Thrombophilia can be identified in about half of all patients presenting with venous thrombosis, and appears to provide at least a partial explanation for a previously poorly explained disease (Weitz et al, 2004). Over the past decades, testing has increased tremendously for various indications (Coppens et al, 2007), but whether the results of such tests aid the clinical management of patients has not been settled. Here, we review the most commonly tested thrombophilic abnormalities, i.e. protein C, protein S, and antithrombin deficiencies, F5 R506Q (factor V Leiden) and F2 G20210A (prothrombin 20210A), and elevated levels of coagulation factor VIII, and their associations with venous and arterial thrombosis. Furthermore, we discuss whether and how this might help in the clinical management of patients.
Thrombophilia can be identified in about half of all patients presenting with venous thrombosis. Testing has increased tremendously for various indications, but whether the results of such tests help in the clinical management of patients has not been settled. Here, we review the most commonly tested thrombophilic abnormalities, i.e. protein C, protein S, and antithrombin deficiencies, the F5 R506Q (factor V Leiden) and F2 G20210A (prothrombin G20210A) mutations, and elevated levels of coagulation factor VIII, and their association with venous and arterial thrombosis as well as pregnancy complications. We conclude that testing for hereditary thrombophilia generally does not alter the clinical management of patients with venous or arterial thrombosis or pregnancy complications. Because testing for thrombophilia only serves limited purpose this should not be performed on a routine basis.
Definition of thrombophilia
Virchow (1856) proposed changes in the blood coagulability as one of the mechanisms that predispose to thrombosis. These changes in the blood coagulability, i.e. thrombophilia, indicate the presence of a hypercoagulable state leading to a thrombotic tendency. Important risk factors for thrombotic disease were mainly identified in studies comprising of families with a high incidence of thrombotic disease. However, acquired factors may also lead to a thrombotic tendency. Thrombophilia may therefore be defined as both an acquired or congenital abnormality of haemostasis predisposing to thrombosis (Table I). Several well-known factors predisposing to thrombophilia are described in more detail below.
|Antithrombin deficiency||Protein C deficiency||Protein S deficiency||F5 R506Q mutation||F2 G20210A mutation||Elevated levels of FVIII*||Lupus anticoagulant*||Anticardiolipin antibodies*||Anti-β2 GPI antibodies||Anti-prothrombin antibodies|
|Prevalence in the general population||0·02%||0·2%||0·03–0·13%||3–7%||0·7–4%||By definition chosen cut-off level||1–8%||5||3·4||14·6|
|Relative risk for a first venous thrombosis||5–10||4–6·5||1–10||3–5||2–3||5||3–10||0·7||2·4||1·4|
|Relative risk for recurrent venous thrombosis||1·9–2·6||1·4–1·8||1·0–1·4||1·4||1·4||1·3–6·7||2–6||1–6||–||–|
|Relative risk for arterial thrombosis||No association||No consistent association||No consistent association||1·3||0·9||3·1||10||1·5–10||–||–|
|Relative risk for pregnancy complications||1·3–3·6||1·3–3·6||1·3–3·6||1·0–2·6||0·9–1·3||1·1–1·2||No consistent data||No consistent data|
Antithrombin is an important natural anticoagulant of the coagulation system and inhibits the coagulation factors IIa, IXa, Xa, XIa, and XIIa. The inhibition, which occurs by the formation of covalent complexes, is accelerated 1000-fold by heparin. Homozygosity for antithrombin deficiency is very rare. Two types of antithrombin deficiency are defined (Lane et al, 1993, 1997). Type I, i.e. the classical and most common deficiency, is a quantitative deficiency with antithrombin plasma levels below 50% of normal. In type II antithrombin deficiency, the plasma levels are within the normal range but functional activity is impaired due to the production of a variant protein. Currently more than 80 different mutations are known in the antithrombin gene (SERPINC1, previously known as AT3) (Lane et al, 1997). The prevalence of antithrombin deficiency in the general population is very low and estimated to be approximately 0·02% in the general Caucasian population (Tait et al, 1994).
Protein C deficiency
Protein C is a vitamin K-dependent glycoprotein that is synthesized in an inactive form in the liver. Upon activation, activated protein C (APC) is an important natural anticoagulant that, together with its co-factor protein S, suppresses thrombin generation by inhibition of coagulation factors Va and VIIIa (Esmon, 2000).
Protein C is activated by thrombin bound to thrombomodulin on cell surfaces. Binding to the endothelial cell protein C receptor (EPCR), a type I transmembrane protein that is highly expressed on the endothelium of large blood vessels, enhances protein C activation more than 10-fold (Stearns-Kurosawa et al, 1996; Taylor et al, 2001). After being released from EPCR, APC is fully activated.
A deficiency in protein C is determined by measurement of plasma levels of protein C activity and antigen. There are two types of protein C deficiency. Type I deficiency is associated with a reduction in protein C activity as well as antigen levels (quantitative protein C deficiency). This is the most common type of protein C deficiency and is most likely to be the result of a reduced synthesis or stability of protein C. In type II deficiency the protein C activity is more reduced than the antigen levels due to the synthesis of abnormal protein C molecules (qualitative protein C deficiency).
Protein C deficiency can be caused by numerous loss-of-function mutations in the protein C gene (PROC). More than 160 mutations in PROC are currently known to cause protein C deficiency, indicating that it is a very heterogeneous disorder (Reitsma et al, 1995). Protein C deficiency is very rare with a prevalence of approximately 0·2% in the general population (Miletich et al, 1987; Koster et al, 1995a).
Protein S deficiency
Protein S is the co-factor of protein C in the inactivation of factors Va and VIIIa. It circulates in a free (∼40%) form or bound to the acute phase C4b-binding protein (∼60%). Bound protein S has no protein C co-factor activity (Dahlback, 1986). Additional to the co-factor activity of protein S, it can inhibit the prothrombinase and tenase complexes independently (Hackeng et al, 1994). Three subtypes of protein S deficiency have been described. In type I deficiency, both the levels of total and free protein S are decreased. In type II, the co-factor activity of protein S is decreased but with normal levels of total and free protein S antigen levels. In type III deficiency (free protein S deficiency), free protein S levels are decreased but with normal levels of total protein S. Thus, type I and III are quantitative protein S deficiencies, whereas type II is a qualitative deficiency. Type II deficiency is very rare and difficult to diagnose. Also for protein S deficiency over 130 genetic mutations in the gene encoding protein S (PROS1) have been described (Gandrille et al, 1997, 2000). The prevalence of protein S deficiency is very low, ranging from 0·03% to 0·13% in the general Caucasian population (Dykes et al, 2001).
For all three natural anticoagulant factors environmental factors may lead to an acquired deficiency. Severe liver disease compromises its capacity to synthesize proteins, and will subsequently reduce the levels of many coagulation factors including antithrombin, protein C, and protein S. Severe vitamin K deficiency, most often intentionally induced by the use of vitamin K antagonists, leads to acquired deficiencies of protein C and protein S (and other vitamin K-dependent coagulation factors, i.e. II, VII, IX and X), a combination that is not expected to be of genetic origin given the rarity of both hereditary deficiencies. A decrease in protein S activity is caused by estrogen excess, such as during pregnancy or use of oral contraceptives (Tans et al, 2000).
F5 R506Q – APC resistance
In contrast with the loss-of-function mutations in PROC, PROS1 and SERPINC1, the gain-of-function mutations in the procoagulant factors of the coagulation system are more homogeneous and more prevalent in the general population.
In 1993, Dahlbäck et al reported that a poor anticoagulant response to activated protein C (APC) was associated with the risk of thrombosis (Dahlbäck et al, 1993). The so-called APC resistance is mainly caused by a mutation in the Arg506 cleavage site of factor Va (Bertina et al, 1994; Voorberg et al, 1994). The activated mutant factor V (F5 R506Q), commonly called factor V Leiden, is inactivated more slowly by activated protein C than wildtype factor Va (Heeb et al, 1995; Kalafatis et al, 1995; Nicolaes et al, 1995; Aparicio & Dahlbäck, 1996). The F5 R506Q mutation is the most prevalent thrombotic risk factor known in the Caucasian population, i.e. 3–7% carry the mutation, but it is very rare in native African and Asian populations (Rees, 1996).
Resistance to APC, although in the majority of cases caused by the F5 R506Q mutation, can also occur in the absence of this mutation. This APC resistance may be caused by currently unidentified factors or by environmental factors. Acquired APC resistance is caused by changes in hormonal status occurring during pregnancy (Cumming et al, 1995), or the use of female hormones, i.e. oral contraceptives (Henkens et al, 1995; Olivieri et al, 1995; Meinardi et al, 1997; Rosing et al, 1997) or hormone replacement therapy (Curvers et al, 2002).
Prothrombin 20210A mutation
Poort et al (1996) described a mutation in the prothrombin gene (F2), i.e. the F2 G20210A mutation, which is associated with an increased level of prothrombin in the circulation. Prothrombin is the precursor of the serine protease thrombin, which is a key enzyme in the blood coagulation. Although not as common as the F5 R506Q mutation, the prevalence of the F2 G20210A mutation is high in the general population with estimates ranging from 0·7% to 4% in Caucasian populations (Rosendaal et al, 1998).
Antiphospholipid syndrome is a non-inflammatory auto-immune disease characterised by thrombosis or pregnancy complications in the presence of antiphospholipid antibodies (Urbanus et al, 2008). Preliminary criteria for the diagnosis of definite antiphospholipid syndrome were formulated at an international consensus meeting in 1999 and updated in 2004 (Wilson et al, 1999; Miyakis et al, 2006). Clinical criteria include having one or more clinical episode of thrombosis, one or more unexplained fetal deaths (>10 weeks of gestation), or having three or more unexplained consecutive miscarriages (<10 weeks of gestation). Laboratory criteria include lupus anticoagulant present in plasma, or medium or high titers of anticardiolipin antibody of IgG or IgM isotype in serum or plasma, or anti-β2 glycoprotein-I antibody of IgG or IgM in serum or plasma. Antiphospholipid syndrome is diagnosed if at least one of the clinical criteria and one of the laboratory criteria is met. To prevent the detection of transiently present antiphospholipid antibodies, laboratory tests should be performed twice, twelve weeks apart, and should be positive on both occasions. Since the clinical criteria as described above are prevalent in the general population, the diagnosis of antiphospholipid syndrome is largely based on laboratory tests. Since data are limited, the prevalence of persistent lupus anticoagulant or antibodies against phospholipid in the general population is not well known. Although some population-based studies have estimated the prevalence of one or more positive tests, in most studies these were only assessed once (Ginsburg et al, 1992; Runchey et al, 2002; de Groot et al, 2005; Naess et al, 2006).
High levels of factors VIII
Elevated levels of coagulation factor VIII, but also of factors IX and XI lead to a hypercoagulable state (Koster et al, 1995b; van Hylckama Vlieg et al, 2000; Meijers et al, 2000). Although a heritable component has been described for these clotting factors, currently no polymorphisms have been discovered that can account for such elevated levels (Kamphuisen et al, 2000; Vossen et al, 2004; Bank et al, 2005). In most laboratories, only factor VIII levels are included in the thrombophilia test panel.
Association between thrombophilia and a first deep venous thrombosis
Initially, interest was mainly focused on the natural anticoagulants of the coagulation system. A deficiency in one of the natural anticoagulants that lead to an increased risk of venous thrombosis, i.e. antithrombin deficiency, was initially described by Egeberg (1965). It was reported that an inheritable deficiency in antithrombin led to lowered blood levels of antithrombin, which could cause a severe tendency to thrombosis in a family with a high incidence of thrombotic diseases. Individuals with antithrombin deficiency have historically been regarded to be at a very high risk of thrombosis, particularly in females during pregnancy. However, this perception is mainly based on small studies in which highly selected thrombophilic individuals were described (Hirsh et al, 1989; Conard et al, 1990). Several studies in families with at least one proband with venous thrombosis and antithrombin deficiency have assessed the risk of a first episode of venous thrombosis in adult antithrombin deficient relatives. The incidence of first venous thrombosis in retrospective and prospective studies ranged between 0·4% and 1·7% per year (Sanson et al, 1999; Simioni et al, 1999; Vossen et al, 2005). In contrast to the relative risk for thrombosis of approximately 8–10 in deficient relatives as compared to those with normal antithrombin levels, a large population based case–control study found that antithrombin deficiency (measured as plasma levels <80 U/dl on two occasions) was associated with a five-fold (95% CI 0·7–34) increased risk of a first deep venous thrombotic event (Koster et al, 1995a).
The first reports of an increased risk of venous thrombosis associated with protein C deficiency appeared in 1981 (Griffin et al, 1981; Bertina et al, 1982; Comp et al, 1984). The prevalence of protein C deficiency is 2·5–6% in patients with a first episode of deep venous thrombosis (Heijboer et al, 1990; Pabinger et al, 1992). Heterozygous protein C deficiency is associated with a 4- to 6·5-fold increased risk of venous thrombosis [odds ratio (OR) 3·8, 95% confidence interval (CI) 1·3–10 if based on repeated measurements of protein C; OR 6·5, 95% CI 1·8–24 if based on DNA diagnosis] (Koster et al, 1995a). Homozygous protein C deficiency results in severe thrombotic complications of the foetus or at very early age, i.e. purpura fulminans (Branson et al, 1983; Marlar et al, 1989). The prevalence of homozygous protein C deficiency is estimated to be one in every 160 000–360 000 births (Miletich et al, 1987).
The risk of venous thrombosis associated with protein S deficiency was first described by Comp et al (1984). Conflicting results have been published on whether protein S deficiency is associated with an increased risk of venous thrombosis. In the Leiden Thrombophilia study, a large population-based case–control study on risk factors for venous thrombosis, no increased risk could be demonstrated (Koster et al, 1995a). However, family studies did show a strongly increased thrombotic risk associated with protein S deficiency [relative risk (RR) 17, 95% CI 7–45] (Simioni et al, 1999). Similarly to protein C deficiency, a homozygous state of protein S deficiency is associated with severe purpura fulminans in neonates (Mahasandana et al, 1990).
The F5 R506Q mutation is the most common hereditary risk factor for venous thrombosis. Heterozygous carriers of the F5 R506Q mutation have an approximately three- to five-fold increased risk of venous thrombosis whereas the risk in homozygous carriers is estimated to be increased 80 times (95% CI 22–289) (Koster et al, 1993; Bertina et al, 1994; Ridker et al, 1995a; Rosendaal et al, 1995; Pomp et al, 2007). APC resistance in the absence of the F5 R506Q mutation is associated with an increased risk of venous thrombosis (Rodeghiero & Tosetto, 1999; de Visser et al, 1999; Tans et al, 2003).
Carriers of the F2 G20210A mutation have a two- to three-fold increased risk of thrombosis (Poort et al, 1996; Brown et al, 1997). This mutation is associated with elevated levels of prothrombin, which is suggested to be the mechanism by which the mutation increases the risk of venous thrombosis. Elevated levels of prothrombin in the absence of the F2 G20210A mutation are associated with a two-fold increased risk of venous thrombosis (95% CI 1·5–3·1 for the highest versus the lowers quartile) (Poort et al, 1996).
High levels of other procoagulant factors have also been associated with an increased risk of venous thrombosis. A high level of factor VIII, i.e. above 150%, is associated with a five-fold increased risk of thrombosis, which was shown to be independent of acute-phase reactions (Koster et al, 1995b; Kraaijenhagen et al, 2000; Kamphuisen et al, 2001; O’Donnell & Laffan, 2001; Bobrow, 2005). Also, high levels of fibrinogen and factors IX and XI have been associated with an approximately two-fold increased risk of thrombosis (Koster et al, 1994; van Hylckama Vlieg et al, 2000; Meijers et al, 2000).
A cause of acquired thrombophilia is the antiphospholipid syndrome. The presence of lupus anticoagulants is most strongly associated with the risk of thrombosis. The presence of antibodies to anticardiolipin, β2-Glycoprotein I, and prothrombin have been associated with thrombosis in some, but not all, studies (Galli et al, 2003; Galli & Barbui, 2005). More recently, it was shown that the presence of anti−β2-Glycoprotein I antibodies was associated with an increased risk of a first episode of deep venous thrombosis (de Groot et al, 2005). This test seems less affected by problems of standardisation (Urbanus et al, 2008). A large prospective follow-up study found an increased risk of venous thrombosis associated with the presence of anti-prothrombin antibodies (Bizzaro et al, 2007).
Deep venous thrombosis is a multicausal disease. Numerous studies indicated that the risk of venous thrombosis is highly increased when more than one risk factor is present within one individual. This became first apparent in families with thrombophilia, in which two or more genetic defects were often found (Zoller et al, 1995; van Boven et al, 1996; Koeleman et al, 1997; Makris et al, 1997; Meinardi et al, 2001, 2002). The finding that the combination of F5 R506Q with oral contraceptive use considerably increases the risk was the first example of a clear gene–environment interaction in the aetiology of venous thrombosis (Vandenbroucke et al, 1994).
Association between thrombophilia and recurrent deep venous thrombosis
Venous thrombosis has a tendency to recur. The cumulative incidence of a second episode is approximately 30% in 8 years (Prandoni et al, 1996a). It has been consistently shown that patients with a clear clinical risk factor eliciting a first deep venous thrombotic event have a very low risk of recurrence (Prandoni et al, 1996a; Baglin et al, 2003; Christiansen et al, 2005). However, the evidence on laboratory testing for thrombophilia to predict the risk of a recurrent thrombotic event is much more challenging than its association with a first thrombotic event.
Conflicting results have been published on the risk of recurrent thrombosis associated with F5 R506Q and F2 G20210A. While some studies reported an increased risk of recurrent thrombotic events in all carriers of F5 R506Q (Eichinger et al, 1997, 2002; De Stefano et al, 1999; Kearon et al, 1999; Lindmarker et al, 1999; Rintelen et al, 1996), others found no increased risk or only in individuals homozygous for the mutation (Ridker et al, 1995b; Marchetti et al, 2000; Simioni et al, 2000; Miles et al, 2001; Vink et al, 2003). Also, double heterozygosity for F5 R506Q and F2 G20210A, or concomitance of other thrombophilic defects appear to increase the risk of recurrent venous thrombosis (De Stefano et al, 1999; Meinardi et al, 2002).
More recently, two large follow-up studies assessed the risk of recurrent venous thrombosis associated with thrombophilic defects. Baglin et al (2003) showed that carriers of a thrombophilic defect, i.e. antithrombin, protein C, or protein S deficiency [hazard ratios (HRs) ranging from 1·0 to 2·9 for individual deficiencies, which were almost all combined with other thrombophilic defects], or F5 R506Q (HR 1·4; 95% CI 0·7–2·8), or F2 G20210A (HR 1·7; 95% CI 0·5–5·6) did not have a highly increased risk of developing a recurrent venous thrombotic event. This was also observed in the large follow-up study by Christiansen et al (2005), which found no clear increased risk of recurrent venous thrombosis the prothrombotic risk factors: F5 R506Q (HR 1·3; 95% CI 0·8–2·1), F2 G20210A (HR 0·7; 95% CI 0·3–2·0), and elevated levels of factor VIII (HR 1·3; 95% CI 0·8–2·1). Only a deficiency in one of the anticoagulants protein C, protein S, or antithrombin, showed a mildly increased risk of recurrent venous thrombosis (HR: 1·8; 95% CI 0·9–3·7). De Stefano et al (2006) reported a similar risk of recurrence associated with deficiencies of the anticoagulants (AT deficiency: HR: 1·9; 95% CI 1·0–3·9, protein C and S deficiency: HR: 1·4; 95% CI 0·9–2·2).
Global assessments could be useful in the prediction of the risk of recurrent venous thrombosis. Elevated levels of D-dimer have been associated with an increased risk of thrombosis. However, with regard to recurrent thrombosis, mainly a high negative predictive value has been reported (Palareti et al, 2002; Eichinger et al, 2003). The usefulness of thrombin generation tests in predicting first or recurrent venous thrombosis remains to be established. Most individual single nucleotide polymorphisms (SNPs) associated with an increased risk of a first deep venous thrombosis, are not associated with the risk of a recurrent event. However, multiple SNP analysis may prove to be useful in the prediction of recurrent thrombosis. Recently, we have reported an increased risk of recurrent venous thrombosis associated with carriers of multiple SNPs that were individually associated with only a mild increased risk of a recurrent event. However, it was concluded that, although the risk of a recurrent event is increased when more than one SNP is present within one individual, the number of carriers is low, indicating that the clinical utility of multiple SNP analysis at present would be limited (van Hylckama Vlieg et al, 2008).
The risk of recurrence in antiphospholipid or anticardiolipin antibodies was investigated in various studies (Prandoni et al, 1996b; Rance et al, 1997; Schulman et al, 1998; Zanon et al, 1999; de Groot et al, 2005; Lim et al, 2006). The outcomes regarding relative risk for recurrence ranged between two- and six-fold. These results are hard to interpret, as these studies did not test the antiphospholipid antibodies or lupus anticoagulant repetitively (as suggested by international guidelines (Miyakis et al, 2006). Moreover, duration of anticoagulant treatment differed substantially.
Association of thrombophilia with arterial thrombosis
Arterial thrombosis is considered to occur when a (until then) subclinical atheroscletosic plaque ruptures. Atherosclerosis is a multi-factorial condition, of which most risk factors don’t overlap with those for venous thrombosis. It has been hypothesised that thrombophilia particularly increases the risk of arterial events in the absence of overt atherosclerotic lesions in the vessel wall. However, whether thrombophilia plays a role in arterial thrombosis remains controversial. Moreover, whether the presence of thrombophilia affects the risk of recurrent arterial thrombosis is unknown.
Evidence of an association between deficiencies of antithrombin, protein C, or protein S and arterial thrombosis is limited to case reports and small studies that are generally hampered by the low prevalence of these thrombophilias. No cases of antithrombin deficiency were found in two studies of young patients with myocardial infarction; one study investigated its prevalence among 70 survivors of myocardial infarction before the age of 36 years (Rallidis et al, 2003), and another study in 75 patients with myocardial infarction before the age of 45 years who had no evidence of coronary atherosclerosis (Dacosta et al, 1998). Although case reports suggest that antithrombin deficiency may be associated with stroke (Arima et al, 1992; Martinez et al, 1993), studies in neonates, children and young adults with ischemic stroke revealed no association with antithrombin deficiency (Gunther et al, 2000; Hankey et al, 2001; Carod-Artal et al, 2005).
Also for hereditary protein C deficiency, there is no strong association with myocardial infarction. Although patients have been described with protein C deficiency and myocardial infarction (Peterman & Roberts, 2003; Tiong et al, 2003), three case–control studies did not demonstrate an increased prevalence of protein C deficiency in young patients with myocardial infarction as compared to controls (Hayashi et al, 1997; Rallidis et al, 2003; Segev et al, 2005). Studies on the relationship between protein C deficiency and ischemic stroke have focused on patient populations with a low atherosclerotic burden, particularly young adults and children. Positive associations have been reported mainly in small studies and case reports (Grewal & Goldberg, 1990; Kohler et al, 1990; De Stefano et al, 1991; Martinez et al, 1993; deVeber et al, 1998). Larger studies, also among unselected patients with ischemic stroke, did not demonstrate a higher prevalence of protein C deficiency in stroke patients (Douay et al, 1998; Ganesan et al, 1998; Munts et al, 1998; Hankey et al, 2001).
Several case reports exist regarding young patients who develop myocardial infarction without evidence of significant coronary artery disease and are then found to have protein S deficiency (Beattie et al, 1997; Manzar et al, 1997). However, larger studies have not found an association between protein S deficiency and myocardial infarction (Rallidis et al, 2003; Segev et al, 2005). Likewise, although small studies and case reports have suggested that protein S deficiency increases the risk of developing ischemic stroke (Girolami et al, 1989; Green et al, 1992), larger studies, also focusing on patient populations with a low atherosclerotic burden including children and neonates, have not confirmed this relationship (Douay et al, 1998; Munts et al, 1998; deVeber et al, 1998; Hankey et al, 2001).
Numerous studies have investigated whether F5 R506Q is a risk factor for myocardial infarction. Several large cohort studies including the Physicians’ Health Study (Ridker et al, 1995b), the Cardiovascular Health Study (Cushman et al, 1998), and the Copenhagen Heart Study (Juul et al, 2002) did not find an association between F5 R506Q and myocardial infarction. In a meta-analysis the pooled odds ratio for myocardial infarction was 1·3 (95% CI 0·9–1·7) (Boekholdt et al, 2001). Several studies have suggested that F5 R506Q may be a risk factor in rare cases where myocardial infarction occurs without evidence of atherosclerosis. The presence of F5 R506Q was significantly higher among people who developed myocardial infarction without evidence of coronary artery disease, than among those with coronary artery disease or healthy controls (Mansourati et al, 2000). Also, a similar phenomenon was observed among young women, who are at very low risk of having significant coronary artery disease (Rosendaal et al, 1997). An interaction between F5 R506Q and other classical risk factors, most notably smoking, was observed. Whereas the presence of one risk factor led to a moderately elevated risk, F5 R506Q carriers who smoked had a 32-fold increased risk of myocardial infarction compared with non-smoking, non-carriers of F5 R506Q (95% CI 7·7–133). Despite these positive findings, other studies did not confirm the association between and myocardial infarction at a young age. For instance, the Italian Atherosclerosis Thrombosis and Vascular Biology Study compared 1210 survivors of myocardial infarction before the age of 45 years with 1210 controls and found no evidence for an association (Atherosclerosis Thrombosis and Vascular Biology Italian Study Group, 2003). The pooled odds ratio in patients developing myocardial infarction before the age of 55 years in the previously mentioned meta-analysis remained 1·3 but became borderline significant (95% CI 1·0–1·6) (Boekholdt et al, 2001). Similar findings exist for the relationship between F5 R506Q and stroke. Large studies investigating unselected patients with ischemic stroke observed no significant association (Ridker et al, 1995b; Cushman et al, 1998; Hankey et al, 2001). Again, results in patients with a low atherosclerotic burden have been inconsistent. Whereas in one study of 106 women with an ischemic stroke before the age of 45 years, F5 R506Q was not a risk factor (Longstreth et al, 1998), a larger study among 468 stroke and transient ischaemic attack patients before the age of 60 years found a 2·6 (95% CI 1·5–4·6) increased risk in F5 R506Q carriers, and an 8·8-fold (95% CI 2·0–38·0) risk in smoking carriers of F5 R506Q as compared to non-smoking non-carriers (Lalouschek et al, 2005). Some studies in children with stroke suggest that F5 R506Q may be a risk factor in this highly selected group (Zenz et al, 1998). An Israeli study among 65 children with ischemic stroke estimated a five-fold higher risk in children with F5 R506Q (OR 4·8, 95% CI 1·4–16·5) (Kenet et al, 2000).
The relationship between F2 G20210A and the risk of myocardial infarction has been subject of a number of studies, but results have been inconsistent. In the Study of Myocardial Infarctions Leiden among 560 men who suffered a first myocardial infarction before the age of 70 years, F2 G20210A was associated with a slightly increased risk of myocardial infarction (OR 1·5, 95% CI 0·6–3·8). However, the risk was mainly increased in the presence of classical cardiovascular risk factors, e.g., smoking, hypertension, diabetes mellitus, or obesity with risk increases between three- and six-fold (95% CIs 1·5–6·7 and 3·0–12·5) (Doggen et al, 1998). A family study revealed that F2 G20210A carriers were at an elevated risk of myocardial infarction as compared with their family members without the mutation (OR 4·7, 95% CI 1·0–22·5) (Bank et al, 2004a). However, a subanalysis of the Physicians’ Health Study observed no significant association between F2 G20210A and risk of myocardial infarction (Ridker et al, 1999). Another study, of 539 patients with myocardial infarction, found an OR of 0·7 (95% CI 0·3–1·6) (Croft et al, 1999). This lack of association was confirmed in a meta-analysis that found an overall OR of 1·1 (95% CI 0·8–1·6) (Boekholdt et al, 2001). However, this meta-analysis estimated a borderline significantly elevated risk of developing myocardial infarction at young age in F2 G20210A carriers (pooled odds ratio 1·9, 95% CI 1·0–3·5), but this was not confirmed by the large Atherosclerosis Thrombosis and Vascular Biology Study among survivors of myocardial infarction before the age of 45 years (Atherosclerosis Thrombosis and Vascular Biology Italian Study Group, 2003). Similar inconsistent observations apply to the relationship between F2 G20210A and stroke. Several large studies among unselected patients with ischemic stroke showed no association with F2 G20210A (Ridker et al, 1999; Hankey et al, 2001; Smiles et al, 2002). Findings in young stroke patients have shown inconsistent results. Patients with documented ischemic stroke before the age of 50 years and without cardiovascular risk factors were five times more likely to carry 2 G20210A than controls (De Stefano et al, 1998). A study of 468 patients with cerebrovascular disease before the age of 60 years found an elevated risk of stroke among male carriers of the F2 G20210A mutation (OR 6·1, 95% CI 1·3–28·3) but not among women (Lalouschek et al, 2005). In contrast, two other large studies observed no significant association between F2 G20210A and risk of developing stroke at young age (Austin et al, 2002; Pezzini et al, 2005).
In conclusion, no firmly established increased risk for arterial cardiovascular diseases, e.g. myocardial infarction and ischemic stroke, and inherited thrombophilia, is present. Some studies suggest that thrombophilia may increase this risk of arterial cardiovascular disease in young patients, particularly when classical risk factors, such as smoking, are present. However, positive studies are equally counterbalanced by negative studies. A clear-cut explanation for the discrepant findings between studies, e.g., differences in patient selection, is not available.
Association with pregnancy complications
Analogous to the clinical manifestations that are part of the antiphospholipid syndrome (Opatrny et al, 2006), family studies in the 1990s were the first to demonstrate that hereditary thrombophilia was also associated with an increased risk for miscarriage (Preston et al, 1996; Sanson et al, 1996; Meinardi et al, 1999). Since then, numerous studies have investigated this issue. From meta-analyses, it can be concluded that the association between hereditary thrombophilia varies depending on type of thrombophilia and timing of fetal loss (Rey et al, 2003) and that there is a modest association also between thrombophilia and other adverse pregnancy outcomes, most notably preeclampsia and intra-uterine growth retardation (Robertson et al, 2006). However, whether this association can be considered causal remains controversial, as many other factors, most notably structural or numeric chromosomal abnormalities, play a role in the risk of pregnancy complications (Rai & Regan, 2006; Middeldorp, 2007a,b; Rodger et al, 2008).
Implications of thrombophilia for clinical management
Many patients and clinicians are motivated to find an explanation for their disease. It should be realised however, that the existence of a thrombophilic defect does not exclude other risk factors, given the multi-causal aetiology of thrombosis. The relevant issue is under what circumstances a positive result on a thrombophilia test helps in the clinical management. Generally, this would consist of the need to adjust the therapeutic regime in thrombophilic patients, and the possibility of identifying asymptomatic family members (for subsequent preventive measures).
Clinical management of patients with venous thrombosis
After a first episode of venous thrombosis, 3–6 months of anticoagulant therapy is considered to have the optimal balance between the risk of treatment, i.e. bleeding, and the benefit, i.e. the prevention of an extension or recurrence of venous thrombosis (Buller et al, 2004). As outlined above, the presence of hereditary thrombophilia in patients with venous thrombosis does not strongly increase the risk of recurrence after discontinuation of anticoagulant therapy. In the absence of trials comparing routine and prolonged anticoagulant treatment in patients that tested positive for thrombophilia, with the current knowledge available prolonged anticoagulant therapy cannot be justified as it may cause more harm than benefit. An attempt to prevent recurrent thrombosis in the long-term without inducing an unacceptable number of major bleeding episodes was investigated in a trial comparing prolonged anticoagulant treatment with vitamin K antagonists with a low intensity (International Normalized Ratio below 2·0) and conventional intensity in patients after a first episode of venous thrombosis, with and without thrombophilia (Kearon et al, 2003). The lower intensity anticoagulation increased the incidence of recurrent venous thrombosis (1·9% vs. 0·6% in the conventional intensity group), without reducing major bleeding complications (0·96% vs. 0·93%).
For patients with antiphospholipid antibody syndrome the issue is more complicated. Patients with a first episode of venous thrombosis and antiphospholipid antibodies (tested once 6 months after the diagnosis of venous thrombosis) had an cumulative incidence of 29% recurrence during 4 years of follow-up, as compared to 14% in the patients without these antibodies (RR 2·1, 95% CI 1·3–3·3) (Schulman et al, 1998). This observation has led to the recommendation to treat patients with known antiphospholipid antibodies for at least 12 months (Buller et al, 2004; Ruiz-Irastorza et al, 2007). Nevertheless, guidelines refrain from giving recommendations regarding the routine testing of consecutive patients with venous thrombosis. Even if the prevalence of persistently positive tests was 10%, 10 patients would need to be tested in order to find one patient with venous thrombosis based on antiphospholipid syndrome in whom prolonged anticoagulant treatment would be installed. To our knowledge, no cost-effectiveness assessments have been performed. Based on retrospective studies with highly selected patients, it was assumed that patients with antiphospholipid antibody syndrome should be treated with a high intensity anticoagulant regime (Khamashta et al, 1995). However, two randomised trials showed that a higher intensity of anticoagulation with vitamin K antagonists in patients with antiphospholipid antibodies did not reduce the risk of recurrence, but led to an increase in the bleeding risk (Crowther et al, 2003; Finazzi et al, 2005).
Identification of asymptomatic family members with hereditary thrombophilia
A potential advantage of testing patients with venous thrombosis for thrombophilia may be the identification of asymptomatic family members of thrombophilic patients in order to take preventive measures if tested positive. As detailed previously, the risk for a first episode of venous thrombosis in relatives with thrombophilia is increased two- to ten-fold. Nevertheless, the overall absolute risk in thrombophilic families has been assessed in many studies and is generally low, even during high-risk situations such as pregnancy, puerperium, surgery, immobilisation, trauma and during the use of oral contraceptives. Estimates are listed in Table II. It is clear that the 2% annual major bleeding risk associated with continuous anticoagulant treatment outweighs the risk of venous thrombosis (van der Meer et al, 1996; Palareti et al, 1996). It is of note that the risk estimates related to surgery, trauma and immobilisation, as shown in Table II, mainly reflect the situation before standard prophylaxis was routine patient care. Low risk during pregnancy in asymptomatic women with thrombophilia does not justify prophylaxis with low-molecular-weight heparin during the nine months of pregnancy. Although the risk of severe complications, such as bleeding, osteoporosis and heparin-induced thrombocytopenia is very small, the nuisance of daily subcutaneous injections and skin reactions in one-third of women is high (Bank et al, 2003; Deruelle et al, 2006). Whether the 80% of pregnancy-related episodes occurring 6–12 weeks postpartum justifies prophylaxis during this period is a matter of the physicians’ and patients’ preference. The number needed to treat to prevent one postpartum thrombosis is 25 in case of a deficiency in the natural anticoagulants and approximately 50 in patients with the common thrombophilias. Finally, it is clear that risk of a first venous thrombosis during use of oral contraceptives should be weighed against the disadvantage of other contraceptive methods.
|Antithrombin, protein C, or protein S deficiency||F5 R506Q mutation||F2 G20210A mutation||Elevated levels of FVIII*|
|Overall (%/year, 95% CI)||1·5 (0·7–2·8)||0·5 (0·1–1·3)||0·4 (0·1–1·1)||0·3 (0·2–0·5)–1·3 (0·5–2·7)|
|Surgery, trauma, or immobilization (%/episode, 95% CI)||8·1 (4·5–13·2)||1·8 (0·7–4·0)||1·6 (0·5–3·8)||1·2 (0·4–1·8)|
|Pregnancy (%/pregnancy, 95% CI)||4·1 (1·7–8·3)||2·1 (0·7–4·9)||2·3 (0·8–5·3)||1·3 (0·4–3·4)|
|During pregnancy, %, 95% CI||1·2 (0·3–4·2)||0·4 (0·1–2·4)||0·5 (0·1–2·6)||1·0 (0·3–2·9)|
|Postpartum period, %, 95% CI||3·0 (1·3–6·7)||1·7 (0·7–4·3)||1·9 (0·7–4·7)||0·3 (0·1–1·8)|
|Oral contraceptive use (%/year of use, 95% CI)||4·3 (1·4–9·7)||0·5 (0·1–1·4)||0·2 (0·0–0·9)||0·6 (0·2–1·6)|
Clinical management of patients with arterial thrombosis
Despite the inconsistent but generally absent association of hereditary thrombophilia with arterial thrombosis, a survey in The Netherlands found that this was the indication for testing in almost a quarter of the ordered tests (Coppens et al, 2007). Given that no differential treatment or secondary prevention will follow from the presence of thrombophilia we strongly argue against testing. Vitamin K antagonists are generally recommended for patients with arterial thrombosis and a definite diagnosis of antiphospholipid antibody syndrome, but there is no evidence from well-designed trials that supports this expert recommendation (Ruiz-Irastorza et al, 2007).
Clinical management of patients with thrombophilia and pregnancy complications
Therapeutic options to prevent pregnancy complications in women with thrombophilia comprise aspirin as well as (low-molecular-weight) heparin. There is currently no evidence supporting treatment for women with recurrent pregnancy loss (Walker et al, 2003; Di Nisio et al, 2005; Middeldorp, 2007a; Rodger et al, 2008). Observational research is hampered by severe methodological flaws or inconsistent results. Two published randomised trials did not use an adequate comparator, i.e. no treatment or placebo (Gris et al, 2004; Brenner et al, 2005). Currently, randomised controlled trials with a no treatment or placebo arm are being carried out and results should be awaited before implementing a potentially harmful intervention in pregnant women with inherited thrombophilia and a history of pregnancy complications. For women with antiphospholipid antibody syndrome, aspirin and low-molecular-weight heparin treatment is often suggested although the evidence is limited (Empson et al, 2005; Jauniaux et al, 2006).
Drawbacks of thrombophilia testing
The disadvantages of testing patients with a venous thrombosis for thrombophilia include be the cost of testing, which is approximately €500, for a complete thrombophilia screen (Machin, 2003). Although some cost-effectiveness studies have been published regarding the testing for thrombophilia, which concluded that in some scenarios testing could indeed be cost-effective, the number of assumptions from inconsistent observational studies seriously hamper their interpretation (Marchetti et al, 2000; Wu et al, 2006). The psychological impact and consequences of a person knowing that they are a carrier of a (genetic) thrombophilic defect is a potential drawback of testing (Cohn et al, 2008). Most studies that focused on impact of testing for thrombophilia showed that patients had experienced low psychological distress following thrombophilia testing (van Korlaar et al, 2005; Legnani et al, 2006). Nevertheless, a qualitative study described several negative effects of both psychological and social origin (Bank et al, 2004b).
Despite the increasing knowledge about the multifactorial etiology of venous thrombosis, testing for hereditary thrombophilia generally does not alter the clinical management of patients with venous or arterial thrombosis or pregnancy complications. There are a few exceptions. For some asymptomatic women at fertile age who come from families with a tendency for venous thrombosis and a known thrombophilic defect, a positive test may lead to the decision to install prophylaxis postpartum in case of pregnancy, or the individual decision to not use oral contraceptives. According to current guidelines, testing for antiphospholipid antibody syndrome may be justified in patients with venous or arterial thrombosis or well-defined pregnancy complications, although no formal studies about its cost-effectiveness have been carried out. There is an intermediate strength recommendation to prolong anticoagulant treatment to 12 months in case of venous thrombosis, and to use vitamin K antagonists instead of aspirin in case of arterial thrombosis in patients who fulfil the laboratory criteria for antiphospholipid antibody syndrome. Finally, women with pregnancy complications and a definite diagnosis of antiphospholipid antibody are usually treated with aspirin and low-molecular-weight heparin.
In conclusion, because testing for hereditary thrombophilia does not affect clinical management of most patients with venous or arterial thrombosis or patients with pregnancy complications, it only serves a limited purpose and should not be performed on a routine basis.