Assessing thrombotic risk
BMJ 1998; 317 doi: https://doi.org/10.1136/bmj.317.7157.520 (Published 22 August 1998) Cite this as: BMJ 1998;317:520
- Michael Laffan (mlaffan{at}rpms.ac.uk), senior lecturer in haematologya,
- Edward Tuddenham, professor of haemostasis.b
- aDepartment of Haematology, Imperial College School of Medicine, Hammersmith Hospital, London W12 0NN
- bMRC Haemostasis Research Group, Clinical Sciences Centre, Imperial College School of Medicine Hammersmith Hospital
- Correspondence to: Dr Laffan
Editorial by Vandenbroucke
Venous thrombosis and venous thromboembolism are major medical problems. Because the time and scope for intervention are limited, the key to reducing the high morbidity and mortality of these conditions is prevention. Increased understanding of thrombotic mechanisms together with the recent identification of subsets of patients who are at increased risk of thrombosis are enabling clinicians to predict risk more accurately and increase the scope for targeted prevention. This article discusses current approaches and future prospects for assessing thrombotic risk and their relation to prophylactic strategies and new treatments.
Background
In hospital practice, postmortem and other studies suggest that venous thromboembolism causes 10% of deaths and contributes to a further 15%. Although it is relatively less common in the community, where the prevalence varies with age between 1 in 10 000 and 1 in 1000 per year, it is still a major cause of concern. Anxiety runs highest when oral contraceptive or hormone replacement therapy is being considered, and a small number of women die each year from thrombosis associated with these agents. The importance of venous thrombosis is often considered largely in terms of risk of death, but this ignores considerable morbidity from the miserable and disabling postphlebitic syndrome, which will develop in >20% of survivors.
In most cases thrombosis arises from a combination of external circumstances and the inherited and acquired predisposition to thrombosis of an individual. Only 30-40% of cases seem to be completely spontaneous. So far we have been better at identifying high risk circumstances than high risk individuals, and so the concept of screening people for risk of thrombosis is relatively new. The tendency of some individuals to develop thrombosis more easily than others is often inherited, and the term thrombophilia has been coined to describe this condition. We can now identify a genetic contribution to thrombosis in roughly half of patients presenting with a first thrombotic episode, although the relative risk associated with these genetic abnormalities varies appreciably (see table). 2 3 As we become better at assessing individual risk, the distinction between those with and without thrombophilia will become blurred and the risk in the population will be seen to present a continuous spectrum. The label thrombophilia is then applicable only to those whose thrombotic risk is greater than some arbitrary value, and thrombophilia screening will be better described as assessing thrombotic risk.
Potential futures
Single step analysis of genetic polymorphisms to determine genetic influence on thrombotic risk
Assessing individual thrombotic risk by analysis of genetic polymorphisms and coagulation factors
Alternatively, direct assessment of thrombotic tendency of plasma samples
With simpler and more accurate assessment methods, screening forthrombotic risk will extend from high risk individuals with personal orfamily history of thrombosis to patients starting treatments known toincrease risk, such as oral contraceptives and hormone replacement therapy
Use of new antithrombotic agents directed against factors Xa and VIIaand plasminogen activator inhibitor, with better risk:benefit ratiosthan warfarin, for extension of prophylactic treatment to low risksituations
Prevalence of prothrombotic factors in a white population and the associated relative risk of venous thrombosis (adapted from Cooper and Krawczack1)
Regulation of coagulation and thrombotic risk
We believe that the principal components of the coagulation network are now known, including both procoagulant and anticoagulant participants. The network results in a balance between these opposing forces that ideally results in formation of a fibrin clot only when physiologically appropriate and which is confined to the region of injury (fig 1). There are five principal anticoagulant factors in this scheme: antithrombin, which neutralises thrombin and other procoagulant enzymes; protein C, which inactivates the cofactors factor V and factor VIII; protein S, the cofactor for protein C; thrombomodulin, which enables thrombin to activate protein C; and tissue factor pathway inhibitor, which neutralises the initiators of coagulation. There are several ways in which the balance of the network may be shifted in favour of thrombosis, but classic thrombophilia usually arises from a deficiency of antithrombin, protein C, or protein S. Deficiency of tissue factor pathway inhibitor might be expected to have a similar effect, but this has not been described in humans (when it is inactivated in mice by genetic mutation the embryos die in utero).
Opposing procoagulant and anticoagulant forces in generation of fibrin clot. Coagulation is initiated by tissue factor VIIa complex, which is neutralised by tissue factor pathway inhibitor. Procoagulant clotting factors are opposed by antithrombin and by protein C with its cofactor, protein S. The fibrinolytic pathway comprises activation of plasminogen to plasmin by tissue plasminogen activator; these factors are neutralised by −2 antiplasmin and plasminogen activator inhibitor respectively. Factor XIII stabilises the fibrin clot but, curiously, has no known inhibitor
When a fibrin clot is formed and cross linked by factor XIII, the fibrinolytic system is activated simultaneously, and this acts in an anticoagulant manner by degrading fibrin. Like the coagulation network, there is a system of profibrinolytic and antifibrinolytic factors—plasmin and plasminogen activator, and antiplasmin and plasminogen activator inhibitor (fig 1). Although a deficiency of fibrinolytic activity would be expected to produce a prothrombotic effect, it has proved difficult to establish any link between risk of thrombosis and either measured concentrations or genetic polymorphisms of fibrinolytic factors. Inheritance of an abnormal fibrinogen molecule (dysfibrinogenaemia) is a rare but well recognised cause of increased thrombotic risk.
Deficiencies of the anticoagulant factors antithrombin, protein C, and protein S are straightforward in physiological terms and cause an estimated 20-fold increase in thrombotic risk. They are, however, rare, accounting for no more than about 8% of patients investigated after a first thrombosis (table).
Newly identified causes of increased thrombotic risk
Recent work has shown that there are other coagulation abnormalities that are much more common among the general population than the anticoagulant factor deficiencies described above but that are associated with smaller risks of developing thrombosis.
Effect of factor V Leiden on thrombin generation compared with that of normal factor V. Thrombin generation is measured in vitro with purified factors to reconstruct the coagulation network. Without anticoagulant factorsprotein C-protein S system and thrombomodulinthere is little difference, but when these are present factor V Leiden results in significantly increased and prolonged thrombin generation. This is explained by the resistance of factor V Leiden to action of activated protein C. (Adapted from van't Veer et al6)
The most important of these is factor V Leiden. In 1993 Dahlback et al observed that in plasma samples from a family prone to thrombosis the anticoagulant effect of the activated form of protein C was less than normal.4 This phenomenon of activated protein C resistance was later shown in>90% of cases to be due to a mutation in the factor V molecule.5 This mutation destroys the cleavage site of activated protein C that is principally responsible for inactivation of factor Va but does not affect the coagulant activity of factor Va. The resultant factor V molecule, now often called factor V Leiden, is thus resistant to degradation by activated protein C. It is easy to envisage how this may result in an increased tendency to thrombosis (fig 2. Factor V Leiden is present in about 5% of most European populations and increases the risk of thrombosis five to 10 times. It is found in 20% of patients with a first episode of venous thrombosis.
Identification of other polymorphisms has followed. Recently, a sequence variation in the 3' untranslated region of the prothrombin gene has been described that is associated with raised concentrations of prothrombin and an increased risk of thrombosis and is present in about 3% of Europeans.7Raised concentrations of factor VIIIc and homocysteine have also been shown to have a close association with thrombosis and to be present in about 20% of patients with venous thrombosis. These abnormalities are difficult to assess because their genetic component is incompletely defined, and they probably arise from a combination of genetic and acquired or environmental factors. Other mutations or polymorphisms, including examples in thrombomodulin, platelet glycoproteins, and factor XIII, are under investigation.
Although these factors are less powerful in their prothrombotic effect than the classic thrombophilias, their higher prevalence makes them just as important, and their identification has greatly increased the number of patients with thrombosis in whom a specific genetic contribution can be recognised.
Multiple genetic factors
The extent to which individuals with the same thrombophilic abnormality are affected by thrombosis varies considerably, even within the same family. Conversely, members of a thrombophilic family have an increased risk of thrombosis even if they do not carry the identified thrombotic trait. These observations suggest the interaction of several factors and that thrombophilia is polygenic in origin.
Because some of the identifiable traits, in particular factor V Leiden, have a relatively high prevalence it has been possible to assess the effect of their interactions. This is shown by the effect of co-inheriting protein C deficiency and factor V Leiden. In one study 78% of subjects with protein C deficiency and factor V Leiden had thromboses compared with 31% of those with protein C deficiency alone.8 Synergy has also been reported between factor V Leiden and protein S deficiency, antithrombin deficiency, raised homocysteine, and raised factor VIII.9
As the list of identifiable genetic abnormalities associated with increased thrombotic risk grows, it will be possible to build up a more complete assessment of an individual's risk. Currently, analysis of genetic polymorphism is expensive and confined to identification of a few single factors, most of which are associated with only a low risk of thrombosis. This is being simplified by the use of multiplex polymerase chain reaction and heteroduplex techniques. If it is to be possible to assess the risk in all individuals then many more genetic traits will be needed and analysis by polymerase chain reaction will be too laborious. To remedy this, new technologies should become available in pathology laboratories within the next five years that will allow high throughput analysis of multiple polymorphisms by hybridisation to arrays of probes distributed on microchips.
Screening for increased thrombotic risk
Investigating patients with thrombosis
With the developments outlined above, it is possible to identify a prothrombotic abnormality (as opposed to a precipitating factor) in roughly half of patients with venous thrombosis. Attempts to increase this proportion by selecting patients for investigation by age or familial or clinical history were not successful because, although the percentage of positive results was increased, many abnormalities were missed. It is now routine in many hospitals to test patients for increased thrombotic risk after a single episode of thrombosis. Investigation is based on the assumption that treatment, for the patient or for relatives, may be guided by the outcome. A recent study supported this assumption by showing that the probability of a recurrence of venous thrombosis was increased 2.4 times in patients in whom factor V Leiden was identified.
Investigating asymptomatic patients
One of the most important questions that such findings pose is whether there is a case for routine screening of all patients who are being given treatment that is known to increase the risk of venous thrombosis. This would obviously include the millions of women who take the oral contraceptives or hormone replacement therapy. A recent analysis concluded that, in relation to factor V Leiden, at least 200 000 and possibly 2 million women would have to be tested to prevent one death and that this was therefore not a useful exercise.10However, this analysis considered only one factor and is not applicable to what we know is a polygenic condition. If we were able to assess an individual's risk simply and reliably then those found to be at high risk of thrombosis could be advised against taking oral contraceptives or hormone replacement therapy, and most women could be reassured. In other circumstances prophylaxis against thrombosis could be modified on the basis of the results.
Extending screening
The association of increased concentrations of factor VIII and prothrombin with increased thrombotic risk suggests that quantitative variations of procoagulant and anticoagulant factors may be important. Thus, a full assessment of thrombotic risk could require assays of numerous factors, the results of which would then be fed into an algorithm as quantitative traits to obtain a quantitative estimate of the thrombotic tendency of the plasma.
The coagulation system has already been modelled in this way. Using multiple differential equations representing the kinetics of the individual enzymatic interactions, Ken Mann has constructed a computer model that reproduces the time course of thrombin generation seen in artificially reconstructed systems. The usefulness of such an approach in predicting thrombotic risk has not been tested, but it does show that coagulation is computable. However, this model does not include any contribution from the endothelium, where, for example, variation in thrombomodulin expression may be critically important. Polymorphic variation has recently been described in the thrombomodulin gene, and this might need to be added to the model.
Direct tests of thrombotic tendency
One way round the problem of trying to integrate multiple yet obviously incomplete tests would be to develop an analysis that directly estimated the thrombotic tendency of patients' plasma or whole blood. This could also be criticised for ignoring the contribution of the endothelium, but it would be a direct attempt to assay the phenotype.
The best known approach is measurement of what have been called activation markers, the most studied of which is prothrombin fragment F1+2. Fragment F1+2 is the peptide cleaved from prothrombin as it is converted to thrombin and is therefore a measure of ongoing thrombin generation. Unfortunately, this is not a good predictor of future thrombosis and is not consistently elevated even in the classic thrombophilias.11
Other tests to measure the “thrombin generation potential” have now been devised, and it may be possible to develop a test that directly assesses a patient's thrombotic tendency. If simplified, this may allow a more quantified assessment of risk before surgery, pregnancy, or taking drugs containing oestrogen. Such a phenotypic assessment would be independent of genetic analysis and might be simple enough to be done at the bedside or in general practice.
Treatment
The benefit of an investigation rests on whether any clinical decision will be altered by the information obtained. The usefulness of relatively refined information on risk of thrombosis is currently limited by the inadequacies of available anticoagulant treatments. After 50 years of use, warfarin remains the only drug available for long term anticoagulation. Standard warfarin treatment is associated with an annual risk of haemorrhage of 7% and of death of 0.2%. To be useful, an investigation must therefore be able to predict a thrombotic risk that outweighs the problems of warfarin treatment; gradations of risk below this are of little use.
New anticoagulant drugs directed against factors Xa and VIIa and plasminogen activator inhibitor now in development promise a better ratio of risk to benefit. This would allow us to take advantage of better assessment of thrombotic risk and consider prophylactic treatment in lower risk situations.
Conclusions
Recently, the details of how the blood coagulation system is regulated have become well understood. In parallel, several abnormalities of this system —including prevalent genetic polymorphisms that render patients and families more susceptible to thrombosis—have also been elucidated. There are still considerable discrepancies between the general risk associated with these abnormalities and the risk experienced by particular individuals, but these will diminish as more contributory factors are identified. We then need to find ways of assessing the multiple factors affecting thrombotic risk and to integrate them via algorithms to predict accurately an individual patient's risk so that appropriate treatment may be selected. This goal is reasonable and not far distant.