4.1. Summary of Available Evidence
Two published, manufacturer-sponsored, open-label, randomized controlled trials (RCTs) were included in this systematic review: EINSTEIN DVT4,5 and EINSTEIN PE.6,7 The two trials evaluated the non-inferiority of rivaroxaban versus an established standard treatment with enoxaparin and a VKA. A meta-analysis was conducted to estimate the NIM; the analysis provided four estimates for the NIM that ranged from 1.54 to 2.0, depending on the effect estimator and scale. The manufacturer selected the most liberal of these estimates: a NIM of 2.0 for the upper limit of the 95% CI associated with the hazard ratio; however, no clear rationale for this choice was provided. A review of the basis of this estimate revealed some discrepancies in data and methods used for the NIM estimation; re-evaluation of the NIM estimate by the CDR reviewers suggested a NIM that ranges from 1.58 and 1.69.
Enoxaparin was administered concomitantly with oral warfarin or acenocoumarol at a dose adjusted to maintain an INR between 2.0 to 3.0, and the enoxaparin was to be discontinued after at least five days of treatment, when the target INR was attained for two consecutive days. Warfarin was the drug selected in most patients in the enoxaparin/VKA treatment groups. Based on the patient’s risk profile and local treatment guidelines, patients were assigned to an intended treatment duration of three, six, or 12 months at the time of randomization. All confirmed events were considered up to one month after the end of the planned treatment period, irrespective of the actual treatment duration.
Both EINSTEIN trials likely had appropriate randomization and allocation concealment strategies, similar treatment groups at baseline, as well as balanced withdrawal rates that were ≤ 20% across treatment groups. Inclusion and exclusion criteria appeared relevant and reasonable, and baseline characteristics suggested that the trial population was likely representative of most real-life patients. Although not ideal, the open-label design is unlikely to have influenced outcome measurement, considering the adjudication of events by a blinded central independent committee.
Patients treated with VKA were within the therapeutic range (INR of 2.0 to 3.0) 58% and 63% of the time in the EINSTIEN DVT and PE trials, respectively. This less than optimal time in the therapeutic range may have biased the results in favour of rivaroxaban; however, this suboptimal TTR is generally reflective of clinical practice, which improves the generalizability of the results.
No data are available to compare the benefits and risks of rivaroxaban with a standard treatment option for more than 12 months, although several patients with irreversible risk factors, such as anticardiolipin antibody, may require chronic therapy. Moreover, patients with idiopathic VTE have moderate to high risk of recurrent events if anticoagulation is stopped (about 10% yearly).40 Between 25% to 46% of first-time VTE patients may have idiopathic VTE and, thus, be eligible for prolonged treatment.41
Another limitation to generalizability is that only a small proportion of patients with active cancer were included in the two trials. Therefore, the findings from the two included trials are not generalizable to cancer patient populations.
4.2. Interpretation of Results
4.2.1. Efficacy
All-cause mortality was a secondary outcome in the EINSTEIN trials. The overall incidence of death was similar for rivaroxaban and enoxaparin/VKA. However, individual trials showed slight differences in death rates between the study treatments, which did not reach statistical significance in either trial.
The primary efficacy outcome in EINSTEIN DVT was symptomatic recurrent VTE, defined as the composite of recurrent DVT or non-fatal or fatal PE. All confirmed events centrally adjudicated by a blinded committee were considered up to one month after the end of the intended treatment period, irrespective of the actual treatment duration.
Results from EINSTEIN DVT met the pre-specified and CDR-estimated NIMs for the upper limit of the two-sided 95% CI associated with the observed hazard ratio for symptomatic recurrent VTE. Results were consistent in both the PP and the ITT populations. As a result, the two treatment options may be considered effective to prevent recurrent VTE in patients with DVT. However, the EINSTEIN PE trial achieved non-inferiority with the pre-specified margin (hazard ratio of 2.0), which was the highest value in the manufacturer’s range of estimated NIMs.8 However, it would fail the non-inferiority criterion if the choice of NIM were conservative (i.e., the lower values of the estimated NIMs from both the manufacturer’s [1.54]8 and CDR’s [1.58] calculations). This difference in rivaroxaban performance between DVT and PE patients was even more obvious when recurrence of DVT and PE was assessed individually in the two trials. Although the EINSTEIN trials were not powered to assess the individual components of the composite outcome, rivaroxaban had better performance in preventing DVT than in preventing PE. Results for the secondary efficacy outcome of recurrent VTE/all-cause mortality were consistent with those for the primary outcome.
While an acute episode of provoked (secondary) VTE is usually treated with anticoagulant therapy for an initial period of three months,3 there is uncertainty around the optimal duration after completing three months of initial therapy in the case of unprovoked or recurrent VTE.42,43 For example, Shulman et al. treated patients who had a second VTE event with either six months or indefinite anticoagulant duration.36 The follow-up of these patients for four years showed a statistically significant recurrence prevention with the prolonged anticoagulant treatment. In the EINSTEIN trials, the majority of patients had a planned duration of treatment of six months, and most of these patients completed the treatment period while receiving the study drug. In this subgroup, results for the primary outcome of recurrent VTE were similar between rivaroxaban and enoxaparin/VKA. The EINSTEIN PE trial provided similar recurrence rates for the two treatments when they were used for 12 months.
Hospitalization was reported in the two trials as part of composite health care resource utilization. In general, resource utilization data did not suggest any major differences between rivaroxaban and enoxaparin/VKA in terms of visits to health care providers, hospitalizations and duration of stay, intensive care unit admissions, rehabilitations, or diagnostic tests. Some small differences were noted between treatment groups; for example, the duration of the hospital stay was often numerically shorter with rivaroxaban than with enoxaparin/VKA. On the other hand, slightly higher proportions of rivaroxaban patients required hospitalization due to recurrent VTE, as well as a numerically higher number of diagnostic tests in the context of suspected VTE and bleeding episodes. However, no statistical analysis was reported, and small sample sizes often limited the interpretation of the results. In addition, there appear to be discrepancies between health care resource utilization data reported in EINSTEIN DVT and Canadian clinical practice: in Canada, few patients with DVT are hospitalized, and LMWH and warfarin are generally administered and monitored in an outpatient setting.
4.2.2. Harms
The overall incidence of AEs, SAEs, and WDAEs during EINSTEIN DVT did not differ significantly between rivaroxaban and enoxaparin/VKA and were not higher than would be expected in this patient population in clinical practice. Although data on hepatotoxicity have led to concern about some oral anticoagulant drugs, the analysis of liver-related events reported by the investigators did not indicate any particular signal related to liver safety for patients treated with rivaroxaban.44 Similarly, cardiovascular events were infrequent (< 1%), and their analysis did not provide signals of rebound events after cessation of rivaroxaban.44
Clinically relevant bleeding was the primary safety outcome and was defined as the composite of major or clinically relevant non-major bleeding. Results for this outcome were similar between treatment groups. Slightly lower event rates were observed with the use of rivaroxaban than with the comparator for major bleeding in the EINSTEIN DVT trial (14 [0.8%] versus 20 [1.2%], respectively); however, in the EINSTEIN PE trial, major bleeding events were significantly lower with rivaroxaban than with comparator treatment (26 [1.1%] versus 52 [2.2%], respectively). However, this particular outcome occurred infrequently, and the EINSTEIN trials were not powered to detect a difference in this individual outcome. Therefore, findings from the included trials suggest that rivaroxaban was likely similar to enoxaparin/VKA with regard to bleeding events.
The management of hemorrhagic occurrences is complicated by the lack of an antidote to reverse the effects of rivaroxaban. Because the drug has a half-life of approximately 5 to 13 hours,2 major bleeding may persist for some time even after the drug is stopped. Hemorrhagic complications would have to be managed through appropriate symptomatic treatment, such as mechanical compression, surgical hemostasis, fluid replacement and hemodynamic support, as well as blood products or platelets, until the bleeding can be controlled.2 This is a general concern with the new anticoagulant drugs, considering also the particularities and issues with blood-product supplies.
In a double-blind, randomized, placebo-controlled extension trial with a maximum follow-up of one year, the incidence of death and major bleeding was low and did not differ significantly between rivaroxaban 20 mg daily and placebo.5 There were no reports of fatal bleeding.5
