Dr Trevor Baglin, Department of Haematology, Addenbrooke’s NHS Trust, Cambridge CB2 0QQ, UK. Tel.: 44 (0)1223 217128; Fax: 44 (0)1223 274871; E-mail: firstname.lastname@example.org
Characterisation of heritable thrombophilic defects has facilitated an understanding of the complex mechanisms influencing risk of venous thromboembolism. In parallel with this, the importance of gene–environment interaction in the development of this disease has become apparent. However, testing for a limited number of heritable thrombophilic defects (first generation thrombophilia testing) has not been shown to predict likelihood of recurrent venous thrombosis to any useful degree. This paradox whereby thrombophilia testing identifies defects associated with an increased risk of a first venous thrombosis but not of a particularly high risk of recurrence is likely the result of limitations imposed by a limited dichotomous testing strategy compounded by test inaccuracy and imprecision. Consequently, the observed intermediate phenotype (defined by limited dichotomous testing) is not concordant with the risk of recurrent venous thrombosis. Whilst a simple dichotomous testing strategy for a limited number of heritable thrombophilic defects has not been shown to have useful clinical predictive value, proof-of-principle is emerging for testing of multiple genetic factors in predicting the likelihood of recurrent thrombosis. In addition, recent studies indicate that measurement of the global activity of the coagulation system using either biomarkers or measuring the thrombin generating potential (second generation thrombophilia testing) may have useful clinical predictive value for recurrent thrombosis. The assessment of the intermediate phenotype by global coagulation tests and genome-wide mutation and SNP (single nucleotide polymorphisms) detection may provide complimentary approaches to the quantification of risk of recurrence and enable a move towards more patient-focussed rather than disease-focussed care.
Venous thrombosis (VT), or venous thromboembolism (VTE), comprises deep vein thrombosis (DVT) with or without symptomatic pulmonary embolus (PE). The incidence of a first episode of VTE is 1.5 per 1000 person-years (Naess et al., 2007) with a per-person lifetime incidence of 5% (Silverstein et al., 1998). After a first episode of VTE, patients are 40 times more likely to suffer a further event compared with previously unaffected individuals (Kearon, 2003). Recurrent DVT or PE can be fatal. In the RIETE registry, fatal PE occurred in 12% of patients presenting with massive PE, 3% of patients with nonmassive PE and 0.6% of patients with DVT alone (Laporte et al., 2008). As the majority of patients with PE are normotensive at presentation, actually 90% of early deaths attributable to PE occur in patients who do not present with massive PE. Preventing recurrent VTE, therefore, prevents death. Recurrent DVT is a major risk factor for postthrombotic syndrome (Kahn et al., 2008), and recurrent PE increases the risk of chronic thromboembolic pulmonary hypertension more than tenfold (Pengo et al., 2004). Therefore, preventing recurrent VTE significantly reduces the burden of disease.
Continued treatment with an oral vitamin K antagonist (VKA), such as warfarin, or direct thrombin inhibitor will prevent more than 95% of recurrent episodes of VTE (Schulman et al., 2009). VKA therapy is monitored by the International Normalised Ratio (INR) with a target INR of 2.5 for the majority of patients, and the risk of VTE recurrence during treatment is higher when the INR is subtherapeutic (Palareti et al., 1997). Importantly, VTE is only prevented for as long as anticoagulation is continued, and so, anticoagulation must be continued indefinitely in patients at high risk to prevent recurrence. In view of the risk of anticoagulant-related bleeding, there is a need to improve risk stratification for recurrent VTE and hence target long-term therapy at patients considered to be at greatest risk. Similarly, accurate individualised estimates of anticoagulant-related bleeding are required to estimate the benefit and risk trade-off associated with continued anticoagulation. The risk of recurrent VTE in an individual can now be predicted to some degree from clinical, laboratory and imaging findings. In relation to the laboratory finding, there is an increasing evidence base for the measurement of the global activity of the coagulation system, using either biomarkers or measuring the thrombin-generating potential as a method of prediction for VTE recurrence. Understanding the interaction and confounding effect of different factors is a target for research to maximise treatment benefits and minimise disease and treatment-associated risks.
The clinical circumstances of the first episode of VTE is the strongest predictor of risk of recurrence
When anticoagulant treatment is stopped after an appropriate duration (at least 3 months), the incidence of recurrent VTE varies according to the clinical risk factors associated with the initial event. Laboratory tests might be used to identify individuals in a low-risk cohort (e.g. a provoked first event) who might have an atypically high risk of recurrence and therefore might benefit from continued anticoagulation. Alternatively, tests might identify individuals in a high-risk cohort (e.g. unprovoked first event) who might have an atypically low risk of recurrence and therefore not require continued anticoagulant therapy. Until recently, it was usual practice to treat patients with a finite period of anticoagulation after a first episode of venous thrombosis not associated with cancer. Consequently, tests were evaluated for their ability to predict a high risk of recurrence that would favour long-term anticoagulation based on a balance of benefit and risk. This was a primary reason for testing for heritable thrombophilia after an episode of venous thrombosis (Greaves & Baglin, 2000) until it became apparent that these tests typically had no clinically useful predictive power (Baglin et al., 2003; Christiansen et al., 2005; Marchiori et al., 2007). Patients with an unprovoked first episode of proximal DVT of the lower limb or PE have an annual risk of recurrence greater than 5% and are now considered for long-term (lifelong) anticoagulation (Kearon et al., 2008b). Patients with provoked venous thrombosis have an annual risk of 5% or less, and long-term anticoagulation is not recommended. Now that it is recommended that long-term anticoagulant therapy be considered after a first episode of VTE the emphasis has changed and tests that predict a low individual risk of recurrence may identify patients in a high-risk group for whom long-term anticoagulant therapy can be withheld.
Quality of anticoagulation and risk of recurrence after stopping treatment
Four studies have evaluated the association between quality of initial oral anticoagulant therapy, measured as time in a defined INR range, and the risk of recurrence after stopping treatment. Linear interpolation was used to measure the percentage of time at any given INR in all 4 studies (Rosendaal et al., 1993).
Palareti et al., (2005) carried out a prospective observational cohort study involving 297 consecutive patients after unprovoked VTE to determine specifically whether the quality of oral anticoagulant control affected the incidence of recurrence after treatment was stopped. Patients were treated for at least 3 months predominantly with warfarin. For the total cohort, the average time spent with an INR <2.0 was 17.6%. The recurrence rate after stopping treatment was 8.5 per 100 patient-years (95% CI 6.2–11.3). Recurrence rates were not influenced by the duration of treatment with no difference in the rate of recurrence in groups of patients with durations from 3 months to more than 12 months. Patients with recurrence had a greater proportion of time at markedly subtherapeutic INR: 1.4% of time with INR <1.5 in first 90 days versus 0% of time in patients without recurrence, P = 0.009. When the patients were divided into quintiles according to time spent with an INR <1.5, the 20% of patients with the most time spent with an INR <1.5 (>3.1% of the time) had a 2.8-fold increased risk of recurrence (recurrences 27%vs. 11%, 16/59 vs. 26/238).
Using a similar design but a different definition of subtherapeutic anticoagulation, the time spent at a subtherapeutic INR was analysed in 266 patients from the Leiden Thrombophilia Study with either unprovoked or provoked VTE (Gadisseur et al., 2007). Patients were treated for at least 3 months predominantly with the relatively longer-acting VKA phenprocoumon. For the total cohort, the average time spent with an INR <2.0 was 9.7%. The recurrence rate after stopping treatment was 2.9 per 100 patient-years (95% CI 2.3–3.4). Time spent below the predefined INR threshold of 2.0 was not associated with a higher rate of recurrence or time to recurrence when anticoagulant therapy was stopped. An INR threshold of 1.5 was not included in the analysis, but the authors commented that very little time was spent below this threshold.
Two further studies with 200 and 182 patients with median follow-up periods of 4.9 and 2.5 years found no difference in the rate of VTE recurrence after stopping treatment in relation to time spent below or above an INR of 2.0 during treatment (Poli et al., 2007; Prandoni et al., 2007).
In conclusion, following standard care with appropriate control of anticoagulation, there is little, if any, association between the initial quality of anticoagulation and subsequent risk of recurrence after stopping treatment. There may be a slightly increased risk in a small proportion of patients with very low INRs (<1.5) in the first few weeks of treatment. It is possible that when the quality of anticoagulant care is neglected, there is an increased risk of VTE recurrence when treatment is stopped.
Heritable thrombophilia describes an inherited tendency for VTE. In addition to case-control studies, family pedigree studies have reported a risk of VTE in association with heritable thrombophilia in thrombosis-prone families. The reported rates of thrombosis in these studies are considered to be overestimates of the true risk owing to selection and publication bias. Nevertheless, a causal link between heritable thrombophilia and venous thrombosis is definite (Rosendaal, 1999; Rosendaal & Bovill, 2002). Five heritable thrombophilic defects have been shown to be unequivocally associated with an increased risk of venous thrombosis (Reitsma & Rosendaal, 2007), i.e. odds ratio of 2 or greater, namely deficiencies of antithrombin, protein C and protein S owing to mutations in the corresponding genes SERPINC1, PROC, PROS and the two common mutations F5G1691A (FV R506Q, factor V Leiden) and F2G20210A (commonly referred to as the prothrombin gene mutation). These five genetic thrombophilias typically constitute the ‘heritable thrombophilia profile’ in coagulation laboratories (Baglin et al., 2010). It is now apparent that:
• Systematic review of the risk of recurrent venous thromboembolism in patients heterozygous for the factor V Leiden mutation indicates a relative risk of recurrence of 1.4 (95% CI 1.1–1.8) and for the prothrombin G20210A mutation of 1.7 (95% CI 1.3–2.3) (Ho et al., 2006). This magnitude of increased risk is modest and does not justify an extended-duration of anticoagulation (Ho et al., 2006).
• Testing for inherited thrombophilia does reduce the incidence of VTE recurrence (Coppens et al., 2008).
• Cohort studies have demonstrated a low risk of primary events in prospectively followed affected asymptomatic relatives of patients with the factor V Leiden or prothrombin mutations, and so, asymptomatic case-finding is not recommended (Langlois & Wells, 2003; Tormene et al., 2004).
Consequently, indiscriminate testing for heritable thrombophilia is not recommended and decisions regarding duration of anticoagulation (lifelong or not) should be made with reference to whether an episode of VTE was provoked and the presence of other risk factors, regardless of whether a heritable thrombophilic defect is present (Baglin et al., 2010).
Antiphospholipid syndrome (APS) is diagnosed when thrombosis or pregnancy loss occurs in association with positive laboratory tests for antiphospholipid antibody. In contrast to the influence of heritable thrombophilia, the risk of thrombosis in patients with APS is increased in the arterial as well as in venous circulation. Whilst there is a higher rate of recurrent VTE in patients with APS, the risk benefit of long-term anticoagulation has still not been clarified. Therefore, it is recommended that the intensity and duration of treatment should be determined on an individual basis, taking into account the presence of additional risk factors, the severity of the presenting event and the risk of bleeding on warfarin, rather than just the diagnosis of APS (Greaves et al., 2000).
Global tests of hypercoagulability
Rather than dichotomise the results of a limited selection of test results, an alternative approach is to quantify risk of thrombosis in relation to the composite effect of candidate thrombophilias, regardless of whether they are known or unknown and regardless of whether they are genetic or acquired (Baglin, 2010). Two approaches have been validated in clinical studies so far: measurement of biomarkers and determination of the thrombin-generating potential. Biomarkers of thrombin generation reflect thrombin generation that has taken place in vivo and D-dimer measurement after completion of a finite period of anticoagulation has been shown to stratify patient risk. Measurement of the thrombin generating potential quantifies the ability to generate thrombin in a plasma sample in vitro in response to a predefined stimulus, usually a low concentration of tissue factor. Various methods for the measurement of the thrombin-generating potential have been developed utilising different substrates and end-point detection methods with varying interpretation of the signal.
The biological variability of global tests will be influenced by different factors. Biomarker measurements are determined by how active the coagulation–anticoagulation–fibrinolysis network has been in the minutes and possibly hours before a blood sample is taken. In contrast, the thrombin-generating potential is determined by the equilibrium between the components of the network in the hours and days before a blood sample is taken. A key area of study is the relationship between different measurements of hypercoagulability and the clinical utility of combining different measures when stratifying patient risk of recurrent venous thrombosis. D-dimer and thrombin generating potential do not correlate indicating that the tests are measuring different aspects of thrombin generation (Besser et al., 2008) and a recent study confirmed independent predictive value of these measurements for recurrent venous thrombosis (Eichinger et al., 2008). It remains to be determined for each assay:
• How many measurements in an individual are required to optimise the predictive value of a test.
• If measurements taken whilst patients are still taking anticoagulant therapy can usefully predict likelihood of recurrence.
D-dimer is a marker of fibrin degradation formed by the sequential action of three enzymes: thrombin, factor XIIIa and plasmin (Adam, Key & Greenberg, 2009). Fibrinogen is a soluble glycoprotein with a tri-nodular structure consisting of a central E domain flanked by two D domains (Figure 1). Fibrinogen is converted to fibrin by cleavage of fibrinopeptides A and B from the central E domain. Thrombin cleavage exposes a cryptic polymerisation site resulting in assembly of fibrin monomers into protofibrils. These protofibrils retain thrombin which activates factor XIII. The resultant XIIIa covalently crosslinks fibrin monomers and protofibrils forming an insoluble fibrin gel with covalently linked D domains. Proteolysis of the fibrin gel by plasmin releases a variety of cleavage products including terminal fibrin degradation products containing D-dimers. The monoclonal antibodies in D-dimer assays detect dimerised D domains which are not present in fibrinogen or noncrosslinked fibrin but each different monoclonal antibody has unique binding characteristics (Dempfle, 2005).
In a series of studies from observation through to a patient management study (PROLONG), Palareti et al. showed that measurement of D-dimer levels following cessation of anticoagulant therapy predicts likelihood of recurrent thrombosis (Palareti et al., 2002, 2003, 2006). A meta-analysis of cohort studies indicated that the annualised risk of recurrence is >5% in patients with an elevated D-dimer and <5% with a low D-dimer after completion of a finite period of anticoagulation after a first venous thrombosis (relative risk 2.4, 95% CI 1.9–3.1) (Verhovsek et al., 2008). In the PROLONG II study, D-dimer measurements were repeated at 2 monthly intervals for 1 year after an initial normal D-dimer 1 month after completion of anticoagulant therapy after a first episode of unprovoked VTE (Cosmi et al., 2010). D-dimer was normal in 68% of patients 1 month after stopping treatment; 14% of patients developed an abnormal D-dimer 2 months after an initial normal result. The rate of VTE recurrence over a mean follow-up of 10.6 months was 22.6% in these patients compared with 4.6% in patients whose D-dimer remained negative. This is an important finding which needs to be replicated in further studies.
The predictive value of D-dimer may be influenced by interacting or confounded factors such as sex and age. Furthermore, the predictive value of D-dimer measurement has typically been evaluated in patients with unprovoked venous thrombosis and is different in patients after provoked events (Baglin et al., 2008). A secondary analysis of the PROLONG study indicated that in patients with a normal D-dimer after completion of anticoagulant therapy after unprovoked venous thrombosis recurrence rates were higher in men than in women (7.4%vs. 4.3% patient-years) and in patients aged 65 years or more (8.4%vs. 3.6%). In the meta-analysis of cohort studies, only male sex had a significant effect on risk of recurrent VTE independent of D-dimer status. Age, hormone therapy use at the time of the index event, body mass index and timing of postanticoagulation testing did not influence the predictive value of the D-dimer test result (Douketis et al., 2010). More studies are required to examine:
• The performance characteristics of different D-dimer assays including quantitative (D-dimer measured as a continuous variable) versus qualitative (D-dimer measured as a threshold) measurement.
• The influence of the clinical profile of the patient on the predictive value of the test result.
• The value of serial measurement, including measurements during and after completion of an initial period of anticoagulant therapy.
Endogenous thrombin potential and thrombin generating capacity
Thrombin generation assays measure the thrombin–time curve which is the enzymatic work potential of thrombin. The Calibrated Automated Thrombogram® (Thrombinoscope BV, Maastricht, The Netherlands) and Technothrombin® TGA (Technoclone, Vienna, Austria) utilise a fluorogenic substrate, and the Endogenous Thrombin Potential Assay® (Siemens Healthcare Diagnostic Inc., Deerfield, IL, USA) and Pefakit Thrombin Dynamics Test® (Pentapharm, Basel, Switzerland) employ a chromogenic substrate. Activation of thrombin generation and interpretation of the thrombin–time curve varies between assays. Various parameters of the thrombin–time curve can be reported including lag time, peak thrombin, time to peak and the area under the curve (AUC, Endogenous thrombin Potential) (Figure 2).
Continuous registration of the thrombin–time curve using slow reacting chromogenic or flurogenic substrates has simplified the process but routinely requires dedicated equipment and a software algorithm to derive the thrombin–time curve (Hemker, Willems & Beguin, 1986; Hemker et al., 1993, 2003; Baglin, 2005; Hemker, 2008). Chromogenic methods require defibrination or use of an inhibitor of fibrin polymerisation to prevent turbidity owing to fibrin gel formation. Thrombin generation measured in defibrinated plasma with a chromogenic substrate is up to 50% higher than in plasma from which fibrinogen/fibrin has not been removed. The replacement of the chromogenic substrate with a slow reacting fluorogenic substrate allows continuous measurement of thrombin generation without the need for defibrination as the signal from the fluorophore is not quenched by turbidity. With a fluorogenic substrate, relationship between thrombin activity and the fluorescent signal is nonlinear. This is because of substrate consumption and the nonlinearity of fluorescence intensity with increasing concentration of fluorescent molecules: the ‘inner filter effect’. This problem can be overcome by monitoring the splitting of the fluorogenic substrate in the patient sample and comparing it with a constant known thrombin activity in a parallel nonclotting sample (Hemker et al., 2003).
A recently developed alternative to these optical detection methods is continuous registration of the thrombin–time curve using an electrochemical biosensor (Thuerlemann, Haeberli & Alberio, 2009). Sensor strips with an amperometric substrate are electrically connected to a measuring unit. Thrombin cleaves the substrate producing an electric current. As electrochemical detection of thrombin activity is not affected by colour or turbidity, measurements can be taken from whole blood samples. Clinical studies utilising this technology have not yet been reported.
The activated partial thromboplastin time (APTT) utilises a nonphysiological activation of thrombin generation and measures the time to clot detection rather than the total thrombin generation. The distinction between a clot-based assay and an assay that detects thrombin independent of clot formation is potentially important as clot formation occurs when only about 5% of the total thrombin has been generated (Mann, Brummel & Butenas, 2003). Whilst the APTT has limitations as a global test of haemostasis, there is an association between a short APTT and the risk of VTE (Tripodi et al., 2004; Hron et al., 2006a; Legnani et al., 2006). In a case-control study of 605 patients and 1290 controls patients who had an APTT ratio less than the fifth percentile of the distribution in the controls, the risk of recurrent VTE was increased 2.4-fold (Tripodi et al., 2004). Ten percent of patients had an APTT ratio less than the predefined APTT ratio threshold. In the cohort of the Austrian Study on Recurrent VTE, the recurrence rate after 4 years was 15.6% in patients with an APTT ratio less than the fifth percentile of all patients compared with 8.5% in those with a higher ratio. It was previously reported in this cohort that recurrent VTE was 1.6 times higher in patients with factor IX levels above the 75th percentile (Weltermann et al., 2003) and seven times higher in those with factor VIII levels above the 90th percentile (Kyrle et al., 2000). A short APTT was associated with higher VIII and IX levels but was still independently associated with risk of recurrence after adjustment for these. In the third study, 628 patients had blood taken approximately 1 month after completing anticoagulant therapy after a first unprovoked VTE (Legnani et al., 2006). After nearly 2 years, the probability of recurrence was 17.5% in patients with an APTT ratio in the lower quartile compared with 7.5% in those with a ratio in the upper quartile. In this study, there was no association between APTT and risk of recurrence after adjustment for factor VIII, IX and XI levels. However, there was still an association after adjustment for age, sex and D-dimer levels.
The Protac-Induced Coagulation Inhibition (PICI) assay (HemosIL Thrombopath, Instrumentation Laboratory, Lexington, MA, USA) is a snake venom-based assay of the protein C anticoagulant pathway. In a study of 190 patients, those with a PICI% of 74% or less were almost three times more likely to suffer recurrent VTE compared with those with a PICI% of more than 87% (Tripodi et al., 2010).
Other tests used in the laboratory to inform clinical management decisions in patients include genotyping for the JAK2V617F mutation and detection of a PNH clone (Paroxysmal Nocturnal Haemoglobinuria). There is conflicting evidence regarding the association between JAK2V617F and thrombosis. In patients presenting with intra-abdominal thrombosis, particularly portal or hepatic vein thrombosis, the mutation is found in approximately one-third of patients. In practice, patients suffering from intra-abdominal thrombosis are often treated with continued anticoagulant therapy and so the relationship between JAK2V617F and recurrent venous thrombosis will be difficult to validate. A PNH clone is associated with venous and arterial thrombosis with the greatest risk in patients with large PNH clones (PNH granulocytes >50%, 44%vs. 5% (Hall, Richards & Hillmen, 2003). Oral anticoagulation effectively prevents thrombosis, and long-term anticoagulation with a target INR of 2.5 has been proposed for patients with large PNH clones (PNH granulocytes >50%) (Baglin, Keeling & Watson, 2006).
Large-scale DNA analysis systems are becoming available that will allow study of venous thrombosis in relation to widespread genomic variation including polymorphic variation. The potential application of genomic DNA analysis to individualised risk assessment is uncertain as the interaction of complex combinations of variants of different genes with environmental factors may be relatively unpredictable at an individual level. There is proof of principle that multiple SNP testing (single nucleotide polymorphisms) quantifies the risk of recurrent VTE (van Hylckama Vlieg et al., 2008) and the number of common coagulation gene variants shown to be possibly related to risk of VTE is increasing (Bezemer et al., 2008; Reiner et al., 2009).
Tests performed on blood samples can measure the potential material contribution of the plasma compartment to risk of thrombosis. This testing will not quantify the influence of the platelet or extravascular compartments, and the risk attributable to each of these compartments is uncertain. As rapid and cheap platforms for personalised genomic risk assessment become available, it is possible that dichotomous testing of multiple candidate gene variants representing all relevant compartments will have a clinical utility that compliments measurement of the plasma-based thrombin generating potential. It remains to be determined in prospective cohort studies how testing of an expanded genetic repertoire and assessment of a plasma-intermediate phenotype will influence clinical decision-making in relation to long-term anticoagulant therapy after an episode of VTE.
Conflict of interest
TB has received educational support to attend symposia and for attending Advisory Board meetings for Sanofi-Synthelabo, Astra-Zenica, Boerhinger, Bayer, CSL Behring and Instrumentation Laboratories.