- Top of page
- Patients and methods
Summary. As hormone replacement therapy is associated with an early excess risk of venous thrombosis, we investigated the effect of different oral hormone replacement therapies on resistance to activated protein C, and on levels of factor VIII antigen (FVIII:Ag) and factor XI antigen (FXI:Ag). In a prospective, randomized, placebo-controlled 12-week study, 60 healthy post-menopausal women daily received either placebo (n = 16) or 2 mg of micronized 17β-oestradiol, either alone (E2, n = 16) or sequentially combined with dydrogesterone 10 mg (E2 + D, n = 14) or trimegestone 0·5 mg (E2 + T, n = 14). Medication was given orally. Normalized activated protein C sensitivity ratios (nAPCsr) were determined by quantifying the effect of activated protein C on the endogenous thrombin potential. FVIII:Ag and FXI:Ag were determined by enzyme-linked immunosorbent assay. Compared with baseline and placebo, the nAPCsr increased (92% to 142%; all P < 0·001) in all active treatment groups after both 4 and 12 weeks. Compared with placebo, hormone replacement therapy was not associated with significant changes in FVIII:Ag. After 4 and 12 weeks, FXI:Ag levels were significantly decreased in the E2 group (mean percentage changes from baseline versus placebo: −15·0%, P = 0·001 at 4 weeks and −16·6%, P = 0·003 at 12 weeks) and in the E2 + D group (−10·4%, P = 0·02 and −10·4%, P = 0·02). In conclusion, all hormone replacement regimens were associated with a large increase in resistance to activated protein C. In contrast, hormone replacement therapy had no effect on FVIII:Ag. Oral E2 and E2 + D had a small, favourable effect on FXI:Ag.
Resistance to activated protein C has been proposed as an explanation for the higher incidence of venous thrombosis in women using third-generation oral contraceptives than in women using oral contraceptives of the second-generation (Rosing et al, 1997, 1999). Moreover, a recent case–control study showed that the risk of developing venous thrombosis during hormone replacement therapy was significantly increased in women with higher resistance to activated protein C in comparison with hormone users with a lower resistance to activated protein C (Lowe et al, 2000). Two recent randomized placebo-controlled studies, in which the effect of oral combined hormone replacement therapy on resistance to activated protein C was investigated, have shown an unfavourable increase in resistance to activated protein C in the hormone replacement groups (Demirol et al, 2001; Høibraaten et al, 2001a).
With regard to the effect of hormone replacement therapy on FVIII, randomized placebo-controlled studies showed no changes (Høibraaten et al, 2001b) or a favourable decrease (Demirol et al, 2001). To our knowledge, only one small study concerning the effect of hormone replacement therapy on FXI levels has been published (Fossum et al, 1999). In this study, FXI decreased after 6 months of unopposed oral oestradiolvalerate.
So far, no randomized placebo-controlled trials comparing the effect of oestradiol, both alone and combined with other therapies or with placebo, on resistance to activated protein C, FVIII or FXI have been published. Moreover, data on the effect of hormone replacement therapy on FXI are scarce. Therefore, we performed a randomized, placebo-controlled study in healthy post-menopausal women to gain more insight in the effect of oral unopposed and opposed oestrogen replacement therapy on resistance to activated protein C, on FVIII antigen (FVIII:Ag) and on FXI antigen (FXI:Ag) levels.
Study design and population. Healthy, non-hysterectomized post-menopausal women were recruited through advertisements in local newspapers; 65 women were enrolled in this 12-week study, which was performed at the out-patient clinic of the Department of Obstetrics and Gynaecology (Van Baal et al, 1999a,b,c, 2000;Post et al, 2002). The investigation conformed to the principles outlined in the Declaration of Helsinki. Written informed consent was obtained from each participant before entering the study. The Institutional Review Board of the VU University Medical Center approved the protocol.
Participants were between 45 and 60 years, smoked less than 15 cigarettes per day, were normotensive (< 160/90 mmHg), had a body mass index (BMI)≤30 kg/m2 and had been amenorrhoeic for 6 months to 5 years with serum follicle-stimulating hormone concentrations above 20 IU/l and oestradiol concentrations lower than 150 pmol/l. None of the women had received hormone replacement therapy for at least 3 months before entering the study and none took cardiovascular medication. Exclusion criteria included a history of cardiovascular, venous thromboembolic, metabolic, endocrinological, and (pre) malignant disease, as well as clinically relevant abnormalities in laboratory tests of haematological, renal and hepatic function. Women with fasting plasma levels of cholesterol and triglycerides > 8 mmol/l and > 4 mmol/l respectively, were also excluded.
Eligible women were randomly assigned to either placebo (n = 17), or to micronized unopposed 17β-oestradiol, 2 mg/d (E2 group; n = 18), or to sequentially combined hormone replacement therapy consisting of micronized 17β-oestradiol 2 mg/d plus either dydrogesterone 10 mg/d (E2 + D group; n = 15; Femoston®, Solvay Pharmaceuticals, Weesp, The Netherlands), or trimegestone 0·5 mg/d (E2 + T group; n = 15; Hoechst Marion Roussel, Romainville-Cedex, France) given for 14 d of each 28-d cycle. The pharmacy at the VU University Medical Center (Amsterdam, The Netherlands) manufactured placebo and oestradiol as capsules with identical appearance. The tablets of the sequentially combined hormone replacement therapy were put into capsules of identical appearance by Hoechst Marion Roussel (Paris, France). After three treatment cycles, women assigned to treatment with unopposed oestradiol were treated dydrogesterone 10 mg/d for 14 d to induce a withdrawal bleed.
In total, 65 participants were initially enrolled. Five women dropped out before the measurement at 4 weeks and were therefore excluded from the analysis (placebo group: n = 1; E2 group: n = 2; E2 + D group: n = 1; E2 + T group: n = 1). Another three women dropped out between 4 and 12 weeks (placebo group: n = 1; E2 + D group: n = 1; E2 + T group: n = 1). In these three cases, the last-observation-carried-forward procedure was applied for the missing values at 12 weeks. Therefore, the analyses were based on 60 participants. Reasons for drop out have been published previously (van Baal et al, 1999b, 2000; Post et al, 2002) and were not associated with venous thromboembolic disease.
Blood collection and handling of samples. At baseline and after 4 weeks (cycle 1, cycle d 24–28) and 12 weeks (cycle 3, cycle d 24–28) of follow up, venous blood samples were taken between 08.00 hours and 10.00 hours. The subjects had fasted and had refrained from smoking for ≥ 10 h and from consuming alcohol for ≥ 24 h before sampling. After 20 min of rest, blood was collected into tubes containing trisodium citrate (1:9 v/v; 0·129 mol/l) (Becton Dickinson, Meyren, Cedex-France) for measurement of resistance to activated protein C and into identical but cooled tubes for measurement of FVIII:Ag and FXI:Ag. The cooled tubes were immediately placed on ice. Tubes were centrifuged within 1 h of collection at 3000 g and 20°C (resistance to activated protein C) or 4°C (FVIII:Ag and FXI:Ag) for 30 min. Plasma was snap-frozen and stored at −70°C until analysis. Normal pooled plasma was obtained from blood of healthy volunteers (six women not using oral contraceptives and nine men, mean age 33 years) and collected in the same manner as described above for the preparation of plasma samples used for measurement of resistance to activated protein C.
Assays. Normalized activated protein C sensitivity ratios (nAPCsr) were determined with a test (Nicolaes et al, 1997) that quantified the effect of activated protein C on the endogenous thrombin potential (ETP) under the conditions described by Rosing et al (1999). In this assay, an increase in nAPCsr represents an increase in resistance to activated protein C. Eight women had a baseline nAPCsr above 1·50 and, as no DNA was available, they were tested for the factor VLeiden (FVLeiden) mutation with a functional test, which, as compared with the DNA test, has been shown to have 100% sensitivity and 100% specificity (Nicolaes et al, 1996). We used commercially available enzyme-linked immunosorbent assays (ELISAs) to measure FVIII:Ag and FXI:Ag (Kordia Life Sciences, Leiden, The Netherlands). To eliminate interassay variation, all samples of a given marker were assayed in a single run. The intra-assay coefficient of variation was 12·4% for nAPCsr, 2·2% for FVIII:Ag and 3·0% for FXI:Ag.
Serum follicle-stimulating hormone was determined with a specific immunometrical (luminescence) assay (Amerlite, Amersham, Little Chalfont, UK). Serum oestradiol was quantified using a double-antibody radioimmunoassay (Sorin Biomedica, Saluggia, Italy) with a lower limit of detection of 18 pmol/l. Serum total cholesterol was measured automatically (Boehringer Mannheim, Germany). Prothrombin fragment 1 + 2 antigen was determined as described elsewhere (van Baal et al, 2000).
Statistical analysis. Statistical analysis was performed using the statistical package for the social sciences 9.0 (SPSS Inc., Chicago, IL, USA). Baseline characteristics (Table I) are given as mean ± standard deviation (SD) when normally distributed or as median (range) when the distribution was skewed.
Table I. Descriptive characteristics of the four groups at baseline.
| ||Placebo group||E2 group||E2 + D group||E2 + T group|
|Number of subjects||16||16||14||14|
|Age (years)||51·4 ± 3·2||53·1 ± 2·8||52·4 ± 3·5||51·7 ± 2·6|
|Duration of amenorrhoea (months)||21·2 (8·2–55·6)||29·2 (5·8–69·1)||17·2 (6·9–64·2)||20·6 (6·2–56·3)|
|Body mass index (kg/m2)||25·0 ± 3·1||24·3 ± 2·9||26·3 ± 2·7||26·7 ± 2·7|
| Systolic (mmHg)||121 ± 9||119 ± 15||124 ± 15||123 ± 14|
| Diastolic (mmHg)||70 ± 8||66 ± 8||70 ± 11||70 ± 9|
|Smokers (n, %)||2 (13)||6 (38)||2 (14)||2 (14)|
|Serum cholesterol (mmol/l)||6·2 ± 1·0||6·0 ± 1·1||6·3 ± 0·9||6·3 ± 1·1|
|Serum FSH (U/l)||54·3 ± 22·9||52·8 ± 20·0||48·1 ± 17·0||55·1 ± 11·4|
|Serum E2 (pmol/l)||18 (18–244)||18 (18–213)||18 (18–173)||18 (18–86)|
nAPCsr and FVIII:Ag levels are given as median and interquartile range (25th and 75th percentile) and FXI:Ag levels are given as mean ± SD. Standard parametric tests were performed on the three haemostatic variables investigated (for nAPCsr and FVIII:Ag on natural-log-transformed data as they had a skewed distribution). The mean of the individual percentage changes from baseline to 4 and 12 weeks are given as mean and 95% confidence interval (geometric values for nAPCsr and FVIII:Ag). Correlations between variables were calculated with Spearman's correlation coefficient. At baseline, BMI differed slightly among the groups (Table I) and therefore a general linear model for repeated measurements, with the baseline values of the variable under consideration and BMI as constant covariates (ancova), was used for comparisons among and between the groups. A re-analysis of the nAPCsr data was performed after exclusion of women who were apparent carriers of the FVLeiden mutation. A two-tailed P < 0·05 was accepted as the level of significance.
- Top of page
- Patients and methods
Table I shows demographic characteristics of the four groups at baseline. In each group, except for the E2 + T group, one woman had an oestradiol level above 150 pmol/l. However, as this occurred in combination with a follicle-stimulating hormone level above 35 IU/l, we did not exclude these women from the analysis. Baseline median nAPCsr was higher in the E2 + T group in comparison with the other groups, and the median baseline level of FVIII:Ag was higher in the E2 + D group than in the placebo group and the E2 + T group. Baseline levels of FXI:Ag did not differ among the groups (Table II). None of the women tested for the FVLeiden mutation was a homozygous carrier, whereas four women were heterozygous for the FVLeiden mutation; two women in the E2 group, one in the E2 + D group and one woman in the E2 + T group. Baseline nAPCsr of these women ranged from 2·32 to 5·67.
Table II. Normalized activated protein C sensitivity ratios and plasma levels of factor VIII antigen and factor XI antigen during the 12-week study period.
| ||Baseline||4 weeks||12 weeks||%Δ†||ancova‡|
| Placebo||0·67 (0·51–0·92)||0·69 (0·45–1·01)||0·56 (0·42–0·65)||−24·2 (−44·5 to 3·5)||< 0·001|
| E2||0·78 (0·44–1·05)||1·43 (1·12–2·31)||1·78 (1·06–2·25)||112·7 (67·3 to 170·3)***|| |
| E2 + D||0·62 (0·41–1·10)||1·16 (1·00–1·33)||1·48 (0·97–2·22)||118·2 (56·8 to 203·7)***|| |
| E2 + T||1·21 (0·56–1·70)||1·87 (1·18–2·38)||1·71 (1·52–2·32)||106·7 (51·4 to 182·1)***|| |
|Factor VIII:Ag (%)|
| Placebo||80·0 (60·7–92·0)||85·5 (65·0–119·0)||80·0 (62·0–93·0)||4·4 (−9·9 to 20·9)|| 0·22|
| E2||92·0 (60·0–116·0)||92·5 (50·8–117·0)||84·0 (57·0–121·0)||1·2 (−14·1 to 19·1)|| |
| E2 + D||98·0 (84·3–122·0)||92·0 (76·0–137·0)||77·0 (69·0–119·0)||−12·0 (−22·8 to 0·32)|| |
| E2 + T||71·0 (55·8–96·3)||95·0 (83·5–114·0)||86·0 (70·5–96·5)||15·7 (−0·44 to 34·4)|| |
|Factor XI:Ag (%)|
| Placebo||132·6 ± 27·1||139·1 ± 24·2||135·5 ± 26·3||4·6 (−2·7 to 12·0)||< 0·001|
| E2||131·8 ± 19·5||120·4 ± 26·0||117·5 ± 26·3||−12·0 (−20·1 to −3·9)**|| |
| E2 + D||130·4 ± 18·6||124·8 ± 22·4||121·2 ± 19·7||−5·8 (−10·2 to −1·5)*|| |
| E2 + T||135·1 ± 20·8||143·5 ± 24·4||132·9 ± 15·2||2·2 (−3·3 to 7·7)|| |
ancova over the 12-week study period showed significant increases in nAPCsr in all treated groups (overall, P < 0·001; E2 versus placebo, P < 0·001; E2 + D versus placebo, P < 0·001; E2 + T versus placebo, P < 0·001). Differences among the three active treatment groups were not statistically significant. As compared with placebo, the mean individual percentage changes from baseline to 4 and 12 weeks were, respectively, +99·1% (P < 0·001) and +136·9% (P < 0·001) in the E2 group; +92·3% (P < 0·001) and +142·4% (P < 0·001) in the E2 + D group; and +93·4% (P < 0·001) and +130·9% (P < 0·001) in the E2 + T group (Table II; Fig 1). A re-analysis was performed after excluding women who appeared heterozygous for FVLeiden. The results were similar (data not shown).
Figure 1. Normalized activated protein C sensitivity ratios of the four study groups at baseline (closed symbols) and at 12 weeks (open symbols). The arrows indicate values of those women who were heterozygous for FVLeiden. E2, micronized 17β-oestradiol 2 mg/d; D, dydrogesterone 10 mg/d; T, trimegestone 0·5 mg/d.
Download figure to PowerPoint
ancova did not show significant differences in FVIII:Ag among the groups (overall, P = 0·22). However, the percentage changes from baseline to 4 and 12 weeks in the E2 + D group were statistically significantly different from the changes in the E2 + T group (respectively P = 0·01 and P = 0·006). The percentage changes from baseline to 4 and 12 weeks were, respectively, −2·2% and −12·0% in the E2 + D group, and + 22·6% and + 15·7% in the E2 + T group (Table II).
For FXI:Ag, ancova showed a significant difference among the groups (P < 0·001). This could be attributed to significant differences between the E2 and placebo groups (P < 0·001), the E2 + D and placebo groups (P = 0·01), the E2 and E2 + T groups (P < 0·001), and the E2 + D and E2 + T groups (P = 0·01). Compared with placebo, the mean individual percentage changes from baseline to 4 and 12 weeks were, respectively, −15·0% (P = 0·001) and −16·6% (P = 0·003) in the E2 group, and −10·4% (P = 0·02) and −10·4% (P = 0·02) in the E2 + D group (Table II).
We did not find significant correlations (all r-values −0·20≤r≤0·23) between either baseline levels of prothrombin fragment 1 + 2 (a marker of net coagulation activity) and nAPCsr, FVIII:Ag, or FXI:Ag. However, the percentage changes from baseline in nAPCsr correlated significantly with the percentage changes in prothrombin fragment 1 + 2 (r = 0·52, P < 0·001 at 4 weeks; and r = 0·34, P = 0·03 at 12 weeks). The exclusion of women who appeared heterozygous for FVLeiden did not change the results of the correlation analysis.
- Top of page
- Patients and methods
The pathogenesis of venous thromboembolism associated with post-menopausal hormone replacement therapy is unclear (Høibraaten et al, 1999). In this 12-week study in healthy post-menopausal women with no history of venous thrombosis a large, adverse increase (of on average, 100%) in resistance to activated protein C was found in all active treatment groups, whereas no major changes were observed in FVIII:Ag. A small, but favourable, decrease was found in FXI:Ag in the unopposed oestradiol group as well as in the oestradiol plus dydrogesterone group. The changes in resistance to activated protein C and FXI:Ag were already apparent after 4 weeks of treatment.
Resistance to activated protein C can be measured with two functional tests that either quantify the effect of activated protein C on the activated partial thromboplastin time (aPTT-based test; effect of activated protein C on clotting times after initiation of the intrinsic pathway) (Dahlbäck et al, 1993) or on tissue factor-induced thrombin generation (ETP-based test; effect of activated protein C on thrombin generation initiated via the extrinsic pathway) (Nicolaes et al, 1997). In the present study, the ETP-based test was used as this assay is particularly sensitive to changes in the hormonal status of women (Curvers et al, 1999, 2001; Sugimura et al, 1999; Høibraaten et al, 2001a). In addition, it is strongly associated with the risk increase of venous thromboembolism reported in oral contraceptive users and in carriers of the FVLeiden mutation (Rosing et al, 1997; Vandenbroucke & Rosendaal, 1997). The different sensitivities of the two tests for hormonal changes in women have been proposed to be owing to the fact that proteins present in plasma differentially modulate the response to activated protein C in the two tests (Curvers et al, 1999; Tans et al, 2000).
The present study is the first to show, in a placebo-controlled manner, that unopposed oral oestradiol increases resistance to activated protein C in post-menopausal women. In addition, we observed a significant increase in women treated with oestradiol sequentially combined with either dydrogesterone or trimegestone. The latter finding is consistent with previous placebo-controlled randomized studies on the effect of combined hormone replacement therapy (Demirol et al, 2001; Høibraaten et al, 2001a). In contrast, in studies lacking a randomized controlled design no effect of oral hormone replacement therapy on resistance to activated protein C was found (Douketis et al, 2000; Lowe et al, 2000; Winkler et al, 2000). Taken together, the available data from studies with a randomized, placebo-controlled design suggest that resistance to activated protein C increases on all oral oestrogen replacement regimens, either unopposed or opposed.
We cannot exclude the possibility that the combination of oral oestradiol with trimegestone significantly increases FVIII:Ag, at least compared with the combination of oestradiol with dydrogesterone (type II error). The increases in percentage change from baseline to 4 and 12 weeks were +24·8% and +27·7% in the E2 + T group when compared with the E2 + D group. Previous studies were not conclusive (Lindberg et al, 1989; Nabulsi et al, 1993; Kroon et al, 1994; Lowe et al, 2000; Winkler et al, 2000; Demirol et al, 2001; Høibraaten et al, 2001b; Lowe et al, 2001). Only two of these studies were placebo controlled, and showed either no change (Høibraaten et al, 2001b) or a reduction (Demirol et al, 2001).
We found a small decrease in FXI:Ag with unopposed oestradiol, which is in agreement with previous data (Fossum et al, 1999). Our study extends these findings by showing that oestradiol sequentially combined with dydrogesterone had the same favourable effect, whereas oestradiol plus trimegestone did not. We have no clear explanation for this difference in effect. Dydrogesterone, a retro-progestogen, is closely related to natural progesterone. It has a high affinity for progesterone receptors, has antioestrogenic activity, but has no oestrogenic or androgenic activity (Foster & Balfour, 1997). Trimegestone (17β-[(S)-2-hydroxypropanoyl]-17-methyl-estra-4,9-dien-3-one) is one of the active metabolites of promegestone, and is a norpregnane progestogen, which has a high relative binding affinity for the progesterone receptor, antioestrogenic activity (Lundeen et al, 2001) and weak antiandrogenic activity (Zhang et al, 2000). The dosages of these progestogens used were based on their efficacy with regard to protection of the endometrium and have not been chosen on the basis of effects on haemostatic variables. Differences in the effect of dydrogesterone and trimegestone have also been observed on several cardiovascular risk factors (Van Baal et al, 1999b; Post et al, 2002). Regardless of why some hormone replacement therapy regimens affect FXI and others do not, the small changes we found are unlikely to affect risk of venous thrombosis in a clinically significant manner (De Visser et al, 1999).
In our study, the nAPCsr doubled in the hormone users. This so-called acquired activated protein C resistance may be an important risk factor for the occurrence of venous thrombosis in users of hormone replacement. In the Leiden Thrombophilia Study, a 50% lower nAPCsr, as determined in an aPTT-based test (similar to the doubled nAPCsr in our study), was associated with a 2·5-fold increase in the risk of venous thrombosis. Previously, we have shown that both oestradiol alone as well as oestradiol combined with dydrogesterone or trimegestone significantly increased prothrombin fragment 1 + 2 (Van Baal et al, 2000), a determinant of in vivo coagulation activity. As changes in prothrombin fragment 1 + 2 correlated with changes in nAPCsr, this suggests that the observed increased coagulation activity is partly a result of the increased resistance to activated protein C.
In conclusion, this randomized, placebo-controlled 12-week study, in which we investigated the effect of oral 17β-oestradiol either unopposed or sequentially combined with dydrogesterone or trimegestone in healthy post-menopausal women, showed a large, unfavourable increase of about 100% in resistance to activated protein C in all treatment groups. We speculate that this large increase in nAPCsr may, at least in part, explain the early increased risk of venous thrombosis during oral hormone replacement therapy, as other prothrombotic changes in users of hormone replacement therapy are usually much smaller.