The risk of recurrence is the highest during the first weeks after an acute venous thromboembolism (VTE) and then declines over time. Recurrence can be prevented by antithrombotic therapy, but the price is severe or sometimes even fatal bleeding. Determining the optimal duration of anticoagulation entails balancing the risk of bleeding against the risk of recurrence, that is, to stop anticoagulation once the risk of bleeding exceeds the risk of recurrence. There are two basic approaches to stratify patients regarding their recurrence risk: identification of patients with a recurrence risk that is lower than the annual risk of severe or fatal bleeding (which is approximately 4% and 0.2%, respectively), or to identify patients with a recurrence risk that is higher than the bleeding risk during anticoagulation. It is well recognized that patients with VTE provoked by surgery, trauma, immobilization, pregnancy or female hormone intake are at low risk of recurrence . All other patients are classified as having unprovoked (also called idiopathic) VTE, and their recurrence risk is approximately 5–7% per year (Fig. 1). About one quarter of patients will suffer recurrence within 5 years, while three quarters will stay recurrence-free. During the last decades, the panel of risk factors of VTE has continuously grown, and their impact amongst others on the risk of recurrence will be discussed.
Summary. Despite considerable advances in the identification of new risk factors for VTE, predicting the risk of recurrence in an individual patient remains a challenge. The impact of many of these risk factors on the recurrence risk is moderate, while the relevance of others is unknown or is regarded controversial. The determination of some markers of thrombophilia either lacks standardization or is too elaborate for routine purposes. Many patients carry more than one risk factor and their combined effect is unknown. Use of global coagulation markers that encompass the effects of clotting and/or fibrinolytic defects may improve risk assessment. In a future step, prothrombotic coagulation alterations need to be integrated with clinical risk factors. Reproducible and simple scoring systems may improve stratification of patients according to their recurrence risk and optimize duration of anticoagulant therapy.
Location of VTE
In one of the first trials on long-term treatment of VTE, all patients with recurrence had initial proximal deep vein thrombosis (DVT), while none with distal DVT had recurrence . Meanwhile, the low recurrence risk of patients with calf vein thrombosis is well recognized (Fig.1). The risk of recurrence is also low in patients with DVT of the upper extremities .
Whether the recurrence risk is higher among patients with pulmonary embolism (PE) is unknown. In the Austrian Study on Recurrent Venous Thromboembolism (AUREC), patients with symptomatic PE had a 2.2-fold (95% CI, 1.3–3.7) higher risk of recurrence than patients with isolated DVT , and had a high risk of PE at recurrence. While some studies also reported an increased recurrence risk associated with PE [5,6], others did not [7,8]. These differences can partially be explained by classification of VTE and inclusion of patients with asymptomatic PE. The relationship between increased risks of PE at recurrence in patients whose first event was PE, has, however, been unanimously confirmed [5–8]. Overall, about 60% of patients with initial PE will have another PE at recurrence.
The risk of recurrence is high among patients with multiple VTE. In a Swedish prospective cohort study, the incidence of recurrence after 5 years was 21.5% (95% CI, 17.7–25.4%) after a first and 27.9% (95% CI, 19.7–36.1%) after a second DVT . A higher risk of recurrence after multiple venous thromboses was also shown in interventional trials . It remains unknown if the increased recurrence risk is also true for patients with multiple provoked VTE. Recurrent ipsilateral DVT increases the likelihood of developing the postthrombotic syndrome, which in turn predicts recurrence [10,11]. Vena caval filters increase the risk of recurrent DVT up to 1.5-fold , and thromboses at the filter site are found in more than 10% of patients.
The risk of recurrence is determined by the patient’s gender. The AUREC investigators were the first to show that men have a higher recurrence risk than women . In a meta-analysis, the pooled estimate of the relative risk of recurrent VTE for men compared with women was 1.6 (95% CI, 1.2–2.0) . An association between increased body weight and recurrence risk was found in retrospective and prospective studies [15 for review] and seems to be linear. Data on the effect of age at the time of first VTE on the recurrence risk are limited and controversial. While an increased risk has been found in some studies [16–18], others failed to confirm an association [9,19]. A positive family history might indicate the presence of known or even unknown genetic defects. However, in a single study, a positive family history did not predict recurrence .
Underlying diseases and drug intake
Patients with cancer have a 3-fold increased risk of recurrence during the first year after an acute VTE, despite anticoagulant treatment . The risk of recurrence of cancer patients who discontinue anticoagulation is regarded as high [7,17]. The pathogenesis of VTE in malignancy is multifactorial, and the spectrum of risk factors is wide. Subgroups of cancer patients with a lower or higher recurrence risk are poorly defined. Metastatic disease and chemotherapy increase the risk for a first VTE and are also regarded as risk factors for recurrence.
Evidence is increasing that the metabolic syndrome increases the risk of VTE. With the exception of obesity (see above), only dyslipidemia has been evaluated. In a prospective cohort study, patients with high levels of apolipoprotein AI or high density lipoproteins had a low risk of recurrence . In a case-control study, lipoprotein(a) levels (>300 mg/L) conferred a 2-fold increased recurrence risk .
In the Leiden Thrombophilia Study (LETS), women with VTE who continued to use oral contraceptives had a higher recurrence rate (28.0/1000 patient-years; 95% CI, 15.9–49.4) than those who did not (12.9/1000 patient-years; 95% CI, 7.9–21.2) . The risk of recurrence is also increased by continued hormone replacement therapy .
Residual vein thrombosis (RVT) and impaired fibrinolysis
Studies on the impact of RVT on the recurrence risk gave conflicting results . In a recent controlled trial, patients with RVT were randomized to either stop or continue oral anticoagulants. In patients without RVT, anticoagulation was stopped and only 1 patient (1.3%; 0.63% person years) had recurrence. Compared with patients without RVT, those with RVT had an adjusted hazard ratio (HR) of 24.9 (95% CI, 3.4–183.6; P = 0.002) . Defects in the fibrinolytic system are of minor importance regarding recurrent VTE [27–29].
Natural coagulation inhibitor deficiencies
Estimates of the recurrence risk in patients with antithrombin, protein C or protein S deficiency are mainly derived from retrospective analyses. The risk seems to be the highest among patients with antithrombin deficiency [30,31]. In LETS, the HR of recurrent VTE for patients with a coagulation inhibitor deficiency was 1.8 (95% CI, 0.9–3.7) . Low levels of TFPI also increase the risk of recurrent VTE .
Antiphospholipid antibodies (APLA)
Evidence for an increased risk of recurrence is weak [33 for review]. Many studies are retrospective, include both arterial and venous thrombotic events and/or use different definitions for the presence of APLA. In a retrospective study, almost half of the patients with either arterial or venous thrombosis and APLA had recurrence after discontinuation of warfarin. In a subgroup analysis of the DURAC studies, the incidence of recurrence among patients with a first VTE and the presence of anticardiolipin antibodies was 29%, compared with 14% among patients without these antibodies (P < 0.01). The risk of recurrence in patients with APLA seems to be lower after VTE than after arterial thrombosis.
Elevated coagulation factors
An association between high factor VIII levels and an increased risk of recurrence was found in several studies [34–36], but the association was weaker  or absent  in other cohorts. Elevated levels of fibrinogen  or factor IX  may also increase the risk of recurrence.
Defects in homocysteine metabolism
The risk of recurrence associated with hyperhomocysteinemia has been evaluated in prospective cohort studies [24,39] and in one interventional trial  and is about 1.5-fold. Several vitamins are involved in the homocysteine metabolism, and an independent prothrombotic effect of these vitamins has been suggested. Indeed, low vitamin B6 levels predicted an increased risk of recurrent VTE that remained unaffected by adjustment for homocysteine .
Gene mutations and polymorphisms
The first publications on the risk of recurrence in patients with a factor V Leiden (FVL) or the factor II (FII) G20210A mutation gave conflicting results. In a recent meta-analysis, the risk ratios of recurrence for heterozygous FVL or FII G20210A were 1.39 (95% CI, 1.15–1.67) and 1.20 (95% CI, 0.89–1.61), respectively . No data are available for homozygous carriers of either mutation or their risk of recurrent VTE. An increased risk of recurrence was found in patients homozygous for Ser128Arg in the E-selectin gene (HR 4.1; 95% CI, 1.5–11.4) , and in heterozygous carriers of long GT-repeat alleles in the hemoxygenase 1 gene [relative risk 2.2 (95% CI, 1.4–3.4)] .
Global markers of coagulation
VTE is a multicausal disease, and the thrombotic risk increases with the number of risk factors. Risk estimates for recurrent VTE are available for some combinations, including double heterozygosity for FVL and FII G20210A , combined elevations of clotting factors [29,38], combined defects in patients with FVL  or a natural coagulation inhibitor deficiency . The idea to estimate the risk of recurrence by measuring global markers of coagulation that might reflect multifactorial thrombophilia is intriguing. Indeed, patients with a short activated partial thromboplastin time (aPTT) have a higher risk of recurrence than those with a longer aPTT . Stratification of patients according to their recurrence risk can be accomplished by an assay that measures the overall function of the protein C pathway .
Discrimination of patients into groups of high and low risk of recurrence has been achieved by measuring D-Dimer. In a systematic review, a negative D-Dimer was associated with a 3.5% annual risk of recurrence, whereas a positive result was associated with an 8.9% risk . Different D-Dimer assays were used, and all provided a precise and reliable estimate of the risk of recurrence. By use of lower cut-off levels, patients with a particularly low risk of recurrence can be identified. In AUREC, patients with a first unprovoked VTE and a D-Dimer <250 ng/mL after discontinuation of anticoagulation had a 60% lower recurrence rate than those with higher levels (HR 0.4, 95% CI, 0.2–0.8) . Two years after anticoagulation, the probability of recurrence was 4% with an upper 95% CI of 6.5%. In a Canadian study, the cut-off of 250 ng/mL proved to be useful to identify women at low recurrence risk . D-Dimer has been integrated in a randomized controlled trial for deciding on the duration of anticoagulation . The study confirms that patients with low D-Dimer after withdrawal of anticoagulation have a low risk of recurrence (4.4 recurrences/100 patient-years). Patients with high D-Dimer in whom anticoagulation was stopped after 6 months had a 5-fold higher recurrence risk than those who continued (10.9 vs. 2.0 recurrences/100 patient-years).
Stratification of patients with VTE according to their recurrence risk can be accomplished by measuring in vitro thrombin generation. In AUREC, the risk of recurrence in patients with low peak thrombin (<400 nM) was as low as 6.5% (upper limit of the 95% CI 8.9%) after 4 years . In the same cohort, an endogenous thrombin potential (ETP) ≥100% conferred a 1.6-fold higher risk of recurrence (95% CI, 1.1–2.3) . In this study, ETP and D-Dimer were independent predictors of recurrence and a combined analysis improved stratification of patients. In an Italian study, patients with high ETP or peak thrombin had an increased risk of recurrence (HR 3.4, 95% CI, 1.3–8.7, and 4.6, 95% CI, 1.7–12.2) . An increased recurrence risk in patients with VTE conferred by high ETP was also found in a British study (HR 4.0, 95% CI, 1.3–11.8) . In LETS, however, ETP did not predict recurrence .
Global markers of coagulation are useful to assess the risk of recurrent VTE. Interventional trials in which the duration of anticoagulation is based on the levels of these markers are under way. The results will show if anticoagulants can safely be stopped in patients who are at low risk of recurrence, and should be continued in those at high risk.
Disclosure of Conflicts of Interests
The authors state that they have no conflicts of interest.