The investigation and management of inherited thrombophilia

Antithrombin deficiency

AT, a member of the serine protease inhibitor (serpin) superfamily, is a 58-kDa single-chain glycoprotein, synthesized primarily in the liver. It is the major plasma inhibitor of thrombin but also inhibits other activated serine proteases involved in blood coagulation, including factors Xa, IXa, XIa, XIIa and kallikrein. The gene for AT is located on chromosome 1q23–25 and includes 13.4 kb of genomic DNA containing 7 exons ( Olds et al. 1993 ). The first case of hereditary AT deficiency was described more than 30 years ago ( Egeberg 1965) and many mutations in the AT gene have since been reported. The most recent update of the AT mutation database ( Lane et al. 1997 ) includes 80 different molecular events.

Hereditary AT deficiency is a heterogeneous and usually autosomal dominant disorder which may be classified as type I (classical) or type II deficiency ( Lane et al. 1993 ). Type I deficiency is characterized by low levels of functionally and immunologically determined AT, typically commensurately reduced to ≈ 50% of normal. Type II deficiency results from the production of a variant AT protein. Circulating AT antigen levels may be normal but functional activity levels are reduced by about 50%. This is because approximately half of the circulating AT antigen is associated with the variant protein. Type II AT deficiency may be caused by functional abnormalities of the AT reactive site (type II RS variants) or the heparin binding site (type II HBS variants). Alternatively it may result from multiple (pleiotropic) functional defects, affecting AT plasma levels, the reactive site and the heparin binding site (type II PE variants). The relative prevalence of type I and type II AT deficiency is uncertain.

Functional AT activity can be measured according to its ability to inhibit thrombin or factor Xa. Residual thrombin or factor Xa in AT assays was originally measured using clotting-based assays, but these have been superceded by the introduction of synthetic, chromogenic substrates, which are now in widespread use. Either heparin cofactor assays, which are performed in the presence of heparin, or progressive activity assays may be used to determine functional AT levels. Progressive activity assays are comparatively slow to perform and lack specificity, as they may reflect the activity of other plasma inhibitors such as α1-antitrypsin and α2-macroglobulin. Heparin cofactor assays are preferred, therefore. The possible influence of heparin cofactor II on heparin-cofactor assays may be minimized by reducing the heparin concentration, by using factor Xa rather than thrombin as a substrate, and by the use of bovine rather than human thrombin ( Perry 1994).

Radial immunodiffusion, electroimmunodiffusion or ELISA have all been used for the immunoassay of AT. Crossed immunoelectrophoresis in the presence of heparin may also be used, to distinguish type II HBS variants from RS variants ( Lane & Caso 1989). This may be of clinical value since heterozygotes for type II HBS variants have a low risk of thrombosis.

For diagnostic screening purposes, plasma AT levels should always be determined by means of a functional assay. Immunoassay alone will not permit the identification of many type II AT variants. Immunoassay is useful in combination with a functional assay for the diagnosis and classification of type II AT deficiency states.

The possibility of an acquired deficiency state should be excluded when considering a diagnosis of hereditary AT deficiency. A large number of conditions are associated with a reduction in plasma AT levels, including heparin therapy, DIC, pre-eclampsia, liver disease, nephrotic syndrome, major surgery and associated complications, and malignancies such as acute promyelocytic leukaemias. The use of l-asparaginase in the treatment of acute lymphoblastic leukaemia may also cause reduced AT levels. Small reductions in plasma AT levels may occur in users of oral contraceptives and in individuals receiving oestrogens for other purposes.

Protein C deficiency

PC is a vitamin K-dependent glycoprotein with a molecular mass of 62 kDa. It is synthesized in the liver as a single-chain polypeptide. Intracellular proteolytic cleavage results in a 2-chain disulphide-linked PC molecule which accounts for ≈ 90% of circulating PC. PC has a central role in the protein C anticoagulant pathway ( Dahlback & Stenflo 1994) where it is activated by thrombin in complex with the endothelial receptor, thrombomodulin. Activated PC (APC), together with its principal cofactor, PS, inactivates membrane-bound factor Va and factor VIIIa.

The PC gene (PROC gene) is located on chromosome 2q13–14, encompassing ≈ 10 kb of DNA and 9 exons ( Patracchini et al. 1989 ). More than 160 different mutations of the PROC gene have been described which result in type I or type II PC deficiency ( Reitsma et al. 1995 ).

PC deficiency associated with familial thrombotic disease was first described in 1981 (Griffin et al. ). The accepted view that heterozygous PC deficiency is a risk factor for venous thrombosis was challenged by Miletich et al. (1987) . In an investigation of more than 5000 healthy blood donors, 79 individuals with low PC levels were identified, none of whom had a personal or family history of venous thromboembolic disease. It was concluded from this study that PC deficiency was not a risk factor for VTE. Blood donors, however, are recruited on the basis of the absence of previous disease, and studies of this nature are likely to underestimate the relationship between a protein deficiency and thrombosis. Furthermore, the risk associated with hereditary factors cannot be estimated by studying healthy individuals only ( Reitsma 1997). The Leiden Thrombophilia Study, a case-control study of unselected patients with a first DVT and population-based thrombosis-free matched case controls, has subsequently clearly confirmed the association of PC deficiency with an increased risk for VTE ( Koster et al. 1995b ) ( Table 1).

Type I PC deficiency is a quantitative disorder with a concordant reduction in both PC activity and antigen. Type II PC deficiency is a qualitative defect due to the presence of a variant PC protein. This results in reduced PC activity but normal levels of PC antigen. Immunoassay (e.g. ELISA) alone will not identify type II variants, and an assay to determine functional PC levels should always be carried out when screening for PC deficiency.

PC is the zymogen of a serine protease, APC, and serine protease activity can only be measured after proteolytic activation. Functional PC assays vary with regard to the PC activator used, including thrombin, thrombin/ thrombomodulin, or snake venoms. Two methods are commonly used to quantify the amount of APC generated, either synthetic chromogenic substrate assays, or tests based on the prolongation of the APTT by the anticoagulant properties of APC ( De Stefano et al. 1996 ). It may be argued that coagulation-based assays are preferable to chromogenic assays as the latter are insensitive to some functional PC defects ( Marlar & Mastovich 1990) ( Table 2). On the other hand, coagulation-based assays for PC may give spuriously low results in the presence of APC resistance.

The possibility of an acquired cause of PC deficiency should be excluded when considering a diagnosis of a hereditary deficiency state. A number of conditions are associated with a reduction in plasma PC levels, including oral anticoagulant therapy, liver disease, DIC and vitamin K deficiency.

Protein S deficiency

PS is a vitamin K-dependent multidomain plasma glycoprotein which is synthesized by hepatocytes and also found in endothelial cells and platelets. It is a single-chain molecule of molecular mass 69 kDa, ≈ 60% of which circulates bound to C4b-binding protein. Unbound, or free, PS has a role as a non-enzymatic cofactor in the protein C anticoagulant pathway ( Dahlback 1991). PS has a high affinity for negatively charged phospholipids and it interacts with APC to propagate the formation of a membrane-bound APC-PS complex. This enhances the inactivation of membrane-bound factors Va and VIIIa by APC, and localizes the anticoagulant response. The C4b-bound fraction of PS does not function as an APC cofactor.

Two homologous genes for PS have been mapped to chromosome 3. The active PROS1 gene, located at 3p11.1–3q11.2 spans over 80 kb and comprises 15 exons ( Borgel et al. 1997 ). A second gene, PROS2, located close to the PROS1 gene, is a pseudogene. It has no open reading frame and contains multiple base changes, stop codons and frameshifts. Genetic diagnosis of PS deficiency is complicated by the high degree of homology between PROS1 and PROS2, and PCR primers must be designed selectively to amplify PROS1 sequences but not those of the pseudogene. PCR-based strategies have been used to identify 69 candidate mutations in type I PS deficiency (low total and low free PS antigen, low PS activity) ( Borgel et al. 1997 ). Gene defects have been identified in only 40% to 90% of PS deficient patients, however ( Mustafa et al. 1995 ; Simmonds et al. 1996 ). It is not known why many patients have no identifiable mutation, and a mechanism other than mutations within coding sequences may be a possible explanation ( Borgel et al. 1997 ). PS type II deficiency (normal free PS, low PS functional activity) is rare and only five candidate causal mutations have been identified. Type III PS deficiency is associated with normal total PS antigen levels and reduced free antigen levels and activity. The coexistence of type I and type III PS deficiency within the same kindred ( Zoller et al. 1995 ) implies that these subtypes may be phenotypic variants of the same genetic disorder. This mechanism has recently been confirmed in one large kindred where a single causative mutation (Gly295Val) was associated with both type I and type III PS deficiency ( Simmonds et al. 1997 ). It was proposed that the Gly295Val mutation causes type I PS deficiency. Total PS increases with increasing age, and this gives rise to a type III PS-deficient phenotype in some individuals.

Familial PS deficiency linked to recurrent VTE was first described in 1984 ( Comp & Esmon 1984; Schwartz et al. 1984 ). Transmitted as an autosomal dominant trait, PS deficiency has been identified in a number of studies to be associated with an increased risk for thrombosis ( Pabinger et al. 1994a ; Pabinger & Schneider 1996). A case-control study reported by the Leiden thrombophilia study group failed, however, to find an association between reduced PS levels and increased thrombotic risk ( Koster et al. 1995b ). It was suggested that thrombosis in affected families may be the result of cosegregating additional genetic defects. A subsequent case-control study ( Faioni et al. 1997 ), however, lent support to the concept that PS deficiency is an independent, albeit mild, risk factor for VTE (relative risk for thrombosis 2.4, Table 1).

The diagnosis of PS deficiency is relatively complex compared with that of AT or PC deficiency. Current commercially available functional assays have low specificity and the presence of APC resistance may give rise to spuriously low PS values. These assays are therefore not recommended ( De Stefano et al. 1996 ). Free and total PS antigen may be measured by immunoassay (usually ELISA), free PS antigen being determined following the precipitation of the PS/C4b-binding protein complex by mixing the test plasma with polyethylene glycol.

Total PS levels in normal subjects and heterozygous PS deficient individuals show substantial overlap, and total PS levels, unlike those of free PS, have also been shown to increase with increasing age ( Simmonds et al. 1997 ). Therefore the measurement of free PS provides a higher diagnostic sensitivity and specificity than the measurement of total PS.

It is important to note that healthy women have lower PS levels than healthy men, and separate male and female reference ranges should be constructed. There are a number of other factors to consider which may confound the diagnosis of a hereditary deficiency state. PS levels are substantially reduced during pregnancy and the post-partum period, by oral contraception and during oral anticoagulant therapy. Free PS levels may be reduced in the presence of a lupus anticoagulant. Liver disease, DIC or vitamin K deficiency may cause acquired PS deficiency and this possibility should be excluded. Conversely, there is a strong positive correlation of PS with alcohol intake ( Woodward et al. 1997 ). C4b-binding protein levels increase during acute phase reactions and this may have a modulating effect on free PS levels in the circulation. The measurement of PS is therefore best avoided close to a thrombotic event.

Activated Protein C resistance

Resistance to the anticoagulant effects of APC (activated protein C resistance, APCR) was initially reported by Dahlback, Carlsson & Svensson (1993) in a family with venous thrombosis. APCR was subsequently shown to be associated, in 90% or more of cases, with a mutation in the factor V gene (Arg506Gln) (Ber tina et al. 1994 ) which gives rise to a variant form of factor V, factor V Leiden (FV Leiden) ( Bertina et al. 1995 ; Dahlback 1995a; Dahlback 1995b; Hillarp et al. 1995 ; Dahlback 1997). FV Leiden has normal procoagulant activity, but is partially resistant to cleavage and inactivation by APC. This gives rise to a hypercoagulable state.

FV Leiden is recognized to be the commonest of the known inherited risk factors for VTE, with a high prevalence in populations of European origin ( Table 1). Based on published surveys, the mean FV Leiden allele frequency throughout Europe is 2.7% ( Rees 1996). Allele frequencies of 7.0% in Greece, 5.9% in Sweden, 3.4% in the UK, and 1.4% in Italy have been reported. The FV Leiden mutation has also been found in North India and Saudi Arabia, but it is absent in indigenous peoples from Asia, America and Australasia ( Rees 1996). Assuming a prevalence of 5–10% for FV Leiden in the general population, homozygosity is expected in 0.06–0.25% ( Dahlback 1995b). It has been suggested that the high prevalence of the FV Leiden mutation in affected populations may be the result of evolutionary selection. Heterozygous carriers of the mutation may have had a conferred survival advantage arising from a reduction in the risk of intrapartum bleeding ( Lindqvist et al. 1998 ).

The most commonly used phenotypic screening test for APCR is based on the prolongation of the APTT by the addition of APC ( Dahlback et al. 1993 ). Results are expressed as an APC ratio, the ratio between the APTTs measured in the presence and absence of added APC. APCR gives rise to a reduced APC ratio. The commercial availability of an APTT-based screening test for APCR ( Rosen et al. 1994 ) resulted in this method being introduced in most laboratories during the mid 1990s. There are a number of important considerations to be taken into account when using APTT-based methods to screen for APCR ( Dahlback 1995c). Different coagulometers give different clotting times, and APC ratios obtained using different instruments should not be directly compared ( Rosen et al. 1994 ). Due to the high prevalence of APCR in the general population, APC-ratio reference ranges should be derived using normal plasma samples excluding individuals with the FV Leiden mutation. Platelet contamination of plasma samples should be avoided as this will reduce the APC ratio (this is particularly the case with frozen and thawed samples). Separate reference ranges are required for fresh or frozen and thawed plasma samples since freezing and thawing may result in a reduced APC ratio.

A variety of factors may influence the APC ratio obtained using the standard APTT-based screening test. In the absence of the FV Leiden mutation, an acquired APCR phenotype may be observed during pregnancy ( Cumming et al. 1995 ), in association with the use of oral contraceptives ( Rosing et al. 1997 ), or in the presence of a lupus anticoagulant ( Halbmayer et al. 1994 ). Small children, with low levels of vitamin K-dependent proteins, may have high APC ratios. The standard screening test cannot be used if patients are being treated with vitamin K antagonists, or with heparin. Generally speaking, results obtained using the standard test may not be reliable unless the baseline APTT is within normal limits.

A modified APTT-based screening test for APCR is now in use in many laboratories. In this modified method, plasma is prediluted in factor V deficient plasma to normalize clotting factor concentrations. This method may be used in anti-coagulated patients ( Jorquera et al. 1994 ; Trossaert et al. 1994 ; Tosetto & Rodeghiero 1995), and it has been shown to allow reliable screening for pre-existing APCR during pregnancy ( Cumming et al. 1996 ). The presence of a lupus anticoagulant may still give rise to anomalous results, however ( Montaruli et al. 1997 ). The modified assay is highly sensitive and specific to the presence of the FV Leiden mutation ( Dahlback 1997). It may therefore be recommended as a suitable screening test for this mutation.

There are a number of other methods which may be used to screen for APCR associated with the FV Leiden mutation ( Tripodi et al. 1997 ), including assays which use activated factor X or Russell viper venom to trigger fibrin formation. The use of a dRVVT-based assay has recently been reported to allow good discrimination for the FV Leiden mutation in the presence of a lupus anticoagulant ( Galli et al. 1998 ). The suitability of these methods for screening purposes remains to be fully evaluated.

A variety of methods have been used to screen the factor V gene for the G1691 A Leiden mutation. These include PCR and restriction enzyme digestion ( Bertina et al. 1994 ), the use of sequence specific PCR primers ( Kirschbaum & Foster 1995; Cumming et al. 1997b ), PCR-mediated site-directed mutagenesis ( Rabes et al. 1995 ), PCR-single strand conformation polymorphism (SSCP) analysis ( Corral et al. 1996 ) and ELISA-based oligonucleotide-ligation assay ( Zotz et al. 1996 ).

Should phenotypic APCR screening tests, or genotypic analysis, or both, be used in the investigation of patients with VTE? Phenotypic screening is relatively inexpensive and easy to perform. The original assay gives some indication of the degree of resistance to APC. The modified assay, requiring the dilution of test plasma in factor V-deficient plasma, approaches 100% sensitivity and specificity for the FV Leiden mutation. It has also been reported to discriminate FV Leiden heterozygotes and homozygotes ( Tripodi et al. 1997 ). FV Leiden genotyping requires specialist molecular biology facilities and expertise, and should not be performed in the absence of these. Compared with phenotypic tests, however, genotyping may seem to be the more definitive method, and sample preparation and handling is not problematic. Neither phenotyping nor genotyping will, in isolation, identify the occasional individual with APCR in the absence of the FV Leiden mutation. A general strategy to screen for the FV Leiden mutation may be proposed on the basis of these observations. Given the high sensitivity and specificity of the modified APTT-based screening test, this method is suitable for use as a first-line screening test. Positive results should be genotyped to confirm heterozygosity or homozygosity for the FV Leiden mutation. Homozygosity is associated with a high risk of venous thrombosis ( Rosendaal et al. 1995 ) and accurate genotyping is essential for appropriate clinical management. Use of the standard APTT-based method may be considered to identify individuals with acquired APCR. Finally, comprehensive lupus anticoagulant screening is recommended in patients with APCR, determined by either standard or modified method, who prove to be FV Leiden negative.

Inherited hyperhomocysteinaemia

Hyperhomocysteinaemia is an established risk-factor for arterial disease, including stroke, myocardial infarction and peripheral arterial disease ( Clarke et al. 1991 ). In recent years it has been recognized that hyperhomocysteinaemia is also a risk factor for venous thrombosis ( Falcon et al. 1994 ; Den Heijer et al. 1995 ; Den Heijer et al. 1996 ).

Homocysteine is a non-protein forming sulphydryl amino acid, arising from the metabolic processing of the essential amino acid, methionine. The intracellular metabolism of homocysteine occurs via two alternative pathways, vitamin B12-dependent remethylation to methionine involving the enzyme methylenetetrahydrofolate reductase (MTHFR), or transulphuration to cysteine by the action of the vitamin B6-dependent enzyme cystathinine β-synthase (CBS) ( D’Angelo & Selhub 1997). Defects in the genes for either MTHFR or CBS are implicated in the development of hyperhomocysteinaemia. The most common cause of severe hyperhomocysteinaemia is homozygous deficiency of CBS, with a frequency in the general population of ≈ 1 : 300 000 ( De Stefano et al. 1996 ). Heterozygous CBS deficiency, which has a high frequency in the general population of 0.3% to 1.4%, is associated with mild or moderate hyperhomocysteinaemia. On rare occasions severe hyperhomocysteinaemia is associated with inherited defects of the remethylation pathway, most frequently homozygous deficiency of MTHFR.

Frosst et al. (1995) reported a common C677T substitution in the MTHFR gene. Homozygosity for this genetic variation gave rise to a thermolabile variant of MTHFR which was associated with mildly elevated levels of plasma homocysteine. Recent reports have indicated that mild hyperhomocysteinaemia due to the presence of the MTHFR C677T variant may arise when plasma folate status is low ( Jacques et al. 1996 ; Ma et al. 1996 ; Girelli et al. 1998 ). Homozygosity for MTHFR C677T does not appear to predispose to hyperhomocysteinaemia when folate status is adequate.

Homozygosity for the C677T variant of the MTHFR gene (677TT) has been reported to be a risk factor for VTE ( Arruda et al. 1997 ). Other studies have not found an increased prevalence of the 677TT genotype among patients with thrombosis, compared with healthy controls ( Salden et al. 1997 ; Tosetto et al. 1997 ; Girelli et al. 1998 ; Kluijtmans et al. 1998 ). A consensus seems to be emerging that homozygosity for the MTHFR C677T variant is not a significant risk factor for VTE, per se.

There is currently no strong argument to include MTHFR C677T genotyping during routine thrombophilia investigations. The question of whether or not homocysteine levels should be measured as part of routine thrombophilia screening remains to be resolved.

Elevated factor VIII levels

Koster et al. (1995a) investigated ABO blood group, von Willebrand factor (VWF) and factor VIII (FVIII) as putative risk factors for deep-vein thrombosis. In univariate analysis, blood group, VWF and FVIII levels were all related to deep-vein thrombosis. In multivariate analysis, however, only FVIII remained as an independent risk factor. This study provided clear evidence that elevated FVIII is an important risk factor for the development of venous thrombosis. Compared with subjects with FVIII levels of 100 IU/dL or less, individuals with levels of greater than 150 IU/dL had a relative risk for thrombosis of 4.8 ( Table 1). This high risk stratum included 25% of patients with first episode deep-vein thrombosis and 11% of matched healthy controls. These findings were confirmed by O’Donnell et al. (1997) , who found elevated FVIII to be the most common abnormality in 260 patients referred for thrombophilia screening (including AT, PC, PS, APCR, FV Leiden, lupus anticoagulant) because of unexplained thromboembolism.

The determinants of elevated FVIII levels are unclear. High levels were not associated with the acute phase reaction in the majority of a group of patients with venous thrombosis ( O’Donnell et al. 1997 ). Using a new test, the familial aggregation test, Kamphuisen et al. (1998) observed that FVIII and VWF levels within families were positively correlated. This suggested a genetic influence. Furthermore, FVIII levels were positively correlated in sister-pairs but not in mother-daughter pairs. Sisters inherit the same paternal X-chromosome and, on average, 50% of sister-pairs will have the same combination of X-chromosomes. This observation therefore was suggestive of a X-linked determinant, possibly the FVIII gene itself.

To summarize, recent studies have shown that a raised FVIII level is an independent risk factor for the development of venous thrombosis. It is likely that there is a genetic contribution to raised levels, and serious consideration should be given to including FVIII assay as part of routine thrombophilia screening. CRP levels should be measured to exclude increased FVIII as an acute-phase reactant.

Prothrombin gene G20210 A variant

Human plasma prothrombin is a single chain vitamin K-dependent glycoprotein with a molecular mass of 72 kDa. During coagulation, prothrombin is converted to the serine protease thrombin by the action of the prothrombinase complex. The prothrombin gene is located on chromosome 11, including 14 exons within 21 kb of genomic DNA ( Degen & Davie 1987).

Towards the end of 1996 a newly identified genetic risk factor for venous thrombosis was reported ( Poort et al. 1996 ). Of 28 selected patients with a familial history of VTE, five (18%) were shown to carry a G to A nucleotide transition at position 20210 in the 3′-untranslated (UT) region of the prothrombin gene. This compared with an incidence of 1% in a group of 100 healthy control subjects. Subsequent reports of the prevalence of the prothrombin gene G20210 A variant in healthy control subjects have varied from 0.7 to 4.0% ( Arruda et al. 1997 ; Brown et al. 1997 ; Cumming et al. 1997a ; Hillarp et al. 1997 ; Rosendaal et al. 1998 ). An overall prevalence of 2.0% was estimated ( Rosendaal et al. 1998 ), with a higher prevalence in southern Europe (3.0%) than in northern Europe (1.7%). The prothrombin variant appears to be very rare in individuals of Asian and African descent, and a recent report suggests a single genetic origin that probably occurred after the divergence of Africans from non-Africans and of Caucasoid from Mongoloid subpopulations ( Zivelin et al. 1998 ).

The prothrombin gene 20210 AG genotype is associated with a higher prothrombin level than the 20210 GG genotype, and elevated prothrombin is itself a risk factor for VTE ( Poort et al. 1996 ). The prothrombotic tendency linked with the 20210 A allele is likely therefore to be mediated through an increase in prothrombin levels. The 5′-and 3′-UT regions of genes are associated with regulation of gene expression, and it may be hypothesized that the G20210 A nucleotide transition in the 3′-UT region of the prothrombin gene could lead to increased expression. There is, however, no direct evidence to support this ( Poort et al. 1996 ). Alternatively, and perhaps more likely, the 20210 A allele may be in linkage disequilibrium with another unknown sequence variation which is itself directly linked to the increased risk for thrombosis.

Even if linkage disequilibrium is the case, the 20210 A allele remains a valid marker for increased thrombotic risk. It may be readily detected by PCR, most commonly using a mutagenic primer designed to introduce a restriction enzyme site in the presence of the variant A allele. In isolation, however, the G20210 A genotype appears not to be a major risk factor for VTE ( Table 1), and routine screening as part of thrombophilia investigations may not be warranted.

The multifactorial nature of thrombophilia

It has been recognized for some years that family members with apparently identical thrombotic-risk genotypes may exhibit markedly different clinical manifestations. This heterogeneity of presentation was explainable in some cases by the additional presence of a circumstantial risk factor, such as pregnancy, surgery, or oral contraceptive use, but in most instances no convincing explanation could be provided.

Heterozygosity for a single genetic risk factor does not usually confer a major increase in the risk of thrombosis ( Table 1). Recently the multifactorial nature of thrombophilia on a genetic level has become apparent, and it is now recognized that more than one genetic risk factor may cosegregate. This influences the thrombophilic phenotype and has important implications for clinical management. For example, an increased risk for venous thrombosis has been reported for the coinheritance of FV Leiden with AT deficiency ( Van Boven et al. 1996 ), PC deficiency ( Koeleman et al. 1994 ; Gandrille et al. 1995 ; Hallam et al. 1995 ; Brenner et al. 1996 ), PS deficiency ( Koeleman et al. 1995 ; Zoller et al. 1995 ; Beauchamp et al. 1996 ), the prothrombin gene G20210 A variant ( Makris et al. 1997 ; Ehrenforth et al. 1998 ; Zoller et al. 1998 ), or hereditary homocystinuria ( Mandel et al. 1996 ).

It has been clearly shown that individuals with two or more inherited thrombophilia risk factors have a high risk of thrombosis. Similarly, subjects with dual inherited defects and a circumstantial risk factor are at very high risk of thrombosis. Such patients require appropriate clinical management and need to be recognized. Their identification is a major justification for appropriately targeted laboratory thrombophilia screening.

Clinical management of inherited thrombophilia

The majority of thrombotic events occurring in adults in association with inherited thrombophilia are venous, although arterial thrombosis is described in patients with inherited defects of homocysteine metabolism ( De Stefano et al. 1996 ). The annual rate of DVT in the general population has been estimated between 48 and 162 per 100 000 and the rate of pulmonary embolism has been estimated at 23–51 per 100 000 ( Geerts & Jay 1996). Rates of both DVT and pulmonary embolism rise with age. Rosendaal recently presented data from the Netherlands showing the incidence of venous thromboembolic disease rising from 0.6 per 100 000 per year in children less than 14 to 74.2 per 100 000 per year in adults aged 40–54 ( Rosendaal 1997).

Pulmonary embolism has been described as the commonest cause of preventable death in hospital ( Morrell & Dunnill 1968). Even in the absence of pulmonary embolism the long-term sequelae of deep vein thrombosis are not insignificant with 67% of patients in one study suffering long-term pain or swelling of the leg following a DVT ( Strandness et al. 1983 ).

Although the first description of inherited thrombophilia was in 1965, deficiencies of AT, PC and PS are all relatively rare ( Table 1). Testing for these risk factors was directed at affected individuals who had presented at a young age with either very atypical thrombosis, recurrent thrombosis or thrombosis in the context of a strong family history. A retrospective analysis by De Stefano et al. (1994) confirmed that the first thrombotic event occurred below the age of 40 in 80% of 238 individuals affected by deficiency of one of those three factors. The same study also concluded that a policy based on knowledge of diagnosis and on implementation of antithrombotic treatment during risk situations appeared to modify the clinical outcome even in the absence of long-term antithrombotic prophylaxis.

The clinical approach to VTE patients has been modified in recent years following the description of the FV Leiden mutation, and to a lesser degree since the identification of the prothrombin gene G20210 A variant. The FV Leiden mutation is associated with a sevenfold increased risk of venous thrombosis in heterozygous carriers, and an 80-fold increased risk in homozygous carriers ( Rosendaal et al. 1995 ). It is found in 20–50% of subjects with VTE ( Bertina et al. 1994 ; Svensson & Dahlback 1994), depending on patient selection criteria. Likewise, case-control studies have found heterozygosity for the prothrombin gene variant in 4–7% of individuals with at least one confirmed episode of VTE ( Poort et al. 1996 ; Arruda et al. 1997 ; Brown et al. 1997 ; Cumming et al. 1997a ; Hillarp et al. 1997 ). Increases in relative risk for venous thrombosis of 2.0- to 6.6-fold were reported. An increased risk for arterial disease has also been recognized in association with the prothrombin gene variant ( Arruda et al. 1997 ; Rosendaal et al. 1997 ). The relative risk for thrombosis in individuals homozygous for the 20210 A allele is not known.

The FV Leiden mutation and the prothrombin gene variant are both recognized to be independent risk factors for VTE. The prevalence of each ( Table 1) is such that there is the potential for enormous numbers of individuals to present for screening. Testing for thrombophilia risk factors is no longer confined to small numbers of laboratories with a specialist interest in thrombosis. There is a need for all hospitals to identify appropriate guidelines for thrombophilia screening that targets patients most at risk without resources being channelled into unnecessary, expensive investigations.

Thrombophilia screening in women using oral contraceptives

Oral contraceptive users were targeted early for investigation of the applicability of thrombophilia screening of selected populations. This group of women consists of large numbers of easily identifiable individuals who have a measurable increased risk of venous thrombosis. Vandenbrouke e t al. (1994) estimated that the incidence of DVT in women not using oral contraceptives was 0.8 per 10 000 women years. This increases to 3 per 10 000 women years among women using oral contraceptives, and to 5.7 per 10 000 women not using the oral contraceptive who are carriers of the FV Leiden mutation. The incidence of DVT among women using the oral contraceptive who are carriers of the FV Leiden mutation is 28.5 per 10 000.

The absolute thrombotic risk for women with mutations of AT, PC or PS who take oral contraceptives is more difficult to calculate as numbers are relatively small. One study (Pabinger e t al. 1994b) demonstrated a convincing increased risk of thrombosis for women with deficiency of AT who were taking the oral contraceptive, with an incidence of thrombosis per patient year of 27.5% against a control incidence of 3.4%. The same study showed a marginal effect for women with PC defects who took oral contraceptives, with an incidence per patient year of 12% against a control of 6.9%. Of 34 patients with PS deficiency studied, during the time of the study 6/17 women taking oral contraceptives had a thrombotic event, and 5/17 women with PS deficiency had a thrombotic event without taking oral contraceptives. This study showed no effect in individuals with PS deficiency but the data should be interpreted with caution as the numbers were small and the background risk was apparently high.

Although there is an undoubted increased risk of thromboembolic disease for users of the oral contraceptive who are also carriers of an inherited thrombophilia mutation, the overall risk to this particular group is relatively small because of their young age. Several authors have concluded that screening all women taking the combined oral contraceptive cannot be justified. Vandenbroucke et al. (1996) estimated that somewhere between 400 000 and 1 million oral contraceptive users would have to be screened in order to prevent one death from pulmonary embolism. Furthermore, if all women identified to carry the FV Leiden mutation were denied the oral contraceptive then somewhere between 80 000 and 200 000 women may be denied effective contraception in order to prevent one fatal pulmonary embolism. Rosendaal (1996) estimated that 2 250 000 women would have to be screened to find these affected women. There is then the risk that some of these women who have been identified and have stopped using the oral contraceptive would become pregnant with an even greater risk of thromboembolic disease.

To complicate the issue further it has become apparent that there is a different risk of venous thrombosis between second generation and third generation combined oral contraceptives, with a higher risk associated with the newer third generation oral contraceptives ( WHO 1995; Spitzer et al. 1996 ). This clinical observation has been supported by a laboratory study demonstrating differences in the effects of APC upon thrombin generation between women using second or third generation monophasic combined oral contraceptives ( Rosing et al. 1997 ). The effect of third generation contraceptives cannot be explained by selective prescribing of these to women with a higher perceived risk of thrombosis ( Andersen et al. 1998 ). It is independently increased for both women with and without the FV Leiden mutation. Bloemenkamp et al. (1995) showed that the highest risk in their study was with oral contraceptives containing desogestrel. The relative risk of venous thrombosis for women taking a desogestrel containing oral contraceptive was shown to be 9.2 for women without FV Leiden. The risk for carriers of FV Leiden using a desogestrel containing oral contraceptive was increased almost 50-fold.

There are strong epidemiological arguments for not undertaking mass thrombophilia screening prior to prescribing oral contraceptives, but what should a clinician do when faced with a concerned individual patient in a clinic? If a woman has already had a DVT then she has an increased risk of further thrombosis, whether or not a thrombophilia screen identifies a defect ( Carter 1994). One study ( Badaracco & Vessey 1974), published before the introduction of thrombophilia screening, demonstrated that the risk of recurrence of venous thrombosis was lower in a group of women who had stopped taking oral contraceptives compared with a group who had never had them. This study would lend support to long-standing clinical guidelines suggesting that women who have had a thromboembolic event should avoid combined oral contraceptives, even if a known thrombophilia risk factor is not identified. The reason for offering such women thrombophilia screening is to attempt to identify the degree of associated risk. It may not alter patient management at the time but may be important for the woman in the future.

If, on the other hand, a woman presents with a family history of venous thrombosis affecting first degree relatives, rather than a personal history of DVT, then thrombophilia screening may add information which will influence the treatment decision. It is important in this situation to attempt to obtain blood from one of the affected relatives to see whether there is any identifiable risk factor.

When all information is available then doctor and patient are in a position to discuss relative risks. As well as wanting effective contraception, many women find that symptoms of either menorrhagia or dysmenorrhoea are greatly improved with the oral contraceptive. They may be using the combined oral contraceptive to limit unwanted, often very unpleasant, associated symptoms of menstruation and may be willing to accept a small risk of thrombosis in this situation. Heterozygosity for FV Leiden or the prothrombin gene G20210 A variant alone should not be regarded as an absolute contraindication to the use of oral contraceptives but women should be given the opportunity to make an informed choice. It should be borne in mind that the risk is for all forms of venous thrombosis and not just DVT. Two recently published studies ( De Bruijn et al. 1998 ; Martinelli et al. 1998 ) have reported a high relative risk of cerebral sinus thrombosis in women using an oral contraceptive who also carry a hereditary prothrombotic condition.

Homozygosity for the FV Leiden mutation presents a much higher risk of venous thrombosis, as does a defect of AT or a combined defect. Individuals who are homozygous for the prothrombin gene G20210 A variant may likewise be at high risk. In these situations the oral contraceptive should be avoided. The risk with defects of PC or PS is less certain and a decision would have to be made taking into account a woman’s personal and family history of thrombosis. The effect of age should be remembered in the discussion as risks to a young woman of 17 are considerably less than those for a woman between the ages of 35 and 40.

Thrombophilia screening in women using hormone replacement therapy

Earlier reports suggested that there was no increased risk of venous thrombosis for women taking hormone replacement therapy after the menopause. More recent epidemiological studies have demonstrated a small increased risk of venous thrombosis, particularly in the first year of treatment ( Barlow 1997). The relative risk of venous thrombosis is reported as between two and three times that of women who are not taking hormone replacement. It is not yet known how much of this increased risk of thrombosis is associated with inherited thrombophilia. Until more information is available clinical guidelines should follow similar lines to these offered to women with inherited thrombophilia who wish to use oral contraception: homozygosity for FV Leiden or the prothrombin gene G20210 A variant, or a combined defect would be regarded as high risk. The risk of an AT mutation in association with hormone replacement therapy is unknown but in view of the high risk in association with the oral contraceptive and with pregnancy it would be safer to assume that there is an increased risk with HRT as well. The risks for heterozygous carriers of the FV Leiden mutation or the prothrombin gene variant are likely to be relatively small. The risks for women with mutations of PC or PS are unknown.

Before a decision is made about treatment, it is important for the clinician to identify the degree of risk to the patient and to weigh this up against the benefit of the hormone replacement therapy, particularly with regard to the effects on bone turnover and the reduced risk of coronary artery disease,

Thrombophilia and pregnancy

Venous thromboembolism is the commonest cause of maternal death in many western countries ( Barbour 1997). Pregnant women have a fivefold increase in venous thrombosis. Earlier studies suggested that the risk of thrombosis was greater in the third trimester and the puerperium. This may have reflected medical management at the time, with a greater emphasis on bed rest, as more contemporary studies suggest a high incidence of DVT in the first and second trimesters as well as in late pregnancy ( Carter 1994; Barbour 1997). There is still evidence of an increased incidence of pulmonary embolism in the puerperium, particularly in association with Caesarean section ( Barbour 1997). This data is slightly at odds with two published studies describing the incidence of thrombosis in pregnancy in women identified to have an inherited thrombophilia risk factor ( Conard et al. 1990 ; De Stefano et al. 1994 ). In both of these studies there appeared to be a significant increase in risk in the puerperium. Both of these studies suffer from being small and retrospective, with the possibility that patients were only identified at the time of their thrombosis in pregnancy. This apparent contradiction is an important one for clinicians to resolve and needs further study. If there is a significant risk of DVT in the first trimester then women with a known thrombophilia risk factor have to be considered for anticoagulation throughout pregnancy. On the other hand, if the risk is concentrated in the post-natal period then anticoagulation could be deferred until after delivery. Until the situation becomes clearer, each case should be considered on an individual basis.

The arguments against routine screening of women for inherited thrombophilia risk factors in pregnancy are similar to those in women taking oral contraception. McColl et al. (1997) calculated that the thrombotic risk for a woman with a FV Leiden mutation was around 1 in 400–500 during pregnancy and the puerperium. If women were screened, and treated with anticoagulants if found to be positive for the FVL mutation, then more than 10 000 women would require screening and between 400 and 500 would be treated with anticoagulants to prevent one thrombotic event.

Increased fetal loss in association with thrombophilia

A study reported in 1996 ( Preston et al. 1996 ) examined 571 women with a thrombophilia defect who had had 1524 pregnancies and compared them with 359 control women with 1019 pregnancies. They reported an increased risk of fetal loss in women with thrombophilia with an overall odds ratio of 1.35. The risk was highest for women with a combined defect, but was also demonstrable for women with AT, PS or PC deficiency. In that study the authors did not demonstrate an increased risk of fetal loss in association with FV Leiden, although subgroup analysis suggested a slight increased risk of stillbirths, but not miscarriages. Sanson et al. (1996) had also demonstrated an increased risk of miscarriage or stillbirth in a group of 60 women with a deficiency of either AT, PC or PS. In this study the overall relative risk was 2.0.

Rai et al. (1996) investigated a group of women with a history of recurrent miscarriage and found an association between second trimester pregnancy loss and APCR that was not demonstrable in women with recurrent first trimester fetal loss. Two studies which examined selected women with a poor obstetric history ( Brenner et al. 1997 ; Rotmensch et al. 1997 ) demonstrated an association between the FV Leiden mutation and first and second trimester fetal loss, and intrauterine death. Risk of fetal loss was also described in association with acquired APCR, raising the possibility that there is a subgroup of women with APCR who have an increased risk of early fetal loss, perhaps due to an associated, as yet unidentified, risk factor.

Data for other risk factors are limited but one recent study also measured homocysteine levels and demonstrated an association between hyperhomocysteinaemia and complications of pregnancy ( De Vries et al. 1997 ). Furthermore, despite normal levels of vitamin B6, B12 and folate, 6 weeks of supplementation with B6 and folate resulted in a decrease in homocysteine levels.

Screening asymptomatic relatives

Asymptomatic relatives of individuals who have had a thrombotic event and been shown to have an inherited risk factor will present to doctors for advice and possible screening. If the affected individual is a first degree relative of the patient it is difficult not to proceed to screening. It is important, however, that the patient is counselled prior to any testing so that they have an understanding of what information is being sought and what the implications are for them and their family. Screening an asymptomatic relative must not result in undue anxiety because a risk factor is identified. Should screening be limited to the factor known to be deficient in the affected relative? Although this is rational and is less expensive and time consuming it is difficult to reassure a patient completely if a full thrombophilia screen has not been done. If limited testing has been done the patient should be informed of this.

Thrombophilia screening in children

Venous and arterial thrombosis does occur in children although it is relatively rare. The majority of events occur in neonates and are frequently associated with an in-dwelling cannula. In older children thrombosis may be associated with malignancy, sepsis or trauma as well as venous catheters. Small studies published have shown that even in children with demonstrable risk factors for thrombosis there is an increased prevalence of inherited thrombophilia and this should be sought in children presenting with thrombosis, even where there is another cause demonstrated ( Manco-Johnson 1997 ; Sutor & Uhl 1997; Uttenreuther-Fischer et al. 1997 ). If a laboratory does propose to offer thrombophilia screening for children it is important that paediatric reference ranges are established as these may vary from adult reference ranges ( Andrew et al. 1992 ; Nowak-Gotti et al. 1994 ; Brandt et al. 1998 ). As the risk of spontaneous thrombosis in children is very small, testing of unaffected children who are relatives of a proband presenting with thrombosis and inherited thrombophilia should be deferred until the child is old enough to give informed consent.

The timing of thrombophilia testing

Patients in whom thrombophilia screening is deemed appropriate can be identified at the time of presentation. Whereas DNA based tests can be undertaken at any time it is better to avoid coagulation-based assays at the time of an acute event ( Lane et al. 1996 ; Kennedy et al. 1995 ). If a patient is to have anticoagulants for a finite time then thrombophilia screening can be deferred until at least 2 weeks after their treatment has been completed. Patients who are to remain on anticoagulants indefinitely can have AT and FVIII measured while on warfarin.

Vitamin K antagonists reduce PC and PS assay results and the diagnosis of hereditary deficiency states is not reliable in patients who are receiving oral anticoagulants. To normalize for the effect of oral anticoagulants, ratios of PC or PS antigen to either FX antigen, prothrombin antigen or factor VII antigen have been used. The diagnostic accuracy of such assay modifications has not been fully validated ( De Stefano et al. 1996 ), and this approach is valid only when the patient to be investigated is stably anticoagulated at the time of investigation. Unless clinical circumstances demand, screening for PC and PS deficiency should be delayed until 14 days after the completion of oral anticoagulant therapy. If it is deemed necessary to test for these deficiencies during the course of oral anticoagulation, unfractionated or low molecular weight heparin may be used to replace oral anticoagulants for a 14-day period. This will allow vitamin K-dependent proteins to return to baseline levels. A blood sample for analysis of PC and PS may be collected before heparin administration on the morning of the 14th day ( Lane et al. 1996 ), after which the patient may be re-established on warfarin. Another approach is to measure the levels of inherited thrombophilia factors in the parents of the patient on anticoagulants, when they are both available.

Accepted for publication 28 October 1998