Laboratory testing for heritable thrombophilia: impact on clinical management of thrombotic disease

Venous thromboembolism has a multifactorial pathogenesis. An acquired risk factor can often be identified in subjects presenting with deep-vein thrombosis or pulmonary embolism. Tissue trauma, including surgery, immobilization, cancer, oestrogen use and pregnancy and the puerperium, are prominent precipitating factors. The attributable risk associated with each of these ranges from 4% to 18%. Recent interest has focused upon genetic risk of venous thromboembolism. It is now commonplace for a broad range of laboratory investigations to be used to determine the presence of heritable thrombophilia (Table I), as a genetic contribution can frequently be identified. For example, activated protein C resistance resulting from heterozygosity for factor V Leiden is present in around 20% of subjects of northern European origin who present with a first episode of venous thromboembolism, the background frequency for the genotype being at least 4%. This ability to ‘explain’ thrombosis has led to the increasing use of thrombophilia testing in the assessment of patients with deep-vein thrombosis and pulmonary embolism. Although such an approach to investigation satisfies the curiosity of the clinician and the patient, whether, in the current state of knowledge, the unselective intensive laboratory investigation of cases of venous thromboembolism is cost-effective is open to considerable doubt. This is mainly because of limitations in the utility of the knowledge gained in determining evidence-based treatment protocols for individual patients, but also because the interpretation of some tests for heritable thrombophilia is problematic. Tests for activated protein C resistance, factor V Leiden and the prothrombin G20210A polymorphism are reasonably robust and reproducible, although results from the UK National Quality Assurance Scheme have revealed some errors in their performance and interpretation. The accurate diagnosis of inherited deficiencies in the physiological anticoagulants antithrombin, protein C and protein S is less straightforward, however. This is because of the considerable overlap between some heterozygotes with a deficiency and unaffected subjects with plasma levels at the lower limit of the normal range. In the case of protein C, for example, this can result in failure to detect 15% of heterozygotes and the misclassification of 5% of normals (Allaart et al, 1993). Although interpretation can be improved through the use of age- and sex-specific normal ranges, this approach is rarely used (Tait et al, 1993; Preston Prandoni et al, 1996) and that of serious bleeding resulting from oral anticoagulant therapy (1% per year, with one-quarter of these being fatal; Palareti et al, 1996). Although the rate of recurrent thrombosis has predictably been shown to be lower during treatment for up to 2 years (Kearon et al, 1999), the risk–benefit ratio of such an approach is unclear. It is therefore only generally used on pragmatic grounds when the risks of further venous thrombosis are perceived to be unacceptably high, for example if a first event was life-threatening pulmonary embolism and was apparently unprovoked. Although some early, small retrospective studies suggested that heterozygotes for factor V Leiden have a higher thrombosis recurrence rate after discontinuation of oral anticoagulant therapy than do subjects with no laboratory evidence of thrombophilia (Ridker et al, 1995; Simioni et al, 1997), subsequent retrospective and prospective investigations have refuted this (Rintelen et al, 1996; Eichinger et al, 1997; Kearon et al, 1999; Lindmarker et al, 1999; De Stefano et al, 1999). Nor is the second most prevalent heritable thrombophilia, heterozygosity for the prothrombin G20210A polymorphism, associated with a relatively increased risk of recurrent thrombosis (Kearon et al, 1999; Lindmarker et al, 1999). Indeed, it is counterintuitive to regard subjects with a previous thrombosis who have one of the common thrombophilic polymorphisms as different from those who do not, as clearly the latter must also have an increased thrombotic tendency. It is therefore illogical to base decisions regarding the intensity or duration of anticoagulant therapy on such laboratory findings. The same considerations may apply to the much less prevalent thrombophilic conditions caused by deficiency of a physiological anticoagulant as, although high recurrence rates have been reported in antithrombin and protein S deficiency (van den Belt et al, 1997), the event rates seem to be no higher than those in unselected subjects with venous thrombosis (Prandoni et al, 1996). Co-inheritance of two inherited thrombophilic conditions, which is most commonly double heterozygosity for factor V Leiden and prothrombin G20210A, does seem to be associated with an increased thrombosis recurrence rate (Zoller et al, 1995; van Boven et al, 1996; Makris et al, 1997; De Stefano et al, 1999), and homozygotes for either polymorphism may also suffer a more aggressive clinical course (Lindmarker et al, 1999), although they often do not develop thrombosis until the fourth decade (Baglin et al, 1998). In families with co-inheritance of factor V Leiden and a deficiency state, thrombosis also tends to occur at a younger age in those individuals with the combined thrombophilia. However, homozygotes and double heterozygotes will represent only around 3% of subjects presenting with a first episode of venous thromboembolism, and frequently have a striking family history of deep-vein thrombosis. Also, they may not represent a discrete group, as some subjects in whom no, or a single, abnormality has been detected may have additional abnormalities that have not been tested or as yet unrecognized genetic or environmental predispositions to thrombosis. Furthermore, whether a favourable risk–benefit ratio exists for indefinite oral anticoagulant prophylaxis in clinically affected individuals remains unknown, and an alternative approach is one in which there is emphasis on avoidance or correction of additional risk factors and use of pharmacological thromboprophylaxis only at times of highest risk. The proportion of the attributable risk of development of venous thromboembolism, which relates to the non-heritable factors that are avoidable or manageable, is at least as great as that resulting from inheritance. Of particular relevance are pharmacological oestrogen exposure, in the form of oral contraception or hormone replacement therapy, or pregnancy. Use of the combined oral contraceptive is associated with an approximately threefold increased risk of venous thromboembolism overall, but there is a multiplicative effect in carriers of factor V Leiden and possibly other inherited thrombophilias. Thus, it has been estimated that a factor V Leiden heterozygote using the combined oral contraceptive is around 30 times more likely to suffer from venous thrombosis than is a woman who has neither risk factor (Vandenbroucke et al, 1995). This has led to consideration of whether testing for heritable thrombophilia could be a useful preventive measure in healthy women before prescription of hormonal contraception (Hellgren et al, 1995). The prevalence of venous thrombosis in women of child-bearing age is, however, very low, and few deaths occur. Indeed, it has been calculated that screening for factor V Leiden before use of oral contraception may require counselling and testing of over 2 million women to prevent one death from pulmonary embolism per year (Rosendaal, 1996; Vandenbroucke et al, 1996). Around 100 episodes of deep-vein thrombosis could also be prevented, but most of these cause no long-term morbidity, and even severe post-phlebitic symptoms improve with time (Prandoni et al, 1999). Furthermore, it is likely that some women advised against the use of hormonal therapy on the basis of test results would become pregnant and be at added risk of venous thrombosis as a result, further undermining, or even reversing, any benefit from thrombophilia screening. A more efficient approach would be to seek factor V Leiden heterozygotes among premenopausal family members of subjects with venous thrombosis. It must be appreciated, however, that even using this more focused approach, it is likely that, in order to prevent a fatal episode of pulmonary embolism, several thousand carriers would have to be advised against the use of the combined oral contraceptive, with unavoidable lifestyle and family planning consequences. If this approach is adopted, it is essential that a procedure for the delivery of expert counselling is in place to facilitate informed decision-making by potential carriers and to minimize unnecessary anxiety. An increasingly common practice is to test women with a putative positive family history of venous thrombosis for familial thrombophilia before hormonal contraceptive use. Although this is occasionally fruitful and may allow informed family planning decisions to be made, this only applies if the thrombophilic condition that segregates with clinical events in the family is known. This information is not usually available, and the potential for imparting false reassurance through the inappropriate interpretation of negative results of laboratory tests is considerable. Conversely, at least 4% of northern European women tested will be found to have a genetic predisposition that may never lead to a clinical event but may influence contraceptive choice. Similar considerations apply to thrombotic risk in women using hormone replacement therapy (HRT), although the background prevalence of venous thrombosis is around twice as great in this age group. The benefits of HRT are significant, ranging from the abolition of symptoms of oestrogen withdrawal to the reduced incidence of myocardial infarction and osteoporosis (Clinical Synthesis Panel on HRT, 1999). Great care must be taken that women are not denied these advantages of treatment through the inappropriate use and uninformed interpretation of laboratory investigations. In relation to pregnancy, there is no evidence yet that the risk–benefit ratio favours the use of thromboprophylaxis in asymptomatic carriers of factor V Leiden or prothrombin G20210A, or in those with anticoagulant deficiency. Even if heparin or warfarin therapy is restricted to the post-partum phase, the period of highest risk, such a policy would of course apply to 4% or more of women in many populations, some of whom would undoubtedly develop complications from the treatment. Knowledge of the presence of a familial thrombophilic state should not alter management in a woman with a previous venous thrombosis, in relation to contraceptive choice and use of pregnancy thromboprophylaxis, as she is already identified as being at increased risk. Exposure to oestrogens as contraceptive or HRT is relatively contraindicated, to a degree depending on the perceived risk–benefit ratio, and many clinicians would consider the use of thromboprophylaxis during, or at least after, pregnancy in such women. The identification of heritable thrombophilia does not influence these decisions. Even in antithrombin deficiency, in which the perceived risk of thrombosis is especially high, although a specific concentrate is available, many women have been managed successfully with conventional anticoagulant therapy. The use of heparin thromboprophylaxis perioperatively offers the possibility of significantly reducing the burden of venous thromboembolic disease and is now routine for patients over 40 years of age. Whether subjects with thrombophilia require more intensive thromboprophylaxis or a lower threshold for its administration has not been studied, but clinical experience does not suggest this to be so. Some additional measures of thrombotic risk are factor VIII, fibrinogen and homocysteine concentrations, each of which has an inherited component. Assays for these contributors to a prothrombotic state are reasonably robust, and the attributable venous thrombosis risk associated with them is similar to that of factor V Leiden and prothrombin G20210A. It is therefore curious that most laboratories concentrate on the classically familial thrombophilic conditions, often to the exclusion of these other important parameters, in attempting to explain thrombosis. Although increased plasma levels of factor VIII, fibrinogen and homocysteine also predict for arterial thrombosis in epidemiological studies, the value of their measurement in individual patient management has not been established in relation to either arterial or venous thromboembolism. Despite an increasing number of requests from clinicians for laboratory testing for factor V Leiden and the prothrombin G20210A mutation as part of the investigation of arterial thrombotic events, these heritable thrombophilias contribute minimally to the overall burden of myocardial infarction and thrombotic stroke, and such investigations are rarely justifiable. In contrast to heritable thrombophilia, the diagnosis of the acquired prothrombotic state that accompanies the presence in plasma of antiphospholipid antibodies is often clinically important. Prospective data indicate that the detection of lupus anticoagulant or anticardiolipin in a subject with thrombosis indicates an exceptionally high risk of future arterial and venous thrombotic events (Greaves, 1999). Indiscriminate application of laboratory investigations is poor clinical practice, as it diverts scarce resources from other areas. It can also be misleading if the clinician is not fully aware of the significance of results, both positive and negative. In the current state of knowledge, identification of a non-modifiable contributory factor to thrombosis is not a worthwhile end in its own right, especially if there is a risk of generating needless anxiety in patients and their asymptomatic relatives or of providing false reassurance. Careful and informed counselling is an essential component of the assessment of a patient with venous thrombosis, and testing for heritable thrombophilia is only occasionally indicated, perhaps, for example, when there is a strong family history of venous thromboembolism. Data from the United Kingdom Quality Assurance Scheme in blood coagulation indicate that over 25 000 tests for activated protein C resistance were performed in the UK in 12 months during 1996/7 (F. E. Preston, personal communication). Many of these samples would also have been tested for prothrombin G20210A and the rare conditions of antithrombin, protein C and protein S deficiency. It is unlikely that the expense and effort involved is justified through improved patient care in relation to the treatment and prevention of venous thromboembolic disease. Subjecting all patients with venous thrombosis or their asymptomatic relatives to a barrage of laboratory investigations, many of which provide no information to guide evidence-based medical practice, cannot be justified. Every patient with venous thromboembolism deserves a thorough clinical assessment, including a search for modifiable lifestyle and environmental risk factors. Before embarking on a search for heritable thrombophilia, it is essential that careful thought be given to any possible value for the patient and family by a clinician with a thorough understanding of venous thrombosis, its causes and management.