Millions of women worldwide use exogenous estrogens for hormonal contraception or postmenopausal hormone therapy (HT) (1, 2). Estrogen in combination with a progestogen effectively prevents pregnancy by suppressing pituitary gonadotropin release and subsequent ovulation. During menopause, HT is useful for preventing short-term symptoms related to estrogen deficiency as well as long term effects, including changes in bone and cardiovascular health (3). Estrogens used for hormonal contraception generally differ from those used for HT. Ethinyl estradiol (EE) is the predominant estrogen used for hormonal contraception, whereas micronized 17β-estradiol (E₂) and conjugated equine estrogens (CEEs) are widely used for HT. In recent years, E₂ and estradiol valerate (E₂V) have also been used in hormonal contraceptive formulations (4, 5); E₂V is rapidly converted to E₂ during the hepatic first pass. Recently, the novel estrogen, estetrol (E₄), has emerged as a promising option for both hormonal contraception and postmenopausal HT (6). The chemical structures of the estrogens used for hormonal contraception and HT are depicted in Figures 1 and 2 .

The most common route of administering estrogens is oral, although different estrogen formulations may offer other routes of administration that can be tailored to patients’ individual needs (5). E₂ can be administered by a variety of different routes, including oral, buccal, sublingual, intranasal, transdermal, vaginal, rectal, intramuscular or subcutaneous injection, and as a subcutaneous implant. CEEs are administered either orally or vaginally. The estrogens are also available in a variety of doses (5).

It is well established that estrogens increase the risk of both arterial and venous thrombosis. Abnormally high levels of some coagulation factors in combination with a decrease in anticoagulation factors contribute to thrombotic risk (7, 8). Following oral administration of estrogen there is a dramatic increase in estrogen-sensitive hepatic proteins, as the highly concentrated estrogen in splanchnic blood is presented to hepatocytes during the hepatic first pass. A variety of proteins are altered, including markers of coagulation, anticoagulation, and fibrinolysis. The exact mechanism by which this occurs is complex and is not understood entirely.

The purpose of the present chapter is to show how different estrogens impact hemostasis in premenopausal and postmenopausal women. We identified articles included in this review by searching in PubMed databases using the following search terms: “estrogens”, “estradiol”, “hemostasis”, “thromboembolism”, and “venous thromboembolism”; all studies were in English.

Estrogens mediate physiologic processes by genomic and nongenomic mechanisms. Both mechanisms involve the binding of estrogens to estrogen receptors (ERs). There are 2 primary ERs, ERα and ERβ, encoded by different genes; ERα is encoded by the ESR1 gene on chromosome 6, whereas ERβ is encoded by the ESR2 gene on chromosome 14 (9). The 2 ERs share common structural components, including DNA binding, ligand-binding, N-terminal, and C-terminal domains. The 2 isoforms vary predominantly in their N-terminal domains and ligand-binding domains. Both ERs are expressed in a wide range of tissues in the body, though their distributions vary across different tissue types and change throughout the lifespan. Recently, a third type of ER has been gaining considerable interest, namely, the G protein coupled estrogen receptor (GPER) (10). In contrast to ERα and ERβ, GPER is a plasma membrane receptor. The physiologic effects of GPER activity are still under study.

The genomic (or classical) mechanism of estrogen action refers to the slow estrogen pathway that takes place over several hours or days (11). In this pathway, estrogen diffuses across the cell membrane into the cytoplasm where it binds to the ERs. The ER complex undergoes a series of coordinated steps that include phosphorylation, homodimerization, and allosteric conformational changes. The activated complex then translocates to the nucleus where it recruits regulatory cofactors (coactivators), resulting in transactivation. Transactivation involves binding of the activated complex to a specific region of DNA, referred to as the estrogen response element (ERE), resulting in the synthesis of messenger RNA (mRNA) by a process called transcription. Afterwards, the mRNA translocates to the ribosomes in the cytoplasm where the genetic code is translated and specific proteins are synthesized. Transcription can also be inhibited by transrepression, in which the ER complex interacts with a corepressor protein, thereby altering recruitment of coactivators, and leading to gene-specific repression.

In contrast to genomic mechanisms of estrogen action, non-genomic mechanisms involve rapid estrogen signaling that takes place within seconds or minutes (11). This occurs via membrane-associated ERs or other cell surface receptors that can activate rapid downstream pathways, including kinase signaling.

Human genome-wide searches for high affinity EREs have demonstrated that they occur in many of the genes belonging to the procoagulant and anticoagulant pathways (12). They include the hepatic-specific coagulation factors II, V, VIII, IX, X, XI, and XII, as well as the anticoagulants protein S and protein C.

Since circulating estrogens affect concentrations of coagulation and fibrinolytic factors, it is important to know their pharmacokinetics, and how pharmacokinetic parameters differ among various estrogens. Much of our knowledge of estrogen pharmacokinetics is based on serum or plasma levels measured by immunoassay methods (13). Although radioimmunoassays (RIAs) with preceding purification steps have provided valuable data on the pharmacokinetics of estrogens, they lack the specificity and sensitivity of mass spectrometry (MS) assays, which are now considered the gold standard for steroid hormone measurements. In addition, direct immunoassays without a preceding purification step have been used in estrogen pharmacokinetic studies. Such assays can overestimate circulating estrogen levels grossly, making it difficult to establish accurate pharmacokinetic parameters (13). Nevertheless, a general idea of how pharmacokinetic parameters differ among the different estrogens used for contraception or HT can be obtained, as summarized below.

Two clinically important pharmacokinetic parameters of a drug are its bioavailability and half-life. The bioavailability of a drug is the extent to which it enters the systemic circulation after undergoing hepatic first-pass metabolism. The half-life of a drug is the time required for a drug’s blood level to fall to 50% of its maximal level. The bioavailability of E₄ is high (14), which is likely due to its limited metabolism. In contrast, EE has a moderate bioavailability (on average, 40-45%) (15), and the bioavailability of E₂ is very low (<2-10%) (16), due to its extensive hepatic first-pass metabolism. As for the bioavailability of E₂ in CEEs, it is not determined since it comprises only 1-2% of the dose. Following oral administration of CEE, E₂ is formed predominantly from estrone sulfate (E₁S), the predominant component of CEE, during hepatic first pass metabolism.

In the circulation, most of the EE and E₄ are loosely bound to albumin, whereas about 61% of E₂ is bound to albumin and 37% with high affinity to sex hormone binding globulin (SHBG) (17). A small percentage (1-2%) of E₂ is non-protein-bound (free). Only the free steroid fraction can enter cells and either undergo metabolism or exert biologic effects (free hormone hypothesis) (18). It is important to note that equine estrogens in CEEs also bind to SHBG and albumin in a manner similar to that of E₂ (19).

The oral bioavailability of different estrogens depends on absorption within the digestive system. E₄ is absorbed very rapidly and reaches a median time (t_max) to reach maximum concentration (C_max) of 0.25-0.5 h following dosing with 1, 10, or 100 mg E₄ in postmenopausal women; the levels then fall sharply (14). EE is also absorbed rapidly, with maximum concentrations attained between 1–2 h following dosing with 0.03 mg EE in premenopausal women (20). In contrast, following oral administration of 1 or 2 mg E₂, circulating E₂ levels rise gradually and C_max is attained around 5 h; the levels are still elevated at 8 h (21). As for orally administered CEE (2 x 0.625 mg), the estrogenic components show slow absorption, with a t_max of 5–9 h (22).

Reported terminal half-lives (t_½) for the different orally administered estrogens are: 5–30 h for EE (23); 13–20 h for E₂ (24); and 28 h for E₄ (25). For key metabolites of CEE, reported mean values include 17.1, 19.1, 11.5, and 13.3 h for E₁, E₂, equilin, and 17β-dihydroequilin, respectively (26).

The potency of an estrogen refers to its strength. It can be defined as a measure of the doses of 2 different drugs required to produce the same pharmacologic effect. The relative potencies of orally administered E₂, CEE, and EE were determined in a study in which their effects on follicle-stimulating hormone (FSH) and estrogen-sensitive hepatic proteins (SHBG, CBG, and angiotensinogen) were evaluated in postmenopausal women who were treated with different doses of these estrogens (27). The results show that on a weight basis EE was by far the most potent estrogen. Compared to E₂, the relative potency of EE was 614, 500, and 331 times greater with regard to the responses of SHBG, CBG, and angiotensinogen, respectively. In contrast, the relative potency of CEE compared to E₂ was 3.2, 1.3, and 5 times greater with regard to the respective hepatic proteins. It is important to realize that ring B unsaturated estrogens such as equilin, equilenin, 17β-dihydroequilin, and 17β-dihydroequilenin are biologically active, and contribute to the estrogenicity of CEE. In fact, 17β-dihydroequilin has a higher relative binding affinity for ERα and ERβ than E₂ (28). The pharmacologic effects of CEE are the result of the sum of the individual activities of its components (29).

The high estrogenic potency of EE is also evident in studies in which it was administered parenterally. Treatment of premenopausal women with a transdermal patch delivering 20 µg/d of EE and 150 µg/d of norelgestromin resulted in significant increases from baseline in serum levels of SHBG, CBG, TBG, and hs-CRP (30). The levels of SHBG and TBG were significantly greater than those associated with a combined oral contraceptive (COC) containing 35 µg EE/250 µg norgestimate, which was also administered in the same study. In addition, it has been shown that vaginal delivery of 15 µg EE combined with 150 µg of segesterone acetate from a contraceptive ring in premenopausal women did not reduce EE-associated increase in hepatic proteins (31). In contrast to these studies, transdermal E₂ treatment in postmenopausal women has negligible effects on SHBG, CBG, and TBG (32).

As for E₄, based on limited data it appears that this estrogen has a considerably lower impact on estrogen-sensitive hepatic proteins. In a study comparing the effect of a COC containing either E₄ combined with drospirenone or EE in combination with drospirenone for 6 cycles, E₄ combined with drospirenone had a significantly lower impact (+ 55%) on SHBG compared to the other formulation (+ 251%) (33).

The process of hemostasis involves several factors that act simultaneously to create a balance between coagulation and anticoagulation. Dysregulation of this system may increase the risk of arterial or venous thrombosis and subsequent tissue ischemia. Although estrogens are known to affect multiple hemostatic markers, the exact molecular mechanism of estrogen-induced thrombosis is not clear. However, it is possible that small changes in those markers by different types, routes, and/or doses of estrogens may increase the risk of thrombosis.

Most studies on the effect of estrogens on hemostasis have been carried out in premenopausal women using COCs; studies in postmenopausal women using HT are limited. The studies with COCs have been carried out predominantly in combination with a progestin. Most of the studies have involved EE-based COCs and considerably fewer studies exist with E₂-based COCs. The effects of E₄-based COCs on hemostasis and risk for thrombosis have been studied only recently, and long-term large-scale studies are still lacking. Also, limited data on the effects of oral E₂ and CEE, and transdermal E₂ used for menopausal HT, on hemostatic parameters exist. Nevertheless, some conclusions can be drawn from those studies regarding how the estrogens compare with respect to their effects on coagulation and fibrinolysis. Studies related to the effect of estrogens on hemostasis among COC and menopausal HT users are discussed separately.

EE is most commonly used in COCs and has been shown to have a dose-dependent effect on thrombosis risk (34). These findings have led to a decrease in prescribed dosages and exploration of alternative estrogen formulations for oral contraception. Agren and coworkers (35) compared the effects of 6 cycles of either E₂ (1.5 mg)/nomegestrol acetate (NOMAC) (2.5 mg) or EE (30 µg)/levonorgestrel (LNG) (150 µg) on hemostatic parameters in 121 premenopausal women. The COC with LNG was chosen because LNG is able to partially counteract the effects of EE on coagulation. Minimal changes from baseline to the end of either treatment were found in the 2 groups in prothrombin, activated factor VII, coagulated activated factor VII, and factor VIII. However, there were significant differences between the 2 groups in the anticoagulant indicators antithrombin III, total protein S, and protein C, and minor differences in the activated partial thromboplastin time (aPTT)-based activated protein C (APC) sensitivity ratio and in free protein S. The EE/LNG group had a substantial significant increase from baseline in the endogenous thrombin potential (ETP)-based APC sensitivity ratio. There was also a small increase in the prothrombin fragment 1 + 2 (F1 + 2) in the EE/LNG group but not in the E₂/NOMAC group, whereas D-dimer did not change in either group. While CRP levels remained within the normal range of values for premenopausal women, there was a greater percent change from baseline in the EE/LNG group.

Another double-blind, randomized study by Gaussem and coworkers (36) compared a lower dose of EE (20 µg)/LNG (100 µg) to E₂ (1.5 mg)/NOMAC (2.5 mg) among healthy reproductive aged women in 3 consecutive cycles. Most of the differences in measured coagulation and fibrinolysis parameters between the 2 groups were relatively small, and values at the end of treatment were within normal ranges.

A randomized, open-label, single-center study by Junge and coworkers reported a comparison of a multiphasic formulation combining E₂V (1–3 mg) and dienogest (DNG) (2–3 mg) with a triphasic formulation combining EE (30-40 µg) and LNG (50-125 µg) among 60 healthy reproductive aged women over a period of 7 cycles (37). Coagulation and fibrinolytic parameter values from baseline to end of treatment showed relatively small changes and remained within normal ranges in both groups.

These short-term studies comparing hemostatic parameters in women receiving either E₂ or EE, each combined with a different progestin, appear to indicate that the E₂- and EE-based COCs generally have minimal effects on most coagulation and fibrinolytic markers, and weaker effects on some markers with E2.

Recently, E₄ has gained interest as an option for oral contraception with a neutral effect on hemostatic parameters. In a randomized open-label exploratory study, Douxfils and coworkers (33) assessed hemostatic effects of an E₄-based COC compared to 2 different EE-based COCs. Thirty-nine premenopausal women received E₄ (15 mg)/DRSP (3 mg), 30 women received EE (30 µg)/LNG (150 µg), and 32 women received EE (20 µg)/DRSP (3 mg) for six 28-day cycles. Median changes in the coagulation factors fibrinogen, prothrombin, factor VII, factor VIII, and Von Willebrand factor were generally small, and not significantly different between the 3 groups, with the exception of factor VII which was significantly increased with EE/DRSP compared to the other 2 formulations. There was a 30% median change in the ETP-based APC sensitivity resistance at cycle 6 in the women receiving E₄/DRSP, which was significantly lower than in those receiving EE/LNG (105%) or EE/DRSP (219%). There were also small changes from baseline to end of treatment in the anticoagulant parameters antithrombin, protein S activity, free protein S, protein C, and the tissue pathway inhibitor (TFPI) with E₄/DRSP and EE/LNG. Changes in protein S activity, free protein S, and protein C were significantly greater in the EE/DRSP group compared to the E₄/DRSP and EE/LNG groups. Assessment of fibrinolytic markers (plasminogen, plasminogen activator inhibitor, tissue plasminogen activator) indicated a weak impact of E₄/DRSP, but any potential hypo-fibrinolytic or hyper-fibrinolytic profile of E₄/DRSP impacting VTE risk could not be determined from these data.

Morimont and coworkers (38) conducted a randomized, open-label parallel study to compare the effects of E₄ (15 mg)/DRSP (3 mg) to EE (20 µg)/DRSP (3 mg) and EE (30 µg)/LNG (150 µg) on thrombin generation over 6 treatment cycles in a cohort of women. They concluded that E₄/DRSP had no impact on thrombin generation measured by lag time, peak, time to peak, ETP, and mean velocity rate index, whereas the EE-containing COCs were associated with a shift to a prothrombotic state.

According to data from the above short-term comparative studies of the effects of different types of estrogens on hemostatic parameters, the E₂- and EE-based COCs generally have similar effects on coagulation and fibrinolytic markers while COCs containing E₄ seem to have less impact on hemostasis, as summarized in Table 1 .

In the absence of standardized assay methods to evaluate coagulability status, there has been misinterpretation of data obtained with 2 activated protein C (APC) sensitivity assays, which quantify the effects of APC on the aPTT and ETP (39). While the aPTT-based APC resistance assay measures the time it takes for blood to clot with and without the addition of APC to reflect reduced sensitivity to APC, the ETP-based APC resistance assay measures the total amount of thrombin generated in the blood sample over time with and without addition of APC. The aPTT assay is highly influenced by prothrombin and factor VIII levels, while the ETP assay is more sensitive to free protein S and free TFPI levels (39, 40). As the latter factors are much more influenced by the former ones, it may in part explain the inconsistent results between the 2 functional APC assays (39). In addition, the normalized APC sensitivity ratio (nAPCsr) has been shown to predict the risk of VTE where this ratio increased (41). Previous studies using this nAPCsr support that E₄/DRSP may be correlated with lower VTE risk compared to EE/LNG and EE/DSG. This evidence suggests the use of the ETP-based APC resistance assay to identify VTE risk among women who are taking estrogen (39, 40, 42, 43). APC resistance is identified among individuals with factor V Leiden, the most common hereditary thrombophilia. APC resistance also occurs with other hereditary (some rare F5 variant) and acquired causes (including hormone-induced, solid tumor, hematologic malignancies). The International Society on Thrombosis and Haemostasis Scientific and Standardisation Committee recommends the ETP-based APC resistance to detect the acquired APC resistance induced by COCs; in particular, the current aPTT-based assays are only sensitive towards factor V Leiden mutation due to the introduction of deficient plasma in the test (44).

Several studies quantified hemostasis in order to examine the association of estrogen used in menopausal HT with thrombosis risk. A meta-analysis of 48 studies including 40–68 years old postmenopausal women (6,229 HT users and 24,974 non-users) explored the association between HT use on coagulation factors (45). The study concluded that HT was associated with significantly decreased fibrinogen, factors VII, antithrombin, protein C, protein S and significantly increased plasminogen levels.

When comparing different type of estrogen, Blondon and coworkers reported from a retrospective cohort study of 140 postmenopausal women that women using E₂ had significant lower thrombin generation and higher total protein S, which suggests less prothrombotic status compared to CEE (46). Studies on the effect of E₄ use in menopausal HT on hemostasis are lacking.

Routes of estrogen administration for menopausal HT have shown different hemostasis effects. A double-blind placebo-controlled study among 152 menopausal women randomized to oral E₂ (1 mg), oral E₂ (1 mg) with gestodene (25 µg), transdermal E₂ (50 µg), or placebo showed a significant increase in nAPCsr in all treatment groups (47). It is important to note that the increase in nAPCsr was significantly higher in the oral compared to transdermal E₂ group. While there were significant changes in fibrinogen, factor VII, thrombin-antithrombin III complexes, tissue-type plasminogen activator and D-dimer with oral E₂, no significant changes of these markers were identified in the transdermal E₂ group (48). The data suggest that transdermal estrogen has a minor effect on hemostasis compared to oral estrogen.

To explore the effect of estrogen doses on hemostasis, Lobo and coworkers (49) measured hemostatic factors in 749 postmenopausal women who were randomized to different doses of CEE (0.625 mg, 0.45 mg, or 0.3 mg) with and without MPA for one year. Overall, CEE was found to increase plasminogen activity and decrease plasminogen activator inhibitor 1 activity, antithrombin III activity, and protein S. The study suggests that lower doses of CEE tend to induce more favorable changes in hemostasis markers towards lower risk of thrombosis. Regarding doses of E₂ and hemostasis, Eilertsen and coworkers (50) conducted a randomized open-label, comparative study of E₂ (2 mg)/norethisterone acetate (NETA) (1 mg), E₂ (1 mg)/NETA (0.5 mg) among 202 menopausal women. D-dimer increased markedly in the conventional-dose HT group but remained unchanged in the low-dose HT group. The reductions in both clotting factors and inhibitors were markedly more pronounced in the conventional-dose HT group compared to the low-dose HT group. The study suggests that low-dose HT is associated with less activation of coagulation than conventional dose HT.

To determine whether biomarkers of thrombosis can identify menopausal women at risk of VTE, a nested case-control study (51) was carried out using the Women’s Health Initiative trial data (52, 53). From the total of 27,347 menopausal women randomized to treatment with CEE with or without MPA or placebo, 215 women who developed thrombosis and 867 women without thrombosis at 1 year were included in the study. The study found that women with thrombosis had lower protein C and free protein S, and higher D-dimer, prothrombin fragment 1 + 2, and PAP. Among these markers, D-dimer was most strongly related to VTE (OR, 6.0; 95% CI, 3.6-9.8). From these findings, the study suggested the potential for clinical use of D-dimer testing to evaluate the risk for thrombosis in menopausal women before prescribing HT.

The increased procoagulant factors and decreased anticoagulatory mechanisms observed with estrogen use have been linked with an increased risk of thrombosis, this includes both venous and arterial thrombosis. The most frequent clinical sign of estrogen-related thrombosis is VTE, affecting the deep veins of the legs or the pulmonary arteries, typically within the first few months of use. Estrogen has also been associated with an increased risk of thrombosis at uncommon sites and with arterial thrombosis (34). The association between estrogens and thrombosis among COC and menopausal HT users is discussed separately.

Recently, there has been an increasing interest in how the types and delivery routes of estrogen influence the risk of VTE. Some studies have proposed that non-oral routes of administration may result in lower serum estrogen concentrations and fewer systemic effects than oral administration. We summarize the studies related to estrogens used for contraception on risk for thrombosis in Table 2 .

For VTE, a systematic review and meta-analysis of 19 clinical trials among 40,410 postmenopausal women (89) reported a similar VTE risk when menopausal HT was initiated within 10 years versus 10 years or more since menopause. Among postmenopausal women within 10 years since menopause, the pooled VTE risk was 1.74 (95% CI, 1.11–2.73) and among those who were 10 years or more since menopause, the pooled VTE risk was 1.96 (95% CI, 1.37–2.80).

For risk of arterial thrombosis, most importantly coronary heart disease (CHD), the timing of HT initiation was associated with the risk. A pooled analysis of 23 clinical trials among 39,049 menopausal women with 191,340 women-years of follow-up (88), showed a different CHD risk in younger versus older menopausal women. While HT significantly reduced CHD in menopausal women less than 60 years old (OR, 0.68; 95% CI, 0.48-0.96), HT had no significant association with CHD in menopausal women older than 60 years old (OR, 1.03; 95% CI, 0.91-1.16). Another meta-analysis of 19 trials with a total of 40,410 menopausal women (89) reported consistent findings of a significant reduction of CHD in younger menopausal women (OR, 0.52; 95% CI, 0.29-0.96) and no effect of HT on CHD in older menopausal women (OR, 1.07; 95% CI, 0.96-1.20).

This review summarizes the mechanism of action, pharmacokinetics, and potency of estrogen to explain the effect of estrogen on hemostasis and risk of thrombosis. Estrogen affects changes in hemostasis that involve complex coagulation, anticoagulation and fibrinolysis cascades, and could explain the estrogen effect on thrombosis. The association of estrogen on thrombosis risk when used for contraception and menopausal HT among healthy women varies by type of estrogen, estrogen dosage, and route of estrogen administration.