Prediction of Recurrent Venous Thromboembolism and Arterial Cardiovascular Events after Discontinuation of Anticoagulation: The R-VTE-predict and MACE-predict Risk Scores.

Patients who had venous thromboembolism (VTE) are not only at increased risk of recurrent VTE but also of major adverse cardiovascular events (MACEs) than the general population. Therefore, the prediction of the risk of these events is important for a tailored prevention and mitigation strategy. We aimed to develop simple scores to estimate recurrent VTE and MACE risks after the discontinuation of anticoagulation in a large cohort of individuals who suffered VTE (EDITH cohort). The primary endpoints were recurrent symptomatic VTE and MACE (composite of non-fatal acute coronary syndrome, stroke and cardiovascular death). Arterial thrombotic event (ATE) exclusively was also considered. Independent predictors of main outcomes were derived from multivariable Cox regression models. Weighted integer points based on the effect estimate of identified predictors were used to derive the final risk scores. A total of 1,999 participants (mean age: 54.78 years, 46.4% male, 43.6% unprovoked VTE) were included in the derivation cohort and 10,000 in the validation cohort (built using bootstrapping). During a median post-anticoagulation follow-up of 6.9 years, recurrent VTE occurred in 29.5% of participants and MACE in 14.8%. Independent predictors of recurrent VTE were male sex, age >65 years, cancer-associated VTE, and unprovoked VTE (vs. transient risk factor-associated VTE). Independent predictors of MACE were age >65 years, cancer-associated VTE, hypertension, renal insufficiency, and atrial fibrillation. The risk of recurrent VTE (moderate vs. low: hazard ratio [HR]: 2.62, 95% confidence interval [CI]: 2.06-3.34; high vs. low: HR: 3.78, 95% CI: 2.91-4.89), MACE (moderate vs. low: HR: 6.37, 95% CI: 3.19-12.69; high vs. low: HR: 12.32, 95% CI: 6.09-24.89), and ATE (based on MACE-predict risk score) increased gradually from the lowest to highest of the respective prediction risk score groups. These results were confirmed in the validation cohort with overall reasonable models' discrimination performance (recurrent VTE C-statistic: 0.62-0.63, MACE and ATE C-statistic: 0.72-0.77). Contemporary simple risk scores based on readily available clinical characteristics can reasonably predict the risk of recurrent VTE and MACE after the discontinuation of anticoagulation. These findings may influence the choice of anticoagulation strategy after the acute phase of VTE and, therefore, need confirmation by further studies.