Thrombophilia: grading the risk

In this issue of Blood, Lijfering and colleagues provide data on the absolute risk for both initial and recurrent venous thromboses in persons with thrombophilia. Based on these data, they subgroup thrombophilic defects into high- and low-risk disorders. They also conclude that some defects are not independent risk factors.

Ever since the first description of antithrombin deficiency as a cause of familial thrombophilia by Egeberg in 1965,¹  the number of reported inherited risk factors for venous thrombosis has been increasing, especially in the past 15 years.²  Most previous studies have reported on relative rather than absolute risks and dealt with single defects. It is clear that thrombophilia is multi-factorial due to gene-gene and gene-environment interactions. While relative risk is useful in terms of learning about the pathophysiology of the disease, the clinician requires absolute risk information to help make decisions about patient management.

In this study involving 3 Dutch hospitals, Lijfering et al investigate 2479 relatives of 877 probands with thrombosis.³  To avoid bias, probands were excluded from the analysis. The risk for a first deep vein thrombosis (DVT) was 1.52% to 1.90% per year for those with deficiencies of antithrombin, protein C or protein S, and 0.34% to 0.49% per year for those with factor V Leiden, the prothrombin mutation or elevated FVIII. Within these 2 groups, the thrombotic risk was similar for each of the individual defects, and in the paper, the 2 groups were classified as high- and low-risk thrombophilias, respectively. The risk of recurrence was 55% at 10 years for defects in the first group, 25% for the second (see table). Although elevated levels of FIX, XI, TAFI, and homocysteine appeared to be associated with an increased thrombotic risk, all were closely linked to concomitant elevated FVIII and were not risk factors in isolation.

In analyzing the risk of venous thrombosis, the authors assume that the risk in persons with antithrombin, protein C, or protein S deficiency is the same, irrespective of the severity of the deficiency or mutation causing the defect. However, as they have recently shown in protein S deficiency,⁴  this may not be the case, and one would logically expect an inverse relationship between deficiency and risk.

Whether or not to test for thrombophilia is controversial because the clinical utility of doing so has not yet been proven.⁵  For FIX, XI, TAFI, and homocysteine, this study suggests the answer is a definitive no, since elevated levels are not associated with increased thrombotic risk independent of elevated levels of FVIII. In the case of heterozygosity for factor V Leiden, or the prothrombin mutation, or for elevated FVIII, the answer is also probably no because the risk of a first DVT at under 0.5% per year is not high enough to warrant primary warfarin prophylaxis and the recurrence risk is no different from that reported for patients with first DVT who have not been tested for thrombophilia.⁶ 

More difficult is the issue of the higher risk thrombophilias represented by the deficiencies of the natural anticoagulants antithrombin, protein C, and protein S. The authors suggest that these are conditions with a much higher risk of first and recurrent thrombosis and should be managed differently from the commoner thrombophilias. Although based on these data alone there would be reluctance to advocate long-term primary prophylaxis, this should certainly be offered at times of additional high risk, such as after surgery, immobility, or pregnancy.

Clinically, the issue of thrombophilia testing and management is more relevant in the setting of patients who have experienced an event already. If testing has been performed and high-risk thrombophilia has been identified, this should certainly be taken into account when deciding on extended anticoagulation, especially for spontaneous events. The issue of whether all patients with a DVT should be screened for high-risk thrombophilia is unresolved⁷  but, for those with a spontaneous event at a young age and a positive family history, this should be considered. Definition of a positive family history is difficult, but the suggestion offered in this paper of more than 20% of relatives affected is not evidence-based and would be dependent on relatives being available for study.⁸ 

Any decision on whether to offer long-term anticoagulation will depend on the risk of bleeding while on anticoagulants as well as the thrombotic risk. This study reports a very low annual bleeding risk at 0.29% but with wide confidence intervals, because it is based on only 2 events. The authors speculate that this may be because the thrombophilic defect reduces the bleeding risk, and this observation certainly requires confirmation. Alternative explanations are the young age of the cohort, the fact that the patients are cared for by expert centers, and the small number of events.