TRAPS
Rivaroxaban vs warfarin in high-risk patients with antiphospholipid syndrome
Clinical Question
Is rivaroxaban noninferior to warfarin in terms of efficacy (prevention of thromboembolic events and vascular death) and safety (major bleeding) in high-risk, triple-positive patients with thrombotic antiphospholipid syndrome?
Study Overview
Objective
To assess whether rivaroxaban is a safe and effective alternative to warfarin for preventing thromboembolic events in high-risk patients with antiphospholipid syndrome (APS).
Study Summary
Rivaroxaban was associated with a higher rate of thromboembolic events, especially arterial thromboses, compared to warfarin in high-risk triple-positive APS patients, leading to early termination of the trial.
Intervention
Randomized, open-label trial comparing rivaroxaban (20 mg daily) to warfarin (INR 2.0–3.0) in patients with high-risk APS (triple positivity for lupus anticoagulant, anticardiolipin, and anti-β2-glycoprotein I antibodies).
Patients per Arm
Rivaroxaban: 59, Warfarin: 61
Bottom Line
The TRAPS trial was prematurely terminated due to an excess of events (primarily arterial thromboembolic events) in the rivaroxaban arm, demonstrating that rivaroxaban was not noninferior to warfarin and was associated with increased risk in high-risk, triple-positive antiphospholipid syndrome patients.
Major Points
- TRAPS is the definitive trial showing that DOACs are dangerous in high-risk (triple-positive) antiphospholipid syndrome — rivaroxaban was associated with a 6.7-fold increase in the composite primary endpoint (thromboembolic events, major bleeding, and vascular death) vs warfarin (HR 6.7, 95% CI 1.5–30.5, P=0.01). Thromboembolic events alone occurred in 7 (12%) rivaroxaban vs 0 warfarin patients (statistical analysis not applicable per Table 4).
- Trial was stopped prematurely by DSMB after 120 of planned 536 patients enrolled, due to an excess of ARTERIAL events in the rivaroxaban arm: 4 ischemic strokes + 3 MIs vs ZERO events with warfarin.
- Triple-positive APS (positive for lupus anticoagulant + anticardiolipin + anti-β2-glycoprotein I) is the HIGHEST-RISK antibody profile — these patients have the most prothrombotic phenotype and require the most reliable anticoagulation.
- Primary composite (thromboembolic events + major bleeding + vascular death): 19% rivaroxaban vs 3% warfarin (HR 6.7, 95% CI 1.5-30.5, p=0.01). The ITT analysis was even more dramatic: 22% vs 3% (HR 7.4, p=0.008).
- Mechanism hypothesis: rivaroxaban inhibits Factor Xa but APS-related thrombosis may involve multiple coagulation pathway activation (tissue factor, complement, thrombin) that warfarin's broader inhibition (Factors II, VII, IX, X) better suppresses.
- Confirmed by similar signals in other APS-DOAC trials: RAPS (rivaroxaban, increased thrombin generation markers) and a subsequent meta-analysis showing DOACs are inferior to warfarin for APS arterial events specifically.
- Dramatically changed clinical practice: all major guidelines now RECOMMEND AGAINST DOACs in triple-positive APS. Warfarin remains standard of care, with target INR 2.0-3.0 (some experts advocate 3.0-4.0 for arterial events).
- Events were predominantly ARTERIAL (stroke and MI) rather than venous — suggesting APS arterial thrombosis has a different coagulation mechanism than venous thrombosis, which may respond to Factor Xa inhibition.
- No deaths in the 'as treated' analysis; the ITT analysis included 1 cardiovascular death in the rivaroxaban arm (occurring 433 days after study drug discontinuation, while the patient was on warfarin). Morbidity burden (4 ischemic strokes and 3 MIs) was substantial and potentially disabling.
- Important caveat: results apply to TRIPLE-POSITIVE APS only. Single- or double-positive APS patients may tolerate DOACs, but this remains unproven. Many experts still prefer warfarin for any APS with arterial events.
Design
Study Type: Prospective randomized phase 3 open-label non-inferiority study with blinded end point adjudication
Randomization: 1
Blinding: Blinded end point adjudication (patients and local investigators were not masked to treatment)
Enrollment Period: November 2, 2014, to January 25, 2018 (terminated prematurely)
Follow-up Duration: Mean follow-up 569 days ('as treated'); 611 days (ITT)
Centers: 14
Countries: Italy
Sample Size: 120
Analysis: Primary analysis to test noninferiority (upper boundary of 1-sided 95% CI for HR < 1.7); Cox proportional-hazards modeling for endpoint analyses. 'As treated' and intention-to-treat (ITT) analyses performed. Statistical significance set at P≤.05.
Inclusion Criteria
- Adult patients between 18 and 75 years of age
- Positive for all 3 antiphospholipid (aPL) tests in the last blood sampling (triple positivity: IgG and/or IgM aCL >40 GPL/MPL or >99th percentile; IgG and/or IgM aβ2GPI >40 U or >99th percentile; and LA test positive; aCL and aβ2GPI ELISA had to be positive for the same isotype)
- History of thrombosis (objectively proven arterial, venous, and/or biopsy-proven microthrombosis).
Exclusion Criteria
- Detailed exclusion criteria are not listed in the primary publication; the paper states 'Exclusion criteria are detailed elsewhere' and refers to the TRAPS design paper (Pengo V, et al. Lupus. 2016;25(3):301-306).
Arms
| Field | Rivaroxaban | Control |
|---|---|---|
| Intervention | Rivaroxaban 20 mg once daily (15 mg once daily if creatinine clearance 30-50 mL/min). Patients previously on warfarin switched when INR < 3. | Warfarin (INR target 2.5). INR maintained between 2.0 and 3.0, checked at least every 4 weeks. For anticoagulation-naive patients, maintenance dose determined by protocol. |
| Duration | Mean follow-up 569 days ('as treated'); 611 days (ITT) | Mean follow-up 569 days ('as treated'); 611 days (ITT) |
Outcomes
| Outcome | Type | Control | Intervention | HR / OR / RR | P-value |
|---|---|---|---|---|---|
| Cumulative incidence of thromboembolic events (arterial or venous), major bleeding, and vascular death. | Primary | 2/61 (3%) events ('as treated') | 11/59 (19%) events ('as treated') | 6.7 | 0.01 |
| Arterial thrombosis ('as treated') | Secondary | 0 | 7 (12%) | ||
| Ischemic stroke ('as treated') | Secondary | 0 | 4 (7%) | ||
| Myocardial infarction ('as treated') | Secondary | 0 | 3 (5%) | ||
| Venous thromboembolism ('as treated') | Secondary | 0 | 0 | ||
| Venous thromboembolism (ITT) | Secondary | 0 | 1 (2%) | ||
| Major bleeding ('as treated') | Secondary | 2 (3%) | 4 (7%) | 2.5 | 0.3 |
| Major bleeding (ITT) | Secondary | 2 (3%) | 4 (7%) | 2.3 | 0.3 |
| Death ('as treated') | Secondary | 0 | 0 | ||
| Cardiovascular death (ITT) | Secondary | 0 | 1 (2%) | ||
| Composite: thromboembolic events, major bleeding, and vascular death (ITT) | Secondary | 2 (3%) | 13 (22%) | 7.4 | 0.008 |
| Major bleeding ('as treated') | Adverse | 2 (3%) | 4 (7%) | 2.5 | 0.3 |
| Major bleeding (ITT) | Adverse | 2 (3%) | 4 (7%) | 2.3 | 0.3 |
| Clinically relevant nonmajor bleeding (led to permanent discontinuation) | Adverse | 0 | 2 | ||
| Metrorrhagia causing acute Hb fall (rivaroxaban) | Adverse | 0 | 1 | ||
| Rectorrhagia requiring transfusion (rivaroxaban) | Adverse | 0 | 1 | ||
| Gastrointestinal bleeding causing acute Hb fall (rivaroxaban) | Adverse | 0 | 1 | ||
| Hb fall (rivaroxaban) | Adverse | 0 | 1 | ||
| Provoked Hb fall (warfarin) | Adverse | 1 | 0 | ||
| Metrorrhagia requiring intervention (warfarin) | Adverse | 1 | 0 | ||
| Note on bleeding events | Adverse | — | — |
Criticisms
- Premature termination (120 of 536 planned) — drastically underpowered for noninferiority testing. The dramatic event imbalance (19% vs 3%) may partly reflect small-sample variability.
- Very small number of total events (11 rivaroxaban, 2 warfarin in 'as-treated') — wide confidence intervals (HR 6.7, CI 1.5-30.5) indicate substantial uncertainty in the true effect size.
- Open-label design — patients and investigators knew treatment assignment. This could influence INR monitoring vigilance, aspirin prescribing, and even event reporting, though endpoint adjudication was blinded.
- Single-country trial (Italy) — Italian APS management patterns, warfarin monitoring quality, and patient demographics may differ from other settings.
- Only tested rivaroxaban — cannot extrapolate to apixaban, edoxaban, or dabigatran. Different DOACs have different mechanisms, and some may perform better in APS. However, RAPS and ASTRO-APS showed similar signals.
- Noninferiority margin derived from observational data rather than a prior RCT — methodologically weaker foundation for the statistical design.
- Aspirin use was not standardized — variable antiplatelet co-therapy confounds interpretation of anticoagulant-specific effects.
- Results apply ONLY to triple-positive APS — the highest-risk subgroup (~30% of all APS patients). Single- and double-positive patients were excluded, leaving a major evidence gap.
- No mechanistic substudy — did not measure thrombin generation, anti-Xa levels, or complement activation to explain WHY rivaroxaban failed. This limits understanding of the optimal anticoagulation target in APS.
Subgroup Analysis
Array
Funding
Bayer S.p.A.
Based on: TRAPS (Blood, 2018)
Authors: Vittorio Pengo, Gentian Denas, Giacomo Zoppellaro, ..., and Alessandra Banzato
Citation: Blood. 2018;132(13):1365-1371
Content summarized and formatted by NeuroTrials.ai.