Practical strategies for hormone replacement therapy and risk of venous thromboembolism


Until recently hormone replacement therapy (HRT), unlike the combined oral contraceptive pill, was not known to be associated with venous thromboembolism1. The difference between HRT and the combined pill with regard to venous thromboembolism was thought to reflect the ‘physiological’ dose of natural oestrogens in HRT in contrast to the ‘pharmacological’ doses of potent synthetic oestrogens in the combined pill. It was acknowledged, however, that the lack of association was due to an absence of evidence rather than an evidence of absence of such an association. Recent case-control studies have clearly indicated a modest increase in the risk of venous thromboembolism in women on HRT containing oestrogen. The first of these was a population-based nested case-control study carried out in the United States in women with idiopathic venous thromboembolism2. The matched relative risk estimate was 3.6 for current users of oestrogens compared with nonusers. The risk appeared to be directly related to the dose of oestrogens used. A UK hospital-based case-control study3 of women in the 45 to 64 year old age group with idiopathic venous thromboeinbolism found an adjusted odds ratio for current users of HRT compared with nonusers of 3.5. The risk appeared to be highest among short term current users. The nurses’ health cohort study estimated the adjusted relative risk of primary pulmonary embolism to be 2.1 in HRT users with no relationship to past use4. A further population- based case control study in the UK utilising the General Practice Research Database found an adjusted odds ratio of 2.1 for venous thromboembolism for current users of HRT5. Interestingly, this study found that the increase in risk was restricted to the first year users with an odds ratio of 4.6 during the first six months5. Despite this increase in relative risk, the absolute risk for venous thromboembolism has been estimated at between 9 to 11 per 100,000 women per year in nonusers compared with 27–32 per 100,000 women per year in users2,3.

Thus, there is consistency in these studies utilising different techniques in a variety of populations with regard to the estimate of relative risk. Diagnostic suspicion bias might overestimate the risk of venous thromboembolism. It is also possible that patient selection led to those with a past or family history of venous thromboembolism being denied HRT. The data are also limited in terms of type and dose of oestrogen replacement, particularly with regard to transdermal administration where no firm conclusion can be drawn on risk of venous thromboembolism. Information has become available over the last decade on congenital and acquired thrombophilic problems. As several of these had not been identified, far less recognised, it is likely that the women suffering from such thrombophilic tendencies would be more likely to have a venous thromboembolism. Thus the groups may be unbalanced with regard to these important variables. Nonetheless, the consistency of the data suggest that this is a real excess risk albeit that the mechanism is poorly explained.

The mechanism may be related to changes in the haemostatic system induced by oestrogen replacement or by unmasking of an underlying thrombotic trait. The menopause is associated with changes in the coagulation system such as an increase in factor VII and fibrinogen which are risk factors for vascular disease, but this is accompanied by an increase in the coagulation inhibitor antithrombin6. Hormone replacement therapy produces significant changes in the haemostatic system including a reduction in fibrinogen, and factor VII7,8. This is accompanied by a reduction in inhibitors of the coagulation system such as antithrombin8, and enhanced fibrinolytic activity9. Some of these changes would appear to be associated with a reduction and others with an increase in thrombotic risk. However, our present knowledge is insufficient to interpret these findings in a clinical context with regard either to mechanisms of oestrogen-associated thrombosis or in determination of risk. We should therefore depend on the case-control studies to estimate the clinical risk of thrombotic complications in association with oestrogen replacement therapy. If the excess risk of venous thromboembolism were simply related to changes in the coagulation system, the frequency of venous thromboembolism would be expected to be higher than that seen in the case control studies. Particular consideration should therefore be given to HRT unmasking congenital thrombophilia. It is possible to screen for many of these underlying conditions, and this is an important factor when prescribing a drug with a thrombotic risk.

Most information exists on the congenital thrombophilias, antithrombin, protein C and protein S deficiencies. These are conditions where the major components of the body's endogenous anticoagulant system are defective or deficient due to quantitative or qualitative defects. These defects are known to cause particular problems in pregnancy10, a situation where oestrogenic changes may predominate. The prevalence of these congenital abnormalities in European populations is between 2 and 5.5 per 1,000. Thus, they are relatively uncommon and will be found in only around 10% of patients investigated for a past history of venous thromboembolism. This contrasts with the more recently identified thrombophilic defect, Factor V Leiden. This defect manifests itself in laboratory terms as a resistance to activated protein C, the endogenous anticoagulant directed against factor Va and factor VIIIa. Activated protein C normally inhibits the coagulation cascade by proteolytic cleavage of these factors. Resistance occurs due to a defect in the factor V gene which is a single mis-sense mutation at the activated protein C cleavage site. This results in normal coagulant activity, but a resistance to activated protein C. Thus a potential hypercoagulable state ensues. The heterozygote gene frequency in the population is 2% to 15% in Western countries11, and investigation of patients with venous thromboembolisin has found an underlying frequency of between 20% and 60% of factor V Leiden12. This defect has been associated with a high proportion of pregnancy and pill related thrombosis13,14. Functional-activated protein C resistance can occur in the absence of the factor V Leiden gene and is associated with the combined contraceptive pill, HRT, increased levels of factor VIII, obesity15 and also in pregnancy16 where factor VIII normally increases.

More recently, a relatively common genetic variation in the prothrombin gene has been found that has been associated with elevated plasma prothrombin levels and an increased risk of venous thrombosis by around three fold17. There is as yet no information about an association with oestrogen therapy and pregnancy. However, the underlying frequency of this genetic variation in the population is over 2%. Hyperhomocysteineaemia is also a clearly identifiable risk factor for venous and arterial thrombosis. However, as HRT containing oestrogen reduces serum homocysteine, this is unlikely to explain the excess risk of venous thromboembolism on HRT18. It is likely that further thrombotic traits will be identified in the near future.

Clearly, the underlying prevalence of congenital thrombophilic defects is in excess of 5% in Western European populations. However, the incidence of venous thromboembolism is substantially lower than this. Thus, the presence of these defects, on their own, does not result in thrombotic problems. In addition, this difference cannot be made up by the contribution of acquired thrombophilia which mainly takes the form of lupus anticoagulant and the antiphospholipid antibody syndrome, although such acquired thrombophilia may represent an additional risk.

Problems occur in the congenital thrombophilias where multiple factors come together to produce a substantial increase in risk of venous thromboembolism14. Examples of this can be found in studies of populations with thrombophilic defects where the frequency of Factor V Leiden is higher in symptomatic families with protein C deficiency than in the general population19. This raises the possibility that symptomatic thrombophilia may be due to multiple gene defects. Alternatively, the combination of a genetic susceptibility and physiological, pathological or pharmacological factors is likely to be important in the clinical expression of thrombophilias. For example, physiological situations, such as pregnancy, pathological problems requiring surgery and pharmacological interventions (eg, the oral contraceptive pill or HRT), may combine with an underlying thrombophilic trait to precipitate venous thromboembolism. There is already evidence of such a potential. For example the combined pill on its own has a relative risk of 3.7, Factor V Leiden alone of 6.9, but when combined the risk increases to 34.714. Given the known effect of the pill on activated protein C resistance such an interaction is not unexpected.

These interactions may explain the finding of the case control studies of venous thromboembolism in the first year of HRT as HRT may ‘unmask’ thrombophilia. Alternatively, HRT may simply be an additional risk factor which, when added to pre-existing factors, precipitates venous thromboembolism possibly through an interaction on activated protein C resistance. At present the use of oestrogen alone cannot be assumed to be the sole factor reponsible for venous thromboembolism in women receiving HRT

How then does the clinician deal with this information about a relationship between HRT and venous thrombotic events? There is no evidence for this problem, and therapeutic strategy depends on clinical logic and opinion. Firstly, should unselected screening for thrombophilia be instituted before commencement of HRT? Given that thrombophilic defects are relatively common and venous thromboembolism relatively rare, and that a link between HRT related venous thromboembolism and thrombophilia has not yet been established, unselected screening should not be employed before commencing HRT. There is an urgent need for information on the natural history of these thrombophilic defects, the mechanisms of oestrogen associated thrombosis and how these two factors are interlinked. It must be appreciated that the absolute risk of venous thromboembolism remains very low, and clearly the benefits in prevention of ischaemic heart disease and osteoporosis will outweigh the small risk of venous disease in the vast majority of women. However, it should be noted that the cardioprotective effect of such therapy, at least with short term use, has been questioned20. While it may be prudent to discuss with the woman at the start of HRT this small risk of venous thromboembolism, it should be placed clearly in the context of the overall benefits. As venous thromboembolism may be dependent on multiple factors coming together it is important to be alert to the potential for excess risk of venous thromboembolism developing due to the presence of pre-existing factors such as obesity combined with HRT. Risk factors for thromboembolism are readily identifiable21. In these situations HRT might be seen as an additional factor best avoided. Again, it is important to review the overall situation as the benefits from HRT in an obese woman at risk of arterial vascular disease may exceed the risk of venous thromboembolism.

There is no doubt that selective screening should be offered to women with a past or family history of venous thromboembolic events and such a history must be actively sought in considering HRT. An attempt should be made to assess the severity of any previous event and whether or not it was objectively confirmed. Clinical diagnosis of deep vein thrombosis and pulmonary thromboembolism are unreliable and objective testing is required for diagnostic accuracy. A history of prolonged anticoagulant therapy would be compatible with a significant previous event. Where there is no underlying thrombophilia and there is no perceived ongoing risk factor or severe previous event, then HRT can be prescribed, but this should be discussed carefully with the woman in order that she is aware of the potential for a thrombotic problem developing. She should also be advised that if symptoms which could be compatible with deep vein thrombosis or pulmonary embolism arise, she should seek medical advice promptly. Where a severe problem, an underlying thrombophilic trait with previous venous thromboembolism or an ongoing significant risk factor is present, it would appear best to avoid HRT if possible unless the woman is on long term anticoagulant therapy. If it has to be prescribed, then consideration should be given to combining it with an antithrombotic strategy, such as Warfarin or possibly low dose aspirin.

Where there is no underlying history of venous thromboembolism but an underlying thrombophilic trait has been picked up through screening (for example, on the basis of a family history) then consideration should be given to the defect and its clinical manifestations in the family. HRT should be avoided unless concomitant antithrombotic therapy is used in high risk situations such as antithrombin deficiency. However, where a problem such as Factor V Leiden or protein S deficiency occurs in isolation, there is insufficient evidence at present to indicate that HRT should be avoided. However, if it is to be used a clear discussion of the potential excess risk should occur with the woman. As this is a controversial area which is rapidly changing with advancing knowledge it might be best, in the interim, that should a thrombophilic defect be identified that advice be obtained from clinicians with expertise in this area.

Finally, HRT is often seen as a risk factor for post- operative thromboembolism. There are no data to support such a view, although the combination of HRT and the changes in coagulation and venous function following surgery might all operate together to provide a significant increase in risk. This risk is likely to be small, and it would appear that virtually all women who are likely to receive HRT will meet the criteria for thromboprophylaxis in guidelines laid down such as those devised by the THRIFT Consensus Group21. It would perhaps be prudent to add HRT as a risk factor for venous thromboembolism when assessing patients preoperatively. At present there is no evidence to support a policy of routinely stopping HRT before surgery, provided appropriate thromboprophylactic measures are taken. This will consist mainly of low dose or low molecular weight heparin with or without thromboembolic deterrent stockings as set out in previous guidelines21.

In conclusion, there is an urgent need for increased information on oestrogen-related thrombosis and their association with thrombophilic traits. Until further information is available, a pragmatic approach is required. There is currently no indication for nonselected thrombophilia screening before HRT, but it is critical that a past and family history of thrombotic problems be obtained and selected screening performed. An assessment of the woman for risk factors for venous thromboembolism should be undertaken before and during HRT, and the risk clearly placed in the context of the overall risk benefit ratio of HRT. HRT should be avoided in high risk women such as those with antithrombin deficiency, unless concomitant prescription of antithrombotic therapy is used. This latter course of action is likely to require specialist input from clinicans with particular expertise in haemostasis and thrombosis.