writing and review of subsequent versions of the paper and accept full responsibility for its overall content. Declaration of interests PP reports receiving consulting and lecture fees from Bayer, Sanofi, Daiichi Sankyo, and Pfizer; AWAL, MG, AFP and MH being employees of Bayer; MHP receiving consulting fees from Pfizer and Daiichi Sankyo; RB receiving consulting and lecture fees from Boehringer Ingelheim, Bristol-Myers Squibb, and Daiichi Sankyo; JB-W receiving grant support, lecture fees, and fees for serving on advisory boards from Boehringer Ingelheim, Daiichi Sankyo, and Pfizer; HB receiving grant 127

Thrombosis Research 168 (2018) 121–129

P. Prandoni et al.

Fig. 2. Kaplan–Meier cumulative incidence differences over time per 10,000 patients for recurrent venous thromboembolism and major bleeding. Panel A, rivaroxaban 20 mg vs aspirin; panel B, rivaroxaban 10 mg vs aspirin.

128

Thrombosis Research 168 (2018) 121–129

P. Prandoni et al.

support and fees for serving on the Thrombosis Research Institute Garfield Registry steering committee, consulting fees from Amgen, and fees for serving on advisory boards from Bayer, Pfizer, and Sanofi Aventis; ATC receiving fees for serving on a committee for Boehringer Ingelheim, grant support and fees for serving on committees from Bristol-Myers Squibb and Daiichi Sankyo, consulting fees and fees for serving on steering committees from Johnson & Johnson and Portola, grant support, consulting fees, and fees for serving on committees from Pfizer, and consulting fees from Sanofi, Janssen, and Ono Pharmaceuticals; BLD receiving consulting fees from Janssen and Portola; BvB receiving lecture fees and fees for serving on an advisory board from Bayer and Daiichi Sankyo and for serving on an advisory board from Bristol-Myers Squibb; PV receiving grant support, lecture fees, and fees for serving on advisory boards from Boehringer Ingelheim and LEO Pharma, lecture fees from Pfizer and Bristol-Myers Squibb, grant support from Sanofi, lecture fees and fees for serving on advisory boards from Daiichi Sankyo, and fees for serving on an advisory board from Portola Pharmaceuticals; PSW receiving grant support, lecture fees, and fees for serving on advisory boards from Bayer, fees for serving on a writing committee from Itreas, consulting fees from Janssen Scientific Affairs, grant support from Bristol-Myers Squibb and Pfizer, and lecture fees from Daiichi Sankyo; ZY and BL being employee of Janssen Pharmaceutical Research & Development LLC; and JIW receiving consulting fees from Bayer, Boehringer Ingelheim, Bristol-Myers Squibb, Daiichi Sankyo, Ionis Pharmaceuticals, Janssen, Johnson & Johnson, Novartis, Portola, Pfizer, and Servier.

[8]

[9]

[10] [11] [12] [13] [14]

[15]

[19]

[20] [21]

[22]

References [1] G.E. Raskob, P. Angchaisuksiri, A.N. Blanco, et al., Thrombosis: a major contributor to global disease burden, Thromb. Res. 134 (2014) 931–938. [2] A.T. Cohen, G. Agnelli, F.A. Anderson, et al., Venous thromboembolism (VTE) in Europe. The number of VTE events and associated morbidity and mortality, Thromb. Haemost. 98 (2007) 756–764. [3] H.R. Yusuf, J. Tsai, H.K. Atrash, S. Boulet, S.D. Grosse, Venous thromboembolism in adult hospitalizations–United States, 2007–2009, MMWR Morb. Mortal. Wkly Rep. 61 (2012) 401–404. [4] C. Kearon, E.A. Akl, J. Ornelas, et al., Antithrombotic therapy for VTE disease: CHEST guideline and expert panel report, Chest 149 (2016) 315–352. [5] P. Prandoni, A.W. Lensing, A. Cogo, et al., The long-term clinical course of acute deep venous thrombosis, Ann. Intern. Med. 125 (1996) 1–7. [6] P.M. Ridker, S.Z. Goldhaber, E. Danielson, et al., Long-term, low-intensity warfarin therapy for the prevention of recurrent venous thromboembolism, N. Engl. J. Med. 348 (2003) 1425–1434. [7] C. Kearon, J.S. Ginsberg, M.J. Kovacs, et al., Extended low-intensity anticoagulation for thromboembolism. Comparison of low-intensity warfarin therapy with

[23]

[24]

[25]

[27]

[28]

129

conventional-intensity warfarin therapy for long-term prevention of recurrent venous thromboembolism, N. Engl. J. Med. 349 (2003) 631–699. S. Schulman, R.J. Beyth, C. Kearon, M.N. Levine, Hemorrhagic complications of anticoagulant and thrombolytic treatment: American College of Chest Physicians evidence-based clinical practice guidelines (8th edition), Chest 133 (2008) 257S–298S. P.E. Marik, R. Cavallazzi, Extended anticoagulant and aspirin treatment for the secondary prevention of thromboembolic disease: a systematic review and metaanalysis, PLoS One 10 (2015) e0143252, , http://dx.doi.org/10.1371/journal.pone. 0143252. The EINSTEIN Investigators, Oral rivaroxaban for symptomatic venous thromboembolism, N. Engl. J. Med. 363 (2010) 2499–2510. S. Schulman, C. Kearon, A.K. Kakkar, et al., Extended use of dabigatran, warfarin, or placebo in venous thromboembolism, N. Engl. J. Med. 368 (2013) 709–718. G. Agnelli, H.R. Buller, A. Cohen, et al., Apixaban for extended treatment of venous thromboembolism, N. Engl. J. Med. 368 (2013) 699–708. The EINSTEIN-PE Investigators, Oral rivaroxaban for the treatment of symptomatic pulmonary embolism, N. Engl. J. Med. 366 (2012) 1287–1297. M.H. Prins, A.W. Lensing, R. Bauersachs, et al., Oral rivaroxaban versus standard therapy for the treatment of symptomatic venous thromboembolism: a pooled analysis of the EINSTEIN-DVT and PE randomized studies, Thromb. J. 11 (2013) 21. C. Chai-Adisaksopha, M. Crowther, T. Isayama, et al., The impact of bleeding complications in patients receiving target-specific oral anticoagulants: a systematic review and meta-analysis, Blood 124 (2014) 2450–2458. J.I. Weitz, R. Bauersachs, J. Beyer-Westendorf, et al., Two doses of rivaroxaban versus aspirin for prevention of recurrent venous thromboembolism. Rationale for and design of the EINSTEIN CHOICE study, Thromb. Haemost. 114 (2015) 645–650. J.I. Weitz, A.W. Lensing, M.H. Prins, et al., Rivaroxaban or aspirin for extended treatment of venous thromboembolism, N. Engl. J. Med. 376 (2017) 1211–1222. S. Schulman, C. Kearon, Subcommittee on control of anticoagulation of the scientific and standardization committee of the international society on thrombosis and haemostasis. Definition of major bleeding in clinical investigations of antihemostatic medicinal products in non-surgical patients, J. Thromb. Haemost. 3 (2005) 692–694. S. Kaatz, D. Ahmad, A.C. Spyropoulos, S. Schulman, Definition of clinically relevant non-major bleeding in studies of anticoagulants in atrial fibrillation and venous thromboembolic disease in non-surgical patients: communication from the SSC of the ISTH, J. Thromb. Haemost. 13 (2015) 2119–2126. J.A. Kline, D. Jimenez, D.M. Courtney, J. Ianus, L. Cao, A.W. Lensing, M.H. Prins, P.S. Wells, Comparison of four bleeding risk scores to identify rivaroxaban-treated patients with venous thromboembolism at low risk for major bleeding, Acad. Emerg. Med. 23 (2016) 144–150. B. Levitan, Z. Yuan, A.G.G. Turpie, et al., Benefit–risk assessment of rivaroxaban versus enoxaparin for the prevention of venous thromboembolism after total hip or knee arthroplasty, Vasc. Health Risk Manag. 10 (2014) 157–167. A.S. Barnett, D.D. Cyr, S.G. Goodman, et al., Net clinical benefit of rivaroxaban compared with warfarin in atrial fibrillation: results from ROCKET AF, Int. J. Cardiol. 257 (2018) 78–83. W. Ageno, A. Samperiz, R. Caballero, et al., Duration of anticoagulation after venous thromboembolism in real world clinical practice, Thromb. Res. 135 (2015) 666–672. J. Simes, C. Becattini, G. Agnelli, J.W. Eikelboom, A.C. Kirby, R. Mister, P. Prandoni, T.A. Brighton, Aspirin for the prevention of recurrent venous thromboembolism: the INSPIRE collaboration, Circulation 23 (130) (2014) 1062–1071.