Approximately 20 to 25 percent of those with an unprovoked venous thromboembolism (VTE) experience a recurrence within five years of the initial event. If it were possible to predict those at risk for recurrence, then, in the absence of unacceptable bleeding risk, anticoagulation might be continued “long-term,” rather than for the three to six months recommended in the CHEST 2012 guidelines.1 Determining optimal duration of anticoagulation, however, is challenging, as VTE recurrence declines and bleeding risk increases over time.2 Further, while individual VTE recurrence risk can be estimated based on individual risk factors (e.g., sex or D-dimer),3 the relationship to bleeding risk, on an individual basis, is unknown as clinical trials have been insufficiently powered to generate evidenced-based recommendations.
To address this problem, Tosetto and colleagues conducted a meta-analysis on pooled data from seven large prospective clinical trials that involved 1,818 patients with unprovoked VTE treated for a minimum of three months with a vitamin K antagonist. Comparing predictors of VTE recurrence and anticoagulation bleeding data, they modeled recurrence risk using Cox regression coefficients to develop a VTE recurrence prediction score. The study was limited to unprovoked proximal VTE or pulmonary embolism, excluding thromboembolic events that occurred during the puerperium period or in the setting of trauma, surgery, immobilization, cancer, or pregnancy. Patients using hormonal therapy at the time of VTE and those with thrombophilia were included, as hormonal therapy was considered a weak risk factor for VTE, and thrombophilia was considered a weak risk factor for recurrence.
DASH Prediction Score Derived From Cox Regression Analysis
| DASH Predictors(N = 1,818 VTE cases) . | β coefficient* . | P-value . | Recurrence score . |
|---|---|---|---|
| 1. D-dimer abnormal, after stopping AC | 0.96 | <0.0001 | +2 |
| 2. Age < 50 yr | 0.43 | 0.002 | +1 |
| 3. Sex - male | 0.58 | <0.0001 | +1 |
| 4. Hormone use at VTE onset | -1.05 | 0.002 | -2 |
| DASH Prediction Rule . | |||
| DASH Score | ≤ 1.0 | 2.0 | ≥ 3.0 |
| Annualized VTE Recurrence Rate | 3.1% | 6.4% | 12.3% |
| DASH Predictors(N = 1,818 VTE cases) . | β coefficient* . | P-value . | Recurrence score . |
|---|---|---|---|
| 1. D-dimer abnormal, after stopping AC | 0.96 | <0.0001 | +2 |
| 2. Age < 50 yr | 0.43 | 0.002 | +1 |
| 3. Sex - male | 0.58 | <0.0001 | +1 |
| 4. Hormone use at VTE onset | -1.05 | 0.002 | -2 |
| DASH Prediction Rule . | |||
| DASH Score | ≤ 1.0 | 2.0 | ≥ 3.0 |
| Annualized VTE Recurrence Rate | 3.1% | 6.4% | 12.3% |
*Cox regression coefficients after backward elimination and optimism correction
Table adapted from Tosetto A, Iorio A, Marcucci M, et al. J Thromb Haemost. 2012;366:1019-1025.
For each individual subject, a VTE recurrence score was calculated based on a sum of scores of four predictors: age, sex, hormone use at the time of the VTE, and D-dimer measured three weeks after stopping oral anticoagulation. The final prognostic score was based on regression coefficients for each predictor (Table). A score of +2 was assigned for abnormal D-dimer, +1 for age < 50 years old, +1 for male sex, and -2 for hormone use at onset of VTE (among women), generating a D2A1S1H-2 score. Therefore, for individuals with an initial unprovoked VTE and a DASH score < 1, the annual VTE recurrence risk is 3.1 percent (Table), which justifies stopping anticoagulation after three to six months, assuming (based on approximation of the annual bleeding incidence) that a VTE recurrence rate of less than 5 percent is an acceptable risk. In contrast, a DASH score of > 2 is associated with a VTE recurrence risk of 6.4 percent or higher, a risk level sufficiently high to warrant prolonged anticoagulation if the bleeding risk is acceptable.
In Brief
Tosetto and colleagues have developed a simple prediction score for VTE recurrence based on D-dimer, age, sex, and hormone therapy. The authors suggest that approximately half of patients with unprovoked VTE fall into the low-risk category (DASH score of < 1), meaning that discontinuation of anticoagulation will be a valid alternative to lifelong anticoagulation in a large group of patients with unprovoked VTE. On the other hand, for patients with a DASH score of 2 or higher, prolonged anticoagulation appears to be indicated, as long as bleeding risk is acceptable. Limitations of the study include short follow-up (22 months), short duration of anticoagulation (6 months), and selection of subjects taking warfarin-only anticoagulation. Notably, this prediction score reflects data from study subjects receiving warfarin anticoagulation. Thus, it would be of interest to develop a recurrence risk score using data from patients receiving low-molecularweight heparins or the newer oral anti-thrombin and anti-Xa inhibitors. For women with unprovoked VTE, the reason estrogen was associated with a two-fold lower VTE recurrence rate is unclear, given that estrogen is a known risk factor for VTE. The older age of the cohort, however, raises the issue of whether the estrogen was lower-dose, post-menopausal estrogen. To address this issue, it would be helpful to reassess the prediction score using data from patients receiving different types and doses of hormonal agents.
References
Competing Interests
Dr. Ragni indicated no relevant conflicts of interest.
