Recruitment of the cohort of families with hereditary deficiencies of protein S, protein C, or antithrombin.
Event-free survival in protein S–, protein C–, and antithrombin-deficient and nondeficient female relatives who ever or never used combined oral contraceptives and in deficient male relatives. The sample sizes represent women at risk. Group 1 indicates deficient men; 2, deficient ever users; 3, deficient never users; 4, nondeficient never users; and 5, nondeficient ever users.
van Vlijmen EFW, Brouwer JP, Veeger NJGM, Eskes TKAB, de Graeff PA, van der Meer J. Oral Contraceptives and the Absolute Risk of Venous Thromboembolism in Women With Single or Multiple Thrombophilic DefectsResults From a Retrospective Family Cohort Study. Arch Intern Med. 2007;167(3):282-289. doi:10.1001/archinte.167.3.282
The risk of venous thromboembolism (VTE) in women taking combined oral contraceptives (COCs) is attributed to changes in coagulation and fibrinolysis. Their impact may be greater in women with preexistent thrombophilic defects.
We assessed the effects of COCs on absolute VTE risk in women with single or multiple thrombophilic defects in a retrospective family cohort study. Female relatives of probands with VTE and hereditary deficiencies of protein S, protein C, or antithrombin were tested for known thrombophilic defects, including the index deficiency. Absolute incidences of VTE were compared in deficient vs nondeficient women, in deficient and nondeficient women who ever or never used COCs, and in deficient and nondeficient women with 0, 1, or more than 1 other thrombophilic defect during exposure to COCs.
Of 222 women, 135 (61%) ever used COCs. Overall, annual incidences of VTE were 1.64% and 0.18% in deficient and nondeficient women, respectively; the adjusted relative risk was 11.9 (95% confidence interval, 3.9-36.2). The risk was comparable in deficient ever and never users (1.73% vs 1.54%). Annual incidences during actual COC use were 4.62% in deficient women and 0.48% in nondeficient women; the relative risk was 9.7 (95% confidence interval, 3.0-42.4). The incidence increased by concomitant thrombophilic defects, from 3.49% to 12.00% in deficient women and from 0% to 3.13% in nondeficient women.
Women with hereditary deficiencies of protein S, protein C, or antithrombin are at high risk of VTE during use of COCs, particularly when other thrombophilic defects are present. They have VTE at a younger age, but the overall risk is not increased by COCs.
Since their introduction in 1960, combined oral contraceptives (COCs), containing a combination of ethinyl estradiol and a progestagen, are associated with an increased risk of venous thromboembolism (VTE).1,2 Despite an up to 10-fold reduction in the dose of both components, this risk, albeit lower, has persisted.3 Up to 1995, the risk of VTE was attributed to ethinyl estradiol, because this component is identical in all COCs, while the type of progestagen included may differ. It is the opinion that the type of progestagen may also influence the VTE risk, because the use of third-generation COCs (containing desogestrel or gestodene) in healthy women was reported to result in a higher risk than the use of second-generation COCs (containing levonorgestrel).4- 6 Nevertheless, because of the low baseline risk of VTE in healthy women of fertile age, ranging from 1 per 100 000 years at the age of 15 years to 10 per 100 000 years at the age of 44 years,7,8 the absolute risk of VTE in COC-using women is low (20-40 per 100 000 pill-years).4- 6
The biological background of this association is complex. Combined oral contraceptives induce changes in coagulation and fibrinolysis but, in healthy women, hemostatic variables remain within their normal ranges.9,10 It is likely that these changes may have more impact in women who are already at increased risk of VTE because of preexisting thrombophilic defects. Although several observational studies11- 17 reported on the interaction of COCs and various single thrombophilic defects, little is known about the absolute risk of VTE attributable to COCs in women with single or combined thrombophilic defects.16,17 Recently, aggregation of multiple thrombophilic defects in families with hereditary deficiencies of protein S, protein C, or antithrombin was demonstrated, and it was shown that the risk of VTE strongly depended on the presence of 1 or more other thrombophilic defects.18 In the present study, we analyzed the data from this large family cohort to assess the effects of COCs on the absolute risk of VTE in protein S–, protein C–, or antithrombin-deficient women and the contribution of other known thrombophilic defects.
The original retrospective study18 was performed in a cohort of protein S–, protein C–, and antithrombin-deficient families. Briefly, probands were consecutive patients with documented VTE, in whom 1 of these deficiencies was demonstrated. First-degree relatives older than 15 years were identified by pedigree analysis. Because the number of antithrombin-deficient probands was small, second-degree relatives with a parent who was deficient were also identified. Relatives were enrolled after informed consent was obtained. Information on previous episodes of VTE, exposure to exogenous risk factors for VTE, and anticoagulant treatment was collected by physicians at the thrombosis outpatient clinic of our hospital, using a validated questionnaire,19 and by reviewing medical records. In women, COC use and obstetric history were documented in detail. Blood samples were taken after collection of clinical data. All relatives were tested for additional thrombophilic defects, including a second deficiency of protein S, protein C, or antithrombin; factor V Leiden, or prothrombin G20210A; increased levels of factor VIII, IX, and XI; lupus anticoagulant; and hyperhomocysteinemia. In the present study, we analyzed female relatives of fertile age from the family cohort. Male relatives of comparable age were used as a reference group. The study was approved by the institutional review board of University Medical Center Groningen.
Venous thromboembolism was considered established when diagnosed by compression ultrasonography or venography (deep vein thrombosis) or by ventilation and perfusion lung scanning, spiral computed tomographic scanning, or pulmonary angiography (pulmonary embolism); or when the patient had received full-dose heparin and vitamin K antagonists for at least 3 months without objective testing when these techniques were not yet available. Venous thromboembolism was classified as secondary when occurring within 3 months after exposure to exogenous risk factors, including surgery, trauma, immobilization for 7 days or more, COC use, hormone therapy, pregnancy and puerperium, and malignancy. In the absence of these risk factors, VTE was defined as primary. Superficial phlebitis was not counted as a thrombotic event, because it was usually not confirmed by an objective technique and consequently might not be accurately classified in the setting of this retrospective study.
Protein S and protein C antigen levels were measured by enzyme-linked immunosorbent assay using reagents from DAKO, Glostrup, Denmark. The activity of protein C (Berichrom Protein C; Dade Behring, Marburg, Germany) and antithrombin (Coatest; Chromogenix, Mölndal, Sweden) was measured by chromogenic substrate assays. Levels of protein S, protein C, and antithrombin were expressed as percentages of the levels measured in pooled plasma set at 100%. Normal ranges were determined in 393 healthy blood donors without a (family) history of VTE, who were neither pregnant nor used COCs in the 3 months before blood sampling. Protein S deficiency type I was defined by total protein S levels below the lower limit of its normal range (<67%), and protein S deficiency type III by lowered free protein S levels (<65%), but normal total protein S levels. After we had demonstrated that protein S deficiency type III was not a risk factor for thrombosis in these families, we excluded families with this deficiency from the analysis.20 Protein C deficiency types I and II were defined by lowered levels of either protein C antigen (<63%) or activity (<64%), and antithrombin deficiency by lowered levels of antithrombin activity (<74%). Deficiencies were considered inherited when confirmed by measurement of a second sample collected after a 3-month interval and demonstrated in at least 2 family members. Factor V Leiden and prothrombin G20210A were demonstrated by polymerase chain reactions.21,22 Factors VIII:C, IX:C, and XI:C were measured by 1-stage clotting assays (Amelung, Lemgo, Germany) and considered increased at levels greater than 150%. Lupus anticoagulant was demonstrated by abnormal dilute Russell viper venom time and activated partial thromboplastin time or tissue thromboplastin inhibition, normalized by adding phospholipids.23 Fasting and post–methionine-loading levels of homocysteine were measured by high-performance liquid chromatography.24 Hyperhomocysteinemia was defined as a fasting homocysteine level greater than 2.50 mg/L (>18.5 μmol/L) and/or a postloading level greater than 7.95 mg/L (>58.8 μmol/L), as described in the Dutch population.25 In probands and symptomatic relatives, blood samples were collected at least 3 months after VTE had occurred. If they were then still treated with a short-acting vitamin K antagonist (acenocoumarol), samples were taken after temporary change of this therapy into subcutaneous nadroparin treatment for at least 2 weeks.
We compared the absolute risk of a first episode of VTE in deficient and nondeficient female relatives, who were divided into women who ever or never used COCs during their fertile lifetime. Probands were excluded from analysis to avoid bias. Annual incidences were calculated by dividing the number of first episodes of VTE by the number of person-years. Person-years were counted from the age of 15 years until the age of 50 years, the first episode of VTE, or the end of the study period. A minimum age of 15 years was chosen because VTE is rare at a younger age, and a maximum age of 50 years was chosen as the end of a fertile lifetime. Event-free survival was analyzed by the Kaplan-Meier method. In this analysis, male-deficient relatives were used as a reference group.
In addition to the overall risk of VTE during fertile lifetime in ever and never users, we calculated the risk of VTE related to actual COC exposure in deficient and nondeficient women. In this calculation, person-years was defined as the number of pill-years, including a 3-month exposure window after COC use was discontinued.
Moreover, we estimated the contribution of concomitant thrombophilic defects in deficient and nondeficient actual COC users. Concomitance was classified as no other thrombophilic defect, 1 other thrombophilic defect, and more than 1 other thrombophilic defect.
Crude relative risks (RRs) were based on annual incidences, and 95% confidence intervals (CIs) were calculated using the binomial probability model (conditional small sample approach).26 To account for clustering of women within families, outcome rates were analyzed by random-effects logistic regression with gaussian distribution, resulting in adjusted RRs and 95% CIs.
Continuous variables were expressed as mean values and standard deviation or median values and range, and categorical data as counts and percentages. Differences between groups were evaluated by the t test or the Mann-Whitney test, depending on the normality of data, for continuous data and by the Fisher exact test for categorical data. A 2-sided P<.05 indicated statistical significance.
Analyses were performed using SAS statistical software, version 8.2 (SAS Institute Inc, Cary, NC), and Stata software, version 9.1 (Stata Corp, College Station, Tex).
The original family cohort contained 91 unrelated families from 39 protein S–deficient probands, 40 protein C–deficient probands, and 12 antithrombin-deficient probands, including 263, 277, and 185 relatives, respectively (Figure 1). Seven families (19 relatives) were excluded because inheritance of the deficiency in the probands could not be established by testing their relatives. The mean (SD) age at the time of VTE was 29 (11) years in the 49 female probands and 38 (14) years in the 35 male probands. Of the female probands, 43 (88%) had secondary VTE (14 because of COC use, 6 because of pregnancy, 16 because of puerperium, and 7 because of trauma, surgery, or immobilization), compared with 8 (23%) of male probands (because of trauma, surgery, or immobilization). Overall, 343 relatives were female, of whom 121 could not be enrolled because of prior death (n = 50), being 15 years or younger (n = 59), no consent (n = 9), or geographic distance (n = 3). Death was possibly related to VTE in 5 women, of whom 2 died at a fertile age. The response rate of eligible women was 95%. The remaining 222 women were analyzed. Their characteristics are summarized in Table 1. One hundred one (45%) of these women were deficient. Fifty-six deficient and 79 nondeficient women had ever used COCs. The mean age at the start of COC use was 22 years. Most women (78%) had only 1 period of COC use. The median duration of COC use was 5 years (range, 0.03-27 years). Thirty-four events were reported, of which 31 (91%) were classified as secondary; 16 events were related to COC use, and 11 to pregnancy or puerperium.
Overall, the annual incidence of VTE was 1.64% vs 0.18% in deficient vs nondeficient women; the adjusted RR was 11.9 (Table 2). The highest risk of VTE was observed in antithrombin-deficient women (2.06%), compared with 1.89% in protein C–deficient women and 1.01% in protein S–deficient women. Antithrombin nondeficient women had the lowest risk (0%); the risk was 0.17% and 0.31% in protein C and protein S nondeficient women, respectively.
The annual incidences were 1.73% vs 0.16% in deficient vs nondeficient ever users of COCs, and 1.54% vs 0.21% in deficient vs nondeficient never users. There was no significant difference between deficient ever and never users (adjusted RR, 1.3; 95% CI, 0.5-3.8). Fifty-five percent of ever users and 58% of never users experienced 1 or more pregnancies. In deficient ever users, 80% of VTE episodes were related to use of COCs, whereas in deficient never users, 67% were related to pregnancy. Event-free survival analysis showed that there were no differences between deficient ever users, deficient never users, and deficient men at the age of 50 years. First events occurred earlier in deficient ever users and, although less pronounced, in deficient never users than in deficient men (Figure 2).
The annual incidence of VTE during actual COC use was 4.62% in deficient COC users, compared with 0.48% in nondeficient COC users (Table 2). The annual incidences were 2.42%, 7.06%, and 5.14% in protein S–, protein C–, and antithrombin-deficient women, respectively, compared with 1.01%, 0.29%, and 0% in nondeficient COC users, respectively (Table 2).
Details of the 16 women who experienced a first episode of VTE while using COCs are presented in Table 3. Thirteen women (81%) were protein S, protein C, or antithrombin deficient. Concomitance of 1 or more other thrombophilic defects was demonstrated in 10 (77%) of the 13 deficient women, and in all 3 nondeficient women. Thirteen women (81%) were first-ever users of COCs. The mean age at onset of VTE was 24 years. Of the women, 69% experienced their first event within 6 months of the start of COC use.
Exposure to third-generation COCs was associated with a higher risk of VTE than exposure to second-generation COCs. Overall, annual incidences were 5.43% vs 1.19%; the RR was 4.5 (95% CI, 1.4-15.7). In deficient women, these were 6.85% and 4.35%, respectively (RR, 1.6; 95% CI, 0.4-6.6); in nondeficient women, 3.57% and 0.31%, respectively (RR, 11.7; 95% CI, 0.9-344.4). However, risk estimates might not be accurate because of the inability to retrieve the brands used in 19 women.
Concomitance of other thrombophilic defects was demonstrated in 57% of deficient women and in 70% of nondeficient women (Table 1). Of these women, a second deficiency of protein S, protein C, or antithrombin, and lupus anticoagulant was not found. The remaining concomitant defects were equally distributed among deficient and nondeficient women. The annual incidences of VTE in deficient actual COC users with 0, 1, or more than 1 concomitant defects were 3.49%, 5.56%, and 12.00%, respectively, compared with 0%, 0.31%, and 3.13% in nondeficient actual COC users, respectively (Table 4).
This study shows a high absolute risk of VTE in protein S–, protein C–, and antithrombin-deficient women, compared with their nondeficient female relatives. Combined oral contraceptives hardly contributed to the overall risk during fertile lifetime. Cumulative event rates at the age of 50 years were similar in women, either ever or never users of COCs, and in men. However, event-free survival curves showed VTE to occur earlier in deficient ever users and, although less pronounced, in deficient never users than in deficient men. These differences between women and men are most likely because of COC use in deficient ever users and because of pregnancy in deficient never users. Therefore, COCs and pregnancy resulted in the occurrence of VTE in deficient women at a younger age, but did not increase the overall risk of VTE during fertile lifetime.
The absolute risk of VTE in deficient women during actual COC use was almost 10-fold higher than in nondeficient COC users (4.62% vs 0.48%). This risk is 115- to 230-fold higher than the published absolute risk in healthy women of fertile age during actual exposure to COCs (0.02%-0.04%).4- 6 The 12- to 24-fold higher risk in nondeficient COC users is explained by the presence of other thrombophilic defects (described later).
Several studies have evaluated the contribution of COC use to the risk of VTE in women with thrombophilic defects. Most studies were addressed to factor V Leiden or prothrombin G20210A, both more prevalent but milder risk factors for VTE than the previously mentioned deficiencies.11- 15 Only 2 studies16,17 reported on the absolute risk of VTE in protein S–, protein C–, and antithrombin-deficient women. The first, a retrospective case-control study,16 compared deficient females who used COCs with those who did not use COCs. The incidence of VTE was increased only in antithrombin-deficient COC users. A comparison with our results is hampered by differences in design because this study also included probands, who by definition had experienced VTE, and patients with prior VTE, whereas superficial phlebitis was counted as an event. The second, a family cohort study,17 reported incidences of 4.3% and 0.7% per year in deficient and nondeficient female relatives who used COCs, respectively, in line with our results.
We demonstrated 1 or more other thrombophilic defects in 57% of deficient and in 70% of nondeficient relatives. These prevalences were higher than is expected in the general population, particularly for factor V Leiden, prothrombin G20210A, increased factor VIII plasma levels, and hyperhomocysteinemia. Apparently, these defects aggregated in these families.
The risk of VTE in deficient women during COC use strongly increased in the presence of 1 or more other thrombophilic defects. The annual incidences of VTE increased from 3.49% (no defects) to 12.00% (>1 concomitant defect). In nondeficient women who used COCs, other thrombophilic disorders increased the risk from 0% (no defects) to 3.13% (>1 defect). The combination of a deficiency with more than 1 other thrombophilic defect exceeded the reported risk of VTE during COC use in healthy women up to 300- to 600-fold. This finding indicates a synergistic interaction of various thrombophilic defects and COC use. Two previous studies27,28 that evaluated the risk of multiple thrombophilic factors in carriers of factor V Leiden or prothrombin G20210A showed a similar effect.
Based on our results, we recommend that the use of COCs be strongly discouraged in women with inherited deficiencies of protein S, protein C, or antithrombin. Even in their nondeficient female relatives, this might be considered, unless they have negative test results for all known thrombophilic defects. Because 45% of female relatives were deficient and 70% of nondeficient women had 1 or more other thrombophilic defects, one might suggest not to use COCs to all female relatives of patients with deficiencies, without further testing. Despite an excessive risk of VTE during the use of COCs in women with deficiencies, the impact of this finding is limited because the prevalence of these deficiencies is low. However, it is likely that COCs will also increase the risk of VTE in women with other single or combined thrombophilic defects.27,28 Although the latter are milder risk factors, their contribution to the risk of VTE in the female population may be greater because they are more prevalent. We speculate that COC-related VTE observed in a small proportion of the female population might be attributed to a preexistent higher risk of VTE as a result of single or combined thrombophilic defects. If so, COCs can safely be prescribed to the remaining majority of women without these defects. We agree that it is neither feasible nor cost-effective to test all women before the use of COCs. In clinical practice, however, testing for hereditary deficiencies of protein S, protein C, and antithrombin should be considered in women with a family history of VTE, especially when VTE had occurred at a young age and was related to the use of COCs.29
This retrospective study has several limitations. The small sample size resulted in wide CIs and unstable estimates. Several events were not established by objective techniques, because these were not available yet. Several relatives were not tested for all concomitant thrombophilic defects, mainly because of insufficient amounts of plasma collected at enrollment. The concomitance of thrombophilic defects was classified by the number of defects, although the thrombotic potency of different defects may vary. However, subgroups of relatives with specific combinations were numerous and consequently too small to allow accurate estimates. We tested 1600 patients to obtain 91 probands, corresponding with a prevalence of hereditary deficiencies of 5.7%, in line with previous series30 of unselected patients with VTE. Therefore, referral bias seems unlikely. Selection bias was avoided by testing consecutive patients with VTE and an extraordinarily high response rate of relatives. Because only a few women died of VTE at a fertile age, the results of this study will not be influenced by an excess of fatal events in deficient relatives. Finally, the risk of VTE might have been overestimated by selecting symptomatic deficient patients, but asymptomatic deficient subjects and their relatives are not identified in clinical practice. Despite these limitations, to our knowledge, this is the first study to address the multicausality of VTE in COC-using women, including all known thrombophilic defects and deficiencies. Further studies are warranted to establish our findings.
In conclusion, the high risk of COC-related VTE in women with hereditary deficiencies of protein S, protein C, or antithrombin strongly depends on frequent concomitance of other thrombophilic defects. These women have VTE at a younger age, but COCs do not increase the overall absolute risk of VTE over fertile lifetime.
Correspondence: Jan van der Meer, MD, PhD, Division of Haemostasis, Thrombosis and Rheology, Department of Hematology, University Medical Center Groningen, Hanzeplein 1, 9713 GZ Groningen, the Netherlands (email@example.com).
Accepted for Publication: October 12, 2006.
Author Contributions:Study concept and design: van Vlijmen and van der Meer. Acquisition of data: van Vlijmen and Brouwer. Analysis and interpretation of data: van Vlijmen, Veeger, Eskes, de Graeff, and van der Meer. Drafting of the manuscript: van Vlijmen. Critical revision of the manuscript for important intellectual content: Brouwer, Veeger, Eskes, de Graeff, and van der Meer. Statistical analysis: Veeger. Administrative, technical, and material support: van Vlijmen, Brouwer, and van der Meer. Study supervision: van der Meer.
Financial Disclosure: None reported.