Thrombophilia testing: A British Society for Haematology guideline.

Abstract

This guideline was compiled according to the BSH process at [https://b-s-h.org.uk/media/16732/bsh-guidance-development-process-dec-5-18.pdf]. The Grading of Recommendations Assessment, Development and Evaluation (GRADE) nomenclature was used to evaluate levels of evidence and to assess the strength of recommendations. The GRADE criteria can be found at http://www.gradeworkinggroup.org. A literature search was carried out using the terms given in Appendix S1 until April 2021. Review of the manuscript was performed by the BSH Haemostasis and Thrombosis Task Force, the BSH Guidelines Committee and the sounding board of BSH. It was also placed on the members section of the BSH website for comment. It has also been reviewed by Royal College of Obstetricians and Gynaecologists, Royal College of Paediatrics and Child Health, Royal College of Physicians and Thrombosis UK, a patient-centred charity dedicated to promoting awareness, research and care of thrombosis; these organisations do not necessarily approve or endorse the contents. This guideline updates and widens the scope of the previous British Society for Haematology (BSH) Clinical guidelines for testing for heritable thrombophilia1 to include both heritable and acquired thrombophilia. The term thrombophilia is generally used to describe hereditary and/or acquired conditions associated with an increased predisposition to thrombosis. Heritable thrombophilia refers to genetic disorders of specific haemostatic proteins. These guidelines focus only on the factors that are identified from laboratory testing and therefore exclude disorders such as cancer, inflammatory conditions and obesity that are associated with thrombosis through multiple mechanisms. The most clearly defined heritable thrombophilias are the factor V Leiden (FVL) variant (F5 G1691A), the prothrombin gene variant (F2 G20210A), protein C (PC) deficiency, protein S (PS) deficiency, and antithrombin (AT) deficiency.2 Important acquired thrombophilias include the antiphospholipid syndrome (APS), paroxysmal nocturnal haemoglobinuria (PNH), myeloproliferative neoplasms (MPN) and the presence of a JAK2 mutation in the absence of an MPN phenotype. Pregnancy is a hypercoagulable state due partly to physiological changes in both the coagulation and fibrinolytic systems. Heritable and acquired thrombophilias can interact to further increase the risk of thrombosis, for example during pregnancy and the puerperium. As there is evidence that some thrombophilias may be associated with pregnancy failure and complications, testing for this purpose is included. Elevated levels of procoagulant factors may increase the risk of thrombosis but the relationship is not straightforward. First, part of the variance is genetic, and therefore lifelong, but some is acquired so that comorbidities such as obesity or inflammation confound the estimate of effect. Second, some factors, most notably factor V (FV), have anticoagulant effects that counterbalance a procoagulant effect from their elevation. A meta-analysis of 12 genome-wide association studies (GWAS) for venous thromboembolism (VTE) identified variants in F2, F5, F11, and FGG (encoding fibrinogen gamma chain) linked to thrombosis as well as non-O alleles of ABO which mediate their effect via elevation of von Willebrand factor (VWF) and secondarily factor VIII (FVIII).3 This approach does not detect rare variants with functional effects increasing thrombotic risk as reported in factor IX (F9), factor II (F2) and fibrinogen-alpha (FGA), fibrinogen-beta (FGB), and FGG.4-6 However, the relevance of these genetic variants to routine clinical practice is not clear at present. A phenotypic analysis was carried out as part of the Multiple Environmental and Genetic Assessment (MEGA) case–control study of VTE. After adjustment for age and sex, levels of factors II, X, IX, XI, VIII and fibrinogen all showed a positive association with risk of thrombosis. After additional correction for FVIII levels, only FIX and FXI retained significance with odds ratios (ORs) for levels >95th centile of 1.8 (95% confidence interval [CI]: 1.1–2.9) and 1.8 (1.1–3.0), respectively. In contrast, the OR for FVIII>95th centile was 16.0 (9.7–26.3) after correction for age, sex, and all the other coagulation factors.7 However, because of interacting heritable and acquired influences on FVIII activity, variability in levels over time, and as yet, lack of evidence of a role in the management of individuals with thrombosis or asymptomatic family members, routine testing for FVIII is not currently recommended. Despite results from animal studies, there remains no genetic or phenotypic8-10 evidence that variation in FXII is associated with thrombosis in humans.11 FXIII has a complex relationship with thrombosis due to interactions with other factors and the effects of genetic variants on FXIII activity assays. Genetic studies showed that the Val24Leu variant was associated with a reduced risk of venous thrombosis (OR: 0.85; 95% CI: 0.77–0.95).12, 13 The associations of PC, PS and AT deficiencies with increased risks of VTE are well-established.14 The degree of deficiency is variable and sensitive to assay type but in general thrombosis risk rises as soon the levels of protein C, S or AT fall below the normal range. In contrast, although tissue factor pathway inhibitor (TFPI), heparin cofactor II, and protein Z-dependent protease inhibitor (ZPI) and its cofactor, protein Z, are also natural anticoagulants, the clinical significance of genotypic or phenotypic variation in these is uncertain and testing for clinical purposes is not recommended. Guidelines on laboratory aspects of testing for deficiencies of natural anticoagulants have recently been published by the British Society for Haematology15 and the International Society on Thrombosis and Haemostasis.16-18 The risk of a first episode of VTE is increased around 15-fold in heterozygous AT deficiency.19 Overall, the risks are similar in those with type I and type II defects with the exception of most type II heparin binding defects, which appear to have a 4-fold lower risk.19 In contrast, homozygous heparin binding site defects appear to be associated with a high thrombotic risk.20 Further differences within antithrombin subtypes have also been observed.21 However, data on differences in risk between and within different subtypes are limited, and findings vary according to study design, the population being studied (family or non-family members), and whether all or only unprovoked venous thrombotic events were included in the analysis. In those with heterozygous PC or PS deficiency, the risk of a first episode of VTE is increased around 5–7-fold.19, 22, 23 There are no clinically useful differences in thrombotic risk between type I and type II PC deficiency15 and no clear evidence of a difference in risk between different subtypes of PS deficiency. These risks for heterozygous PC and PS deficiency are similar to or greater than those associated with FVL variant or F2 G20210A variant, but deficiencies of the natural anticoagulants are much less common (population prevalence of <0.5% for each deficiency), at least in those of European origin, and contribute relatively little to the population burden of VTE. Deficiencies of physiological anticoagulants interact with acquired risks and a transient provoking factor is present in approximately 50% of episodes of VTE in genetically predisposed individuals.24, 25 Since deficiencies of these natural anticoagulants are caused by multiple different genetic variants, clinical laboratory assessment is generally based on measurement of plasma activities or concentrations rather than molecular analysis.15 Acquired causes of deficiencies (Table 1) should always be considered before testing and when interpreting results as, if present, it may not be possible to reliably diagnose a heritable deficiency. Acquired problems include warfarin and the potential assay-dependent impact of direct oral anticoagulants (DOACs).15 When the decision has been made to test for deficiencies of physiological anticoagulants, this should be performed only after 3 months of anticoagulation for acute thrombosis, as there is uncertainty over the validity of the results obtained earlier, leading to repeat testing and increased costs, and with there being no evidence that it influences acute management. Protein C activity Chromogenic assay Protein S Free protein S antigen Antithrombin activity Chromogenic assay Physiological reduction Neonates and children (different normal range from adults) Other causes of reduction Vitamin K antagonists (e.g., warfarin) Vitamin K deficiency Liver disease Disseminated intravascular coagulation Severe sepsis Artefactual increase DOACs or heparin if using clotting-based assay Artefactual decrease Factor V Leiden if using clotting-based assay Physiological reduction Neonates (Different normal range from adults) Pregnancy and puerperium Other causes of reduction Vitamin K antagonists (e.g., warfarin) Vitamin K deficiency Liver disease Nephrotic syndrome Disseminated intravascular coagulation Severe sepsis Recent thrombosis Oral oestrogen therapy (e.g., combined oral contraceptive pill or hormone therapy) Acute phase response Sickle cell disease Artefactual increase DOACs or heparin if using clotting-based assay. Artefactual decrease Factor V Leiden if using clotting-based assay Physiological reduction Neonates (Different normal range from adults) Late pregnancy, early postpartuma a James et al. 2014.176 Other causes of reduction Liver disease Disseminated intravascular coagulation Nephrotic syndrome Severe sepsis Recent thrombosis Heparin therapy L-asparaginase therapy Artefactual increase DOACs: Xa inhibitors – if using Xa-based assay Thrombin inhibitors – if using thrombin-based assay The FVL and F2 G20210A variants are the most commonly tested genetic variants predisposing to VTE.29 These are detected using polymerase chain reaction (PCR)-based methods. Their prevalence varies in populations of different ethnicity. For example, heterozygosity for FVL is present in about 5% of individuals of European descent but is rare or absent in peoples from sub-Saharan Africa, East Asia and indigenous populations of the Americas and Australia. Similarly, heterozygosity for the prothrombin gene variant is present in 1%–2% of Europeans and is rare or absent in other ethnic populations.30 The FVL variant abolishes a cleavage site for activated PC in factor V increasing procoagulant activity. The prothrombin gene variant is a point mutation (G20210A) in the 3′ untranslated region of the gene31 causing increased levels of prothrombin.32 These variants result in increased relative risks for first venous thrombosis of 5- and 3-fold, respectively.33 A large number of variants in other genes with a wide range of prevalences have been reported to confer an increased risk of thrombosis. These include variants of methylenetetrahydrofolate reductase (MTHFR), SERPINE1 (encoding plasminogen activator inhibitor type 1) (PAI-1) and factor XIII as well as variants linked to the quantitative changes in procoagulant factors discussed above.28 However, either their association with thrombosis is not convincingly consistent or their effect is too small to alter management and they should not be included in thrombophilia panels at present. Although it has been shown that multiple variants present in an individual can combine to identify a significant risk of recurrence,34 this requires validation and we do not yet know how and when to introduce this oligogenic model into practice. Paroxysmal nocturnal haemoglobinuria (PNH) and myeloproliferative neoplasms (MPN) are acquired genetic traits that increase the risk of thrombosis. PNH is an acquired clonal stem cell disorder characterised by the expansion of a population of blood cells deficient in glycosylphosphatidylinositol anchored proteins (GPI-AP) due to PIGA gene mutation resulting in a deficiency or absence of all GPI-anchored proteins including CD55 and CD59 on the cell surface. Absence of CD59 leads to chronic complement activation resulting in the classical clinical features of intravascular haemolysis and thrombosis.35 Up to 10% of patients with PNH will present with thrombosis. The neutrophil clone size correlates best with thrombosis risk and patients with a clone of over 50% have a cumulative 10-year incidence of thrombosis of 34.5% compared to 5.3% in those with a clone of <50%. MPNs are characterised by clonal expansion of an abnormal haematopoietic stem/progenitor cell and include polycythaemia vera (PV), essential thrombocythemia (ET), and primary myelofibrosis (PMF). MPN or presence of a clone characterised by a JAK2 mutation in the absence of an MPN phenotype are associated with arterial and venous thromboses.36 The thromboses associated with PNH and MPN can occur anywhere in the venous or arterial systems but particularly in unusual sites for example, splanchnic vein thrombosis (SVT) (which includes portal vein (PVT), mesenteric vein (MVT) and splenic vein thrombosis, and the Budd–Chiari syndrome (BCS)) and cerebral venous sinus thrombosis (CVST).37, 38 In MPN, thrombosis often precedes disease recognition. Molecular abnormalities, primarily the V617F mutation in JAK2 exon 14, are found in 95% of PV (and an exon 12 mutation in most remaining patients) and in 60%–70% of ET and PMF patients.39 Isolated JAK2 mutations occur in approximately 0.1%–0.2% of the general population without an MPN phenotype and in 2.9%–5.6% of patients with CVST with no MPN phenotype40 (Table 2). A proportion of patients positive for JAK2 mutation with normal full blood count at presentation progressed into MPN during follow-up.41 Mutations of MPL exon 10 are present in about 5% of those with ET or PMF.42-44 In patients without JAK2 or MPL mutations, 67%–71% of those with ET and 56%–88% of those with PMF are positive for a calreticulin gene (CALR) mutation.45 In a study by Rumi et al. of 1235 consecutive patients diagnosed with ET or PV, the incidence of thrombosis associated with JAK2-mutated patients with ET and PV was similar; 7.1 and 10.5% respectively and was four times that of patients with ET and the CALR mutation (2.8%). The incidences of thrombosis associated with the JAK2 exon 12 and MPL mutations are not well documented due to the small number of patients with these mutations. Testing for JAK2, CALR, MPL variants in peripheral blood is sensitive and bone marrow samples are not required.39 Detailed guidance on assays used for detection of JAK2 mutations is available in separate guidelines.46 Diagnosis of PNH is based on flow cytometric analysis using antibodies directed against GPI-AP.47 The diagnosis of APS is dependent on the presence of at least one clinical feature (thrombosis or pregnancy morbidity) and at least one laboratory feature of antiphospholipid antibodies (aPL) which include lupus anticoagulant (LA), immunoglobulin (Ig) G or IgM anticardiolipin antibodies (aCL) or anti-β2-glycoprotein-I (anti-β2GPI) antibodies).48 The aPL need to be persistent, that is, present on two or more occasions at least 12 weeks apart.49 Of the three tests, a positive LA appears to be the most strongly associated with recurrent thrombosis, but individuals who are positive for all three assays (“triple positives”) have the highest thrombotic risk.50-52 Although the BSH guidelines (2012) on the investigation and management of antiphospholipid syndrome stated that in patients with thrombosis, measuring IgM antibodies does not add useful information,53 both IgG and IgM aCL and anti-β2GPI are part of the international consensus on laboratory diagnostic criteria for APS.49 There is increasing evidence that IgM anticardiolipin and anti-β2GPI antibodies have a pathogenic role in patients with APS.54-57 In patients with thrombotic APS, uncertainties remain as to the recurrence risk in patients with an initial unprovoked, compared to provoked, VTE and in those with venous compared to an initial arterial thrombosis.58 There is increasing evidence that the recurrence risk of VTE provoked by minor risk factors is similar to that with unprovoked VTE.59, 60 Therefore, such patients may also benefit from extended anticoagulation therapy as in those with unprovoked VTE. As the presence of antiphospholipid antibodies may alter management including choice of antithrombotic therapy in these patients, it may be reasonable to test for antiphospholipid antibodies. Catastrophic APS (CAPS) is a rare, but potentially fatal, variant of APS characterised by sudden onset of extensive microvascular thrombosis at multiple sites leading to multiorgan failure.61 CAPS tends to occur usually in patients with triple positive APS. Recommendations on the timing of, and indications for, antiphospholipid antibody testing following venous or arterial thrombosis are provided in the Addendum to British Society for Haematology Guidelines on Investigation and Management of Antiphospholipid Syndrome (2020).62 In asymptomatic individuals with triple positive antiphospholipid antibodies (mostly identified because of a prolonged activated partial thromboplastin time or presence of an autoimmune disorder), the incidence of first thrombotic events (which were equally distributed between venous and arterial thrombosis) was estimated to be 5% per year.52 Lower incidences of thrombosis of 1% and 0.5% annually respectively have been described in asymptomatic single antibody positive individuals and in women with the obstetric antiphospholipid syndrome.63, 64 In situations where the clinical utility of testing is not clear, testing is clearly not mandatory (clinical utility is defined as the ability of a test to improve clinical outcome). It is important that patients are counselled in advance of any decision on whether or not to undertake testing. This should include discussion of the aims of testing and how it might alter management decisions. What is the utility of identifying a heritable thrombophilic trait in a patient who has had a venous thrombotic event in modifying their future management or the management of asymptomatic family members? The relative risk of thrombophilic traits for recurrent VTE is less than that for a first episode of thrombosis because the comparator group is different. Moreover, the risk is managed differently, and no clinical trials have been undertaken. There are conflicting data on the association of FVL and F2 G20210A variants with risk of recurrence in the overall population of patients with VTE.33, 65 Observational data suggest that FVL Leiden but not F2 G20210A is associated with an increased risk of recurrence.33, 65 However, in a study with of 354 consecutive patients aged ≥65 years with a first unprovoked VTE, 9.0% of patients had FVL and 3.7% had a F2 G20210A variant.66 After adjustment for age, sex, and periods of anticoagulation as a time-varying covariate, at 3-year follow up neither the FVL (HR 0.98; 95% CI: 0.35–2.77) nor the F2 G20210A mutation (HR 1.15; 95% CI: 0.25–5.19) was associated with recurrent venous thromboembolism compared to controls.66 Patients with natural anticoagulant deficiencies were excluded from prospective studies from which predictive models for recurrent VTE after completion of treatment for a first event were derived.67 A meta-analysis of individuals with AT deficiency concluded the odds of recurrence were increased 2-4-fold with an absolute annual recurrence risk without long-term anticoagulant therapy of 8.8% (95% CI: 4.6–14.1) for AT-deficient and 4.3% (95% CI: 1.5–7.9) for non-AT-deficient VTE patients.19 A further cohort study in which AT was measured in percentage points on only one occasion found the odds of recurrent VTE were increased 3.7-fold (95% CI: 1.4–9.9) in those with AT activity <70% (fifth centile 87%) and 1.5-fold (95% CI: 1.0–2.3) in those with AT activities of 70%–87%.68 In a prospective study of familial thrombophilia, the annual risk of recurrent VTE in patients who did not receive long-term anticoagulant treatment was 5.1% (95% CI: 2.5–9.4) in those with PC deficiency and 6.5% (95% CI: 2.8–11.8%) in those with PS deficiency.69 In a meta-analysis, the odds of recurrent VTE were increased 2.9-fold (95% CI: 1.4–6.0) in PC deficient patients and 2.5-fold (95% CI: 0.9–7.2) in those with PS deficiency (25). At 10 years, the rates of recurrence were 31, 43 and 41% among patients with FXI activity <34th centile, between the 34th and 67th centiles, or >67th centile, respectively.70 Patients with the highest factor VIII level category (>200 iu/dL−1) had a hazard ratio for recurrence of 3.4; (95% CI: 2.2–5.3) compared to those with FVIII ≤100 iu/dL−1.71 In absolute terms this corresponded to a recurrence rate of 5% per annum compared to 1.4% per annum. Although these effects are significant, their utility is limited. Clinical history, in conjunction with simple tests such as D-dimer in selected patients, can identify those whose risk of recurrence is high enough to warrant long-term anticoagulation and which is not lowered significantly by the absence of a thrombophilic trait. These factors also identify patients with low risk of recurrence not requiring long-term anticoagulation, even in the presence of heritable thrombophilic traits.72-75 There is no evidence that the presence of heritable thrombophilia influences the intensity, choice or the monitoring of anticoagulant therapy when treating thrombosis except potentially in those with AT deficiency.76 In AT deficiency, diagnosis makes specific treatment (antithrombin concentrate) available,77 which can be valuable and can also facilitate interpretation of laboratory monitoring of heparin. Nonetheless, this is a rare disorder and so routine testing is not advised in the absence of a strong family history (defined as two or more first-degree relatives with VTE).78 For patients with a strong personal and/or family history of thrombosis in the absence of a clear risk factor, genetic analysis via Genomics England Limited (GEL) is available as noted above and should be combined with phenotypic testing where available. The likelihood of detecting a genetic trait increases with the strength of the family history.26 The major heritable thrombophilic traits follow Mendelian inheritance albeit with variable penetrance. Levels of FVIII and FXI have clear genetic components but also significant acquired modifiers so the likelihood of relatives being affected is less certain. Identification of a heritable trait in a family member does not indicate a risk of thrombosis high enough to warrant anticoagulation and does not alter most thromboprophylaxis regimens. However, some guidelines include knowledge of heritable thrombophilic traits in their risk assessment schemes with a consequent impact on management.79 Absence of that trait in a family member significantly reduces their risk of thrombosis but does not return it to normal and the utility of testing will depend on their personal circumstances and the circumstances of the proband's VTE event.80, 81 Overall, the recurrence risk for VTE is determined by the clinical situation (e.g., provoked vs. unprovoked) along with non-Mendelian risk factors (e.g., body mass index and age) rather than the inherited thrombophilia panel. Therefore, when a patient is known to have a heritable thrombophilic trait, it may be reasonable to consider selective testing of first-degree relatives when this will alter their management choices, for example, highly penetrant deficiencies of PC, PS or AT deficiency in a woman of childbearing age. Routine screening for FVL is not required in women with a first degree relative with FVL but no history of thrombosis (i.e., mother or siblings) prior to starting combined oral contraceptive pills or oestrogen replacement therapy.82, 83 However, the influence of family history of thrombosis, thrombophilia testing and risk of thrombosis related oestrogen-progesterone content of therapies should be discussed with all women to determine whether they will alter their therapy choices and should be documented clearly. Investigation and management of thrombosis at unusual sites are discussed in another BSH Guideline.84 For thrombosis at unusual sites, which often involves local or systemic conditions triggering the event, testing for thrombophilia should be reserved for selected patients with unexplained events. The association of MPN and PNH with thrombosis at unusual sites, especially SVT which includes portal, mesenteric, splenic vein thrombosis and the Budd-Chiari syndrome, has been demonstrated in many studies85, 86 and these disorders should be tested for in the absence of a clear reason for the SVT, such as abdominal sepsis, cancer or cirrhosis. Analysis of data from pooled incidence-cases found that in 19% of patients, splanchnic vein (hepatic, mesenteric, portal, splenic, inferior vena cava) thrombosis preceded the diagnosis of PNH.87 For the remaining patients, visceral thrombosis occurred at a median of 5 years (range, 0–24) after diagnosis. Diagnosis of PNH and MPN is important because these diseases have specific treatments in addition to anticoagulation to prevent recurrent thrombosis. In a systematic review and meta-analysis of nine small observational studies to assess the prevalence of heritable thrombophilia in patients with PVT and BCS (total 4 studies), the pooled prevalence of AT, PC, and PS deficiencies were 3.9, 5.6, and 2.6% in PVT, and 2.3, 3.8, and 3.0% in BCS, respectively. Only three studies compared the prevalence of heritable thrombophilia between PVT patients and healthy individuals. The pooled odds ratios of heritable AT, PC and PS deficiencies for PVT were 8.89 (95% CI: 2.34–33.72, p = 0.0011), 17.63 (95% CI: 1.97–158.21, p = 0.0032), and 8.00 (95% CI: 1.61–39.86, p = 0.011), respectively.88 These studies are only for the first thrombotic event and the risk of recurrent events associated with heritable thrombophilia and thrombosis at unusual sites is not well established but seems to be low. Therefore, the value of testing for heritable thrombophilia is unknown and testing should be considered only if the thrombotic event occurs in the absence of a clear risk factor for the index event at a young age (median ~46 years).88 CVST is a rare entity accounting for <1% of all strokes.89 The majority (85%) of CVST patients will have an identifiable risk factor, the most common of which are oestrogen-containing oral contraceptive use and pregnancy.90 Other rare causes that can contribute to CVST include APS, vasculitis, MPN, PNH, chronic inflammatory disorders, and local factors such as infection, malignancy, trauma or surgery.90 CVST is reported in 2%–8% of patients with PNH91, 92 and around 3.8% of patients with MPN.93 Around 2.6%–5.6% of patients diagnosed with CVST are found to have a JAK2 mutation with normal full blood count at presentation41, 94-96 (Table 2). CVST are reported in 2% to 8% of patients with PNH.90, 97-99 However, it is not clear how many of these patients had a normal full blood count at presentation with CVST. Several studies have shown the presence of aPL increases the risk of thrombosis at unusual sites such as SVT and CVST.100, 101 As the type and duration of anticoagulation are affected by the presence of antiphospholipid antibodies, testing for these antibodies is recommended in an updated BSH guideline.62 In the absence of a clear risk factor, patients with CVST may need long-term anticoagulation and routine testing for heritable thrombophilia is not required. There is no evidence to suggest an association of heritable thrombophilia with retinal vein occlusion (RVO). The pathogenic role of antiphospholipid antibodies in RVO is uncertain. A meta-analysis of 11 studies showed that presence of antiphospholipid antibodies was significantly associated with incidence of RVO (OR = 5.18, 95% CI: 3.37, 7.95).102 A more recent study that included 331 consecutive patients with RVO and 281 controls, also showed that antiphospholipid antibodies were more prevalent in RVO-patients than in controls (33, 10% vs. 12, 4.3%; OR 2.47; 95% CI: 1.25–4.88; p = 0.009) with RVO-APS patients having more frequently lupus anticoagulant or triple positive antiphospholipid antibody than controls.103 Testing for aPL may be considered in patients without local risk factors and no other explanation for RVO such as diabetes, hypertension, and hpercholesterolaemia as those with persistently positive aPL would be considered for anticoagulation. There is conflicting evidence with respect to the presence and the strength of associations between FVL and F2 G20210A variant and arterial thrombosis. Although some observational studies demonstrated a moderate increased risk of myocardial infarction (MI) in patients with FVL or F2 G20210A variants, this has not been reproduced in others. Table S1 summarises meta-analyses on the association of the FVL and/ or F2G20210A polymorphism with MI. These variants are common in the European population and will be found in many patients with cardiovascular disease. Whether their presence reflects a causal role for cardiovascular events is not known and is difficult to determine from these meta-analyses. When statistically significant associations have been found, these have been too modest to be of clinical significance and there are no clinical trials to suggest that management should be influenced as a result of the presence of these variants. Because heritable deficiencies of AT, PC and PS are rare, observational studies with sufficient statistical power to assess potential associations with risk of arterial thrombosis are lacking. Overall, there is no evidence to support an association between heritable thrombophilia and arterial thrombosis in adults a