Investigation and management of heritable thrombophilia


The Secretary BCSH, British Society for Haematology, 2 Carlton House Terrace, London SW1Y 5AF, UK.

There is no internationally accepted definition of thrombophilia. The British Committee for Standards in Haematology (1990) suggested that the term thrombophilia be used to describe ‘disorders of the haemostatic mechanisms which are likely to predispose to thrombosis’. This definition is widely used but has disadvantages. Firstly, as new thrombophilic conditions have been described, it has become evident that many individuals who carry these defects remain asymptomatic. Secondly, even though there is an increase in the number of abnormalities recognized as likely to enhance the risk of thrombosis, detailed laboratory investigations fail to detect any abnormality in at least 50% of patients who present with a history of thrombosis. In North America, the term thrombophilic is frequently used by clinicians to describe patients who have developed venous thrombosis either spontaneously or of a severity out of proportion to any recognized stimulus, patients who have recurrent venous thrombotic events and patients who develop venous thrombosis at an early age. This may be a more clinically useful concept. Individuals who have a thrombophilic defect identified on laboratory testing and who have a family history of proven venous thrombosis are at greater risk of thrombosis than individuals who have a thrombophilic defect but no personal or family history of venous thrombosis (Lensen et al, 1996).

Inevitably, with the developing interest in the role of prothrombotic abnormalities in thrombosis risk, haematologists and other clinicians have come under pressure to initiate laboratory tests on an increasing number of patients. Performance of a comprehensive range of laboratory tests for thrombophilia has thus become commonplace in subjects presenting with deep vein thrombosis (DVT) or pulmonary embolism. Detection of a heritable prothrombotic state may also lead to testing of family members in an attempt to identify asymptomatic relatives who may be at increased risk of venous thromboembolism (VTE). There are several important issues relevant to the clinical utility and cost effectiveness of this approach. Clinicians may tend to overestimate the risk of thrombosis associated with thrombophilias and to underestimate the risks associated with anticoagulation. Frequently, this has led to the belief that prophylactic anticoagulation is a safer option than clinical surveillance. As evidence about the risk of venous thrombosis associated with thrombophilias accumulates, it is becoming clear that for most patients this is not the case. Furthermore, there is often a lack of recognition that failure to identify a defect in an individual is not proof that no defect exists, only that the particular defects for which tests have been performed are probably not present. The patient with ‘negative’ test results may well have as yet unidentifiable prothrombotic abnormalities, which increase his or her risk of thrombosis. Reassuring patients with normal test results may constitute false reassurance, which may present a real risk for the patient if negative laboratory results lead to ignoring or underestimating the clinical history.


Heritable thrombophilias

Prothrombotic states may be heritable, acquired or mixed – the result of the environment (e.g. oestrogen use, obesity or other lifestyle factors) interacting with genetic background. To date, a limited number of genetic variants are proven to be independent risk factors for venous thromboembolism. These include mutations in the genes encoding the natural anticoagulants antithrombin, protein C and protein S, and the clotting factors fibrinogen, prothrombin and factor V.

Antithrombin deficiency Antithrombin (previously called antithrombin III) is synthesized by the liver. Its inhibitory effect is not confined to thrombin. It also inhibits the activated clotting factors IXa, Xa, XIa, XIIa and tissue factor-bound factor VIIa. Heparins markedly accelerate the rate of complex formation between antithrombin and the serine proteases.

Within the last decade, our understanding of the basis of familial antithrombin deficiency has been greatly facilitated by advances made in the molecular biology and functional characterization of this inhibitory glycoprotein. Of particular significance has been the recognition that the antithrombin molecule possesses two important functional regions – a heparin-binding domain and a thrombin-binding domain. Two major types of heritable antithrombin deficiency are recognized. Type I is characterized by a quantitative reduction of qualitatively (functionally) normal antithrombin protein. Type II deficiency is due to the production of a qualitatively abnormal protein. In both types of antithrombin deficiency, antithrombin activity is reduced to a variable extent. Type II deficiency is subclassified according to the site of the molecular defect:

(a) Reactive site (RS) – abnormalities residing in the reactive (thrombin binding) site.

(b) Heparin binding site (HBS) – abnormalities residing in the heparin binding site.

(c) Pleiotropic effect (PE) – abnormalities residing in both reactive and heparin binding sites.

Antithrombin assays Only functional assays of heparin cofactor activity will detect both type I and type II antithrombin deficiencies. For routine clinical purposes it is recommended that a heparin cofactor activity assay be used in the initial screen.

The distinction between the subtypes of antithrombin deficiency is of clinical relevance as the incidence of thrombosis is higher in association with type I deficiency and type II deficiency in which the mutation affects the reactive site than in type II deficiency in which the mutation affects the heparin binding site (Finazzi et al, 1987). An initial classification into type I or type II can be made by comparing the result of an immunological assay with the result of the heparin cofactor assay.

Type II heparin-binding variants are associated with a lower risk of thrombosis than type II reactive site defects. However, a heparin-binding variant may increase the attributable risk of an additional thrombophilic defect, such as the factor V Leiden mutation. A short incubation, of 30 s or less, with a low concentration of heparin is required for detection. As most currently used antithrombin activity assays utilize a long incubation, heparin-binding defects are not detected. The distinction between type II defects is therefore only an issue for those centres using assays specifically to detect heparin-binding defects. Crossed immunoelectrophoresis with heparin is a simple tool that may be used to detect type II heparin-binding site variants in those centres that need to identify these defects. Although many mutations associated with antithrombin deficiency have been described, identification of the mutation is not usually necessary for clinical purposes.

Normal ranges and variations Age- and sex-related variations in antithrombin activity (Tait et al, 1993a) and antigen levels are minor, so the reference ranges in healthy populations are narrow. Antithrombin levels are slightly lower in premenopausal women than in men of similar age or post-menopausal women and are slightly lower in women using combined oral contraceptive pills than in non-pill-using women (Tait et al, 1993a). More significant decreases in antithrombin activity are observed in patients on heparin treatment (Marciniak & Gockerman, 1977) and in those with current thrombosis. Profound decreases in plasma antithrombin are seen in disseminated intravascular coagulation, liver disease and the nephrotic syndrome.

Prevalence of antithrombin deficiency and risk of thrombosis The prevalence of type I antithrombin gene mutations in the general population is of the order of 0·02% (Tait et al, 1994). Family studies suggest that antithrombin deficiency is a more severe disorder than deficiencies of protein C or protein S with the majority of patients suffering thrombosis before the age of 25 years (Thaler & Lechner, 1981; Hirsh et al, 1989; Demers et al, 1992). In studies of unselected patients with thrombosis, antithrombin deficiency was reported in 1% (Heijboer et al, 1990) and 0·5% (Mateo et al, 1997). The relative risk of venous thromboembolism is around 25–50-fold for individuals with type I antithrombin deficiency (Rosendaal, 1999).

Protein C deficiency Protein C is a vitamin K-dependent glycoprotein that is synthesized in the liver. Before activation by the thrombin–thrombomodulin complex on the endothelial cell surface, it circulates as a two-chain zymogen. By degrading activated clotting factors Va and VIIIa, activated protein C (APC) functions as one of the major inhibitors of the coagulation system. Activated protein C also reduces platelet prothrombinase activity by degrading platelet-bound factor Va at the receptor for factor Xa. The inhibitory effects of activated protein C are facilitated through the cofactor activity of protein S.

As with antithrombin deficiency, familial protein C deficiency can be classified into two types on the basis of phenotypic analysis using functional and immunological assays. Type I is characterized by parallel reductions of functional and immunoreactive protein C. In type II the functional level is substantially lower than that of the antigen. In contrast to antithrombin deficiency, in which type II deficiency is more common than type I, type I protein C deficiency is more common than type II. The anticipation that the underlying genetic variant and associated phenotype might be predictive of the degree of thrombotic risk has not been realized and phenotypic classification of protein C deficiency therefore serves no useful clinical purpose.

Protein C assays Most functional assays of protein C use the specific activator Protac which is derived from snake venom. The activated protein C formed can be quantified by clotting or chromogenic methods. Both are available in kit form from commercial manufacturers. A standard calibrated against the current International Standard for protein C must be used. Chromogenic assays are simple to perform and will detect all type I defects and the vast majority of type II defects. In the presence of factor V Leiden, misleadingly low protein C activity levels may be obtained with the clotting method (Faioni et al, 1996). Clotting methods also underestimate protein C activity in patients with elevated plasma factor VIII levels (De Moerloose et al, 1988) and in the presence of hyperlipidaemia. Measured protein C activity by clotting assay may be unreliable in the presence of a lupus inhibitor (Simioni et al, 1991).

Protein C antigen assays are available widely and will help distinguish between type I and type II deficiency. As there appears to be no clear relationship between the type of protein C defect and the risk of thrombosis, there is no clinical justification for this extra investigation or for molecular studies to identify the specific mutation.

Reference ranges and variations There is a wide overlap in protein C activity between heterozygous carriers and their unaffected relatives in families with protein C deficiency (Allaart et al, 1993). Protein C activity levels appear to be related to age and sex (Tait et al, 1993b), but this is explained by blood lipid levels (Rodeghiero & Tosetto, 1997). Reduced protein C activity is observed in patients with disseminated intravascular disease (DIC) and in liver disease. Protein C activity is reduced markedly by coumarins.

Prevalence of protein C deficiency and risk of thrombosis The prevalence of heritable protein C deficiency in the general population is approximately 0·2–0·3% (Miletich et al, 1987; Tait et al, 1995) and in unselected patients with venous thromboembolism is around 3% (Heijboer et al, 1990; Koster et al, 1995a; Mateo et al, 1997). The relative risk of venous thromboembolism is between 10- and 15-fold for individuals with protein C deficiency (Rosendaal, 1999).

Protein S deficiency Protein S, another vitamin K-dependent protein, is a cofactor for activated protein C. Approximately 65% of the total plasma protein S is complexed with C4b-binding protein (C4bBP) and has no cofactor activity. The remaining 35%, designated free protein S, remains uncomplexed and is the active moiety. The bioavailability of protein S is closely linked to the concentration of C4bBP, which acts as an important regulatory protein in the activated protein C:protein S inhibitory pathway.

Three types of protein S deficiency are described. In accordance with the classification of antithrombin and protein C deficiencies, type I protein S deficiency is a quantitative defect caused by genetic abnormalities which result in the reduced production of structurally normal protein. Both total and free protein S antigen levels are reduced. Type II protein S deficiency has been characterized as a qualitative (functional) defect, but it has become evident that some individuals with inherited or acquired APC resistance have been incorrectly diagnosed as having type II protein S deficiency (Faioni et al, 1993). In type III deficiency, although free protein S antigen is reduced, the total protein S antigen level is normal. It has been suggested that type I and type III protein S deficiencies may be phenotypic variants of the same genetic disorder (Zoller et al, 1994a; Simmonds et al, 1997).

Protein S assays Three main types of assay are available: for functional protein S, and for total immunoreactive protein S and free immunoreactive protein S.

Functional protein S assays are based on the cofactor activity. Ideally these assays should reflect only free protein S activity, but this is not always the case as separation of free protein S from C4bBP-complexed protein S is not performed in many of the available methods. Functional assays detect all types of protein S deficiency, but some functional assays of protein S are non-specific and have been shown to be sensitive to the inherited APC resistance associated with factor V Leiden and the acquired APC resistance observed in some patients with antiphospholipid antibodies (Faioni et al, 1993).

Several techniques are available for the determination of total immunoreactive protein S. In addition to Laurell assays, these include enzyme-linked immunosorbent assays (ELISAs) and radioimmunoassays. There are a number of methodological problems associated with the Laurell technique and it is therefore not recommended. An increasing number of commercial kits and reagents are available for the measurement of total protein S, the majority of which are ELISAs. It is important to establish that the chosen method for measuring total immunoreactive protein S is not influenced by the concentration of C4bBP.

Free protein S is the active moiety of the total protein S. Consequently, free protein S assays are frequently performed tests in the investigation of heritable thrombophilia. In the most widely used method, separation of free protein S from the C4bBP-bound protein S is achieved by precipitation with polyethylene glycol followed by centrifugation. Results are expressed relative to a pooled plasma calibrated against the current International Standard, either as a proportion of the total protein S or, preferably, against the free protein S content. Assays using monoclonal antibodies for distinct epitopes of free protein S allow direct measurement of free protein S in citrated plasma without the need for a precipitation stage (Aillaud et al, 1996).

It is recommended that in cases in which a functional protein S assay is used as an initial screening test for protein S deficiency, low results should be further investigated with an immunoreactive assay of free protein S.

Reference ranges and variations Protein S levels are slightly higher in men than in women. Protein S levels fall progressively during pregnancy (Clark et al, 1998) and are reduced to a lesser extent in women using oestrogen-containing oral contraceptives or hormone replacement therapy (Comp et al, 1986; Malm et al, 1988; Lowe et al, 1999). Overdiagnosis of protein S deficiency is therefore a risk. Acquired protein S deficiency is also seen in patients on coumarins, in those with antiphospholipid antibodies and in disseminated intravascular coagulation and liver disease.

Prevalence of protein S deficiency and risk of thrombosis The prevalence of protein S deficiency in the general population remains unknown. It has been suggested that the best way to determine protein S deficiency is by measurement of free protein S antigen (Faioni et al, 1997). In the Leiden Thrombophilia Study (Koster et al, 1995a) and in a subsequent population-based case–control study reported from Italy (Faioni et al, 1997), low levels of free protein S antigen were found in around 3% of patients with venous thromboembolism and in 2·1% and 1·3% of the controls, respectively, suggesting that low free protein S levels have a mild effect on the risk of venous thrombosis – increasing the risk by only around twofold. However, the conclusion that protein S deficiency is a mild risk factor may be flawed as it is possible that the prevalence of protein S deficiency in the general population is much lower than 1–2%. The currently available evidence indicates a substantial difference in risk associated with protein S deficiency in thrombophilic families and in unselected consecutive patients, suggesting that the effect in families is the result of interaction with other defects (Koeleman et al, 1995; Zoller et al, 1995).

Activated protein C (APC) resistance and factor V Leiden APC resistance is defined as an impaired plasma anticoagulant response to APC added in vitro. The phenomenon of APC resistance first attracted widespread attention when it was reported that APC resistance co-segregated with thrombosis in families with familial venous thromboembolism (Dahlback et al, 1993). Shortly thereafter it was demonstrated that the majority of patients with familial APC resistance have the same point mutation in the gene for clotting factor V (1691G-A), the ‘factor V Leiden’ mutation (Bertina et al, 1994). Although factor V Leiden is the most common cause of inherited APC resistance, other changes in haemostasis cause acquired APC resistance, e.g. increased plasma levels of factor VIII or the presence of antiphospholipid antibodies. Activated protein C resistance increases with age and in women who use oestrogen-containing contraceptive pills (Olivieri et al, 1995) or hormone replacement therapy (Lowe et al, 1999). The effect is more marked with third-generation than second-generation oral contraceptives. Activated protein C resistance also increases during pregnancy (Cumming et al, 1995; Mathonnet et al, 1996; Clark et al, 1998).

APC resistance tests The most commonly used test system is the activated partial thromboplastin time (APTT). Samples are tested with and without added APC and the resultant clotting times are expressed as a ratio (Dahlback et al, 1993; De Ronde & Bertina, 1994) – the so-called APC sensitivity ratio (APC:SR). When testing for APC resistance it is important to avoid platelet contamination and activation. Reagents, coagulometers and concentrations of APC affect the results obtained using the APTT method. It has been suggested that the methodological variability associated with APTT-based tests for the detection of APC resistance can be reduced by ‘normalizing’ the results by dividing the patient's APC:SR by the APC:SR of pooled normal plasma. If this system is adopted it is important to establish that the normal plasma pool does not include a contribution from an individual who carries the factor V Leiden mutation, as even a single affected donation is sufficient to affect the APC:SR of the pool (Tripodi et al, 1998).

The originally described APC resistance test is abnormal in subjects with acquired APC resistance and in those who have a prolonged baseline APTT due to, for example, clotting factor deficiencies or anticoagulant therapy. It is therefore not diagnostic of factor V Leiden. Predilution of the test plasma in factor V-deficient plasma increases the sensitivity and specificity of the APTT-based APC: SR as a screen for factor V Leiden (Jorquera et al, 1994; Trossaert et al, 1994). This modification makes the test close to 100% specific and sensitive to factor V Leiden in both healthy controls and patients with suspected VTE (Svensson et al, 1997; Tripodi et al, 1997) and may be reliably used if DNA analysis is not available. However, there is evidence that the APC:SR determined with the original unmodified test correlates with venous thrombosis risk, irrespective of whether or not factor V Leiden is present (Zoller et al, 1994b; Bertina et al, 1995; De Visser et al, 1999; Rodeghiero & Tosetto, 1999). The specificity of the modified APC:SR test means that individuals who have increased APC resistance for reasons other than the possession of the factor V Leiden mutation will be overlooked if the original APC:SR test is omitted from the screening procedure. The clinical value of detecting APC resistance in the absence of factor V Leiden is unknown.

Detection of the factor V Leiden mutation Detection of the factor V Leiden mutation relies on amplification of the nucleotide region close to the exon–intron boundary in exon 5 of the factor V gene from either genomic DNA or from mRNA followed by a mutation detection step.

Prevalence of the factor V Leiden mutation and risk of thrombosis In Caucasian populations, factor V Leiden is much more common than any of the other heritable thrombophilias having a reported prevalence of between 2% and 15% (Rees et al, 1995) and is more prevalent in individuals of Northern European extraction than in those from Southern Europe (Rosendaal et al, 1995; Ridker et al, 1997). Depending on patient selection, factor V Leiden is found in 20–50% of patients presenting with a first episode of venous thromboembolism (VTE) (Koster et al, 1993; Rosendaal et al, 1995) and in more than 50% of probands from selected families with familial thrombophilia (Griffen et al, 1993). Heterozygous carriers have a three- to eightfold increased risk of venous thrombosis (Koster et al, 1993; Ridker et al, 1995a; Rosendaal et al, 1995) and homozygotes have an 80-fold increased risk (Rosendaal et al, 1995).

Familial APC resistance in the absence of the factor V Leiden mutation Occasionally familial APC resistance occurs in the absence of factor V Leiden (Zoller et al, 1994b; Bertina et al, 1995). One identified cause is a mutation in the 306 APC cleavage site (factor V Cambridge) (Williamson et al, 1998). A specific factor V gene haplotype (HR2) has been shown to occur more frequently in individuals with APC resistance ratios beneath the 15th percentile than in those with higher ratios or in normal controls (Bernardi et al, 1997). Co-inheritance of this HR2 haplotype with factor V Leiden may increase the risk of VTE above that associated with factor V Leiden alone (Faioni et al, 1999).

Prothrombin G20210A mutation The G→A transition at nucleotide 20210 in the 3′-untranslated region of the prothrombin gene is associated with elevated plasma prothrombin levels and an increased risk of venous thrombosis (Poort et al, 1996). In the absence of a specific phenotypic test for the presence of the variant 20210A allele, DNA-based procedures are required. The 20210A transition is not associated with the introduction or loss of a specific restriction enzyme recognition site and detection methods that do not require the use of restriction enzyme digestion of the amplified polymerase chain reaction (PCR) product have been devised.

Prevalence of the prothrombin G20210A mutation The prevalence in Northern Europe is around 2% in the healthy population and 6% in unselected patients with a first thrombosis (Poort et al, 1996). Higher prevalences have been reported in Southern Europe in which prothrombin G20210A is the most prevalent heritable thrombophilic defect (Souto et al, 1998). The risk of venous thrombosis in heterozygous carriers of the 20210A allele is estimated to be around three times that in non-carriers.

Dysfibrinogenaemia Over 250 cases of heritable dysfibrinogenaemia have been reported. Most cases are asymptomatic and found coincidentally, but in about 20% there is an increased tendency to arteriovenous thromboembolism and in 25% a bleeding tendency (Haverkate & Samama, 1995). In some there is a low fibrinogen functional level. The prevalence of fibrinogen abnormalities in patients with venous thrombosis is low (0·8%), but a high incidence of post-partum thrombosis and an increased risk of pregnancy loss have been reported in women with thrombophilic fibrinogen variants (Haverkate & Samama, 1995). Several mechanisms to explain the thrombotic tendency have been proposed, including defective lysis of the abnormal fibrin or defective binding of thrombin to the abnormal fibrin with consequent elevated plasma thrombin levels.

Other defects It is probable that there are other as yet unidentified heritable abnormalities of haemostasis associated with an increased risk of thrombosis. Some additional candidates such as dysplasminogenaemia and heparin cofactor II deficiency have been studied. However, none has been demonstrated conclusively to contribute to heritable thrombophilia.

Mixed aetiology defects

In some instances the prothrombotic changes in haemostasis are the result of interactions between genetic and environmental factors.

Elevated factor VIII Factor VIII levels are governed by both genetic and environmental factors. Factor VIII levels above 150 IU/dl are associated with a sixfold increased risk of venous thromboembolism compared with factor VIII levels of less than 100 IU/dl (Koster et al, 1995b).

Hyperhomocysteinaemia Case–control studies have demonstrated an approximately 2·5-fold increased risk of venous thrombosis in individuals with homocysteine levels exceeding 18·5 µmol/l and a three- to fourfold risk associated with levels exceeding 20 µmol/l (Den Heijer et al, 1996, 1998; Simioni et al, 1996). Genetic analysis is not indicated as, although the common thermolabile methylene tetrahydrofolate reductase variant contributes to hyperhomocysteinaemia, it is not itself associated with venous thromboembolism. Other mutations are uncommon.

Combined thrombophilias

Because of the high prevalence of factor V Leiden and the prothrombin 20210A mutation in many populations, individuals and families with more than one heritable thrombophilic condition are encountered. The prevalence of thrombosis is often higher in family members with combinations of thrombophilic conditions.

Thrombophilia case finding

Initial assessment

The initial assessment of a patient presenting with a clinical picture suggestive of thrombophilia must commence with a carefully taken personal and family history. In addition, a clinical examination with appropriate laboratory, imaging and other investigation should be performed. Patients should be asked specifically if they have a personal history of venous thromboembolism or a family history of venous thrombosis. Ideally these events should have been objectively confirmed but, as less weight was placed on the requirement to confirm a clinical diagnosis of deep vein thrombosis or pulmonary embolism in the past, a degree of pragmatism is necessary. Thus, if a venous thrombosis has not been confirmed by objective testing, it would seem reasonable to accept the clinical diagnosis if the history is plausible and if it resulted in the patient being given more than a few (7–10) days of anticoagulant treatment. Additional risk factors should be clearly documented. These include advancing age, a past history of venous thrombosis, immobility, trauma, surgery, nephrotic syndrome, inflammatory disorders, hormone use, pregnancy, post-partum state and obesity. Occult cancer should be considered, although invasive investigation is not routinely indicated. A full blood count should be performed to exclude myeloproliferative disorders. Depending on the initial assessment of the patient and on the clinical management decisions to be addressed, laboratory testing for heritable thrombophilia can be considered.

When to collect samples for thrombophilia testing

Some tests for heritable thrombophilia (for example, assays of antithrombin, protein C and protein S) are affected by the acute post-thrombotic state and by anticoagulant use. Also, finding a thrombophilic abnormality almost never influences the management of an acute thrombotic event. There is little point in striving to obtain samples for tests for heritable thrombophilia when the patient presents with an acute thrombotic event. Testing is usually best delayed until at least 1 month after completion of a course of anticoagulation. If possible, testing for heritable thrombophilia should be avoided during intercurrent illness, pregnancy, use of a combined oral contraceptive pill or hormone replacement therapy. If this is impossible, then it is essential that the individual interpreting the screen is aware of the presence and potential influence of these various acquired factors on the components of the test results. PCR-based tests for FV Leiden and the prothrombin 20210A allele are unaffected by the above factors.

Recommendations for laboratory tests and interpretation

When testing is indicated, it should include assays for heritable defects: deficiency of antithrombin, protein C or protein S, factor V Leiden and prothrombin G20210A mutations and for antiphospholipid antibodies.

• The activated partial thromboplastin time (APTT), prothrombin time and thrombin clotting time should be incorporated in the initial screening. The APTT may identify some patients with antiphospholipid antibodies (depending on the sensitivity of the APTT reagent used), but is not sufficient alone to exclude antiphospholipid antibodies (Greaves et al, 2000). The thrombin clotting time will allow identification of dysfibrinogenaemia and heparin contamination. The prothrombin time is useful in the interpretation of low protein C or protein S results

• Functional assays should be used to determine antithrombin and protein C levels

• Chromogenic assays of protein C activity are less subject to interference than clotting assays and are preferable

• Immunoreactive assays of protein S antigen are preferable to functional assays. If a protein S activity assay is used in the initial screen, low results should be further investigated with an immunoreactive assay of free protein S

• The modified APC:SR test (predilution of the test sample in factor V-deficient plasma), as opposed to the original APC:SR test, should be used as a phenotypic test for the factor V Leiden mutation

• PCR-based testing for prothrombin G20210A is required, as there is no screening test

• Laboratories must establish their own reference ranges for the assays and tests that they use, including antithrombin, protein C and protein S and modified APC:SR

• Comprehensive assays for antiphospholipid antibodies (both lupus inhibitors and anticardiolipin antibodies) should also be performed (Greaves et al, 2000)

• Rigorous internal quality assurance and participation in accredited external quality assessment schemes are mandatory

• The interpretation of thrombophilia test results is difficult and fraught with pitfalls, which occasionally lead to underdiagnosis and frequently to overdiagnosis of defects. It is strongly recommended that thrombophilia testing is supervised by and results are interpreted by an experienced clinician who is aware of all relevant factors that may influence individual test results in each individual.

Who should be tested for heritable thrombophilia?

Widespread testing for thrombophilia has become commonplace, but there are important issues relevant to the clinical utility and cost effectiveness of testing that must be addressed in considering who should be tested (Greaves & Baglin, 2000).

The diagnostic accuracy of laboratory tests for thrombophilia

Although genetic tests for factor V Leiden and prothrombin G20210A are robust and highly reproducible, results from the UK National External Quality Assurance Scheme have identified some errors in performance and interpretation (Preston et al, 1999). As described above, there are significant difficulties in the accurate diagnosis and classification of deficiencies of the natural anticoagulants – antithrombin, protein C and protein S. Diagnostic uncertainty is thus commonly encountered. Over- and underdiagnosis of thrombophilia may result in misleading advice and potentially harmful clinical and lifestyle decisions.

Clinical usefulness of case finding

There is no evidence that the detection of a heritable thrombophilic defect is useful in guiding clinical decision-making in relation to the choice of anticoagulant or to the intensity or duration of anticoagulant therapy to treat a thrombotic event. Seeking laboratory evidence of heritable thrombophilia in most cases of venous thrombosis is unlikely to lead to information of value in the clinical management of the individual case.

Interpretation of results

When laboratory investigation for heritable thrombophilia is pursued, it is essential that facilities are in place for the provision of informed and detailed advice based on a clear appreciation of the limitations of the laboratory tests and the major uncertainties regarding the clinical usefulness of the results obtained.

In the case of the more prevalent heritable thrombophilic defects, many heterozygotes will never suffer a thrombotic event and most events will be non-life-threatening. An inexperienced counsellor may engender considerable anxiety on the one hand and on the other may risk giving false reassurance from negative test results. An example is the use of laboratory tests to predict the risk of oral contraceptive-related thrombosis in women with a family history of venous thromboembolism – absence of laboratory evidence of a heritable thrombophilic defect is informative only if a genetic abnormality segregating with clinical thrombosis has already been identified within the kindred.

There is evidence that the risk of venous thrombosis is increased in women with thrombophilia when they use combined oral contraceptives and there is compelling evidence that, at least in the case of factor V Leiden, the polymorphism interacts with combined oral contraceptives to produce a relative risk of venous thrombosis that is considerably greater than would be predicted from the additive effect of the relative risks (Vandenbroucke et al, 1994). It has been suggested that there may be some value in the identification of factor V Leiden heterozygotes to aid risk counselling and contraceptive choice (Hellgren et al, 1995). However, it has been estimated that if population screening was adopted and oral contraception eschewed in all heterozygotes, over 2 million women would need to be tested to prevent one death from oral contraceptive-related pulmonary embolism each year, assuming that 1% of all venous thromboembolic events would be fatal (Rosendaal, 1996). Clearly a greater number of non-fatal thromboses may be avoided but, as the combined pill is the most efficient form of contraception, some additional pregnancies with associated thrombotic risk would occur, potentially negating any benefit (Vandenbroucke et al, 1996).

Similar calculations can be made in relation to assessment of thrombotic risk in pregnant women (Vandenbroucke et al, 1996). The highest risk is during the post-partum period. Factor V Leiden heterozygotes could be offered anticoagulant prophylaxis for 6 weeks after delivery. However, this approach suggests the need for pharmacological thromboprophylaxis in around 3·5% of women in many populations. Use of coumarin would undoubtedly result in some fatal bleeds and there is no evidence for overall clinical benefit from such a strategy of widespread thromboprophylaxis for all subjects with factor V Leiden or indeed other heritable thrombophilias.

There is accumulating evidence that heritable thrombophilia is associated with other pregnancy complications including an increased risk of late fetal loss (Preston et al, 1996; Grandone et al, 1997), pre-eclampsia and intrauterine growth restriction (Dizon Townson et al, 1996). There is only preliminary evidence regarding the utility of antithrombotic therapy to prevent pregnancy loss (Brenner, 2000) and testing for heritable thrombophilia is not generally justifiable in this patient population.

Heritable thrombophilia and arterial thrombosis

There is no good evidence currently available to support the hypothesis that heritable thrombophilias increase the risk of arterial disease. A number of studies have reported no increased prevalence of the factor V Leiden or prothrombin G20210A mutation in patients with myocardial infarction or stroke (Longstreth et al, 1998) but there may be an amplification of the risk of myocardial infarction in patients with factor V Leiden or prothrombin G20210A who have other coronary atherosclerosis risk factors such as smoking (Rosendaal et al, 1997; Doggen et al, 1998). Some small studies have suggested that antithrombin, protein C or protein S deficiencies may be associated with an increased risk of non-haemorrhagic stroke, but larger studies have not supported these findings in adults (Douay et al, 1998; Munts et al, 1998).

Recommendations on selection of patients for thrombophilia case finding

Indiscriminate application of laboratory investigations is clinically inappropriate, a waste of scarce resources and can be misleading. Diagnostic uncertainty is frequent.

• Testing of unselected patients is inappropriate and should be avoided

• Haematologists must give clear guidance to clinical colleagues on the selection of patients for testing and should oversee requests

• There is little if any clinical value in testing for heritable thrombophilia if the appropriate mechanisms for tracing, careful informed counselling and testing of at risk relatives are not in place, as there is a risk of engendering confusion, misinformation, false reassurance and unnecessary anxiety

• In relation to heritable thrombophilia, in most instances any value of laboratory testing will relate to the possibility of preventing a first venous thromboembolic event in affected relatives. The effectiveness and risks of this approach have not been formally assessed. The potential for any such benefit should form part of the criteria for testing. For example, testing for thrombophilia is unlikely to be informative in an elderly subject with a first venous thromboembolism in whom the family history is negative. In contrast, where unprovoked venous thrombosis occurs in a subject with a positive family history or in a young subject with children or siblings, especially female children or women of child-bearing age, identification of heritable thrombophilia may assist in counselling of affected relatives regarding the avoidance of risk. In adopting this approach it must be borne in mind that, in most instances, heritable thrombophilia represents a non-life-threatening late-onset genetic disorder that will not manifest clinically in a large proportion of affected individuals. The presence of a strongly positive family history of thrombosis may serve as a useful indicator of possible benefit from counselling and case-finding

• Testing for heritable defects and, in particular, genetic testing should be avoided in children unless there is a very strong clinical indication for it

• Although it is widely accepted that women with a history of three or more consecutive pregnancy losses should be screened for antiphospholipid antibodies, it would be premature to recommend that testing of these women or women with a history of pre-eclampsia or intrauterine growth retardation should be extended to include testing for heritable defects.

Management of acute venous thromboembolic events

Management of a first acute thrombotic episode

 • The initial management of deep vein thrombosis or pulmonary embolism in patients with heritable thrombophilia is, in general, no different from the management of venous thrombosis in any other patient. (Grade B recommendation).

For the majority, initial anticoagulation with unfractionated or low-molecular-weight heparin for a minimum of 5 d is followed in the non-pregnant by oral anticoagulation for 6 months at a target International Normalized Ratio (INR) of 2·5 (range 2·0–3·0) (British Committee for Standards in Haematology (BCSH), 1998). Very occasionally thrombolytic therapy may be appropriate.

Choice of anticoagulant and intensity of therapy Although heparin resistance and thrombus progression are theoretical risks in antithrombin deficiency, retrospective data suggest that these are infrequent problems in clinical practice (Schulman & Tengborn, 1993). Also, although warfarin-induced skin necrosis may complicate the introduction of coumarin therapy in subjects deficient in protein C or protein S and is, in theory, more likely to occur if there is under-anticoagulation with heparin when the oral agent is introduced, the condition is extremely rare. Therefore, there is no reason to believe that any change to the standard regimen for induction of anticoagulation is indicated in these deficiency states.

A target INR of 2·5 is generally deemed to be appropriate for the management of acute venous thromboembolism unless there is thrombosis recurrence on treatment, in which case a higher target INR is recommended (British Committee for Standards in Haematology (BCSH), 1998).

• Although there is no randomized study that addresses the issue, published reports and clinical experience do not suggest that more intensive oral anticoagulant therapy is usually indicated for the treatment of venous thromboembolism in patients with protein C, protein S or antithrombin deficiency (Grade C recommendation)

• Based on retrospective data, the presence of the most common heritable thrombophilias, heterozygosity for factor V Leiden or prothrombin G20210A, should not influence the usual recommendation to use a target INR of 2·5 for the treatment of acute venous thromboembolism (Grade C recommendation).

Duration of anticoagulant therapy• After a first venous thromboembolism, anticoagulant therapy is generally administered for 6 months. A shorter period of treatment may be acceptable when the thrombus is confined to distal veins (calf veins) and if there is evidence of a temporary risk factor that is no longer present. It is recommended that when there is a persisting thrombotic risk factor such as cancer or already identified high-risk thrombophilic defects (e.g. type I or type II reactive site antithrombin deficiency or combined defects), consideration should be given to extending the usual period of anticoagulation on an individual patient basis (Grade B recommendation)

• Identification of the most prevalent forms of heritable thrombophilia, heterozygosity for factor V Leiden or prothrombin G20210A, should not influence decisions about the duration of anticoagulant therapy (Grade B recommendation).

The primary aim in increasing the intensity of anticoagulation or extending the duration of anticoagulation is to minimize the risk of thrombus recurrence. The risk of recurrence after discontinuation of oral anticoagulant therapy after a first venous thromboembolism in unselected patients is substantial, amounting to around 15–20% by 2 years (Prandoni et al, 1996). Although this recurrence rate can be reduced by continuation of anticoagulation for at least 24 months (Kearon et al, 1999), this is associated with an inevitable and unavoidable increased risk of haemorrhage. On standard anticoagulant therapy, major haemorrhage occurs at a rate of around 1% per year of treatment and one quarter of these bleeds are fatal (Palareti et al, 1996). For the patient who has had a first venous thrombotic event, the benefits of long-term treatment (more than 6 months) have not yet been shown to outweigh the risks.

Because deficiency of the natural anticoagulants antithrombin, protein C or protein S, is relatively uncommon, there is a lack of reliable data on the risk of recurrent venous thromboembolism in affected individuals. Although recurrent venous thrombosis after discontinuation of warfarin therapy has been stated to be high in antithrombin and protein S deficiency (Van den Belt et al, 1997), other studies have suggested that the recurrence rate may be similar to that in unselected patients with venous thrombosis (Prandoni et al, 1996).

The risk of thrombosis recurrence after stopping oral anticoagulant therapy in subjects heterozygous for factor V Leiden has been compared with that in subjects with no detectable heritable thrombophilia. Two studies suggested a higher recurrence rate in factor V Leiden heterozygotes (Ridker et al, 1995b; Simioni et al, 1997), but one was retrospective and involved small numbers of subjects, and in neither study did the strength of the findings support the use of long-term anticoagulation for all factor V Leiden carriers. In four other studies, including three of prospective design, the thrombosis recurrence rate after discontinuation of anticoagulant therapy was not increased in factor V Leiden heterozygotes (Rintelen et al, 1996; Eichinger et al, 1997; Kearon et al, 1999; Lindmarker et al, 1999), although homozygotes may be at increased risk (Lindmarker et al, 1999).

There is also evidence that the risk of recurrence is no higher for heterozygotes with the prothrombin G20210A mutation than in those without (Eichinger et al, 1999; Kearon et al, 1999; Lindmarker et al, 1999).

Patients who have combinations of thrombophilias seem to be at additional risk of venous thromboembolism (Zoller et al, 1995; Van Boven et al, 1996, 1999; Makris et al, 1997). The risk of recurrent events in such subjects is not known and whether they merit a longer duration of anticoagulation is open to question. These patients may, however, not form a discrete group as far as treatment decisions are concerned, as many subjects in whom no or a single heritable condition has been detected may well have additional, as yet unrecognized, genetic or environmental predisposition to thrombosis.

Management of recurrent venous thrombosis

 • When recurrent events have occurred while the patient was no longer anticoagulated, it is sufficient to reintroduce coumarin at a target INR of 2·5 after initial treatment with heparin, but when a recurrent event has occurred while the patient was on anticoagulants and their INR was within the target range of 2·0–3·0, an increase in the intensity of anticoagulation to a target INR of 3·5 (range 3·0–4·0) is indicated (Grade C recommendation)

• In general, patients who have had two or more apparently spontaneous venous thrombotic events require consideration for indefinite anticoagulant thromboprophylaxis (Grade C recommendation).

However, patients who have had recurrent thrombotic events in association with identifiable prothrombotic triggers (for example pregnancy, surgery, oestrogen use) and in whom those prothrombotic triggers are no longer present may not require indefinite anticoagulant thromboprophylaxis but do require prophylaxis during high-risk situations.

Thrombosis prevention

There is no evidence to support a policy of long-term pharmacological primary thromboprophylaxis of asymptomatic family members found to have a thrombophilic genotype. The risk of serious or fatal haemorrhage considerably outweighs the risk of a fatal venous thrombotic event even for patients with the most severe types of thrombophilias, for example type I antithrombin deficiency or combinations of thrombophilic conditions.

As patients who have had a thrombotic event come towards the end of their period of anticoagulant treatment, they should be counselled about the signs and symptoms of thromboembolism and should be aware of the requirement to seek medical attention early if they suspect a recurrence. They should also be advised that they should mention their history of venous thromboembolism to medical attendants so that appropriate decisions about short-term thromboprophylaxis at times of increased thrombotic risk can be made.

The use of an identification card stating the nature of the thrombophilia may be considered for those with a past personal history of VTE or a family history of VTE co-segregating with an identified deficiency or genotype.

• All patients with a past history of VTE (with or without evidence of a thrombophilic defect) merit consideration for short-term thromboprophylaxis to cover periods of increased thrombotic risk, for example surgery, trauma, plaster casts or immobilization (Grade C recommendation)

• In addition, affected (with an identifiable thrombophilia) but asymptomatic relatives of thrombophilic patients who have had a venous thrombotic event merit consideration for short-term thromboprophylaxis to cover similar periods of increased thrombotic risk (Grade C recommendation).

There is no evidence to suggest that patients with thrombophilia in general merit more intense regimens of anticoagulant prophylaxis or prophylaxis for a longer duration than other patients of similar age and in similar clinical circumstances.

A link between long journeys involving lengthy periods of immobility and venous thromboembolism risk has been proposed but, to date, the evidence is only circumstantial (Geroulakos, 2001). Any advice given to patients therefore has to be empirical and based on their individual history.

Contraceptive advice

Most family planning clinics would advise against the use of a combined oral contraceptive pill as the first-choice contraceptive for women who have a personal history of venous thrombosis and many may also suggest that a woman who has a family history of at least one first-degree relative with a history of proven venous thrombosis should consider using a contraceptive method other than the combined pill as their first choice. Testing for heritable thrombophilia rarely helps in decision-making in this context and is only fully informative if a family study is performed. It must be borne in mind that 6% or more of asymptomatic women are heterozygous for factor V Leiden or prothrombin G20210A, that the combined pill is the most acceptable and effective form of contraception for many women, and that the thrombotic risk associated with its use is low, even in these heterozygotes.

The progestagen content of combined oral contraceptive pills appears to be important in determining the level of thrombotic risk. Preparations containing third-generation progestagens are associated with a higher risk of VTE than pills containing second-generation progestagens. Progestagen-only preparations used to treat menstrual disorders are associated with increased venous thrombosis risk (Poulter et al, 1999) but, in the general population, progestagen-only pills used for contraception appear not to be associated with increased venous thrombosis risk (Vasilakis et al, 1999). Although there is no published information about the risk of venous thrombosis in women with thrombophilia using progestagen-only preparations, many clinicians would consider prescribing progestagen-only pills or progestagen-bearing intrauterine devices for women with a personal or family history of venous thromboembolism, including those with heritable thrombophilia.

Management of pregnancy

Risk of pregnancy-associated venous thromboembolism

Venous thromboembolism is the major cause of maternal mortality in the developed world. The most recent report of the Confidential Enquiries into Maternal Deaths in the United Kingdom demonstrated an increase in the number of deaths in the triennium 1994–96 compared with the preceding triennium (Confidential Enquiries into Maternal Deaths., 1998). The incidence of pregnancy-associated venous thromboembolism is approximately 0·9 per 1000 deliveries, the risk being greater post partum than ante partum (McColl et al, 1997). In approximately two-thirds of patients who suffer a pregnancy-related VTE, acquired thrombotic risk factors including age (over 35 years), high parity (four or more), intercurrent illness, immobility and caesarean section, can be identified (McColl et al, 1997).

In a retrospective study of unselected consecutive patients with confirmed venous thromboembolism associated with pregnancy, around 30% of patients were found to have a heritable thrombophilia (McColl et al, 1997). A number of studies have suggested that, in the absence of anticoagulant prophylaxis, patients with antithrombin deficiency are at high risk of pregnancy-associated thromboembolism (Conard et al, 1987, 1990; Hellgren et al, 1992; De Stefano et al, 1994; Pabinger & Schneider, 1996). All these early studies were subject to bias as the patients studied came from symptomatic kindred. In the study reported by McColl et al (1997), the patients were unselected and previously uninvestigated. It was calculated that patients with type I antithrombin deficiency had an approximately 35% chance of developing venous thrombosis associated with the index pregnancy. The incidence of pregnancy-associated thrombosis in women with protein C or protein S deficiency appears to be considerably lower than that for antithrombin-deficient women (Conard et al, 1990; De Stefano et al, 1994; Friederich et al, 1996; Pabinger & Schneider, 1996; McColl et al, 1997). In an early report, Hellgren et al (1995) suggested that 60% of women who had had pregnancy-associated venous thromboembolism had evidence of APC resistance. More recent studies have, however, demonstrated that the prevalence of factor V Leiden in women with a history of pregnancy-associated venous thrombosis is considerably less than 60% and may be between 10% and 15% (Hough et al, 1996; McColl et al, 1997).

It is essential that in assessing thrombotic risk associated with pregnancy, acquired factors, as well as genetic predisposition, are taken into account. Patient age, parity and weight have each been shown to be associated with pregnancy and puerperal venous thrombotic risk. As in the non-pregnant, immobilization and serious medical disorders increase thrombotic risk and, in the post-partum period, the risk of venous thrombosis is increased in women who have had a caesarean section (particularly as an emergency) or a difficult instrumental delivery. Formal assessment of venous thrombotic risk should be made in individual patients at each antenatal review, at delivery and post partum.


Thrombophilia clinics looking after young women must have close liaison with an obstetric unit familiar with the management of thrombophilic women. Women with identified heritable thrombophilia from symptomatic kindred should be given information about the perceived risk of pregnancy-associated venous thrombosis in relation to their particular defect and clinical history and allowed the opportunity to discuss plans for managing future pregnancies. Women who are on long-term oral anticoagulants should understand the risk of fetal complications associated with maternal coumarin ingestion and should be strongly encouraged to report any expected menstrual period that is overdue by 3 d or more as soon as possible. Immediate pregnancy testing should be offered and, if the pregnancy test is positive, ultrasound confirmation of pregnancy sought if possible by 5–6 weeks gestation. For women on long-term oral anticoagulant prophylaxis, a number of possible approaches may be considered. For a very few, it may be reasonable to discontinue oral anticoagulation and institute self-administered heparin when the patient starts trying to conceive. This approach has the significant potential disadvantage of exposing the woman to an extended period of heparin therapy before pregnancy is achieved and should be reserved for use only in women who have already demonstrated the fertility of their partnership and who are unwilling to accept the possibility of risk to their fetus from continuing coumarin for even a very limited period after conception. A second approach is to continue the oral anticoagulant until conception is confirmed and only then to substitute heparin. This approach requires the full co-operation and understanding of the patient, her thrombophilia clinic and the obstetric unit to ensure that there will be no delay in diagnosing and confirming pregnancy so that coumarin can be discontinued no later than 6 weeks gestation.

Prenatal diagnosis

To date, prenatal diagnosis has been performed in a few instances only in an attempt to avoid severe thrombophilia (Millar et al, 1994; Lane et al, 1996). It can be considered appropriate only in the rare cases in which the fetus may be expected to be homozygous or compound heterozygous for coagulation inhibitor (antithrombin, protein C or protein S) defects.

Classification of patients at increased risk of pregnancy-associated venous thromboembolism

The management of pregnant women with known thrombophilic defects and no prior venous thromboembolism remains controversial because of the lack of information about the natural history of the various thrombophilias and the lack of randomized trials of thromboprophylaxis in these patients. Type I and type II reactive site antithrombin deficiencies appear to be associated with a higher risk of pregnancy-associated venous thrombosis than other heritable thrombophilias. Accordingly, many experts suggest that women with these defects be managed more aggressively than those with other heritable thrombophilias. Pregnant women with a history of previous venous thrombosis may be at increased risk of a recurrent event. The magnitude of the risk is not known but estimates range from 0% to 15% (Howell et al, 1983; De Swiet et al, 1987; Tengborn, 1989). The risk may be lower than suggested by some of these studies as in the past it was uncommon to seek objective confirmation of venous thrombosis and recurrence rates may have been overestimated. In a prospective study of 125 women with a past history of a single previous (objectively confirmed) venous thrombosis in whom antenatal thromboprophylaxis was withheld, the overall rate of recurrence was 2·4%. There were no recurrences in women who did not have an identifiable thrombophilic defect and in whom the previous event was associated with a temporary and no longer present acquired thrombotic risk factor. The recurrence rate in women who had a thrombophilic defect and/or in whom the previous event was apparently spontaneous was 5·9% (Brill-Edwards, 2000). Based on these results, antenatal thromboprophylaxis may be unwarranted in women with no evidence of thrombophilia in whom a previous venous thrombosis was associated with a temporary and no longer present risk factor. The risk of venous thromboembolism is greater post partum than ante partum (McColl et al, 1997). Accepting that women with heritable thrombophilia and women with a past history of venous thrombosis are at increased risk of pregnancy and/or puerperal venous thromboembolism, they may be classified arbitrarily as being at high, moderately or slightly increased risk.

(i) Women at high risk of pregnancy-associated venous thromboembolism.

• Women who are on long-term anticoagulant thromboprophylaxis and women who have type I antithrombin deficiency or a type II reactive site antithrombin defect (whether or not they have already had a thrombotic episode) should be considered to be at high risk of antenatal or postnatal thrombosis.

(ii) Women at moderately increased risk of pregnancy-associated venous thromboembolism.

• Women who have a previous history of venous thromboembolism and who have evidence of a thrombophilic defect and women who have a previous history of spontaneous venous thromboembolism (whether or not a genetic predisposition has been identified), who are no longer on long-term anticoagulant prophylaxis, should be considered to be at moderately increased risk of pregnancy associated thrombosis

• Women who have no personal history of venous thrombosis but give a family history of proven venous thrombosis and who have therefore been screened for thrombophilia and been found to be heterozygous for protein C deficiency, or homozygous for factor V Leiden or the prothrombin G20210A mutation, or to have combinations of defects (excluding type I or type II reactive site antithrombin deficiency), should be considered to be at moderately increased risk of pregnancy associated thrombosis.

(iii) Patients at slightly increased risk of pregnancy-associated venous thromboembolism.

• The slightly increased risk category includes women who have no personal history of venous thrombosis but who have been screened for thrombophilia because they have offered a family history of venous thrombosis and as a result have been found to be heterozygous for protein S deficiency, heterozygous for factor V Leiden or heterozygous for prothrombin G20210A

• Also included in this risk category are women with a history of previous venous thrombosis occurring in association with a temporary and no longer present acquired risk factor and no identifiable thrombophilic defect.

Management during pregnancy

 • All women in the high and moderately or slightly increased risk categories should be encouraged to wear graduated compression stockings throughout their pregnancy and for 6–12 weeks after delivery (Grade C recommendation).

(i) Women at high risk of pregnancy-associated venous thromboembolism.

These women should usually be offered antenatal anticoagulant prophylaxis throughout their pregnancy, changing to heparin or introducing heparin as soon as pregnancy is confirmed. Many obstetric units now prefer to use low-molecular-weight (LMW) heparin rather than unfractionated heparin for antenatal thromboprophylaxis (although no LMW heparin is specifically licensed for use in pregnant women). There is a lack of consensus on appropriate dosage regimens for unfractionated or LMW heparin for this high-risk group of women. Pre-filled syringes simplify the outpatient management of women at increased VTE risk and reduce the risk of inadvertent overdosing.

• It is suggested that, for women at high thrombotic risk, adjusted doses of low-molecular-weight or unfractionated heparin, higher than those usually used for the prevention of venous thrombosis, may be offered (Grade C recommendation)

• If a LMW heparin is used it could be introduced at a dose of around 75 anti Xa units/kg early pregnancy weight 12 hourly subcutaneously (12 hourly dose rounded to the nearest 1000 anti Xa units). This dose may be expected to give peak plasma anti Xa activity by chromogenic substrate assay of between 0·35 and 0·5 units/ml 3 h after injection

• If subcutaneous unfractionated heparin is used, doses could be administered 8–12 hourly subcutaneously and adjusted to give a peak APTT ratio equivalent to a plasma heparin activity of 0·2–0·4 units/ml by protamine sulphate titration, equivalent to 0·35–0·70 units/ml by anti Xa assay 4 h after injection (Grade C recommendation). This is often difficult to achieve and maintain (Kitchen & Preston, 1996; Kitchen et al, 1996)

• In some patients it may be reasonable to use fixed prophylactic doses of unfractionated or LMW heparin (4000–5000 anti Xa units once daily subcutaneously or 7500 IU unfractionated heparin 8–12 hourly subcutaneously)

• The platelet count should be checked before introducing heparin and at around 4–8 d of treatment (Olson et al, 1998).

There is a poor relationship between anti Xa levels and the risk of bleeding or thrombosis. Empirically it has been suggested that, if a low-molecular-weight heparin is used, 12 hourly injection may be preferable to once daily injection as the clearance of heparins is increased during pregnancy. Although in non-pregnant patients routine monitoring of anti Xa activity is not necessary, during pregnancy the peak plasma anti Xa activity should be checked after the first month's use and 4–6 weekly thereafter to ensure that the expected anti Xa activity level has been achieved and not exceeded.

Women who are on long-term thromboprophylaxis because they have a prosthetic heart valve require intensive anticoagulation during pregnancy, but this is beyond the scope of this paper.

(ii) Women at moderately increased risk of pregnancy-associated venous thromboembolism.

• Women in the moderately increased risk category merit consideration of antenatal thromboprophylaxis. Fixed prophylactic doses of unfractionated or LMW or unfractionated heparin (4000–5000 anti Xa units LMW heparin once daily subcutaneously or 7500 IU unfractionated heparin 8–12 hourly subcutaneously) are usually adequate (Grade C recommendation)

• Monitoring of anti Xa activity or the APTT or is not generally necessary, but the platelet count should be checked at baseline and 4–8 d after introduction of heparin.

There is debate about when this prophylaxis should be started. Some obstetricians introduce prophylaxis early in pregnancy, whereas others prefer to wait until the second trimester. It should be borne in mind that a significant proportion of pregnancy-related VTE occur in the first trimester. Other thrombotic risk factors such as intercurrent disease, immobilization or dehydration resulting from hyperemesis must be taken into consideration.

(iii) Patients at slightly increased risk of pregnancy-associated venous thromboembolism.

• In general, these women do not require anticoagulant thromboprophylaxis antenatally but consideration should be given to anticoagulant prophylaxis following delivery

• In deciding which women should receive pharmacological thromboprophylaxis, consideration should be given to the details of the family history of thrombosis, including age at presentation and presence of additional risk factors in affected relatives, as well as in the subject in question. Based on this information, many women can be regarded as being at lower risk and pharmacological prophylaxis is not mandatory.

Management of delivery

Concern has been expressed about the risk of peripartum haemorrhage in women given heparins during pregnancy. One early study suggested that prolongation of the APTT may persist longer than expected in women given unfractionated heparin (Anderson et al, 1991). More recent studies have shown no increased risk of clinically significant haemorrhage in women given low-molecular-weight heparins during pregnancy compared with unanticoagulated women (Sanson et al, 1999). In previous guidelines it was recommended that epidural or spinal anaesthesia could be used safely in women who had been given unfractionated heparin during pregnancy, providing their coagulation screen was within normal and their platelet count was greater than 80 × 109/l (British Committee for Standards in Haematology, 1993). Recently, there has been considerable alarm raised about the possible increased risk of significant spinal bleeding after neuroaxial block in patients on low-molecular-weight heparins (Horlocker & Heit, 1997). While it is clearly sensible to avoid inserting or removing an epidural or spinal catheter at peak heparin levels, it should be appreciated that exposure to general anaesthesia (to allow a non-elective caesarean section) may be associated with a greater risk than the risk of epidural haematoma.

• In spite of considerable debate, it remains unclear exactly what period of time should elapse between the last prophylactic or therapeutic dose of LMW heparin and insertion or removal of an epidural or spinal catheter or how long the time interval should be until the next dose. Practically, it may be reasonable to allow at least 12 h to elapse after the last fixed prophylactic dose of LMW heparin before inserting an epidural or spinal catheter, but a delay of up to 24 h may be appropriate in patients on higher adjusted or full therapeutic doses of LMW heparins. Local policies should be decided after discussion with the anaesthetists providing the service (Grade C recommendation).

Post-partum management

 • All women at high or moderately increased risk of pregnancy-associated venous thrombosis require thromboprophylaxis during the puerperium and consideration should be given to post-partum thromboprophylaxis for women in the slightly increased risk category

• Providing the patient is not bleeding excessively, unfractionated heparin or LMW heparin may be introduced or reintroduced within 12 h of completion of delivery in fixed prophylactic doses, as described above for the management during pregnancy of women at moderately increased thrombotic risk (Grade C recommendation)

• If the patient does not wish to continue injections throughout her puerperium she may change to an oral anticoagulant. Generally, oral anticoagulants may be introduced on the first or second post-partum day and the heparin withdrawn when the INR is within the recommended therapeutic range (usually 2·0–3·0) for two consecutive days (Grade C recommendation)

• In general, post-partum anticoagulation should be continued for 6 weeks in women at high, moderately or slightly increased risk of pregnancy-associated venous thrombosis (Grade C recommendation).

Heparin is not secreted in breast milk and warfarin has a high degree of protein binding and is not secreted in any large quantity from the breast.

• Women using heparin or coumarin post partum may be encouraged to breast-feed.

Treatment of VTE during pregnancy

The management of venous thromboembolism occurring during pregnancy in a woman with thrombophilia is no different from the management of venous thromboembolism in any other pregnant woman (Grade C recommendation).

Most experts consider continuation of full therapeutic doses of heparin for the remainder of the pregnancy and anticoagulation for at least 6–12 weeks post partum the treatment of choice. The total duration of anticoagulation should usually be no less than 6 months (Grade C recommendation).

Hormone replacement therapy

Until recently it was widely believed that hormone replacement therapy (HRT) was not associated with an increased risk of venous thromboembolism. However, a series of recently published studies provide clear evidence linking hormone replacement therapy to increased venous thrombosis risk (Daly et al, 1996; Grodstein et al, 1996; Jick et al, 1996; Gutthann et al, 1997). One study found a significantly increased risk of VTE in women with a history of previous VTE using HRT compared with women with no history of previous VTE (Hoibraaten et al, 2000).

There are no published studies examining VTE risk in women with thrombophilia using HRT. In a recent publication, Lowe et al (2000) reported a follow-up case–control study of women aged 45–64 years with a history of idiopathic VTE. At follow-up, blood samples were obtained to allow measurement of 20 variables of haemostasis. The relative risk of VTE showed significant associations with certain thrombophilic markers, including increased APC resistance, low antithrombin and low protein C. For HRT users with abnormalities of haemostasis, the relative risks of VTE were around 3·5–4·0 times the risk in non-HRT users with the same defect(s).

There is limited information about the risk of VTE in users of Selective Estrogen Receptor Modulators (SERMs), but in a randomized placebo-controlled trial the relative risk of VTE in users of Raloxifene was 3·1 (95% CI 1·5–6·2), suggesting that the risk is similar to that with oestrogen-containing HRT (Ettinger et al, 1999).

Because of a lack of evidence about the risk of HRT-related VTE in women with other thrombotic risk factors, advice and therapeutic strategies have to be based on clinical logic and expert opinion (Greer, 1998; Royal College of Obstetricians and Gynaecologists, 1998).

There is widespread agreement that, on presently available evidence, universal screening of women for thrombophilc defects prior to the prescription of oestrogen in combined oral contraceptives or in hormone replacement therapies is inappropriate and should be discouraged.

Women with an established requirement for oestrogen replacement who have a personal history of previous VTE present a particular challenge. Recently published evidence suggests that women with a personal history of previous VTE may be at increased risk of a recurrence if they use oestrogen replacement therapy (Hoibraaten et al, 2000). In women with a past history of VTE who have a clear indication for HRT (for example, severe and debilitating menopausal symptoms), some clinicians recommend simultaneous use of oral anticoagulant prophylaxis, but the risk of anticoagulant-related bleeding must be taken into account in the risk:benefit analysis. On standard anticoagulant thromboprophylaxis (INR 2·0–3·0), major haemorrhage occurs at a rate of around 1% per year of treatment and one quarter of these bleeds are fatal (Palareti et al, 1996). Women who have no personal history of VTE but who have a family history of proven VTE may be offered thrombophilia screening after careful counselling regarding its risks and limitations. In these women, if a thrombophilic defect is found which segregates with thrombosis in family members or if the defect is a ‘severe’ defect (e.g. antithrombin deficiencies or combinations of defects), HRT is relatively contraindicated. When its use is considered to be essential, combined oral anticoagulant prophylaxis has been used. However, exposure of a woman who has never suffered thrombosis to the inevitable bleeding risk associated with oral anticoagulant therapy requires careful consideration. The risk of anticoagulant-related haemorrhage probably outweighs the risk of HRT-related venous thrombosis in women with a family history of VTE but no personal history of VTE, who have no identifiable thrombophilic defect or who have one of the defects usually associated with a lesser risk of VTE (heterozygosity for factor V Leiden or the prothombin 20210A polymorphism).

Although there are no epidemiological studies comparing the risk of VTE in users of transdermal and oral HRT preparations, transdermal preparations are associated with lesser prothrombotic changes in haemostasis.

All women commencing HRT should be counselled about the risk of VTE, should be aware of the signs and symptoms of VTE, and should be able to access medical help rapidly if they suspect that they have developed a thrombus. Women using anticoagulants should also be aware of the danger of haemorrhage and the requirement for regular dose monitoring.


The members of the Haemostasis and Thrombosis Task Force wish to thank the following for their comments and advice: Dr G. Dolan, Profesor I. A. Greer, Dr J. Keidan, Professor G. D. O. Lowe, Dr J. A. Murray, Dr K. Ryan, Mrs B. Smith, Dr H. G. Watson and Dr E.J. Watts.


  1. on behalf of the Haemostasis and Thrombosis Task ForceBritish Committee for Standards in Haematology


This Guideline has been written by the Haemostasis and Thrombosis Task Force of the British Committee for Standards in Haematology, a subcommittee of the British Society for Haematology. Members of the Haemostasis and Thrombosis Task Force are M. Greaves (Chair), I. D. Walker, F. E. Preston, T. Baglin (Secretary), S. Machin, T. W. Barrowcliffe, M. Winter and S. Kitchen.

The recommendations are not intended to dictate an exclusive course and the guidance must be evaluated in the context of the individual patient. The principal authors are practising clinical haematologists with special expertise in the investigation and management of thrombophilia. The manuscript was reviewed by a ‘sounding board’ of 12 practising clinicians representing a number of clinical specialties, e.g. internal medicine and obstetrics and gynaecology, by nurse specialists and by a patient representative prior to submission for publication.

Targeted literature searches with Medline and Embase were used to gather evidence on which to base recommendations. The evidence used was graded using the scheme below and the recommendations formulated with a standardized grading scheme.

Classification of evidence

• Ia: Evidence obtained from meta-analysis of randomized controlled trials

• Ib: Evidence obtained from at least one randomized controlled trial

• IIa: Evidence obtained from at least one well-designed controlled study without randomization

• IIb: Evidence obtained from at least one other type of well-designed quasi experimental study

• III: Evidence obtained from well-designed non-experimental descriptive studies, such as comparative studies, correlation studies and case studies

• IV: Evidence obtained from expert committee reports or opinions and/or clinical experience of respected authorities.

Grades of recommendations

• A: Requires at least one randomized controlled trial as part of a body of literature of overall good quality and consistency addressing the specific recommendation (evidence levels Ia, Ib)

• B: Requires the availability of well-controlled clinical studies, but no randomized clinical trials on the topic of the recommendation (evidence levels IIa, IIb, III)

• C: Requires evidence obtained from expert committee reports or opinions and/or clinical experience of respected authorities. Indicates an absence of directly applicable clinical studies of good quality (evidence level IV).

Although the advice and information contained in these guidelines is believed to be true and accurate at the time of going to press and to be valid until December 2003 when they will be reviewed, neither the authors nor the publishers can accept any legal responsibility or liability for any errors or omissions.