ACTION-CVT
Year of Publication: 2022
Journal: Stroke
Link: https://doi.org/10.1161/STROKEAHA.121.037541
PDF: https://www.ahajournals.org/doi/10.1161/STROKEAHA.121.037541
PDF: Clinical Question
In patients with cerebral venous thrombosis, are DOACs as effective as warfarin for preventing recurrent venous thrombosis and do they have a favorable safety profile?
Study Overview
Objective
Direct oral anticoagulants vs warfarin — to compare effectiveness and safety in a real-world multicenter cohort of patients with cerebral venous thrombosis across 27 centers in 4 countries.
Study Summary
- Retrospective multicenter cohort of 845 CVT patients at 27 centers in US, Europe, New Zealand (2015-2020).
- Rate of recurrent venous thrombosis (5.68/100 patient-years overall) was similar with DOACs vs warfarin (aHR 0.94; p=0.84).
- Major hemorrhage was lower with DOACs (aHR 0.35; 95% CI 0.15-0.82; p=0.02) — the main clinical advantage.
- Recanalization rates (partial or complete) were similar: 86.0% DOAC vs 84.1% warfarin (p=0.56).
- Propensity score matching and multiple sensitivity analyses were directionally consistent.
- Largest real-world CVT study to date — supports DOACs as a reasonable alternative to warfarin.
Intervention
Observational comparison of oral anticoagulation strategies in CVT: direct oral anticoagulants (dabigatran, rivaroxaban, apixaban) vs warfarin (INR monitoring per standard of care; no target INR range specified in paper). Treatment chosen by treating clinicians; no randomization.
Patients per Arm
DOAC only 279; Warfarin only 438; Both at different times 128 (analyzed as as-treated crossovers, not excluded). Full cohort N=845 used for clinical outcomes.
Bottom Line
In 845 patients with CVT across 27 international centers, DOACs were associated with similar rates of recurrent venous thrombosis (aHR 0.94; p=0.84), death (aHR 0.78; p=0.70), and recanalization (aOR 0.92; p=0.79) as warfarin but with significantly lower major hemorrhage risk (aHR 0.35; 95% CI 0.15-0.82; p=0.02). Provides the largest real-world dataset supporting DOACs as a reasonable alternative to warfarin for CVT.
Major Points
- Retrospective multicenter international observational study at 27 centers in US, Italy, Switzerland, New Zealand (Yaghi 2022)
- N=1025 CVT patients screened; 845 met inclusion (CVT treated with oral anticoagulation, excluding APS, active cancer, pregnancy)
- 33.0% DOAC only, 51.8% warfarin only, 15.1% both at different times (crossovers analyzed as-treated; full N=845 used for clinical outcomes)
- Median follow-up 345 days (IQR 140-720); primary analysis via inverse probability of treatment weighted Cox regression
- Primary outcome: recurrent venous thrombosis — aHR 0.94 (95% CI 0.51-1.73); p=0.84 — SIMILAR (5.26 DOAC vs 5.87/100 PY warfarin)
- Secondary safety outcome: major hemorrhage — aHR 0.35 (95% CI 0.15-0.82); p=0.02 — LOWER with DOACs (2.44 DOAC vs 4.70/100 PY warfarin)
- All-cause death: aHR 0.78 (95% CI 0.22-2.76); p=0.70 — similar (1.81 DOAC vs 1.90/100 PY warfarin)
- Recanalization (partial/complete): aOR 0.92 (95% CI 0.48-1.73); p=0.79 — similar (adjusted model N=448; the full recanalization analysis cohort of 525 had unadjusted OR 1.16, p=0.56)
- Propensity score matching with replacement (N=720) was directionally consistent but non-significant for major hemorrhage (aHR 0.42; 95% CI 0.16-1.06; p=0.07); matching without replacement (N=532) was significant (aHR 0.34; 95% CI 0.12-0.95; p=0.04)
- Sensitivity analyses excluding deep CVT, baseline hemorrhage, or the antiphospholipid-antibody variable from the models, and separately excluding COVID-19 patients (n=6), all showed consistent findings
- Consistent with earlier studies (RE-SPECT CVT, meta-analyses) on efficacy; in contrast to prior work, showed a significantly LOWER major hemorrhage risk with DOACs in CVT
- Limitations: retrospective, non-blinded outcome ascertainment, unequal group sizes, heterogeneous imaging follow-up
Design
Study Type: Multicenter international retrospective observational cohort study
Randomization:
Blinding: Unblinded (non-central outcome adjudication)
Follow-up Duration: Median 345 days (IQR 140-720)
Sample Size: 845
Screened: 1025
Analyzed: 845
Analysis: Inverse probability of treatment weighted Cox regression (primary); propensity score matching with/without replacement (sensitivity); multiple adjusted models. Crossovers between drug classes analyzed as-treated.
Inclusion Criteria
- Adult patients with imaging-confirmed CVT
- Admitted between January 1, 2015 and December 31, 2020
- Treated with oral anticoagulation (warfarin or DOAC)
- Available clinical outcomes
- Identified via ICD-9/10 codes with chart and imaging confirmation
Exclusion Criteria
- Antiphospholipid antibody syndrome (warfarin typically preferred)
- Active cancer (DOACs typically preferred)
- Pregnancy (oral anticoagulation contraindicated)
- Not treated with oral anticoagulation
- For recanalization analysis: endovascular treatment or use of both drug classes before follow-up imaging
Baseline Characteristics
| Characteristic | Control | Active |
|---|---|---|
| N | 438 | 279 |
| Age | 44.3 ± 16.1 | 46.1 ± 17.4 |
| Female | 63.0% | 67.4% |
| History of VTE | 6.6% | 15.4% |
| Antiphospholipid antibody (incidental) | 12.1% | 6.8% |
| Intracranial hemorrhage at baseline | 36.4% | 37.8% |
| LMWH use | 77.9% | 33.3% |
Arms
| Field | Control | DOAC only |
|---|---|---|
| N | 438 | 279 |
| Intervention | Warfarin with INR monitoring per standard of care (no target INR range specified in the paper) | Direct oral anticoagulant (dabigatran, rivaroxaban, or apixaban) at clinician discretion |
| Duration | Median 202.5 days | Median 194 days |
Outcomes
| Outcome | Type | Control | Intervention | HR / OR / RR | P-value |
|---|---|---|---|---|---|
| Recurrent cerebral or systemic venous thrombosis on oral anticoagulation | Primary | 5.87 per 100 patient-years (warfarin) | 5.26 per 100 patient-years (DOAC); unadjusted P=0.61 | aHR 0.94 (IPTW adjusted) | p=0.84 |
| Major hemorrhage (intracranial or extracranial) | Secondary | 4.70 per 100 PY (warfarin) | 2.44 per 100 PY (DOAC) | aHR 0.35 (95% CI 0.15-0.82) | p=0.02 (LOWER with DOACs) |
| All-cause death | Secondary | 1.90 per 100 PY (warfarin) | 1.81 per 100 PY (DOAC); unadjusted P=0.97 | aHR 0.78 (95% CI 0.22-2.76) | p=0.70 |
| Partial or complete recanalization — unadjusted (n=525) | Secondary | 291/346 (84.1%) | 154/179 (86.0%) | OR 1.16 (95% CI 0.70-1.94) | p=0.56 |
| Partial or complete recanalization — adjusted (n=448) | Secondary | Reference | Similar | aOR 0.92 (95% CI 0.48-1.73) | p=0.79 |
| Unadjusted major hemorrhage | Secondary | Reference | Lower | HR 0.47 (95% CI 0.21-1.04) | p=0.06 (trend; adjusted analyses stronger) |
| Propensity score matched with replacement — major hemorrhage (N=720) | Secondary | Reference | Lower, non-significant | aHR 0.42 (95% CI 0.16-1.06) | p=0.07 |
| Propensity matched without replacement — major hemorrhage (N=532) | Secondary | Reference | Lower | aHR 0.34 (95% CI 0.12-0.95) | p=0.04 |
| Recurrent CVT within 90 days | Secondary | Reference | Similar | aHR 1.13 (95% CI 0.44-2.87) | p=0.80 |
| Excluding COVID-19 patients (n=6) - sensitivity | Secondary | Reference | Findings unchanged | findings remained unchanged | Robust |
| Major hemorrhage rate per 100 PY | Adverse | 4.70 | 2.44 | aHR 0.35 (95% CI 0.15-0.82) | p=0.02 |
| Recurrent venous thrombosis rate per 100 PY | Adverse | 5.87 | 5.26 | aHR 0.94 | p=0.84 |
| All-cause death rate per 100 PY | Adverse | 1.90 | 1.81 | aHR 0.78 | p=0.70 |
| Intracranial hemorrhage rate per 100 PY | Adverse | 3.51 (warfarin) | 1.52 (DOAC) | P=0.05; 23 ICH total (19 symptomatic, 4 asymptomatic across full cohort N=845) | |
| Major extracranial hemorrhage rate per 100 PY | Adverse | 1.15 (warfarin) | 0.91 (DOAC) | P=0.64; 9 events total across N=845 | |
| Asymptomatic hemorrhage on imaging | Adverse | Likely underdetected (ascertainment bias) | Likely underdetected | Sensitivity analysis excluding asymptomatic ICH from major hemorrhage still favored DOACs (aHR 0.40; 95% CI 0.16-0.995; P=0.049) | |
| Progression of CVT (captured in recurrence outcome) | Adverse | Captured | Captured | No difference |
Subgroup Analysis
Sensitivity analyses excluding deep CVT and baseline hemorrhage (not established predictors of recanalization/hemorrhage) confirmed similar recanalization and persistently lower major hemorrhage with DOACs (aHR 0.35; p=0.02). Because presence of antiphospholipid antibodies is not a known predictor of recurrence, sensitivity analyses excluded the antiphospholipid-antibody variable from the recurrent-VTE models (not APS-positive patients themselves) and yielded consistent findings. Excluding COVID-19 patients (n=6) and those lost to follow-up before 90 days also produced consistent findings. Time-of-recurrence subgroups (<90 days vs >90 days from anticoagulation initiation, and 2-week and 4-week cutoffs) showed similar DOAC vs warfarin effect, suggesting no differential risk window. The younger age of the cohort (mean 45 years) aligns with CVT epidemiology and with the age group showing greatest DOAC safety advantage over warfarin in other contexts.
Criticisms
- Retrospective observational design — residual confounding cannot be excluded despite IPTW and propensity matching
- Non-central, non-blinded outcome ascertainment — imaging and event adjudication varied by center
- DOAC subtype heterogeneity (dabigatran, rivaroxaban, apixaban) — outcomes not stratified by agent
- Unequal group sizes (279 DOAC vs 438 warfarin) reflect practice patterns and may introduce selection bias
- Follow-up imaging timing was heterogeneous, limiting precision of recanalization analysis
- Asymptomatic hemorrhage likely underdetected due to ascertainment bias (routine follow-up imaging not standardized)
- Part of the study period overlapped the COVID-19 pandemic; only 6 CVT-with-COVID patients were included and excluding them left findings unchanged
Funding
Partially supported by the Italian Ministry of Health Ricerca Corrente — IRCCS MultiMedica. Individual author support and disclosures listed in the paper (grants from NIH, AHA, and industry consultancy for various authors).
Based on: ACTION-CVT (Stroke, 2022)
Content summarized and formatted by NeuroTrials.ai.